Phase II Study of the Liposomal Formulation of Eribulin (E7389-LF) in Combination with Nivolumab: Results from the Gastric Cancer Cohort

In 2020, there were approximately 1.1 million new cases of gastric cancer worldwide, and 770,000 deaths due to gastric cancer (1). Because of a lack of efficacy and long-term survival seen in previously studied therapies, new therapies for pretreated advanced gastric cancer are warranted. Trastuzumab deruxtecan showed promising efficacy in patients with previously treated HER2-positive gastric cancer (2), but it is approved for patients who have received a prior trastuzumab-based regimen (3).

Eribulin mesylate (eribulin) is a microtubule dynamics inhibitor that is approved in more than 80 countries worldwide, including Japan, the United States, and countries in Europe and Asia for the treatment of patients with breast cancer or sarcoma (4). E7389-LF is a new formulation that uses liposomes to encapsulate eribulin, which is anticipated to improve eribulin concentration in tumor tissues. Preclinical pharmacokinetic studies found that plasma AUC of eribulin by E7389-LF administration was more than 600-fold higher than that of eribulin by aqueous formulation (5). Moreover, in vivo studies demonstrated that antitumor activity was stronger with E7389-LF than with eribulin, with a more favorable pharmacokinetic profile (6). This model also suggested that E7389-LF may be involved in altering the tumor microenvironment. Preclinical safety assessments determined that E7389-LF had similar toxicity to eribulin, possibly as a result of the pharmacokinetic profile (7). In a phase I study of E7389-LF in patients with solid tumors, partial responses (PR) were observed in 10.3% of patients (six of 58), and no new or unexpected toxicities were observed compared with eribulin (8). It was determined that the pharmacokinetic profile of E7389-LF was superior to eribulin and that the dose of 2.0 mg/m² (as free eribulin) every 3 weeks should be considered for several types of solid tumors, including gastric cancer (9).

A preclinical study examined E7389-LF in combination with anti-programmed cell death-1 (PD-1) treatment and found that the combination showed enhanced antitumor activity relative to each monotherapy. This potential synergistic effect may be due to cytotoxic actions and modulation of the tumor microenvironment (6). E7389-LF and nivolumab, a PD-1 inhibitor, have both shown efficacy as monotherapies in pretreated gastric cancer (10, 11). In a phase I study of E7389-LF in patients with advanced gastric cancer (n = 34), the objective response rate (ORR) was 17.6% [95% confidence interval (CI), 6.8–34.5; ref. 11]. All six patients who had a partial response (PR) had received prior anti–PD-1 therapy. In a phase III trial of nivolumab in patients with advanced gastric cancer or gastroesophageal junction cancer, the ORR was 11.2% (95% CI, 7.7–15.6) in patients who received nivolumab and 0% (95% CI, 0–2.8) in patients who received placebo (10).

We assessed the efficacy and safety of E7389-LF in combination with nivolumab in patients with gastric cancer in the gastric cancer cohort of the phase II part of Study 120 (NCT04078295).

This expansion cohort of the phase Ib/II open-label study of E7389-LF in combination with nivolumab was conducted in Japan and focused on patients with gastric cancer. The primary objective of the phase II part of Study 120 was to evaluate ORR in patients who received E7389-LF in combination with nivolumab. Secondary objectives included assessment of safety, progression-free survival (PFS), and pharmacokinetics. Exploratory objectives included assessment of overall survival (OS), disease control rate (DCR), and clinical benefit rate (CBR), and evaluation of potential biomarkers.

The gastric cancer cohort of Study 120 enrolled patients with unresectable gastric cancer and at least one lesion measurable by Response Evaluation Criteria in Solid Tumors version 1.1 (RECIST v1.1). Lesions previously treated with radiotherapy or locoregional therapies must have shown evidence of progressive disease to be deemed a measurable lesion. Patients had to have disease progression following treatment with a combination therapy including a platinum drug plus fluoropyrimidine (first-line therapy) and a taxane-containing regimen (second-line therapy). Patients needed to have an Eastern Cooperative Oncology Group performance status (ECOG PS) of 0 or 1. Patients who had previous treatment with any anti–PD-1, anti–PD-ligand 1 (PD-L1), anti–PD-L2, anti–CD137, or anti–CTLA-4 antibody, or any other antibody or drug specifically targeting T-cell costimulation or checkpoint, or cancer vaccine therapy that resulted in a grade ≥3 immune-related adverse event (irAE) or needed to discontinue treatment due to an irAE of any grade were excluded from the study.

