Randomized Phase II Study of PARP Inhibitor ABT-888 (Veliparib) with Modified FOLFIRI versus FOLFIRI as Second-line Treatment of Metastatic Pancreatic Cancer: SWOG S1513

This article is featured in Highlights of This Issue, p. 6279

Metastatic pancreatic ductal adenocarcinoma takes an immense toll on life (1). After progression from first-line therapies, second-line regimens include 5-fluorouracil (5-FU), folinic acid with liposomal irinotecan or irinotecan (FOLFIRI) with median survival of 6 months (2, 3). Pancreatic cancer is characterized by genomic instability and may harbor loss-of-function mutations in the BRCA–Fanconi anemia (FA) pathway: Breast cancer genes 1 and 2 (BRCA1 and BRCA2), partner and localizer of BRCA2 (PALB2), and FA genes (4, 5). Additional homologous recombination DNA damage repair (HR-DDR) gene defects contribute to homologous recombination deficiency (HRD) in approximately 15% of patients (6). The high lethality of pancreatic cancer stems partially from its resistance to cytotoxic agents. PARP enzymes contribute to repair from topoisomerase 1–associated DNA damage (7), and PARP inhibition increases cytotoxicity from camptothecins (8, 9).

PARP inhibitors have modest single-agent activity for previously treated or refractory pancreatic cancers associated with germline BRCA1/2/PALB2 loss-of-function mutations, with overall response rates (ORR) of 0%–22%, progression-free survival (PFS) averaging 2–4 months, and overall survival (OS) up to 10 months (10–13). Nevertheless, olaparib received FDA approval for patients with germline BRCA1/2-mutated cancers as maintenance therapy after response or stable disease from first-line platinum chemotherapy, due to improved PFS (7.4 months with olaparib vs. 3.8 months with placebo, P = 0.004; ref. 14). Studies of veliparib with 5-FU, folinic acid, and oxaliplatin (FOLFOX; ref. 15), or with gemcitabine plus cisplatin (16, 17) demonstrated response rates of 30%–77% and survival rates of 9–23 months for patients with germline BRCA1/2/PALB2-mutated cancers. The combination of FOLFIRI with veliparib was tested in a phase I trial among 92 patients with refractory solid tumors (18). Veliparib was dosed from 10 to 270 mg BID (twice daily). To maximize PARP inhibition around the time of FOLFIRI administration and to minimize toxicities, especially myelosuppression, veliparib was dosed intermittently on days 1–5 and 15–19, with FOLFIRI dosed on days 1–3 and 15–17 in 28-day cycles. Dose-limiting toxicities were grade 4 neutropenia (n = 2 at 160 and 270 mg BID), grade 4 febrile neutropenia (100 mg BID) and grade 3 gastritis (270 mg BID). To improve tolerability, the 5-FU bolus was discontinued, and the recommended phase 2 dose was veliparib 200 mg BID with modified (m) FOLFIRI (FOLFIRI without 5FU bolus). Response and stable disease rates among 14 patients with refractory pancreatic cancers without known BRCA1/2 mutations were 14% (n = 2, lasting 10 and 25 months) and 43% (n = 6), respectively, and the 6-month time-to-progression was 27%. Given the preclinical synergism between veliparib and irinotecan (7–9) and phase I data demonstrating safety and preliminary efficacy of veliparib combined with mFOLFIRI, including for cancers without known BRCA1/2/PALB2 mutations (18), we designed a randomized phase II study to assess whether veliparib improves survival when added to mFOLFIRI compared with FOLFIRI alone for second-line treatment of metastatic pancreatic ductal adenocarcinoma. Secondary objectives were to evaluate ORR, PFS, safety, and tolerability, with translational objectives to evaluate whether germline or somatic BRCA1/2 mutations and other HR-DDR gene alterations were associated with efficacy outcomes in each arm.

This study was conducted by the SWOG Cancer Research Network and the National Cancer Institute (NCI) National Clinical Trials Network (NCTN), with 28 participating institutions. The applicable regulations and guidelines governing clinical study conduct were followed and the study was performed in compliance with the Declaration of Helsinki. Participating site institutional review board approval and patient written informed consent were obtained before enrollment. Pertinent eligibility criteria were age ≥18 years, adequate organ function, Zubrod performance status 0–1, no prior irinotecan or PARP inhibitors, and either one prior therapy for metastatic pancreatic cancer, or progression to metastatic disease within 3 months of gemcitabine/nab-paclitaxel treatment for localized disease. Patients must have been able and willing to undergo pre-treatment biopsy and submit tumor and blood samples for research purposes; archival tumor sample, if available, was required. The trial protocol is available in the Supplementary Appendix S1.

