key: cord-0762942-y128d6hb
authors: nan
title: Society of Surgical Oncology SSO 2020 – International Conference on Surgical Cancer Care
date: 2020-03-02
journal: Ann Surg Oncol
DOI: 10.1245/s10434-020-08278-z
sha: b5f002b942937b3a7099c1d591dd7770d4d17b3d
doc_id: 762942
cord_uid: y128d6hb

nan

Background For gastrointestinal stromal tumors (GISTs) amenable to macroscopically complete resection, risk of recurrence and decisions on adjuvant therapy administration are based upon tumor size, location, mitotic rate, and tumor genotype. To improve risk stratification, we explored gene expression analysis for potential biomarkers to predict clinical outcomes. Methods We characterized transcriptional profiles in clinically annotated GIST lesions attempting to identify risk-associated gene expression. The biological relevance of identified markers was assessed in vitro using CRISPR-mediated gene disruption. For lead biomarkers, we developed immunohistochemical (IHC) based assays to detect expression in FFPE samples from a cohort of archival cases. Results The transcription factor HAND1 was differentially expressed between localized and recurrent or metastatic GIST samples by RNA-seq, with unique expression in recurrent or metastatic tumors. HAND1 was found to be necessary for normal GIST cellular proliferation in vitro. In an analysis of 331 archival GIST samples, HAND1 was expressed in tumors harboring KIT mutations (118/235, 50%), but not those bearing PDGFRA (0/30) or SDH subunit mutations (0/12). HAND1 expression was seen primarily in tumors arising from the small bowel (95 of 115, 82%), and less frequently in gastric tumors (10/153, 7%). HAND1 positivity was associated with tumor risk stratification, and relapse free survival (RFS) following surgical resection was significantly worse in primary tumors expressing HAND1 (hazard ratio 2.13, 95% CI 1.01-4.49), with a 5-year RFS rate of 48% in HAND1-positive and 70% in HAND1-negative tumors. In the metastatic setting, HAND1 was expressed in a majority (55/92, 60%) of tumors. Conclusions We identified the transcription factor HAND1 as a potential biomarker of high-risk and metastatic GIST. Additional research is required to determine whether HAND1 expression might add independent information to clinical decisions regarding the use of adjuvant kinase inhibitor therapy.

Introduction: Cholangiocarcinoma (CCA) is an aggressive malignancy with no approved targeted agents to date. Our group previously identified the overexpression of Exportin 7 (XPO7) in CCA-associated biliary exosomes. Here, we evaluate the prognostic significance of XPO7 in CCA, and investigate its biology to identify therapeutic targets. Methods: Immunohistochemical (IHC) staining was completed on a human CCA tissue microarray (TMA). Short hairpin RNA-mediated knockdown of XPO7 in human CCA cell lines was used to assess its phenotypic significance. Mass spectrometry was used to identify XPO7 binding partners immunoprecipitated from CCA tumorspheres. Screening of small molecule SLK inhibitors was completed using the Reaction Biology platform. Results: IHC evaluation of a human CCA TMA (n=161) demonstrated an association between cytoplasmic XPO7 and reduced patient survival (median OS 23 vs 41 months, p=0.036). XPO7 knockdown leads to reduced tumorsphere growth and orthotopic tumor size in mice (median 4 vs 1 mm, p=0.009). Moreover, tumors that developed from XPO7 knockdown cells displayed markedly divergent histology, reminiscent of well-differentiated, cystic tumors. Interrogation of XPO7's interactome revealed that it imparts this phenotype through the cytoplasmic shuttling and stabilization of Ste20-like kinase (SLK). Knockdown of SLK, in turn, results in markedly abrogated tumor growth (median 4 vs 0.5 mm, p<0.001) and altered tumor morphology through reduced activation of several oncogenic targets, including AKT. To validate this relationship in human tumors, we stained the CCA TMA and demonstrated an association between cytoplasmic XPO7 positivity and AKT activation (median p-AKT IHC score 25 vs 11, p<0.001). Next, we conducted a high-throughput screen to identify small molecule inhibitors with activity against SLK. In doing so, we identified GZD-824, and demonstrated its potent cytotoxicity toward XPO7-expressing CCA cell lines. Conclusions: Biliary exosomal XPO7 can be used as a liquid biomarker for aggressive biology in CCA. Tumors with high XPO7 expression appear to be vulnerable to SLK inhibitors. Future work will center on pre-clinical and clinical evaluation of SLK inhibitors.

Purpose: Using the long-term survival data from the C9343 trial as a temporal reference point we sought to determine updated radiation therapy (RT) treatment trends for older patients with early stage breast cancer before and after the long-term data were published. We also examined rates of adherence to adjuvant endocrine therapy (AET) as this is a pertinent factor when clinical decisions are made to omit RT. Methods: The Surveillance, Epidemiology, and End Results (SEER)-Medicare database was utilized to identify women diagnosed with breast cancer from 2007-2016 whose clinical characteristics matched the inclusion criteria of the C9343 trial. Bivariate associations were calculated to determine variable characteristics by time frame (Group 1: 2007 (Group 1: -2012 vs Group 2:2013 Group 2: -2016 . We used multivariate logistic regression to estimate the effect of group on use of RT and AET adherence. Temporal rates of both RT and AET adherence at 1 year, 2 years, and 3 years were plotted. Results: The final study cohort included 12,222 Medicare beneficiaries. There was a statistically significant difference in RT use between groups with a higher proportion omitting RT in the later time period (25% Group 2 versus 20% Group 1, p <0.001). After adjusting for covariates, in both groups, those who omitted art were statistically less likely to adhere to AET (Group 1: OR 0.74, p <0.001; Group 2: OR 0.66, p = <0.001). In all study years, women who omitted RT have statistically significant lower rates of 1, 2, and 3-year AET adherence. Conclusion: Fifteen years following the publication of the of the C9343 results suggesting older women with early stage cancer can safely forgo RT and subsequent NCCN guideline revision, our findings suggest that there has been minimal change in clinical practice, as most patients receive RT. Importantly, AET adherence was significantly lower in the non-RT group, and decreased over time from year 1 to year 3. For women who meet criteria for omission of adjuvant RT, non-adherence of AET could result in undertreatment of their breast cancer and RT should not be considered overtreatment in this population.

A two-sample t-test was used to determine if there is a significant difference between the means of two groups Introduction: Choosing Wisely® is an initiative from the American Board of Internal Medicine Foundation aiming to reduce the use of unnecessary tests and treatments across specialties using evidence-based recommendations. In 2016, SSO recommended against the routine use of sentinel lymph node biopsy (SLNB) in clinically node negative women 70 years old with early stage hormone receptor positive breast cancer. The extent to which SLNB has been de-implemented in this population remains largely unknown. Methods: Using the Michigan Radiation Oncology Quality Consortium (MROQC) registry, we evaluated rates of SLNB in patients 70 years old undergoing breast conservation therapy for early stage (I, II) hormone receptor positive (ER or PR) breast cancer. MROQC includes 26 radiation practices (5 academic, 21 community) across the state of Michigan. All patients received adjuvant whole breast radiotherapy as an MROQC eligibility requirement. Results: Between 2012 to 2018, 1633 patients meeting the eligibility criteria for SLNB omission were treated at MROQC sites. SLNB utilization remained high throughout the study period with more than ninety percent of patients undergoing SLNB in each study year. Annual site-specific utilization ranged from 66.7% to 100%. Conclusion: SLNB remains highly utilized in women 70 years old with early stage hormone receptor positive breast cancer after release of the Choosing Wisely recommendation. Persistently high utilization suggests either misalignment regarding the value of SLNB in these scenarios between SSO and the clinical treatment team, inadequate dissemination of recommendations, or other patient, provider or organization barriers to de-implementation.

Background The impact of length of time to surgery (TTS) on oncologic outcomes following neoadjuvant chemotherapy (NAC) in breast cancer patients is unclear. This study investigates the relationship between TTS on residual cancer burden (RCB) score and ultimate oncologic outcomes. Methods Females with nonmetastatic breast cancer who received NAC from January 2011 to December 2017 were identified. The effect of TTS on recurrence, survival, and RCB score was examined with multivariate analysis, adjusting for clinical stage at diagnosis, lymphovascular invasion (LVI), receptor status, and RCB score/final pathologic stage. Descriptive statistics and Cox proportional hazards models were used. Results We identified 463 patients. Median TTS was 29 days (range 11-153). Median follow-up was 47 months (range 2-93 months). Five-year estimates for overall survival (OS), recurrence-free survival (RFS), and locoregional recurrence-free survival were 82.6%, 74.4%, and 91.4%, respectively. On multivariate analysis, TTS 4-6 weeks was associated with worse RFS (HR 1.998, p<0.01) . TTS >6 weeks was strongly associated with worse RFS (HR 4.614, p<0 .0001) and OS (HR 2.461, p<0.01) . Mean RCB score was higher in patients with TTS >6 weeks compared to <4 weeks (2.192 vs 1.426, p<0.0001) . On multivariate analysis, TTS >6 weeks was associated with a positive size of effect (SOE) on RCB score of 0.59 (p<0.0001). Additionally, hormone receptor positive disease, LVI, and clinical stage III disease were associated with higher RCB score (SOE of +1.02, +0.89, and +0.70, respectively; p<0.0001 for all). Human epidermal growth factor receptor 2 (HER2) positive disease was associated with lower RCB score (SOE -0.89, p<0.0001) and on subgroup analysis was the only group not impacted by longer TTS. Conclusion Increasing TTS beyond four weeks is a modifiable risk factor for adverse oncologic outcomes following NAC for breast cancer. Longer TTS is strongly associated with higher RCB score and likely mediates the association with RFS and OS. Surgery should be performed within 4 weeks following NAC for optimal oncologic outcomes in HER2-negative patients.

Introduction NAC has been proven to increase rates of BCS in randomized trials, but conversion rates from BCS-ineligible (BCSi) to BCS-eligible (BCSe) in pts with an unfavorable tumor-to-breast ratio (UTBR) are unknown. We sought to evaluate conversion to BCS eligibility with modern NAC and identify pts likely to benefit. Methods Consecutive pts with stage I-III breast cancer treated with NAC from 11/2013-03/2019 were identified from a prospective database. BCS-eligibility pre-/post-NAC was prospectively determined by the treating surgeon. Patients deemed BCSi before NAC due to UTBR were studied. NAC included ddAC-T in 93%, and dual blockade in 99% of HER2+ pts. Univariate associations were evaluated using t-tests and chi-square tests; independent predictors were evaluated using logistic regression. Results Of 1354 pts, 626 were BCSi pre-NAC with UTBR; 69% were non-BCS candidates, 31% were borderline-BCS (bBCS) candidates. Compared to bBCS candidates, non-BCS candidates had larger tumors (4.9cm v 3.6cm, p<0 .001), non-ductal histology (p=0.009), and cN+ status (p=0.003). Of non-BCS candidates, 66% became BCSe after NAC; 66% chose BCS, and 90% were successful. Among bBCS candidates, 86% were BCSe after NAC, with 73% choosing BCS with a 96% success rate. On univariable analysis, bBCS candidacy, lower cT stage, cN0 status, absence of mammographic calcifications, HER2+/triple negative (TN) receptor status, poor differentiation, ductal histology, and breast pCR were predictive of conversion to BCS-eligibility. On multivariable analysis, receptor status was independently associated with post-NAC BCS-eligibility (HR+/HER2-ref, odds ratio [OR] HER2+ 2.15; HR-/HER2-1.97, p=0.008), while non-BCS candidacy (OR 0.36, p<0 .001), cN+ status (OR 0.55, p=0.004), and mammographic calcifications (OR 0.52, p<0.001) predicted failure to downstage (Table) . Conclusion In pts with UTBR, conversion rates to BCS eligibility are high with NAC, particularly in bBCS candidates. HER2+/TN receptor status predicts for successful downstaging to BCS, while failure is associated with cN+ status and mammographic calcifications; these factors should be considered when selecting pts for NAC. Table. Predictors of conversion to BCS-eligibility with NAC in patients presenting with unfavorable tumor-to-breast ratio §unknown: n=17 clinical tumor size, n=1 differentiation, n=1 histology. *Other includes invasive mucinous (n=1), metaplastic (n=4), and anaplastic (n=1). Abbreviations: NAC, neoadjuvant chemotherapy; BCS, breast-conserving surgery; CI, confidence interval; HR, hormone receptor; pCR, pathologic complete response Introduction Neoadjuvant chemotherapy (NAC) downstages the axilla in up to 80% of patients (pts) depending upon breast cancer phenotype. This allows for the potential to perform a sentinel node biopsy as compared to an axillary dissection, significantly reducing morbidity. ACOSOG Z1071 established that dual tracer-guided sentinel lymph node biopsy (tr-SLNB) has a false negative rate (FNR) of >10%, but tr-SLNB combined with resection of the lymph node (LN) marked with a clip at initial presentation can decrease the FNR. Directed SLNB (d-SLNB) with I-125 radioactive seed localization (RSL) of the biopsy proven LN is a method to ensure retrieval of the clipped LN at the time of surgery. We hypothesize that d-SLNB alone can determine axillary status after NAC. Methods A single-institution, prospective NAC database was retrospectively reviewed for pts with percutaneous biopsy proven axillary LN metastasis pre-NAC. Pts received NAC and underwent combined d-SLNB and tr-SLNB from July 2014-August 2018. All pts received dual tracer technetium(Tc-99) and blue dye for tr-SLNB combined with d-SLNB. Demographics, clinicopathologic features and surgical outcomes were evaluated. Results 151 pts underwent a d-SLNB with a mean age of 51. d-SLNB was successful in 146/151 (97%) pts. In 131 pts for whom tracer uptake status of the d-SLNB was known, 23 (18%) had no tracer uptake in the RSL node. For the successful d-SLNB cohort, 73/146 (50%) had an axillary pathologic complete response (pCR). Of the 73 pts who did not have axillary an pCR, 68 were persistently node positive on d-SLNB and 5 had a negative d-SLNB but had other positive axillary LNs identified on either tr-SLNB or axillary dissection. Therefore, d-SLNB alone accurately characterized the axilla in 141/146 better determine the mechanism of resistance, single cell RNA-sequencing of a parent tumor identified three distinct populations, and analysis from one cluster identified 12 functional pathways with altered gene expression, including immune pathways, transcriptional changes, and ribonucleoprotein interactions. The analysis also identified a limited population of cells that expressed ERBB2, ESR1, and FGFR1, genes associated with anti-estrogen therapy resistance, underscoring the necessity to compare expression signatures observed in CTCs with those observed in primary tumors. Conclusion: FISH-CTCs found in patients on anti-estrogen therapy represent the best and earliest opportunity to evaluate molecular mechanisms of resistance. The molecular characterization of these cells may allow the addition of targeted secondary therapies to decrease clinical recurrence.

Breast Cancer D. Liang,* C. Hall, S. Meas, B. Narendran, V. Sarli, A. Lucci. MD Anderson Cancer Center, Houston, TX. Background: Inflammatory breast cancer (IBC) is a rare, aggressive subtype of breast cancer with a poor prognosis, which may be due to the dissemination of micrometastatic cancer cells that are responsible for relapse and distant metastasis. It was previously shown that the presence of circulating tumor cells (CTCs) in stage III IBC patients after neoadjuvant chemotherapy (NACT) was associated with shortened relapse-free survival (RFS), although there was no correlation with pathologic complete response (PCR) rates. In this pilot study, we analyzed IBC patients' blood samples to enumerate CTCs prior to initiation of NACT and after completion of NACT to determine if changes in CTCs can predict PCR, overall survival (OS), and RFS. Methods: We performed a retrospective analysis of data collected in a prospective IRB-approved study that analyzed blood for CTCs in 51 stage III IBC patients using the CellSearch TM System. Changes in CTC counts were correlated with RFS and OS using Log-rank test and Cox regression analysis. Results: Of 51 patients, 10 patients had an increase in the number of CTCs when comparing before and after NACT time points. There was no statistically significant difference in PCR rates of patients whose CTCs increased with NACT compared to patients with no increase in post NACT CTCs (30.0% vs. 43.9%, p=NS). However, during the median follow-up time of 1.7 years, patients with increase in CTCs had significantly lower OS (log-rank p=0.012, hazard ratio 4.6, 95% CI 1.14-18.59) as well as a trend of lower RFS (p=NS). Furthermore, of 20 patients who had one or more CTCs prior to initiation of NACT, 14 patients no longer had any CTCs after completion of NACT. For these patients, although not statistically significant, there was a trend for improved survival, when compared to those who had persistent CTCs. Conclusions: Changes in CTC enumeration in response to systemic therapy may have prognostic significance in IBC patients. We are accruing additional patients to determine if CTCs can be used to monitor response to NACT and to identify patients at high risk of relapse who may benefit from additional adjuvant therapy. Figure 1 . Venn Diagram of 718 Differentially Expressed Genes: ER positive and TNBC A white color gradient (top) represents the upregulated genes, and a black color gradient (bottom) represents the downregulated genes. The circles of the diagram represent ER positive (left) and TNBC (right). There were 718 differentially expressed genes whose expression was statistically significant. From that group, 97 genes were upregulated in both tumor types (top middle), and 115 genes were similarly downregulated (bottom middle). In ER positive primary breast cancer, axillary lymph node (ALN) metastasis had 251 uniquely upregulated genes (top left), and 210 uniquely downregulated genes (bottom left). In TNBC, ALN metastasis had 26 genes uniquely upregulated genes (top right), and 19 uniquely downregulated genes (bottom right).

via methylation of promotors on DNA and aberrantly silences tumor suppressor genes related to oncogenesis. We hypothesize that the level of EZH2 and DNMT1 will correlate to the higher Oncotype DX (ODX) scores, a validated gene-based diagnostic for prediction of breast cancer recurrence and response to chemotherapy. After IRB approval, tumor specimen was obtained from 32 patients with ER/PR+, HER2-breast cancer who underwent resection with their respective ODX scores between 2009 and 2012. Patient slides were stained for EZH2 and DNMT1 expression and correlated to their ODX scores. MCF-7 and T-47D breast cell lines were treated with EZH2 inhibitor 3-Deazaneplanocin A (DZNep) and/or DNMT1 inhibitor 5-aza-2'-deoxycytidine . Cell proliferation assay was performed as well as gene expression of tumor suppressors via qRT-PCR. Of 32 slides, 7 patients had high ODX scores, 13 patients had medium ODX scores, and 12 patients had low ODX scores. High ODX scores had higher IHC staining intensity of EZH2 and DNMT1 expression than tumors with medium ODX score (p<0.01) and low ODX scores (p<0.01). Treatment of breast cell lines MCF-7 and T-47D with both DZNep and 5-AZA demonstrated reductions in cell proliferation by 31 and 58%, respectively (p=<0.01). Additionally, there was upregulation of tumor suppressors p53 (4.4 rel. fold; p<0.01) , and CDKN1b (1.7 rel. fold; p = 0.05) in T-47D and p53 (1.5 rel. fold; p=0.35) , and CDKN1b (5.6 rel. fold; p=0.29) in MCF-7. Epigenetic dysregulation has a direct correlation to higher ODX recurrence scores. Epigenetic therapy can reduce cell proliferation and upregulate previously silenced tumor suppressors by epigenetic dysregulation. These clinically available epigenetic therapies warrant further research for potential clinical trial implications for reducing recurrence in patients with early-stage breast cancer.

Background Tumor genomic prognostic assays estimate 10-year local recurrence risk in Ductal Carcinoma in Situ (DCIS) and can guide treatment decisions. Our aim is to evaluate which patients treated with breast conserving surgery (BCS) for DCIS underwent genomic assay testing with DCIS score and the influence of assay results on adjuvant treatment recommendations. Methods We identified patients with DCIS treated with BCS from the National Cancer Database (NCDB) from 2010-2016 and DCIS score use over time, assessing factors associated with testing and comparing adjuvant treatments in relation to score. Analyses included chi-square tests and multivariable logistic regression. Results Of 141,047 patients, 4,255 (3.0%) had DCIS score performed, increasing from 0.3% in 2010 to 5.8% in 2016 (p<0.001) . Patients more likely to undergo DCIS score assessment had more favorable tumor features: smaller size, negative margins, estrogen receptor (ER) positive and lower grade. Testing was less likely in: age 70+, Charlson-Deyo comorbidity score 1, black race, and those uninsured or with Medicaid. DCIS score testing result was documented in 3,888 patients (91.4%): 70.5% low-, 14.9% intermediate-, 14.6% high-risk. Patients with low-risk scores were less likely (each p 0.001) to have radiation (XRT) than those with intermediate or high-risk for both ER+ (35.0% vs 71.0% and 81.1%) and ER-disease (48.1% vs 77.0% and 85.5%). In ER+ disease, patients with high-and intermediate-risk score were most commonly treated with both XRT and hormone therapy (HT) (57.1% and 52.2%), while the most common treatment for those with low-risk DCIS score was HT alone without XRT (37.1%). Compared to patients without it, patients with DCIS score were less likely to get adjuvant therapies overall (Table 1) . Conclusion Use of the DCIS score increased over time, predominantly for favorable DCIS. Patients with a low-risk score were significantly less likely to receive radiation, supporting an impact of the DCIS score on decisions about treatment de-escalation. Rosenberger, 8 C. Parker, 9 K.K. Gallagher, 10 L.K. Jacobs, 11 S. Feldman, 12 P. Lange, 4 S.D. DeSantis, 4 S.J. Schnitt, 1 T.A. King. 1 1. Surgery, Brigham and Women's Hospital, Boston, MA; 2. Beth Israel Deaconess Medical Center, Boston, MA; 3. Memorial Sloan Kettering Cancer Center, New York, NY; Boston, MA; 5. Indiana University, Carmel, IN; 6. University of Pittsburgh, Pittsburgh, PA; 7. Georgetown, Washington, DC; 8. Duke University, Durham, NC; 9. University of Alabama, Birmingham, AL; 10. University of North Carolina, Chapel Hill, NC; 11. Johns Hopkins University, Baltimore, MD; 12. Montefiore, Bronx, NY. BACKGROUND: Intraductal papilloma (IP) is a common breast lesion with variable clinical presentations. Retrospective data suggest that asymptomatic IPs (i.e. without a palpable mass or pathologic nipple discharge) found on core biopsy (CB) are associated with carcinoma at excision in 0-12% of cases. We conducted a prospective multi-institutional trial (TBCRC 034) to determine the upgrade rate to ductal carcinoma in situ (DCIS) or invasive cancer at excision for asymptomatic IP on CB. METHODS: Patients (pts) with a CB diagnosis of IP and BI-RADS <=4 were prospectively identified and consented to excision. Cases deemed discordant or with additional lesions requiring excision were excluded. Upgrade rates to cancer were determined based on local and central pathology review. Sample size and confidence intervals were based on exact binomial calculations. RESULTS: All 116 registered pts underwent excision (median age 56 years, range 24-82 years). The most common imaging abnormality was mammographic density/distortion (n=62, 53%), all were BI-RADS<=4 and concordant. Per local pathology review, 2 cases (1.7%; 95% CI 0.2% to 6.1%) were upgraded to DCIS at excision. On central pathology review, IP was confirmed in 85 of 116 reviewed cases (73%, 95% CI 64%-81%). The remaining cases had benign changes other than IP (n=21) or atypical IP (n=2) or atypical ductal hyperplasia (ADH) adjacent to IP (n=8). Central pathology review of the excision specimens revealed no upgrades: the 2 locally upgraded cases were considered to have ADH on central review, although in 1 case not all excision slides were available for review. If the latter case were found to have DCIS on central pathology review, among 85 cases with confirmed IP without atypia on CB, the upgrade rate would be 1/85 (1.2%; 95% CI 0.03% to 6.4%). CONCLUSION: In this prospective study of 116 pts with IP on CB, the upgrade rate was 1.7% by local pathology review and up to 1.2% by central pathology review, suggesting that routine excision is not indicated for pts with IP without atypia or adjacent ADH on CB with concordant imaging findings. 1 1. Surgery, Brigham and Women's Hospital, Boston, MA; 2. Dana Farber Cancer Institute, Boston, MA. Background: The significance of residual axillary disease after neoadjuvant endocrine therapy (NET) for hormone receptor positive HER2 negative (HR+HER2-) breast cancer is uncertain. The objective of this study was to examine nodal disease burden, management and outcomes after NET among cN0-1 pts. Methods: Patients with stage I-III HR+HER2-breast cancer receiving NET, Dec 2015-Sept 2018, were identified from a prospectively maintained institutional database. A similar cohort of pts from the National Cancer Data Base (NCDB) 2010-2015 were identified for comparison and overall survival (OS) analysis. Results: From the institutional data, 95 (4%) of 2191 HR+HER2-pts received NET followed by surgery. Presenting nodal stage was cN0 in 71 (75%) and cN1 in 24 (25%). 46 (65%) cN0 pts underwent axillary surgery [Table] ; 31 (67%) were ypN0, 13 (28%) had 1-2 pos nodes and 2 (4%) had 3 pos nodes. 9/15 (60%) ypN+ pts underwent ALND. All cN1 pts underwent axillary surgery; 21/22 (95%) ypN+ pts underwent ALND and 2 (8%) achieved a nodal pCR. At a median follow up of 29mths, there have been no locoregional recurrences, and 4 distant recurrences (2 in cN0; 2 in cN1 pts). In the NCDB, 3640 (2%) of 215,530 HR+HER2-pts received NET followed by surgery. Presenting nodal stage was cN0 in 3031 (83%) and cN1 in 609 (17%). 2921 (83%) cN0 pts underwent axillary surgery [Table] ; 2208 (76%) were ypN0, 547 (19%) had 1-2 pos nodes and 161 (6%) had 3 pos nodes. 370/708 (52%) ypN+ pts had ALND. 603/609 (99%) of cN1 pts underwent axillary surgery; 450/513 (88%) ypN+ pts had ALND and 68 (11%) achieved a nodal pCR. At a median follow up of 37 months, there were no differences in 3-yr OS based on type of axillary surgery in cN0-1 pts with negative nodes (96% SNB vs 96% ALND) or 1-2 pos nodes (95% SNB vs 90% ALND). Conclusion: In cN0 HR+HER2-pts selected for NET, only ~5% had 3 pos nodes. In contrast, >40% of cN1 pts had 3 pos nodes and nodal pCR was rare. The absence of a survival difference by type of axillary surgery among cN0-1 patients found to have 1-2 positive nodes after NET suggests an opportunity to de-escalate treatment of the axilla post-NET in patients with limited nodal disease.

Introduction Metaplastic breast cancer(BC) constitutes a rare but unique histologic entity with poor prognosis. We hypothesized that the unique genetic make-up and tumor immune microenvironment are the key differentiating factors of metaplastic BC. Methods Metaplastic BC cases were identified through a total of 10,865 cases from the Cancer Genome Atlas Data Set(TCGA) and the AACR -GENIE(Genomics Evidence Neoplasia Information Exchange) project. The tumor infiltrated immune cells and CYT score, were estimated by RNA sequenced data. Baseline clinical characteristics were compared, and gene set enrichment analysis (GSEA) were performed. Results Out of 1102 breast cancer cases on the TCGA data set and 9763 on the GENIE cohort, 9 (0.8%) and 59 (0.6%) metaplastic subtypes were identified. Metaplastic BC was predominantly ER negative (7 (77.8%) vs 211 (22.7%), p=0.001), triple negative (5 (55.6%) vs 142 (15.5%), p=0.007), and node negative (8 (88.9%) vs 403 (43.4%), p=0.01) compared to non-metaplastic subtypes in the TCGA cohort. Tumor mutation burden was lower in the metaplastic subgroup compared to the non-metaplastic group, median: 0.4 vs. 1.6 /Mb in TCGA-TNBC (p=0.67) and 3.0 vs. 4.0 / Mb (p=0.1) in GENIE-cohort. TP53 mutation was more frequent in MBC (66.7 % vs. 31 % in TCGA (p=0.19 ) and 62.7 % vs. 38.8% in GENIE (p=0.002)). Cell proliferation related gene sets like MYC target and E2F targets were down-regulated in metaplastic BC compared to non-metaplastic BC (p=0.04 and p=0.05, respectively). However, increased intratumor heterogeneity (p<0.001) was observed in metaplastic BC. Signaling gene sets such as androgen response (p<0.001), estrogen response (p<0.001) and TGF-beta (p=0.02) gene sets were not enriched in the metaplastic group. Metaplastic BC had decreased lymphocyte infiltration (p=0.002) and increased M2 macrophage (p<0.001). Immune checkpoints such as CTLA4 and TIGIT were significantly less in the MBC group (p=0.05 and p=0.03, respectively). Conclusions High intratumor heterogeneity, less lymphocyte infiltration, and infiltration of M2 macrophage represent several likely factors leading to poor outcomes related with metaplastic BC. with BC; 83 patients (33.7%) had a personal history and/or strong family history of BC. We defined a weak family history as either having no relatives with DGC or only a single relative with DGC. Within this CDH1 high-risk BC cohort, 51 (61%) patients also had a weak family history of DGC. These patients came from 24 families, while the majority were Caucasian (94.1%) and female (82.4%). Median age of personal or family member BC diagnosis was 47 years (range 32-85). Of this 51 patient cohort, 50 went on to have surveillance endoscopy and/or prophylactic total gastrectomy and 68% (N=34/50) were found to have signet ring cells (SRCC, T1a) on pathology. This compares to an 86% rate of finding SRCC on CDH1 patients undergoing prophylactic total gastrectomy regardless of family history. Conclusions: CDH1 families with predominant BC and weak DGC family history remain at high risk of developing early stage gastric cancer as evidenced by high rates of gastric SRCC pathology. Therefore, CDH1 carriers with even a limited family history of DGC should still receive counseling regarding elevated gastric cancer risk. Purpose: To report an interim analysis of a trial of WHBI following breast conserving surgery for stage 0-II breast cancer. Methods: Patients had negative margins, 0-3 positive axillary nodes and met prespecified criteria for being medically underserved. There were no restrictions on age, breast size, grade, receptor status, or systemic therapy. WHBI was delivered to the whole breast using tangential beams with no elective coverage of lymph nodes. Dose was initially 30 Gy in 5 weekly fractions but later reduced to 28.5 Gy following a protocol amendment. A partial breast boost was permitted. The primary endpoint was ipsilateral breast tumor recurrence (IBTR). Secondary endpoints were distant disease-free survival (DDFS), recurrence free survival (RFS), overall survival (OS), adverse events and cosmesis. Kaplan-Meier method was used to estimate cumulative incidence in the absence of competing risk. Multivariable analysis using Cox regression was used to investigate the prognostic significance of clinicopathologic factors. Results: From 2011 to 2015, 158 received WHBI and were eligible for analysis. Median follow up was 4.3 years (range, 0. 2-8.1) . Stage distribution was DCIS 22%; invasive pN0 68%; invasive pN1 10%. Median age was 59 (range, 30-84) . 83.8% were hormone receptor positive, 63.3% were grade 1/2, 31% received chemotherapy and 12.7% a boost. 80 patients received 30 Gy and 78 received 28.5 Gy with median follow up times of 5.6 and 3.7 years, respectively. There were 5 IBTR events, all in the 30 Gy group. [5] [6] [7] year risks of IBRT for all patients were 2.2% (95% CI 0. 6-5.8 ) and 6.0% (95% CI 1. 1-17.2) , respectively. The 7-year rates of DDFS, RFS, and OS were 96.3%, 91.5% and 93.7%, respectively. Improvement in IBTR-free time was seen in DCIS, lobular histology, low grade, and 28.5 Gy dose (all p<0.0001). Toxicities are reported in the Table. Cosmetic outcome was good/excellent in 78.5% and fair/poor in 21.5%. Conclusions: Disease-specific outcomes after WHBI are favorable and parallel those seen with daily irradiation and accelerated partial breast irradiation. Adverse events and cosmetic outcomes are acceptable.

Discharge T. Moo,* M. Assel, R. Yeahia, R. Nierstedt, K. Van Zee, L. Kirstein, M. Morrow, R. Twersky. Memorial Sloan Kettering Cancer Center, New York, NY. Background Opioid analgesics are overprescribed after surgery. In 8/2018 we replaced routine discharge opioid prescription with an NSAID for patients having lumpectomy or excisional biopsy (lump/ex). Opioids were prescribed only for medical contraindications to NSAID or clinical reasons. Here we compare patient-reported post-discharge pain scores among patients treated before and after the change in routine discharge medication. Methods All breast surgery patients are sent an electronic survey on postoperative day (POD) 1-5. Pain is assessed daily as: 0=none; 1-3=mild; 4-6=moderate; 7-8=severe; 9-10=very severe. We used multivariable generalized estimating equations to test association between pain severity and discharge in the first (routine opioids) vs second (routine NSAID) study period. Results 1606 patients had lump/ex between 12/2017-6/2019. 789 (49%) reported pain scores and were analyzed (328 first period, 461 second). Demographic/intraoperative variables were similar in both periods, except for higher intraoperative acetaminophen use (97% vs 94%) and lower post-anesthesia care unit narcotic dosing after adoption of routine NSAIDs (Table) . Opioid prescription at discharge decreased from 96% in the first study period to 14% in the second. 1% of patients discharged with NSAID were later prescribed an opioid. Maximum reported pain score on any POD for all patients was: very severe 0%; severe 3.8% (30); moderate 28% (217); mild 54% (430); none 14% (112). There was no difference in proportion of patients reporting any pain (mild-severe) among those treated in the first vs second study period (OR 1.02, p=0.9) . Further, there was a non-significant trend toward more frequent reports of moderate/severe pain among patients discharged with an opioid prescription vs NSAID over both time periods (OR 1.15; p=0.3) . Conclusions In patients undergoing lump/ex, we found no clinically meaningful difference in reported post-discharge pain scores between the study periods where routine opioids or NSAIDs were prescribed at discharge. Routine discharge with opioid analgesics offers no benefit in patients undergoing lump/ex. Table 1 . Demographic and perioperative treatment characteristics Abbreviations: MME, morphine milligram equivalents; NSAID, nonsteroidal anti-inflammatory drug; IQR, interquartile range; BMI, body mass index; ASA, American Society of Anesthesiologists; monitored anesthesia care; PACU, post-anesthesia care unit Background: Cancer surgery, while necessary, contributes to recurrence and metastases. We established that Natural Killer (NK) cells are greatly suppressed after surgery, with surgery-induced myeloid derived suppressor cells (MDSCs) in a mediating role leading to increased metastases. Peri-op arginine immunonutrition reduces post-op complications, but how arginine relates to MDSCs and post-op NK function is not known. Methods: To assess the peri-op role of arginine we used two murine models of surgery-induced lung metastases (implanted B16F10 melanoma or spontaneously metastasizing 4T1 breast cancer) in which mice were fed either a control or arginine enriched diet (AED). Surgical stress was established via laparotomy and nephrectomy. For human studies, blood was collected from cancer patients pre-/ post-op. NK function (cytotoxicity, IFN production, markers) and MDSCs were assessed by flow cytometry. Arginine levels were measured via LCMS. Results: After surgery, a rapid drop in arginine levels was observed in mice (180 to 95 mol, 4h post-op) and patients (27 to 10 mol, 24h post-op) with a concomitant increase in MDSCs (>2-fold in mice and humans). These MDSCs also had higher activity of Arginase-1. In mice, pre-op depletion of myeloid cells ( Gr1) or inhibition of Arginase-1 (with CB1158) prevented the drop in arginine, suggesting that MDSCs are regulators of post-op arginine. AED fed mice had 1.7-fold higher arginine than control diet mice. In both mouse models there was a significant reduction in post-op lung metastases in AED Introduction: Colorectal cancer is a common malignancy that can be cured when detected early, but treatments for advanced disease are needed. Immunotherapies targeting tumor-specific antigens are an emerging option, and human endogenous retroviruses (HERVs) can encode such antigens. However, absence of annotated HERV sequences from reference assemblies of the human transcriptome is a major barrier to target discovery. Methods: Sequences for 91 HERV-K transcripts were obtained from NCBI and incorporated into a custom transcriptome based on the GRCh38 cDNA reference, and a joint analysis of pooled transcriptomic data from all available bulk RNA-seq datasets of healthy, histologically normal tumor-adjacent, and tumor colorectal tissue was performed. Transcript abundance was estimated with Salmon controlling for GC bias, and downstream analysis was performed in R 3.5.1. Results: Expression levels of annotated human transcripts as well as HERV-K transcripts were assessed in 834 independent samples from eight high-quality datasets, including 462 healthy, 61 tumor-adjacent, and 311 tumor samples. Most HERV-K transcripts were expressed at levels lower than a standard deviation below non-HERV transcripts across all phenotypes, but six were expressed within a standard deviation of median non-HERV transcript expression. Of those, all but one were significantly under-expressed in tumoradjacent and tumor samples. HK_3q12.3 had the highest median expression in healthy samples and had the second-largest fold change decrease in tumor samples. A single transcript, HK_8q24.3b, had moderate tumor-specific over-expression (FDR-adjusted p=0.023), and a subset of high-outlier tumor samples was observed. Conclusions: Novel insights into HERV-K expression were generated in this mega-analysis of colorectal transcriptomes, which is the largest unbiased assessment of HERV transcripts in colorectal tissue to date. Several HERV-K transcripts were consistently under-expressed in tumor samples, which could suggest a functional role for HERVs in maintenance of healthy colorectal epithelium, and a single transcript was specifically overexpressed in tumor tissue, providing a potential target for immunotherapy.

Introduction Despite improvements in the multimodality treatment for patients with locally recurrent rectal cancer (LRRC), oncological outcomes remain poor. This study evaluates the effect of induction chemotherapy and subsequent chemo(re)irradiation on the pathologic response and the rate of resections with clear margins (R0 resection) in relation to the long-term oncological outcomes. Method All consecutive patients with LRRC who underwent an intentionally curative resection in the Catharina Hospital Eindhoven between January 2010 and December 2018 after treatment with induction chemotherapy and subsequent chemo(re)irradiation were identified from a prospectively kept database and retrospectively reviewed. Induction chemotherapy generally consisted of three cycles CAPOX or four cycles FOLFOX. Chemoradiotherapy was administered according to standard of care. Endpoints were the extent of pathologic response, R0 resection rate and the overall, disease free, local recurrence free and metastasis free survival (OS, DFS, LRFS, MFS) . Results A pathologic complete response (pCR) was observed in 22/132 patients (17%), a "good" response ( in 74 patients (56%) and a "poor" response in 27%. A R0 resection was obtained in 83/132 patients (63%). Three-year overall survival was 92% for patients with a pCR. The degree of pathologic response was linear correlated with the R0 resection rate. After uni-and multivariable analyses, pathologic response remained an independent predictor Background: Prognostic significance of circulating tumor cells (CTCs) in colorectal cancer is controversial. Tumor drainage mesenteric venous blood is an ideal sample without dilution or filtration by the liver. This study aimed to investigate clinical significance of CTCs in the drainage mesenteric venous blood in clinical stage II-III rectal cancer treated with and without neoadjuvant chemoradiotherapy (nCRT). Methods: Blood samples were collected intraoperatively before resection from the inferior mesenteric vein and peripheral vein in patients with clinical stage II-III rectal cancer who underwent curative resection with and without nCRT. Number of CTCs was measured per 7.5mL blood sample using the Cell Search System. Results: A total of 47 patients were enrolled. Twenty patients received nCRT with 5 achieving a complete response (20.0%). In an overall cohort, CTCs were more frequently detected in the mesenteric venous blood compared to the peripheral blood (48.9% vs. 8.5%, p<.001). Presence of CTCs and number of CTCs in the mesenteric venous blood were both associated with pathological T, N and TNM stages, lymphovascular invasion and preoperative CEA, while the association was not significant in the peripheral blood. Patients treated with nCRT had significantly lower number of CTCs in the mesenteric venous blood compared to patients without nCRT (median 0, IQR 0-0 vs. median 3, p<.001) . This finding was similarly observed when focusing on patients with pathological stage III disease (median 0, IQR 0-6 vs. median 4.5, p=0.032) or T3/4 disease (median 0, IQR 0-1.5 vs. median 4, IQR 1-13, p=0.018) . Conclusions: In clinical stage II-III rectal cancer, CTCs are frequently detectable in the mesenteric venous blood in contrast to peripheral blood, and CTCs in the mesenteric venous blood are a significant marker of advanced disease. Patients treated with nCRT have marked reduction of CTCs in the mesenteric venous blood even when focusing on patients with residual cancer. This finding may support oncological safety of long waiting after nCRT in the presence of residual cancer.

Introduction Tumor border configuration (TBC) is a prognostic factor in colorectal adenocarcinoma. However, the role of TBC is not well documented in colon adenocarcinoma. Our aim is to study the effect of TBC on overall survival (OS) and recurrence-free survival (RFS) in stage II and III colon adenocarcinoma. Methods This is a single institution retrospective cohort study of patients with pathologic stage II and III colon adenocarcinoma who were surgically treated at a tertiary medical center between 2004 and 2015. We excluded patients who had received neoadjuvant chemotherapy. We stratified the patients into four groups based on stage and TBC (Table 1) . A Cox regression model was used to control for cofounders including group, histologic grade, the American Society of Anesthesiologists (ASA) score, age, adjuvant therapy, and extramural vascular invasion. The outcomes of the study were OS and RFS. Results The cohort consisted of 725 patients (376 stage II and 349 stage III). The mean age of the patients was 68 15 years. There were 382 females (53%) and 342 males (47%). Infiltrating TBC was evident in 222 stage II cases (45%) and 268 stage III cases (55%). Infiltrating tumor border configuration was associated with increased extent of invasion (P<0.001), increased regional lymph node involvement (P<0.001), and extramural vascular invasion (P<0.001), but not histologic grade (P=0.7). Compared with pushing TBC, infiltrating TBC increased the hazard of death by a factor of 1.6 (1.1-2.4, P=0.01) and 1.6 (1.0-2.4, P=0 .002) in stage II and III patients, respectively. The hazard of death in patients with stage II disease (infiltrating TBC) and patients with stage III disease (pushing TBC) was not significantly different (aHR, 0.8; 95% CI, 0. 5-1.2; P=0.3) . Infiltrating TBC was also associated with worse recurrence-free survival compared to pushing TBC stage II disease (aHR, 1.5; 95% CI; 1.72.1, P=0.02) and stage III disease (aHR, 2.2; 95%; CI; 1.4-3.3; P<0.001). Conclusion Infiltrating TBC is a high-risk feature in patients with stage II and III colon adenocarcinoma. Stage II diease patients with infiltrating TB should be considered for adjuvant chemotherapy. Table 1 : Results of multivariable regression analysis for overall survival. * All comparisons are adjusted for the extent of invasion, regional lymph node metastasis, histologic grade, age, ASA, adjuvant therapy, and extramural vascular invasion.

Background: The management of newly diagnosed pancreatic neuroendocrine tumors (pNETs) is limited by a lack of sensitivity of existing biomarkers for prognostication. Our goal was to investigate the potential utility of genetic markers as a preoperative predictor of metastatic disease and postoperative outcomes. Methods: Patients from a prospectively maintained pNET database with fresh frozen surgically resected tissue available were identified. Whole exome sequencing was performed using the xGen exome research panel and libraries were prepared with the KAPA Hyper prep kit. Variant calling and genomic analysis were performed using the Broad GATK pipeline. Additionally, a further 98 pNET samples with genetic and clinicopathologic information available was obtained from the ICGC pNET cohort. Results: A total of 40 patients were identified with both normal and tumor tissue available for sequencing and sequencing resulted in a mean of 133.8 million reads and 99.2% exomic coverage. With a median of 4.5 years of follow-up, 9/40 (22.5%) patients developed a recurrence. Genomic analysis revealed similar patterns to prior studies of pNETs with few somatic gene mutations but numerous instances of copy number changes and complex phenomena including kataegis and chromothripsis. Analysis of genomic and clinicopathologic outcomes from a combined dataset of our study as well as the ICGC pNET cohort (n = 138 patients) revealed that genomic alterations including recurrent whole chromosome loss (RPCL, OR 1.3) as well as alterations in the TP53 (OR: 1.6) and PI3K (OR: 1.3) oncogenic pathways were independently associated with the risk of recurrence. A model of preoperative genetic predictors was able to predict the risk of recurrence with a balanced accuracy of 77%. Conclusions: Analysis of multiple genomic features revealed a recurrent pattern of whole chromosome loss associated with alterations in two specific oncogenic pathways (TP53 and PI3K) that was associated with a significantly increased risk of recurrence after surgical resection. Further studies are needed to determine if these genomic features would provide useful preoperative predictors of patient outcomes after surgery.

depending on cyst size. The primary aim of this study was to determine the intermediate and long-term incidences of malignancy and growth kinetics of SB-IPMNs. METHODS: 559 patients with pancreatic cysts labeled as SB-IPMNs were identified at our institution (2/2000-11/2018 ). Patients were grouped as intermediate longitudinal observation (5) (6) (7) (8) (9) (10) or long-term LO (LT-LO; 10 years). The incidence of crossover to resection, malignant transformation and growth kinetics were measured relative to cyst size. RESULTS: 415 and 144 patients underwent I-LO and LT-LO, respectively. The cumulative incidences of malignancy for I-LO and LT-LO patients with SB-IPMNs were 5/415 (1.2%) and 0/144, respectively. Crossover to resection was similar between the two groups -1.9% for I-LO and 2% for LT-LO -with 4/11 (36.4%) of resected cysts yielding high-grade dysplasia. The median cyst growth rates were 0.44 mm/year (-3.2 to 11.4 mm/year) in I-LO and 0.38 mm/year (-1.7 to 2.5 mm/year) in . In I-LO with malignant degeneration, the median rate of cyst growth was 2.2 mm/year (range -2.4 to 4.4 mm/year, p=0.012) . In both groups, approximately 73.2% of cysts increased in size while 7.3% remained stable and 19.5% decreased in size. The overall variance in SB-IPMN growth rates decreased after 10 years (p<0.0001). CONCLUSIONS: Malignant transformation and crossover to resection were low-frequency events ( 2%). The vast majority of observed SB-IPMNs grew 2 mm/year and growth rates became stabilized after 10 years of observation. Cyst regression was common in both groups with unknown significance. Collectively, these data suggest the incidence of malignant conversion of SB-IPMNs to be lower than other large series, requiring confirmation from additional prospective, multi-institutional longitudinal observation studies.

Comparison of cyst growth rates during intermediate longitudinal observation (I-LO; 5-10 years), long-term LO (LT-LO; 10 years), and resected cysts (left), and histogram of cyst growth rates and number of cases (right).

parenchyma demonstrates increased transcription of myelopoietic cytokines, with GM-CSF being expressed nearly 1000 fold (P<0.001). Therapeutic blockade with anti-GM-CSF mab slowed disease growth and onset of multifocal disease, and prolonged survival in KPPC mice (P=0.01). Flow cytometry of treated and untreated KPPC tumors demonstrated a significant decrease in TAMs, and increased MHC-II expression by TANs, suggestive of increased antigen presentation. Human CCA avidly stains for GM-CSF by epithelial and stromal cells compared to normal liver. CCA is dominated by suppressive myeloid leukocytes, future therapies which target these populations through blockade or repolarization represent a novel treatment strategy. Introduction: Tobacco smoking is an established risk factor for multiple cancers. The precise mechanisms through which this promotes cancer are relatively unknown. Our previous studies (Sethi, Kurtom et al, Gastroenterology, 2018) have demonstrated that the gut microbiota promotes cancer in multiple mouse models through the immune system. We, therefore, sought to investigate the role of smoking-associated microbiota in promoting cancer. Methods: C57BL/6 mice were randomized to receive antibiotics or saline in conjunction with/without nicotine-derived nitrosamine ketone (NNK). NNK has been shown to mimic the effects of tobacco smoke exposure and antibiotics were used for gut sterilization. Mice were injected subcutaneously with cancer cell lines: KPC (pancreatic), MC38 (colon), MB49 (bladder), LLC (lung) and tumors were measured serially. We also tested the effect of NNK with/without microbiota in a model of autochthonous cancer through A/J mice. Tumors were immunophenotyped using flowcytometry and experiments were repeated in T/B cell lacking RAG1 knockout mice. Fecal microbiota transplant (FMT) was used to see a direct effect of NNK-associated stools on tumorigenesis. 16S rRNA sequencing was used to evaluate microbial changes. Results: Across all models, treatment with NNK promoted tumor growth in mice with intact gut microbiota but failed to do so in the absence of the gut microbiota. FMT with NNK-associated stools directly promoted cancer in untreated mice as compared to FMT of control stools. 16S rRNA analysis revealed that NNK caused a drastic shift in gut microbiome and led to enrichment of Bacteroidea. NNK-associated tumors had significantly less infiltration of CD8+ T cells as well as PD1+ cells. Finally, the effect of NNK-associated tumor promotion was abrogated in Rag1 knockout mice thereby implicating the immune system as an important player in cancer promotion. Conclusion: Our studies demonstrate that an important smoking-associated carcinogen promotes cancer in part through a change in the gut microbiome. Ongoing studies are characterizing the microbial taxa directly responsible for this tumor promotion as well as using smoking chambers to reproduce findings. Nucleic acid sensor agonists are promising anticancer therapies. They activate innate immunity to enhance tumor immunosurveillance. However, recent studies indicate they can inhibit cancer progression in a tumor cellautonomous manner. We show activation of STING, a cytosolic DNA sensor that initiates type I interferon (IFN) production, inhibits growth of pancreatic ductal adenocarcinoma (PDAC). We propose a novel mechanism where IFNs alter nucleotide metabolism and sensitizes PDAC tumors replication stress response (RSR) inhibitors. The STING agonist cGAMP was transfected in PDAC cells and RT-qPCR for IFN performed. A LC-MS metabolomics assay identifying >130 unique metabolites was performed on PDAC cells exposed to IFN . Intracellular deoxyribonucleotide triphosphates (dNTPs) generated from were quantified using LC-MS. Immunoblots were performed assessing activation IFN signaling (STAT1) and RSR (pCHEK1) markers. Where indicated, cells were treated with IFN or VE822, an inhibitor of ATR, the effector kinase of the RSR pathway. Doxycycline-inducible constitutively active STING and shRNA knockdown of ATR cell lines were created. Orthotopic PDAC models were created by injecting these luciferase-tagged cells in pancreata of NSG mice and growth assessed with bioluminescence measurements. 3D organoids were grown using U-bottom, low attachment 96-well plates. STING agonists result in robust IFN production in PDAC cells (a). A global metabolomics analysis of IFN -exposed PDAC cells revealed decreases in nucleotide pools as the most prominent changes in metabolites (not shown). This was confirmed with a targeted analysis of dNTP pools (b). This was associated with activation of the replication stress response pathway (c). A combination of STING activation and RSR inhibition decreased PDAC progression in both in vitro 3D organoid models (d, e) and in vivo orthotopic models (f,g). We show STING activation leads to IFN production and RSR activation in PDAC by depleting dNTP pools. STING activation combined with RSR inhibition synergistically suppresses in vitro and in vivo PDAC models, suggesting a viable treatment strategy for a disease resistant to traditional chemotherapy. Introduction: Cholangiocarcinoma (CCA) is the second most common primary liver malignancy, with increasing incidence. Surgical resection offers the only chance for cure. However the prognosis remains poor in part due to high rates of unresectability, recurrence, and poor response to conventional therapy. Thus new systemic therapies represent an unmet medical need. Few preclinical models exist for identifying and testing new targeted or immune based therapies. Here we present our findings of the immune infiltrate in human CCA tumor microenvironment (TME) and a spontaneous murine model that faithfully recapitulates human disease. Methods: Histology and immunohistochemistry (IHC) staining was performed on human CCA and adjacent normal liver. Mice with targeted hepatic Kras activation and loss of p53 (KPPC) spontaneously develop CCA. KPPC hepatic tumors and normal livers from littermate controls underwent histological and gene expression studies. Flow cytometry analysis was performed on bone marrow, spleen, peripheral blood, CCA tumors and normal littermate livers. Results: Digital IHC quantification of archival human CCA specimens demonstrated elevated levels of CD15 + CXCR2 + granulocytic myeloid derived suppressor cells (G-MDSC) compared to adjacent normal liver (p = <0.05). In addition, the CXCR2 ligand, CXCL5, was significantly elevated in CCA tumors compared to adjacent normal liver. In KPPC mice, flow cytometry analysis of hepatic tumors showed an abundance of CD45 + leukocytes comprised of immunosuppressive G-MDSC vs normal littermate controls (p < 0.01) which recapitulates human disease. qRT-PCR demonstrated significantly increased expression of G-csf, Csf1, Cxcl1, Cxcl2, and Cxcl5 (p < 0.01) in KPPC CCA tumors compared to normal livers. Accordingly, granulocytes in KPPC mice were elevated in both the bone marrow and blood compared to normal littermate controls. Conclusion: These data suggest CCA co-opts the ELR+ cytokine/ CXCR2 axes to mobilize and recruit immunosuppressive G-MDSC to the TME. Targeted therapy against tumor infiltrating neutrophils can be tested in this pre-clinical model to inform clinical translation.

Cancer and Breast Cancer H. Takahashi,* M. Asaoka, E. Katsuta, L. Yan, K. Takabe. Surgical Oncology, Roswell Park Comprehensive Cancer Center, Buffalo, NY. Introduction: Lymphovascular invasion (LVI) is one of pathologically defined aggressive tumor characteristics. Although LVI has been considered as a risk factor for distant metastasis in various cancers, its underlying molecular biology is poorly understood. On the other hand, we recently found that PDACs with mature blood vessels (high CD31 expression) have better survival likely due to increased anti-cancer immune cells infiltration. Herein, we hypothesized that tumors with LVI have the distinctive molecular biological profiles such as enhanced lymphangiogenesis or angiogenesis, leading to worse clinical outcomes. Methods: Genomic profiles and clinical information were obtained from The Cancer Genome Atlas (TCGA) pancreatic ductal adenocarcinoma (PDAC) and breast invasive carcinoma (BRCA) cohorts. Presence of LVI was obtained through Text Information Extraction System (TIES) and manually matched to the TCGA data. Results: LVI was positive in 85 patients (65.4%) in PDAC and 242 patients (37.8%) in BRCA. Positive LVI was associated with aggressive clinical characteristics in both cohorts; perineural invasion (p=0.023), higher AJCC T (p=0.017) and N categories (p<0.001) in PDAC, as well as higher histological grade and MKI-67 expression in BRCA (all p<0.001). Further, positive LVI was associated with worse OS in PDAC (p=0.014) and in advanced breast cancers (p<0.001). Contrary to the previous reports, no significant difference was identified between LVI and lymphangiogenesis or angiogenesis, detected by PDPN, LYVE1, and VEGFs gene expression in both cohorts. Lastly, Gene set enrichment analysis revealed that cell proliferation and DNA replication gene sets were enriched in both cohorts; mitotic spindle (Normalized enrichment score (NES)=1.76, p=0.016) and G2/M checkpoint (NES=1.75, p=0.036) in PDAC, and mTORc1 signaling (NES=2.27, p<0.001), G2M checkpoint (NES=2. 15, p<0 .001), E2F targets (NES=2.14, p<0.001), and MYC targets (NES=2.08 p=0.002) in BRCA. Conclusion: Transcriptomic analyses revealed that aggressiveness of LVI results from enhanced cell proliferation and DNA replication, independent of lymphagiogenesis.

Histopathological growth patterns (HGPs) classify the invasive margin of hepatic tumors. Previous studies in breast and colorectal cancer liver metastasis demonstrated that patients with hepatic tumors fully encapsulated by the desmoplastic type (dHGP) exhibit superior survival compared to patients with any observed non-desmoplastic type (non-dHGP). A single center retrospective cohort study was performed to assess HGPs in surgically treated non-cirrhotic hepatocellular carcinoma (HCC) patients. The HGP was determined on H&E tissue sections of resected HCC specimens. Overall (OS) and disease-free (DFS) survival was estimated by Kaplan-Meier and multivariable Cox regression analysis. Patients that died postoperatively, defined as within 30 days of surgery, were excluded. Effect modification for subgroups was investigated by interaction terms. The HGP was determined in 121 patients who underwent first resection of HCC between 2000 and 2018, 7 of whom died postoperatively. In one third of patients (n=41, 34%), the hepatic tumors were fully encapsulated by dHGP. Any non-dHGP was observed in the remaining two-thirds (n=80, 66%). Apart from patients with dHGP being older, no statistically significant differences were observed for gender, number and size of lesion(s), alpha-fetoprotein levels, microvascular invasion (MVI), and comorbidities. Median OS and DFS were 68 & 39 months in dHGP, and 67 & 16 months in non-dHGP (overall log-rank: p=0.56 and p=0.14). After correction, non-dHGP was associated with a hazards ratio (HR) [95% Confidence Interval (CI)] of 1.58 [0.85-2.93 [1.54-7.77 ] (p<0.01) in non-dHGP. This study demonstrated for the first time that HCCs of non-cirrhotic patients display HGPs as described in other hepatic tumor types. Furthermore, MVI seems only clinically relevant in patients with non-dHGP (figure 1), as significant effect modification for OS was found between the HGP and MVI status. Figure 1 . Kaplan-Meier survival estimates for overall and disease-free survival stratified histopathological growth pattern (HGP), or by HGP and microvascular invasion (MVI). Reported p-values represent the overall log-rank test. dHGP: desmoplastic type HGP, non-dHGP: any observed non-desmoplastic type HGP.

Hepatectomy or Transplantation F.S. Dahdaleh, 1 * S. Naffouje, 2 G.I. Salti. Naperville, IL; 2. Moffit Cancer Center, Tampa, FL. Background: Biopsy to achieve tissue diagnosis (TD) of hepatocellular carcinoma (HCC) carries risk of needle tract seeding. Chest wall and peritoneal recurrences have been reported after TD which, in essence, denote upstaging. We hypothesized that TD adversely affects overall survival (OS) when compared to clinical diagnosis (CD) in HCC. Methods: The National Cancer Database (NCDB) Participant User File for liver cancer was utilized. Only patients with non-metastatic HCC treated with major hepatectomy or transplantation were included. CD patients were matched 1:1 to TD patients per propensity score. Survival was examined in the unmatched and matched cohorts. Results: Of 172,283 cases, 16,366 met the inclusion criteria. 12,100 (73.9%) were males and mean age was 60.8 10.4 years. Curative procedures were divided equally between hepatectomies and transplantation (48.4% and 51.6%, respectively). 70.4% of cases had CD and 29.6% underwent TD. After matching, 4,251 patients were selected from each group. Patients who underwent TD had decreased OS compared to the CD group (median 85. 6 4.6 vs. 65.5 2.7 months, respectively; P<0.001) . Similarly, five-year survival was lower in the TD group (47.6% vs 60.9% P<0.001, Figure) . Cox regression confirmed the diagnostic method as an independent predictor of OS in addition to age, Charlson-Deyo score, grade, delay of surgery, lymphovascular invasion, nodal stage, and procedure type, favoring transplantation over hepatectomy. Conclusion: In this large, national dataset, OS was decreased in HCC patients who underwent preoperative TD compared to CD. This study supports exhausting non-invasive methods to diagnose HCC prior to attempting biopsy.

Kaplan-Meier graph for HCC patients who had clinical diagnosis (CD) vs. tissue diagnosis (TD).

Evaluating the Immune Modulatory Effects of A Targeted Anticancer Therapy, ACXT-3102, in a Murine Model of Pancreatic Cancer U.Y. Panni, 1 * J.M. Herndon, 2 D. Spitzer, 1 D. DeNardo, 2 W. Hawkins. 1 1. Hepatobiliary Surgery, Washington University in St. Louis, St.Louis, MO; 2. Washington University in St. Louis, St.Louis, MO. Background: Pancreatic cancer (PC) is expected to become 2nd deadliest malignancy in the US by 2030. Modern approaches, including immunotherapy, have shown minimal efficacy in human PC. Recent studies have demonstrated that conventional and targeted anticancer drugs exhibit immunomodulatory effects on the tumor microenvironment in addition to their direct cytotoxic effects. This promising strategy can improve the efficacy of novel agents by combining them with select immunotherapeutic drugs and vice versa. ACXT-3102 is a small molecule therapeutic agent which has shown to provide efficient tumor control and prolonged survival with minimal off-site toxicities in murine PC models. It consists of an Erastin derivative, des-methyl Erastin, conjugated to a sigma-2 receptor ligand, SV119. ACXT-3102 binds to cancer cell-specific sigma2-receptors and blocks intracellular uptake of cysteine at cysteine-glutamate transporter, leading to an increase in intracellular accumulation of reactive oxygen species, and causing ferroptosis. Here we assess the immunomodulatory effects of ACXT-3102 on tumor immune infiltrates in a murine PC model. Methods: A spontaneously derived orthotopic murine PC model (KP2.0) was established in age-matched C57BL/6 mice. Mice were treated with an every other day dosing regimen of ACXT-3102 for 10 days. At the end of the treatment, tumors were harvested and were analyzed using flow-cytometry. Results: We observed a significant reduction in tumor volume (p= 0.04) and gross tumor weight (p=0.0259) in the mice treated with ACXT-3102. Simultaneously there was an increase in CD8+ T cell numbers, activation, and proliferation by flow-cytometry. We also noted a significant increase in the CD4+ T cell population with a decrease in the regulatory T cell population. Conclusion: Our results uncover the significant impact ACXT-3102 has on the T cell infiltration and activation in the tumor microenvironment, in addition to controlling tumor growth. We believe that exploring these findings further in combination with immunotherapy may be a useful strategy in the treatment of human PC. Background: Pancreatic Ductal Adenocarcinoma (PDAC) has less than a 10% five year survival and will become the second leading cause of US cancer mortality in the next decade. Immunotherapy, such as checkpoint inhibition against anti-Programmed death-ligand 1 (PD-L1) has not been successful in the treatment of PDAC patients. Both tumor associated macrophages (TAMs) and the TGF-protein are ubiquitous in PDAC tumors. We hypothesize that TGF-increases the overall number of TAMs and degree of PD-L1 expression of TAMs in PDAC. Methods: Our lab has a mouse pancreatic cancer cell line derived from a genetically engineered mouse model (KPC mice that spontaneously form PDAC tumors that are similar to human PDAC). We orthotopically implanted this cell line into the pancreas of immunocompetent C57BL/6 (B6) mice. In groups of 5 each, mice were treated with saline (control) or TGF-f. We investigated tumor burden, the number of TAMs (CD45 + , CD11b + F4/80 hi , Ly6C -, Ly6G -/lo ) in the tumors with flow cytometry and the percentage of TAMs expressing PD-L1 in the pancreas and metastatic lesions in the liver. Results: As a percent of leukocytes in the tumor, PDAC liver metastases had more TAMs compared to tumors in the pancreas (33 5% vs 10 4%, P=0.001). Compared to controls, TGF-treatment significantly increased the percent of PD-L1 expressing TAMs (32 6 % vs 12 5%, P=0.013, see Figure) in the pancreas but no effect was evident on TAM density. In liver metastases, treatment with TGF-decreased the overall TAM density (P=0.039) but did not affect the number of PD-L1 positive TAMs. Conclusions: TGF-plays pivotal role in the progression of PDAC and demonstrates context dependent activity. Our results suggest that an immunosuppressive effect mediated by PD-L1 expression on TAMs may be initiated by TGF-. Future investigations will focus on understanding the role of the PDAC -TAM interaction to develop effective immune therapies for PDAC patients.

Platelets Drive Anti-Tumor Immunity in Non-Alcoholic Fatty Liver Disease Associated Hepatocellular Carcinoma L.P. Diggs, 1 * Q. Fu, 3 C. Ma, 1 B. Heinrich, 1 Z.J. Brown, 2 B. Ruf, 1 T.F. Greten. 1 1. Surgical Oncology Program, National Cancer Institute, National Institutes of Health, Washington, DC; 2. Robert Wood Johnson Medical School, Rutgers University, New Brunswick, NJ; 3. Food and Drug Administration, Rockville, MD. Introduction: Non-alcoholic fatty liver disease (NAFLD) is a leading cause of hepatocellular carcinoma (HCC) worldwide. Recent studies demonstrate that NAFLD is associated with higher levels of hepatic platelet aggregation and activation. Anti-platelet agents are commonly prescribed in such patients to prevent thromboembolic events. Platelets modulate anti-tumor immunity but their role in NAFLD associated HCC (NAFLD-HCC) remains unclear. We aimed to elucidate the role of platelets in NAFLD-HCC immune surveillance and the impact of their inhibition on this tumor's development. Methods: HCC tumors were induced in mice using DEN, hepatocyte MYC overexpression, and orthotopic injections of the mouse HCC line Hep55.1c. NAFLD was induced using choline-deficient L-amino acid-defined (CDAA) or methionine-choline deficient (MCD) diets. Platelet depletion was achieved using anti-GPIBa antibody. We then tested the P2Y12 platelet receptor inhibitors (PRIs) clopidogrel and ticagrelor, the cyclooxygenase inhibitor aspirin, and the phosphodiesterase III cilostazol. Tumor weight, number, and tumor:liver ratios were compared in NAFLD mice with and without normal platelet function. Results: In our 3 NAFLD-HCC models, platelet depletion caused a significant increase in liver tumor burden (LTB) compared to mice with intact platelet function. Regardless of platelet function, no difference in LTB was noted among mice with healthy livers. These results were recapitulated using P2Y12 PRI clopidogrel ( Figure 1 ). Conversely, aspirin or cilostazol treatment in NAFLD-HCC mice was not associated with a difference in LTB. To confirm that platelet-driven immune modulation is P2Y12 mediated, mice received ticagrelor which also resulted in larger LTB. Conclusion: Using three NAFLD-HCC models in different mouse strains, our results consistently demonstrate that platelet activation induces anti-tumor immunity. Specifically, P2Y12 mediated platelet activation appears to be essential to optimal immune surveillance in NAFLD. P2Y12 PRI usage is high in NAFLD patients and therefore careful selection of anti-platelet agents may be warranted when caring for NAFLD and NAFLD-HCC patients.

Response to anti-platelet therapy in NAFLD-HCC Mice. NAFLD was induced in C57/BL6 mice. These mice then underwent intrahepatic injection of Hep55.1c tumor cells and were treated with clopidogrel or negative control. Two weeks after tumor cell injection, mice were sacrificed and tumor weight to liver weight ratio was calculated and compared between mice with and without normal platelet function.

Assessment of the Long-term Impact of Pancreaticoduodenectomy on Health-Related Quality of Life Using the EORTC QLQ-PAN26 Module Z. Fong,* Y. Sekigami, M. Qadan, C. Fernandez del-Castillo, A.L. Warshaw, K.D. Lillemoe, C.R. Ferrone. Surgery, Massachusetts General Hospital, Boston, MA. Introduction: We have previously reported favorable general quality of life (QoL) in 5 year survivors after pancreaticoduodenectomy (PD). However, there has been a paucity of studies utilizing pancreas-specific modules for QoL assessment, which may uncover disutility that general modules are not designed to detect. Methods: We administered the EORTC QLQ-PAN26 questionnaire to patients who underwent PD for neoplasms from 1998 to 2011 and compared their scores with published baseline scores of patients with pancreatic cancer. Clinical relevance (CR) of differences was scored small (5) (6) (7) (8) (9) (10) , moderate (10-20), or large (>20) based on validated interpretation of clinically important differences. Results: Of the 1,266 patients who underwent PD during the study period, there were 305 5-year survivors with valid contact information, of whom 248 responded to the questionnaire (response rate 81.3%). Median follow-up was 9.1 years (range 5.1-21.2 years). Compared to baseline scores, patients reported higher pain scores (41.7 17.6 vs 18.1 20.5, p<0 .001, CR large), greater frequency of digestive symptoms (26.3 29.5 vs 18.7 27.8, p=0 .002, CR small), altered bowel habit (37.6 30.6 vs 20.0 24.5, p<0.001, CR moderate), and sexual dissatisfaction (63.0 37. 5 vs 35.1 34.3, p<0 .001, CR large) scores ( Figure) . Additionally, there was a higher incidence of bloating, indigestion, and flatulence, but lower incidence of future health worry (71.7% vs 89.6%, p<0.001) and limitation in planning activities (30.1% vs 48.3%, p<0.001) at 5 years compared to baseline scores at the time of diagnosis. Conclusion: While we reported that patients who had a PD had better global QoL at 5 years, a more granular, pancreas-specific questionnaire uncovered important continued abnormalities. While it is important to distinguish perception of QoL and functional outcomes, these data can better inform clinical decision-making and potential areas for improvement. Introduction: Over 840,000 new patients are diagnosed with primary liver cancer including hepatocellular carcinoma (HCC) each year. Understanding the HCC microenvironment is crucial for the development of novel therapeutic strategies and drug regimens. 3D-bioprinting is useful for studying how cells behave, with advantages over traditional 2D cell culture. We developed and tested a novel scaffold-free 3D-bioprinted liver cancer model perfused with sorafenib to investigate tumor-support cell interactions in the setting of chemotherapy. Methods: Primary HepG2 human HCC and hepatic stellate cells (HSC) were co-cultured to form spheroids. Spheroids were bioprinted using the Regenova 3D-Bioprinter into liver tissue constructs, as HSC-only (control) or HCC:HSC combination. Constructs were continuously perfused for 7 days with media-alone or media-with-sorafenib (8.40 M/L) in the FABRICA bioreactor. Real-time PCR was used to measure gene expression of AFP, COL1A1, CYP2B6, TGFB1 for tumor growth, metastatic potential, and HepG2 cell differentiation in spheroids/constructs on days 0, 1, 3, 7. Results: The 3D-HCC model was successfully bioprinted and perfused with media or sorafenib for 7 days. Sorafenib negatively affected HepG2 and HSC cell viability and extracellular matrix formation in the HCC:HSC construct, while robust structural integrity was maintained in the sorafenib-treated HSC construct ( Figure 1C ). Sorafenib perfusion was associated with decreased AFP expression in HCC-containing spheroids, decreased COL1A1 in HSC-only and combined spheroids, rapidly decreased CYP2B6 in combination spheroids, and early-phase decrease and late-phase increase in TGFB1 in combination spheroids. All constructs had higher COL1A1 than corresponding day 7 spheroids. Conclusion: Development of an inexpensive and efficient 3D-HCC model enables us to study its microenvironment for expression of cancer markers, collagen formation, liver metabolism, and invasion/metastasis. Future studies will be conducted to create a comprehensive liver/cancer microenvironment for HCC and other cancers, focusing on efficiently testing multiple drugs against a unique tumor. Introduction: Patients with ICC generally have a poor prognosis, yet there can be heterogeneity in patterns of presentation and associated outcomes. We sought to identify clusters of ICC patients based on preoperative characteristics that may have distinct outcomes based on differing patterns of presentation. Methods: Patients undergoing curative-intent resection of ICC between 2000-2017 were identified using a multi-institutional database. A cluster analysis was performed based on preoperative variables to identify distinct patterns of presentation. A classification tree was built to prospectively assign patients into cluster assignments. Results: Among 1,004 patients with ICC, 4 distinct presentation patterns were revealed. Specifically, cluster 1 (n=412, 41%) consisted of mostly men (65.1%), non-White patients (68.7%) who had a single (100%), small size ICC (median=4.4cm) with no LNM on imaging; cluster 2 (n=256, 25.5%) consisted of more female (54.3%), non-White patients (57.1%) who had single ICC (100%) with no metastatic LNs; yet larger size tumors (median=9cm); cluster 3 (n=193, 19.2%) was largely non-White patients (61.1%) with single ICC (100%), medium sized tumors (median=6cm), yet all patients had suspicious/metastatic LN on imaging; cluster 4 (n=143, 14.3%) consisted mainly of White patients (75.4%) who had multiple tumors (>2, 100%), with a median tumor size of 7cm and almost one third (30.8%) had suspicious/metastatic LNs on imaging. Cluster 1 patients had the best 5-year OS compared with other clusters (48.5% vs 41.9% vs 25.8% vs 24.3%, p<0.001, Fig 1a) ; patients in clusters 3 and 4 had comparable outcomes (p>0.05). The classification tree used to assign patients into different clusters had very good agreement with actual cluster assignment ( =0.86, 95%CI 0.82-0.89) (Fig 1b) . Conclusion: Machine learning classification tree analysis identified 4 distinct prognostic clusters based solely on preoperative characteristics among patients with ICC. Characterizing preoperative patient heterogeneity with machine learning tools can help physicians with preoperative selection and risk stratification of patients with ICC. Bartlett, A. Zureikat. Surgical Oncology, University of Pittsburgh Medical Center, Pittsburgh, PA. Introduction: Alteration in the tumor suppressor gene SMAD4 is one of the key molecular underpinnings that contribute to pancreatic ductal adenocarcinoma (PDAC). The clinical significance of SMAD4 status in metastatic lesions, however, remains to be determined. The objectives of our study were to analyze the oncological outcomes of SMAD4 expression in recurrent PDAC. Methods: A retrospective review of patients who presented with a recurrence after resection of PDAC between 2008-2019 was performed. SMAD4 status in the primary and recurrent tumor was determined and classified as either preserved or loss of expression. Survival analysis using Kaplan-Meier estimates and Cox-regression was performed. Results: A total of 382 patients were identified: 154 (40%) had preserved SMAD4 and 228 (60%) had loss of expression. There was no significant difference in the overall survival (OS) (24. 6 VS 23.4 months, p=0.877) or recurrence-free survival (RFS) (13 vs 12 months, p=0 .858) between both groups. However, in patients with lung first recurrence, and on multivariate analysis, survival remained significantly longer than those with abdominal first or synchronous lung/abdominal recurrence, regardless of SMAD4 expression (18 VS 7.8 vs 6.5 months, HR=1.72, 95% CI, 1.22-2.44 ). Among patients who had SMAD4 testing in the recurrent lesions (n=61), 31% were SMAD4 preserved and 69% had loss of SMAD4 expression. When SMAD4 status of the recurrent lesions was compared to that of the primary tumor, 15% had a change in the SMAD4 expression, all of which converted from preserved in the primary to loss in the recurrence. By Kaplan-Meier estimates, this mutational change corresponded to a significant effect on OS (26 vs 44 months, p<0.012). Conclusion: Although the primary site of metastasis is the principle determinant of survival, a change of SMAD4 from preserved in the primary to loss in the recurrence may correlate with more aggressive systemic disease. Analysis of the metastatic lesion may identify more indolent recurrences that can potentially benefit from aggressive systemic and local control therapies.

Kaplan-Meier Overall-Survival Estimates comparing recurrent lesions with no change in SMAD4 expression versus those with loss of SMAD4 expression compared to the primary tumor.

Fibrotic Response to Neoadjuvant Chemoradiation Predicts Survival in Human PDAC D. Erstad, 1 * M. Sojoodi, 1 M.S. Taylor, 2 C.T. Farrar, 3 A.L. Axtell, 1 N.J. Rotile, 3 C. Jones, 3 K.A. 3 D.S. Ferreira, 3 T. Michelakos, 1 F. Kontos, 1 A. Chawla, 1 S. Li, 1 S. Goshal, 1 Y. Chen, 3 G. Arora, 1 V. Humblet, 3 V. Deshpande, 2 N. Bardeesy, 1 C.R. Ferrone, 1 M. Lanuti, 1 K.K. Tanabe, 1 P. Caravan, 1 B. Fuchs. 1 1. Surgery, Massachusetts General Hospital, Boston, MA; 2. Pathology, Massachusetts General Hospital, Boston, MA; 3. Martinos Center for Biomedical Imaging, Massachusetts General Hospital, Boston, MA. Introduction: Neoadjuvant chemoradiation for PDAC is associated with a dense fibrotic response. Treatment-induced tumor fibrosis is objectively quantifiable, and we hypothesize that the degree of fibrosis associates with overall (OS) and disease-free survival (DFS) in surgically resected PDAC patients.

Methods: Records of patients with borderline resectable PDAC treated with neoadjuvant FOLIFIRINOX and radiation therapy followed by surgical resection were analyzed. Sirius Red collagen staining of tumor specimens was performed, and fibrosis was digitally quantified as the percent collagen proportional area (CPA). Results: 93 patients were eligible. The median age was 62 years, 49 (51%) were female, median OS was 41.3 months (mo), and median DFS was 12.4 mo. Post-neoadjuvant tumor fibrosis levels were normally distributed with a mean CPA of 69. 4 8.0% . CPA was associated with increased OS (HR 0.94, p=0.001) and DFS (HR 0.95, p=0.009) . A 10% increase in CPA was associated with an approximately 50% reduction in risk of death or recurrence. The median OS in patients with low (<61.4%, <1 SD of mean), moderate (61.4-77.4%, 1 SD), or high fibrosis (>77.4%, >1 SD) was 24. 7, 41.3, and 55.4 mo, respectively (p=0.0072) ; median DFS in these groups was 8.2, 13.5, and 37.7 mo, respectively (p=0.031) (Figure 1 ). Elevated serum CA-19-9, tumor diameter, T stage 3-4, N stage 1-2, poor response by pathologic regression grading, and perineural invasion were associated with reduced survival. On COX multivariable analysis, only CPA and tumor diameter at diagnosis remained significantly associated with OS and DFS. Patients with both pathologic regression (complete or moderate grades) and increased post-neoadjuvant tumor fibrosis (CPA 70%, threshold determined by ROC analysis) experienced significantly better survival outcomes than patients lacking one or both of these tumor features (median OS 55 vs. 33 mo, p=0.0008; median DFS 37 vs. 11 mo, p=0.018). Conclusion: CPA is an objectively quantified marker that more significantly associated with survival than regression grading, the current clinical standard for treatment response. CPA can be readily performed by pathologists and may improve PDAC prognostication. Introduction: The use of the robotic platform for complex hepatobiliary surgery is increasing. Majority of the literature have only reported outcomes on robotic minor non-anatomical liver resection. This study was undertaken to examine our institutional perioperative outcomes, safety, and feasibly with robotic major liver resection. Methods: 150 consecutive patients undergoing robotic liver resections were prospectively followed since 2016. Major hepatectomy is defined as a resection of 3 or more segments. For illustrative purposes, data are expressed as median (mean + SD) when appropriate. Results: Of all patients undergoing robotic liver resections, 85 patients underwent major resection. Median age was 63 years (62 12.8), 46% were women, BMI was 29 (29 6.2) kg/m 2 and ASA Class was 3 (3 0.6) . Alcohol use was seen in 12 patients (14%). Of the 85 operations that were undertaken, 30% were for hepatocellular carcinoma, 28% for metastatic adenocarcinoma, 9% for cholangiocarcinoma, and 5% for metastatic neuroendocrine tumor. Regarding the type of resection, 11 patients (13%) had central hepatectomy, 22 patients (25%) had formal right, 23 patients (27%) had formal left, 11 patients (13%) had non-anatomical right, 9 patients (11%) had non-anatomical left, and 9 patients (11%) had posterior superior resection. Prep time (in the room until incision) 58 minutes (62 18.4) , Extraction time (incision until specimen extraction) 124 minutes (146 99.5) , Console time 198 minutes (210 123.9), Closure time (extraction until dressing placement) 109 minutes (131 93.8), Operative duration was 246 minutes (269 123.2) and time under anesthesia 330 minutes (353 109.6) . Estimated blood loss was 200 mL (246 266.3) and length of stay was 4 days (5 4.3) . 7 patients experienced postoperative complications (4 ileus, 1 pneumonia, 1bile leak, 1-gram negative bacteremia). 13 patients were readmitted within 30 days with one death after readmission, due to aspiration. Conclusion: Application of the robotic platform to major liver Introduction: For patients (pts) with surgically resected cutaneous melanoma with positive sentinel lymph nodes (SLN), the Multicenter Selective Lymphadenectomy Trial 2 (MSLT2) demonstrated equivalent melanomaspecific survival with active surveillance (AS) using nodal basin ultrasound (US) versus immediate completion lymphadenectomy (CLND). Methods: We retrospectively evaluated adoption of AS and early oncologic outcomes in pts with +SLN from 6/2017-6/2019. Using Cox proportional hazards models, we explored the impact of AS and adjuvant systemic therapy (ADJ) on overall recurrence (OR) and isolated nodal basin recurrence (NBR). Results: Across 15 institutions 4194 SLNB were performed, of which 782 had +SLN and negative staging for distant disease (19%) . 167 pts underwent CLND (21%) while 615 (79%) received AS. CLND was used more often for younger pts (median 58 vs 62, p=0.048), head/neck vs trunk or extremity tumors (34 vs 20 or 19%, p<0 .001), more +SLNs (median 1.63 vs 1.27, p<0 .001), extra-nodal extension (ENE) (38 vs 20%, p=0.003), and larger nodal deposits (median 1.5 vs 0.5mm, p<0.001). For those in AS, 11 of 15 (73%) institutions used US while 4 of 15 (27%) used CT scans to follow at-risk nodal basins. 294 patients (38%) received ADJ (84% anti-PD1, 10% BRAF/MEK, 6% other). ADJ correlated with adverse tumor and nodal features including ulceration (p<0.001), more +SLNs (p=0.002), ENE (p<0.001), larger nodal deposits (p<0.001), and higher pathologic stage (p<0.001). At 10.3 months median follow-up 42 patients had isolated NBR (post-CLND 6/167 (3.6%), AS 36/615 (5.9%)) and 107 (13.7%) recurred at local, distant, or multiple sites. In adjusted analyses CLND and ADJ were associated with decreased OR but not NBR (Table) . Conclusions: With high adoption of AS, short-term isolated nodal basin failure was uncommon. Higher risk patients were more likely to receive CLND and ADJ, which were both associated with reduced overall recurrence. While nodal observation appears appropriate in most patients, future work should seek to clarify when CLND remains beneficial and which AS patients benefit from ADJ.

Association of patient, tumor, and treatment factors on oncologic outcomes in adult patients with cutaneous melanoma and positive sentinel lymph node(s) Cox proportional hazard models performed for outcomes of any recurrence and isolated nodal basin recurrence, adjusted for the above variables; HR = Hazard Ratio; CI = Confidence Interval; *denotes statistically significant findings; values <1 associated with decreased recurrence; values >1 associated with increased recurrence Introduction: While the staging value of sentinel lymph node (SLN) biopsy in melanoma is clear, the therapeutic value of the procedure is frequently overlooked. One clear therapeutic effect is in regional node basin disease control, even in the absence of completion node dissection. We examined factors related to long-term in-basin control after SLN biopsy alone in the second Multicenter Selective Lymphadenectomy Trial (MSLT-II) Methods: Subjects randomized to observation in MSLT-II following removal of a positive SLN were examined for factors related to an absence of regional node recurrence. Examined factors included age, sex, primary tumor thickness, ulceration, number of positive SLNs, node basin site, and SLN tumor burden (measured by maximum diameter of largest focus or percent area of node). Results: Among 855 observed basins (subject n=823), at 10-years (actuarial) 80% of basins were free of nodal recurrence. By univariable analysis, freedom from regional nodal recurrence was associated with age <50 (OR 0.49), non-ulcerated primary (OR 0.36), lesser tumor thickness (<1.5mm OR 0.46), axillary basin site (OR 0.61), fewer positive SLNs (1 vs. 3 HR 0.32) , and smaller SLN tumor burden (measured by either diameter (<1mm OR 0.36) or area (<5% OR 0.36) (all p-values <0.05). By multivariable analysis, younger age (OR 0.57), thinner primary (OR 0.48), axillary basin (OR 0.55), and smaller SLN metastasis (diameter OR 0.52, area OR 0.58) were independently associated with basin control. Including age ( 50), ulceration, Breslow >3.5 mm, non-axillary basin, max diameter 1mm, and metastasis area 5% as risk factors and excluding missing value cases, basin disease-free rates at 5-years were 96%, 89%, 86%, 80%, 61%, and 54% for 0, 1, 2, 3, 4, and 5 factors, respectively. Conclusion: To our knowledge, this is the largest prospective evaluation of long-term basin control after SLN Background: While sentinel lymph node (SLN) biopsy is the standard technique used to identify patients at risk for recurrent melanoma, it fails to accurately identify certain patients at high or low risk. Since processes in the SLN determine the effectiveness of anti-tumor immune responses, we hypothesize that patterns of immune gene expression in SLNs may distinguish patients at high risk for recurrence. Methods: NanoString nCounter® is an RNA-based technology that allows for digital quantification of multiplexed target molecules. The nCounter PanCancer Immune Profiling Panel was used to analyze expression of 730 immune genes in formalin-fixed paraffinembedded SLN specimens (n=60) from a retrospective melanoma patient cohort. Patients were selected based on complete follow up, available SLN tissue, and SLN status (30 positive [pSLN] , 30 negative [nSLN] ). Stepwise variable selection for Cox regression models was used with alpha levels of 0.05 for insertion and 0.1 for deletion to identify immune genes differentially expressed in the SLNs of patients who went on to develop recurrence to develop a risk model predictive of recurrence. Results: Mean Breslow depth was 3.85 mm for pSLN vs 2.45 mm for nSLN. At a median follow up of 6.3 years, 14 pSLN (46.7%) patients and 6 nSLN (20.0%) patients recurred. Among all patients, regardless of SLN status, a model incorporating ten immune genes accurately predicted risk of recurrence (C-index 0.9919). The 10 genes and direction of expression included increased expression of TIGIT, an immune checkpoint, and decreased expression of CXCL16, a cytokine important in promoting interaction between dendritic cells and T cells. Independent of SLN status, a recurrence risk score using these 10 genes was able to stratify patients into cohorts at high and low risk for recurrence with high concordance (Figure 1 , log-rank p<0.001). Conclusions: In this study, we identified immune genes associated with melanoma recurrence which may be used in a risk stratification model to identify patients at high risk for recurrence regardless of SLN status, potentially enhancing patient selection for adjuvant therapy.

Kaplan Meier curves showing recurrence free survival (RFS) in patients with positive (pSLN) or negative (nSLN) sentinel lymph node status stratified by calculated recurrence risk score into high (above median) or low (below median) risk cohorts.

Background: Trials demonstrating the safety of nodal observation after sentinel lymph node (SLN) metastases required ultrasound (US) in follow up. However, the contribution of US to the safety of observation has not been clearly defined. We sought to identify the number of nodal recurrences in MSLT-II detected by US as the first detection modality and any differences in clinical characteristics or outcomes in that group. Methods: Patients with histopathology positive SLN randomized to observation underwent nodal US at 4-month intervals for 2 years, then 6-month intervals through 5 years. Abnormal US characteristics were length:depth ratio of <2, hypoechoic center, loss of hilar vascular echoes, or hypoechoic foci with increased vascularity. Physical exams occurred at the same intervals, and other imaging was performed according to each center's standard of care. Results: Among patients assigned to observation, 823 were eligible and accepted their assigned treatment. With the most recent follow up, 187 had nodal disease at initial recurrence. Of those, 90 (48%) were found by US alone, 32 (17%) were found by US and physical exam simultaneously, and 13 (7%) were found solely by other imaging modalities. Among obese patients, 65% were detected by US only. Time to nodal recurrence detection was similar for US-only detected disease relative to other methods (median 14.6 mo vs. 12.9 mo, p=0.65), as was the number of involved nodes (1.3 vs. 1.3, p=0.52) . Melanoma-specific survival (MSS) was not significantly different between patients with nodal metastases detected by US vs. other means (8yr MSS 55. 8 vs. 44.9, p=0.23) . US-detected patients were non-significantly less likely to have concomitant distant metastases (6.7% vs. 12.4%, p=0.19) . Conclusion: In MSLT-II, approximately half of nodal recurrences were detected by US before other methods. A statistically significant survival benefit was not observed with US detection, though a clinically significant benefit cannot be ruled out. Until additional safety data are available, US should remain part of the follow up of patients with SLN metastases undergoing nodal observation.

Is Therapeutic Lymph Node Dissection of Value for Lymph Node Recurrence in Melanoma? A.K. Wilson,* S. Stern, P.D. Lorimer, N.A. Lee, J.B. Ramiscal, T.D. Fischer, F. Leland, R. Essner. Surgical Oncology, John Wayne Cancer Institue, Santa Monica, CA. Background: Although completion lymphadenectomy is no longer recommended for patients with tumor positive sentinel lymph nodes, therapeutic lymph node dissection (TLND) is still recommended for patients with lymph node recurrences. This study evaluated the potential survival benefit of TLND in patients with nodal recurrence following a previously negative sentinel lymph node dissection. Method: Utilizing our institutional melanoma database of over 15,000 patients and data from the Multicenter Selective Lymphadenectomy Trial (MSLT-1), we identified patients with nodal recurrences following a negative sentinel lymph node biopsy. Patients with concomitant local or distant recurrence were excluded. Patients who underwent TLND were compared to those who underwent biopsy alone. Clinicopathologic characteristics, distant metastasis-free survival (DMFS) and melanoma-specific survival (MSS) were compared between the groups. Results: During the study period, 1991-2017, 172 patients with lymph node recurrence in a previously negative sentinel lymph node basin were identified. Median follow up was 30 months from the time of recurrence. TLND was performed in 134 (78%), and 38 (22%) were treated with lymph node biopsy alone. Five year DMFS was 50.4 9.4 % for patients who had a biopsy alone compared to only 39.4 4.7% for those having TLND (p=0.13). Overall, in the biopsy group and 45.9 4.7% in the TLND group (p=0.10). In the TLND group, 54 patients (40.3%) had one tumor positive lymph node and 80 (59.7%) had two or more. For both subgroups, MSS was lower than the biopsy alone group (52. 8 7.3% and 41.2 6 .0%, respectively, p=0.07). There was not a significant difference in use of systemic treatment or immunotherapy between the groups. Conclusion: TLND did not improve DMFS or MSS in patients with lymph node recurrence, and in fact, there was a trend towards worse outcomes following TLND. Presumably, at least some patients in the biopsy alone group had multiple involved lymph nodes, yet they still had at least equivalent outcomes to patients with one positive node treated with TLND. This data further suggests a lack of therapeutic value for TLND in melanoma. Introduction: The treatment landscape for stage III melanoma has changed dramatically with the increased use of adjuvant immunotherapy and decreased use of immediate completion lymph node dissection (CLND) for patients with occult lymph node metastases. However, adjuvant immunotherapy outcomes have not been reported for patients with a positive sentinel lymph node biopsy (SLNB) who did not undergo immediate CLND. The purpose of this study was to evaluate the use of adjuvant immunotherapy and outcomes for patients with stage III melanoma who did not undergo immediate CLND. Methods: This is a retrospective review of a single institution database. Patients with a positive SLNB from 2016-2019 who did not undergo immediate CLND were included. Disease-free survival (DFS) and nodal recurrence-free survival (nRFS) were compared between patients who received adjuvant immunotherapy versus observation alone, using the Kaplan-Meier method and Cox Proportional Hazards regression. Results: 73 patients were included with a median follow-up time of 14 months. 41 patients (56%) received adjuvant immunotherapy while 32 patients (44%) underwent observation alone. Patients who received adjuvant immunotherapy were younger (mean age: 54.5 vs 63.4, p=0.02), had greater SLNB disease burden (>1 positive SLN: 29.3% vs 6.3%, p=0.01), and trended towards more advanced stage (stage IIIC: 48.8% vs 31.3%, p=0.18). Despite more adverse features in the immunotherapy cohort, nRFS was equivalent between groups (1-year nRFS: 92% vs 92%; p=0.71) and adjuvant immunotherapy was associated with longer DFS than observation alone (1- year DFS: 92% vs 79%; p=0.08) ( Figure 1) . On multivariable analysis, controlling for age and stage, adjuvant immunotherapy was associated with improved DFS (hazard ratio, 0.22, p=0.018). Conclusion: For patients with a positive SLNB who did not undergo immediate CLND, adjuvant immunotherapy was commonly used and associated with a significant improvement in DFS. The benefit of adjuvant immunotherapy in this analysis compares favorably to that observed in previous randomized trials in patients who underwent immediate CLND. Background: Although melanoma is more frequent in older patients and is associated with cumulative UV radiation exposure, younger patients have better prognosis. This study evaluated melanoma age-related genomic differences and identified potential targetable mutations. Methods: We analyzed next-generation sequencing data in the AACR Genomics Evidence Neoplasia Information Exchange Database (GENIE). We excluded tumors profiled with "hotspot" mutational panels. Mutational frequencies and mutational load were compared by age groups (<40, 40-60, and >60 years of age), and targetable alterations were identified using the Precision Oncology Knowledge Base levels of evidence (OncoKB). Results: We found 2,500 melanoma patients in the latest GENIE release (v.6.1). After filtering for incomplete mutational panels, we identified 1,194 patients with a common set of 30 genes sequenced across all exons. The mutational load for these 30 genes increased with age, with a mean 2.39 mutations per sample in <40 years of age, 2.92 in 40-60 years of age, and 3.67 in >60 years of age. Across all age groups, the three most commonly mutated genes were BRAF, TP53, and NRAS, but with differences in the frequency of mutations. The top 5 mutated genes in patients <40 years of age (N=98) were BRAF (59%), TP53 (31%), NRAS (17%), PTEN (14%), and APC (13%); in 40-60 years of age (N=354) were BRAF (51%), NRAS (30%), TP53 (26%), APC (13%), and KDR (12%); and in >60 years of age (N=742) were BRAF (38%), NRAS (33%), TP53 (26%), KDR (19%), and APC (18%). We identified 6 targetable alterations meeting OncoKB criteria for some level of evidence predictive of drug response specific to the melanoma indication: BRAF (level 1), KIT (level 2A), NRAS (level 3A), and ATM, MET, and FGFR2 (all level 4) . Additionally, alterations were found in 4 genes with level 1 OncoKB level of evidence in other cancer indications: FLT3, PDGFRA, ABL1, and PIK3CA. Conclusions: Mutational load in melanoma increases with age. Independently, the frequency of different mutations varies across age groups and demonstrates targetable mutations for personalized therapies. Further research is needed to elucidate the age-related biologic differences of melanoma and any impact in outcomes. 1 1. Surgical Oncology, Moffitt Cancer Center, Tampa, FL; 2. University of Pennsylvania, Philadelphia, PA; 3. Emory University, Atlanta, GA; 4. University of North Carolina at Chapel Hill, Chapel Hill, NC. Background: TVEC is an oncolytic virus approved for the treatment of unresectable recurrent melanoma. Recent studies have reported an overall response rate (ORR) of 57% with TVEC monotherapy and an ORR of 67% in combination with immunotherapy (IO). The goal of this study is to characterize the efficacy of TVEC after failure of IO in patients with unresectable metastatic melanoma. Methods: We performed a multi-institutional review of AJCC 7 stage IIIB-IV patients treated with TVEC after failure of IO at 4 US centers from October 2015 to May 2019. Failure was defined as disease progression or recurrence after treatment with IO. Primary outcomes were in-field progression free-(IPFS) and in-field disease-free survival (IDFS) after TVEC with a secondary outcome of overall DFS. Kaplan Meier survival analysis was performed for all outcomes. Results: Of 87 patients, median age was 69 years (interquartile range [IQR] 59-77.5); 47 patients (54%) were male. Prior to TVEC, most patients received 1 IO regimen (67%) as opposed to 2 or more. Most patients received TVEC for in-transit disease (n=69, 79%) of the lower extremity (n=46, 53%). Median follow-up was 7 months (IQR 3-12). Administration of TVEC concurrently with IO or sequentially after failure of IO did not significantly affect in-field response (Table) . In-field ORR after TVEC was 43% (23% CR, 20% PR). Median estimated IPFS was 31 months (95% CI 14.9-NE) and IDFS was 17 months (95% CI 13.5-NE). 19 patients (22%) had no evidence of disease after TVEC of which 4 eventually relapsed (median DFS not estimable). Sequential TVEC after IO yielded ORR of 45% and concurrent TVEC with IO had an ORR of 41%. Conclusions: After failure of initial systemic IO for unresectable metastatic melanoma patients, treatment with TVEC with/without concurrent IO have an in-field ORR of 43% and an in-field CR of 23%. TVEC with IO as second-or later-line therapy is an efficacious treatment option for these patients. Myeloid-derived suppressor cells (MDSC) have been linked to loss of immune effector cell function through a variety of mechanisms such as the generation of reactive nitrogen species. Our group has shown Bruton's tyrosine kinase (BTK) is important for MDSC function. Ibrutinib inhibits BTK activation and is currently used for the treatment of B-cell malignancies. We hypothesized that targeting BTK in MDSC would inhibit their function and enhance immune checkpoint blockade therapy. MDSC-like cell line MSC2 and MDSC isolated from tumor-bearing mice were treated first with ibrutinib and then lipopolysaccharide (LPS). Cell lysates were probed for phosphorylated BTK (p-BTK) and total BTK. Cells were treated with escalating doses of ibrutinib, and viability was assessed. Using a syngeneic murine model (B16F1 in C57BL/6), mice were treated with either anti-PD-L1 therapy or control, with or without ibrutinib. Tumors were then sorted for MDSC (Gr1 + / CD11b + ), and profiled using a Nanostring profile of 730 genes and normalized. Blood samples from melanoma patients in an IRB-approved prospective clinical registry were sorted for MDSC (CD33 + /CD11b + /HLA-DR low/neg ). Relative gene expression of MDSC was studied. Ibrutinib inhibited BTK phosphorylation in both murine and human MDSC as measured by immunoblot (p-BTK:BTK ratio 1.25 in DMSO, 0.6 with Ibrutinib). Ibrutinib was not directly cytotoxic to MDSC in vitro (<10% cell apoptosis), and was noted to reduce expression of inducible nitric oxide synthase (iNOS) by 1.5-fold (p=0.0017). Nanostring revealed that ibrutinib is associated with increased IL-17 signaling. Ibrutinib decreased the levels of circulating MDSC in vivo (8.66% vs 3.73%, p=0.049) and significantly increased the efficacy of single-agent anti-PD-L1 antibody treatment (p=0.014). Compared to control, single-agent anti-PD-L1 antibody therapy showed a 16% reduction in tumor volume, whereas combination therapy resulted in a 66% reduction (p=0.072). These results show that BTK is important in MDSC biology and its downstream effector functions. Targeted inhibition of MDSC is a novel strategy to enhance the immune therapy of advanced melanoma. Background: Although imaging surveillance of high-risk melanoma can lead to early detection of recurrence, intensive imaging remains controversial because its survival impact is unknown. We investigated the impact of imaging intensity on rates of asymptomatic surveillance-detected recurrence (ASDR) and subsequent treatment outcomes in patients with access to immunecheckpoint inhibitors (ICI) and targeted therapy (TT). Methods: Patients were identified with resected malignant melanoma, AJCC (7th edition) stage IIB-IIIC, undergoing imaging surveillance at The Ottawa Hospital Cancer Centre between January 2006 to January 2016. Surveillance and recurrence characteristics were retrospectively collected, including imaging, symptom, treatment, and survival data. Univariate (t-test, 2) and multivariate Cox regression analyses were conducted. Results: Of 353 high-risk melanoma patients (stage IIB 24%, IIC 19%, IIIA 27%, IIIB 16%, IIIC 14%) , 71 (45%) had ASDR, and 88 (55%) had symptomatic recurrence (SR). Shorter imaging intervals identified more ASDR (57% 0-6 months; 34% 6-12 months; 33% >12 months (p=0.03)). ASDR had better prognostic factors than SR (<3 metastatic sites (43 vs 21%, p=0.003), normal LDH (53 vs 38%, p=0.09), brain metastases (11 vs 40%, p<0.001)) and received more systemic treatment (72 vs 49%, p=0.003; ICIs 55 vs 31%, p=0.002; TT 8 vs 13%, p=0.41). ASDR had better survival outcomes than SR on ICI treatment (2-year OS 56 vs 31%, p<0.001). Median overall survival from surveillance start was 39.6 (ASDR) vs 22.8 (SR) months (p<0.001). ASDR was independently associated with survival (HR 0.47, 95% confidence interval 0.29-0.78, p=0.003) adjusting for stage, sex, age, disease burden, LDH, era of recurrence, brain metastases, and ICI/TT treatment. Conclusions: These real-world data support further study on intensified imaging surveillance protocols for high-risk resected melanoma, as ASDR was associated with superior survival outcomes from ICI therapy. Durham, NC; 4. Emory University, Atlanta, GA; 5. Fox Chase Cancer Center, Philadelphia, PA; 6. Peter MacCallum Cancer Centre, Melbourne, VIC, Australia; 7. The Alfred Hospital, Melbourne, VIC, Australia; 8. Princess Alexandra Hospital, Brisbane, QLD, Australia; 9. University of Texas Medical Branch, Galveston, TX; 10. Melanoma Institute Australia, Sydney, NSW, Australia. Introduction: ILI is a minimally invasive procedure for delivering highdose chemotherapy to extremities affected by locally advanced or in-transit melanoma. This study compares outcomes of melanoma patients treated with ILI in the US and Australia. Methods: AJCC 7 stage 3B/3C melanoma patients treated with ILI at 4 US and 5 Australian centers between 1992-2018 were identified. Demographic and clinicopathologic characteristics were collected. Primary outcomes were treatment response, in-field (IPFS), distant (DPFS) progression-free survival and overall survival (OS) evaluated by Kaplan-Meier method. Multivariate analysis evaluated whether availability of immunotherapy (once approved) in each country affected outcomes. Results: 687 patients were identified; median age 71 yrs (IQR 62-79); majority female (n=412, 60%). At the time of ILI, 383 (56%) patients were stage 3B and 304 (44%) were 3C, most with lower extremity disease (n=610, 88%). More ILI were performed in Australia for stage 3B disease compared to the US (62% vs 46%; p<0.001). US patients were younger (median age 68 vs 73 yrs, p<0.001) and had greater median limb volume (7.5 vs 6 L, p<0 .001), longer circulation (30 vs 21 min, p<0.001) and ischemia times (58 vs 49 min, p<0.001) but with no significant differences in median melphalan dose (43 mg, p=0.9) (Table) . Overall response rate (complete+partial response) was higher among Australian patients (73% vs 53%, p<0.001). Among stage 3B patients, the Australian cohort had better IPFS (p=0.001) with similar DPFS and OS. Among stage 3C patients, the US cohort had better OS (p<0.0001) with similar IPFS or DPFS. Availability of immunotherapy did not affect IPFS or DPFS in either country; in the US, stage 3C patients who received ILI after ipilimumab approval in 2011 had significantly improved OS (HR=0.63, p=0.02). Conclusions: In Australia, patients were treated at an earlier disease stage (3B) than the US with better IPFS only for stage 3B. In the US, stage 3C patients had better OS associated with approval of immunotherapy. Differences in procedural circulation time may also contribute to differences in response. Introduction: Succesful treatment of patients with colorectal peritoneal carcinomatosis (PC) highly depends on complete surgical tumor resection. Surgical outcomes can be improved by intraoperative imaging using a tumor-targeting antibody conjugated to a fluorophore and a radiotracer. This enables preoperative radionuclide imaging, real-time intraoperative fluorescence imaging and gamma detection. In this study we investigate the feasibility, accuracy and safety of CEA-targeted preoperative SPECT/CT and intraoperative fluorescence imaging in patients with colorectal PC. Methods: In this single arm intervention study patients will receive an intravenous dose of the CEA-targeting tracer 111 In-DOTA-labetuzumab-IRDye800CW. Four days after injection a SPECT/CT is performed to determine intraabdominal tumor load. Two days later surgical resection extended with fluorescence imaging and radioguidance is performed. After surgery, the peritoneal cavity will be examined for residual disease with fluorescence imaging. Resected specimens are analyzed microscopically, immunohistochemically (CEA) and by gamma counting. First, a dose escalation with 3 doses (n=15) is performed. In the expansion cohort, 14 patients will undergo multimodal image-guided surgery with the optimal tracer dose. Results/Discussion No adverse events were noticed in the first 7 patients. Fluorescence imaging was feasible for real-time visualization of peritoneal tumour deposits in all patients. Fluorescence imaging was able to distinguish vital tumor lesions from fibrotic lesions after neoadjuvant systemic therapy. Pre-operative SPEC/CT imaging revealed extra peritoneal nodal metastasis not seen on conventional imaging modalities in two patients. Immunohistochemical analysis showed specific binding of the fluorescent and radiolabelled tracer to CEA-expressing Introduction: Treatment of peritoneal surface malignancies with cytoreductive surgery and HIPEC has substantial risk of local recurrence and causes off-target toxicity. We sought to reinvent HIPEC utilizing a gold nanorod (GNR) platform to provide precision delivery of mild hyperthermia (42C) to increase chemotherapeutic efficacy and reduce toxicity. Methods: A biocompatible hydrogel film containing GNRs and cisplatin (CisPt) was designed to create controlled, mild hyperthermia that can be generated from NIR laser treatment of GNRs. The concentration of GNRs and duration of hyperthermia was determined using dose-response curves to titrate to goal temperature parameters. Loading and release efficiency of CisPt at an IC50 dose was quantified. SNU-16 was treated with GNR-, CisPt-or GNR+CisPt-film and NIR laser to determine cytotoxicity. Results: Thirty seconds of NIR irradiation to film with GNR concentration of 2.35x10 10 particles/ml achieved mild hyperthermia (42C) in a controlled fashion. CisPt-film was loaded with 80% efficiency and released with 90% efficiency. Treatment of SNU-16 with GNR+CisPt-film and NIR laser to 42C resulted in significantly increased cytotoxicity compared to CisPt-film (p<0.05). Conclusion: NIR laser treatment of hydrogel film containing gold nanorods and chemotherapy improved the efficacy of cisplatin in gastric cancer cells. Preclinical evidence of improved efficacy utilizing a biocompatible hydrogel GNR-CisPt film could lead to clinical trials in peritoneal malignancies. If successful, treatment of other cancers could benefit from this film as a method to decrease the incidence of local recurrence in any surgical resection bed. Background Peritoneal metastasis is common in gastrointestinal and pancreatic cancer. Current treatment and life expectancy of patients with peritoneal carcinomatosis (PC) are limited. CMP-001, a virus-like particle composed of Q bacteriophage capsid protein encapsulating a Toll-Like Receptor 9 agonist CpG-A oligodeoxynucleotide, activates plasmacytoid dendritic cells (pDC) and triggers interferon alpha (IFN ) release, leading to a cascade of anti-tumor immune response. Methods To evaluate the ability of CMP-001 for triggering immune response in patients with PC, an ex vivo model was established. Malignant ascites or peritoneal lavage fluids were collected from patients with PC. Immune profiles of isolated cells were obtained using flow cytometry. Peritoneal cells were stimulated by CMP-001 and IFN release was quantified by ELISA. To evaluate the anti-tumor response in vivo, murine PC models were generated using mouse cancer cell lines (Panc2 & MC38) and treated with intraperitoneal (ip) CMP-001 or saline on days 5, 9 and 13 after tumor challenge. Kaplan Meier method was used for survival analysis. Necropsy and immunophenotyping of peritoneal tumor microenvironment were performed on day 26. Results pDC was found in 1% (0.1-3.9%, N=15) of the isolated peritoneal cells. Ex vivo CMP-001 stimulation of the peritoneal cells released on average 1 ng/ml IFN (1.8-4700 pg/ml, N=9) . IFN concentration was directly proportional to %pDC present in the peritoneal cell mixture (r=0. 8, p=0.012) . In murine PC models, ip CMP-001 treatment elicited a robust  anti-tumor immune response including an increase in chemokines (RANTES,  MIP1 , MIP1 ), proinflammatory cytokines (IFN , IL6, IL12), and peritoneal/  tumor immune infiltrates (CD4+/CD8+ T cells & NK cells) . This immune response led to an improved survival in both Panc2 (median: 35 vs 28 days) and MC38 (median: 57 vs 35 days) PC models (P<0.05). Conclusion CMP-001 is a novel immunotherapeutic agent currently on clinical trial for advanced melanoma with impressive treatment response. Our data suggest CMP-001 can be used to treat patients with PC. Clinical development of CMP-001 in PC is under discussion.

Tumor Vessel Characterization in Peritoneal Carcinomatosis Using Intravital Microscopy E.M. Gabriel,* S.P. Bagaria, K. Knutson, M. Wallace. Surgery, Mayo Clinic Florida, Jacksonville, FL. Introduction Intravital microscopy (IVM) allows for the real-time, direct observation of tumor-associated vessels at the time of tumor resection. We have previously shown that IVM is feasible in the observation of tumor vessels associated with melanoma and sentinel lymph nodes. The purposes of this pilot study were to determine the feasibility of IVM in peritoneal carcinomatosis at the time of cytoreductive surgery and to analyze vessel characteristics in relation to tumor responses to systemic therapies. Methods This was a pilot study of IVM for patients with peritoneal carcinomatosis of any histology. Fluorescein (iv) was used to enhance the tumor vessel observations. The fluorescent microscope was supplied by Mauna Kea Technologies. Postobservational analyses were used to measure tumor blood flow, vessel diameter, dye uptake as a measure of vessel functionality, and fluorescein diffusion rate into the surrounding tissues. Results Twenty-five patients have been enrolled to date. Histologies included disseminated peritoneal adenomucinosis (DPAM) of appendiceal or ovarian origin (8) , mucinous colorectal carcinoma (6) , ovarian carcinoma (6), endometrial carcinoma (1), gastric adenocarcinoma (1), sarcoma (1 dedifferentiated liposarcoma and 1 pleomorphic sarcoma), and peritoneal mesothelioma (1) . Excluding DPAM histology, all patients received neoadjuvant chemotherapy and/or targeted therapy, and 13/17 (76%) had a partial response. Tumor vessels were visible in all patients except for those with DPAM, yielding a feasibility rate of 68%. Tumor vessels were characterized by aberrant architecture, bidirectional flow, and rapid dye extravasation. Up to 50% of tumor vessels did not support flow as noted by the absence of fluorescein uptake. Interestingly, in areas of tumor response, vessels showed normalization of blood flow and architecture similar to non-tumor vessels. Conclusions This is the first study to show that IVM permits the observation of tumor vessels associated with peritoneal carcinomatosis. Our early results suggest that tumor vessel normalization may be related to systemic therapy responses. Long-term analysis will determine if these findings relate to survival outcomes. Background: Noninvasive tests for peritoneal metastasis (PM) detection lack sensitivity. Genome-wide mapping of 5-hydroxymethylcytosine (5hmC) on nanogram quantities of peripheral plasma circulating cell-free DNA (PcfDNA) was previously shown to differentiate non-metastatic colorectal cancer from healthy subjects. We aimed to examine if patients with colorectal cancer (CRC), high grade appendiceal cancer (HGA) or low grade appendiceal cancer (LGA) with PM have distinct signatures of 5hmC in PcfDNA compared to each other and to patients matched for tumors without PM. Methods: We analyzed plasma samples from a prospectively collected tissue bank. To correlate 5hmC signatures with intraoperative findings, only patients who underwent abdominal surgery in proximity to plasma collection were selected. Key steps of PcfDNA processing included extraction from plasma, nano-hmC-Seal chemical labeling and enrichment of 5hmC-modified fragments, next-generation sequencing, and mapping to the reference human genome. DESeq2 R package was finally used to compare relative 5hmC enrichment and detect distinct 5hmC signatures according to disease histology and PM presence. Results: Plasma samples were collected between 11/2016 -3/2019 from 46 patients with CRC (n=21), HGA (n=17) and LGA (n=8). Of those, 32 (70%) had PM based on intraoperative findings (median peritoneal cancer index score = 15, range 2-39) and 14 did not have PM. An average of 24.1 million paired-end reads were sequenced for each sample. Four samples (8.7%) were excluded from the analysis due to low sequencing coverage or high duplication level. Unique 5hmC enrichment patterns were found to discriminate with p < 0.05 between CRC PM and HGA PM (n=616 differentially hydroxymethylated genes (DHMGs)), CRC/HGA PM and LGA PM (n=1074 DHMGs), and CRC/HGA PM and CRC/HGA patients without PM (n=1576 DHMGs, Image 1). Conclusions: Distinct signatures of 5hmC in PcfDNA could differentiate patients with CRC/HGA/LGA PM from each other and from patients with similar tumor histologies without PM. Thus 5hmC signatures in PcfDNA might potentially serve as a sensitive marker of occult PM.

5-hydroxymethylation gene enrichment map (right) and principal component analysis (left) of plasma samples collected from patients with colorectal / high grade appendiceal peritoneal metastasis, vs. patients with colorectal / high grade appendiceal cancer without peritoneal metastasis. Introduction: Cytoreductive surgery with hyperthermic intraperitoneal chemotherapy (CRS/HIPEC) is an accepted treatment for Appendiceal Mucinous Neoplasm (AMN) that could potentially confer long term survival. This major operation should be reserved for appropriate candidates in whom the benefits outweigh the risks. Clinical decision making is more challenging when complete cytoreduction is not feasible. Nevertheless, some patients still benefit from long term survival after incomplete CRS/HIPEC. There is a lack of a robust predictive tool that can assist in clinical decision making. Methods: We have quantified tumor gene expression with a custom Nanostring 148-gene panel. Using signed non-negative matrix factorization algorithms, we have classified AMN into 3 molecular subtypes with distinct prognostic signatures: Immune Enriched (IE), Mixed (M) and Oncogene Enriched (OE). Expression of Nanostring panel of 82 AMN patients with incomplete CRS/ HIPEC (R2 resection) was assessed and genetic signatures were identified. Log-rank test and Kaplan-Meier survival curves were used to compare the overall survivals (OS). Cox proportional hazard models were used to identify the genes associated with OS. Results: We studied 82 patients with R2 resection. Genes associated with survival were distinct for each pathologic grade. They were involved in various signaling pathways associated with innate and adaptive immunity, carcinogenesis cascades and metastatic pathways including ERBB signaling, Wnt/ -catenin, PI3K/Akt, mitogen-activated protein kinase and Epithelial-mesenchymal transition. Median OS were 4.8yr (IE), 3.5yr (M) and 1.4yr (OE) . OE was associated with a significantly lower survival for OE compared to others (HR=2.53; 95%CI: 1.40-4.57, p=0.002) . This was more prominent in the high-grade subset (Median survival 1yr, HR=2.79; p=0.015) . Conclusion: Genetic signatures can identify AMNs with poor outcomes who would less likely benefit from incomplete CRS/HIPEC. This has potential clinical implications as a molecular tool that can be utilized for clinical decision making after prospective validation Background: While parenchymal hepatic metastases were previously considered a contraindication to cytoreductive surgery (CRS) and heated intraperitoneal chemotherapy (HIPEC), liver resection (LR) is increasingly performed concomitantly with CRS/HIPEC. As this practice continues to expand, identification of preoperative factors associated with poor outcomes is paramount. Methods: Patients from the US HIPEC Collaborative (2000) (2001) (2002) (2003) (2004) (2005) (2006) (2007) (2008) (2009) (2010) (2011) (2012) (2013) (2014) (2015) (2016) (2017) with invasive appendiceal or colorectal adenocarcinoma undergoing primary, curative intent CRS/HIPEC with CC0-1 resection were included. Liver resection was defined as a formal parenchymal resection. Primary endpoints were postoperative complications and overall survival (OS). Results: 658 patients were included. Average age was 54 years and 45% were male; 83 (15%) underwent liver resection of colorectal (58%) or invasive appendiceal (42%) metastases. Liver resection patients had more complications (81 vs 60%; p=0.001), greater number of complications (2.3 vs 1.5; p<0.001), and required more reoperations (22 vs 11%; p=0.007) and readmissions (39 vs 25%; p=0.014) than non-liver resection patients. Liver resection patients had decreased OS (2- year OS 62% vs 79%, p<0.001), which persisted on multivariable Cox regression when accounting for PCI and histology type. Preoperative factors associated with decreased OS on multivariable analysis in patients undergoing liver resection included age <60 years (HR:3.61), colorectal histology (HR:3.84), and multiple liver tumors (HR:3.45) (all p<0.05). When assigning one point for each factor, there was an incremental decrease in 2-yr survival as the risk score increased from 0 to 3 (0: 100%; 1: 91%; 2: 58%; 3: 0%; p<0.001, Figure 1 ). Conclusions: As concurrent liver resection with CRS/ HIPEC has become more common, we created a simple risk score to stratify patients considered for CRS/HIPEC with liver resection. These data aid in striking the balance between an increased perioperative complication profile with the potential for improvement in overall survival. Introduction: For patients with colorectal peritoneal metastases (CRPM) cytoreductive surgery with hyperthermic intraperitoneal chemotherapy (CRS-HIPEC) can offer significant survival benefit. Early recurrence remains a significant hurdle. Preoperative identification of patients at risk for early recurrence and poor survival following CRS-HIPEC would enable optimal treatment decisions and improve patient outcomes. Methods: Patients who underwent CRS-HIPEC for CRPM from 2000-2018 were included. A minimum p-value approach was used to determine an optimal definition of early post-CRS-HIPEC recurrence based on prediction of overall survival. Patients were divided into early (ER) and late recurrence (LR) cohorts. Risk factors for ER were assessed by logistic regression. A predictive score for ER was generated via beta coefficient method using preoperative variables. Results: 402 patients were identified, 339 (84.3%) had documented recurrence. Optimal length of post-operative RFS to distinguish ER (n=137, 40.4%) versus LR (n=202, 59.6%) was 8 mos (p<0.001). ER patients had significantly shorter median OS post-CRS-HIPEC compared to LR (13.8 vs 32.4 mos, p<0 .001). Preoperative risk factors for ER included female sex (OR 1.81, p=0.009), hepatic lesions (OR 1.71, p = 0.045), poorly differentiated tumors (OR 1.68, p=0.04), PCI scores 20 (OR 1.44, p=0.10), progression on neoadjuvant (OR 2.03, p=0.023), and <6 mos from presentation to diagnosis of CRPM (OR 1.45, p=0.10 ). Using this model, patients were assigned risk scores from 0-9. Scores 6 were classified as high risk (FIGURE) . High risk patients (n=29, 7.2%) had an observed rate of ER of 79%. Overall 2-and 5-year survival rates for high risk patients were 13% and 3% respectively. The model showed fair discrimination (AUC 0.7) and good calibration (Hosmer-Lemeshow GOF p= 0.81). Discussion: Early recurrence, defined here as recurrence within 8 mos of CRS-HIPEC, predicts markedly worse OS following surgery. Preoperative factors can accurately determine patients at high risk for ER using a simple risk score and identify patients in whom CRS-HIPEC for CPRM is futile.

Introduction: Failure to thrive (FTT) has come to describe a complex syndrome of nutritional failure, functional decline, and loss of independence in adults. Readmission following cytoreductive surgery and hyperthermic intraperitoneal chemotherapy (CRS-HIPEC) is often a harbinger of functional decline and poor oncologic outcomes. Rates of FTT following CRS-HIPEC are unknown. This study aimed to determine underlying causes, risk factors, and prognostic significance of FTT for patients undergoing CRS-HIPEC. Methods: We retrospectively reviewed patients who underwent CRS-HIPEC from 2010 to 2018. All unplanned readmissions within 90 days of discharge were examined. FTT was determined by FTT ICD codes and chart review. We compared baseline characteristics, oncologic data, perioperative outcomes, and survival among three patient groups: FTT readmission, non-FTT readmission, no readmission. Results: 1,068 discharges following CRS-HIPEC were identified. 379 patients (36%) were readmitted within 90 days, of which 134 (35%) were labeled as FTT. Nutritional failure (n=70, 52%) and infection (n=44, 33%) were the primary co-diagnoses for FTT. Patients with FTT were older, had a higher proportion of colorectal primaries and worse peri-operative functional status. They had greater tumor burden with more complex resections, longer hospital stays, and more postoperative complications (all p<0.05). Stoma creation (OR 4.0, p<0.001), VTE in hospital (OR 2.4, p=0.01), and age over 50 (OR 2.1, p=0 .002) were the strongest predictors for FTT. FTT readmission was associated with worse median overall survival compared to non-FTT readmission or no readmission on both univariate (30 mos vs 48 mos and 80 mos respectively, p<0.001) and multivariate analysis (HR 1.4, p=0.04) after controlling for oncologic, hospitalization, and baseline factors. Conclusions: FTT is a common occurrence following CRS-HIPEC which adversely affects both quality and quantity of life, and appears to be associated with baseline functional status and extent of cytoreduction. Peri-operative strategies for improving nutrition and activity in this high-risk cohort along with early intervention in FTT may improve both surgical outcomes and overall survival.

