key: cord-1012210-htzjz822
authors: McKay, Siobhan C.; Drake, Thomas M.; Spiers, Harry VM.; Bassi, Claudio; Giovinazzo, Francesco; Citterio, Davide; McGahan, William; Isaac, John; Satoi, Sohei; Glaseby, James C.; Bhangu, Aneel; Harrison, Ewen M.; Roberts, Keith
title: Outcomes of patients undergoing elective liver and pancreas cancer surgery during the SARS-CoV-2 pandemic: an international, multicentre, prospective cohort study
date: 2022-03-16
journal: HPB (Oxford)
DOI: 10.1016/j.hpb.2022.03.002
sha: c96d8d42f5a2be05217ed56e2979f79ccc9bacbf
doc_id: 1012210
cord_uid: htzjz822

BACKGROUND: The effect of SARS-CoV-2 infection upon HPB cancer surgery perioperative outcomes is unclear. Establishing risk is key to individualising treatment pathways. AIM: Identify the mortality rate and complications risk for HPB cancer elective surgery during the pandemic. METHODS: International, prospective, multicentre study of consecutive adult patients undergoing elective HPB cancer operations during the initial SARS-CoV-2 pandemic. Primary outcome was 30-day perioperative mortality. Secondary outcomes included major and surgery-specific 30-day complications. Multilevel cox proportional hazards and logistic regression models estimated association of SARS-CoV-2 and postoperative outcomes. RESULTS: Among 2,038 patients (259 hospitals, 49 countries; liver n=1,080; pancreas n=958) some 6.2%, n=127, contracted perioperative SARS-CoV-2. Perioperative mortality (9.4%, 12/127 vs 2.6%, 49/1911) and major complications (29.1%, 37/127 vs 13.2%, 253/1911) were higher with SARS-CoV-2 infection, persisting when age, sex and comorbidity were accounted for (HR survival 4.15, 95% CI 1.64 to 10.49; OR major complications 3.41, 95% CI 1.72 to 6.75). SARS-CoV-2 was associated with late postoperative bleeding (11.0% vs 4.2%) and grade B/C postoperative pancreatic fistula (17.9% vs 8.6%). DISCUSSION: SARS-CoV-2 infection was associated with significantly higher perioperative morbidity and mortality. Patients without SARS-CoV-2 had acceptable morbidity and mortality rates, highlighting the need to protect patients to enable safe ongoing surgery.

Cancer pathways within healthcare systems worldwide have been severely impacted by the 84 SARS-CoV-2 pandemic. Widespread cancellation of elective surgery led to an estimated 28.4 85 million cancelled or postponed operations, including cancer surgery, in part due to a scarcity of 86 intensive care capacity (1-3). Fear of peri-operative mortality among patients undergoing major 87 liver and pancreas cancer surgery in the setting of SARS-CoV-2 has also affected allocation of 88 surgery. Early studies reporting 30-day mortality from perioperative SARS-CoV-2 as high as 89 23.8% have been a key driver for this (4-7). Patients undergoing surgery for pancreatic and liver 90 cancers are already at higher risk of poor peri-operative outcomes compared to other cancers (8). 91

They also have particularly time critical disease where treatment delays are associated with 92 disease progression (9-15). As the pandemic continues and new variants emerge, a better 93 understanding of how SARS-CoV-2 infection impacts peri-operative outcomes is especially 94 important for this group of patients. 95

With the SARS-CoV-2 pandemic ongoing, and the continued threat of the development of new 97 variants of the virus, it is vital that the risk of perioperative SARS-CoV-2 for patients with 98 resectable liver and pancreas cancer is better understood. Firstly, this will enable risk 99 stratification of elective patients prior to elective surgery (5). Secondly, clinicians will be better 100 equipped to inform patients of the risk of perioperative infection, and subsequently balance that 101

This international, multicentre, observational cohort study included consecutive elective patients 116 with liver and pancreas cancer who underwent surgery with curative intent during the COVID-19 117 pandemic. Local principal investigators were responsible for obtaining clinical audit, institutional 118 review board, or ethical approval in line with local and national regulations. In the United 119

