Association of Tumor Mutational Burden with Efficacy of Pembrolizumab±Chemotherapy as First-Line Therapy for Gastric Cancer in the Phase III KEYNOTE-062 Study

In many cancers, tumor cells acquire somatic mutations that can give rise to the formation of neoepitopes, which act as neoantigens and can be recognized by T cells (1, 2). Oncogenic viruses such as the Epstein-Barr virus (EBV) also contribute to the pool of neoantigens by generating viral antigens distinct from those that result from somatic mutations (1, 2). Furthermore, tumors with high microsatellite instability (MSI-H) are associated with high tumor mutational burden (TMB-H; ref. 3). Consequently, tumors with high TMB, high neoantigen load, or both may have a larger pool of tumor-specific T cells that can be re-activated by immune checkpoint inhibition (1–3).

TMB and neoantigen load have been shown to predict response to immune checkpoint inhibitors, including the anti–programmed death 1 monoclonal antibody pembrolizumab, across multiple tumor types (4–12). Exploratory analyses from studies in non–small cell lung cancer have demonstrated an association between high TMB and improved clinical outcomes in patients treated with pembrolizumab monotherapy but not in those treated with pembrolizumab in combination with chemotherapy (pembrolizumab-chemotherapy) or chemotherapy alone (13, 14). In a recent exploratory analysis of the phase III KEYNOTE-061 trial, tissue TMB analyzed using the FoundationOne® CDx next-generation sequencing (NGS)–based panel (Foundation Medicine) also demonstrated an association for clinical outcomes with pembrolizumab but not paclitaxel as second-line therapy in patients with gastric/gastroesophageal junction (G/GEJ) cancer (15). Furthermore, in a prospectively planned analysis across a range of solid tumor types from the KEYNOTE-158 study, pembrolizumab-treated patients with tumors identified as TMB-H [≥10 mutations per megabase (mut/Mb) by the FoundationOne CDx] demonstrated an objective response rate (ORR) of 29% and durable responses (>50% of patients had response durations of ≥24 months; ref. 16). These findings from KEYNOTE-158 led to the FDA approval of pembrolizumab for the treatment of patients with unresectable or metastatic TMB-H (≥10 mut/Mb) solid tumors who experienced disease progression while receiving previous treatment and who have no satisfactory alternative treatment options (17).

The KEYNOTE-062 study was a randomized, controlled, partially blinded, phase III trial that evaluated pembrolizumab, pembrolizumab-chemotherapy, and chemotherapy alone as first-line treatment in patients with advanced G/GEJ adenocarcinoma (18). Pembrolizumab monotherapy was non-inferior to chemotherapy for overall survival (OS) in patients with a programmed death ligand 1 (PD-L1) combined positive score (CPS) of ≥1 and showed clinically meaningful improvement in patients with a CPS of ≥10 (18). Pembrolizumab-chemotherapy did not show clinically meaningful benefit in OS or progression-free survival (PFS) compared with chemotherapy alone (18).

Herein, we present the results of a prespecified exploratory analysis from KEYNOTE-062 that evaluated the association between TMB status and clinical outcomes in patients with advanced G/GEJ adenocarcinoma treated with pembrolizumab, pembrolizumab-chemotherapy, or chemotherapy alone as first-line therapy.

The phase III KEYNOTE-062 study was conducted at 200 centers in 29 countries; details about the study design and the eligibility criteria have been reported (18). Briefly, key eligibility criteria included the following: age 18 years or older; histologically confirmed locally advanced/unresectable or metastatic disease; HER2-negative, PD-L1 CPS ≥1 tumors; measurable disease as per RECIST v1.1 by investigator assessment; and Eastern Cooperative Oncology Group performance status (ECOG PS) score of 0 or 1. Patients had to provide either a newly obtained or an archival tumor sample for PD-L1 analysis.