Outcomes of Palliative CRS/HIPEC for Patients with Advanced Peritoneal Carcinomatosis from Appendiceal Cancer R.J. Hendrix, 1 * C. Corban, 1 J.M. Cloyd, 2 A. Ahmed, 2 F.M. Johnston, 3 J. Greer, 3 T.E. Grotz, 4 J.L. Leiting, 4 K.F. Fournier, 5 A.J. Lee, 5 S. Dineen, 6 S. Dessureault, 6 J. Veerapong, 7 J. Baumgartner, 7 C.N. Clarke, 8 H. Mogal, 8 C.A. Staley, 9 M.Y. Zaidi, 9 S. Patel, 10 V. Dhar, 10 D.E. Abbott, 11 C. Pokrzywa, 11 K. Lafaro, 12 B. Lee, 12 L. Lambert. 13 1. Surgery, University of Massachusetts Medical School, Worcester, MA; 2. The Ohio State University Wexner Medical Center, Columbus, OH; 3. Johns Hopkins University, Baltimonre, MD; 4. Mayo Clinic, Rochester, MN; 5. University of Texas MD Anderson Cancer Center, Houston, TX; 6. Moffitt Cancer Center, Tampa, FL; 7. University of California San Diego, San Diego, CA; 8. Medical College of Wisconsin, Milwaukee, WI; 9. Winship Cancer Institute, Emory University, Atlanta, GA; 10. University of Cincinnati College of Medicine, Cincinnati, OH; 11. University of Wisconsin, Madison, WI; 12. City of Hope National Medical Center, Duarte, CA; 13. Hunstman Cancer Institute University of Utah, Salt Lake City, UT. Background: The impact of cytoreductive surgery (CRS) and hyperthermic intraperitoneal chemoperfusion (HIPEC) on the palliation of advanced peritoneal carcinomatosis (PC) from appendiceal cancer is unknown. Methods: The US HIPEC Collaborative was retrospectively reviewed to compare the outcomes of patients with appendiceal cancer who underwent CRS/HIPEC. Patients were stratified by operative intent and completeness of cytoreduction (CC) into three categories: curative (curative intent and CC 0/1 resection), incomplete (curative intent and CC 2/3 resection), and palliative (palliative intent and CC 2/3 resection). Results: CRS/HIPEC was performed in 1,311 patients with PC of appendiceal origin: 1,086 (83%) curative; 182 (14%) incomplete; 43 (3%) palliative. Mean age was 55 years and 58% were female. Peritoneal Cancer Index correlated with CC score and operative designation (curative13.9 vs. incomplete 24.5 vs. palliative 31.9, p<0.0001; Table 1 ). Length of surgery was significantly longer for curative CRS/HIPEC (7.7 vs. 6.5 vs. 6.5 hours, p<0.0001 ). There were no significant differences in hospital length of stay (11.9 vs. 12.4 vs. 12.2 days), 30-day complications (57% vs 56% vs 53%, p=0.89), Clavien-Dindo grade >III complications (18% vs. 21% vs. 30%, p=0.26), reoperations (10% vs. 9% vs. 12%, p=0.82), or readmissions (18% vs. 15% vs. 19%, p=0.62) . Median survival after palliative CRS/ HIPEC was 2.5 years. After palliative CRS/HIPEC, 80% experienced partial/ complete resolution of ascites symptoms, 100% partial/complete improvement in obstructive symptoms, and 90% had improvement in pain symptoms. Conclusion: Palliative intent CRS/HIPEC provides high rates of symptom control and acceptable survival in patients with advanced appendiceal PC. Appropriate patient selection is essential and efforts should be made to improve operative times, length of stay and minimize complications. Objective. Low-grade appendiceal mucinous neoplasms (LAMN) are tumors that frequently present with peritoneal spread of either acellular (AM) or cellular mucin (CM). We aimed to determine how mucin type and distribution affect recurrence and survival. Methods. A retrospective singlecenter cohort study was conducted using a prospective database. Newly diagnosed LAMN patients with AM vs CM treated with cytoreductive surgery/ hyperthermic intraperitoneal chemotherapy (CRS/HIPEC) were compared. Postoperative pathology reports were reviewed to assess each altered abdominal zone. Survival was analyzed with the Kaplan-Meier method. Results. Of 202 LAMN patients identified, 81 with iterative, palliative, and aborted procedures, or who lacked pathology information were excluded. Of 121 included patients, 50 (41%) had AM and 71 (59%) had CM. Peritoneal cancer index was lower in AM vs CM (21 vs 31, p=0.004), but the majority had a complete cytoreduction (CC) (98% vs 96%, respectively [p=0.64]). The 5-year progression-free survival (PFS) was higher in AM vs CM (96% vs 69.8%, p=0.002). The 5-year overall survival (OS) was not significantly different between AM and CM (96% and 88%, p=0.24). Patients in the CM group had zones with epithelial cells as well as zones with only mucin. CM subgroup analysis showed significant differences in 5-year PFS between patients with 1-3, 4-7 and 8-10 zones with cells, 95.2%, 68.4%, and 35.7%, respectively (p<0.001). However, 5-year PFS was not significantly different based on numbers of all altered zones (1-3, 4-7, and 8-10) Introduction: The Leapfrog group identified surgeon and hospital procedure volume standards for high-risk surgical oncology procedures. However, variation in outcomes among high-volume surgeons at highvolume hospitals that meet this standard is unknown. Methods: The Medicare 100% Standard Analytic File (2013-2017) was queried for esophagectomy, lung resection, pancreatectomy, and proctectomy for cancer. Mixed-effects analyses assessed the association between Leapfrog annual volume standards for surgeons (esophagectomy 7, lung resection 15, pancreatectomy 10, proctectomy 6) and hospitals (esophagectomy 20, lung resection 40, pancreatectomy 20, proctectomy 16) and major complications and 90-day mortality. Results: Among 6, 933 esophagectomies, 67, 808 lung resections, 18, 196 pancreatectomies, and 27, 501 proctectomies, Despite this association, for the average patient at the average high-volume hospital, there was a two-fold difference in the adjusted complication rate between the best and worst-performing high-volume surgeon for all four operations (esophagectomy: 22-53%; lung resection: 7.9-17.3%; pancreatectomy: 14.3-33.5%; proctectomy: 12.7-23.6%). Additionally, there was four-fold variation in the adjusted 90-day mortality rate (range=2.7-10%) among patients undergoing esophagectomy by a high-volume surgeon at a high-volume hospital. Conclusions: While there is an association between the Leapfrog surgeon/hospital volume standards and improved postoperative outcomes, there is wide variation in complication and mortality rates across high-volume surgeons. Quality metrics should be continuously evaluated across surgeons and hospital systems. Introduction: Sixty percent of hospitals today are part of larger health systems. Proponents of hospital consolidation tout its potential to reduce health spending and improve outcomes, but available evidence suggests this promise is unrealized. Variation in costs and outcomes within systemsmay highlight opportunities for collaborative quality improvement and practice standardization. To assess this potential, we sought to measure variation in episode spending within and across hospital systems among Medicare beneficiaries undergoing complex cancer surgery. Methods: Using 100% Medicare claims data, we identified fee-for-service Medicare patients undergoing elective pancreatectomy, esophagectomy, lung resection, or proctectomy for cancer from 2014-2016. We calculated risk-adjusted, price-standardized payments for the surgical episode from admission through 30 days post discharge. We then assessed reliability-adjusted variation at hospital and system levels. We defined centralization as the proportion of operations performed by the highestvolume center within each system. Results: The average within-system variation ranged from $1,282 (6%) for proctectomy to $3,858 (9%) for esophagectomy. Greater centralization was associated with lower cost for all cohorts. For example, there was a $474 reduction in mean episode cost for esophagectomy for each 20% increase in the degree of centralization within systems. Index hospitalization payment explained 37% and 48% of the variation in spending between hospitals for pancreatectomy and esophagectomy, respectively. For lung resection and proctectomy, 37% of the variation between hospitals was explained by post-acute payment Conclusion: In this analysis of Medicare patients undergoing complex cancer surgery, we found wide variation in surgical episode spending both within hospital systems. Additionally, we found significant reduction in mean episode payments with increased centralization. System leaders may seek to better understand variation in practices between their hospitals in order to standardize care and reduce variation in outcomes, utilization, and costs.

Notes: Hospital and system characteristics were weighted at the episode level, to reflect differences in hospital and system volume. To measure consolidation within each system we calculated a Herfindahl-Hirschman Index (HHI), which reflects the degree of concentration within each system. An index below 0.15 indicates an unconcentrated industry, between 0.15 to 0.25 indicates moderate concentration, and above 0.25 indicates high concentration. Hub concentration is defined centralization as the proportion of operations performed by the highest-volume center within each system. Percentages of total episode payment difference were calculated by subtracting each payment component in the lowest cost hospital within the system from the payment component in the highest cost hospital within the system and dividing this by the difference in total episode payments between the lowest and highest cost hospitals within the system. BACKGROUND: Prior studies have shown that care fragmentation after surgery is associated with worse outcomes. To aid in improving outcomes by targeting interventions to at-risk patients, we evaluated patient-level variables associated with care fragmentation in readmissions following hepatopancreatobiliary (HPB) and gastric oncologic surgery. METHODS: Patients undergoing HPB or gastric oncologic surgeries were identified from select State Inpatient Databases from the Healthcare Cost and Utilization Project from 2006-2014. Follow-up ended at 90 days after discharge. The primary exposure was readmission to any hospital other than the index surgical hospital ('outside hospital', OSH). Kruskal-Wallis and logistic regression tests were used. RESULTS: Of 7,536 patients readmitted within 90 days of discharge following HPB and gastric oncologic surgery, 28% (n=2,124) were readmitted to an OSH, where mortality was 50% higher than mortality for patients readmitted to the index surgical hospital (8.0% vs 5.4%; OR 1.5, 95% CI 1. 2 -1.9) . Patients readmitted to an OSH were older (median age 67 years vs 65; p<0.01), resided more than twice as far from the index surgical hospital (median distance of 24 miles vs 10; p<0.01), and were readmitted much later (median of 25 days after discharge vs 12; p<0.01). Charlson Comorbidity Index median scores were identical for these groups (14). After adjustment for differences in age and distance from the index surgical hospital, socioeconomic variables -race, insurance status, and income by patient ZIP code -were not significantly associated with care fragmentation. CONCLUSIONS: In this large multi-state cohort, care fragmentation in readmissions following HPB and gastric oncologic surgery was associated with a 50% increase in mortality. Increasing age, living a greater distance from the index surgical hospital, and delayed presentation were associated with care fragmentation. This data can be used to target post-discharge interventions to at-risk patients, with the goal of reducing care fragmentation and improving patient outcomes.

Introduction: There are currently no nationally endorsed quality measures for gastric cancer care. Given the importance of quality multimodality therapy (MMT) for patients with locally advanced disease, the intent of this study was to evaluate whether hospital performance in the delivery of MMT might be a potential quality target in gastric cancer care. Methods: National cohort study of 4,412 patients with resected, locally advanced gastric cancer (i.e.: cT2 or greater and/or cN+) treated at 908 hospitals within the National Cancer Data Base (2006) (2007) (2008) (2009) (2010) (2011) (2012) (2013) (2014) (2015) . Hospitals were categorized based on performance in achieving an adequate lymphadenectomy [AL], defined as 15 nodes, and administration of adjuvant therapy [AT], defined as pre-or perioperative chemotherapy or postoperative chemotherapy and radiation. High performing hospitals in the delivery of quality MMT were those achieving both an AL 80% of the time (defined by the Commission on Cancer) and delivering AT to 75% of patients. The association between hospital performance and overall risk of death was evaluated using Cox shared frailty modeling. Results: Over the course of the study, 45.6% and 56.5% of hospitals met the AL and AT benchmarks, respectively, and 28.2% of hospitals were high performers in delivering MMT. Over time, the proportion of hospitals meeting the AL ( Introduction: The focus of centralized tertiary medical centers as the backbone of oncologic care has increased with more specialized treatment. Little data exists examining outcomes in patients who undergo surgery at highvolume centers (HVC) and seek out coordinated multicenter care (CMC) for their additional treatment. Methods: Patients were identified from the 2004-2015 ACS NCDB® who were diagnosed with AJCC 8 th Ed Stage II-IV rectal adenocarcinoma and underwent resection at an HVC, defined as 16 cases/ year reported every year. CMC was defined as receiving care at more than one facility. Outcomes were 1-year (1YS) and 5-year (5YS) survivorship. CMC and single-center care (SCC) cohorts were propensity-score matched (PSM) 1:1 for survival analysis. Results: Of the included 14,852 patients, CMC occurred in 13,130 (88.4% CMC occurs in the majority of patients receiving surgical care at HVC for rectal adenocarcinoma and is non-inferior to SCC for short-and long-term outcomes. Delivery of complex oncologic treatment can be effectively administered through a multicenter approach, but should be tailored with shared decision making between providers and patient to optimize quality of life without compromising care. Background: Adherence to semi-annual screening guidelines of at-risk patients for hepatocellular carcinoma (HCC) at safety-net hospitals is abysmally poor. Only 23% of patients at Grady Health System had a screening exam within 1-year of a new HCC diagnosis. For those undergoing initial screening, adherence to continued imaging is only 15-20% nationally. Hepatitis C (HCV) induced fibrosis/cirrhosis remains the most common etiology of HCC (75%) in safety-net populations. We aimed to improve screening of patients with HCV fibrosis/cirrhosis. Methods: Utilizing a newly-formed HCV treatment clinic that sees approximately 25 new patients/month, an HCC screening program was initiated in April 2018 for HCV patients with grade 3/4 fibrosis/cirrhosis consisting of q6month imaging alternating MRI with ultrasound. Hospital resources were allocated to hiring a nurse practitioner, medical assistant, patient navigator, and database manager to support the program. Results: In the 17-months since initiation, 194 patients have been enrolled in the screening program, with an average age of 61 years, 79% Black race, and 87% having public or no insurance. Adherence to imaging dramatically improved with 84%, 60%, and 55% of patients completing all recommended screening for their first, second, and third evaluation, respectively. Eleven patients (6%) were diagnosed with a new HCC and all were referred for definitive treatment. Conclusions: Implementation of a successful screening program for HCC at a safety-net hospital is feasible. Quality improvement initiatives are underway to elucidate patient barriers, optimize adherence, and to expand to other highrisk populations. Efforts to implement this model at other safety-net hospitals are forthcoming. Background: Machine learning (ML) is a subfield of artificial intelligence that builds models, constructed from raw data, to perform computational tasks. A potential application for ML is outcome prediction using big data. Recent technological advances enabled data to be readily analyzed with these techniques to draw inferences from specific populations. To implement ML models, a national database of colon cancer patients was selected to predict negative outcomes following surgery. Methods: Surgical colon cancer patients were identified using the 2013 National Cancer Database (NCDB). The negative outcome was defined as a composite of 30-day unplanned readmission and 30-and 90-day mortality. Predictive variables were selected and missing values were imputed. Data was split into training and test sets. ML models, including random forest and XGBoost, were built and compared with conventional, parametric penalized logistic regression. To account for unbalanced outcomes, a synthetic minority oversampling technique (SMOTE) was implemented and applied using XGBoost. Results: Analysis included 528,060 patients. The negative outcome occurred in 12% of patients. Model building utilized 30 variables. SMOTE included 342,272 patients and increased the negative outcome to 43%. Receiver operating characteristic (ROC) curves were built. The primary metric was area under the curve (AUC). In comparison to penalized logistic regression (AUC 0.730, 95% CI: 0.725-0.735), AUC's for ML algorithms ranged between 0.748 and 0.757, with the random forest model (AUC 0.757, 95% CI: 0.752-0.762) outperforming XGBoost (AUC 0.756, 95% CI: 0.751-0.761) and XGBoost trained on SMOTE data (AUC 0.748, 95% CI: 0.743-0.753). Conclusions: We show that a large, national cancer registry of patients undergoing surgery for colon cancer can be utilized by applying ML algorithms to predict negative clinical outcomes with differential levels of discriminative ability. Random forest and XGBoost models outperformed the logistic regression model. Our results support the implementation of ML models to analyze clinical datasets to accurately define patient risk.

Introduction The American Society of Anesthesiologists classification (ASA) and age are the most established predictors of frailty in outcome assessment, but ASA is prone to misclassification in cancer patients. The predictive performance of the ACS-NSQIP modified frailty index (mFI) in determining failure to rescue (FTR) an anastomotic leak (AL) following a colectomy for colorectal cancer has not been established. Methods Patients undergoing colectomy for colorectal cancer were identified from the 2012-2016 ACS-NSQIP database. Primary outcome was FTR after AL. The mFI is comprised of functional status, diabetes, history of COPD, history of congestive heart disease (CHD) and hypertension requiring treatment. The predictive performance of a model using age and ASA (model 1) to predict FTR after AL was compared to a model using age, ASA and mFI (model 2) as well as to the ACS-NSQIP mortality prediction using all variables collected (model 3). Model comparison was done by area under receiver operator curve (AUROC) assessment. Results We identified 50,944 patients who underwent colectomy for colorectal cancer during the study period (FTR 1.41%), 1755 patients experienced an AL (3.46%) with a FTR of 6.44%. Of those who leaked, the mean age was 65.6 (95% CI 65.28 -65.58), median ASA was 3 (IQR 2-3), 51 (2.92%) were partially or totally dependent, 366 (20.86%) were diabetic, 105 (5.98%) had a history of COPD, 32 (1.82%) had a history of CHD and 966 (55.04%) were on hypertensive treatment. The ACS-NSQIP mean mortality prediction was 2.3% (95% CI 2.05-2.55). In the multivariate logistic regression only age and ASA were independent predictors FTR after AL. The performance of model 1 (AUROC 0.76; 95% CI 0.72-0.81), model 2 (AUROC 0.77; 95% CI 0.72-0.81) and model 3 (AUROC 0.79; 95% CI 0.75-0.83) to predict FTR were no different (p=0.37). Conclusions Age and ASA remain the most reliable predictors of failure to rescue an anastomotic leak after colectomy for colorectal cancer. Addition of the modified frailty index, or all variables collected by NSQIP, did not significantly improve predictive performance.

Predictive perfromance of models used to predict failure to rescue after an anastomotic leak.

Estimation of the Proportion of Socioeconomic Status-based Survival Disparities that Could be Eliminated by Equalizing Use of Treatment for Stage I-II Pancreatic Adenocarcinoma D.S. Swords,* C.L. Scaife. Surgery, University of Utah, Salt Lake City, UT.

BACKGROUND Higher socioeconomic status (SES) is associated with both higher rates of treatment and longer survival in stage I-II pancreatic ductal adenocarcinoma (PDAC) in the United States. The extent to which raising treatment rates for lower SES patients to those of highest SES patients would eliminate survival disparities is unknown. METHODS We analyzed the 2007-2015 SEER census tract-level database and the 2007-2014 NCDB. We included patients ages 18-80 years with stage I-II PDAC and excluded patients with treatment contraindications/refusal, missing SES, and death within 2 months of diagnosis. SES was defined using composites measured at small area levels (see Table footnotes ). We used causal mediation analysis techniques to calculate the proportion eliminated (PE), a policy-relevant measure that captures what is predicted to happen to the effect of an exposure (SES) on the outcome (overall survival [OS]) if the intermediate (treatment) were fixed to the same level for all persons. Continuous composite treatment variables were created in each cohort, and models examined the effect of increasing the mean treatment level in lower SES strata to that of the highest SES strata. RESULTS Among 15, 333 patients in SEER and 38, 427 in the NCDB, there were graded associations between increasing neighborhood SES and both increasing use of treatment (Table parts 1a & 2a) and OS (Table  parts 1b & 2b). The proportion of SES-based survival disparities which could be eliminated by increasing use of treatments were estimated to be between 40-50% for all lower SES strata (Table parts 1c & 2c). CONCLUSIONS We estimate that approximately 45% of SES-based survival disparities could be eliminated if treatment rates were increased to those in highest SES areas. This may be an underestimate of the actual PE since SES was not measured at the individual-level and some aspects of treatment quality were unmeasured. These findings simultaneously point to the opportunity to improve outcomes by increasing use of treatment use in lower SES patients and the importance of non-treatment-related social determinants of health. Table   * SES was defined in SEER using quintiles of a time-dependent composite score constructed from seven variables that measure different aspects of the SES of a census tract. § The treatment mediator variable stratified OS fairly well in the SEER cohort with a concordance statistics of 0.66. †These models adjusted for age, sex, race/ethnicity, personal cancer history, and year were adjusted for. Insurance status and marital status were handles as intermediate variables and were not adjusted for. **SES was defined in the NCDB using a previously published 7-level SES composite formed from median income and education levels. This 7-level index was collapsed to 4 categories in this analysis because treatment use and OS were very similar for several neighboring levels. § § The treatment mediator variable stratified OS fairly well in the NCDB cohort with a concordance statistics of 0.65. ‡ High volume facilities were defined as those performing > 16 resections/year for stage I-II PDAC, which corresponded to the top two volume quartiles. † †These models adjusted for age, sex, race/ethnicity, Charlson-Deyo score, and year. Insurance status and evaluation at >1 CoC facility were considered as intermediate variables and were not adjusted for.

Introduction : The recently reported PRODIGE 24 trial demonstrated the benefit of adjuvant mFOLFIRINOX in patients with resected pancreatic ductal adenocarcinoma (PDAC). While FOLFIRINOX results in decreased recurrence and improved survival, the regimen is costly and associated with significant toxicities. We sought to assess the cost-effectiveness of mFOL-FIRINOX compared to gemcitabine (GEM). Methods: We created a costeffectiveness model to simulate the costs and outcomes for a theoretical population of 1,000 patients with resected PDAC initiating pharmacologic therapy and their downstream management followed for 5 years. We obtained all model inputs from the literature. Outcomes included were disease recurrence, death, discontinuation of therapy due to toxicity, cost and quality-adjusted life years (QALYs). The strategy was considered cost effective if the incremental cost-effectiveness ratio (ICER) was less than a willingness-to-pay threshold of $100,000 per QALY, which is an accepted standard for economic analyses in the United States. Results: In our theoretical cohort, for patients treated with adjuvant mFOLFIRINOX compared to GEM, there was an additional cost of $29,284,767, despite 44 fewer recurrences, 127 fewer deaths, and 255 additional QALYs (Table) . We found that mFOLFIRINOX was not cost effective compared to GEM, with an ICER of $117,140 per QALY, which was greater than the willingness-to-pay threshold. In sensitivity analysis, we found that the model was cost effective only if the monthly cost of mFOLFIRINOX was lower than $785 (baseline estimate: $1,382) or the monthly cost for managing adverse effects was lower than $7,131 (baseline estimate: $7,728). Conclusion: We found that mFOLFIRINOX was not cost effective compared with GEM at a willingness-to-pay threshold of $100,000 per QALY. Our projected findings highlight the potential survival and disease control benefit of mFOLFIRINOX in terms of patient outcomes compared to GEM. However, they also underscore the necessity of developing cost-effective toxicity management to control over all treatment related costs.

Outcomes in a theoretical cohort of 1000 patients with resected PDAC managed with one of the following strategies: Gemcitabine or mFOLFIRINOX USD:US dollars, QALY: quality-adjusted life year, ICER: incremental cost-effectiveness ratio * The ICER is calculated by dividing the difference in cost (USD) by the difference in QALYs. In cost effectiveness analyses, a strategy is cost -effective if the ICER is less than the willingness to pay threshold (difference of $ 100,000 per QALY's between therapies).

What is the Cost-effective Treatment of Melanoma Patients with a Positive Sentinel Node? A.R. Hersh,* M.H. Taylor, J. Vetto, A. Caughey, D. Han. Oregon Health & Science University, Portland, OR. Introduction: Melanoma-specific survival for positive sentinel lymph node (SLN) patients does not differ significantly after completion lymph node dissection (CLND) versus observation (OBS), but studies also show a benefit for treating positive SLN patients with CLND and checkpoint inhibitors (CIs) such as pembrolizumab (PEM). We evaluated the cost effectiveness of CLND, OBS, and CLND with PEM (CLND-P) in melanoma patients with SLN metastases. Methods: A Markov model was designed in TreeAge Pro 2019 software using data from the literature to simulate treatment of a theoretical cohort of 1,000 positive SLN patients per therapy with 5 years of follow-up.

An intervention was cost effective if the incremental cost-effectiveness ratio was below the willingness-to-pay threshold of $100,000 per quality-adjusted life year (QALY) between therapies. Results: OBS compared with CLND or CLND-P resulted in 178 fewer cases of lymphedema but in more locoregional and distant recurrences (Table) . Compared with OBS, CLND had higher costs and lower QALYs. Therefore, CLND was dominated and not cost effective. Although CLND-P had the lowest number of recurrences and deaths compared with OBS or CLND, it had the highest cost and QALYs that were lower than OBS, thus also making CLND-P not cost effective compared with OBS. However, in removing the effects of CLND from CLND-P, allowing evaluation of PEM effects alone (P-Alone), the resulting QALYs were the highest, but P-Alone was still not cost effective compared with OBS ($939,663 per QALY). P-Alone would become cost effective by reducing drug cost to below $44,634 (baseline input: $128,597) per patient. Conclusions: OBS was cost effective compared with CLND, CLND-P, and P-Alone for managing positive SLN patients when followed for 5 years. While CLND-P and P-Alone result in less recurrences and deaths, the quality-of-life decrement from CLND and high drug cost make these therapies not cost effective, particularly in strategies using CLND. However, this is the first study to assess the potential use of adjuvant CIs without CLND for positive SLN patients, which is a strategy that may become cost effective if the drug cost can be reduced.

Outcomes in a theoretical cohort of 1,000 melanoma patients with a positive sentinel lymph node managed with one of the following strategies: observation, completion lymph node dissection (CLND), CLND with pembrolizumab (CLND-P) or pembrolizumab alone (P-Alone).

CLND: completion lymph node dissection, USD: US dollars, QALY: quality-adjusted life year, ICER: incremental cost-effectiveness ratio * The ICER is calculated by dividing the difference in cost (USD) by the difference in QALYs. Each strategy was compared with observation. In cost-effectiveness analyses, a strategy is cost effective if the ICER is less than the willingness-to-pay threshold (difference of $100,000 per QALY between therapies). A strategy is dominant if it is lower in cost and higher in effectiveness, making the strategy cost effective in comparing therapies. A strategy is dominated if it is higher in cost and lower in effectiveness, making the strategy not cost effective in comparing therapies.

INTRODUCTION: Within the context of cancer care, the use of patientcentered decision-making (PCDM) by surgeons varies widely and may occur in less than half of clinical encounters, increasing the risk of negative clinical outcomes. We sought to compare surgeons' use of PCDM strategies within the pre-operative cancer treatment conversation. METHODS: Approaches to PCDM were assessed using a cross-sectional survey containing clinical vignettes. Each vignette assessed the surgeon's likeliness (0=not at all likely, 100=very likely) to utilize PCDM strategies such as directiveness (Q1), spending equal time discussing treatment options (Q2), making an explicit recommendation (Q3), perceiving that the patient prefers an active role (Q4), and agreeing with patient decisions (Q5). Analysis included repeated measures mixed-effects linear regression. RESULTS: Among 208 surgeons, the mean age was 51.6 years (SD=9.87); most were male (65.4%), Caucasian (82.0%), and in practice 10 years (66.8%). Specializations included breast (18.9%), HPB (21.4%), and melanoma (14.6%) cancer. Surgeons at non-academic medical centers (versus academic) were more likely to spend time discussing all treatment options (Q2: 82.3 vs. 77.9), make an explicit recommendation (Q3: 71.7 vs. 67.8), and agree with patient decisions (Q5: 73.4 vs. 70.2; all p<0.01). Surgeons working in an urban versus rural setting were more likely to perceive that patients preferred an active role in decision-making (Q4: 66.6 vs. 61.9) and agree with patient decisions (Q5: 73.2 vs. 70.4; both p<0.05). Surgeons that specialized in breast cancer (versus other surgical specialties), in practice 10+ years (versus < 10 years), and female (versus male) were more likely to spend equal time discussing all treatment options (Q2: 82. 8 vs. 77.3, p<.001; 81.6 vs. 78.6, p=0.013; and 82.1 vs. 78.0, p=0.003, respectively) . A summary of PCDM approaches are in Table 1 . CONCLUSIONS: Overall, approaches to PCDM among surgeons varied. Factors such as surgeon sex, years in practice, and practice setting were associated with differences in the likelihood that surgeons would utilize certain strategies during the PCDM process.

Surgeons' patient-centered approaches to treatment decision-making stratified by demographic and occupational factors Background: While better outcomes at high-volume hospitals(HVH) have driven regionalization of complex surgical care, access to high-volume centers often requires travel over long distances. While patients may travel further distances to undergo surgery, some patients may bypass high volume hospitals while driving to their destination hospital. We sought to characterize travel patterns of patients undergoing complex oncologic surgery, as well as identify possible disparities in care relative to travel distance. Methods: The California Office of Statewide Health Planning database was used to identify patients who underwent pancreatectomy(PD), esophagectomy(ES), proctectomy(PT), or lung resection(LR) for cancer between 2014-16. Total minutes(m) traveled, as well as whether a patient bypassed the nearest hospital that performed the operation to get to a higher-volume center was assessed. Data were merged with the American Community Survey for information on income and educational attainment. Results: Overall 25,865 individuals(PD:10.5%; ES:4.9%; PT 29.1%; LR: 55.6%) underwent a complex oncologic operation. Median travel time was 13m (IQR: ; ES 22m (IQR: 10. 6-46.9) ; PT 15m (IQR: 8.1-28.4) ; LR 14m (IQR: 7.8-27.0)]. Nearly three-quarter of patients(PD: 72%; ES 34%; PT: 81%; LR: 73%) underwent an operation at a high-volume hospital. Hispanic patients as well as Medicaid beneficiaries were less likely to undergo surgery at high-volume hospitals; African American patients were more likely to bypass a high-volume hospital on the way to a lower volume destination hospital (Table 1) . Increased zip code median income and education levels were associated with increased travel time. Conclusion: Although nearly three-quarter of patients underwent a complex oncologic operation at a high-volume hospital, disparity-based differences in access to high-volume care were noted. Specifically, African American, Hispanic, and Medicaid beneficiaries were more likely to receive care at lower volume hospital. In addition, patients living in zip codes with lower income and education levels were less likely to travel to reach high volume centers. MD Anderson Cancer Center, Houston, TX. Background: The gap between initial education and retention is a common barrier in quality improvement implementation. In a previous study from a departmental opioid reduction education program, there was an immediate 1-month change in perceptions of opioid needs and recommendations in opioid prescriptions for surgical oncology patients. This study's aim was to re-evaluate if those early trends were retained 1 year later. Methods: Surgical Oncology attendings, fellows, and advanced practice providers (APPs) at a high-volume Comprehensive Cancer Center were surveyed 1 year after a Aug 2018 opioid reduction education program, to compare departmental and individual opioid prescribing habits for five index operations (open abdominal resection, laparoscopic colectomy, wide local excision, thyroidectomy, port). The 1-year follow-up questions matched the pre-education (Aug 2018) and post-education (Sep 2018) surveys. Responses were compared as an entire department for Aug 2018 vs. Sep 2019, and by individuals for those who completed both the Sep 2018 and Sep 2019 surveys. Results: The Sep 2019 response rate was 54/93 (58%): 13 attendings, 14 fellows, and 21 APPs, with 41 completing both the post-education and 1-year follow-up surveys. Both the overall departmental cohort and the matched cohort continued to recommend a lower quantity of discharge opioids for all 5 index operations (by at least 50%) and less days to zero opioid needs (by at least 2 days), when compared to pre-education perceptions (Table) . Providers continued to agree that discharge opioid prescriptions for inpatient operations should be based on a 5x multiplier of a patient's last 24 hours of inpatient opioid use rather a fixed procedurespecific quantity or a provider's "usual" amount. There was universal agreement that each respondent's actual opioid administration has decreased in the past year. Conclusions: In dramatic contrast to pre-education survey results and similar to initial changes measured 1 year ago, our Surgical Oncology providers continue to recommend less discharge opioids, faster weaning to zero opioids, and patient-specific discharge opioid volume calculations. INTRODUCTION: National Institutes of Health (NIH) is the primary public funding source for surgical research in the United States. Surgical oncology is a highly academic career yet its role in NIH funding is poorly understood. We sought to analyze the participation of surgical oncologists (SOs) as primary investigators (PI) in NIH-funded projects, assess for potential gender disparities, and correlate with scholarly productivity and academic rank. METHODS: The NIH RePORTER (Research Portfolio Online Reporting Tools Expenditures and Results) was queried to identify R01-and-equivalents grants (2008) (2009) (2010) (2011) (2012) (2013) (2014) (2015) (2016) (2017) (2018) awarded to Departments of Surgery(DOS). PI's data were extracted from publicly available information from their institutions' webpage or social media. RESULTS: Of 1,101 DOS awarded grants, 510(46.3%) were led by surgeons. Among these, general surgeons(31%) accounted for most grants, followed by SOs(20.8%), cardiothoracic(8.4%), vascular(6.3%) and transplant surgeons(5.1%)(P<0.001). Women represented only 211(24.1%) of the projects, although female participation was higher among SOs compared to other surgeons(30.0% vs. 16.1%; P=0.001). SOs who obtained grants also required less years of experience (YoE) (12 [IQR 8.75] vs. 13[IQR 13] ; P=0.003), had fewer senior (>24YoE) investigators(4.0% vs. 18.9%; P<0.001), and fewer PhD-degree holders(30.8% vs. 65.5%; P<0.001) compared to the overall cohort. SOs-led projects accounted for 1,121(14.1%) publications, with a higher proportion of high impact(IF>10) articles(26.3% vs. 9.7%; P<0.001), and were more likely to become a registered patent on univariate(6.6% vs 2.8%; P=0.025) and multivariate analysis (OR 3.30; ;P=0.016) analysis. CONCLUSION: SSO-accredited surgeons accounted for a large share of the NIH grants to surgical investigators, were younger in their career, did not require a PhD degree and had a higher impact scholarly productivity. Fewer proportion of women received NIH grants, however a greater proportion of female SO obtained funding compared to other specialties. Fellowship programs should continue to stimulate groundbreaking research by integrating grant-writing training and special mentorship programs during fellowship.

Introduction: Despite increasing numbers of women in academic surgery, female underrepresentation in surgical societies remains an ever-present issue. Participation in society meetings serves as a proxy for academic success and is important for a surgeon's career development. We aimed to investigate and report the gender breakdown of presenters at recent SSO meetings. Methods: Genders of all presenters for poster, parallel, plenary, and video sessions at SSO meetings from 2014 through 2019 were collected. Data for first and last authorship were collected for each session type. These data were further broken down to first/last authorship relationships including female/female, female/male, male/female, and male/male. The proportions of female to male presenters were compared for each session type. A sub-analysis was performed to look at non-breast surgery specialties. Statistical significance was set at p<0.05. Results: From 2014 through 2019, there were 3110 presenters and 46.7% were female. Approximately 26% of SSO members and 37% of recently matched fellows are female. Women were listed as first authors more often for the poster session (47.9%) compared to all other sessions but this was not significant (p=0.07) (Figure 1a ). Women were also listed more often as senior authors for the poster session (29.3%) and this was statistically significant (p=0.04). Of note, there were significantly fewer female senior authors compared to male senior authors across all session types ( Figure 1b) . Subanalysis of first and senior authors for non-breast surgery sessions showed no significant differences (Figure 1c , 1d). The most common combination of first and senior authors was male-male (41.4%), followed by female-male (29.4%), female-female (18.8%), and male-female (10.4%). Conclusion: Female representation at SSO is comparable to society demographics. Female first authorship is relatively equal to male first authorship in poster sessions. Opportunities to improve gender equality in senior authorship positions, especially as mentors to trainees and junior faculty, should be explored. Medical Center, Dallas, TX; 5. Johns Hopkins University School of Medicine, Baltimore, MD; 6. H. Lee Moffit Cancer Center & Research Institute, Tampa, FL; 7. Memorial Sloan Kettering Cancer Center, New York, NY. Introduction: The learning curve for robotic HPB surgery is long and may limit adoption. We previously validated a robotic curriculum using virtual reality (VR) and biotissue drills as an effective tool to obtain proficiency during fellowship at University of Pittsburgh (UPMC). Here we evaluate the feasibility and efficacy of implementation at other Society of Surgical Oncology (SSO) fellowships. Methods: 4 SSO programs were targeted based on robotic case volumes. UPMC faculty and staff performed a webinar, site visit, pre-and post-test for each institution. All sites had simulators, and kits were provided to perform 19 suture/biotissue drills. Prior robotic experience, curriculum compliance, and performance were ranked low, moderate, or high. The external SSO fellows were compared to UPMC fellows. Results: Three sites accepted curriculum implementation. All fellows completed the pre-and post-test. There was no difference in pre-test scores between UPMC and SSO. Only 7 of 15 SSO fellows completed the VR curriculum (47% compliance). UPMC had higher curriculum times (217 vs 93 mins, p=0.007) and % mastery (86 vs. 55%, p=0.14). Time spent on curriculum was associated with % mastery (p=0.01). Both groups showed improvement between pre-and post-test. Post-test VR scores trended higher for UPMC (221 vs 180, p=0.06). Among SSO fellows, prior experience correlated with performance (p<0.001), but not compliance or curriculum time. Between the SSO sites, there was no difference in prior experience (p=0.33), performance (p=0.441), or curriculum time (p=0.09), but there was a difference in compliance (p=0.03) and % mastery (p=0.03). During that 2-year period, 0/15 SSO fellows performed the biotissue drills, while 5 contemporary UPMC fellows did 253 biotissue drills. Approximately 140 UPMC faculty and 300 staff hours were spent on the pilot. Conclusions: A proficiency curriculum results in mastery of robotic skills. However multiple barriers to implementation exist: industry representatives to acquire scores, availability of a training robot, protected time for fellows, and buy-in from leadership to support fellows and add value for faculty to train in a modern era of technology.

Site Infection Following Inguinal Lymph Node Dissection G. Frenda, 1 * L. Baker, 2 Y. Zhang, 2 C. Nessim. 2 1. General Surgery, McGill University, QC, Canada; 2. University of Ottawa, Ottawa, ON, Canada. Surgical site infection (SSI) following ILND is common. Post-operative SSI is associated with decreased quality of life, significant health care resource utilization and delays to initiation of adjuvant oncology treatment. Modification to surgical technique to a two-incision technique with skin bridge (SB) from the traditional lazy-S (LS) incision, as well as utilization of prophylactic incisional negative pressure wound therapy (iNPWT) are theorized to reduce the risk of SSI following Inguinal lymph node dissection (ILND). A retrospective review of patients undergoing ILND pre and post-implementation of ILND bundle aimed at reducing the incidence of SSI between October 2013-January 2019 was performed. The bundle was implemented in September 2016 and involved a SB incision, running subcuticular skin closure, and prophylactic placement of a iNPWT. Prior to this, a LS incision was used, with stapled skin closure, without iNPWT. Postoperative adverse events were compared. Thirty-four ILND were preformed in thirty-three patients over the study period (n=15 ILND pre-implementation and n=19 ILND post-implementation of bundle). The cohorts were similar across baseline demographics. Implementation of a bundle was associated with reduction in SSIs (31.6% vs 73.3%, p=0.036) and elimination of wound dehiscence (0 vs 33.3%, p=0.0108). Multivariable logistic regression demonstrated the bundle was associated with a 5.8 fold reduction in SSI (OR: 0.173; 95% CI: 0.038-0.796, p=0.0243). Trends towards reduction in post-operative antibiotic prescription (43% vs 65%, p=0.223) following implementation of the bundle were observed. Preliminary data demonstrates a reduction in SSIs following implementation of an ILND bundle. The use of a SB incision in combination with iNPWT demonstrates potential reduction in post-operative morbidity. Further research is needed to determine if the potential improved outcomes are secondary to the incision, the iNPWT or both therapies used in combination. Introduction: Opioids are often over-prescribed after cancer surgery, leading to persistent use and potential public dissemination. The objective of this study was to assess the current practices and perceptions surrounding opioid use by surgical oncologists to inform quality improvement initiatives targeting perioperative opioid reduction. Methods: After approval from the SSO Research Committee, an online survey was distributed to its membership. Five sample oncologic procedures were used to assess prescribing habits. Participants were asked to consider facilitators to reducing opioid prescriptions. Data were summarized and compared by self-reported demographics and geography. Results: Of 175 total participants, 149 (85%) identified as attending/faculty, 24 (14%) as trainee, and 2 (1%) as advanced practice provider. Most participants (76%) practiced in academic/research programs; 21% practiced in non-U.S. locations. Few differences in prescribing habits were found based on clinical role, academic rank, or years in practice. Within the U.S., providers in the Northeast recommended lower initial opioid prescriptions compared to other regions for 4/5 operations. Compared to non-U.S. providers, those practicing in the U.S. reported higher perceived patient-reported pain scores at discharge, recommended higher initial opioid prescriptions, and estimated longer postoperative opioid use for almost every procedure (Table 1) . Compared to non-U.S. providers, fewer U.S. providers agreed that discharge opioids should not be given to patients already opioid-free in their last 24 inpatient hours (80% vs. 50%, p=0.001). Education was rated "very important" for lowering opioid prescriptions by the vast majority of respondents (90% for "education for patients" and 86% for "education for providers"). Conclusions: Compared to their international counterparts, U.S. surgical oncology providers perceived greater needs for perioperative opioids for their patients. Educating U.S. surgical oncology providers on multimodal bundles, accelerated opioid weaning protocols, and standardized discharge prescribing protocols could impact our national opioid epidemic. Introduction: A rising metastasectomy rate, together with improved survival outcomes, may be associated with an increasing prevalence of ventral hernias in patients with intra-abdominal metastases (IAM). While these hernias are frequently not addressed given other oncologic considerations, they can affect quality of life and lead to sequelae necessitating an emergent operation. We sought to determine the national trend in elective (E) and nonelective (NE) ventral hernia repairs (VHR) in patients with IAM. Methods: Patients with IAM who underwent VHR-E/VHR-NE were identified using the National Inpatient Sample (2003-15) . Average annual percent change (AAPC) was estimated for the entire study period, and annual percent change (APC) for a portion of the period when a significant (P<0.05) breakpoint in trend was present. Perioperative outcomes were determined using multivariable regressions. Results: An estimated 947,112 VHR were performed nationally between 2003 and 2015, of which 5602 (0.6%) occurred in patients with IAM. Among those with IAM, 40.1% had a VHR-NE, mean (standard deviation) age was 64 (12) years, and 65.1% were women. During the study period, the total number of VHR performed nationally did not change (AAPC 0.062, P=0.84 ). VHR-NE was also associated with a 41% (95% CI 29-54%) longer hospital stay and 25% (95% CI 14-37%) higher charges. Conclusion: VHR-NE procedures in patients with IAM has increased significantly and is associated with worse outcomes and higher cost. Focus should be placed on abdominal closure techniques that mitigate hernia development and, in the appropriate clinical setting, such as a favorable cancer prognosis, consideration of an earlier non-emergent intervention. Introduction: Neoadjuvant chemotherapy (NAC) increases breast conserving surgery (BCS) rates with comparable locoregional control and survival outcomes to adjuvant therapy. Studies suggest increased margin involvement after NAC, though a "no-ink-on-tumor" margin may be acceptable. This study evaluated the effect of NAC on margins in an era of lower re-excision rates after the adoption of "no-ink-on-tumor" in primary BCS (PBSC). Methods: An institutional review board approved database was queried for women undergoing BCS for invasive breast cancer after March 2014. Re-excision after NAC was at surgeon's discretion. Patients were dichotomized as NAC followed by BCS vs. PBCS. Groups were compared along demographic, tumor, treatment, and outcome variables using univariate and multivariable analysis; p<0.05 was considered significant. Results: Of 168 patients, 57 received NAC while 111 had PBCS. NAC patients were younger, more likely to be hormone receptor negative, HER2 positive, and higher stage at diagnosis. 39% had a complete pathologic response after NAC. NAC patients had higher clinical T stage at diagnosis (T2 68% NAC vs. 13% PBCS, p<0.001), but smaller tumors on final pathology (0.7 cm NAC vs. 1.1 cm PBCS, p<0.001). There was no difference between the two groups with respect to complications requiring intervention (1.8% NAC vs. 0.9% PBCS, p=1.00), margin positivity (2% NAC vs. 5% PBCS, p=0.42), re-excision rate (2% NAC vs. 9% PBCS, p=0.07), or conversion to mastectomy (0% NAC vs. 5% PBCS, p=0.50). However, NAC patients had smaller resection volumes at lumpectomy (median 41.8 cm 3 NAC vs. 47.1 cm 3 PSBC, p=0.045). On multivariable analysis NAC was not an independent predictor of lumpectomy volume < 47.1 cm 3 , rate of margin positivity, re-excision, conversion to mastectomy, or complications requiring intervention when controlling for age and T stage at diagnosis. Conclusion: BCS after NAC is not associated with differences in surgical outcomes, when compared to a contemporary PBCS cohort in the "no-ink-on-tumor" era. With NAC in selected patients, surgeons can expect similar re-operation, complications requiring intervention, and lower volumes of excision. BACKGROUND: High neutrophil-to-lymphocyte ratio (NLR) has been reported to be associated to poor prognosis in several tumors. Data on NLR in soft tissue sarcomas are limited to small series, with lack of information for retroperitoneal sarcoma (RPS). OBJECTIVE: To investigate the prognostic role of NLR in a mono-institutional series of RPS. METHODS: Consecutive primary RPS operated at a single referral center between 2002 and 2016 were extracted from a prospective database. Baseline NLR (i.e. before treatment) was retrospectively computed. Correlation between NRL and clinicopathologic variables was studied by Kruskal-Wallis test. Overall survival (OS) was analyzed by Kaplan-Meier. The prognostic effect of NLR was adjusted for age, tumor FNLCC Grade and histology subtype in a multivariate Cox regression model. RESULTS: In the study period 463 patients underwent surgery for primary RPS at our Institution. Median follow up was 57 months; 147 (31.7%) patients received preoperative chemotherapy and/or radiation therapy; baseline NLR before treatment was available in 422/463 (91.1%) patients. Median baseline NRL was 3.3 (IQR, and did not differ according to age, sex, tumor size, while it differed across histologic subtypes [median 3.2 (IQR, in liposarcomas, 3.1 (IQR, in leiomyosarcomas, 3.4 (IQR, in solitary fibrous tumors, and 4.0 (IQR, in other RPS; p<.001] and increased with malignancy grade [median 2.7 (IQR, in G1 RPS, 3.5 (IQR, 4.0 (IQR, in G3 RPS; p<.001]. 5 year-OS was 81.9% (SE 5.8) for NLR 2; 71.4% (SE 3.8) for 2<NLR 4 and 55.6% (SE 4.9) for NLR>4 (p<.001, Figure 1 ). At multivariate analysis, NLR was independently associated to OS with an hazard ratio of 1.05 (95%CI, 1.01-1.04, p=.01), when adjusted for existing prognostic factors, meaning a 5% increase of the risk of death for each 1 unit increase of NLR. CONCLUSIONS: Baseline NLR is a readily available biomarker, which seems to be independently associated to survival also in patients affected by primary RPS. Further investigations on the prognostic and predictive role of NLR are warranted as well as its association with other inflammatory markers. ON, Canada; 2. Brigham and Women's Hospital, Boston, MA; 3. Istituto Nazionale Tumori Milano, Milan, Italy; 4. Royal Marsden Hospital, London, United Kingdom; 5. Gustave Roussy, Villejuif, France; 6. Institute Bergonié, Bordeaux, France; 7. Institut Curie, Paris, France; 8. The Netherlands Cancer Institute, Amsterdam, Netherlands; Warsaw, Poland. Background Recent series show an actuarial 10-yr overall survival (OS) of 46% in patients with primary RPS who undergo resection in reference centers. Here, we aim to characterize actual 10-yr survivors in order to investigate potential contraindications to resection of primary RPS, and appropriate follow-up (FU) strategies for long-term survivors. Methods Consecutive patients with primary RPS who underwent curative intent resection between 2002 and 2017 across 10 sarcoma centers were included. Patients were stratified into 4 cohorts, according to their actual survival (<2 yrs, 2-5 yrs, 5-10 yrs and >10 yrs). Actual survival <2yrs indicates death or last FU within 2 yrs of resection. High risk features were derived from a nomogram for OS endorsed by the AJCC. Results For the entire study group of 1944 patients (median FU 61mos, IQR 34-106), the actual survival was 0-2 yrs in n=544, 2-5 yrs in n=722, 5-10 yrs in n=513, and over 10 yrs in n=165. There was a decline in high risk features (tumor multifocality, high grade, high-risk histology, intraoperative tumor rupture, incomplete resection) and corresponding enrichment for low risk features, across the 4 actual survival time cohorts (Table) . Within the actual 10-yr survivors cohort, the primary RPS was multifocal in 4.5%, grade 3 in 12.1%, ruptured intraoperatively in 5.5%, and incompletely resected (R2) in 3.0%. For the actual 10-yr survival group of 165 patients, median FU after the 10-yr mark was 19 mos (IQR 7-39). At last FU, 119 were NED, 28 AWD, 15 DOD and 3 DOC. Following the 10-yr mark, 9 patients developed local recurrence (8 LPS, 1 MPNST) and 2 developed distant metastasis (1 LMS, 1 SFT) as first events, at a median of 139 and 133 mos, respectively, following primary resection. Conclusions Amongst patients who undergo resection of primary RPS despite high risk features, a significant proportion become actual 10-yr survivors. Thus an attempt at initial resection may be appropriate despite high grade and multifocality. Actual 10-yr survivors remain at risk of recurrence, particularly at a local intra-abdominal site, and FU should be continued beyond 10 yrs.

Little is known about grade changes between a primary RPS and its subsequent local recurrence (LR). The French grading system was developed for primary sarcoma and prognosis is unknown if grade worsens at the time of LR. For primary RPS, higher grade is associated with worse metastatic risk and overall survival (OS). We evaluated the frequency of patients whose grade changed between primary RPS and susequent first LR, and whether this change affected outcome. METHODS: Data were collected from 22 TARPSWG centers from 2002-2011. All patients undergoing resection for a first LR were selected. Main outcome was the prognostic effect of grade change between the primary and recurrent RPS. Patients were categorized in three subgroups: improving, stable and worsening grade at the time of first LR. Overall and disease-free survival (DFS) were performed using the Kaplan-Meier method. Crude cumulative incidence (CCI) of second local recurrence (LR2) and distant metastases (DM) were calculated in a competing-risk framework. RESULTS: Of 681 patients, 481 had information on both primary and recurrence grades. 57 patients had grade improvement, 88 had worsening grade and 336 remianed stable (127 grade I, 69 grade II, 140 grade III). With median follow-up >60 months for all 3 subgroups, LR2 and DFS but not DM were significantly worse for worsening grade patients compared to the other subgroups (p=0.04). There was a trend towards a worse OS in the same subgroup, although not statistically significant (p=0.17). The 5-yr OS for the improving, stable and worsening grade patients were 57%, 61% and 49% respectively. DISCUSSION: In our series of RPS sarcoma, the outcome to LR surgery was dominated by high risk of further loco-regional recurrences, rather than DM. Among the 481 patients, approximately 20% experienced a worsening grade. However, while this is associated with worse DFS and further LR risk, the DM risk remained low and not different from that of patients with improved or stable grade. A worsening grade seems to be only associated with a more aggressive local course.

INTRODUCTION: Textbook outcome (TO) has recently been introduced as a single composite measure intended to evaluate the quality of surgical care in complex operative procedures. This metric has not been previously characterized among patients undergoing surgical resection for retroperitoneal sarcoma (RPS). METHODS: All patients who underwent resection for RPS between 2000-2016 from 8 academic institutions were included. Using expert consensus, TO was defined as a patient with R0/R1 resection that was discharged to home and was without blood transfusion, reoperation, grade 2 complications, hospital stay >50th percentile, or 90-day readmission or mortality. Univariate and multivariable analyses were used to determine factors associated with TO. RESULTS: Among 777 patients who underwent RPS, 54.4% were female and the median age was 59 years. The most common histologies include liposarcoma (43.9%), leiomyosarcoma (23.2%), and undifferentiated pleomorphic sarcoma (7.3%). The median tumor size was 11.5 cm. A majority of RPS resections were performed for primary tumors (61.9%) vs. recurrent disease (38.1%). A minority of patients achieved a TO (35.8%) ( Figure) . On multivariable analysis, factors associated with not achieving a TO were tumor size >20cm vs. <10cm (Odds Ratio (OR): 0.3), left colon/ rectal resection (OR 0.3), adrenalectomy (OR: 0.1), major venous resection (OR: 0.1), and psoas fascia/muscle resection (OR: 0.2) (all p<0.05). TO was associated with increased overall survival (median: 153.1 months vs. 69.1 months, p<0.01). This effect persisted when splitting the cohort between primary tumor resection (median 151.9 months vs. 72.7 months, p<0.01) and resection for recurrent disease (median: not reached vs. 38.7 months, p<0.01). CONCLUSIONS: Among patients undergoing surgical resection for RPS, failure to achieve a TO is common and associated with significantly worse overall survival. The use of composite metrics such as TO may improve patient counseling by informing expectations and can serve as a simple measure for patient-level hospital performance and assessing quality improvement efforts.

Severe Postoperative Adverse Events Affect Survival in Primary and Recurrent Retroperitoneal Sarcoma E. Nizri,* Y. Netanyahu, S. Shamai, O. Sher, O. Merimsky, G. Lahat, J. Klausner. Surgery, Tel-Aviv Sourasky Medical Center, Tel-Aviv, Israel. Background: Retroperitoneal Sarcoma (RPS) surgery entails multivisceral resction and may be associated with severe postoperative adverse events (SAE). Our aim was to assess the effects of SAE on survival and identify their predictors in primary and recurrent RPS. Methods: We retrospectively analyzed our institutional database. Patient comorbidities and SAE were calssified according to the Charlson comorbidity index (CCI).and Clavien-Dindo classification ( 3) . Results: From 2008 to 2018 we operated on 171 patients with . Patients in the recurrent RPS group had higher rate of neoadjuvant chemotherapy and multifocal tumors (39.2% vs. 12.9%, p<0.001 and 48.9% vs. 18.3%, p<0.001), while high grade (3) tumors were less abundant (58.1% vs. 36.8%, p=0.05). Surgical extent was more limited in the recurrent group as judged by the median number of organs resected (1 vs. 2, p=0 .07), weighted organ score (1 vs. 2, p=0 .01) and transfusion requirements (20.2% vs. 34.1%, p=0.04). Overall mortality and SAE rates were 5.3% and 36.8%, respectively, and did not differ between the primary and recurrent groups. Rates of reoperation, anastomotic leak, respiratory failure and thromboembolic events were comparable. Preoperative serum albumin and CRP did not differ between patient with and without SAE (40 vs. 39, p=0.69; 25 vs. 24, p=0.92 required. The aim of our study was to establish whether diagnostic sensitivity can be improved by targeting solid areas of tumor on percutaneous biopsy (PB). Methods: Between July 2016 and 2019, data on patients with suspected primary retroperitoneal sarcoma who underwent a PB were extracted from a prospectively maintained database at two major institutions. Sensitivity and specificity were calculated to quantify the diagnostic accuracy of PB vs. final pathology, with respect to tumor histology and grading. These data were then compared to our previously reported biopsy series from 2005 -June 2016. For DDLPS tumors, comparisons were then made between PBs that had targeted the solid component and those that did not, as well as between the types of image guidance (US vs. CT). Results: Data were available for 123 patients in the current series, and 238 from the previous study. The proportion of PBs returning a histological subtype that was concordant with post-operative pathology was 83% in the current series, which was a significant improvement over our previous study (67%, p=0.001). For the diagnosis of DDLPS, the sensitivity of PB improved from 40% to 74% (p<0.001), with an increase from 13% to 50% (p=0.006) where G3 DDLPS was treated as a separate disease ( Table 1 ). Within the new study, targeted biopsy yielded a sensitivity of 100% for identifying DDLPS, compared to only 10% in non-targeted biopsy (p<0.001). Both CT-and US-guided targeted biopsy had 100% specificity for differentiating G3 from G2 DDLPS, with no significant difference in sensitivity (67% vs. 50%, p=0.650). Conclusion: Systematic targeting of solid areas of tumor within suspected retroperitoneal liposarcoma has improved sensitivity for detection both of DDLPS and G3 tumors on PB. This approach minimises the risk of under-diagnosis of patients with DDLPS who may potentially benefit from neoadjuvant chemotherapy. 

Loss of 13q and Reduced Expression of MYCBP2 and IRS2 is Associated with PPAR Dysregulation and Liposarcoma Dedifferentiation A. Lofthus, 1 * J. Hu, 2 C. Antonescu, 2 N. Socci, 2 S. Singer, 2 A. Crago. 2 1. Medstar Georgetown University Hospital, Washington, DC; 2. Memorial Sloan Kettering Cancer Center, New York, NY. 12q13-15 is amplified and encodes oncogenic drivers in well-differentiated and de-differentiated liposarcoma (WD/DDLS). Other copy number alterations have been identified in subsets of tumors; but their role in liposarcomagenesis and progression is poorly understood. Analysis of matched WD and DDLS samples from the same patient showed 13q loss is associated with tumor progression. This study aimed to identify tumor suppressors encoded by 13q. WDLS (n=49) and DDLS (n=56) tumors and cell lines were analyzed by array comparative genomic hybridization (aCGH) and U133A gene expression arrays. Adipogenesis was simulated in WD cell lines by supplementing media with biotin, insulin, and a PPAR agonist. Proliferation, protein expression and adipogenic differentiation were analyzed by CyQuant, immunoblot, and Oil Red O staining, respectively. Loss of 13q was identified in 20% of DDLS but only 3% of WDLS. Integrated analysis showed dedifferentiation was specifically associated with reduced expression of 13q genes IRS2 (FC= -2.04, FDR <0.001) and MYCBP2 (FC= -4.11, FDR <0.001). Both have potential to regulate the insulin signaling pathway, essential for adipogenesis. Two cell lines without 13q loss were identified (WD4847 and WD7525). While knockdown of neither IRS2 nor MYCBP2 increased cellular proliferation, both decreased protein levels of PPAR compared to a control in cell lines grown in adipogenic media. Cells treated with both shRNAs sequentially exhibited further PPAR additive protein reduction. Consistent with PPARy's known role in adipogenic differentiation, IRS2 and MYCBP2 shRNA also decreased Oil Red O staining (21% and 14% respectively, vs. 53% in scramble control, p<0.05). This finding was again amplified in cells treated with both shRNAs (13% vs. 52% in a scramble control, p<0.05). Inhibition of the 13q-encoded IRS2 and MYCBP2 genes downregulates PPAR and leads to a blunted response of WDLS cells to pro-adipogenic differentiation factors. These putative tumor suppressor genes both have potential to affect differentiation suggesting modulation of insulin signaling may be exploited for therapeutic benefit. Grünhagen, 5 C. Verhoef. 5 1. University Medical Center Utrecht, Utrecht, Netherlands; 2. Netherlands Cancer Institute, Amsterdam, Netherlands; 3. Netherlands Comprehensive Cancer Organization (IKNL), Utrecht, Netherlands; 4. Leiden University Medical Center, Leiden, Netherlands; 5. Erasmus Medical Center, Rotterdam, Netherlands. Background: Despite curative intents of treatment in localized malignant peripheral nerve sheath tumors (MPNST), prognosis remains poor. This study investigated survival and prognostic factors for overall survival in nonretroperitoneal and retroperitoneal MPNSTs in the Netherlands. Methods: Data were obtained from the Netherlands Cancer Registry and the Dutch Pathology Database. All primary MPNSTs were collected. Pediatric cases (age 18 years) and synchronous metastases were excluded from analyses. Separate Cox proportional hazard models were made for retroperitoneal and nonretroperitoneal MPNSTs. Results: A total of 629 localized adult MPNSTs (35 retroperitoneal cases, 5.5%) were included for analysis. In surgically resected patients (88.1%), radiotherapy and chemotherapy were administered in 44.2% and 6.7% respectively. In retroperitoneal cases significantly less radiotherapy and more chemotherapy were applied. In non-retroperitoneal MPNST, older age (60+), presence of NF1, size >5cm, and deep-seated tumors were independently associated with worse survival (Figure 1 ). In retroperitoneal MPNST male sex and age 60+ years old were independently associated with worse survival. Survival of R1 and R0 resections were similar for any location, while R2 resections were associated with worse outcome. Radiotherapy and chemotherapy administration were not associated with survival. Conclusion: In localized MPNST, risk stratification for survival can be done using several patient-and tumor specific characteristics. Resectability is the most important predictor for survival in MPNST. No difference is present between R1 and R0 resections in both retroperitoneal and non-retroperitoneal MPNSTs. The added value of radiotherapy and chemotherapy is unclear. Background: Lymph node (LN) metastasis is uncommon among gastrointestinal stromal tumors (GISTs) and it is not utilized in prognostication for this disease. The purpose of this study was to study the factors associated with GIST LN metastases and its impact on survival within a large national dataset. Methods: Patients undergoing surgical resection of GIST with nodal evaluation performed were selected from the National Cancer Data Base. Logistic regression was utilized to evaluate factors associated with LN metastases. Survival was assessed for patients with and without nodal involvement and Cox regression was used to evaluate the impact of LN metastases while adjusting for other prognostic factors. Results: Out of 4904 patients selected, 353 (7.2%) were noted to have LN involvement. LN metastases occurred most frequently among tumors originating in the small bowel (8.3%), followed by the stomach (6.6%), and the colorectum (6.2%). LN involvement occurred more frequently in the setting of concomitant liver metastases (10.8% vs 2.4%, p < 0.001). On multivariable analysis, male sex (OR 1.88, p < 0.001), high mitotic rate (OR 1.97, p < 0.001), and tumor size (OR 1.03, p < 0.001) were all significantly associated with LN metastases. LN metastases were associated with decreased OS for tumors arising in the stomach (114.0 vs 136.0 months, p = 0.0320), the small bowel, 59.7 vs 115.5 months, p < 0.001), and the colorectum (18.0 vs 122.4 months, p < 0.001). Within a multivariable Cox regression model adjusting for multiple prognostic factors including age, tumor size, mitotic rate, and primary site, LN metastatic disease remained independently associated with an increased risk of death (HR 1.91, p = 0.003). Conclusions: For GISTs, LN metastasis occurs more often in males and is associated with greater tumor size and mitotic activity. LN involvement is associated with worse outcomes, independent of other more well-established prognostic factors. Introduction: Sarcoma research is limited by heterogeneity of tumor biology, lack of comprehensive preclinical models and rarity of disease. We have hypothesized that engineering hydrogel based sarcoma organoids directly from the patient and without xenogeneic growth factors, is routinely feasible on a clinical setting and allows rare tumors, from individual patients to remain viable for personalized research. Methods: Surgically obtained sarcoma specimens were dissociated, and incorporated into an ECM-based hydrogel system and biofabricated into 3D patient-specific sarcoma organoids. Cells were not sorted for tumor, as to preserve tumor heterogeneity. Organoid sets were screened in parallel with chemotherapy and immunotherapy agents for 72 hours. Enrichment with immune system was performed with patient specific peripheral mononuclear cells. Quantification of live/dead staining and metabolism assays, recorded relative drug efficacy while light microscopy and immunohistochemistry were used for tissue verification. Results: Biospecimens from seven patients, were collected from April 2018 to May 2019. Histologies included, cutaneous (n=2) and metastatic to the liver angiosarcomas, leiomyosarcoma, myxofibrosarcoma, dermatofibrosarcoma protuberans with sarcomatoid changes, and pleiomorphic pelvis sarcoma. Successful establishment rate of viable organoid sets was 100% (7/7), although organoid number yield varied with biospecimen size. Average time from organoid development to initiation of drug testing was 5 days. Conclusion: Routine development of sarcoma hydrogel based organoids directly from the operating room, is a feasible platform, allowing for such rare tumors to remain viable for drug sensitivity analysis, drug development and personalized translational research.

Patient specific testing of metastatic angiosarcoma to the liver response to immunotherapy. LIVE/DEAD imaging of angiosarcoma organoids without (top) and with (bottom) peripheral mononuclear cells from the same patient. Insignificant viability differences were observed without the presence of immune system (top panel). Check point inhibitors nivolumab and ipilimumab were effective only when a supplemental immune cell population was incorporated (bottom panel). Green marks viable cells, red marks dead cells Background:Surgical resection is a widely accepted therapy for sarcoma lung metastases and has been associated with improved overall survival(OS). Our aims were to identify preoperative factors associated with OS and to develop a risk score to stratify patients being considered for lung metastasectomy. Methods:Pts who underwent curative-intent resection of sarcoma lung metastases were identified from the US Sarcoma Collaborative. Pts with extrapulmonary metastatic disease or R2 resection of either primary tumor or metastases were excluded. Primary endpoint was OS. Results:352 pts met inclusion criteria. Mean age was 53yrs and 57% were male. Location of primary tumor was truncal/extremity in 85%(n=270) and retroperitoneal in 15%(n=49). The most common histiotypes were undifferentiated pleomorphic sarcoma(26%), leiomyosarcoma(18%), and synovial sarcoma(15%). 49%(n=171) of pts had solitary and 51%(n=180) had multiple lung metastasis. Medan number of lung nodules was 2. 28% received chemotherapy. Median OS was 49 months, 5-year OS was 42%. On multivariable analysis, age 55(HR 1.77), retroperitoneal tumor location(HR 1.67), R1 primary tumor resection(HR 1.72), and multiple lung nodules(HR 1.77) were independently associated with decreased OS(all p<0.05). Assigning one point for each factor, we developed a risk score from 0-4. Pts with 0 and 1 risk factor had similar 5-year OS(61vs48% p=0.335) and pts with 2, 3, and 4 risk factors had similar 5-year OS(17vs10 vs0%,p=0.245 and p=0.645, respectively). Thus, pts were divided into two risk groups: low(0-1 factor) and high (2-4 factors) . When stratifying pts by risk, the low-risk group (n=159) had significantly better 5-year OS compared to pts in the high-risk group(n=108)(51vs16%, p<0.001) (Fig. 1) . Conclusion:Using our multi-institutional cohort, we created a novel risk score for pts with sarcoma lung metastases and identified four characteristics that portend a worse OS. Given that pts in the high-risk group have a projected OS of <20% at 5 years, this risk score is an important tool to aid in preoperative counseling and patient selection for pulmonary metastasectomy. Introduction: Locally advanced cutaneous squamous cell carcinoma (cSCC) involving the scalp, cranium and brain are not well described. We evaluated patients with T3 and T4 cSCC treated with extensive resection to evaluate recurrences and outcomes. Methods: We reviewed AJCC 8 stage 3/4 cSCC patients treated from January 2010 -March 2019. Extent of excision was guided by pre-operative imaging and intraoperative evaluation of tumor invasion. Stage 3 and 4 patients were compared using Kruskal Wallis, Chi-square and Fisher's exact tests. Overall (OS) and disease-free survival (DFS) were analyzed using Kaplan Meier curves. Results: 76 patients were identified. The median age at diagnosis was 75 years (IQR 66-82); 68 (89%) were male (Table) . 42 (52%) were treated for primary cSCC of the scalp as opposed to recurrent disease. 93% of patients had cardiac comorbidities; 21 (27%) patients were immunosuppressed due to solid organ transplantation (n=10) or immunosuppressive malignancy (n=11). The frontal (n=30, 39%) and parietal scalp (n=33, 43%) were the most commonly affected areas. 66% of patients had resection of the periosteum or tangential craniectomy of the outer table of the cranium, 22% had craniectomy without dural resection and 12% had craniectomy with dural resection; there were no intraoperative mortalities. Final pathology revealed 33 (43%) patients with stage 3 disease and 43 (57%) stage 4 disease (T4a: 36/43, T4b: 7/43). The median area of resection was 76 cm 2 (36-120) and all patients received rotational or free tissue flap reconstruction. 14 patients required titanium mesh cranioplasty for reconstruction in addition to flap closure. 23 patients received post-operative radiation (87%) or systemic chemotherapy (13%). Over median follow-up time was 11.5 months, 18 patients (24%) had a local recurrence; 3 required additional calvarial resection. Median OS and DFS were 22 and 27 months. Conclusion: Locally advanced cSCC is more common in elderly male patients. Radical resection with craniectomy is well-tolerated with adequate local control. The resulting defect requires flap surgery for closure.

Characteristics of patients with locally advanced cutaneous squamous cell carcinoma of the scalp 43 Changing Volume-Outcome Association for Esophagectomies -Do Prior Volume Thresholds Still Apply? K. Jogerst,* Y. Chang, B. Pockaj, C. Stucky, P. Cronin, N. Wasif. General Surgery, Mayo Clinic Arizona, Phoenix, AZ. Introduction: There is a well-known volume-outcome relationship for complex oncology surgeries, including esophagectomy. This association has inspired recommended minimum volume thresholds nationally. Improvements in perioperative care and surgical advancements have decreased postoperative mortality over time, leading to attenuation of the volume-outcome relationship. We hypothesize this attenuation changes the current minimum recommendation of 20 cases annually per hospital for esophagectomy. Methods: Patients with esophageal cancer undergoing surgery from 2005-2015 were identified in the National Cancer Database. High (HV), medium (MV), and low-volume (LV) hospital strata were defined by quartiles. Adjusted odds ratios (OR) and adjusted 30d mortality between low, medium and high volume hospitals were calculated and trended over time. Logistic regression analyses were used to perform risk adjustment. Results: The proportion of hospitals with 20 mean annual cases was 1.1% and the unadjusted 30day mortality for esophagectomy was 3.8% overall. Both unadjusted and adjusted 30day mortality for all three volume strata trended down from 2005-2015, with a disproportionate decrease for low and medium volume vs. high volume hospitals. Adjusted OR between volume strata decreased over time and by 2015 the odds of 30d mortality was not significantly different in MV vs. HV hospitals (OR 1.35, ). For hospitals with 20 cases annually the adjusted 30d mortality was 2.7%. To achieve this same 30d mortality the volume threshold shifted from 20 cases overall to 7 in 2015; Figure 1 . Conclusion: Only 1.1% of NCDB hospitals meet current recommended volume standards for esophagectomy. Differential improvements in postoperative mortality in low and medium volume hospitals compared to high volume ones has led to an annual volume of 7 cases in 2015 achieving the same adjusted 30d mortality as 20 cases in the 11-year cohort. In contemporary practice lower volume thresholds for esophagectomy may be appropriate and will likely increase the proportion of hospitals able to meet volume standards.

For hospitals with 20 cases annually the 30d mortality was 2.7% overall. To achieve this same 30d mortality the volume threshold shifted to 7 cases annually in 2015.

Characteristics on Clinical Outcomes J.P. Landry, 1 * B. Voros, 1 P. Boudreaux, 1 R. Thiagarajan, 1 R. Ramirez, 2 E. Woltering. 1 1. Surgery, LSU HSC New Orleans, Thibodaux, LA; 2. Ochsner Medical Center Kenner, New Orleans, LA. Background: Pancreatic neuroendocrine tumors (PNETs) are rare neoplasms that exhibit a wide range of malignant potential. Many patients with PNETs present with advanced disease; therefore, effective treatment for metastasis is essential. Recent studies report increased survival in patients with neuroendocrine tumors after surgical cytoreduction. Further studies are needed to evaluate if patients with PNETs benefit from surgical cytoreduction. Methods: A retrospective review was performed for a consecutive series of patients with PNET who presented to our tertiary referral center from 1998-2019. Patient and tumor characteristics, treatments, and outcomes were evaluated. Patients with additional malignancies were excluded. Results: In total, 301 patients were included with a median age at diagnosis of 53 years (range, 20-81 years). Median follow-up was 3.3 years (range, 28 days-26.7 years) and a total of 200 (66%) patients had resection of their primary PNET. By last follow-up, 231 (77%) patients had metastatic disease and 125 (54%) of these patients had surgical cytoreduction of their metastatic disease. Patients with primary tumors >2cm, lymphovascular or perineural invasion, intermediate or high grade tumors, or positive lymph nodes were significantly more likely to present with metastatic disease at diagnosis. On multivariate analysis, patients with high grade tumors (HR: 3.27, p<0.0001) or metastatic disease at diagnosis (4.39, p<0 .0001) had worse overall survival. Median progression-free survival after primary resection was 24 months and median overall survival for patients with metastatic disease at diagnosis was 9.1 years. Patients with high grade tumors had improved survival with cytoreduction of their metastatic disease (HR: 0.41, p=0.008) with a median survival of 7 vs 1.5 years, respectively. We found no difference in survival by tumor functionality or tumor size in patients with metastatic disease. Conclusions: In our cohort, metastatic disease and tumor grade were significant predictors of survival and cytoreduction showed significantly improved survival in patients with high grade tumors. Background: Early resection of the primary tumor in metastatic small bowel neuroendocrine (SB-NET) remains controversial. Conflicting data exist on its clinical and survival benefits. We compared long-term outcomes of upfront small bowel resection (USBR) to non-operative management (NOM) for metastatic SB-NET. Methods: A population-based analysis of patients with SB-NET metastatic at diagnosis between 2001-2017 was conducted by linking administrative datasets. USBR was defined as resection within 6 months of diagnosis. Primary outcomes were subsequent unplanned acute care admissions and small bowel related surgery. Secondary outcome was overall survival (OS). USBR and NOM patients were matched 2:1 with a propensity-score of age, sex, year of diagnosis, socioeconomic status, institution academic status, and functional status. We used time-to-event analyses with cumulative incidence functions and Andersen-Gill regression for primary outcomes, and Kaplan-Meier methods and Cox regression for OS. E-value methods assessed the potential for residual confounding. Results: Of 1000 patients identified, 785 (78.5%) had USBR. The matched cohort included 348 patients with USBR and 174 with NOM. Matched groups were well balanced with standardized mean differences <10% for matched variables. Patients with USBR had lower 3-year risk of subsequent admissions (72.6% vs 86.4%, p<0.001) than those with NOM, with hazard ratio (HR) 0.72 (95%CI 0.57-0.91). USBR was associated with lower risk of subsequent small bowel related surgery (15.4% vs 40.3%, p<0 .001), with HR 0.41 (95%CI 0.30-0.56). OS was superior for USBR patients compared to NOM (HR 0.55, 95%CI 0.41-0.74). E-values indicated it was unlikely that the observed risk estimates could be explained by an unmeasured confounder. Sensitivity analysis excluding emergent resections to define USBR did not alter the results. Conclusions: USBR for SB-NETs in the presence of metastatic disease was associated with clinical benefits over NOM, in terms of decreased subsequent admissions and interventions, and improved survival. USBR should be considered for metastatic SB-NETs to improve patient outcomes.

Cumulative incidence of unplanned admissions for patients with upfront small bowel resection compared to non-operative management, in the matched cohort. Background: PRRT was approved by the FDA for use in the US in January 2018. There is a paucity of data regarding clinical outcomes and toxicity of this modality since its US inception. Methods: We reviewed all patients treated at the Medical College of Wisconsin with PRRT for metastatic neuroendocrine tumors. Prior to treatment, all patients underwent Gallium-68 DOTATATE PET/CT to demonstrate receptor avidity. 177 Lu-DOTATATE was administered at 200 mCi over 30 minutes for 4 cycles given 8 weeks apart. Antiemetics were given 30 minutes prior to amino acid infusion. A solution of Arginine HCl (2.5%) and Lysine HCl (2.5%) in 1L NS was administered over 4 hours beginning 30 minutes prior to radiotherapy treatment. CT/MRI imaging was obtained after cycles 2 and 4 to assess response. Results: We have treated 35 patients, all had evidence of disease progression prior to treatment. Median age was 62 yrs (IQR: 53-69) and 17 (49%) were female. Primary tumor sites were GI (10; 29%), pancreas (19; 54%), lung (1; 3%), unknown (3; 9%), and paraganglioma (2; 6%). Tumor grade was G1 in 14 (40%), G2 in 19 (54%), G3 in 1 (3%), and in 1 (3%) patient grade was unknown. All 4 planned cycles were completed in 27 (77%) patients; 7 (20%) had therapy halted due to disease progression; 1 (3%) patient is still in treatment. Imaging to assess response was available in 33 patients; 9 (27%) showed response, 12 (36%) had stable disease, and 12 (36%) demonstrated disease progression. Transient toxicity was noted in 19 (54%) patients; leukopenia (WBC <3000/ L) in 10 (29%), thrombocytopenia (platelets <75000/ L) in 4 (11%), anemia (HgB <10 g/dL) in 10 (29%), nephrotoxicity (eGFR< 60mL/min) in 4 (11%), and hyperbilirubinemia (T.bili>1.5 mg/dL) in 1 (3%). Toxicities resulted in only 1 treatment delay. Grade 1/2 adverse events (AE) included nausea, fatigue, and diarrhea in 3 patients; 1 patient developed alopecia, and 1 developed Grade 3/4 fatigue and thrombocytopenia. Conclusion: PRRT was well tolerated in our early experience with reversible toxicity and few AE. Disease response or stabilization was seen in two-thirds of patients.

Management of Appendiceal Neuroendocrine Tumors: Metastatic Potential of Small Tumors J.P. Landry, 1 * B. Voros, 1 R. Ramirez, 2 P. Boudreaux, 1 E. Woltering, 1 R. Thiagarajan. 1 1. Surgery, LSU HSC New Orleans, Thibodaux, LA; 2. Ochsner Medical Center Kenner, New Orleans, LA. Background: Appendiceal neuroendocrine tumors (ANETs) are rare neoplasms usually discovered incidentally during appendectomy. ANETs <2cm were thought to have no metastatic potential and this dogma has driven management. Our aim was to evaluate the metastatic potential of ANETs <2cm and identify potential risk factors for metastatic disease and worse outcomes. Methods: A comprehensive retrospective review was performed in a consecutive series of patients with an ANET who presented to our tertiary referral center from 1998-2019. Demographics, tumor characteristics, treatment, and clinical outcomes were evaluated. Patients with additional malignancies were excluded. Results: In total, 114 patients were included. Median follow-up was 3.3 years (range, 21 days-15 years). At last follow-up, 34 (30%) patients had positive regional lymph nodes and 20 (18%) patients had metastatic disease. Of the 20 patients with metastatic disease, 11 (55%) had primary ANETs <2cm. Patient age >40 years at diagnosis, ANETs with serosal invasion, lymphovascular invasion, intermediate tumor grade, or positive lymph nodes were significantly more likely to present with metastatic disease at diagnosis. We found no significant difference in the rate of lymph node metastasis, metastatic disease at diagnosis, or overall survival when patients were stratified by tumor size or type of resection (appendectomy vs right hemicolectomy). On multivariate analysis, patients with metastatic disease at diagnosis had worse overall survival (HR=24.4, p=0.008). Conclusions: In our cohort, tumor size was not a significant risk factor for metastatic disease or worse outcome as many patients with ANETs <2cm developed metastatic disease. Appendectomy alone was sufficient surgical management for most ANETs. Patients with identified risk factors for metastatic disease, regardless of primary ANET tumor size, should be evaluated thoroughly and counseled for further management and surveillance.

Introduction Tumor biomarkers (TBMs) reflect disease burden and have implications for survival in patients with resected small bowel neuroendocrine tumors (SBNETs). The NCCN recommends evaluating only chromogranin A (CgA) when monitoring for disease progression, while some question the utility of using any markers for surveillance. This study sought to determine the performance of CgA, pancreastatin (PST), neurokinin A (NKA), and serotonin (5HT) in the follow-up of resected SBNET patients. Methods Patients undergoing surgery for SBNETs were identified from a single institutional database. Serum CgA, PST, NKA, and 5HT levels were assessed as categorical (normal vs. elevated) and continuous variables for association with progression-free (PFS) and overall survival (OS) by the Kaplan-Meier method with Cox multivariable models adjusting for confounders. Sensitivity, specificity, and predictive values of TBM levels in identifying imaging-confirmed progression were calculated. Results There were 218 patients (44% female, 90% node+, 73% metastatic, 97% G1 or G2). Higher levels of CgA, PST, NKA, and 5HT correlated with higher grade disease, node positivity, and metastatic disease at presentation (p<0.05 for all). Elevated pre-or post-operative CgA, PST, and NKA were significantly associated with lower PFS and OS (p<0.01, median follow-up 49.6 mos.; Table) . Normal levels of CgA, PST, and NKA were present in 23. 3, 16.9, and 72 .6% of patients with progression, while elevated levels were present in 47.7, 24.8, and 1.3% of patients without progression. Using TBMs to determine progression revealed the superiority of PST (78.8% correctly classified) over CgA (64.0%) or CgA and PST together (60.3%). Conclusion Surveillance recommendations for patients with resected SBNETs include TBMs and imaging. New therapies are considered when progression occurs. Performance of 5HT was variable and NKA was seldom elevated. Post-operative CgA and PST values correlated with PFS and OS, but PST outperformed CgA in predicting progression. Using both markers added little to using PST alone. Results from this large series of surgically resected SBNETs suggest that PST should replace CgA for disease surveillance. Adrenocortical carcinoma (ACC) is an aggressive cancer associated with poor 5-year survival. No effective systemic treatments in patients with advanced disease are available to date. Because of the lack of anti-tumor efficacy of current regimens, we aimed to identify synergistic drug combinations using the quantitative high-throughput screening of drugs with known toxicity profiles in humans. In this study, we screened 4,991 compounds on SW13 and NCI-H295R ACC cell. Of 85 drugs that passed selection criteria, the maternal embryonic zipper kinase inhibitor (OTS167) and cyclin-dependent kinase inhibitor (CDK) (RGB286638) drugs were selected. Both drugs 1) had high anti-proliferative effects in both cell lines at achievable serum levels in human and 2) target cell cycle progression and mitosis via different mechanisms. We analyzed multiple publicly available databases and found that MELK and CDK1-2 were overexpressed in ACC and independently associated with advanced ACC, metastasis, recurrence, and mortality. In vitro single-drug treatment with OTS167 and RGB286638 in ACC cell lines showed decreased MELK expression and increased G2M cell cycle arrest, respectively. The combination treatment of OTS167 and RGB286638 were synergistically effective in cell proliferation assay, tumor spheroid formation, and clonogenic assay. We found increased G2/M arrest, p21, and increased caspase-dependent apoptosis in combination treatment compared to the single-drug. We demonstrated the novel pathway that this combination decreased AXIN2, GSK3-, FOXM1, and -catenin. To investigate the effects on cancer invasiveness, we found a lower expression of epithelial-mesenchymal-transition (EMT) markers such as fibronectin and SNAIL2, compare to the single-drug groups. In summary, the combination of OTS167 and RGB286638 is synergistically effective in ACC via AXIN2/GSK3-/FOXM1/ -catenin and the reduction of EMT markers. Anaplastic thyroid cancer (ATC) remains a lethal disease with an average life expectancy of 2-6 months. Response to BH3 mimetics in solid tumors is usually limited and short-lived due to their reliance on multiple anti-apoptotic proteins. Our work shows that small molecule, BTSA1 that directly binds to and activates BAX effectively induces apoptosis in ATC cells; this effect is potentiated with the addition of a BH3 mimetic. We postulate that these effects can be seen in the in vivo mouse model. Dose-response profiles of BH3 mimetic, ABT-263 (Navitoclax) and BTSA1, single agent and combination were generated on a panel of 10 ATC cell lines. Co-IP followed by western blot was performed to assess BAX-bound proteins following treatments. We utilized an immunodeficient mouse model orthotopically implanted with human ATC cell lines for in vivo studies. We developed a novel thyroid window in order to visualize ABT-and BTSA1-induced apoptosis in vivo with cell lines infected with stably expressed RFP-apoptosis inducible GFP as a reporter. We found BTSA1 induces prompt apoptosis in all cell lines. However, co-IP assays showed that BTSA1-activated BAX can still be sequestered by BCL-XL, which likely reduces its efficacy. Thus, we tested the ability of the ABT-263 to increase BTSA1 efficacy by inhibiting BCL-XL. In fact, we found a dramatic decrease in the BTSA1 IC50 in the presence of a low dose of ABT-263 that has no effect as single agent. Also, we found that the efficacy of ABT-263 was equally amplified by a low dose of BTSA1. Time-course analysis of caspase 3/7 activity confirmed that these compounds induce prompt apoptosis. These drugs were found to be similarly effective in 3D spheroid tissue culture, suggesting in vivo efficacy. Applying these drugs to our infected RFP-inducible GFP cell lines further confirms drug-induced apoptosis and presents a novel model of visualizing these effects in vivo. These data suggest that there is a promising therapeutic opportunity in targeting the apoptosis pathway in ATC with a BAX activator accompanied by a BH3 mimetic. Confirming in vivo efficacy substantiates these drugs' translational potential. Background: Anaplastic thyroid cancer (ATC) and poorly differentiated thyroid cancer (PDTC) are rare but devastating malignancies with ineffective current treatments. Glucose and glutamine are essential nutrients for cancer cell growth that are metabolized through the glycolysis and glutaminolysis pathways, respectively. We hypothesized that suppression of these metabolic pathways using glycolytic inhibitor 3-bromopyruvate (3-BP) and glutamine inhibitor 6-diazo-5-oxo-L-norleucine (DON) would attenuate tumor cell growth, migration, and invasion. Methods: ATC cell line 8505C and PDTC cell line BCPAP were grown in media containing high (25 mM) (HG) or low (3 mM) (LG) glucose concentrations with supplemental administration of 3-BP (100-200 uM) and/or DON (250 uM). Cell proliferation, migration, and invasion assays were performed. Results: For 8505C, LG versus HG significantly lowered cell proliferation by 33% (P = 0.01), migration by 39% (P < 0.0001), and invasion by 30% (P = 0.001). Adding 3-BP and DON to LG produced even greater inhibition of proliferation by 94% (P < 0.0001), migration by 85% (P < 0.0001), and invasion by 72% (P < 0.0001) compared to HG. For BCPAP, LG versus HG did not suppress cell proliferation, migration, or invasion. However, supplementation of 3-BP and DON to LG led to significant reductions in cell proliferation by 73% (P < 0.0001), migration by 48% (P < 0.0001), and invasion by 26% (P < 0.01) compared to HG. In both cell lines, concomitant use of 3-BP and DON yielded greater anti-tumor effects than use of either inhibitor alone. Conclusions: Inhibition of glycolysis and glutaminolysis pathways using 3-BP and DON, respectively, significantly reduced in vitro cellular proliferation, migration, and invasion in a low glucose environment for ATC and PDTC cell lines. These results suggest that targeting suppression of glucose and glutamine metabolic pathways may represent an effective strategy in the treatment of ATC and PDTC. Introduction This study compares oncologic outcomes of patients who receive neoadjuvant chemotherapy (NAC) and neoadjuvant chemotherapy plus chemoradiation (NACXRT) for resectable gastric adenocarcinoma (GA). Methods We analyzed data from a prospective database of patients who underwent resection for gastric cancer at our institution from 7/1995 to 7/2018. Rates of microscopic (R0) resection, pathologic complete response (pathCR), and pathologic stage were assessed. Comparisons were made between propensity matched cohorts based on age, sex, race, histology, and clinical stage. Kaplan-Meier survival analyses were used to examine effects on overall survival (OS) and disease-free survival (DFS). Results 438 patients underwent gastrectomy following neoadjuvant therapy. Age was 61 12y, 62% were male, and 55% were white. Of the entire cohort, 343 (78%) received NACXRT while the remaining 95 (22%) underwent NAC. 65 patients with GA who underwent NAC were effectively matched to 65 patients who underwent NACXRT. Of the patients who received NAC, reasons for XRT exclusion were that of tumor related variables (n=17, 26%), patient preference due to logistical/financial reasons (n=7, 11%), toxicity (n=5, 8%), or other/physician preference (n=36, 55%). Patients who underwent NACXRT vs NAC had similar rates of R0 resection (9.2% vs 6.2%, p=0.74) yet significantly higher rates of pathCR (27.7% vs 1.5%, p<0.001). Although the matched cohorts had similar preoperative clinical staging, those who underwent NACXRT had significantly lower final pathologic stage (p=0.002). The Figure depicts OS and DFS based on NAC/NACXT of the propensity matched cohorts. Median DFS for NAC and NACXRT was 50.9(95% CI: 4.65-97.2) vs 122.1(69.0-175.1) months (p=0.07); median OS was 70.7(23.9-117.5) vs 122. 1(68.7-175.4 ) months (p=0.21), respectively. Conclusions NACXRT is associated with higher rates of pathCR and subsequent lower overall final pathologic stage, yet differences in DFS and OS did not reach significance. Ongoing investigation is needed to delineate the optimal preoperative therapeutic regimen. Background: Neoadjuvant therapy is the standard of care for locally advanced esophageal and gastroesophageal junction (GEJ) adenocarcinoma, with most patients receiving neoadjuvant chemoradiation (CRT). CRT can be delivered concurrently or sequentially after induction chemotherapy. The purpose of this study was to compare pathologic complete response (pCR) and overall survival (OS) among patients who received concurrent versus sequential CRT. Methods: Using the National Cancer Database (NCDB), patients who received neoadjuvant CRT and underwent curative intent esophagectomy for esophageal or GEJ adenocarcinoma from 2006-2015 were included. Patients with clinical T4 or metastatic disease were excluded. Concurrent CRT was defined as radiation treatment starting within 6 weeks of chemotherapy start. Sequential CRT was defined as radiation treatment starting greater than 6 weeks after chemotherapy start. Propensity weighting was conducted to balance patient, disease, and facility covariates between groups. Results: 12,460 patients met inclusion criteria. 11,880 (95%) patients received concurrent CRT and 580 (5%) patients received sequential CRT. Patients who received sequential CRT were significantly younger (mean age: 60.7 vs 62.2 years), had higher clinical nodal stage (N2-3: 14.7% vs 10.1%), and were more often treated at academic/research hospitals (67.1 vs 55.5) (all p 0.001). pCR was achieved in 16.2% of patients who received sequential CRT and in 14.0% in patients who received concurrent CRT (p=0.131). Following propensity weighting, OS was significantly improved among patients who received sequential versus concurrent CRT (HR 0.82; 95% CI 0.74-0.92; p<0.001) with a median OS for the sequential cohort of 41.4 months versus 29.4 months for those who received concurrent CRT. Conclusion: In this retrospective study from a large national database of patients who received neoadjuvant CRT for esophageal and GEJ adenocarcinoma, sequential CRT is associated with a significant OS benefit. These results merit consideration of a well powered prospective multi-institutional randomized clinical trial to further evaluate this observed difference.

Propensity-weighted Kaplan-Meier survival curve comparing concurrent vs. sequential neoadjuvant chemoradiation therapy for esophageal and gastroesophageal junction adenocarcinoma. Before PSM, most patient characteristics differed significantly between MIS patients undergoing the two types of surgery. Between the two cohorts, each including 889 propensity score-matched patients, actual 5-year OS did not differ significantly: 54.0% following minimally invasive and 50.4% after open gastrectomy, respectively (p = 0.205). Subgroup analysis confirmed that survival was similar between surgical cohorts among patients with each stage of GA and between those undergoing distal or total/proximal gastrectomy. On multivariable analysis, surgical approach was not an independent prognostic factor. Conclusions: Following PSM of US and Chinese patients with GA undergoing gastrectomy, long-term survival does not significantly differ between patients undergoing minimally invasive vs. open gastrectomy. Introduction: For multiple GI cancers, appropriate LAN is associated with improved local disease control and cancer-specific OS. For colon cancer, a minimum of 12 examined lymph nodes (exLNs) is standard for an adequate LAN to accurately identify stage II (T3-4N0) pts; fewer than 12 exLNs is associated with higher risk of recurrence and is an indication for adjuvant chemotherapy (AC). SBA is a rarer entity for which management is not standardized and the importance of an adequate LAN has not been established.

Methods: We used the National Cancer Database to identify pts diagnosed and treated for node negative stage II SBA (T3-4N0) from 2004 to 2016 (N = 2,709). We compared pts with <8 exLNs (N = 1,259) to those who had 8 or more (N = 1,450). We examined the effect of co-variates including age, race, Charles-Deyo score, tumor grade, T-stage, margin status, primary site, number of exLNs, and receipt of AC on actuarial 7-y OS in stage II (T3-4N0) SBA pts. Results: On univariate analysis, <8 versus 8 exLNs was associated with decreased 7-y OS (38.9% vs. 56.7% respectively, p<0.0001). On multivariate analysis, examination of <8 LNs (HR:1.66, p <0.0001), T4 tumors (HR:1.61, p <0.0001), high-grade tumors (HR:1.25, p=0.0028), and positive margin status (HR:1.85, p<0.0001) were each associated with decreased OS. In pts with <8 exLNs, high-grade tumors, T4 tumors, and positive margin status were associated with decreased OS (HR: 1.37, 1.49, 1.87; respectively) . For pts with 8 exLNs, T4 tumors and positive surgical margins were associated with decreased OS (HR:1.66, 1.91, respectively). The use of AC was associated with an improved 7-year OS in pts with <8 exLNs (HR: 0.781, p= 0.019) but not in pts with 8 exLNs (HR: 1.125, P: 0.35). Discussion: In stage II SBA, pts with <8 exLNs had poorer OS compared to those with 8 exLNs. Pts with <8 exLNs had an associated survival benefit with chemotherapy while those with 8 exLNs did not. These data suggest that <8 exLNs is inadequate for accurate staging of pts with SBA. In stage II SBA pts, a minimum of 8 LNs examined is necessary for accurate staging and selection for AC. Background: A "sandwich" approach involving pre-operative chemotherapy (preCTX), resection, and post-operative chemotherapy (postCTX) is a standard of care treatment sequence for patients with resectable gastric cancer (GC). However, the necessity and impact of the postCTX on overall survival (OS) is not well defined. Methods: The NCDB (2006-2014) was queried for patients who received preCTX and resection for GC. Analysis was performed to identify factors influencing receipt of postCTX. The influence of postCTX on survival was evaluated in propensity-matched groups. Results: Among patients who received preCTX and resection for GC (n=3449), 1091 (31.6%) received additional postCTX and 2358 (68.4%) did not. Characteristics associated with receipt of postCTX included younger age, Asian race, private insurance, a comorbidity index of 0, poorly differentiated tumors, lymphovascular invasion, T3 tumors, 15LNs examined, and distal GC (all p<0.05). After propensity matching to establish well balanced sociodemographic and clinicopathologic cohorts (n=1091/group), median OS was 56.8 vs 52.5 months for those who did and did not receive postCTX, respectively (p=0.131). Subset analysis according to tumor grade, lymphovascular invasion, number of nodes evaluated, and T and N class did identify an improvement in OS for patients with N1 disease who received postCTX (79.6 months, n=222) vs those who did not (41.3 months, n=211), p=0.025. However, no other subgroup had a significant survival benefit with postCTX. Conclusions: Additional postCTX is administered to a minority of patients who receive preCTX and gastrectomy for GC and its influence on OS appears limited to a small subset of patients. Future trial design should aim to more clearly define patients who will benefit from additional postCTX. In addition, consideration should be given to shifting toward planned total neoadjuvant CTX. BACKGROUND: Neoadjuvant chemoradiotherapy is considered standard of care for resectable esophageal cancer patients. Recent evidence debates whether neoadjuvant radiation adds survival benefit especially with its added associated postoperative morbidity. This study aims to assess evidence from Randomized Controlled Trials (RCTs). METHODS: A comprehensive online search of MEDLINE, EMBASE, PubMed, SCOPUS and the Cochrane database was conducted (Jan 2000 -Oct 2019 . RCTs comparing neoadjuvant chemotherapy to chemoradiotherapy in resectable esophageal cancer patients were reviewed. Pooled risk ratios (RR) and 95% confidence intervals (CI) were calculated. Heterogeneity and bias were assessed. RESULTS: Overall, 5 RCTs with a total of 769 patients were included; 381 patients who had neoadjuvant chemotherapy alone were compared to 388 patients who had chemoradiotherapy. Demographic variables, comorbidities and tumor stages were similar in the two groups. Meta-analysis of included studies showed that patients treated with neoadjuvant chemoradiotherapy had slightly better 3-Yr survival (RR:0.8; CI: 0.65-0.99; p=0.04), higher rate of R0 resection (RR:0.88; CI: 0.83-0.93; p<0.001), higher complete pathological response rate (RR:0.18; CI: 0.09-0.33; p<0.001). However, they also had significantly higher postoperative pulmonary, cardiovascular and anastomotic leakage rates (p<0.05). In addition, there was no difference in 5-Yr overall survival compared to those who had chemotherapy alone (RR:0.8; CI:0.63-1.02; p=0.07). CONCLUSION: Despite improving pathological response and resection margins; given the added morbidity of neoadjuvant chemoradiation and the lack of definitive improvement in 5-year survival over chemotherapy alone, individual selection of patients for multimodality therapy may further improve outcomes. Background. The incidence of lower esophageal and gastroesophageal junction adenocarcinoma has sharply increased over the past several decades and is a serious public health problem. Preoperative therapy with either chemotherapy or chemoradiation is recommended, but the optimal regimen is unknown. We used the National Cancer Database (NCDB) and propensity score matching to investigate whether preoperative chemoradiation therapy offers an advantage over chemotherapy alone for patients with these tumors. Methods. From the NCDB esophageal and gastric dataset we selected patients with either lower esophageal or gastric cardia adenocarcinomas who had undergone definitive resection after chemotherapy or chemoradiation. We used propensity score matching to balance groups based on the preoperative treatment they received. We then used conditional multivariable logistic regression and cox proportional hazard models to examine the association between preoperative therapy regimen and pathological response, overall survival (OS), and postoperative outcomes. Results. Our study included 16,969 patients; 14,735 (86.8%) had received preoperative chemoradiation. Propensity score matching created 2,221 pairs. Patients receiving chemoradiation were 2.3 times (95% confidence interval [CI], 1.89-2.87 times) more likely to achieve complete response in the primary tumor than were those receiving chemotherapy alone; however, chemoradiation was not associated with improved OS (hazard ratio, 1.02; 95% CI 0.94-1.11). Short-term outcomes (length of stay, mortality, and readmissions) were similar between the two groups. Conclusions. Preoperative chemoradiation was associated with a higher complete response rate in the primary tumor but not with improved OS in lower esophageal and gastroesophageal junction adenocarcinoma. Background Upfront surgery is standard of care for stage I gastric cancer. Despite this, many clinicians offer preoperative therapy for clinical stage I disease with signet ring cell histology, given its aggressive biology. We aimed to assess the validity of this practice. Methods The National Cancer Database(2004-15) was reviewed for pts with non-metastatic signet ring cell gastric cancer who underwent treatment with surgery alone, perioperative chemotherapy, neoadjuvant therapy, or adjuvant therapy. Analysis was stratified by preoperative clinical and pathologic stage. Primary outcome was overall survival(OS). Results Of 3000pts, median age was 61(IQR:51-70). 34% were clinical stage I(n=1018) of which 53% received surgery alone(n=542), 5% perioperative chemotherapy(n=47), 12% neoadjuvant therapy(n=125), and 30% adjuvant therapy(n=304). Median follow-up was 26mos. For clinical stage I disease, surgery alone was associated with improved median OS(108mos) when compared to perioperative chemotherapy(80mos), neoadjuvant therapy(41mos), or adjuvant therapy(73mos, all p<0.001)(Fig1A). For pathologic stage I, surgery alone had equivalent survival to perioperative and adjuvant therapy p=0.22)(Fig1B) . Concordance between clinical and pathologic stage I was 56%, specifically 41% of clinical stage I pts were upstaged to pathologic stage II(44%) and stage III(56%). Adjuvant therapy for these pts was associated with improved median OS compared to pretreatment(perioperative chemotherapy/neoadjuvant therapy) for those upstaged to pathologic stage II(122vs37mos, p<0.001) or stage III(40vs18mos, p<0.001) disease(Fig1C-1D). Conclusions Our stage-stratified study demonstrates improved survival with upfront surgery for clinical stage I signet ring cell gastric cancer. Despite 41% of clinical stage I pts being upstaged to stage II or III on final pathology, adjuvant therapy offers a favorable rescue strategy, with improved outcomes compared to those treated preoperatively. Surgery alone also affords similar survival for pathologic stage I disease compared to multimodal therapy. This study challenges the intrinsic bias to over-treat stage I signet ring cell gastric cancer. Background: Patient derived xenografts (PDX) can be a useful platform to perform preclinical testing of novel therapeutic strategies. The Wnt pathway is dysregulated in nearly 30% of gastric cancer and represents a promising target for a novel class of therapy. However, while Wnt inhibitors are in clinical trials for several cancers, none include gastric cancer. To test WNT974, a commercially available, first-in-class, picomolar inhibitor the Wnt acyltransferase Porcupine, as a therapy for gastric cancer, we established PDX from human gastric cancer patient samples. Methods: Consent was obtained from gastric cancer patients undergoing pre-treatment endoscopic biopsies or postneoadjuvant therapy gastrectomies. Tissue samples (~10mg from biopsies and 100 mm 3 from resections) were coated with Matrigel and subcutaneously implanted into the flanks of NSG mice. Successful engraftment was defined as growth of tumor after transplantation from the initial implant of human tissue into a second murine passage. Two PDX lines were used to test WNT974, which was given daily at 5mg/kg via oral gavage. Results: The engraftment rates were 41% (31 of 75 samples) for EGD biopsies and 61% (20 of 33) for resection samples. Median time to first passage was 9.8 weeks (range: 5.3 to 23) for EGD biopsies and 17.8 weeks (5.9 to 49.1) for post-treatment samples (P < 0.01). Almost all samples were able to be serially passaged and reconstituted after deep freeze. Histology was generally maintained through passages. Median tumor cellularity in engrafted samples was 60% (range: 0 to 95%) and 42.5% in non-engrafted tissue (5-100%; non-significant). Diffuse type cancers engrafted at a lower rate than intestinal and mixed-type tumors (34.9%, 48.9%, and 81.8%, respectively; P=0.02). Patient gender, clinical stage, tumor location, ethnicity/race were not associated with engraftment. After three weeks of treatment, there was no difference in tumor growth between vehicle and WNT974 groups. Conclusions: We have established multiple PDX lines from gastric cancer patients that are tractable for preclinical testing. Further study will be needed to determine the efficacy of WNT974 for gastric cancer.

Endoscopic Therapy Versus Esophagectomy for Stage 1 Esophageal Cancer: Less is More J.C. McCarty, 2 R.K. Parker, 3 R.J. Vidri, 1 * K.A. Robinson, 4 S. Gangadharan, 4 P. Iyer. Portland, ME; Boston, MA; 3. Tenwek Hospital, Tenwek, Kenya; Boston, MA; 5. Mayo Clinic, Rochester, MN. Introduction: Current guidelines recommend consideration of endoscopic therapy (ET) when treating select early stage esophageal cancers. The proportion of early stage esophageal cancers treated with ET compared to esophagectomy has increased over time. Overall and cancer-specific survival have not been shown to be superior with ET in prior population-based studies. We thus evaluated cancer-specific survival comparing patients treated with ET compared to esophagectomy. Methods: We performed a retrospective cohort study using the Surveillance, Epidemiology, and End Results (SEER) Database from 2004-2015 of patients with node-negative, superficial (T1a/T1b) esophageal cancer treated with ET or esophagectomy. Competing-risks models were used to compare cancer-specific survival. Cox proportional hazards models were used to asses overall survival. Subgroup analysis was performed comparing time periods 2004-2009 and 2010-2015 Conclusion: Endoscopic therapy was associated with improved cancerspecific survival compared to esophagectomy in stage I esophageal cancer. This advantage was more pronounced for patients treated after 2009, potentially due to increasing clinician expertise in performing endoscopic therapy and patient selection. Introduction: Minimally invasive esophagectomy (MIE) has emerged as an alternative to traditional open techniques. Our objectives were to evaluate perioperative and oncologic outcomes in our robotic-assisted MIE (RAMIE) series and compare outcomes with open esophagectomy (OE) using contemporary controls. Methods: We retrospectively reviewed patients who underwent RAMIE for malignancy. Patient characteristics, perioperative outcomes, and survival were evaluated. Contemporary controls who underwent OE were identified and propensity-score matched from NSQIP and NCDB databases and utilized for perioperative and oncologic outcome comparison. Results: We identified 350 patients with who underwent robotic Ivor-Lewis esophagectomy between 2010 and 2019. Median BMI was 27.4, 32% demonstrated a CCI 4. Nodal disease was identified in 50% of patients and 74% received neoadjuvant chemoradiotherapy. Adenocarcinoma was the predominant diagnosis (87%). Mean operative time and blood loss were 424 minutes and 232 mL, respectively. Anastomotic leak occurred in 16% of patients, 2% requiring reoperation. Median length of stay (LOS) was 9 days, and 30-day mortality was 3%. A median of 21 nodes were dissected with 96% achieving an R0 resection. Median survival was 67.4 months with 1-and 5-year survival estimates of 89% and 53%. 127 RAMIE were matched 1:1 to the NSQIP OE control. RAMIE demonstrated decreased LOS (10 vs. 11 days, p=0.041) and re-operative rates (3.2 vs. 11%, p=0 .025) and longer operative time (427 vs. 312 min, p=0.039) in comparison to NSQIP cohort. There was no difference in leak rate or mortality. 322 RAMIE were matched to an NCDB cohort. There was no difference in mortality, extent of lymphadenectomy or overall survival between RAMIE and OE. Conclusion: We defined the perioperative and oncologic outcomes in the largest reported RAMIE series to-date. RAMIE is associated with a longer operative time but is associated with a decreased LOS, likely reducing resource utilization. RAMIE demonstrates comparable oncologic outcomes to OE without increased major complications. Our data support continued use and study of this approach. INTRODUCTION: Most patients in the ACOSOG Z0011 trial were > 40 years of age, with ER positive, grade 1-2, invasive ductal cancers with no lymphovascular invasion (LVI). We evaluated treatment patterns and clinical outcomes in patients with positive sentinel lymph node biopsy (+SLNB) underrepresented in the Z11 trial but treated by Z11 criteria. METHODS: The NCDB was queried for female patients from 2012 to 2015 who met Z11 criteria and were grouped by the following characteristics: age < 40 years, ER negative, LVI, grade 3, invasive lobular (ILC) histology and facility type. Patients treated according to Z11 (no axillary lymph node dissection (ALND) after +SLNB) were compared to patients undergoing ALND after +SLNB. Kaplan-Meier method was used to estimate overall survival (OS), and Cox proportional-hazards regression model was used to identify predictors of survival. RESULTS: 15,260 patients were identified, of which 208 were age < 40 years, 2,394 had ER negative disease, 8,728 had LVI, 6,612 had grade 3 tumors, and 2,232 had ILC. Management according to Z11 criteria was lowest at community cancer programs (CCP) (31.8% for age <40 years, 56.8% for ER negative, 62.3% for LVI, 64.0% for grade 3, and 71.3% for ILC). Highest rates of Z11 application were at academic/research programs for ER negative (69.4%), LVI (74.4%), grade 3 (71.1%), and ILC (78.1%) groups, and at integrated cancer programs for age < 40 years (84.0%). Median OS did not differ between SLNB-alone vs completion ALND, when patients were stratified by age < 40 years, ER negative disease, LVI, grade 3 tumor, or lobular histology. Older age and higher grade and T stage were associated with increased risk of mortality. ER positivity, receipt of hormone therapy or chemotherapy, and care at a non-CCP were associated with decreased risk of mortality. CONCLUSIONS: Our study demonstrates that patients underrepresented in the Z11 trial have no differences in survival after +SLNB according to treatment with SLNB alone vs SLNB + ALND. Moreover, academic/ research programs and integrated cancer programs adopt non-Z11 inclusion criteria at highest rates.

Variation in Axillary Surgery in Sentinel Lymph Node Positive Breast Cancer Patients in the Netherlands J. van Steenhoven, 1 * M. van Maaren, 2 A. Kuijer, 3 S. Siesling, 2 M. Smidt, 4 T. van Dalen. 1 1. Surgery, Diakonessenhuis Utrecht, Amsterdam, Netherlands; 2. Netherlands Comprehensive Cancer Organisation, Utrecht, Netherlands; 3. St. Antonius Hospital Nieuwegein, Nieuwegein, Netherlands; 4. University Medical Center Maastricht, Maastricht, Netherlands. Introduction: Following publication of the Z0011 results in 2010, axillary dissection was abandoned gradually over the subsequent years in patients with tumor positive lymph nodes (SLN N+). In this study we addressed patient-, tumor-, and institutional factors in relation to the chance of axillary surgery being abandoned in SLN N+ patients. Methods: All primary invasive breast cancer patients treated between 2011-2016 who were staged as SLN N+ (Nmi or N1a) were selected from the Netherlands Cancer Registry (NCR). Descriptive analyses were performed to evaluate geographic variation in the performance of an ALND. Multivariable logistic regression analyses examined the influence of tumor-, patient-, and institutional factors on the performance of ALND. Regarding institutional factors, hospital type, patient volume and the propensity to apply novel treatment insights were addressed. For the latter, we used the percentage of gene-expression profile use (GEP) within a hospital, in patients eligible for GEP use, as a proxy for the hospital's innovative character. Results: A total of 11,392 SLN N+ patients were identified. The performance of an ALND in Nmi patients decreased from 44% of patients in 2011 to 3% of patients in 2016 and from 82% to 12% in N1a patients. Substantial variation in axillary surgery was observed between geographic regions (range 13-47%). High grade, multifocality, triple negative subtypes and young age were associated with a higher probability to undergo an ALND. Regarding institutional factors high patient volume and academic setting were associated with a higher probability of omitting an ALND [OR 0.86 95% CI 0.77-0.96 and OR 0.44 95% CI 0.35-0.56, respectively]. The deployment of innovative techniques strategies was associated with a propensity to abandon an ALND . Conclusion: In this study, we observed substantial regional variation in axillary surgery. A larger patient volume, an academic setting and a propensity to use other innovative techniques in the same patient category of breast cancer patients, were hospital factors that identify early adopters in terms of de-escalating axillary surgery.

Extranodal Metastases in the Axillary Fat Indicate the Need for Axillary Dissection Among T1-2cN0 Patients with Positive Sentinel Nodes A. Mamtani,* A. Barrio, D. Goldman, H. Wen, A. Vincent, M. Morrow. Memorial Sloan Kettering Cancer Center, New York, NY. Introduction The ACOSOG Z0011 trial demonstrated the safety of omitting axillary dissection (ALND) in T1-2cN0 patients with <3 positive sentinel nodes (SLN) undergoing breast-conservation therapy. While microscopic extracapsular extension (mECE) of >2mm is associated with increased nodal burden, the significance of extranodal metastases (EM) in the axillary fat in this population is uncertain. Methods Consecutive patients with T1-2cN0 breast cancer undergoing sentinel node biopsy (SLNB) and ALND for SLN metastasis from 1/2010-12/2018 were identified. Those who had neoadjuvant therapy, recurrent, or bilateral disease were excluded. EM were defined as intravascular tumor emboli or discrete metastatic deposits in the axillary fat. Clinicopathologic characteristics and nodal burden were compared by EM status with Fisher's exact, Wilcoxon Rank sum, and logistic regression. Results Among 1114 patients, 113 (10%) had EM: 81 (72%) were intravascular tumor emboli and 32 (28%) were soft tissue deposits. Patients with EM had larger tumors (median 2.2 vs 2.1cm, p=0.033) and more often had mECE (83% vs 44%, p<0.001) (Table) . On univariable analysis, the presence of EM (OR 9.7, 95% CI 6.4-14.7), larger tumor size (OR 1.5 95% CI 1.3-1.7), and mECE (OR 10.7, 95% CI 6.9-16.8) were associated with 4 additional positive non-sentinel nodes (NSLN) at ALND (all p<0.001). On multivariable analysis, EM remained associated with 4 positive NSLN (OR 5.5, 95% CI 3.5-8.9, p<0.001). There was no interaction between mECE and EM (p=0.8). Intravascular tumor emboli and soft tissue deposits had similar associations with 4 positive NSLN (OR 5.4, , respectively. In an interaction model, EM remained strongly associated with 4 positive NSLN (OR 7.2, 95% CI 4. 1-12.8) in patients with 1-2 positive SLN (n=925). Conclusions Among patients with T1-2cN0 breast cancers with positive SLN, EM are strongly and independently associated with 4 positive NSLN at ALND. Even among those who may otherwise meet criteria for omission of ALND, the presence of EM in the axillary fat warrants consideration of completion axillary dissection.

Clinicopathologic characteristics and nodal disease burden HR+: hormone receptor positive; HER2: human epidermal growth factor receptor 2; TNBC: triple negative breast cancer; ECE: extracapsular extension; SLN: sentinel lymph nodes; ALND: axillary lymph node dissection; LN: lymph nodes * Lymphovascular invasion unknown in n=22 patients ** Patients with unknown extent of ECE (n=14) were not included in calculation £ P-value derived from univariable logistic regression with any additional positive lymph nodes at ALND as outcome ££ P-value derived from univariable logistic regression comparing 4 vs <4 additional positive lymph nodes at ALND Introduction: Controversy remains regarding which mastectomy patients with 1-2 positive sentinel nodes (+SLNs) nodes can forego completion axillary dissection (cALND). Our aim was to evaluate nodal treatment patterns and recurrence rates in patients treated with mastectomy with 1-2 +SLNs. Methods: From our prospective breast surgery database, we identified clinically node negative (cN0) breast cancer patients treated with mastectomy and SLN surgery from 2008-2018 with 1-2+SLNs. Patients treated with neoadjuvant therapy were excluded. Intraoperative frozen section analysis of SLNs were performed; patients with 1-2 +SLNs not detected by frozen section (FS-SLN) were compared to patients with +SLNs identified on frozen section (FS+SLN) using chi-square or Wilcoxon rank-sum tests. Recurrence rates were estimated using the Kaplan-Meier method. Results: Of 2295 cN0 mastectomy patients, 340 patients (15%) had 1-2 +SLNs. 109 patients (32%) were FS-SLN and 231 patients (68%) were FS+SLN. In FS+SLN cases, use of cALND was higher (97% vs 41%, p<0.001) as was the rate of additional nodal disease on cALND (31% vs 11%, p=0.006). FS+SLN group had higher pT category (63% vs 45% pT2-4, p=0.003), larger SLN metastases (median 5 vs 1.2 mm, p<0.001), were more likely to have lymphovascular invasion (34% vs 17%, p=0.001), and SLN extranodal extension (34% vs 6%, p<0.001). Across the 340 patients, PMRT was performed in 41% with SLN surgery only and 47% of patients with cALND (p=0.37) [ Table 1 ]. With 61 months median follow-up there were no axillary recurrences in the FS-SLN group, whereas in the FS+SLN there were 3 (all in the cALND group). In the FS+SLN group 5-year freedom from recurrence and regional recurrence were 85% and 94%, respectively, and were not significantly different from the FS-SLN group (p 0.15). Conclusion: Mastectomy patients with 1-2 FS+SLNs have higher nodal disease burden; majority underwent cALND and 52% received PMRT with good oncologic outcomes. A substantial proportion of FS-SLN patients successfully avoided both cALND and PMRT (36%) with low recurrence rates. FS analysis of SLN can help stratify risk and guide de-escalation of axillary management. Introduction: In 2016 as part of the Choosing Wisely campaign to reduce low value care, SSO recommended against routine use of sentinel lymph node biopsy (SLNB) in clinically node negative women 70 years old with hormone receptor positive breast cancer. The extent to which this recommendation has impacted clinical practice is unclear. Our objectives were to (1) evaluate trends in SLNB use in patients eligible for omission, (2) understand the role of Oncotype Dx in adjuvant therapy decisions, and (3) compare adjuvant therapy receipt in women with and without SLNB. Methods: We retrospectively analyzed an institutional database of women 70 years old with clinically node negative, ER/PR+, HER-invasive breast cancer who received surgical treatment between 2014 and 2018. Primary outcomes included utilization of SLNB, Oncotype Dx testing, and adjuvant therapy receipt in women with and without SLNB. Results: In total, 262 patients met eligibility criteria for SLNB omission. SLNB was performed in 66% of patients. Time series analysis showed a decrease in SLNB rates from 2014 to 2018, but the SSO recommendation did not have a statistically significant impact on this trend. Patients who received SLNB were significantly younger with higher rates of mastectomy and lobular histology compared to patients without SLNB. Patients who received SLNB were more likely to receive Oncotype testing and all adjuvant therapies. Fifteen percent of patients had a positive SLNB. Eighteen percent of SLNB-positive patients received chemotherapy without Oncotype Dx. An additional 42% of SLNB-positive patients were candidates for chemotherapy and had Oncotype Dx testing, but all patients had a score <30 and omitted chemotherapy. From 2016 to 2018, only 2 SLNB-positive patients received chemotherapy (1.3% of the overall population). Conclusion: Since 2014, there has been increasing de-implementation of SLNB in women 70 years old with early stage ER/ PR+, HER2-breast cancer. Overall, when combined with Oncotype Dx testing for risk-stratification, only 6% of patients who received a SLNB and 18% of patients with a positive SLNB received chemotherapy. Introduction: Omission of axillary dissection (AD) is supported for patients with clinically negative but sentinel node positive disease who meet Z11 criteria; and is being studied for those who are clinically node positive but downstaged with neoadjuvant chemotherapy (NAC). Data is lacking for patients with clinically positive nodes who undergo surgery upfront, so AD remains standard of care. Methods Patients who underwent AD for breast cancer between 2012-2018 at our institution were retrospectively identified. Only patients with positive axillary nodes confirmed preoperatively were included. Patients who had sentinel node biopsy, received NAC or had surgery for axillary recurrence were excluded. Clinical and pathological factors associated with nodal stage were evaluated in univariable ordered logistic regression analyses and subsequent multivariable analyses (MVA) if p<0.1. Results Of 117 patients who met inclusion criteria, 62.4% had a palpable node on exam. The majority of tumors were ER positive (98.1%), and 21.4% were of lobular histology. At AD, 37.6% had minimal nodal disease (N1). Lobular histology, tumor size, metastasis size and matted nodes were associated with >N1 disease in univariable analyses. On MVA (n=105), patients with tumor size >5cm (OR 3.8, p=0 .028) and matted nodes (OR 9.4, p=0 .009) had significantly greater odds of having >N1 while the effect for lobular vs. ductal histology failed to reach significance (p=0.058). In a separate MVA (n=27), increased metastasis size was associated with a greater likelihood of having >N1 disease (OR 1.3, 95%CI 1.06-1.69, p=0.014). Conclusion: In our study, over one third of breast cancer patients who presented with clinically positive nodes prior to surgery ultimately had minimal nodal disease (N1 stage) noted on AD. Additionally, palpable nodes on exam did not predict for higher nodal stage. As indications for nodal radiation are expanding to include patients with N1 disease, it is imperative that we explore whether some patients who present with clinically positive nodes are being overtreated with AD and could be treated with more targeted lymph node surgery.