Kingdom the study did not require research ethics approval and was registered as a clinical audit 120 as only routinely collected anonymized data were recorded. Research Center. SARS-CoV-2 rate was divided into peak and post-peak (post-peak defined as a 180 sustained fall to below 20% of the maximum number of cases) (24). 181 182 183

Frequencies and percentages were used to summarize categorical data. Continuous data were 185 summarized using mean (standard deviation) or median (interquartile range) depending on 186 distribution. We used Cox-proportional hazards models for the primary survival outcome and 187 multilevel logistic regression models for secondary outcomes to adjust for other explanatory 188 variables on the outcome of interest. For time to death, day of surgery was taken as day zero. 189

Discharge from hospital was considered an absorbing state (once discharged, patients were 190 considered no longer at risk of death). Discharged patients were not censored and included in the 191 risk set until the end of follow-up, thus discharge did not compete with death. Multilevel logistic 192 regression models were specified with level 1 fixed effects (patient and disease characteristics) 193 and level 2 random effects (centre level). Clinically plausible factors were entered into these 194 models and final model selection was guided through minimization of the Akaike information 195 criterion (AIC). Models were checked for any relevant first order interactions and any significant 196 interactions were retained. Effect estimates are presented as hazard ratios for the primary time-197 to-event outcome (HR) and odds ratios for binary outcomes (OR). Effect estimates are presented 198 alongside their corresponding 95% confidence intervals (95% CI). As a sensitivity analysis to 199 help healthcare planners decide when elective surgery for hepatopancreatobiliary cancer may be 200 safe, we compared risk of death andSARS-CoV-2 at during the first peak and after the first peak. 201

All analyses were performed in R version 3.6.3 (R Foundation for Statistical Computing, Vienna, 202 AUT) using the tidyverse and finalfit packages. Between the emergence of SARS-CoV-2 and 31st August 2020, 2,038 adult patients undergoing 206 elective surgery for liver (1,080/2,038, 53.0%) and pancreas cancer (958/2,038, 47.0%) were 207 enrolled across 259 hospitals in 49 countries ( Figure 1 ). Two patients without survival outcomes 208 were excluded. Most patients were from United Kingdom (15.0%, n =306), Italy (14.6%, n = 209 298), Spain (8.3%, n =170), and Germany (7.9%, n =161) (Supplementary Table S1 ). Patient 210 demographic data is described in Table 1 , disease characteristics of patients with pancreatic or 211 liver cancers in Tables 2 and 3 respectively. 212

Overall mortality and complications 214

The overall 30-day mortality was 3.0% (61/2,038), being higher among patients undergoing 215 pancreatic resection (4.1% 39/958) compared to liver resection (2.0% 22/1,080) (Table 4 ). 216

Mortality rates by extent of liver surgery are described in Supplementary Table S2 (Table 4 ). The 228 30-day postoperative mortality rate was over three times higher for patients with SARS-CoV-2 229 than those without (9.4%, 12/127 vs 2.6%, 49/1,911; p<0.001). Following adjustment for 230 explanatory variables, development of SARS-CoV-2 infection remained associated with 231 significantly worse survival (Figure 2 ). Performance status of 2 or greater was the only 232 explanatory variable found to have an independent association with 30-day survival. SARS-233 length of stay (Table 4 ). After adjustment for explanatory variables, SARS-CoV-2, male gender, 236 higher performance status and pancreatic surgery were independently associated with higher 237 rates of major complications ( Figure 3 ). As expected, patients who tested positive for SARS-238

CoV-2 had a higher rate of respiratory complications (40.9%, 52/127 vs 5.8% 110/1,911; 239 adjusted OR 13.98, 95%CI 6.60 to 29.63, Table 4 and Supplementary Figure S2) . 240

