Pembrolizumab alone or in combination with chemotherapy as first-line therapy for patients with advanced gastric or gastroesophageal junction adenocarcinoma: results from the phase II nonrandomized KEYNOTE-059 study

Background The multicohort, phase II, nonrandomized KEYNOTE-059 study evaluated pembrolizumab ± chemotherapy in advanced gastric/gastroesophageal junction cancer. Results from cohorts 2 and 3, evaluating first-line therapy, are presented. Methods Patients ≥ 18 years old had previously untreated recurrent or metastatic gastric/gastroesophageal junction adenocarcinoma. Cohort 3 (monotherapy) had programmed death receptor 1 combined positive score ≥ 1. Cohort 2 (combination therapy) received pembrolizumab 200 mg on day 1, cisplatin 80 mg/m2 on day 1 (up to 6 cycles), and 5-fluorouracil 800 mg/m2 on days 1–5 of each 3-week cycle (or capecitabine 1000 mg/m2 twice daily in Japan). Primary end points were safety (combination therapy) and objective response rate per Response Evaluation Criteria in Solid Tumors version 1.1 by central review, and safety (monotherapy). Results In the combination therapy and monotherapy cohorts, 25 and 31 patients were enrolled; median follow-up was 13.8 months (range 1.8–24.1) and 17.5 months (range 1.7–20.7), respectively. In the combination therapy cohort, grade 3/4 treatment-related adverse events occurred in 19 patients (76.0%); none were fatal. In the monotherapy cohort, grade 3–5 treatment-related adverse events occurred in seven patients (22.6%); one death was attributed to a treatment-related adverse event (pneumonitis). The objective response rate was 60.0% [95% confidence interval (CI), 38.7–78.9] (combination therapy) and 25.8% (95% CI 11.9–44.6) (monotherapy). Conclusions Pembrolizumab demonstrated antitumor activity and was well tolerated as monotherapy and in combination with chemotherapy in patients with previously untreated advanced gastric/gastroesophageal junction adenocarcinoma. Clinical Trial ClinicalTrials.gov NCT02335411 Electronic supplementary material The online version of this article (10.1007/s10120-018-00909-5) contains supplementary material, which is available to authorized users.

Tumor cells frequently use the programmed death receptor 1 (PD-1) pathway to evade immune surveillance [13,14]. Binding of PD-1 to its ligands, PD-L1 and PD-L2, inhibits effector T-cell function, resulting in suppression of antitumor immune response [13,14]. PD-L1 is frequently upregulated in gastric cancer [15][16][17][18][19]. Pembrolizumab is a selective, humanized, immunoglobulin G4κ monoclonal antibody that blocks interactions between PD-1 and its ligands [20]. The phase II KEYNOTE-059 clinical trial, a multicohort study designed to evaluate pembrolizumab alone or in combination with chemotherapy in patients with recurrent or metastatic G/GEJ adenocarcinoma, included a cohort of patients whose disease had progressed while they previously received at least two systemic therapies for advanced disease (cohort 1; pembrolizumab monotherapy) and two cohorts of patients who had not previously received systemic therapy for advanced disease. Data from cohort 1 have already been published and demonstrate durable responses with pembrolizumab monotherapy, with higher ORR and longer response durations in patients with PD-L1-positive tumors [combined positive score (CPS) ≥ 1] than with PD-L1-negative tumors [21]. Based on these results, the US Food and Drug Administration approved pembrolizumab for the treatment of patients with recurrent or metastatic G/GEJ adenocarcinoma whose tumors express PD-L1 and whose disease has progressed on or after ≥ 2 prior lines of therapy [20]. In the phase III, KEYNOTE-061 study, pembrolizumab did not significantly improve OS, versus paclitaxel, as second-line therapy in patients with advanced G/GEJ cancer and PD-L1 CPS ≥ 1 [22].
Chemotherapeutic agents such as cisplatin and 5-fluorouracil enhance immunogenicity of cancer cells and increase susceptibility to immune-mediated cytotoxicity [23][24][25]. Combining pembrolizumab with chemotherapy has demonstrated efficacy and manageable safety in other cancers [26,27]. Cohorts 2 and 3 of the KEYNOTE-059 study were designed to further explore pembrolizumab in combination therapy and as monotherapy in the PD-L1-positive population. Herein, we present results from cohorts 2 and 3 of the KEYNOTE-059 study, which evaluated first-line therapy with pembrolizumab plus chemotherapy (cohort 2) or alone (cohort 3) in advanced G/GEJ cancer.

Materials and methods
KEYNOTE-059 is a multicenter, open-label, nonrandomized, 3-cohort, phase II study. Results for cohorts 2 and 3 are presented; results for cohort 1 were published separately [21]. The study was conducted in accordance with Good Clinical Practice guidelines and the Declaration of Helsinki. The protocol was approved by institutional review boards and independent ethics committees for each institution. All patients provided written, informed consent. This trial was registered with ClinicalTrials.gov, number NCT02335411.