The Technique of Sentinel Lymph Node Biopsy After Neoadjuvant Treatment in Breast Cancer is Crucial to Reduce False-negative Results: The ILINA Trial C. Siso, 2 A. Esgueva, 3 I. Miranda, 5 C. Sobrido, 4 J. Salazar, 5 M. Espinosa-Bravo, 2 I.T. Rubio. 1 * 1. Breast Surgical Oncology. Clinica Universidad de Navarra, Madrid, Spain; 2. Breast Surgical Oncology. Hospital Universitario Vall de Hebron, Barcelona, Spain; 3. Breast Surgical Oncology. Clinica Universidad de Navarra, Madrid, Spain; 4. Breast Imaging Unit. Clinica Universidad de Navarra, Madrid, Spain; 5. Breast Imaging Unit. Hospital Universitario Vall de Hebron, Barcelona, Spain. Background. Intraoperative ultrasound (IOUS) for excising clipped axillary nodes after neoadjuvant treatment (NAT) has shown to be an accurate technique. Identifying residual disease in the axilla after NAT is important to select patient for sparing an axillary node dissection (ALND). The aim of the study is to refine the IOUS technique to improve outcomes. Methods. Women with node positive breast cancer undergoing NAT are included in this prospective study. After NAT, IOUS-guided excision of the US visible clipped node along with SLNB (dual tracer) was performed. ALND was completed in the majority of patients as part of the initial study. Results. Of 148 patients with a clip placed in the positive node, 12 patients (8%) had a direct ALND performed for suspicious or biopsy proven axillary nodes before surgery. The clip was not found in 6 patients (4,05%). The clipped node was successfully removed by IOUS-guided excision in 130 patients (95%). Axillary pathologic complete response was achieved in 48 patients (36.9%). Of this, 12 patients (25%) did not undergo ALND. The SLN was the clipped node in 96 (74%) patients. Identification rates for SLN, and for clipped nodes were 94.6% and 98.5% respectively. False negative rate for SLN was 20.8%, for the clipped node was 9.7% and for using both was 4.2%. Mean number of SLNs retrieved was 2.8. Careful technique is needed when thoracic nerve block is used, as the anesthetic injected interfere IOUS visibility. After a median follow-up of 20 months (range 11-32 months), 6 patients (4.6%) developed metastasis. Of this, 2 (33.3%) had also regional recurrence. Both patients had an ALND and residual disease in the axilla. Three patients (4.1%) died from the disease. Five year disease free and overall survival was 82% and 95.6% respectively. Conclusions. IOUS-guided excision of the axillary clipped node in combination with SLN after NAT is a feasible, safe and accurate technique. ALND may be spared in 40% of patients with no axillary residual disease. Residual disease after NAT remains a prognostic factor, so refining technical issues are important to improve outcomes. Background: For breast cancer patients presenting with clinically node positive disease, targeted axillary dissection (TAD) after neoadjuvant therapy has been proposed, with pretreatment marking of nodal disease and subsequent removal of the clipped node along with SLNB. Using this approach, selected patients have been offered TAD with the potential to avoid morbidity of ALND. This study addressed the impact of nodal quantity and tumor biology on axillary pathologic complete response (AXpCR) in the implementation of TAD. Methods: A prospective IRB-approved database was reviewed to identify clinically node positive patients who underwent neoadjuvant therapy followed by ALND. From 2002-2017, 274 patients were identified with abnormal nodal findings on pretreatment axillary ultrasound. Thirty-eight patients were excluded for inability to delineate the number of abnormal nodes. The remaining 236 patients were stratified into groups by number of abnormal axillary nodes (1-2 vs 3 or more) and biologic subtype. AXpCR post-treatment was evaluated. Results: For the study population, there was no significant difference in AXpCR between patients with 1-2 versus 3 or more abnormal nodes on pretreatment imaging (n = 125 vs 111; 43% vs 39%, p=0.51). AXpCR did vary among the biologic subtypes. When comparing response to treatment in the 1-2 vs 3 or more nodal groups respectively, regardless of nodal burden, highest AXpCR was identified in the HER2-enriched subtype (86% vs 67%, p=0.28), followed by luminal B (47% vs 33%, p=0.26), basal (39% vs 35%, p=1.0), and luminal A (10% vs 0%, p=0.52). When luminal B patients were further stratified by HER2 status, HER2 positive patients had a significantly higher AXpCR compared with the HER2 negative patients. Conclusion: Nodal burden identified by pretreatment imaging was not indicative of AXpCR. Notably, biologic subtype did correlate with treatment response, with HER2 positive patients having the highest AXpCR. As patient selection and counseling for the implementation of TAD continues to evolve, consideration for tumor biology may be helpful to set expectations for definitive axillary management.

Introduction The Alliance 11202 trial currently is evaluating whether axillary lymph node dissection (ALND) can be avoided in breast cancer with axillary nodes (LN) that remain positive after neoadjuvant chemotherapy (NCT). We have previously demonstrated, in patients not undergoing NCT, that finding a single (rather than multiple) suspicious LN on axillary ultrasound (axUS) predicts N1 disease on ALND. We hypothesized, for patients without a nodal pathologic complete response (nPCR) to NCT, that a pre-NCT axUS showing a single suspicious node would predict N1 disease. Methods From a database of newly diagnosed breast cancer (BC), we selected patients presenting from 2011-2015 who underwent NCT, had a suspicious pre-NCT axUS with LN metastasis shown by needle biopsy, and had SLNB and/or ALND after NCT showing persistent LN metastasis. We used a scatter plot to examine the relationship between axUS findings and pathologic N category. Results Of 303 cases with invasive BC and axUS, 49 had persistent LN metastasis after NCT and formed the study cohort. Ten (20.4%) were triple negative, 9 (18.4%) were HER2 positive, and 30 (61.2%) were ER positive/HER2 negative. Forty-five (91.8%) had ALND. Median LN removed was 12 (interquartile range 7-17). Thirty (61.2%) had N1, 15 (30.6%) had N2 and 4 (8.16%) had N3 disease on surgical staging after NCT. Overall, scatter plot analysis did not demonstrate a relationship between number of suspicious nodes on axUS and number of positive LN on surgical staging ( Figure) . Within the triple negative subtype, a single suspicious node on axUS appeared to correspond to N1 disease, but the number of triple negative cases was very small. Conclusion For patients without a nPCR, pre-treatment axUS showing minimal disease burden does not predict N1 disease and cannot be used to identify patients who could avoid ALND. While awaiting Alliance 11202, ALND should be recommended for most patients with positive LN after NCT. Further study is needed to determine if there may be a relationship between axUS findings and nodal burden in the triple negative subtype. Background: Neoadjuvant chemotherapy (NAC) may downstage the axilla in breast cancer patients with positive lymph nodes and allows for sentinel lymph node biopsy (SLNB) in select patients with clinical N1 disease.

We examine axillary response rates after NAC in patients with clinical N2 disease to evaluate the potential for SLNB in this population. Methods: Breast cancer patients with clinical T0-4N2M0 (AJCC 7th Edition) disease who received NAC followed by a modified radical mastectomy or partial mastectomy + axillary lymph node dissection were selected from our institutional tumor registry (2009) (2010) (2011) (2012) (2013) (2014) (2015) (2016) . Data was collected for patient and tumor characteristics, NAC, treatment response, and surgical management. Clinical and pathologic axillary response rates were assessed. Results: Of 52 patients, median age was 55 years, all were female, and 40.4% were Hispanic ( Table 1) . Distribution of clinical stage at diagnosis was:T0-1N2 8 patients(15.4%), T2N2 13 patients(25.0%), T3N219 patients (36.5%), and T4N2 12 patients(23.1%). Approximated breast cancer subtype was hormone receptor (HR+) and human epidermal growth factor receptor-2(HER2) negative in 27 patients(51.9%), HER2+ in 17 patients(32.7%), and triple negative(TN) in 8 patients(15.4%). All patients received multiagent NAC and all patients with HER2+ tumors received HER2-targeted therapy(trastuzumab+/-pertuzumab). Clinical complete response was observed in 10 patients(19.2%), and 5 had a pathologic complete response (pCR) in the breast and axilla (2TN, 3HER2+). Clinical partial response was seen in 31 patients(59.6%), and 1 had a pCR in the axilla alone(HER2+). Of the 46 patients with residual nodal disease, the median number of positive nodes was 4 (1-3 in 15 patients, 4-9 in 23 patients, and >9 in 8 patients). Conclusion: The majority of breast cancer patients with clinical N2 disease at diagnosis did not have a pCR in the axilla and significant residual nodal burden was identified. In patients who have a clinical complete response, re-imaging and repeat biopsy of the axilla may be considered to assess eligibility for SLNB. Selection of patients for SLNB in this setting must be carefully evaluated. Background:Axillary lymph node dissection (ALND) can be avoided in node positive patients who receive neoadjuvant chemotherapy (NAC) if > 2 negative sentinel lymph nodes (SLNs) are retrieved. We sought to identify predictors of identification of > 2 SLNs and nodal pathological complete Abstracts: Poster Presentations S69 response (pCR). Methods:Consecutive patients with cT1-3, biopsy-proven N1 breast cancer treated with NAC between 11/2013 and 7/2019 were identified from a prospective database. All cN0 patients after NAC had sentinel lymph node biopsy (SLNB) with dual mapping using isotope and blue dye. ALND was performed for any positive SLNs and if < 3 SLNs were retrieved. Groups were compared using Fischer's exact test and Wilcoxon rank-sum test. Multivariable logistic regressions (MVR) were used to calculate the odds ratio (OR) of finding >2 SLNs and achieving nodal pCR. Results:630 N1 patients were eligible for axillary downstaging with NAC; 573 (91%) converted to cN0 and had SLNB. 531 patients (93%) had >2 SLNs identified. cT3 stage, lymphovascular invasion (LVI) and higher BMI were significantly associated with failure to identify >2 SLNs (Table) . In MVR, only LVI (OR 0.46, 95% CI 0.24 -0.87, p = 0.02) and increasing BMI (OR 0.77, 95% CI 0.62 -0.96 per 5 unit increase, p = 0.02) were significantly associated with failure to identify >2 SLNs. Nodal pCR was achieved in 237 (45%) patients with >2 SLNs retrieved and was significantly associated with histology, receptor status, LVI and grade (Table) 1. Division of Surgical Oncology, Jewish General Hospital Segal Cancer Center, McGill University Medical School, Montreal, QC, Canada; 2. Division of Radiation Oncology, Jewish General Hospital Segal Cancer Center, McGill University Medical School, Montreal, QC, Canada. Background: The oncologic safety of sentinel lymph node biopsy (SLNB) alone in clinically node-positive (cN1-2) patients who convert to pathologic node-negative (ypN0) following neoadjuvant chemotherapy (NAC) is not well established. Methods: We retrospectively identified 244 consecutive patients with a diagnosis of cT1-3cN0-2 breast cancer who underwent NAC followed by SLNB at our institution between 2013 and 2018. The Kaplan Meier method was used to compare locoregional recurrence rates following SLNB alone in women with ypN0 disease according to clinical nodal status at presentation. Results: Among 244 patients who underwent NAC followed by surgery with SLNB, 111 (45.5%) were cN0 at presentation while 133 (54.5%) had biopsy-proven cN1-2 disease and converted to cN0 following treatment. In patients presenting with cN0 disease, 101 (91.0%) were ypN0 on SLNB pathology, compared to only 72 (54.1%) of patients who presented as cN1/2 and converted to cN0 following NAC (p<0.001). Regional nodal irradiation was administered to 72.3% of cN1-2/ypN0 patients compared to 4% of cN0/ypN0 patients (p<0.001). In cN0/ypN0 patients, five-year local and regional recurrence rates after SLNB alone were 5.0% (95% CI, 1.9-13.0) and 1.1% (95% CI, 0. 15-7.24 ). In cN1-2/ypN0 patients, the five-year local and regional recurrence rates were 11.5% (95% CI, 4.1-30.1) and 0%, with no axillary recurrences diagnosed. Conclusion: Sentinel lymph node biopsy following NAC is associated with low and acceptable short-term axillary recurrence rates. Additional follow-up from prospective clinical trials are needed to confirm long-term oncologic safety and define optimal local therapy recommendations. BACKGROUND The aim of this study is to compare the safe resection margin for breast cancer patients between using us-guided indocyanine green fluorescence (ICG-F) marking and sono-guided skin marking in breast conserving surgery (BCS). METHODS From April 2016 to March 2019, we enrolled 114 patients who underwent BCS using US-guided ICG-F marking prospectively and we compared these result with the 300 patients who underwent BCS using US-guided skin marking from the same surgeons from January 2012 to December 2016. The study compared the identification rate, the status of frozen resection margin, the type of intra-operative positive margin, with additional resection during operation, final pathologic margin status and re-operation rate. RESULTS The ICG-F identification rate was 100% (114/114). The mean time of margin approach in ICG -F using group was about 13 minutes. The positive rate of frozen resection margin was 2.6% using ICG-F and 12% using sono-guided skin marking (p=0.004). The rate of intraoperative additional resection was significantly lower in the ICG-F using group than sono-guided skin marking group (8.8% vs. 23.3%, p=0.001). In both groups, ductal carcinoma in situ was the most common in intraoperative margin positive type (ICG-F using group 1.8% and sono-guided skin marking group 7%, p=0.018). The positive rate of permanent resection margin was 2.6% in the ICG-F using group and 8.7% using sono-guided skin marking group (p=0.013). The rate of re-operation was 4.4% in the ICG-F using group and 4% in sono-guided group (p=0.789%). After operation using ICG-F, no complications including skin necrosis, skin discoloration, and allergic reaction occurred. CONCLUSION This study is the prospective trial using ICG-F for surgical margin in BCS. Using ICG-F could be safe method for BCS. This method could be a feasible and may be approach that even beginners can easily obtain safe margin in breast conserving surgery.

Breast-conserving Surgery B.L. Smith, 1 * C.R. Lanahan, 1 B. Kelly, 1 M.C. Specht, 1 C.L. Brown, 1 J.E. Korotkin, 1 K.S. Hughes, 1 J.M. Ferrer, 2 E.F. Brachtel, 1 T.L. Rice-Stitt, 1 M.A. Gadd. 1 1. Surgical Oncology, Massachusetts General Hospital, Boston, MA; 2. Lumicell, Inc., Newton, MA. Introduction A safe lumpectomy for breast cancer requires microscopically clear margins. However, real-time margin assessment options are limited, and 20-40% of lumpectomies have positive margins that need re-excision. New tools for intraoperative margin assessment are needed. The LUM Imaging System previously showed 84% sensitivity and 73% specificity for tumor detection during breast lumpectomy surgery. We tested the properties of the LUM System further. Methods Data was obtained from an IRB-approved, prospective, non-randomized study in lumpectomies for breast cancer. The LUM Imaging System (Lumicell, Newton, MA) uses LUM015, an injectable protease-activated fluorescent imaging agent detected in the surgical site by the LUM optical head. Fluorescent cavity images were collected in real-time and analyzed digitally. Results 55 patients, median age 60 years (range 44-79), received IV LUM015 (0.5 or 1.0 mg/kg) and had a lumpectomy for 31 (56.4%) invasive ductal carcinomas (IDC) DCIS, 5 (9.1%) invasive lobular carcinomas (ILC), 5 (9.1%) invasive cancers with ductal and lobular features and 14 (25.4%) DCIS. 29% of patients were pre-or perimenopausal; 62% of breasts were heterogeneously or extremely dense. All tumor types were distinguished from normal tissue, with mean tumor:normal (T:N) signal ratios of 3.81 for IDC DCIS, 3.98 for ILC, and 5.69 for DCIS (Table) . T:N ratios were 4.45 in non-dense and 4.00 in dense breasts (p=.31) and were 3.52 in premenopausal and 4.59 in postmenopausal women (p=0.10). Histopathology, IHC, and FISH testing were not affected. Falsely positive readings were more likely when tumor was present <2 mm from the adjacent specimen margin. LUM015 signal was stable in vivo at least 6.5 hours post injection, and ex vivo at least 4 hours post excision. Conclusion Intraoperative use of the LUM Imaging System detected all breast cancer subtypes with robust performance independent of menopausal status and breast density. There was no impact on surgical workflow or pathology evaluation. We found a potential gradient of LUM015 activation around tumor deposits which may help achieve wider negative margins.

Tumor-to-normal fluorescence ratios by tumor histological types, mammographic breast density, and patient menopausal status a. calculated with a one-way analysis of variance ANOVA b. calculated with a 95% confidence interval, single-tailed Student's t-Test Introduction Historically, more than 1/3 of women with DCIS treated with breast-conserving surgery (BCS) had either re-excision or conversion to mastectomy. The SSO-ASTRO guidelines advise 2mm margins for women with DCIS having BCS and whole-breast radiation (WBRT). Here we examine guideline impact on additional surgery and factors associated with re-excision. Methods Women treated with BCS for pure DCIS from 8/2015-12/2017 were identified. Guidelines were adopted on 9/1/2016, and all women had separately submitted cavity-shave margins. Clinicopathologic characteristics, margin status, and rates of additional surgery were examined. Results Among 650 women with DCIS who attempted BCS, 50 (8%) converted to mastectomy. Of 600 who had BCS as final surgery, 336 (56%) received WBRT and comprised our study group. 128 (38%) were treated pre-guideline and 208 (62%) post-guideline. Characteristics and margin status were similar between groups (Table) . Re-excisions declined from 38% to 29% (p=0.09) after guideline adoption, with 91% having only 1 re-excision (86% pre-vs 95% post-guideline, p=0.11). Re-excision despite 2mm margins was uncommon, and similar between eras (6% pre-vs 5% post-guideline). Re-excision for negative but <2mm margins was also similar (65% pre-vs 63% postguideline), respectively. On multivariate analysis, younger age (OR 0.96, 95% CI 0.94-0.98, p<0.01) and larger DCIS size (OR 1.42, 95% CI 1.2-1.7, p<0.01) were predictive of re-excision; guideline era was not. Women who converted to mastectomy were younger (median 49 vs 58yrs, p<0.01), had larger DCIS size (median 1.8 vs 0.9cm, p<0.01), high grade (40% vs 24%, p=0.03), and preoperative MRI (46% vs 29%, p=0.02). Younger age (OR 0.93, 95% CI 0.9-0.97, p<0.01) and larger size (OR 1.64, 95% CI 1.3-2.1, p<0.01) were predictive of conversion to mastectomy, but residual tumor burden was low (median size 0cm, range 0-1.8cm). Conclusion The SSO-ASTRO guidelines did not significantly change re-excision rates for DCIS in our practice, likely since re-excision for margins 2mm was uncommon prior to guideline adoption. Younger age and larger DCIS size were associated with additional surgery. Table: Clinicopathologic characteristics and margin status of patients treated with breast-conserving surgery and whole-breast radiotherapy, stratified by guideline era *Represents n = 235 cases; n = 73 had slides counted and n = 28 had DCIS present with no measurement of extent. Abbreviations: US, ultrasound; MRI, magnetic resonance imaging; ER, estrogen receptor Utilization of Selective Shave Margins in Breast Conserving Surgery After Neoadjuvant Chemotherapy INTRODUCTION: Excision of selective shave margins (SM), defined as representative sampling of tissue from the margins of a lumpectomy cavity, has been associated with reduced rates of positive margins relative to conventional lumpectomy (CL). There are limited data on the utility of SM in breast conserving surgery (BCS) after neoadjuvant chemotherapy (NAC). METHODS: From a prospectively maintained database, 125 patients with stage I-III breast cancer who were treated with NAC followed by BCS from 2008-2017 were identified. Patients were divided into two groups based on technique of margin excision: SM and CL. Method of margin determination was according to the standard practice of the surgeon. Clinicopathologic characteristics, margin status, response to therapy, reexcision, and local recurrence rates were evaluated. RESULTS: Median age was 53 years (range 28 to 87). Ninety-three patients had SM and 32 had CL. The two groups did not differ in age, stage, histology, estrogen receptor and HER2 status. Clinical response to NAC was seen in 91% in the SM group and 84% in the CL group (P=0.43). Pathologic complete response in the breast occurred in 41 (44%) patients in the SM group, but only in 7 (22%) patients in the CL group (P=0.03). Extent of residual disease was similar between groups (SM:1.15 cm and CL:1.78 cm, P=0.07). There was no difference between groups with respect to positive margin or re-excision rates (SM:14% and CL:16%, P=1.0, and SM:15% and CL:16%, P=0.94, respectively). Additionally, volume of lumpectomy specimen was smaller in the SM group (82.6 cm 3 ) than the CL group (146.2 cm 3 ), P=0.02. In-breast recurrence occurred in 3 (3%) in the SM group and 2 (6%) in the CL group (P=0.45). CONCLUSIONS: Use of SM in the neoadjuvant setting was associated with similar rates of positive margins and re-excision when compared to CL. Both methods of margin excision were associated with very low local recurrence rates.

Introduction The use of neoadjuvant chemotherapy (NAC) has enabled more patients to be eligible for breast-conservation surgery (BCS). Achieving negative lumpectomy margins, however, may be challenging after NAC due to changes in tissue density and the potential for residual carcinoma to be scattered in the tumor bed. Data regarding patients undergoing BCS after NAC has shown variable re-excision rates. MarginProbe (Dune Medical Devices Ltd, Israel) has been shown to identify positive resection margins intraoperatively and reduce the number of re-excisions in primary BCS, but has not been previously studied in NAC+BCS cases. The purpose of our study was to investigate the clinicopathologic characteristics, including margin status, and re-excision rates in patients who had NAC+BCS with and without the use of MarginProbe at our institution. Methods The Institutional Breast Cancer Database was queried for all patients who received NAC and had subsequent BCS from 2010-2019. Variables of interest included demographics, tumor characteristics, pathologic complete response (pCR), MarginProbe use, treatment and outcomes. Statistical methods included Chi-Square and Fisher's Exact tests. Results A total of 214 patients had NAC in our study population, and 61 (28.5%) of those patients had NAC+BCS. The median age was 53.5 years. A total of 19 (31.1%) patients had pCR. Of the remaining 42 patients, 9 (21%) had close or positive margins that required re-excision. Re-excision was associated with a larger residual tumor size (p=0.025), and ER-positive disease before NAC (p=0.041). Breast density and the presence of palpable disease did not differ between the groups. MarginProbe use was associated with a lower re-excision rate in patients with NAC+BCS (6% vs. 31%, respectively). Conclusion The use of NAC may enable more patients to undergo BCS. Patients with larger residual tumor burden and ER-positive disease were at an increased risk for inadequate margins at the time of surgery. The use of Margin-Probe was associated with a lower re-excision rate. Techniques to reduce the need for re-excision will support the use of BCS after NAC. Background: In 2015, a randomized controlled trial on shave margins for partial mastectomy showed significantly less positive final margins and re-excision rates when shave margins were taken, leading to shave margins being adopted into practice. We sought to investigate the prognostic factors for residual disease in shave margins. Methods: Patients with invasive breast carcinoma who had global shave margins excised during lumpectomy were abstracted from a retrospective database from 2015-2018. We defined mammographically occult disease (MOD) as residual disease in the shave margin when the lumpectomy margin was negative or residual disease in the shave margin when the corresponding lumpectomy margin was negative. We identified the frequency of MOD in this cohort and conducted logistic regression analyses on patients with invasive carcinoma to identify factors associated with MOD. Variables included age, T stage, grade, positive lymph nodes, hormone receptors, histology, positive lumpectomy margin, MRI, and tomosynthesis. Regression analysis was used to evaluate the association of recurrence with MOD. Results: 130 patients met inclusion criteria with shave margins taken. 95 patients (73.1%) received endocrine therapy and 87 patients (66.9%) received radiation therapy. The median age was 64 (range 32-87). MOD was found in shave margins in 31 patients (23.8%) ( Table 1 ). In univariate analyses, a positive lumpectomy margin was predictive of MOD with an odds ratio (OR) of 3.08 (p = 0.008), and invasive lobular carcinoma was predictive with an OR of 3.83 (p = 0.021). In multivariate analysis, a positive lumpectomy margin remained a significant predictor of MOD, p = 0.025. Invasive lobular carcinoma was not significant in multivariable analysis, but there was a trend, p = 0.076. With a median follow-up of 28 months, the overall recurrence rate was 6.15%. The presence or abscence of MOD was not associated with local recurrence (p=0.26). Conclusion: A positive margin on the primary lumpectomy specimen was a predictor of residual disease in a shave margin. The presence or absence of additional disease in the shave margin did not affect the local recurrence rate. Table 1 a. Invasive ductal carcinoma; b. invasive lobular carcinoma; c. ductal carcinoma in situ * Some patients with both DCIS and invasive carcinoma in shave margin Introduction Randomized trials have demonstrated that survival in women age 70y with stage 1 hormone-positive breast cancer is not changed by axillary staging (AS), in part due to competing mortality risk. The 21-gene recurrence score (RS) is commonly used to predict adjuvant chemotherapy benefit, but its impact in women >70y is unclear. We evaluate how AS and RS testing influenced decisions for chemotherapy in these patients. Methods Patients age 70y with cT1N0/ER+/HER2-tumors undergoing breastconserving therapy (BCT) or mastectomy from 1/2011-12/2013 were identified from a prospectively maintained database. Clinicopathologic features were evaluated. 10-year estimated mortality reported as likelihood of death was calculated using the Suemoto index (SI). The primary outcome of interest was chemotherapy receipt. Results 361 patients were identified; 333 (92%) had BCT, 28 (8%) had mastectomy. Median age was 76y (70-101), median tumor size was 1cm (0.2-1.9), mean RS was 15.1 (0-43), and mean SI was 43. 320 women (88.6%) underwent AS; 9 (2.8%) had nodal metastases. The median SI of those having/not having AS was 41 and 74, respectively (p<0.01). Overall, 164 (45.4%) women had RS performed, with no difference in use based on nodal status (p=0.9). There were no associations between patient age or SI and obtaining a RS. 18/164(11%) women had high RS (>25); 12(67%) of these women received chemotherapy. 146 women had RS score 25. In total, 25 patients (6.9%) received chemotherapy; 12 with a high RS, 8 with a RS 25 and 5 who did not have a RS. The percent of women undergoing chemotherapy after AS was 7.8%. Table 1 summarizes use of AS, RS and chemotherapy by age. Conclusion AS use in women age 70y with clinical Stage 1 breast cancer is still frequent, but nodal metastases were rare and had little influence on the use of chemotherapy. A high RS is uncommon in this patient population and not all women with high RS received chemotherapy. Careful consideration should be used prior to performing AS and obtaining RS in elderly patients with early-stage breast cancer to increase the value of care delivered to these women. Background: Concordance between Oncotype DX recurrence scores (RS) in synchronous ipsilateral invasive breast carcinomas (BCs) has been reported at 87% in tumors with similar histology. The impact of age on concordance is unknown. We examined RS concordance in synchronous ipsilateral BCs of any histology in women ages 50 and >50 years, for whom risk stratification varies with respect to chemotherapy benefit according to TAILORx data. Methods: From 2009-2018, women with synchronous ipsilateral invasive estrogenreceptor positive, lymph node negative BCs with 2 RS were retrospectively identified from an institutional database. Women were stratified by age 50 and >50 years and concordance between RS was assessed. For age >50, risk groups were categorized as low (RS 25) and high (RS>25). For age 50: low (RS<15), intermediate and high (RS >25) risk groups were categorized. Results: Overall, 120 patients with multifocal or multicentric synchronous ipsilateral invasive BCs were identified: 96 (80%) were >50 years old and 24 (20%) were 50 years old. Tumor characteristics are shown in the Table. Of women >50 years old, 90/96 (93.8%) had recurrence scores in the same risk category. Concordance was 97.1% (66/68) in tumors with similar histology compared to 84.0% (21/25) in synchronous tumors of both ductal and lobular histologies (excluding 3 tumors of other mixed histology). In women 50 years old, risk category concordance fell to 58.3% (14/24). Among the 10 women 50 years old with discordant RS, the higher score classified 29.2% (7/24) and 12.5% (3/24) in the intermediate and high categories, respectively. Conclusion: Synchronous ipsilateral invasive BCs are highly concordant in women >50 when both tumors are morphologically similar. These findings suggest that RS testing of one tumor focus is sufficiently prognostic and predictive, although our sample is too small for meaningful statistical comparison. Discordance was more common in women 50 with morphologically similar tumors and in both age groups with morphologically distinct tumors, suggesting that testing of individual tumor foci is still warranted. Introduction: Oncotype DX is used to predict the risk of recurrence in patients with hormone receptor positive, Her2-neu negative, invasive breast cancer (IBC), which constitutes 73% of all IBC. Recurrence scores (RS) are reported as a low (0-25) or high (26-100) risk. The low risk group receives no benefit from adjuvant chemotherapy, while the high risk group receives a substantial benefit. Ki67 is a proliferation marker and one of the 21 genes used in calculation of RS. At the time of diagnosis, 90% of IBC is T1 or T2, and 20% have a Ki67 10% or less. We investigated the relationship favorable Ki67 and low risk RS in an effort to more judiciously predict benefit from chemotherapy and evaluate its impact on cost. Methods: A retrospective review of all patients with available RS was performed. Patients were omitted if tumors were greater than 5 cm or if they had node-positive disease. Data was collected on these patients including Ki67 in biopsy specimen, final tumor grade, maximum tumor diameter, and RS results. If given a range, the Ki67 value was assigned to the highest number in the range. These characteristics were used to predict patients who will have a low risk RS. Potential financial impact and cost saving were then calculated. A Fisher's exact test was used to determine if favorable Ki67 is associated with low risk RS. Results: RS was available for 251 patients who met the above criteria. Of those, 120 patients had an available Ki67 value in a biopsy report. The data is shown in Table 1 . All patients with favorable Ki67, defined as 10% or less, had low risk RS. The p-value is 0.10. With 230,000 new diagnoses of IBC annually and a test cost of $3,873, potential national cost saving could be as high as $117,049,806. Conclusion: RS is useful in assessing potential need for adjuvant systemic therapies. Most new IBC diagnoses qualify for testing. The tests appears to be low yield in tumors with favorable Ki67. Although statistical significance was not achieved, trends in benefit as related to cost has become evident. As stewards of evidence-based science, we need to be conscientious of value and cost. Current guideline recommendations for genomic testing may be too broad. Background: Recently, there has been a trend toward de-escalation of breast surgery in favor of less aggressive surgical techniques. Z0011 trial has decreased the need for axillary dissection in many cases and thereby decreased the rates of lymphedema. Because of this growing trend, the goal of this study was to determine whether a low Oncotype DX score (less than or equal to 11) and negative axillary lymph nodes on MRI would correlate to negative lymph nodes on sentinel lymph node biopsy, therefore negating the need to perform sentinel lymph node biopsy in these patients. MRI was selected due to its high sensitivity to detect disease and Oncotype Dx was selected as a way to account for tumor biology. Methods: Through a retrospective IRB approved study, our tumor registry was queried for patients from 2011-2017 that had an MRI in the preoperative setting, and who later got an Oncotype DX. It was found that 248 patients had an Oncotype of less than or equal to 11. Results: Of the 248 patients, 219 (91%) had sentinel lymph node biopsy. Of these patients, 34 (13.7%) had an Oncotype of 11 or less and at least one positive lymph node metastases on sentinel lymph node biopsy. Of these 34 patients, 28 (82.4%) had MRIs that read axillary nodes as negative for suspicious findings. Six of the 34 (17.6%) had a suspicious MRI indicating lymph node involvement. One of the 34 (2.9%) had >3 lymph nodes positive on surgical resection. Conclusions: Sentinel lymph node biopsy remains the standard of care in staging the axilla. The use of MRI plus Oncotype Dx does not accurately predict the status of the axilla. Our findings did show that 17.6% of suspicious lymph nodes on MRI were positive for metastatic disease; this correlates with what we know about MRI accuracy of detection of breast lesions. The addition of Oncotype DX did not seem to impact propensity for lymphatic involvement. Our study shows that, even in low risk patients (<11score), the assay does not allow omission of sentinel lymph node biopsy, and this should continue as the standard of care for the immediate future. Intro: Nodal positivity has been an indication for chemotherapy, but the 21 gene recurrence score (Oncotype DX; 21GRS) has refined its use. Given waning indications for axillary dissection, this study was performed to determine whether sentinel lymph node biopsy (SLNB) is superfluous when the 21GRS is used to determine candidacy for chemotherapy. Methods: The National Cancer Database was queried for patients having ER positive, HER2 negative, stage I-IIIa breast cancer who underwent a 21GRS and SLNB in 2012-2016. Based on the years of study, 21GRSs were defined as low (<18) and high ( 31). Findings were concordant if they had: a negative SLNB with a low 21GRS and a positive SLNB with a high 21GRS. A subset of Z0011-eligible patients (cT1/2, 1-2 positive sentinel nodes, and breast conservation with radiotherapy) was also analyzed. Results: 121,877 patients were included. SLNBs were negative in 100,216 (82.23%) patients, while 21GRSs were low risk in 65,582 (53.81%) patients. Overall, 21GRS was concordant with SLNB in 56,901 (66.02%; Table 1 ) patients. Predictors of concordance were Black race, diagnosis year, tumor grade, lymph node status and days from diagnosis to first surgery. The 21GRS was far more predictive of adjuvant chemotherapy administration than SLNB (OR 20.4 vs 1.8). The 21GRS was predictive of adjuvant chemotherapy administration in the Z0011 subset, while the sentinel node status was not (OR 28.1 vs 0.7). Additional predictors of chemotherapy administration included grade, lympho-vascular invasion, size, and time to first surgical procedure. In those discordant, with a positive SLNB and low 21GRS, chemotherapy declined from 23.05% in 2012 to 14.42% in 2016. In Z0011 candidates discordant with a positive SLNB and low 21GRS, chemotherapy declined from 18.30% in 2012 to 9.31% in 2016. Conclusions: In breast cancer patients undergoing the 21GRS, SLNB is discordant in 1/3 of cases. The addition of SLNB changed adjuvant chemotherapy administration in the minority of patients, while in Z0011 candidates this is even less likely. With waning axillary dissection indications and use of 21GRS in both N0 and N1 disease, the value of SLNB continues to decline over time. INTRODUCTION: The revised 8th edition of the AJCC staging for breast was developed to be a better outcome predictor when applied to individual patients. While genomic profiling could be included to derive the pathologic prognostic stage, the staging system was developed independent of it. Genomic profiling is frequently used as a prognostic tool and to determine treatment options. Our analysis sought to determine if the pathologic prognostic stage correlated better with genomic profiling when compared to the anatomic stage. METHODS: Patients that had genomic profiling-Oncotype Dx and Mammaprint-in 2013 to 2017 were included and were re-staged according to the new staging system. Oncotype Dx scores were categorized according to the findings from the TAILORx trial-high risk= >25. RESULTS: Among 321 patients, the majority had an Oncotype Dx test, were white and between 50 and 74 in age (73%, 76% and 80% respectively). Most patients (208; 65%) were a stage IA. All anatomic stage IA patients remained at IA on the pathologic prognostic stage. Most of the remaining anatomic stages were down-staged on the pathologic prognostic stage (96/113; 85%). In the entire cohort of patients, 249 (77.6%) had a low-risk score. Among all patients who had a high-risk score, 29% (21/72) were down-staged on the pathologic prognostic stage. When those with stage IA were excluded (all concordant on anatomic vs. pathologic prognostic stage), 88% were discordant and, hence, down-staged in the low-risk score category compared with 75% in the high-risk score category (table). No difference was found when the results were further stratified by race or age-50 vs. >50. CONCLUSION: The pathologic prognostic stage derived independent of genomic profiling did not correlate entirely with the risk score and was found to down-stage a significant number of patients with a high genomic profile risk score when the anatomic stage was higher than IA.

Concordance between anatomic and prognostic stage per risk score excluding stage IA P=0.089 Background: The alarming rise in the incidence of gastric (GC) and colorectal (CRC) adenocarcinomas in young adults (YA) over the past three decades is not well understood. How YA populations differ from older patients with the same gastrointestinal malignancies warrants investigation. Methods: We retrospectively analyzed the California Cancer Registry and the Office of Statewide Health Planning and Development data for all GC and CRC cases from 2000 to 2012. Pearson's Chi-square analysis was used to analyze differences in demographic, clinical and histopathologic features and log-rank test to compare survival between young ( 40 years old) and older adults (40-90 years old) with GC or CRC. Results: Of the GC (n=19,368) and CRC (n=117,415) patients included in the study, YA accounted for 4.5% (n=883) of GC and 3.2% (n=3723) of CRC. Hispanic ethnicity was more common in Background: Despite the increasing incidence of gastroesophageal junction adenocarcinoma (GEJA), the optimal treatment strategy for the disease remains unknown. The objective of this study was to describe treatment patterns for GEJA in the United States. Methods: The National Cancer Database was searched to identify all patients who underwent resection of the lower esophagus, abdominal esophagus, and/or gastric cardia for GEJA between 2006 and 2016. Patients were grouped by clinical disease stage: early localized (L;T1-2/N0), locally advanced (LA;T3-4/N0), regional (R; T1-2/N+), or regionally advanced (RA;T3-4/N+), and compared by treatment strategies. Results: The search identified 28,852 GEJA patients. The dominant age range was 60-69 years (39%). Most patients were men (85%), and most were white (92%). Most L patients (69%) underwent upfront surgery, whereas most LA, R, and RA patients received neoadjuvant therapy (NAT; 86%, 80%, and 90%, respectively). The use of NAT has constantly increased over the study period in all clinical stages (p<0.001, Figure 1 ). Among patients who received NAT, 85% received chemoradiotherapy, 2% chemotherapy alone, and 13% radiation alone. The higher proportion of neoadjuvant chemoradiotherapy as the choice modality was similar across all clinical stages. The use of adjuvant therapy was relatively rare across all groups (15-20%) . In the LA, R, and RA groups, the survival duration of patients who received NAT were significantly longer than those patients who underwent upfront surgery (p<0.001). Conclusion: Use of NAT has constantly increased over the past decade. With the exception of patients with localized, node-negative disease, most GEJA patients receive neoadjuvant chemoradiotherapy despite lack of prospective trials reporting survival benefit over chemotherapy alone. Background: Esophageal cancer is increasing in incidence worldwide. It is estimated in 2019 there will be 17,650 new cases of esophageal cancer and 16,080 deaths from disease. There has been an increase in younger patients diagnosed with esophageal cancer. We sought to evaluate the outcomes in younger patients diagnosed with esophageal cancer. Methods: Utilizing the National Cancer Database we identified patients with esophageal cancer. We then stratified by age <50, 51-60, 61-70, and >70 years. Baseline univariate comparisons were made for continuous variables using both the Mann-Whitney U and Kruskal Wallis tests. Pearson's Chi-square test was used to compare categorical variables. Unadjusted survival analyses were performed using the Kaplan-Meier method. All statistical tests were two-sided and p <0.05 was significant. Propensity score matched analysis was performed and only exact matches were allowed. Results: We identified 20,324 patients (<50, 2157), (51-60, 5387), (61-70, 7853), and (>70, 4927). T-stages and N stages were higher in the younger age groups p<0.001 and p<0.001 respectively. Median lymph nodes positive were highest in the <50 group (2.2 vs 1.8 vs 1.6 vs 1.7) p<0.001. Additionally, tumor size was largest in this age cohort (3.7 vs 3.5 vs 3.5 vs 3.2) p=0.002. Neoadjuvant therapy (NT) was administered in 69.4% of patients in the <50, 68.5% (51-60), 65.5% (61-70), and 49.5% >70 patients, p<0.001. The <50 age group was however more likely to receive adjuvant therapy (AT) (22.9% vs 20.5% vs 16.9% vs 13.4%) p<0.001. Median and overall survival was 49.8 mo and 45% (<50), 45.4mo and 43% (51-60), 45.4mo and 43% (61-70), and 35.8mo and 39% >70, p<0.001. After propensity score matching, multivariate analysis we found that age <50, male gender, GEJ tumor location, grade, Charleson Deyo score, T-stage N stage, margin, facility volume, NT and AT were predictors of survival. Conclusions: Although younger patients present with larger tumors, higher T-stages, and N stages, they are more likely to receive neoadjuvant and adjuvant therapies. It is these therapies which are most likely contributing to improved survival compared to their older counterparts. Background: Patients undergoing esophagectomy frequently experience malnutrition, which in combination with the catabolic effects of surgery can result in loss of muscle mass and function. Safe swallowing requires the preservation of muscle mass. Modified barium swallow (MBS) enables assessment of postoperative swallowing impairments. We assessed the incidence and risk factors of swallowing dysfunction post-esophagectomy. Methods: Patients with a MBS post-esophagectomy were identified between January 2015-June 2019 at Levine Cancer Institute at Carolinas Medical Center. Swallowing was evaluated with the Penetration Aspiration Scale. Muscle loss was evaluated with pre-operative hand-grip strength (HGS) and skeletal muscle index (SMI) and skeletal muscle density (SMD) from axial CT images. Uni-and multivariable GLM analyses were performed. Results: 91 patients (79 men, 12 women) underwent esophagectomy with an average age of 64.0 + 10.1. Pre-operative HGS, SMI, and SMD all decreased with age. Significant differences existed between sexes in HGS, SMI, and SMD, so the cohort was stratified by sex for analysis. Univariate analysis of male patients revealed older age, lower body mass index (BMI), smoking history, prior feeding tube, and lower pre-operative HGS and SMI were associated with aspiration or penetration on MBS. Among women, no factors analyzed were significantly associated with swallowing dysfunction. Conclusions: Swallowing dysfunction after esophagectomy is correlated with increased age and lower BMI. The role of muscle loss in the risk of aspiration after esophagectomy is not clear. Further research is needed to determine the relationship between these factors with the goal of enabling preoperative physiologic optimization and patient selection. Objective: Ontario regionalized thoracic surgery to designated centers to provide high-volume care for patients undergoing esophageal cancer resection. The objective of this study was to assess variation in treatment patterns and outcomes across thoracic centers, and to compare their performance to non-thoracic centers. Methods: A retrospective, population-based cohort study (2002) (2003) (2004) (2005) (2006) (2007) (2008) (2009) (2010) (2011) (2012) (2013) (2014) was conducted in Ontario, Canada (population 13.6 million), examining adults with resected esophageal cancer. Case mix, use of neoadjuvant therapy, surgical outcomes (lymph node yield and margin rates) and survival were described across thoracic centers. Multivariable regression was used to estimate the effect of having surgery at a regionalized thoracic center on perioperative (in-hospital & 90-day post-discharge) mortality and long-term survival, adjusting for case mix. Results: Of 3,880 patients meeting study criteria, 2,213 had pathology data available and were included in the analysis. Average age was 64 years, 85.7% had adenocarcinoma, 50.2% were pT3, and 38.4% were pN0. Most (82.6%) had surgery at one of 15 thoracic centers. Across thoracic centers, rates of neoadjuvant therapy varied from 16.4-81.6%, positive margin rates varied from 8.2-29.6%, median lymph node harvest varied from 7-20 nodes, perioperative mortality varied from 2.6-20.5%, and 2-year survival varied from 48-80%. There was a trend toward reduced perioperative mortality, but no difference in long-term survival, with having surgery at a thoracic center. Conclusions: Even at designated thoracic centers, there is significant variability in treatment patterns, surgical outcomes, and survival. Looking beyond center volume, and translating best practices from high-performing hospitals to other hospitals, may improve patient outcomes. Background: Young women with breast cancer (BC) have a higher risk for locoregional recurrence (LRR) regardless of surgical treatment choice. We sought to assess LRR by molecular subtype, accounting for treatment including local therapy, among a multicenter cohort of young women. Methods: Between 2006-2016, the Young Women's Breast Cancer Study enrolled 1,302 women diagnosed with BC at age 40. Treatment information and incident LRR (ipsilateral breast/chest or lymph node recurrence) and contralateral breast cancer (CBC) were self-reported on study surveys and confirmed by chart review; molecular subtype was determined by central pathology review. Cumulative incidence of LRR was calculated using the Kaplan-Meier method; hazard ratios were estimated by Cox proportional hazards regression. Results: Among 1231 women with Stage 0-III BC, median follow-up was 85 (range: 7 -156) mos. Among patients with local therapy and subtype data available (n=1170), 377 (32%) were Luminal A, 497 (42%) Luminal B, 96 (8%) HER2+, and 200 (17%) triple negative (TN). 356 (30%) of women had breast conserving surgery (BCS), 304 (37%) unilateral mastectomy, 510 (44%) bilateral mastectomy. Of women who had mastectomy, 52% had radiation. Overall cumulative incidence of LRR was 6% (95% CI 5%-8%); 2% developed CBC. In unadjusted analyses stratified by local therapy strategy (Table) , LRR rates differed by subtype only for the mastectomy with radiation group, (p=0.03). For BCS, LRR were numerically higher for HER2+/TN subtypes however overall group differences were not significant (p=0.14). In multivariable analyses (MVA), adjusted for stage, chemotherapy, and local therapy, only TN (vs. Luminal A) was associated with an increased LRR risk (HR: 2.36, 95% CI 1.16-4.79). Conclusion: In young women with BC, risk of local recurrence varies by subtype and in this observational cohort was most pronounced among those who received mastectomy with radiation. However on MVA, TN was the only subtype associated with LRR. Introduction Chemotherapy is recommended for the majority of patients with triple negative breast cancer (TNBC), but comorbidities and competing causes of death in the elderly can complicate clinical decision making. The purpose of this study was to explore the frequency of nodal positivity in elderly patients with TNBC and the impact of nodal disease on receipt of systemic therapy in two dedicated cancer centers. Methods Patients 70 years of age with stage I-III TNBC treated with upfront surgery including axillary staging at Memorial Sloan Kettering Cancer Center (MSK) and at the European Institute of Oncology (EIO) between 2000-2012 were identified from prospective databases. Clinicopathologic features were retrieved and comparisons between nodal categories were made using t-tests for continuous variables and chi-square tests for categorical variables. Results We identified 230 patients treated at MSK and 194 patients treated at EIO with a median age of 75 years (range 70-97) ( Table 1 ). The majority of patients in both cohorts were node-negative (159 of 230, 69.1% at MSK and 113 of 194, 58.2% at EIO). Nodal positivity was associated with receipt of chemotherapy for the overall study population (p=0.002) and in the MSK cohort (p=0.04) but not in the EIO cohort (p=0.13). Overall, there were 152 total nodepositive patients of which 103 (68.2%) received chemotherapy, including 42 of 71 (59.2%) patients at MSK and 60 of 81 (74.1%) at EIO. Chemotherapy receipt in 2 (1.3%) node-positive patients was unknown. There were 13 patients with pT1aN0 disease and none of these patients received chemotherapy. Of the remaining 272 node-negative patients, 71 of 146 (48.6%) nodenegative patients at MSK and 70 of 112 (62.5%) node-negative patients at EIO received chemotherapy. Conclusions Despite nodal positivity being an indication for chemotherapy and being associated with receipt of chemotherapy in the overall study population, around a third of node-positive patients did not receive chemotherapy. In elderly patients with TNBC, factors other than nodal positivity seem to affect receipt of chemotherapy. Introduction Cancer with certain gene mutations can have aggressive phenotypes. On the other hand, large amounts of mutations in tumors attract anti-cancer immune cells due to released neoantigens in the tumor immune microenvironment (TIME). Although breast cancer (BC) is known to have low mutation, we hypothesized that BC with high mutation rates demonstrate aggressive phenotypes with enhanced immunogenicity. Methods Genomic and clinical data were obtained from The Cancer Genome Atlas (TCGA). Tumors with high mutation rates defined as greater than 3 counts/Mbps non-silent mutation. The METABRIC cohort was used as a validation cohort. Results Although BCs with high mutation rates were associated with aggressive clinical features, survival was not different between high and low mutation rates (p=0.383). While there were no differences in tumor size or clinical stage, mutation rates were higher in Age 50 (p=0.03), in ER-negative tumors (p<0.01), in triple-negative subtype (p=0.03), and BCs with high mutation rate were associated with increased MKI-67 mRNA expression (p<0.01). Tumors with high mutation rates were associated with APOBEC3B, a well-known DNA mutator in BC, and homologous recombination deficiency (all p<0.01), which contributed to increased neoantigen loads in TIME (p<0.01). Given increased neoantigens, we performed CIBERSORT to estimate the immune cell composition, which demonstrated increased lymphocytes and anti-cancer M1 macrophage infiltration, as well as T-cell receptor diversity in BCs with high mutation rates (all p<0.01). Gene Set Enrichment Analysis revealed that cell proliferation (E2F targets, mTORc1 signaling, and G2M checkpoint) and immune activity (Interferon gamma response, and Inflammatory response) related gene sets were enriched in BCs with high mutation rates. Additionally, cytolytic activity score (CYT) and T-cell exhaustion marker expression were significantly elevated in BCs with high mutation rates (all p<0.01). The majority of the above results were validated by the METABRIC cohort. Conclusion BCs with high mutation rates demonstrated biologically aggressive phenotypes, but it appeared to be counterbalanced by enhanced immunity associated with increased neoantigens. Introduction: Metabolic Syndrome (MS) is a group of medical conditions defined by National Cholesterol Education Program Adult Treatment Panel III (NCEP/ATP III) as having at least 3 of 4 components: obesity, dyslipidemia, hypertension(HTN) and elevated fasting plasma glucose. The mechanism by which MS affects prognosis of breast cancer patients is not clear but is postulated to include higher level of estrogen, insulin, insulin like growth factor and chronic inflammatory changes. Increasing body mass index (BMI) and diabetes have been associated with worse breast cancer survival. The impact of MS on the triple negative breast cancer (TNBC) has not been evaluated. Methods: An IRB approved database review of TNBC from 2007-2013 identified 243 patients. Patients were excluded due to incomplete treatment, stage IV at diagnosis or pathologic stage unknown, leaving 181 patients for whom disease-free survival (DFS) could be calculated. Patient characteristics collected were stratified by age, BMI, race, pathologic stage, treatment as well as time to treatment from diagnosis. Results: 181 TNBC patients met the inclusion criteria. Fifty patients (27.6%) were identified as having MS. The 5-year DFS was 78.8% for the total cohort. For MS DFS was 70.0% and was 82.3% for those without MS. The MS cohort was significantly older at the time of diagnosis 67 versus 57 years (p <0.001). In the univariable Cox models, age at diagnosis, HTN, MS, and pathologic stage III were associated with worse prognosis. After controlling for age and pathologic stage, HTN but not MS was significantly associated with worse prognosis (HR=3.65, 95% CI: 1.34-9.95). Patients with a BMI over 30 unexpectedly experienced improved prognosis compared to those of normal weight (HR=0.30, 95% CI: 0.12-0.78). Conclusion: Originally MS seemed to decrease DFS for patients with TNBC. However in univariate and multivariate analysis HTN not MS decreased DFS in the TNBC population. Efforts should be directed at treating HTN as part of the TNBC patient's overall care. Background. Accumulating evidences revealed that autophagy played vital role in breast cancer (BC) progression. Thus, this study aimed to investigate the potential prognostic role of autophagy-related genes and develop a genomic-clinical model to predict overall survival (OS) in BC patients. Methods. Using the mRNA mining approach, we conducted mRNA expression profiling in a large BC cohort (n = 1007) from The Cancer Genome Atlas (TCGA) database. The association between patients' OS with autophagy-related genes was analyzed by the LASSO and Cox regression analyses. A predictive model was established based on the independent prognostic variables. Then, the C-index, calibration curve and decision curve analysis were calculated to determine its predictive accuracy, calibration ability and clinical utility. Results. Four autophagy-related genes with prognostic value (ATG4A, IFNG, NRG1 and SERPINA1) were confirmed in breast cancer patients using the LASSO and multivariate Cox regression analyses. A genomic-clinical model was then constructed based on the four autophagy-related genes, age and TNM stage, dividing patients into low-risk and high-risk groups. The OS time was higher in the low-risk group than that in the high-risk group [P <0.0001]. Timedependent receiver operating curve at 5 years indicated that the genomicclinical model had better prognostic accuracy than American Joint Committee on Cancer TNM stage in the training cohort (C-index: 0.731 vs 0.640, P<0.01) and in the validation cohort (C-index: 0.804 vs 0.671, P<0.01). The decision curve uncovered that the genomic-clinical model had good clinical utility across the wider range of threshold probability Conclusion. In conclusion, we built and validated a novel genomic-clinical model, a credible approach to 5-year OS prediction in BC patients, which can assist oncologists in devising more efficient therapeutic strategies. Background In patients with a positive sentinel lymph node (SLN) after neoadjuvant chemotherapy (NAC), the likelihood of residual disease at axillary dissection (ALND) is high. Whether non-SLN metastasis frequency varies based on tumor subtype and SLN metastasis size is uncertain. We examined the association between tumor subtype and non-SLN metastasis frequency in patients with SLN micro vs macrometastases after NAC. Methods Patients who had SLNB after NAC from 7/2008-7/2019 were identified. Those with a positive SLN and completion ALND were stratified based on tumor subtype. Association between SLN disease volume and frequency of non-SLN metastasis was examined using Fisher's exact or Chi-square tests for categorical variables and Kruskal-Wallis tests for continuous variables. Results Of 1008 patients who met study criteria, 273 had positive SLNs and ALND. Mean age was 51y, 87% (238) of tumors were ductal, 80% (219) clinically node positive, and 85% (231) cT2-3 (Table) . SLN metastases were more frequent in HR+/HER2-tumors after NAC (62%, 168/273). At ALND, the frequency of positive non-SLNs was non-significantly higher in HR+/HER2-(61%) vs HER2+ (52%) and TN tumors (45%) (p=0.09). SLN metastases size distribution differed among the subtypes. While most SLN metastases were macrometastases for all subtypes, TN tumors had lower micrometastases rates (9%) compared to HR+/HER2-(17%) and HER2+ (34%) (p=0.015). There was no association between SLN disease volume and non-SLN metastases frequency for any subtype. Among patients with SLN micrometastases, 64% of HR+/ HER2-, 41% of HER2+ and 60% of TN tumors had at least 1 positive non-SLN, similar to rates seen in patients with SLN macrometastases (64%, 55% and 43% respectively; p=0.4). Further, there was no difference in number of positive non-SLNs for micro or macrometastasis in any tumor subtype (p=NS) ( Table) . Conclusion In patients with a positive SLN after NAC, the likelihood of residual non-SLN disease at ALND was high across all tumor subtypes and not significantly different for SLN micro vs macrometastases. ALND is recommended for SLN micro and macrometastases after NAC in all tumor subtypes. Background: Major hepatectomy with combined resection of hepatic artery and/or portal vein for perihilar cholangiocarcinoma is technically demanding and challenging procedure. The aim of this study is to review our experience with major hepatectomy with combined vascular resection in patient who have advanced perihilar cholangiocarcinoma. Material and Methods: Two hundred eight patients undergoing surgical resection for perihilar cholangiocarcinoma until 2016 were studied. Of them, vascular resection was performed in 78 patients. Surgical outcome was compared between each procedure. Results: Resected vessels were hepatic artery alone in 30, portal vein alone in 38 and both in 12. Surgical procedure was left hepatectomy in 32, left trisectionectomy in 22, right hepatectomy in 16 and right trisectionectomy in 8, respectively. Positive dissected margin was detected in 2 (2.6%) patients with vascular resection and 1 (0.6%) patients without vascular resection (p=0.246). Morbidity was almost the same between each group; Clavien-Dindo classification >3, 47% in patients with vascular resection and 46% in patient without vascular resection (p=0.999). There were a few vascular resection related complication; portal vein thrombosis in 2 and liver abscess in one. Survival of patients with vascular resection (median survival time (MST): 36 months) was significantly worse than that of patients without vascular resection (MST: 61 months) (p=0.009), but significantly better than that of patients with locally advanced perihilar cholangiocarcinoma treated with chemotherapy (MST: 11 months) (p<0.001). Among patients with vascular resection, there were no significant differences between each procedure; MST, 41 months in arterial resection, 38 months in portal vein resection and 34 months in both arterial and poral vein resection.

There are fourteen 5-year survivor. Conclusion: Hepatectomy with combined vascular resection for perihilar cholangiocarcinoma was technical demanding procedure but could be performed with acceptable short-term and long-term outcomes in a high-volume center.

Association Between Biopsy and Development of Peritoneal Carcinomatosis in Perihilar Cholangiocarcinoma V.G. Aveson, 1 * J. Chou, 1 M. Gonen, 1 V. Balachandran, 1 T.P. Kingham, 1 J.A. Drebin, 1 W. Jarnagin, 1 R. Chandwani, 2 M. D'Angelica. 1 1. Surgery, Memorial Sloan Kettering Cancer Center, New York, NY; 2. Weill Cornell Medical Center, New York, NY. Background Perihilar Cholangiocarcinoma(PHC) is a disease with a poor prognosis and few effective treatment options. Case reports suggest transperitoneal (TP) biopsy is associated with dramatically higher rates of peritoneal carcinomatosis(PC) than intraluminal(IL) biopsies, but no large study has been performed. We aimed to identify any association between biopsy type and development of PC in a cohort of resected patients. Methods Consecutive patients with biopsy-confirmed PHC who received R0/R1 surgery at Memorial Sloan Kettering and at least 6 months of follow up after biopsy were included. Instances of PC were determined by radiographic findings or biopsy. Overall survival(OS) in months from surgery was estimated using Kaplan -Meier methods. Cumulative incidence of PC was estimated using competing risks methods and compared between subgroups using Gray's test. 5.8-24 .3%] respectively at 2 years. We did not detect a significant difference in the cumulative incidence rates of PC after surgery among biopsy types (p=0.638). Cumulative incidence rates of PC after surgery were also compared between positive and negative biopsies of each type and no biopsy and we again did not detect a significant difference (p=0.541 Figure 1 ). Discussion In our cohort, we found a high rate of peritoneal recurrence after R1/R0 resection, suggesting that PC is an important part of the natural history of hilar cholangiocarcinoma. The type of biopsy patients underwent was not associated with a higher risk of PC. The risk of peritoneal seeding from TP biopsies may not be as high as case studies suggest. Background: While surgery remains a mainstay in the treatment of patients with intrahepatic cholangiocarcinoma (ICC), the role of neoadjuvant therapy (NT) for this cancer has not been proven. We sought to investigate the trends in NT utilization, evaluate factors associated with NT use, and study its impact on overall survival (OS) for patients with surgically resected ICC. Methods: This is a retrospective cohort study of 4,456 surgically resected ICC patients within the National Cancer DataBase (2006) (2007) (2008) (2009) (2010) (2011) (2012) (2013) (2014) (2015) (2016) . NT included chemotherapy alone and/ or (chemo)radiation. Standard descriptive statistics were used to describe the cohort. Multivariate logistic regression models were used to examine factors associated with decision to administer NT. Survival was estimated using the Kaplan-Meier method. Results: Utilization of NT has increased over time (11% in 2006 to 16% in 2016, trend Five-year OS for patients who received NT was better than that for those who did not (37.2% vs 31.6%, log-rank p<0.001). Among 118 cN+ patients who received NT, 34 (29%) had pathologically negative nodes (ypN-) on surgical pathology. Fiveyear OS for these patients was better than that for patients who remained ypN+ after NT ( Figure 1 ). Conclusions: NT is used with increasing frequency in the treatment of ICC, and seems driven by factors suggestive of more aggressive tumor biology. Nearly one-third of patients with cN+ disease who received NT were downstaged to ypN-. Further work is needed to inform guidelines for NT use in ICC patients with potentially resectable disease.

Overall survival in surgically-resected patients with intrahepatic cholangiocarcinoma, based on clinical and pathologic nodal status and receipt of neoadjuvant therapy BACKGROUND: Intrahepatic cholangiocarcinoma (ICC) is a primary liver cancer that is often unresectable at diagnosis and has a high risk of recurrence when resected. ICC is increasingly arising in the setting of nonalcoholic fatty liver disease (NAFLD). We sought to describe the patterns and risk factors for recurrence and survival, including NAFLD, after hepatic resection at a multidisciplinaryr program within an NCI Comprehensive Cancer Center. METHODS: We identified n=52 patients with ICC who underwent resection with curative intent from 2004-2017. Clinicopathologic data were analyzed and oncologic outcomes were assessed. Overall survival (OS) and recurrence-free survival (RFS) were evaluated with Kaplan-Meier and Cox proportional hazards models. RESULTS: Median patient age was 64-years-old with an equal male to female distribution. Median BMI was 30.3, with 23% of patients (n=12) having histologically confirmed NAFLD with evidence of ongoing inflammation at time of resection. Median tumor size was 5.5 cm (range 0.6-18cm). At median follow-up of 47 months, n=26 (50%) had recurred. The most common pattern of recurrence was intrahepatic alone (n=14, 54%). All recurrences were within 51 months of resection. Median OS was 49 months with a 5-year OS of 42%. Perioperative chemotherapy was administered in 24 patients (46%). On univariate analysis, factors associated with worse RFS were tumors >5 cm (P<0.05) and the presence of NAFLD with inflammation (P=0.05). Tumors >5cm, tumor multifocality, and NAFLD with inflammation were associated with worse OS (all P<0.05). After adjusting for disease stage and margin status, only NAFLD with active inflammation was associated with worse OS (HR 3.59, P=0.012), and suggested worse RFS, but was inconclusive due to small sample size (HR 2.31, P=0.1). CONCLUSION: In our institutional experience, resected ICC most commonly recurs in the liver and uniformly within 5 years. The presence of NAFLD with active inflammation is independently associated with recurrence and should be considered an indication for intense surveillance with integration of systemic therapy into treatment planning. Background The current understanding of the genomic landscape of hepatobiliary cancer (HBC) is limited. Recent genomic and epigenomic studies have demonstrated that various cancers of different tissue origins can have similar molecular phenotypes. Therefore, the aim of this study is to evaluate the genomic alterations of HBCs as a first step towards creating a novel molecular subtype classification. Methods A multidimensional analysis of nextgeneration sequencing for the genomics landscape of HBCs was conducted using mutational data from the AACR-Genomics Evidence Neoplasia Information Exchange database (v. 5.0). From 61 gene mutation platforms, we found 42 genes common to all HBC cases. Associations between histomolecular characteristics of HBCs (hepatocellular (HCC), cholangiocarcinoma (CCA), and gallbladder carcinoma (GBC)) with gene mutations (classified by COSMIC CENSUS) were analyzed. Results A total of 1,017 alterations were identified in 61 genes (516 missense variant, 157 gene amplifications, 101 inactivating mutations, 106 truncating mutations, 84 upstream gene variants, 37 gene homozygous deletions, 16 gene rearrangements) in 329 patients: 115 (35%) CCA, 87 (26.4%) GBC, and 127 (38.6%) HCC. The majority 77.8% (256) of tumors harbored at least two mutations and 38.9% (128) had at least one alteration, with GBC having a higher average number of alterations (3.28) than HCC (3.23) and CCA (2.49) However, HCCs had the higher maximum number of alterations compared to CCA and GBC (p<0.05). The ten genes most frequently altered across all the HBCs were TP53, TERT, CTNNB1, KRAS, ARID1A, CDKN2A, IDH1, PIK3CA, MYC, and SMAD4 with disparities in the distribution of genes altered repeatedly observed (p<0.001). IDH1 mutations were associated with CCA, CTNNB1 and TERT mutations with HCC, and TP53 mutations with both HCC and GBC. Conclusion HBC subtypes appear to have unique mutational landscapes, but also a significant overlap of genetic signatures. Therefore, further exploratory genetic and epigenomic research is needed to develop a histomolecular classification algorithm that can be used for prognostic and therapeutic stratification of these cancers. Introduction: Accurate interpretation of anatomy is crucial for preoperative planning, and it relies heavily on the surgeon's ability to mentally reconstruct 2-dimensional (2D) axial radiologic images into 3D constructs. This is especially the case in liver surgery, and given the complex surgical anatomy of the liver, this mental exercise can be challenging for surgical trainees. The objective of this study is to determine whether a 3D model of pre-operative axial imaging improves the surgical trainee's assessment of liver lesions and aids in surgical planning. Materials and Methods: High-quality, multiphase, liver-protocol computerized tomographic (CT) scans of patients with hepatic tumors were obtained and reconstructed into 3D PDF models accessible using Adobe Acrobat Reader. A questionnaire was administered to surgical residents at a single institution to assess interpretation of surgical anatomy, query a proposed surgical plan and evaluate residents' perceived confidence in their answers. Residents completed the questionnaire twice, initially with CT scan only and then again with the CT scan and 3D model accessible. Participants had 4 minutes per case to complete the questionnaire. Results: 32 surgical trainees participated in the study. Access to the 3D model resulted in a higher percentage of correct answers on objective questionnaire versus CT scan only (49% vs 76% correct, p<0.0001). The improvement was observed in both junior residents (PGY 1-2; 42 vs 70%, p<0.001) and senior residents (PGY 3-5; 60% vs 83 %, p=0.003). Overall percent correct increased in 10 of 11 questions when 3D model was accessible. 72% of participants reported higher degree of confidence in answers when the 3D model was accessible versus using the CT alone. All participants reported the 3D model to be helpful overall in interpretation of each patient case. Conclusion: 3D modeling of liver anatomy is a helpful tool to facilitate accurate anatomic interpretation and surgical planning for surgical trainees. The usefulness appears to be present regardless of the level of training, background knowledge or recent HPB experience.

Background The optimal level of pedicle ligation during proctectomy for rectal cancer, either at the origin of the inferior mesenteric artery (IMA) or the superior rectal artery (SRA), is still debated. Reasons for IMA ligation include facilitating a tension-free anastomosis and improved clearance of regional lymph nodes. Our aim was to determine if SRA ligation portends inferior outcomes. Methods The US Rectal Cancer Consortium database (2007) (2008) (2009) (2010) (2011) (2012) (2013) (2014) (2015) (2016) (2017) was reviewed for pts with primary, non-metastatic rectal adenocarcinoma who underwent treatment with low anterior resection or abdominoperineal resection. Primary outcomes were anastomotic leak rate and lymph node (LN) harvest. Secondary outcomes were locoregional recurrence-free survival (LRFS), recurrence-free survival (RFS), and overall survival (OS). Results Of 877 pts, median age was 59 years (IQR52-67) and 62% were male (n=541). 86% received an IMA ligation (n=755) while 14% SRA (n=122). 12% were pathologic stage 0 (n=101), 33% stage I (n=281), 24% stage II (n=206), and 31% stage III (n=269). Median follow-up was 34 mos. SRA ligation was more common in stage III disease (43vs30%, p=0.005) while IMA ligation was more often performed with a minimally invasive approach (70vs42%, p<0.001). SRA ligation was associated with a nearly identical anastomotic leak rate compared to IMA (9vs8%, p=1.0). Similarly, the median number of LNs removed was the same between both ligation groups (15vs15, p=0.38). On multivariable analysis accounting for an open approach, advanced pathologic stage, and positive resection margin status, SRA ligation was not associated with increased anastomotic leak rate or reduced LRFS, RFS, or OS (all p>0.1). Conclusions If a tension-free anastomosis is feasible, SRA ligation is not associated with either a worse technical outcome or inferior lymph node harvest. Furthermore, all cancer survival metrics are similar between SRA and IMA ligation. Given that either approach is safe and feasible from both a technical and oncologic standpoint, this study questions the routine practice of IMA ligation. Introduction: Low-ligation (LL) of the inferior mesenteric artery (IMA) with preservation of the left colic artery is used as a method to increase perfusion to colorectal anastomoses over ligation of the IMA at its origin (high ligation, HL). This LL method has been criticized for its potential to leave apical IMA nodes in situ, so has been paired with apical node dissection (AND). For patients undergoing LL+AND, we quantified the number of apical nodes resected, the proportion positive for malignancy, and the proportion of patients with increased stage by including the apical nodes. We compared nodal retrieval, local recurrence and overall survival between HL and LL+AND. Methods: In this retrospective cohort study we identified all patients receiving a rectosigmoid resection for cancer between January 2012 and July 2018 from a prospectively maintained institutional database. We excluded patients with synchronous metastatic disease and those with LL but without AND. Demographic and clinical outcomes were compared between groups using a student's t-test, chi-square analysis, and Mann-Whitney U test, as appropriate. AND was processed separately to quantify the number of nodes, presence of malignancy, and upstaging. We performed survival analysis using a Cox proportional hazard model. Results: Of the identified patients, 84 had HL and 89 LL+AND with median follow-up of 20 months. In the LL+AND group, a median of 2 (IQR=1-3) apical nodes were resected; 4.1% were positive for malignancy resulting in increased prognostic stage in 25% of LL patients with positive apical nodes. Between HL and LL+AND there were no differences in number of nodes retrieved (16 vs 17, p=0.19), any complication (29.8% vs 38.2%, p=0.31), anastomotic leak (3.1% vs 4.5%, p=0.97), local recurrence (7.1% vs 2.2%, p=0.24), or overall survival (HR 0.55, 95%CI 0.16-1.66). Conclusion: No detectable differences were identified between HL and LL+AND with respect to perioperative complications, local recurrence, or overall survival. 4.1% of resected apical lymph nodes were positive for malignancy, suggesting that if a LL is performed, that it should be paired with an AND. Surgical treatment of locally advanced colon cancer (LACC) can be challenging due to tumor size and the need for multivisceral resection. The role of laparoscopic resection in LACC is controversial. The aim of this study is to compare outcomes for laparoscopic versus open surgery for LACC. A retrospective review was conducted of all patients with LACC referred to a Provincial Cancer Centre treated between 2005 and 2015. Patients with non-metastatic, T4 colon cancers > 16 cm above the anal verge were included. Descriptive analysis was used to define the study population. Survival and recurrence analyses were performed using Kaplan-Meier curve estimation and Cox-proportional Introduction: Colorectal cancer (CRC) is the third most commonly diagnosed malignancy in the world according to the World Health Organization GLOBOCAN database and approximately 20% of patients have metastatic disease at diagnosis. There is a growing interest in the surgical management of peritoneal metastasis with cytoreductive surgery and hyperthermic intraperitoneal chemotherapy (CRS and HIPEC). Here, we study the outcomes after CRS and HIPEC, in a subgroup of CRC patients who have synchronous colorectal peritoneal metastasis (CPM), Methods: A retrospective analysis of a prospectively maintained database on CPM patients treated with CRS and HIPEC at the National Cancer Centre Singapore between January 2003 to January 2018 was performed. All patients who had peritoneal disease diagnosed at presentation who underwent CRS & HIPEC at the time of primary surgery or within six months from primary surgery were included into the study. Results: Twenty patients with synchronous CPM underwent CRS and HIPEC during the study period and were included in the analysis. Histological review found 85% (n=17) of tumours were at least moderately differentiated with a T stage of 3 or 4. Patients had a median peritoneal cancer index (PCI) score of 9 , and completeness of cytoreduction (CC) score of 0 was achieved in 95% (n=19) of patients. The median 3 year overall survival of these patients was 27.5% (95% CI 4.5-58.4%), and the progression free survival was 28.8% (95% CI 5.2-59.1%) in 3 years. Conclusion: CPM is a challenging disease and patients who are diagnosed with synchronous CPM tend to have an overall poor prognosis. This may be explained by the differences in their cancer biology, and suggest that synchronous CPM may require a different approach in their management as compared to metachronous CPM.

Background: Although potentially associated with increased infections, intraoperative pelvic drains are often placed during low anterior resection (LAR) to evacuate postoperative fluid collections and identify / control potential anastomotic leaks. Our aim was to assess the validity of this practice in a large dataset of patients undergoing LAR for rectal cancer. Methods: Patients from the US Rectal Cancer Consortium (2007-17) who underwent curative-intent LAR for a primary rectal cancer were included. Patients were categorized as receiving a closed suction drain intraoperatively or not. Primary outcomes were superficial surgical site infection(SSI), deep SSI, intraabdominal abscess, anastomotic leak, and need for secondary drain placement. Three subgroup analyses were conducted in patients who received neoadjuvant chemoradiation, had a diverting loop ileostomy (DLI), and had low tumors <6cm from the anal verge. Results: Of 996 pts, average age was 58 yrs, 61% were male, and 67% (n=551) received a drain. Drain patients were more likely to be male (64vs54%), have a smoking history (25vs19%), have received neoadjuvant chemoradiation (73vs61%), have low tumors within 6cm of the anal verge (56vs36%), and have received a DLI (80vs71%) (all p<0.05). Drains were associated with an increased anastomotic leak rate (14vs8%, p=0.041), although there was no difference in the need for a secondary drainage procedure to control the leak (82vs88%, p=0.924). These findings persisted in all subset analyses. Drains were not associated with increased superficial SSI, deep SSI, or intraabdominal abscess in the entire cohort or each subset analysis. Reoperation (12vs10%, p=0.478) and readmission rates (28vs31%, p=0.511) were similar. Conclusions: Although not associated with increased infectious complications, intraoperatively-placed pelvic drains after low anterior resection for rectal cancer are associated with an increase in anastomotic leak rate and no reduction in the need for secondary drain placement or reoperation. Routine drainage should be abandoned. Background: Laparoscopic and robotic approached in liver resections comprises the most widely used surgical platforms. The purpose of our study was to provide a current assessment of the perioperative benefits of the minimally invasive surgery (MIS) based on the extent of surgery. Methods: We used the ACS_NSQIP database targeted for hepatectomy to identify patients who underwent liver resection of colorectal liver metastases (CRLM) during 2014-2016. A 1:1 propensity score analysis was performed to match patients who underwent open vs. MIS based on age, ASA classification, race, gender, liver texture, neoadjuvant therapy, BMI, pre-operative serum albumin level, and chronic steroid use. The short -term outcomes in open vs. MIS groups were compared. Results: From a cohort of 3801 patients, 1278 patients were matched with 639 in each group. Overall morbidity was greater among patient in open group compared with MILS (32.1% vs 20.0%, p= <0.001). There was no difference in mortality. MIS was associated with significant reductions in operative time (235.94 vs 253.97 min, p = 0.006) and the length of stay (4.65 vs 6.55, p = <0.001) overall. The observed benefit was significant for non-anatomical resections: decreased operative time (215.65 vs. 232.00, p = 0.020), LOS (4.12 vs 5.85, p < 0.0001) and overall morbidity (17.4% vs. 28.1%, p< 0.001). On multivariate logistic regression analysis, the extent of surgery (non-anatomical resection vs major hepatectomy), open approach (p <0.001) were associated with increased odds of overall morbidity. Open surgeries required more transfusions (19.2% vs 8.6%, p < 0.001) and led to more organ space infections (6.4% vs. 3.4%, P =0.019), septic shock (1.9% vs. 0.5%, p=0.034) and higher readmission rate (10.6% vs. 6.6%, p = 0.012) Conclusions: MIS approach in colorectal liver metastases was associated with a shorter operative time and reduced postoperative morbidity, even after controlling for important patient and clinicopathologic confounders, compared to open approach. The difference in outcomes was more prominent for non-anatomical resections. INTRODUCTION: The role of intraoperative margin treatment for a positive margin in curative-intent hepatectomy is unclear. This study aims to examine the effect of margin treatment on recurrence and survival in colorectal cancer patients with liver metastases. METHODS: A retrospective review of a prospective hepatobiliary database was performed of 335 patients receiving hepatectomy for metastatic colorectal carcinoma from November 1996 to June 2017. Surgical margin treatment was defined as either re-resection or ablation of resection margin. Patients were stratified into three groups: R0 resection, R1 resection without margin treatment, and R1 resection with margin treatment. Variables analyzed included age, sex, extent of surgery, ablation, number of liver lesions, maximum size of liver lesion(s), and complications. RESULTS: Analysis included 335 patients undergoing 368 procedures. 286 (85.37%) patients underwent R0 resection, 28 (8.36%) patients underwent R1 resection without margin treatment, and 21 (6.27%) patients underwent R1 resection with margin treatment. Rates of local hepatic recurrence (p=0.15), new hepatic recurrence (p=0.70), and extra-hepatic recurrence (p=0.24) did not differ between the three groups. Median OS for all three groups was 44.2 months, 26.0 months, and 20.3 months for the R0 group, R1 without margin treatment group, and the R1 with margin treatment respectively (p=0.0008). On multivariate analysis, margin group (p=0.045), gender (p=0.033), and ablation (p<0.0001) were all significantly associated with OS. CONCLUSION: Intraoperative margin treatment for positive margin during resection of colorectal liver metastases did not have any effect on recurrence. R1 resection is associated with decreased OS compared to R0 resection, regardless of margin treatment. Introduction: Gall bladder cancer(GBC) is the most common malignancy of the biliary tract (1) . Preoperative staging is the most important tool in deciding the management protocol of the patients. Though PET-CT and Diagnostic laparoscopy have been performed increasingly, the exact role of each modality in the staging of GBC patients is not clear. In the present study, we evaluated the utility of PET-CT and Diagnostic laparoscopy in the staging of Gall bladder cancer patients. Materials and Methods: This prospective study was conducted in the Department of Surgical Oncology (SO), at AIIMS, New Delhi from May 2018 to May 2019. Those patients diagnosed with GBC or suspicious of GBC deemed operable in CECT, were included in the study. Both PET-CT and Diagnostic laparoscopy were done in all the patients and evaluated for staging of GBC. Results: Among the total of 50 patients (CECT wise operable), PET detected definitive inoperability (2 patients) and peritoneal and omental metastasis (3 patients) in 5 patients (10%). Among the 45 patients in whom PET-CT showed operability, 16 patients (35.6%) were found to be surgically inoperable either due to metastasis or local unresectability. When CECT is considered as sole modality, 21 (42%) patients would have been surgically inoperable against PET-CT which showed inoperability in 16 (35.6%) patients. Among those patients who were deemed operable by PET-CT, diagnostic laparoscopy detected inoperability/ metastatis in 8 patients (17.8%). Out of the 16 patients who were surgically inoperable, diagnostic laparoscopy reduced the rate of futile laparatomies in 8/16 (50%) patients. The sensitivity, specificity, positive predictive value, and negative predictive value of PET-CT in detecting the lymph node involvement in Gall bladder cancer were 57.1%, 79.1%,30.8% and 86.1% respectively. In overall cohort, This rate would further increase 11/16 (68.7%) if the PET-CT detected metastasis (3 patients) was included in the Diagnostic laparoscopy group. Conclusion: PET-CT and Diagnostic laparoscopy should be considered as an additive staging modality in the operable GBC patients. Futile laparotomy was avoided in half of the inoperable patients by doing diagnostic laparoscopy after PET-CT. It will facilitate early post operative recovery as well as early administration of chemotherapy. Background: The ability to recapitulate the complexities of solid human tumors for the purposes of drug development and testing remains a major obstacle in the progress of cancer care. Despite efforts expended on pre-clinical optimization with existing models, most drugs simply fail to demonstrate efficacy when subjected to phase III clinical trial scrutiny. Our interpretation is that current available model systems lack the appropriate clinical predictive power. Methods: Through an iterative process using porcine livers, discarded donor livers, and tumor-bearing partial hepatectomies, we developed a perfusion model of human metastases using resected tumor-bearing liver. The Liver Assist device from Organ Assist was re-crafted to support long term oxygenation of hemi-livers, outfitted with a mechanical compression device powered by a ventilator, which recreated diaphragmatic mechanical forces for lymphatic flow. The perfusate was engineered for maximal oncotic pressure, including use of a cell-free oxygen carrier solution. Physiologic parameters, transcriptomics, immune cell populations and tissue/tumor architecture were assessed. Results: Four trials were completed with porcine hemi-livers that remained viable at normothermic conditions up to 50 hours. Five tumorbearing, human liver segments were kept viable from 10 to 49 hours. All perfused segments were maintained at physiologic conditions, showed autoregulation of pH (7.3-7.7) , lactate (0.5-1.5mmol/L first 24hr), and oxygen extraction ( pO 2 >300mmHg) throughout perfusion, while maintaining synthetic function (urea 10-45mg/dL/hr; bile 0.1-0.5mL/hr). Normal tumor architecture was noted on histology. Helper and cytotoxic T-cells were cultured and exhibited secretion of IFN-(147counts/well). Applications of the model were verified with pembrolizumab drug delivery. Conclusions: Our system will accelerate the ability to characterize tumors and test new therapies in an unfettered platform free from the restrictive limitations associated with interventions in patients. The platform will transcend multiple disciplines in translational medicine to permit interrogations and manipulations not previously possible.

BACKGROUND: Hepatic arterial infusion (HAI) of FUDR has been demonstrated to improve survival in select patients with colorectal liver metastases (CRLM). Use of HAI FUDR is associated with biliary sclerosis (BS) in 5-30% of patients. While there is not a standardized definition, some centers define BS as a stricture from HAI therapy requiring biliary stenting. We believe this definition does not truly capture the clinical or therapeutic impact of FUDR biliary toxicity. Herein, we examine the patterns of BS at a newly established HAI program and propose a definition of BS. METHODS: Between 2016 and 2018, n=22 consecutive patients had an HAI pump placed for treatment of advanced CRLM at a multidisciplinary program within an NCI Comprehensive Cancer Center. HAI with FUDR and dexamethasone was initiated and adjusted according to an established protocol. Clinicopathologic data and overall survival (OS) were evaluated. RESULTS: HAI pumps were placed in 15 patients with technically unresectable CRLM and 7 patients with completely resected CRLM; all patients received concurrent systemic therapy. Of the 21 patients that received 1 dose of FUDR, 4 unresectable patients (27%) and 1 adjuvant patient (17%) developed BS (total: n=5, 24%) after a median of 5 HAI cycles (range: 3-10). Only 3 of the 5 patients (60%) required biliary stenting and underwent a median of 10 (range: 8-21) biliary drainage procedures. Patients with BS received more HAI cycles before their first dose reduction (median 2 vs 1) and had higher cumulative doses of FUDR (median 423 vs 263 mg; P=0.02) ( Figure) . There was no significant difference in median OS between patients that developed BS and those that did not (54 vs 39 months, respectively; P=0.36). CONCLUSIONS: Although development of BS did not shorten OS in our patient population, there is a subset of patients that have significant morbidity. To better capture the full spectrum of clinically relevant biliary toxicity that impacts HAI therapy, we propose BS be defined as "new biliary dilatation and/or a nonmalignant fixed narrowing of a bile duct segment resulting in a hold or termination of HAI therapy, regardless of need for biliary stenting." Background: Robotic hepatectomy is gaining increasing acceptance. However, most studies to date have focused on minor hepatectomy comparisons and data on robotic versus open major hepatectomy ( 3 Couinaud segments) is still lacking. The objective of this study was to compare the perioperative and short-term outcomes between robotic and open major hepatectomy from our institution. Methods: Between 2016 and 2019, data on robotic (n=40) and open (n=40) major hepatectomy patients was retrospectively analyzed after 1:1 individual matching. Both cohorts were matched based on demographics, body mass index (BMI), ASA status, underlying liver disease and type of hepatectomy. Results: Robotic major hepatectomy (RMH) was comparable to open major hepatectomy (OMH) in overall complication rate, postoperative bile leak, negative tumor margin status, 30-day readmission rate and mortality (1 patient in each group). Mean operative time was significantly longer in the robotic group (302 minutes vs 197 minutes, p<0.0001). However, RMH was associated with lower intraoperative blood loss (198 mL vs 357 mL, p=0.002) and lower rate of perioperative blood transfusion (OR 0.243, p=0.044). Length of hospital stay was significantly shorter for RMH when compared to OMH (5.1 days vs 6.9 days, p=0.009). Conclusion: This study represents one of the largest series to date of matched comparison between robotic and open major hepatectomy. The robotic approach offers similar safety and feasibility as the open approach and is associated with lower intraoperative blood loss and transfusion rate along with a significantly shorter hospital stay. INTRODUCTION The identification of relevant biological pathways and mutations is integral to improving outcomes in appendix peritoneal metastases (APM). Interrogation of cancers with NGS 50-gene mutation panels has become more widely utilized identifying prognostic and actionable mutations. This study is a dedicated analysis of the value of expanded mutation analysis in APM. METHODS The IRB approved study included 50 APM patients where data was retrospectively collected from a CRS/HIPEC registry treated 2012-2018. Standard clinical 50-gene NGS analysis was performed in CLIA approved lab. All patients underwent CRS/HIPEC with mitomycin C delivered for 90 minutes at 41-42 C. PCI, CC-score, length of stay (LOS), PFS, OS were collected along with the rates and types of mutation in APM. OS and PFS analysis was performed on all, high grade (HG), and low grade (LG) APM, specifically evaluating the impact of smad4 and p53 mutations on survival. RESULTS Eighty-two percent of APM had a mutation identified with 52% of cases harboring 2 mutations. Kras was most frequent, 75% of APM (90% LG 40% HG) and GNAS was identified in 90% of LG APM. As shown in Table 1 , Smad4 or p53 mutation occurred in 21% of APM and a significant reduction in OS in all APM (55 vs 88 months p=0.045) and HG APM (20 vs 50 months p=0.049) was observed. Smad4 mutation was also associated with a significant reduction in PFS in LG-APM (p=0.049) and HG-APM (p=0.044). Actionable mutations were identified in majority of APM. CONCLUSIONS Smad4 and p53 mutations were associated with more aggressive APM and maybe a useful tool in patient selection and outcome. Extended mutation profiles is valuable in APM and further application is warranted. Research in Kras, p53 and smad4 pathways and drug development will benefit APM. Background: Cytoreductive surgery (CRS) with or without hyperthermic intraperitoneal chemotherapy (HIPEC) is indicated for patients with appendiceal cancer with peritoneal dissemination. The role of neoadjuvant chemotherapy (NAC) prior to CRS-HIPEC is controversial and its impact on survival outcomes is unclear. Methods: A retrospective review of adult patients who underwent CRS +/-HIPEC for metastatic appendiceal cancer at 12 U.S. institutions from 2000-2017 was performed. Patients with non-invasive neoplasms were excluded. Patients who did and did not receive NAC were compared before and after 1:1 propensity score matching (PSM) using age, differentiation, presence of signet ring cells, and peritoneal carcinomatosis index (PCI) score. Results: Among 803 patients with appendiceal cancer who underwent CRS+/-HIPEC, 225 received NAC and 578 underwent surgery first (SF). Median patient age was 55 years old and and most patients had appendiceal adenocarcinoma (95%). Median PCI score was 16 (IQR 10-23), the majority of patients underwent complete cytoreduction (CC-0/1, 75%) and had well (56%) or moderately (24%) differentiated tumors. A minority of patients (27%) had signet ring cells (SRCs) present. Overall (NAC: 19 vs. SF: 29 months, Figure 1A , p<0.001) and recurrence-free (12 vs. 20 months, p<0.001, Figure 1B ) were worse among the NAC group. After PSM (n=186) there was no significant difference in matched factors or in median overall survival (OS) (NAC: 40 vs. SF:56 months Figure 1C , p=0.210), but recurrence-free survival (RFS) was worse among patients undergoing NAC (14 vs. 22 months, Figure 1D , p=0.007). On multi-variable Cox regression analysis, receipt of NAC was associated with worse OS (HR 1.81, 95% CI 1.03, 3.18, p<0.001) and RFS (HR 1.93, 95% CI 1.25, 2.99, p=0.003). Conclusion: In this multi-institutional retrospective analysis of patients with appendiceal cancer and peritoneal dissemination, the use of NAC prior to CRS-HIPEC was associated with worse OS and RFS even after PSM and multivariable regression. Immediate surgery should be considered for patients with disease amenable to complete cytoreduction. INTRODUCTION: The completeness of cytoreduction (CC) score, which assesses the amount of residual tumor, is a major prognostic factor when treating appendiceal carcinomatosis with cytoreductive surgery (CRS) and hyperthermic intraperitoneal chemotherapy (HIPEC). It assigns a score of 0-3, with CC-0 and CC-1 both considered complete cytoreductions (CC-0/1) and associated with the best outcomes. We aimed to analyze if the CC-0/1 definition is reliable across appendiceal histopathologic subtypes. METHODS: A prospective database of patients who underwent CRS/HIPEC for appendiceal carcinomatosis from 1998-2019 was reviewed to identify patients with CC-0/1 at first CRS/HIPEC. Kaplan-Meier overall survival (OS) and progression-free survival (PFS) by CC-score within each histopathologic subtype were calculated. RESULTS: Of 335 CRS/HIPEC patients, 297 (89%) had CC-0/1 scores. Mean age was 54 12 years with 67% female. Histopathologic subtypes were low-grade mucinous carcinoma peritonei (LGMCP) (45%, n=133), high-grade MCP (HGMCP) (27%, n=81), HGMCP with signet ring cells (HGMCP-S) (20%, n=59), and goblet cell adenocarcinoma (GCAC) (8%, n=24). Median peritoneal cancer index (PCI) was 27 (interquartile range: 11-34) with CC-0 and CC-1 achieved in 63% and 37% of patients, respectively. CC-0 and CC-1 was achieved in 57% and 43% of LGMCP, 65% and 35% of HGMCP, 68% and 32% of HGMCP-S, and 79% and 21% of GCAC, respectively. PCI was significantly higher in patients with CC-1 versus CC-0 across all histopathologic subtypes. OS and PFS were statistically longer for CC-0 versus CC-1 in HGMCP-S (p=0.001 & p=0.012, respectively) and GCAC (p<0.001 & p<0.001), but not for LGMCP (p=0.098 & p=0.398) or HGMCP (p=0.167 & p=0.356) ( Table 1) . CONCLUSIONS: Survival outcomes for CC-0 and CC-1 after CRS/HIPEC are different for HGMCP-S and GCAC, but not for LGMCP and HGMCP. In HGCMP-S and GCAC, only CC-0 should be considered a 'complete cytoreduction' and analyzed separately from CC-1. This distinction is key for surgeons treating this condition to achieve the best outcomes. Background: Peritoneal carcinomatosis from appendiceal neoplasms is a rare disease usually found unexpectedly at the time of appendectomy. Most practicing surgeons may never, or will rarely encounter this in their career. Thus, quality reporting of key clinical information to specialty referral centers is often suboptimal. Herein, we evaluate the quality of referral operative and pathology reports to a major center. Methods: Retrospective review of our prospectively maintained appendiceal peritoneal carcinomatosis database evaluating index operative and pathology reports was completed. Quality was scored the operative report by two standards; general quality of the operative report determined by the Royal College of Surgeons "Good Practice Guidelines". Specific peritoneal carcinomatosis standard determined by thoroughness of peritoneal evaluation and reporting of disease. Results: A total of 857 cases over a 26 year period were reviewed. General quality of the index operative reports was excellent with nearly 90% of reports encompassing all the RCS quality metrics and the majority of the rest of the reports missing less than 3 quality metrics. However, the assessments of peritoneal metastasis/ carcinomatosis quality was uniformly poor. 48% of the reports limited description of evaluation only to the appendix itself, while 38% only involved partial peritoneal evaluation. Only 14% of all the reports described a complete peritoneal evaluation. 44% of the pathology reports from the referring institution had discrepancies with final pathologic findings. Of the discordant pathology reports, 30% misdiagnosed low-grade and 70% misdiagnosed high-grade. Conclusions: This review finds that referral operative reports description of the technical procedure is usually complete. However, oncologic parameters and descriptions of peritoneal metastases is frequently lacking. Further, pathology reports from outside institutions are frequently misleading. We propose a structured peritoneal evaluation to assist with accurate reporting of burden of disease. Referral pathology is also prone to inaccuracy. . A completeness of cytoreduction score (CC) of 0/1 was achieved in 79%. Median follow-up was 4.9 (IQR 2-7) years. 95 patients (54%) had no evidence of disease. 48 (27%) were alive with disease recurrence. 34 (19%) patients were deceased: disease specific mortality was 65%. Median overall survival (OS) for the entire cohort was 9.3 years. Univariate analysis showed a significant association between CC and survival (HR 1.90, 95% CI 1.40-2.59; p<0.001; Table 1 ). On multivariable analysis controlling for both clinically relevant and statistically significant covariates, incomplete cytoreduction (CC 2/3) yielded an adjusted hazard ratio (aHR) of 2.15 (95% CI 1.40-3.29; p<0.001). PCI score did not have an effect on long-term survival. Conclusion: CRS/HIPEC is effective for the management of LAMN with peritoneal metastases. Completeness of cytoreduction is the greatest predictor of survival and offers patients the potential for long-term survival. PCI score should not be used as a means to exclude patients who can achieve a complete cytoreduction from surgery. Background Neoadjuvant chemotherapy (NAC) given every three weeks for 3-4 courses followed by interval debulking surgery (DS) + three or more cycles of adjuvant chemotherapy or, six cycles of NAC followed by DS is an alternative approach for patients with advanced epithelial ovarian carcinoma (AEOC) in those initial surgery (IS) is not achievable. Choosing between these two options is taken on the probability of getting complete resection of all macroscopic disease. Objective To evaluate NAC 3-4 cycles versus (vs) six or more courses + DS in AEOC, and compare with primary surgery followed by adjuvant chemotherapy. Methods A retrospective analysis of patients with AEOC from 2011 to 2016 that underwent to NAC was performed to compare complete cytorreduction (R0), surgical complications (SC), progression-free survival (PFS) and overall survival (OS) in relation to the number of NAC cycles. Women with unresectable disease were excluded. Results We include 225 cases with AEOC, 47 (20.9%) had IS followed by adjuvant chemotherapy (group 1), 104 (46.2%) had DS after 3-4 cycles of NAC (group 2) and 74 (32.9%) had cytoreductive surgery after 5 or more cycles of NAC (group 3), median of initial Ca-125 was 335 (112-1181), 1729 (542-3895) and 2013 (909-4381) in group 1,2 and 3 (p=0.01). We found differences between group 1 and the other two groups in the surgical time (p=0.01), blood loss (p=0.00), length of hospital stay (p=0.02) and postoperative complications (12.8% vs 4.8% and 5.4% of group 2 and 3) (p= 0.049). The rate of re-intervention during 30 day after DS was 6.4% in the primary surgery group vs 1.9% of interval DS (p<0.001). R0 was achieved in 89.4%, 92.3% and 91.6% in group 1, 2 and 3, respectively (p=NS). Median PFS was 18.5, 10.5 and 12.6 months, and OS was 57, 44.5 and 46 months for groups 1,2 and 3 respectively with no significance after adjusted by clinical stage (p=NS). Conclusion Our study demonstrates that timing for DS does not have an impact in terms of PFS and OS. In those cases with poor response to initial treatment, completing 6 cycles of chemotherapy could be an option but a randomized clinical trial is imperative to confirm our results. RIFLE score, thrombocytopenia, pancytopenia, length of stay (LOS), PFS, OS, peritoneal relapse and all relapse events were collected and compared from the date of surgery. RESULTS A total of 34 recurrent and AOC patients had CRS, 21 treated with HIPEC. Mean PCI was 25 for both CRS and CRS/HIPEC, 92% HIPEC and 90% no HIPEC had CC0/1. No major nephrotoxicity occurred. Thrombocytopenia (platelet < 50K) occurred in 11% of HIPEC patients. LOS was 9.5 days for both groups. OS was 20 vs. 56 months for CRS vs. CRS/ HIPEC, p<0.01. There was no difference in median OS in HIPEC group treated at recurrence or first CRS. Peritoneal recurrence was 69% for CRS vs 33% for CRS/HIPEC, p<0.01 (Table 1) . CONCLUSION This data demonstrates that Carboplatin HIPEC has similar efficacy to cisplatin without the nephrotoxicity. Carboplatin HIPEC for recurrent and AOC is safe and efficacious. The survival benefit may be attributable to peritoneal disease control and peritoneal relapse free survival may be a viable endpoint in future HIPEC clinical trials.

E.W. Beal, 1 * J. Chen, 1 A. Ahmed, 1 C. Kimbrough, 1 T.E. Grotz, 2 J.L. Leiting, 2 K.F. Fournier, 3 A.J. Lee,

A.N. Bruce, 1 * Y. Song, 1 E.C. Paulson, 2 R.E. Roses, 1 R.R. Kelz, 1 D.L. Fraker, 1 J.T. Miura, 1 G.C.

Background Traditional treatment of advanced epitelial ovarian cancer is cytoreductive surgery and platinum-based systemic chemotherapy. Recently, several studies have reported similar survival with chemotherapy administered before surgery but with less surgical morbidity. Because most patients develop intraperitoneal recurrence, we conducted a pilot study evaluating HIPEC as consolidation therapy after complete neoadjuvant systemic therapy. Methods We included patients with advanced epitelial ovarian cancer (FIGO stage IIIC and IV with only malignant pleural effusion). After informed consent, patients received complete treatment of platinum-based chemotherapy and cytorreductive surgery was performed at the end of the last cycle. If surgery was considered optimal, patients underwent 90 min HIPEC with cisplatin (100mg/m 2 ) at 40-43 C. Patients were followed every 3-months until recurrence or death. Descriptive statitics were performed. Protocol was IRB aproved. Results From January 2015 through July 2019, twenty-five patients with a median age of 54 (range 37-73 years) were included. All patients had a peritoneal cancer index (PCI) >20 at diagnosis. Most patients (92%) had stage IIIC disease at diagnosis and 84% had high grade papillary serous carcinoma. Median CA-125 at diagnosis was 4228 U/mL (range 28-21141 U/mL). Patients received a median of 6 cycles of carboplatin/paclitaxel and 12% developed toxicity grade 3-4. Median PCI at the time of surgery was 8 (range 1-17) and all patients underwent optimal cytorreduction. Median surgical time was 294 min. Three patients (12%) had severe surgical morbidity (Clavien 3-4), with no surgical mortality. Median lenght of hospital stay was 11 days (range 5-42). With a median follow-up of 22 months after cytoreductive surgery and HIPEC, six patients (24%) experienced recurrent disease and two patients died (one of unrelated causes). Half of recurrences ocurred out of peritoneal cavity. Conclusion In patients with very high tumor burden, HIPEC appear to reduce incidence of peritoneal recurrence, but overall recurrence rate seem to be similar to that reported in other series. Background: Current cross-sectional imaging demonstrates poor sensitivity for peritoneal disease which often requires laparotomy/laparoscopy to accurately detect and quantify peritoneal tumor burden. We report interim results of a novel method utilizing high resolution (HR) MRI to detect peritoneal mesothelioma. Methods: Patients with peritoneal mesothelioma undergoing laparoscopy/laparotomy were enrolled in a single arm Phase II prospective clinical trial (NCT03867578). During the exploratory phase, MRI scans obtained from the first 5 patients were used to optimize coil positions, sequence parameters and contrast timing. Novel elements of our finalized MRI protocol included: double dose injection of Dotarem; pre-contrast free breathing HR coronal T2 weighted sequences without fat suppression; and 3D T1 HR coronal sequences, acquired in 3 breath holds at multiple post-contrast time points (range 2-18 minutes) using mDixon technique and focused on the right diaphragm / liver dome. Sensitivity was assessed by a blinded radiologist on the remaining cohort. Post scanning image processing is currently being performed. Results: Ten patients (out of planned 21) with epithelioid type MPM (7 males, median age 57 (range 43-67) years, median BMI 30.4 (range 23.9-38.9) kg/m 2 ) were enrolled between 2/2019-9/2019. MRI was performed 3 (range 1-17) days before laparoscopic (n=6) or open (n=4) surgery. The median intraoperative peritoneal cancer index score was 34 (range 9-39). One patient was excluded from the analysis due to failed laparoscopy. Review by a blinded radiologist yielded a per-region sensitivity of 8/9 (89%) for the right diaphragm region and 6/6 (100%) for the pelvis. During the study period, no procedure-related or contrast-related adverse events were reported. Conclusions: Our HR MRI protocol is tolerable, safe and may increase the diagnostic sensitivity of MPM detection. Final results from this ongoing study will allow to compare sensitivities with standard of care imaging. Introduction: Laparoscopic hyperthermic intraperitoneal chemotherapy (LS-HIPEC) has been proven safe in patients with gastric adenocarcinoma. LS-HIPEC, however, has shown limited success in inducing conversion to negative cytology, making patients eligible for resection, and improving overall survival. Thus, we examined factors associated with improved survival and resection rates. Methods: Prospectively collected data for all patients undergoing LS-HIPEC between June 2014 and November 2018 was analyzed for associations with survival and resection using uni-and multivariate logistic regression, Cox proportional hazards models, and Kaplan-Meier survival functions. Results: A total of 71 patients underwent LS-HIPEC procedures for stage IV gastric adenocarcinoma. Of these, 43 (61%) patients received a 2-drug protocol: mitomycin C and cisplatin, while 27 (39%) were treated with 3 drugs: mitomycin C, cisplatin, and paclitaxel. There were no statistically significant demographic or oncologic differences between groups, though the 3-drug group had lower radiation therapy use (58% vs 15%, p<0.01). On univariate analysis, high histologic grade (Cox hazard [1.56-6.41 ]; p<0.01). On further analysis, 17 (38%) of 45 patients without ascites went on to resection, while none of the patients with ascites were able to proceed to resection (p<0.01). Conclusion: We were unable to demonstrate a benefit for 3-drug LS-HIPEC over the 2-drug regimen on short-term follow-up. The presence of ascites, however, is a newly described prognostic factor. Given the high rate of ascites upon initial laparoscopy, we hope our findings will assist clinicians in identifying patients who are unlikely to proceed to resection and are candidates for novel therapeutic approaches or clinical trials.

Kaplan-Meier survival curve comparing patients with and without ascites at the time of staging laparoscopy. Median survival durations were 9.3 months and 18.2 months respectively. Background: A significant proportion of patients with malignant peritoneal mesothelioma (MPM) carry pathogenic germline mutations (GM), but the surgical phenotype of MPM patients with GM has not been studied previously. We aimed to investigate if MPM patients with GM have distinct intraoperative and baseline characteristics when compared to those without GM. Methods: MPM patients were selected from an ongoing prospective study that conducts germline testing of 85 susceptibility genes. Patients with benign mesothelioma or ongoing genetic testing were excluded. Germline status was correlated with surgical data obtained from a prospectively collected database using univariate analyses. Results: Out of 66 MPM patients enrolled between 2009-2019, 15 GMs (23%) were identified. BAP1 was the most prevalent GM (n=9, 14% of all patients). The remaining genes were WT1, CDKN2A, CHEK2, ATM, SDHA and BRCA2. Surgical procedures (n=63) were performed in 52 patients, the most common of which were cytoreductive surgeries with hyperthermic intraperitoneal chemotherapy (n=52). Baseline characteristics (gender, age, asbestos exposure and proportion of bicavitary disease) were similar between GM and no GM (n=51) groups. Prevalence of other prior cancers was higher among GM patients (69.2% vs. 26.0%, p=0.007). All GM patients had MPMs with epithelioid histology, whereas 5.8% of patients without GM had biphasic type (p=0.3). Distinct baseline and intraoperative characteristics of patients with BAP1 GM, compared to those without BAP1 GM, are presented in Table 1 . Median overall survival did not differ significantly between GM and no GM patients (not reached vs. 44 months, respectively, p=0.74). Conclusions: GM drive tumorigenesis in a substantial proportion of MPM patients. Higher tumor burden, involvement of the diaphragm or multiple peritoneal surfaces, and loss of BAP1 protein on immunohistochemistry are suggestive of BAP1 GM and should prompt genetic testing.

Baseline and intraoperative characteristics of malignant peritoneal mesothelioma patients with BAP1 germline mutation (GM) * Analysis restricted to patients who underwent peritonectomy procedures. ** Diaphragmatic involvement defined as bicavitary disease or full thickness diaphragmatic resection.

Background: We previously reported that surgical palliation maintained patient quality of life (QOL) while improving solid food intake with an acceptable surgical safety at least for the first 3 months after surgery in patients with bowel obstruction caused by peritoneal dissemination of gastric cancer (ASCO-GI 2019). We evaluated the associations of QOL change and improved food intake with overall survival (OS). Methods: Eligibility included (1) no oral intake or liquids only requiring parenteral nutrition (2) aged 20 (3) surgically fit (4) ECOG PS of 0-2 and (5) written IC. Patients underwent resection of small intestine/colon, bypass of small intestine/colon, or ileostomy/colostomy (placement of endoscopic stent was not allowed). Validated instruments (EuroQol-5D and EORTC QLQ-STO22) assessed QOL at baseline, 2 weeks (wks), 1 month (m), and 3 months following the surgical palliation, and postoperative improvement of oral intake was evaluated according to the GOOSS (gastric outlet obstruction scoring system). Univariate and multivariate survival analyses were conducted regarding the QOL changes and improved food intake. Results: Median survival time of the enrolled 63 patients was 6.70 (95% CI: 4.96 -10.28) months. Changes in QOL scores at 1m after surgery had significant impact on OS, but not at 2wks and 3m. Improved oral intake up to GOOSS 3 (low-residues or full diet) had positive impact on OS at 2wks, 1m and 3m postsurgery. In univariate analysis, OS was significantly long in patients with postoperative chemotherapy, better baseline EQ5D score, lower baseline CRP, and improved oral intake. Multivariate analysis selected postoperative chemotherapy, lower baseline CRP, and improved oral intake of GOOSS 3 as independent prognostic factors. Conclusion: In patients who received surgical palliation for bowel obstruction caused by peritoneal dissemination of gastric cancer, improved oral intake could predict better OS. Background: Cytoreductive surgery (CRS -gastrectomy combined with metastasectomy) for non-palliative indications is controversial for patients with metastatic gastric adenocarcinoma (MGA). We hypothesized that CRS in addition to systemic chemotherapy is associated with an improved survival when compared to patients with MGA receiving chemotherapy alone. Methods: Patients with MGA who received systemic chemotherapy between 2004-2016 were identified using the National Cancer Database (NCDB). Nearest neighbor 1:1 propensity score matching of demographic, tumor-related and treatment-related factors was used to create comparable groups. Overall survival (OS) was compared between subgroups using Kaplan-Meier analyses. Immortal bias analysis was performed among those that survived at least 90 days. Results: We identified 29,728 chemotherapy-treated patients who were divided into 4 subgroups: No surgery (NS, n=25,690), metastasectomy alone (n=1170), gastrectomy alone (n=2248) and CRS (n=620) with a median OS of 8.6, 10.9, 14.8 and 16.3 months, respectively (p<0.001 BACKGROUND Metastatic gastric adenocarcinoma is associated with poor long-term survival. Palliative chemotherapy is the mainstay of treatment, but tolerance is often limited due to tumor-related symptoms. The aim of this study was to evaluate the role of selective palliative surgery (PS) in incurable gastric cancer. METHODS A prospectively-maintained database was queried to identify all patients with stage IV and unresectable locally advanced (T4b) gastric adenocarcinoma treated at a single high-volume centre from 03/2006-01/2019. Patient characteristics and treatment outcomes were confirmed by consulting the electronic medical record. Results are reported as median (interquartile range). Mann-Whitney U tests were used to compare outcomes between PS and no surgery (NS) groups. RESULTS Over the study period, 129 patients met the inclusion criteria [age: 62(54-73) years; 79(61%) male]. Sites of metastases were: peritoneum (80; 62.0%), non-regional lymph nodes (37; 28.7%) and solid organ (22; 17.1%). Twelve patients (9.3%) presented with unresectable locally advanced (T4b) disease. Fifty-four (41.9%) patients underwent palliative surgery which involved: total gastrectomy (24; 44.4%), subtotal gastrectomy (18; 33.3%); proximal gastrectomy (2; 3.7%); gastrojejunostomy bypass (4; 7.4%) and feeding jejunostomy (6; 11.1%). Reasons for surgical intervention were: palliation of obstruction (46; 85.2%), treatment of uncontrolled bleeding (7; 13.0%) and perforation (1; 1.9%). Surgical complications within 30-days occurred in 29 (53.7%) cases, of which 10 (18.5%) were severe (CD grade^3). Mortality within 30-days occurred in 1 case (1.9%). Post-surgery, 31 (57.4%) went on to receive palliative chemotherapy and 25 (46.3%) were alive at one year. For the PS and NS groups respectively, median OS was 9.8 vs 9.4 months (p=0.43) and median PFS was 7.3 vs 6.7 months (p=0.08) after a follow up of 7.3 (4.7-13.1) vs 7.8 (2.6-13.4 ) months (p=0.46). CONCLUSION Targeted surgical intervention for incurable gastric cancer can be used to palliate symptoms and facilitate continuation of systemic therapy with acceptable risks and comparable survival to non-operative management but should be reserved for carefully selected patients.

Clinicopathologic Characteristics of Oligometastases from Esophageal Cancer and Long-term Outcomes of Resection Y. Ohkura,* H. Udagawa, M. Ueno. Gastroenterological surgery, Toranomon hospital, Tokyo, Japan.

Background: Recurrent esophageal cancer after radical therapy is usually thought to be incurable and treated with a palliative-intent systemic therapy. However, it is empirically known that surgical resection may be effective in selected patients, though there has been no consensus on the efficacy of surgery for recurrent esophageal cancer. This study sought to identify a group of patients in whom surgical resection is thought to be effective. Methods: 206 patients who had recurrence after radical therapy for esophageal cancer at a single center were included in this study. Prognostic factors after recurrence were identified and efficacy of surgery was analyzed according to whether the recurrent lesions were oligometastases (i.e., 5 lesions in a single domain), or not. Results: In multivariate analysis, oligometastatic presentation was the only factor associated with survival after recurrence (hazard ratio 6.29; 95% confidence interval, 4.10-9.71). Actuarial survival rates in patients with oligometastases were 59.5% at 3 years and 51.7% at 5 years. Survival rates at 3 and 5 years were significantly higher in patients who underwent resection (64.3% and 55.6%, respectively) than in those who did not (both 10.0%) and in patients with multiple metastases (9.8% and 0%, respectively). The survival rates of the patients with oligometastases without resection were comparably low to the patients with multiple metastases Conclusion: Oligometastatic presentation at recurrence was associated with better survival outcomes among patients developing recurrence after radical treatment for esophageal cancer, and surgical resection could be a choice of treatment in this group of patients. Over one third of gastric cancer patients in the United States have stage IV disease at the time of diagnosis, and the prognosis is poor, with a 5-year overall survival (OS) rate of only 5.3%. This emphasizes the need to improve multimodality treatment strategies. Surgical resection for patients with asymptomatic metastatic gastric cancer (mGC) is controversial. However, recent retrospective data implicates a potential survival benefit of surgery in select patients with mGC. We evaluated the relationship of surgery of both the primary and metastatic sites to OS in patients with mGC. We conducted a retrospective review of patients receiving chemotherapy for metastatic gastric adenocarcinoma using the National Cancer Database (2004) (2005) (2006) (2007) (2008) (2009) (2010) (2011) (2012) (2013) (2014) (2015) . Patients were grouped according to surgery of: 1) the primary site (PS) only, 2) primary and distant sites (PDS), 3) distant site only (DS), or 4) no surgery (NS). A propensity score adjustment was used to compare all clinical and demographic baseline characteristics. Overall survival was compared after multivariate regression using Cox proportional hazard models. We identified 18,772 patients with the following in each group: 962 PS, 380 PDS, 984 DS, and 16,446 NS. Surgery of the primary site or primary plus distant sites was associated with improved OS as compared to no surgery (HRs 0.489, 0.583, p<0.001) (Figure 1 ). Median OS (months) was 15.8 and 15.9 for the PS and PDS groups versus 8.6 for NS patients. There was no difference in OS between patients who had surgery of the primary site alone versus the primary and distant sites, and no benefit of distant site surgery only versus no surgery. Gastrectomy with or without metastasectomy was associated with improved survival in stage IV gastric cancer patients receiving chemotherapy. These data suggest that surgery of the primary site may improve overall outcomes in these patients; further research should study criteria for patient selection and generate prospective, randomized analyses. Introductions For treatment of stage IV gastric cancer, conversion surgery (CVS) after chemotherapy has received attention as a treatment strategy instead of combination chemotherapy alone. However, the clinical value of CVSs and evidence for prognostic factors remain controversial. This study aimed to evaluate the prognostic value of ypTNM stage and the oncologic outcomes in patients with locally advanced but unresectable or stage IV gastric cancer who underwent conversion surgery. Methods Clinicopathologic findings and oncologic outcomes of 116 patients who underwent CVS with curative intent after combination chemotherapy between January 2000 and December 2015 were retrospectively reviewed. Results Twenty six patients (22.4%) underwent combined resection of another organ and 12 patients received paraaortic lymphadenectomy (10.3%). Pathologic complete remission (CR) was confirmed in 11 cases (9.5%). Stage 3 tumor (34.5%) was the most common followed by stage 2 and 1. The median overall survival and disease free survival time was 35.0 and 21.3 months, respectively. In multivariable analysis, ypTNM stage was a sole independent prognostic factor (P = 0.042). Kaplan-Meier curves showed that the 3 year OS rate of patients with pathologic CR and those with CR of primary tumor but residual node metastasis was 81.8% and 80.0%, respectively. Overall survival of stage 1 patients was 65.8% and followed by stage 2 of 49.8% and stage 3 of 36.3%. Conclusions ypTNM staging is a significant prognostic factor in patients who underwent CVS for localized unresectable or stage IV gastric cancers.

Prognostic factors for disease-free survival after conversion surgery in patients with irresectable or stage IV gastric cancers HR and CI refer to a hazard ratio and confidence interval, respectively. * OI, invasion to an unresectable organ; LN, distant lymph node metastasis; PS, peritoneal seeding; SO, solid organ metastasis † TNM stage was based on the American Joint Committee on Cancer 7th edition. ‡ NRT and NRT N(+) refer to no residual tumor and those with residual metastatic lymph node, respectively. Purpose Identifying and preserving the parathyroid gland during thyroidectomy is crucial. The purpose of this study was to generate a preliminary data using a parathyroid localization algorithm developed through deep learning technology. Methods Both upper and lower parathyroids were searched for and high quality video clips were recorded using a commercial camera and a 10mm 30 degrees' endoscopic camera. Still cut images of confirmed parathyroids were prepared for deep learning and region of interest was identified. After the preprocessing stage, deep learning was performed using the Retinanet deep learning model based on ResNet152 backbone. The learning model was based on Deconvolution Network architecture and trained with 6~12 batch sizes and 100 epochs. True object detection condition was defined as intersection over union (area of overlap/area of Union) 0.3 and probability 0.5. A 10-fold cross validation was performed for learning.(Fig1) The deep learning algorithm was developed for all images (camera, 30 + endoscope, 45), camera only and endoscope only and their true positive, false negative, false positive numbers and sensitivity were calculated and compared. Results Based on still cut frame numbers, the camera + endoscope algorithm produced 40501 true positive, 42696 false negative and 24864 false positive detections which resulted in a sensitivity of 0.49. The camera algorithm made 25041 true positive, 26098 false negative, and 18679 false positive detections demonstrating a 0.49 sensitivity. Finally, the endoscope model produced 10726 true positive, 21332 false negative and 5079 false positive detections translating into a 0.33 sensitivity. The camera + endoscope algorithm produced 7 true positive and false negative cases which resulted in a sensitivity of 0.95. The camera algorithm detected 27 true positive and 3 false negative cases which translates to a 0.9 sensitivity. Finally, the endoscope model produced 41 true positive and 4 false negative detections which translates to a 0.91 sensitivity. Conclusion The preliminary results demonstrate feasible results. Further learning is necessary to fine-tune the performance of the parathyroid detection algorithm.

Detecting Introduction: Indeterminate cytology after fine needle aspiration (FNA) of thyroid nodules remains a therapeutic challenge. NCCN guidelines recommend molecular analysis of biopsy specimens as it has improved diagnostic accuracy compared to standard histologic assessment and can help to prevent unnecessary surgery. The two most widely utilized products, Afirma and Thyroseq, have both undergone significant updates in recent years, yet no study has compared the accuracy of these products. Methods: An online database search was performed for all articles in which the Afirma Genomic Sequence Classifier (GSC) and/or the Thyroseq v3 Genomic Classifier (TGC) were used to evaluate indeterminate thyroid FNA specimens (Bethesda III/IV lesions). Patients had to subsequently undergo surgical resection for validation of FNA results. Sensitivity and specificity analyses were performed with calculation of 95% confidence intervals. Results: Five studies totaling 694 nodules (Afirma: 437, Thyroseq: 257) were eligible for meta-analysis with four evaluating Afirma and one for Thyroseq. Overall, there was a greater percentage of female patients compared to male patients (Afirma = 76%, Thyroseq = 80%). Test performance for thyroid nodules categorized as Bethesda III and IV were pooled and compared. Afirma demonstrated a sensitivity range of 91% (95% CI: 0.79-0.96) -97% (95% CI: 0.76-1.00) versus a sensitivity of 94% (95% CI: 0.86-0.98) for Thyroseq. Afirma specificity ranged from 20% (95% CI: 0.08-0.42) -68% (95% CI: 0.60-0.75) and there was statistical heterogeneity between studies (p<0.01). This was compared to a specificity of 82% (95% CI: 0.75-0.87) for Thyroseq [ Fig. 1 ]. Conclusion: The Afirma GSC and Thyroseq V3 products have been available for several years, but have not been compared head to head. Meta-analysis demonstrated similarly high sensitivities, but decreased specificity for Afirma compared to Thyroseq. Both Afirma and Thyroseq remain effective tests for ruling out thyroid cancer in indeterminate biopsies. Although Thyroseq demonstrated higher specificity, both tests remain limited in their ability to identify patients with cancer. Background: Current guidelines recommend an appendectomy (AP) for carcinoid tumors (AC) 2cm in size. However the management of tumors with lymphovascular invasion (LVI) remains controversial. We evaluated the prognostic value of LVI in AC 2cm in size. Design: The National Cancer Database was queried for all patients with AC 2cm in size who underwent either an AP or a right hemicolectomy (RHC) One of the challenges in transitioning from in vitro to in vivo models is studying drug pharmacodynamics at the cellular level in a living organism. Intravital imaging has proven to be a vital tool for studying biological processes such as cancer progression and metastasis by capturing sub-cellular, high-resolution images. We have developed an implantable imaging window and surgical protocol to study orthotopically implanted thyroid tumors, live and in real time. We utilize this technique to study drug uptake, delivery, and tumor response in anaplastic thyroid tumors. SV-129 mice were orthotopically injected with N794, GFP+ syngeneic murine anaplastic thyroid cancer (ATC) cells. Tumors were allowed to grow and at day 7-10 post-injection, an imaging window was surgically implanted allowing intravital multiphoton microscopy of the live tumor. Mice were sacrificed 72 hours post-window implantation and tumor specimens were collected to evaluate for signs of window-induced inflammation. Rag2 -/mice were orthotopically injected with human BHT-101 ATC cells constitutively labeled with a red fluorescent protein (RFP) and expressing a biosensor for apoptosis (FlipGFP). Tumors were allowed to grow until palpable below the skin, and a thyroid window was implanted. Single-cell resolution intravital imaging was performed to determine the uptake of and impact of systemic chemotherapy (doxorubicin). Optimal window implantation involves fully exposing the anterior aspect of the thyroid tumor, suturing the right salivary gland to prevent window obstruction, and ensuring best possible contact of the tumor with the window. Positioning and fixturing the animal on the microscope is critical to obtaining stable, high resolution images. Windows can be successfully implanted when the tumor is palpable. Window bearing mice remain viable for up to 3 days post implantation allowing studies of drug uptake, delivery, and effect to be adequately evaluated. Surgical implantation of a thyroid imaging window is a novel technique that allows in vivo studies of thyroid tumors. Potential applications include studying drug delivery and uptake as well as bridging the gap between in vitro assays and in vivo results. Background: We hypothesized that there are geographic areas of increased cancer incidence in Alberta, and that these are associated with high densities of oil and gas(O+G) infrastructure. Our objective was to describe the relationship between O+G infrastructure and incidence of solid tumours on a population level. Methods: We analyzed all patients >18 years old with urological, breast, upper GI, colorectal, head and neck, hepatobiliary, lung, melanoma, and prostate cancers identified from the Alberta Cancer Registry from 2004-2016. Locations of active and orphan O+G sites were obtained from the Alberta Energy Regulator and Orphan Well Association. Orphan sites have no entity responsible for their maintenance. Arcmap 10.6.1(ESRI, Toronto, Ontario) was used to calculate the distribution of O+G sites in each census distribution area (DA). Patient residence at diagnosis was defined by postal code. Incidence of cancer per DA was calculated and standardized. Poisson regression was done on O+G site density as a categorical variable with cutoffs of 0 and 30 wells/100km 2 , compared to areas with 0 sites. Results: 125,316 patients were identified in the study timeframe;58,243 (46.5%) were female, mean age 65.6 years. Breast (22%) and prostate (19.8%) cancers were most common. Mortality was 36.5% after a median of 30 months follow up . For categorical density of active O+G sites, RR was 1.02 for 0-30 sites/100km 2 (p=0.53, 95% CI=0.95-1.11) and 1.15 for >30 sites/100km2 (p<0.0001, 95%CI=1.11-1.2). For orphan sites, 0-30 sites RR was 1.25 (p<0.0001, 95%CI=1.16-1.36) and 1.01 (p=0.97, 95%CI=0.7-1.45) for >30 sites. For all O+G sites, RR for 0-30 sites was 1.03 (p=0.4328, 95%CI=0.95-1.11) and 1.15 (p<0.0001, 95%CI=1.11-1.2) for >30 sites. Conclusion: We report a statistically significant correlation between O+G infrastructure density and solid tumour incidence in Alberta. To our knowledge this is the first population-level study to observe that active and orphan O+G sites are associated with increased risk of solid tumours. This finding may inform policy on remediation and cancer prevention. We are unable to comment on a mechanism to account for this observation. Introduction: Disparities in cancer outcomes for minority and socio-economically disadvantaged populations have been demonstrated for patients with potentially curable cancers. These outcome disparities correlate with presentation at later stages, lower rates of cancer screening, difficulty in accessing high quality care, and receiving non-guideline concordant care. These disparities in oncological treatment delivery are also seen in Stage IV cancer. However, less is known about how treatment disparities impact outcomes for patients with Stage IV cancer. Methods: Patients with Stage IV pancreatic, colorectal, lung, breast, and prostate cancer were identified from 2004-2015 in the National Cancer Database. Patients' demographic characteristics and likelihood of cancer treatment modalities were compared across cancer types. Cox proportional hazard models were used to quantify how baseline demographic characteristics and treatments received impacted overall survival. Results: 903,151 patients were compared across the five cancer types. Patients who were younger, non-Hispanic white, with private insurance, higher income (>$46,000 annual), or who received their care at an academic center were more likely to receive surgery, chemotherapy, and/or radiation therapy for their Stage IV pancreatic, colorectal, lung, breast, or prostate cancer (p<0.001). Overall, Black patients, those with Medicare, Medicaid, or no insurance, as well as patients with lower annual income have lower survival rates across all five cancer subtypes (p<0.001). On multivariable Cox models, surgery, radiation and chemotherapy attenuated, but did not completely eliminate this worse survival across the five cancer types (p<0.001). Conclusion: Differences in the delivery of life-prolonging therapy result in disparities in Stage IV cancer survival in the United States. Black patients, uninsured patients, and those with lower annual income have lower survival rates. Differences in the treatments received explained some, but not all of this discrepancy, suggesting cancer treatment standardization could attenuate, but likely not eliminate Stage IV cancer survival disparities. Introduction: Regionalization of care to specialty centers results in improved procedural and survival outcomes. There is significant geographic diversity in the Veterans Integrated Service Networks that mirror the US. Our high complexity center serves 10 states and 540,000 square miles, with ~1,000,000 patients. Rural location has been linked to adverse cancer outcomes due to lack of specialists and distance to treatment centers. We hypothesize that VA patients from rural locations have decreased utilization of treatment and worse outcomes for hepatocellular carcinoma (HCC). Methods: Data was prospectively collected on patients referred to the liver multidisciplinary clinic at our VA from 2006 to 2019. Data include age, sex, date of diagnosis, receipt of liver directed treatment (TACE, Ablation, Resection, Transplant) and survival from diagnosis. Patients were stratified as urban or rural by Federal Office of Health Policy definitions. Fishers exact test was used to determine the effect of rural and urban location on utilization of treatment. Mantel-Cox log rank test was used to calculate survival curves. Results: 742 patients were seen during this time; 371 patients were diagnosed with HCC by the treatment team. 284 patients (76.5%) were urban and 87 rural (23.5%). The utilization of treatment was 77% for urban vs 66% in the rural patients (p<0.05). Median survival between urban and rural did not differ (859 days for urban vs 649 days for rural, p NS. Patients that underwent treatment had significantly increased median survival (939 days) compared to those that did not (243 days, p<0.0001) regardless of location. Conclusions: Rural location is a risk factor for decreased utilization of treatment for HCC. Survival is significantly decreased in patients who do not undergo treatment for HCC. These data demonstrate that rural patients referred for HCC evaluation have a decreased utilization of care compared to urban counterparts. However, when treatment is received survival is not different. These data warrant additional studies to examine decreased utilization of care in the rural population.

