key: cord-0018161-0mpaurr7
authors: Yagishita, Shigehiro; Kato, Ken; Takahashi, Mami; Imai, Toshio; Yatabe, Yasushi; Kuwata, Takeshi; Suzuki, Mikiko; Ochiai, Atsushi; Ohtsu, Atsushi; Shimada, Kazuaki; Nishida, Toshirou; Hamada, Akinobu; Mano, Hiroyuki
title: Characterization of the large‐scale Japanese patient‐derived xenograft (J‐PDX) library
date: 2021-06-04
journal: Cancer Sci
DOI: 10.1111/cas.14899
sha: edddd185ca98cbf1b2b056b808c53916ec965faa
doc_id: 18161
cord_uid: 0mpaurr7

The use of patient‐derived xenografts (PDXs) has recently attracted attention as a drug discovery platform with a high predictive clinical efficacy and a preserved tumor heterogeneity. Given the racial differences in genetic variations, it would be desirable to establish a PDX library from Japanese cancer patients on a large scale. We thus tried to construct the Japanese PDX (J‐PDX) library with a detailed clinical information for further clinical utilization. Between August 2018 and May 2020, a total of 1126 cancer specimens from 1079 patients were obtained at the National Cancer Center Hospital and National Cancer Center Hospital East, Japan, and were immediately transplanted to immunodeficient mice at the National Cancer Center Research Institute. A total of 298 cross‐cancer PDXs were successfully established. The time to engraftment varied greatly by cancer subtypes, especially in the first passage. The engraftment rate was strongly affected by the clinical stage and survival time of the original patients. Approximately 1 year was needed from tumor collection to the time when coclinical trials were conducted to test the clinical utility. The 1‐year survival rates of the patients who were involved in establishing the PDX differed significantly, from 95.6% for colorectal cancer to 56.3% for lung cancer. The J‐PDX library consisting of a wide range of cancer subtypes has been successfully established as a platform for drug discovery and development in Japan. When conducting coclinical trials, it is necessary to consider the target cancer type, stage, and engraftment rate in light of this report.

In recent years, there has been much interest in the use of patientderived xenografts (PDXs), a cancer-bearing mouse model generated by transplanting a patient's tumor directly into an immunodeficient mouse, as a platform for anticancer drug development. 1, 2 Compared with traditional cell line and cell line xenograft models, a PDX is reported to better reflect the efficacy of treatment in the clinic. 3 It is expected to increase the probability of success, which is estimated to be around 5% in cancer drug development, reduce the cost of development by hundreds of millions of dollars, and shorten development times. 4, 5 Large PDX libraries have already been created and widely used by EuroPDx, the National Cancer Institute Patient-Derived Models Repository (NCI-PDMR), and several commercial companies, mainly in Europe and the USA. 6, 7 In Japan, however, there is no large-scale PDX library, and considering racial differences in oncogenes, it is hoped that a library will be established as soon as possible. 8 There are several possible issues in using PDX models for drug development. In the preclinical phase, it is expected that a PDX model will be used to evaluate the efficacy of a drug in a "PDX basket trial" to confirm proof of concept and to select target cancer types and fractions. In the clinical development phase, pretreatment PDXs in parallel with clinical trials, called "coclinical trials,"

can be used to search for biomarkers of drugs and identify factors associated with refractory response. Post-treatment PDXs in the drug-resistant phase can be used to explore the mechanism of resistance, provide a basis for the next phase of drug discovery and development, and enable the search for combination therapies. After regulatory approval, it is also expected that PDXs can be used to evaluate new indications. Coclinical trials are particularly important because of the potential to compare the effects of treatment in PDXs and original patients. They are expected to become increasingly important soon. In short, the use of PDX models has the potential to provide the ultimate in personalized medicine 1, 6 ; if a PDX model can be pregenerated as a patient's avatar with the ability to predict the effects of a drug before it is administered to a patient, it could not only have a significant impact on patient treatment decisions but also provide an essential platform for drug discovery and development.