Patients
Eligibility criteria for both cohorts were age ≥ 18 years, histologically or cytologically confirmed recurrent or metastatic G/GEJ adenocarcinoma, no prior systemic therapy for advanced disease, measurable disease per Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST v1.1) by central review, HER2/neu-negative status, Eastern Cooperative Oncology Group performance status 0/1, life expectancy ≥ 3 months, and adequate organ function.
For cohort 2, patients had to provide new or archival tumor tissue for PD-L1 analysis. For cohort 3, patients had to have PD-L1-positive tumors as per the PD-L1 IHC 22C3 pharmDx assay (Agilent Technologies, Carpinteria, CA, USA). PD-L1 positivity was defined as CPS ≥ 1, where CPS is the number of PD-L1-positive cells (tumor cells, lymphocytes, and macrophages) divided by the total number of tumor cells multiplied by 100 [21]. Exclusion criteria are presented in Online Resource Table S1. We assessed DNA mismatch repair across 5 mononucleotide repeat markers (NR21, NR24, BAT25, BAT26, MONO27) using DNA extracted from formalin-fixed, paraffin-embedded tumor samples and blood (normal control) using the MSI Analysis System, version 1.2 (Promega). In microsatellite instabilityhigh tumors, 2 or more markers were changed, compared with normal controls.

Study treatment
In cohort 2, patients received pembrolizumab 200 mg by intravenous infusion over 30 min on day 1, cisplatin 80 mg/ m 2 on day 1 (for up to 6 cycles), and 5-fluorouracil 800 mg/ m 2 by continuous infusion on days 1-5 of each 21-day cycle. In Japan only, patients received capecitabine 1000 mg/m 2 twice daily orally on days 1-14 of each 21-day cycle instead of 5-fluorouracil. Up to two dose reductions were allowed per chemotherapy agent (Online Resource Table S2). In cohort 3, patients received pembrolizumab monotherapy 200 mg by intravenous infusion over 30 min on day 1 of each 21-day cycle.
Pembrolizumab dose reduction was not permitted. Treatment was continued for up to 35 cycles (~ 2 years) or until confirmed disease progression, unacceptable toxicity, complete response, or patient or investigator decision to withdraw. Clinically stable patients with radiographic progressive disease (PD) could continue treatment at the investigator's discretion until confirmed PD. Patients in cohort 2 who discontinued chemotherapy could continue pembrolizumab if they had not met criteria for discontinuation.

Outcomes and assessments
For cohort 2, the primary end points were safety and tolerability. Secondary end points included ORR (percentage of patients who experienced complete or partial response) per RECIST v1.1 by central imaging review and evaluation of the relationship between PD-L1 expression and efficacy. For cohort 3, primary end points were ORR and safety and tolerability. Secondary end points included disease control rate (DCR; percentage of patients who experienced complete or partial response or stable disease for ≥ 6 months), duration of response, progression-free survival (PFS; time from study treatment initiation to first documented instance of PD or death) per RECIST v1.1 by central imaging review, and OS (time from initiation of study treatment to death from any cause).
Tumors were assessed by computed tomography or magnetic resonance imaging at 9 weeks after study treatment initiation, every 6 weeks for year 1, and every 9 weeks thereafter (or more often if clinically indicated). Response was confirmed by repeat radiographic assessment ≥ 4 weeks after first documentation. Patients were followed up every 12 weeks for survival status until death, withdrawal of consent, or study end.
Patients were monitored for adverse events (AEs) until 30 days after the last dose of treatment (90 days for serious AEs and events of clinical interest). AEs were graded using National Cancer Institute Common Terminology Criteria for Adverse Events, version 4.0; investigators determined treatment attribution.

Statistical analysis
For cohort 2, the sample size was based on clinical experience. A sample size of ~ 18 patients was planned to provide the reasonable two-sided 95% confidence interval (CI) of AE incidence (Online Resource Table S3). For cohort 3, a sample size of ~ 25 patients with PD-L1 CPS ≥ 1 was estimated using Bayesian sample size methods assuming a uniform prior for the response rate and a true ORR of < 40% (ORR based on standard of care). Safety and efficacy were analyzed in all patients who received ≥ 1 dose of study treatment. Safety was analyzed using descriptive statistics. Analyses were conducted using SAS statistical software version 9.3 (SAS Institute, Inc). ORR and DCR with 95% CI were calculated using the exact method based on binomial distribution (Clopper-Pearson method [28]). Patients with missing data were considered nonresponders. Time-to-event end points were estimated using the Kaplan-Meier method, with censoring at the last date of assessment for patients with missing data.