Patients were randomly assigned in a 1:1:1 ratio to receive pembrolizumab 200 mg every 3 weeks, pembrolizumab plus chemotherapy (cisplatin 80 mg/m² on day 1 plus 5-fluorouracil 800 mg/m² on days 1 to 5 or capecitabine 1,000 mg/m² twice daily on days 1 to 14, all every 3 weeks), or placebo plus chemotherapy. Randomization was stratified by region (Europe/North America/Australia vs. Asia vs. rest of world), disease status (locally advanced vs. metastatic), and chosen fluoropyrimidine (5-fluorouracil vs. capecitabine). Treatment continued until documented disease progression, unacceptable toxicity, or physician/patient withdrawal, or 35 administrations (2 years) of pembrolizumab or placebo.

The study protocol and all amendments were approved by the institutional review board or ethics committee at each participating institution. The study was conducted in accordance with the protocol, its amendments, the ethical principles originating from the Declaration of Helsinki, and Good Clinical Practice guidelines. Written informed consent was provided by all patients before enrollment.

In this exploratory analysis, prespecified primary objectives included assessment of whether TMB as a continuous variable (transformed to the square root scale) is associated with improved clinical outcomes of pembrolizumab or pembrolizumab-chemotherapy. The association between TMB and clinical outcomes of chemotherapy alone was also evaluated. The relative treatment effects of pembrolizumab or pembrolizumab-chemotherapy versus chemotherapy alone were estimated in patients with TMB-H (≥10 mut/Mb) and in patients whose tumors were non–TMB-H (<10 mut/Mb). The secondary objectives were to test whether TMB and PD-L1 are independent predictors of response to pembrolizumab, pembrolizumab-chemotherapy, and chemotherapy alone, separately, in a multivariable model and to evaluate the consistency of TMB testing results and clinical utility of TMB ≥10 mut/Mb in the non–MSI-H subgroup [defined as one unstable locus of five or no microsatellite instability (MSI) detected (i.e., microsatellite stable) or unknown MSI status].

Clinical outcomes of interest assessed in this analysis included confirmed ORR and PFS by blinded central radiology review as per RECIST v1.1 and OS. Response was assessed by radiographic tumor imaging using CT or MRI, which was performed every 6 weeks.

TMB was assessed by NGS using the FoundationOne CDx. PD-L1 expression was centrally assessed during screening using PD-L1 IHC 22C3 pharmDx (Agilent); PD-L1–positive tumors had CPS ≥1, which was calculated as the number of PD-L1–staining cells (tumor cells, macrophages, and lymphocytes) divided by the total number of viable tumor cells, multiplied by 100. MSI/DNA mismatch repair across five mononucleotide repeat markers (NR21, NR24, BAT25, BAT26, and MONO27) was assessed using DNA extracted from formalin-fixed, paraffin-embedded tumor samples and blood (normal control) using the MSI Analysis System, version 1.2 (Promega). Tumors with high levels of MSI were those for which ≥2 markers changed compared with normal controls.

This exploratory analysis included all treated patients in KEYNOTE-062 with available TMB data that passed quality control; analyses followed a scientific and statistical analysis plan written before merging clinical and FoundationOne CDx–based TMB data, specifying where statistical testing would be used and TMB cutoffs that would define subgroups for pembrolizumab or pembrolizumab-chemotherapy versus chemotherapy efficacy comparisons. TMB endpoints were generated before the merging of TMB and clinical data; thus, TMB data were masked to treatment group and clinical outcomes. The TMB cutoff of 10 mut/Mb used for this analysis was based on the analysis of KEYNOTE-158 in a range of advanced solid tumor types, which demonstrated that TMB-H (≥10 mut/Mb) was associated with higher tumor response rates with pembrolizumab monotherapy in the study population (16).

For testing of the association between TMB and clinical outcomes, logistic regression (ORR) and Cox proportional hazards regression (PFS and OS) models were used, adjusting for ECOG PS (0 vs. ≥1). Nominal P values were calculated using the one-sided Wald test for pembrolizumab monotherapy and pembrolizumab-chemotherapy (hypothesized positive association) and the two-sided Wald test for chemotherapy alone (no assumed direction hypothesized). A prespecified subgroup analysis using a TMB cutoff was performed to understand the potential clinical utility, categorizing patients into two groups using the predefined TMB cutoff of 10 mut/Mb and estimating the efficacy of pembrolizumab or pembrolizumab-chemotherapy compared with chemotherapy in those TMB subgroups. Within each TMB population, a Cox proportional hazards regression model was used to estimate the HR and the 95% confidence interval (CI) for pembrolizumab monotherapy compared with chemotherapy and pembrolizumab-chemotherapy compared with chemotherapy.