This study was conducted in accordance with standard operating procedures of the sponsor, which were based on the Principles of the World Medical Association Declaration of Helsinki, all applicable Japanese Good Clinical Practices and regulations, and the Pharmaceutical Affairs Law for studies conducted in Japan. Written informed consent forms were obtained from all participants; these and the study protocol were reviewed and approved by the applicable institutional ethical review board.

In the phase Ib part of Study 120, the recommended phase II dose was E7389-LF 2.1 mg/m² (eribulin mesylate content) in combination with nivolumab 360 mg every 3 weeks, administered as intravenous infusions (9). E7389-LF could be dose-reduced in consecutive steps to 1.7, 1.1, and 0.8 mg/m². Dose reductions of nivolumab were not permitted. Tumor responses were assessed every 6 weeks by the investigators per RECIST v1.1; best overall responses (BOR) of complete response (CR) or PR required confirmation by a subsequent assessment at least 28 days following the initial assessment. Patients were permitted to continue treatment beyond disease progression as long as investigator-assessed clinical benefit was observed and the patient was tolerating the study drugs. Figure 1A and Supplementary Fig. S1 include data from beyond disease progression. Post hoc efficacy assessments were also conducted for several subgroups: PD-L1 combined positive score (CPS) status, prior gastrectomy, liver metastases at baseline, and prior immune checkpoint inhibitor (ICI) treatment. Safety assessments consisted of monitoring and recording all adverse events (AE), and grading was conducted according to the National Cancer Institute Common Terminology Criteria for Adverse Events version 5.0 (RRID:SCR_010296).

For the biomarker analyses, PD-L1 expression (by an approved IHC 28–8 pharmDx assay, Agilent Dako; Agilent Technologies, RRID: SCR_013575) and PD-L1 CPS (percent, based on PD-L1 expression in viable tumor cells and immune cells) were assessed in tumor samples. Plasma samples were collected predose on cycle 1 day 1 (C1D1), day 8, day 15 of cycles 1 and 2, day 1 of cycles 3 to 9, and off-treatment, and biomarkers were investigated with AngiogenesisMAP, Multiplex, and Simoa systems. For patients who underwent both screening and cycle 2 day 1 biopsies, IHC was conducted on tumor samples for PECAM1 (CD31) and panCK/CD8 to categorize immune phenotypes and to measure microvessel density and number of CD8-positive tumor-infiltrating lymphocytes.

Immune phenotypes were assessed using methods modified from those previously used in the ABACUS trial (12, 13). Phenotypes were determined by a pathologist using a density proportion score (DPS) for panCK-CD8 (9). The DPS is based on the density of inflammatory cells within a tumor and can be used to score areas that range from those with minimal inflammatory cells to those with a higher percentage of inflammatory cells; scores were assessed in two compartments or “nests,” the tumor stroma and tumor epithelial nests. Patients in which ≥20% of tumor epithelial nests had a moderate to high density of infiltrating CD8 cells were considered “immune-inflamed,” and if patients had a lower DPS in tumor epithelial nests, they were considered “immune-excluded” if ≥20% of the tumor stroma had a moderate to high density of infiltrating CD8 cells. If patients had no or only low densities of infiltrating CD8 cells in both tumor epithelial nests and tumor stroma, they were categorized as “immune-desert.”