Results
The study was conducted at 57 sites in 17 countries. For cohort 2, 25 patients were enrolled at 10 sites in the Republic of Korea (South Korea), Japan, Israel, the United States, and France between March 2, 2015, and December 14, 2015. For cohort 3, 31 patients were enrolled at 18 sites in Canada, Chile, Israel, Japan, South Korea, and the United States between July 23, 2015, and December 2, 2015. Patient characteristics are presented in Table 1.
Of 28 patients with measurable disease at baseline who had ≥ 1 evaluable postbaseline assessment, 24 (85.7%) experienced reduction in target lesion size from baseline (Fig. 1c). Decrease in tumor burden was generally maintained over several assessments (Fig. 1d; Online Resource Figure S2).

Discussion
Cohort 1 of the KEYNOTE-059 study confirmed that pembrolizumab monotherapy provided durable responses with manageable safety in patients with advanced G/GEJ adenocarcinoma that had progressed following second-line treatment [21]. Although patient numbers in the current analyses were small, results from cohorts 2 and 3 of the KEYNOTE-059 study demonstrated manageable safety and antitumor activity for pembrolizumab as monotherapy and as combination therapy with cisplatin plus 5-fluorouracil or capecitabine in previously untreated advanced G/GEJ adenocarcinoma. Although it is difficult to draw definitive conclusions regarding the clinical impact of pembrolizumab based solely on these data, the results, along with those from cohort 1 of the study [21], support the growing body of evidence for the potential use of pembrolizumab in patients with G/GEJ adenocarcinoma. Additional large-scale clinical trials are needed to confirm efficacy and safety in these patients.
In cohort 2, the incidence and severity of TRAEs with pembrolizumab plus cisplatin plus 5-fluorouracil or capecitabine were generally consistent with those of known toxic effects of cisplatin plus fluoropyrimidine [7,9,10,12]. The incidence of grades 3 and 4 chemotherapyrelated neutropenia (48% and 16%, respectively) in this cohort was somewhat higher than expected; previous studies of first-line chemotherapy in patients with advanced gastric cancer have reported grade ≥ 3 neutropenia occurring at rates of 20-44% [29][30][31], although these data were collected from larger study populations that may not be directly comparable. It is notable that neutropenia levels were not elevated in either cohort 3 or cohort 1 [21], but, with a population of just 25 patients, it is difficult to draw definitive conclusions as to the cause of the slightly greater incidence in cohort 2. The incidence of pembrolizumabrelated stomatitis (12% overall; 8% grade 3) also initially appeared to be higher than expected, but this was likely an artifact given the small number of patients involved (3 of 25). Although a similar incidence of pembrolizumabrelated stomatitis (2 of 23; 9%) was reported in a recently published study of patients with advanced colorectal carcinoma [32], the risk associated with pembrolizumab appears to be small compared with the number of patients in whom chemotherapy-related stomatitis develops (52% in the current study; 16% grade ≥ 3). Immune-mediated AEs were similar to those reported with pembrolizumab plus platinum-based doublets in non-small cell lung cancer [26], suggesting a degree of constancy across nonhematologic malignancies. In cohort 3, the safety profile of pembrolizumab monotherapy was consistent with that previously reported [21,33]. No new safety signals were identified. Relative to standard chemotherapy, pembrolizumab had lower rates of grade 3/4 AEs and discontinuation because of AEs, suggesting potential for a greater therapeutic index with pembrolizumab than with available therapies [7][8][9][10][11][12] and the possibility of combining with other therapies.
Sixty-four percent of patients in cohort 2 were PD-L1 positive (CPS ≥ 1), compared with an overall prevalence of PD-L1 positivity in advanced gastric cancer of 30-65%, although different methods, antibodies, and cutoff values were used across studies [15][16][17][18][19]. Analysis of patients with PD-L1 CPS ≥ 1 in cohort 2 suggests that PD-L1 expression enhances response to pembrolizumab plus chemotherapy; however, larger prospective clinical trials are necessary to substantiate these results.
One patient in cohort 3 who had microsatellite instabilityhigh tumors experienced a partial response. Although this partial response contributed to the ORR in cohort 3, it does not account for the ORR in its entirety. This finding is consistent with that reported for pembrolizumab monotherapy in the cohort of patients who have received ≥ 2 prior therapies for G/GEJ cancer in this KEYNOTE-059 study [21].
Overall, pembrolizumab plus cisplatin and 5-fluorouracil or capecitabine demonstrated manageable safety and promising antitumor activity as first-line therapy in advanced G/GEJ adenocarcinoma, regardless of PD-L1 expression. Pembrolizumab monotherapy also demonstrated encouraging antitumor activity and acceptable safety in patients with PD-L1 CPS ≥ 1. Results support continued evaluation of pembrolizumab in clinical trials.

Compliance with ethical standards
Ethics standards All procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1964 and later versions. Informed consent to be included in the study, or the equivalent, was obtained from all patients.