The clinical data cutoff date for this analysis was March 26, 2019. This trial is registered with ClinicalTrials.gov: NCT02494583.

Merck Sharp & Dohme LLC (MSD), a subsidiary of Merck & Co., Inc., Rahway, NJ, is committed to providing qualified scientific researchers access to anonymized data and clinical study reports from the company's clinical trials for the purpose of conducting legitimate scientific research. MSD is also obligated to protect the rights and privacy of trial participants and, as such, has a procedure in place for evaluating and fulfilling requests for sharing company clinical trial data with qualified external scientific researchers. The MSD data sharing website (available at: http://engagezone.msd.com/ds_documentation.php) outlines the process and requirements for submitting a data request. Applications will be promptly assessed for completeness and policy compliance. Feasible requests will be reviewed by a committee of MSD subject matter experts to assess the scientific validity of the request and the qualifications of the requestors. In line with data privacy legislation, submitters of approved requests must enter into a standard data-sharing agreement with MSD before data access is granted. Data will be made available for request after product approval in the United States and the European Union or after product development is discontinued. There are circumstances that may prevent MSD from sharing requested data, including country or region-specific regulations. If the request is declined, it will be communicated to the investigator. Access to genetic or exploratory biomarker data requires a detailed, hypothesis-driven statistical analysis plan that is collaboratively developed by the requestor and MSD subject matter experts; after approval of the statistical analysis plan and execution of a data-sharing agreement, MSD will either perform the proposed analyses and share the results with the requestor or will construct biomarker covariates and add them to a file with clinical data that is uploaded to an analysis portal so that the requestor can perform the proposed analyses.

Between September 18, 2015, and May 26, 2017, 763 patients were randomly assigned to pembrolizumab (n = 256), pembrolizumab-chemotherapy (n = 257), or chemotherapy (n = 250). In the overall study population, the median time from randomization to data cutoff was 29.4 months (range, 22.0–41.3).

Overall, 306 of 763 patients (40.1%) had evaluable TMB data (pembrolizumab, n = 107; pembrolizumab-chemotherapy, n = 100; chemotherapy, n = 99) and were included in this analysis (Fig. 1).

Baseline characteristics in the TMB-evaluable population were well balanced between treatment groups and were generally similar to those of the overall study population (Table 1). Compared with the overall study population of KEYNOTE-062, the TMB-evaluable population had a lower proportion of patients from Asia (Table 1). ORR and PFS outcomes in the TMB-evaluable population were generally consistent with those observed in the overall study population; however, there were inconsistencies in OS in the pembrolizumab monotherapy group (median OS, 8.9 months) compared with the overall study population (median OS, 10.6 months; Table 1).

TMB was significantly associated with ORR, PFS, and OS in both the pembrolizumab monotherapy and the pembrolizumab-chemotherapy treatment groups (one-sided P < 0.05), but not in the chemotherapy group (two-sided P > 0.05; Table 2). The area under the receiver operating characteristics curve (AUROC; 95% CI) for TMB discriminating objective response was 0.82 (0.69–0.95), 0.58 (0.47–0.69), and 0.44 (0.33–0.56) in the pembrolizumab, pembrolizumab-chemotherapy, and chemotherapy treatment groups, respectively (Table 2,³).

The distribution of TMB by response status and treatment group is shown in Supplementary Fig. S1, with MSI-H status indicated. In general, most patients in this analysis with MSI-H tumors had TMB ≥10 mut/Mb (22 of 25 patients; 88%; Supplementary Table S1; Supplementary Fig. S1). Across treatment groups, 22 of 50 patients (44%) who had TMB ≥10 mut/Mb also had MSI-H tumors. However, among patients with TMB <10 mut/Mb (n = 256), only three (1%) had MSI-H tumors.