Temporal patterns of SARS-CoV-2 242 SARS-CoV-2 diagnoses occurred most frequently between March and April 2020, when most 243 countries had reached a peak number of infections or were building up to their peak (6.9% peak, 244 124/1,797 versus 0.8% after peak 22/240, Figure 4 ). 11.8% (240/2,038) of patients had their 245 operation after the peak and had equivalent demographics (including age, sex, performance 246 status and BMI) to those during the peak (Supplementary Table S3 and S4). Higher mortality 247 rates were seen during national peaks when compared to after the peak (3.2% vs 0.8%, Figure  248 4B). These rates were primarily driven by patients with SARS-CoV-2 (8.8% mortality in peak 249 time, versus 2.8% in those who did not have SARS-CoV-2 during same time period). In people 250 who sustained a major complication at times of peak infection rates, the mortality rate was 251 22.7% (58/256) compared with 6.0% (2/33) after the peak. 252

Elective operating during the SARS-CoV-2 pandemic has been a source of concern for patients 254 and clinicians. To our knowledge, this is the largest study of elective major liver and pancreas 255 surgery undertaken during the pandemic, with 2,038 patients undergoing liver and pancreas 256 surgery, including 127 patients with perioperative SARS-CoV-2. This prospective international 257 cohort study undertaken during the first wave of the SARS-CoV-2 pandemic demonstrates that 258 perioperative SARS-CoV-2 infection in patients undergoing elective surgery for liver and 259 pancreas cancer is associated with significantly poorer postoperative outcomes. One in every 16 Early studies during the pandemic reported significant mortality rates associated with 266 perioperative SARS-CoV-2 infection. One study found an overall mortality rate of 23.8%, 267 however, of the 294 patients undergoing elective surgery in the study, including a small cohort of 268 30 undergoing major liver or pancreas resections, the mortality rate was 19.8% (4). This was 269 corroborated by meta-analysis of 2,947 patients with perioperative SARS-CoV-2, which reported 270 a mortality rate of 20%, although these were mixed specialty emergency and elective procedures 271 of varying operative complexity (25). Prior to our study, data on mortality rates of perioperative 272 SARS-CoV-2 in liver and pancreas surgery was lacking, as was data pertaining to the risk of 273 operating on patients without SARS-CoV-2 during the pandemic. It has therefore proven 274 challenging for clinicians and patients to fully understand the perioperative risk of major liver 275 and pancreas surgery during the pandemic. The mortality rate for those with SARS-CoV-2 was 276 9.4% compared to 2.6% without. Although this is considerably lower than previous reports, it 277 remains unacceptably high, highlighting the need to protect elective surgery patients from 278 contracting SARS-CoV-2, to ensure safe surgery can continue for this rapidly progressive The overall 30-day mortality rate for patients undergoing elective surgery for liver or pancreas 285 cancer during the pandemic was 3.0%. This is in keeping with recent national studies of 286 unselected hospitals reporting in-hospital mortality rates after pancreatic surgery, ranging from 287 3.2 to 8.6% (26, 27) and 3.4 to 5.8% for liver surgery (28, 29). However, when mortality rates 288 were stratified by peak and post-peak pandemic time periods, there was considerable variation in 289 mortality (peak of first wave 3.1% compared with 0.8% after the first wave). This is surprising as 290 there were no significant differences in reported patient demographics between groups and is 291 likely multi-factorial. It is possible that there was greater patient selection post-peak with lower 292 risk patients undergoing surgery not reflected in the observed variables, explaining the lower 293 mortality rate. This observation was identified by a UK study, showing one third of pancreatic 294 centres changes their management strategy for patients with borderline-resectable venous disease 295 from a surgery-first approach to neo-adjuvant therapy at the beginning of the pandemic (7). 296

When resources are constrained, operating upon patients more likely to have a good outcome 297 reduces the strain on precious critical care resources and risk stratification should be considered. 298

It is also possible that the higher peak mortality rate was related to a failure to rescue patients 299 following complications due to resource constraints caused by the pandemic, also seen in a 300