Each 2-week cycle of FOLFIRI consisted of irinotecan 180 mg/m², folinic acid 400 mg/m², 5-FU bolus 400 mg/m² and 5-FU infusion 2,400 mg/m² according to published guidelines (19). mFOLFIRI consisted of irinotecan 180 mg/m², folinic acid 400 mg/m², and 5-FU infusion 2,400 mg/m² with no 5-FU bolus (18). Veliparib (ABT-888, NSC-737664) was provided by the NCI's Division of Cancer Therapy Evaluation Program (CTEP). To maximize PARP inhibition before and during chemotherapy dosing, veliparib was administered orally 200 mg BID on days 1–7, with mFOLFIRI on days 3–5 of every 2-week cycles. If irinotecan or 5-FU infusions were held for toxicity, veliparib was held. Dose reductions for FOLFIRI and mFOLFIRI followed published guidelines (19). Veliparib could be reduced to 150, 100, and 50 mg BID for toxicities.

Adverse events were assessed for severity using NCI Common Toxicity Criteria for Adverse Events (CTCAE) version 4.0 (20). Patients were evaluated for tumor response every 8 weeks according to modified RECIST version 1.1 (21). Blood and tumor samples from fresh biopsies, and archival samples if available, were collected before starting treatment. Blood and tumor DNA was sequenced using BROCA-HR, a targeted capture, massively parallel sequencing test developed at University of Washington (22). BROCA-HR captures >99% of deleterious BRCA1/2 mutations (including gene rearrangements and copy number variations), and includes other genes in the FA-BRCA HR, mismatch repair, nucleotide excision repair, non-homologous end joining pathways as well as surveillance and modifier genes of interest (Supplementary Table S1; ref. 23). HR-DDR repair gene defects were categorized in four groups based on degree of HRD conferred by each gene: (i) BRCA1/2; (ii) core non-BRCA1/2: PALB2, ATM, RAD51C/D, BRIP1, BARD1; (iii) non-core genes that could impact DNA repair: For example, FANCA-M, CDK12, CHEK2, BLM, SLX4, ERCC1/4; (iv) wild-type, which includes non–HR-DDR genes: For example, EZH2, MSH2, MSH6, POLD1, POLE, RIF1, WRN.

The primary outcome was OS defined as time from randomization to death of any cause. Secondary outcomes included toxicity; PFS defined as time from randomization to disease progression, symptomatic deterioration, or death by any cause; ORR defined as confirmed and unconfirmed complete and partial response, disease control rate defined as confirmed and unconfirmed complete and partial response and confirmed and unconfirmed stable disease; and duration of response defined as time from first documentation of response to PFS failure. Translational secondary outcomes were to evaluate the impact of germline and somatic BRCA1/2 and other HR-DDR genetic defects on efficacy outcomes in each arm.

Patients were enrolled by the treating physicians at the participating hospitals. Patients were randomly assigned in equal proportions to the treatment arms using a dynamic balancing algorithm with stratification by prior systemic treatment for metastatic pancreatic cancer (yes vs. no; ref. 24). In this open-label study, patients, investigators, and the study team were not masked to study treatment.

We hypothesized the median survival with FOLFIRI to be 6 months (2). Assuming a one-sided type I error of 10%, approximately 28 months accrual, and 10 months follow-up, 128 eligible patients provided 80% power to detect a 33% increase in median survival to 9 months [hazard ratio (HR), 0.67] with veliparib plus mFOLFIRI. Approximately 110 deaths were expected at the time of the final analysis. An interim futility analysis of PFS was planned when 35% of the expected events (approximately 40 events) in the control arm were observed; evidence suggesting early closure consisted of rejection of the alternative hypothesis (HR, 0.67) at one-sided P < 0.05. For OS and PFS and duration of response, censoring occurred at the date of last contact. The study did not require selection or stratification based on BRCA1/2 mutation status.