All statistical analyses were performed and plots were generated using SAS v9.4 (SAS Institute Inc.). The safety analysis set included all patients who received at least one dose of E7389-LF or nivolumab; efficacy analyses were also conducted on the safety analysis set. The primary objective of the phase II part was to assess ORR, with success defined as Bayesian posterior probability over 85% beyond a clinically meaningful value (15%) for ORR (i.e., seven or more responders of planned 30 patients enrolled). A clinically meaningful value for ORR was set based on the ORR of nivolumab monotherapy (14.0%; 95% CI, 8.5–21.2) in Japanese patients who received nivolumab as third-line or later therapy for gastric cancer in the ATTRACTION-2 trial (14). BOR were summarized and ORR, DCR, and CBR were calculated with 95% CI by cohort; two-sided exact 95% CI were calculated based on the Clopper–Pearson method. ORR was the proportion of patients who had BOR of CR or PR. DCR was the proportion of patients who had BOR of CR or PR or stable disease (SD, ≥5 weeks after C1D1). CBR was the proportion of patients who had BOR of CR or PR or durable SD (≥23 weeks of SD).

For PFS and OS, medians were estimated by the Kaplan–Meier method, and 95% CI were calculated based on the Greenwood formula. Estimates for survival follow-up time were calculated in the same way as the Kaplan–Meier estimates of OS but with the meaning of “censor” and “event” status indicator reversed. Summary statistics of neutrophil count from C1D1 to C2D1 were calculated.

The number and percentage of patients with all AE and serious AE observed after first dosing were listed and summarized.

The effect of E7389-LF with nivolumab combination therapy on soluble and tumor tissue biomarkers was assessed. Plasma biomarkers were analyzed in biomarkers in which ≥75% of the samples demonstrated levels above the lower limit of quantification. Percent changes from C1D1 were assessed using the one-sample Wilcoxon signed-rank test. For the percent change in biomarkers analyses using all markers, P values were adjusted using the Benjamini–Hochberg procedure for FDR control, with the number of biomarkers analyzed at each time point. For microvessel density and CD8-positive cell density, percent changes from screening to C2D1 were assessed using the one-sample Wilcoxon signed-rank test. Dichotomized analyses (cutoff at 33% percentiles, patients with plasma values <33% were categorized as “low” and patients with plasma values ≥33% were categorized as “high”) of plasma levels of each plasma biomarker at each time point and PFS were performed using univariate Cox regression and log-rank test to investigate possible prognostic biomarkers for PFS.

Eisai Inc. commits to sharing data from clinical trials upon request from qualified scientific and medical researchers. Data requests are reviewed and authorized by an independent review panel on the basis of scientific merit, and data are anonymized with respect to applicable laws and regulations. Requests for data that have not yet been uploaded to the database can be made by choosing "Click here to submit an Enquiry" at: https://www.clinicalstudydatarequest.com/SearchAllPostings.aspx.

Among the 31 patients with gastric cancer who entered the treatment phase, 18 patients were male and 13 were female (Table 1). The median age was 63.0 years, and 12 patients (38.7%) had a previous gastrectomy. This study population is generally representative of the population of patients with gastric cancer (Supplementary Table S1). Among all patients, 20 had not received any prior ICI, one was known to have received a prior ICI, and 10 patients had unknown prior ICI status. All patients had received at least two previous anticancer medications. Specific previous therapies received are listed in Supplementary Table S2. By the data cutoff date (May 31, 2022), 29 patients discontinued treatment. Of these, 26 patients discontinued for the primary reason of disease progression, and three patients discontinued for the primary reason of AE.

The ORR was 25.8% (95% CI, 11.9–44.6) and the DCR was 71.0% (95% CI, 52.0–85.8; Table 2). Bayesian posterior probability for ORR was 95.9%, which was beyond the planned criteria (85%) that ORR would be greater than the clinically meaningful threshold of 15%. Among patients with a CPS of <5 (n = 20), ORR was 25.0% (95% CI, 8.7−49.1); the ORR of patients with CPS ≥5 (n = 9) was 22.2% (95% CI, 2.8−60.0). In patients who had previously had a gastrectomy (n = 12), the ORR was 16.7% (95% CI, 2.1–48.4), and in patients who did not have a gastrectomy (n = 19), the ORR was 31.6% (95% CI, 12.6–56.6; Supplementary Table S3). In patients who had liver metastasis at baseline (n = 12), the ORR was 25.0% (95% CI, 5.5–57.2), and the ORR for the 19 patients who did not have liver metastasis at baseline was 26.3% (95% CI, 9.1–51.2). Patients who had not received any prior ICI (n = 20) had an ORR of 30.0% (95% CI, 11.9–54.3). Patients’ sums of diameters of target lesions from baseline over time and maximum change from baseline at nadir are shown in Fig. 1A and B. Doses received over time are shown in Supplementary Fig. S1A. PR were generally sustained through dose reductions of E7389-LF (i.e., 1.7 and 1.1 mg/m²).