On the other hand, the major challenges in using PDX models and conducting coclinical trials are the low transplantation rate, the long transplantation time, and most importantly, the paucity of basic information. Previous reports have shown an engraftment rate of 20%-50%, depending on the patient's cancer type, stage, amount of specimen transplanted, and type of immunodeficient mouse used. 9 It takes 6-12 months after transplantation before a PDX can be evaluated for drug efficacy. 10 Several retrospective studies have been reported comparing anticancer drugs' efficacy in established PDX models and in the original patients. [11] [12] [13] [14] [15] [16] [17] In addition, a high concordance rate of anticancer drug efficacy between PDXs and the corresponding clinical trials has been reported. 1, [18] [19] [20] [21] However, these reports are studies of specific drugs in specific tumors and are rarely systematic.

There is no doubt that the acquisition of preclinical proof of concept will play a role in the success or failure of anticancer drug development in the development of treatments. As the importance of coclinical trials with PDXs is expected to increase, there is an urgent need to obtain basic information for preparing coclinical trials on PDX engraftment rates, time to engraftment, and time to drug efficacy studies for different types of cancer.

We initiated the J-PDX library project in 2018 to create a PDX library from Japanese cancer patients and to innovate in drug development. In this project, we recruited patients focusing on advancedand recurrent-stage cancers resistant to standard treatment and eligible for early clinical trials. In addition, unmet medical needs of pediatric cancers and rare cancers were also set as priority targets.

Over the 21 months to May 30, 2020, we have enrolled 1126 cases and established nearly 300 PDXs. This paper reports on our experience in establishing cross-cancer PDX libraries in the J-PDX library project, cancer type-specific engraftment rates, and time to engraftment. Moreover, the probability of patient survival in conducting a coclinical trial is presented, and insights into planning coclinical trials are discussed.

Between August 22, 2018 and May 31, 2020, a total of 1126 specimens from 1079 patients were enrolled at the National Cancer Center Hospital and National Cancer Center Hospital East in Japan. The protocol was approved by the institutional review consisting of a wide range of cancer subtypes has been successfully established as a platform for drug discovery and development in Japan. When conducting coclinical trials, it is necessary to consider the target cancer type, stage, and engraftment rate in light of this report. 

After confirming consent acquisition, surgical specimens were separated in the pathology department, and biopsy specimens were received in the collection department, including the endoscopy room.

The following samples were used: 2 to 10-mm 3 surgical specimens, 1-2 punctures for needle biopsies, 1-3 tissues for endoscopy, and more than 20 mL of pleural and ascitic fluid. Tissues were immediately soaked in storage solution (Theliokeep, Bio Verde Inc) after collection and stored at 4°C. The specimens were anonymized after receipt and transported to the National Cancer Center Research Institute. 

Pathological tissues were routinely subjected to hematoxylin-eosin staining, human CD45 (clone, D9M8I, Cell Signaling Technology) staining, human COX IV (clone, 3E11, Cell Signaling Technology), and rodent COX IV (clone D6I4K, Cell Signaling Technology) staining to confirm replacement by lymphoma outgrowth and murine tumors. If human CD45 was determined to be 3+ in any of the TG1-3 samples, the PDX was determined to be a lymphoma outgrowth. The percentage of human tissue was confirmed by human COX IV staining, and if the percentage of human COX IV-positive cells was low, staining was performed with rodent COX IV to confirm the presence of murine tumor.

YAGISHITA eT Al.

Patients' characteristics were collected, including age, sex, Eastern Cooperative Oncology Group (ECOG) performance status (PS), smoking history, medical history, and family history. According to the Union for International Cancer Control classification for each tumor, tumor characteristics were noted, including histology and tumor-node-metastasis (TNM) stage. Biomarker data were collected from medical records, including gene analysis results of companion diagnostics, and clinical sequencing. Prior treatment characteristics, including surgery, surgical procedure, radiation dose, chemotherapy regimens, cycles of chemotherapy, best response, progression-free survival, and overall survival (OS), were collected. OS was defined as the time from the day of study enrollment to the last day on which the patient was confirmed alive or dead from any cause.

All information on tumor volume and body weight changes related to PDX implantation, passage, and establishment, as well as biomarker analysis and drug administration results using PDX samples, were aggregated in a database. As an assessment of the time to the growth of PDX tumors, the time to reach a tumor volume of 

The results are expressed as means ± SEM. OS differences were analyzed using the Kaplan-Meier method, and the log-rank test was used to compare survival. Logistic regression analysis was used to investigate the associations between passageable tumors and factors related to patient characteristics. Analyses were performed using STATA SE 

From August 22, 2018 to May 31, 2020, 1126 specimens were received from 1079 patients, with three specimens from one patient and two specimens from 45 patients. Due to the SARS-CoV-2 pandemic's impact, enrollment was temporarily suspended from March 30, 2020 ( Figure 1A ).