Spanish study of emergency general and gastrointestinal surgery patients (30). 301 302 Patients with SARS-CoV-2 infection had significantly higher rates of major complications 303 including late postoperative bleeding, POPF and bile leak. However, our study lacks the 304 temporal data around time of SARS-CoV-2 infection to definitively attribute complications to 305 SARS-CoV-2 infection, we therefore can report the association between the factors, but not 306 definitive causality. Patients sustaining complications are more likely to require longer hospital 307 stays, therefore increasing the risk of developing nosocomial SARS-CoV-2 infection, thus 308 potentially giving the appearance of a higher rate of surgical complications in the SARS-CoV-2 309 group. Similarly, the development of complications frequently shares common risk factors with 310 development of severe SARS-CoV-2 infection (i.e., being male, high BMI)(31). Coagulopathy 311 and massive activation of the fibrinolytic system is frequently observed in patients with SARS-312 patients with liver and pancreas cancers. Particularly if these occur at home, or at healthcare 315 facilities with limited capacity to rescue severe complications which could be a concern if critical 316 care resources are already stretched by SARS-CoV-2. Greater research in this area would prove 317 clinically useful, as preventative interventions for those at risk of developing complications 318 could be deployed, strict SARS-CoV-2 free pathways used or even implementation of 319 ambulatory care or telemedicine to avoid exposure to SARS-CoV-2 in patients who do develop 320 complications. 321

The low perioperative morbidity and mortality for patients without SARS-CoV-2 infection 323 supports the view that surgery for liver and pancreas cancer during the pandemic can be safely 324 performed. It is clearly desirable to ensure SARS-CoV-2 free hospital pathways reduce the 325 likelihood of perioperative SARS-CoV-2 infection, and the optimal perioperative pathway is yet 326 to be determined (33). Different approaches are likely to work in different settings and will 327 depend on a variety of local factors including flexible staffing, critical care availability, local 328 immunisation rates and local rates of SARS-CoV-2. However, the effectiveness of complete 329 segregation of the operating theatre, critical care, and inpatient ward areas in reducing 330 perioperative SARS-CoV-2 infection has been demonstrated on a global scale (18, 34, 35) . 331

Worldwide, liver and pancreas cancer surgery has been impacted by availability of postoperative 332 ICU support, with only 5%-14% units reporting normal operative during pandemic the peak of 333 the pandemic (7, 36). Providing safe pathways to protect patients from perioperative SARS-334

CoV-2 infection and providing ICU care where needed is a clear priority. When these pathways 335 cannot be provided, alternative cancer management strategies are essential as liver and pancreas 336 cancers are rapidly progressive (9, 37, 38). Initial advice suggested moving away from surgery-337 first treatment to reduce caseload of critically stretched ICUs, and due to concerns that patients 338 with cancer were at higher risk of severe SARS-CoV-2 infection due to cancer-related 339 immunosuppression (39-42). However, non-operative pathways such as neoadjuvant 340 FOLFIRINOX in pancreatic cancer, have also been compromised as they represent a risk to 341 patients through immunosuppression and multiple healthcare institution visits where the burden 342 of SARS-CovV-2 infection may be high (43, 44) . Anecdotally, this led to delivery of less This study acknowledges several limitations. Firstly, only patients who had an operation were 345 included, and previous work has identified that large numbers of elective operations were 346 cancelled (7, 46, 47) . Our data may focus on those representing either low SARS-CoV-2 347 infection incidence areas, or those prioritized for surgery owing to their underlying disease (1). 348

The effect of delays in surgery on long-term outcomes in patients with liver and pancreas cancers 349 remains unknown, and the balance of reduced deaths from SARS-CoV-2 infection against the 350 cancer-related deaths arising from delays to surgery, needs characterising fully (48) J o u r n a l P r e -p r o o f J o u r n a l P r e -p r o o f Length of stay (days) Mean (SD) 11.6 (9.9) 11.2 (9. 