Using the prespecified FoundationOne CDx–based TMB cutoff of 10 mut/Mb, the overall prevalence of TMB-H was 16%. For pembrolizumab monotherapy compared with chemotherapy, patients in the TMB ≥10 mut/Mb subgroup (n = 35) had greater ORR (55.6% vs. 41.2%), PFS (median, 11.1 vs. 7.0 months; HR, 0.52; 95% CI, 0.24–1.13), and OS (median, 31.6 vs. 13.4 months; HR, 0.34; 95% CI, 0.14–0.82) benefit than did patients in the TMB <10 mut/Mb subgroup [n = 171; ORR, 6.7% vs. 47.6%; median PFS, 2.6 vs. 7.1 months (HR, 1.73; 95% CI, 1.26–2.38); median OS, 7.5 vs. 12.6 months (HR, 1.41; 95% CI, 1.02–1.95); Fig. 2].

When pembrolizumab-chemotherapy was compared with chemotherapy, patients in the TMB ≥10 mut/Mb subgroup (n = 32) had greater ORR (73.3% vs. 41.2%), PFS (median, 31.1 vs. 7.0 months; HR, 0.62; 95% CI, 0.39–0.97), and OS (median, 31.1 vs. 13.4 months; HR, 0.54; 95% CI, 0.33–0.89) benefit than did patients in the TMB <10 mut/Mb subgroup [n = 167; ORR, 45.9% vs. 47.6%; median PFS, 6.9 vs. 7.1 months (HR, 0.97; 95% CI, 0.82–1.14); median OS, 11.4 vs. 12.6 months (HR, 1.01; 95% CI, 0.86–1.20); Fig. 3]. The analyses reported in Figures 2 and 3 were exploratory in nature and not powered to make claims of statistical significance with P values; therefore, P values are not reported.

After adjustment for PD-L1 CPS, TMB remained associated with ORR, PFS, and OS following treatment with pembrolizumab and pembrolizumab-chemotherapy (Table 3). TMB and CPS demonstrated a weak correlation, with an overall r = 0.23 among all patients (Fig. 4).

After patients with MSI-H tumors were excluded (Supplementary Tables S2 and S3), the positive association between TMB and objective response remained in the pembrolizumab monotherapy group, with statistical significance (one-sided P = 0.001); however, associations with PFS and OS were no longer significant (one-sided P > 0.05; Supplementary Table S2). The association between TMB and improved efficacy (ORR, PFS, and OS) in the pembrolizumab-chemotherapy group was attenuated after the exclusion of patients with MSI-H tumors (one-sided P > 0.05; Supplementary Table S2).

This exploratory analysis from the KEYNOTE-062 trial demonstrates evidence of an association between TMB and clinical outcomes in patients receiving pembrolizumab monotherapy and pembrolizumab-chemotherapy as first-line treatment for advanced G/GEJ adenocarcinoma. The results of this analysis also highlight the lack of association between TMB and clinical outcomes with chemotherapy alone. Our findings also suggest that there may be an OS benefit in pembrolizumab-treated patients with TMB ≥10 mut/Mb. The magnitude of benefit versus chemotherapy was higher in patients receiving pembrolizumab monotherapy than in those receiving pembrolizumab-chemotherapy. Furthermore, TMB remained associated with clinical outcomes after adjustment for PD-L1 CPS, suggesting that TMB is an independent predictor of clinical outcome beyond PD-L1 status in advanced G/GEJ adenocarcinoma. In this analysis, 28 of 50 patients with TMB ≥10 mut/Mb tumors were also classified as non–MSI-H. When the analysis was limited to the non–MSI-H subgroup, both the positive association between clinical outcomes with pembrolizumab or pembrolizumab-chemotherapy and TMB as a continuous variable and the clinical utility of pembrolizumab (with or without chemotherapy) versus chemotherapy by TMB cutoff was attenuated.