Objective: While hepatocellular carcinoma (HCC) is ideally diagnosed outpatient by screening at-risk patients with cirrhosis, many are diagnosed in the Emergency Department (ED) due to undiagnosed liver disease and/or limited access to healthcare. We explored factors associated with the diagnosis of HCC in the ED to identify patients who may benefit from improved access to care. Methods: All patients with HCC diagnosed in the ED or outpatient setting[Primary Care Physician (PCP) or hepatologist] from the US Safety-Net Collaborative(2012-14) database were included. Sociodemographics, symptomatology, tumor characteristics and healthcare access were analyzed. Multivariable regression identified factors associated with being diagnosed with HCC in the ED. Results: Among 1,620 patients, median age was 60, and 76% were male. 68% were diagnosed outpatient and 32% in the ED. ED patients were more likely male, Black/Hispanic and uninsured (Table 1) . ED patients were less likely to receive HCV treatment, have a PCP, have seen a PCP or hepatologist in the last year or have a patient navigator (PN). ED patients presented more often with decompensated liver disease(ascites and variceal bleeding), more symptoms(abdominal pain and weight loss), advanced clinical stage and more aggressive features(nodal involvement and macrovascular invasion). On multivariable analysis, independent predictors for presentation to the ED were male[OR1.5, 95%CI(1. The sociodemographic and clinical profile of patients diagnosed with HCC in the ED varies significantly from those diagnosed in an outpatient setting. Those diagnosed in the ED were more likely a racial/ethnic minority, uninsured, and have limited access to healthcare. This is the first multi-institutional study to reveal disparities in presentation at the time of HCC diagnosis. It highlights the importance of improved screening and access to care in an already vulnerable population. Introduction: Guidelines recommend hepatitis B (HBV) testing in individuals from endemic areas, and if positive, screening for hepatocellular carcinoma (HCC). While screening programs are well established in the Asian immigrant population in New York City (NYC), less is known about the characteristics of HBV/HCC among the African immigrant community. Methods: A retrospective review was performed of HCC cases from 2005-2018 at our institution. Country of origin was not documented in the electronic medical record; therefore, African immigrant status was approximated using self-identified race/ethnicity, positive HBV status, and an online registry to determine country of origin based on last name. Surnames with the greatest prevalence or density in an African country were considered. Results: Among 4,400 patients with HCC, 472 identified as non-Hispanic black; of these, 86 were HBV+. Based on surname, it was estimated that 33 individuals were likely immigrants from Africa. In this group, median age at HCC diagnosis was 48 years (IQR 43-55). In patients with a known date of HBV diagnosis (n=24), 17 (71%) were unaware of their HBV status when they presented with HCC, and only 5 patients (21%) were on anti-viral treatment. Zero patients were diagnosed with HCC through routine screening. Evaluation of symptoms led to HCC diagnosis in 66% of patients; in 17%, tumors were detected incidentally on imaging obtained for unrelated complaints. Twelve patients (36%) underwent resection or transplantation; the remaining 64% were non-operable. Of the 26 patients with follow-up data, 18 (69%) died of disease or were critically ill at last documented encounter; of these, 14 (77%) died within one year of HCC diagnosis. Conclusion: Most African immigrants in NYC with HBV/HCC were unaware of their hepatitis status. No patients were enrolled in routine HCC screening; the majority were diagnosed based on imaging obtained for symptoms. Most individuals presented with inoperable disease, and the majority died within months of diagnosis. African immigrants are at high risk of both HBV and HCC-related mortality and would benefit from enhanced efforts in education and advocacy. Introduction: In low-and-middle income countries, travel time is a significant determinant of access to cancer care. This has a disproportionate effect on poor households. We used a geoinformation system model to map physical access to Nigeria's comprehensive cancer centers among households living on less than $2 USD/day. Methods: Population-level travel times to eight comprehensive cancer centers, outlined in the National Cancer Control Plan (NCCP) (2018-2022), were estimated using a cost-distance model. The model was constructed using open source road infrastructure data and travel speeds verified by the authors. Geolocated population estimates for households living on <$2 USD/day were merged with the cost-distance model to calculate travel times to the nearest comprehensive cancer center. Results: At four-hours of one-way, continuous vehicular travel, 80,393,235 Nigerians (66.6%) living on less than $2 USD/day have access to a public comprehensive cancer center. At two-hours, the proportion of individuals with access decreases by 34.7%. There is significant variability in access between geopolitical zones (p<0.001). The North East has the lowest access at four-hours (31.5%) compared to 85.0% in the South East. Conclusion: The Nigerian NCCP supports investment in eight public comprehensive cancer centers. Strengthening these centers will allow the majority of poor households in Nigeria to physically access a full complement of multidisciplinary care within a reasonable timeframe. This population is particularly sensitive to regressive out-of-pocket costs, such as those associated with travel. Heterogeneous population-level access between the North and South has the potential to produce divergent outcomes for cancer patients of similar need. This must be considered as ongoing investments are made in this rapidly growing cancer system.

Cancer Care Delivery for Incarcerated Populations G.G. Kasumova,* A.P. Loehrer. Surgery, Dartmouth-Hitchcock Medical Center, Lebanon, NH. Introduction Over 2-million Americans are incarcerated in the United States at any given time and this incarcerated status has been associated with considerable disparities in overall health and treatment for infectious disease, mental health, and substance use disorders. However, little is known about treatment of operable (and potentially curable) cancer diagnoses. Methods This cohort study uses the California Office of Statewide Health Planning and Development publicly-available database to evaluate all hospital admissions for cancer of the lung, hepatobiliary system, colon, rectum and genitourinary system between 2012 and 2014. ICD-9 diagnosis and procedure codes were used to determine oncologic conditions and procedural intervention during hospitalization. Our primary outcomes were rates of admission and procedures for cancer for incarcerated individuals. The population of incarcerated individuals in California was determined using United States Bureau of Justice Statistics data. Secondary outcomes include odds of undergoing surgical intervention and difference in length of stay compared to non-incarcerated patients. Results We identified admissions with cancer for 1,515 incarcerated and 391,098 non-incarcerated patients. Incarcerated patients were significantly more likely to have unscheduled admissions for all reasons (75.6% vs. 70.7%; p<0.001) and for surgical admissions (41.9% vs 18.4%; p<0.001). Incarcerated patients had considerable comorbidities including Hepatitis B or C virus (29.6%), cirrhosis (19.2%), and substance abuse (18.7%). Table demonstrates rates of select cancer diagnoses and procedures as well as odds of receiving operation at time of admission and length of stay compared to non-incarcerated patients. Conclusions In a large and aging population of incarcerated individuals, admissions with oncologic diagnoses are common, largely with lower odds of receiving surgery. Incarcerated patients also had longer postoperative length of stay. Additional studies are needed to determine the quality of complex, multidisciplinary cancer care and identify opportunities for improved cancer care for this vulnerable group. Introduction: Previously we have shown the "tweet" to be a valid tool for reflection among trainees. From these tweets, a dominant experience was exposure to patients at end-of-life. To provide better context and training for these impactful patient encounters, we incorporated didactic sessions into the core curriculum, testing if students would be better equipped to identify barriers to providing palliative care for advanced cancer patients. Methods: During the surgical clerkship, all third-year students were asked to "tweet" a reflection following their end-of-life didactic session, with maximum characters set at 280. With IRB approval, these were coded by two surgical residents, an education coordinator, and a PhD education expert. Themes from 126 tweets from 126 students were identified across three broad categories -provider, patient, and system. Sub-categories were a combination of both proscribed and a priori features. Results: During the 2018-2019 academic year, 110 of 126 students successfully completed the assigned task by identifying at least one barrier to palliative care. Of those, 13 out of 110 (11%) identified burnout and time constraints as significant factors. Cultural differences including language barrier and approach to life-sustaining measures were identified in 11 (10%) the tweets. The chief concern among medical students was their own emotional reaction to the prospect of delivering bad news to a patient --46 students directly cited memories and anxieties interfering with the ability to communicate effectively with the patient. Discussion: Students identified barriers in a majority of cases, showing that they can be critical thinkers and assets to the surgical team. Burnout, cultural disparities and emotional anxieties are key hurdles to providing palliative surgical care to cancer patients. Extrapolation of this tool's potential can be extended to residents and surgical oncology fellows. We herein demonstrate the effectiveness of the "tweet" reflection in recognizing barriers to surgical care at end-of-life, and in fostering trainees to become insightful about these significant patient encounters. Background: SEER data are widely used to study rural-urban disparities. However, no studies have directly assessed how well the rural areas covered by SEER represent the broader rural United States. We hypothesized that rural SEER (R-SEER) areas would differ from rural non-SEER (R-nSEER) areas with respect to underlying measures of sociodemographics, health behaviors, cancer incidence, and health care access. Methods: The Area Health Resource File, County Health Rankings, and NIH State Cancer Profiles tool were used to calculate population adjusted county level sociodemographics (age, race/ ethnicity, sex, poverty, unemployment), health behaviors (smoking, obesity, physical inactivity, alcohol use), health access measures (uninsured, physician density, breast/colorectal screening), and all cause age adjusted cancer incidence. Driving time from each census tract to nearest Commission on Cancer certified facility was quantified using ArcGIS. R-SEER and R-nSEER counties were compared using standard differences. Driving time was log-transformed and compared across R-SEER and R-nSEER areas using linear regression and controlling for region. Results: R-SEER and R-nSEER counties were similar with respect to age, race, sex, poverty, health behaviors, provider density, and cancer screening. Cancer incidence was similar across R-SEER and R-nSEER for all races/ethnicities. Incidence was higher in R-SEER vs R-nSEER for White, Hispanic, and Asian patients, but similar for Black patients. Overall unadjusted median travel time was 25 mins (IQR: 13-46) in R-SEER and 28 mins (IQR: 14-47) in R-nSEER census tracts. Linear modeling showed shorter travel times across all levels of rurality in R-SEER vs R-nSEER census tracts when controlling for region (Large Rural: 32% shorter 95% CI: 26-38%; Small Rural: 30% shorter 95% CI: 19-39%; Isolated Rural: 19% shorter 95% CI: 7.3-29%). Conclusions: The rural population covered by SEER data are comparable to R-nSEER areas and should be representative of the broader rural population. However, when using travel time as a measure of health access, studies using SEER should recognize travel times in R-SEER areas underestimate travel times for the broader rural population in the US. Intro: Low health literacy (HL) is associated with increased resource utilization and poorer outcomes in medical and surgical patients (pts) with various diseases. Surgery for cancer is an especially complex experience of shared decision-making and therapy options. This study was designed to determine the feasibility of measuring HL in the post-op setting, the prevalence of low HL for surgical cancer pts at an NCI-designated CCC, and correlations of HL with sociodemographic variables and clinical outcomes Methods: Pts admitted post-op (n=155) for GI cancers (58%), bladder cancers (19%), or sarcomas (23%) from 4/19-9/19 completed surveys prior to discharge. Pts reported HL (via a validated 4-question Brief Health Literacy Screen; 4-12=low, 13-16=marginal, 17-20=adequate) and sociodemographic (SD) variables (age, race, education, income, etc). Clinical data (Charlson score, surgery, Clavien-Dindo, length of stay (LOS), ED visits, readmissions, etc) were abstracted 90 days post-op. Pearson correlations or Wilcoxon-Mann-Whitney tests examined associations of HL with SD and clinical variables; clinical variables significantly associated with HL were examined via multivariate (MV) linear regression Results: Feasibility was shown by the number of pts consenting to participate who completed the survey (100%). 20 eligible pts declined participation. In this sample, HL was low in 17%, marginal in 28%, and adequate in 55%. Low HL was associated with older age (p=0.015), and higher Charlson score (p=0.003). Longer hospital LOS was the only outcome variable associated with low HL (p=0.035). In MV analysis, HL did not predict LOS (p=0.09); instead, age was associated with LOS (p=0.03). Other SD and clinical variables did not reach significance (P>0.06) Conclusions: Post-operative survey administration was feasible in this study. In this sample, HL was "low" in 17% and "marginal" in 28% of pts receiving complex cancer surgery at this NCI-designated CCC. Results suggest that higher age, higher Charlson scores, and longer LOS are observed in pts with low HL. Targeted interventions for pts with low HL may decrease resource utilization

Patients L. Abraham, 1 * E. De Jager, 2 Introduction Socioeconomic and racial disparities exist in surgical outcomes for colon cancer patients. Few studies, however, have examined the role of median household income (MHI) and insurance in patients under the age of 65. This study aims to determine the effect of insurance group and neighborhood income on the presentation of colon cancer patients undergoing surgery. Methods Conducted a retrospective review of the National Inpatient Sample from 2000 to 2014. Patients selected were between 50 and less than 65 years of age, underwent colorectal surgery and had a primary diagnosis of colon cancer. Patients were stratified based on insurance group (Medicaid, commercial insurance, and uninsured) and MHI. The primary outcome was presentation for non-elective colorectal surgery. Multivariate logistic regression was performed to determine the effect of insurance and MHI on the odds of presenting with the outcome, while adjusting for age, gender, race, perforation/obstruction/metastatic disease, hospital teaching status, and co-morbidities. Results 136,089 patients met our inclusion criteria. 50.5% were male and median age was 58 (SD4.23). The distribution by insurance group was 17.4% Medicaid, 75% commercial insurance, and 7.6% uninsured. For MHI, the distribution by quartile was 22.6% lowest, 24.7% low middle, 24.8% low high, and 27.9% highest. The overall percentage of patients that required non-elective colorectal surgery was 57.8%. The percentage of patients undergoing non-elective surgery varied by insurance group (22.7% Medicaid, 68.1% commercial, 9.2% uninsured [p<0.01] ) and income quartile (26.5% lowest, 24.9% low middle, 24% low high, and 24.6% highest [p<0.01]). In the adjusted analysis, the odds of non-elective surgery varied by both insurance group and household income ( Figure 1 ). Conclusion Our study demonstrates that uninsured patients and those with Medicaid are more likely to present in nonelective status when compared to the commercial group. It also demonstrates that in patients under 65, as MHI increases within each insurance group, there is a decreased likelihood of requiring non-elective surgery for colon cancer. Introduction: Rising healthcare costs have resulted in an increase in medically related bankruptcies. The objective of this study is to understand the relationship between neighborhood bankruptcy incidence, socioeconomic status (SES) and overall mortality among breast, colon, lung and prostate cancer patients in Indiana. Methods: Patients in the Indiana Cancer Registry diagnosed with breast, lung, colon and prostate cancer from 01/01/2007-12/31/2014 were identified. Neighborhood bankruptcy incidence was calculated by dividing the number of bankruptcies filed in a given zip code by the total population (census 2010) in that zip code. Using zip codes linked to county level data on occupation, income, poverty, wealth, education and crowding, an Agency for Healthcare Research Quality (AHRQ) neighborhood SES index was calculated. The data was divided into 3 groups based on bankruptcy incidence (no bankruptcy, low = 1-4%, high= 5%) for univariate analysis. The log rank test and Cox proportional hazards models were used to evaluate overall mortality. Results: The study included 165,287 patients. Every 10 units increase in SES status reduced the rate of bankruptcy by 2.02 per 1000 population (correlation coefficient = -0.1028; p<0.001). Unadjusted analysis showed residence in a high bankruptcy neighborhood was associated with a higher overall mortality than a low or no bankruptcy neighborhood (p<0.0001). Multivariable analysis showed no association between neighborhood bankruptcy incidence and overall mortality (no bankruptcy-ref, low HR 1.05 95%CI (0.96-1.15), high HR 0.99 95% CI (0.72-1.35). However, low SES (HR 1.18, 95%CI 1.08-1.28) was associated with an increase in overall mortality compared to high SES. On subset analysis, only lung cancer patients showed an association between neighborhood bankruptcy incidence (low bankruptcy HR 1.15 95%CI (1.03-1.28), no bankruptcy-ref) and overall mortality. Conclusion: In our ecological study of Indiana cancer patients with breast, colon, lung or prostate cancer, living in a low SES neighborhood was associated with an increased incidence of bankruptcy and worse overall mortality. Introduction Quality and practice guidelines from the American College of Surgeons Commission on Cancer (CoC) and the National Comprehensive Cancer Network (NCCN) recommend neoadjuvant chemoradiation (NCR) for locally advanced rectal cancer (LARC). We examined guideline adherence in a healthcare system serving a region with significant socioeconomic disparities and poor cancer outcomes. Methods We performed a retrospective analysis of patients from 2005-2014 with stage II and III rectal cancer in our local 5-hospital healthcare system in the Mid-South region of the US. We examined the associations between guideline adherence and patient demographic, socioeconomic, and clinicopathologic data. Results Of 157 stage II/III RC patients, 96 (61.1%) received NCR. By univariate analysis, factors associated with receipt of NCR included white race versus non-white (OR=2.14, p=0.024), private insurance versus no or public insurance (OR=2.70, p=0.005), employed status versus unemployed (OR=2.29, p=0.031), age at diagnosis (OR=0.74, p=0.032), and appropriate local staging with EUS or MRI (OR=6.67, p=<0.0001). Those who were diagnosed and treated later in the study period were more likely to get NCR (OR per 1 year=1.20, p=0.006). In addition, receipt of NCR was protective against death at time of follow up (OR=0.41, p=0.009). By multivariate analysis, those with private insurance (OR=2.51, p=0.023), younger age at diagnosis (OR per 10 years=1.39, p=0.048) and with appropriate local staging (OR=6.67, p<0.0001) were more likely to have received NCR. Conclusion Guideline adherence for LARC in our system, which serves a population with socioeconomic disparities, is relatively low. Sociodemographic factors such as employment and race influenced receipt of NCR, while insurance status and age were independent determinants. Despite some increase in compliance over time, adherence to NCR for LARC remains lower than expected and represents an important target for improvement efforts. Introduction: Cytoreductive surgery and hyperthermic intraperitoneal chemotherapy (CRS-HIPEC) can offer significant survival advantage over systemic therapy alone for select patients with colorectal peritoneal metastases (CRPM). Low socioeconomic status (SES) has been correlated with disparities in access to care for this and other malignancies. We analyzed the impact of SES on operative management and survival at a tertiary CRS-HIPEC center. Methods: We conducted a retrospective cohort study examining all patients who underwent CRS-HIPEC for CRPM from 2000-2018. Patients were grouped according to their SES based upon geographic data. Baseline characteristics, perioperative outcomes, and survival were examined between groups. Results: 226 patients were identified: 107 (47%) low and 119 (53%) high SES patients. High SES patients were younger (52 vs 57 years, p=0.005) and more likely to be white (95.0% vs 91.6%, p=0.064), employed (65% vs 47%, p<0.05), and privately insured (83% vs 57%, p<0.001). They also traveled significantly further for treatment (310 vs 83 miles, p<0.01) and had lower burden of comorbidities and frailty (p=0.001). There were no differences in rates of previous resection or receipt of chemotherapy, but low SES patients more often presented with synchronous carcinomatosis at diagnosis (48% vs 35%, p=0.05). Following CRS-HIPEC, low SES patients had longer length of stay and higher rates of post-operative complications, readmission, and 90-day mortality (all p<0.05). Median survival following CRS-HIPEC for low SES patients was markedly lower (17.8 vs 32.4 mos, p=0.022) despite comparable tumor burden (PCI), completeness of cytoreduction, and pathologic tumor characteristics. On multivariate analysis, high SES remained a significant predictor of improved survival (HR=0.68, p=0.039). Conclusions: Despite improving therapies for CRPM, low SES patients remain at a significant disadvantage. Even patients who overcome hurdles of access and undergo treatment at a high-volume, tertiary referral center experience worse short-and longterm outcomes. Improving access and addressing these disparities is crucial to ensure equitable outcomes and improve patient care. Purpose: Sarcopenia is associated with poor long-term outcomes in many gastrointestinal cancers, but its role in anal squamous cell carcinoma (ASCC) is not defined. ASCC frequently develops in the context of comorbid conditions that have high rates of sarcopenia, including HIV and solid organ transplant. We hypothesized that sarcopenic anal cancer patients experience worse outcomes. Methods: A retrospective review of ASCC patients treated at a tertiary academic medical center from 2006 to 2017 was performed. Of 134 ASCC patients, 64 underwent PET/CT prior to chemoradiation and were included in the analysis. Skeletal muscle was measured at the third lumbar vertebral body transverse process. Musculature was selected using a range of 25-125 Hounsfield units; extraneous tissue was manually excluded. Skeletal muscle index was calculated as total L3 skeletal muscle divided by height squared. Sex-specific thresholds of sarcopenia were 52.4 cm 2 /m 2 for men and 38.5 cm 2 /m 2 for women. Sarcopenic patients were compared to the remainder of the cohort. Cox regression analysis was performed to assess overall and progression-free survival. Results: Sarcopenia was present in 25% of patients (n=16). Demographics were similar between groups. Sarcopenic patients had lower body mass index (22.2 vs 28.6 kg/m 2 , p=0.001) and were more likely to be white (88% vs 60%, p=0.046). There was no difference in clinical stage or comorbidities, completion of chemoradiation, or recurrence between groups. On multivariate analysis, factors associated with improved survival included white race (HR 0.17, p=0.006) and completion of chemoradiation (HR 0.04, p<0.001). Factors associated with worse overall survival were male gender (HR 5.94, p=0.010) and sarcopenia (HR 12.0, p=0.001) ( Figure 1 ). Completion of chemoradiation was associated with improved progression-free survival (HR 0.21, p=0.007) while male gender was associated with worse progression-free survival (HR 2.79, p=0.011). Conclusions: Sarcopenia is associated with worse overall survival in anal cancer patients. Further studies are indicated to determine if survival can be improved with increased attention to nutritional status in sarcopenic patients.

Kaplan-Meier curve of overall survival. MD Anderson Cancer Center, Houston, TX; 2. Kwame Nkrumah University of Science and Technology, Kumasi, Ghana; 3. Komfo Anokye Teaching Hospital, Kumasi, Ghana; 4. New York University, New York, NY. Introduction: Colorectal cancer (CRC) is a leading cause of death worldwide and becoming increasingly common in Sub-Saharan Africa, having increased in incidence over 30% over the last 5 years, and estimated to increase an additional 60% in the next 15 years. Currently, the practices and effectiveness of CRC care in this region have not been described. We sought to describe presentation, stage, surgical and adjuvant care, and survival for CRC patients presenting to a major tertiary hospital in West Africa. Methods: Records of CRC patients admitted to a tertiary care center in West Africa from 2013-2016 were prospectively reviewed for data on presentation, staging, management, and mortality. Patients were followed with biannual telephone calls to establish their post-discharge status. Survival analysis was performed for patients with pathologic or radiologic confirmation of cancer with adequate staging. Results: 53 CRC patients were included in analysis. Median age at diagnosis was 54(range:23-84) years; 36% were male. Common presenting complains were abdominal pain (23;43%), weight loss (20;38%), and bleeding per rectum (14;26%). Over half of patients (28;53%) were treated with palliative intent as compared to 25(47%) patients treated with curative intent. Overall, 35 (66%) patients received surgery, 6(11%) received chemotherapy, and 7(13%) received radiation therapy. Only 33(62%) patients had adequate staging, of which 25(76%) were stage IV. Three-year overall survival for all patients was 33%(95%CI:20-48%) and survival was found to be independent of stage (Adjusted Hazard Ratio ( INTRO Texas is home to the largest population of uninsured patients in the United States. Improving care for this population requires accurate, complete data, and an assessment of the effectiveness of existing public funding. Here we describe data quality disparities in the Texas Cancer Registry (TCR) and colon cancer outcomes across payor groups. METHODS TCR data for patients < 65 years old diagnosed with colon cancer between 2012 and 2016 were assessed for data completeness in key fields. We compared stage at presentation and overall survival (OS) among uninsured, privately insured, and publicly funded patients (Medicare, Medicaid, Tricare, Military, VA). RESULTS 14,855 patients were identified. Compared to privately insured patients, uninsured and publicly insured patients had higher rates of unknown data values in TNM stage (23% vs 17%, p<.001), surgery performed (7% vs 2%, p<.001), and node positive patients receiving chemotherapy (26% vs 23%, p=.02). In patients with complete staging data, being uninsured and publicly insured was associated with a higher incidence of clinical stage 4 presentation (46% vs 42%, p<.001). Diagnosis at autopsy was noted in 16% of uninsured patients vs 4% for both Public and Privately insured patients (p<.001). Overall survival was equivalent for uninsured and publicly insured patients (Figure, log rank p=.5). Privately insured patients had a significantly higher OS compared to uninsured and publicly insured patients (Figure, log rank p<.001).

Only one of four metropolitan service areas >1 million persons demonstrated an improved OS in publicly funded vs uninsured patients (log rank, p=.002). CONCLUSION Uninsured patients were more likely to have unknown values in TCR data, which could impact efforts to assess quality and improve care in this patient population. Public funding of medical care does not appear to be associated with improved OS in colon cancer in the non-elderly population, suggesting that opportunities may exist to improve resource utilization for this group of patients. However, incomplete data diminishes our ability to assess whether survival differences are due to discrepancies in treatment by stage or other non-treatment-related factors. Background: Despite support from the ACS, teaching surgical residents about palliative care has not been universally implemented. We previously tested the efficacy of palliative education via survey and now revisit this question after 10 years. We hypothesized that results of this new survey would closely correlate with the original, also expecting that senior residents would score better in terms of knowledge and comfort level regarding aspects of palliative surgical care. Methods:10-question surveys with Likert scale responses identical to our previous pre-test survey plus educational didactic sessions, post-test surveys, and analysis of pre-and post-test case scenarios involving surgical oncology standardized patients. Results: 65 surveys were completed. Compared to a decade earlier in which 100% of residents agreed that palliative care can include surgery, this diminished to 63.4% among interns and 75% among PGY2s. In 2007 residents felt comfortable discussing 'hospice care' at a rate of 68%. Recently, this rate was lower among interns(41.5%) and PGY2s(58.3%), but seniors(PGY3=100%, PGY4=80%, PGY5=75%). In 2007 91% of residents disagreed that "keeping the patient alive as long as possible is the ultimate goal of palliation." Although senior residents at this 2017-19 survey disagreed(PGY3-5 = 100%) only 75% of PGY2s and 92.7% of interns expressed recognition of this incorrect statement. Conclusions: Overall, senior residents scored similarly to the original 2007 cohort. PGY1 results likely reflect more medical school education rather than intern-year experience due to the early administration of the survey within the academic year. PGY2 results may suggest a varied clinical exposure to these principles at still a relatively early stage of training. As interest and knowledge expand in these areas of communication and palliative surgical principles, it is evident that more structured education is needed. We demonstrate here the pilot scaffolding of what could become a standardized curriculum for all resident and fellow trainees across the nation. Background: Unconscious bias often negatively impacts academic advancement and is reinforced by subordinating language. Recent reviews of medical conferences showed that women were less likely to receive a formal introduction compared to men. We examined speaker introductions at the Society of Surgical Oncology (SSO) annual meeting to determine if similar biases exist within our organization. Methods: A retrospective observational study of video-archived speaker introductions at the 2018 and 2019 SSO annual meetings was conducted. Professional address was defined as professional title followed by full name or last name. Multivariable logistic regression was used to identify factors associated with form of address. Results: A total of 499 speaker introductions were reviewed. The majority of speakers were men (290, 58%); 238 (48%) were post-graduate trainees (residents and fellows). A non-professional form of address was used to introduce 148 (30%) speakers. Post-graduate trainees received a non-professional address most often (33%). Full professors were more likely than junior faculty to introduce speakers with a non-professional form of address (18% of assistant professors vs 34% of associate professors vs 37% of full professors, p<0.001). In multivariable regression analysis including session topic, gender, home institution region, and ethnicity of the speaker and introducer, these findings remained strongly significant. Trainees were 2.7 times more likely to receive a non-professional address (p=0.003). As introducer academic rank increased, the odds of using a non-professional form of address increased in a stepwise fashion (Figure) . Use of a non-professional introduction, however, did not significantly vary by either speaker or introducer gender. Conclusions: Residents and fellows were more likely to receive a non-professional speaker introduction. Non-professional introductions increased with rising seniority of the introducer. The manner of speaker introduction did not vary by gender in our organization. More research is needed to explore the influence of these disparities on academic advancement for the next generation of surgical oncologists. Background: Gender bias has been identified in letters of recommendation for surgical residency and several surgical fellowships. We analyzed letters of recommendations from a single complex general surgical oncology fellowship over several years to determine whether a bias persisted based on the applicant or letter writer gender. Methods: Letters of recommendation were analyzed from applicants selected for interview at a single institution for the 2013-2020 surgical oncology fellowship application cycles. The letters were analyzed using Linguistic Inquiry and Word Count (LIWC2015), a validated text analysis program. Letters were compared by gender of applicant and letter writer using multivariable analysis. Correlation between letters for the same applicant was accounted for using generalized estimating equations and a repeated measures model, adjusting for applicant and recommender gender, application year, and recommender rank. Results: 841 letters (87% male authors) from 219 applicants (54% male) were included. Of the 41 word categories analyzed, significant differences were identified in only three. Insight, positive emotion, and perception processes were identified more often in letters by female writers in reference to male applicants (p = 0.05, 0.05, and 0.01 respectively). The remaining word categories including power, work, drive, and achievement were equivalent between male and female applicants and writers. The summary variables of authenticity, clout, analytic thinking and emotional tone showed that for female applicants, male writers were more likely to use clout (p = 0.04) and for male applicants, female writers were more likely to use emotional tone (p = 0.004). Authenticity and analytic thinking were equivalent between groups. Conclusion: There are very few significant differences in letters of recommendation for interviewees at a single surgical oncology fellowship over the last 8 years based on applicant or letter writer gender. Further work is needed to elucidate if this demonstrates progress based on recent work and educational effort to reduce gender bias. It may also be the result of selection bias as only letters from interviewed applicants were analyzed. Introduction: Podcasts have been extremely popular in multiple areas of education, society, and culture. Despite their popularity, no evidence based, free, surgical oncology podcast geared towards medical students, residents, and potentially interested non-medical professionals currently exist. Methods: We created an evidence-based surgical oncology podcast, covering diverse topics relevant to trainees. Our stated goal for each episodes is to cover the topic at hand and address: current and standard therapies, history of the disease, and future directions and trials. In addition, we provide a written brief summary of each episode and links to the relevant literature on our website. We host our podcast through wordpress.com and on the website www.surgoncfiles.com. We use Blubrry podcasting services to track all downloads. All recording is done via Audacity to record and edit episodes. Results: Beginning in 2017, we have created 20 episodes, for a total of 26,809 downloads, from 92 countries ( Figure) , or ~1350 downloads/episode (range 746-1743). In addition, we have had 14,611 unique website views. Episodes range from 20-60 minutes. Production costs were $378/ year. Our listening base is composed of ~75% residents of all levels, 20% attending physicians, and 5% other medical or non-medical professionals. The most predominant places where people listen to our podcast is during exercise or during their commute (50 and 73%, respectively). When asked "Do you prefer podcasts over traditional book or journal-based learning," 54% of respondents said they preferred or strongly preferred podcasts, while only 13% of people preferred traditional learning modalities. Conclusion: We have created a surgical oncology podcast covering broad topics reaching a large, international audience base. This has been performed with low costs and has engaged experts in our field. Further work in developing podcast-based education has the potential to supplement traditional online and print-based didactics with advantages of low cost, easily updatable content, and broad reach. Background: Teaching hospitals that train both general surgery residents and fellows in complex general surgical oncology have become more common. Despite ACGME dictums, attending surgeons may favor either residents or fellows assisting on operations of greater complexity, depending upon a variety of factors, including local surgical culture. This study investigates whether participation of a senior resident versus a fellow impacts outcomes of complex cancer surgery. Methods: Patients who underwent esophagectomy, gastrectomy, hepatectomy, or pancreatectomy with assistance from either a senior resident (PGY-4 or 5) or a fellow (PGY-6 or 8) were identified from the American College of Surgeon's National Surgical Quality Improvement Program (2007) (2008) (2009) (2010) (2011) (2012) . Analyses were performed separately for each operation. Propensity-scores were created for the odds of undergoing the operation assisted by a fellow. Patients were matched based on propensity score. Results: In total, 1,160 esophagectomies, 2,432 gastrectomies, 4,670 hepatectomies, and 7,519 pancreatectomies were identified. Senior resident participation was reported in 60.2%, 86.6%, 75.1%, and 86.0% of these operations, respectively. After matching, rates major complication rates were higher for patients who underwent esophagectomies involving a resident compared to fellow ( This study suggests that senior resident participation in complex cancer operations does not negatively impact operative time or outcomes, compared to involvement of a surgical oncology fellow. These findings indicate that senior residents should be given the same opportunities as fellows to participate in these potentially more challenging operations.

The Disclosure Slide -Informative or Obligatory, Five Years of SSO Cancer Symposium Presentations W.T. Merritt,* D. Roberge Bouchard, C. Ford, N. Petrelli, G. Tiesi. Helen F. Graham Cancer Center and Research Institute, Christiana Care Health System, Newark, DE. Introduction: Disclosure of financial support is a staple of presentations at scientific meetings and highlights potential conflicts of interest. Despite its importance, there has been minimal focus on it. Our aim was to review trends in financial disclosure (FD) at the Society of Surgical Oncology (SSO) Annual Cancer Symposium. Methods: A retrospective review of SSO Virtual Meeting presentations from 2015-2019 were accessed via the ExpertEd@SSO online portal. Presentations were evaluated for FDs given or not, number of FDs, time spent on or discussing the slide content and FD relevance. FDs were considered relevant if explicitly stated or listed on slide. Results: A total of 963 presentations were reviewed. 463 (43%) had no verbal mention of FDs and 331 (34%) had no slide and no verbal mention. 164 (17%) listed at least one FD and there was no difference in this rate during the study period (p=0.9). The average number of FDs per presentation from 2015 -2018 was 3.6 +/-4.3. Presentations from 2019 had a higher average number of FDs per presentation (5.1 +/-4.8) which led to an overall statistical difference amongst the groups (p<0.001). Presenters with FDs averaged 6.6 +/-6.7s on the slide and 4.6 +/-4.9s discussing the content; these both decreased over time (p<0.013/0.003) [Fig 1] . Without FDs, presenters averaged 2.0 +/-0.2s on the slide and 1.0 +/-0.1s discussing the slide; these did not decrease over time (p=0.155/0.94). There was a statistical difference in time spoken (p= 0.005) and time on the slide (p=0.004) for presentations with and without FDs. Relevance of FDs were stated in only 22 (13.4%) of presentations with FDs. Conclusions: Disclosures of financial support were often not included or acknowledged during oral presentations at the SSO Annual Cancer Symposium. Little time was spent discussing financial disclosures or their relevance. Despite a significant increase in the number of disclosures per talk, presenters spent less time discussing their financial disclosures. Greater emphasis on financial disclosure at the SSO Cancer Symposium should be undertaken to help members identify areas of potential conflict of interest. Introduction: Germline CDH1 loss of function mutations confer up to 70% lifetime risk of diffuse gastric cancer (DGC), therefore prophylactic total gastrectomy (PTG) is recommended. Despite this, many patients elect for yearly surveillance endoscopy (EGD) with biopsy. We evaluated factors associated with choosing PTG and examined patients' decision-making. Methods: A single institution, prospective study of patients with CDH1 mutation treated from January 2017 to August 2019 was utilized. An additional survey was administered to patients about their decision regarding PTG. Categorical and continuous variables were analyzed using chi square and independent t-test, respectively. All variables with p<0.1 were included in multivariable regression. Results: 100 patients were included in this analysis. Most were Caucasian (N=99, 99%), and female (N=69, 69%), with median age 45 years. All patients underwent EGD, of whom 36 (36%) had gastric biopsy positive for occult signet ring cancer cells (SRCC). Patients were analyzed in two cohorts: those who proceeded with PTG after EGD (N=44, 44%) and those who did not (N=56, 56%). There was no difference between cohorts according to age, gender, race, EGD biopsy pattern, clinical findings on EGD, patient symptoms, family history of DGC, or personal history of breast cancer. Patients who proceeded to PTG were more likely to have SRCC detected at EGD (p=0.01) and family history of breast cancer (p=0.06). Detection of occult SRCC at time of EGD was an independent predictor of proceeding to PTG (OR 3.1, CI 1. 3 -7.4, p=0 .01). Sixty-two (62%) patients have responded to the survey. Top reasons cited to forego PTG (N=26) were family history (N=11, 42%) and/ or EGD results (N=7, 27%). Patients who underwent PTG and answered the survey (N=36) reported their decisions were based on family history (N=27, 75%) and/or EGD biopsy results (N=5, 14%). Conclusion: CDH1 mutation carriers often incorporate EGD biopsy results into decision making regarding PTG. Although consensus guidelines offer surveillance EGD as an alternative to PTG, it appears that negative EGD biopsy results may influence patient decision making regarding PTG. Introduction: Hereditary diffuse gastric cancer attributed to germline CDH1 mutation imparts up to a 70% lifetime risk of gastric cancer. Occult signet ring cancer cell (SRCC) foci are found frequently in the lamina propria of the stomach (T1a) of asymptomatic carriers at time of prophylactic total gastrectomy. Annual surveillance endoscopy is recommended for patients who decline gastrectomy, however this carries a high false negative rate of SRCC detection. Confocal endomicroscopy (CEM) obtains in vivo histologic images of gastric mucosa. Our goal was to determine SRCC detection rate using CEM in CDH1 carriers. Methods: A phase II single institution study (NCT# 03648879) was initiated in February 2019 for patients with CDH1 germline mutations. All patients underwent white-light, high definition, upper endoscopy using the Cambridge method in addition to probe-based CEM biopsies. Thirty-six were enrolled for 89% power to rule out a 24% Cambridge detection rate in favor of 45% CEM detection rate. Results: Thirty-four asymptomatic patients with CDH1 mutations underwent endoscopic surveillance per protocol. Patients were majority Caucasian (91%) and female (73%); median age was 48 (range 25-74). The median number of biopsies was 30 for Cambridge method and seven for CEM. Seven patients (20%) had SRCC detected in at least one biopsy. Four patients (11%) had SRCC detected utilizing Cambridge (control) method, one of which was not detected by CEM. Six patients (17%) had SRCC detected by CEM, three of which were not detected by Cambridge method. Fourteen patients (41%) went on to prophylactic total gastrectomy. All 14 (100%) had SRCC detected on final pathology; 12 had negative pre-operative biopsies using Cambridge method (false negative rate 85%), while 10 had negative biopsies by CEM (false negative rate 71%). Conclusions: Confocal endomicroscopy may yield a lower false negative rate of SRCC detection in asymptomatic CDH1 mutation carriers undergoing surveillance when compared to the Cambridge method. Evaluation of more sensitive endoscopic methods for gastric cancer surveillance is needed for patients who decline prophylactic total gastrectomy. Background Gastric cancer often presents at an advanced stage. Guidelines recommend diagnostic laparoscopy in routine staging. Advancements in CT scan technology, and logistical limitations likely explain low rates of laparoscopy reported in some studies. This questions whether a more selective use of laparoscopy may be warranted. We aimed to determine the rate of positive staging laparoscopy in the treatment of gastric cancer within our region. Secondary outcomes were factors predictive of positive laparoscopy. Methods A survey was sent to all members of the Alberta Association of General Surgeons in Alberta, Canada to identify surgeons treating gastric cancer. Staging laparoscopies done for gastric cancer between July 2007 and February 2019 by participating surgeons were reviewed retrospectively. The primary outcome was positive laparoscopy and/or cytology. Univariate and multivariate analysis were performed to identify factors predictive of positive laparoscopy. Results There were 116 cases reviewed from 5 surgeons at 4 Centers. In 91% of cases laparoscopy was performed as a separate procedure before chemotherapy. Metastatic disease (positive biopsy and/or positive cytology) was identified in 29%. On univariate analysis, the following preoperative CT findings were associated with an increased risk of metastatic disease at laparoscopy: visualization of the primary tumor (OR 9.8, p=0 .03), presence of abdominal lymphadenopathy (>1cm) (OR 2.4, CI 1. 1 -5.4,p=0.04) , and presence of ascites (OR 19.1, p=0.007) . The presence of ascites maintained statistical significance on multivariate analysis (p=0.006). Conclusions The rate of positive laparoscopy was 29%. This is higher than previously reported, despite advancements in imaging technology and suggests laparoscopy should still be used routinely in gastric cancer staging. Our study identifies several preoperative imaging findings associated with positive laparoscopy, however further study with larger numbers is needed to establish robust predictors of positive and negative laparoscopy prior to advocating for a selective laparoscopy approach. We have established a prospective gastric cancer database to further study this. Figure) . CONCLUSION: Histology and neoadjuvant therapy are not associated with adequate lymphadenectomy in esophageal cancer. Use of a minimally invasive approach and treatment at a high volume center are associated with adequate lymphadenectomy. The likelihood of achieving adequate lymphadenectomy has increased over time, but only half of surgical patients achieve this threshold; thereby justifying the continued use of this generalized quality measure.

Cancer M. AKABANE.* TORANOMON HOSPITAL, Tokyo, Japan.

Background: A suitable treatment strategy for esophageal cancer after DCRT for T4 cases has not been established and remains unclear. The aim of this study was to clarify the independent prognostic factors, surgical indications and optimal extent of lymphadenectomy for T4 esophageal cancer. Methods: Of 803 patients who underwent esophagectomy for esophageal cancer at our institution from 2006 to March 2018, 33 patients who underwent salvage esophagectomy with locally advanced T4 cancer were included. We examined the baseline attributes and treatment results of these cases, and evaluated the prognostic factors and treatment strategies. Results: Independent favorable prognostic factors in T4 cancer(T4a/T4b=11/22) included non-T4b status(hazard-ratio[HR]:15.311, 95%confidence-interval[CI]:1.277-183.5) and R0 resection (HR:14.706, 95%CI: 1.193-166.67 ). In cases in which R0 resection(n=14) was possible, both 1-,5-year survival rates were 90.9%, while the 1-,5-year survival rates of cases without R0 dissection(n=19) was 44.9% and 0%, respectively. In univariate analysis, cases that underwent 2-or3-field lymph node dissection tended to have a better prognosis(p=0.062), and those with 60 or more lymph nodes dissected had a significantly better prognosis(p=0.038). In cases that underwent salvage esophagectomy with typical lymph node dissection, the rate of complications with Clavien-Dindo grade III was not increased, at 33.3%, indicating that the procedure was relatively safe. Conclusions: The results showed that in salvage esophagectomy for T4 esophageal cancer, R0 resection led to improved prognosis. Because typical 2-or3-field lymph node dissection, including prophylactic dissection could be performed safely and led to better prognosis in salvage esophagectomy, typical esophagectomy including prophylactic lymph node dissection should be performed if possible. Introduction: Surgical resection is critical for improving survival in patients with esophageal cancer. With the advancement of minimally invasive surgery, there is still debate on the best approach for esophagectomy. We report a modern analysis of outcomes with transthoracic (TT) versus transhiatal (TH) esophagectomy. Methods: A prospectively managed esophagectomy database was queried for patients undergoing transthoracic or transhiatal esophagectomy between 1996 and 2018. Propensity score matching was performed based upon age and stage. Continuous variables were compared using the Kruskal Wallis or the ANOVA tests as appropriate. Pearson's Chi-square test was used to compare categorical variables. Unadjusted survival analyses were performed using the Kaplan-Meier method comparing survival curves with the log-rank test. Results: We identified 846 patients who underwent esophagectomy with a mean age of 64 10 years. There was no difference in EBL for TT and TH, but the mean OR times were longer for TT vs. TH (p<0.001) and the number of retrieved lymph nodes was higher for TT vs. TH (p<0.002). Postoperative complications occurred in 207 (29.0%) patients who underwent a TT approach versus 59 (44.7%) who underwent a TH approach, (p<0.001). These were: anastomotic leaks: 4.3% vs 9.8%, (p=0.01), anastomotic stricture 7% vs 26.5%, (p<0.001) and pneumonia 12.6% vs 22.7%, (p<0.002), aspiration (p<0.001), wound infections (p=0.004), and pleural effusions (p<0.001). Median survival was significantly improved in patients undergoing TT (62 months) vs TH (39 months) p=0.03. After matching there were 131 in the TT and 131 in the TH groups. Post-operative complications remained lower in the TT (32.1%) vs TH (44.3%), p=0.04. Anastomotic strictures (p<0.001), all pulmonary complications (p=0.006), aspirations (p<0.001), and pleural effusions (p<0.001) were lower in the TT cohorts. Conclusion: In this modern propensity score matched analysis of TT versus TH esophagectomy we found that a TT approach was associated with a lower morbidity but an improvement in nodal harvest. Survival was also significantly improved in patients who underwent transthoracic esophagectomy.

Introduction Gene mutations, such as Kirsten rat sarcoma viral oncogene homologue (KRAS), are used to direct clinical management and prognosticate for patients with metastatic colorectal cancer (mCRC). We seek to investigate pathologic and survival differences based on KRAS exon mutation for patients with colorectal liver metastases (CRLM). Methods This retrospective, single-center study includes patients with R0 or R1 resection of CRLM from 1992-2016. Patients with persistent extrahepatic disease, liver-first resection, and unknown mutation status were excluded. Clinicopathological characteristics were compared between mutant and wild type KRAS patients. Pathologic features, overall survival (OS) and recurrence free survival (RFS) were assessed from time of CRLM resection and stratified by exon 2, 3, and 4. Fisher's exact test, Wilcoxon Rank Sum test, and Log-rank test were used, where appropriate. Results 938 patients were identified with median age of 57 (range 19-91). KRAS mutations were present in 47% of patients (445/938). Of these, 407 (91%) had a mutation on exon 2, 14 (3%) on exon 3, and 24 (5%) on exon 4. Median OS was 71.4 months (95%CI: 66.1-76.5). Follow up time in survivors was 63.0 months (range: 0.3-300). Patients with KRAS mutations had worse OS (median 55.5 months (95%CI: 50.9-62.9)) compared to wild type (median 91.3 months (95%CI: 76.5-111.3)), p<.001. No significant difference was seen in OS based on exon, with five-year OS of 45.7% (95% CI: 0.40-0.51) for exon 2, 39.1% (95% CI: 0.11-0.68) for exon 3, and 68.8% (95%CI: 0.45-0.84) for exon 4 (p = 0.12) (Figure 1 ). While patients with KRAS mutations had worse RFS (median 10.8 months (95%CI: 9.6-11.8)) compared to wild type (median 15.8 months (95%CI: 13.8-20.2)), p<.001, no difference in RFS distribution was found based on exon (p=0.14). No significant association was seen between tumor size and KRAS exon (p=0.10) or between number of metastases and KRAS exon (p=0. Purpose: Metabolic alterations in cancer are considered important targets for potential therapeutic intervention. The cystine/glutamate antiporter (xCT) plays a crucial role in providing building blocks for redox homeostasis through the generation of glutathione (GSH). We have previously developed an erastin-based xCT inhibitor -conjugated sigma-2/erastin (ACTX-3102). In an attempt to discover pathways that enhance pharmacologic xCT blockade, we investigated the role of malic enzyme 1 (ME1). ME1 maintains redox balance through the generation of NADPH and regeneration of intracellular GSH. Importantly, ME1 has been shown to be deficient in 5% of all human tumors. In our current study, we investigate the underlying biology of ME1 negative tumors, and the role of ME1 in the context of ACXT-3102-mediated blockade as a discriminator toward more efficient cancer therapy. Experimental Design: ME1(+) and ME1(-) non-isogenic tumor cell lines were evaluated for sensitivity to ACTX-3102. Drug sensitivities using isogenic tumor cell lines, via stable ME1 inhibition (shRNA) or reconstitution (cDNA transfection) were performed in vitro and in vivo. Finally, the impact of ME1 expression on xCT activity was determined in vitro utilizing fluorescently-labeled cystine. Results: ME1(-) tumors displayed substantially higher sensitivity to ACXT-3102 than ME1(+) tumors (non-isogenic). These findings were further corroborated in the isogenic setting via ME1 knockdown and reconstitution. ME1 deficiency was associated with more rapid cystine influx kinetics, indicative of increased xCT activity. Preclinical studies in a mouse model confirmed our prediction that ME1-deficient tumors required decreased ACTX-3102 dosing to achieve tumor control when compared to ME1-expressing tumor subtypes. Conclusion ME1 deficient tumors are characterized by an increased reliance on cystine import and are thus more vulnerable to cell death by pharmacologic xCT inhibition. Our data provide a strong rational for including ME1 as a possible prognostic indicator for both patient and dose selection in a phase I clinical trial. Background: Patients with stage III colorectal cancer are generally recommended for adjuvant chemotherapy. Even with adjuvant chemotherapy, 20-30% of stage III patients develop recurrences. The risk of recurrence for patients with stage III colorectal cancer is unclear. Preoperative systemic inflammation index has been linked to poor prognosis in colorectal cancer. However, postoperative systemic inflammation index for patients with stage III colorectal cancer and recurrence is unclear. This study aimed to evaluate the association between preoperative and postoperative systemic inflammation index and the recurrence for patients with stage III colorectal cancer Methods: 133 stage III colorectal cancer patients were included in this study. Preoperative and postoperative C-reactive protein/albumin ratio (CAR), neutrophil to lymphocyte ratio (NLR), and platelet to lymphocyte ratio (PLR) were evaluated. Then the relationship between these factors and recurrence was analyzed. Results: The optimal cutoff values of the systemic inflammation index were determined by the ROC curve. In the multivariate analyses, N-stage, postoperative complication, preoperative NLR, and postoperative CAR were independent predictive factors for the recurrence-free survival (RFS). The postoperative CAR was also independent predictive factors for the overall survival (OS). In patients of postoperative CAR^0.035, those patients without adjuvant chemotherapy was associated with shorter RFS and OS than adjuvant therapy, but there were not significantly different for RFS and OS in patients of postoperative CAR<0.035. Conclusions: Postoperative CAR was strongly associated with poor prognosis in patients with stage III colorectal cancer and this is a useful biomarker that can be evaluated whether adjuvant chemotherapy should be given. Introduction: Human papillomavirus (HPV) infection is associated with substantial alterations in host DNA methylation which may contribute to carcinogenesis. By epigenome-wide analysis, we previously identified genes that are differentially methylated during the progression of HPV-associated anal and cervical cancers. ASCL1, a lineage-specific tumor suppressor gene is one target of particular interest. ASCL1 can be transcriptionally repressed by DNA promoter methylation or by increased NOTCH1 activity. We sought to confirm the methylation-mediated transcriptional regulation of ASCL1 and examine its association with NOTCH signaling pathway expression. Methods: Expression of ASCL1 was evaluated by reverse transcriptase PCR and Western Blot analysis in a comprehensive panel of anal (HPV+: ACC), cervical (HPV+: SiHa, HeLa; HPV-: C33), and OP (HPV+: CRL3212, HPV-: HTB43) cancer cell lines. The methylation status of ASCL1 was evaluated by quantitative methylation specific PCR (qMSP) with re-expression confirmed by demethylation. Expression of NOTCH1, NOTCH2, and NOTCH3 was evaluated by Western Blot analysis. Results: ASCL1 was significantly downregulated in ACC and HPV+ cervical cancer cell lines with corresponding high levels of promoter methylation. ASCL1 expression was maintained in OP and HPV-cervical lines with no significant methylation. ASCL1 re-expressed after demethylation of non-expressing cell lines with 5 Aza-2-deoxycitidine. NOTCH1 and NOTCH3 were downregulated in HPV+ anal and cervical cancer cell lines but maintained expression in HPV-cervical and OP cancer. In contrast, NOTCH2 demonstrated an inverse relationship to ASCL1 expression, with expression in HPV+ anal and cervical cancer cell lines, and downregulation in OP cells. Conclusion: We have confirmed that ASCL1 downregulation is mediated by DNA promoter methylation in HPV+ anal and cervical cancer cell lines and likely not related to NOTCH1 signaling. The potential of ASCL1 as an epigenetic driver in these cancers and its interplay with other NOTCH signaling elements warrant further investigation. ASCL1 may play a different lineage-specific role in HPV-associated head and neck cancers. Introduction: For patients with locally advanced rectal cancer (LARC), neoadjuvant chemoradiation (CRT) followed by surgery remains the standard of care. However, all patients do not benefit equally from this treatment with up to 30% showing minimal to no response. Having the ability to predict benefit from CRT would allow for more customized care. Here we examine the correlation between tumor mutation status and response to CRT in patients with LARC. Methods: After IRB approval, we enrolled 39 patients with LARC, defined as T3-4NanyM0 or TanyN1-2M0. All patients underwent staging via endorectal ultrasound (EUS) and/or pelvic MRI. Tissue obtained from EUS was sent for targeted genomic sequencing. Patients received 4-6 weeks of standard of care CRT, which consisted of 5-fluorouracil/capecitabine with concomitant external beam radiation. Those without a complete clinical response (CCR) were treated with surgical resection as recommended by the treatment team. Surgical specimens were examined for treatment response, and categorized as poor, minimal, moderate or complete. Results: Thirtynine patients were enrolled and 34 were analyzed. 5 patients were excluded from analysis secondary to decline in performance status or death causing discontinuation of curative treatment or being lost to follow up. Average age at diagnosis was 57.9 (26-81) and all had ECOG PS 0 or 1. Seven (20.5%) patients had a CCR (2) or pathologic complete response (PCR, 5) . In patients with a complete response, the mean number of mutations was 5.19 ( 2.6) compared to 6.86 ( 3.7) in those without complete response (p=0.26). The 4 most commonly mutated genes were APC, TP53, KRAS, SMAD4. These were mutated in 82%, 23%, 13% and 3% of patients, respectively. These mutations were not associated with complete or poor response. Conclusion: Greater than 20% of our patients with LARC who underwent CRT had a CCR or PCR. Neither mutational burden nor a specific mutated gene was associated with complete or poor response to CRT. Further investigation is needed to examine if specific mutational profiles, rather than individual mutations, are associated with response. Introduction Peritoneal metastasis from gastric adenocarcinoma (GCa) portends a poor prognosis and resistance to conventional systemic treatment. High expression of Plk4 has been identified as a molecular predictor of aggressive behavior in patients with breast, pancreas and colorectal cancers. We hypothesize that Plk4 facilitates peritoneal metastasis in GCa by enhancing cell migration and invasion of the peritoneum, potentially through the Rac1 activating Guanine Exchange Factor (GEF) PREX2. Methods Parental GCa cell lines AGS, N87 and Hs746T were engineered to stably express shRNAs for Plk4, PREX2, or Luciferase by lentiviral infection, and depletion of Plk4 or PREX2 was confirmed vs shLuc control. Real-time scratch wound healing migration assays were performed on confluent GCa cells with images obtained q1h. A mesothelial cell clearance assay was developed to model the interaction between GCa cells and the mesothelial cell layer of the peritoneum. 3D GCa spheroid invasion through Matrigel was assessed over 72h. Results Plk4 knockdown suppressed 2D wound healing, mesothelial cell clearance and 3D invasion in all 3 GCa cell lines (p<0.05 vs. shLuc control, n=3 biologic replicates per experiment). PREX2 knockdown mimicked the effects of Plk4 knockdown. Co-immunoprecipitation assays revealed a physical interaction between Plk4 and PREX2 in cellulo. Furthermore, Plk4 was shown to phosphorylate PREX2 in an in vitro kinase assay, indicating a potential functional interaction whereby PREX2 mediates the effects of Plk4 on GCa cell migration, mesothelial cell clearance and invasion into the underlying matrix. Conclusions In assays that model the steps of the peritoneal metastatic cascade, Plk4 shows the ability to promote GCa cell peritoneal metastasis. We also demonstrate a novel interaction between Plk4 and the GEF PREX2, implicating the Rac1 pathway as a potential target for therapeutic intervention. Introduction: Contralateral prophylactic mastectomy (CPM) rates have risen significantly nationwide over the past decade. The Society of Surgical Oncology issued an updated position statement regarding CPM in 2017, with emphasis on counseling about alternative risk-reduction approaches. We sought to analyze the trend of CPM at a single academic institution with dedicated breast surgeons and genetic counselors. We hypothesized a peak in CPM associated with lay media exposure would decline with surgeon-guided consent underscoring the lack of overall survival advantage to CPM in patients without genetic mutation risk factors. Methods: Using our prospective tumor registry database, women who underwent CPM for unilateral breast cancer were identified. Chart review was performed to verify registry data. Age, genetic testing status and genetic mutation presence were recorded. Statistical analysis was performed by chi-squared and Fisher's exact test. Results: Between 2005 and 2016, 1,492 women had a mastectomy and 347 (23%) had a CPM. The median age was 48.7 years, with a range of 28-83. The frequency of CPM varied significantly over the years (p<0.0001), but an ascending linear trend was not supported. There were 202 (58%) CPM patients that underwent genetic testing, which varied significantly over time (p <0.056) with an ascending linear trend. Forty-six (23%) of these patients were found to have a BRCA 1/2 mutation (p 0.260). While BRCA mutation positivity remains constant the number of patients tested has increased. Conclusion: The trend of CPM at our academic institution follows the trend for genetic testing and BRCA positivity over time. CPM more closely follows genetic testing rather than detection of a deleterious BRCA mutation. Continued efforts to secure adequate patient education regarding the risks and benefits of CPM is warranted. A total of 48 genes were analyzed on average. ATM (23 patients) and CHEK2 (20 patients) mutations were most common. BRCA1 (16 patients), BRCA2 (10 patients) followed. Also noted were Lynch syndrome mutations (MLH1, MSH2, MSH6), MRE11, MUTYH, and APC. A subset of patients had multiple deleterious and VUS mutations. Of the patients with VUS, 54/116 (46.5%) had mastectomies (14 unilateral and 40 bilateral) versus 41/56 (73%), 12 unilateral and 29 bilateral of patients with deleterious mutations. Overall mastectomy rate at our institution over the same time period was 52.5%. Conclusions: In our patient population, mutations in ATM and CHEK2 were more common than BRCA1 or BRCA2. There were also several actionable mutations found that are not included in small selected panels. Finding a VUS did not increase the percentage of mastectomies performed. Panel testing should be the standard of care as many patients may have multiple mutations outside of BRCA1 and BRCA2. INTRODUCTION: There is a paucity of data regarding inherited mutations associated with phyllodes tumors (PT). Multiple germline mutations have been reported including TP53, BRCA1, VHL, SDHB, NF1, and RB1; however, a PT diagnosis does not meet current NCCN criteria for genetic testing, including Li-Fraumeni Syndrome (TP53). As no study has systematically evaluated germline mutations for PT, we sought to determine the prevalence of mutations associated with these rare tumors. METHODS: We performed an 11 institution retrospective review of contemporary (2007-2017) PT practice. We recorded multi-generational family cancer history, personal history of genetic testing, specific genes tested and mutations identified. We identified patients that met criteria for genetic referral, irrespective of PT diagnosis, based on strict NCCN criteria (e.g. unaffected individual with a 1 st or 2 nd -degree relative with ovarian cancer). Logistic regression was used to estimate the association of select covariates with likelihood of undergoing genetic testing. RESULTS: Of 550 PT patients, 59.8% (N=329) had a close family history of any cancer, 34.0% (N=112) of whom had 3 family members affected. Only 6.2% (N=34) had genetic testing, 38.2% (N=13) of whom had only BRCA1/BRCA2 tested, excluding multiple previously reported PT associated gene mutations. Selection for testing was not associated with age (OR=1.01, p=0.55) or PT size (p=0.12), but was associated with PT grade (malignant vs benign: OR=9.17, 95% CI 3.97-21.18), and meeting NCCN criteria (OR=3.43, 95% CI 1.70-6.94). Of 34 patients tested, 3 (8.8%) had a deleterious mutation (1 BRCA1, 2 TP53), and 2 (5.9%) had a BRCA2 VUS. Notably, an additional 86 (15.6%) patients met NCCN criteria for further genetic evaluation, yet had no genetic testing completed. CONCLUSIONS: Very few patients with PT undergo germline testing. Of those tested, few received multi-gene testing; however, a deleterious mutation was identified in roughly 10%, consistent with other rare tumors. Multi-gene testing of a PT cohort would present an opportunity to discover the true incidence of germline mutations in PT patients. Background: Many breast imaging centers offer risk assessment to women undergoing screening mammography, and 15-17% are expected to carry a 20% lifetime risk of breast cancer. Contrary to mutation carriers, there is a paucity of data on outcomes of MRI screening in this population. This study assesses outcomes of baseline MRI screens in non-mutation carriers from a high-risk breast clinic (HRBC). Methods: We retrospectively reviewed patients seen at our institution's HRBC between Mar 2018-Feb 2019. Those with prior breast cancer, known genetic mutation or missing data were excluded from analysis. We determined baseline breast MRI results and rates of MRI-detected high-risk lesions (HRL, defined as atypical hyperplasia or LCIS) and MRIdetected breast cancers. Results: 319 women attended our HRBC; median age was 48 (range: 22-75), 4.7% had prior atypia or LCIS. 94.4% were white, 9.1% had Ashkenazi Jewish heritage, and 54.9% had heterogeneous or extreme breast density. Most had 1 (49.5%) or 2 (29.2%) close relatives with breast cancer and 39.5% had a relative affected at age <50. Screening MRI was recommended for 282 (88.4%) patients; 39 (13.8%) had prior MRI screening and of the remaining, 156 (64.2%) underwent a screening MRI. Among 189 baseline breast MRIs, a BIRADS 3 or 4 finding occurred in 37 (19.6%) patients; 25 (12.7%) required additional diagnostic breast imaging, 16 (8.5%) had 1 core biopsy and 4 (2.1%) had an excisional biopsy after initial core (Figure 1 ). Another 7 (3.7%) patients had an incidental non-breast finding. A HRL was identified in 2 (1.1%) patients (atypical ductal and lobular hyperplasia, respectively), both of which were excised. 2 (1.1%) patients with normal screening mammograms were diagnosed with T1N0 ER+ breast cancers. Conclusions: In the setting of a HRBC, approximately two thirds of women with a 20% lifetime risk of breast cancer pursued screening MRI when recommended. On baseline screen, the rate of MRI-detected breast cancer was low (1%); however, malignancies were mammographically occult and identified at an early stage. Further follow-up is needed to assess the impact of continued MRI screening in this population. Introduction While lobular carcinoma in situ (LCIS) increases the risk of developing breast cancer (BC) in either breast, the risk for bilateral cancer among women with a history of LCIS is poorly understood. We sought to determine the incidence of synchronous and metachronous bilateral BC (SBBC; MBBC) in this population. Methods Women diagnosed with LCIS between 1980-2017 who subsequently developed BC were retrospectively identified from a prospectively maintained database. Bilateral BC diagnosed <6 months or 6 months apart were defined as SBBC or MBBC, respectively. Comparisons were made between unilateral BC (UBC), MBBC and SBBC to assess for risk factors associated with bilateral BC among women with prior LCIS. The Kaplan-Meier method was used to estimate 5-year MBBC risk. Results Among 1650 women with LCIS, 241 developed BC. Median age of LCIS diagnosis was 50y; median age of first BC diagnosis was 55y. Index BCs were 26% DCIS, 45% IDC, 26% ILC, 3% other histology and 93% ER+. 13% (n=31) received chemoprevention before BC diagnosis. With a median follow-up from first BC of 3.2y, 17% (n=33) developed bilateral BC, including 17 SBBC and 16 MBBC [incidence of 7.1%, 6.6%]. There were no significant differences between UBC vs SBBC patients in age, tumor features, MRI screening, breast density or chemoprevention use (all p>0.05). Women with MBBC were more likely to have DCIS as first BC (56% v 30%,p=0.01), less likely to receive endocrine therapy for first BC (31% v 64%,p=0.02), and more likely to receive chemotherapy for initial invasive BC (62% v 26%,p=0.04). Estimated 5-year risk of any MBBC was 6.4% (95% CI 2-11%). ER+ and PR+ BC were the only factors on univariate analysis associated with decreased chance of MBBC (HR 0.13 p=0.02; HR 0.25 p=0.05, respectively). There was no difference in development of MBBC by tumor histology (ILC v other,p=0.28). Conclusions Bilateral BC rates among women with a history of LCIS appear modestly elevated compared to reported rates in the general BC population. Women with ER+ disease have a reduced MBBC risk, highlighting the durable protection of endocrine therapy, while index ILC was not associated with an increased risk. Background Delayed age-related lobular involution (LI) in benign breast biopsies predicts increased breast cancer risk. We developed and tested an automated computational pathology algorithm to assess LI. Methods Benign breast biopsies from 165 participants in a retrospective cohort were evaluated with digital images of H&E stained sections. A pathologist annotated < 10 representative lobules per image, and visually classified normal lobules into three categories of LI (none, partial, or complete). The algorithm was developed with deep learning using 72 lobules from 27 subjects to quantify digital image features (based on pixel counts) hypothesized to correlate with LI: lobule area, epithelial area, acini count, and average acini area. Algorithm performance was validated in the remaining 138 subjects. Association of each LI metric with visual LI categories was evaluated using generalized linear mixed effects models. ROC curve analysis of area under the curve (AUC) was used to assess performance of each LI metric in discriminating subjective categories of LI. Results Median age (52) was similar in both sample sets. The percentages of visually assigned LI levels were: none(20), partial(40), complete(40) in development set (N=72 lobules); and none (12), partial(40), complete(48) in validation set (N=599 lobules). Lobule area correlated strongly with epithelial area (Spearman correlation coefficient 0.95), less strongly with acini count (0.80) and minimally with average acini size (0.38). Lobule area, epithelial area, acini count, and average acini size each demonstrated inverse and significant associations with visual LI categories, with smaller values associated with more complete LI (p 0.001). Three measures demonstrated excellent discrimination across the three LI categories: epithelial area (AUC: 0.86), lobule size (AUC: 0.85), and acini count (AUC: 0.84), whereas average acini size showed poor discrimination of LI category (AUC 0.58). Conclusion Automated quantitative computational pathology assessment of LI achieves excellent agreement with expert assigned classification of LI that predicts breast cancer risk after benign biopsy.

Introduction Frailty in surgical patients is associated with increased morbidity, mortality, and failure to rescue. However, the economic impact of frailty is poorly understood. Methods Cost data for elective surgery were linked to a NSQIP targeted population of colorectal, esophageal, hepatic, and pancreatic operations performed in 2017 and 2018 at an academic medical center. The Modified Frailty Index-5 (MFI) categorized patients into non-frail (NF): MFI 0, pre-frail (PF): MFI 1-2, and frail (FR): MFI >2. Coarsened exact matching was used to reduce imbalances between groups. Results Of 613 patients, the majority were white, female, and had median age of 64 years. Frail patients were more likely to have a longer hospital stay and be discharged to a skilled care facility. There was no difference in rate of complications (NF 14.93%, PF 14.46%, FR 21.05%, p=0.734). Unadjusted mean total cost of care was $19,413.06. The average increase in cost from a complication was $12,732.72; however, this varied among groups: NF $12, 182.36, PF $12, 768.77, and FR $16, 340.88 . Adjusted analysis resulted in a cost increase of $2,945.71 for PF (p=0.007), and $8,337.26 (p=0.008) for FR; when compared to NF. The presence of a complication added $12,656.67 (p<0.001) regardless of frailty category. Multivariable models using NF as a reference do not suggest an association between an increase in total cost and increasing frailty category: PF OR 1.4 (CI 0.96 -2.04) and FR OR 2.87 (CI 1.00 -8.23). The association between the presence of a complication and increased total cost persisted (OR 2.68, . Matched analyses, after categorizing patients into frail (MFI>2) and non-frail, suggest that frailty is not associated with increased cost (OR 1.54, ). However, the presence of a complication was strongly associated with an augmented cost of care (OR 6.69, CI 2.69 -16.61). Similar results were encountered for direct and indirect care costs. Conclusions Complications are the major driver of total cost of care after gastrointestinal surgery. In frail patients the cost of a complication is considerably higher than in their non-frail counterparts.

Box plot depicting total cost of care (Dollars) over frailty category (MFI) and the presence or absence of a complication (Yes or No).

An NCDB Analysis: Patterns of Adjuvant Chemotherapy Use Among Octogenarians with Pancreatic Adenocarcinoma that Underwent Curative Resection W.T. Mehtsun, 1 * L.M. Pak, 2 J. Wang. 3 1. Dana Farber Cancer Institute, Brigham and Women's Hospital, Massachusetts General Hospital, Boston, MA; 2. Brigham and Women's Hospital, Boston, MA; 3. Dana Farber Cancer Institute, Brigham and Women's Hospital, Boston, MA. Introduction: Perioperative morbidity and mortality following pancreaticoduodonectomy have improved markedly over the last decade, consequently a larger proportion of elderly patients with pancreatic adenocarcinoma(PDAC) are being offered curative resection. Yet, the use of chemotherapy following curative resection has not been examined at a population level. We sought to investigate the use of adjuvant chemotherapy for elderly patients that underwent resection for PDAC and to examine outcomes associated with its use. Methods: Using the National Cancer Database, we identified patients with PDAC, 80 years or older, diagnosed between 2004-2014, and who underwent a pancreaticoduodonectomy. The proportion of patients who received adjuvant chemotherapy was examined over time. We compared clinical, pathologic, and demographic characteristics between patients who received adjuvant chemotherapy and those who did not using chi 2 and t-test statistics. The hazard of death was examined using Kaplan-Meier Cox proportional hazards multivariable models Background:Functional outcomes are central to decision-making by older adults (OA), but long-term risks of dependency following cancer surgery have not been described beyond 1 year. We evaluated healthcare dependency by examining homecare use following cancer surgery for OA. Methods: We conducted a population-based study of adults >=70 years old undergoing cancer surgery over 2007-2017. Outcomes were receipt and intensity of homecare from post-operative discharge to 5 years after surgery. Receipt of homecare was a dichotomous recurrent event, with censoring for admission to long-term care facility or death. Multivariable Andersen-Gill and Poisson regression models examined predictors of receipt and intensity of homecare. Results: Of 82,037 patients, 52,119 (63.5%) were female and 6,443 (7.8%) were frail, 5,865 (7,1% were admitted to long-term care facilities, and 34,044 (41.5%) died during follow-up. Homecare use was highest (43.7% of eligible patients) in post-operative month-1. Need for homecare subsequently decreased to stabilize at a new baseline between year-1 (13.9%) and year-5 (12.6%). Among patients not receiving pre-operative homecare, 10.9% became long-term users at year-5 after surgery. Advancing age, female sex, frailty, high intensity surgery, more recent period of surgery, and receipt of pre-operative homecare were associated with increased hazards of post-operative homecare. Intensity of homecare went from 10.3 to 10.1 days per patient-month from month-1 to year-1 and reached 12.0 days per patient-month at year-5. Type of homecare services changed from predominant nursing care in year-1 (51.9%) to increasing personal support services from year-2 (69.6%) to year-5 (77.5%). Conclusion: Receipt of homecare increased long-term after cancer surgery for OA, with a peak in the first 6 months and plateau thereafter at a new baseline. One-tenth of patients without pre-operative homecare became long-term homecare users post-operatively, indicating healthcare dependence, focused on personal support services from year-2 to 5. This information is important for patient counselling regarding surgery and preparation for recovery. By 2030, 1/5 of the US population will be older than 65. We aim to identify preoperative risk factors contributing to postoperative complications in geriatric surgical oncology patients. We hypothesize patients undergoing neoadjuvant treatment, with 4 co-morbidities, and advanced age will have increased incidences of postoperative complications. We performed a single institution retrospective study of patients 65 years old who underwent surgery for an intra-abdominal malignancy from 1/2018-12/2018. We excluded patients < 65 years, minor surgery, emergent and non-cancer operations. Patients were stratified by age in 5-year intervals. Patient characteristics including concurrent comorbidities, polypharmacy ( 5 home mediations), neoadjuvant therapy, ASA score and operation were collected. Perioperative outcomes (complications, discharge location and 90-day readmission rate) were compared. Statistical analysis was performed with SAS. We compared differences in outcomes using Chi-squared tests. We identified 202 patients, 167 (83%) underwent formal resection for an alimentary or hepatopancreatobiliary malignancy. The perioperative complication rate was 50% (n=83). 90-day readmission rate was 23% (n=38). The perioperative mortality rate was 4% (n=7) and 13% (n=22) were discharged to an acute inpatient rehabilitation/skilled nursing facility (AIR/SNF). Patient age, race, surgical anatomic location or receipt of neoadjuvant therapy did not statistically increase perioperative outcomes. Four or more comorbidities (p=0.03), morbid obesity (p=0.03), polypharmacy (p=0.01) were associated with increased rates of perioperative complications. Discharge to an AIR/SNF was more likely in patients with 4 comorbidities (p=0.03). Male patients (p=0.04) and patients with polypharmacy (p=0.03) were more likely to be re-admitted within 90 days of discharge. This study shows the impact of comorbidities and polypharmacy contribute significantly to perioperative outcomes rather than age or neoadjuvant therapy alone. Multidisciplinary preoperative assessment of geriatric oncology patients may allow tailored preoperative interventions to improve postoperative outcomes in these patients. Background: Institution-free time alive and at home is a patientcentred outcome of particular relevance to older adults. Long-term outcomes are currently unavailable. We examined institution-free survival (IFS) over 5 years following cancer surgery for older adults. Methods: We conducted a population-based cohort study of older adults (>70 years old) undergoing cancer resection between 2006-2018 using linked health administrative data. We defined IFS as <14 institution-days within one year, previously identified as clinically important. Institution-days included any day spent in emergency or inpatient acute care, outpatient procedures, inpatient mental health care, inpatient rehabilitation, or skilled nursing facilities. We estimated IFS using the Kaplan-Meier method from date of surgery. We examined factors associated with inferior IFS using multivariable Cox models. Results: We included 82,037 individuals with median follow-up of 46 (IQR,23-80) months. Overall, patients spent most days alive at home: median 98 (IQR,94-99) days at home per 100 days alive in postoperative year 1 and 99 (IQR,97-100) in years 2 to 5. IFS at 1 and 5 years were 70.3% (95%CI,70.0%-70.6%) and 53.2% (95%CI,52.8%-53.5%), respectively. IFS dropped most in year 1 and stabilized over years 2 to 5. Increasing age, male sex, preoperative frailty, high intensity surgery, and later period of surgery were independently associated with inferior IFS. Institution-days were most commonly due to emergency or inpatient acute care. Acute care days were largely cancer-related in year 1 (58.2% of institution days), but not in subsequent years (17.6% in year 2 to 11.5% in year 5). Conclusion: Following cancer surgery, the majority of older adults are surviving and having <14 institution-days. The greatest loss in IFS occurred within year 1, owing to cancer-related acute care days, and stabilized thereafter. Most institution-days were in acute care; targeting those at increased risk of inferior IFS for greater transitional care planning and home supports may improve patient care and experience. This is important patient-centred information for preoperative counselling and preparedness planning. Introduction: Merkel Cell Carcinoma (MCC) is a rare, aggressive skin cancer which typically presents at an older age. In this population wide study, we aimed to examine the differences in presentation and outcomes between younger patients and older patients with MCC. Methods: All patients >18 years of age diagnosed with MCC between 2003 and 2014 were identified in the National Cancer Database. Patients with 1 or more comorbidities were excluded to eliminate immunosuppression related MCC. Patients were divided into age groups (<60 or 60) to compare sociodemographic, tumor, and survival differences by age. Results: Of 6560 patients, 14.1% were <60 years old; 62.1% were male with equal gender distribution between age groups. Patients <60 years of age were significantly more likely to present with MCC of the trunk and extremities compared to adults >60 years (47.6% vs 37.6%, p<0.001), who were more likely to present with head and neck MCC (17.4% vs 33%, p<0.001). Younger patients were also more likely to undergo axillary staging (45% vs 37%, p <0.001), which was subsequently associated with an increase in pathologic nodal disease compared to older patients (p<0.001). However, older patients were more likely to have positive surgical margins (10.2% vs 6.6%, p<0.001). Three-year (<60: 74%, 60: 51.5%, p<0.001) and 5-year (<60: 65.5%, 60: 39.3%, p<0.001) overall survival was significantly lower in older patients. After controlling for competing factors, older age significantly increased the risk of death after an MCC diagnosis (HR:2.21, CI:1.88-2.6) along with positive margin status (p<0.001) and increasing stage (p<0.001), whereas female sex (HR:0.73, CI: 0.68-0.79) and extremity location (HR:0.7, CI:0.64-0.77) improved OS in all patients. Conclusions: Excluding patients with comorbidities, younger and female MCC patients have an independently improved survival compared to older and male patients with MCC. Younger patients are also more likely to present on the extremities and with nodal disease, the latter of which may reflect increased utilization of axillary staging and a more aggressive approach in the younger age group. Aim: Glycolysis has gained attention as a new therapeutic target for inhibiting tumor growth and angiogenesis. The aim of this study was to clarify the role of phosphofructokinase-2/fructose-2, 6-bisphosphatase 3 (PFKFB3) in tumor cells and tumor endothelial cells of hepatocellular carcinoma (HCC). Methods: The relationship between the long-term prognosis and the PFKFB3 expression in tumor cells and tumor endothelial cells were examined in a total of 99 consecutive patients who underwent curative surgery for HCC in our Hospital. Introduction: The utility of transarterial chemoembolization (TACE) for unresectable hepatocellular carcinoma (HCC) in the setting of advanced cirrhosis is questionable. Current HCC prognostic models include parameters of liver dysfunction and those related to HCC. This heterogeneity may impact the clinical utility of current prognostic treatment models. We hypothesize that the serum albumin level at diagnosis independently predicts 90-day mortality of HCC. Methods: Our institutional registry was queried for all HCC cases treated with TACE, chemotherapy and best supportive care (BSC). Patients undergoing surgery or ablation were excluded. Standard demographic and clinical variables were included. Inverse probability of treatment weighted propensity scores was created using age, stage, gender, AFP, albumin, Model for End-Stage Liver Disease (MELD) score at diagnosis and time period were modelled to estimate survival. A logistic regression model was created to identify variables associated with survival less than 90 days Results: 1127 HCC patients comprised this study. TACE was performed on 277 (27%), chemotherapy only 198 (16%), BSC 682 (56%) and 123 patients received both (10%). In those treated with TACE, variables associated with survival<90 days included MELD (OR 1.06, 95%CI:1.03-1.10, p<0.01) and regional stage (OR 2.27, 95%CI:1.35-3.82, p<0.01), albumin (OR 0.58, 95%CI:0.43-0.78, p<0.01). For those receiving TACE and albumin<3 at diagnosis, the probability of survival<90 days by stage was 25% (regional) and 38% (metastatic) (Figure 1 ). The estimated mean survival was 11.4 months (95%CI:9.91-12.8, p<0.01) for patients treated with BSC. TACE increased mean survival an additional 8.8 months (95%CI:5.44-12.2, p<0.01). Conclusion: Locoregional TACE significantly improved survival when compared to BSC. As MELD score increases and albumin decreases, the TACE survival benefit diminishes significantly for unresectable HCC. Using this model serum albumin at diagnosis is predictive of 90 day survival across all stages of HCC.

Relationship between serum albumin at diagnosis and survival probability < 90 days for all stages of hepatocellular carcinoma To understand potential molecular mechanisms that drive hepatocellular carcinoma (HCC) progression, we performed RNA sequencing from paired adjacent non-tumor liver and HCC tumor tissues of nine local HCC patients which revealed significant involvement of a gene called Six Transmembrane Epithelial Antigen of Prostate 2 (STEAP2). STEAP2 is a metalloreductase involved in the reduction of iron and copper ions. We found significantly higher levels of total copper in HCC tissues. The upregulation of STEAP2 expression in HCC was confirmed in other published HCC gene expression datasets. Hepatic copper overload is associated with Wilson's disease and a known risk factor for HCC, therefore, we hypothesize that STEAP2 upregulation and copper accumulation contribute to HCC progression. Paired HCC and adjacent non-tumor tissues were collected for RNA sequencing, metal ion measurement, and measurements of STEAP2 levels with RT-qPCR and Western blot. Public HCC datasets were queried for STEAP2 expression. HCC cell lines with knockdown (KD) and overexpression (OE) of STEAP2 were created to perform mechanistic studies including measurements of copper levels and MAP kinase activities, cell proliferation, migration, and anchorage independent growth in vitro and tumor growth in vivo. STEAP2 is significantly upregulated in various HCC gene expression datasets; its expression is positively associated with tumor grade. STEAP2 KD in HCC cell lines decreased cell growth, migration, invasion, and xenograft tumor growth, while STEAP2 OE showed opposite effects. STEAP2 KD in HCC cells also reduced intracellular copper levels and activation of stress-activated MAP kinases including p38 and JNK. Treatment with copper rescued the reduced HCC cell migration due to STEATP2 KD and activated p38 and JNK. Furthermore, treatment with p38 or JNK inhibitors significantly inhibited cell migration. Thus, STEAP2 appears to play a malignant-promoting role in HCC cells by driving migration/invasion via increased copper levels and MAP kinase activities. Our study uncovered a novel molecular mechanism contributing to HCC malignancy and a potential therapeutic target for HCC treatment. INTRODUCTION 30-50% of patients undergoing hepatectomy for hepatocellular carcinoma (HCC) develop an early intrahepatic recurrence (EIR). Current diagnostic modalities have poor sensitivity for small HCC lesions which contribute to a high EIR rate. Diagnostic modalities with higher sensitivity for HCC are needed. Our group developed an antibody against glypican-3 (GPC3), a cell-surface protein that is expressed in HCC, which specifically identified tumors with zirconium-89 ( 89 Zr) immuno-positron emission tomography (PET). This study aimed to validate 89 Zr-aGPC3 immuno-PET, determine its accuracy in identifying small HCC lesions and measuring tumor size in an orthotopic xenograft HCC model. METHODS A murine orthotopic xenograft model of HCC was generated by subcapsular injection of human hepG2 cells, which natively express GPC3. Serum alpha-fetoprotein (AFP) concentration, an established surrogate for tumor size in our model, was measured. Animals were then injected with 89 Zr-aGPC3 and imaged with PET/CT scanner. Gross tumor volume (GTV) was measured by a combination of fixed thresholding and manual segmentation methods by a user blinded to experimental conditions of each mouse. RESULTS 58 tumors were imaged by immuno-PET with a mean GTV of 0.26cm 3 (SD 0.39cm 3 ), corresponding to a spherical tumor diameter of 8mm and a mean serum AFP of 235,724 ng/mL (SD 291, 438) . The smallest GTV was 0.001cm 3 , corresponding to a spherical tumor diameter of 1.2mm. The largest GTV was 1.54cm 3 , corresponding to a spherical tumor diameter of 14.4mm. GTV measurements accurately correlated with serum AFP (Figure 1 ) confirming accuracy of tumor volume measurement. 30% of tumors measured were less than 5mm in diameter, the lower limit of sensitivity for magnetic resonance imaging for HCC. CONCLUSION Immuno-PET using 89 Zr-aGPC3 specifically identifies sub-centimeter HCCs and accurately measures tumor volume in a GPC3 expressing orthotopic tumor model. With improved sensitivity and specificity of HCC detection using this technology, the EIR rate may be reduced and selection of patients who would benefit from upfront surgery versus neoadjuvant therapy may be determined. Introduction: The overall (OS) and recurrence-free survival (RFS) benefit of negative margins for hepatocellular carcinoma (HCC) has been clearly demonstrated. However, there is no consensus regarding the optimal resection margin width. We aimed to assess the impact of hepatic resection margin width for solitary HCC on OS, RFS and liver-specific recurrence-free survival (LSRFS). Methods: Clinicopathologic data were retrospectively collected for solitary HCC patients who underwent a negative margin hepatectomy between 7/1992 and 12/2015. Margin width was categorized in tertiles as "narrow" (</=0.3cm), "intermediate" (0.31cm-1.0cm) or "wide" (>1.0cm) and OS, RFS and LSRFS were compared between width groups. Results: Of the 268 HCC patients who underwent a hepatectomy, 178 had a solitary lesion and negative margins. Most were male (76%) with a median age of 63y, median MELD score 8 and median pathologic tumor size of 5.2cm. Almost all (93%) were Child-Pugh class A. Median margin width was 0.5cm (range: 0.1-5.0cm). There was no significant difference in survival between margin width groups: median OS was 52, 74 and 77 months (p=0.76), RFS 31, 34 and 16 months (p=0.36), and LSRFS 33, 45 and 35 months (p=0.76) for the narrow, intermediate and wide margin groups, respectively. In multivariate analyses, margin width was not an independent predictor of OS (p=0.77), RFS (p=0.74), or LSRFS (p=0.92). Findings were similar in all subgroups analyzed (tumor </=5cm or >5cm, vascular invasion, T1 or T2/T3). Similar results were obtained when margin width was dichotomized or analyzed as a continuous variable. Conclusion: Narrow margins appear to be oncologically safe and the feasibility of achieving wide margins should not be a determinant of resectability.

Kaplan-Meier overall survival curves. Narrow (</=0.3cm) vs. Intermediate (0.31cm-1.0cm) vs. Wide (>1.0cm) resection margin width groups. P-values were derived from log-rank tests. Background: Fibrolamellar carcinoma (FLC) is a rare liver tumor that affects young, non-cirrhotic patients due to a DNAJB1-PRKACA fusion protein.

Lymph node and peritoneal metastases are frequent at diagnosis, and late recurrences are common. Immune checkpoint inhibitor therapy has been proposed, but not enough is known about the immune response to FLC to guide treatment approaches. Methods: Using Nanostring, we analyzed the gene expression patterns of 4 FLC tumors as compared to paired non-tumor liver (NTL). We performed multiplex immunohistochemistry (mIHC) staining for CD8, CD4, CD68, FoxP3, and CK18 on 11 tumor samples. Using fresh FLC tissue, we cut 250 m slices with a vibratome and cultured the slices with either IgG control antibody, anti-PD-1, IgG + AMD3100 (a CXCR4 small molecule inhibitor), or anti-PD-1 + AMD3100 for 24 hours. The slices were incubated with SR-FLICA (a reagent that fluorescently labels cells undergoing apoptosis) and then stained for EpCAM, CD8, and nuclei. The slices were imaged live using a confocal microscope to determine CD8 + cell migration and antitumor cytotoxicity. Results: CD4 + cells comprised the majority of leukocytes in FLC (4.5% of all cells in the tumor compared to 0.3% CD8 + and 0.09% CD68 + ), but were lower in FLC compared to NTL by mRNA (p=0.01) and IHC (p=0.10). Regulatory T cells were 5% of CD4 + cells and 0.2% all cells. Higher densities of CD4 + and CD8 + cells were found in the intratumoral stroma compared to the nonstromal tumor by IHC (p=0.01, p=0.16). Treatment with anti-PD-1 + AMD3100 resulted in a significant increase in CD8 + cell migration to within 20 m of carcinoma cells compared to control or AMD3100 monotherapy (p=0.01, p=0.07). Cytotoxic T cell activity was suggested by the finding that carcinoma cell apoptosis in the combination treatment group was only seen when CD8 + cells were in close proximity (<20 m) to carcinoma cells (p=0.008). Conclusions: CD4 + cells are the predominate lymphocyte in FLC and are significantly more concentrated in the intratumoral stroma. Despite lower CD8 + cell numbers, treatment with combination anti-PD-1 and anti-CXCR4 therapy potentiates CD8 + migration and cytotoxic activity in FLC slices. Introduction Successful complete cytoreductive surgery(CRS) and hyperthermic intra-peritoneal chemotherapy(HIPEC) in peritoneal malignancies is dependant on pre-operative imaging, with the view to allow prediction of disease burden, and aid in patient selection. We aim to evaluate the accuracy of pre-operative computed tomography(CT) imaging and validate previously proposed CT prognostic factors(CT-PF), and their role in predicting the success rate of complete CRS and HIPEC. Methods A retrospective analysis of patients who were planned for CRS and HIPEC from 2000-2014 was performed. A comparison of pre-operative radiological findings between patients who successfully underwent complete CRS and HIPEC and those in whom planned procedure was not completed was performed. Additionally, based on our previously published CT-PF score, 8 pre-operative imaging CT-PFs were selected, and correlated with intra-operative findings. Results A total of 123 patients were planned for CRS and HIPEC during the study. Eighty-two(66.7%) underwent conventional CT and 41(33.3%) underwent positive-emission tomography CT. Twenty-five(20.3%) patients demonstrated small bowel disease, 26(21.1%) had gross ascites, 30(23.8%) had lesions larger than 5cm, 15(11.9%) had omental caking, 3(2.44%) had bowel obstruction, 31(25.2%) had perihepatic nodules, 14(11.4%) had ureteric obstruction and 1(0.81%) had biliary obstruction. A total of 84(68.3%) patients successfully underwent CRS and HIPEC, while 39(31.7%) were treated with palliative intent due to extensive disease, not amenable to complete CRS. Pre-operative imaging findings of small bowel disease, gross ascites and omental caking were significant factors for predicting incomplete CRS and HIPEC (table 1) . The mean CT-PF calculated was 0.87 for those that underwent complete CRS and HIPEC, and 1.87 for those unable to undergo the complete operation. Conclusion Pre-operative radiological findings of small bowel disease, gross ascites and omental caking are factors that predict failure of complete CRS. Higher CT-PF scores predicted failure for complete CRS and is a useful tool that can be adopted for pre-operative counselling and surgical planning.

Analysis of pre-operative imaging factors between patients who successfully completed CRS and HIPEC and those who were unable to Background: Cytoreductive surgery with hyperthermic intraperitoneal chemotherapy (CRS/HIPEC) is used to treat peritoneal surface malignancies of various histopathologic origins. Due to the complexity of CRS/HIPEC, which often involves multi-visceral resection, patients experience a longer hospital length of stay (LOS) than typical surgical patients. We sought to use predictive modeling to identify factors associated with prolonged LOS to better recognize patients who would benefit from enhanced recovery protocols. Methods: Primary outcome was prolonged LOS (> median LOS). Multivariable logistic regression was performed and model performance was assessed using k-folds cross-validation. Area under the receiver operating characteristic curve (AUC) and the Hosmer-Lemeshow (HL) test were calculated to determine discriminatory ability and goodness-of-fit. Odds ratios (OR) and 95% confidence intervals (CI) were reported. Results: There were a total of 366 patients who underwent CRS/HIPEC and met inclusion criteria during the study period. The median (quartiles) LOS for these patients was 10 days (7, 12 days) . Multivariable logistic regression revealed factors predictive of increased LOS including symptomatic peritoneal metastasis, intraoperative blood loss, case duration, completeness of cytoreduction score, splenectomy, hepatic resection, and anastomosis created. A risk calculator constructed using these variables demonstrated high PPV and NPV in predicting increased LOS. Additionally, after stratification by the extremes of hospital LOS, younger patient age (< 65 years) was found to be correlated with hospital LOS < 7 days while factors indicative of high disease burden were associated with hospital LOS > 12 days. Conclusions: Factors associated with prolonged LOS following CRS/HIPEC were identified and integrated into a risk calculator that can facilitate post-operative triaging of resources and stratification of risk. Patients of older age or with symptomatic peritoneal metastasis may be considered for more rigorous prehabilitation and enhanced recovery protocols given their increased LOS in the hospital. Background Lymphocyte-to-Monocyte Ratio (LMR) is associated with poor overall survival in patients with colorectal cancer, but this association has not yet been studied in the setting of peritoneal metastasis. This retrospective study explores whether LMR is a prognostic marker for survival for patients with colorectal cancer or appendiceal cancer undergoing HIPEC and CRS. Methods We identified all patients who underwent CRS/HIPEC for colorectal or appendiceal adenocarcinoma at our institution between 2007 and 2017. The relationship between LMR, NLR, PLR, and MPV were examined with respect to clinicopathologic variables and analyzed with Kaplan-Meier log-rank survival analyses and multivariable Cox regression models with overall survival (OS) and disease-free survival (DFS). Results We identified 216 patients who underwent CRS/HIPEC for colorectal or appendiceal adenocarcinoma. On univariable analysis, decreased LMR (<3.71) was associated with worse ECOG performance status (p 0.05) and higher ASA (p 0.05).

Five-year overall survival for patients with LMR <3.71 was 35.2% compared to 60.4% for LMR >3.71 (HR of 2.0; 95% CI 1.1-3.5, p=0.017, see Figure 1 ). On multivariable Cox-regression, only LMR<3.71 and PCI were significantly associated with overall survival (p 0.05, p<0.001 respectively). Conclusions Preoperative LMR is an independent preoperative predictor of overall survival in patients undergoing CRS/HIPEC for colorectal and appendiceal cancer. To our knowledge, this is the first study to identify LMR's potential role as a noninvasive marker that may serve as a preoperative surrogate for disease burden and help guide discussion regarding the severity of disease. Background Compared with colon primaries, peritoneal carcinomatosis (PC) from rectal cancers have been associated with poorer outcomes after Cytoreductive Surgery and Hyperthermic Intraperitoneal Chemotherapy (CRS-HIPEC). 1 We aim to determine if the site of primary disease affects disease free survival (DFS) and overall survival (OS) in our tertiary centre. Methods 119 patients were analysed from 1999 to 2018. 104 had colon cancers -75 left-sided, 26 right-sided and 3 of the transverse colon. 15 had rectal cancers. Kaplan-Meier (KM) and Cox regression analyses were performed. Results Median OS was 38 months in the rectal group and 37.2 months(95% CI 25. 5-45.8) .7) for right-sided tumors(p=0.069). The left-sided tumor group had better 1, 3 and 5 year survival rates. On univariate analysis for OS there were no significant factors but for DFS, age, race, pre-operative albumin, PCI score, duration of peritonectomy, intra-operative blood loss and post-operative blood transfusion were significant. All factors except PCI score and intra-operative blood loss remained significant on multi-variate analysis. Conclusion There is no significant difference in OS and DFS between the rectal and nonrectal, and left-and right-sided tumors. Patients with PC from rectal cancers should be considered for CRS-HIPEC. 1 Introduction: Cytoreduction and hyperthermic intraperitoneal chemotherapy (CRS/HIPEC) remain the mainstay of treatment for peritoneal carcinomatosis. With advancements in pre-operative imaging it is possible to establish peritoneal carcinomatosis index (PCI) scores. However, the significance of these scores is uncertain. Additionally, situations where CRS is abandoned due to unresectable disease can lead to significant morbidity. Considering the potential for high morbidity without any oncologic benefit, we aim to identify patients in the pre-operative phase who are at high risk for incomplete CRS leading to aborted HIPEC. Methods: All patients scheduled for CRS/HIPEC at the University of Kansas Medical Center between the years 2009 and 2018 were included irrespective of successful procedure completion. Clinical data was abstracted from patient charts. CT-PCI scores were calculated by a team of staff radiologists with body-reading sub-specialization. Multiple regression analysis provided relative weights to the clinical variables. These were used to generate a nomogram for clinical use. Results:There were 329 patients included in this analysis. The aborted procedure rate was 33.1%. CT-PCI score was predictive of aborted CRS (OR 1.105/pt, p<0.01). The cancer type category was predictive of aborted CRS (p=0.005) with colon cancer, rectal cancer, and "other" cancer types each independently predicting failure to complete the procedure (OR 3.22 p=0.007, OR 12.36 p=0.002, and OR 8.62 p<0.01 respectively). A predictive nomogram was generated using total PCI score and cancer type to predict the likelihood of aborted CRS. The predictive model was associated with an AUC-ROC of 0.881 signifying strong predictive ability of the model. Conclusions: Clinical data was successfully used to create a predictive nomogram for office use to predict the likelihood of aborted CRS/HIPEC. This model still requires external validation; however this can be used to guide surgeons in the patient selection process. Certain patients with advanced peritoneal disease are more likely to be unresectable based on cancer type. Background Epithelial ovarian carcinoma (EOC) is the leading cause of gynecological cancer-related death. Prognostic factors related with poor outcomes as clinical stage, histological subtype and grade, size of residual disease after debulking surgery and platinum response were found to be reliable and independent predictors, due to the fact that most of the patients are diagnosed in advanced stages other factors related to metabolism, nutrition, and inflammatory response have been evaluated in recent studies. Objective To evaluate the clinical and prognostic significance of the pretreatment prognostic nutritional index (PNI), inflammation-associated blood cell markers: neutrophil-lymphocyte ratio (NLR) and platelet-lymphocyte ratio (PLR) and monocyte-lymphocyte ratio (MLR) in EOC stages III-IV. Methods A retrospective analysis from records of patients with advanced EOC from 2011 to 2016 was performed. PNI was calculated according to the following formula: 10 serum albumin (g/L) + 0.005 lymphocyte count (per mm). And We calculated the ratio of NL, PL, and ML. Univariate Cox proportional hazards analyses were performed on continuous data leukocyte, neutrophils, lymphocyte and Platelet counts, PLR, NLR MLR, and PNI. The cut-off value was calculated by ROC analyses and defined as 40.0 for PNI, <3 for NLR, <270 for PLR, and <0.33 for MLR. We used the Youden index to further confirmation of our results Results Of 244 cases identified with EOC stages III-IV, all of them were treated with debulking surgery and neoadjuvant or adjuvant chemotherapy. AUC for PNI was 0.616, decreased PNI was associated with shorter OS (p=0.011), but we didn't find differences in PFS (0.249). None of the other factors analyzed showed correlation with OS an PFS. Multivariate analysis confirmed PNI is an independent prognostic factor associated with overall survival. Conclusion Decreased Pretreatment PNI was found to be an independent factor of poor prognostic in patients with advanced EOC in terms of overall survival. The analyses of the other systemic inflammatory markers were not useful as prognostic markers in this retrospective study. Objectives -To investigate any disparities in terms of use, post-operative complications and survival between the two most frequently administered chemotherapeutic regimens during hyperthermic intraperitoneal chemotherapy (HIPEC) for colorectal peritoneal metastases: Mitomycin C (MMC) and Oxaliplatin. Methods -All patients diagnosed between 2005 and 2017 with synchronous peritoneal metastases of colorectal origin who underwent cytoreductive surgery (CRS) and HIPEC were selected from the Netherlands Cancer Registry. As the application of Oxaliplatin only started in 2014, a subgroup of patients treated in 2014-2017 was created in which overall survival (OS) was compared between patients treated with either MMC or Oxaliplatin. Cox regression analyses were used to assess the individual influence of HIPECregimen on OS. Results -The study population comprised 579 patients, of whom 297 were diagnosed in 2014-2017. Among them, 177 (59.6%) received MMC and 120 (40.4%) received Oxaliplatin. Patient-and tumor characteristics were similar in both groups, except for primary tumor location (p=0.048). 1-, 2-and 3-year OS were 84.6% vs. 85.8%, 61.6 vs. 63.9% in patients treated with MMC and 44.7% vs. 53.5% in patients treated with Oxaliplatin. Median OS was 30.7 months in the MMC group vs. 46.6 months in the Oxaliplatin group (p=0.181). After correction for co-variables, differences remained nonsignificant (adjusted HR 0.77 [0.56-1.12]). Conclusions -The current study demonstrated no significant differences in overall survival between patients treated with either MMC or Oxaliplatin. Based on these results, the choice of either HIPEC-regimen seems justified in the treatment of peritoneal metastases from colorectal cancer. Background: Preliminary data from the PRODIGE 7 study has questioned the benefit for intraperitoneal chemotherapy (IPC) in the management of peritoneal carcinomatosis secondary to colorectal cancer (CRC). In the United States, despite smaller series showing improved survival, national outcomes are unknown. We sought to evaluate trends and survival outcomes of patients undergoing CRS-IPC using the National Cancer Database (NCDB). Materials and methods: CRS-IPC cases for colon and rectal cancer were extracted from the NCDB from 2004-2016. The number of CRS-IPC operations performed for peritoneal carcinomatosis secondary to colorectal cancer were analyzed for four respective time periods-2004-2007, 2008-2010 and 2011-2013, and 2014-2016 . Clinicopathologic data were examined and survival analyzed using Kaplan Meier analysis. Results: A total of 2,459 patients underwent CRS-IPC during the study period for CRC with 96% of cases being performed for colon cancer. Median age was 55 years with 50.4% being female, and 87% white. Operations were most often performed at academic centers (60%), or comprehensive community cancer programs (18%). There was an increase in the number of CRS-IPC performed for all institutions in the study period (p=0.001). The use of CRS-IPC increased over subsequent time periords for patients with intermediate/poorly differentiated tumors (4%, 19%, 29%, and 43%, p=0.002) but not those with lymph node positive disease or LVI (p>0.05).

Median overall survival for the entire cohort was 53.0 and remained stable over time (median OS 56.6mo, 92.2mo, 76.2mo, and NR p=0.07 ). In addition, 90day mortality was 2.3% with no significant change over time (p>0.05). Conclusions: There has been a demonstrable increase in the number of CRS-IPC cases performed for CRC in the US over the last decade. Furthermore, CRS-IPC is increasingly being performed for patients with more biologically aggressive disease without poorer outcomes. These data suggest that the role of CRS-IPC needs further investigation and data from the PRODIGE 7 study should be interpreted with caution. Purposes: Although systemic chemotherapy has been improved, peritoneal carcinomatosis remains a factor of poor prognosis in patients with colorectal cancer. Cytoreductive surgery combined with hyperthermic intraperitoneal chemotherapy became a curative option for peritoneal metastases in selected patients. The conventional intraperitoneal chemotherapeutic drugs have been encountered with many drawbacks, for instance, high cytotoxicity, lack of target specificity, and very short serum half-life. Cisplatin and some of its platinum complexed derivatives are among the most active therapeutic agents. Our purpose of this study is to establish the new drug to overcome many drawbacks of previous intraperitoneal chemotherapeutic agents. Methods : Through rational drug design, we have developed a novel cell-penetrating peptide (CPP) and platinum drug conjugate, to overcome previously mentioned issues. We experimented with various colorectal cancer cell lines, and also used the MTT assay to verify the effectiveness. We also confirmed the internalization directly using the microscope. Results : Cell-penetrating peptide facilitates intracellular delivery of various cargos through diverse pathways and acts as a delivery vehicle in various colorectal cancer cell lines. MTT assay confirmed that the survival rate of colorectal cancer cell lines and the resulting IC 50 values of the conjugate were much reduced than those of parent drug. Cellular internalization was observed by confocal microscopy, which has confirmed that the conjugate was well internalized. Conclusions : Our data demonstrate that CPP-mediated platinum conjugate is very promising and efficient for cancer therapy. Thus, our results suggested that CPP-mediated platinum conjugates may have great potential applications in intraperitoneal chemotherapy for peritoneal metastasis with colorectal cancer.

Metastases from Colorectal Origin: A Dutch Populationbased Study C. Bakkers, 1 * R.v. Meer, 2 R. Roumen, 2 R. Lurvink, 1 V. Lemmens, 3 F.v. Erning, 3 I.D. Hingh. 1 1. Surgical Oncology, Catharina Hospital, Eindhoven, Netherlands; 2. Maxima Medical Center, Veldhoven, Netherlands; 3. Netherlands Comprehensive Cancer Organization, Utrecht, Netherlands. Objective: The aim of this nationwide study was to provide insight in the incidence, risk factors, treatment and survival of patients with ovarian metastases from colorectal cancer (CRC). Methods: Data from the Netherlands Cancer Registry were used. All newly diagnosed female CRC patients between 2008 and 2016 were included. Treatments were categorized as follows: cytoreductive surgery followed by hyperthermic intraperitoneal chemotherapy (CRS-HIPEC); resection of the primary tumor; palliative treatment; and no treatment. Overall survival (OS) was investigated using Kaplan-Meier and multivariable Cox Introduction: The incidence of peritoneal carcinomatosis (PC) after curative resection of stage II/III colon cancer varies widely in the literature. Certain features, such as T4 tumors, mucinous or signet ring histologies, are considered high risk for PC. However, the impact of these high-risk features on PC incidence is unclear. Methods: A retrospective analysis was performed on patients 18 years old with surgically-resected stage II/III colonic adenocarcinoma treated at our institution from 2008-2018. Clinicopathologic features, treatment characteristics, time to and sites of recurrence were recorded. Patients with reported high-risk features (pT3N0-2 with mucinous or signet ring components, any pT4, any pN1c, clinical/pathologic perforation) were identified. The remaining stage II/III patients without high-risk features were used for comparison. Results: Of 97 patients identified, 36/97 (37.1%) were stage II and 61/97 (62.9%) were stage III. Median follow-up time was 27 (1-146) months. Adjuvant systemic treatment was administered to 57/97 (58.8%) patients. Overall incidence of PC was 9/97 (9.3%) and the median time to PC was 20 (7-37) months. At time of PC identification, 6/9 (66.7%) were isolated PC, while 3/9 (33.3%) were PC with additional metastatic sites. Non-PC local recurrences occurred in 6/97 (6.2%), liver metastases in 10/97 (10.3%) and lung metastases in 3/97 (3.1%). The high-risk group and the comparison group contained 61 and 36 patients, respectively. Mucinous or signet ring histology was present in 25/61 (41.0%) patients and perforation occurred in 40/61 (65.6%). Incidence of PC was not significantly different between the high-risk and comparison groups (high-risk 9.8% vs. comparison 8.3%, p=0.81). Median time to PC was also not significantly different (high-risk 23 (14-37) months vs. comparison 20 (7-36) months, p=0.74). Conclusion: The overall incidence of PC in patients with stage II/III colon cancer patients was 9.3%. Because PC incidence was not significantly higher in patients with risk factors previously associated with PC, future studies designed to evaluate PC in stage II/III colon cancer should include all patients regardless of high-risk features. Introduction: Patients with colorectal peritoneal carcinomatosis (CRPC) are increasingly treated with cytoreductive surgery and hyperthermic intraperitoneal chemotherapy (CRS/HIPEC), as there is improved survival compared with systemic therapy alone. Unfortunately, data defining preoperative risk factors for poor oncologic outcomes after aggressive surgical intervention are limited. We therefore sought to determine the prognostic value of preoperative CEA, CA125, and CA19-9 on disease progression or recurrence after CRS/ HIPEC. Methods: Patients with CRPC treated with curative intent CRS/HIPEC from twelve participating sites in the United States from 2000-2017 wereidentified. Advancement-free survival (AFS), defined as progression or recurrence of disease, was the primary outcome. Results: In 279 patients who met inclusion criteria, the mean age was 53.5 years (+/-12), the median PCI was 11 (IQR 11), and the rate of disease advancement was 63.8%. Elevated CA 19-9 was associated with dismal AFS at two years [8.9%: elevated vs. 30%: normal; p=0.007]. There was no statistically significant association between AFS and either elevated CEA or elevated CA 125. All patients with concurrent elevation of CEA and CA 19-9 had experienced advancement of their disease within two years. In 113 patients who underwent CRS/HIPEC without prior neoadjuvant therapy (NAT), CA19-9 emerged as the sole independent predictor of worse AFS [HR 2.88 p=0.048]. No association between tumor markers and AFS was noted in the NAT + CRS/HIPEC subgroup. Conclusion: Elevated CA19-9 is associated with decreased AFS in patients with CRPC. All patients with concurrent elevation of CA19-9 and CEA had disease advancement within two years of diagnosis, potentially highlighting a subset with aggressive disease biology. While traditionally CEA is the main tumor marker assessed in colon cancer, we found that CA19-9, in combination with CEA, may inform preoperative risk stratification for poor oncologic outcomes; however, prospective studies are required to confirm this association. Introduction: Cytoreductive Surgery (CRS) with Hyperthermic Intraperitoneal Chemotherapy (HIPEC) is increasingly utilized for peritoneal malignancies, despite significant costs related to the procedure and hospitalization. To address potentially modifiable factors contributing to excessive cost, we sought to determine factors predictive of high cost of care for patients undergoing CRS/HIPEC. Methods: A merged institutional fiscal and CRS/HIPEC database was queried for adult patients from 2014-2018. Hospital cost was defined as cost of the index procedure and hospitalization; high cost cases were defined as > 75 th percentile for costs. Bivariate analyses for high cost cases were performed using ANOVA and Chi Squared as indicated. All significant tumor-related, patient and surgeon specific (p<0.05) variables were entered in a multivariable linear regression for hospital cost. A separate linear regression was performed for length of stay (LOS). Results: In total, 59 patients underwent 61 CRS/HIPEC procedures. The median cost was $30,552 (25,122-42,570) for high cost and $17,650 (15, 042) for non-high cost cases. Median LOS was 8 (7-11.5) days and ICU stay was 1 (1-1.5) day. LOS, ICU stay length, EBL, and the comorbidity index Risk of Mortality (ROM) 3 were predictive of hospital cost on multivariable analysis (Table 1) . Multivariable factors associated with LOS were OR time (p=0.04), ROM and Severity of Illness (SOI) 4 (p=0.04 and p<0.01 respectively), ASA 3 (p=0.01) ostomy creation (p=0.045) and discharge to SNF vs. home (p=0.03). Notably, complications by Clavien-Dindo grade, tumor-related factors, such as PCI and CCR, and surgeon were not predictive of cost nor LOS. Discussion: Our results, the first to identify predictors of high cost of CRS/HIPEC-related care, reveal cost was largely related to length and intensity of care during the index hospitalization. In turn, these drivers were primarily influenced by patient comorbidities rather than complications, tumor biology or surgeon related factors. As payors and trackers of quality metrics continue to scrutinize high resource care, these data are important in demonstrating that not all high cost HIPEC care is modifiable. Background Postoperative complications(POCs) are associated with worse oncologic outcomes in various cancer histologies. The impact of POCs on the survival of patients with appendiceal or colorectal cancer after cytoreductive surgery/heated intraperitoneal chemotherapy(CRS/HIPEC) is unknown. Methods US HIPEC Collaborative(2000-17) was reviewed for pts who underwent CCR0/1 CRS/HIPEC for appendiceal/colorectal cancer. Analysis was stratified by non-invasive appendiceal neoplasm vs invasive appendiceal/ colorectal adenocarcinoma. POCs were grouped into infectious, cardiopulmonary, thromboembolic, and intestinal dysmotility. Primary outcomes were overall survival(OS) and recurrence-free survival(RFS). Results Of 1304pts, median age was 55yrs(IQR 47-64), 41% were male(n=537), 33% had noninvasive appendiceal(n=426) and 67% had invasive appendiceal/colorectal adenocarcinoma(n=878). In the non-invasive appendiceal cohort, POCs were identified in 55%(n=233), and 3 yr OS and RFS did not differ between patients who experienced a complication and those who did not(OS 94vs94% p=0.26 Fig1A; RFS 68vs60% p=0.15 Fig1B). In the invasive appendiceal/colorectal adenocarcinoma cohort, however, POCs (63%; n=555) were associated with decreased 3 yr OS(59vs74% p<0.01 Fig1C) and RFS(32vs42% p<0.01 Fig1D). Infectious POCs were most common(35%; n=196). On MV analysis accounting for gender, PCI and incomplete resection(CCR1), infectious POCs in particular were associated with decreased OS compared to no complication(HR 2.08 95%CI 1.48-2.93 p<0.01) or other types of complications(HR 1.7 95%CI 1.28-2.25 p<0.01). This association persisted for infectious POCs and reduced RFS(HR 1.61 95%CI 1.23-2.10 p<0.01). Conclusion Postoperative complications are associated with decreased OS and RFS after CRS/HIPEC for invasive histology, but not for an indolent disease like non-invasive appendiceal neoplasm. Of all complication types, infectious complications are the main driver for this association. The exact mechanism is not known, but may be immunologic. Efforts must target best practices and standardized prevention strategies to minimize infectious POCs. INTRODUCTION: Composite measures may be superior to individual metrics for evaluating quality for complex surgical procedures. While recent studies have introduced the concept of a textbook outcome (TO) as a novel composite metric for measuring postoperative outcomes, the incidence and importance of a TO has not been previously characterized among patients undergoing cytoreductive surgery (CRS) for peritoneal surface malignancies (PSM). METHODS: All patients who underwent CRS hyperthermic intraperitoneal chemotherapy (HIPEC) between 1999-2017 from 12 academic institutions were included. Using expert consensus, TO was defined as: completeness of cytoreduction (CC) 0/1, no reoperation, no readmission within 90-days, no mortality within 90-days, no grade 2 complications, no prolonged hospital stay >75th percentile, and discharge to home. Univariate and multivariable analyses were used to determine factors associated with TO. RESULTS: Among 1963 patients who underwent CRS, 43.1% were male, the median age was 55 years, the most common histology was appendiceal (65.0%), the mean peritoneal carcinomatosis index (PCI) was 14.6, and 86.0% underwent CC0/1 resection. Only 28.9% of patients achieved a TO, limited mostly due to postoperative complications (Figure) . On multivariable analysis, factors associated with achieving a TO were albumin 3. TO was associated with increased overall survival (median 156 months vs 55.7 months, p<0.01) even after controlling for confounders on Cox regression analysis (hazard ratio: 2.3, p<0.01). CONCLUSIONS: Among patients undergoing CRS HIPEC for PSM, failure to achieve a TO is common and associated with significantly worse overall survival. The use of composite metrics such as TO may enhance patient counseling by informing expectations and can serve as a valuable tool for measuring patient-level hospital performance. Background. During cytoreductive surgery and hyperthermic intraperitoneal chemotherapy (CRS/HIPEC), surgeons are reluctant to perform unprotected pelvic anastomosis despite lack of supporting data. We analyzed pelvic anastomosis outcomes and factors that influence ostomy creation in CRS/HIPEC patients. Methods. A single-center, descriptive study of patients with rectal resection during CRS/HIPEC was conducted using a prospective database. Surgical variables were reviewed. Multinominal logistic regression of outcomes (end or protective ostomy) was performed with pre-and intraoperative factors as predictors. Results. Overall, 274 of 789 CRS/HIPEC patients underwent rectal resection, including 243 (89%) with pelvic anastomosis [232 (85%) without ostomy, 11 (4%) with protective ileostomy] and 31 (11%) with no anastomosis [16 (6%) with end colostomy, 15 (5%) with end ileostomy]. Median age was 57 and 29% (79) were male. Of 243 pelvic anastomosis patients, 3 (1.2%) had rectal anastomotic leaks, including 1 with a protective ileostomy. Other anastomotic leaks occurred in 3.6%. Overall, 13% had Clavien-Dindo complications IIIB and the readmission rate was 30%. Mortality at 30-days and 100-days was 0.4% and 2.2%, respectively. Male gender and primary rectal cancer were associated with protective ileostomy (odds ratio (OR)=7.01, 95%CI: 1.6-31.5, p=0.011, and OR=16.4, 95%CI: 3-88.4, p=0.001, respectively). Male gender and prior pelvic surgery were associated with end colostomy (OR=13.9, 95%CI: 3.7-53, p<0.0001, and OR=17.2, 95%CI: 3.8-78.6, p<0.0001). Conclusions. Pelvic bowel reconstruction without protective or end ostomy during CRS/HIPEC is safe. Protective ileostomy is associated with male gender and primary rectal cancer. End colostomy is associated with male gender and prior pelvic surgery. Background: Cytoreductive surgery (CRS) with hyperthermic intraperitoneal chemotherapy (HIPEC) requires a large abdominal incision for complete exposure of the peritoneal cavity. For patients with low-volume disease, a laparoscopic approach may be feasible and may result in reduced postoperative morbidity. Methods:Patients with a peritoneal surface malignancy of appendiceal, colorectal, or primary peritoneal origin with an estimated peritoneal carcinomatosis index (PCI) of </= 10 were enrolled in this single-arm, prospective study and underwent laparoscopic exploration for cytoreductive surgery and HIPEC. A balanced, two-staged trial design was utilized with the primary end-point of 30-day postoperative morbidity. 18 patients were needed to detect a hypothesized 28% decrease in postoperative morbidity (from 43 to 15%) with 81% power. Results:A total of 31 patients were enrolled in the study from January 2015 through December 2018. There were 13 screen failures due to findings of a higher disease burden at surgery (12) or ineligible primary tumor site (1) . Eighteen patients went on to receive laparoscopic CRS/HIPEC. Median age was 57 years (range 29 -78), BMI was 24.7 (17.5 -38.1), and PCI 3 (0 -10). 15 patients (83%) had appendiceal and 3 (17%) had colon primary tumors. 9 patients (50%) had adenocarcinoma histology and 9 had low-grade mucinous appendiceal primaries. All patients had complete cytoreduction. Minor 30-day complications occurred in 3 patients (17%). At a median follow-up of There were no major complications or deaths. Table 1 shows operative and postoperative variables between the patients on study and patients with similar PCI who underwent open surgery. Estimated blood loss, length of stay, and inpatient morbidity were significantly lower in the laparoscopic group. Conclusions:Laparoscopic CRS/HIPEC is feasible to evaluate prospectively but screen failure is common. The findings in this study suggest that it can be performed safely with improved short-term postoperative outcomes compared to open surgery for patients with low-volume peritoneal surface malignancy.