The CONSORT diagram is shown in Figure S1 . Of the 1126 specimens enrolled, 16 specimens were unavailable due to inability to obtain specimens after consent. Ten hematologic tumor specimens are awaiting transplantation because the method of transplantation differs from that of solid tumors. As a result, 1100 transplanted specimens were determined to be "Totally Assessable Specimens".

A total of 476 specimens were defined as "Discontinued": 290 specimens that failed to grow in TG1, 39 specimens that failed to grow in TG2, 10 specimens with a murine tumor, and 137 specimens with lymphoma outgrowth. As of May 31, 2020, 326 specimens were TG1 ongoing. Finally, a total of 298 specimens were judged to be "Passageable tumors": 54 TG2 ongoing specimens, 43 TG3 ongoing specimens, and 201 specimens established up to TG3.

A wide range of major cancer types was enrolled, including 258 samples of colorectal cancer, 245 samples of lung cancer, 70 samples of breast cancer, and 54 samples of gastric cancer ( Figure 1B) .

Moreover, rare cancers and sarcomas were also collected in 306 samples from 69 cancer types ( Figure S2 ).

The patients' background characteristics are shown in Table 1 1.27 days, P = .005). The average time to transplantation for specimen type was 0.87 days (0-11 days) for biopsy specimens vs 1.40 days (0-10 days) for surgical specimens and significantly shorter for biopsy specimens (P <.001). The longer time to transplantation was due to the end-of-year holidays and consecutive holidays. Details on the distribution of stages and specimen types by carcinoma are shown in Table S1.

The specimens' establishment status is shown in 

Forty-five cases had two specimens, and one case had three specimens enrolled in this study. Forty-five cases with two specimens enrolled are summarized in Table S2 . Fourteen cases had multiple specimens collected simultaneously (the same day or the next day), and 31 cases had specimens collected over time. Of the cases in which multiple specimens were collected simultaneously, four were from synchronous tumors and 11 were from different sites of the same tumor. Of the 31 cases collected over time, 17 were collected before and after chemotherapy for the same tumor, 11 were collected over time with no intermediate treatment, and three were collected from metachronous tumors. The patients who registered three specimens were combined resection of colorectal cancer and gastric cancer. They provided two specimens from colorectal cancer and one specimen from gastric cancer. Finally, 20 PDXs generated from 10 cases were able to be established in pairs over time.

Typical histopathological images for each carcinoma are shown in Figure S3 . Overall, the histopathological structures of the PDX tumors were retained even after passaging up to TG3 compared with the original tumors.

Representative immunostaining images of lymphoma outgrowth and murine tumor are shown in Figure S4 . Figure S4 A shows a tumor generated from a surgical specimen of colorectal cancer. In TG1, there were only a few CD45-positive cells, but in TG2, all of the human cells were positive for CD45, and it was judged to be a lymphoma outgrowth. Figure S4B is a tumor generated from a biopsy specimen of pancreatic cancer, which showed almost no human COX IV-positive cells from TG1, and was judged to be a murine tumor. F I G U R E 1 Enrollment progress and enrolled cancer types. A, Graph depicting the enrollment progress from August 22, 2018 to May 31, 2020. Blue bars indicate the number of enrollments per month, red lines indicate the total number of enrollments, and green lines indicate the total number of engrafted PDXs. As of May 31, 2020, there were a total of 1126 cases enrolled and 298 engrafted. B, The breakdown of enrolled cases by cancer type. A breakdown of rare cancers is shown in Figure S2 TA B L E 1 Patient characteristics 

For the 201 specimens that could be established up to TG3, the TTV200 in each cancer type is shown in Figure 2A . Overall, the mean TTV200 in each passage was 72.9 days in TG1, 44.9 days in TG2, and 43.1 days in TG3. Figure S5 shows the tumor growth curves up to the passage at TG1, 2, and 3 for each cancer type. The TTP is shown in Figure 3B . 

To explore factors affecting tumor growth in PDX, 774 specimens classified as Passageable tumor and Discontinued were examined in relation to their clinical background characteristics ( prior chemotherapy, and OS were found to be significant factors on univariate analysis, and the significance of stage and OS was confirmed on multivariate analysis (P = .046 and P = .001, respectively).