The primary analysis of OS was conducted in all eligible and evaluable patients according to the intent-to-treat principle. Probabilities of OS, PFS, and duration of response were estimated using the Kaplan–Meier method, with statistical differences in event rates between treatment arms assessed via stratified log-rank test. ORR and disease control rate were compared across treatment arm by χ² or Fisher's exact test in patients with measurable disease. Eligible patients who received at least one dose of any drug were included in the safety analysis. Prespecified analyses between HR-DDR gene defects and efficacy outcomes were conducted. Associations of HR-DDR gene defects groups: Any (core + non-core), and wild-type with OS and PFS were estimated via the Kaplan–Meier method and Cox regression where feasible. Clinical benefit response was defined as overall response plus stable disease lasting for at least 4 months from randomization, and differences in clinical benefit response between HR-DDR gene defects groups were assessed via the Fisher's exact test. All P values were two-sided with significance level of 5%, apart from the planned interim futility analysis. Analyses were carried out using SAS version 9.4 and R version 3.5.2.

A total of 123 patients were randomized between September 1, 2016 and December 13, 2017 (Fig. 1). Four patients were deemed ineligible due to not receiving prior systemic therapy for metastatic pancreatic cancer or for localized disease with metastatic progression within 3 months from treatment (n = 3) or inadequate hematologic function (n = 1); two eligible patients refused their assigned FOLFIRI treatment and follow-up and were excluded from efficacy and safety analyses. Baseline characteristics for 117 eligible patients were balanced (Table 1). Among all eligible patients enrolled, 55 (93%) and 50 (86%) in the veliparib and control arm, respectively, received prior treatment with gemcitabine plus nab-paclitaxel. No patients were previously treated with FOLFIRINOX. The study was closed to accrual following review of the interim futility analysis by the SWOG Data and Safety Monitoring Committee, given the results suggested that veliparib plus mFOLFIRI was unlikely to be more effective than FOLFIRI (control), based on rejection of the PFS alternative hypothesis. All patients assigned to receive veliparib were advised to discontinue veliparib but were allowed to continue on study with mFOLFIRI alone.

The safety analysis included 106 patients, because 11 eligible patients never started treatment. Treatment-related grade 3/4 toxicities observed more often with veliparib versus control were neutropenia, fatigue, nausea, and diarrhea (Table 2). No treatment-related grade 5 toxicities were reported. More patients experienced ≥grade 3 toxicities in the veliparib arm versus control: 69% versus 58% (P = 0.23). Patients had similar treatment exposure: 4 (IQR 3–8) versus 5 (IQR 3–14) cycles (P = 0.15) with veliparib versus control (Table 3), and similar proportions had at least one dose reduction, dose held for toxicity, or discontinuation of chemotherapy: 66% versus 56% (P = 0.39).

Efficacy analyses were conducted in 117 patients, with median follow-up of 9 months (IQR 1–27). Median survival did not vary significantly between treatment arms: 5.4 months [95% confidence interval (CI), 3.7–7.2] with veliparib and mFOLFIRI versus 6.5 months (95% CI, 5.6–7.8) with FOLFIRI (HR, 1.23; 95% CI, 0.85–1.78; P = 0.28; Fig. 2A). Median PFS was 2.1 months (95% CI, 1.9–2.4) with veliparib and mFOLFIRI versus 2.9 months (95% CI, 2.2–4.2) with FOLFIRI (HR, 1.39; 95% CI, 0.96–2.02; P = 0.09; Fig. 2B). Among 106 patients treated, 103 patients (55 in the veliparib arm and 48 in the control arm) had adequate response assessment. Six (11%; 95% CI, 2.7%–19.2%) versus five patients (10%; 95% CI, 1.8%–19.1%; P = 0.99) had partial response (confirmed or unconfirmed), and 14 (25%; 95% CI, 13.9%–37%) versus 21 patients (44%; 95% CI, 29.7%–57.8%; P = 0.06) had stable disease with veliparib versus control, respectively. Disease control rates were 36% (95% CI, 24%–49%) versus 54% (95% CI, 40%–68%; P = 0.07), and median duration of response was 5.1 months (95% CI, 2.2–5.2) versus 3.4 months (95% CI, 2.6–3.4; P = 0.99) with veliparib versus control, respectively.