Elective surgery cancellations due to the COVID-19 pandemic: global 433 predictive modelling to inform surgical recovery plans

Critical Supply Shortages -The Need for Ventilators and 435 Personal Protective Equipment during the Covid-19 Pandemic

Critical care crisis and some 438 recommendations during the COVID-19 epidemic in China

AHPBA) SoAGaESSTAH-P-BA. SAGES -AHPBA RECOMMENDATIONS 444 REGARDING SURGICAL MANAGEMENT OF HPB CANCER PATIENTS DURING THE 445 RESPONSE TO THE COVID-19 CRISIS 2020

Global guidance for surgical care during the COVID-19 pandemic. Br 448 J Surg

Impact of SARS-CoV-2 455 Pandemic on Pancreatic Cancer Services and Treatment Pathways: United Kingdom 456 Experience

Comparing 458 textbook outcomes among patients undergoing surgery for cancer at U. S. News & World 459 Report ranked hospitals

Management of Pancreatic Cancer During COVID-19 Pandemic: To 461 Treat or Not to Treat?

Effect of time to 463 surgery in resectable pancreatic cancer: a systematic review and meta-analysis. Langenbecks 464 Arch Surg

Time to Surgery of Colorectal Liver Metastases on Survival

Does the 468 interval from imaging to operation affect the rate of unanticipated metastasis encountered during 469 operation for pancreatic adenocarcinoma?

Impact of the 471 time interval between MDCT imaging and surgery on the accuracy of identifying metastatic 472 disease in patients with pancreatic cancer

The value of liver magnetic resonance imaging in patients with 474 findings of resectable pancreatic cancer on computed tomography

A reduced time 16

Consent During the COVID-19 Pandemic

How should surgeons obtain consent during the covid-19 pandemic? 482

Elective Cancer Surgery in COVID-19-Free 484

Surgical Pathways During the SARS-CoV-2 Pandemic: An International

Comparative Cohort Study

Considerations for the treatment of pancreatic cancer during the COVID-19 pandemic: the UK 488 consensus position

Guidelines for pre-operative COVID-19 testing for elective cancer surgery

The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE)

Statement: guidelines for reporting observational studies

Research electronic 497 data capture (REDCap)--a metadata-driven methodology and workflow process for providing 498 translational research informatics support

Postoperative mortality among surgical patients with 502 COVID-19: a systematic review and meta-analysis

Textbook Outcome: Nationwide Analysis of a Novel Quality Measure in 505

Pancreatic Surgery

Mortality Following Pancreatic Surgery in Germany is Higher than Anticipated

Mortality after liver 510 surgery in Germany

Group AC-FHS. Incidence and risks of 512 liver resection: an all-inclusive French nationwide study

Analysis of outcomes of emergency general and gastrointestinal surgery during the COVID-19 516 pandemic

Features 518 of 20 133 UK patients in hospital with covid-19 using the ISARIC WHO Clinical Characterisation 519 Protocol: prospective observational cohort study

Immunity, endothelial 521 injury and complement-induced coagulopathy in COVID-19

Results 529 of COVID-Minimal Surgical Pathway During Surge-Phase of COVID-19 Pandemic

Delivery of hepato-pancreato-biliary surgery during the COVID-19 pandemic

African Hepato-Pancreato-Biliary Association (E-AHPBA) cross-sectional survey

Time to Progression of Pancreatic Cancer: Evaluation with 536

Multi-Detector Computed Tomography

Time to progression of 538 pancreatic ductal adenocarcinoma from low-to-high tumour stages

SARS-CoV-2 Transmission in Patients With Cancer 540 at a Tertiary Care Hospital in Wuhan, China

Cancer patients in SARS-CoV-2 542 infection: a nationwide analysis in China

Risk of COVID-19 for patients with cancer

Do patients 546 with cancer have a poorer prognosis of COVID-19? An experience in New York City

Cancer Radiation Therapy Recommendations in Response to COVID-19

Systemic 554 Treatment of Patients With Gastrointestinal Cancers During the COVID-19 Outbreak: COVID-555 19-adapted Recommendations of the National Cancer Institute of Milan

Elective surgery cancellations due to the COVID-19 pandemic: global 558 predictive modelling to inform surgical recovery plans

Effect of COVID-19 pandemic lockdowns on planned cancer surgery 560 for 15 tumour types in 61 countries: an international, prospective, cohort study