Categorical variables reported as N (%); continuous variables reported as median (range). Background Patient age is often a significant factor in preoperative selection for major abdominal surgery. Its association with postoperative outcomes in patients undergoing cytoreductive surgery (CRS) and hyperthermic intraperitoneal chemotherapy (HIPEC) remains ill-defined. Methods The US HIPEC Collaborative database (2000-2017) was reviewed for patients who underwent a CCR0/1 CRS/HIPEC. Age was categorized into <65 or >65 yrs. Primary outcomes were postoperative major complications, readmission, 30-day mortality, and non-home discharge (NHD). Analysis was stratified by disease histology: non-invasive (appendiceal LAMN/HAMN), and invasive (appendiceal/colorectal adenocarcinoma). Results Of 1090 patients identified, 22% were >65 yrs (n=240), 59% were female (n=646), 25% had non-invasive (n=276) and 51% had invasive (n=555) histology. Median PCI was 13 (IQR7-20). Patients >65 yrs had a higher rate of major complications (37vs26%, p=0.02), readmission (28vs22%, p=0.05), 30-day mortality (3vs1%, p=0.02), and NHD (12vs5%, p<0.01) compared to those <65 yrs. On multivariable analysis accounting for extent of disease as measured by PCI, for non-invasive histology, age >65 yrs was an independent predictor for NHD (OR: 2.54, 95% CI: 1.08-5.99, p=0.03), but not major complications. For invasive histology, even when accounting for PCI, age >65 yrs was an independent predictor for both NHD (OR: 2.54, 95% CI: 1.08-5.98, p=0.03) and major complications (OR: 2.04, 95% CI: 1.16-3.59, p=0.05). Age was not associated with hospital readmission or 30-day mortality for any histology. Conclusions Regardless of histology, patients >65 yrs are nearly at three-fold increased risk for non-home discharge after CRS/HIPEC. For invasive histology, age >65 yrs is also associated with increased major complication rates, but the procedure seems to be better tolerated when performed for indolent biology. These data inform preoperative counseling and risk stratification. Early planning for discharge disposition in this high-risk population can potentially translate to cost savings. Introduction: The definition of resectability in pancreatic adenocarcinoma (PDAC) has been a topic of great debate. An R0 resection is important since it impacts overall survival. It is unclear if the distinction between borderline resectable (BR) and locally advanced (LA) disease at the time of diagnosis can predict resectability in the era of neoadjuvant therapy with FOLFIRINOX and chemoradiation. Methods: Clinicopathologic data was retrospectively collected for all patients who received total neoadjuvant therapy [TNT, FOLF-IRINOX followed by chemoradiation (25 or 50.4 Gy)] for PDAC and underwent resection between 2011 and 2018. Patients were classified as BR or LA by a multidisciplinary team according to the AHPBA/SSO/SSAT guidelines. Results: A total of 161 patients were included, of whom 69 (42.9%) were classified as BR and 92 (57.1%) as LA at diagnosis. Median age of the cohort was 63 yo (IQR:57-69), 50% were male, median tumor size on CT was 2.3cm (IQR:1.2-3.5), an R0 resection was achieved in 126 (80%), and 53 (34%) had positive lymph nodes. There was no difference between the two groups in baseline clinicopathologic factors, including the rate of R0 resection [BR 54 (79%) vs LA 72 (80%), p=0.955], except for tumor size on CT at diagnosis (BR 3.1cm vs LA 3.3cm, p=0.024). LA PDAC was associated with a worse median DFS (LA 21m vs BR 31m, p=0.038) and OS [LA 41m vs BR not reached (56% of patients were alive at the end of the observation period), p=0.018]. On multivariate analysis, high CA19-9 at diagnosis (p=0.025), bigger tumor size (p=0.039), lymph node positivity (HR:2.4,p=0.01) and R1 resection (HR:1.6,p=0.017) were associated with poor OS, but not BR/LA status (p=0.098). Regarding DFS, multivariate analysis demonstrated that only lymph node positivity (HR:1.7,p=0.007) and R1 resection (HR:1.7,p=0.004) were associated with poor outcomes. Conclusion: In the era of neoadjuvant FOLF-IRINOX and chemoradiation, the original classification of PDACs as BR or LA does not predict resectability if the patient had a clinical response and was explored in the operating room and is not an independent predictor of survival. Background: Conversion of frozen section (FS) R1 pancreatic neck margin (NM) to permanent section (PS) R0 by additional resection (FS:R1 PS:R0; NM "conversion") during upfront pancreaticoduodenectomy (PD) for pancreatic ductal adenocarcinoma (PDAC) has not been shown to prolong overall survival (OS). The impact of neoadjuvant chemotherapy (NAC) on this practice and OS is unknown. We sought to reevaluate the role of NM "conversion" in a multi-institutional cohort of PDAC patients receiving NAC. Methods: We reviewed PDAC patients (80% borderline resectable/locally advanced [BR/LA]) undergoing PD after NAC at 7 academic centers (2010-2018). Multivariable models examined the association of PS:R0, PS:R1, and NM "conversion" with OS by NM status alone and by all evaluable margin (NM, SMA, bile duct) status. Results: All 272 patients received at least 2 (median 4) cycles of NAC (71% FOLFIRINOX or gemcitabine/nab-paclitaxel) and 37.1% received neoadjuvant radiation (NXRT). At median follow-up of 42m PS:R0 (n=238, 87.5%) resection was associated with improved OS compared with PS:R1(n=34, 12.5%) resection (median 25 vs 16m, P=0.02). When assessed by NM status alone, no significant difference in OS was observed between PS:R0-NM, PS:R1-NM, and NM "conversion" cohorts (P=0.26; Figure 1A ). When assessed by all evaluable margins (NM, SMA, bile duct), PS:R0-all was associated with improved OS compared with PS:R1-all and NM "conversion" groups (P=0.023; Figure 1B) , with no OS difference between PS:R1-all and NM "conversion" groups (P=0.9). The discordance in OS of the PS:R1 cohort when evaluated by NM-only or all margin status was driven by SMA margin status. On Cox modeling, SMA positivity (HR 2.4, P=0.008), but not pancreatic NM positivity (HR 1.1, P=0.87), was associated with worse OS. NXRT was not significantly associated with OS (HR=1.1, P=0.62). Conclusion: In a cohort of predominantly BR/LA PDAC patients undergoing PD following NAC, R1 rate is seen in a minority of patients. As with patients undergoing upfront PD, pursuing a negative NM after NAC and PD appears to be inconsequential. Background: Carbohydrate antigen 19-9 (CA19-9) is a surrogate for disease burden and neoadjuvant therapy (NT) response for patients with pancreatic adenocarcinoma (PDAC). CA19-9 normalization is ideal, but the importance of varying CA19-9 trajectory is undefined. We aimed to create CA19-9 response categories that offer prognostic stratification. Methods: Data for resected PDAC patients treated with NT (7/2011-10/2018) were abstracted from a prospective database. Patients with 3 CA19-9 time points (concurrent bilirubin <2mg/dL) were grouped by CA19-9 trajectories. Nonproducers (<1U/ml) were excluded, and 35U/ml was considered normal(ized). "Classification and regression tree" analyses were performed for trajectory association with post-resection progression-free survival (PFS) and overall (OS). Results: Of 281 patients, 162 had eligible CA19-9 values. Median baseline CA19-9 was 101. The overall 2-yr post-resection PFS and OS were 37% and 63%. Patients with normalization (53%) had better 2-yr PFS (47% vs. 28%, p=0.01) and OS (75% vs. 49%, p=0.01), compared to non-normalization. CA19-9 trajectory was analyzed by shape, direction, and ultimate normalization. Patients were fit into 5 response types ("A-B-C-D-E"; Fig. 1 ). Type A was "always" decreasing to normal, B was "bidirectional" but eventually normal, C was "consistently" normal, D was any "decrease" without normalization, and E was "elevated" always without normalization. These response types were associated with differential PFS (p=0.02) and OS (Fig 1; p=0 .04). Type A-B responses resulted in the best 2-yr PFS (53%,30%) and OS (72%,100%). Response types that never normalized (D-E) had worse outcomes, with Type E at high risk for early recurrence. Type C had intermediate outcomes. Conclusions: Categorizing PDAC patients into novel A-B-C-D-E strata, both CA19-9 response trajectory and normalization were associated with prognosis. These data have implications for guiding treatment decisions, specifically for surgeons and patients estimating the post-resection benefit in patients with borderline biology and/ or borderline medical operability. Background: For borderline resectable (BRPC) or locally advanced pancreatic cancer (LAPC), neoadjuvant (NAT) FOLFIRINOX or gemcitabine plus nab-paclitaxel (GnP) are standard treatment options and these regimens have shown a survival advantage over single-agent gemcitabine. However, the role of these modern therapeutic regimens in elderly patients is debatable. In this analysis, we evaluated the outcomes of neoadjuvant treatment (NAT) with combination chemotherapy in elderly patients. Methods: 230 consecutive patients who underwent neoadjuvant treatment for BRPC/ LAPC discussed and planned for NAT at the University of Colorado Cancer Center from January 2011 to March 2019 were reviewed. 214 patients who received FOLFIRINOX (n=143) or GnP (n=71) were eligible for analysis. We divided all patients into three groups (<70, 70-74, 75 years) and compared the short-term and long-term outcomes. Results: Of 214 patients, patients <70 (n = 147) received FOLFIRINOX more frequently than the other groups Background Postoperative pancreatic fistula (POPF) drives morbidity and mortality following pancreatectomy. Over the past decade, use of neoadjuvant chemotherapy (NAC) has increased in the treatment of potentially-resectable pancreatic ductal adenocarcinoma (PDAC). Previous studies have reported lower POPF rates with NAC. This study examined the effect of NAC on POPF rates and postoperative outcomes in PDAC. Methods The 2014 to 2017 NSQIP Targeted Pancreatectomy dataset was queried to identify patients with non-metastatic PDAC who underwent curative-intent resections. Propensity score matching was used to stratify patients by receipt of NAC. The effect of NAC on POPF rates and perioperative morbidity was estimated using conditional logistic regression. Multivariate analysis was used to identify predictors of POPF. Results 6,863 patients met the study's inclusion criteria, of those 1908 (27.8%) received NAC and 4955 (72.2%) did not (NNAC). 2,062 patients were matched 1:1 from each group with adequate adjustment. After matching, there were no differences in age, BMI, gland texture, or pancreatic duct size among both groups (P>0.05). Patients receiving NAC had significantly lower POPF rates (9.0% vs. 14.5%; P<0.001), and the majority were categorized as grade I (5.1% vs 9.5%). NAC patients had significantly decreased overall 30-day morbidity (40.4% vs 49.5%; P<0.001). Specifically, pneumonia (2.3% vs 4.1%), organ space infections (7.9% vs 13.2%), sepsis (5.2% vs 8.0%) and delayed gastric emptying (10.1% vs 14.8%) occurred less frequently in the NAC group (P<0.05). Postoperative mortality and unplanned reoperations were similar (0.8% vs 1.0% and 4.3% vs 4.5%, respectively; P>0.05). On multivariate analysis, receipt of NAC was an independent predictor of decreased POPF rates (HR: 0.73 [0.56-0.94]; P=0.016). Other factors included gland texture, duct size, male gender and lower BMI. Conclusions In this propensity-matched, population-based cohort study of PDAC patients, NAC was associated with lower POPF rates and overall major complications. Those findings reaffirm the safety of NAC in appropriately selected patients. Introduction: As the rates of neoadjuvant chemotherapy (NT) prior to resection of pancreatic cancer increase, the role of adjuvant chemotherapy (AT) remains unknown. Minimal evidence exists, coming only from small single-institutional studies, about factors influencing who receives AT after NT. We present the first population study using a large national database to study current determinants of AT in this patient population. Methods: Data from the National Cancer Data Base was used to identify patients with pancreatic adenocarcinoma stage I-III who underwent NT followed by surgical resection. Univariate and multivariate analysis was used to compare patients who received NT alone and those who received AT after NT. Results: A cohort of 6,728 patients was identified. Of the patients who received NT followed by surgical resection, 35.3% also received AT. AT after NT increased from 5.2% in 2006-2008 to 41.8% in 2015-2016 . Multivariate analysis found that demographic and socioeconomic factors associated with increased odds of receiving AT including younger age (OR 1.21 p=.003), Caucasians (OR 1.43 p=.006), private insurance (OR 1.16 p=.012), and treatment at a community center (OR 1.25 p=.0005). Pathologic factors associated with AT included >20 nodes recovered (OR 1.51 p<.0001), pathologic stage 2 (OR 1.22 p<.0001), and positive nodes (OR 1.45 p<.0001). On univariate analysis, positive surgical margin was associated with receipt of AT (p=.0092), however on multivariate analysis, surgical margin status was no longer statistically significant (OR 1.06 p=.40). Recent year of diagnosis, 2015-2016 vs 2006-2008, was associated with a 50% increase likelihood of receiving AT (OR 1.5 p=.0004). Conclusion: As the rates of NT in localized pancreatic cancer increase, the role of addition systemic treatment remains extremely important. We found that despite lack of Level 1 evidence of the benefit of AT, the practice has increased nationally from 2006 to 2016. Both demographic and pathologic factors influence receipt of AT after NT. Further research is required to fully elucidate the real-world practice patterns, role of, and benefits from AT after NT. BACKGROUND: Metastatic foregut cancers (MFC) are frequently associated with debilitating symptoms that have significant negative impact on patients' quality of life. Palliative care (PC) is effective in mitigating disease-, psychosocial-, and treatment-related effects. However, PC remains heavily underutilized. The aim of our study was to characterize the rate of PC utilization in MFC and determine the impact of various clinicopathologic and socioeconomic factors associated with the receipt of PC. METHODS: We conducted a retrospective review of 277,957 National Cancer Database patients diagnosed with MFC between 2004-2013. Chi-squared tests were used to analyze differences between groups. Logistic regression was performed to assess the impact of factors on the likelihood of receiving PC. RESULTS: PC utilization increased among all groups over time (12.3% 2004-2006 vs. 14.7% 2007-2010 vs. 16.4% 2011-2013 for all cancers). Female sex, Medicaid, median income < $46,000/year, higher education level, higher Charlson/Deyo Score, and pancreatic or biliary cancers were associated with increased likelihood of PC interventions. Additionally, patients treated at an academic center or integrated network cancer program were more likely to receive PC than patients treated in the community setting. When receipt of PC was stratified by race, Hispanics were significantly less likely to have undergone palliative interventions compared to non-Hispanic Whites (OR 0.70, 95% CI 0.66-0.73). Patients with Medicare or private insurance were less likely to receive PC than uninsured patients (OR 0.92, 95% CI 0.87-0.97 and 0.81, 95% CI 0.77-0.89, respectively). CONCLUSION: Differences in palliative care receipt rates exist with regards to racial/ethnic and socioeconomic factors such as insurance status, median household income, and education level. Patients receiving care in the community setting were also less likely to receive palliative care than those treated at academic or integrated network cancer program centers. Further studies are needed to delineate why these disparities exist with regards to palliative care utilization. Background: Evidence to guide management of locally advanced pancreas cancer (LAPC) is limited to retrospective reports. Numerous studies suggest attitudes regarding management are inconsistent between surgeons and across institutions. We sought to examine the influence of surgeon volume on attitudes regarding LAPC management. Methods: An extensive electronic survey was distributed by email to an international cohort of pancreas surgeons. Data collected included practice characteristics, preferences for staging and management, and 6 vignettes (with detailed videos of post-neoadjuvant arterial and venous CT imaging) to assess attitudes regarding eligibility for exploration. Surgeons were classified according to self-reported annual pancreatectomy volumes: >10 cases (high volume [HV]), >25 cases (double volume [DV]), and >50 cases (very high volume [VHV] ). The propensity to consider exploration for 6 vignettes was examined across multiple volume threshholds: HV for the primary analysis, and DV and VHV for sensitivity analyses. Results: A total of 153 eligible responses were received from 4 continents. Median duration of practice was 12 years (IQR 6-20). A majority of participants (N=132, 86.3%) were considered HV, while 58.2% (N=89) were classified as DV, and 22.2% (N=34) as VHV. HV surgeons had practiced longer (median 12 vs. 7 years, P=0.01), were more likely to "always" attend multidisciplinary conference (72.0% vs. 52.4%, P=0.03) and "always" recommend neoadjuvant chemotherapy (75.0% vs. 52.4%, P=0.006), and were more likely to offer minimally invasive surgery (37.1% vs. 9.5%, P=0.01), and vein resection (99.2% vs. 80.9%, P<0.001). No differences were found across HV status in the propensity to offer exploration for any of 6 vignettes, the rationale to avoid exploration, or the alternative treatment recommended. On sensitivity analyses, these findings were unchanged. Conclusions: In an international survey of pancreas surgeons, the propensity to consider exploration for LAPC was not associated with multiple categories of self-reported surgeon volume. Better evidence is needed to define the optimal management approach to LAPC. 

Background Pancreatic cancer is a disease of the elderly. In this population, maximizing post-operative recovery is imperative as surgical morbidity has been associated with decreased long-term survival. Recently, the NSQIP-based modified frailty index has been shown to correlate with clinical outcomes. We hypothesize that a new modified frailty score (MFS) of five variables can be used to determine outcomes after pancreatectomy. Methods Patients who underwent pancreatic resection were identified from the NSQIP Participant Use File from 2014-2017. A five-factor score was determined using variables from the Canadian Study of Health and Aging 70-item index which also corresponded to NSQIP variables. The MFS was categorized as low (1), intermediate (2) , and high ( 3). Bivariable and multivariable analyses were performed to evaluate the impact of frailty on discharge status, complications, readmission, and mortality. P values of < 0.05 were considered statistically significant. Results There were 24,674 patients included in this analysis. MFS significantly predicted discharge to a facility, as did other factors such as ASA class, albumin, weight loss, and complications. The odds of discharge to a facility were doubled with a high compared to a low MFS (OR 1.24 vs 2.68, p < 0.01). The major complication rate was 22%. Clavien-Dindo complication class > III were significantly associated with a low or high MFS (RR 1.09 vs 1.29, respectively, p < 0.05). This was also significantly associated with obesity, ASA of > 4, open approach, radiation, and drain placement. Readmissions were significantly associated with high MFS (RR 1.6; 95% CI 1.197, 2.035; p = 0.001), male sex, obesity, drain placement, chemotherapy, and radiation. 30-day mortality was associated with an intermediate MFS (OR 1.6; 95% CI 1.006, 2.558; p = 0.048). Conclusions The modified frailty score was associated with discharge status, complications, readmissions, and postoperative mortality in patients after pancreatectomy. Assessment of frailty may be useful in identifying patients at risk for poor outcomes after surgery. Intro: Colloid carcinoma (CC) of the pancreas is a histopathologic variant of pancreatic cancer associated with intraductal papillary mucinous neoplasms. Data for this uncommon entity are limited and discordant. We describe the clinicopathologic characteristics, treatment, and outcomes for CC relative to infiltrating duct carcinoma (IDC), intraductal papillary mucinous carcinoma (IPMC), and adenocarcinoma not otherwise specified (AC-NOS). Methods: Adults diagnosed with stage I-IV CC, IDC, IPMC, or AC-NOS between 2004-2015 in the National Cancer Database were eligible for inclusion. Patients missing information on treatment (n=10,067) were excluded. Kaplan-Meier curves were used to assess median overall survival for each subtype. Results: Overall, 251,681 patients were included: 6,787 (3%) had CC, 27,681 (11%) had IDC, 908 (<1%) had IPMC, and 216,305 (86%) had AC-NOS. Patients with CC were more likely to present with Stage IV disease than IDC and IPMC (46% vs. 14% and 7%) but were similar to AC-NOS (49%). Patients with CC were less likely to undergo surgery than IDC and IMPC (27% vs. 73% and 76%) but more likely than AC-NOS (14%). Only 65% of stage I and II CC patients underwent surgery; 80% of the rest were not offered surgery. Patients with CC had shorter median survival compared to IDC and IPMC (9 mo. vs. 16 mo. and 33 mo.) but better than AC-NOS (6 mo.). Among stage I-II patients, median survival for CC treated with chemotherapy and surgery was significantly better than IDC and AC-NOS (36 mo. vs. 25 mo. and 24 mo.) and slightly shorter than IPMC (40 mo.). Conclusions: Colloid carcinoma often presents at an advanced stage with poor prognosis similar to most pancreatic adenocarcinoma. However, early stage CC patients undergoing multimodality therapy have better survival than those with IDC and AC-NOS. Surgery is underutilized in this group, and treatment should be guided by early histopathologic diagnosis and multimodality therapy with both surgery and chemotherapy for patients with colloid carcinoma. Background: Reports on the safety and benefits of minimally invasive pancreaticoduodenectomy compared to open pancreaticoduodenectomy (OPD) have demonstrated mixed results. One multi-institution study comparing robotic pancreaticoduodenectomy (RPD) versus OPD have demonstrated decreased complications associated with RPD. We sought to evaluate the overall morbidity of RPD versus OPD using ACS NSQIP. Methods: This is a retrospective cohort study from 2014-2017 using the ACS NSQIP multiinstitutional clinical registry. Procedure-targeted pancreatectomy PUF was queried for all patients undergoing PD. Factors associated with morbidity and mortality were evaluated using multivariate logistic regression (MVA) and propensity score matching (PSM). Results: Of 13,110 PDs performed over the study period (55% pancreatic ductal adenocarcinoma [PDAC]): 12,612 (96.2%) were OPD and 498 (3.8%) were RPD. Patients who underwent RPD versus OPD were less likely to have any complications (46.8% vs. 53.3%; p=0.004), surgical complications (42.6% vs. 48.6%; p=0.008), wound complications (6.2% vs. 9.1%; p=0.029), clinically relevant postoperative pancreatic fistulas (11.9% vs. 15.6%; p=0.026), sepsis (6.2% vs. 9.3%; p=0.019), and pneumonia (1.6% vs. 3.8%; p=0.012). On MVA, OPD was associated with increased complications compared with RPD ( Introduction: The association between high procedural volume and improved outcomes is generally accepted for most high risk, low volume procedures, such as pancreatic or esophageal surgery, which supports the beneficial effects of the concentration of these procedures in high volume centers. Several cutoffs have been reported to define a high volume center for pancreatic surgery; however, the definition of high volume center for distal pancreatectomy still needs to be determined. The purpose of this study was to evaluate the association between hospital procedure volume and patient mortality for patients undergoing distal pancreatectomy using a large database to determine an evidence-based threshold of hospital volume associated with improvement in postoperative mortality. Methods: Patients who underwent distal pancreatectomy were identified using the National Cancer Database from 2004 to 2015. The relationship between hospital volume and 90-day mortality was assessed. Logistic regression analysis and restricted cubic spline regression analysis was performed to determine the linear and non-linear association between mean hospital volume and mean 90-day mortality. Results: 13,307 patients underwent distal pancreatectomy at 1,081 unique hospitals. 30-and 90-day mortality of the study population were 1.77% (n=236) and 4.07% (n=542), respectively. Baseline characteristics and mean annual mortality of individual hospitals were determined. A logistic regression analysis was performed, which demonstrated that institutional volume is significantly associated with decreased 90-day mortality. The maximum improvement in 90-day mortality was seen if the annual hospital volume was greater than 6 (p<0.0001, OR=2.057 (1.697-2.493). We further explored the non-linear association between institutional volume and 90-day mortality, which demonstrated continued improvement in 90-day mortality with an increase in average hospital volume Fig1. Conclusion: This data suggest that hospital case volume has a direct impact on 90-day mortality after distal pancreatectomy. Based on our results, we recommend defining a high volume center as hospitals performing seven or more distal pancreatectomy cases per year. Background: Black-white disparities in pancreatic cancer outcomes are well documented. We aim to examine the effect of racial segregation on the diagnosis, management, and outcomes of pancreatic cancer in black patients. Methods: Black patients with pancreatic cancer in urban counties were identified using data from the 2018 submission of the Surveillance, Epidemiology and End Results program and 2010 Census. The racial index of dissimilarity (IoD), a validated proxy of racial segregation, was used to assess the evenness with which whites and blacks are distributed across census tracts within each county. Analyses were restricted to patients from areas within the lowest and highest quartile of IoD. Multivariable analyses were performed predicting advanced stage at diagnosis (AJCC stage III-IV) and the resection of localized disease (AJCC stage I-II). Propensity scores were created, and patients were matched, for the odds of living in a highly segregated county. After matching, survival analysis was performed via Kaplan-Meier method. Results: 2,911 black patients with pancreatic cancer were identified. Median IoD was 83.4 (IQR, 77.3-83.4) and 53.9 (IQR, 52.6-61.2) for patients in the lowest and highest segregated counties, respectively. Counties with high segregation were associated with a higher incidence of uninsurance (15.0% vs. 12.9%; p<0.001) and less than high-school level education (14.9% vs. 12.7%; p<0.001). On multivariable analysis, patients living in highly segregated counties were significantly more likely to be diagnosed at advanced stage (OR, 1.94; 95% CI, 1.21-3.10; p=0.006). However, a high index of dissimilarity was not predictive for the likelihood of undergoing pancreatic surgery in patients with localized disease (OR, 1.07; 95% CI, 0.81-1.40; p=0.649). After matching patients of all stages, those in the most segregated counties had significantly lower overall median survival compared to patients in less dissimilar counties (4 vs. 7 months; p<0.001). Conclusion: Black patients with pancreatic cancer in the most segregated counties are significantly more likely to be diagnosed at an advanced stage and have lower survival rates, regardless of socioeconomic status. Background: Most pancreatic ductal adenocarcinoma (PDAC) are diagnosed with advanced disease at which point tumors are mostly unresectable and have a dismal prognosis. This leaves core biopsy or endoscopic ultrasound-guided fine needle aspiration (EUS-FNA) as the sole source of tumor tissue for molecular characterization. Currently, these small tissue fragments are not considered suitable for genomic analysis, which is why this precious snapshot usually remains unexplored. Material and Methods: Applying an EpCAM-enrichment strategy, we show the feasibility and reproducibility of in-depth, molecular-barcoded, whole-exome sequencing (WES) analysis in real-world samples. Results: In a pilot phase, we used five resected tumors and three core biopsies to successfully test our enrichment strategy. This methodology also resulted in a significant increase of KRAS mutant allele frequency (MAF) (28.0% vs. 13.9%, p=0.03) in an independent set of 23 EUS-FNA samples. Validation experiments showed an excellent correlation between digital-droplet PCR and sequencing based MAF for KRAS and GNAS as well as for MYC amplification levels (all p<0.001). Despite low DNA level input, genomic landscape resembles previously described patterns from high-quality tissue sources. Potentially actionable mutations were present in 34% (8/23) of patients which could be used for trial inclusion. Additionally, significantly increased tumor mutational burden (TMB) correlated with single-nucleotide variants in DNA damage repair (DDR) genes (p<0.001) and showed prognostic significance. Correspondingly, high aneuploidy had prognostic implications (progression free survival: 104 days vs. 356 days, p=0.004). Response to platinum-based therapy was estimated by evaluating sample's homologous repair deficiency status using an integration of DDR genes, mutational signature analysis and aneuploidy score. Conclusion: Collectively, these results emphasize the value, feasibility and reproducibility of real-world small biopsies for molecular characterization of PDAC and their possible value for patient's treatment decisions. Introduction: While benchmarks for morbidity have been described for pancreatic surgery using composite measures of patient outcomes, there are no established short-term oncologic benchmarks that can be used to determine optimal surgery for pancreatic adenocarcinoma. Methods: We selected patients who underwent pancreaticoduodenectomy (PD) and distal pancreatectomy (DP) for pancreatic adenocarcinoma using the National Cancer Database between 2010 and 2016. Optimal oncologic surgery was defined as the absence of readmission and 90-day mortality, negative margins, retrieval of 12 lymph nodes, receipt of chemotherapy and a length of stay of 12 days for PD and 7 days for DP. We compared three time periods (2010-2012, 2013-2014 and 2015-2016) using open, laparoscopic and robotic approaches. Results: We identified 20 942 pancreaticoduodenectomies of which 30% had optimal oncologic surgery. Over the three-time periods, the percent of optimal oncologic surgery increased for all pancreaticoduodenectomies (24% vs 30% vs 36%, p<0.01), for open approach (24% vs 31% vs 25%, p<0.01) and for laparoscopic approach (28% vs 28% vs 39%, p<0.01). There was no statistical change for robotic pancreaticoduodenectomies (29% vs 38% vs 38%, p=0.21). On Cox regression, the year of surgery was associated with improved overall survival with HR = 0.91 (95% CI: 0.85-0.97) for 2015 compared to 2010. We identified 3 964 distal pancreatectomies of which 28% had optimal oncologic surgery. Over the three-time periods, the percent of optimal oncologic surgery increased for all distal pancreatectomies (17% vs 24% vs 27%, p<0.01), for open approach (16% vs 21% vs 24%, p<0.01) and for laparoscopic approach (20% vs 24% vs 28%, p=0.04). There was no change for robotic distal pancreatectomies (36% vs 32% vs 33%, p=0.87). On Cox regression, the year of surgery was associated with improved overall survival with HR = 0.88 (95% CI: 0.70-0.98) for 2015 compared to 2010. Conclusions: This new benchmark can be used as quality metric for optimal oncologic surgery for pancreatic adenocarcinoma. Outcomes of surgery for pancreatic cancer are improving, however there is still significant room for further improvement. Background: While the incidence of suicide has increased over the last decade, the prevalence of self-harm among patients with pancreatic cancer is poorly understood. We sought to define the incidence of self-harm behavior including suicide among patients with pancreatic cancer and identify factors associated with self-harm. Methods: Patients with pancreatic adenocarcinoma were identified in the Surveillance, Epidemiology, and End-Results-Medicare database. Patients were considered to have committed suicide if the cause of death was coded as "suicide and self-inflicted injury." Suicidal ideations were identified using ICD codes. Logistic regression was used to determine factors associated with self-harm, including suicide. Results: Among 43,576 patients (median age: 73y, IQR 67-79; 49% male) with pancreatic adenocarcinoma, the suicide rate was 103.3 per 100,000 people. Specifically, men with pancreatic cancer had a suicide rate of 196.5 per 100,000 individuals whereas women had a rate of 13.5 per 100,000 individuals. In contrast to death by suicide, the rate of suicidal ideation following diagnosis of pancreatic cancer was 172.0 cases per 100,000 people. Overall, the median time to first documented case of suicidal ideation was 111d (IQR 22-372) following diagnosis. Among individuals who underwent curative-intent surgery, the median time to self-harm behavior including suicide was 165d (IQR 9-272) following surgery. On multivariable analysis, patients who underwent surgical resection had higher odds of self-injurious behavior including death by suicide (OR 2.12, 95%CI 1.27-3.52). Also, the likelihood of self-injurious behavior was higher among males (OR 3.16, 95%CI 2.08-4.81) and patients with a history of mental illness (OR 4.87, . Conclusion: Individuals with pancreatic cancer had a suicide rate almost 7 times higher than the general U.S. population. Certain populations of patients including men, patients with a history of mental illness, and patients undergoing surgical resection were at higher risk of self-harm. Caregivers need to be aware of the mental stress present among patients with pancreatic cancer, which may increase the risk of suicidal ideation and suicide. Background: Among patients with pancreatic cancer, the association of mental illness with operative outcomes remains unknown. We sought to analyze how pre-existing mental illness was associated with outcomes following pancreatectomy. Methods: Patients with pancreatic cancer who underwent pancreatectomy were identified in Medicare Standard Analytic Files. Patients were classified as having mental illness if an ICD-CM code for anxiety, depression, bipolar disorder, schizophrenia or other psychotic disorder was recorded in one inpatient or two outpatient claims before cancer diagnosis. Outcomes included complication rate, length-of-stay, 30-and 90-day readmission and mortality rate. Results: Among 14,495 patients (median age: 73, IQR: 69-78; 53.2% male) who underwent a pancreatectomy, 1544 (10.7%) individuals had an existing mental illness diagnosis prior to pancreatic cancer diagnosis. Among individuals with pre-existing mental illness, 35% were diagnosed with only anxiety, 37% with only depression and 22% with depression and anxiety; a smaller subset (6%) was diagnosed with severe mental illness. Individuals with pre-existing mental illness were more likely to be white (93% vs 89%), female (61% vs 44%) and have a higher comorbidity burden (4, IQR 2-8 vs 3, IQR 2-8) (all p<0.05). Of note, patients with a pre-existing mental illness were more likely to have a complication following surgery (33% vs 29%) and die within 30-(6.6% vs 5.1%) and 90-days (11% vs 8.8%) (all p<0.05). On multivariable analysis pre-existing mental illness remained associated with risk of complication (OR 1.29, 95%CI 1.15-1.45), prolonged LOS (OR 1.21, 95%CI 1.07-1.37), 30-day readmission (OR 1.15, 95%CI 1.01-1.32), death within 30-(OR 1.37, 95%CI 1.11-1.71) and 90-days (OR 1.26, 95%CI 1.06-1.50) (all p<0.05). Discussion: One in ten patients with pancreatic cancer undergoing curative-intent surgery had pre-existing mental illness. Individuals with mental illness were more likely to have worse operative outcomes. Surgeons should address mental health concerns among cancer patients by providing adequate support services to improve outcomes for these high-risk patients. Background: Lymph node (LN) metastases affect overall survival (OS) in pancreatic cancer (PC). However, a LN sampling threshold does not exist. We examined the impact of nodal sampling on overall survival (OS). Methods: Patients with Stage I-III PC 55 years old who underwent curative resection from 2004-2016 were identified from the National Cancer Database (NCDB). After adjusting for age, gender, ethnicity, grade, stage, and Charlson-Deyo score (CDCC), multiple binomial logistic regression analyses assessed the impact of the LN ratio (LNR) on OS. LNR was defined as the number of positive LN over the number of LN examined. Regression analyses, a Cox-Regression, and a Kaplan-Meier (KM) survival curve assessed how many LN should be sampled. Results: A total of 13,673 patients, median age 69 years (55-90), were included. Most were Caucasian (86.6%) males with CDCC scores 1 (90.3%) and moderately to poorly differentiated PC (90.1%). Median number of LN examined was 15 (1-75) with a median of 1 positive LN (0-35). As expected, increased number of positive LNs was associated with reduced OS, p < 0.001. After data normalization, an increasing LNR was associated with a 12-fold likelihood of death [OR: 11.9, p < 0.001 (CI 6.0, 23.7)]. Subsequent regression models established evaluation of 16 LNs as the greatest predictor of OS. A regression model evaluating < or 16 lymph nodes correctly classified 74.5% of cases with a specificity of 99.6% (p < 0.001) (Figure 1 ). Examination of < 16 LN, Caucasian race, grade, stage, and higher CDCC were significantly associated with decreased OS. If 16 LNs were examined, patients had a 1.5-fold likelihood of improved OS, p < 0.001 (CI 1.4, 1.6 Background: Peritoneal recurrence after pancreatectomy for pancreatic cancer is not uncommon. Here we aim to evaluate if peritoneal cell-free tumor DNA present in the intraoperative lavage fluid can be used as a predictive biomarker of peritoneal recurrence. Methods: Under IRB-approved protocol, pre-and post-resection peritoneal lavage fluids were collected from pancreatic cancer patients undergoing curative pancreatectomy. Malignant ascites or peritoneal lavage fluids from patients with peritoneal carcinomatosis derived from KRAS-mutant cancers were also collected as positive control. Cell-free DNA was extracted from peritoneal fluids using QIAGEN DNA extraction kit. Droplet digital PCR (ddPCR) was performed using BioRad ddPCR KRAS G12/G13 screening kit according to manufacturer's instructions. Results: Mutant KRAS DNA was detected in all malignant ascites or peritoneal lavage fluids from patients with carcinomatosis (N=6) with a mutant allele frequency (MAF) of 0.11% to 4.68%. Mutant KRAS DNA was detected in 7/15 (47%) pre-resection and 8/11 (73%) post-resection peritoneal lavage samples with a MAF of 0.04% to 0.34%. With a median follow-up of 8 months, 6/10 (60%) patients with MAF of 0.11% in either pre-or post-resection peritoneal lavage fluid had peritoneal recurrence with a median time to recurrence of 3.8 months, while none of the patients with MAF of < 0.11% recurred (Figure 1 ). Conclusions: This pilot study suggests that detection of peritoneal cell free tumor DNA by ddPCR may be a useful test to predict peritoneal recurrence after pancreatectomy for pancreatic cancer. Background: SMAD4, a tumor suppressor gene, is inactivated or deleted in 60-90% of pancreatic adenocarcinomas (PDA). Loss of SMAD4 allows tumor progression by limiting cell cycle arrest and apoptosis and increasing metastases. SMAD4 deficient PDA cells are resistant to radiotherapy by upregulation of autophagy, a cell survival mechanism that allows intracellular recycling of macromolecules and organelles. Hydroxychloroquine (HCQ) is a known autophagy inhibitor, suggesting that HCQ treatment in SMAD4 deficient PDA may prevent therapeutic resistance induced by autophagy upregulation. Methods: We retrospectively analyzed the SMAD4 status of PDA patients enrolled in two prospective clinical trials evaluating preoperative HCQ. The first dose escalation trial demonstrated the safety of preoperative gemcitabine with HCQ (NCT01128296). More recently, a randomized trial of gemcitabine/nab-paclitaxel +/-HCQ evaluated Evans Grade histopathologic response (NCT01978184). Immunohistochemistry of resected specimens for SMAD4 was previously performed. Patients not treated at the max HCQ dose (n=5), not resected (n=2) or with SMAD4 staining unavailable were excluded (n=10). The effect of SMAD4 loss on response to HCQ and chemotherapy was studied for association with clinical outcome. Fisher's exact test and log-rank test were used to assess response and survival. Results: 52 patients receiving HCQ with neoadjuvant chemotherapy and 24 patients receiving neoadjuvant chemotherapy alone were studied. Of the HCQ group, 25 patients had SMAD4 loss (48%), compared with 15 control patients (63%, p=0.32). 76% of HCQ treated patients with SMAD4 loss obtained a histopathologic response 2A, compared to only 37% with SMAD4 intact (p=0.006). In the control group, loss of SMAD4 was associated with a nonsignificant detriment in 3 year OS (25% vs. 78%, p=0.3) that was less apparent in patients treated with HCQ (46% vs. 47%, p=0.18). Conclusions: The addition of HCQ to neoadjuvant chemotherapy in PDA may improve treatment response in patients with SMAD4 loss. Further study of the relationship between SMAD4, autophagy and treatment outcomes in PDA is warranted. Response to cancer immunotherapy in primary versus metastatic disease has not been well-studied. We found primary pancreatic ductal adenocarcinoma (PDA) is responsive to diverse immunotherapies whereas liver metastases are resistant. We discovered divergent immune landscapes in each compartment using a combination of flow cytometry, immunohistochemistry, and single cell RNA sequencing. Compared to primary tumor, liver metastases in both mice and humans are infiltrated by highly anergic T cells and MHCII lo IL10 + macrophages that are unable to present tumor-antigen. Moreover, a distinctive population of CD24 + CD44 -CD40 -B cells dominate liver metastases. These B cells are recruited to the metastatic milieu by Muc1 hi IL18 hi tumor cells, which are enriched >10-fold in liver metastases. Recruited B cells drive macrophagemediated adaptive immune-tolerance via CD200 and BTLA. Depleting B cells or targeting CD200/BTLA enhanced macrophage and T-cell immunogenicity and enabled immunotherapeutic efficacy of liver metastases. Our data detail the mechanistic underpinnings for compartment-specific immunotherapyresponsiveness and suggest that primary PDA models are poor surrogates for evaluating immunity in advanced disease. Introduction: Mortality due to pancreatic ductal adenocarcinoma (PDAC) continues to rise. Within the next decade, it will become the second most common cause of cancer death in the US. Both primary and metastatic, PDAC is an inflammatory disease with TGF-playing roles as both tumor suppressor and promoter. Modulation of its activity is being evaluated in treatment of the disease. We hypothesize that modulation of TGF-activity will vary the disease burden of primary and metastatic tumors after orthotopic implantation of KP1 cells, a murine pancreatic cancer cell derived from a genetically engineered mouse model. Methods: KP1 cells were marked with luciferase and grown in culture media with four treatment groups: PBS, Gemcitabine, Gemcitabine + TGF-, and Gemcitabine + Galunisertib (TGF-inhibitor). They were grown for 24 hours before being isolated and implanted orthotopically within the pancreas of B6 mice. The luminescence was measured at 7 and 14 days after implantation and the signals standardized against the PBS group. The mice were sacrificed after two weeks and were grouped into categories of those with macroscopic metastatic disease and those without. Results: In our analysis of the luminescence we found, at 14 days, with the score of the PBS group set at 1.0, that the relative signal of the Gemcitabine, Gemcitabine + TGF-, and Gemcitabine + Galunisertib groups were, 0.52 0.08 (P = 0.22), 0.78 0.12 (P = 0.36), 0.12 0.05 (P = 0.018) respectively. When the mice were sacrificed, 80% of mice in the PBS group had metastases, 60% in the Gemcitabine group, 100% in the Gemcitabine + TGF-group, and 20% in the Gemcitabine + Galunisertib group (P=0.046). There were no differences in the sizes of the tumors in the pancreas. Conclusion: In our model, mice treated with the combination therapy of Gemcitabine + Galunisertib had a significantly decreased overall tumor burden and metastatic tumor burden. There was limited effect to the primary tumor site. Combination therapy represents a significant area of interest that needs further exploration due to potentially different effects between the primary site of disease and metastases. Introduction: Accurate staging of patients with pancreatic ductal carcinoma (PDAC) is imperative for appropriate treatment. One third of patients may have peritoneal dissemination (PD) at diagnosis that is undetectable with standard imaging. Diagnostic laparoscopy is utilized by some to identify grossly visible disease. Addition of peritoneal lavage (PL) with cytology to laparoscopy may increase detection of PD but is limited by cellular yield and low sensitivity. Since 1/2018, we have used a ddPCR cell-free DNA assay to detect mutant KRAS (mKRAS) in peripheral blood to identify possible hematogenous metastases, as >90% of PDAC tumors harbor mKRAS. Our aim was to determine the utility of this assay to detect mKRAS in PL fluid. Methods: All radiologically-localized PDAC patients undergoing staging laparoscopy with PL by a single-surgeon under an IRB-approved prospective trial were included. Patients underwent PL with instillation of 1000ml of NS which was agitated, aspirated, and sent for cytologic examination. PL fluid was then sent for ddPCR mKRAS cfDNA assay. mKRAS was considered positive if any mKRAS was detected. mKRAS status was compared to gross findings and cytologic results. Results: Eighteen patients undergoing staging laparoscopy and PL as part of a pilot feasibility cohort were included. At laparoscopy, 3/18 (17%) patients had gross disease, 2/18 (11%) had positive cytology, and 8/18 (44%) had mKRAS in PL fluid with a mean 99.3 mKRAS copies/20 l ml. Four (22%) patients were clinically positive (grossly and/or cytologically). mKRAS was detected in 75% of clinically positive cases. Of 3 grossly positive patients, cytology was positive in only 1 (33%). Five additional patients were clinically negative but had detectable mKRAS with a mean 2.4 mKRAS copies/20 l. Conclusion: To our knowledge, testing PL fluid for mKRAS has never been performed. Preliminary findings show a significant number of patients with detectable mKRAS in PL fluid. With these results we have expanded our trial to assess sensitivity, specify, and ability of this assay to detect PD in patients with clinically non-metastatic PDAC. Background: Pancreatic ductal adenocarcinoma (PDAC) is associated with limited response to systemic therapy (ST). Elevated tumor interstitial fluid pressures (IFP) inhibit penetration of ST. Regional Pressure Enabled Drug Delivery has recently demonstrated improved response for liver tumors in a clinical trial. However, this delivery method has not been evaluated in PDAC. We compared gemcitabine (Gem) by systemic delivery vs. a novel pressurized Pancreatic Retrograde Venous Infusion (PRVI) method in an orthotopic PDAC mouse model. Methods: PDAC murine cell line (KPC4580P) tumors were transplanted onto the pancreatic tail of C57BL/6J mice. Groups of 15 mice were randomly assigned to PRVI Gem, PRVI saline (Control), or intraperitoneal Gem (Systemic) groups. Tumor IFPs were measured throughout RPVI. Five mice from the PRVI and Systemic groups were randomly selected after one hour post infusion to evaluate Gem tumor concentrations by liquid chromatography -tandem mass spectrometry (ng/mg), and the remainder of mice were euthanized after 7 days to evaluate treatment response. Results: PRVI resulted in peak mean increase of 27mmHg in the IFP. Tumor concentrations of Gem were significantly higher following PRVI compared to Systemic (128 vs. 19, p < 0.01) at one hour after treatment ( Figure) . Seven days after treatment, PRVI Gem mice demonstrated lower mean tumor volume (mm 3 ) than Systemic Gem and Control mice (274 vs. 857 vs. 629, p <0.01), respectively. Histologic evaluation of tumors demonstrated decreased cellularity in the PRVI Gem mice compared to Systemic and Control mice (35 vs. 78 vs. 71%, p = 0.01), respectively. No differences were seen in Ki67% or immune cell infiltrate between groups. Conclusions: PRVI delivery resulted in increased PDAC Gem concentrations and improved treatment responses with decreased tumor burden and cellularity. These findings suggest that pressurized regional chemotherapy infusion overcomes the elevated PDAC IFP and justifies additional translational pre-clinical studies with other chemotherapeutics (including immunomodulating antibodies) with different physicochemical properties. Background: Pancreatic adenocarcinoma (PDAC) with a conspicuous change in enhancement at the interface between tumor and parenchyma (highdelta) have more aggressive biology and exhibit worse clinical outcomes compared to inconspicuous (low-delta) tumors. Here, we hypothesized these imaging-defined biophysical subtypes of PDAC would exhibit spatial histologic differences in immune and stromal characteristics. Materials and Methods: Baseline CT scans of 44 treatment-naïve patients with PDAC who underwent pancreatectomy were classified into high and low delta. Corresponding slides stained for -SMA, VEGF, and H&E were digitally scanned (20x) and analyzed using Definiens-TS to quantify expression and micro-vessel density. For H&E, tissue was segmented into tumor and stroma regions and the cells were phenotyped into lymphocytes, fibroblasts and tumor-cells. Spatial analysis (K-function, R software) characterized distributions of tumor-cells and lymphocytes. Kaplan-Meier, cox-proportional-hazards, logistic regression and t-test were used for statistical analysis. Results: Patients were 50% male, with a mean age of 66.3 years. When compared to high-delta tumors, low-delta tumors were significantly associated with better overall survival (33.5 vs. 19 months, p=0.007) and recurrence free survival (25 vs. 7 months, p=0.003). On multivariate-analysis, delta was an independent prognostic factor for survival (HR: 2.3, p=0.01). Pathology analysis revealed significant associations between low-delta tumors and low micro-vessel density (p=0.002) and low stromal -SMA expression (p<0.0001), when compared to high-delta tumors. Spatial analyses showed that clustering of tumor-cells and lymphocytes at closer distances (<110 , p=0.03) was significantly associated with low-delta tumors. Conclusion: Biophysical imaging subtypes of PDAC are associated with aggressive biology and clinical outcomes. The pathological differences of the immune and stromal properties of these subtypes may enable the use of the imaging-based classification to guide therapeutic decisions for pancreatic adenocarcinoma. Introduction: We have reported that conjugated bile acids (CBAs)activated Sphingosine 1-Phosphate (S1P) receptor 2 (S1PR2) is a key regulator of lipid metabolism in hepatocytes (Hepatology 2015), and that the same signaling pathway promotes cholangiocarcinoma growth by cholestasis (Hepatology 2014), using our newly established murine long-term obstructive jaundice (Bile duct ligation; BDL) model (JSR 2016). We hypothesized that CBAs via obstructive jaundice promote metastatic pancreatic cancer progression through S1PR2 activation. Methods: Pancreatic adenocarcinoma cell lines, Panc02luc (murine) and AsPC-1 (human), Taurocholate (TCA), CYM5520 (S1PR2 agonist), and JTE-013 (S1PR2 antagonist) were used. Panc02-luc cells were implanted in the left lobe of the liver or injected intraperitoneally. Results: TCA and CYM5520 promoted significant cell growth in dose dependent manner on Panc02-luc and AsPC-1 cell lines (p<0.001), both of which significantly expressed S1PR2, whereas JTE-013 inhibited cell growth despite additional TCA stimulation (p<0.001). The same was observed for cell migration by the scratch assay (p<0.05). BDL resulted in significantly increased tumor burden in liver metastasis as well as peritoneal carcinomatosis. BDL group showed increased number of peritoneal nodules, more ascites, increased Ki-67 expression, as well as significantly shorter survival (all p<0.05) compared to the sham group in the peritoneal carcinomatosis model. Lastly, treatment with CYM5520 significantly increased carcinomatosis compared to the sham model (p<0.05). Conclusions: CBAs with obstructive jaundice model promoted metastatic pancreatic cancer progression in the cells as well as in the animal models. Introduction: Pancreatic adenocarcinoma (PDAC) is a highly aggressive cancer that utilizes multiple mechanisms to evade the immune system. CD155 is a transmembrane receptor that is known to be overexpressed in PDAC. The interaction of CD155 with CD226, a natural killer (NK) and T cell activating receptor, enhances anti-tumor response. The interaction of CD155 with the T cell receptor TIGIT (T cell immunoglobulin and ITIM domain) dampens the anti-tumor response. Our objective was to evaluate if intra-tumoral expression of CD155, TIGIT and CD226 was associated with outcome in patients with PDAC. Methods: A tissue microarray (TMA) with short-term (cancer-related death within 12-months of resection, n=63) and long-term survivors (>30 months, n=82) who underwent resection of PDAC was utilized. Immunohistochemical (IHC) staining for CD155, CD3 (T cells), CD226 and TIGIT was performed. CD155 staining was scored manually by a pathologist. Other markers were scored by pixel quantification using ScanScope software. The relative expression of each marker was estimated (stongly positive pixels/total pixels) and compared between STS and LTS using Wilcoxon test. Results: 145 patients were evaluated. STS and LTS were similar with respect to median age (STS, 69 vs LTS, 66 yrs, p=.07) and T-stage (T3: STS, 98.4% vs LTS, 93.9%, p=.54). STS were more often node positive (84.1% vs 64.6%, p=.02). CD155 was expressed in 99.2% of tumors, and was not associated with survival (LTS, 60% vs. STS, 61.6%, p=.88). LTS had higher immune cell infiltration (median CD3: 0.56% vs 0.17%, p<.01) and CD226 expression (0.75% vs 0.24%; p<.01). These associations were not dependent on CD155 expression (r<0.2). TIGIT expression was also not associated with survival outcome (LTS, 0.41% vs STS, 0.43%, p=0.90). Patients with lymph node positivity had lower CD3 infiltration (mean 1.38% vs 0.61%, p=.04). Conclusion: In this study, CD155 was broadly expressed in most PDAC tumors but was not associated with outcome. CD226 expression was associated with improved survival, however, this finding was independent of CD155 expression. Novel therapies enhancing anti-tumor CD226 activity through interaction with CD155 should be explored. Introduction: Pancreatic ductal adenocarcinoma (PDAC) is predicted to be the second leading cause of cancer death in the US by 2030. Previously, we developed an ex-vivo, live tissue sensitivity assay (LTSA) that can quantify the response of patient derived xenograft (PDX) tumors to standard of care (SOC) regimens (Roife et al., Clin Cancer Res, 2016) . In this study, we modified the LTSA for use as a high-throughput drug screening platform to test novel agents alone or in combination with SOC agents in PDAC PDX models. Methods: Nine drugs not currently in clinical use against PDAC were screened as panel of 45 treatment combinations. 8 low-passage PDAC PDX were harvested and 2mm x 200mm-thick tissue slices were prepared and deployed in a 96-well plate. After the addition of the indicator dye, resazurin, drugs were added and incubated for 72 hours. Experiments were performed at three doses and in triplicate. Sensitivity to treatment was defined by a 30% reduction in viability compared to controls. The combination index (CI) method was used to identify responder drug treatments that were then validated in a panel of 10 PDX tumors. Results: A wide range of responses were observed across 8 PDX models when treated with single and paired drug combinations. We observed no dominant drug or drug combination that was effective against all 8 PDX models and the response profile of each PDX tumor appeared unique. However, four drug pairs were observed to be effective and synergistic at initial and subsequent screens: sunitinib+trametinib, palbociclib+trametinib, panobi-nostat+trametinib, and BEZ235+panobinostat. Of these, sunitinib+trametinib was the most globally effective, displaying synergy in 8/10 PDX tumors on the second screen. Conclusion: Using the LTSA as a drug screening platform resulted in the identification of new compounds and combinations that may have efficacy in PDAC. High-throughput drug treatment studies in wellcharacterized, PDAC PDX models are possible in a short timeframe. The variability of tumor responses to each treatment underscores the heterogeneity of this disease and the need for the development of personalized therapies. 

Background: Pancreatic ductal adenocarcinoma (PDAC) is a nearly universally lethal disease with an 8% five-year survival. Efforts to extend the benefits of checkpoint immunotherapy to PDAC have been unsuccessful, suggesting that an alternative approach is needed in pancreatic cancer. While initially thought to be an immune suppressive cytokine, high levels of IL10 have recently been shown to augment T cell effector function in other cancer types. Here-in we utilize a modified form of IL10, IL10Fc, to stimulate effector function in tumor infiltrating T cells to promote an anti-tumor immunity in pancreatic cancer. Methods: Murine T cells were cultured in increasing concentrations of IL10Fc and analyzed via flow cytometry and RT-PCR for production of effector cytokines including Gzmb, IFNg, TNFa. In a genetically engineered p48-CRE/ LSL-Kras G12D /p53 flox mouse model of PDAC, IL10Fc was administered and mice were analyzed for tumor burden. KI cancer cells were injected into the pancreata of FvBn mice and treated with IL10Fc. Mice were analyzed at 4 and 8 days for immune changes by flow cytometry. Mice were analyzed at 28 days for tumor burden and separately for difference in overall survival. Results: IL10Fc directly increases effector function in CD8 + T cells in vitro including the production of GzmB, IFNg and TNFa. These effects were also observed in vivo, where a single dose of IL10Fc produced significant increases in CD8 + T cell production of GzmB, IFNg and TNFa. Additionally, we observed that IL10Fc depleted total numbers of FoxP3+ Tregs and Granulocytic MDSCs. As monotherapy it significantly slowed tumor progression compared to vehicle at 28 days. It also increased median overall survival compared to vehicle. Conclusion: We demonstrate that IL10Fc can increase effector cytokine production in T cells in murine PDAC and that these effects lead to decreased tumor burden and improved overall survival as monotherapy. Additional studies are ongoing to investigate further changes to the tumor microenvironment and best combination regimen to maximize efficacy. IL10Fc is a novel T cell immunotherapy that shows promise in the setting of pancreatic cancer. Introduction Increased stromal tumor infiltrating lymphocytes have been shown to predict response to neoadjuvant chemotherapy (NAC) as well as survival outcomes in breast cancer patients. Eosinophils are reported to secrete chemokines to support T-cell homing to tumor. We investigated the impact of tumor-associated tissue eosinophilia (TATE) as it relates to neoadjuvant chemotherapy response. Methods The Cancer Genome Atlas Data Set was analyzed for the eosinophil signatures in breast cancer specimens. Descriptive analyses, including the tumor infiltrating cell composition using CIBERSORT, score and gene set enrichment analysis (GSEA), as well as baseline clinical characteristics, were performed. Several GEO cohorts were then subsequently analyzed for immune cell composition in relation to complete pathologic response (pCR). Results Out of 1069 cases analyzed, 40 (3.7%) were found to have tissue eosinophils (TATE group). TATE was noted in 3.6% luminal, 5.1% HER2+, 5.3% TNBC. Patients in the TATE group were predominantly Caucasian 72.5%, 27.5% stage I, 52.5% stage II, 20% stage III. Specimens with TATE had a statistically significant higher number of monocytes, p<0.001, T follicular helper cells, p=0.005. There was no difference in complete pathologic response in the cohorts analyzed between specimens with TATE vs not, regardless of the regimen used. After neoadjuvant taxane-anthracycline in GSE25066 pCR was (17 (20%) vs 82 (20.4%); p=1), after paclitaxel and fluorouracil, doxorubicin, and cyclophosphamide (T/FAC) in GSE20271 (0 vs 26 (15%); p=1), after neoadjuvant T/FAC in GSE20194 (7 (15.5%) vs 43 (21.2%); p=0.5), after T/FAC in GSE22093 (26 (32.5%) vs 2 (11.8%); p=0.14), after doxorubicin/paclitaxel (AT) followed by cyclophosphamide/ methotrexate/fluorouracil (CMF) in GSE50948 (4 (33.3%) vs 49 (34%); p=1), after Docetaxel-Capecitabine in GSE22358 (10 (23.3%) vs 20 (25.3%); p=1). Conclusions Eosinophil expression was noted in 3.7% of breast cancers, more frequently in HER2+ and TNBC. There was no correlation between eosinophil presence and complete pathologic response in several cohorts analyzed after several neoadjuvant chemotherapy regimens.

Background: Pembrolizumab (pembro) + chemotherapy (chemo) as neoadjuvant treatment had manageable safety and promising antitumor activity in patients (pts) with locally advanced triple-negative breast cancer (TNBC) in the phase 1b KEYNOTE-173 and the phase 2 I-SPY 2 studies. The phase 3 KEYNOTE-522 study (NCT03036488) of pembro+chemo vs placebo (pbo)+chemo as neoadj therapy, followed by pembro vs pbo as adjuvant (adj) treatment in pts with early TNBC showed that neoadj pembro+chemo significantly improved the pathologic complete response (pCR) rate. Methods: From March 2017 to September 2018, 1174 pts from 21 countries were enrolled. Key eligibility criteria included age 18 years; previously untreated, early, non-metastatic, centrally confirmed TNBC (stage T1c N1-2 or T2-4 N0-2 per AJCC by investigator); and ECOG PS 0-1. Pts with bilateral or multifocal primary tumors and inflammatory breast cancers were allowed. Pts were randomized 2:1 to the experimental and control arms to receive 4 cycles of [pembro (200mg Q3W) or pbo (normal saline)] + paclitaxel (80mg/m2 QW) + carboplatin (AUC 5 Q3W or AUC 1.5 QW) in the first 12 wks, followed by 4 cycles of pembro/pbo + [doxorubicin (60mg/m2 Q3W) or epirubicin (90mg/m2 Q3W)] + cyclophosphamide (600mg/m2 Q3W) in the subsequent 12 wks prior to surgery (neoadj part). After definitive surgery, pts received radiation therapy as indicated and adj pembro or pbo every 21 days for 9 cycles (adj part) or until recurrence/unacceptable toxicity (whichever came first). All enrolled patients were stratified by tumor nodal status (pos vs neg), size (T1/ T2 vs T3/T4), and schedule of carboplatin administration (Q3W vs QW). The dual primary endpoints are pCR rate, defined as ypT0/Tis ypN0, and EFS. Key secondary endpoints are safety, OS, and pCR rates, defined as ypT0 ypN0 and ypT0/Tis. Safety was evaluated per NCI CTCAE v4.0. Results: pCR rates in key patient subgroups will be reported. Introduction: IT is increasingly used in combination with CT in the neoadjuvant treatment (Trx) of TNBC and HER2 with the goal of increasing rates of pCR. In this study we assessed the effectiveness of MRI in the evaluation of tumor response after the use of several neoadjuvant IT approaches in combination with CT (NAICT). Methods: We retrospectively reviewed the clinicopathological data of 76 women treated at a single institution who had undergone definitive surgery after IRB approved NAICT protocols. 27 pts were excluded due to a lack of pre or post therapy MRI and 5 were excluded that were ER+HER2-. We analyzed 44 TNBC and HER2 NAICT pts from 4 distinct Trx protocols including: (1) 14 TNBC pts treated with intratumoral Talimogene laherparepvec (TVEC) in combination with weekly paclitaxel (T) followed by dose dense Adriamycin and Cytoxan (ddAC); (2) 13 ER+HER2+ pts treated with subcutaneous interferon gamma (IFN-) in combination with weekly T with trastuzumab and pertuzumab (HP); (3), 6 ER-HER2+ pts treated with 3 weeks of HER2 pulsed dendritic cell vaccines (DC1) followed by Taxotere, Carboplatin, and HP; and (4), 11 pts on ISPY2 trial; 8 pts randomized to Trx with pembrolizumab with weekly T followed by ddAC, 1 pt treated with SGN-LIV1A followed by ddAC, and 2 pts treated with durvalumab, and T, followed by ddAC. Results: The median age of the pts was 47 (range 27-70), the median clinical tumor size was 3.6 cm (0.7-11). Overall pCR was 27/44 (61%). Within the NAICT groups, TVEC pCR was 10/14 (71 %), IFNwas 13/13 (100 %,) DC1 was 2/6 (33.3%), pembrolizumab was 1/8 (12.5%). pCR rates for TNBC and HER2+ breast cancers were 12/25 (48%) and 15/19 (79%) respectively. The sensitivity and specificity of MRI in predicting residual disease was 59%, while the PPV and NPV were 45.5 and 70%. Conclusion: The addition of IT with standard CT regimens in TNBC and HER2 BC has improved pCR but reduced the predictive capacity of MRI suggesting MRI may not be accurate enough to make further surgical decisions. Recent trials have demonstrated the efficacy of checkpoint inhibitors in metastatic triple negative breast cancer, which may signal a more significant role for immunotherapy in breast cancer (BC) in general. We demonstrated previously the safety of SV-BR-1-GM, a whole-cell immunotherapy composed of irradiated metastatic BC cells transfected with a GM-CSF plasmid. Here, we present results of a phase I/IIa trial evaluating SV-BR-1-GM/pembrolizumab combination therapy in the treatment of metastatic/recurrent BC. Female patients 18yo with metastatic/recurrent BC having failed 1 line of therapy were enrolled. Treatment cycles consisted of cyclophosphamide ( 300mg/m 2 ) 48-72h prior to inoculation, SV-BR-1-GM inoculation (10-40x10 6 cells/inoculation), followed by interferon-alpha-2b on post-inoculation day (PID) 2 and 4, and pembrolizumab on either PID 2 or 4. Cycles were repeated every 3 weeks. Safety was assessed. Immune response was determined by development of delayed type hypersensitivity (DTH) reactions to SV-BR-1 (non-transfected) or SV-BR-1-GM. Data are reported as median (interquartile range [IQR] ) and compared with Kruskall-Wallis tests. Eleven patients have enrolled thus far; 9 have completed 1 cycle (4, IQR 2-6). Patients have a median age of 62 years and have failed a median of 10 (IQR 5-15) prior systemic regimens. Initial cancers were ER/PR+ in 6 (55%) and HER2+ in 7 (64%) patients. Six patients (55%) experienced 35 total adverse events (AEs) including four Grade 3 events: two occurred following cyclophosphamide and two were unrelated to the study. There were no toxicities > Grade 2 that were related to combination therapy, and no withdrawals due to AEs. Of 8 patients with follow-up data, all developed a delayed type hypersensitivity (DTH) reaction; 2 have not progressed on therapy; median time to progression was 71 days (IQR 67-106). Combination therapy with SV-BR-1-GM and pembrolizumab is safe and consistently produces anti-cancer immunity as measured by DTH reactions. Further study is needed to determine the efficacy of combination therapy in the treatment of aggressive BC.

Combination Therapy Using Adoptively Transferred T-Cells from Tumor-draining Lymph Nodes and Checkpoint Blockade, Chemotherapeutic, and Immune-modulating Agents Improve Therapeutic Efficacy in a Syngeneic Murine Model J.L. Lyons, 3 * M. Zhang, 1 Z. Senders, 3 J. Kim Background While adoptive immunotherapy (AIT) has shown promising results in the treatment of liquid and solid tumors, its efficacy in solid tumors is limited largely due to the immunosuppressive tumor microenvironment. Immune modulating agents have also shown efficacy in treating solid tumors by potentiating the body's own immune response. While our lab has shown efficacy using immune modulating agents in combination with AIT in a xenograft murine model, this study was designed to investigate the efficacy of combination therapy in a syngeneic model with high clinical relevance. Methods Tumor draining lymph nodes (TDLNs) were harvested from mice inoculated with 4T1, a highly tumorigenic and metastatic murine breast cancer cell line which highly mimics human breast cancer. T-cells from the TDLNs were preferentially activated, grown ex-vivo, and tested for specific anti-4T1 activity. Mice with established 4T1 tumors were then treated with the resulting t-cells both alone and in combination with a murine PD-1 inhibitor, gemcitabine, and the novel immune modulator BG34. The primary endpoint was overall survival. Results The adoptively transferred t-cells showed higher INFrelease (151.7 pg/ml) when exposed to 4T1 cells than the t-cells alone (16.6 pg/ml) or 4T1 cells alone (0.99 pg/ml). Mice treated with t-cells alone, BG34 alone, and gemcitabine alone all showed improved survival over control mice (p<0.01, p=0.02, p<0.01 respectively). Treatment with PD-1 inhibitor alone did not show a statistically significant improvement in survival (p=0.18). Mice treated with the combination of t-cells and BG34, t-cells and gemcitabine, and t-cells and PD-1 inhibitor all showed improved survival over the corresponding monotherapy (p=0.04, p<0.01, p<0.01 respectively). The combination of gemcitabine and t-cells produced the most durable improvement in survival. Conclusion Using adoptively transferred t-cells from TDLNs in combination with immune modulators show improved survival than either monotherapy alone in a syngeneic murine model. Background Since William Halsted described the radical mastectomy technique in the 19th century, breast cancer surgery has evolved towards less invasive techniques. Last decades have been of major advances in breast conservative surgery (BCS) and immediate reconstruction (BR) after multiple studies demonstrated its oncologic safety. The aim of this study is to characterize the population of patients undergoing breast cancer surgery at our Institution and to examine time trends in surgical management. Methods Previous IRB approval, we analyzed a retrospective cohort of breast cancer patients who underwent surgery as primary treatment at National Cancer Institute of Mexico from 2012 to 2018. Sociodemographic, tumoral and surgical variables were analyzed. Frequencies and percentages were graphed to obtain time trends in surgical variables. Results A total of 4854 patients were analyzed. We found that reconstructed patients had earlier clinical stages (54.9% vs 31.9%; p<0.001), less incidence of Diabetes Mellitus (3.5% vs 9.3%; p<0.001) and systemic hypertension (4.5% vs 9.9%; p<0.001). Reconstructed patients usually have no history of chemotherapy or radiotherapy (p<0.001) and the most common reconstructive method was alloplastic (66.5%). BCS increased substantially between 2014 and 2016, and after 2017 it has remained constant. There is an in crescendo pattern seen in immediate BR, with a 162.2% increase in 7 years. Autologous BR increased substantially from 2012 to 2014 and have decreased progressively thereafter; while alloplastic BR have increased since 2014 with a sharp increment in last years. The creation of the Oncologic Reconstruction and Microsurgery training program in 2018 explains the sharp increase in BR in this year. We found that immediate BR overall survival rate was 95.8% at 10 years. Conclusions We have noticed an important growth in BCS and immediate breast reconstruction, specially in earlier clinical stages and healthy women. There is no negative impact of breast reconstruction in the overall survival rate, so we consider it is a safe procedure with great benefits in terms of quality of life and emotional status for these patients. Background: Synchronous metastatic melanoma, clinically defined as multiple lesions diagnosed within 6 months, carries a poor prognosis. Despite recent advances in immune checkpoint inhibitors, only a minority of stage IV melanoma patients respond. Inherent tumor heterogeneity secondary to genetic instability between metastases results in distinct tumor microenvironments (TMEs) and unpredictable clinical lesion-specific immunotherapeutic responses (where some lesions respond and others do not). While intertumoral heterogeneity has been clinically correlated with differential responses to treatment, no mechanism has been identified due to the lack of a good preclinical animal model. Methods: We have generated a novel murine synchronous metastatic melanoma model using parental YUMM 1.7 and its UVB-irradiated derivative YUMMER 1.7 cell lines. The two lines contain the same Braf V600E /Pten -/-/Cdk2n -/driver mutations but only share approximately 40% of somatic mutations. We characterized individual lesion responses to PD-1 (programmed cell death protein 1) inhibitor and tumor immune microenvironmental changes via flow cytometry. Results: Over time, YUMM 1.7 and YUMMER 1.7 generated distinct immune TMEs when synchronously implanted onto opposite flanks of the same mouse. The determination of TME immunomodulatory capabilities (pro-T cell exhaustion or anti-T cell exhaustion) occurred at the point of tumor escape. Systemic PD-1 inhibitor treatment of the synchronous model led to lesion-specific responses as well as differential changes in intratumoral CD8+ T cell activation status and overall immune infiltration. Conclusions: A novel preclinical model of synchronous metastatic melanoma model demonstrates that intertumoral genetic heterogeneity generate distinct, localized TMEs. Consequentially, local TMEs determine lesion-specific responses to systemic PD-1 blockade. Further genetic and immune characterization of this model could provide mechanistic insight underlying heterogenous tumor responses to immunotherapy in synchronous metastatic melanoma patients and provide potential therapeutic targets to improve treatment efficacy.

Histotripsy Tumor Ablation -An Opportunity to Induce Abscopal Immunity to Cancer A.E. Felsted,* A.L. Pepple, T. Worlikar, R. Hubbard, A.A. Kevelin, B. Bredbeck, A. Ganguly, Z. Xu, C.S. Cho. Surgery, University of Michigan, Plymouth, MI. Background Despite its profound benefits, most patients remain unresponsive to checkpoint inhibition immunotherapy. Efforts to use tumor-directed therapies to stimulate systemic anti-tumor immune responses have had limited success. Histotripsy is a novel non-invasive technology that uses focused ultrasound to induce acoustic cavitation in tumors. This mechanical cellular disruption enables histotripsy to stimulate potent local tumor-specific adaptive immune responses. We sought to characterize the ability of histotripsy to induce abscopal immune responses. Methods Immunocompetent mice bearing bilateral flank melanoma tumors were treated with unilateral sham or histotripsy tumor ablation. At various timepoints, treated and untreated tumors were analyzed by flow cytometry, immunohistochemistry, and immunofluorescence to characterize immune responses. Results Compared to sham ablation, histotripsy-treated tumors demonstrated an acute release of the damage-associated molecular pattern high mobility group box protein 1, and a strong inflammatory response with increased levels of CRP, IL-2, CXCL10, and CCL6. Leukocytic infiltration was dominated by neutrophils and macrophages and densely concentrated at the periphery of ablation zones. By 7 days, the infiltrate evolved to be CD8 + T cell dominant, with increased levels of CD4 + T cells and B cells also. In contrast, contralateral untreated (abscopal) tumors exhibited early, durable, and homogenous NK cell infiltration and progressively increased CD8 + and CD4 + T cell and B cell infiltration. Compared to sham ablation, unilateral histotripsy ablation resulted in a significant inhibition of contralateral tumor growth, attesting to the induction of a meaningful abscopal anti-tumor immune response. Conclusions Histotripsy ablation produces a rapid, massive local inflammatory response that quickly induces local adaptive immune responses. This local immune response then systemically reorients the immune system, enabling it to target additional sites of disease in a tumor-specific manner. Histotripsy's unique ability to induce systemic anti-tumor immune responses may be a means to increase the numbers of patients who benefit from cancer immunotherapy. Vaccination with a mixture of 6 melanoma helper peptides (6MHP) induced both CD4 + T cell and IgG antibody (Ab) responses to 6MHP. These Ab responses are associated with improved survival in patients with advanced melanoma. We hypothesized that 6MHP-induced Abs would bind multiple epitopes on the same peptide, forming large antigen-Ab immune complexes (IC) and activating complement. We also hypothesized that vaccination would predominantly induce IgG subtypes that support antigen opsonization (IgG3 and IgG1). Patients were vaccinated with 6MHP and incomplete Freund's adjuvant (IFA, n=10), IFA+ polyICLC (n=5), IFA+metronomic cyclophosphamide (mCy, n=5), or IFA+polyICLC+mCy (n=10). Abs reactive to fragments of 6MHP peptides were detected with ELISA. Serum IgG1, IgG2, IgG3 and IgG4 Ab titers to 6MHP were also determined by ELISA at time of peak IgG total response. A positive Ab response was defined as a reciprocal titer above 100. Vaccine-site biopsies were evaluated by direct immunofluorescence for IgG and complement C1q. IgG Abs to 6MHP were identified in 27/30 patients (90%), with 73% and 37% of patients having positive IgG3 and IgG1 titers, respectively. IgG2 and IgG4 responses were not detected. Mean titers for IgG total , IgG3 and IgG1 were 4960, 1050, and 520, respectively. Adding polyICLC to IFA significantly increased IgG total (p=0.004) and IgG3 (p<0.001) titers ( Figure 1 ). Adding mCy did not enhance IgG responses. Serum Ab responses were detected to multiple fragments of the same 6MHP peptide, and both IgG and C1q were detected at vaccination sites. These data provide evidence that vaccination with 6MHP can induce memory B cells that undergo IgG isotype switching in melanoma patients. The IgG subtypes produced were IgG3 and IgG1, which can both facilitate antigen opsonization. These data also suggest that 6MHP peptides contain multiple epitopes that may allow Abs to form large ICs, triggering complement activation and IgG deposition in vaccination sites. Finally, these data show that the TLR3 agonist polyICLC enhances induction of IgG and IgG3 Ab responses. Future studies will investigate whether peptide-antigen presentation is enhanced by these class-switched Ab responses. 7 1. Surgery, Massachusetts General Hospital, Boston, MA; 2. University of Oxford, Oxford, United Kingdom; 3. University of Liverpool, Liverpool, United Kingdom; 4. University of Miami, Miami, FL; 5. Providence Cancer Institute, Portland, OR; 6. West Cancer Center, Germantown, TN; 7. The Institute of Cancer Research, London, United Kingdom; 8. Replimune, Inc., Woburn, MA. RP1 is a novel oncolytic HSV-1 expressing a fusogenic glycoprotein (GALV-GP R-), designed to enhance immunogenic cell death, and GM-CSF, which is designed to promote anti-tumor immunity. The safety, dosing and feasibility of combining RP1 with nivolumab in cancer patients is unknown. A phase 1 clinical trial (NCT03767348) was conducted with the objectives of defining RP1 safety alone and with nivolumab, determining the recommended phase 2 dose, and assessing biologic activity, biomarkers, and initial efficacy. Eligible patients with advanced solid tumors were treated by intra-patient dose escalation of RP1 (up to 10mL of 10 4 -10 8 PFU/mL) by intratumoral injection into a single tumor Q2W up to 5 times followed by 12 patients dosed up to 8 times at the RP2D combined with nivolumab (240mg Q2W for 4 months from the second RP1 dose, then 480 mg Q4W for 20 months). Image guidance was used for deep/visceral lesions. Pre-and on-treatment tumor biopsies were obtained for biomarker analysis, viral shedding and anti-HSV antibody titers. A total of 36 patients with previously treated metastatic tumors were enrolled. Overall, treatment was well tolerated with low-grade constitutional and local injection site adverse events, including fever, nausea, fatigue, local inflammation and erythema being the most common side effects reported. A similar profile was seen with nivolumab and no immune-related adverse events were seen. No clear differences were seen between superficial and visceral dosing. RP1 was detected at the injection site and in blood for up to 14 days (next injection), suggesting virus replication. All HSV seronegative patients seroconverted after three injections. The RP2D was selected as up to 10mL of 10 6 PFU/mL followed Q2W by multiple doses of 10 7 PFU/ mL. Injected tumors demonstrated infiltration of CD8+ T cells and increased PD-L1 expression. Three patients experienced an objective response. RP1 is safe and well tolerated alone and combined with nivolumab. Biomarker data supports induction of anti-tumor immunity and further evaluation of RP1 and nivolumab is underway in a phase 2 trial for melanoma, non-melanoma skin cancer, bladder cancer and MSI-H solid tumors. Background: Heat shock proteins (hsp) are intracellular chaperones but also have extracellular immunostimulatory properties when complexed with antigens. Preclinical data with a recombinant hsp110-gp100 chaperone complex vaccine showed antitumor response and prolonged survival in mice bearing B16 melanoma tumors. A Phase I dose escalation study of a recombinant human hsp110-gp100 vaccine in advanced stage melanoma patients was performed to evaluate toxicity, immunostimulatory potential, and clinical response. Methods: Patients with pretreated, unresectable Stage IIIB/C/ IV melanoma received the chaperone complex vaccine in a dose escalation protocol; 3 vaccinations over a 43-day period. Tumor response (RECIST criteria) and clinical toxicity were measured as was immune cell reactivity from peripheral blood samples. Results: Ten patients (8 female, median age 70) were enrolled in the protocol. Doses ranged from 30 to180 mcg. Two patients had grade 1 adverse events (AE); minor skin rash and fever. There was no grade 3 or higher AE. Median progression free survival was longer in patients receiving lower vaccine doses as compared to the maximum dose of 180 mcg (4.5 vs 2.9 months, p=0.018). The two patients at the lowest doses (30 and 60 mcg) had clinical tumor responses (1 partial response, 1 stable disease). ELISPOT of IFN-production in these patient samples demonstrated T cell reactivity to gp100 antigen (Fig 1) . In samples drawn at later time points, increasing regulatory T cell populations were noted as was increased PD-1 expression on CD 8+ cells, a finding which correlated with increasing tumor growth clinically. Across the entire cohort, there was no significant change in expression of CTLA-4, Lag 3, or Tim-3 over time. Production of gp-100 specific IgG antibody by B-cells was not noted. Conclusions: A fully recombinant human hsp110-gp100 chaperone complex vaccine had minimal toxicity, measurable tumor responses at lower doses, and produced peripheral CD8+ T-cell activation in patients with advanced, pretreated melanoma. Future directions include combination with currently available immunotherapies and multiantigen vaccine development. Surgery, Brooke Army Medical Center, Fort Sam Houston, TX; 2. John Wayne Cancer Institute, Santa Monica, CA; 3. University of Utah, Salt Lake City, UT; 4. Rutgers Cancer Institute of New Jersey, New Brunswick, NJ; 5. Mayo Clinic, Rochester, MN; 6. University of Cincinnati, Cincinnati, OH; 7. University of Arizona, Tucson, AZ; 8. Orbis Health Solutions, Greenville, SC; 9. Cancer Vaccine Development Program, San Antonio, TX. The TLPLDC vaccine uses yeast cell wall particles (YCWP) to load tumor lysate into autologous dendritic cells (DC). In a phase IIb trial, TLPLDC was well tolerated and improved 24 month (mo) disease free survival (DFS) in the per treatment analysis. A novel vaccine strategy, known as tumor lysate, particle-only (TLPO), involves loading autologous tumor lysate into YCWP, then capping them with silicate. This strategy is simpler and more costeffective than the TLPLDC vaccine, but comparable safety and efficacy are unknown. Here, we present the results of an interim analysis of a phase II trial comparing the TLPO and TLPLDC vaccines. Stage III/IV patients who were clinically disease-free after standard surgical resection were randomized 2:1 to TLPO vs. TLPLDC. Both vaccines were administered monthly x3 followed by booster at 6, 12, and 18 mo. The primary endpoints are safety as assessed by CTCAEv4.03 and efficacy measured as 24 mo disease-free survival (DFS). Fifty-nine patients were randomized (39 TLPO, 20 TLPLDC). Groups were well-matched except there were more males in the TLPO arm (90%) compared to the TLPLDC (62%, p=.022). Overall, both vaccines were well-tolerated with over 90% of adverse events (AE) being grade 1-2. In the TLPO group, 33% of patients experienced at least one AE compared to 55% in the TLPLDC group. There was a similar rate of grade 3/4 AEs between groups (7.7% vs 10.0%, p=0.763, Table 1 ). No patients withdrew from the study due to adverse events. At the time of this interim analysis, median follow-up was 3.85 mo, and there was no significant difference in 24 mo DFS (p=0.359). The TLPO vaccine is as safe and well-tolerated as the TLPLDC vaccine. At this early interim analysis, it appears to have similar efficacy. Longer follow-up will demonstrate whether the simpler and more cost-effective antigen delivery via TLPO is as effective as external loading of dendritic cells as a vaccination strategy. ON, Canada; 3. University of Manitoba, Winnipeg, MB, Canada; 4. Health Outcomes and PharmacoEconomic Research Centre, Toronto, ON, Canada. Background: Only 10% of patients with pancreatic adenocarcinoma (PAC) undergo curative-intent resection (pancreatoduodenectomy or distal pancreatectomy), however many patients may have some type of non-therapeutic operation. This may delay initiation of systemic chemotherapy, lead to worse oncologic outcomes, and add significant burdens to patients and Conclusions: A significant number of patients undergo non-therapeutic operations for PAC, with poor outcomes, and a high cost to the system. Diagnostic laparoscopy was associated with better outcomes and lower healthcare cost than more invasive operations. Further research should investigate predictors for non-therapeutic operations, and how this burden to patients and the healthcare system can be reduced.

Overall survival of patients with pancreatic adenocarcinoma stratified by surgical procedure (PD: pancreatoduodenectomy, DP: distal pancreatectomy, DL: diagnostic laparoscopy, EL: exploratory laparotomy, BP: biliary/enteric bypass, OO: other operation). Introduction: Negative margins are an important prognostic factor for patients undergoing resection for pancreatic adenocarcinoma (PDAC). For successfully resected patients with close or positive margins, intraoperative radiation (IORT) can be used to improve local control and survival. While it is accepted that a microscopically positive margin (R1) has a negative impact on survival, it is unclear whether an R1 resection has the same prognostic impact in patients who have received total neoadjuvant therapy and IORT. Methods: Clinicopathologic data was retrospectively collected for all patients who received total neoadjuvant therapy (TNT; FOLFIRINOX followed by chemoradiation (25Gy or 50.4 Gy)) for PDAC and underwent resection between 2011 and 2018. Kaplan-Meier curves were used to plot the distribution of disease-free survival (DFS) and overall survival (OS). Results: A total of 197 patients received TNT and were resected with a median DFS of 24 months and OS of 50 months. Median age of the cohort was 57 and 49.2% were male. Eighty-one patients (41%) received IORT (10Gy) at the time of resection. Of the 81 patients, 64 (79%) underwent an R0 and 17 (21%) an R1 resection. There was no significant difference in any clinicopathologic factors between the IORT and no-IORT groups, except for surgeons' concern for a positive/ close margin. An R1 resection was associated with a worse DFS (R0: median 29m vs R1: median 18m; p = <0.001) and OS (R0: median 65m vs R1: median 30m; p = <0.001). However, among patients who received IORT there was no difference in DFS (R0: median 28m, IQR 15-43 vs R1: median 25m, IQR 16-29; p = 0.29) or OS (R0: median 48m, IQR 24-not reached vs R1: median 46m, IQR 30-not reached; p = 0.336) between patients who underwent an R0 vs an R1 resection. In multivariate analysis, within the IORT group, R1 resection was not associated with DFS or OS. Conclusions: IORT may mitigate the adverse effect of a microscopically positive margin (R1) on DFS and OS in PDAC patients. Figure. Kaplan-Meier overall survival curves. R0 vs. R1 resection margin groups among patients who received IORT. P-values derived from log-rank tests.

Introduction Positive microscopic (R1) resection margins for pancreatic cancer (PDA) are common (up to 80%). Although neoadjuvant therapy (NAT) is shown to increase margin negative resections, its impact on disease dependent survival after an R1 resection is not established. We aimed to study the impact of NAT on survival and recurrence in patients following R1 and R0 resections. Methods Retrospective, single center analysis of consecutive PDA patients who underwent pancreatectomy between 2008-2017. Patients were staged based on the AJCC 8 th edition and stratified as R0 or R1 (carcinoma 1mm from the margin) and as NAT or surgery first (SF) approach. Conditional survival analysis was performed using Kaplan Meier estimates and Coxregression. Results Of 580 patients analyzed (mean age 67.5, female=50%), 59% received NAT and 41% were SF. NAT patients were younger, had less co-morbidities and higher tumor size at diagnosis (all p<0.05). They also had smaller tumor size, less lymph node disease (LN+), LN ratio, lymphovascular and perineural invasion on resection (all p<0.05). R1 resection (50% of patients) was not associated with NAT (p=0.653 Introduction/Background: More biologically active chemotherapeutics have led to improved systemic control in patients with locally advanced pancreatic cancer (LAPC), ushering in more aggressive surgical strategies such as distal pancreatectomy with celiac axis resection (DP-CAR). We sought to determine outcomes, procedure specific complications, and patterns of recurrence in patients undergoing DP-CAR. Methods: A prospectively maintained single institution database of all pancreatic resections was queried for patients undergoing distal pancreatectomy with arterial resection/reconstruction and chart review was undertaken to identify patients undergoing DP-CAR and to abstract clinical variables. Patients from 2008-2018 were included. Results: Forty-eight patients that met inclusion criteria were identified with a median age of 62.3yrs. Most patients received neoadjuvant therapy (92%). Intraoperatively 21% required arterial reconstruction and 27% had concomitant visceral resection. Median length of stay was 7 days and median transaminanses were mildly elevated postoperatively (ALT 66, AST 86U/L). The R0 resection rate was 92%. 42% of patient experienced a complication, with chyle leak (15%) and gastric perforation (6%) being higher than expected with standard DP. Ninety day mortality was 2%. 75% of patients recurred with a median follow up of 19.9mo, with local recurrence being most common (29%), followed by lung only (15%) and intraabdominal (15%). Interestingly liver only recurrence was uncommon (8%). Median recurrence free survival was 10.7mo and overall survival was 27.7mo. Conclusions: Application of DP-CAR in the setting of modern neoadjuvant approaches appears to be safe, with sound oncologic outcomes, and may be associated with improved overall survival. These patients have disparate patterns of recurrence when compared with standard pancreatic resections and are at risk for unique post-procedure complications, namely chyle leak and gastric perforation. Further prospective study is required to document continued safety and oncologic benefit in this subgroup of patients with LAPC. Patients with confirmed BR-PDAC of the head of the pancreas, which was defined as having vessel involvement and without metastases, were included. Poisson regression models with robust standard errors were performed to evaluate the relative risk (RR) of undergoing a pancreatic resection (PD or extended PD) among non-white patients (Black, Asian, and non-white Hispanic) compared to white patients. The multivariable model was adjusted for potential confounders (age, gender, Charlson-Deyo score, tumor size, neoadjuvant therapy (NAT), median income, insurance status, urban/rural location, and facility type). A multivariate, Poisson regression model was then used to evaluate interaction between race and facility type. RESULTS: There were 15,482 patients (median age 68 years, IQR 60-76 years; 48.6% male) with BR-PDAC who were predominantly white (84.3%, n=13,058; non-white, 15.7%, n=2,424) . Overall, 18.4% (n=2,853) had a pancreatic resection. Non-white patients had significantly lower likelihood of undergoing a pancreatic resection for BR-PDAC when compared to white patients on univariable and multivariable analysis (RR, 0.75, 95% CI 0.68-0.83, P<0.001). In the interaction analysis, these findings persisted with the exception of non-white patients who received their care at an integrated network cancer program (RR, 0.99, P=0.97) . If the non-interaction analysis was confined to patients who received NAT, there were no significant differences in the likelihood of resection between non-white and white patients. CONCLUSIONS: Non-white patients were 25% less likely to undergo a pancreatic resection for BR-PDAC when compared to white patients. If patients were treated at an integrated network cancer program or with NAT, the racial disparity in likelihood of resection of BR-PDAC disappeared.

Resectable Pancreatic Cancer A. Barenboim, 1 * G. Lahat, 1 I. Nachmany, 1 E. Brazowsk, 1 R. Geva, 1 I. Wolf, 1 t. Golan, 2 J. Klausner, 1 N. Lubezky. 1 1. Surgery, Tel Aviv Sourasky Medical center, Tel Aviv, Israel; 2. Tel Hashomer Medical Center, Ramat Gan, Israel. Background: Standard treatment of patients with borderline resectable pancreatic cancer (BRPC) includes neoadjuvant FOLFIRINOX followed by curative-intent resection. Our aim was to evaluate potentialpreoperative clinical, radiological, and biochemical parameters, as well aspathological factors that can predict oncological outcomes in patients undergoing curative resection after FOLFIRINOX. Methods: A retrospective review at two institutions, including all patients with BRPC that underwent curative intent resection after neoadjuvant FOLFIRINOX. We evaluated clinical response (weight change, resolution of pain), tumor markers response, radiological tumor response (RECIST classification), and pathological tumor grade response to chemotherapy. We assessed correlation of response parameters with overall survival (OS)and progression free survival (PFS BACKGROUND: The use of implants for breast reconstruction and augmentation is increasing each year, with a projected increase in use of over 10% by 2025. Subsequently, there has been increased reporting of BI-ALCL cases. The risk and magnitude of this problem have not been adequately evaluated. METHODS: A comprehensive online search of MEDLINE, EMBASE, PubMed and SCOPUS was conducted. Worldwide studies reporting on BI-ALCL occurring with different types of implants between January 1990 and October 2019 were systematically reviewed. Descriptive statistics were used to summarize the findings. RESULTS: Overall, 15 reports with a total of 926 patients diagnosed with BI-ALCL were included. Mean age was 52 years. A history of breast cancer treatment was reported in 244 (26.4%) patients. Mean interval between implant insertion and BI-ALCL diagnosis was 9.5 1.08 years. BI-ALCL was unilateral in 282/300 (94%) and bilateral in 18/300 (6%) reported patients. The most common presentations were seroma/effusion in 316/456 (69.2%) patients and palpable mass in 67/452 (14.8%). Out of 949 implants inserted, 572 (60.3%) were textured implants while 377 (39.7%) (p<0.05) were other types such as smooth or polyurethane coated. Demographic variables and comorbidities were similar between both implant type groups. Reported BI-ALCL specific mortality was 18/373 (4.8%).

CONCLUSION: Textured breast implants pose a higher risk for development of BI-ALCL. BI-ALCL can be associated with significant morbidity and mortality and should be discussed with patients as a potential complication of implant insertion. Background Additional surgery after mastectomy with reconstruction is a known occurrence. The frequency, timing, and incidence of subsequent surgeries are not well described. To better counsel patients on surgical options, we characterized all additional surgical procedures among patients having mastectomy with reconstruction. Methods We retrospectively reviewed patients undergoing mastectomy with implant or autologous tissue reconstruction from 2008-18 for in-situ and invasive carcinoma. Subsequent surgery was categorized as "planned" for multi-stage procedures (i.e. expander to implant exchange), and "unplanned" for complications or cosmetic deformity. Patient characteristics underwent univariate and multivariate analysis. Results A total of 548 patients were identified with a median follow-up of 4 years. Average age was 52 years (range 22-82). Most underwent nipple-sparing (55%) or skin-sparing (41%) mastectomy. Majority (83%) had implant-based reconstruction (Table 1) . Only 27% of patients had no subsequent surgery, while 25% had only planned operations, and 48% had 1 unplanned procedure. The incidence of unplanned procedures by type of reconstruction included: Autologous tissue (52%), 1-Stage Implant (44%), and 2-Stage Implant (53%), with 43% of surgery occurring 0-6 months from the index operation. Total average of unplanned procedures was 1.2 (range 0-7). TRAM flap reconstruction had the highest average subsequent operations (3.6 2), while 1-Stage implant was lowest (2.1 1.4) . For implant reconstruction, 1-Stage had more unplanned procedures (p=0.0002). The only factor associated with 1 unplanned procedure was active smoking (p=0.017); no treatment characteristics increased the incidence of unplanned procedures. On multivariate analysis, risk factors for 2 total subsequent operations included 2-Stage reconstruction (OR= 4.8, , flap ischemia (OR= 7.9, 95% CI 3.9-16), hematoma (OR=4.2, 95% CI 1.9-9.6), infection (OR=2.9, 95% CI 1. 4-5.7) , and capsular contracture (OR=2.4, 95% CI 1. 1-5.4) . Conclusion Seldom "one and done," the likelihood of additional surgery after breast reconstruction is significant and warrants discussion with patients. Introduction Axillary lymph node dissection (ALND) is a procedure associated with high morbidity in breast cancer patients, mainly linked to the development of lymphedema. New techniques have arisen to combat this problem, such as immediate lymphatic reconstruction (ILR), a procedure that is performed at the same time as ALND to restore lymphatic flow after the removal of axillary lymph nodes. This technique requires anastomosing divided draining arm lymphatics to recipient veins to restore flow. Successful ILR depends on the breast surgeon's awareness of the axillary anatomy and preservation of superficial venous structures during the ALND. We sought to describe the current outcomes of ILR and determine the most commonly used superficial veins. Methods An institutional prospective database was established to monitor outcomes of ALND with ILR. We retrospectively reviewed this lymphatic database from 2016-2019. Breast surgery patients who underwent ALND and ILR were included. The success rate of ILR, the number of lymphatic channels bypassed, and the anatomical veins used for lymphatic reconstruction were described. Results In our review, 97 patients underwent attempted ILR after ALND. Of these patients, 81.4% (79/97) had a successful ILR. The median number of channels bypassed was 1 (IQR 1, 2) . The most commonly used vein was the accessory vein in 59.5% (47/79) of the cases, followed by the thoracodorsal vein 13.9% (11/79). (Table 1 ). Conclusion Immediate lymphatic reconstruction is a promising technique for lymphedema prevention after ALND. Successful ILR is dependent on the breast surgeons' technique during ALND and careful preservation of lymphatics and veins for successful bypass. The most commonly used vein for ILR is the accessory vein. Better understanding of the anatomy required for ILR is important for continuing to improve the success of this technique. Background Surgical site infection (SSI) in breast reconstruction is one of the most feared complications for oncologic teams. It could provoke expander/ implant or flap to be removed, consequently putting in hold the reconstructive process and generating emotional distress for the patient. It can also delay oncologic therapy and jeopardize patient's outcome. Several studies have found a relationship between the use of drains and SSI. In this study we aim to determine if there is an association between drain duration and SSI. Methods We performed a retrospective chart review, previous IRB approval, of all patients with breast cancer undergoing reconstructive surgery and diagnosed surgical site infection (SSI) at the National Cancer Institute of Mexico from 2016 to 2019. We performed a 3:1 statistical matching with randomly selected patients from the same database who had no SSI. For the bivariate analysis we included demographic, tumor, surgical and drain duration variables and the outcomes of interest were SSI and reconstruction failure rates. Results Of the total of 456 patients who underwent breast reconstruction, 10% had SSI clinically diagnosed. We performed a subanalysis of the infected group and found that 35 (52.2%) had reconstruction failure. It showed greater proportions of diabetes mellitus (10.5% vs 2.7%; p=0.01) and smoking (17.9% vs 9.1%; p=0.05). We observed a significant difference in drain duration between groups (17.4 days 11.6 vs 13 11.4; p=0.005); reintervention rate (56.7% versus 8%; p=0.00) and the duration of surgery (5.1 hours 3.8 versus 4.5 hours 4.1; p=0.01). There were no differences in age, clinical stage, oncologic treatment, timing of reconstruction or last daily drainage volume. In the multivariate analysis, we found that the odds of having SSI in patients with drain duration >10 days was greater (OR 3.68; p=0.003), the same happened for smoking (OR 3.68; p=0.009). Autologous reconstruction was found a protective factor. Conclusion Patients undergoing breast reconstruction are at risk of SSI when smoking is active and with drains being removed after >10-day. Autologous reconstruction is a protective factor and should be encouraged in high risk patients.

Introduction Invasive lobular carcinoma (ILC) is the second most common type of breast cancer and has unique clinical features. Lacking the adhesion protein E-cadherin, ILC grows in a diffuse pattern and has high positive margin rates, leading some to advocate for more extensive surgical excision in this tumor type. There are scarce data comparing positive margin rates for nipple sparing versus non-nipple sparing mastectomies specifically in ILC. We therefore sought to compare positive margin rates and time to recurrence by mastectomy type in a large cohort of patients with ILC. Methods We queried an institutional database of 700 cases of ILC and included all those with stage I-III disease undergoing mastectomy between 1981-2019, with 6 months follow up. Our predictor was type of mastectomy (total skin sparing mastectomy [TSSM], skin sparing [SSM], or simple). The primary endpoint was positive margin rate, and secondary endpoints were time to local and distant recurrence. Data were analyzed in Stata 14.2. Results Of 342 cases evaluated, 119 (34.8%) had TSSM, 53 (15.5%) had SSM, and 170 (49.7%) had simple mastectomy. Patients undergoing simple mastectomy were significant older, more likely to have lymphovascular invasion (LVI), and had the longest follow up time (Table 1 ). Positive margins rates for TSSM, SSM, and simple mastectomy cases were 12.7%, 17.6%, and 8.5% respectively, with no statistical difference between groups. There was no difference in time to local recurrence by both univariate and multivariate analyses. On univariate analysis, patients undergoing simple mastectomy had significantly shorter time to distant recurrence (p=0.0201). However, when adjusting for age, stage, receptor subtype, grade, LVI, and receipt of radiation, type of mastectomy was no longer associated with distant recurrence. Conclusions Despite a high incidence of positive margins overall in patients with ILC, there is no difference in positive margin rates between TSSM, SSM, or simple mastectomy. Additionally, TSSM is not associated with differences in time to local or distant recurrence. These data support the safety of TSSM in ILC, a distinct subtype of breast cancer.

Type of mastectomy and tumor characteristics, pathology, and recurrence. * Data only available on 324 patients. **Data only available on 323 patients. Background: While obesity is thought to increase complication rates in general surgery procedures, its effect in mastectomy patients remains to be fully elucidated. Obese patients often have advanced disease at presentation warranting neoadjuvant therapy and may have concomitant comorbidities that may affect their postoperative course. We sought to determine if obesity is associated with a higher complication rate and length of stay after mastectomy, independent of clinicopathologic and treatment factors. Methods: Medical records of breast cancer patients undergoing mastectomy at our institution between January 2010 and April 2018 were retrospectively reviewed. Patients were separated into obese (BMI 30) and non-obese (BMI < 30) categories, and compared using non-parametric statistical analyses. Results: Of the 927 patients in this cohort, 292 (31.3%) were obese. Sociodemographic and clinicopathologic factors of the obese and non-obese patients are shown in the table below. In general, obese patients have more complications (26.4% vs. 19 .8%, p = 0.032) and greater number of complications per patient (p = 0.016) than their non-obese counterparts. In particular, they are more likely to have infections (11.0% vs. 3.5%, p = 0.004), flap thrombosis/necrosis (5.5% vs. 2.4%, p = 0.018), and skin breakdown/wound complications (8.9% vs. 4 .4%, p = 0.010). Additionally, obese patients are more likely to have longer hospital length of stay (LOS; LOS > 2 days: 77.4% vs. 65.3%, p < 0.001). Controlling for the potential confounders of race, insurance type, smoking, hypertension, diabetes, type of mastectomy, neoadjuvant chemotherapy, and node positive disease, obesity remained associated with a higher rate of LOS > 2 days (OR = 1.94; 95% CI: 1.34-2.82, p = 0.001), infection (OR = 2.47; 95% CI: 1.21-5.07, p = 0.013), and flap thrombosis/necrosis (OR = 2.89; 95% CI: 1.13-7.42, p = 0.048). Conclusion: Obese breast cancer patients undergoing mastectomy tend to have more complications than non-obese patients, regardless of other comorbidities and clinicopathologic/treatment factors. Background: Over the past decade, therapy targeting the BRAF-MEK signaling pathway has led to a treatment revolution in advanced BRAF-mutant (BRAF mut ) melanoma. However, nearly all patients develop resistant disease. Mechanisms of resistance are not clear; thus, further targeted therapy regimens are currently unavailable. Methods: We created a scRNA (single cell RNA) atlas of 286,003 cells over 19 melanoma tumor specimens from 16 patients. In parallel, we generated tumor-derived cell lines from two BRAF mut samples who were not on BRAF-MEK inhibitor therapy, followed by creation of resistant cell lines after in-vitro BRAF-MEK inhibitor treatment. To characterize features of treatment-resistant cell lines, the original lines and treatment-resistant cells were both subjected to scRNA sequencing analysis via the 10x Genomics platform. For in-vitroviability analysis, Cell-Titer Glo Cell Viability Assay was performed. Results: In BRAF-MEK treatment resistant lines, MET and TRIB3were highly expressed ( Figure 1A ). In addition, a subpopulation of resistant cells demonstrated upregulation of KRAS and JAK2 signaling, suggesting heterogeneity of treatment-resistant cells. Metformin was used to target the TRIB3 signaling pathway, which led to increased cell death for BRAF-MEK resistant melanoma cells, as compared to wild type ( Figure  1B) . Conclusion: scRNA sequencing of BRAF-MEK treatment-resistant cell lines reveals novel mechanisms of resistance. Targeting these pathways appears to overcome resistance and may be incorporated into future treatment regiments for targeted therapy in melanoma. Further work is underway to explore other potential mechanisms of resistance and to develop rationally designed, precision medicine treatment strategies targeting identified pathways. INTRODUCTION: Melanoma of unknown primary (MUP) is characterized by regional or distant metastases without identification of an established primary. The aim of this study is to evaluate MUP biology as a function of presentation, molecular profiling, and response to systemic therapy (Rx). METHODS: A prospectively maintained melanoma database (2008-2019) at a tertiary cancer center was queried for demographics, molecular aberration, treatment, and outcomes. RESULTS: Of 816 melanoma cases, 67 (8.2%) were MUP of which 67% were in men. The most frequent sites of presentation were: nodal (28%), visceral (25%), brain (16%), and subcutaneous (10%). Tumor genomic profiling was performed in 42 of 67 (63%) of patients (pts). Of these, BRAF mutations were identified in 22/42 (52%): V600E 55%, V600K 27%, other 7%, not specified 7%. KIT mutations were found in 3/26 (12%) of tumors tested. Panel testing was performed in 22 pts, with 20/22 having a non-BRAF/KIT mutation that was likely pathogenic, most commonly in the TERT promoter (50%), TP53 32%, CDKN2A 23%, PDGFRA 14%, and NRAS 14%. Of pts with a single site of disease, only 22 (37%) developed a second site of metastasis: 8/19 (42%) pts with nodal MUP, none of 7 pts with subcutaneous MUP, 8/16 (50%) pts with visceral MUP, 3/11 (27%) pts with brain MUP, and 3/5 for all other MUP. Systemic Rx was administered to 37 of 64 (54%) pts: checkpoint inhibitors (25 pts), BRAF/MEK inhibitors (5 pts), cytotoxic chemotherapy (3 pts), interferon (2 pts), and IL-2 (1 pt). Stratified by date of diagnosis, 42% of pts diagnosed between 2008 and 2015 received systemic Rx compared to 84% diagnosed after 2015. Twenty-seven patients received radiation: 74% to the brain, 19% to nodal basins, and 7% to bone. CONCLUSION: Molecular profiling identified likely pathogenic mutations in 90% of MUP pts tested. More than half of pts with MUP did not go on to develop secondary metastatic sites, highlighting the unique biology of this disease, the value of molecular profiling, and the appropriate context of surgical resection in these patients. Immune surveillance in MUP remains under investigation but is likely important in preventing new metastatic disease. Introduction: Approximately 40-50% of melanomas harbor activating BRAF mutations. Similar to checkpoint immunotherapies, BRAF+MEK targeted therapy is associated with significant benefit in BRAF V600mutant high-risk resectable or unresectable/metastatic melanoma. National and international guidelines agree and recommend that all patients with stage III and IV melanoma should be tested for BRAF mutations. Here, we report the resulted BRAF test rates in a real-world setting by utilizing the Flatiron Health electronic health record (EHR)-derived database to highlight implications for melanoma patient care in US oncology practice. Methods: This retrospective observational study utilized Flatiron Health's nationwide longitudinal database comprised of de-identified EHR structured and unstructured data, curated via technology-enabled abstraction. Patients with pathologic stage III or IV cutaneous melanoma at initial diagnosis, or a locoregional or distant recurrence, with 2 clinic encounters in the database on or after January 1, 2011 were included in the advanced melanoma cohort. Data was searched for any evidence of BRAF testing with reported results. Results: A total of 9027 patients from 163 US cancer clinics were in the advanced melanoma cohort (~80% community/~20% academic). In 2018, only 56% of patients in the adjuvant setting and 64% in the first-line (1L) metastatic setting had evidence of known BRAF status before systemic treatment selection, compared to 91% in second-line (2L) metastatic setting. Importantly, within 30 days after treatment initiation, BRAF status was reported in an additional 17% of patients who were receiving 1L metastatic therapy, and in 10% who were receiving adjuvant therapy. Conclusion: BRAF mutation status, prior to the initiation of therapy, is known in approximately 60% of patients in the adjuvant and 1L metastatic settings. These results indicate an opportunity for surgical oncologists in facilitating earlier BRAF testing for patients with advanced melanoma in an effort to better inform patient-specific treatment plans. Surgical oncologists could play an important role in testing patients with stage III or IV melanoma.

Cohort Analysis M. Meneveau, 1 * K. Lynch, 1 C. Slingluff, 1 M. Rutkowski. 2 1. Surgery, University of Virginia, Charlottesville, VA; 2. Dept. of Microbiology, Immunology, and Cancer Biology, University Of Virginia, Charlottesville, VA. Background: Toll-Like Receptors (TLR) play a role in regulating immune responses to cancer. TLR agonists can improve tumor infiltration by lymphocytes and show promise as intratumoral therapies. TLRs are expressed on cancer cells, but their function within them remains unclear. Some suggest that TLR activation promotes cancer cell survival, immune tolerance, and germline mutations of TLR5 are known to limit cancer progression. We hypothesize that somatic mutations of TLR pathway genes in human cancers are associated with improved survival. Methods: Whole exome sequencing data of human cancers were obtained through the Oncology Research Information Exchange Network (ORIEN). The database was searched for mutations in 13 TLR pathway genes: TLR1-10, IRAK1, IRAK4, and MYD88. Tumor mutation burden (TMB) and survival data were collected. Welch's T-test, logrank test, andCox-proportional hazards regression were performed. Results: We identified 10,576 patients with tumor mutation data. 18% harbored one or more TLR pathway mutations. Mutation prevalence by gene ranged from 0.5% (MYD88) to 5% (TLR8). Overall survival was longer in patients with TLR mutations vs those without for the entire cohort (p<0.01), and for colorectal cancer (p=0.03), head and neck cancer (HNC) (p=0.01), melanoma (p=0.04), and pancreatic cancer (p=0.01, Fig. 1a) . However, those with sarcomas harboring TLR mutations trended toward worse survival (p=0.075). TLR mutation was associated with TMB (p<0.01, Fig. 1b) , but inHNC, TLR mutation was independently associated with better survival when controlling for TMB (HR 0.45, 95%CI 0.21, 0.96; p=0 .04, Fig. 1c) . Conclusions: Somatic TLR pathway mutations are associated with longer survival in solid tumors. This effect correlates with TMB but is independently associated with better survival for HNC. Intratumoral therapy with TLR agonists has promise, but 18% of cancers have TLR mutations that may mitigate immune escape. Furthermore, our findings suggest that intact TLR signaling may support tumor progression. Further studies are needed to elucidate the TLRs' function in cancer cells and their interaction with the oncobiome.

Introduction The prognostic and predictive value of intratumoral T and natural killer (NK) cells in soft tissue sarcoma (STS) is unclear. Our objective was to characterize the phenotype of these cells in a cohort of patients undergoing surgery. Materials and Methods Matched blood and tumor specimens were obtained from 11 STS patients. Using flow cytometry, we assessed the frequency of tumor T and NK cells along with expression of activation (Ki67, CD69, NKG2D, HLA-DR) and checkpoint markers (PD-1, TIGIT). We analyzed numbers of tumor T and NK cells, activation status of cells from blood and tumor, and the association of these immune markers with outcome. Results 45% were women, and mean age was 51 16 years. Nine (82%) tumors were high grade, 6 (55%) were dedifferentiated liposarcoma, and 6 (55%) were retroperitoneal. Three patients (27%) experienced disease progression during a median follow-up of 1.7 months (range 0. 6-13.5) . Among total tumor cells, T cells comprised 6 3% while NK cells were 1.5 1% (P>0.05). Ki67 expression was <5% in both T and NK cells. CD69 expression was 41 16% on tumor T cells vs 43 16% on NK cells (P>0.05). In contrast, peripheral expression of CD69 was 4 1% on T cells and 8 2% on NK cells (P<0.05 compared to tumor). NKG2D was highly expressed on both tumor and peripheral T and NK cells (>80%). PD-1 was <10% on NK cells, but highly expressed on both tumor (57 6%) and peripheral (35 4%, P=0.03) T cells. Notably, PD-1 expression was significantly increased on tumor T cells compared to peripheral T cells in patients who experienced STS progression (60 6% vs 27 4%, P=0.01). Inhibitory marker TIGIT was also highly expressed on tumor T (36 4%) and NK (47 5%) cells and was higher on tumor T cells in patients with progression than without (49 6% vs 28 4%, P=0.01). Conclusion In this heterogenous STS cohort, intratumoral T and NK cells express activation markers but also express inhibitory checkpoint markers consistent with dysfunction, which is associated with worse prognosis. Additional studies are needed to validate intratumoral T and NK phenotype as a biomarker of STS prognosis along with strategies to reverse immune dysfunction. Introduction: Natural killer (NK) cells can kill transformed cells without prior antigen sensitization, highly appealing for cancer immunotherapy. Dogs link studies between mice and humans to translate NK immunotherapy from bench-to-bedside. Here, we used RNA sequencing to compare transcriptional profiles of resting and activated canine and human NK cells, and cells isolated from dogs on a clinical trial of inhaled recombinant human (rh) IL-15 to treat metastatic OSA. Methods: Isolated dog and human NK cells were activated ex vivo with 3 days of rhIL-15, with 14 days of co-culture with irradiated K562 cell line, or not activated. We conducted differential gene expression (DGE) analysis to determine global transcriptional profile of treatment groups and visualized differential expression using ratio intensity (MA) plots. Principal component analysis (PCA) determined degree of variance. Gene ontology enrichment analysis evaluated regulators/pathways of NK activation. We used qPCR to verify changes in negative regulators of NK activation. Results: DGE analysis revealed distinct transcriptional profiles between canine NK cell groups with high variation on PCA (PC1 68%, PC2 23%), however human cells had a remarkably lower degree of variance (PC1 79%, PC2 7%, see Figure) . Significant differences in gene expression exist for inhibitory (KIR/ Ly49) and activating (CD56, KLRB1) NK receptors (P<0.05) across conditions and species. TIGIT was markedly upregulated post activation in both species (3.7-fold versus 6.2-fold, P<0.05), whereas we detected no evidence of PD-1 expression. We identified novel upregulated genes (P<0.05) in both species: TOP2A (topoisomerase) and ASPM (mitotic function) in humans and SPP1/OPN (chemotaxis, adhesion) and CCL5/RANTES (NK cell activation, proliferation) in dogs. Preliminary data demonstrate unique gene signatures of responders/non-responders from dog OSA patients treated with inhaled rhIL-15. Conclusion: These data identify gene signatures of resting and activated dog and human NK cells and can be used to generate a predictive signature of responding/non-responding patients treated with NK-based immunotherapies. Background: Perioperative complications, especially seroma formation, are frequent when performing inguinal lymphadenectomy. We have employed a 2-layer negative pressure wound closure technique (2-LNPWT) as a method to reduce seroma rate and perioperative complications, by simultaneously acting as a closed suction drainage system and a compression dressing. We present the outcome of our initial experience with 2-LNPWT and compared the outcomes of its use with traditional closed suction drains (CSD). Material and methods; A 4-year, non-randomised retrospective case-control series analysis was undertaken. Surgeons performing inguinal lymphadenectomy utilised either the 2-layer negative pressure wound (2-LNPWT) therapy or traditional CSDs according to their practice preference. Data included patient demographics, procedure indication and extent, tumour burden, length of stay in hospital, drainage period, seroma formation rate and other complications. Results: The cohort included 111 patients. The cohorts were wellmatched for gender, disease burden, body mass index and co-morbidities. The 2-LNPWT cohort was significantly younger (57.5 v. 69.0 years; p<0.007). There was a significant association with better post-operative outcomes using the 2-LNPWPT compared to CSD in terms of incidence of seroma formation (26.9% v. 49.4%; p<0.03), period of drainage (15 days v. 20 days; p=0.005) and return to theatre rate (0% v. 16.4%; p<0.03). The overall seroma rate was 52.9%. The only significant association was the type of drainage system used (2-LNPWT: 31.2% versus CSD: 58.3%; p<0.03; OR:3.0). There was no significant difference in overall or disease-specific survival detected between the 2 groups. Conclusion: This retrospective non-randomised case-control study has demonstrated the safe use of a novel application of negative pressure wound therapy that substantially improves the incidence of seroma formation and post-operative complication rate for inguinal lymphadenectomy for melanoma. Given the very encouraging data we would recommend that a prospective phase III RCT is undertaken to validate our findings.

Comparison of patient demographics, tumour characteristics and burden, pathology and perioperative course between 2-LNPWT and CSD group *The Clavien-Dindo classification of surgical complications. Objective: Surgical management of melanoma is highly dependent on pathology features (Breslow depth, mitotic rate, ulceration) that determine both the extent of surgical margins and necessity of performing a sentinel node (SLN) biopsy. We sought to assess the impact of dermatopathology second opinion from our affiliated University dermatopathologists on surgical decision making. Design: Patients with newly diagnosed melanoma had pathology reports and slides from referring physicians re-evaluated by our tertiary care dermatopathology group. This was done consecutively on all new referrals. Setting: University-Affiliated Community Skin Cancer Center Main Outcome Measures: Changes in Clinical Stage and Surgical Management Results: From January 2017-July 2019 there was 210 secondary pathology reviews for patients with a new diagnosis of invasive melanoma or melanoma in situ (MIS). There were 134 men (63.8%) and 76 women, mean age 63. Among 210 patients evaluated for primary excision +/-SLN biopsy, secondary pathologic interpretation led to significant changes in the pathology report in 40 (19.0%). A change in clinical stage occurred in 23 (10.9%; 6 patients were upstaged and 17 downstaged), and a change in surgical treatment occurred in 19 (9.0%). For 11 patients (5.2%) the pathology re-interpretation lead to a change in plan regarding SLN biopsy: 8 patients for whom SLN biopsy was planned were changed to wide excision only and in 3 upstaged patients the planned procedure was changed to include a SLN biopsy. For 5 other patients (2.4%) the diagnosis of melanoma was downstaged to either atypical nevus or MIS, decreasing planned margins of surgical excision in 4. Conclusions: Surgeons need to have awareness of the significant challenges of pathologic interpretation of melanoma and other pigmented lesions and should have a low Introduction Infantile melanoma is a rare, poorly described disease, essentially limited to case reports with diverse presentations that often differ from other age groups, including prepubescents. This study examines a national dataset to identify disease characteristics and outcomes in infantile and prepubescent melanoma. Methods Cases of primary cutaneous melanoma in patients £ 12 years of age in the National Cancer Database (NCDB) were identified. Patients were divided into age-based cohorts: <2 years of age (infantile) and 2-12 years of age (prepubescent). Patient demographics, primary tumor characteristics, and overall survival (OS) were compared between groups. Results Of 492 melanoma cases, 67 were infantile and 425 prepubescent. Overall, 55% of the total cohort were female and 77% white non-Hispanic, with almost identical proportions in the infantile and prepubescent groups. Infantile melanomas were most often found on the trunk (34.3%), followed by head and neck (H&N: 25.4%) and lower limb (22.4%). Conversely, in prepubescent patients, tumors were most common in the H&N (27.3%), followed by lower (25.9%) and upper limbs (25.2%) (P=0.02). Infantile (52%) and prepubescent (57%) patients were most likely to present with thick tumors (T3 or T4 disease). Prepubescent patients were more likely to have nodal metastases (32% vs 16%, p<0.001). Three and 5-year OS was significantly different between the infantile (3yr: 85.4%, 5yr: 81.8%) and prepubescent (3yr: 96.9%, 5yr: 96.6%) groups (P<0.001). Age was the only factor that impacted survival on multivariable analysis-infantile melanoma having an over 400% risk of death compared to prepubescent melanoma (HR:4.25; 95%CI: 1.78-10.17). Discussion Infantile and prepubescent melanomas are most often thick lesions with similar distributions across sex and race. However, fewer infantile patients have positive nodes, yet they have significantly decreased survival compared to prepubescent patients. This contradicts the classic observation that thicker lesions tend to have higher rates of nodal positivity and that those patients have decreased survival. Further research is needed to elucidate the likely biological differences of prepubertal and infantile melanoma.

Utility of Staging Imaging in Clinical Stage II Melanoma J.L. Thompson, 1 * L. Kiiskila, 2 O. Oboh, 2 A. Prentice, 1 T. Truong, 2 G.P. Wright. 1 1. General Surgery, Spectrum Health, Grand Rapids, MI; 2. Michigan State University College of Human Medicine, Grand Rapids, MI. Background:National guidelines suggest reserving imaging for patients with pertinent systemic symptoms in the setting of clinical stage II melanoma. Despite this recommendation, interest in preoperative imaging for "high-risk" clinical stage II melanoma has increased in the era of effective systemic therapy. We sought to examine the yield of imaging in this population. Methods:Retrospective review was carried out of consecutive patients with clinical stage II melanoma within a large community health system. The primary outcome was the incidence of metastasis identified on imaging, either pre-or postoperative. Univariate analysis was performed for variables associated with identification of metastasis. A logistic regression model was constructed to identify independent variables associated with metastasis identification on imaging. Results:There were 243 patients with clinical stage II melanoma during the study period; IIA=46.3%, IIB=36.1%, IIC=17.6%. Median age was 67 years (IQR 55-79) and locations varied between head/neck (n=50, 20.6%), trunk (n=95, 39.1%), upper extremity (n=52, 21.4%), and lower extremity (n=46, 18.9%). Median Breslow depth was 2.6mm and ulceration was present in 133 (59.6%). Staging imaging was performed preoperatively in 115 (47.5%) and metastases were identified in 6 patients (5.2%). Sentinel lymph node biopsy was performed in 217 (90.4%) and positive for metastasis in 59 (27.2%). Postoperative imaging was performed in 77 patients with additional metastasis found in 7 patients (9.1%). Overall rate of positive imaging in patients who had imaging performed was 6.8% and 5.3% among the entire cohort. Location of positive imaging findings included regional lymph nodes (n=7), distant metastasis (n=5), and both regional and distant metastasis (n=1). The only variable associated with metastases identified on imaging was presence of lymphovascular invasion (12.5% vs 2.8%, p=0.047). There were no significant independent variables associated with metastases identified on imaging on multivariable analysis. Conclusion:Staging imaging in clinical stage II melanoma is of low clinical value. Judicious utilization of imaging should be exercised according to current national guidelines.