The median number of somatic mutations differs across tumor types, and TMB thresholds are studied in the context of response to immunotherapies. To establish the clinical utility of TMB for individual tumor types, standardizing the TMB cutoff for individual tumors may be desirable (19). However, owing to the variety of available panel types, numerous genetic analysis methods, and the different distribution of TMB for various tumors, setting different TMB cutoffs for varying tumors is a theoretically desirable but very difficult process to implement in practice. In the current study, we used the clinically validated TMB cutoff that is part of the FDA approval for use of pembrolizumab in patients with TMB-H solid tumors. On the basis of the KEYNOTE-158 study, FoundationOne CDx, using a TMB cutoff of 10 mut/Mb, is the FDA-approved companion diagnostic for pembrolizumab, irrespective of solid tumor type (16).

The findings from our exploratory analysis are consistent with those reported from the KEYNOTE-061 study of second-line therapy for G/GEJ adenocarcinoma (15). In the analysis of KEYNOTE-061, tissue TMB was assessed with FoundationOne CDx with the same TMB cutoff (≥10 mut/Mb); tissue TMB was significantly associated with clinical outcomes in patients treated with pembrolizumab but not in patients treated with paclitaxel (15). Similarly, the association between tissue TMB and clinical efficacy was maintained after adjustment for PD-L1 CPS (15). Taken together, the findings of our exploratory analysis suggest that TMB-H status has clinical utility in predicting response to pembrolizumab in patients with G/GEJ adenocarcinoma, as has been shown with other tumor types (16).

A few limitations should be considered when interpreting the results of this analysis. First, the TMB-evaluable population represented only 40% of the overall KEYNOTE-062 study population, and an imbalance in the proportion of patients from Asia was observed between the two populations. Second, OS for the pembrolizumab monotherapy group differed between the TMB-evaluable population and the overall study population. Third, although a trend favoring pembrolizumab continued to be evident in the TMB ≥10 mut/Mb subgroup, when MSI-H tumors were excluded, this effect was lessened, suggesting that the clinical utility of TMB ≥10 mut/Mb is attenuated in non–MSI-H tumors. Finally, exploration of dual biomarkers is necessary but was limited by the small sample sizes of subgroups in this analysis.

In addition to PD-L1 expression, MSI status, and TMB status, many studies have evaluated other predictive biomarkers for immunotherapy in G/GEJ cancer. One study has suggested that EBV is a predictive biomarker for the efficacy of immune checkpoint inhibitors (20). However, other studies have reported no such trend (21–23). Further, analysis of patient samples from KEYNOTE-061, which used whole-exome sequencing to determine EBV status, showed no differential impact on pembrolizumab response (24). Therefore, whether EBV positivity is a predictive biomarker for immune checkpoint inhibitors remains controversial. Other suggested biomarkers include gene expression profiles such as the 6-gene IFNγ signature (25) or the 18-gene T-cell–inflamed signature (26), quantification of the tumor microenvironment (27), and alterations in DNA damage response and repair genes (28). The role of these biomarkers is not yet confirmative, and more related studies are warranted.

In conclusion, results from this prespecified exploratory analysis from the randomized phase III KEYNOTE-062 trial suggest that associations exist between TMB (MSI-H included) and clinical outcomes with first-line pembrolizumab monotherapy and, potentially, with pembrolizumab-chemotherapy, in patients with advanced G/GEJ cancer. Because of the exploratory nature of this analysis, additional prospectively planned analyses of larger TMB-evaluable data sets are needed to confirm these findings.