To investigate the association between patient survival and PDX engraftment in detail, the differences in OS between Passageable tumors and Discontinued were examined using Kaplan-Meier curves.

In all, Passageable tumors had a shorter OS (P = .003), especially lung cancer and rare cancer/sarcoma (P = .004, respectively) ( Figure 3 ). 

Overall, as many as 1100 specimens from more than 50 cancer types were transplanted, and 298 PDXs were successfully generated. failed to grow in TG2, and 0 (0/771, 0%) that failed to grow in TG3, confirming the high probability of implantation after TG2 if TG1 was successfully grown.

The incidence of lymphoma outgrowth was 12.5%, with the highest incidence for colorectal cancer (27.5%), gastric cancer (21.1%), thymic carcinoma (16.7%), and lung cancer (9.9%). Reports on the incidence of lymphoma outgrowth varied widely, but they were generally comparable to previous reports. 19, [24] [25] [26] [27] [28] Furthermore, some mice whose tumors were not grown in TG1 or TG2 showed dermatitis and rapid weight loss, and their autopsy findings showed the accumulation of human CD45-positive cells in dermatitis sites and hepatosplenomegaly. These are thought to be part of the xenograftassociated lymphoproliferative disorder (XALD) of mice caused by human lymphocytes in the transplanted tumor tissue. It has been reported that XALD has a significant effect on the PDX engraftment rate, and that pretreatment with rituximab is effective. 29, 30 It is necessary to investigate the mechanism of XALD and how to prevent it in the future.

Different time to tumor growth has been reported in the past, but there has been no report of a single institute with multiple cancer types in a systematic manner, to the best of our knowledge. In the present study, the TTV200 and the TTP were evaluated as endpoints to assess differences among tumors. We previously conducted a coclinical trial for patients with very poor prognostic histology in uterine cancer. As a high rate of postoperative recurrence was predicted in this population, we prepared for a coclinical trial by generating PDXs from surgical specimens in advance, and we successfully conducted a coclinical trial with multiple patients. We will conduct coclinical trials in collaboration with hospital physicians, basic researchers, and pharmaceutical company researchers with basic and clinical data such as histological type, molecular profile, therapeutic efficacy, and survival status in PDXs and the original patients. Moreover, we will use these data to validate the utility of PDXs and conduct drug discovery and development research using PDXs.

One of the issues we need to work on in the future is understanding the molecular profiles of original patient tumors and corresponding PDX tumors. It has been reported that different genetic mutations in lung cancer led to different PDX engraftment rates. 17 It has also been reported that copy number variations in PDX tumors increase with passaging, and that major genetic mutations in the original tumor are inherited. 32, 33 The effects of these genetic changes associated with passaging in the original and PDX tumors on growth characteristics and drug sensitivity will need to be evaluated in the future.

A key challenge in using PDX for patient avatars and personalized medicine is that tumor characteristics are constantly changing in the patient's body. A PDX created from a treatment-naïve tumor is expected to be different from the tumor status of a patient who has relapsed after chemotherapy. Conversely, PDXs generated from postchemotherapy tumors will be different from treatmentnaïve tumors. In addition, considering that it takes several months to a year to be able to evaluate the efficacy of PDXs and that the establishment rate is about 30%, there is a good chance that the patient's tumor characteristics will change while the PDXs are being established and that the PDXs cannot be established. In the future, it is necessary to study the possibility of improving the establishment rate and shortening the establishment period of PDXs, as well as to examine phenotypic changes with PDXs created over time in the same patient. In the J-PDX, subcutaneous transplantation is used for all patients except breast cancer in order to standardize and unify the technique. It is known that the intratumor microenvironment and growth rate differ between orthotopic and xenotopic transplantation, and orthotopic transplantation is preferred to preserve tumor biology. Because of the difficulty of assessing tumor size in orthotopic transplantation, it is necessary to examine the differences in biology and drug response between orthotopic and xenotopic transplantation.

We hope to use the J-PDX library to promote drug discovery and development research with pharmaceutical companies around the world, as well as biological research with academia and other researchers. However, standardization of evaluation methods is needed to predict the clinical effects in humans from PDX results.

The number of PDX mice to be used, the method of drug administration, the timing of drug administration initiation, the number of passages to be used, the method of determining efficacy, and pharmacokinetics/pharmacodynamics considerations will all need to be developed with clinical efficacy as an endpoint. Our J-PDX library has been constructed to focus on patients with advanced or recurrent stage and resistance to standard treatment. We have also compiled a database of clinical drug administration history and treatment response. This makes it a unique platform to contrast PDX results with those of the original patients.