Introduction Advanced primary melanoma is associated with poor survival which often represents aggressive tumor biology but may also reflect competing medical comorbidities. We sought to compare patients with clinical stage IIC melanoma who did and did not undergo sentinel lymph node biopsy (SLNB). Methods Patients treated at a single institution between January 1999 and June 2019 presenting with an ulcerated primary melanoma >4mm in thickness without clinically detected lymphadenopathy or distant metastases were identified from a prospectively maintained database. Melanoma-specific survival (MSS) and overall survival (OS) were estimated with adjustment for clinicopathologic factors using Cox regression. Results A total of 652 patients were identified, of whom 453 (69.5%) underwent WE (Wide excision)/SLNB and 199 WE alone (30.5%). Patients who were selected for WE alone were older (median 74 years versus 62 years, p<0.001), had thicker primary lesions (median 7.0 vs 6.0 mm, p<0.01), and higher mitotic rates (median 10/hpf vs 8/ hpf, p<0.01). Among patients who underwent WE/SLNB, 51% had a positive SLNB. With a median follow-up of 28 months, a positive SLNB was prognostic for 5-year MSS (72.4% if SLNB negative versus 50.9% if SLNB positive, p<0.01). Patients undergoing excision alone had equivalent MSS to patients with a WE and SLNB (65.0% vs 63.2%, p=0.143) at 5 years. However, OS was significantly worse in the WE group (44% vs 29%, p<0.01) at 5 years. Figure 1 . Selection for WE alone remained independently associated with OS (HR 1.7, 95%CI=1.2-0.97) but not MSS (HR 0.96, 95%CI=0.63-1.5) after multivariate adjustment for era of presentation (pre-2010, versus post-2010), age, thickness of primary, and mitotic rate. Within 2 years of surgery, 38% of WE alone patients had died and 66% of the deaths were attributable to non-melanoma causes. Conclusion A substantial proportion of patients with stage IIC melanoma are selected to undergo WE alone. Consistent with MSLT-1, performance of SLNB yields prognostic information, but is not associated with improved MSS. Surgeons treating patients with advanced primary melanomas must consider the competing risks prevalent in this population.

Introduction Cutaneous melanoma and the immune system are intimately related. We aimed to evaluate survival in patients who are exposed to immunosuppressive medication (ISM) prior to or after diagnosis of melanoma, at a population level. Methods Patients diagnosed with invasive cutaneous melanoma (2007) (2008) (2009) (2010) (2011) (2012) (2013) (2014) (2015) were identified from the Ontario Cancer Registry and linked to identify demographics, stage at diagnosis, exposure to medications, and outcomes. Patients eligible for Ontario's Drug Benefit Plan were included to ensure accurate prescription data. The primary outcome was overall survival. Cox Proportional Hazards Regression models identified factors associated with survival, including use of ISM as a time-varying covariate. Sensitivity analysis excluded patients whose exposure to ISM consisted only of steroids. Results 4962 patients had a diagnosis of cutaneous melanoma, with 1609 having exposure to ISM. Steroids were the only ISM used in 1344 (83.5%) patients. Median age [IQR] was 74 [68] [69] [70] [71] [72] [73] [74] [75] [76] [77] [78] [79] [80] in those with and without ISM use. Overall 58.4% of patients were male (60.5% in non-immunosuppressed, 54.1% in those using ISM, p<0.001). In all patients, Stage I disease was most common (50%); however immunosuppressed patients had significantly higher stage at diagnosis (p<0.001). At the end of the study period, 81.6% of patients with no ISM were alive, while 62.6% of patients on ISM were alive and the distribution for the cause of death was statistically different between the two groups (p<0.001). The use of any immunosuppression was associated with an increased hazard of death (Hazard Ratio (HR) 9.1, 95% CI 7.3-11.3, p<0.0001)) when controlling for factors including age, stage, anatomic site, comorbidity, and treatment. Other factors associated with of death were increasing age, male sex, increased stage, truncal location of primary melanoma, and inadequate treatment. In sensitivity analyses, non-steroid ISM was also associated with an increased HR for death (HR 4.9, p 0.0015) . Conclusions The use of immunosuppressive medications in melanoma is associated with worse overall survival. Use of ISM should be limited in at-risk patients when possible. Background: Merkel Cell Carcinoma (MCC) is an aggressive neuroendocrine cutaneous malignancy with a propensity for regional and distant metastasis. Because of the relative infrequency of this disease, the patterns of metastasis in MCC are understudied, which has implications for surveillance and treatment strategies. Methods: Patients with biopsy-proven MCC (1/1/2008-2/28/2018) treated at a single academic institution were identified, and stage at diagnosis was categorized according to the American Joint Committee on Cancer 8 th edition. The first site of metastasis was classified as regional lymph node (LN) basin, in-transit, or distant. Distant metastasis-free survival (DMFS) curves were estimated in patients who presented with stage I-III MCC using the Kaplan-Meier method. Results: Of 133 study patients, 64 (48%) were stage I, 13 (10%) stage II, 48 (36%) stage III, and 8 (6.0%) stage IV at presentation. Among 95 patients with clinically localized disease, 69 (73%) underwent sentinel lymph node biopsy (SLNB) and 18 (26%) were SLN positive. Of 117 patients who underwent cross-sectional imaging at diagnosis, 30 (26%) had regional and 8 (7%) had distant metastases. However, only 2 (2%) patients had isolated distant metastases without evidence of regional disease. After a median (interquartile range) follow-up time of 30 (14-54) months, 78 (59%) patients had developed regional and/or distant metastases. First site of metastatic disease was regional in 87% and distant in 13%. In total, 37 (28%) patients eventually developed distant disease, which most commonly included the abdominal viscera (51%) and distant LNs (46%). The lung (0%) and brain (2.7%) were rarely the first distant sites. Five-year DMFS (95% confidence interval) rates were 71% (56-90%) for stage I, 61% (33-100%) for stage II, and 62% (47-80%) for stage III patients (P=0.13). Conclusion: The regional LN basin is the most common first site of metastasis in MCC, confirming the importance of nodal evaluation in these patients. Isolated distant metastasis at presentation is rare. Certain sites, such as the brain and lung, occur uncommonly as the first distant metastasis, which may help guide surveillance strategies.

Metastasectomy for Malignant Melanoma: Expanding the Indications E. Lilley.* Department of Surgery, Brigham and Women's Hospital, Brookline, MA. Introduction Historically, metastasectomy for melanoma has been reserved to palliate symptomatic disease. However, recent developments in targeted and immunotherapy agents have improved survival and expanded the role for surgery. We sought to describe survival after metastasectomy among patients who received preoperative systemic treatment for metastatic melanoma. Methods This retrospective single-institution chart review study utilized a tissue bank database to identify patients with metastatic melanoma who underwent surgical metastasectomy from 2010-2019 after preoperative systemic treatment. Surgical intent was categorized as curative (resection of all metastatic disease), resection of isolated progression, or palliative (for symptom relief). Results We identified 87 patients who underwent 127 procedures after one or more systemic agents: immunotherapy (95%), targeted therapy (17%), or chemotherapy (22%). The cohort was 64% male with median age of 59 years (interquartile range 52-71 years). Cutaneous melanoma was the most common primary (67%). Surgical resection sites included skin/soft tissue (37%), lymph nodes (33%), intestine (12%), brain (11%), bone (5%), and lung (2%). Surgery for palliative intent (47%) was more common than isolated progression (35%) or curative intent (17%). Overall survival after metastasectomy was 67% at 1-year and 38% at 5-years. Resection for palliative intent was associated with reduced survival at 1-year (46%) and 5-year survival (27%) compared with resection of isolated progression (86% and 43%, hazard ratio 0.43 [95% confidence interval 0.24-0.78]) and resection for curative intent (94% and 61%, hazard ratio 0.24 [0.09-0.60]) (see Figure) . Conclusion In the setting of targeted therapy and immune checkpoint blockade for metastatic melanoma, it is reasonable to expand surgical indications to include resection of persistent or progressing disease sites in patients with at least partial response. Further work is needed to determine which patients will derive the greatest benefit from metastasectomy.

Introduction With a median overall survival (OS) of less than 15 months, patients with metastatic uveal melanoma (UM) derive little benefit and high morbidity from current medical and liver-directed therapies. Genetic profiling has identified recurrent mutations in UM, some of which have been linked to metastatic potential, such as those in SF3B1 and BAP1. Little is known about their prognostic value for long-term oncologic outcomes. Methods We reviewed a prospective cohort of patients with UM referred to the NIH (2013-2016) and University of Pittsburgh (2017-present) for clinical trial evaluation. All underwent metastatic tumor resection and gene expression profiling. The primary endpoint was OS. Time to metastases (TTM) and time with metastatic disease were secondary endpoints. Results 57 consecutive patients with UM were included. Liver metastases were the most common source for genetic analysis in N=33 (58%), followed by subcutaneous (n=9) and omental (n=5). Median age was 56 years, and 51% were female. At a median follow-up of 141 months, n=16 are alive (72% mortality). SF3B1 was mutated in n=23 (40%). BAP1 was mutated in n=20 (35%). Overall, the median TTM was 31 months (2.59 years). TTM was significantly longer for SF3B1 mutated UM (5.86 vs. 2.33 years; HR 3.8, p<0 .0001) and shorter for BAP1 mutation (1.77 vs. 4 .83 years; HR 0.49, 95% CI 0.23-1.03, p=0.02). Median overall survival was 4.66 years. OS was significantly longer for those with an SF3B1 mutation (11.7 vs. 4.5 years; HR 4.0, 95% CI 2.09-7.52, p<0.0001).

Presence of a BAP1 mutation was associated with worse overall survival (4.65 vs. 8.07 years; HR 0.51, 95% CI 0.23-1.02, p=0.02). SF3B1 mutation was also associated with longer survival after metastases (p=0.005), while BAP1 mutation had no association with survival after metastases (p=0.25). Conclusion This is one of the first studies to demonstrate an OS advantage for mutated SF3B1 in metastatic UM. We confirm mutated SF3B1 as a genetic marker associated with favorable metastatic prognosis. Standardized genetic screening to risk-stratify metastatic patients will facilitate selection for clinical trials and overall oncologic management. Introduction: Malignant melanoma is one of most common cancers in the US, and the incidence is increasing. Approximately 50% of patients with advanced melanoma will develop brain metastasis which significantly contributes to overall morbidity and mortality. Molecular profiling has made advancements in the management of melanoma, however, there is still much that is unknown about the significance of certain mutations. Using molecular profiling, we hypothesized that certain markers may predict survival among patients with brain metastases. Methods: We performed a retrospective review of patients who were treated for melanoma at Fox Chase Cancer Center, an NCI designated cancer center, from 2014 to present. Patients with metastatic disease to any site with molecular profiling performed were identified. Univariate analysis and Cox proportional hazards models were performed. Results: 126 patients were identified with metastatic disease who underwent molecular profiling. 115 patients (91.2%) had cutaneous disease, with 67.8% developing brain metastasis. Thirty-seven percent were female, and the mean age at diagnosis was 59.3. In patients with cutaneous melanoma with brain metastases, the median thickness of the primary tumor was 1.83mm with 27 patients (39.7%) presenting with ulceration. The median time for patients to develop brain metastasis was 33.5 months. The most frequently seen mutation was any BRAF mutation in 32% of patients, followed by NRAS mutation (24.3%), and TP53 (11.5%). In patients with brain metastasis, overall survival was worse in patients who displayed a mutation in BRAF V600K (HR 6.74, p=0.018), GNAS (HR 42.5, p=0.008), or MET (HR 4.69, p=0.045) with a median overall survival of 22.3, 12.1, and 14.2 months respectively as compared to 51.6 months with any other mutation. Conclusion: Mutations in BRAF V600K, GNAS and MET in patients with cutaneous melanoma and brain metastases are associated with worse overall survival. Our data suggest these patients may require a more aggressive or individualized treatment strategy. Further research should investigate additional prognostic factors associated with these mutations and development of new targeted therapies. Surgery, Royal Adelaide Hospital, Adelaide, SA, Australia; 2. Moffitt Cancer Center, Tampa, FL; 3. Duke University Hospital, Durham, NC; 4. Winship Cancer Institute, Emory University, Atlanta, GA; 5. Fox Chase Cancer Center, Philadelphia, PA; 6. Peter MacCallum Cancer Centre, Melbourne, VIC, Australia; 7. The Alfred Hospital, Melbourne, VIC, Australia; 8. Princess Alexandra Hospital, Brisbane, QLD, Australia; 9. University Texas Medical Branch, Galveston, TX; 10. Melanoma Institute Australia, Sydney, NSW, Australia. Introduction: Isolated limb infusion (ILI) is a minimally-invasive procedure for delivering high-dose regional chemotherapy to patients with locally advanced or in-transit melanoma confined to a limb. The aim of this international multi-center study was to identify predictive factors for toxicity and response after ILI for AJCC 7 th edition stage IIIB/C limb melanoma. Patients and Methods: Data from patients who underwent a first ILI between 1992 and 2018 were collected at five Australian and four US tertiary referral centers. Primary outcome measures were toxicity and response. Results: After a first ILI 487 (71%) of 687 patients experienced grade I/II limb toxicity, 165 (24%) grade II, and 27 (4%) grade IV limb toxicity (missing n=8; 1%). No patient had grade V limb toxicity (necessitating amputation). Predictive factors for more severe limb toxicity were female gender, lower limb procedures, higher melphalan dose and increased drug circulation and ischemia times. An overall response was observed in 441 patients (64.1%), of whom 199 (28.9%) had a complete response (CR). Patients with a CR had a lower burden of disease, earlier stage of disease (stage IIIB vs. IIIC) and a lower Breslow thickness of the primary melanoma. In addition, an overall response was associated with female gender, lower limb ILI, lower actinomycin-D doses and a delayed serum creatine phosphokinase peak. Upon multivariate analysis, a higher melphalan dose (for toxicity), and stage of disease and burden of disease (both for overall response) remained predictive factors. Conclusion: This international multi-center study shows that ILI is safe and effective to treat locally advanced limb melanoma. Knowledge of the predictive factors for toxicity and response will improve patient selection and allow optimization of intra-operative factors to increase response rates, while keeping toxicity low.

Background: The phase 3 study CheckMate 238 demonstrated improved relapse-free survival (RFS) with NIVO 3 mg/kg vs IPI 10 mg/kg in patients (pts) with resected stage III/IV melanoma. Sustained efficacy benefit at 24 mo of follow-up with NIVO vs IPI was previously reported. Here we present a 36-mo analysis of efficacy and biomarker data from this study. Methods: Pts aged 15 y with completely resected stage IIIB/C or IV melanoma were randomized 1:1 to receive NIVO (3 mg/kg Q2W; N = 453) or IPI (10 mg/kg Q3W for 4 doses; Q12W thereafter; N = 453) for 1 y or until disease recurrence/unacceptable toxicity. RFS was the primary endpoint; exploratory endpoints included distant metastases-free survival (DMFS) in pts with stage III disease and potential biomarkers of efficacy. Results: With 36 mo of follow-up, NIVO continued to demonstrate superior RFS vs IPI (HR, 0.68; P < 0.0001; 3-y RFS rates, 58% vs 45% and 188/453 vs 239/453 pts with events, respectively). Prespecified subgroup analyses demonstrated a consistent pattern similar to that of the 24-mo analysis, with HRs favoring NIVO (table) . DMFS analysis also continued to favor NIVO (table). High levels of all biomarkers analyzed (interferon-gamma gene expression signature, tumor mutational burden, and CD8+ T-cell infiltration by immunohistochemistry) showed an association with improved RFS for both NIVO and IPI. Median RFS (mo; 24-mo follow-up) by high vs low values for each biomarker for NIVO was 30.8 vs 24.1, not reached (NR) vs 30. 8, and 30.8 vs 24.9, respectively; and for IPI was NR vs 15.9, NR vs 18.3, and NR vs 13.8, respectively. Conclusions: With 36 mo of follow-up, NIVO continued to demonstrate superior efficacy over IPI in pts with stage III/IV melanoma at high risk of recurrence across pt subgroups. Additional analyses using composite scoring of biomarker combinations will be presented.

Modern Treatment of Inflammatory Breast Cancer: Analysis from the National Cancer Database M. Grova,* E.E. Navajas, p. strassle, K.K. Gallagher, D. Ollila, S. Downs-Canner, P.M. Spanheimer. Surgery, University of North Carolina, Chapel Hill, NC. Background: Inflammatory breast cancer (IBC) is an aggressive subset characterized by high rates of locoregional and distant failure. Using a modern cohort to assess national trends in treatment patterns, we investigated the yield of pathologically positive lymph nodes and prognostic value of routine axillary lymph node dissection after neoadjuvant chemotherapy (NAC) by receptor subtype in IBC. Methods: Adult women diagnosed with clinical stage T4d, N0-N3, M0 inflammatory breast cancer from 2012 to 2016 in the National Cancer Database were included. Kaplan Meier survival curves and Cox regression were used to assess mortality by receptor subtype and nodal status. Results: We identified 6,039 patients, of which 2,150 (36%) were hormone receptor (HR)+/HER2-, 2,0239 (34%) were HER2+, and 1,486 (25%) were triple negative (TN). NAC was received by 70% of patients, 90% in total received chemotherapy, and 68%received radiation (RT). Mastectomy was performed in 79% of patients, 2% had lumpectomy, and 19% did not have resection of the primary tumor. In patients that were resected, 87% had axillary lymph node dissection (ALND) and 6% had sentinel lymph node biopsy (SLNB) without ALND. For the entire cohort, 5-year overall survival was 50% for all patients, 49% for HR+, 64% for HER2+, and 33% for TN, figure  1A . Pathologic complete response in the breast and axilla after NAC occurred in 6% of HR+, 36% of HER2+ patients, and 18% of TN patients. After NAC, 76% of HR+ were ypN+ compared to 37% of HER2+, and 27% of TN. In patients that were cN0, 50% of HR+, 21% of HER2+, and 27% of TN were ypN+. In cN+ patients, 72% of HR+, 34% of HER2+, and 53% of TN patients were ypN+. ypN+ correlated with survival in all subtypes ( figure 1B-D) , with the most pronounced risk stratification for TN (p<0.001). Conclusions: Fiveyear overall survival for IBC is 50% in this modern cohort for all comers and the HER2+ subtype has the best prognosis. The probability of positive nodes after NAC varied by subtype, but is sufficiently high, even in cN0 patients, to support continued routine pathologic assessment in IBC. More accurate, less morbid alternatives to ALND are needed to identify the ypN0 axilla. Introduction: The Commission on Cancer (CoC) issues Cancer Program Practice Profile Reports (CP3R) that set standards for the provision of highquality care. Three metrics for breast cancer (BC) include: radiation administered within 1 year of diagnosis for women <70 receiving breast conserving surgery (BCSRT), radiation recommended within 1 year after mastectomy for women with 4 positive regional lymph nodes (MASTRT), and hormonal therapy recommended within 1 year of diagnosis of a stage IB-III hormone receptor-positive BC (HT). Our study evaluates national trends in compliance with these quality metrics. Methods: The National Cancer Database (NCDB) was queried from 2004 to 2015 to identify all patients who met criteria for the three quality metrics. National trends in compliance were compared by CoC facility type. Multivariate logistic regression was used to identify factors associated with patient compliance. Results: 1,597,971 patients who qualified for BSCRT (n=965,553), MASTRT (n=116,089), or HT (n=516,329) were identified. In 2015, 80.3% of patients met BCSRT, 69.9% met MASTRT, and 81.1% met HT. BSCRT compliance rates were similar between community and academic programs (79.3% vs 79.3%, p=0.85), while MASTRT and HT rates were lower in community hospitals compared to academic programs (66.5% vs 72.9%, p<0.01, and 76.8% vs 82.0%, p<0.01, respectively). On multivariate analysis, patients receiving care at academic programs (OR 1.17, 95% CI 1.1-1.2), insured patients (OR 1.59, 95% CI 1.4-1.7), and higher income patients (OR 1.07, 95% CI 1.02-1.1) had higher odds of BCSRT compliance. Older patients ages 66-90 (OR 0.8, 95% CI 0.69-0.9), longer travel distance (OR 0.81, 95% CI 0.77-0.84), and minorities (OR 0.79, 95% CI 0.77-0.83) had lower odds of BCSRT compliance. Similar results were seen for MASTRT and HT. Conclusion: Compliance rates for breast cancer quality metrics BCSRT, MASTRT and HT are significantly lower than the 90% CoC standard. Patients with insurance and those receiving care at academic programs had higher odds of meeting quality metric goals, while patients who traveled farther and minority populations had lower odds. Purpose: Occult primary breast cancer (OPBC) is a rare occurence in which a patient has breast cancer in regional lymph nodes but no breast lesion is noted. As such, it remains unclear how locoregional treatment affects outcome. Methods: The National Cancer Database was used to evaluate women 18 years and older, diagnosed with carcinoma of breast origin, from 2004 to 2016. Patients were pathologically node positive, with an unknown primary and no evidence of distant metastasis (cTX or cT0, N1-3, M0). Patients were divided based upon locoregional therapy modality: axillary surgery (AS) with radiation therapy (RT) or mastectomy (TM), AS+TM, AS+RT, and AS+TM+RT. Univariable analyses were conducted to determine the association of locoregional treatment modalities, including RT and TM on overall survival (OS). Multivariable Cox proportional hazard models were used to adjust for patient age, year diagnosed, histology, grade, and race. Results: 813 patients met inclusion criteria. 120 received no locoregional therapy other than AS, 202 received AS+TM, 237 received AS+RT, and 254 received AS+RT+TM. 97% of patients had at least one lymph node removed, with a median 13 nodes removed. Women in the mastectomy group were younger than those that did not receive TM (58.1 vs 61.3 years; p <0.001). There was a decrease in the percentage of patients treated with surgery during the period studied. In 2004, 75% of patients were treated with surgery versus 49% in 2015. No statistical differences were noted in treatment between patients with different grade tumors, histology, or race. Women who underwent any locoregional therapy had longer median OS compared to those who did not (38.4 months vs 45.8; p=0.016). Among women who received any local treatment, there were no significant differences in OS. Conclusions: A national trend exists towards omitting mastectomy for OPBC patients. This trend is supported by data that demonstrates no association between mastectomy and improved OS. However, those patients who receive local therapy (surgery or radiation therapy) have better OS than those treated with systemic therapy alone. Introduction: Invasive papillary carcinoma (IPC) of the breast is a rare form of breast cancer, which is thought to have a better prognosis than invasive ductal carcinoma (IDC). The aim of this study was to investigate the clinical or pathologic characteristics and survival between IPC and IDC using National Cancer Database. Method: Female patients diagnosed with malignant IPC and IDC between 2005-2014 were included. Clinical or pathologic parameters included age, race, insurance status, income, education, residence location, treating facility type, primary tumor site, laterality, grade, stage, regional LN status, ER/PR/HER2 status, and adjuvant treatments. Overall survival (OS) was compared between IPC and IDC. Result: Overall, 308,426 patients were identified, of whom 1,147 had IPC and 307,279 had IDC. Median follow-up time 53 vs. 57 months. IPCs occurred in older black women ( 50 years old), with lower grades, early stage disease, higher rates of ER and PR positivity, decreased lymph node involvement, and were less likely to receive radiation and chemotherapy than patients with IDC. IPC had a five-year OS rate of 86.8%, which was similar as compared to IDC 88.7% overall (P=0.06). After propensity score (PS) matching, no significant difference was observed in the five-year OS rate between IPC and IDC (86.6% vs. 87.6%, P=0.33). Insurance status, axillary LN status, staging, radiation and chemotherapy treatment was shown to be independent prognostic factors of IPC on multivariate analysis. Conclusion: IPC has a similar prognosis as compared to IDC, suggesting that these patients should follow the same treatment protocols. Introduction Axillary lymph node dissection (ALND) is an important component of breast cancer staging and regional control. Neoadjuvant chemotherapy (NAC) is being utilized more frequently as several studies have shown the predictive advantage of systemic therapy prior to surgery. However, there is controversy within the literature whether NAC decreases the total axillary lymph node yield. Four small single-institution studies showed that total ALND lymph node yield was lower in patients who underwent NAC. However, a larger retrospective single-institution review of 698 patients showed no difference in the number of lymph nodes in ALND after NAC. With these variable results, we designed a study to review the national cancer database and identify if there is a difference between the total axillary lymph node yield in patients with and without NAC. Methods Review of the National Cancer DataBase (NCDB) database from 2010-2015 was performed. Breast cancer patients who underwent ALND with or without prior sentinel lymph node dissection (SLN) were included. Number of nodes included the total nodes from SLN, if performed, and ALND. Patients who underwent NAC were identified. Descriptive statistics and comparisons using the Wilcox Rank-Sum test were used for the summary analysis. Results Review of the NCDB database included 129,890 patients. Mean age was 55 years old and ranged from 18-90 years old. Of the cohort included, 25.5% (33,085/129,890) of patients underwent NAC before their lymph node surgery and 74.5% (96,805/129,890) did not have any listed chemotherapy prior to lymph node surgery. Type of tumor receptor subtype distribution is as shown in Table 1 . Median total lymph nodes in patients with NAC was 11 (IQR 6,16) and 11 (IQR 6,17) without NAC. (Table 1 ) Conclusion NCDB database review of 129,890 patients showed that there is no difference in total axillary lymph node yield after NAC. This is currently the largest review of patients and helps to further verify that NAC does not compromise oncologic staging or lymph node yield. Background: Complete pathologic response (pCR) to neoadjuvant therapy (NAT) is variable among patients with HER2 overexpressing (HER2+) tumors. The association between HER2 protein overexpression or gene amplification and response to NAT has been inconsistently reported in the literature. The objective of this study was to determine if the degree of HER2 protein overexpression and/or gene amplification may be used to predict the extent of response to NAT. Methods: Patients with operable HER2+ breast cancer who underwent NAT followed by surgical resection between 2007 and 2017 were identified from a prospectively maintained database. HER2 status was determined by immunohistochemistry (IHC) and/or FISH obtained from core needle biopsy. The HER2/CEP17 ratio and average HER2 copies/tumor cells were recorded, Histopathologic characteristics, clinical T stage, pathologic T stage, and residual disease were evaluated. Two-sample t tests were used to compare continuous variables while Fisher's exact test was used to compare categorical variables. Results: Of the 160 patients with HER2+ breast cancer identified, 85 (53%) had a pCR after NAT. In addition to standard chemotherapy, 105 patients (66%) received neoadjuvant Trastuzumab/Pertuzumab dual therapy, 47 (29%) received Trastuzumab alone, while the treatment was unknown in 8 (5%). The anti-HER2 regimen did not correlate with pCR (p = 0.30). 121 patients had tumor HER2 overexpression determined by IHC, with 103 reported as 3+ and 16 reported as 2+. Two tumors were 1+ on IHC but found to have gene amplification on FISH. Patients with 3+ on IHC were more likely to obtain a pCR (p = 0.001). 102 patients had FISH testing performed; however, only 97 patients had both the HER/CEP17 ratio and average HER2 copies reported. When HER2 gene amplification was examined, we found that those with a higher HER2/CEP17 ratio and higher average HER2 copies/tumor cell were more likely to have a pCR following NAT (p = 0.009 and p=0.003, respectively). Conclusion: For patients with HER2+ breast cancer who undergo NAT, the degree of HER2 protein overexpression and gene amplification were associated with response to NAT.

Average FISH score comparison of patients who obtained pCR vs. those who did not *p <0.05

Analysis P. Jadeja, 1 * V. Jadeja, 3 R. Ha, 2 L. Wiechmann, 2 R. Rao, 2 B. Taback. 2 1. Breast Surgery, Summit Medical Group, Livingston, NJ; 2. Columbia University Medical Center, New York, NY; 3. Independent, Morristown, NJ. Background: Sequence of treatment for early-stage triple negative (TNBC) and HER2 positive breast cancers is a greatly debated topic. Proponents of neoadjuvant chemotherapy (NAC) cite de-escalating axillary surgery and opportunities for additional treatment. This study utilized the National Cancer Database (NCDB) to evaluate outcomes in early stage breast cancer patients undergoing breast conserving surgery (BCS). Methods: This study analyzed patients from the NCDB with T1, N0, M0 TNBC or HER2 positive breast cancer who underwent BCS from January 2010 through 2015. The patients were divided into the following groups: NAC versus adjuvant chemotherapy (AC). Margins, nodal status, overall survival, age, and time to treatment were analyzed. Comparison analyses using SPSS were performed with p < 0.05 deemed significant. Results: Among the 7864 patients who underwent BCS surgery for T1 breast cancer, 76.2% demonstrated TNBC and 23.8% HER2 positive subtypes. The majority of patients in both groups underwent AC (94.8%). There was no statistically significant difference in margin positivity (p=0.198) or days to first treatment (p=0.062). Patients undergoing NAC were more likely to be younger (53.5 years vs 58.5 years, p<0.001). Sentinel lymph nodes (SLN) were more likely to be negative among patients undergoing NAC (97.8% vs 87.8%, p<0.001). Overall SLN and non-SLN positivity was greater amongst patients undergoing AC (13.5% versus 4.9%, p <0.001). Node negative patients undergoing AC had a significantly greater OS (94.6% vs 92.1%) with a mean survival of 80.4 months vs. 75.9 months (p=0.03) when compared to node negative patients following NAC. Conclusions: Patients with earlystage TNBC or HER2 positive cancers are more likely to have negative SLN and undergo fewer completion axillary lymph node dissection when treated with NAC. However, margin status is unaffected and OS is higher in patients receiving AC. Therefore, treatment with upfront surgery may allow for better prognostication by determining more accurate nodal status and therefore optimizing systemic therapy in these high-risk patients. Introduction Recent studies suggest that breast cancer (BC) patients are at increased risk for locoregional recurrence (LRR) after neoadjuvant chemotherapy (NAC) versus surgery first. Our hypothesis is that LRR after NAC is affected by tumor subtype, NAC response, and extent of breast surgery. Methods We performed a single-institution retrospective review of female, node-positive BC patients with minimum follow-up (F/U) 60 months (mo) treated with NAC and breast/lymph node surgery from 2000-2014. Clinical, pathologic, treatment, and outcome data were collected. Statistical analyses performed using Wilcoxon rank sum test and Pearson's correlation. Results After median F/U of 88mo (range 60-226), 225 NAC cases (224 patients) were reviewed; median age 51 years (range 20-76) and median BMI 28 kg/m 2 (range 14-55). Most (168, 75%) received anthracycline-based NAC. Overall 13 (8%) LRR; no LRR after NAC that included targeted Her2 agents (0/57)(p=0.14). No differences in LRR by tumor subtype (p=0.87), hormone receptor status (p=0.46), grade (p =0.49), or histology (p=0.67) (Table) . After NAC, 172 (76%) had mastectomy and 161/172 (94%) post-mastectomy radiation (PMRT), 63/161 (39%) with regional nodes (RNRT). 53 patients had lumpectomy and 51/53 (96%) adjuvant radiation (RT), 28/51 (55%) with RNRT. There was no difference in LRR comparing in-breast partial response, pathologic complete response (pCR), and no response to NAC (p=0.75). Patients without nodal pCR (152/225, 66%) had increased risk of LRR vs patients with nodal pCR (p=0.05). Cases with ypN2 (8/45, 18%) or ypN3 (3/23, 13%) had higher risk of LRR than ypN0 (1/73, 1.4%) or ypN1 (1/84, 1.2%)(p=0.001). Among 13/255 (6%) LRR patients, 6 had lumpectomy and RT (2/6 with RNRT) at median time to LRR of 27mo (IQR 21-33). The other 7 received PMRT (5/7 with RNRT), with median time to LRR of 33mo . Risk of LRR was higher after lumpectomy with recurrence rate of 11% vs 4% (p=0.048). Conclusions Overall incidence of LRR after NAC and 5+ years F/U is comparable to published LRR rates after surgery first. In-breast response did not affect LRR risk. Our study found higher ypN stage and breast conserving surgery were significantly associated with LRR.

Variables and outcomes by breast cancer subtype. IQR: interquartile range; IDC: invasive ductal carcinoma; T: size of tumor at presentation; pCR: pathologic complete response; *1 patient with insufficient pathologic information to assign subtype Introduction: Neoadjuvant chemotherapy (NAC) reduces the extent of in-breast surgery and is increasingly applied to reduce axillary surgery after NAC as well. We hypothesize that complete in-breast response to NAC does not predict complete axillary nodal response. Methods: We performed a singleinstitution review of female breast cancer patients with nodal metastases treated with NAC, resection and lymph node surgery between 2000-2019. Clinicopathologic and outcome data were collected. Pearson correlation (R) and a linear regression model were used to assess relationship between in-breast and axillary response to NAC. Results: We reviewed 711 cases (707 patients) with median age of 51 years (IQR 43-59) and median BMI 27.4 kg/m 2 (23.5-31.6) at diagnosis. Most tumors were T2 (50%), grade 3 (64%), invasive ductal (IDC) (88%), Estrogen (ER) (62%) and Progesterone receptor (PR) (51%) positive and HER2 negative (66%) with biopsy proven nodal metastases (89%) treated with mastectomy (81%) and axillary dissection (93%). After NAC, 207 patients (29.1%) had in-breast pathologic complete response (pCR) and 254 (36%) had nodal pCR; 157 (22%) had pCR in both (Table) . A linear regression model of nodal versus in-breast response demonstrated better nodal response to NAC associated with better in-breast response ( =1.28). IDC type, grade 3, ER and PR negative, HER2 positive tumors were associated with both in-breast and nodal pCR (p 0.001). Conversely, there was moderate correlation of breast pCR to nodal pCR (R=0.52, p<0.01). After median follow-up of 39 months (IQR 22-71), 176 patients (25%) recurred (28 [4%] local, 10 [1%] regional, 6 [<1%] locoregional (LRR), 131 distant [18%]). All patients who developed LRR had received prior adjuvant chest wall or whole breast radiation. Conclusions: We identified characteristics significantly associated in-breast and nodal pCR, however response to NAC was not symmetric. Interestingly, our study suggests that nodal pCR is significantly correlated to in-breast pCR, but in-breast response to NAC is not predictive of nodal response. Therefore, clinical in-breast response after NAC should not decrease the importance of appropriate axillary clearance. Background: HER2+ breast cancers often present with prominent calcifications or non-mass enhancement (NME) on imaging which suggest the presence of extensive disease. However, with neoadjuvant HER2-targeted therapies, a significant proportion of patients experience pathologic complete response at time of surgery. The objective of this study was to correlate pre-treatment radiographic findings with pathologic extent of disease at surgery in patients with HER2+ breast cancer treated with neoadjuvant HER2-targeted therapy. Methods: The institutional Breast Surgery QI database was queried from January 2016 through January 2018 for patients with HER2+ breast cancer who underwent neoadjuvant HER2-targeted therapy followed by definitive surgery. An attending breast radiologist reviewed all preoperative imaging and the radiographic extent of disease was compared with the pathologic response seen at time of surgery. Results: A total of 182 patients have been identified including 64 that have been reviewed for this analysis (Table) . 36 patients had both mammogram and MRI prior to initiating therapy and 28 had mammogram only. Calcifications were identified on mammogram in 36 (56.3%) and NME was identified on MRI in 22 (34.4%). In total, 62.5% (40/64) had calcifications and/or NME on pre-treatment imaging. Of the 40 patients with pre-treatment calcifications and/or NME, 12 (30.0%) had no invasive or in-situ disease at time of surgery, 9 (22.5%) had only in-situ disease, and 19 (47.5%) had residual invasive disease. Of the 24 patients without calcifications or NME, 9 (37.5%) had no residual disease, 2 (8.3%) had only in-situ disease, and 13 (54.2%) had residual invasive disease. Conclusions: Nearly two-thirds of HER2+ patients have imaging findings at diagnosis suggestive of extensive disease. However, 30% of these patients have a complete pathologic response, defined as no residual invasive or in-situ disease, following neoadjuvant HER2-targeted therapy, suggesting that the pre-treatment extent of disease may not preclude conservative surgery. Attention should be paid to both pre-and post-treatment extent of disease when determining extent of surgical resection. Introduction Occult primary breast cancer (OPBC), defined by carcinoma in axillary lymph nodes with no evident primary breast lesion despite full physical and radiographic assessment, accounts for <1% of all newly diagnosed breast cancers. The management of this rare entity has evolved from a modified radical mastectomy to axillary lymph node dissection (ALND), breast radiation therapy and appropriate systemic therapy. Neoadjuvant chemotherapy (NAC) is increasingly used for patients with an intact primary breast cancer and N+ disease to downstage the axilla. Patients who convert from cN1 to cN0 following NAC are eligible for a sentinel lymph node biopsy (SLNB) and omission of ALND if pN-. We sought to determine the frequency of nodal pathologic complete response (pCR) following NAC in a cohort of patients with OPBC. Methods We retrospectively identified 27 patients with biopsy-proven cN+ stage II/III OPBC treated between 1/2008-9/2019. Eight were excluded; 3 underwent lymph node excisional biopsy and 5 did not receive NAC. The remaining 19 were cN1-3, had pretreatment lymph node needle biopsy and received NAC. Results Of 19 OPBC patients, median age was 53 years and 79% were of ductal histology. Eight (42%) were triple negative (TN), 6 (32%) were ER+/HER2-and 5 (26%) were HER2+. Among the 14 cN1 patients, 13 converted to cN0 following NAC; 8 (62%) underwent SLNB, all of whom achieved a nodal pCR (100%). The remaining 5 (38%) underwent ALND; all achieved nodal pCR (100%). Patients with cN2-3 disease underwent ALND; 2 achieved pCR while 3 remained pN+. Table 1 shows the pCR rates by receptor profile: 3/6 ER+/HER2-, 7/8 TN, and 5/5 HER2+. There have been no nodal recurrences among women treated with SLNB alone. Conclusion Among women with OPBC treated in the modern era, we see rates of nodal pCR that are excellent and numerically exceed those in the literature with an overall nodal pCR rate of 79%-which varies by receptor profile. In this small cohort of patients, all women who converted from cN1 to cN0 had a pCR and could be spared an ALND. For women presenting with this uncommon clinical entity, a large proportion will be eligible for downstaging of axillary surgery following NAC. Introduction: Response to preoperative therapy is known to correlate with overall survival in patients with resected gastric cancer. However, it is unknown whether the degree of response to therapy correlates with locoregional (LR) or distant recurrence (DR). Methods: We reviewed all patients at our institution who underwent resection of gastric adenocarcinoma following preoperative chemoradiation from 1995 through 2015. Response to therapy was graded using the percentage of viable tumor in the resected specimen (tumor regression grade [TRG] 0 = 0%, 1 = 1-2%, 2 = 3-50%, 3 = >50%). Relationships between TRG and recurrence-free survival (RFS), LR, and DR were examined. Results: 247 patients met inclusion criteria. Tumors were poorly differentiated (71%), located in the gastric body (41%) and gastroesophageal junction (27%), and of advanced clinical stage (cT 3/4, 78%; cN+, 57%). TRG0-3 was distributed as follows: TRG0 (n = 52, 21%), TRG1 (n = 49, 20%), TRG2 (n = 98, 40%), and TRG3 (n = 48, 20%). DR occurred in 32% of patients, and LR occurred in 6%. Five-year RFS rates were 79%, 62%, 54%, and 42% for patients with TRG0, 1, 2, and 3, respectively (p = .003, log-rank test, Figure 1 ). However, in multivariable analysis TRG was not independently associated with RFS, LR, or DR. ypN status was independently associated with DR (hazard ratio [HR] 2.44, p < .001) but not LR. R1 resection was independently associated with LR (HR 17.85, p < .001) but not DR. The rate of R1 resection (6-15%, p = .213) and incidence of LR (6-8%, p = .892) were not significantly different in TRG1-3 patients. However, none of the 52 patients with TRG0 experienced LR. Conclusion: No LR was observed among patients with TRG0, indicating excellent local control when complete pathologic response to chemoradiation is achieved. However, there were no differences in the R1 resection rate among patients with TRG1-3, and R1 resection rate continues to define the risk of LR. As a result, TRG is not independently associated with LR. The risk of DR is defined by ypN status, not TRG. 

First-line Chemotherapy Versus Chemoradiation for Resectable Distal Esophageal Adenocarcinoma S.W. de Geus, 1 * S. Hirji, 2 S. Ng, 1 K. Suzuki, 1 T.E. Sachs, 1 V.R. Litle, 1 J.F. Tseng. 1 1. Surgery, Boston Medical Center, Boston, MA; 2. Brigham and Women's Hospital, Boston, MA. Background: Multiple randomized trials have shown that both neoadjuvant chemotherapy (CT) and chemoradiation (CRT) convey survival benefit as compared to upfront surgery in patients with locoregional advanced esophageal adenocarcinoma. The present study compares the pathologic response rates and overall survival of first-line CT versus CRT for patients with distal esophageal adenocarcinoma. Methods: Patients with clinical stage T2-T3, N0-N+ esophageal adenocarcinoma originating in the distal esophagus, who received first-line CT or CRT were identified from the National Cancer Data Base (2006) (2007) (2008) (2009) (2010) (2011) (2012) (2013) (2014) . Propensity-scores were created for the odds of receiving CRT. Patients were matched 1:1 based on propensity score. Subset analysis was performed in patients who underwent esophagectomy. Pathological complete response was defined as ypT0N0M0. Results: In total, 709 and 8,877 patients who received first-line CT and CRT were identified, respectively. Unadjusted, patients who received first-line CT had a significantly lower clinical stage (cT2: 27.2% vs. 23.3%; p=0.017) and were more often treated at a high-volume center (27.2% vs. 20.2%; p<0.001) compared to patients who received first-line CRT. After matching, resection rates were comparable for patients who received first-line CT or CRT (62.2% vs. 63.7%; p=0.545); however, median overall survival was slightly lower for patients who receive CT compared to CRT (23.7 vs. 28.4 months; p=0.044). Among patients who underwent esophagectomy, time to surgery (135 vs. 134 days; p=0.689) and median overall survival (37.0 vs. 40.5 months; p=0.630) was similar between matched cohorts. However, complete response (15.8% vs. 25.8%; p<0 .001) and negative margin (88.9% vs. 94.3%; p=0.004) rates were significantly lower after CT compared to CRT. Conclusion: In all patients with esophageal adenocarcinoma, first-line CRT results in significant survival benefit compared to CT. In patients who proceed to surgery, first-line CRT also demonstrates higher pathological complete response and negative resection margins rates. These findings suggest that first-line CRT is preferable over CT when tolerated in patients with distal esophageal adenocarcinoma. Introduction: In patients who undergo curative intent surgery for gastric cancer, current NCCN guidelines call for a minimum lymph node yield (LNY) of 15 lymph nodes. We investigated the impact of neoadjuvant therapy on LNY. Methods: We selected patients diagnosed with gastric adenocarcinoma who underwent surgical resection between 2004 and 2016 from the National Cancer Database (NCDB). Using the chi-square and Kruskal-Wallis tests, we performed univariate, multivariable, and stratified analyses to assess the impact of neoadjuvant therapy on LNY. Results: A total of 57,895 patients underwent surgical resection for gastric cancer and we then selected those who received neoadjuvant therapy. On univariate analysis, factors significantly associated with a lower median LNY included: primary tumor site in antrum/pylorus, neoadjuvant chemoradiation, treatment at facility type other than academic/research cancer program, subtotal gastrectomy, open surgical approach and favorable features such as lack of lymphovascular invasion and well differentiated grade. We stratified by patients into those receiving neoadjuvant chemotherapy (NC) v. neoadjuvant chemoradiation (NCXRT) v. no neoadjuvant therapy, we found that 7% received NC, 51% received NCXRT, and 42% had no neoadjuvant therapy (see Table) . On this stratified analysis, NCXRT patients consistently had a lower LNY compared to NC and no neoadjuvant therapy, across primary site, facility type, type of surgery and surgical approach (median 15 vs 19 and 17, p< 0.0001). Lastly, when we performed a multivariable analysis to evaluate predictors of LNY less than 15, we found that receipt of neoadjuvant chemotherapy was significant (OR 2.1, 1.2-2.5, p=0.009), as well as facility type (OR 0.6, 0.5-0.8, p=0.0002). Conclusion: Patients who receive neoadjuvant chemoradiation, in general, have a lower LNY when compared to patients who receive neoadjuvant chemotherapy or no neoadjuvant therapy. However, receipt of neoadjuvant chemotherapy is also a significant predictor of LNY <15. This study suggests that neoadjuvant therapy of any type may impact LNY; however, further studies are needed to validate these findings and evaluate impact on survival. Introduction: Perioperative chemotherapy is the standard of care for most patients with resectable gastric adenocarcinoma. We hypothesized that patients with cT2N0 cancers undergoing upfront surgery and adjuvant therapy have similar outcomes to those receiving pre and/or perioperative chemotherapy. Methods: The National Cancer Database was used to identify all patients with cT2N0M0 invasive gastric adenocarcinoma who underwent radical gastrectomy. Patients were categorized into 3 treatment groups: 1) pre/perioperative chemotherapy, 2) postoperative chemotherapy and 3) postoperative chemoradiation. Overall survival was compared using the Kaplan-Meier method and multivariate Cox-regression analysis, controlling for relevant pretreatment and operative confounding factors. Results: 4002 patients met inclusion criteria. The mean age was 67.2 12.4 years, 2554 (64%) were male, 2168 (54%) had high-grade disease and 48% had appropriate lymphadenectomy ( 15 nodes). 61%, 20%, 15% and 3% had final pathologic stage I, II, III and IV disease, respectively. 482 (12%) of patients underwent pre/perioperative chemotherapy, 246 (6%) had postoperative chemotherapy, 585 (15%) underwent postoperative chemoradiation. There was no significant difference in age, comorbidity, tumor grade, and margin status among treatment groups. Patients undergoing preoperative treatment had lower final pathologic stage compared to the postoperative chemotherapy and chemoradiation groups (78%, 58% and 65% pathologic stage I disease, respectively, p<0.05). Patients undergoing postoperative chemoradiation had a lower rate of appropriate lymphadenectomy (55%) compared to preoperative (61%) and postoperative (64%) chemotherapy (p<0.05). Kaplan Meier analysis and multivariate Cox regression analysis controlling for age, comorbidity, grade and nodal harvest demonstrated no significant differences in overall survival in the 3 treatment groups. Conclusion: Patients with clinical T2N0 gastric adenocarcinoma who undergo upfront surgery with adjuvant therapy have similar overall survival compared to those who undergo preop/perioperative chemotherapy. Background: Nodal downstaging after preoperative therapy for gastric cancer has been shown to impart excellent prognosis, but this has not been validated in a national cohort. The role of preoperative chemoradiation (CXRT) in nodal downstaging remains unclear, when compared to chemotherapy alone. Furthermore, it is unknown whether the prognostic implications of nodal downstaging differ by preoperative regimen. Methods: Using the National Cancer Database, we compared OS duration among natural N0 (cN0/ypN0), downstaged N0 (cN+/ypN0), and node-positive (ypN+) gastric cancer patients. Factors associated with nodal downstaging were examined in a propensity score-matched cohort of cN+ patients, matched 1:1 by preoperative CXRT and chemotherapy alone. Results: Of 7426 patients, 58.2% received preoperative CXRT. 1858 patients were natural N0 (25.4%), 1813 were downstaged N0 (24.4%), and 3755 were ypN+ (50.4%). The median OS durations of downstaged N0 (5.1 years) and natural N0 (5.6 years) patients were similar, and longer than that of ypN+ patients (2.1 years) (p < 0.001). In the matched cohort of cN+ patients, more recent diagnosis (2010-2015 versus 2004-2009 ) (OR 2.57, p < 0.001) and preoperative CXRT (OR 2.02, p < 0.001) were independently associated with nodal downstaging. The median OS duration of downstaged N0 patients was significantly shorter after CXRT (4.7 years) than after chemotherapy alone (8.9 years) (p=0.003). Conclusions: Downstaged N0 patients have the same prognosis as natural N0 patients. Nodal downstaging occurred more frequently after preoperative CXRT, however the survival implication of nodal downstaging after CXRT may not be the same as when achieved by chemotherapy alone. Background Phase III trials demonstrate improved short-term outcomes after laparoscopic gastrectomy for LAGC compared to open. We hypothesized that robotic D2 gastrectomy(RG) after NAC yields benefits across multiple outcome domains vs standard of care treatment for LAGC. Methods Single institution, interrupted time series comparing SOC+open gastrectomy (OG,2008 (OG, -2013 for LAGC(T 2-4 N any /T any N + ) vs universal neoadju-vant+RG(2013 neoadju-vant+RG( -2018 . Treatment burden was a composite of adverse events affecting:90 d postoperative resource utilization(length of stay, reoperation/ readmission), oncology(positive margins, <16 resected nodes), cumulative major morbidity(90 d complication index) and pain(narcotic consumption). Predictors were evaluated via multivariate modeling; overall survival was estimated by Kaplan-Meier/log-rank test. Learning curve analysis was done using CUSUM starting at RG case#1. Results After exclusions, 87 LAGC subjects were treated via SOC+OG(n=55) or NAC+RG(n=32) with equivalent baseline characteristics:demographics, BMI, comorbidity(Charlson), tumor size, and clinical AJCC staging aside from male sex(69% vs 44%OG,p=0.02). NAC increased from 35%OG to 100% in RG. All four domains of resource utilization, oncologic efficacy, morbidity, and pain(narcotic use) improved in the NAC+RG cohort. Treatment burden declined from 86% in SOC/OG to 56% after NAC+RG(p=0.003). Multivariable modeling demonstrated OR 0.23 for treatment burden in NAC+RG compared to SOC/OG(CI 0.07-0.72,p=0.012) whereas extent of resection(total/subtotal), tumor size, T stage, and sex had no effect. Differences in OG vs RG treatment burden persisted in subgroup analysis for NAC(n=51). No detrimental effect of NAC+RG was observed on 2-year overall survival(80%RG vs SOC60%, p=0.048). These RG outcomes were achieved in the earliest phase of the learning curve(case#1-32). Conclusion After NAC, robotic D2 gastrectomy was associated with decreased treatment burden relative to OG. Frequencies of unfavorable hospitalization, adverse oncological outcomes, major morbidity, and narcotic consumption all improved in this interrupted time series. Prospective trials are needed. Background: Because of the high likelihood of distant metastatic disease, it is generally recommended that patients with poorly differentiated gastroenteropancreatic neuroendocrine tumors (GEP-NET) not be offered an operation. However, some patients with no evidence of distant metastatic disease will nonetheless undergo surgical resection and their outcomes are not known. Objective: To determine whether surgery confers any survival advantage over systemic therapy alone in poorly differentiated GEP-NET who are clinically 'M0' (i.e. non-disseminated). Methods: A retrospective cohort study of patients in the National Cancer Database (NCDB) with poorly differentiated (WHO Grade 3) GEP-NET without clinical evidence of distant metastasis (M0) between 2004 and 2014. Patients who underwent surgery were compared against those treated by systemic therapy. We performed an adjusted Coxproportional hazards model to assess the risk of death by treatment. Results: Of a total of 2,245 patients (mean age 64 14.3 years, 44.8% female, 20.6% pancreatic, 79.4% gastrointestinal), 1,549 (69%) were treated with surgical resection while 696 (31%) were treated with either systemic therapy or palliative measures alone. Median survival was 31 months after surgery versus 9 months after non-operative therapy (log-rank test, p<0.001). Rates of 5-year overall survival were 43.0% after surgery while only 17.6% were alive in the non-operative group at 5 years. Adjusting for age, sex, comorbidity status, and primary tumor location, patients treated with surgery had a 60.2% lower likelihood of death compared with non-operative therapy (HR: 0.40, p<0.001). Restricting to those patients who were found to have no distant metastasis intraoperatively (i.e. pathologically 'M0'), 5-year survival after surgery reached 50.2%. Conclusions: While patients with poorly differentiated GEP-NET have a poor prognosis, among patients with no evidence of disseminated metastatic disease at presentation (clinically 'M0'), resection confers significant improvements in long-term survival. Future guidelines might reflect this survival advantage. Background For unresectable small bowel neuroendocrine liver metastasis (SB-NELM), NCCN guidelines recommend interventional treatments (IVTs), including debulking liver resection, when disease progresses under somatostatin analog therapy. We evaluated the significance of IVT prior to liver-progression for patients at high risk. Methods This singleinstitution, retrospective study included 127 patients who were diagnosed as SB-NELM between 2007 and 2018. IVTs were defined as liver resection, ablation, intraarterial therapy, or peptide receptor radionuclide therapy. Coxproportional hazard regression analysis was used to identify prognostic factors for liver-progression-free survival (LPFS). Overall survival (OS) from the date of diagnosis was compared stratified by early-IVT prior to liver-progression, late-IVT after liver-progression, and non-IVT among patients at high risk for liver-progression. Results The median LPFS time was 20.1 months. The presence of a large mesenteric mass > 2 cm [LMM, hazard ratio (HR): 2.55, 95% confidence interval (CI): 1.23-5.85], peritoneal metastasis (HR: 2.19, 95% CI: 1.03-4.54), and G2 (HR: 2.21, 95% CI: 1.10-4.53) were negative prognostic factors for LPFS. The median LPFS times of patients with LMM and those without LMM were 16.2 and 33.3 months, respectively. Among patients with LMM, 32 patients underwent early-IVT, 17 underwent late-IVT, and 33 did not undergo IVT. Proportions of patients who underwent liver resection differed significantly between the early-IVT and late-IVT groups (59% and 24%, respectively, P = 0.015). OS of the early-IVT group was significantly better than the late-IVT and non-IVT groups (5-year OS rates of 88.6%, 55.9%, and 38.8%, respectively). After adjusted by liver resection, peritoneal metastasis, and grade, early-IVT was associated with significantly reduced risk of death against non-IVT (HR: 0.14, 95% CI: 0.02-0.71) among patients with LMM, whereas late-IVT was not (HR: 0.50, 95% CI: 0.13-1.52). Conclusions Early IVT prior to liver-progression might benefit patients with LMM > 2 cm, who are at high risk for liver-progression. Introduction: Lymph node (LN) dissection for small pancreatic neuroendocrine tumors (PNETs) has been heavily debated over the years. We aimed to investigate the role of LN dissection for small PNETs in a large single-institution retrospective study. Methods: Clinicopathologic data of patients with PNET operated on at our institution between 1992-2019 were collected. Analyses were restricted to PNETs measuring up to 3cm on imaging. Survival was compared between patients with positive, negative and not resected LNs. Results: A total of 371 patients underwent a PNET resection, of which 225 (61%) had tumors </=3cm on imaging. Of the latter, 46% were female; median age was 56y (interquartile range [IQR] 48-62); 79% of the tumors were non-functional, 71% were of WHO grade 1, and their median size was 1.9cm (IQR 1.3-2.4) and 1.9cm (IQR 1.4-2.5) on imaging and pathology, respectively. The majority were located at the pancreatic tail (43%). LNs were negative in 138 (61%), positive in 36 (16%) and not resected in 51 (23%) patients. Patients who did not undergo LN resection had smaller tumors (median size 1.5cm, IQR 1.1-1.9 vs 2.1cm, IQR 1.5-2.7; p<0.001) of lower WHO grade (WHO grade 1 87% vs 67%, p=0.03), which were less frequently located at the head (10% vs 32%, p<0.001). LNs were not resected in 56% vs 11% (p<0.001) of patients who underwent a pancreas-preserving procedure (enucleation or middle pancreatectomy) vs a classic resection, respectively. Among patients who underwent LN resection, LN positivity was associated with poor OS (mean OS 144 vs 212 months, p=0.034) and DFS (mean DFS 117 vs 210 months, p<0.001). There was no difference in OS or DFS between patients who had negative LNs and those who did not undergo LN resection (mean OS: 191 vs 212 months, p=0.496; mean DFS: 186 vs 211 months, p=0.632). Similar results were obtained when the analyses were restricted to patients with non-functional tumors. Conclusion: Lymph node dissection in patients with PNETs </=3cm was not associated with a survival benefit. These results indicate that less extensive approaches could be followed for patients with small PNETs.

Kaplan-Meier overall survival curves. Negative lymph nodes vs no lymph node resection among patients with PNETs measuring up to 3cm on imaging. P-values were derived from log-rank tests.

Signature with Human Relevance and Potential Novel Therapeutic Avenues C. Dudgeon, 1 C. Harris, 1 A. Casabianca, 5 * Y. Chen, 1 A. Sharma, 1 M. Shah, 2 A. Roberts, 1 E. Collisson, 3 V. Balachandran, 4 D.R. Carpizo. 1 1. Rutgers Cancer Institute of New Jersey, New Brunswick, NY; 2. Dept. of Surgical Oncology, Emory University, Atlanta, GA; 3. Dept of Medical Oncology, University of California at San Francisco School of Medicine, San Fransisco, CA; 4. Dept. of Surgery, Memorial Sloan Kettering Cancer Center, New York, NY; 5. Dept of General Surgery, Rutgers Robert Wood Johnson Medical School, New Brunswick, NJ. Pancreatic cancer (PC) is a highly lethal malignancy with a proclivity for early metastatic dissemination. While surgical resection is a treatment option for localized cancers, the majority of these patients (>80%) recur indicating the presence of disseminated tumors cells (DTCs) at the time of pancreatectomy. These DTCs are referred to as minimal residual disease (MRD). While most patients recur early, a subset recur late which is due to the reactivation of dormant tumor cells. The mechanisms that govern cancer dormancy are poorly understood in part because of a lack of animal models that reflect the biology of the resected patient. Here we introduce a novel model of PC dormancy that mimics outcomes in resected patients. Using single cell transcriptomics and an assay for chromatin accessibility, we found dormancy is characterized by a distinct cellular state from pre or post-dormant cancer cells. Mechanisms of dormancy include upregulation of the transcriptional repressor Dec2, which functionally drives quiescence, and monoallelic suppression of mutant KRAS by DNA methylation. Pathway analysis of dormant tumor cell transcriptomics indicated an upregulation of enzymes involved in linoleic acid metabolism. Linoleic acid treatment of non-dormant tumor cells inhibits proliferation and induces Dec2. Dormant PC cells increase expression of cancer stem cell markers and exhibit features of chemoresistance. We found that cancer dormancy is characterized by large scale transcriptomic changes and global chromatin remodeling which substantiate the conclusion that dormancy is driven largely by epigenetic mechanisms. We identified a dormancy gene signature distinct from non-dormant PC cells and found that this correlates with human resected patients with long term survivals, and inversely in patients with short term survivals. We developed methods for isolating DTCs from the livers of early stage PC patients undergoing resection and found that the murine dormancy signature correlated with the gene expression of the liver DTCs and not primary tumor cells. These results indicate that this dormancy signature has human relevance.

After Pancreaticoduodenectomy: A 10-year Experience from a Single Academic Center N. ARKADOPOULOS,* G. Gemenetzis, P. KOKOROPOULOS, P. VASSILIU, V. SMYRNIOTIS. 4th DEPT OF SURGERY, NATIONAL AND KAPODISTRIAN UNIVERSITY OF ATHENS, Chaidari, Greece.

Introduction: Despite advances in surgical technique and perioperative care, pancreaticoduodenectomy remains a challenging procedure with relatively high morbidity. Various surgical variations in anastomotic techniques have been used in attempts to reduce the rate of complications such as pancreatic fistula and delayed gastric emptying. Currently, gastrointestinal tract reconstruction (pancreaticojejunostomy, hepaticojejunostomy, and gastrojejunostomy) in a single loop is the most frequently utilized approach in high-volume centers. In this study we present a single surgical team 10-year long experience with an uncommon approach to alimentary tract reconstruction that was primarily devised in order to reduce pancreatic fistula rates. Methods: Our approach following a standard pylorous-preserving pancreaticoduodenectomy includes-as depicted in the figure-an end to side, duct to mucosa pancreatico-jejunostomy (A) and an end to side gastro-jejunostomy (B) on the short limp of a Roux-en-Y reconstuction (C) while our hepatico-jejunostomy (D) is placed on the long limp. The theory behind this approach is that activation of pancreatic enzymes that leads to anastomotic disruption, is likely to be avoided by the placement of the pancreatic anastomosis far from the biliary drainage. Results: Between January 2009 and December 2018 our anastomotic technique was used in 170 patients with malignancy, 101 male/69 female, mean age 64 years. A standard perioperative care protocol was used. Mean operative time was 208 min and mean intraoperative blood loss was 350 mL. Mean tumor size was 3,4 cm and mean lymph node harvest was 14 nodes. Overall morbidity was 26%. Pancreatic fistula rate was 9,4% (4,1% grade A, 3,5% grade B and 1,8% grade C). Delayed gastric emptying was recorded in 6% of cases. Mean hospital stay was 10 days and 90-day mortality was 2,3%. Conclusion: The present series is probably one of the largest with the described anastomotic technique which appears to be associated with considerable reduction of two common postoperative complications, i.e pancreatic fistula and gastric delay. I The Multicenter Selective Lymphadenectomy Trial (MSLT-II) demonstrated no difference in survival when comparing ultrasonographic surveillance (US) to completion lymph-node dissection (CLND) for node positive melanoma (SLN+). However, this was contingent on a structured surveillance schedule with experienced ultrasonographers and radiologists. We examine US adherence for SLN+ melanoma patients. M A retrospective review (6/2017-9/2019) was performed for SLN+ patients from three different hospitals. Patients who underwent sonographic nodal surveillance, had six months elapse since SLN biopsy, and had initial negative whole-body staging imaging were included. Requisite surveillance ultrasound (RSU) adherence (every four months for the first two years per MSLT-II protocol), ultrasound reports and patterns of recurrence are described. R Fifty-nine SLN+ patients (26 CLND and 33 US) were treated over the study period. Of 21 US patients who met inclusion criteria, mean age was 63, 43% were female, and primary location was extremity in 48%. Median Breslow depth was 2.5mm, 33% were ulcerated, and none had extracapsular extension. Median follow-up was 13 months. Of the 63 recommended RSUs for the 21 US patients, 35 were actually performed (56%). Of those performed, 26% were more than 30 days late. Eight patients (38%) had no US performed while 8 had all RSUs completed. Eleven of the 28 missing RSUs were substituted with cross-axial imaging. An average of 2.4 out of 5 radiographic descriptors (shape, size, hypoechoic appearance, hilum, cortical vascularity) were reported upon review of all 35 RSUs. Five patients recurred (one local, one regional, one distant, and two both regional and distant) with a median time to recurrence of 7.9 months. A PD-1 inhibitor was administered to 76% of patients. Of the 13 patients that did not undergo all recommended RSUs, factors were patient-related (n=5), provider-related (n=5) or unknown (n=3). C RSU adherence of SLN+ patients was suboptimal and demonstrates the need for appropriate patient selection, education, and systems for follow-up. Standardization of radiographic reporting may further improve MSLT-II RSU protocol implementation. Background In the post MSLT II era of melanoma care most patients are not receiving completion lymph node dissection (CLND). The objective here is to evaluate the role of CLND in a high risk patient population of thick (>3.5 mm thick) melanoma as this group of patients has not been well studied. Methods We utilized the prospectively collected databases from the multicenter selective lymph node dissection I and II melanoma trials to determine the outcome of patients with thick (>3.5 mm) primary melanoma who underwent CLND or observation alone following identification of a tumor-positive sentinel lymph node biopsy (SNB) from either of these two trials. Results 487 patients were identified from the two databases with melanomas >3.5mm thick. About half of the patients had a CLND (52.0%), the remainder were observed. There was no difference in gender (p=0.13), age (p=0.23), primary thickness (p=0.44), or site (p=0.93) between the two treatment groups. Those undergoing CLND were more likely to present with distant recurrence versus local recurrence (p<0.001). 7-year melanoma specific survival (MSS) was 56.0%+/-3.8 for observed patients and 49.4%+/-3.5 after CLND. On multivariate analysis increasing age (>60 years), increasing primary tumor thickness, and ulceration, and all were associated with worsening DFS and MSS. Lymphocytic infiltration at the primary and regression were also associated with worse OS. Conclusion: CLND does not lead to an improvement in MSS or OS in the subset of thick melanomas, and there is a trend toward long term decrease in survival following CLND. Features of the primary tumor may be more indicative of survival and not the extent of surgery. Introduction: Adjuvant radiation therapy (RT) has been shown to decrease lymph node basin (LNB) recurrences in patients with clinically-evident melanoma lymph node (cLN) metastases, but its role in the era of immune checkpoint inhibitors and BRAF/MEK-targeted therapies is less clear. We sought to compare recurrence and survival outcomes in patients with resected cLN who received adjuvant RT and/or modern systemic therapies (ST; immune checkpoint inhibitors or targeted therapies). Methods: Patients at two academic centers who underwent complete lymph node dissection (1/1/2005-6/30/2018) for cLN and either received adjuvant RT or met the indication, but received only adjuvant ST, were identified. Dates preceding the modern ST era were included to capture patients receiving only adjuvant RT. The primary outcome was the cumulative incidence of LNB recurrences estimated using the Kaplan-Meier method. Secondary outcomes included in-transit/distant recurrences, disease-free survival (DFS), and melanoma-specific survival (MSS). Results: Of 116 patients, 51 (44%) received ST, 50 (43%) received RT, and 15 (13%) received ST and RT. Patients who received RT or ST/RT were less likely than those receiving ST only to have inguinal compared to axillary or head/neck involvement (inguinal: 37% ST, 14% RT, 7% ST/RT, P=0.013), but more likely to have extracapsular extension (47% ST, 68% RT, 80% ST/RT, P=0.028). The median (interquartile range) follow-up time was 24.5 (13.9-43.5) months. The 3-year LNB recurrence rates (95% confidence intervals) differed significantly by treatment group and were 25% (11-37%), 14% (3-25%) , and 0% in the ST, RT, and ST/RT groups, respectively (log-rank P=0.047). Comparing specifically the ST and ST/RT groups, the incidence was significantly lower among ST/RT patients (log-rank P=0.045). The treatment groups did not differ in in-transit/distant recurrences (log-rank P=0.29), DFS (log-rank P=0.49), or MSS (log-rank P=0.75). Conclusion: Adjuvant RT may still have a valuable role in the era of modern ST in reducing LNB recurrences. Further research is needed to identify patient populations most likely to benefit from the addition of adjuvant RT to modern ST. Background: The Multicenter Selective Lymphadenectomy Trial II (MSLT-II) demonstrated equivalent melanoma-specific survival in patients with sentinel lymph node biopsy positive (SLNB+) melanoma undergoing either active surveillance (AS) or completion lymphadenectomy (CLND). Since its publication, the adoption and implementation of AS has varied. We evaluated patient and disease-specific determinants of CLND in the MSLT-II era. Methods: Patients with SLNB+ cutaneous melanoma between 2016 and 2019 were identified. Patient and disease-specific factors were collected and analyzed for association with CLND. Factors significant on univariate analysis or with clinical significance posited to influence decision-making for CLND were included in multivariate analysis. Results:A total of 236 SLNB+ patients met inclusion criteria. Rates of CLND decreased from 68% of SLNB+ patients in the year preceding MSLT-II, 34% in the year following publication of MSLT-II and 16% in the second year following MSLT-II. Increasing BMI, increasing size of nodal metastasis and head and neck (HN) primary melanoma were associated with CLND; advancing age was associated with decreased likelihood of CLND. Patients with a BMI>30 were 37% more likely to have CLND compared to patients with normal BMI. (OR: 1.37, p<0.05). Patients with HN primaries were more than 3 times as likely to receive CLND (OR 3.31, p<0.01) compared to primaries of the lower extremity. For every 1 mm increase in largest nodal tumor, patients were 11% more likely to undergo CLND (OR: 1.11, p<0.01). For every one-year increase in age there was a 2% lower likelihood of CLND (OR: 0.98, p<0.05). Conclusions: Factors predicting maximum benefit from CLND have been proposed and include nodal burden and primary tumor features, but there remains significant variability in implementation of AS since MSLT-II. We identified lower adoption of AS in patients with HN primaries, potentially due to under-representation of HN patients in the original trial, lower morbidity from elective CLND and greater morbidity of lost local control compared to trunk and extremity nodal basins. The increased use of CLND in obese patients is an area of interest that requires additional investigation. Background Multiple randomized controlled trials in melanoma demonstrated improved survival with the use of immune checkpoint inhibitors (CPI) in patients with Stage IV disease. This study looks at the response rate and durability of response in patients with advanced Stage III melanoma. Methods A query of the prospectively maintained melanoma database identified 175 patients with Stage IIIb to IIId melanoma (AJCC Version 8), who also received CPI over the time-period of 2008 to 2018. Response rate was determined by retrospective review of of radiographic scans, after first cycle of CPI. Progression free survival (PFS) and overall survival (OS) were estimated from the start of CPI to date of progression or death. Results Primary characteristics are in Table 1 . The majority of patients (67%) were initially treated with surgery alone for melanoma, and then presented with local/regional recurrence, without distant metastasis. The remaining 33% of patients presented for initial treatment with Stage III disease. Of the 175 patients in the cohort, most were stage IIIC (68%, n=120) or IIID (18%, n=37) at initiation of CPI. Treatment consisted of single agent CTLA-4 blockade 43% (n=76), PD-1 blockade 37% (n=65), or combination 19% (n=34). Response (complete (CR) or partial (PR)) after the first cycle of treatment occurred in 42% (95% CI: 34.9, 50) of patients, of which 18% (n=31) had a CR on imaging. Median follow-up among survivors was 41 months, with a median OS of 87 months (95% CI: 46.8, Not Reached). Progression free survival was 7 months. Among 79% (n=138) of patients who progressed, the initial site of progression was locoregional in 65%, and systemic in 35%. At last follow-up, 34% (n=60) of patients were NED, 42% (n=25) of whom had an initial CR on imaging. After CPI, surgery was performed in 50 patients (28% of all patients) and 82 patients (47% of all patients) received additional therapies, including targeted therapy, additional immunotherapy, clinical trial, and radiation. Conclusion As clinicians continue to be faced with the decision of whether to operate or give CPI in recurrent Stage III disease, this study provides real world data on the response rate and durability of responses.

Introduction After a sentinel lymph node biopsy (SLNB), nodal disease can be detected via immediate completion lymphadenectomy (iCLND) or at the time of nodal recurrence. Lymphadenectomy may be performed in a delayed fashion after node-only recurrence in patients undergoing observation of a positive SLNB (dCLND) or for node-only recurrence following a false negative SLNB (fnCLND). We sought to determine the prognosis associated with these three distinct presentations. Methods A prospectively collected singleinstitution database was retrospectively queried to identify iCLND, dCLND, and fnCLND patients from 1995-2018. The primary endpoint of interest was melanoma-specific survival (MSS) from the time of SLNB. Cohorts were compared using univariate statistics and multivariate cox regression modeling. Results A total of 49 iCLND patients, 39 dCLND patients, and 57 fnCLND patients were identified. Across all cohorts, the median age was 59, the median Breslow thickness was 2.8mm, and 46% of patients had ulcerated melanomas. There were no significant differences identified in the clinicopathologic characteristics of the three cohorts. With a median follow up of 39 months, MSS in patients with iCLND was 51% (95% CI 35-65%), dCLND was 75% (95% CI 51-89%), and fnCLND was 65% (95% CI 47-78%, p=0.008). Figure 1 . After multivariate adjustment, fnCLND (HR=2.8, p=0.03) and iCLND (HR=4.1, p=0.002) were significantly associated with decreased MSS compared to patients with dCLND. Ulceration (HR=3.02, p=0.001) was the only other prognostic factor identified as significantly associated with MSS in this cohort. Conclusions Selection for lymphadenectomy by nodal observation (dCLND) is associated with a substantially improved survival compared to iCLND in patients found to have a positive non-SLN. Despite a period of observation to select for "nodotrophic" biology, fnCLND is associated with a significantly worse survival than dCLND. This may be consistent with a distinct biologic behavior associated with false-negative SLN biopsies. Introduction: While there is evidence of the benefits of minimally invasive surgery (MIS) for colorectal cancer (CRC) resection, data specific to older adults (OA) lacks. Functional outcomes central to decision-making by OA are not described beyond 1 year after surgery. We compared long-term healthcare dependency outcomes of OA undergoing MIS and open CRC resection. Methods: A population-based analysis of patients > 70 years old undergoing CRC resection between 2007-2017 was conducted by linking administrative datasets. Outcomes were receipt of homecare and institution-free survival (IFS), defined as <14 institution-days within one year, in the 5 years after surgery. We used time-to-event analyses accounting for competing risk of death. Homecare was analyzed as recurrent dichotomous outcome with Andersen-Gill multivariate models, and IFS with Kaplan-Meier methods and Cox multivariate models. Results were similar when stratifying by primary tumor site and presence of stoma, with higher homecare needs and lower IFS for stoma and rectal cancer site. Conclusions: MIS was used in less than half of OA undergoing CRC resection. Compared to open surgery, it was associated with lower homecare needs, indicating reduced long-term functional dependence. Superior IFS with MIS indicated increased number of days alive and at home which is an established important patient-centred endpoint reflecting overall long-term treatment burden on mortality and morbidity. Neoadjuvant chemoradiation treatment strategies in rectal cancer have led to significant improvements in both oncologic outcomes and sphincter preservation rates; however, the morbidity associated with proctectomy remains a challenge, particularly in elderly patients. Pathologic complete response (pCR) rates after neoadjuvant treatment range from 10-25%. In patients who achieve pCR, non-operative approaches, coined "Watch-and-Wait," are gaining momentum as a treatment option with acceptable salvage rates. Although geriatric patients often have worse perioperative outcomes, there is a paucity of data on the impact of age and perioperative outcomes specific to proctectomy. We sought to evaluate the impact of age on surgical outcomes after proctectomy using the American College of Surgeons National Surgical Quality Improvement Program (ACS-NSQIP) database. METHODS: Patients undergoing proctectomy from 2016-2017 were analyzed using the ACS-NSQIP database. Patients were grouped according to age greater or less than 75. Peri-operative and 30 day outcomes were then compared using univariable logistic regression, pairwise analysis, and linear regression, adjusting for key covariates. RESULTS: All patients undergoing protectomy (n = 8648) were included and divided into age categories less than 75 years (n = 7365) or greater than 75 years (n = 1283).). Mean operative time and gender were comparable between the two groups. Older patients were more likely to be nonindependent, and have pre-operative diabetes, hypertension, and COPD. These patients also had significantly more perioperative complications, including higher rates of pneumonia (2% vs. 3.9%, p < 0.0001), anastomotic leak (2.9% vs. 2.7%, p < 0.01), septic shock (1.8% vs. 4.3%, p <0.0001), blood transfusion(12.5 % vs. 18.6%, p< 0.0001), renal failure (0.4% vs. 1.0%, p < 0.01), and myocardial infarction (0.5 vs. 1.8, p < 0.0001). CONCLUSIONS: Patients over the age of 75 have more comorbidities and are at higher risk for perioperative complications after proctectomy. Older patients who have achieved pCR may benefit more from a "Watch-and-Wait" treatment approach compared with younger patients. Background: Although studies have identified demographic and clinical factors associated with quality colorectal cancer care, the association between patient-reported experience of care and quality of care is unknown. Our primary aim was to assess the relationship between patient-reported experience of care and receipt of guideline-concordant colon cancer (CC) treatment. Methods: Fee-For-Service Medicare beneficiaries with resected stage I-III CC (2003) (2004) (2005) (2006) (2007) (2008) (2009) (2010) (2011) (2012) (2013) were identified in the linked SEER registry and Consumer Assessment of Healthcare Providers and Systems patient experience survey (SEER-CAHPS) dataset. Patient-reported ratings were compared based on receipt of guideline concordant care [resection of 12 lymph nodes (LN) (stage I-III) and receipt of adjuvant chemotherapy (stage III)]. Linear regression was performed to compare mean patient experience scores by receipt of guideline concordant care, adjusting for patient and hospital factors. Results: 1010 patients with stage I-III CC were identified (mean age 76.7, SE 6.9). Of these, 58.4% of stage I (n=192/329) and 73.4% of stage II (n=298/406) patients underwent resection of 12 LN. Among stage III patients, 76.0% (n=209/275) underwent resection of 12 LN and 52.4% (n=144/275) received adjuvant chemotherapy. By multivariable analysis, patient-reported ratings of health care quality, personal and specialty physicians, customer service, physician communication, getting needed care, and getting care quickly were similar among patients who received guideline-concordant treatment compared to those who did not. However, mean ratings of overall health care quality [91.3 (SE 2.0) vs. 82.4 (SE 1.7), p=0.0004] and getting needed care [92.8 (SE 2.4) vs. 86.8 (SE 2.0), p=0.047] were higher among stage III patients who received guideline concordant care compared to those who did not. Conclusions: Patient-reported ratings of health care quality and ability to get needed care are associated with guideline concordant cancer care among elderly patients with stage III CC. Further investigation is needed to determine if patient-reported experience correlates with other clinical measures of quality of colorectal cancer care. Aim: Surgery represents the only curative option for most patients with colon cancer. However, despite the recommendations of their physicians, some patients refuse surgery. The purpose of this study is to identify the incidence, trends, risk factors associated with refusal of surgery in patients with colon cancer and estimate the impact on survival. Methods: A retrospective review of the National Cancer Data Base (NCDB) between 1998-2012 was performed. We identified 924,290 patients with colon cancer with potentially resectable tumors (pretreatment clinical stage I, II or III). Patients who underwent cancer surgery were compared with patients in whom surgery was recommended but refused. Multivariate models were employed to identify factors predicting failure to undergo surgery and assess the impact on survival. Results: Of those patients who were recommended surgery, 7,152 patients (0.8%) refused surgery for resectable colorectal cancer. On multivariable analysis, patients were more likely to refuse surgery if they were older [odds ratio (OR)=1.14; 95% confidence interval (CI) 1.14-1.15], female (OR=1.20; 95% CI 1.12-1.68), African American (vs White), OR=2.30; 95% CI 2.10-2.51), on Medicaid (vs private, OR=3.06; 95% CI 2.49-43.77), higher Charlson-Deyo score (2 vs 0, OR=1.85; 95% CI 1.70-2.01). Patients who were recommended surgery but refused had significantly worse survival for every stage compared to those who received surgery [median survival 6.8 vs 24 months, Cox hazard ratio (HR) 3.41; 95% CI 3.12-3.60] Conclusions: The percentage of patients refusing surgery for operable colorectal cancer remains a significant issue that affects survival. Refusal of surgery is independently associated with several variables including gender, race, and lack of insurance, raising the concern that socioeconomic factors may drive patients to forego potentially life-saving care. To mitigate national disparities in surgical care, future studies should focus on exploring potential reasons for refusal and developing communication interventions. BACKGROUND: Performance status (PS) is traditionally used to predict tolerance and morbidity associated with CRC treatment. Monitoring activity level during therapy using a wearable fitness tracker (Fitbit™) may provide a more accurate estimate of a patients overall PS and help predict treatment related toxicities. METHODS: With IRB approval, we prospectively enrolled CRC patients undergoing therapy into 2 cohorts, medical (M) and surgical (S). Primary aim was to assess feasibility of Fitbit™ to assess activity level and toxicity. After documenting baseline ECOG PS, M and S patients wore Fitbit™ for 4 days while receiving chemotherapy or prior to surgery, respectively. The surgical patients also wore the device for 4 days post discharge. Patients' mean steps per day (SPD) were calculated, excluding days Fitbit™ was worn <12 hours. To stratify prediction of toxicity risk, a cutoff of 5000 SPD was selected and any post-operative complication (S patients) or grade 3 toxicity (M patients) was counted as toxicity. The study met accrual of 80 patients. RESULTS: 80 patients were evaluated for the primary aim. Sixtyeight (85%) had at least 3 days with 12 hours of Fitbit™ usage, meeting the 75% feasibility endpoint. Seventy-six patients (37 M, 39 S) had at least 1 day with 12 hours of Fitbit™ usage, and had data available for analysis. Mean SPD for PS 0 and PS 1 was 6313 and 2925, respectively (p=0.0003). Twelve M patients and 9 S patients experienced toxicity (Table) . The rate of toxicity was 25% (13/52) in those with PS 0 and 33% (8/24) in PS 1. With SPD as cutoff, toxicity rate was 21% (7/33) in patients with >5000, compared to 32% (14/43) in patients with <5000. In S patients, there was a decrease in SPD by 3364 ( 3760, p <0.001) in the post-discharge setting. This was not associated with toxicity. CONCLUSION: We observed high rates of compliance with Fitbit™ use in CRC patients. We also saw that SPD serves as a useful identifier for toxicity and provides rationale to study SPD in conjunction with PS for risk stratification of patients undergoing therapy, and can possibly be incorporated into pre-and rehabilitation selection in high risk groups.

Introduction The development of tyrosine kinase inhibitors (TKI) serves as an inflection point in the treatment of gastrointestinal stromal tumours (GIST). However, its role in the neoadjuvant setting and the optimal duration of therapy remains poorly defined. As such, we aim to evaluate the impact of neoadjuvant TKI on oncological and functional outcomes in our cohort of patients with rectal GIST. Method A retrospective analysis of 36 consecutive patients who underwent treatment for rectal GIST at the National Cancer Centre Singapore from February 1996 to October 2017 was analysed. Surgical, recurrence and survival outcomes between the groups who underwent neoadjuvant therapy and those who underwent upfront surgery were compared. Result Neoadjuvant TKI was commonly offered due to upfront unresectability (55%) and when high morbidity surgery was expected (36%). Patients who received neoadjuvant treatment had significantly larger tumours (median size 7.1 vs. 6.0 cm, p = 0.04) and lower mitotic count (> 10 per 50 HPF, 14% vs. 70%, p = 0.03) when compared to the non-neoadjuvant group. With neoadjuvant TKI (median duration 8.8 months), the majority of patients (82%) achieved at least partial response to the therapy coupled with a significant downsizing effect of up to 39% (median size 7.1 to 3.6 cm), resulting in similar rates of sphincter-sparing surgery (75% vs. 76%, p = 0.94) when compared with the non-neoadjuvant group. In general, the neoadjuvant group had lower rates of local recurrence (0% vs. 69%, p = 0.04) and higher overall survival (7.4 vs. 5.7 years, p = 0.03) compared to the non-neoadjuvant group. Conclusion Neoadjuvant TKI has the benefit of downsizing rectal GIST. This opens up the option of sphincter-sparing procedure for a subset of patients who would have deemed unresectable or undergo radical excision right from the onset of the initial diagnosis. Neoadjuvant therapy may also improve local recurrence and improves overall survival rates. Introduction: Although malignant bowel obstruction (MBO) is often a terminal event, a subset of patients has long term survival and optimal multimodality care remains unclear. We sought to examine the impact of chemotherapy after the diagnosis of MBO, including in relation to medical or surgical management of MBO. Methods: Using Surveillance, Epidemiology, and End Results (SEER) Medicare, we retrospectively analyzed patients 65 years with a primary gastrointestinal or genitourinary cancer and metastatic disease at diagnosis who were hospitalized with a diagnosis of MBO from 2008 to 2012. We used multivariable Fine and Grey competing risk models and Cox proportional hazards analyses to determine factors associated with post-MBO chemotherapy and overall survival (OS). Results: Of the 2,983 MBO patients identified, 1,511 (50.7%) underwent surgery, while 1,472 (49.3%) were managed medically. There were significant differences in age, gender, medical comorbidities, preoperative ascites and sepsis, and cancer type among medical and surgical patients (p<0.05 all). Surgical patients had higher rates of colorectal cancer (60.6% vs 36.9%), while medical patients had higher rates of pancreatic (22.0% vs 11.1%), ovarian (16.6% vs 10.7%), and gastric cancer (7.3% vs 2.3%; p<0.0001 all). There were no differences in rates of post-MBO chemotherapy among surgical and medical patients in univariate (40.2% vs 38.1%, p=0.23) and multivariable analyses (SM HR 1.12, 95%CI 1.00-1.26, p=0.06). Surgery followed by chemotherapy (Figure 1 ) was associated with greater survival compared to medical management followed by chemotherapy (HR 1.40, 95%CI 1.24-1.59, p<0.0001), surgery alone (HR 2.97, 95%CI 2.65-3.34, p<0.0001) and medical management alone (HR 4.56, p<0.0001) . Subgroup analysis of the colorectal, pancreatic and ovarian patients showed similar results. Conclusion: Chemotherapy appears to play an integral role in maximizing oncologic outcome in select patients with stage IV cancer and MBO. Awareness of these data are critical to optimizing multimodality care for MBO patients. Introduction: Pancreaticoduodenectomy is associated with high morbidity and the complexity of the procedure increases with vascular resection. With the advancement in minimally invasive surgical techniques, laparoscopic and robotic pancreatic surgery is becoming more common. A growing body of evidence exists demonstrating the equivalence or benefit of minimally invasive pancreaticoduodenectomy (MIPD) when compared to open pancreaticoduodenectomy (OPD) regarding the short term outcomes and safety. The purpose of this study is to determine the postoperative outcomes of patients undergoing vascular resection with OPD or MIPD approach using a large, multicenter cohort. Methods: All patients undergoing elective pancreaticoduodenectomy, including OPD and MIPD (robotic and laparoscopic, including open-assisted and unplanned open conversion), with vascular resection in the ACS NSQIP database were included in the study. Patients were stratified into those who underwent OPD vascular resection and those who underwent MIPD vascular resection. Patient covariates and outcomes were compared using standard statistical methods. Results: 2233 patients underwent OPD with vascular resection (17.8%) and 149 patients underwent MIPD with vascular resection (13.8%). The mean OR time was significantly longer in patients undergoing MIPD and vascular resection compared to OPD and vascular resection (485. 4 13.30 vs. 427 152.90, p-value <0.0001) . We next compared postoperative complications and found that patients undergoing OPD vascular resection were more likely to have postoperative sepsis compared to MIPD vascular resection (10.17% vs. 4.70%, p-value= 0.0299). There was a trend towards a decrease in superficial surgical site infection, organ space infection and length of stay and an increase in 30-day readmission in the MIPD vascular resection group; however, these differences were not statistically significant. On multivariate analysis, postoperative sepsis was independently associated with OPD (p-value= 0.0446). Conclusion: MIPD with vascular resection is safe and feasible. Future studies are needed to determine the effect of MIPD with vascular resection on long term outcomes and survival.

Introduction Emergency readmissions following Cytoreductive Surgery and Hyperthermic Intraperitoneal Chemotherapy (CRS & HIPEC) pose a challenge to peritoneal surface malignancy centres worldwide. Rates of emergency readmissions (ER) range from 10 to 30%, incurring a significant burden to both patients and healthcare systems. With the increasing recognition and utilisation of CRS & HIPEC in the management of various peritoneal malignancies, it is important to address the factors that might predict ER, with the aim to reduce ER post CRS & HIPEC. Methodology Patients who underwent CRS & HIPEC at a single institution between January 2001 and January 2019 were identified from a prospectively maintained database. Emergency readmissions (ER) were defined as any unplanned hospital stay following CRS & HIPEC. Early readmissions (EER) occurred within 30 days of discharge while late readmissions (LER) occurred between 30 to 90 days. Results 475 patients underwent CRS & HIPEC during the study period. The rates of EER and LER were 12.6% (n =60) and 5% (n=25) respectively. The most common reasons for ER were gastrointestinal related complications (45%, n=38 -poor oral intake/acute kidney injury, high stoma output, adhesions related intestinal obstruction) and deep space infections (24%, n=20). Major post-operative morbidity, prior CRS & HIPEC surgery, and stoma creation were predictors of EER but not LER. Median time to adjuvant chemotherapy was 3.5 vs 1.7 months in the ER and no ER groups respectively. Conclusion Gastrointestinal complications and deep space infections are major culprits resulting in ERs post CRS & HIPEC. Effective strategies to reduce ER occurences & its accompanying socioeconomic implications should be targeted as these areas. Introduction: Community hospitals provide convenience and access and are increasingly offering complex minimally invasive (MIS) procedures. This study compares the quality and safety of MIS pancreatic surgery in the community compared to academic centers. Methods: 2010-2016 National Cancer Data Base queried for cancer patients who underwent MIS (laparoscopic or robotic) pancreatic resection, excluding those with metastasis. Surgeries performed at community centers compared to those performed at academic centers. Multivariate regression models constructed to evaluate association between hospital type and oncologic and short-term post-operative outcomes. Results: 3,294 patients undergoing MIS pancreaticoduodenectomy (PD) and 2,793 distal pancreatectomy (DP) at 314 community and 199 academic centers identified. 23.5% of PDs and 25.2% of DP performed in the community. Trends in Figure 1 . Patients in community more likely to be white (p=0.013), have a lower median income (p=0.0004), a Charlson score of > 1 (0.017), be Stage 1 (p=0.038), and travel a shorter distance (<0.001.) Median hospital volume/ year for community vs. academic centers was 1.14 vs. 5.57 and 0.97 vs. 5.22, for PD and DP, respectively. For PD, community centers more likely to have > 10 day LOS (41% vs. 32%, p<0.001), convert to open (27.5% vs. 22.4, p=0.003) and less likely to examine 12 nodes (58.9% vs. 79.4%, p=0.001.) For DP, community centers had higher unadjusted 90-day mortality (3.7% vs. 1.9%, p=0.019), positive margins (11.8 vs. 8.9%, p= 0.027), and were less likely to examine 12 nodes (33.7% vs. 50%, p = 0.001.) Upon adjustment, for PD, community hospital status remained significant predictor of LOS > 10 days (OR 1.43, p=0.001) , conversion to open (OR 1.28, p=0.04) and < 12 nodes examined (OR 0.53, p<0.001.) For DP, community status a significant predictor of positive margin (OR 1.59, p=0.016) and < 12 nodes examined (OR 0.67, p=0.001.) Conclusions: A quarter of all MIS pancreatic cancer surgery is performed in the community, yet 90% of community hospitals are in the bottom quartile for volume. Nonetheless, surgery in the community appears to be safe. However, oncologic outcomes may be inferior.

Trends in number of minimally invasive pancreatic cancer resections performed by academic vs. community centers over time (lines.) Proportion of total cases performed by community centers over time (bars.)

Dissemination of Robotic Surgical Approaches to Resection for Gastrointestinal Malignancy P.R. Varley, 1 * C. Schlegel, 1 S. Tohme, 1 C. Shen, 2 A. Tsung. 2 1. Surgical Oncology, UPMC, Pittsburgh, PA; 2. OSU, Columbus, OH. Introduction: Robotic-assisted surgery has rapidly increased in popularity among both patients and providers. The goal of this project was to assess the varying degrees of dissemination for robotic approaches in selected gastrointestinal resections and compare basic operative outcomes to other approaches. Methods: The National Cancer Data Base was queried for patients undergoing resections for esophageal, gastric, rectal and pancreatic adenocarcinoma from 2010-2015. A nonparametric test for trend was used to assess changes in surgical approach over time. Lymph node yield was evaluated using negative binomial regression, while 90-day mortality and margin positivity were assessed using logistic regression. Results: From 2010 to 2015 there were 25,067 gastrectomies, 12,317 esophagectomies, 71,285 rectal resections, and 22,687 pancreatectomies available for analysis. For each surgical procedure there was a significant trend toward increased utilization of both laparoscopic and robotic approaches (Figure 1 ). When analyzing factors related to surgical resection, robotic approach was associated with a significant increase total lymph node yield for gastrectomy (IRR 1.22, p <0.001), esophagectomy (IRR 1.20, p < 0.001), and rectal resection (IRR 1.07, p < 0.001) when compared to an open approach. Significant increase in lymph node yield was also identified in robotic approaches to gastrectomy and rectal resection when compared to laparoscopic approaches (p = 0.004 and p = 0.032, respectively). Robotic approaches were also associated with significant lower rate of margin positivity in esophagectomies (OR 0.53, p = 0.003) and rectal resections (OR 0.65, p < 0.001). Robotic approach was not associated with 90-day mortality in any of the four malignancies. Conclusions: There is evidence for continued increase in the use of robotic approaches to gastrointestinal resection for malignancy, though the pace of dissemination is variable between resection type. Improved lymphadenectomy and lower margin positivity for some resection types may be related to visualization and precision provided by robotic approaches. However, clear impact on cancer-specific survival has yet to be demonstrated. Background: Surgical resection is the cornerstone of treatment of gastrointestinal cancer, however, it entails substantial morbidity. To estimate this risk, it is crucial to use a tool based on objective parameters. Surgical Apgar Score (SAS) is a practical instrument of 3 intra-operative variables (estimated blood loss, lowest mean blood pressure and lowest heart rate) that provides instant feedback. The aim of this study was to evaluate the performance of the SAS in predicting complications at 30 days in patients with primary GI cancer undergoing curative surgery at a national reference center. Methods: A prospective observational study was conducted. The patients were classified as low ( 4) and high ( 5) SAS. Complications were defined as any event that met the criteria of the Clavien-Dindo system (II to V). Bivariate and multivariate analyzes were performed through Cox regression, considering significant a p value < 0.05. Results: One hundred and fifty patients with a mean age of 58.64 years were included. The most common cancer site was colorectal (46.0%). Post-operative morbidity was 50.0%. Eighty-six percent of patients were classified as ASA 3. In the multivariate analysis, body mass index (HR: 3.351, 95% CI: 1.218-9.217, p = 0.019), SAS (HR: 0.266, 95% CI: 0.077-0.922, p = 0.037), surgical time (HR: 3.170, 95% CI: 1.092-9.198, p = 0.034) and the requirement of ephedrine administration (HR: 0.356, 95% CI: 0.144-0.880, p = 0.025) demonstrated a significant association with the development of adverse outcomes. Conclusions: SAS is an independent predictive factor of post-operative morbidity at 30 days in the surgical management of GI cancer and, in a high-risk population with ASA 3, it seems to offer a reliable sub-stratification. Background: The use of neoadjuvant endocrine therapy (NET) is increasing. The goal of this study was to examine patient selection and local-regional treatment patterns in patients who received NET on and off clinical trials at the Dana Farber Cancer Institute (DFCI). Methods: Patients presenting with stage I-III estrogen receptor positive HER2 negative (ER+HER2-) breast cancer were identified from a prospectively maintained database. Detailed tumor and treatment characteristics were abstracted from the medical record. Results: From 12/2015-9/2018, 2191 patients presented with stage I-III ER+HER2breast cancer. 322 (15%) patients received neoadjuvant systemic therapy; 95 (4%) received NET. 38 (40%) NET patients were treated on a clinical trial (31 [81%] PELOPS, 6 [16%] ALTERNATE, 1 [3%] DFCI protocol). The median age of NET patients was 65yrs (range 37-95) and the median length of NET was 6.6 (2.4-11.7) months. Patients treated on-trial were significantly younger, more likely to have invasive lobular carcinoma, cN+ and multifocal disease. Grade, tumor size, and initial breast conserving surgery (BCT) eligibility were not statistically different in those treated on and off-trial [Table] . Overall, 21 (22%) patients were initial BCT candidates, and 56 (59%) patients had successful BCT after NET. Patients treated on-trial were less likely to achieve BCT than those treated off trial (45% vs 68%, p<0.01). 70 (74%) patients underwent axillary surgery (97% on trial vs 58% off trial, p<0.01). Adjuvant chemotherapy use was higher in patients treated on trial (47% vs 7%, p<0.01). Radiation use did not differ. At a median follow-up of 29.4 months (range 7.2-48.0), there have been no locoregional recurrences and 4 (4%) distant recurrences (2 on-trial, 2 off-trial, p=0.68) [Table] . Conclusions: In this modern cohort of patients with ER+HER2-breast cancer, 4% were selected for NET. Patients treated with NET off-trial were older, more likely to achieve BCT, and less likely to have axillary surgery or adjuvant chemotherapy. Despite differences in patient and treatment characteristics, short-term oncologic outcomes are similar between groups.

WBRT whole breast radiation therapy; PMRT post mastectomy radiation therapy; RNI regional nodal irradiation

Postoperative Pain Control C. Morin, 1 * K.E. Rojas, 1 M. Javid, 1 Y. Patel, 1 P. Flom, 2 C. Andaz, 1 D. Manasseh, 1 P.I. Borgen. 1 1. Maimonides Medical Center, Brooklyn, NY; 2. Peter Flom Statistical Consulting, New York City, NY. Background A previous pilot study demonstrated that the postoperative opioid prescription could be successfully eliminated for lumpectomy patients without inferior pain control. In the present study, we sought to demonstrate improved postoperative pain control in a larger group of patients undergoing lumpectomy who received multimodal analgesia without an opioid prescription at discharge. Methods A prospective cohort of patients undergoing lumpectomy with varied axillary management received an opioid-sparing ERAS protocol. In-hospital and discharge opioids were compared to a historical cohort that received usual care (non-ERAS) using oral morphine equivalents (OMEs). Postoperative day one and week one pain scores were compared using Kruskal-Wallis test. Results Between 2017 and 2019, 730 patients received the ERAS protocol and were compared to a similar historical control group of 624 patients who underwent surgery between 2015 and 2018. Of the ERAS patients, 634 underwent lumpectomy for both benign and malignant indications with varied axillary management. Average pain scores were significantly lower in the ERAS group when compared to the control group both the day after surgery (1.6 vs. 2.7 p <0.001) and the week after surgery (1.1 vs. 1.7 p<0.001). Furthermore, ERAS patients were significantly less likely to report severe pain (7-10 of 10-point scale) on the day after surgery when compared to non-ERAS patients (7.5% vs. 15.2% p=0.019). Patients in the ERAS group were discharged with a median of 0 OMEs (range 0-150), while the non-ERAS group were discharged with a median of 90 OMEs (range 0-360) (p<0.001). Conclusion Postoperative prescriptions play a major role in the opioid crisis. We found that implementation of an opioid-sparing ERAS protocol in patients undergoing lumpectomy resulted in significantly improved postoperative pain scores without the routine prescription of postoperative opioids. Surgeons can improve their own patients' outcomes while addressing the larger societal issue of the opioid crisis by adopting similar protocols that decrease the quantity of narcotics available for diversion within our patients' communities. Background: Preoperative multimodal interventions appear to mitigate the functional decline attributed to intra-abdominal surgery. However, there is limited data on prehabilitation for breast cancer patients. Patients who are undergoing neo-adjuvant chemotherapy (NAT) have a prolonged time period prior to surgery and could benefit from a prehabilitation strategy. Prior to initiating a full-scale trial assessing prehabilitation in a breast cancer population receiving NAT, it is important to assess feasibility and barriers to recruitment. Methods: Patients undergoing NAT for breast cancer were randomized to intervention or standard care at a single site. The prehabilitation protocol uses 5 core components: total body exercise, locoregional exercise pertinent to treatment-related deficits, nutritional optimization, stress reduction/psychosocial support, and smoking cessation. A full-scale trial would be considered feasible on the basis of this study if recruitment rate of 25% of eligible patients is achieved. Results: Between August 2018 and September 2019, 71 patients were approached for study enrollment. Initial consent to participate was obtained from 31 patients. 7 patients withdrew or were subsequently excluded from the study, for a final cohort of 24 patients (34%). Patients who participated in the study have a median age of 50.5 years (31-69), and 69% live < 30 km from the treatment site. Patients who declined to enroll have a median age of 51 years (32-75) and 64% live < 30 km from the treatment site. Baseline physical scores and demographics of enrolled patients are presented in Table 1 . Strategies such as combining study visits with medical appointments, offering options for follow-up strategies (in person or by telephone) and providing parking vouchers to patients facilitated recruitment. Conclusion: A full-scale randomized controlled prehabilitation study is feasible in our study population. Distance to hospital and patient age do not seem to be barriers to recruitment. Minimizing the burden of study involvement is important, to facilitate participation in study. 3 1. Spectrum Health General Surgery Residency Program, Grand Rapids, MI; 2. West Michigan Anesthesia, Grand Rapids, MI; Spectrum Health Medical Group, Grand Rapids, MI. Background: Recent attention has focused on regional methods of analgesia in an effort to reduce the use of narcotics and their associated complications; however, sparse high-quality comparative data exist regarding specific techniques. The goal of our study was to compare the efficacy of the paravertebral nerve block (PVB) versus pectoral nerve block (PEC) for patients undergoing breast surgery. Methods: The study was a single institution randomized control trial. Patients were randomized between PVB or PEC/serratus block preoperatively. Patients undergoing total or partial mastectomy, with or without reconstruction, were eligible. All patients received intravenous sedation only. The primary outcome measures included perioperative use of narcotic/sedative medications and a 2 week post-op pain rating. Secondary outcomes included length of stay, conversion to general anesthesia, and any surgical or block complications. Results: 101 patients participated in the study. 12 patients were excluded from analysis due to protocol violations. 89 patients were analyzed with 40 patients receiving PVB block and 49 receiving PEC/serratus block. The population had a mean age of 64 11 years. The most common procedure was partial mastectomy (85.5% of PEC vs 90% of PVB) with sentinel lymph node biopsy (87.8% PEC vs 92.3% PVB). There were no statistically significant differences among the two groups in regards to tumor histology, co-morbidities, pre-op diagnosis of chronic pain, or use of narcotics pre-op. There was no significant difference in conversion to general anesthesia (p = 0.62), use of extra sedative/narcotic medications (p = 0.57) or additional local anesthetic (p = 0.39). There was no difference in the total morphine equivalents used during surgery (p = 0.11) or PACU (p = 0.79), and no difference in the 2-week post-op pain rating (p = 0.11). Two adverse events were recorded in the PEC group including one surgical site infection and one post-operative hematoma, but no block-related complications. Conclusion: There was no statistically significant difference in the efficacy of the paravertebral block vs the pectoral nerve block in patients undergoing breast surgery. Introduction: Telemedicine is enhancing or replacing many aspects of traditional in-office healthcare. The role of telemedicine in the complex care of breast cancer (BC) patients remains unclear. Methods: Via a survey at their initial BC appointment, patients self-reported demographics such as age, race, distance from hospital and education level. Patients answered ten Likert questions about perceptions of telemedicine. Wilcoxon signed rank tests assessed whether overall the respondents reported more or less than neutral interest or concerns due to each of the ten considerations. Wilcoxon rank sum tests and Spearman's rank correlations were used for univariate analysis to assess if responses varied by patient characteristic. Multiple regression assessed for associations. Results: 51 female BC patients completed the survey. 31 patients were White, 19 were <60, and 17 lived over 20 miles from the hospital. 30 patients answered telemedicine could save time, 34 answered telemedicine could improve their access to care, and 35 answered telemedicine could decrease wait time, with Likert responses of "4"/Agree or "5"/ Strongly Agree. No significant associations were observed between interest in telemedicine (gauged by responses to the Likert score questions) and respondent characteristics. The questions that were associated with greater interest in telemedicine had to do with easier access, less waiting time and reduced exposure to infectious diseases. Each of these variables had a median Likert response of "4"/Agree with signed rank test p-values 0.010. The question which denoted the most concern about telemedicine involved the use of technology, although the median for that question was "3"/neutral. Finally, the overall median of the directionally aligned Likert responses was 3.4 partway between neutral and agree (indicating more interest/less concern). The signed rank test p-value for this average was 0.009. Conclusions: BC patients are interested in telemedicine. Since age, race, distance from hospital, and education level, are not predictors of interest in telemedicine, providers can consider offering these opportunities to everyone, as healthcare continues to become more personalized.

Introduction Improving patient safety and quality of care are priorities in health care and the responsibility of every physician. The study of malpractice cases represents a potential opportunity to identify area for quality improvement. While malpractice cases in breast surgery are often multifactorial given the multidisciplinary team approach and commonly related to delays in care, there is not much clarification or literature detailing the reasons of malpractice claims when surgeons are the primary responsible service. We sought to determine the factors associated with malpractice in breast cancer when surgeons were the primary responsible service. Methods Using the CRICO Comparative Benchmarking System, a national database representing approximately 30% of U.S. malpractice cases, closed claims between 2008-2019 involving breast cancer surgeries were reviewed. Only cases where a surgical oncologist, general surgeon, or plastic surgeon were identified as the primary responsible service were included in this analysis. Results A total of 174 malpractice cases were captured in the timeline reviewed. Total amount paid for closed malpractice cases was $5,347,316. Plastic surgeons were the most commonly named primary responsible service in 63.8% (111/174) of these cases, followed by general surgeons 30.5% (53/174), and surgical oncologists 5.7% (10/174). The most common allegations were errors in surgical treatment (152/174 or 84.4%), followed by diagnosis-related (10/174 or 5.7%) ( Table 1 ). The most common procedures in these malpractice cases were breast reconstruction (93/174 or 53.4%) followed by mastectomy procedures (33/174 or 19%) . Conclusion Claims related to mastectomy and reconstructive procedures were predominately responsible for malpractice among surgeons involved in breast cancer treatment. Gaining a better understanding of the malpractice cases involving surgeons in breast cancer care may provide practical insight to guide future initiatives aimed at improving patient outcomes. Introduction Carcinoembryonic antigen-related cell adhesion molecules (CEACAM) are cell surface proteins upregulated during organogenesis, as well as tumor progression and metastasis. CEA (CEACAM5) used as a tumor marker, is most well-known, and is a promising tumor-specific probe for fluorescence-guided surgery (FGS). Other CEACAM's (CEACAM1, CEACAM6) are upregulated in GI malignancies including pancreatic cancer. 6G5j is a novel anti-CEACAM antibody binding human CECAM 1, 3, 5, 6, 8. We evaluated 6G5j as a potential tumor-specific probe for FGS of pancreatic cancer. Materials/Methods 6G5j monoclonal antibody was conjugated to LICOR-IRDye800. Cells (5 x 10 6 ) or tumors fragments (2 mm 3 ) of human BxPC3 pancreatic cancer were injected intraperitoneally or surgically implanted into pancreatic tail of nude mice respectively (n=5). 6G5j-800 (50 ug) was injected intravenously 4 weeks after implantation. Mice were imaged with the following devices 48 hours after injection: LI-COR Pearl Trilogy, Curadel FLARE RP1, and Intuitive DaVinci Firefly laparoscope. Fluorescence intensity was quantified using the FLARE software. Mean fluorescence intensity (MFI) was measured at the tumor, liver, bowel and skin. The tumor-to-background ratio (TBR) was calculated. Results Images obtained 48 hours after injection of 6G5j-800 clearly labeled the tumor using both benchtop and clinical imaging devices ( Fig. 1 A-F) . A heat map image shows the highest intensity within the tumor. MFI is as follows: Tumor 447. 3, Liver 109.6, Bowel 86, Skin 82.3. At 48 hours, TBR was 5.43 (SEM= 0.76) . In the peritoneal-implant model, otherwise occult submillimeter gastric and abdominal wall deposits were detected with the Intuitive Firefly laparoscope. Conclusions The use of an anti-CEACAM antibody, conjugated to LICOR-IRDye800CW can selectively label pancreatic cancer in clinically relevant nude mouse models using laboratory and clinical imaging devices. Previous work has shown that an individual target of CEAM5/CEA is effective for FGS. The present work shows the potential and feasibility of targeting multiple CEACAM's for high resolution tumor labeling.

Fluorescence in-vivo imaging using 6G5j-800 in a BxPC3 human pancreatic cancer orthotopic and peritoneal xenograft nude mouse model. A tumor is clearly detectable on color overlay (A), black-and-white (B), and heat map (C) modes using the LICOR-Pearl imager and FLARE RP1. Even sub-millimeter deposits in a peritoneal metastasis model were detectable using 6G5j-800 with the Intuitive Firefly laparoscope (D-F). Intro: Pancreatic cancer comprises a complex system of interacting compartments including normal and neoplastic pancreatic cells, immune cells, and stroma. We have shown that deconvolution of the tumor transcriptome into cellular compartments defines tumor subtypes, informing prognosis and treatment response. Treatment options for pancreatic cancer are limited and outcomes poor, so identification of factors to guide developing therapies is paramount. We have developed a framework called DECODER to perform deconvolution and fraction weight estimation of cellular compartments in tumor samples. This study examines the association of tumor location with compartments defining the tumor microenvironment in pancreatic cancer. Methods: RNA-seq datasets of primary pancreatic cancer from The Cancer Genome Atlas (n=146) and the International Cancer Genome Consortium (n=65) were combined for meta-analysis. Using DECODER, fractions of previously identified tumor and microenvironment compartments were determined. Patient clinical and demographic variables were analyzed against compartment weights and ratios. Kaplan-Meier curves with log-rank test were used to assess survival. Results: Tumors of the pancreatic head compared to the body/tail were more likely to have a higher fraction of normal stroma (p=0.009) and immune compartments (p=0.023). Head tumors also had lower ratios of activated to normal stroma (p=0.019) and activated to immune (p=0.011). While compartment weights for normal stroma (p=0.043), immune (0.038), and activated/immune (0.039) were all significantly associated with survival, tumor location was not associated with survival (p=0.25). Conclusions: Pancreatic tumor location is associated with stromal composition and immune infiltration. Tumors in the head are associated with more favorable normal stroma and immune-high subtypes. Although survival was not significantly different in this cohort, our results may explain prior studies demonstrating tumor location in the pancreatic head associated with better survival. These results suggest that tumor location may have predictive value for emerging therapies targeting the immune system and tumor microenvironment. BACKGROUND: Multidisciplinary care has become the gold standard for the management of complex malignancy. An institutional culture of collaboration between different specialists may have profound implications for patient care. To better define this, we compared the multidisciplinary care delivered for pancreatic adenocarcinoma (PAC) at two major cancer centers within the same city. Cancer center A (CCA) is university-affiliated and within a private hospital network. It has a robust and well-attended weekly GI multidisciplinary conference for oncologists with remote video access. Center B (CCB) is a university-based program that serves as a major tertiary care institution for the region. Conversely, it has a weekly GI tumor board with variable degrees of participation by individual oncologists. Both centers refer to the same pancreatic surgeons. METHODS: Patient databases, from 2000-18 for CCA, and, 2009-19 for CCB, were queried for stage 1 and 2 (resectable) PAC. Patient demographic and treatment data were extracted. Rates of surgical referral and attempted resection were measured. Chi-square and T-tests were used to compare the institutions. RESULTS: 1,687 cases of resectable PAC from CCA and 845 from CCB were identified. CCA had a significantly greater percentage of patients discussed at multidisciplinary conferences and greater attendance. Over 19 years, there were only 16 cases of resectable PAC (0.1%) at CCA who elected to pursue treatment for their disease that did not have an evaluation for surgical therapy. In 10 years at CCB, 75 patients (8.9%) with resectable disease were identified who, despite receiving chemotherapy, never had an evaluation by a surgeon (p<0.001). All of these patients were managed principally by medical oncologists and were ultimately treated with definitive chemotherapy with or without radiation. CONCLUSION: An institutional culture of collaboration ensures that patients reach appropriate definitive care more consistently. Programs to increase the attendance and quality of multidisciplinary conferences are worthwhile endeavors that facilitate communication between specialists and can greatly impact patient care in complex malignancy.

Association of Surgical Pathology with Multimodal Therapy and Survival in Pancreatic Cancer E. Olecki, 1 * K. Stahl, 1 J.S. Peng, 2 M. Dixon, 2 N. Gusani, 2 C. Shen. 1 1. Division of Outcomes, Research, and Quality, Department of Surgery, Pennsylvania State College of Medicine, Hershey, PA; 2. Program for Liver, Pancreas, and Foregut Tumors, Department of Surgery, Pennsylvania State College of Medicine, Hershey, PA. Introduction: Current guidelines recommend multimodal therapy (MT) in the form of surgery and chemotherapy for early stage pancreatic cancer. Some providers, however, may be tempted to omit chemotherapy based upon favorable pathologic results. In this study, we investigate how final pathology after pancreatectomy is related to receipt of MT versus surgical resection only (SR) and its association with survival. Methods: We obtained data from the National Cancer Data Base to identify patients who underwent surgery with stage I/II pancreatic cancer. Chi-square and logistic regressions were used to determine differences between patients receiving MT vs SR. Kaplan-Meier survival curves were used among groups with favorable pathology (node negative, low grade, tumors <2cm, pathologic stage I) to compare survival of those who received MT vs SR. Results: A cohort of 22,924 was identified. Logistic regression found significant increase in the likelihood of receipt of MT in patients who were younger, had fewer commodities, higher incomes, more recent diagnosis, and received treatment in the Northeast (all p<.001). Those with pathology of stage II or greater were nearly twice as likely to receive MT versus stage I (OR 1.93 p<.001). Patients with positive lymph nodes were more likely to receive MT than those with negative lymph nodes (OR 1.25 p<.001). When survival of patients was compared by MT and SR among those with favorable pathology, there was significant improvement in median survival months of patients receiving MT with negative lymph nodes (MT 30.23, SR 22.18) , low grade (MT 25.56, SR 16.56) , tumor size <2 cm (MT 22.9, SR 13.14) , and stage I pathology (MT 35.98, SR 34.46 ) (all p values<0.01). Conclusion: Patients with favorable pathology of negative lymph nodes and stage I after pancreatectomy are less likely to receive MT. Undertreatment is associated with worse survival when comparing MT versus SR among groups with favorable pathology. These results suggest that pathologic results play a role in the decision to forego MT after pancreatectomy. This bias, in the form of decreased MT and deviation from current guidelines, appears to have an impact on survival.

The Utility of the Surgical Apgar Score in Pancreatic Cancer and Modification L. Zhou.* Sir Run Run Shaw Hospital, School of Medicine, Zhejiang University, Hangzhou, China. Objective The Surgical Apgar Score (SAS, a 10-point score calculated using limited intraoperative data) can predict the incidence of complications in different surgical fields. However, it is rarely studied in pancreatic cancer. The aim of the present study was to assess the predictive value of the SAS in pancreatic ductal adenocarcinoma (PDAC), and then propose a modified SAS which was more suitable for pancreatic cancer patients. Methods A prospective cohort of 160 PDAC patients was studied. The primary endpoint was 30-day major complication. The SAS was calculated as described. The overall discriminatory power of the score was analyzed using Receiver Operating Characteristic (ROC) curves and the Area Under the Curve (AUC) with respect to major complication or death. Results It showed a significant predictive value of SAS in major complications or death in PDAC (p = 0.020, AUC=0.606), especially in complication of pneumonia and pleural effusion (p = 0.022, p =0.023). In addition, the SAS exert significant predictive value in distal pancreatectomy (DP) group, but it has a weak predictive value for pancreaticoduodenectomy (PD) group. On multivariable analyses, occurrence of major postoperative complications was associated with lowest mean arterial pressure (MAP), estimated blood loss (EBL) and operative time (OT). Interestingly, as a characteristic of SAS, lowest heart rate (HR) was not involved. The modified SAS we proposed including MAP, EBL and OT increased AUC from 0.606 to 0.743. Conclusion SAS can be a simple, rapid scoring system that effectively predicts major postoperative complications. Besides, the modified SAS we proposed in this study, based on OT instead of HR, exert a better predictive value in PDAC patients. Background: In literature, percentages of complete pathologic response (pCR) in patients presenting with resectable (RES), borderline resectable (BLR) or locally advanced (LA) pancreatic cancer (PaC) after neoadjuvant treatment are variable, reaching 33%. Those data come mostly from retrospective reviews of single centers. The objective of this systematic review is to assess the incidence of pCR. Methods: Following the criteria of the PRISMA statement, a literature search was conducted looking for prospective papers focusing on neoadjuvant treatment in PaC. Retrospective papers, other than ductal carcinoma histologies and trials including metastatic or only resected patients were excluded from the present review. The quality of the available evidence was assessed using the Cochrane Collaboration's tool for assessing risk of bias. Data extraction was carried out by two independent investigators. The meta-analysis was carried out with ProMeta3 Software (Internovi, 2015) . PROSPERO rregistration ID: CRD42018095641. Results: The literature search of Embase, Cochrane and Medline with the terms "neoadjuvant OR preoperative" and "pancreatic OR pancreas" and "cancer OR adenocarcinoma OR tumor" led to the identification of 3128 papers. We restricted the search to humans, last 10 years and English language articles resulting in 1158 eligible articles to review. Extended papers revision led to the inclusion of 27 papers. Studies included 1129 patients, 71,51% were surgically explored, 62,44% resected. In more than half papers no pCR was observed, overall it ranged 0-11.11%, at the metanalysis 4% (95%CI 3-5%); in prospective studies 0-9.09% whereas in prospective databases 1.63-11.11%. Subgroups analysis suggets a better response in radiation treated patients and with mFOLFIRINOX regimens. Conclusions: Pathologic complete response in pancreatic cancer is actually infrequent: high quality studies provide a more reliable picture of neoadjuvant effects, high rates of pCR are reported in selected retrospective studies but it is over-estimated. Background: Sarcopenia has been associated with post-operative complications and length of stay (LOS) in patients undergoing esophagectomy. A variety of methods exist to measure muscle mass and strength, with few comparisons between methods. We compared hand-grip strength (HGS), muscle mass and intramuscular adipose tissue as predictors of post-operative outcomes. Methods: Patients with esophageal cancer undergoing esophagectomy were identified between January 2015 -June 2019 at Levine Cancer Institute. Skeletal muscle index (SMI) and skeletal muscle density (SMD), a measure of intramuscular adipose tissue, were derived from CT. HGS was measured using a dynamometer. Uni-and multivariable GLM analyses were performed Results: 115 patients (100 male, 15 female) underwent esophagectomy with an average age of 64.3 +/-9.8. The analysis was stratified by sex due to significant differences in HGS, SMI, and SMD. Among men, univariable analysis revealed a significant association between pre-operative HGS <25 kg and increased risk of post-operative pneumonia (p=0.02), ventilation >48hrs (p=0.02), LOS (p=0.002), discharge to home (p=0.001), and one-year mortality (p=0.005). All associations except discharge home remained significant in multivariable analyses (Table 1) . Among women, no factors analyzed were significantly associated with postoperative outcomes. Conclusion: HGS is a more powerful predictor of postoperative complications and LOS than either muscle mass or intramuscular adipose tissue among men undergoing esophagectomy. HGS is cost-effective and easily incorporated into routine clinical care, allowing for preoperative intervention to optimize patients for esophagectomy. To better understand the implications in women, additional research with a larger cohort is needed. Kukar, S.N. Hochwald. Roswell Park Comprehensive Cancer Center, Buffalo, NY. Introduction: Minimally invasive esophagectomy (MIE) has become the procedure of choice over open esophagectomy (OE). The National Surgical Quality Improvement Program (NSQIP) risk calculator does not enable prediction for MIE due to lack of current procedural terminology (CPT) codes for these procedures until recently. Furthermore, NSQIP is limited to 30-day follow-up while 90 days is now considered a more meaningful endpoint. We compared institutional MIE to NSQIP predicted OE outcomes. Methods: Patients who underwent MIE from June 2012 to April 2019 were included. Patient characteristics and 90-day postoperative outcomes were collected. The institutional 90-day outcomes were compared to NSQIP predicted 30-day outcomes using the equivalent CPT code for OE. Results: Ivor Lewis (IL) MIE was performed in 115 patients, McKeown in 83, and transhiatal (TH) in 20. Mean age was 62.2 8.5 years and body mass index was 28.2 5.2 kg/m2. Malignancy was present in 214 patients (98.2%), including 186 adenocarcinomas (86.9%). Preoperative stage was stage III in 155 patients (72.4%) and 189 patients (88.3%) received neoadjuvant therapy. For the abdominal portion, 204 (93.6%) were completed minimally invasively, with 5 planned laparotomies and 9 conversions (4.2%). Thoracic access was performed in 198 patients and 170 (85.9%) were performed minimally invasively, with 17 planned thoracotomies and 11 conversions (6.1%). The mean lymph node yield was 18.8 7.0 and negative margins were obtained in 204 patients (95.3%).