K.W. Lee reports grants from MSD (to institution for conducting clinical trials) during the conduct of the study; grants from AstraZeneca, Ono Pharmaceutical, Merck KGaA, Pfizer, BeiGene, Astellas Pharma, Zymeworks, ALX Oncology, Macrogenics, Five Prime Therapeutics, Seagen, Bolt Therapeutics, Trishula Therapeutics, Oncologie, Pharmacyclics, LSK BioPharma, MedPacto, Green Cross Corporation, ABL Bio, Y-BIOLOGICS, Genexine, Daiichi Sankyo, Taiho Pharmaceutical, InventisBio, and Leap Therapeutics (to institution for conducting clinical trials); and personal fees from ISU Abxis, Bayer, Daiichi Sankyo, MSD, Bristol Myers Squibb, Vifor Pharma (consultation), Ono Pharmaceutical, and Boryung (honorarium) outside the submitted work. E. Van Cutsem reports grants from Amgen, Bayer, Boehringer Ingelheim, Bristol Myers Squibb, Celgene, Ipsen, Lilly, MSD, Merck KGaA, Novartis, Roche, and Servier paid to institution and personal fees from participation on advisory boards for AbbVie, Array, Astellas Pharma, AstraZeneca, Bayer, BeiGene, Biocartis, Boehringer Ingelheim, Bristol Myers Squibb, Celgene, Daiichi, Halozyme, GSK, Helsinn, Incyte, Ipsen, Janssen Research, Lilly, MSD, Merck KGaA, Mirati, Novartis, Pierre Fabre, Roche, Seattle Genetics, Servier, Sirtex, Terumo, Taiho Pharmaceutical, TRIGR, and Zymeworks outside the submitted work. Y.J. Bang reports grants from MSD during the conduct of the study; personal fees from MSD, Merck Serano, Daiichi Sankyo, Astellas Pharma, BeiGene, Samyang Biopharm, Alexo Oncol, Hanmi Pharmaceutical, Daewoong Pharmaceutical, and Amgen and grants from Genentech/Roche, MSD, Merck Serano, Daiichi Sankyo, Astellas Pharma, BeiGene, and Amgen outside the submitted work. C.S. Fuchs reports personal fees from Amylin Pharmaceuticals, AstraZeneca, Bain Capital, CytomX Therapeutics, Daiichi Sankyo, Eli Lilly, Entrinsic Health, EvolveImmune Therapeutics, Genentech, Merck, Taiho Pharmaceutical, and Unum Therapeutics outside the submitted work. C.S. Fuchs served as a Director for CytomX Therapeutics and owns unexercised stock options for CytomX and Entrinsic Health. C.S. Fuchs is a cofounder of EvolveImmune Therapeutics and has equity in this private company. C.S. Fuchs provided expert testimony for Amylin Pharmaceuticals and Eli Lilly. C.S. Fuchs is an employee of Genentech and Roche. H.C. Chung reports grants from MSD during the conduct of the study; grants from Lilly, GSK, Merck Serono, BMS-Ono, Taiho Pharmaceutical, Amgen, BeiGene, Incyte, and Zymeworks and personal fees from Taiho Pharmaceutical, Celltrion, MSD, Lilly, BMS-Ono, Merck Serono, Gloria, BeiGene, Amgen, and Zymeworks outside the submitted work. J. Lee reports grants from Merck, AstraZeneca, and Lilly during the conduct of the study; and grants from OncXerna outside the submitted work. H.R. Castro reports grants from MSD during the conduct of the study. J. Chao reports nonfinancial support and other support from Merck during the conduct of the study; grants and personal fees from Merck; personal fees from Lilly, AstraZeneca, Foundation Medicine, Daiichi Sankyo, Amgen, Bristol Myers Squibb, Astellas Pharma, Turning Point Therapeutics, Roche, Silverback Therapeutics, Novartis, Coherus BioSciences, and Geneos outside the submitted work. Z.A. Wainberg reports other support from AbbVie, Amgen, Bayer, AstraZeneca, Daiichi, Roche, Merck, Novartis, Seagen, BMS, Ipsen, and Incyte outside the submitted work. Z.A. Cao reports other support from Merck & Co., Inc., Rahway, NJ, other support from Merck & Co., Inc., Rahway, NJ, during the conduct of the study; and other support from BMS outside the submitted work. D. Aurora-Garg reports other support from Merck & Co., Inc., Rahway, NJ, during the conduct of the study and other support from Merck & Co., Inc., Rahway, NJ, outside the submitted work. J. Kobie reports other support from Merck & Co., Inc., Rahway, NJ, outside the submitted work. R. Cristescu reports other support from Merck & Co., Inc., Rahway, NJ, during the conduct of the study. P. Bhagia reports other support from/employed by Merck & Co., Inc., Rahway, NJ, during the conduct of the study and other support from/employed by Merck & Co., Inc., Rahway, NJ, outside the submitted work. S. Shah reports employment at Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ. J. Tabernero reports personal fees from Array Biopharma, AstraZeneca, Avvinity Therapeutics, Bayer, Boehringer Ingelheim, Chugai, Daiichi Sankyo, F. Hoffmann-La Roche Ltd., Genentech Inc., HalioDx SAS, Hutchison MediPharma International, Ikena Oncology, Inspirna Inc., IQVIA, Lilly, Menarini, Merck Serono, Merus, MSD, Mirati, NeoPhore, Novartis, Ona Therapeutics, Orion Biotechnology, Peptomyc, Pfizer, Pierre Fabre, Samsung Bioepis, Sanofi, Seattle Genetics, Scandion Oncology, Servier, Sotio Biotech, Taiho Pharmaceutical, Tessa Therapeutics, TheraMyc, Imedex, Medscape Education, MJH Life Sciences, PeerView Institute for Medical Education, and Physicians Education Resource (PER) outside the submitted work. K. Shitara reports grants from Astellas Pharma, Chugai Pharma, Medi Science, and Eisai; personal fees from Eli Lilly and Company, Bristol Myers Squibb, Takeda Pharmaceuticals, Pfizer Inc., Novartis, AbbVie Inc., GSK, Boehringer Ingelheim and Janssen; and grants and personal fees from Ono Pharmaceutical, Daiichi Sankyo, Taiho Pharmaceutical, Merck Pharmaceutical, and Amgen outside the submitted work. L. Wyrwicz reports personal fees and other support from MSD during the conduct of the study; and personal fees from BMS, BeiGene, AstraZeneca, and Roche outside the submitted work. No disclosures were reported by the other authors.