Of 117 patients who underwent BROCA-HR testing, 115 were eligible, and 102 had adequate response assessment. Twenty-two patients (19%) had cancers with HR-DDR gene defects and 10 patients (9%) had cancers with other non–HR-DDR gene defects (Supplementary Table S2)

In the FOLFIRI arm, patients with cancers with any HR-DDR gene defects (n = 10) versus wild-type (n = 46) had median OS of 10.1 months (95% CI, 3.8–17.6) versus 5.9 months (95% CI, 4.9–7.7; HR, 1.63; 95% CI, 0.81–3.26; P = 0.17), median PFS of 7.3 months (95% CI, 2.0–9.7) versus 2.5 months (95% CI, 2.1–3.8; HR, 1.98; 95% CI, 0.98–3.99; P = 0.05), response rates of 20% (95% CI, 0.01%–44.8%) versus 8% (95% CI, 0%–16.9%; P = 0.29), and clinical benefit response rates of 70% (95% CI, 41.6%–98.4%) versus 35% (95% CI, 19.8%–50.5%; P = 0.07; Fig. 3A, C, and E).

In the veliparib plus mFOLFIRI arm, patients with cancers with any HR-DDR gene defects (n = 12) versus wild-type (n = 47) had median OS of 7.4 months (95% CI, 1.7–16.3) versus 5.1 months (95% CI, 3.3–6.2; HR, 1.78; 95% CI, 0.89–3.55; P = 0.10), median PFS of 2.0 months (95% CI, 0.8–8.4) versus 2.1 months (95% CI, 1.9–2.5; HR, 1.19; 95% CI, 0.61–2.33; P = 0.62), response rates of 18% (95% CI, 0%–41%) versus 9% (95% CI, 0.60%–17.6%; P = 0.59), and clinical benefit response rates of 45% (95% CI, 16%–74.9%) versus 25% (95% CI, 12.2%–37.8%; P = 0.27; Fig. 3B, D, and E). Efficacy outcomes for individual carriers of core and non-core HR-DDR gene defects (n = 22) treated with veliparib and mFOLFIRI or FOLFIRI are shown in Fig. 3E and Supplementary Table S2.

SWOG S1513 was a randomized phase II study to assess whether the PARP inhibitor veliparib improved survival when added to second-line FOLFIRI chemotherapy for patients with metastatic pancreatic cancer. The study closed to accrual after the interim analysis (123 patients accrued of 143 patients planned) demonstrated futility with the addition of veliparib. Veliparib did not improve OS in combination with mFOLFIRI compared with FOLFIRI alone (5.4 vs. 6.5 months; HR, 1.23; P = 0.28). Notably, the median survival of 6.5 months with FOLFIRI was comparable with the survival of 6.2 months observed with the NAPOLI-1 regimen nal-Iri/5-FU (3).

This was the first study performed through the NCI sponsored National Clinical Trials Network that prospectively tested the presence of germline and somatic DDR defects in patients with metastatic pancreatic cancer and assessed their effects on outcomes with chemotherapy with or without a PARP inhibitor. Although the presence of germline or somatic BRCA1/2/PALB2 mutations was not required for study enrollment, 117 patients (95%) provided blood and tumor samples that were analyzed for HR-DDR gene defects. Twenty-two patients (19%) had germline or somatic defects, including five (4%) patients with BRCA1/2/PALB2 mutations. Small numbers preclude definitive conclusions, but the addition of veliparib did not benefit patients with BRCA1/2 or any HR-DDR defects: median OS 7.4 versus 10.1 months (P = 0.65), median PFS 2.0 versus 7.3 months (P = 0.15), and response rates 18% versus 20% (P = 0.99) with veliparib versus control, respectively.

Toxicities, especially neutropenia, nausea, vomiting, diarrhea, dehydration, and fatigue, were increased with veliparib, although no differences reached statistical significance. Although neither the median number of cycles received, nor the incidence of chemotherapy dose modifications varied significantly between arms, numerically fewer chemotherapy cycles were administered in the veliparib arm, which may have led to inferior outcomes. We have not conducted a pharmacokinetic analysis in this study, but previous studies noted no drug–drug interactions between veliparib and irinotecan or veliparib and FOLFIRI, suggesting that veliparib should not have influenced FOLFIRI drug exposure (18, 25).

On the basis of phase 1 data (18), we dosed veliparib at 200 mg BID on days 1–7 with mFOLFIRI on days 3–5, every 14 days. O'Reilly and colleagues (17) tested first-line treatment with cisplatin/gemcitabine with or without veliparib 80 mg BID on days 1–12 of every 21-days cycle in gBRCA1/2/PALB2 mutations carriers and observed high responses (65%–74%) and survival rates (medians, 15.5–16.4 months) in both arms, but no benefit from the addition of veliparib. Veliparib has been the only PARP inhibitor that could safely combine with chemotherapy, albeit at lower doses than used as monotherapy (400 mg BID), but no evidence of clinical synergism has been observed to date in patients with metastatic pancreatic cancer.