Postoperative outcomes compared to NSQIP predicted risk are detailed in Table 1 . Conclusions: Use of the NSQIP calculator for OE underestimates the risk of overall morbidity after MIE, due to restriction to 30-day outcomes and lack of esophagectomy specific complications. The 30-day OE risk calculator was a good predictor of 90-day MIE readmission, pneumonia and myocardial infarction rates, and length of stay, but it overestimated the risks for surgical site infection, reoperation, and mortality. A 90-day MIE specific risk calculator is needed for better predictions of postoperative morbidity and mortality.

Cancer J.S. Peng,* M. Brady, W. Ji, K. Attwood, S.N. Hochwald, M. Kukar. Roswell Park Comprehensive Cancer Center, Buffalo, NY. Introduction: Smoking is a known risk factor for malignancy and postoperative complications. Studies have also suggested that smoking is associated with worse oncologic outcomes for patients undergoing treatment for a number of cancers including pancreatic, prostate, and urothelial cancers. We evaluated the impact of smoking status for patients treated for esophageal and gastroesophageal junction (GEJ) cancers. Methods: Patients who underwent esophagectomy for distal esophageal and GEJ cancers between June 2012 and April 2019 were included. Univariate comparisons of baseline characteristics and outcomes were performed for former/current smokers versus never smokers. Multivariate propensity weighted analyses were performed for outcomes of disease-free survival (DFS) and overall survival (OS) adjusting for age, body mass index (BMI), American Society of Anesthesiologists (ASA) classification, and pathologic stage. Kaplan-Meier methods were used to compare survival. Results: A total of 210 patients were treated for malignancy with known smoking status; 155 patients were former/current smokers, including 19 (12.3%) active smokers at the time of surgery and 42 (27.1%) who had quit within 6 months prior to surgery. Univariate comparisons of baseline characteristics, operative, postoperative, and long-term outcomes are detailed in Table 1 .

Postoperative outcomes were similar except for a higher rate of stricture after 90 days in former/current smokers (13.5% vs 3.6%, p=0.043). Median OS for former/current smokers versus never smokers was 35.3 versus 57.0 months, and DFS was 22.2 versus 57.0 months, although this difference did not reach statistical significance. Conclusions: Never smokers experienced similar postoperative outcomes compared to former/current smokers, but had lower rates of anastomotic strictures long-term. The cause of this difference should be investigated further but may represent subclinical leaks in the postoperative period. In addition, although there were trends toward better median DFS and OS in never smokers, the difference did not reach statistical significance and requires evaluation in a larger cohort of patients.

Risk Factors for Postoperative Chylothorax After Radical Subtotal Esophagectomy K. FUJISAWA,* Y. Ohkura, M. Ueno, H. Udagawa. GI surgery, Toranomon Hospital, Tokyo, Japan. Background Chylothorax is one of the complications of esophagectomy for esophageal cancer. The treatment of this condition has been well discussed, but the risk factors for postoperative chylothorax remain unclear. Methods A retrospective review of 294 patients who underwent esophagectomy for esophageal cancer was conducted. These were patients with squamous cell carcinoma or adenocarcinoma of the esophagus including Siewert type I tumor of the esophagogastric junction who underwent subtotal esophagectomy with two-field or three-field lymphadenectomy. Of these, 24 patients who were diagnosed with chylothorax as a postoperative complication were allocated to the chylothorax group and the other 270 patients were allocated to the nonchylothorax group. Results Univariate analysis showed a significant difference in 3 factors: resection of thoracic duct, post-chemoradiotherapy, and high intraoperative fluid balance. Multivariate analysis revealed that post-chemoradiotherapy (hazard ratio [HR] = 3.430; 95% confidence interval [CI]: 1.364-8.625 ) and high intraoperative fluid balance (HR = 1.569; 95% CI: 1.2.7-2.039) were all independent factors predicting chylothorax. In addition, resection of the thoracic duct may be a predictor of chylothorax after esophagectomy (HR = 3.389; 95% CI: 0.941-12.201, p=0.062). Receiver operating characteristic curve analysis of intraoperative fluid revealed that the sensitivity was 62.5%, specificity was 74.1%, and the cutoff value was 6.55 mL/kg/h. Conclusions This study revealed that post-chemoradiotherapy and high intraoperative fluid balance are the predictors of chylothorax after esophagectomy. The elucidation of clinicopathological factors that can predict the incidence of chylothorax will help establish more effective perioperative management for esophageal cancer patients.

Length of Gastric Extension of Gastroesophageal Junction Adenocarcinoma and Incidence of Peritoneal Failure After Resection N. Paez Arango, 1 * K.G. Mitchell, 2 M. Blum, 3 J. Ajani, 3 P. Das, 4 B. Minsky, 4 J. Estrella, 5 P. Mansfield, 1 B. Badgwell, 1 W.L. Hofstetter, 2 N. Ikoma. 1 1. Dept Surgical Oncology, MD Anderson Cancer Center, Houston, TX; 2. Dept Thoracic and Cardiovascular Surgery, MD Anderson Cancer Center, Houston, TX; 3. Dept Gastrointestinal Medical Oncology, MD Anderson Cancer Center, Houston, TX; 4. Dept Radiation Oncology, MD Anderson Cancer Center, Houston, TX; 5. Dept Pathology, MD Anderson Cancer Center, Houston, TX. Background The incidence of peritoneal recurrence after resection of gastroesophageal junction adenocarcinoma and the patterns of such recurrence are not well reported. The aim of this study was to identify the incidence and patterns of recurrence after resection in patients with Siewert type II/III adenocarcinoma. Methods We queried a prospectively maintained database for patients with Siewert type II/III gastroesophageal junction adenocarcinoma who underwent curative-intent resection in 2000-2015 at our institution, and we analyzed clinical variables of patients who experienced recurrence. Results Among 299 patients included in this study, 225 (75%) had Siewert II and 74 (25%) had Siewert III tumors. Length of gastric tumor extension was <2 cm in 85 patients (28%), 2-4 cm in 140 (47%), and 4 cm in 74 (25%). A total of 203 patients (68%) had clinical T3 tumors, 156 (52%) had clinically positive lymph nodes, and 238 (80%) received neoadjuvant treatment. During follow-up, median duration of which was 4.8 years, 91 patients (30%) experienced recurrence, and 140 (47%) died. Only 4 patients (1%) who were alive and without recurrence by the end of the study period had a follow-up length shorter than 2 years. Overall, distant locations were the dominant recurrence sites (n=69, 23%); peritoneal recurrence (n=24, 8%) and locoregional recurrence (n=8, 3%) were relatively rare. The incidence of peritoneal recurrence was significantly higher when the length of gastric tumor extension exceeded 4 cm (15%, vs. 7% for <2 cm and 5% for 2-4 cm; p=0.001; Figure 1 ). Siewert type did not affect recurrence patterns. Conclusion Among patients with Siewert II/III adenocarcinoma, peritoneal recurrence is relatively rare but is more common in patients with gastric tumor extension 4 cm than in patients with shorter gastric tumor extension, demonstrating the importance of preoperative staging laparoscopy for gastroesophageal junction adenocarcinomas. (2005) (2006) (2007) (2008) (2009) (2010) (2011) (2012) (2013) (2014) (2015) (2016) . Logistic regression analyses were used to evaluate associations between tumor/patient characteristics, and serum biomarker levels (as continuous variables). Cut point analysis was performed using the Liu method. Kaplan-Meier analysis was used to identify median survival stratified by age and other patient-specific factors, and Cox proportional hazard models were used to identify associations between factors and overall survival. Results: Of included patients, 1054 were metastatic at the time of diagnosis. On multivariate analysis urinary 5-HIAA (OR 1.05, 95%CI 1.00-1.1, per 50mg/24h, p=0.001) and chromogranin A (OR 1.1, 95%CI 1.00-1.1, per 50 ng/mL, p<0.001) were associated with presence of metastasis in SBNET (Table 1) . Cut point analysis demonstrated higher sensitivity for metastatic disease with 5-HIAA (71% at 12.5mg/24hours) as compared to chromogranin A (58% at 101.4ng/mL); both were equally specific (65% Introduction: Gastric cancer is often diagnosed at advanced stages, when patients have limited treatment options. Palliative care is increasingly recognized as an important modality for patients at all stages of cancer, but especially in advanced disease. There is limited research on the use of palliative care nationally among gastric cancer patients. We analyzed the most recent National Cancer Database (NCDB) data to better determine the utilization of palliative treatments in patients with advanced gastric cancer. Methods: We identified patients with stage IV gastric adenocarcinoma from the NCDB data between 2005 and 2016. We used chi-square tests to examine subgroup differences between patients based on the receipt of palliative care. Multivariate analyses including logistic regression were used to examine the treatment patterns and demographic factors influencing utilization of palliative care for these patients nationally. Results: Our final cohort included 36,757 patients, in whom there was a gradual trend towards increased receipt of some form of palliative care over the 12-year period. Initially, only 15.6% of patients were utilizing palliative care services in 2005 -2006 . This increased to 25.1% in 2015 -2016 . Palliative chemotherapy utilization increased from 4.7% in 2005 -2006 to 12% in 2015 -2016 , while palliative surgery use decreased from 2.8% to 1.3% over the same time period (p<.001). We also found that elderly patients, females, White Hispanics, those from a rural setting and those treated at community cancer programs were less likely to receive palliative care (p <.001). Of note in more recent years (2015-2016 vs 2005-2006) patients were almost two times as likely to receive palliative care (OR 1.837, p<.001) . Conclusions: Our study shows increased utilization of palliative care in patients with advanced gastric cancer over this 12-year time period. However, 74.9% of patients diagnosed in 2015-2016 were still not receiving any palliative care, indicating much room for improvement in dissemination of the message that palliative care is important for all cancer patients, especially those with latestage disease. Background: Advanced directives (AD) are an important aspect of healthcare, especially for patients with advanced cancer. Surgery is a life-altering event and it is desirable for patients to have considered, and documented their advanced directives prior to it. We examine the effectiveness of an intervention designed to increase the rate of documentation of AD for patients with advanced cancer seen in a surgical oncology clinic. Methods: From July 2018-present, the surgical oncology clinic at the University of Chicago has included an embedded palliative care physician (PC), who co-visits with patients with emotional or physical distress. We designed an intervention to include a standard conversation by a navigator, trained in counseling and advanced directives, to encourage patients to complete AD during their clinic visit. Reminders were included at subsequent visits. Physicians and advanced providers also encouraged patients to complete their AD prior to surgery. Standard (electronic and telephonic) follow up was undertaken for patients requiring time to process the request. At admission to the hospital, patients were inquired about their AD. Results: 192 patients were seen between Jan-June 2019, with diagnoses of metastatic cancer of colon (28%), appendix (21%), and peritoneal mesothelioma (15%). 61 (32%) of patients documented AD. Qualitative barriers to obtaining documented AD were incomplete forms, patient reluctance, difficulty with electronic medical records, multiple health systems and non-return of forms. Majority of patients were willing to discuss advanced care planning. Conclusion: Despite aggressive pursuit of documentation of AD, the actual rate of documented plans was 32%. Innovative workflows are necessary to increase adoption and documentation of advanced care planning. INTRODUCTION: Spirituality and religion (S&R) may be valuable resources for patients during cancer care. Despite varied prognoses and therapies, most research does not evaluate S&R needs by diagnosis-related differences. We sought to characterize S&R needs and beliefs among cancer survivors who underwent cancer-directed surgery stratified by disease and treatment-level factors. METHODS: A cross-sectional survey was completed by patients with a 4-month history of cancer who had undergone cancerdirected surgery. Data on demographics, clinical characteristics, and S&R identity/beliefs/needs were analyzed. Factors associated with S&R beliefs were added to logistic regression models with S&R identity to determine if they increased the likelihood that patients wanted to include S&R in their cancer treatment. RESULTS: Among 236 respondents, the mean age was 58.8 (SD=12.10) years; most were female (75.8%) and white/Caucasian (94.1%). While the majority (78.4%) identified as currently cancer-free, common malignancies included breast (43.2%), male reproductive (8.9%), skin (8.5%), and GI (7.2%). Patient S&R identity varied, with moderately S&R being the most frequent category (33.5%; Figure) . Patients who identified as highly or moderately S&R were much more likely to want S&R to be part of their cancer care (p<.001). Patients who underwent an out-patient procedure were 47% less likely to believe that R&S played a role in their cancer care versus patients in an in-patient setting (OR CI:0.30, 0.97; p=.037) . However, surgical setting did not increase the odds that a patient wanted S&R incorporated into their cancer care (p=.260) or desired resources in the hospital (p=.205). Stage of disease, cancer-free status, and age at diagnosis were not associated with S&R needs/beliefs relative to their care (all p > 0.05). CONCLUSION: Perhaps not surprisingly, patients that identified as highly or moderately S&R wanted these resources incorporated into their cancer care. While stage of disease, age, and disease-free status were not associated with S&R needs, patients undergoing surgery in the outpatient setting were less likely to believe S&R played a role in their care. Introduction: Interhospital transfers have been shown to be associated with increased 30-day mortality, increased hospital acquired conditions (HACs), and worse patient outcomes. They remain a necessary part of the care we provide to our patients, however. We aimed to analyze the interhospital transfer patients at our institution to develop a predictive model to improve the safety of accepting interhospital transfer patients. Methods: A retrospective review of all patients transferred from an outside emergency department (ED) or inpatient unit to our NCI designated comprehensive cancer center was conducted. Patients demographics and comorbidities, clinical variables, and transfer data were collected. Analyses were conducted using univariate and multivariate logistic regression models. Results: Between June 1, 2018 and May 31, 2019, 819 patients were transferred to our center; these included 229 surgical patients and 590 medical patients. Twelve patients had never been seen at our center before. Eighty-four percent of medical patients and 82.5% of surgical patients arrived during the night shift (p=0.518). One hundred twenty one patients (14.8%) died within 30 days of transfer, 37 (4.5%) died during the hospitalization, and 13 (1.6%) died within 48 hours of transfer. Transfer to ICU within 24 hours of arrival at our institution occurred in 19 patients (2.3%). On univariate analyses, increased HR, decreased hgb, and increased lactate were associated with increased risk of 30 day mortality (p<0.05). Distance travelled from the outside institution and the time of transfer did not affect mortality. Conclusion: Commonly collected clinical parameters can be predictors of poor outcomes in patients undergoing interhospital transfers. We are now using these data to prospectively launch a scoring system to help identify patients at highest risk for poor outcomes following interhospital transfer.

Regression of 30d mortality, univariate analysis:

Declining Use of Red Blood Cell Transfusions for Gastrointestinal Cancer Surgery: A Population-based Analysis J. Zuckerman, 1 * N. Coburn, 2 J. Callum, 3 A. Mahar, 4 V. Zuk, 5 Y. Lin, 3 R. McLeod, 6 A. Turgeon, 7 E. Pearsall, 8 G. Martel, 9 J. Hallet. 2 1. Division of General Surgery, Department of Surgery, University of Toronto, Toronto, ON, Canada; 2. Odette Cancer Centre, Sunnybrook Health Sciences Centre, Toronto, ON, Canada; 3. Sunnybrook Health Sciences Centre, Toronto, ON, Canada; 4. University of Manitoba, Winnipeg, MB, Canada; 5. Sunnybrook Research Institute, Toronto, ON, Canada; 6. Cancer Care Ontario, Toronto, ON, Canada; 7. CHU du Quebec, Quebec, QC, Canada; 8. Department of Surgery, Toronto, ON, Canada; 9. The Ottawa Hospital, Ottawa, ON, Canada. Background: Perioperative anemia is common in gastrointestinal (GI) cancer surgery patients and is often treated with red blood cell transfusion (RBCT), which carries risks for inferior oncologic outcomes. Despite level-1 evidence for restrictive transfusion strategies, RBCT use is often not consistent with guidelines leading to a high rate of unnecessary transfusions. Understanding of RBCT use at the population-level is necessary to develop system-level efforts to minimize perioperative RBCT for cancer. We sought to evaluate the secular trends of transfusion in a large North American population. Methods: We conducted a population-based retrospective cohort study of patients undergoing GI cancer resection between 2007-2018 using linked administrative health datasets in Ontario, Canada. Primary outcome was administration of any RBCT during the hospitalization. Temporal RBCT trends were analyzed with Cochran-Armittage tests for trend. Modified Poisson regression assessed trends while controlling for potential confounders. Results: Of 79,764 patients undergoing GI cancer resection, median age was 69 (IQR: 60-78) years old and 55.5% were male. The most frequent cancer site was colorectal cancer (n=63,243), followed by esophago-gastric (n=7,307), hepato-pancreato-biliary (n=6,510), and small bowel (n=2,704). 30% of patients received RBCT. The proportion of patients transfused decreased from 26.5% in 2007 to 18. 9% in 2018 (p<0.001) . This trend remained consistent when stratified by sex, age, cancer type, operative approach, procedure setting, and institution teaching status. After adjusting for patient and institution factors, the time period was associated with receipt of RBCT with a relative risk of 0.94 (95% CI 0.91-0.96) for 2011-14 and 0.75 (95% CI 0.73-0.78) for 2015-2018 compared to the period of 2007-10. Conclusions: Over eleven years, we observed a decrease in RBCT for GI cancer resection. These findings may reflect the dissemination of clinical guidelines and implementation of patient blood management programs. An evaluation of institutional variation and the relationship with outcomes is warranted to identify opportunities for further improvement.

Quality of Colorectal Cancer Care within a Local VA Cohort: Small Wins and Room for Improvement G. Edwards, 1 * R.L. Martin, 1 L. Samuels, 1 C. Bailey, 1 C.L. Roumie. 2 1. Surgery, Vanderbilt University Medical Center, Nashville, TN; Geriatric Research Education and Clinical Center, Nashville, TN. Introduction: The Veterans Health Administration (VHA) is the largest integrated provider of cancer care in the United States, and colorectal cancer (CRC) is the third most common cancer in VHA. Despite improving trends in CRC mortality, discrepancies remain between adherence to National Comprehensive Cancer Network (NCCN) guidelines. We evaluated changes in NCCN adherence over time for non-metastatic CRC provided within our local VA system. Methods: We performed a retrospective cohort study of patients with stage I-III CRC who underwent resection between January 1, 2001 and December 31, 2015 at a local VA system. Clinical information was abstracted from the electronic health record using a structured form by two clinicians (kappa>0.90). The exposure was year of CRC resection. The outcome was complete adherence to applicable year-specific NCCN metrics during three phases: pre-operative, intra-operative, and post-operative. We evaluated the association between year of CRC resection and adherent care using logistic regression models adjusting for the following covariates: age, race, stage, and evidence of prior hospitalization within the preceding 365 days. Results: A total of 430 patients met inclusion criteria (346 with colon cancer and 84 with rectal cancer). For the pre-operative period, we excluded 58 cases which presented as surgical emergencies and thus were ineligible for pre-operative evaluation. Overall, optimal adherence to NCCN metrics was achieved among 31%, 75%, and 27% of patients within the pre-, intra-, and post-operative phases, respectively. After covariate adjustment, significant improvement was observed over time for the pre-and intra-operative phases, but not for the post-operative phase (Table 1) . Within the post-operative phase, while median time from resection to surveillance colonoscopy decreased over the study period, time to receipt of adjuvant chemotherapy if indicated significantly increased over the study period (P=0.04). Conclusion: While adherence to national guidelines improved for pre-operative evaluation and intra-operative surgical quality, there is room for improvement in ensuring that Veterans receive timely adjuvant chemotherapy. Introduction: Unpublished cancer registry data from 2013 to 2015 reveals that a large proportion of melanoma patients in Saskatchewan, Canada, did not receive guideline directed surgical care, according to their Tumour-stage; likely representing an issue in the quality of care being delivered. We set out to diagnose and understand problems that currently lead to poor rates in the achievement of guideline directed surgical care for melanoma patients. Methods: A hypothetical reference case scenario for a patient who did not receive guideline directed surgical care, based on an amalgamation of real cases, was reviewed in semi-structured interviews with ten melanoma stakeholders to identify weaknesses, strengths, and areas for improvement in melanoma care. The interview content was themed, and the human factors methods of the Systems Engineering Initiative for Patient Safety (SEIPS) and a Lovebug diagram applied. Results: A Lovebug diagram was produced ( Figure 1 ) and three key themes were identified: (1) melanoma care occurs within a system with a lack of standardization and controls, (2) an essential element to improving melanoma care will be specialist support of primary care physicians in addition to reducing misunderstanding between these professionals, (3) improved communication and coordination of care amongst physicians who care for melanoma patients would be beneficial. Conclusion: This research has led to an understanding of the context, barriers, and causes, leading to the poor rates in the achievement of guideline directed surgical care for melanoma patients in Saskatchewan. This knowledge will be beneficial when designing improvement interventions to improve the quality of care delivered to melanoma patients going forward. Introduction: The American Cancer Society (ACS) makes clear recommendations for physical activity (PA). Cancer Survivors (CS) with ostomy may experience difficulties with performing PA. We describe the frequency and duration of PA for CS with ostomy participating in an intervention to improve their quality of life (QOL). Methods: Data were analyzed from a multisite, randomized trial of an ostomy telehealth intervention for CS. Ostomy type was obtained by EMR. This analysis utilized baseline surveys utilizing the ACS question on PA and the Chronic Disease Self-Efficacy Scale (CDSES). Survey data were collected via mailed questionnaires or via the web using REDCap. "Weighted PA minutes" were calculated by moderate PA time plus 2 times strenuous PA time. Data were stratified by ostomy type, summarized descriptively, and tested using one-way ANOVA. Results: The majority of CS had urostomies (N=55; 53.9%), followed by colostomies (N=38; 37.3%), and ileostomies (N=9; 8.8%), with mean age (72.5, 63.3, 65.1) and male gender (74.6%, 44.7%, 66.7%), respectively. CS with urostomies reported an average of 55.9 min./week for weighted PA minutes, 10.9 min./week for strenuous PA, and 34.0 min./week for moderate PA. CS with colostomies reported an average of 136.7 min./week for weighted PA minutes, 30.5 min./week for strenuous PA, and 75.6 min./week for moderate PA. CS with ileostomies reported an average of 66.8 min./week of weighted PA minutes, 11.1 min./week for strenuous PA, and 44.6 min./week for moderate PA. Eighteen CS (17.6%) met the ACS PA guideline: 9 CS with urostomies, 7 CS with colostomies, and 2 CS with ileostomies. CS with urostomies reported average self-efficacy scores of 7.0/10 and 6.6/10 for confidence in the ability to engage in gentle or aerobic exercise, respectively; CS with colostomies reported average scores of 7.4/10 and 7.3/10, respectively; and CS with ileostomies reported average scores of 7.2/10 and 6.3/10, respectively (nonsignificant). Conclusion: Few CS did meet the recommended ACS PA guidelines despite PA self-efficacy ratings indicating confidence in the ability to engage in exercise. Strategies should be devised to ensure CS with ostomy engage in PA.

Participant characteristics by ostomy type. 1 One-way ANOVA findings showed nonsignificance.

Patients J.T. Cohen,* E.A. Fallon, K.P. Charpentier, W.G. Cioffi, T.J. Miner. Surgery, Brown University, Norwood, MA. Introduction: The value of palliative surgery (PS) is optimized by maximizing symptom improvement and quality of life while minimizing toxicity. Value is further augmented in patients experiencing longer than expected survival. The impact of PS is enhanced when symptom severity is addressed in a manner consistent with patients' care goals. Methods: All procedures performed from 2003 to 2017 to palliate symptoms of advanced cancer were identified from a comprehensive palliative surgery database. Outcomes measured included symptom resolution, complications, and fall risk factors. Patients were observed for >90 days or until death. Results: Symptoms, which included enteral obstruction (41.9%), pain (27.5%), biliary obstruction (9.6%) and local disease control (6.0%), were improved in 146 (87.4 %) of the 167 identified patients. The top quartile of patients had 565 days of survival. There were no significant differences in the cohorts' demographics or degree of symptom improvement when comparing the top quartile to the lower three quartiles. Patients in the top quartile for survival were significantly more likely to require re-operation for new or recurrent symptoms (41.5% vs 20.2%, p=0.007), however they experienced significantly longer re-intervention free survival (416 (IQR 226 -705) vs 59.5 days, p<0.0001). Multivariate analysis demonstrated that the ability to rise from a chair was independently associated with top quartile survival (HR 7.61, p=0 .008), as was the need for re-operation (HR 2.81, p=0.0012) . Patients who were able to rise from a chair had significantly prolonged overall survival (320 (IQR 147 -undefined) vs 87 days, p<0.001) (Figure 1 ). Conclusions: Although not the primary goal, long-term survival can be achieved following PS and is associated with the ability to rise from chair. Many long-term survivors go on to require additional PS to manage symptoms, reflecting the success of multimodal treatment and favorable tumor biology. Taken together, this highlights how a high value PS can offer a large positive impact on advanced cancer patients.

C.G. Boudreau, 1 * T. Levatte, 1 A. Gareau, 2 S. Legere, 1 M. Bezuhly. 3 1. Medicine, Dalhousie University, Halifax, NB, Canada; 2. Calgary Lab Services, Calgary, AB, Canada; 3. Dalhousie University Department of Plastic Surgery, Halifax, NS, Canada. Background: Abdominal adhesions (AA) are fibrous bands formed in response to surgical trauma which connect visceral and/or peritoneal surfaces, leading to possible long-term complications. This study uses a murine model of AA formation to determine the anti-fibrotic effect of a novel selective angiotensin II type 2 receptor agonist, compound 21 (C21), in reducing adhesion formation. Methods: Laparotomy was performed on female BALB/c mice and cecum and overlying parietal peritoneum abraded with sandpaper. Systemic (oral gavage) or local (intraperitoneal injection) administration groups were treated with C21 (10 g/kg) or saline (vehicle) daily for 7 days. Mice were sacrificed 8 days post-surgery, adhesions graded by a blinded observer, and peritoneal fluid collected for TGFb ELISA. Laparotomy incisions were excised for histological study. Parietal peritoneal fibroblasts and visceral mesothelial cells were isolated and scratch-wound assays performed using C21 (10 mM), angiotensin II (AngII, 1 mM), or both. Results: Systemic and local administration of C21 reduced the formation of AA in vivo and amount of TGFb in peritoneal fluid. Histological analysis of surgical incisions revealed no difference in the number of CD31+ vessels, CD68+ cells, collagen I/III distribution, total collagen density, and dermis thickeness, while aSMA expression was reduced in C21-treated animals. Migration of parietal peritoneal fibroblasts and visceral mesothelial cells in vitro was reduced with C21 compared to control or AngII. Conclusions: C21 reduced or completely prevented adhesion formation both with local and systemic adminstration. These findings may be attributed to decreased levels of TGFb in vivo and decreased migration of peritoneal fibroblasts and visceral mesothelial cells with C21 treatment. Importantly, C21 did not have histologically quantifiable effects on laparotomy wounds. This suggests that C21 could reduce AA without comprimising laparotomy healing.

Implementation of an EMR Integrated Pathway for the Management of Malignant Bowel Obstruction D. Schuitevoerder,* C. Vining, M. White, C. Hoppenot, Y. Berger, I. Lazo, S.K. Sherman, A. Kamm, L. Chavez, P. Kallakuri, E. Fenton, J. Male, L. Capicchioni, S. Tun, O. Ahmed, C. Semrad, D. Radovanovic, D. Micic, N. Lee, B. Polite, O.S. Eng, K. Turaga. University of Chicago, Chicago, IL. Background: Despite published evidence based interventions for malignant bowel obstruction (MBO), implementation of a standard pathway is challenging. We hypothesized that using industrial engineering techniques and a modified dynamic sustainability framework for implementation, we can implement an electronic medical record (EMR) based pathway in the management of MBO. Methods: A workflow in the management of MBO was developed using iterative meetings from 8/2018 to 4/2019 including gateway stakeholders (surgical oncology, gynecological oncology and medical oncology), interventional stakeholders (gastroenterology, interventional radiology) and supportive stakeholders (hospital medicine, palliative care, nutrition, nursing). Industrial engineers were utilized to study human factors, and perform a method study. EMR integration was performed using EPIC systems Agile MD pathway and educational materials were created. Interventions such as early placement of gastrostomy tubes, total parenteral nutrition and medications were protocolized. Results: Since implementation, over 6 months the pathway and order set has been activated 56 times. Orders have been employed 21 times through the AgileMD pathway demonstrating a pathway drift of 62.5%. Educational materials have been accessed routinely during this time. Conclusion: Feasibility of implementing an EMR integrated MBO pathway is demonstrated with early suggestion of pathway drift. Utilizing tools of implementation science are necessary to facilitate widespread adoption of evidence based interventions in the management of patients with MBO. BACKGROUND: Signet-ring cell carcinoma (SRC) is a distinct malignancy occurring across tubular gastrointestinal tract (tGIT) and associated with worse prognosis. Given paucity of data we comprehensively examined outcomes of SRC treatment across tGIT. METHODS: We analyzed sitespecific adjusted hazard ratios (aHR) derived from Cox models on all actively treated SRC cases recorded in the National Cancer Data Base between 2004 and 2015 and compared to no-other-specified adenocarcinoma (ADC). RESULTS: There were 41,686 SRC (4.6%) and 871,373 ADC patients (95.4%) recorded in the NCDB and meeting study criteria. SRC patients were younger (63.1 14.7 versus 67.0 13.4 years, p<0.001). Disease sites differed in their respective proportion of SRC as compared to ADC (appendix 30.4%, stomach 23.6%, small bowel 4.7%, esophagus 4.5%, colon 1.8%, rectum 1.1%). Due to broad differences in cancer frequency among sites, however, stomach (n=24,433) and colon (n=9,914) contributed highest number of SRC cases. Fraction diagnosed initially with stage IV proportion was higher among SRC patients (42.5% versus 24.5%, p<0.001). Multivariate regression model for overall survival including age, sex, comorbidity, year of diagnosis and stage demonstrated that SRC histology is associated with worsened survival (aHR=1.377, p<0.001) while adjusted for other factors, including disease site (esophagus=reference, stomach aHR=0.910, small bowel aHR=0.808, colon aHR=0.417, rectum aHR=0.442, appendix aHR=0.448; all p<0.001) . CONCLUSIONS: Patients diagnosed with SRC are commonly younger as compared to those with ADC. SRG occurrence is site-dependent, with highest proportion in vermiform appendix but highest frequency from gastric carcinoma. Even when adjusted for stage, SRG histology represents a negative prognostic factor with approximately 37% reduction of overall survival. That being said, cure among patients with non-metastatic disease remains possible.

OS stratified by metastatic status (M0 vs M1) and adenocarcinoma NOS (solid lines) versus SRC carcinoma (dashed lines).

Disease-associated Colorectal Cancer: A Population-based Study J. Bogach,* G. Pond, C. Eskicioglu, H. Seow. Health Research Methods, Evidence and Impact, McMaster University, Hamilton, ON, Canada. Background Recommendations regarding the extent of surgical resection (proctocolectomy, total colectomy, segmental resection) among patients with Inflammatory Bowel Disease (IBD) and Colorectal Cancer (CRC) are inconsistent. We aim to identify population-level patterns of resection and their association with survival. Methods Patients with CRC and IBD between 2000-2015 were identified through ICES. IBD patients were classified as Ulcerative Colitis (UC) or Crohn's Disease (CD) using a validated algorithm. Included patients had surgical resection, which was defined as segmental, total colectomy or proctocolectomy. Cox Proportional Hazards Regression Models looked at impact of extent of resection, adjusted for other factors prognostic for Overall Survival (OS). Results From 2000 -2015 or CD (366) underwent surgery for CRC. Median age at diagnosis was 60-64 in UC and 55-59 in CD (p=0.078). There was no difference in primary site, stage, grade or comorbidity between those with UC or CD. Five year OS is 63.7% in UC and 57.5% in CD (p=0.033). Patients with UC most commonly had segmental resection (45.6%), followed by proctocolectomy (36.1%) then total colectomy (18.4%). In UC, in univariable analysis, proctocolectomy is associated with improved OS when compared to segmental (HR 0.63, p<0.001) . However, in multivariable regression there is no difference between segmental resection and proctocolectomy ]. Patients with CD had segmental resection (68.0%) followed by proctocolectomy (19.4%) and total colectomy (12.6%) (p<0.001). In univariate analysis type of resection does not impact survival in CD (p=0.48). In multivariable analysis, total colectomy has worse survival in both IBD subtypes p<0.001] . Conclusion Most IBD patients with CRC are having segmental resection. When other factors are controlled for, overall survival is similar in segmental resection and proctocolectomy. Total colectomy is associated with worse survival in both IBD subtypes. Unmeasurable factors that contribute to selected resection type may impact these survival outcomes. Background: Neoadjuvant chemoradiotherapy (nCRT) followed by resection and postoperative chemotherapy is standard of care in the treatment of locally advanced rectal cancer (LARC). A novel treatment approach known as total neoadjuvant therapy (TNT) has been increasingly employed, in which patients receive both chemotherapy and nCRT prior to resection. The aim of this study is to assess the nationwide utilization and potential benefits of TNT use in LARC. Methods: Patients with clinically staged LARC (Stage II-III) treated with nCRT or TNT were identified from the National Cancer Data Base (NCDB) between 2004 and 2015. TNT was defined as multi-agent chemotherapy initiated 90 days prior to the start of nCRT. Temporal, pathologic and survival outcomes were compared. Propensity score analysis was used to account for indication bias. Results: 24,926 patients met the inclusion criteria, 24,439 (98%) of which were treated with nCRT and 487 (2%) were treated with TNT. Patients were more likely to receive TNT if they were younger, had private insurance, had a higher median income, were treated at academic cancer centers, resided in the Northeast and were diagnosed recently (P<0.0001). Utilization of TNT increased steadily throughout the study period with the most recent 2 years of analysis comprising 47.4% of cases. Patients treated with TNT were more likely to have clinical T4 lesions, N1-2 disease, larger tumor size, undergo a sphincter preserving procedure or have 12 or more LNs retrieved (P<0.0001). After propensity score matching, nCRT and TNT had similar OS at 5-year survival (73% vs, 72%, respectively.) TNT however did demonstrate a higher risk of mortality when compared to nCRT at 5 years (HR 1.78, 95% CI 0.99-3.17; P<0.05). Conclusions: Nationwide utilization of TNT in the treatment of LARC is increasing. Patients are more likely to receive TNT if they had advanced disease, were younger or had better access to healthcare. Despite worse presenting clinical disease, patients treated with TNT had better pathologic outcomes and similar 5-year survival to nCRT. Randomized controlled trials are underway to truly define the role of TNT P104 Impact of Histology on Survival in Colon Cancer with Peritoneal Carcinomatosis: Does It Really Matter? L. Hendrick,* S. Chintalapani, P. Dickson, E.S. Glazer, D. Yakoub, D. Shibata, J. Deneve. Surgery, University of Tennessee Health Science Center, Memphis, TN. Background: Peritoneal carcinomatosis (PC) is a pattern of metastasis from colon adenocarcinoma and associated with a poor prognosis. Mucinous adenocarcinoma is associated with PC and has a worse outcome when compared to non-mucinous adenocarcinomas. We sought to investigate the relationship of histologic subtype and overall survival in colon cancer patients with PC. Methods: The 2004-2015 National Cancer Database (NCDB) was queried for cases of colon cancer with PC. Trends in treatment and survival outcomes were examined. Factors affecting overall survival (OS) were assessed with Kaplan-Meier log rank test and Cox regression analysis. Results: Of 781,292 CRC cases, 23,735 patients (3.0%) with PC were identified [median age 69 years, male (49.5%)]. Most cancers were non-mucinous (87.2%), with 10.5% mucinous and 2.3% signet ring cell adenocarcinoma. The majority of cancers were low grade (69.8%). There was a slight right sided predominance (46.4%) vs. left (39.8%) and transverse (10.1%). The median OS for the entire cohort was 37.4 months, with an OS of 38. 3, 33.6, and. 19.9 months for non-mucinous, mucinous, and signet ring cell subtypes, respectively. On univariate analysis, age, race, insurance status, Charlson/Deyo comorbidity index, tumor grade and location, nodal and margin status, receipt of chemotherapy, and histologic subtype were associated with OS. On multivariate analysis, increasing age (p<0.001), public or no insurance (p<0.001), high grade histology (p<0.001), R1 or R2 resection (p<0.001), and KRAS positivity (0=0.023) were associated with worse outcomes while left sided tumors (p<0.001) and receipt of chemotherapy (p=0.013) conferred a survival benefit. After adjusting for covariates, histologic subtype was not associated with OS in patients with PC (p=0.895). Conclusions: In colon cancer patients with peritoneal disease, histologic subtype does not impact overall survival. Right sided and KRAS positive tumors with PC demonstrate worse OS regardless of histology, while patients with R0 resection and those receiving chemotherapy had a survival benefit. Consideration for aggressive surgical resection and multimodal therapy should be made when possible.

Time to Surgery and Colon Cancer Survival in the United States S. Khan, 1 * C. Shen, 2 C. Kaltenmeier, 1 M.S. David, 1 G.A. David, 1 A. Zureikat, 1 D. Bartlett, 1 S. Tohme. 1 1. University of Pittsburgh Medical Center, Pittsburgh, PA; 2. Ohio state university, Columbus, OH. Introduction: Time to surgery (TTS) is of concern to patients diagnosed with cancer and their physicians. Controversy surrounds the impact of TTS on colon cancer survival. There is limited national data evaluating the association, thus, our aim was to estimate the overall survival (OS) impact from increasing time to surgery (TTS) for patients with colon cancer. Methods: Using the National Cancer Data Base (NCDB), we assessed overall survival as a function of time between diagnosis and surgery by evaluating intervals encompassing <7, 7-30, 31-60, 61-90, 91-120 , and 121-180 days in length. All patients were diagnosed with nonmetastatic colon cancer and underwent surgery as initial treatment. Our main outcome was overall survival as a function of time between diagnosis and surgery, after adjusting for patient, demographic, and tumor-related factors using Cox regression models and propensity score-based weighting. Results: A total of 514,103 patients diagnosed between 1998-2012 were included. Individuals having <7, 7-30, 31-60, 61-90, 91-120, and 121-180 days between diagnosis and surgery comprised 35.4%, 45%, 15.1%, 2.9%, 1%, and 0.6% of the patients, respectively. There was a steady increase in median TTS across the years. On multivariable analysis, TTS greater than 30 days or within the first week independently increased mortality risk. There was a significant increase in mortality with TTS 31-60 (HR 1.13), 61-90 (HR 1.49), <7 (HR 1.56), 91-120 (HR 2.28), and 121-180 (HR 2.46) compared to surgery performed 7-30 days after diagnosis (p<0.001). Conclusion: TTS independently affects survival, and this represents a public health issue that should be addressed at a national level. Although time is required for evaluation prior to surgery, efforts to reduce TTS should be pursued to enhance survival. Background. Administration of combined approaches such as NCRT and total mesorectal excision has reduced locoregional recurrences. However, patients with lymph node metastases after NCRT have a significant rate of locoregional and distant recurrences associated with low survival rate. Objective. To identify factors associated with locoregional and distant recurrences in a group of patients treated with NCRT. Materials and methods. Between 2005 and 2018, 195 patients with stage III rectal adenocarcinoma treated with neoadjuvant chemoradiation. All patients received neoadjuvant Capecitabine +/-oxaliplatinum and 45 -50.4 Gy (CT/RT). Total Mesorectal Excision (TME) and postoperative CT. Factors associated with outcome and recurrence were analyzed by logistic regression analysis and Kaplan-Meier method. Results. Were 107 males and 88 females, mean age was 56 yrs. Mean distance from distal tumor margin to the anal verge was: 4.5 cm. Surgical procedures included: low anterior resection: 85; Pelvic exenteration: 53, APR: 46 and Colo-anal anastomosis: 11. Median follow-up was 70 months. Median analyzed and + lymph nodes were 15 and 3, respectively. Pathologic tumor stage were: IIIA: 24, B: 102 and C: 69. Recurrences were: local: 2%, locoregional: 9.7% and distant: 31.6%. 6-year disease-free survival was: IlIA: 78%, B: 49% and C 30% (p= 0.02). Adverse factors associated with locoregional and distant recurrence were: pre-treatment CEA >5 ng/mL; >4 + lymph nodes, poor differentiation (p=0.01) and adjuvant chemotherapy with one drug (p=0.04). Conclusion. In our experience, Stage III rectal cancer patients treated with NCRT + standard TME, locoregional failure is low whereas distant failure was significant. Total neoadjuvant chemotherapy should be considered in patients with elevated CEA (>5 ng/mL), poor differentiation, and N2 disease on pre-treatment staging imaging. Background: The role of cytoreductive surgery in treating colon cancer with peritoneal metastasis (PM) has become less controversial, but treatment sequence remains debated. We aimed to examine treatment sequencing trends and predictors of treatment using the National Cancer Database (NCDB). Methods: The NCDB was queried to identify colon cancer patients with PM between 2010-2015. Clinical stage IV colon cancers with known primary site and without bone, brain, liver or lung metastasis were included. Cases with neuroendocrine histology, isolated nodal metastasis, radiotherapy, immunotherapy, and missing treatment or follow-up data were excluded. Cochran-Armitage tests were used for the sequencing trend analyses, and logistic regression models for multivariable analysis. Results: A total of 3798 patients with PM were identified; 63%, 21%, and 15% received surgery, chemotherapy, or no treatment as first approach, respectively (Fig. 1A) . Of those who received both surgery and chemotherapy, <10% received chemotherapy in the neoadjuvant setting. Over the period, a significant downward trend was observed for patients receiving surgery and adjuvant chemotherapy (46% to 26%) and an upward trend for no treatment (14% to 21%) and chemotherapy only (13% to 20%, p<0.01, Fig. 1B) . On multivariable analysis, facility type, age, sex, insurance, primary site, histology and year of diagnosis were significant. Notably, community (OR 1.85) and integrated (OR 1.91) cancer programs were at increased odds of performing surgery first compared to administering chemotherapy. Patients with mucinous adenocarcinoma tended to have surgery first versus chemotherapy (OR 1.47), conversely patients with signet-ring cell carcinoma tended to have chemotherapy first (OR 1.79) when compared to adenocarcinoma. Patients with age 70 or public insurance were less likely to receive any treatment. Conclusion: While there was an overall trend towards utilizing palliative treatment approaches, only a small number of stage IV-PM patients received neoadjuvant chemotherapy. Age, facility, and insurance were the major determinants of treatment received, possibly suggesting underlying healthcare disparities.

Thin Melanoma: Early Results From a Multi-Institutional Collaborative M. Mavros, 1 * M. Moncrieff, 2 C. Nessim, 3 C. Seo, 3 N. Look-Hong, 1 F. Wright. 1 1. University of Toronto, Toronto, ON, Canada; 2. Norfolk and Norwich University Hospitals (NHS Foundation), Norwich, United Kingdom; 3. The Ottawa Hospital, Ottawa, ON, Canada. Background: Consideration of sentinel lymph node biopsy (SLNB) is currently recommended for patients with T1b (Breslow depth 0.8-1mm or <0.8mm with ulceration) and T1a (Breslow depth <0.8mm without ulceration) cutaneous melanomas with high-risk features, however this recommendation leads to an exceedingly large number of negative biopsies. In this regard, we aimed to identify clinicopathologic factors predicting SLNB positivity by examining a large multi-institutional population, and create a model for prediction of SLNB positivity which will assist in every-day decision making. Methods: Data were extracted on adult patients with thin (T1) cutaneous melanoma who underwent a SLNB between 2005 and 2018 in 3 tertiary referral centers in Europe and Canada (data from additional centers are pending). Univariable and multivariable logistic regression analyses were performed to identify predictors of SLNB positivity. Results: A total of 488 patients were analyzed. The median age was 59 years (interquartile range 47-67 years) and 53% were female. Most of the patients had high-risk features: Breslow thickness 0.75-1 mm in 83%, ulceration in 7.6%, mitotic rate 1 (per mm 2 ) in 85.2%, Clark's level 4 in 33.3%, lymphovascular invasion in 1.6%, nodular histology in 1.8%, tumor regression in 21.5%, and absence of tumor infiltrating lymphocytes in 14.3%. Thirty-three patients (6.8%) had a positive SLNB. On multivariable analysis, Breslow thickness and mitotic rate were independently associated with SLNB positivity. A predictive model will be developed and validated once the full cohort is available. Conclusions: Our preliminary results confirm that Breslow thickness and mitotic rate remain independent predictors of SLNB positivity in patients with T1 cutaneous melanoma. Even with these refinements, the number needed to diagnose is 14:1 indicating that more work is required to identify accurate biomarkers for this important group of patients. I The Multicenter Selective Lymphadenectomy Trial (MSLT-II) demonstrated no survival benefit to immediate completion lymph node dissection (CLND) after positive sentinel lymph node biopsy (SLNB) when compared to regional nodal surveillance ultrasound (RNS). However, patients with positive non-sentinel lymph nodes (NSLN) undergoing CLND did benefit from improved staging and regional disease control. Criteria to identify subclinical NSLN-positive patients is lacking. M Gene expression profiling (GEP) was prospectively performed on 570 melanomas from 2/2015 to 8/2019, identifying specimens as low-risk Class 1 (GEP1), or high-risk Class 2 (GEP2). Patients who underwent SLNB were stratified by stage and examined for rates of nodal positivity within each GEP class. SLN positive (SLN+) patients who underwent CLND or regional nodal surveillance were also examined for risk factors and outcomes based on GEP classification. R Of 378 patients who underwent SLNB, GEP2 predicted SLN+ when compared to GEP1 (31.9% vs 7.0%; p<0.00001). Of 150 patients 55 years of age with T1 and T2 tumors, 2.5% of GEP1 patients were SLN+, compared to 21.9% of GEP2 patients (p=0.0008). 59 patients were SLN+ (17 GEP1, 42 GEP2), of which 26 underwent CLND and 33 underwent regional nodal surveillance (RNS). Of the 26 patients who underwent immediate CLND after SLN+ with negative staging, positive NSLNs were identified on CLND in 1 of 9 GEP1 patients and 10 of 17 GEP2 patients (11.1% vs 58.8%; p=0.036). Two additional GEP2 patients had positive NSLN on post-operative staging imaging and percutaneous biopsy but refused CLND. Of 23 patients who underwent RNS with at least six months of follow-up, 1 of 5 GEP1 patients and 10 of 18 GEP2 patients had a recurrence (20.0% vs 55.6%; p=0.32). The single GEP1 recurrence was regional; recurrences in GEP2 patients included 4 regional, 4 distant, and 2 local. C GEP may be useful to identify a higher risk of positive NSLN after +SLNB, thereby avoiding unnecessary CLND in GEP1 patients, and inappropriate nodal surveillance in GEP2 patients. Further study is necessary to examine the utility of GEP in risk stratification based on age and tumor thickness.

Validation of a Gene Expression Profiling Assay in Primary Cutaneous Melanoma A.W. Kangas-Dick,* A. Greenbaum, V.A. Gall, R. Groisberg, J. Mehnert, V. Koshenkov, A. Berger. Rutgers Cancer Institute of New Jersey, New Brunswick, NJ. Introduction: A significant proportion of deaths from cutaneous melanoma are patients initially diagnosed with stage I or II disease. The Decision-Dx Melanoma (DDM) 31 gene assay allows patients to be stratified by risk of recurrence. We present the largest single institution series to date demonstrating the accuracy of gene expression profiling (GEP) in predicting melanoma recurrence. Methods: A retrospective chart review was conducted from September 2013 to September 2019 of all patients who underwent surgery for cutaneous melanoma at a large academic cancer center upon whom the DDM had been performed (n=359). Patient demographics, tumor pathologic characteristics, AJCC stage, sentinel node status, GEP class, and recurrence free survival were collected. Results: Median follow-up was 15 months. Mean age was 65 years. 32.5%(n=117) were female and 67.4%(n=242) were male. At presentation, 49%(n=177) of patients had stage I disease, 31.5%(n=113) were stage II, and 18.4%(n=66) had stage III. The majority of primary tumors were located on the extremities (50.1%(n=180), 18.1%(n=65) on the head or neck, and 31.2%(n=112) truncal. Sentinel node biopsy was performed in 71.6%(n=257), and was positive in 53 patients (21%). Overall, 18.4%(n=66) of patients had a recurrence. There were 15 stage IA/IB patients who recurred; of these 7(47%) were GEP class 2B, 2(13%) were class 2A, 1(6%) was class 1B, and 5(33%) were class 1A. Median recurrence free survival for patients in the GEP class 1 group was 66 months vs 45 months in GEP class 2 patients. 5 year RFS was 68% vs 30% (Figure) . Median RFS was 68, 63, 60, and 43 months in GEP class 1A, 1B, 2A, and 2B patients respectively, and 5 year RFS was 74%, 57%, 31%, and 23%. On multivariate regression analysis, age, GEP class, Breslow thickness, tumor regression, and sentinel node positivity were significantly associated with RFS (all p<0.05). Conclusion: Recurrence in stage I/II cutaneous melanoma is frequent, and results in significant morbidity and mortality. Genetic profiling of these tumors is helpful in determining the need for close surveillance and adjuvant therapy in this population, as recurrence risk may be higher than previously estimated. John's Health Center, Santa Monica, CA; 2. Providence HealthCare Center, Portland, OR. Introduction: We recently developed a logistic regression model to predict sentinel lymph node status as the first step to devise a machine-learning algorithm to guide physicians in the personalization of melanoma care. The objective of this study is to create an externally validated melanoma recurrence model to advance the machine-learning algorithm. Methods: A multi-center randomized trial (RCT) database was reviewed to identify patients who underwent SLNB. Covariates of recurrence (local-regional (LR) and distant (M)) evaluated by Cox-regression included gender, age, primary site, ulceration, Breslow thickness, Clark level, +SLNB, histology, lymphatic invasion, microsatellites, lymphocytic infiltration, regression, and mitosis. A Cox-regression model for 5-year recurrence (LR and M) was created, and model performance was assessed with external validation, discrimination (c-index), and calibration (predicted probabilities). Results: Of the 1,640 SLNB patients identified, 530 (32.3%) had a -SLNB and 1,110 (67.7%) had a +SLNB. Breslow thickness and +SLNB were the covariates most likely to be associated with both +LR and +M recurrence while primary site-trunk was least likely to be associated with +M only ( Table 1 ). The models constructed for +LR and +M had good predictive c-indexes (0.736 and 0.730, respectively), calibration, and were externally validated against a multicenter RCT database with 809 (81.7%) -SLNB and 181 (18.3%) +SLNB, indicating high performance. Conclusion: The Coxregression recurrence-model is the last step in the creation of a comprehensive machine-learning algorithm to assist in complex melanoma decisions. This algorithm can help physicians individualize surgical care and predict who should receive adjuvant therapy by providing an individual risk assessment of metastatic recurrence. The utility of the model to impact outcomes should be assessed prospectively.

A.M. Holder, 1 * A. Ziemys. 2 1. Surgery, Houston Methodist Hospital, Houston, TX; 2. Houston Methodist Research Institute, Houston, TX. Introduction: In staging of melanoma patients without clinical nodal disease, the status of the sentinel lymph node (SLN) is the most important prognostic indicator; yet, the false negative rate of SLN biopsy (SLNB) is 15%. Even still, up to 25% of patients with negative SLNB will go on to develop metastatic disease. Therefore, we addressed this need to develop a more accurate method of predicting the metastatic potential of melanoma. Methods: Five melanoma patients without clinical nodal disease who qualified for SLNB underwent intraoperative radiotracer measurements. Repeated measurements (A-C) of gamma counts between the primary tumor and SLN basin were entered into our innovative analytical model to determine the lymphatic transport efficiency through apparent diffusion coefficient, and results correlated with clinicopathologic data. Results: There was no association of lymphatic transport efficiency with age, BMI, or primary tumor location. There was a strong correlation of lymphatic transport efficiency with Breslow depth and SLN status, with the most efficient lymphatic transport correlating with a positive SLN. Although the coefficient of variation was ~80%, the lymphatic transport efficiency of patients with positive SLNB was 50x greater (6.0 vs 0.1 cm 2 /s) than that of patients with negative SLNB (power=1). Conclusion: Lymphatic transport efficiency of radiotracer from intraoperative gamma counts predicted SLN status of patients with clinical Stage I-II melanoma. Our straightforward intraoperative methodology will allow surgeons to rapidly determine a patient's lymphatic transport efficiency to provide personalized prognostic information more accurately and more quickly than SLNB. This method of predicting metastatic potential in melanoma can improve risk stratification and staging, prevent delays in adjuvant treatment, and increase our understanding of the metastatic process, potentially leading to innovative treatments in melanoma and other cancers.

Proteomic Markers of Immune-checkpoint Inhibitorrelatedd Toxicities in Melanoma K. Sierra-Davidson, 1 * A. Mehta, 1 X. Bai, 1 D. Frederick, 1 G.G. Kasumova, 1 M. Kim, 1 M. Rucevic, 2 L. Hultin Rosenberg, 2 D. Lieb, 3 N. Hacohen, 1 K. Flaherty, 1 R. Sullivan, 1 G.M. Boland. 1 1. Department of Surgery, Massachusetts General Hospital, Boston, MA; 2. Olink Proteomics, Watertown, MA; 3. Broad Institute of MIT and Harvard, Cambridge, MA. Background: Therapeutic inhibition of programmed cell death receptor-1 (PD-1) has revolutionized treatment of melanoma. Immune checkpoint inhibitors (ICI) target specific interactions in the immune cascade, ultimately activating tumor-specific T cells to promote tumor destruction. This mechanism of action gives rise to potential immune responses against self-antigens that can lead to the emergence of immune-related adverse effects (irAEs). There is growing evidence that patients who suffer from a subset of tissue-specific irAEs may derive greater benefit from therapy. Methods: We have undertaken a retrospective analysis of 146 melanoma patients treated at the Massachusetts General Hospital, generating detailed response and toxicity clinical data. In parallel, we have generated proteomic data from plasma of these patients via a proximity extension array (Olink®), examining ~1000 markers of protein expression. Samples from before and during treatment have been analyzed. Correlation with response is complete, and comparative analysis with toxicity is underway. Results: Of our entire cohort, 56% had irAEs. The occurrence of irAEs was associated with significantly better objective response rate (59.5% vs. 22.2%) (P<0.001), longer progression free survival (PFS, median, 65.1 vs. 11.4 weeks) and overall survival (OS, median, not reached vs. 74.7 weeks) (all P<0.001; Fig 1) . A subset of tissue-specific irAEs correlated with longer PFS and OS (cutaneous [n=37, 25%], musculoskeletal [n=31, 21%]), while other irAEs did not (colitis [n=21, 14%]). Discussion: Unbiased proteomic profiling and correlation with tissue-specific irAEs may yield insight into shared and exclusive markers of toxicities. We hypothesize that a distinct subset of immune cells is responsible for organ-specific toxicities. We are examining proteomic profiles of three groups of patients: (1) cutaneous and musculoskeletal toxicities (hypothesized shared mechanism of toxicity and response), (2) colitis (hypothesized distinct mechanism of toxicity), and (3) those with no toxicities. We have identified proteomic markers of ICI response (data not shown) and are now utilizing this data set to nominate immunologic markers of irAEs. Figure 1 : A) Overall survival in melanoma patients who did and did not experience immune-related adverse events (irAEs); B) Correlation of tissue-specific irAEs with objective response rate (ORR), progression free survival (PFS) and overall survival (OS).

Subtype Predominant and Prognostic Related C. Li,* G. Zhang, Y. Wang, B. Chen, L. Guo, K. Li, L. Cao, C. Ren, L. Wen, M. Jia, H. Mok, X. Li, N. Liao. Department of Breast cancer, Guangdong Provincial People's Hospital, Guangzhou, China. Background: MAP3K1 is a serine-threonine kinase in the mitogenactivated protein kinase (MAPK) family involved in the frequent mutation of human cancers. Regarding the fact that the correlation between MAP3K1 mutation and clinicopathological characteristics and prognosis in breast cancer remains unclear in Chinese population, we perform a retrospective study in our center with an attempt to explore the possible role and function of MAP3K1 in breast cancer. Methods: Data obtained from 412 consecutive breast cancer patients were selected from Guangdong Provincial People's Hospital (GDPH) in this study. Mutations were tested by next generation sequencing (NGS). Relationship between MAP3K1 mutations and clinicopathologic features were analyzed to further compare with populations in Molecular Taxonomy of Breast Cancer International International Consortium (METABRIC) cohort and The Cancer Genome Atlas (TCGA) cohort. Results:In GDPH cohort, MAP3K1 was mutated in 8.9% (n=37) of 412 cases with 45 incidences, compared to 9.4% (n=337) in METABRIC and 8.3% (n=133) in TCGA. The majority of the mutations were truncating mutation, followed by missense mutation in three cohorts. Alterations of MAP3K1 were predominant in the luminal A, followed by luminal B, breast cancer subtypes in both TCGA and METABRIC data sets (p<0.001), but we observed no significant difference in GDPH (p=0.288) . In addiction, pathologic grade was lower in MAP3K1 mutant group in GDPH cohort. In METABRIC cohort, patients with MAP3K1 mutation had a favorable overall survival (OS) when compared to those without MAP3K1 mutation (P=0.006). When focusing on HR+ breast cancer, we found a more significant favorable OS in MAP3K1 mutation group (P<0.001). Moreover, MAP3K1 alterations in breast cancer are mutually exclusive with those of MAP2K4 and partially overlap with PIK3CA. Conclusions: Our findings demonstrate the MAP3K1 mutations of breast cancer in Chinese population and further compares significant MAP3K1 mutation differences in race and ethnicities distribution. MAP3K1 mutations are mainly detected in HR+/ HER2-breast cancer and has a possible prognostic effect.

Somatic mutation of gene MPA3K1 mutations in breast cancer. Green dots represent missense mutations, while black dots represent truncating mutations.

Young Age is an Independent Predictor of Chemotherapy Recommendation in Breast Cancer E. Graham,* D. Boselli, A. Heckshner, A. Voci, D. Sarma, M. Forster, T. Sarantou, R.L. White, L. Hadzikadic-Gusic. Breast Surgical Oncology, CMC Atrium Health, Charlotte, NC. Introduction: Oncotype DX (oDX) is a 21 gene expression assay validated to predict distant recurrence and benefit of chemotherapy for patients with early stage breast cancer. Although chemotherapy benefit is determined by tumor molecular biology, younger age is associated with variation in chemotherapy administration. We examined whether age was an independent predictor for chemotherapy recommendation (CR) regardless of oDX recurrence score (RS). Methods: The National Cancer Database was queried for patients with primary breast cancer from 2010 to 2016 with documented oDX RS. Analysis included female patients 18 years with hormone positive, HER2/neu negative, pathologic T1-T3/node negative tumors; patients with metastatic disease, borderline hormone status, and neoadjuvant chemotherapy were excluded. Younger age was defined as 50. oDX RS was considered low (0-10), intermediate (11) (12) (13) (14) (15) (16) (17) (18) (19) (20) (21) (22) (23) (24) (25) [1.01, 1.02] ). Analysis by decade found odds of CR decreased for all sequential decades in the intermediate and high RS groups 30 years, and in the low RS group until 50 years (p<0.0001). Conclusion: Younger age at diagnosis is an independent predictor for increased CR. Within the low oDX RS group analyzed by decade, age significantly impacts odds of CR until age 50. Despite molecular profiling, age is a risk factor for discordant use of oDX for treatment strategies in breast cancer. Background: The average time from diagnosis to surgery (TTS) among breast cancer patients has been steadily increasing since the 1990s. While no clinical guideline currently exists for allowable TTS, studies have indicated that delays of surgery negatively impact survival of breast cancer patients. Therefore, we investigate the association between TTS and upstaging in tumor size (T-upstaging), nodal status (N-upstaging), and stage group. Methods: Using the National Cancer Database, we analyzed locoregional upstaging among women diagnosed with clinical stage I breast cancer between 2010 and 2015. Adjusted logistic regression models were used to examine the relationship between upstaging (stage group, T-upstaging, and N-upstaging) for pre-and post-menopausal women (i.e. <50 years old, >50 years old). Results: Pre-menopausal women had a higher proportion of group upstaging than post-menopausal women (22% vs. 17%). Multivariable analysis revealed a significant association between increasing TTS and upstaging. Compared to TTS <30 days, TTS over 90 days was found to have a significant association with T-upstaging (OR 1.3; 95%CI 1. , and stage group (OR 1.2; 95%CI 1.1-1.3) among postmenopausal women, while only T-upstaging (OR 1.3; 95%CI 1.0-1.7) was significant among premenopausal women. A differential pattern in the relationship between longer TTS and upstaging was also seen by hormone receptor status. Tumor characteristics including grade (moderate to poor differentiation) and histology (non-ductal) were significantly associated with upstaging of any kind as well as African American race and non-breast conserving surgery. Conclusions: Among postmenopausal women, TTS > 90 days remained significantly associated with tumor size, node involvement, and group upstaging after adjusting for clinical and sociodemographic characteristics. Among premenopausal women, this relationship persisted only for tumor size upstaging.

Suboptimal Treatment After Lumpectomy in Triple-negative and Her2+ Breast Cancer: Age Bias G.C. de Oliveira,* J. Johnson, E. Navajas, P.M. Spanheimer, K.K. Gallagher, D. Ollila, p. strassle, S. Downs-Canner. Surgery, University of North Carolina Chapel Hill, Chapel Hill, NC. INTRODUCTION: In patients with Stage II triple-negative (TN) and Her-2 positive (Her2+) breast cancer (BC) undergoing lumpectomy as first treatment, adjuvant radiation and chemotherapy or chemotherapy plus targeted therapy are standard. We hypothesized that disparities exist in the adjuvant treatment in women with early TN and Her2+ BC. METHODS: We identified women 40 and older with pT2N0 TN or Her2+ BC treated with lumpectomy first in the National Cancer Database from 2012-2016. We excluded those treated with any neoadjuvant therapy or with missing treatment data. Multivariable log-binomial regression was used to identify factors associated with receiving optimal treatment (OT) (surgery plus adjuvant systemic therapy and radiation) compared to sub-optimal (SO) treatment (any variation of OT). Systemic therapy included chemotherapy and/or Her2 targeted therapy. RESULTS: Of the 9,716 patients in the study 67.8% received OT. The majority of patients had ductal carcinoma (OT 86%, SO 80%) and were TN subtype (OT 70%, SO 62%). OT patients were significantly younger (median [IQR] 59 [52, 66] vs 71 [60, 81] years, p<0.0001); older age was associated with decreasing rates of OT with 76%, 53%, and 17% of patients 60-69, 70-79 and 80-89 receiving OT, respectively. Of the 2279 (33% of cohort) patients who did not receive systemic therapy, 44% were not offered therapy, 41% refused, and 13% had a contraindication. Of the 1815 (18.7% of cohort) patients who did not receive radiation, 47% were not offered therapy, 44% refused and 6% had a contraindication. After adjustment, only age was associated with receipt of SO treatment; race, insurance status, median income, educational attainment, cancer center accreditation, region, and distance to care did not have an effect on SO treatment. CONCLUSION: In women with T2N0 TN or Her2+ BC treated by lumpectomy first, 1/3 of patients did not receive OT. Besides older age, no socioeconomic or demographic factors contributed to SO treatment. Patients who had SO treatment most commonly were not offered or refused OT. Incorporating assessments of life expectancy beyond chronological age and patient/ provider education may improve delivery of OT. Introduction: The ACOSOG Z0011 trial has altered the routine use of axillary lymph node dissection (ALND) in women with breast cancer. The National Cancer Care Network still recommends completion ALND in women with positive sentinel lymph nodes (SLN) unless they fit the strict Z0011 criteria or only have micrometastases. This analysis aimed to assess utility and survival benefit of ALND in clinical T3/T4 node negative patients. Methods: An IRB approved study evaluated patients from the National Cancer Database from 2006-2016. Women with clinical T3 and T4 tumors that otherwise fit the original Z0011 criteria were identified in the National Cancer Database. Exclusion criteria included neoadjuvant systemic therapy and clinical nodal disease. Patients were grouped into SLNB alone (1-5 lymph nodes examined) and ALND ( 10 lymph nodes examined). Results: A total of 230 women were identified who fit inclusion criteria with 36% undergoing ALND. Clinicopathologic characteristics were similar between groups. On Cox regression analysis, increased age (HR 1.05, p<0.01) and ER-breast cancer (HR 5.55, p<0 .01) were associated with an increased risk of death. On Kaplan-Meier survival analysis there was no survival benefit for ALND compared to SLNB alone. Subgroup analysis of triple negative, HER2+ and ER+ breast cancers also failed to show a survival benefit of ALND. Conclusion: Currently, NCCN recommends axillary lymph node dissection in node positive patients with macrometastases or those who are outside the ACOSOG Z0011 criteria.

However, our findings indicate that the Z0011 results may be applicable in women with larger tumors, and it may be reasonable to omit ALND in women who otherwise fit Z0011 criteria.

A Contemporary Analysis of Breast Angiosarcoma from the SEER Database A.U. Friedrich,* E. Reisenbichler, J.M. Le Blanc, T.S. Park, B.K. Killelea, D.R. Lannin. Yale New Haven Hospital, New Haven, CT. Introduction: Angiosarcoma of the breast can occur de novo or after treatment of breast cancer. Our objective was to analyze differences between primary and secondary angiosarcoma, as well as risk factors for development of secondary angiosarcoma after breast cancer treatment. Methods: We searched SEER 18 (Surveillance, Epidemiology, and End Results) database for patients with angiosarcoma of the breast, trunk, shoulder and upper arm from 1974 to 2016. Primary and secondary angiosarcoma cases were identified and compared. Secondary tumors were then matched with the patient's prior breast cancer and differences among patients who did and did not develop angiosarcoma were compared. Results: 904 patients with angiosarcoma were identified. Of these, 313 (34.6%) were primary tumors. Patients with secondary tumors were older (71 vs 52 years), more likely to be white (91.2 vs 80.5%, p<0.001), and to have regionally advanced disease (28.2 vs 10.5%, p<0.001) or tumors of higher pathologic grade (Grade 3: 72.5 vs 44.1%, p<0.001). Of the 591 secondary angiosarcoma patients, 420 patients had previously been diagnosed with breast cancer. Mean time to develop angiosarcoma was 8.2 years with 78% of tumors occurring by ten years. Among women with breast cancer and at least ten years of follow-up, patients who developed angiosarcoma were more likely to be white (90.0% vs 85.0%, P=.006); to be older at diagnosis of breast cancer (65.1 vs 57.5 years P<0.001); to have an invasive breast tumor (94.3 vs 80.8%, p<0.001); to have more than 11 lymph nodes removed (23.3 vs 17.5%, P<0.001); to have been treated with lumpectomy (92.7 vs 56.3%, P<0.001) and radiation (83.3 vs 44.0%, p<0.001); and to be ER positive (83.6 vs 79.3%, p=0.049) or PR positive (75.2 vs 69.2%, P=0.017). In multivariate analysis, independent risk factors included white race (p=0.039), invasive tumor (p=0.028), number of lymph nodes removed (p<0.001), lumpectomy (p<0.001), radiation treatment (p<0.001), and older age (p<0.001). Conclusion: Epidemiological, clinical, pathological and treatment factors help distinguish primary from secondary angiosarcoma and contribute to a breast cancer patient's risk of developing angiosarcoma in the future.

Predicting Positive Margins in Pancreatic Head Adenocarcinoma Following Neoadjuvant Therapy: Investigating Disparities in Quality Care Using the National Cancer Database C. Suraci, 1 * K. Young, 2 J. Dove, 2 J. Blansfield. 2 1. Geisinger Commonwealth School of Medicine, Clarks Summit, PA; 2. Geisinger, Danville, PA. INTRODUCTION: In pancreatic cancer, surgical resection with neoadjuvant therapy improves survival, but survival relies significantly on the margin status of the resected tissue. This study aimed to develop a model that predicts margin positivity, and then to identify facility-specific factors which influence the observed-expected (O/E) ratio for positive margins among facilities. METHODS: This was a retrospective review of patients with pancreatic head adenocarcinoma (TNM Stage I and II) who received neoadjuvant therapy for a pancreaticoduodenectomy within the National Cancer Database (2004 Database ( -2016 . Logistic regression was used to develop a model that predicts margin positivity, and this model was used to identify outlier facilities with regards to the O/E ratio. Hospital volume was defined as total number of pancreaticoduodenectomies per year. RESULTS: A total of 4,085 patients were included, with 16.8% of patients having positive margins. Most patients (64%) had a tumor size between 2-4cm and approximately 51% of patients did not have positive lymph nodes upon resection. A logistic regression model showed being male, having a larger tumor size, and having positive lymph nodes to be predictors of positive margins following resection with neoadjuvant therapy. This model was validated to yield a bootstrap-corrected concordance index of 0.632 O/E ratios were calculated with the model, identifying 12 low-and 17 high-O/E outlier facilities among 401 studied hospitals. There was no difference in facility type (i.e. academic vs integrated network) between outlier hospitals, but there was a significant difference between yearly hospital volume (Low-outlier 20.6 vs high-outlier 10.7, p=.008). CONCLUSIONS: A disparity in care was found due to lower volume facilities being associated with higher than expected rates of positive margins. This disparity was identified by the use of an O/E ratio as a quality indicator for facilities. Facilities can gauge the efficiency of their own practices by referencing their own O/E, and they can also improve their practices by analyzing the framework of low O/E facilities. While patient-specific estimates of perioperative complications are critical in the operative decision-making for cancer patients, the SRC's accuracy has yet to be validated in the general surgical oncology population. We sought to determine the performance accuracy of the SRC in cancer patients. Methods: We performed a retrospective study of consecutive patients who underwent an elective inpatient surgical oncology operation at a Comprehensive Cancer Center during 2018. Complications were identified through administrative data and surgeon self-reporting and then were reviewed by the research team. For each patient who underwent an operation, the 21 SRC patient-specific factors were entered in the online SRC and a risk level and outcome probability were generated. The predictions from the online SRC were compared to the actual patient outcomes to assess the predictive performance of SRC in this cancer patient population. Results: A total of 402 operations and 541 procedures in 384 patients were included. At least one of the SRC defined complications were identified in 98 of the operations (24.3%). There were 4 deaths within 30 days of the index operation (0.9%). Table 1 describes the disease site and stage distribution in our cohort. Overall, the SRC had an Area Under Receiver Operating Curve (AUROC) of 0.51 in predicting any complication using categorized risk levels of "Average," "Above Average," or "Below Average." The SRC had an AUROC of 0.63 in predicting any complication when using outcome probabilities. Conclusion: The SRC performs poorly in predicting risk of complications in the general surgical oncology population. Cancer-specific variables, such as stage and preoperative therapy, as well as adjusted variable weights may improve the performance of the SRC. BACKGROUND: Surgical emergenices in metastatic patients are increasingly common. Selecting appropriate management is crucial to palliating symptoms and maximizing quality of life. The purpose of this study was to describe the outcomes of medically and surgically treated Stage IV cancer patients presenting with malignant bowel obstruction, with the aim of developing a prediction model to identify patients who would most benefit from palliative bowel surgery. METHODS: This is a retrospective, single-center study, evaluating clinical characteristics and outcomes of any metastatic cancer patient admitted to The Ottawa Hospital with MBO from 2008-2017. The Mann-Whitney-U test was used to compare continuous variables, the Fischer'sexact test for dichotomous variables, and the Log-rank (Mantel-cox) for comparison of survival curves. In the surgical group, a stepwise logistic regression identified predictors of death within 3 months following surgery for MBO. RESULTS: Of the 465 patients identified, 170 were managed surgically and 295 medically. The mean survival time was 5.65 12.4 months in the medical group, compared to 13.7 17.8 months in the surgical group (p<0.001). We created a model to explain 45.1% of variability in outcome following surgery. Variables in this model are age; ASA score; albumin; lung primary; hepatobiliary primary; ascites; lymph node, liver, lung or peritoneal metastases; the presence of greater than three sites of metastases; proximal colon obstruction; multiple sites of obstruction; and medical management prior to surgery (R-squared=0.451). Independent predictors of death within 3 months were albumin, and greater than three sites of metastatic disease. CONCLUSION: We identified multiple variables that contribute to death following surgical intervention in patients admitted with MBO. Our results identified albumin and more than three sites of metastatic disease as independent predictors of death within three months of surgical intervention. This is the preliminary step to creating a model for prediction of outcome following palliative bowel surgery in MBO, which will assist in every-day decision making. Background Interest in cytoreductive surgery for advanced intraperitoneal malignancies continues to rise. Many studies have reported significant morbidity from these operations, yet also with variation or heterogeneity in practice patterns. Whether utilization of the usual hospital quality measures of perioperative outcomes are applicable in this patient population is unknown. Methods Cytoreductive operations performed for cancer between January 1, 2013 and June 30, 2018 were identified in the American College of Surgeons National Surgical Quality Improvement Program registry (via CPT codes such as 49203, 49204, 49205). Risk-adjusted hospital-level variation in 30-day death, serious morbidity, reoperation, readmission, and a composite of death or serious morbidity (DSM) were evaluated using hierarchical models. National Cancer Institute (NCI)-designated Cancer Center (NCI-CC) status was also explored. Results A total of 6,125 operations were included across 588 hospitals, 56 of which were NCI-CCs. Overall rates of death, serious morbidity, reoperation, readmission, and DSM were 1.5%, 12.8%, 3.5%, 8.6%, and 13.3%, respectively. Hierarchical covariance parameters for death, serious morbidity, reoperation, readmission, and DSM were 0.24 (p=0.28), 0.04 (p=0.19), 0.21 (p=0.05), 0.002 (p=0.47), and 0.04 (p=0.17), respectively. When compared to other hospitals, NCI-CCs had better risk-adjusted 30-day mortality (p<0.001), but not for serious morbidity (p=0.13), reoperation (p=0.21), readmission (p=0.60), or DSM (p=0.14). Conclusion In these data, hospital-level variation was unable to be detected using the usual measures of perioperative outcomes. Given the morbidity associated with cytoreductive surgery, this demonstrates a clear opportunity to better improve the way quality is measured for patients undergoing cytoreductive operations for cancer. Purpose: Cancer treatment is a significant driver of healthcare costs worldwide, however, the economic impact of treating patients with anti-neoplastic agents is poorly elucidated. Hence, we conducted a systematic review and meta-analysis to estimate the direct costs associated with administering intravenous chemotherapy in an outpatient setting. Methods: We systematically searched four databases from 2010 to present and extracted hourly administration costs and the respective components of each estimate. Separate analyses were conducted of Canadian and United States (US) studies, respectively, to address a priori hypotheses regarding heterogeneity amongst administration cost estimates. The Drummond checklist was used to assess risk-of-bias. Data were summarized using medians with interquartile ranges and five outliers were identified; costs were presented in 2019 USD. Results: A total of 44 studies were analyzed, including sub-analyses of 19 US and seven Canadian studies. 26/44 studies were of moderate or high quality. When components of administration cost were evaluated, physician costs were reported most frequently (24 studies), followed by lab tests (13) and overhead costs (9) . The median cost estimate when outliers were excluded was $142/hour (IQR=$103-166). Sensitivity analyses determined the median administration cost in the US was $149/hour (IQR=$118-158), and was $128/hour (IQR=$102-137) in Canada. Conclusions: There is currently a paucity of literature addressing the costs of chemotherapy administration, and existing studies utilize a patchwork of reporting methodologies which renders direct comparison challenging. Our results demonstrate that the cost of administering chemotherapy is approximately $125-150/hour, globally. This value is dependent upon the region of analysis, inclusiveness of cost subcomponents as well as the methodology used to estimate unit prices, as described here. Background: Patients undergoing oncologic resection experience more than one risk factor attributing to higher rates of venous thromboembolism (VTE) and this can lead to increased morbidity and higher hospital costs. Frequently, low molecular weight heparin (LMWH) is recommended as extended thromboprophylaxis and has been shown to reduce rates of VTE. However, recent literature suggests that ambulatory cancer patients have significantly lower rates of VTE with the use of apixaban therapy. Methods: This study is a retrospective review of consecutive patients undergoing resection for oncologic indications at a single institution from May 2016 to May 2019. This study evaluated the use of apixaban (2.5mg twice daily for 1 month) as extended thromboprophylaxis (ETP) at discharge. The primary outcomes were deep vein thrombosis (DVT), pulmonary embolism (PE), or mesenteric/portal venous thromboembolism at 30, 60, and 90 days post-operatively. Results: A total of 601 patients underwent resection for oncologic indications, 450 patients received no ETP at discharge and 151 patients received apixaban. Patients discharged on therapeutic anticoagulation or LMWH were excluded. At 90 days, VTE occurred in 6.1 % (26/428) of patients without ETP and 4.1% (6/145) of patients discharged with apixaban (p=0.380). More specifically, PE occurred in 1.1%, 1.9%, and 2.8% of patients without ETP and 0%, 0%, and 0.7% of patients in the apixaban group (at 30, 60, and 90 days, respectively). DVT occurred in 1.8%, 2.3% and 2.8% of patients without ETP and 0%, 0%, and 1.4% in the apixaban group.

Re-admission for bleeding occurred in 2.8% of patients without ETP and 3.3% of patients in the apixaban group. Including all VTE at 30 days, the total cost per 100 patients was US $68,171 in the apixaban group compared to $67,620 in the group without ETP; apixaban costs $551 more for every 100 patients. Conclusion: Apixaban therapy as ETP did not result in a statistically significant reduction in VTE when compared to no ETP in patients undergoing oncologic resection. However, given the substantial cost of VTE, even a small reduction in VTE rates with apixaban could be considered cost effective. Introduction: Colonoscopy is widely used to screen for colorectal cancer but its effectiveness varies based on the quality of the procedure. In small and rural hospitals, colonoscopy is often performed by providers with varied endoscopy training. Our objectives were to describe colonoscopy quality and identify factors associated with colonoscopy quality in rural and underserved hospitals. Methods: Trained abstractors from hospitals in rural and underserved areas participating in a statewide surgical collaborative prospectively collected data for validated colonoscopy quality metrics from 2016-2019. Provider characteristics were obtained from the AMA Masterfile. Quality measures were benchmarked with established gastroenterology multi-society guidelines. Patient and provider characteristics associated with colonoscopy quality metric performance were identified. Results: A total of 4,219 colonoscopies (1,866 for screening) were performed at 8 hospitals. Of 31 providers, 19 were general surgeons (GS), 10 were gastroenterologists (GI), and 2 were family practitioners (FP). Individual procedure volume ranged from 15 to 540 (median = 93; IQR 45-168). In screening colonoscopies (Table 1) , adequate bowel preparation was documented in 90.6% (goal 85%), cecum photodocumentation was recorded in 93.1% (goal 95%), mean withdrawal time was 8.1 minutes (goal 6), and adenoma detection rate (ADR; the best measure of colonoscopy effectiveness) was 26.6% (goal 25%). On multivariate analysis, there was no association between ADR and bowel preparation (AOR 0.87, 95% CI 0.58-1.30) or cecum photodocumentation (AOR 0.63, 95% CI 0.36-1.08), but increased ADR was associated with physician specialty (36.9% GI vs. 22.5% GS; AOR 2.30, 95% CI 1.40-3.77). Conclusion: Physicians performing colonoscopy at rural and underserved hospitals meet most national quality metrics. However, there were differences in quality among specialties and there were higher and lower quality providers in all specialties. As these providers are often the sole option for rural and underserved patients, providing opportunities for individualized and targeted educational interventions are needed for lower quality providers. Consortium(2007-17) was reviewed for pts with primary rectal adenocarcinoma who underwent R0/R1 low anterior resection(LAR) or abdominoperineal resection(APR). 90 day POCs were categorized as major vs minor and grouped into infectious, cardiopulmonary(CP), thromboembolic(TE), renal, or intestinal dysmotility. Primary outcomes were 5 yr overall survival(OS) and recurrence-free survival(RFS). Results Of 1136 pts, median age was 59yrs(IQR51-67), 61% were male(n=693), median f/u was 31 mos(IQR 13-54). 70% underwent LAR(n=799) and 30% APR(n=337). Complication rate was 46%(n=527), with 63% minor(n=330) and 32% major(n=170). Of all POCs, infectious complications comprised 20% (n=105), cardiopulmonary 3%(n=14), thromboembolic 5%(n=25), renal 9%(n=46) and intestinal dysmotility 19%(n=100). When compared to minor or no POCs, major POCs were associated with both worse RFS(48vs63vs76% p<0.01 Fig1A) and OS(64vs76vs80% p<0.01 Fig1B). While a single POC was associated with worse RFS(61vs76% p<0.01), multiple POCs were associated with worse OS(62vs79% p=0.02). Regardless of complication grade, infectious POCs were associated with worse RFS(56vs76% p<0.01 Fig1C) while CP and TE POCs were associated with worse OS(CP 40vs78% p<0.01; TE 63vs78% p<0.01). Postoperative renal dysfunction was associated with both worse RFS(26vs76%, p<0.001 Fig1D) and OS(62vs78% p=0.01). This persisted on MV analysis for OS when accounting for pathologic stage, receipt of neoadjuvant therapy, and final margin status(CP: HR 3.6 p=0.01; TE: HR 19.4 p<0.01; renal: HR 2.4 p=0.01) and for RFS(infectious: HR 2.1 p<0.01; renal: HR 3.2 p<0.01). Conclusions Major complications after proctectomy for cancer are associated with decreased recurrence-free and overall survival. Given the association of infectious complications and postoperative renal dysfunction with earlier disease recurrence, efforts must be directed towards defining best practices and standardizing care. Introduction: Circumferential resection margin (CRM) has been shown to be one of the most important determinants of local control in patients with rectal cancers. However, the extent to which CRM involvement exists after colon cancer surgery within the United States is unknown. The aim of this study was to identify rates of CRM positivity and factors associated with this following colon cancer resection. Methods: Colon cancer cases with surgical resection between 2010 and 2015 were identified from the National Cancer Data Base. CRM status was the primary outcome. Comparisons were made between cases with or without positive CRM ( 1 mm) using chi-squared tests and multivariable logistic regression. Hospital-level analysis was then performed, adjusting for case mix using observed-to-expected (O/E) ratios to determine facility performance. Results: A total of 170,022 cases were identified: 150,291 cases with negative and 19,731 with positive margins (11.6% positive CRM rate). Overall, pathologic T-category was the greatest predictor of positive CRM, with higher positive CRM rates in pathologic T4 (25.8%), T4A (24.7%) and T4B (31.5%) versus T1(4.5%), T2(6.3%) and T3 ( S. colon cancer cases is high. T-category is an independent predictor. In T4 cancers, the lowest rates of CRM positivity are achieved in high volume academic/research facilities. While tumor biology is a major factor, CRM represents an area of needed focus for improvement in quality of care. A coordinated approach with multimodality therapy and centralization of locally advanced cases may drive improvement. Introduction Hospital readmission after colorectal cancer resection results in higher cost of treatment and is associated with higher risk of one-year mortality. The potential impact of readmission on patient outcomes could be an incentive for health care providers to reduce readmissions. Therefore, the aim of our study was to determine the rate, etiology and risk factors associated with 30-day unplanned readmission after colorectal cancer resection, and evaluate the impact of readmission on one-year overall survival and tumor recurrence rates. Methods A retrospective review of colorectal cancer patients who underwent curative intent tumor resection between 2004 and 2016 at our institution was conducted. Patients who had stage IV cancer were excluded. Those who died at index admission were excluded from readmission and tumor recurrence analysis. Logistic regression was used to determine predictors of readmission, one-year overall survival and tumor recurrence. Results A total of 248 patients were included in the study. Of these, 12 (4.8%) patients died at index admission. 30-day unplanned readmission occurred in 16 (6.8%) patients. Developing any post-operative complication was independently associated with higher risk of readmission (Odds ratio 6.09; 95% CI 1.29-28.71, p= 0.022). However, one-year overall survival and tumor recurrence were not significantly associated with readmission. In multivariate analysis of patient outcomes, higher risk of tumor recurrence was observed in patients with surgical site infection (OR 3.212; 95% CI 1.224-8.428, p = 0.018), post-operative blood transfusion (OR 3.066; 95% CI 1.145-8.208, p = 0.026) and higher tumor invasion, (OR 4.619; 95% CI 1.863-11.452, p = 0.001). Factors that significantly predicted poor overall survival were surgical site infection (OR 0.043; 95% CI 0.003-0.618, p = 0.021), emergency surgery (OR 0.028; 95% CI 0002-0.315, p = 0.004) and higher tumor invasion (OR 0.056; 95% CI 0.006 -0.525, p = 0.012). Conclusion Our study found no significant association between readmission and patient outcomes. However, interventions to reduce postoperative complications are key to reduce readmission. INTRODUCTION: Despite 40 published RCTs, there is controversy regarding surgical site infection (SSI) rates for oral antibiotics (OA) and mechanical bowel preparation (MBP) prior to elective colon surgery. Guidelines vary, with recommendations for OA, OA+MBP, and no MBP. There is conflicting evidence for studied groups, and no randomised evidence comparing OA with no MBP. The National Surgical Quality Improvement Program (NSQIP) has increased quality and efficiency of real-time data collection. The REthinking Clinical Trials (REaCT) platform has improved enrollment for RCTs that compare standards of care. In this novel, multi-centre, RCT protocol, we will compare OA with no MBP for SSI rate in elective colon surgery. With clinical trial enrollment at <10% of eligible patients, this RCT is the first to streamline recruitment and data collection by combining REaCT with NSQIP. METHODS: We predict a 50% relative risk reduction of SSI, with an improvement in secondary outcomes of length of stay, hospital costs, and quality of life, with no increase in C. Difficile or antibiotic-resistant infections. A cost-utility analysis, and quality of life assessment will be performed. The analysis will be both intention-to-treat and per-protocol. Consent will occur verbally during a single clinical visit. Permuted block randomisation, using a web-based app, will occur at the same visit. This integrated consent and randomisation model is a hallmark of the REaCT platform. Data collection will be streamlined by real-time collection at NSQIP centres. RESULTS: In an 8-month feasibility study, 67/73 eligible patients (92%) were enrolled, an efficiency benchmark rarely met. 63/67 enrolled patients (94%) completed consent, enrollment, and randomisation during a single clinical visit, reducing cost and attrition. 2/67 patients (3%) were non-compliant. CONCLUSION: This is the first prospective RCT protocol that combines real-time NSQIP data collection and the expedited REaCT recruitment and randomisation platform. This unique design can easily be applied to other surgical interventions to improve enrollment and data collection.

Trial enrollment flowchart highlighting the REaCT integrated oral consent, web based randomisation, and real-time data collection through NSQIP. BACKGROUND: Cytoreductive surgery (CRS) followed by intraperitoneal chemotherapy (IPC) may prolong survival for select patients with appendiceal and colorectal peritoneal carcinomatosis (PC). However, little has been reported regarding outcomes following incomplete cytoreduction. METHODS: Patients who were explored with the intention of optimal CRS but had incomplete CRS (CC 2/3) for appendiceal and colorectal PC between 2008 and 2015 at a single academic center were reviewed. Demographics, surgical details, perioperative complications, and overall survival (OS) were analyzed. Subgroup analysis was performed comparing well-differentiated appendiceal cancer (WD), moderate/poorly-differentiated appendiceal cancer (M/PD), and any colorectal cancer (CRC). RESULTS: Fifty-eight patients were included (11 CC2, 47 CC3). Patients in the WD group were less likely to receive neoadjuvant systemic chemotherapy and had longer operative times compared to patients with M/PD or CRC. The estimated 3-year OS from diagnosis for the WD, M/PD, and CRC subgroups was 86%, 33%, and 11%, respectively (p=0.010). Twenty (11.6%) patients required at least one subsequent palliative procedure, and the time to that intervention was significantly longer for the WD group compared to the M/PD or CRC groups. Grade 3-4 complications occurred in 5.8% of patients and were not associated with a difference in OS. Four percent (4%) of patients received palliative IPC postoperatively and those patients had similar overall survival compared to those who did not receive postoperative IPC. CONCLUSION: Incomplete CRS is associated with low morbidity, a low incidence of subsequent palliative procedures, and better than expected overall survival. Oncologic outcomes correlated with histologic subtype; patients in the WD group had superior outcomes. These data may help guide expectations in the setting of incomplete CRS. Background Currently, trials are investigating active surveillance in esophageal cancer compared to standard esophagectomy after neoadjuvant chemoradiotherapy (nCRT). If non-inferiority is reported, patients will be imposed on the choice between active surveillance or immediate esophagectomy. The aim of this study was to identify subgroups of patients with different clinical and tumor characteristics that could potentially benefit from such an active surveillance strategy. Methods HRQOL was measured using EORTC-QLQ-C30 and QLQ-OES24 questionnaires prior to nCRT and three, six, nine and twelve months postoperatively. Subgroups were defined from patients with different preoperative and predefined clinical (global HRQOL, WHO-status) and tumor characteristics (histology, disease stage and location of the tumor). High and low global-HRQOL were defined as global health scores 75 and <75, respectively. Cohen's d effect-sizes were determined, 0.5-0.8 was considered a medium and >0.8 considered a large effect. The primary endpoints were physical functioning and eating problems. Secondary endpoints were global HRQOL, fatigue and emotional problems. Results In total, 363 patients received HRQOL questionnaires. All subgroups showed impaired HRQOL up to twelve months postoperatively for both physical functioning and fatigue with a medium to large Cohen's d effect size (P<0,004). No difference were found between subgroups except that patients that reported a high global HRQOL prior to nCRT had a significantly worse deterioration in HRQOL compared to patients reporting a low global HRQOL prior to nCRT on all endpoints except for physical functioning (Cohen's d 12 months postoperatively: 0.85 for eating problems, -1.25 for global HRQOL, 0.75 for fatigue and 0.65 for emotional problems). Conclusion All predefined subgroups reported impaired HRQOL for both physical functioning and fatigue up to twelve months after surgery. The results suggest that patients that report a high global HRQOL before neoadjuvant chemoradiotherapy may benefit more from active surveillance if non-inferiority for active surveillance is established.

Analysis of 393 Patients Y. Hu,* E. Vos, M. Schattner, D.G. Coit, V. Strong. Surgical Oncology, Memorial Sloan-Kettering Cancer Center, New York, NY. INTRODUCTION: Gastrectomy is integral to the treatment of gastric cancer, however, long-term quality of life outcomes are sparse within Western populations. The purpose of this study was to identify factors associated with quality of life after gastrectomy. We hypothesized that postoperative quality of life recovery would be associated with extent of resection. METH-ODS: Patients anticipated to undergo gastric cancer resection were prospectively invited to complete the EORTC-QLQ C30 and STO22 surveys in the preoperative setting and at 1-3 months (early), 4-8 months (intermediate), and 9-18 months (late) following resection. Patients who underwent wedge resection and those who did not complete both a preoperative and early postoperative survey were excluded. Quality of life recovery for each patient was measured as paired differences between postoperative and preoperative survey results. Multivariable linear regression was used to identify factors associated with degree of change in quality of life following resection. RESULTS: Across participants (N = 393), median age was 65 and 42% were female. Response rates at the intermediate and late postoperative time points were 58% (N = 228) and 71% (N = 277), respectively. Relative to baseline, global health scale decreased in the early (-15.0 pts, p <0.001) and the intermediate (-4.8 pts, p = 0.005) postoperative surveys, but recovered by the late time point (-0.8 pts, p = 0.606). Relative to distal/subtotal gastrectomy, proximal/total gastrectomy was associated with worse recovery in both the short-and long-term (Table 1) . Older patients and those with locally-advanced tumors (T3-T4) had lower preoperative quality of life scores, and more readily recovered to baseline. Surgical complications were associated with worse postoperative recovery. CONCLUSIONS: Following major gastrectomy for cancer, most patients recover to baseline function within 1 year. Patients with locally-advanced tumors tend to have poorer baseline quality of life which may be partially improved through resection. When feasible, avoidance of a total or proximal gastrectomy is associated with long-term quality of life. Background Gastric outlet obstruction (GOO) in gastric adenocarcinoma (GA) represents advanced disease, and the optimal management approach is unclear. The purpose of this study was to describe the presentation of GOO, evaluate management approaches, and examine outcomes in this patient population. Methods GA patients who presented with GOO to our institution in January 1995 -April 2019 were identified from a prospectively maintained institutional database. Clinicopathologic data were obtained from the electronic medical record. The GOO management approach and clinical course of each patient were retrospectively reviewed. Cox regression and Kaplan-Meier survival analyses were used to examine effects of clinical factors and management strategy on overall survival (OS). Results We identified 60 patients with GOO secondary to GA. Mean age was 64y; 57% were male. At presentation, 22% of patients had imaging evidence of metastasis, and 55% had peritoneal disease on staging laparoscopy or at attempted resection. Management approaches to GOO were: upfront resection (38%), feeding jejunostomy (JT) with or without decompressive gastrostomy (GT) (42%), surgical gastrojejunostomy (8%), and endoscopic intervention including stent or pyloric dilation (12%). After receiving chemotherapy, chemoradiotherapy, radiation, and/or hyperthermic intraperitoneal chemotherapy (HIPEC), 7/25 patients initially treated with palliative GT/JT, and 3/7 initially treated endoscopically, ultimately underwent resection. Overall, 20/60 and 13/60 patients underwent curative or palliative resection, respectively. The figure depicts OS based on final resection status. The median OS durations were 7.5 (0.4-26.2) months for patients without resection, 21.4 (4.1-69.7) months for palliative resection (p=0.003), and 26.4 (3.9-221.1) months for curative resection (p<0.001). Conclusion Patients presenting with GOO secondary to GA have a high rate of metastatic disease and demonstrate improved survival with resection. OS does not differ between patients undergoing curative or palliative resection, reflecting the poor prognosis associated with GOO and the need for novel treatment approaches and clinical trials in this population. Surgery, Kansai Rosai Hosipital, Amagasaki, Japan; 2. Kindai University, Japan; 3. National Hospital Organization Osaka National Hospital, Osaka, Japan; 4. Osaka Police Hospital, Osaka, Japan; 5. Kansai Medical University, Hirakata, Japan; 6. Higashiosaka City Medical Center, Higashiosaka, Japan; 7. Ikeda City Hospital, Ikeda, Japan; 8. Yao Municipal Hospital, Yao, Japan; 9. Kaizuka City Hospital, Kaizuka, Japan; 10. Toyonaka Municipal Hospital, Toyonaka, Japan; 11. Osaka Rosai Hospital, Sakai, Japan; 12. Tottori University, Tottori, Japan; 13. Nippon Medical School, Tokyo, Japan; 14. Wakayama Medical University, Wakayama, Japan. Backgrounds: We have already shown that daily nutritional intervention with an oral elemental diet (ED) attenuated the short-term postoperative percentage of body weight loss (%BWL) in post-gastrectomy patients, especially in underwent total gastrectomy (TG).(KSES-001 study, Imamura et al. Ann Surg Oncol 2016) This additional study was conducted to evaluate the postoperative BWL and long-term skeletal muscle mass loss(SML) of nutritional intervention. Methods: One hundred and eleven patients were randomly allocated to receive the control or ED diet in the original trial. Control (C) group received the regular diet alone after gastrectomy, while ED group received 300 kcal of ED plus their regular diet for 6-8 weeks. Only 107 patients who had CT imaging data at 1 year after surgery were eligible in this study. The postoperative percentage of skeletal muscle mass loss (%SML) was calculated by measuring lumbar skeletal muscle area at L3 level of both presurgical and at 1year after surgery by using SYNAPSE VINCENT. %BWL and %SML from the presurgical bodyweight to that at 1 year after surgery were analyzed by surgical type. Results:There was not significant difference in the %BWL between the two groups (9.13 7.72 % vs. 7.09 7.49 %, respectively; p=0.171) . The %BWL at one year after surgery was significantly lower in ED group than in control group among patients who underwent TG (n=19 and 17, respectively; 9.66 5.98% vs. 15.11 6.78%, respectively; p=0.015), but not in patients who underwent distal gastrectomy (DG) (n=38 and 32, respectively; 5.81 7.91% vs. 5.96 6.20%, respectively; p=0.933). There was not significant difference in the %SML between not only the whole each group (-8.0 9.92% vs. -10.1 10.6%,respectively; p=0.443), but also by surgical type (TG: -12.5 11.2% vs. -17.5 13.0%, respectively; p=0.406 / DG: -5.6 8.4% vs. -6.2 6.4%, respectively; p=0.784). Conclusions: Postoperative nutritional intervention with ED could not reduce SML at 1 year after surgery, but could reduce BML at 1 year in patients who underwent total gastrectomy. Introduction: Esophageal and gastroesophageal junction (GEJ) cancers represent aggressive malignancies and trimodality therapy for locally advanced tumors is the current standard of care. We examined our institutional outcomes to identify predictors of early disease recurrence after resection. Methods: Patients who underwent treatment including esophagectomy for distal esophageal and GEJ cancers between June 2012 and April 2019 were included. Early recurrence was defined as evidence of local or distant disease recurrence within 6 months of esophagectomy. Pathologic confirmation was not required for diagnosis if recurrence was evident by imaging. Patients with less than 6 months of follow-up or benign disease were excluded. Univariate comparisons were performed for patients with and without early recurrence. A multivariate logistic regression model was created for the outcome of early recurrence using weight loss > 5 kg at diagnosis, margin status, lymph node ratio, and postoperative conduit dilation as predictors. Kaplan-Meier methods were used to compare overall survival. Results: Of 218 patients screened, 179 patients were included for analysis while 39 patients were excluded due to benign diagnosis or inadequate follow-up. Early recurrence within 6 months of surgery occurred in 28 patients (15.6%) . Univariate comparisons of the early recurrence group to the remaining patients are shown in Table 1 . Significant associations with early recurrence included weight loss > 5 kg at diagnosis, increased nodal stage, and positive margins. Multivariate logistic regression modeling confirmed lymph node ratio (LNR) as a significant predictor of early recurrence (OR 1.58, 95% CI 1.14-2.18, for every 10% increase in LNR), and non-significant trends for weight loss > 5 kg at diagnosis (OR 1.69, CI 0.69-4.17) and positive margin (OR 2.70, ). The median OS was lower in the early recurrence group (7.5 vs. 49 .5 months, p < 0.001). Conclusions: Pathologic lymph node ratio is a strong predictor of early recurrence after resection for esophageal and GEJ cancers. Patients with early recurrence have significantly decreased overall survival.

A New Postoperative Pain Management Leads to Enhanced Recovery After Esophagectomy: A Propensity Score-Matched Analysis Y. Nakano,* Y. Ohkura, M. Ueno, H. Udagawa. Gastroenterological Surgery, Toranomon Hospital, Tokyo, Japan. Purposes: To investigate the efficacy of postoperative scheduled intravenous acetaminophen to reduce the opioid use and enhance the recovery after esophagectomy. Methods: A propensity score-matched population was created using the 93 and 69 consecutive patients who underwent esophagectomy for esophageal cancer before and after the introduction of postoperative scheduled intravenous acetaminophen, and the short-term clinical outcomes were compared. Results: Significant defervescence was demonstrated in the Acetaminophen group (A-group) compared with control group (C-group) during the perioperative period (p < 0.05), whereas no significant differences were observed in the postoperative inflammatory parameters. The incidence of postoperative complications was similar between the groups. The number of PCA pushes and the frequency of using other non-opioid analgesics were significantly smaller in the A-group than in the C-group (p < 0.05). Both daily and cumulative opioid uses were significantly smaller in the A-group than in the C-group (p < 0.05). The time to first flatus was significantly shorter in the A-group than in the C-group (p < 0.001). The day of first walking after surgery was significantly earlier in the A-group than in the C-group (1.0 versus 2.0 days, p = 0.003). The ICU stay (2.86 versus 3.61 days, p < 0.001) and the hospital stay (21.5 versus 26.0 days, p = 0.061) tended to be shorter in the A-group than in the C-group. Conclusions: Postoperative scheduled intravenous acetaminophen decreased the rate of opioid use without increasing the intensity of postoperative pain and may be a feasible new pain management option in the enhanced recovery after surgery protocol following esophagectomy. Introduction: The incidence of ductal carcinoma in situ (DCIS) has risen with the increasing use of screening mammography but the mortality from DCIS is low. Treatment is excision with or without adjuvant endocrine therapy (ET) and radiation therapy (RT). The aim of this study is to identify the effects of ET and RT on overall second breast events (SBEs) in different time frames and on rates of second invasive breast events (SIBEs) and second DCIS breast events (SDBEs). Methods: A retrospective chart review was performed using an institutional cancer registry to identify patients who were diagnosed with DCIS between 1990 and 2018. All variables were confirmed by chart review. Cox regression with time-varying coefficients for SBEs and competing risks regressions for SIBEs and SDBEs were performed with the statistical software R. Results: 460 patients met inclusion criteria. Median follow up time was 5.6 years. 322 (70%) patients received RT and 212 (47%) ET. In the multivariate model RT reduced SBE risk in the first 5 years (hazard ratio (HR)=0.4, p=0.01) while ET reduced risk for both 0-5 and 5-10 years (HR=0.3 and HR=0.2 with p=0.02 and p=0.02, respectively). By treatment group, patients receiving both ET and RT had significantly reduced risk of SBEs in the 0-5 and 5-10 year periods and the RT only group had reduced risk in the 0-5 year time frame. On multivariate competing risks analysis ET reduced the risk of SIBEs (HR=0.33, p=0.03) but not SDBEs; RT reduced the risk of ipsilateral SDBEs (HR=0.26, p=0.0008) but not SIBEs. By treatment group, patients receiving ET and RT had a reduced risk of SIBEs (HR=0.23, p=0.03) whereas the ET only (HR=0.44, p=0.29) and RT only (HR=0.80, p=0.62) groups did not. Conclusions: Our findings suggest that ET alone and ET with RT are preferable to RT alone. ET reduced the risk of SIBEs and reduced the risk of SBEs for up to 10 years, whereas RT predominantly reduced the risk of SDBEs and did so mainly in the first 5 year period. The treatment group receiving ET and RT received the greatest benefit in risk reduction, including SBEs in both 0-5 and 5-10 year periods, as well as in both SIBEs and SDBEs. INTRODUCTION: Long term data analyzing accelerated partial breast irradiation (APBI) in node negative, early stage breast cancer using intracavitary brachytherapy by any of the established guidelines for patient selection is limited. We seek to report our outcomes. METHODS: Patients treated with intracavitary brachytherapy APBI using the SAVI (Strut-Adjusted Volume Implant) at a single academic tertiary referral center from January 2009 Introduction: IORT is an attractive option for breast conserving therapy. Trials show trends toward higher local failure rates than whole breast irradiation which, despite not meeting statistical significance, have led to controversy about appropriate use. We enrolled on a prospective registry trial for X-ray IORT and herein report our 3-year results. Methods: 100 women >/= age 40 with tumor size </= 3 cm, with cTis-2N0 breast cancer were treated with lumpectomy followed by immediate IORT from October 2013 to May 2017. IORT was withheld in the event of positive sentinel nodes on frozen sections. IORT dose was 20 Gy to the balloon surface using a 50 kVp source. Median age was 65; no patient was younger than 45. Eight women were lost to follow-up after BCT, leaving a total of 92 patients evaluable for recurrence data. Median follow-up was 35.3 months. Results: Of the 100 women treated, 70 had invasive ductal carcinoma, 20 had DCIS, and 10 had invasive mammary/mucinous carcinoma. Seven patients were N+, only 2 having macroscopic nodal disease. Cosmesis was satisfactory, with a 6% rate of any cosmetic changes. In-breast tumor recurrence rate (IBTR) in the 92 evaluable patients was 3.26% (3 patients); 1 patient recurred in the contralateral breast. No trends in tumor size, markers, or nodal status were seen among women who recurred, likely due to small number of events. It was noted that 2 of the 3 IBTR were in patients with initial DCIS, and that all 3 recurrences were in the setting of close (1 mm) margins. Two recurrences were adjacent to the lumpectomy cavity, and one was in a different quadrant. Median time to recurrence was 35 months (range 30-42). See table for details of initial and recurrent disease in those who recurred. Conclusion: IBTR rate was 3.26% at median follow-up of 35 months. Median time to recurrence was 34.5 months. Given the natural history of breast cancer, one would assume more recurrences will happen; the hazard rate of 1% per year is consistent with other types of breast conservation. We encourage other groups to join us in following the results of IORT for longer periods to demonstrate a more robust hazard rate curve over time for this technique.

Does Failure to Achieve Pathologic Complete Response with Neoadjuvant Chemotherapy Identify Node-negative Patients Who Would Benefit from Postmastectomy Radiation or Regional Nodal Irradiation? A. Crown,* M. Gonen, T. Le, M. Morrow. Breast Service, Department of Surgery, Memorial Sloan Kettering Cancer Center, New York, NY. Background: Nodal positivity before or after neoadjuvant chemotherapy (NAC) is an indication for postmastectomy radiation (PMRT)/regional nodal irradiation (RNI). Triple negative (TN) cancers and failure to achieve pathologic complete response (pCR) with NAC increase the risk of locoregional recurrence (LRR). Whether the absolute risk of LRR is high enough to justify PMRT/RNI in node-negative patients who do not have a pCR with NAC is uncertain. Methods: Patients with cT1-T3N0 and pN0 disease treated with NAC who did not have breast pCR were identified from a prospective database. Patients were eligible if they had mastectomy or breast conserving therapy (BCT) with negative margins and whole breast radiation. Those receiving PMRT or RNI were excluded. Actuarial rates were estimated using the cumulative incidence function. Results: Among 227 patients, the mean age was 51.4 12.6 years and 81 (35.7%) were TN (Table 1) . With a median follow-up of 29 months, 15 recurred: 7 (3.1%) LRRs and 8 (3.5%) distant recurrences. The overall crude and actuarial 3-year LRR rates were 3.1% and 4.2%, respectively. The crude LRR rate for TN patients was 3.7% with a 3-year actuarial rate of 5.3%. ER/PR+ Her2-patients had a 2.8% crude LRR rate and a 3-year actuarial rate of 3.9%. Her2+ patients had a 2.7% crude LRR rate and a 3.2% 3-year actuarial rate. Mean age of patients with LRR was 44.3 years, compared to 51.7 years in patients without LRR. Conclusions: LRR is a rare event even in node-negative patients who fail to achieve pCR following NAC. Even among patients with TN disease, 3-year LRR rates are not high enough to justify PMRT/RNI, but further follow-up is needed for definitive conclusions. Background: Radiation therapy (RT) is used to treat many adolescent and young adult (AYA) and childhood cancer patients. Additionally, it is a known risk factor for secondary breast cancer (BC). However, no studies have evaluated how RT impacts clinical characteristics and survival for premenopausal BC in AYA and childhood cancer survivors. Methods: Female patients ages 12-50 diagnosed with primary BC during 1988-2014 (n=107,751) were obtained from the California Cancer Registry and compared with secondary BC patients of the same age whom had RT treatment for their primary tumor (n=1,147). We examined the association of sociodemographic factors and clinical characteristics, with breast cancer specific survival (BCSS) for women with primary and secondary BC. Regression models also assessed the associations of secondary versus primary BC on BCSS for all women and in subgroups defined by race/ethnicity, age, histologic grade, lymph node (LN) status, stage, and tumor receptors. Results: Compared with primary BC, women with secondary BC were more likely to be Hispanic or non-Hispanic Black, have earlier stage tumors, be higher grade, have no LN involvement, and be ER/PR negative. Most secondary BCs were treated with mastectomy (63.0%) but were less like to receive chemotherapy (55.4% vs 62.2%) or RT (26.3% vs 45.5%) than primary BC. BCSS was decreased for Black women with primary BC, while Hispanic women had decreased BCSS with secondary BC. All women showed worse BCSS for large tumors, LN involvement, and tumor marker negativity. BCSS was worse for women with secondary vs primary BC, both overall (HR: 1.98; 95% CI 1.77-2.23) and in all subgroups including Asian/Pacific Islanders, LN negative patients, and ER/PR positive disease. Conclusions: We found that BCSS is significantly decreased among all childhood and AYA cancer survivors treated with RT, even in the setting of early-stage BC or other characteristics (such as LN negativity and ER positivity) which are considered protective. Therefore, we may need to consider alternative and even more aggressive treatment in what were considered low risk populations previously. York, NY; 4. Memorial Sloan Kettering Cancer Center, New York, NY. Introduction: Breast cancer is now the leading cause of cancer related death in Nigeria. This pilot study reports the total out-of-pocket cost for breast cancer diagnosis and management and the associated rate of catastrophic healthcare expenditure at a public institution in Nigeria. Methods: Patients presenting with a new diagnosis of breast cancer between December 2017 -August 2018 were identified from a prospectively maintained database at Obafemi Awolowo University Teaching Hospital in Ile-Ife, Nigeria. Patients were excluded if treatment was still ongoing or if no active treatment was received after initial consultation. A questionnaire was administered to capture patient-reported income and cost of care, including direct and indirect expenses (e.g. cost of travel). A threshold of 25% of annual household income was used to define a catastrophic healthcare expenditure. Monetary units were converted to international dollars ($) using the World Bank's 2016 purchasing power parity conversion factor. Results: Data was collected from 22 eligible patients. The mean age was 50.6 years and 77.3% presented with locally advanced disease (63.6% Stage III, 13.7% Stage IV). Thirty-two percent had some form of pre-paid health insurance. The mean cost of diagnosis and treatment was $5,073 (SD $4,590). Chemotherapy represented 45.9% of the direct cost of care. Immunohistochemistry for receptor-status was performed on 22.7% (5/22) of patients. None of the patients received HER-2 targeted or radiotherapy. In total, 86% of households experienced a catastrophic healthcare expenditure as a result of care. Fourteen percent of patients with potentially resectable disease did not undergo surgery due to cost. Conclusion: In this small pilot study on the patient-reported cost of breast cancer care, pre-paid health insurance was uncommon, which resulted in a catastrophic healthcare expenditure for over 80% of households. The high out-of-pocket cost associated with care had a significant impact on treatment course. Greater financial protection is essential as the burden of breast cancer increases in Nigeria. Background Mounting evidence has suggested that acute pancreatitis (AP) is a risk factor of pancreatic cancer (PC) and may accelerate oncogenesis in PC, but its role in survival in PC patients, especially in patients with resectable pancreatic cancer has not been investigated. The objective was to investigate the association of history of AP with survival among PC patients who have undergone surgical resection in a retrospective cohort-based study. Methods In this retrospective cohort of 662 pancreatic cancer patients who underwent radical surgical resection, we evaluated survival by acute pancreatitis history prior to PC diagnoses using Kaplan-Meier methods and log-rank tests. Multivariate analyses for mortality were estimated using Cox proportional hazards model, adjusted for age, sex, tumour location, tumour grade, tumour stage, diabetes mellitus status, baseline CA19-9 level, adjuvant chemotherapy and adjuvant radiotherapy. Results The log-rank tests showed that the patients with history of acute pancreatitis had worse overall survival compared to those without acute pancreatitis history (p = 0.007). The multivariable-adjusted HR for mortality comparing participants with acute pancreatitis to those without was 1.709 (95% CI: 1.187-2.459, P-value = 0.004). After categorizing patients with history of acute pancreatitis into recent (< 2 years) and remote history ( 2 years) of AP, patients with recent history of AP, rather than patients with remote history of AP, were found to have significantly worse survival (P-value = 0.007) than those without AP history. Conclusion Our findings indicate that history of acute pancreatitis, especially recent history of acute pancreatitis, is associated with worse survival in resectable pancreatic cancer patients. Introduction: A strengthening consensus exists for neoadjuvant therapy (NAT) in borderline resectable pancreatic adenocarcinoma (PA), but the utilization of NAT in resectable stage I PA remains controversial. Many cancer centers are using NAT for these patients (pts), but others continue to offer upfront surgery and adjuvant therapy (AT). We hypothesized that NAT would improve margin negative resection in clinical stage I resectable PA. Methods: We utilized the IRB approved 2016 national cancer database for pancreas to establish a cohort of stage I PA pts. We divided this subset into pts who underwent NAT vs AT. We compared demographics. Primary endpoint was surgical margins. Results: 10,453pts from 2004 to 2016 had clinical stage I resectable PA: 8483pts (81.1%) AT and 1970pts (18.9%) total or partial NAT. There was a statistical difference in age (64.9 9.9years NAT and 66.2 9.9years AT, p<0.001), but no difference in Charlson comorbidity score (p=0.1693). NAT pts had significantly higher margin negative resection rates (84.5%) than AT pts (79.4%) (p<0.0001). Final pathologic staging was available for 10,237 pts: 8369 (81.8%) AT and 1868 (18.2%) NAT. Significantly fewer pts were upstaged on final pathology to stage II or greater (73.5%) in the NAT group than the AT group (84.1%) (p<0.001). Conclusions: NAT leads to significantly higher margin negative rates for resectable clinical stage I PA than surgery followed by AT. The majority of pts for both groups were upstaged suggesting that we continue to clinically understage the majority of pts. Overall, total or partial NAT for clinical stage I resectable PA provides a better chance for margin negative resection. Further study in the form of a randomized control trial is necessary. Introduction: Neoadjuvant therapy (NAT) is a growing treatment strategy for pancreatic cancer (PDAC), including patients with radiographically resectable disease. Since elderly patients are at increased risk of treatment withdrawal due to functional decline, the beneficial impact of NAT on survival in this cohort needs to be outlined. The objective of this study was to compare outcomes of elderly patients with resectable head PDAC who underwent NAT or a surgery first (SF) approach. Methods: We conducted a retrospective review of all patients 75 years and older with radiographically resectable (NCCN criteria) PDAC who underwent pancreaticoduodenectomy at a single institution from 2008-2017. Baseline characteristics and perioperative outcomes were compared between SF and NAT cohorts. Recurrence free (RFS) and overall survival (OS) were analyzed by treatment strategy. Results: A total of 158 patients were identified: SF=90 (57%) and NAT=68 (43%). Patients in the SF cohort were older (80 versus 78 years; p=0.01), but there were no differences in preoperative comorbidities or frailty indices. SF patients had a trend toward higher rates of major (Clavien 3) complications (37.8% vs 23.5%, p=0.06) with higher comprehensive complication index totals (CCI 20.9 vs 20, p=0.03) for their hospitalizations. There were similar rates of adjuvant therapy receipt (SF 48% vs NAT 51%, p=0.713) and number of cycles received (SF 5 vs NAT 3, p=0.12) between the two groups. NAT was associated with significantly longer OS (24.6 vs 17.6 mos, p=0.007) and RFS (16.2 vs 12.4 months, p=0.001). On MVA, NAT remained an independent predictor of OS (HR 0.66; 95% CI 0.44 -0.97, p=0.007) and RFS (HR 0.56; 95% CI 0.38 -0.80, p=0.002) after adjusting for baseline, oncologic, and perioperative factors. Conclusion: These data suggest NAT is a safe strategy for elderly patients. Patients receiving NAT had fewer complications after surgery and were equally as likely to receive adjuvant therapy. Most importantly, NAT was associated with improved RFS and OS. This suggests that the NAT strategy may offer significant benefit without undue risk for older patients. Introduction: Current clinical consensus guidelines recommend 4 weeks of venous thromboembolism prophylaxis (VTE ppx) in high risk cancer patients undergoing abdominal surgery. Compliance with this recommendation is unknown following surgery for pancreatic cancer (PC). Methods: We analyzed stage I-II PC patients 65 years of age who underwent pancreatic resection using the Surveillance, Epidemiology, & End Results (SEER) database merged with Medicare data (2009) (2010) (2011) (2012) (2013) . VTE ppx included warfarin, low molecular weight/unfractionated heparin, or factor Xa/IIa inhibitors. Exclusion criteria included: absence of Medicare Parts A/B/D coverage, length of stay 4 weeks, and preoperative conditions of: anticoagulation, atrial fibrillation, VTE, hypercoagulopathy or heart valve replacement. Fisher's exact test, t-tests, and multivariable logistic regression models compared clinicopathologic characteristics and outcomes with the use of prolonged VTE ppx. Results: Of the 1,003 PC patients who underwent pancreatic resection, only 4.3% (44) were prescribed VTE ppx at discharge. There were no significant clinicopathologic differences in patients who were and were not prescribed VTE ppx (all p>0.05). Overall incidence of VTE within 30-days of surgery was 2.9%; No difference in VTE was observed for those patients prescribed prolonged VTE ppx (2.3% vs. 2.9%, p=1.00). There were also no differences in postoperative hemorrhage (4.6% vs. 6.3%, p=1.00), overall complications (40.9% vs. 44.8%, p=0.64), 30-day readmissions (22.7% vs. 35.6%, p=0.10), 30-day mortality (0% vs. 3.8%, p=0.62), or overall survival (median survival 23 months vs. 21 months, p=0.32) between those prescribed and not prescribed VTE ppx. Conclusion: In this analysis of older adult patients with resectable PC, there was low compliance with recommended guidelines for prolonged chemical VTE prophylaxis following PC surgery. The overall incidence of VTE within 30 days of surgery was 2.9% raising questions about the current recommendation extending beyond the perioperative period. Introduction: Accumulating evidence demonstrates a significant role for circadian disruption in cancer development and progression. However, investigations of circadian disruption in pancreatic ductal adenocarcinoma (PDAC) are limited. We sought to determine if circadian disruption played a role in PDAC development through use of a genetically engineered mouse model. Methods: Kras G12D/+ mice (K) were crossed with PDX-1 Cre (C) on a C57BL/6 background to generate KC mice. At 4 weeks of age, matched mice were housed in standard lighting conditions (KC normal circadian [KC NC ]) or subjected to a chronic jet-lag protocol known to induce circadian disruption (KC CD ), with light/dark cycles shifted forward 8 hours every 2-3 days. Mice were sacrificed at 9 months. Histologic analysis was performed by a boardcertified gastrointestinal pathologist. Categorical Comparisons were made with Fischer's Exact Test, and Kaplan Meier log rank survival analysis for overall survival. Results: Histology for 27 KC NC and 7 KC CD mice were analyzed. Previous studies report all KC mice exhibit chronic pancreatitis and PanIN-1. Concordantly, all KC NC and KC CD mice exhibited PanIN-1 and chronic pancreatitis (100%; p=1.00) (Fig. 1 ). However, rates of acute pancreatitis were significantly higher in KC CD (86% vs 4%; p<0.01). In evaluating progression to PDAC, we observed significantly higher rates of PanIN-2 in KC CD than KC NC (43% vs 7%; p=0.04). PanIN-3 (14% vs 7% p=0.48) and PDAC (29% vs 16% p=0.6) were also increased but this did not reach statistical significance. Fewer KC CD mice survived to 9 months (55% vs 70%, p=0.44), but this was not different on log rank analysis. Conclusion: Circadian disruption appears to play a role in many cancers, but has not been adequately studied in PDAC. Despite a limited sample size, we have demonstrated that phenotypic differences in the rates of pancreatic pathology exist in mice experiencing circadian disruption. Currently we are expanding the number of KC CD mice to increase study power and determine if differences in PanIN-3 and PDAC truly exist. Future work will explore the molecular basis of circadian disruption in PDAC, which may lead to identification of therapeutic targets. Background: We have demonstrated that physical activity during pancreatic cancer treatment remodels tumor vasculature and improves quality of life and physical fitness. Here we sought to assess survivors' adherence to established exercise guidelines and to examine factors associated with guideline adherence. Methods: We surveyed all survivors who underwent pancreatectomy at our center for an adenocarcinoma or neuroendocrine tumor 2000-2017 to assess their frequency, intensity and duration of exercise. Additional surveys were administered to measure motivation, barrier self-efficacy, QOL, and fatigue. Self-reported exercise was compared to American College of Sports Medicine (ACSM) guidelines for cancer survivors: 150 min/wk moderate-to-vigorous (or 75 min/wk vigorous) aerobic exercise and 2x/wk muscle strengthening. We constructed multivariable models to evaluate associations between clinicodemographic and psychosocial variables and guideline adherence, and between guideline adherence, QOL, and fatigue. Results: Two hundred and sixty two (51%) of 518 eligible survivors returned surveys a median of 64 (range 9 -209) months following pancreatectomy (see Table) . Only 50 (19.1%) survivors reported meeting aerobic and strengthening guidelines; 117 (45%) reported meeting neither. Adjusted analyses demonstrated that higher autonomous motivation was associated with adherence to each guideline (both p<.05). Higher barrier self-efficacy and age were associated with aerobic guideline adherence (p<.05). We identified no significant associations between guideline adherence and other clinicodemographic characteristics including tumor type, time since surgery, cancer status or recent cancer therapy (all p>.05). We did find statistically significant and favorable associations between aerobic guideline adherence and both QOL and fatigue and between strengthening guideline adherence and QOL (all p<.05). Conclusion: Almost half of survivors in this cross-sectional study met neither aerobic nor strengthening exercise guidelines. To maximize adherence and related benefits, interventions should aim to increase autonomous exercise motivation and help participants overcome barriers.