The median PFS was 2.69 months (95% CI 1.91–2.99); PFS rates at 3, 6, and 9 months were also calculated (Fig. 2A). The median PFS of patients with CPS <5 was 2.60 months (95% CI, 1.18–4.93), and the median PFS of patients with CPS ≥5 was 2.79 months [95% CI, 1.18–not estimable (NE)]. The median follow-up time for OS was 11.96 months (95% CI, 10.38–14.46), and median OS was 7.85 months (95% CI, 4.47–NE); OS rates at 3, 6, and 9 months were also calculated (Fig. 2B). The median OS of patients with CPS <5 was 6.62 months (95% CI, 2.83–NE), and the median OS of patients with CPS ≥5 was 7.85 months (95% CI, 4.40–NE).

Poststudy anticancer treatments were received by 17 patients of the 29 patients (58.6%) who discontinued E7389-LF in combination with nivolumab; specific medications are listed in Supplementary Table S2. Timing and duration of subsequent treatments are shown in Supplementary Fig. S1B.

Among the 31 patients in the gastric cancer cohort, 16 (51.6%) experienced at least one treatment-related treatment-emergent AE (TEAE) that led to dose reduction of E7389-LF. The most common TEAE leading to E7389-LF dose reduction were neutropenia (29.0%), febrile neutropenia (22.6%), and leukopenia (9.7%). TEAE that led to study drug withdrawal of either E7389-LF or nivolumab occurred in four patients and included upper gastrointestinal hemorrhage (grade 3), decreased appetite (grade 2), cerebral hemorrhage (grade 5), asthma (grade 3), and pulmonary edema (grade 3). Each event occurred in a single patient. Of these events, decreased appetite and asthma were attributed to both drugs. The others were considered not related to either study drug.

The most common treatment-related TEAE of any grade were neutropenia (77.4%), leukopenia (74.2%), and decreased appetite (51.6%; Table 3). The most common treatment-related TEAE of grade ≥3 were neutropenia (71.0%), leukopenia (51.6%), and febrile neutropenia (22.6%). When examining hematologic treatment-related TEAE at the highest level (grade 3), leukopenia was observed in 12 patients (38.7%), febrile neutropenia in seven (22.6%), anemia in six (19.4%), neutropenia in four (12.9%), thrombocytopenia in three (9.7%), and lymphopenia in two (6.5%). Treatment-related hematologic events with the highest grade of 4 were also assessed: neutropenia occurred in 18 patients (58.1%), leukopenia occurred in four patients (12.9%), and thrombocytopenia occurred in one patient (3.2%). TEAE of any grade that occurred with an incidence of >10% are listed in Supplementary Table S4. Two grade 5 TEAE were observed (one case of liver disorder and one case of cerebral hemorrhage, both within 30 days after study treatment discontinuation); however, neither of these TEAE were deemed related to either study drug.

Myelosuppression was reversible, as shown by absolute neutrophil counts (ANC). ANC decreased from baseline between C1D8 and C1D15 in patients without prophylactic pegylated granulocyte colony-stimulating factor (peg-GCSF) administration but had recovered by C2D1, when the next dose of study drugs was administered. ANC decreases did not affect the next administration in the Q3W regimen (Supplementary Fig. S2).