K.W. Lee: Data curation, formal analysis, investigation, writing–original draft, writing–review and editing, approved final manuscript. E. Van Cutsem: Data curation, validation, writing–review and editing, approved the final manuscript. Y.J. Bang: Conceptualization, data curation, formal analysis, investigation, writing–review and editing, approved final manuscript. C.S. Fuchs: Conceptualization, data curation, formal analysis, validation, investigation, writing–review and editing, approved final manuscript. I. Kudaba: Data curation, writing–review and editing, approved final manuscript. M. Garrido: Data curation, formal analysis, validation, writing–review and editing, approved final manuscript. H.C. Chung: Data curation, validation, writing–original draft, writing–review and editing, approved final manuscript. J. Lee: Data curation, writing–review and editing, approved final manuscript. H.R. Castro: Data curation, writing–review and editing, approved final manuscript. J. Chao: Data curation, writing–review and editing, approved final manuscript. Z.A. Wainberg: Data curation, formal analysis, validation, writing–review and editing, approved final manuscript. Z.A. Cao: Conceptualization, validation, investigation, writing–review and editing, approved final manuscript. D. Aurora-Garg: Data curation, formal analysis, writing–original draft, approved final manuscript. J. Kobie: Formal analysis, validation, writing–original draft, writing–review and editing, approved final manuscript. R. Cristescu: Data curation, formal analysis, validation, writing–original draft, writing–review and editing, approved final manuscript. P. Bhagia: Validation, writing–review and editing, approved final manuscript. S. Shah: Conceptualization, validation, investigation, writing–review and editing, approved final manuscript. J. Tabernero: Conceptualization, data curation, formal analysis, validation, investigation, writing–review and editing, approved final manuscript. K. Shitara: Conceptualization, data curation, validation, investigation, writing–original draft, writing–review and editing, approved final manuscript. L. Wyrwicz: Conceptualization, formal analysis, validation, investigation, writing–review and editing, approved final manuscript.

This work was supported by Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ. The funder participated in the study design, data analysis and interpretation, and manuscript writing, and maintained the study database. All authors had full access to the data and had final responsibility for the decision to submit for publication. The authors thank the patients and their families and caregivers as well as the primary investigators and site personnel for participating in the study. Medical writing and/or editorial assistance was provided by Melanie Sweetlove, MSc, Maxwell Chang, and Holly C. Cappelli, PhD, CMPP, of ApotheCom. This assistance was funded by Merck Sharp & Dohme LLC, a subsidiary of Merck & Co., Inc., Rahway, NJ.

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.