We have successfully established a library of 300 Japanese cancer patient-derived PDXs as a platform for drug discovery and development in Japan. These data provide important information on survival rates of the original patients for planning coclinical trials.

We will continue to actively promote coclinical trials to establish the use of PDXs and improve the probability of drug discovery and development ( Figure S6) .

We are grateful to all the patients, oncologists, allied healthcare professionals, research concierges, and research assistants who participated in this study. We also thank Mr Hideaki Kakinuma from LSI Medience Corporation for maintaining PDX mice.

The authors declare no potential conflicts of interest.

Shigehiro Yagishita https://orcid.org/0000-0002-1277-1203

Mami Takahashi https://orcid.org/0000-0001-5062-811X

Toshio Imai https://orcid.org/0000-0002-5339-079X

Yasushi Yatabe https://orcid.org/0000-0003-1788-559X

Atsushi Ochiai https://orcid.org/0000-0001-6857-4009

Hiroyuki Mano https://orcid.org/0000-0003-4645-0181

Patient-derived xenograft models: an emerging platform for translational cancer research

Examining the utility of patientderived xenograft mouse models

High-throughput screening using patient-derived tumor xenografts to predict clinical trial drug response

How to improve R&D productivity: the pharmaceutical industry's grand challenge

Parsing clinical success rates

Interrogating open issues in cancer precision medicine with patient-derived xenografts

PDX-MI: minimal information for patient-derived tumor xenograft models

Gene aberrations for precision medicine against lung adenocarcinoma

A systematic review of the validity of patient derived xenograft (PDX) models: the implications for translational research and personalised medicine

Patient-derived xenografts effectively capture responses to oncology therapy in a heterogeneous cohort of patients with solid tumors

A renewable tissue resource of phenotypically stable, biologically and ethnically diverse, patient-derived human breast cancer xenograft models

Tumorgrafts as in vivo surrogates for women with ovarian cancer

Clinical utility of patientderived xenografts to determine biomarkers of prognosis and map resistance pathways in egfr-mutant lung adenocarcinoma

A new model of patient tumor-derived breast cancer xenografts for preclinical assays

Tumor engraftment in nude mice and enrichment in stroma-related gene pathways predict poor survival and resistance to gemcitabine in patients with pancreatic cancer

Anticancer drug response and expression of molecular markers in early-passage xenotransplanted colon carcinomas

Patient-derived first generation xenografts of non-small cell lung cancers: promising tools for predicting drug responses for personalized chemotherapy

Inhibition of MEK and PI3K/ mTOR suppresses tumor growth but does not cause tumor regression in patient-derived xenografts of RAS-mutant colorectal carcinomas

A patient tumor transplant model of squamous cell cancer identifies PI3K inhibitors as candidate therapeutics in defined molecular bins

A molecularly annotated platform of patient-derived xenografts ("xenopatients") identifies HER2 as an effective therapeutic target in cetuximab-resistant colorectal cancer

Characterization of a large panel of patient-derived tumor xenografts representing the clinical heterogeneity of human colorectal cancer

Humane endpoints and cancer research

Humane endpoints for laboratory animals used in regulatory testing

Early development of human lymphomas in a prostate cancer xenograft program using triple knock-out immunocompromised mice

The ability to form primary tumor xenografts is predictive of increased risk of disease recurrence in early-stage non-small cell lung cancer

Setting up a wide panel of patientderived tumor xenografts of non-small cell lung cancer by improving the preanalytical steps

Patient-derived tumor xenografts are susceptible to formation of human lymphocytic tumors

Evaluation of the risk of lymphomagenesis in xenografts by the PCR-based detection of EBV BamHI W region in patient cancer specimens

Prevention of human lymphoproliferative tumor formation in ovarian cancer patient-derived xenografts

Rituximab treatment prevents lymphoma onset in gastric cancer patient-derived xenografts

Prioritizing phase I treatment options through preclinical testing on personalized tumorgraft

Patient-derived ovarian tumor xenografts recapitulate human clinicopathology and genetic alterations

Patient-derived xenografts undergo mouse-specific tumor evolution

Additional supporting information may be found online in the Supporting Information section