The VELIA/GOG-3005 study in ovarian carcinomas showed that veliparib 150 mg BID with carboplatin/paclitaxel, followed by 400 mg BID maintenance significantly improved PFS in patients with BRCA1/2-mutated high-grade serous ovarian carcinomas, as well as in cancers with or without HRD compared with chemotherapy alone (PFS 34.7 vs. 22, 32 vs. 20.5, and 23.5 vs. 17.3 months, respectively; ref. 26). Although it is possible that the maintenance (100% dose intensity) rather than the concurrent administration with chemotherapy (37.5% dose intensity) conferred the PFS benefit, to date, VELIA/GOG-3005 is one of the few clinical trials where a PARP inhibitor added to chemotherapy remarkably improved efficacy (P < 0.001). Veliparib dosed intermittently at 120 or 100 mg BID, respectively, improved PFS when added to carboplatin/paclitaxel in gBRCA1/2-mutated breast cancers (medians 14.5 vs. 12.6 months, P = 0.0016; ref. 27), or with cisplatin/etoposide in small-cell lung cancers (medians 6.1 vs. 5.5 months, P = 0.06; ref. 28), but the benefit was modest. Studies to date in pancreatic cancer have not tested continuously dosed veliparib with standard chemotherapy. It is not clear whether the lack of benefit from veliparib in SWOG S1513 was due to the lower than standard dose (50%) and/or intermittent dosing, or, alternatively the lower incidence of HR-DDR defects (approximately 20%) compared with high-grade serous ovarian carcinomas (50% HRD) that are inherently more susceptible to PARP inhibition.

It has been postulated that mechanistic differences in catalytic activity and the capacity for PARP1 trapping on DNA account for differential PARP inhibitors' cytotoxicity or ability to combine with chemotherapy (29), with olaparib and rucaparib having increased PARP1-trapping capacity compared with veliparib. Few patients seem to benefit from PARP inhibitors after progression from first-line therapy (10–13). Nevertheless, PARP inhibitors are active as first-line maintenance for platinum-sensitive pancreatic cancers. Olaparib maintenance after response or stable disease from first-line platinum chemotherapy in germline BRCA1/2-mutated pancreatic cancers improved PFS (medians 7.4 vs. 3.8 months, P = 0.004), and response rates (23% vs. 11.5%), but conferred no survival advantage compared with placebo (medians 19 vs. 19.2 months; P = 0.35; refs.12, 30). Rucaparib maintenance has also demonstrated encouraging activity, with response rates of 41.7%, median PFS and OS of 13.1 and 23.5 months, respectively, among patients with germline or somatic BRCA1/2/PALB2-mutated, platinum-sensitive, advanced pancreatic cancers without radiologic or CA19–9 progression within 8 weeks after stopping treatment with platinum chemotherapy (31).

In SWOG S1513, patients with cancers with any germline or somatic HR-DDR gene defects versus wild-type treated with FOLFIRI and with veliparib plus mFOLFIRI had numerically longer OS (10.1 vs. 5.9 and 7.4 vs. 5.1 months, respectively), and higher clinical benefit responses (70% vs. 35%, and 45% vs. 25%, respectively), suggesting a prognostic impact from treatment with FOLFIRI. Nevertheless, the small proportion of cancers with BRCA1/2 and other HR-DDR gene defects makes it difficult to determine their predictive value in each treatment arm. Golan and colleagues (32) have recently described classifiers for an HRD signature and the importance of biallelic inactivation for BRCA1/2 and PALB2 in pancreatic cancers, which may be better predictors of response to DNA-damaging agents and/or PARP inhibitors and should be tested in future studies.

Given the increased hematologic and gastrointestinal toxicity with chemotherapy and PARP inhibitors, new clinical trials, including SWOG S2001, are testing novel combinations of PARP inhibitors with immune checkpoint blockade, RAS/PI3K/AKT/MEK pathways inhibitors, cell cycle or other DDR inhibitors, hoping to enhance efficacy without overlapping toxicity (33–37).

In conclusion, SWOG S1513 demonstrated that second-line treatment with veliparib added to mFOLFIRI did not improve survival for patients with metastatic pancreatic cancer and found that HR-DDR genetic defects may be associated with improved prognosis with FOLFIRI chemotherapy. Identifying susceptibility biomarkers beyond BRCA1/2/PALB2 mutations to predict benefit from DNA-damaging agents and PARP inhibitors is feasible, and novel treatment approaches for patients with distinct molecular signatures are urgently needed.