Clinicodemographic characteristics and exercise guideline adherence *Persistent or recurrent pancreatic cancer or secondary cancer, ** Last measured BMI during clinic visit (median 3.5 [0 -178] months prior to survey), *** Calculated using self-reported data from ACSM Exercise Risk Questionnaire Introduction: RP LMS predominantly arise from smooth muscle-walled veins, including named veins and smaller tributaries. IVC LMS can present technical treatment challenges and tumor cells have ready access to the pulmonary vasculature which may facilitate metastasis. Here we investigate whether presentation, treatment and outcomes differ between IVC and non-IVC LMS managed at our institution with a multidisciplinary approach. Methods: Consecutive patients with primary RP LMS who underwent resection between 10/97 and 11/18 were identified from a prospective database and charts retrospectively reviewed. Survival curves were constructed by the Kaplan-Meier method and compared by log rank analysis. Crude cumulative incidence of local and distant relapse were calculated within a competing risks framework. Results: In 60 patients who underwent resection for primary RP LMS, the site of origin was IVC in 32 and non-IVC in 28 (Table) . There was no significant difference in age, gender, or FNCLCC grade between the groups, but non-IVC derived tumors were larger (p=0.003). The majority of patients received neoadjuvant therapy: 80% had external beam radiation, and 36% chemotherapy, with no difference between IVC and non-IVC groups. There were no R2 resections, and though there was a trend to increased R1 margin status in the IVC group, this was not statistically significant. In the IVC group, major vascular reconstruction was performed in 84% of patients. Mortality and major morbidity did not differ between the two groups. Median postoperative follow-up was 52 mos . While approximately half of patients developed distant metastases by 5yrs, 5yr DSS was 76%. The rate of local relapse was low. There were no differences in DSS, OS, DFS or local and distant relapse rates between IVC and non-IVC groups. Conclusions: Our experience demonstrates the feasibility of multi-modality therapy for IVC LMS, with similar short-and long-term post-resection outcomes as patients with non-IVC derived RP LMS. These results suggest that there is no inherent difference in tumor biology between RP LMS that originate in the IVC and those that arise elsewhere. Introduction Primary inferior vena cava leiomyosarcoma (IVC LMS) is a rare malignant tumor with poor prognosis. The 5-year progression-free survival (PFS) is only 6% with up to 50% of patients developing distant metastases postoperatively. The impact of perioerative treatment modalities on IVC LMS outcomes needs to be studied. The aim was to assess the impact of different treatment strategies on long-term outcomes for patients undergoing en bloc resection of IVC LMS. Methods After IRB approval, data has been collected retrospectively on patients who had IVC LMS between 1991 and 2019 at a tertiary medical center. The data included patient demographics, tumor characteristics, perioperative treatment modalities (neoadjuvant, intraoperative and adjuvant), surgical resection data, short-term and long-term outcomes. The primary outcomes were the PFS and the overall survival (OS) in months. The Kaplan-Meier survival and Cox proportional hazards regression analyses have been used as appropriate. Results Sixty-three patients had en bloc surgical resection of IVC LMS. One-third of patients received neoadjuvant systemic induction chemotherapy and/or chemoradiation and 2/3rds received adjuvant chemotherapy. The median PFS for IVC LMS who received neoadjuvant therapy was longer than those who did not receive neoadjuvant therapy (p=0.051). Adjuvant treatment was not associated with improved PFS (p=0.72) and was associated with worse OS (p<0.01). Background: Primary retroperitoneal leiomyosarcoma (RP LMS) often requires extensive resection for complete extirpation of disease, and has a high propensity for distant recurrence. Preoperative therapy is often given to limit the extent of resection and mitigate the risk of distant recurrence. Whether radiographic response (RR) or pathological response (PR) to therapy correlates with outcome is unknown. Methods: We reviewed an institutional database to identify primary, non-metastatic RP LMS patients who underwent resection following preoperative chemotherapy and/or radiation therapy (RT). RR was based on Response Evaluation Criteria in Solid Tumors (RECIST). PR was categorized as minimal, moderate, and major based on the percentage of residual viable tumor (> 50%, 11% to 50%, and 10%, respectively). We investigated relationships between RR, PR, and outcome. Results: 47 patients were identified (1996) (1997) (1998) (1999) (2000) (2001) (2002) (2003) (2004) (2005) (2006) (2007) (2008) (2009) (2010) (2011) (2012) (2013) . 22 (47%) had tumors involving the inferior vena cava, and 18 (38%) had tumors at least 10 cm in diameter. All patients received preoperative chemotherapy, and 16 (34%) received preoperative RT. By RECIST criteria, 5 (11%) patients had partial response, 38 (81%) had stable disease, and 4 (7%) had progressive disease. PR was minimal in 31 (66%) patients, moderate in 11 (23%), and major in 5 (11%). There was no association between RR and PR (Fisher exact test, p = 0.603). 33 (70%) patients had R0 resection. 22 (47%) patients experienced recurrence: 15 (32%) distant only, 5 (11%) both local and distant, and 2 (4%) local only. Median overall survival (OS) was 5.3 years; median recurrence-free survival (RFS) was 2.5 years. Neither RR nor PR was associated with OS. By log-rank test, RR was associated with RFS (p = 0.019, see Figure) , but PR was not. Neither RR nor PR predicted RFS on univariate cox regression. Conclusion: In this series of patients with RP LMS, major RR and PR was uncommon. RR and PR did not correlate with OS. RR-but not PR-was associated with an improved RFS, and may help select patients for surgery. The utility of neoadjuvant treatment for patients with RP LMS remains unclear using currently available systemic chemotherapeutic agents. Background: Local, rather than distant, recurrence is the major concern after resection of well-differentiated liposarcoma, especially in the retroperitoneum. Variant histologies, namely sclerosing and inflammatory types, have been suggested as increasing the risk for local recurrence. Therefore, we sought to analyze the association between variant histology, recurrence, and tumor location in patients with primary, well-differentiated (WD), liposarcoma. Methods: This is a single institution, retrospective analysis of patients who had undergone resection of well-differentiated liposarcoma (without dedifferentiation) from 2000-2017. Patients were excluded if they had a resection at another facility, received neoadjuvant therapy, or their slides were not available for review. All resected specimens underwent pathologic review by a sarcoma pathologist. The presence and extent of sclerosing and inflammatory variant histology were specifically noted and, if the dominant histology (>40%), was considered a variant specimen. The primary end point was time to recurrence, measured by log-rank comparison of Kaplan-Meier curves. Results: Of 243 resected cases of WD liposarcoma, 150 cases (21 retroperitoneal) were available for pathologic review. Median duration of follow-up was 4 years. Recurrences occurred at a median time of 3.3 years and were local (rather than distant) in 90%. Variant histology was the dominant histology in 40% of RP tumors (vs 6% of non-RP tumors, p<0.0001), and was associated a shorter time to local recurrence than those tumors without variant histology (p<0.0001). Conclusion: For patients with well-differentiated liposarcoma, variant histology is common in the retroperitoneum, but rare elsewhere. When it is the dominant histological type, variant histology suggests a more locally aggressive phenotype. Further study is warranted. INTRODUCTION: How well radiologic size agrees with pathologic size of gastric GISTs is unknown. Guidelines risk stratify GISTs based on location, pathologic size and mitotic rate, with controversy regarding surveillance, biopsy, or surgery for small gastric GISTs ( 2-3cm) . To inform preoperative risk-stratification, the discrepancy between imaging and pathologic size was assessed. METHODS: CT, EUS and pathology reports were reviewed for 116 Gastric GISTs. Bland-Altman analysis was used to assess agreement of radiologic and pathologic size for all GISTs and GISTs 3cm. Regression analyses assessed interactions between imaging features and mitotic rate. The effect of growth rate on results was analyzed by non-linear best fit curve of radiopathologic size difference (cm) vs. the imaging to surgical delay (days). Patients who changed risk category due to size discrepancy were identified. Neoadjuvant therapy, metastases, rupture, and surgical delay > 16 weeks were excluded. RESULTS: For all GISTs, pathologic size was underestimated by CT (n=110) and EUS (n=50) by an average of 0.42cm (95%CI 0.11 to 0.73) and 0.54cm (95%CI 0.25 to 0.82) respectively. For GISTs 3cm, pathologic size was underestimated by CT (n=28) and EUS (n=26) by 0.24cm (95%CI 0.01 to 0.47) and 0.56cm (95%CI 0.27 to 0.84) respectively. Spearman's correlation ( ) was >.92 for all groups (p <.001). Growth during surgical delay was not significant. For GISTs <=3cm, 4/28 (14.29%) on CT and 7/26 (26.9%) on EUS moved to a higher risk category on final pathology. No patients moved to a lower risk category. Uni-and multivariate analyses revealed no relationship between imaging features and mitotic rate (p >0.05 for all). RFS and OS were not evaluated due to low event rate (4 recurrences-2 local and 2 distant and 8 non-GIST-related deaths). Median follow-up was 48.5 months. CONCLU-SION: EUS and CT underestimate gastric GIST size despite strong correlation. This effect is robust for all sizes and more pronounced on EUS. The average size underestimation for small GISTs suggests caution when using imaging alone to select gastric GISTs for surveillance in the 3cm cohort, especially if only imaged with EUS. Introduction: Gastrointestinal stromal tumor (GIST) is a rare mesenchymal neoplasm usually found in the stomach. Previously published work demonstrated that up to two thirds of GISTs are not captured by cancer registries and therefore would not be included in large databases. Therefore, the true incidence of GIST and the outcomes of surgically resected incidental GISTs are unknown. Objective: To determine the extent of under-reporting and the characterization of GISTs omitted by the cancer registry. Methods: This was an IRB approved retrospective study that identified patients with gastric and small intestinal GISTs at our institution. Patient and tumor characteristics were evaluated. Results: Between 2010 and 2017 there were 63 cases of GIST from the stomach (55.4%, n=42) and small intestine (28.4%, n=21). Mean age was 62 years (31 to 93 years) and 58.7% were female (n=37). Patients were 46.0% Caucasian (n=29), 19.0% Black (n=12), 22.2% Hispanic (n=14), 7.9% Asian (n=5), and 4.8% other race/ethnicity (n=3). The cancer registry captured 31.7% of GIST cases (n=20) while the majority was only in the pathology database and not reported to the cancer registry. Mean size of cancer registry GISTs was 10.7cm (3.5 to 30.0cm) and mean size of non-reported GISTs was 5.8cm (0.15 to 24cm). Cancer registry versus non-reported cases were low-grade with <5/50HPF mitoses in 55.0% (n=11) vs 55.8%(n=24) respectively. However, 5.0% (n=1) vs 9.3%(n=4) had 5-10/50HPF mitoses and 30.0% (n=6) vs 4.7% (n=2) had >10/50HPF mitoses, respectively. There were unknown mitotic rates for 10.0% (n=2) vs 30.2% (n=13). Conclusions: Two-thirds of GISTs were not captured by the cancer registry. This underreporting of GISTs makes it difficult to determine true incidence and outcomes and to interpret existing literature using such databases. The use of a dichotomous classification system, benign vs. malignant, likely contributes to this. We suggest that all GISTs should be reported regardless of size and mitotic index and the benign classification should be removed. INTRODUCTION:Consensus guidelines endorse core needle biopsy (CNB) in the work-up of a suspicious RP mass. Our group has demonstrated the safety of CNB in terms of short-term complications (2%, all <CD grade 3) as well as needle tract seeding (<0.5%). Here we investigate the effect of biopsy technique on diagnostic yield and accuracy. METHODS:Patients who had CNB of a suspicious RP mass at 2 sarcoma centers (The Ottawa Hospital and Mount Sinai Hospital, Toronto) from 2000-2015 were included. Biopsy technique details, including imaging modality, needle gauge, # and length of cores were obtained. Diagnostic yield was calculated as the number of biopsies with a definitive diagnosis, divided by the total number performed. Diagnostic accuracy was calculated as the number with correct histologic subtype and grade, divided by the number of patients who underwent resection of RPS. Receiver operating curves determined cut-off points for CNB features. Logistic regression models were run to understand contributors to diagnostic yield and accuracy. RESULTS:395 biopsies were done on 322 patients. The diagnostic yield was 91.7%. On multivariable analysis, ^3 cores (p=0.000) and length of specimen ^2cm (p=0.004) predicted improved diagnostic yield. 228 CNB were done on 204 patients who underwent RPS resection. The diagnostic accuracy, correct histologic subtype and grade vs. correct histologic subtype irrespective of grade was 81.1% and 90.4%, respectively. In 14% the grade on the resection specimen was higher than on CNB. 74% of patients had preoperative radiation, potentially diminishing the extent of upgrading. On multivariable analysis, ^3 cores (p=0.002) and length of specimen ^2cm (p=0.017) predicted improved diagnostic accuracy, both for histologic subtype and grade. There was no difference between CT or US guidance. Too few CNB were performed with a 16G needle to determine the potential effects vs.18G. CONCLUSION:

In patients with suspected RPS, both the diagnostic yield and accuracy of CNB analysed in our centers were high. We recommend that at least 3 cores of at least 2 cm in length be obtained to optimize diagnostic yield and accuracy. Background: Given the complex nature of retroperitoneal sarcoma (RPS) resection, perioperative red blood cell transfusions (PBT) are often required. PBTs have previously been associated with poor oncologic outcomes in colon, liver, and gastric cancer. This study aims to identify factors that predict PBT and to assess the association of PBT on outcomes in RPS patients. Methods: 121 patients with primary RPS resected between 2006-2018 were retrospectively reviewed. Patients receiving PBT from surgery up to the date of discharge were included. Uni-and multivariable analyses with Wald's test were used to identify clinicopathologic factors associated with PBT requirement, and to assess the association of PBT with oncologic outcomes. Local and distant recurrence-free and overall survival rates were analyzed using Cox proportional hazard modelling. Results: 46 patients (38.0%) received PBT either intraoperatively or postoperatively. Multivariable analysis revealed tumour grade (p=0.0222) and intraoperative blood loss (p=0.0061) were independently associated with PBT. Duration of surgery (p=0.0829) and low pre-operative albumin (p=0.0742) approached significance. Preoperative anemia was not significantly associated (p=0.1881). Median follow-up time was 37.2 months. In survival analysis, PBT was significantly associated with reduced overall survival (p=0.0115), but not local (p=0.1068) nor distant (p=0.5933) recurrence-free survival. As a univariate predictive factor, PBT was associated with post-operative complications (Clavien-Dindo score 3, p=0.0049) and overall survival (p=0.0285), but not local (p=0.3434) or distant (p=0.9934) recurrence-free survival. In multivariable analysis, PBT was no longer significantly associated with post-operative complications (p=0.8933) or overall survival (p=0.9538). Conclusion: Our data suggest that larger studies are needed to assess the association of PBT on oncologic outcomes, due to covariate and confounding factors. However, strategies aimed at mitigating low albumin, operative time, and intraoperative blood loss are warranted to reduce the incidence of perioperative transfusion. Background: Radiation therapy for retroperitoneal sarcoma remains controversial. This study compares survival outcomes for patients with retroperitoneal sarcoma who did and did not received neoadjuvant or adjuvant radiation therapy by tumor grade. Methods: National cancer data base (2004) (2005) (2006) (2007) (2008) (2009) (2010) (2011) (2012) (2013) (2014) data were obtained for patients who underwent surgery for locoregional advanced retroperitoneal soft tissue sarcoma with or without radiation therapy. Patients with low-and high-grade sarcoma were analyzed separately. To control for selection bias in treatment allocation, propensity scores were created for the odds of receiving radiation therapy. Patients were matched 1:1 based on propensity score. Landmark analyses were performed at 3-months to control for survivorship bias. Survival analysis were performed using the Kaplan-Meier method. Results: In total, 1,662 low-grade and 3,298 highgrade sarcoma patients were identified, with 18.0% and 35.1% of patients with low-and high-grade tumors receiving radiation therapy, respectively. In patients with low-grade sarcoma, radiation therapy did not significantly improve survival before (5-year survival rate; 84.5% vs. 80.7% vs. overall log-rank p=0.5787) and after matching (5-year survival rate: 84.8% vs. 85.2%; p=0.6293). However, in patients with high-grade sarcoma radiation therapy resulted in significant survival benefit before (5-year survival rate: 60.3% vs. 51.1; p<0.001) and after matching (5-year survival rate: 60.4% vs. 52.4%; p<0.001). Treatment sequence did not significantly (overall log-rank p=0.291) impact survival in patients with high-grade sarcoma, resulting in a 5-year survival rate of 59.4% after neoadjuvant compared to 63.5% after adjuvant therapy in the subset high-grade sarcoma cohort. Conclusions: The findings of this study, while non-randomized, suggest that radiation may be associated with survival benefit for patients with high-grade sarcoma. However, may not have a role in the treatment of patients with low-grade retroperitoneal sarcoma. Introduction: It remains unclear how the treatment patterns and outcomes of patients with retroperitoneal sarcoma (RPS) are related to whether treatment was received at an academic facility (AF) or non-academic facility (NAF). Methods: The National Cancer Database (NCDB) was used to identify patients diagnosed with RPS years 2006-2015. The sample was stratified by the type of facility at which treatment was received: AF or NAF. Chi-square tests were used to compare categorical variables. Logistic regression was used to identify factors associated with a significant likelihood of receiving treatment at an AF. Kaplan-Meier survival estimates were calculated, and Cox regression used to determine the risk of death associated with treatment at an AF. Results: The sample included 4,759 patients with RPS. The mean age was 63.6 years, 49.2% were male and 84.1% were white. Most were treated at an AF (61.7% ; 2,936/4,759) . Compared to patients treated at a NAF, those treated at an AF were younger (mean age 62.8 versus 65.0; p <0.001), less likely to be female than male (OR = 0.82; 95% CI: 0.73-0.93), and more likely to have private insurance (OR = 1.48; 95% CI: 1.06-2.06), Medicaid (OR = 1.51; 95% CI: 1.01-2.27) or Medicare (OR = 1.44; 95% CI: 1.01-2.04) rather than no insurance. Patients treated at an AF were more likely to have undifferentiated rather than well differentiated tumors (OR = 1.46; 95% CI: 1.16-1.84). Also, patients treated at an AF were less likely to have positive margins following surgical resection (OR = 0.81; 95% CI: 0.69-0.94). There was no significant change in the odds of receiving chemotherapy, but patients treated at an AF had a decreased likelihood of receiving radiation (OR = 0.78; 95% CI: 0.68-0.89).

Five-year survival was 46.5% for patients treated at NAFs and 51.1% for those treated at AFs (p < 0.01). Similarly, patients treated at AFs had a lower risk of all-cause death (HR = 0.86; 95% CI: 0.79-0.95). Conclusion: Patients with RPS treated at an AF appear to have superior survival compared to those treated at a NAF. These findings support the notion that cancer treatment at specialized academic centers leads to better outcomes. Background: Surgical outcomes vary widely among hospitals and surgeons, generating the need for more accurate quality indicators, especially among patients undergoing complex operative procedures. Recently, the concept of textbook outcomes (TO) has emerged as one novel effort to construct a benchmark that reflects multiple domains of quality. The aim of the current study was to define TO for patients with retroperitoneal sarcoma (RPS) and assess the variation in rates of TO and its association with hospital procedural volume. Methods: Patients aged 18 years or older who underwent resection of a primary RPS diagnosed between 2004 and 2015 were identified in the National Cancer Database (NCDB). The primary outcome was TO. TO was defined as a composite of endpoints: hospital length of stay < 75 th percentile, survival >90 days from the date of surgery, no readmission within 30 days of discharge, and grossly negative margins at time of resection. Results: Of the 11,032 patients included in the final analysis, 54.0% had a TO. Of patients who did not achieve TO, 45.0% had an extended length of stay, and 55.9% did not have grossly negative margins. Only 2.3% patients had no more than one TO criterion. Compared with Medicare and privately insured patients, patients insured by Medicaid had lower probability of TO (relative risk [RR]=0.88, 95% confidence interval [CI]: 0.80 -0.97, p=0.007). Among patients who had a TO, 57.8% were treated in a hospital with a higher procedural volume (more than 2 cases per year). Finally, to evaluate the association between textbook outcomes and overall survival, we subset patients with gross negative margins and without 90-day mortality. After adjustment for these factors, having a TO was still associated with 31.5% longer survival time (95% CI: 1.217 -1.422, p<0.001). Conclusions: This study proposes a novel definition of TO for patients undergoing resection of RPS, utilizing variables available in the largest national cancer database. The concept of TO is a promising tool for measuring patient-level hospital performance and may be useful for identifying important variations in care for patients with this complex and rare malignancy.

Introduction Multi-visceral resection (MVR) is often utilized in the treatment of primary retroperitoneal sarcoma (RPS). Morbidity following distal pancreatectomy for primary pancreatic cancer has been well-documented; however, it is not known whether distal pancreatectomy in the context of MVR for primary RPS has similar outcomes. The objective of this study was to evaluate postoperative pancreatic fistula (POPF) following distal pancreatectomy for primary RPS. Methods Twenty-six sarcoma centers that are members of the Trans-Atlantic Australasian Retroperitoneal Sarcoma Working Group (TARPSWG) participated in the study. Consecutive patients who underwent distal pancreatectomy for primary RPS from January 1, 2008 to December 31, 2017 were identified and charts retrospectively reviewed. Outcomes measured were overall 90-day Clavien Dindo 3 complications and POPF fistula rate. Results A total of 278 patients underwent distal pancreatectomy for primary RPS. The most common histologic subtypes were dedifferentiated liposarcoma (57%), well-differentiated liposarcoma (19%) and leiomyosarcoma (13%). Preoperative radiation was administered to 77 (28%) patients. RPS invaded the pancreas in 38% of cases and the pancreas R0 margin rate was 89%. The 90-day overall Clavien Dindo 3 complication rate was 39%. POPF grade B and C complication rates were 19% and 5%, respectively. Administration of preoperative radiation, use of somatostatin analogues, mode of pancreatic division (staplers, sealants, or tissue flaps), level of pancreatic division (neck, body, tail), and use of drains did not impact the risk of developing a POPF. Discussion The absolute rate of clinically meaningful postoperative pancreatic fistula following distal pancreatectomy in the context of multi-visceral en bloc resections for primary RPS compares favourably to the fistula rate following distal pancreatectomy for primary pancreatic cancer. Resection of the distal pancreas is a reasonable approach to facilitate complete resection of primary RPS when appropriate. Introduction: The intraoperative localization of non-palpable primary and recurrent soft tissue sarcoma is a technical challenge that has important oncologic implications. Non-radioactive magnetic seeds have FDA approval for use in non-palpable breast cancer, and have been recently approved for soft tissue lesions. Here we report an early experience with the use of magnetic seeds for the localization of resectable soft tissue sarcoma. Methods: A prospective database was initiated in January 2019 for patients with soft tissue sarcoma undergoing preoperative placement of magnetic seeds for lesion localization. Data captured include the indication for surgery, method and rationale for seed placement, technical success (seed recovery), and oncologic success (margin negative resection). Results: To date, 10 patients have had magnetic seeds placed prior to resection. The indication for operation was primary (n = 4, 40%) and recurrent (n = 6, 60%) disease. 8 patients received preoperative therapy (chemotherapy and radiation [n = 3], chemotherapy [n = 3], radiation therapy [n = 2]) prior to seed placement. The anatomic locations included the abdominal wall (n =2), groin/pelvis (n = 2), back (n = 2), arm (n = 1), chest wall (n = 1), and scalp (n = 1). The lesion was described as either a discrete mass (n = 6) or an ill=defined mass (n =4) based on radiographic characteristics. 13 lesions were identified in the 10 patients, and a total of 20 seeds were deployed. Median time between placement of the seed and surgery was 5.5 days (range 1-42). Seeds were placed in the center (n = 7) of the lesion, or at the margins (n = 3) for bracketing of the lesion. All seeds were recovered. Resection margins were negative in 7 of 8 patients with margins assessed, including the 2 patients in whom multiple seeds were placed around an illdefined mass (Figure 1 ). Discussion: Magnetic seed localization of primary and recurrent soft-tissue sarcoma is feasible and can facilitate the intraoperative location of non-palpable lesions. Magnetic seeds can be used to identify the center of the mass, or may be used to bracket the margins of an ill-defined mass to facilitate complete resection. Figure 1 : Demonstration of magnetic seed soft tissue localization for recurrent sarcoma in a patient with multifocal recurrence after resection of a primary sarcoma. (a) Serial MRI sections demonstrating two tumor nodules (red arrows) adjacent to a postoperative seroma cavity (green arrow). These lesions were non-palpable, biopsy proven recurrences. Two seeds were placed to intraoperatively localize these nodules. (b) Transcutaneous localization of the magnetic seeds prior to incision. A 2cm margin is marked around the previous incision and encompassing the localized seeds. (c) Specimen radiograph demonstrating capture of the seeds. Introduction: Surgical resection of non-palpable soft tissue tumors presents a unique challenge, particularly in the setting of recurrent disease in which surrounding tissue has been surgically manipulated and often irradiated. SAVI SCOUT is a localization device that was developed to aid in resection of non-palpable breast lesions and was recently FDA-approved for localization of soft tissue tumors. We sought to evaluate feasibility and describe our early experience using SAVI SCOUT to facilitate resection of non-palpable soft tissue tumors. Methods: We assembled a single-institution retrospective case series of patients with trunk and extremity sarcoma resected with SAVI SCOUT guidance by five surgeons from December 2018 through September 2019. In each case, a single SAVI reflector was placed using ultrasound or computed tomography guidance. Clinical variables were abstracted from the electronic medical record including treatment history, pathology, and early oncologic outcomes. Results: There were seven SAVI SCOUT-localized sarcoma resections performed. Five resections were for locally recurrent disease, of which four (80%) had prior radiation as a component of their treatment history. One resection was for a primary tumor that became non-palpable after neoadjuvant chemotherapy, and one was an oligo-metastasis responsive to chemotherapy. The SAVI SCOUT detector was used intraoperatively to localize the target lesion, which was resected along with the tissue tract used for reflector placement. SAVI SCOUT facilitated resection in all cases, and reflector removal was confirmed radiographically and/or by lack of signal in the tumor bed. With a median follow-up of 77 days, one of seven patients recurred locally 7.5 months after resection, requiring re-resection. One patient developed new metastatic disease, and one had progressive metastatic disease. Conclusion: SAVI SCOUT technology facilitated resection of non-palpable recurrent sarcoma of the trunk and extremities in all cases attempted. There was only one instance of local failure in this high-risk population. Center, Nuevo Leon, Mexico; 3. Hospital Angeles Lomas, Mexico, Mexico; 4. Centro Oncologico Estatal, Chihuahua, Mexico; 5. Hospital Regional de Alta Especialidad del Bajio, Leon, Mexico. Background: Soft tissue sarcomas (STS) are uncommon tumors: they are rare, heterogeneous and with variable clinical behavior. Lymph node metastasis is uncommon and histology related. Because of the low incidence of regional lymph node metastasis, node-positive soft-tissue sarcoma patients remain poorly characterized. Our objective was to describe our experience in STS with lymph node metastasis Methods Retrospective cohort of patients treated for extremity STS from Jan/01/1990 to Dec/31/2017. 813 consecutive patients were analyzed. Clinicopathologic data and outcomes were examined to assess the incidence and clinical significance of lymph node metastasis Results 813 patients: 417 (51.3%) females and 396 (48.3%) males were included. Mean age was 45y (18-91). Median tumor size 12.4cm (1.8-78 cm) . Lower limb (specifically the thigh) was the most common location (49.2%), followed by the leg (15.8 %) and forearm (8%). Synovial sarcoma was the most frequent histology (26.2%), 24.6% cases were liposarcoma and 12.4% were UPS. Stage distribution was IIIB (27.1%); IV (26%); IIIA (23.7%). 44 (5.4%) cases had lymph node positive disease at presentation. Synovial sarcoma 14(31%), Epithelioid sarcoma (5)11.3%, UPS (5)11.36%, Rhabdomyosarcoma (4)9%, Clear cell (3) 6.8%. Median OS was 90 vs 17 (p=<0.001) months in N0 and N1 patients, it is to be noted that M1 patients had a better prognosis, their Median OS was 22.2 months, perhaps related to the frequency of chemotherapy received; 7(30.%) of N1 patients received adjuvant CT. 29% of M1 patients had node positive disease, and 25% of N1 patients had distant metastasis at presentation. Conclusion Lymph node metastasis cases are poorly addressed in most trial, in our cohort their prognosis was even worse than M1 patients, maybe related to low awareness of the outomes in this cases, and to the fact that few patients received CT and the high association with M1 stage and functional outcome of TSTS in a mono-institutional series focusing on full-thickness wall resections. METHODS: Patients affected by primary localized TSTS operated between 2000 and 2017 were extracted from institutional prospective database. Clinicopathologic variables and details on technique were retrospectively reviewed for patients who underwent a full-thickness wall resection. Reconstruction strategy and functional outcome were analyzed according to the following groups: abdominal wall (AW), thoracic wall (TW), paravertebral region (PV), groin (GR), and pectoral region (PE). 5-year local recurrence-free survival (LRFS) was calculated. RESULTS:151/549 (27.5%) TSTS patients were operated with a full-thickness wall resection and mesh repair. Median age was 50 years (IQR 34-64); 55.6% were male, and 44.4% female. Anatomic distribution was 86 AW, 26 TW, 11 PV, 25 GR, 3 PE. The most frequent histologic type was desmoid (21.2%) followed by undifferentiated pleomorphic sarcoma (9.9%). Tumor was deeply located in 60.3% cases; median size of wall defect was 108 cm 2 (IQR 40-520). The most used mesh material was polypropylene in 74.2%. There was no association between the type of mesh(p=0.519), anatomic size(p=0.415), or wall defect size(p=0.32) and the development of hernia. After a median follow up of 67.6 months(IQR 26-104) incisional hernia was experienced in 12 cases. 5-year LRFS for the whole cohort was 67.5%. LRFS for AW was 73.3%, TW 73.1%, PV 63.6%, GR 44% and PE 66.7%. DISCUSSION: Full-thickness resection is required in approximately 25% of TSTS. Although desmoids are differently treated than sarcomas, they represent the majority of full-thickness resections. Groin tumors are at higher risk of recurrence. Wall defect reconstruction is feasible, with an overall incidence of incisional hernia of 7.9%. Technical improvement for reconstruction is foreseen particularly in AW. Introduction: Alveolar soft part sarcoma (ASPS) is a sarcoma subtype that is often diagnosed in young patients following metastasis. ASPS is rare and represents <1% of soft tissue sarcomas. Due to the rarity of ASPS, previous studies have been limited to small series or national databank analyses; therefore lacking clinical information that is essential to patient management. In this study we evaluated patients with ASPS for potential risk factors of local recurrence, metastasis, disease-free survival, and overall survival. Methods: A comprehensive retrospective review of patient and tumor characteristics, treatments, and outcomes was performed in a consecutive series of patients with ASPS who presented to The University of Texas MD Anderson Cancer Center from 1977 -2019 . Results: Median follow-up was 2.75 years (range, 21 days-32 years). Sixty-two (41%) patients had localized disease and 90 (59%) had metastases at diagnosis (5-year overall survival (OS) of 74% vs 37%, respectively). Patients with a retroperitoneal primary ASPS had increased local recurrence rates (HR=12.13, p=0.02). Increased rate of metastasis was observed in patients with primary tumors >5cm (n=88, 58%) or with tumors locally invading bone (n=25, 16%) (p<0.0001). Reduced risk of local recurrence was observed in patients with primary tumor size <5cm or location in the head/neck (p=0.03 and p=0.01, respectively) with 5-year OS of 69% vs 19% and 57% vs 33%, respectively. Patients with metastatic disease who received primary tumor resection, lung metastectomy, or immunotherapy had improved OS compared to patients who received no primary resection, no lung metastectomy, or no immunotherapy 79% vs 27%, and 59% vs 32%, respectively) . Conclusions: Early diagnosis is essential as ASPS primary tumors <5cm predicted improved disease-free survival (DFS) compared to patients with larger tumors respectively) . Surgical management as well as immunotherapy show potential for improved survival in patients with metastatic disease. This information is useful for ASPS patient prognosis, counseling, management, and follow-up.

Introduction: Sentinel lymph node biopsy (SLNB) is recommended for intermediate thickness melanoma, but for thick melanoma, guidelines state that SLNB may be recommended only after a thorough discussion. We present a large multi-institutional study on thick melanoma evaluating for prognostic factors. Methods: The Sentinel Lymph Node Working Group database was queried for thick melanoma cases that had a SLNB from 1993 to 2018. Clinicopathologic characteristics were correlated with SLN status and melanoma-specific survival (MSS). Results: There were 1235 patients, and median follow up was 28 months. Median thickness was 5.9 mm, with 956, 175 and 104 cases having a thickness of >4-8, >8-12 and >12 mm, respectively. SLN metastases were seen in 439 of 1235 (35.5%) cases and in 33.9%, 40.6%, and 42.3% of melanomas >4-8, >8-12 and >12 mm, respectively. MSS was significantly worse overall and in each thickness group for positive compared with negative SLN cases (all p<0.005). For melanomas >4-8, >8-12 and >12 mm, 5-year MSS for positive SLN cases was 53.6%, 42.4% and 35.1%, respectively, and was 74.6%, 70.8% and 78.6%, respectively, for negative SLN cases. Multivariable analysis showed that SLN metastasis, male gender, increasing thickness, lymphovascular invasion (LVI) and microsatellitosis (MS) significantly predicted worse MSS for melanomas >4-8 mm, with SLN metastasis having the greatest risk (HR 2.17, 95% CI 1.64-2.87, p<0.0001). For all melanomas >8 mm, only SLN metastasis significantly predicted MSS (>8-12 mm: HR 3.93, 95% CI 2.00-7.73, p<0.0001; >12 mm: HR 3.58, 95% CI 1.56-8.22, p<0.0027) . Multivariable analysis also showed that age, thickness, ulceration, LVI, MS and histologic subtype predicted SLN status (all p<0.05). Conclusions: Thick melanoma patients with SLN metastasis have significantly worse MSS compared with negative SLN patients, even in the thickest cases, and SLN status is the most powerful and/or only predictor of MSS. Moreover, 5-year MSS of >70% is seen for thick melanoma patients who have a negative SLNB. These findings show that SLNB has important prognostic value in this population and is indicated for localized thick melanoma.

Recurrence Patterns After Negative Sentinel Lymph Node Biopsy for Melanoma S. Soelling,* K.A. Delman, M.C. Lowe. Emory University School of Medicine, Atlanta, GA.

The implications of a failed sentinel lymph node biopsy (SLNB) are reportedly significant, but data focusing on clinical nodal disease is variable with modern analyses failing to detail recurrence patterns. Earlier studies implied similar 5-year survival between those with false negative and true positive SLNB. This study aims to detail nodal basin recurrence characteristics and impact on survival. Retrospective review of patients with a diagnosis of cutaneous melanoma undergoing SLNB and subsequent lymphadenectomy between 1994 and 2018 was done. The cohort was divided into negative SLNB and mapped nodal basin recurrences (termed false negatives for this study) and those with positive SLNB (true positives). Demographics, clinicopathologic characteristics, patterns of recurrence, and survival were compared between groups. Data were examined with chi-squared tests and Kaplan Meier curves with log-rank test. 419 patients were identified, 69 (16.4%) had false negative (FN) SLNB and 350 (82.9%) had true positive (TP) SLNB. Mean age of FN at SLNB was 55.9 (14.9); 47 (68.1%) were male. For TP mean age was 51.5 (15.3) , 247 (70.6%) were male. Mean Breslow depth was 2.8 for FN and 3.3 for TP. 21 (30.4%) were ulcerated in FN and 149 (42.6%) were ulcerated in the TP group. FN had 11 (15.9%) local recurrences, 69 (100%) regional, and 36 (52.2%) distant; TP had 36 (10.3%) local, 42 (12%) regional, and 135 (38.6%) distant. Distant recurrences were higher in false negatives (p=0.04). The mean number of pathologically detected nodes in FN on lymphadenectomy for recurrence was 3.01 (3.4) . At median follow-up of 33 months, recurrence-free survival at 1, 2, 3, 4 and 5 years was 76.5%, 45.6%, 17.7%, 10.3% and 8.8% for FN and 76.8%, 59.2%, 49.2%, 43.8%, and 38.9% in TP, respectively (p<0.001). 5-year overall survival was 86.2% in FN and 68.6% in TP (p=0.11). Recurrence patterns in failed SLNB demonstrated more distant recurrences and worse recurrence-free survival than those with true positive SLNB. However, the overall survival was not significantly different between the two groups. Introduction: NCCN guidelines recommend that SLNB be discussed with patients that have thin melanomas at higher risk for lymph node metastasis (T1b or T1a with positive deep margins, lymphovascular invasion, or high mitotic index). We sought to examine the association between SLNB on resource utilization in patients with higher risk thin melanoma. Methods: A retrospective cohort of patients that underwent wide local excision (WLE) for higher risk thin melanomas from 2009-2018 at a tertiary care center was identified in the electronic health record, which was then merged with a patient-level, all-payer hospital financial database. Patients with higher risk thin melanomas who underwent SLNB were compared to those who did not undergo SLNB with regards to resource utilization, including total hospital cost and operative time. Multivariable linear regression was performed for these variables adjusting for clinically relevant patient and operative factors. Results: 83 patients met criteria and were included in the final analysis. 61 patients had T1b disease (73.4%), 21 patients had T1a lesions with positive deep margins (25.3%), and one patient had a T1a lesion and high mitotic index (1.2%). There were no T1a patients with lymphovascular invasion. 62 patients (74.7%) underwent SLNB. On bivariate analysis, WLE with SLNB was associated with increased hospital cost ($6,601 2,089 vs $3,972 1,827, p<0.01) and increased operative time (75.3 25.4 vs 38.6 16.7 minutes, p<0.01). These differences persisted on multivariable analysis (Table 1) . Of patients who underwent SLNB, 58 patients had a negative SLN (93.5%), 3 patients (4.8%) had a positive SLN, and in one patient the tracer did not localize to a nodal basin (1.6%). The mean cost to identify a single positive SLN was $54,333. Conclusion: In patients with thin melanoma at increased risk for nodal metastasis, SLNB is associated with increased cost and operative time. The probability of SLN positivity was low in this cohort, with a high cost incurred for the potential benefit of few patients. Reconsideration of SLNB, including evaluating survival outcomes and the cost of ultrasound surveillance, may be indicated in this clinical scenario. Durham, NC; 2. Rutgers Cancer Institute, New Brunswick, NJ; 3. University of Texas Medical Branch, Galveston, TX; 4. Thomas Jefferson University, Philadelphia, PA. Introduction: Preoperative lymphoscintigraphy is essential to localize sentinel lymph nodes (SLN) in melanoma especially in locations with ambiguous drainage patterns such as head, neck, and trunk. In 2014, technetium Tc 99m tilmanocept (tilmanocept) replaced technetium Tc 99m-sulfur colloid (Tc-99m SC) as the preferred lymphoscintigraphy agent at two institutions. This study compares the effectiveness of the two mapping agents in terms of mapping times, intra-operative identification of SLN, and false negative (FN) rate. Methods: Patients with truncal or head and neck melanoma who underwent sentinel lymph node biopsy (SLNB) between 2010-2018 were identified at the two study institutions. Retrospective patient demographic, tumor, and imaging data was stratified by receipt of Tc-99m SC (2010 -2018 or tilmanocept (2014-2018, n=134) for lymphoscintigraphy. Student's t-test and Mann-Whitney U test were used to compare characteristics and outcomes, including number of SLN identified, mapping times and FN rate, defined as FN/(FN+true positive). P values of <0.05 were considered significant. Results: The tilmanocept and Tc-99m SC groups were similar in age, gender and primary tumor characteristics including Breslow depth, presence of ulceration, and number of mitotic figures. Both tilmanocept and Tc-99m SC identified a similar total number of SLN (3.09 vs 3.25, p = 0.408) with a resulting similar total number of SLN removed intraoperatively (3.09 vs 3.04, p = 0.902). However, tilmanocept required significantly shorter lymphoscintigraphy mapping times (37.9 vs 85.6 minutes, p <0.01). Within the first ten minutes of mapping, tilmanocept identified 59.7% of total SLN compared to Tc-99m SC only identifying 29.1% within ten minutes (p<0.01). There was no significant difference in the number of patients with positive SLN (tilmanocept 17.9 vs Tc-99m SC 11.1%, p=0.083). The FN rate at median follow up time 23.2 months was similar for both agents, 33% for tilmanocept and 32% for Tc-99m SC. Conclusion: In comparison to Tc-99m SC, tilmanocept requires significant less lymphoscintigraphy mapping time while identifying similar number of SLN and maintaining similar FN rates.

Clinical Significance of SLN Benign Capsular Nevi in Patients with Melanoma T. Szabo Yamashita, 3 * B. Pockaj, 1 S.P. Bagaria, 2 T.J. Flotte, 3 A.S. Fahy, 3 R.U. de Azevedo, 3 W.S. Harmsen, 3 J.W. Jakub. 3 1. Mayo Clinic, Scottsdale, AZ; 2. Mayo Clinic, Jacksonville, FL; 3. Mayo Clinic, Rochester, MN. Background: Benign capsular nevi (BCN) are not infrequently identified in SLNs in patients with melanoma. These are not felt to have any prognostic significance; however, the literature on this is very scant. This study sought to define the clinical significance of incidentally found BCN in this patient population. Methods: Multi-institutional retrospective observational study (2000) (2001) (2002) (2003) (2004) (2005) (2006) (2007) (2008) (2009) (2010) (2011) (2012) (2013) (2014) (2015) (2016) . All patients who underwent SLN biopsy for cutaneous melanoma at Mayo Clinic destination sites were included. Patients were divided into four groups according to their SLN biopsy results: 1) Patients with negative SLN biopsy and no BCN; 2) Patients with negative SLN biopsy and presence of BCN; 3) Patients with positive SLN on IHC only; and 4) Patients with a single positive SLN via H&E analysis. Demographics and pathology features were analyzed and, hazard ratios (HR) for survival and recurrence rates are reported using multivariable analysis and Kaplan Myer curves. Results: One thousand two hundred and fifty-three patients were identified who met the inclusion criteria (Group 1= 978, Group 2=56, Group 3=32, and Group 4=187). 57.6% of patients were male, and mean age was 59.3 years, with no significant differences between groups. The most common melanoma subtype was superficial spreading (47.3%), followed by epithelioid (20%) and nodular (19.7%), with no difference in their distribution among groups. BCN was identified in 77 patients (6.2%), 72% in node negative group (56/77). On a multiple variable model, using the negative SLN biopsy group as reference, patients with BCN only had a decreased risk of recurrence, HR of 0.46 (p=0.06). HR for patients with IHC only disease was 2.38 (p=0.02), and patients with H&E disease an HR of 2.01, P<0.0001. Recurrence rates and overall survival Kaplan Meier survival curves are shown. Conclusion: Patients with BCN and negative SLN biopsy had lower recurrence rates than patients with a negative SLN biopsy and no BCN. Our data is the largest report of BCN in patients with melanoma and suggests a possible protective effect over recurrence.

Introduction MSLT have established SLN management in extremity/trunk melanoma, reproducibly demonstrating a 16% +SLN rate and 14% +NSLN in CLND. CLND improved DSS without benefit in OS with increased morbidity. The results of MSLT guide H&N melanoma but H&N site only represented Introduction: Esophagogastric cancer (EGC) carries a heavy mortality burden owing largely to high rates of unresectable disease at diagnosis. Among patients not undergoing curative-intent therapy, access to care may vary. We examined the geographic distribution of care delivery and survival across a province, and its relationship with distance to cancer centres (CCs), for non-curative EGC. Methods: We conducted a population-based analysis of adults with non-curative EGC from 2005-2017 using linked administrative healthcare datasets. Outcomes were medical oncology consultation, receipt of chemotherapy, and overall survival (OS). We used geographic information system analysis to map locations of CCs and outcomes across census divisions. Regions of discordance between care use and OS were identified with bivariate choropleth maps. Multivariable modified Poisson models assessed the relationship between distance to the nearest CC and outcomes, adjusting for demographic, clinical, and socioeconomic factors. Subgroup analysis assessed the relationship between distance and chemotherapy use among patients who first received medical oncology consultation. Results: Of 10,228 patients surviving a median of 5.1 months (IQR: 2.0-12.0), 68.6% had medical oncology consultation and 32.2% received chemotherapy. CCs providing higher-level care were distributed unevenly throughout the province and were concentrated in regions with higher oncology consultation, chemotherapy use, and OS. Greater distance from location of residence to the nearest CC ( 10, 11-50, 51-100, and 101 km) was associated with lower likelihood of seeing medical oncology and receiving chemotherapy, and inferior OS. Within the subgroup that received medical oncology consultation, only the greatest distance ( 101 km) was associated with lower chemotherapy use. Conclusions: Location of residence influenced access to care and OS, with inferior outcomes for those living further from a CC. In designing interventions to reduce disparities in access to care and outcomes for non-curative EGC, improving access to medical oncology consultation may decrease barriers to receipt of therapy.

Socioeconomic Deprivation Does Not Impact Gastric Adenocarcinoma Survival at a High-Volume Cancer Center B. Powers, 1 * M. Kocab, 2 S. Saeed, 1 R. Mehta, 1 J. Frakes, 1 S. Hoffe, 1 S. Dineen, 1 J.M. Pimiento, 1 P. Hodul. 1 1. Gastrointestinal Oncology, Moffitt Cancer Center, Tampa, FL; 2. University of South Florida, Tampa, FL. Introduction: Studies have shown that low socioeconomic status (SES) leads to less treatment and worse overall survival (OS) for gastric adenocarcinoma (GC) patients. However, these studies employed large geographic areas (zip codes/counties) for SES geocoding. To overcome these challenges, we used the Area Deprivation Index to assess 1) the degree of socioeconomic deprivation (SED) and 2) the impact of SED on OS in a cohort of resected GC patients. Methods: This retrospective cohort study included patients who underwent curative-intent resection for GC at a high-volume cancer center between 2000 and 2017. The ADI is a publicly-available, standardized measure of SED using 17 different SES variables, including income, education, and housing quality at the census block group level. Patient addresses were used to obtain ADI rankings, identifying 226 patients. ADI was categorized based on SED (Low SED 1-3, Moderate SED 4-6, High SED 7-10). The impact of neighborhood SED on patient and tumor characteristics was assessed with frequency and survival analyses. Results: The majority of patients were from high SED neighborhoods 88 (39.0%). Patients from low and moderate SED neighborhoods comprised 57 (25.2%) and 81 (35.8%) of the cohort, respectively. ADI was associated with racial disparities; Blacks and Hispanics had higher rates of SED than White patients (p=0.04). However, age, sex, insurance, ASA score, neoadjuvant or adjuvant therapy, pathologic T or N stage, grade, perineural invasion, lymphovascular invasion, and mortality were not associated with SED. Median OS was 46 months. After adjusting for potential confounders, SED was not a driver of OS ( Table 1) . Predictors of increased mortality were higher T and N stage, R1 resection, and multiorgan resection. Neoadjuvant and adjuvant treatment predicted decreased mortality. Conclusions: While nearly 40% of our patients come from high SED neighborhoods, this was not predictive of OS. SED was also not associated with neoadjuvant or adjuvant treatment. This study suggests that at a high-volume cancer center employing a standardized clinical pathway, stage and treatment quality are the primary drivers of OS, not SED. Introduction National administrative studies have shown that patients with resectable esophageal (EC) and gastric adenocarcinoma (GC) often fail to receive surgery. However, these reports used broad inclusion criteria and none explored the consequences of failure to resect. We assessed the rate of curative-intent surgery for resectable EC and GC patients and the population-based impact of failure to receive surgery. Methods The NCDB was queried from 2004-2015 for patients with clinically resectable EC and GC. Patients who were not surgical candidates (due to comorbidities, refused or died before surgery) were excluded. Frequency and survival analyses were performed. Using Cox regression models for each cancer type, we generated 5-year overall survival (OS) prediction models based on rate of surgery. Results 38,974 EC and 15,287 GC patients were identified as surgical candidates. 61.2% of EC and 72.2% of GC patients underwent curative-intent resection. Clinically significant demographic differences between patients who received surgery and those who did not were not observed. Patients treated at academic centers and high volume facilities ( 10 surgeries/year) received surgery more frequently for both cancer types. After adjusting for potential confounders (including chemotherapy and radiation), the greatest driver of mortality for both cancers was increasing stage, followed by failure to receive surgery (HR 1.67 (EC); HR 2.44 (GC)). Increasing receipt of surgery was associated with increased 5-year OS (p<0.001; Figure 1 ). For each 1% increase in resection rate, there was a 0.7% and 0.5% increase in 5-year OS for EC and GC respectively. Thus, a 20% increase in national surgery rates would improve 5-year OS by 14% and 10% for the resectable EC and GC population. Discussion Over 1/3 of EC and 1/4 GC patients with resectable disease do not receive indicated surgery. The reasons for this public health failure are unknown. If receipt of surgery increased, there would be a significant improvement in 5-year OS rates. Additional research is critical to guide policy interventions to increase delivery of standard of care treatment-curative-intent surgery-to EC and GC patients. BACKGROUND. Transoral endoscopic thyroid surgery vestibular approach (TOETVA) is a technique involving no skin incision. Since its first use in 2015, TOETVA has been adopted by several hospitals worldwide. TOETVA is expected to be a safe alternative to open surgery for certain patients and has been used increasingly by several surgeons around the world for the past years. The purpose of this paper is to review our 2-year head and neck oncology team experience and describe in detail our preoperative considerations, patient selection, operating room distribution, anesthetic considerations, evolution and modification in the surgical technique to make it more efficient in tissue management and operative time, postoperative management, and outcomes. METHODS. Between July 2017 and September 2019, 100 thyroid patients underwent TOETVA by a single endocrine surgeon. Analyses were performed on clinical and pathological findings, complications, and surgical completeness in total thyroidectomy cases, as indicated by the serum thyroglobulin (Tg) level. Data were collected prospectively and analyzed retrospectively. RESULTS. We have performed 100 thyroid surgeries (74 thyroid lobectomies and 26 total thyroidectomies) in 100 patients. Mean age 42 years. The percentage of females was 96%. Operation time (min) for lobectomy 95 and total thyroidectomy 135, the shorter OT were in the last third group (3 months to date). The mean modifications preoperative was the adition of arnica tablets 3 days before and etamsylate IV 1 hour before surgery. The operative techniques modification were efficient reflected in a shorter OT, less bleeding and better conservation of thyroid tissue and surrounding muscles. Percentage of stimulated Tg below 1.0 ng/ml was 85.0%. CONCLUSIONS. TOETVA is feasible in selected thyroid patients, not only because it is cosmetically advantageous but also because it is oncologically safe and faster recovery than open surgery.

Metastases R. Zheng-Pywell, 1 * H. Chen, 1 D. Abraham. 2 1. General Surgery, UAB, Birmingham, AL; 2. Christian Medical College Vellore, Vellore, India. 35-50% of patients with medullary thyroid cancer (MTC) present with lymphadenopathy. Serum calcitonin and CEA correlate with extent of disease. We were interested if calcitonin and CEA levels can predict mortality in those who presented with lymphadenopathy. 42 patients presented with lymphadenopathy only and underwent surgery for MTC between 2008-2014. They were divided into 3 cohorts: no recurrence, recurrent lymphadenopathy only, and recurrence with metastases. Pearson's 2 -test, Kolmogorov-Smirnov test, and Kruskal-Wallis test were used to compare cohorts. Of the 42 patients with lymphadenopathy: 21 had no recurrence, 4 had recurrent lymphadenopathy, 11 had new metastases, and 6 were lost to follow-up. Out of the 32 patients who followed up, 7 patients died. When comparing pre-op calcitonin and CEA levels amongst the mortality, no recurrence, recurrent lymphadenopathy, and recurrent metastases groups, no differences were found (p=0.67 and p=0.17 respectively). Similarly, no differences were found in post-op calcitonin or CEA levels (p=0.16 and p=0.10 respectively). In comparison to patients who initially presented with metastases (n=11, 0% mortality), there was a greater mortality in patients with recurrent metastases (p=0.02, 45% mortality). Those with recurrent metastases also had greater mortality than those without recurrence (p=0.01, 10% mortality), but there was no difference when compared to those with recurrent lymphadenopathy (p=0.10, 0% mortality). Between those whose disease recurred with new metastases and those who had initially presented with metastases, there was no difference noted in pre-op calcitonin and CEA (p=0.99 and p>0.99 respectively) or post-op calcitonin and CEA (p=0.06 and p=0.98 respectively). Mortality is significantly different between stable Stage IVC disease and recurrent Stage IVC disease. No statistically significant differences were seen in serum marker levels between those who died and those who have not. Serum markers are not good predictors of mortality. Differences in mortality between these two may reflect differences in disease biology despite similar staging.

Background Adenocarcinoma and squamous cell carcinoma of the esophagus have distinct outcomes, treatment strategies, and response profiles to therapy. Adenosquamous carcinoma (ASC) is thought to behave more aggressively than each of its counterparts. Our aim was to determine if ASC is best managed as adenocarcinoma or squamous cell carcinoma. Methods The National Cancer Database(2004-15) was queried for pts with non-metastatic, esophageal ASC. Analysis was stratified by clinical node negative(cN0) or node positive(cN1-3) according to AJCC 8 th edition. Treatment types were categorized into chemoradiation alone, surgery alone, or preoperative chemoradiation followed by surgery. Primary outcome was overall survival(OS). Results Among 352 pts, median age was 67 yrs, 80% were male(n=281). Median f/u was 46 mos. 43% were cN0(n=151), 57% were cN1-3(n=201). 55% had chemoradiation alone(n=194), 15% had surgery alone(n=53) and 30% had preoperative chemoradiation(n=105). Of pts who had preoperative chemoradiation, 20% had pathologic complete response(n=17). For either cN0 or cN1-3, Charlson-Deyo Comorbidity Index did not differ among the treatment groups(all p>0.05). On KM analysis for cN0 disease, treatment with surgery alone had a comparable 5-yr OS to preoperative chemoradiation(47vs34% p=0.5 Fig1A) and each had improved 5-yr OS compared to chemoradiation alone(30%; p=0.02; p=0.06 Fig1A). On UV analysis for pts with cN0 disease, clinical T-stage was not associated with OS. For pts with cN1-3 disease, however, preoperative chemoradiation was associated with improved 5-yr OS when compared to chemoradiation alone or surgery alone(27vs19vs0% p<0.01 Fig1B). This persisted even when accounting for age and clinical T-stage(HR 0.45 p<0.001). Conclusion Esophageal adenosquamous carcinoma behaves more like adenocarcinoma both in response to chemoradiation and survival outcomes based on the treatment modality. The complete response rate to chemoradiation is only 20% unlike what has been shown for squamous cell carcinoma, where chemoradiation is an acceptable definitive therapy. Esophageal adenosquamous carcinoma should be managed like adenocarcinoma and not squamous cell carcinoma.

(HR 1.66, 1.33-2.06) and not living in proximity to a metro area (HR 1.27, 1.10-1.47). Conclusions: Socioeconomic factors shown to have worse OS in patients with mNETs were lower income, lower education, treatment at community cancer program and not living in proximity to a metro area. Patient demographic and socioeconomic factors play an important role in OS for patients with mNETs and access to care must be considered and optimized in this subpopulation of cancer patients. Background: Non-melanoma skin cancers (NMSC), including cutaneous squamous cell (cSCC) and basal cell carcinomas (BCC) are among the most prevalent malignant neoplasms. Treatment options for advanced disease are limited. The poliovirus receptor, CD155, is an immunoregulatory cell-adhesion molecule overexpressed in many malignancies including melanoma and head and neck SCC. It can be associated with an unfavorable prognosis and has been suggested as a potential therapeutic target but has not been studied in NMSC. Methods: Histologically confirmed cSCC (n=18) and BCC (n=24) FFPE specimens were collected at two institutions. Immunohistochemistry staining was performed and the percentage of tumor cells expressing CD155 and PD-L1 and lymphocytes expressing PD-1 were analyzed manually by a pathologist. Density of CD45+ lymphocytic infiltrate was manually graded as high or low. Relative marker expression is reported as the mean +/-standard error of the mean and are compared using nonparametric tests. Correlations of marker expression were assessed using Pearson correlation. Results: Both cSCC and BCC tumor cells expressed similarly high levels of CD155, with mean expression of 74.1% +/-4.7 and 56.7% +/-6.5 respectively. PD-L1 expression on tumor cells was also higher in cSCC vs BCC (21.0% +/-7.0 vs 0.9% +/-0.5, p<0.0001) while PD-1 expression was similar in both tumor types. An increased proportion of cSCCs had high levels of CD45 infiltration compared to BCC. CD155 and PD-L1 expression trended towards a positive correlation (r=0.31, p=0.051) and CD155 expression was higher in tumors that were found to be CD45 high. Conclusions: In this study we characterize expression of CD155 in BCC and cSCC. Similar to previously reported tumors, NMSCs highly express CD155, which may correlate with other markers of immune dysregulation. CD155 may be a novel therapeutic target for patients with locally advanced NMSCs, with potential for synergistic effects when combined with existing strategies of immune checkpoint blockade.

Representative hematoxylin and eosin (H&E) and immunohistochemistry stains for markers CD155, PD-L1, PD-1 and CD45 are shown for cutaneous squamous cell (cSCC) and basal cell carcinomas (BCC).

Increased BMI is Protective Against Hypocalcemia in Patients Undergoing Total Thyroidectomy for Thyroid Cancer D.A. Mahvi,* R.G. Witt, H. Lyu, M. Nehs, G. Doherty, N. Cho. Brigham and Women's, Boston, MA. Introduction: Hypocalcemia following total thyroidectomy is a common event due to parathyroid dysfunction after surgery. Vitamin D deficiency is closely associated with obesity. We hypothesized that obese patients would have higher rates of hypocalcemia following total thyroidectomy. Methods: We performed a retrospective study of all total thyroidectomies performed from 2016 to 2019 at our institution after the implementation of a new post-thyroidectomy calcium supplementation protocol. Patient characteristics and outcomes were measured including post-operative hypocalcemia (Ca<8.3), operative time, surgical complications, and hypocalcemic symptoms. Results: A total of 288 total thyroidectomies were performed for thyroid cancer during the 3-year study period. Overall, hypocalcemia requiring calcium supplementation adjustment and an afternoon calcium level check (Ca<8.3) occurred in 86 patients (30%), and symptoms of mild hypocalcemia occurred in 65 patients (22.6%). Patients with overweight or obese BMI had higher mean postoperative Ca compared to normal BMI patients (8.62 vs. 8.48, p=0.042) . In a subset of 228 patients who had a preoperative Ca level, increased BMI correlated with a smaller decrease in postoperative Ca compared to normal BMI (-0.80 vs. -0.97, p=0.025). On multivariate analysis, patients with central neck dissection (OR 2.124; p=0.036) and lower preoperative Ca (OR 0.240; p=0.001) were more likely to develop hypocalcemia. BMI was found to be protective against hypocalcemia on multivariate analysis (OR 0.942, p=0.02). Conclusions: Despite the association between obesity and vitamin D deficiency, we found that increased BMI does not correlate with postoperative hypocalcemia following total thyroidectomy and may be protective. Risk factors associated with hypocalcemia included central neck dissection, low preoperative calcium and lower BMI. These higher risk groups may benefit from preoperative supplementation with calcium and/or rocaltrol to decrease symptomatic hypocalcemia postoperatively and length of stay.

Fluorescent-Image Guidance in Robotic Subtotal Gastrectomy N. Ikoma,* B. Badgwell, P. Mansfield. Surgical Oncology, MD Anderson Cancer Center, Houston, TX. Robotic surgery technology has significant advantages, but its limitations include lack of tactile feedback. Fluorescent-imaging technology, part of the da Vinci robotic surgery system, helps to overcome this lack of feedback and improve safety. This video demonstrates the utility of fluorescent-image guidance in robotic subtotal gastrectomy. First, peritumoral injection of indocyanine green helps localize the primary lesion. Second, it enables visualization of sentinel lymphatic flow to guide the extent of lymph node dissection. Third, intravenous injection of indocyanine green confirms adequate tissue perfusion of the remnant stomach. Lastly, fluorescent angiography also visualizes the jejunal arterial arcade in the mesentery. In summary, we consider fluorescent-image guidance is useful to maintain safety and quality of robotic subtotal gastrectomy. INTRODUCTION Small, well differentiated, neuroendocrine tumors (NET) of the duodenum that are not amenable to endoscopic resection provide a unique challenge in management. Resection with negative margins is sufficient for good oncologic outcomes, but these tumors may be hard to localize with minimally invasive surgery (MIS). Here, we describe an approach combining endoscopic and MIS modalities, allowing for an adequate resection of a small, duodenal NET in a morbidly obese gentleman with significant medical comorbidities. CLINICAL PRESENTATION The patient was a 47-year-old, morbidly obese, male with a history of diabetes and obstructive sleep apnea who was incidentally found to have a 1-2 cm mass in his duodenal bulb. Endoscopic ultrasound (EUS) with biopsy was performed and cytology showed a well-differentiated NET, grade 1, Ki67 2%. His CT scan showed no evidence of metastatic disease. Submucosal endoscopic resection was attempted but was aborted due to a concern for potential incomplete resection. RESULTS The patient underwent a combined endoscopic and robotic full thickness resection of this duodenal bulb NET. The DualKnife™ electrosurgical knife was used to perform a submucosal dissection around the mass. Full thickness resection was then performed using a combination of the electrosurgical knife and the robotic Harmonic ACE® scalpel. The resulting defect in the duodenum was closed with a running suture. Final pathology was consistent with 1.5 cm, welldifferentiated, NET, grade 2, Ki67 8%, 1 mitoses/10 hpf. All margins were negative. The patient was discharged home on postoperative day 2. DISCUS-SION Small, well-differentiated, NET of the duodenum do not require radical resection and lymphadenectomy. Some lesions cannot be endoscopically resected due to location, size, and/or depth. We have described an approach combining endoscopic resection with minimally invasive robotic resection.

A.D. Wisneski,* C.U. Corvera. Surgery, Univ CA San Francisco, San Francisco, CA.

Background: Robotic-assisted techniques in surgical oncology benefit patients who can undergo minimally invasive resection for tumors in difficult anatomic locations. Visualization and articulation of the robotic instruments enhance dissection and resection. Gastrointestinal stromal tumors (GIST), given their typically benign nature and locations in the foregut, are often amenable to robotic-assisted procedures. Methods: We detail three cases of robotic GIST resection performed by a single surgeon at our institution: an endophytic gastric GIST, an exophytic gastric GIST, and a duodenal GIST. Results: Case 1: A 72yo male with prior stroke on clopidogrel had a 3cm endophytic gastric GIST adjacent to the gastroesophageal junction. PET-CT demonstrated an avid lesion with no metastasis. The mass was successfully resected with pathology revealing low grade tumor with <5 mitotic figures/5mm 2 . Patient was discharged home on postoperative day (POD) 3. Case 2: A 64yo female was diagnosed with a 7cm exophytic gastric GIST, adjacent to the splenic hilum. Patient was discharged home on POD 1. The tumor was high grade, with 15 figures/5mm 2 , and adjuvant imatinib was initiated postoperatively. Case 3: 63yo female with coronary artery disease was found to have a 4cm duodenal GIST after evaluation for abdominal pain. Neoadjuvant imatinib was started as the patient underwent cardiac clearance, and tumor demonstrated shrinkage. Tumor was low grade with 3 mitotic figures/5mm 2 . An upper GI series confirmed no leak from duodenostomy on POD 4; patient was discharged home on POD 7. Conclusions: All patients underwent successful GIST resection with aide of the robot, with negative margins and no post-operative morbidity. Robotic-assisted GIST resection can be performed effectively, adhering to principals of oncologic resection. Though traditional laparoscopic methods could be used, we strongly believe that robotic techniques enhance the surgeon's capabilities when resecting tumors in difficult anatomic locations.

Robotic Excision of a Large Gastroesophageal Junction Mass K. Yendamuri,* J.S. Peng, S.N. Hochwald, M. Kukar. Surgical Oncology, Roswell Park Comprehensive Cancer Center, Buffalo, NY. In this video, we describe our technique for a robotic resection of a large complex multilobulated leiomyoma at the gastroesophageal junction (GEJ). 3D visualization and robotic instruments dexterity helped us accomplish this safely in a challenging anatomical location. The patient is a 32 year old female, who was found to have large GEJ mass on upper endoscopy (EGD) and subsequent CT scan, for upper abdominal discomfort. Endoscopic Ultrasonography and fine needle aspiration cytology was consistent with leiomyoma. Safe resection of the leiomyoma was accomplished utilizing 4 robotic ports, 1 assistant 5 mm port and a nathanson liver retractor. After careful removal of the tumor, the mucosal layer was completely intact and the muscle layer was closed over it using interrupted 2-0 silk sutures. The closure was reinforced with pedicled omental flap. EGD and leak test was negative. Post operatice course was unremarkable and final pathology was consistent with Leiomyoma. Introduction: We present a patient who underwent a robotic total gastrectomy with intracorporeal stapled anastomoses Methods: The patient is a 71-year-old male who presented with dysphagia and abdominal pain. He was diagnosed with a Siewert type III gastric adenocarcinoma, T1aN0. He underwent endoscopic mucosal resection with positive lateral and deep margins, pT1bN0. We elected to proceed with surgical resection Results: Key steps of the operation are demonstrated in the accompanying video. The gastrocolic ligament is divided, and the transverse colon and mesocolon separated from the omentum. The greater curvature is mobilized by dividing the short gastric vessels. The duodenum is mobilized and the gastroepiploic vessels are dissected and divided. The right gastric artery is dissected and divided. The duodenum is divided. The lesser omentum is divided. The left gastric artery and vein are dissected and divided. The right and left crura are dissected and exposed. The phrenoesophageal ligamenet is incised and the anterior body of the esophagus exposed. The distal esophagus is mobilized. The distal esophagus is divided and the specimen extracted through a Pfannenstiel incision. Frozen section on the proximal margin was noted to be negative for cancer. The proximal jejunum is divided 25 cm from the ligament of Treitz, and brought retrocolic to the esophagus. A 25 mm Orvil is passed through the mouth and through the esophageal staple line. The staple line on the small bowel is removed. An EEA stapler was passed through a gelport at the Pfannenstiel incision and through the end of the small bowel. A stapled esophagojejunostomy is created. The common channel is closed. At 50 cm distal to the esophagojejunostomy, a jejunojejunostomy is created. Completion endoscopy demonstrated intact esophagojejunostomy and leak test is negative. Final pathology demonstrated T1b adenocarcinoma with negative margins and 0 of 24 lymph nodes positive. The patient recovered uneventfully and was discharged home on postoperative day 7. Conclusion: Minimally invasive robotic total gastrectomy with intracorporeal reconstruction can be offered to patients with safe technical and oncologic results.

Robotic Pre-peritoneal Iliac, Obturator and Inguinal Lymphadenectomy C. Contreras.* The Ohio State University, Columbus, OH.

Introduction: A 30-year-old woman presents with a mole of her midline gluteal cleft. Her initial biopsy shows a 3.1 mm depth ulcerated melanoma with elevated mitoses. Physical examination reveals a 2 cm palpable right inguinal mass. Imaging shows isolated right inguinal disease without distant metastasis. Fine needle aspirate of the right inguinal mass confirms melanoma. She then undergoes robotic pre-peritoneal iliac, obturator and inguinal lymphadenectomy. Methods: After resection of the primary melanoma site, the patient is placed on a split-leg table in the supine position. A dissecting balloon establishes the preperitoneal space. Three robotic arms are utilized for the iliac and obturator lymphadenectomy. The specimen is withdrawn and attention is turned to the femoral triangle. The subcutaneous inguinal space is initially developed with standard hand-held electrocautery dissection. Three robotic arms are utilized for the inguinal lymphadenectomy. The specimen is withdrawn and a single 19-French Blake drain is placed within the inguinal field. Conclusions: The patient tolerated the operation without complication and is discharged home on the first postoperative day. Her final pathology report indicates negative margins of the primary melanoma resection site. There were two of eight metastatic nodes in the iliac and obturator basin, and 2 of 13 metastatic nodes in the inguinal basin. The patient is currently receiving adjuvant PD-1 therapy. BACKGROUND: Laparoscopic versus open hepatocellular carcinoma (HCC) resection has been shown to reduce morbidity without a compromise in oncologic safety. Moreover, in the subgroup of cirrhotic patients, a decreased risk of prolonged postoperative ascites and liver decompensation has been reported. PATIENT: A 54-year-old homeless, deaf, male, with chronic alcoholism, hepatitis C and advanced cirrhosis was found to have a caudate tumor. Imaging showed a 3.6 cm HCC in the caudate lobe, associated with the Inferior Vena Cava (IVC). TECHNIQUE: With the patient in reversed, modified French position, the liver was mobilized, and the hepatocaval space dissected. Portal and short hepatic vein branches were individually controlled, and the caudate lobe dissected off the IVC. At the superior portion of the Spiegel process the tumor was inseparable from the IVC, necessitating en-bloc segment I with IVC resection. The IVC was reconstructed laparoscopically following a preplanned approach. Pathology report confirmed R0 resection of a moderately differentiated hepatocellular carcinoma without vascular or perineural invasion (pT1bN0M0). CONCLUSION: Laparoscopic caudate lobectomy in cirrhotic patients with IVC resection is technically demanding. It therefore requires a strategic and preplanned approach with dedicated instrumentation and laparoscopic skills available. While the laparoscopic caudal view along the axis of the IVC facilitates dissection, a laparoscopic approach necessitates particular attention to central venous pressure management (intravenous fluid and respiratory tidal volume management), meticulous control of portal and short hepatic vein branches, and availability of specialty laparoscopic instrumentation to ensure procedural safety. INTRODUCTION: This is the case of a 35 year old woman who presented with right upper quadrant pain. She was diagnosed with a Type I choledochal cyst. She elected to proceed with a robotic choledochal cyst resection and Roux-en-Y hepaticojejunostomy. METHODS: The operation was performed with a combination of laparoscopy and use of the DaVinci Si Robotic Surgical System PROCEDURE: We began laparoscopically with the mobilization of a Roux limb of small bowel and creation of a side-to-side jejuno-jejunal anastomosis. This was followed by a careful dissection of the biliary tree. A roughly 2cm choledochal cyst was identified with its distal margin very close to the bifurcation of the hepatic duct. This required an extensive dissection at the base of the liver and at the proximal margin of the cyst near the duodenum. The cyst resection and hepaticojejunostomy were performed after the robot was docked. A two-layered anastomosis was created and confirmed to be leak and tension free. The patient tolerated the procedure well and was discharged on postoperative day 3. Her follow up imaging demonstrates a patent anastomosis without signs of stricturing. CONCLUSION: This was a laparoscopic and robotic choledochal cyst resection with hepaticojejunostomy. The case presented unique challenges, including a difficult dissection of the distal biliary tree and a nearly intrahepatic anastomosis. The robotic approach allowed for high technical precision in a difficult anatomic location and also allowed for excellent visualization. This video is of a robotic total anatomical left hepatectomy with caudate lobe resection and microwave ablation of segments VII and VIII. The patient is a 76-year-old man with a history of colon resection status post neoadjuvant chemotherapy. Pre-operative CT scan showed left sided liver lesions with two additional lesions in the right side of the liver. The left hemi-liver was mobilized from the diaphragm and abdominal wall by dividing the falciform, left coronary, and triangular ligaments. The patient had a superficial left hepatic cyst which was unroofed. The gastrohepatic ligament was taken down, exposing the caudate lobe. Lesion was confirmed in caudate lobe via ultrasound. The short hepatic veins were individually isolated with hook cautery prior to division between clips. Hook cautery and an extended vessel sealer device were utilized to transect the caudate lobe. With great care, the left hepatic artery and left portal vein were isolated for inflow control. Staying medial to the left hepatic vein, preserving the middle hepatic vein the superficial liver parenchyma was undertaken with hook cautery, followed by a vessel sealer for deeper liver parenchyma. A linear stapler was used to carefully transect the left hepatic bile duct and left hepatic vein, completely removing the left hemi-liver. Cholecystectomy was undertaken for cholelithiasis. Using real-time ultrasound guidance, the lesions in segments VII and VIII were identified and microwave ablated with 90 watts of energy for 5 minutes, resulting in a 3.5 cm zone of ablation. This video demonstrates a safe and efficacious application of the robot to an extremely technical and complex abdominal operation. Introduction: Advances in breast cancer diagnosis and management have significantly improved cancer related survival. Patient's satisfaction with their post-surgery quality of life and psychosocial wellbeing is prudent for women and these changes focus from the surgeon's to patient's perspective. Objective of this study was to develop a new patient-reported outcome measure to assess outcomes of breast reconstruction in an Indian cohort. Material and methods: A prospective study is being conducted in patients of breast cancer who underwent Latissimus dorsi flap reconstruction from June 2017 till present. A score has been developed including all parameters pertaining to first preoperative interaction to post surgery quality of life and psychosocial wellbeing. Total 25 parameters were included. Each parameter was rated with three categories with score of one, two and three respectively. Score value ranged from 25 to 75, 25 being the worst and 75 the best and final outcome was categorized as unsatisfactory (<25), good (25 to 50) or excellent (51-75). It was applied after 6 weeks of radiation therapy and named as 'Rajeev Garima Latissimus Dorsi Score' (RGLDS). Results: Score has been applied to 100 patients who underwent latissimus dorsi flap reconstruction followed by radiotherapy. Mean RGLDS was 52±4. Conclusion: The RGLDS can be used to study impact and effectiveness of breast reconstruction from patient's point of view. By including various parameters present score has potential to support an evidence-based approach to modern surgical practice. Background. Focus has moved to de-escalate of axillary surgery in the management of early breast cancer. Based on the American College of Surgeons Oncology Group-Z0011, our team created the nomogram to be useful for identifying patients who do not require intraoperative analysis of the sentinel lymph node. The nomogram could identify as patients likely to three or more positive axillary lymph nodes by preoperative imaging. But the nomogram had several limitations. This study investigated a developed nomogram by excluding the chest computed tomography (CT) and adding 18F-Fluorodeoxyglucose (FDG) positron emission tomography (PET)/CT as a modality. Methods. Patients underwent preoperative ultrasonography (US) and PET/CT. The training set consisted of 1030 patients with clinical T1-2 and node-negative invasive breast cancer. Factors associated with 3 involved axillary lymph nodes (ALNs) were evaluated by logistic regression analysis. The validation set consisted of 781 independent patients. The nomogram was applied to 1,067 patients who met the selection criteria of the ACOSOG-Z0011. Results. Of the 1,030 patients, 89 (8.6%) had 3 positive nodes. Multivariate analysis according to be used for making a new nomogram to predict the probability of involvement 3 nodes, there was significantly associated with larger tumor size, higher grade ultrasonographic ALN classification, and findings suspicious of positive ALN on PET/CT. The areas under the receiver operating characteristic curve of the nomogram were 0.856 (95% confidence interval [CI], 0.815 to 0.897) for the training set and 0.866 (95% CI, 0.799 to 0.934) for the validation set. Application of the nomogram to the patients who met the ACOSOG-Z0011 showed that 90/1067 (8.4%) had scores above the cut-off and the false-negative rate was 37/977 (3.8%). And the specificity was 93.8%, and the negative predictive value was 96.4%. Conclusion. The upgraded nomogram improved predictive accuracy, and that was possible using only US and PET/CT. It reduced operation time and cost, with a very low re-operation rate. This nomogram is useful for identifying patients who do not require intraoperative analysis of sentinel lymph nodes. Introduction: There has been an increase in breast-conserving surgery (BCS) with oncoplastic techniques in patients with early stage breast cancer. Due to advances in neoadjuvant and adjuvant therapies, selection criteria for patients who are candidates for BCS have been amplified. Perks of BCS over mastectomy include: less morbidity, more aesthetic results and an increase in patient satisfaction. There has been a greater rate of complications reported for oncoplastic techniques compared to traditional BCS. In the case of positive margins in oncoplastic surgery, the widening could result difficult due to rearrangement of the tissue during oncoplastics. Nevertheless, the oncoplastic surgery permits wider surgical margins, so the risk of having positive margins should be reduced. Objective: Report early post-surgical complications and the assessment of positive margins in the pathological report in BCS and compare the results between oncoplastic and conventional techniques. Materials and Methods: Medical records of 108 patients with early-stage breast cancer undergoing BCS were reviewed. Post-surgical complication rates were compared between both techniques. As well as positive surgical margins. Categorical variables were analyzed using ². The results are presented as percentages. For data analysis, the SPSS v23 program was used. Results: A total of 108 records of patients diagnosed with early stage breast cancer treated at a Breast Diseases Institute undergoing BCS were analyzed. They were divided into conventional(n=75) and oncoplastic technique (n=33). The oncoplastic techniques performed were 20 with a circular pattern (60.6%), 6 with a lateral pattern (18%), 5 with a double branch pattern (15%), 1 with a vertical branch pattern (3%) and in one patient a bilateral mastopexy was performed. Regarding post-surgical complications in conventional BCS: seroma (22.7%), necrosis (1.3%), wound dehiscence (1.3%), no infection or re-intervention due to bleeding was reported. Regarding post-surgical complications in BCS with oncoplastic technique: seroma (15.2%), infection (6.1%), wound dehiscence (6.1%), no necrosis or re-intervention due to bleeding was reported. Positive margins in conventional BCS were 2.7% and 3% in oncoplastic surgery. Conclusion: Results similar to those already described in the literature for the oncoplastic and conventional technique group are shown. There is a significant difference (p= 0.031) when comparing post-surgical infections; the oncoplastic technique being unfavorable in the studied sample. 3 Centro Oncologico Estatal. 4 Background: Access to quality healthcare in LMICs has gained global attention and is common knowledge that lower income patients have poorer outcomes. In Mexico there are 1027 board certified surgical oncologists meaning 1/221 breast cancer (BC) cases. Initial diagnosis/treatment is often provided by non-oncologists. National regulations do not specify who should treat BC. Guideline adherence measurement is not common in private practice settings. It is not known if initial workup and treatment are better or worse in this group. Material and Methods: Retrospective review of a series of nonmetastatic BC cases referred to a private center in northern Mexico during Sept 1, 2017 to Sept 1, 2019. Patients with no follow up, and non-epithelial tumors were excluded. We compared guideline adherence in diagnostic workup and initial treatment between board certified surgical/gynecologic oncologists and general surgeons/gynecologist. Descriptive analysis is presented. Results: 101 cases were identified and 14 eliminated. All females with a median age of 53yo. Only 31% were diagnosed by screening. 87.5% of patients had a private health insurance plan and 90% had a college degree. A core needle biopsy was done in only 55% cases, an excisional biopsy with immediate radical mastectomy occurred in 41%. 52(59%) cases had no ER/PR prior to first treatment. 25% of cN0 received an axillary dissection. Despite a median tumor size of 2cm, only 25% patients had a breast conserving surgery, MF disease was uncommon 4(4.5%). 60% cases were first approached and treated by non-oncologists, who adhered to 2019 Mexican guidelines and the NCCN guidelines in only 22% cases; cases diagnosed/treated by board certified oncologists did so in 90%. 10% received an immediate contralateral mastectomy, not one of them had BRCA mutations or a high-risk family history. Discussion: Disparity of access to quality health care in LMICs is a reality, however in a population with Background: Current clinical guidelines recommend the routine use of adjuvant chemotherapy (AC) for locally advanced rectal cancer (LARC) patients. However, the effects of adjuvant chemotherapy in patients with pathological complete response (pCR) after neoadjuvant chemoradiotherapy (NACRT) and radical surgery showed discrepancies in different investigations. Objective is to determine whether patients with LARC who achieve pCR after NACRT and resection benefit from the administration of AC. Methods: A retrospective analysis of prospectively maintained data was performed at a tertiary cancer center in the Indian subcontinent. All consecutive patients who received curative treatment for adenocarcinoma rectum, stage T3/T4 and node positive disease from January 2012 to December 2017 were included in the study. Non adenocarcinoma, metastatic patients and patients who received NACT were excluded from the study. Depending on whether adjuvant chemotherapy was used, the patients were categorized into observation arm (Obs-Arm) and adjuvant chemotherapy arm (AC-arm). Long-term outcomes were analyzed. Result: A total of 93 patients were included in the study, 66 patients (70.1%) in AC-arm and 27 patients (29.9%) in Obs-arm. The demographic and disease characteristics were comparable in both the groups. Study duration was 48 months. The mean follow-up (FU) was 39 months. The pCR rate in our study was 17.9%. Three-year disease-free survival (DFS) for AC-arm was 69.2%, whereas those of Obs-arm was 82.1% (P = 0.563). Though Obs-arm had better 3-year DFS but it was not statistically significant. There was no difference in 3-year overall survival (OS) in both the groups (P = 0.787). There was no statistically significant OS or DFS difference between two groups. However, more early events were noted in AC-arm. Adding adjuvant chemotherapy in pCR patients did not improve DFS or OS. Conclusion: The administration of adjuvant chemotherapy in patients with rectal cancer with pCR has not shown any improvement in survival. Debate of administration of adjuvant chemotherapy in pCR patients is still exist.

Although this study did not suggest a beneficial effect of adjuvant treatment on survival in patients with pCR, these results are limited by the presence of potential unmeasured confounding in this nonrandomized study. This is the largest Indian dataset on this topic (to the best of our knowledge). There is a need for large prospective studies or randomized controlled trial, preferably multi-centric and collaborative. Background: Lymph node yield is an important factor to ensure adequate staging in rectal cancer. The minimal nodal yield to be achieved has often been established to be 12. The effect of neo-adjuvant radiation on nodal yield has showed contrary results, with some studies hinting at better outcomes for those patients with lesser nodal yield. This study intends to assess nodal yield in rectal cancer after neo-adjuvant therapy, and its prognostic implications with regard to disease recurrence, thereby hopefully shining some light on this dilemma. Methods: Patients who underwent curative resection of rectal cancer after neo-adjuvant therapy in the Department of Surgical Oncology, Regional Cancer Center, Thiruvananthapuram, Kerala, India from 2016-2018 were included. Nodal yield was assessed and factors including patient characteristics, disease parameters and treatment details were collected and analyzed retrospectively. Patients were then followed-up and oncological outcomes were correlated with nodal yield. Results: A total of 501 consecutive patients who fulfilled the criteria were included in the study period of 3 years. 53.9 % had lower 1/3rd primaries and 60% underwent sphincter preserving resection. 87.9% of patients had pre-op NACRT. Mean nodal yield was 12.3 nodes with a range of 5-42. 13.4 % patients achieved pCR whereas 67% were pN0. On multi-variate analysis, nodal yield significantly increased with increasing pathological stage (p=0.002) and positive lymphovascular invasion (p=0.021). No difference was found between various surgical approaches, site of primary or NACRT Vs SCRT. After a Median follow-up of 2 years, 12.5% recurred of which locoregional recurrence comprised a mere 1.7%. Disease free survival probability at 3 years was 84.1% for those with 12 nodes (p= 0.253). This trend was more so among those who attained pCR (96.1% Vs 92.9%; p= 0.677). Conclusions: Nodal retrieval from rectal resections is the result of interplay between multiple factors. It has been found to be strongly associated with the presence of LVE and increasing pathological stage. <12 lymph nodes may not be a detrimental factor, rather, may be a surrogate marker of better response to neo-adjuvant therapy and thereby may indicate better oncological outcomes. Purpose: The wide implementation of multi-disciplinary approach in cancer care has been shown to decrease variability between physicians' recommendations, greater likelihood of care delivery in accordance with national standards and clinical practice guidelines and improved quality of life and survival. Such tumor board discussions allow efficient and better communication between different physicians, decreased errors and duplicate tests, and provide clear treatment plan. The purpose of the current study is to evaluate the impact of MDT discussion on patient selection for CRS and HIPEC in patients with peritoneal surface malignancy. Methods: The study period was from 3/2016 through 9/2019. Patients with peritoneal surface malignancies from primary gynecological (ovarian, uterine, cervical and vaginal) and colorectal malignancies diagnosed at the NCI, Cairo, Egypt were presented at a newly formed weekly MDT conference where relevant medical history, co-morbidity, radiological and pathological information were discussed and recommendations recorded. Adequacy of the members' attendance and adherence to the final recommendation were also reported.

Results: During the study period, 780 patients were prospectively enrolled and discussed in a weekly MDT conference. Attendance of required specialties was 95%. These included patients with primary gynecological malignancies and 330 patients with colorectal cancer. Most referrals for MDT conference discussion were from surgical oncology (342 patients, 76%), radiation oncology (54 cases, 12%), medical oncology (29 cases, 6.4%) and radiology (25, 5.6%). None of these patients had prior CRS and HIPEC. Of those 780 patients, 195 (25%) were previously evaluated by a single specialty and deemed candidate for CRS and HIPEC (118 gynecological & 77 CRC) patients. After MDT conference review and discussion of these 195 patients, only 20 patients (9.7%) were deemed appropriate candidate for CRS & HIPEC (12 gynecological & 8 CRC) patients. Conclusion: The diverse clinical expertise of the MDT members is essential and critical in identifying patients with peritoneal surface malignancies who would benefit the most from CRS & HIPEC. This is of paramount importance for achieving positive objective impact on patients' prognosis and quality of life, cancer service efficiency and cost effectiveness in countries with limited resources. Introduction. The rectal cancer (RC) is the fourth cancer disease among the world between men and women. The advances on the treatment have accomplished give them more survive, making necessary the study on the quality of life in this area. Because of the surgical treatment the complication rates of several series around the world, its having a 25% of frequency. That is why, it is time to ask about the quality of life, if this could be strong enough to impact on the decision about the surgical treatment. The European Organization Research on Cancer Treatment (EORT), has the Quality of life (QOL) Core 30 (C30) and Core 29 (C29) questionnaires, that contains 30 items about symptoms, functional status, global quality of life. The C29 contains items that evaluate specific symptoms and function of RC patients. Materials and Methods. In a year of RC treatment (2017) the EORT C30 and C29 questionnaires were applied to the patients who wanted to participate in the trial (10 patients), with the inclusion criteria of being surgically treated with abdominoperineal resections or anterior low resections of the rectum and coloanal anastomosis with colostomy. Results. The study can be analized in 3 gropus: symptoms, functional status and quality of life. According to the C30, the symptoms with more impact on quality of life were the financial difficulty (36.7%), fatigue (29.1%) and pain (25.8%). The less often were constipation (7.3%) and anorexia (3.5%). On the symptom scale in C29, which predicts directly morbidity of colorectal surgery, the more and less often were dyspareunia (50%), urinary frequency (38.5%), anxiety (29.4%), and troubles with stomal care (0%). The functional scale on C30 inquest, showed an adequated preserved: cognitive function (98.1%), emotional function (87.09%), social function (83.3%), physical function (80.7%). C29 by its side, showed the corporal image preserved (91.6%), sexual function on men (52.8%) and woman (67%). The global scale on quality of life on C30, showed a promedy of 73.1%, it means, a 26.1% of quality of life reduction, with a minimum value of 58% and a maximum of 91.6%.

Thyroid Malignancies Amongst Patients with Thyroid Swellings -A Clinicopathological Profile from Eastern India Singh VP,* Ranjan S, Ghosh SR, Kumar Vipon. Command Hospital (EC) , Kolkata.

Introduction: The annual incidence of thyroid cancer varies considerably in different registries, ranging from 1.2-2.6 per 100,000 individuals in men and from 2.0-3.8 per 100,000 in women. However, there is no definite data from Indian subcontinent. This study is an effort to understand the clinicopathological profile of Thyroid swellings in Eastern India. Materials & Methods: Prospective Observational study, carried out at this Armed Forces Tertiary care Hospital in Eastern India. All patients presenting with Thyroid swelling from Jan 2016 till Jun 2019 were studied. Patients with proven malignancies were excluded. All patients underwent surgery. Results:

In this study a total of 123 patients with thyroid swelling, were evaluated using clinical, radiological and final Histopathological evidence. These patients were in the age group between 20 to 80 years of age and qualified for inclusion in the study based on the criteria defined earlier. A total of 42 participants, ie, 34.14% were eventually found to have thyroid malignancy. The remaining 81 patients, ie, 65.85% of the study group had benign thyroid neoplasms. USG Thyroid with TIRADS and USG guided FNAC were most important determining factors. Conclusion: When ordering for preoperative US scan of thyroid, a surgeon should specifically ask for the visualization of Echogenicity, Nodularity, Microcalcification, Lobar involvement and Presence of neck node. Preoperative TSH levels should be documented amongst all patients being planned for surgery. Whenever possible FNAC should be performed under US guidance. BACKGROUND. Single institution retrospective study to find out the patient characteristics, tumor factors and early postoperative outcome and postoperative histopathology of cases of periampullary and pancreatic cancer who presented at tertiary level center in region of south India. METHODS. The Hospital Record Database was searched for cases of periampullary and pancreatic carcinoma admitted between (2006) (2007) (2008) (2009) (2010) (2011) (2012) (2013) (2014) (2015) (2016) . Multivariate data analyses were performed for 5 variables: demographic, tumor characteristics, preoperative staging, early post-operative outcomes and postoperative histopathology. Chi square test and ANOVA statistical test were the statistical tests used. RESULTS. Out of 160 pancreatic cancer patient maximum incidence was seen in the age group of 40 to 60 yrs. (Mean 53.28 yrs.) with male preponderance (67%). At presentation, 72.5% (97) patients had metastasis and 25% (40) evaluated for surgery, out of which 7.5% (12) were found to be resectable, 11.25% (18) border line resectable and 6.25% (10) patient were diagnosed with locally advanced carcinoma. The head of pancreas being the most common site of origin 30 days early postoperative outcome in 23 patients who had under pancreaticoduodenectomy was analysed. The morbidity rate was 56 %. The most common early postoperative complication were Postoperative Pancreatic Fistula (26 %), Postpancreatectomy Haemorrhage (13%). Mortality rate was 13.03%. CONCLUSION. Pancreatic cancer is not a common cancer in our demography, presentation was more in an advanced stage because of lack of free accessibility to Medicare and curative treatment was possible in a dismally low proportion of patients. At our center which is not an organ specific center, even though postoperative morbidity among resected patients was equivalent to reports from other centers, the mortality rate was almost double. This was likely due to inability to salvage patients with serious postoperative complications. Better knowledge about salvage techniques / referral of such patients to appropriate centers after complication will improve results.

the Kaplan Meier method. Univariate and multivariable logistic regression analyses were performed to investigate predictors of R2 resection. Results. Of 1946 RPS patients who underwent resection, 1850 (95.1%) had R0/R1 resection while 96 (4.9%) had R2 resection. As expected, R2 resection was associated with a lower 5-year OS (37.5% vs. 69.1%) and a shorter median OS (23 vs. 129 months) than R0/R1 resection. On univariate analysis, male gender (p =0.003), dedifferentiated liposarcoma (DDLPS) histologic subtype (p=0.03), multifocality (p <0.0001) and Grade 3 histology (p=0.0068) were associated with R2 resection; Grade 3 histology (p=0.04) and multifocality (p <0.0001) retained significance on multivariable analysis. The combination of multifocality, Grade 3 and DDLPS histology was highly significant in predicting R2 resection (p < 0.0001); 25% of such patients underwent an R2 resection. Conclusion. Predictors of R2 resection include male gender, multifocality, Grade 3 and DDLPS histology. Preoperative decisionmaking should take these variables into account, and neoadjuvant strategies considered. Background. Sarcomas are a heterogeneous group of neoplasms originating from mesenchymal tissue, which affects soft and bone tissues; predominantly in soft tissue extremities. Surgery is the mainstay of treatment and the addition of radiotherapy increases locoregional control. However, it is not clear if the surgical margin influences as a prognostic survival factor. Objective. To assess whether the surgical margin influences as an independent prognostic factor of survival of soft tissue sarcomas as well as their influence on local recurrence, distant metastases at the Oncology Hospital of the SXXI National Medical Center attended during the period from 2006 to 2014. Materials and Methods. It is a retrospective, longitudinal and analytical study. One hundred and twenty-eight patients were included during the period from January 1, 2006 to December 31, 2014 with a diagnosis of limb soft tissue sarcoma. Multivariate analysis was performed to identify the factors related to local and metastatic recurrence. Survival was assessed through Kaplan and Meier curves. Results. Liposarcoma occupies 43.8% followed by mixofibrosarcoma 14.1%, malignant fibrous histiocytoma 11.7%, synovial sarcoma 10.9%, malignant tumor of the peripheral nerve sheath 6.3%, 13.3% other histologies, 71.1% were treated with surgery exclusively, 53.9 presented a tumor larger than 16cm, 45.3% was grade I. In 22.7% the surgical margin was positive. In the bivariate analysis The histological grade p = .002, the surgical limit p = .000, and the stage p = .000, were significant to DFS. The histological type (p = .032), treatment (p = .030), histological grade (p = .009), surgical limit (p = .003) and CD (p = .000). were significant to SG. In relation to the multivariate analysis. PLE and SG , the ILV (p = .022, OR 12.941), the histological type (p = .036, OR 2.994), histological grade 3 (p = .000, OR 6.170) and the surgical limit (p = .000 , OR 11,812).were significant. Conclusion. The surgical limit has a controversial role in global survival, but in our trial the margen was significant to SG. However, a clinical trial is necessary to obtein more information. Objectives: Robotic assisted thoracic surgery (RATS) has been increasingly utilized for lobectomy. Other than small institutional series, little data exits comparing current VATS techniques with RATS in regard to perioperative outcomes. In this analysis we sought to compare both RATS vs. VATS in lung cancer patients who underwent lobectomy. Methods: Pubmed, ScienceDirect, and Embase databases were queried and we included only reported RATS vs. VATS lobectomy for primary lung cancer. The primary endpoint was in-hospital/30-day ortality. Secondary endpoints were operative time, conversion to open, tumor size, clinical and pathological non-stage I, lymph node (LN) retrieval, LN stations retrieved, air leak > 5 days, estimated blood loss (EBL), length of stay (LOS), and complication rate. The odds ratio and standard mean differences (OR and SMD) were the summary statistic using the random model. Leave-one-out analysis was done. Results: Among 1,312 articles, 9 were eligible with 39,696 patients included (7, 271 RATS and 32, 424 VATS) . Pooled analysis indicated that there was no difference in the rate of conversion to open thoracotomy, tumor size, clinical and pathological non-stage I, LN numbers retrieved, number of LN stations retrieved. Prolonged air leaks (>5days) were similar between both groups as well as LOS, complication rates and in hospital/30 days mortality. On meta-regression, age, sex, smoking and FEV1 had no effect on perioperative mortality. Although RATS was associated with a longer OR-time (SMD= 0.82, 95%CI=0.55-1.09, P<0.0001) there was a trend towards less EBL (SMD= -0.36, 95%CI=-0.73 -0.01, P=0.06). Conclusions: This current analysis demonstrates that both RATS and VATS can be performed safely with comparable peri-operative outcomes. Randomized clinical trials comparing both approaches are required to elucidate possible RATS advantages. Purpose: Poor nutritional status and hypoalbuminemia affects a large percentage of esophageal cancer patients. Serum albumin level is a widely used method for nutritional assessment and acts as a prognostic and inflammatory marker. We evaluated the effect of postoperative day 1 (POD1) albumin levels on surgical recovery after esophageal surgery. Methods: 713 patients operated for esophageal cancer over a period of 8 years (2009-2016) were retrospectively reviewed, albumin levels on postoperative day 1 were recorded and its correlation with clinico-pathological characteristics, surgical outcomes, complication rates and mortality were evaluated. Results: Postoperative hypoalbuminemia was significantly associated with female sex, low preoperative serum albumin, advanced stage, greater operating time and intra-operative blood loss. Hypoalbuminemic patients (< 2.5 g/dl) had a higher rate of major surgical complications such as sepsis, respiratory insufficiency, blood transfusion, prolonged hospital stay and post operative mortality. Conclusion: Low postoperative albumin levels correlate with adverse clinical outcomes. POD 1 albumin < 2.5 g/dl may identify a cohort postoperatively that are at an increased risk and require close monitoring. Its promising role as an early marker for surgical stress response deserves further prospective evaluation.

The Influence of Postoperative Complications on Survival and Tumor Recurrence of Patients Who Underwent Curative Resection of Gastric Cancer Ayato Obana,* Motoi Koyama, Kenta Kitamura,Tomonori Matsumura, Yoshinobu Sato, Norimasa Koide, Shinsuke Usui, Kazuhiro Karikomi, Tatsushi Suwa. Kashiwa Kousei General Hospital. Background. In modern times patient safety in gastrectomy has significantly improved due to the innovation of surgical devices and of operative techniques. On the other hand, in modern time postoperative complications are still troubling problems and their influences on longterm prognosis are unclear. Method. 258 patients with gastric cancer who underwent curative resection from April 2009 to December 2017 in our facility were retrospectively analyzed and group comparisons were performed between 59 patients with post-op complications ranked equal to or greater than Clavian-Dindo class II and 199 patients without them. Results. There was a significant difference in overall survival (OS) (with complication, 3-year: 57.6%, 5-year: 50.0%, without complication 3-year: 84.3%, 5-year: 75.6%, p<0.001), but no significant difference in relapse-free survival (RFS) (with complication, 3-year: 89.3%, 5-year: 87.3%, without complication, 3-year: 85.8%, 5-year: 85.8%). Death through other diseases was significantly higher in the complication group (p<0.001) and more than half of them died of pneumonia. Cox proportional hazard ratio indicated that adverse prognostic indicators for OS included ASA physical status greater than or equal to III (hazard ratio [HR] 2.93, 95%CI; 1.53-5.59), transfusion (HR: 2.81, 95%CI; 1.49-5.31 ), open surgery (HR: 3.23, 95%CI; 1.49-6.67), male (HR: 2.22, 95%CI; 1.17-4.17), total gastrectomy (HR: 1.82, 95%CI; 1.01-3.33), past medical history of neuro/psych disease (HR:2.01, 95%CI; 1.03-3.96). Conclusion. Even after discharge, patients who experienced postoperative complications are likely to suffer from pulmonary diseases which can lead to severe consequences. To achieve long-term prognosis after gastrectomy, we should avoid invasive procedures such as open surgery or total gastrectomy for patients with poor physical status. In addition, patients who experienced postoperative complications after gastrectomy should take precautions to prevent pulmonary disease after discharge.

ypT and ypN Staging Impact on Gastric Cancer Patients' Survival Compared to pT and pN Staging after Upfront Surgery. A TNM stage Comparison by Treatment Alex de A L Barbosa,*¹ Victor Hugo F de Jesus,² Tiago C Felismino, MD,² Héber S C Ribeiro,¹ Alessandro L Diniz,¹ André Luís de Godoy,¹ Igor Correia de Farias,¹ Silvio Melo Torres,¹ Wilson L da Costa Jr,³ Felipe J F Coimbra.¹ A. C. Camargo Cancer Center, São Paulo, Brazil. Background: In its most recent edition, The American Joint Committee on Cancer (AJCC) Tumor Node Metastasis (TNM) Manual has presented a different stage grouping for patients receiving neoadjuvant therapy (ypTNM). Studies have showed it has a significant prognostic impact. What remains unclear is how this staging compares to the one of patients treated with upfront surgery (pT and pN), and if that impacts treatment strategies. Objective: We aimed to compare the overall survival between the ypTNM and pTNM stages of gastric cancer patients who were treated with perioperative chemotherapy or upfront surgery. Method: This was a retrospective cohort study that included 632 patients with non-metastatic gastric or gastroesophageal (GE) junction adenocarcinoma treated with curative surgery with or without neoadjuvant chemotherapy (CTx) in the Department of Abdominal Surgery, from 2002 to 2016. Results: Among the study population, 221 received neoadjuvant CTx and 411 were treated with upfront surgery. Most patients were male (55.4%), with a mean age of 61.4 years and 76.4% were classified as ASA 1/2. Total gastrectomy was performed in 44.6% and a subtotal one in 55.4%, with D2 lymphadenectomy in 89.1%. Postoperative major morbidity was 10.1%. Regarding pathological staging groups, stage I subjects in the neoadjuvant group had overall 5-year survival of 93%, versus 86% in the upfront surgery group (p <0.06). In stages II and III, survival differences were not statistically significant (p values of 0.81 and 0.11, respectively), but among stage III individuals the perioperative CTx group had numerically worse 5-year (37% vs. 42%) and median survival (29 months vs. 42 months), respectively. When ypT and ypN stages were analyzed separately, the ypT0-2 and pT0-2, and ypT3-4 and pT3-4 group comparison identified similar survival numbers, with p values of 0.14 and 0.51, respectively. And in lymph node staging, pN0 patients had equivalent results to ypN0 (82%). Conclusion: The comparison of stage groups showed that post-CTx stage I and II subjects had same survival results when compared to the upfront surgery group, but stage III ones (nonresponders) seemed to do worse. Also, no survival differences were observed in the comparable T and N stages.

Tumor Mutation Burden Can Be a Good Biomarker for Chemotherapy Responsiveness and Long-Term Survival in Gastric Cancer Kyungho Pak,* 1 Jae-Ho Cheong. 2 Hallym University, 1 Yonsei University. 2 Background. High tumor mutation burden (TMB) is an emerging biomarker of sensitivity to immune checkpoint inhibitors in various cancer. However, clinical characteristics and prognostic significance of TMB have not been elucidated yet. This study aims to delineate the clinical usefulness of TMB in gastric cancer. Methods. 36 gastric cancer patients who have underwent gastrectomy were included in this study. All tumor and normal gastric fresh frozen tissue from them were collected. Whole exome sequencing (WES) was processed and TMB, in silico MSI (microsatellite instability), DNA polymerase gene mutation and DNA mismatch repair gene (MMR) mutation were analyzed. For chemotherapy responsiveness and long-term survival analysis, 21 patients who underwent curative gastrectomy and complete follow-up were enrolled. Fischer exact test, Mann-Whitney U test, multiple logistic regression and Kaplan-Meier test with Cox regression analysis were used for statistical analysis. Results. All patients were divided to high TMB (TMB-H, median 20.7) and low TMB (TMB-L, median 7.7) by the median value of TMB (12.5). Almost all parameters (age, sex, smoking, alcohol, location, differentiation, tumor size, T stage, N stage, TNM stage, lympho-vascular invasion, perineural invasion, HER2 expression, neutrophillymphocyte ratio, chemotherapy treatment. However, TMB-H group showed better chemotherapy response (P=0.03) and lower recurrence (P=0.06). TMB-H showed longer disease-free survival (DFS) (P=0.03) and independent prognostic factor (P=0.03, HR 21.4 (1.39-329) . In addition, MSI was well matched with in silico MSI genotyper and DNA polymerase gene was well correlated TMB (P<0.01). Conclusions. TMB can be a good predictive biomarker for chemotherapy responsiveness and prognostic marker for DFS in gastric cancer patients.

As required by the Accreditation Council for Continuing Medical Education (ACCME) and in accordance with the Society of Surgical Oncology (SSO) policy, all educational planners, presenters, instructors moderators, authors, reviewers and other individuals in a position to control or influence the content of an activity must disclose all relevant financial relationships with any commercial interest that have occurred within the past 12 months. This includes the disclosure of all financial relationships with a commercial interest of a spouse or partner. A commercial interest is any entity producing, marketing, reselling, or distributing health care goods or services consumed by, or used on, patients. ACCME does not consider providers of clinical service directly to patients to be commercial interests. The ACCME considers financial relationships to create conflicts of interest when individuals have both a financial relationship with a commercial interest and the opportunity to affect the content of CME about the products or services of that commercial interest. All identified conflicts of interest must be resolved and the educational content thoroughly vetted for fair balance, scientific objectivity, and appropriateness of patient care recommendations. It is required that disclosure be provided to the learners prior to the start of the activity. Individuals with no relevant financial relationships must also inform the learners that no relevant financial relationships exist. Learners must also be informed when off-label, experimental/investigational uses of drugs or devices are discussed in an educational activity or included in related materials. Disclosure in no way implies that the information presented is biased or of lesser quality. It is incumbent upon course participants to be aware of these factors in interpreting the program contents and evaluating recommendations. Moreover, expressed views do not necessarily reflect the opinions of the SSO. (Please note that Posters are not certified for credit.)

The Global Posters Program Abstract Authors and Presenters have reported that they have no relevant financial relationships with commercial interests to disclose. 

Is Total Skin-sparing Mastectomy Safe in Invasive Lobular Carcinoma of the Breast?

SF3B1 Mutation is Associated with Improved Long-term Survival in

Adjuvant Nivolumab (NIVO) Versus Ipilimumab (IPI) in Resected Stage III/IV Melanoma: 3-Year Efficacy and Biomarker Results From the Phase 3 Checkmate 238 Trial

Veneto Institute of Oncology IOV -IRCCS

Division of Surgical Oncology

Postoperative Pancreatic Fistula Following Distal Pancreatectomy for Primary Retroperitoneal Sarcoma: An Analysis by the Trans-Atlantic Australasian Retroperitoneal Sarcoma Working Group (TARPSWG) S.P. Bagaria, 1 * C

16 E. Stoeckle, 17 M. Almquist, 18 N. Ahuja, 19 N. Klemen

European Institute of Oncology

Oncologic and Functional Outcome of Truncal Soft Tissue Sarcoma Requiring Full-thickness Wall

Gronchi, 3 M. Fiore. 3 1. Hospital Angeles Lomas / European School of Soft Tissue Sarcoma

INTRODUCTION: Truncal soft tissue sarcomas (TSTS) arise from the abdominal and thoracic wall and account for roughly 20% of all sarcomas. Surgery often requires full-thickness wall resection and complex repair. Available data are scant for this location

Non-resection strategies in rapidly progressive Desmoid tumors: Think of TNF-Based Isolated Limb Perfusion

Background Desmoid Fibromatosis (DF), a monoclonal fibroblastic proliferation arising in deep soft tissues. Is characterized by infiltrative growth, local recurrence and no distant metastasis. Its uncommon and affects young patients, becoming a function-treathening condition. Observation has been proposed for asymptomatic lesions, rapidly progressive tumors are considered for surgery

At follow up none of the patients treated with ILP/Observation had progressed and all had achieved full symptomatic relief. Patients treated with ILP/Surgery all relapsed and/or had symptoms. Conclusion Observation is commonly used for asymptomatic or stable disease in DF. Symptomatic or rapidly progressive disease is an indication for resection, outcomes in this scenario have not been addresed. We achieved a 100% rate of symptomatic control in patients treated only with ILP, no lesions progressed during follow up

Sentinel Lymph Node Biopsy is Prognostic in the Thickest Melanoma Cases and Should Be Performed for Thick

124 T1b out of 289 cases were identified with SLN injection and 108 SLN dissections performed. All CLND were functional dissections. Complication rates following SLN and CLND, T-stage, rates of +SLN, +NSLN in CLND were calculated, as well as death due to disease (DOD), progression free survival (PFS), along with rates of local (LR), nodal (LNR), and systemic (SR) recurrence. Results T-stage for 108 SLN cases was 41% IB, 23% IIA, 28% IIB, 8% IIC. Nerve (sensory/motor) complication rates was 4% for SLN and 11% for CLND. The -SLN group survival is 93% compare to survival of 70% for +SLN group with median follow-up of 40 months. The rate of +SLN was 29% (24% IB, 30% IIA, 33% IIB, 33% IIC) and the +NSLN rate for CLND was 50%

Access to Care and Outcomes for Non-Curative Esophagogastric Cancer: A Population-Based Geographic Study

INTRO Time from cancer diagnosis to first treatment (TimeTx) is an important quality of care indicator. Identifying predictors of delayed TimeTx for gastric cancer (GC) patients may reveal opportunities to reduce disparities. METHODS National Cancer Database identified adults (>18 years) with gastric adenocarcinoma surgically treated 2004-2014. Patients with unknown TimeTx, stage IV GC, or surgical treatment within 14 days of diagnosis (emergencies) were removed. Descriptive statistics compared those with and without delayed TimeTx (>8 weeks). Logistic regression models identified predictors of delayed TimeTx for the entire cohort

vs. 67%), without private insurance (68 vs. 59%), and with >=1 comorbidities (34 vs. 30%). They were treated in academic centers (55 vs. 45%), the Northeast (50 vs. 45%), and metro areas (83 vs. 82%)

Methods: A retrospective cohort study was performed of patients with clinical stage I or II gastric adenocarcinoma using the National Cancer Database from years 2004-2014. Trends were assessed using the Cochrane-Armitage test. Multivariable logistic models were employed to identify factors predicting refusal. Propensity scores were created for the odds of refusing surgery, and patients were stratified based on score. Overall survival was evaluated using the Kaplan-Meier method. Results: Of the 12,185 patients identified with gastric adenocarcinoma stage I or II, 3.56% (n=434) refused surgery. The number of patients who refused did not change significantly over time

After stratification, patients who refused surgery had significantly worse survival compared to those who underwent resection. Conclusions: Patients who refused resection of gastric cancer had diminished overall survival and were more likely to be of older age, black race, have less comprehensive insurance, and a disparate treatment location. Further studies should investigate reasons for refusal and elucidate barriers in the physician-patient relationship

Methods: We identified patients diagnosed with GC between 2004 and 2016 using the NCDB. Socioeconomic factors examined were race, insurance status, income, and education. Outcomes of interest were 30-day mortality, readmission rate, and time from diagnosis to treatment. Multivariable logistic regression models were created to examine the associations between socioeconomic factors and the outcomes of interest. Results: We identified 151,016 patients diagnosed with gastric cancer, 129,898 (86%) of whom underwent surgical resection. Overall 30-day mortality and readmission rates were 3.94% and 3.29%. Overall 30-day mortality was significantly higher in patients without insurance

Conclusion: Black race, lack of insurance, and lower income are significantly associated with higher 30-day mortality, readmission, and longer time from diagnosis to treatment in patients diagnosed with GC. Better and more accessible support measures are needed to improve the care for patients that are socially and economically disadvantaged

Associations between clinical and pathologic features and outcomes in ACC were examined. Logistic regression analyses were used to evaluate associations between tumor/patient characteristics and the choice of minimally invasive surgery. Propensity score matched analysis was performed, matching on characteristics predictive of a minimally invasive approach. Cox proportional hazard models were used to identify associations between factors and overall survival. Results: 795 (66.8%) patients underwent open surgery and 395 (33.2%) underwent minimal access surgery

After propensity score matching for these factors, 293 pairs were identified. In the propensity matched cohort, minimally invasive surgery was not associated with increased rate of 30-day readmission (OR 0.6, 95% CI 0.3-1.2, p=0.2), 30-day mortality

Minimally Invasive Resection of Pheochromocytoma: Is a Retroperitoneoscopic Approach Better After All? A.M

Site-specific Molecular Comparison Between Primary Gastric (GC) and Gastroesophageal Junction (GEJ) Cancer Emphasizing Peritoneal Metastases (PM)

Results: 1366 cases were identified: 1041 GC (707 P, 98 PM, 236 OM) and 325 GEJ (248 P, 5 PM, 72 OM). PM were increased in GC versus GEJ (9% v. 2%, p < 0.0001). 91% GC and 93% GEJ were adenocarcinoma (AD); GC were more likely signet ring (SR) histology versus GEJ (11% v. 3%, p < 0.0001) and GC PM were more likely SR versus OM or P (13% v. 12% v. 7%, p = 0.067). The mean age of PM pts (57 years) was younger than primary GC (63, p = 0.002) and OM (61

OM (9%, p = 0.041) had more CNA in CCNE1 than PM (2%, p = 0.041) or P (5%, p = 0.002). Conclusions: Compared to P and OM GC, PM pts were younger, more likely female and had a higher incidence of SR histology. PD-L1, HER2 IHC and ERBB2 CNA were reduced in PM versus P

Should Adenosquamous Esophageal Carcinoma be Treated Like Adenocarcinoma or Squamous Cell Carcinoma?

Medullary Thyroid Cancer in Patients Over 45-Years -A Nationwide Trend Over Four Decades S

Most patients had private insurance (50.5%) or Medicare (41.3%). Patients typically lived in larger metropolitan areas (51.5%) and in zip codes with median household incomes >$48,000 (60.7%). Only 14.7% lived in zip codes where >20% did not graduate high school (no HSD). The majority were treated at community comprehensive cancer centers (43.8%) or academic/research centers (35.2%). Overall, 3358 (24.5%) presented with metastasis at diagnosis. The 5-year OS for the entire cohort was 78.5% and was worse in patients with lower median income (p<0.0001), lower education (p<0.0001), in those patients not living in proximity to a metro area

Methods: Retrospective, observational and descriptive study. 405 files of patients with BC diagnosis from 2010-2019. ST was defined as that diagnosed in the same time and until 6 months at diagnosis. MT as that diagnosed at more than 6 months after. Results. Of the 450 files, 13 cases with BCB (2.88%), 6 cases were ST and 7 MT. ST group the age average was 62 years

Is Adjuvant Chemotherapy Necessary for Locally Advanced Rectal Cancer Patients with Pathological Complete Response after Neoadjuvant Chemoradiotherapy and Radical Surgery?

P142, P208, P249, P253, P259

P82, P87, P105

P163, P245, P247, P299, P309

P249, P297

Botty Van den Bruele

P156, P240, P241

P184, P328, P336, P358

P43, P86, P104, P125, P128

95, 140, P249, P253, P259, P261, P262, P265

92, 95, P249, P253, P259, P261

139 gangi, a. P116, P121

P1, P16

92, 95, P249, P253, P259

92, 95, P249, P253, P259

P141, P148

95, P243, P249, P253, P259, P261

P249, P253, P259, P261, P262

45 N Nachmany, I

P224, P298, P320, P321

95, P249, P253, P255, P259, P261

P141, P148

134 sON

P63, P79, P224, P235

P43, P104, P125, P128

39 yamamoto,

P79, P274, P276

P82, P87, P105, P130, P149

ABSTRACT BACKGROUND: A consensus on the best minimally invasive approach for resection of pheochromocytoma has not been established. Individual and institutional factors still determine operative preferences. Retroperitoneoscopic adrenalectomy (RPA) and transabdominal laparoscopic adrenalectomy (TLA) have been well described. We aimed to objectively compare short-term operative outcomes between these techniques. METH-ODS: A comprehensive online database search of MEDLINE and EMBASE was performed; key bibliographies were reviewed. Studies comparing operative data of RPA and TLA were included. Meta-analysis of pooled data with calculation of mean differences (MD) and the corresponding 95% confidence intervals (CI) by random and fixed effects models was performed. Funnel plots were used to evaluate publication bias. Study quality was assessed using STROBE (Strengthening the Reporting of Observational Studies in Epidemiology) criteria. RESULTS: Overall, 6 studies with a total of 810 patients were included (268 had RPA while 542 had TLA). Average age was 49 years. Demographic variables, comorbidities and tumor characteristics, including tumor size, were similar in both groups. Meta-analysis of included data showed that RPA had significantly shorter operative time (MD:-29.18, CI:-51.55 to -6.81, p=0.01), less blood loss (MD:-83.14, CI:-139.12 to -27.15, p=0.004), and shorter length of hospital stay (MD:-1.66, CI:-2.77 to -0.56, p=0.003). There was no significant difference between the groups in terms of conversion rates to open operation, overall complication rates, intraoperative hypertensive episodes or vasoactive medication use. CONCLUSION: RPA for pheochromocytoma may be a faster operation and result in a shorter hospital stay when compared with TLA. Given similar conversion and complication rates without increased intraoperative hemodynamic instability, prospective randomized controlled trials could help further characterize these two procedures and lead to a wider inclusion of RPA in surgical training.Background: Somatic mutations are principally implicated in the majority of patients with medullary thyroid carcinoma (MCT) aged 45 years. This patient population has a distinct presentation and survival characteristics compared with their younger cohorts. Methods: Using the SEER registry from 1973 -2016, all patients aged 45 years with MTC were included for analysis. We estimated the incidence rates by demographics and tumor characteristics, followed by analysis of its impact on survival. Kaplan-Meier survival was analyzed for categorical variables and Cox regression was performed to predict independent risk factors associated with poor survival. Results: A total of 2533 patients aged 45 years with MTC were identified. Mean age of MTC in this patient group was 61.46 10.77 years (45 -99 years). Over this time period, the incidence of sporadic MTC is has increased statistically by 10.8% / year ( Figure Tumor was most commonly localized to thyroid gland and extrathyroidal extension was seen only in 9.4 % of patients. Surgery offered modest but significant survival benefit as compared to patients who did not undergo surgery (170.99 6.92 months vs 163.76 2.59 months, p < 0.01). On Cox regression analysis, increasing age (HR 1.05, 95%CI 1.03 -1.07, p <0.01), tumor grade (p<0.01), SEER stage (p<0.01) and tumor extension (p<0.01) were associated with poor survival. Conclusions: These data demonstrate that the national incidence of MTC in patients aged 45 years is increasing. Surgery does offer modest survival advantage. However, tumor associated factors such as advanced grade, SEER stage, and extra-thyroidal extension were the only independent risk factors associated with poor survival. Introduction: India has an estimated incidence of more than one million cancers annually. Oral, breast, and cervical account for over one-third of newly diagnosed cases. With the introduction of pilot cancer screening programs in India, little is known about potential psychosocial and cultural barriers that may hinder acceptance of screening and subsequent care and treatment. We sought to identify and quantify knowledge gaps and misconceptions about cancer, as well as attitudes and stigmas surrounding cancer diagnosis. Methods: A baseline survey was conducted in Assam, India, of residents age 30 years, as part of the Detect Early and Save Her/Him (DESH) program, a mobile screening program for oral, breast, and cervical cancer that was to be implemented in the region. Data was collected on respondent demographics, knowledge about cancer etiology, and attitudes towards cancer screening, diagnosis, and treatment. Results: 923 residents participated in the baseline survey (table) . Respondents demonstrated high awareness of certain carcinogens, such as tobacco and alcohol (69-78%). Low-medium awareness was demonstrated regarding the carcinogenic effects of betel nuts (n=433, 47%) and of genetic contributors (n=151, 16%). While less than 10% of respondents had misconceptions about the etiologies of cancer, 42-57% agreed with statements endorsing a negative cancer stigma, including its long-term detrimental effects on personal, occupational, and familial life. However, majority of respondents (68-96%) agreed with statements endorsing positive community support and medical care for cancer patients. Conclusions: This study identifies actionable targets for intervention in cancer education and awareness within a large rural Indian population. Education to address preventable causes of cancer, such as betel nut and tobacco consumption, and to correct misconceptions and stigma is a critical component in ensuring the successful implementation of cancer screening programs.