E.G. Chiorean reports grants from National Cancer Institute of the National Institutes of Health during the conduct of the study as well as personal fees from AstraZeneca, Bayer, Celgene, Eisai, Ipsen, Legend, Noxxon, Pfizer, Seattle Genetics, Sobi, and Stemline, and grants from Boehringer–Ingelheim, Bristol Myers Squibb, Celgene, Clovis, Corcept, Fibrogen, Halozyme, Incyte, Lilly, MacroGenics, Merck, Rafael, Roche, and Stemline outside the submitted work. K.A. Guthrie reports grants from NIH/NCI during the conduct of the study. E.M. Swisher reports grants from NIH during the conduct of the study. M.J. Pishvaian reports grants and personal fees from Merck and AstraZeneca/Medimmune, as well as personal fees from Caris Life Sciences, Perthera, and Sirtex Medical; grants from ARMO Biosciences, Bavarian Nordic, Bayer, Bristol-Myers Squibb, Calithera Biosciences, Celgene, Celldex, Curegenix, Fibrogen, Genentech, Gilead Sciences, and GSK; grants and personal fees from Halozyme; and grants from Karyopharm Therapeutics, Novartis, Regeneron, Pfizer, Pharmacyclics, Tesaro, and Seattle Genetics outside the submitted work; in addition, M.J. Pishvaian reports a patent for AbbVie pending to Myself and Perthera issued to Myself. J. Berlin reports grants from National Cancer Institute during the conduct of the study as well as grants from Atreca, Pfizer, Astellas, Lilly, Boston Biomedical, I-MAB, Symphogen, Karyopharm, Dragonfly, Immunomedics, and BMS and personal fees from Seagen, QED, Mirati, Pancreatic Cancer Action Network, Novocure, Ipsen, Bayer, Eisai, EMD Serono, and Clovis outside the submitted work. M.S. Noel reports other support from Ipsen outside the submitted work. I. Garrido-Laguna reports grants from Navire, BMS, Novartis, Trishula, Redhill Therapeutics, Tolero, Amgen, GSK, Bayer, Array, Medimmune, Rafael, Lilly, and Seattle Genetics, and personal fees from Pfizer, Eisai, and Sotio outside the submitted work. D.B. Cardin reports personal fees from AbbVie outside the submitted work. A.M. Lowy reports personal fees from Pfizer, Kinnate, Merck, Huya, Halozyme, Rafael, and Corcentric, and grants from Syros outside the submitted work. H.S. Hochster reports personal fees from Bayer and Genentech and personal fees and non-financial support from Elion and TRIGR outside the submitted work. No disclosures were reported by the other authors.

E.G Chiorean: Conceptualization, resources, data curation, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. K.A. Guthrie: Conceptualization, data curation, formal analysis, validation, investigation, methodology, writing–original draft, writing–review and editing. P.A. Philip: Conceptualization, resources, data curation, supervision, funding acquisition, investigation, project administration, writing–review and editing. E.M. Swisher: Conceptualization, resources, formal analysis, validation, investigation, methodology, writing–original draft, writing–review and editing. F. Jalikis: Resources, investigation, writing–review and editing. M.J. Pishvaian: Resources, investigation, writing–review and editing. J. Berlin: Resources, investigation, writing–review and editing. M.S. Noel: Resources, investigation, writing–review and editing. J.M. Suga: Resources, investigation, writing–review and editing. I. Garrido-Laguna: Resources, investigation, writing–review and editing. D.B. Cardin: Resources, investigation, writing–review and editing. M.R. Radke: Resources, investigation, writing–review and editing. M. Duong: Formal analysis, writing–review and editing. S. Bellasea: Formal analysis, writing–review and editing. A.M. Lowy: Conceptualization, resources, funding acquisition, investigation, writing–review and editing. H.S. Hochster: Conceptualization, resources, supervision, funding acquisition, investigation, project administration, writing–review and editing.

Research reported in this publication was supported by the National Cancer Institute of the National Institutes of Health under award numbers CA180888, CA180819, CA180820, CA180821, CA233230, CA189821, CA180818, CA189830, CA189856, CA189858, CA180826, CA190002, CA189957, CA180828, CA189971, CA180835, CA189809, CA46368, CA189960, CA180834, CA239767, CA189860, CA189861, CA189854, CA189872, CA189958, CA180801, and CA189848.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.