Immune phenotypes were categorized as inflamed, excluded, and desert, based on presence and placement of CD8-positive T cells (Fig. 3A). In patients with both screening and C2D1 biopsy samples available, changes in immune phenotypes were seen by C2D1 (Fig. 3B). E7389-LF in combination with nivolumab tended to increase the presence of CD8-positive cells at C2D1 (P = 0.039; Fig. 3C) and increase microvessel density within the tumor cell region (P = 0.055; Fig. 3D). Of the 78 biomarkers tested for plasma biomarker analysis, 49 met the criterion of ≥75% of the samples with levels above the lower limit of quantification. Several key vascular and IFNγ-related biomarkers showed significant increases from baseline; although patient numbers were limited, there appeared to be a trend of higher levels of IFNγ at earlier timepoints in patients with partial responses compared with patients who did not have a partial response (Fig. 3E). Among the biomarker population (n = 8), one partial response was observed; the immune phenotype of this patient changed from immune-excluded to inflamed at C2D1 and increases in both CD8⁺ cells and microvessel density were observed.

After treatment with E7389-LF 2.1 mg/m² in combination with nivolumab 360 mg every 3 weeks, six of seven vascular markers and four IFNγ-related markers showed increases from C1D1 (Supplementary Figs. S3A–S3S3C). Specifically, for vascular markers (collagen IV, TIE2, ICAM1, PECAM1, VEGFR3, endoglin), pharmacodynamic changes were confirmed at the dose of E7389-LF 2.1 mg/m², and with continuous treatment, pharmacodynamic changes were maintained at later cycles (Supplementary Fig. S3C). IFNγ-related biomarkers (IFNγ, MIG, IP10, ITAC) had peaks 7 days after dosing, at C1D8 and C2D8 (Supplementary Fig. S3C). Dichotomized analyses were performed to compare PFS between patients with high and low biomarker levels (patients with plasma levels <33% were categorized as “low” and patients with plasma levels ≥33% were categorized as “high”). At baseline, the only IFN-related biomarker associated with PFS was MIG. Patients with higher levels of collagen IV, PECAM1, TIE2, VEGFR3, IP10, and ITAC after treatment showed longer PFS (P < 0.05) at one or more time points (Supplementary Table S5; Supplementary Fig. S3D).

Study 120 was an open-label phase Ib/II study of E7389-LF in combination with nivolumab in patients with solid tumors. This analysis focused on the gastric cancer cohort of the phase II part of the study, in which eight of 31 patients had an objective response, resulting in an ORR of 25.8%, which exceeded the planned success criteria. The median PFS was 2.69 months (95% CI, 1.91–2.99) and the median OS was 7.85 months (95% CI, 4.47–NE). The ORR in patients with CPS ≥5 was 22.2% and ORR in patients with CPS <5 was 25.0%, demonstrating promising antitumor activity across CPS levels. Results of this study in Japan are noteworthy because the European Medicines Agency has approved nivolumab in combination with chemotherapy for first-line treatment with gastric, gastroesophageal, or esophageal adenocarcinoma in patients with CPS ≥5 (15). Assessment of some subgroups (i.e., prior gastrectomy and liver metastases at baseline) indicated that E7389-LF in combination with nivolumab showed efficacy across these groups, as well as efficacy independent of CPS at baseline.

The ATTRACTION-2 trial assessed nivolumab versus placebo in patients with unresectable or advanced gastric cancer, who were confirmed to have disease refractory to, or intolerant of, standard therapy (10). Patients who had received two or more regimens of chemotherapy in the advanced or recurrent settings were eligible. In the ATTRACTION-2 trial, patients who received nivolumab (n = 268) had an ORR of 11.2% (95% CI, 7.7–15.6). In Japanese patients who received nivolumab (n = 152) in ATTRACTION-2, ORR was 14.0% (95% CI, 8.5–21.2; ref. 14). Median PFS in the nivolumab arm was 1.7 months. Median OS in the nivolumab arm was 5.4 months; and the OS rate was 76.2% (95% CI, 68.6–82.2) at 3 months, 47.1% (95% CI, 38.9–54.8) at 6 months, and 35.8% (95% CI, 28.2–43.4) at 9 months.

The Japanese population of ATTRACTION-2 represents a more relevant comparison to the entirely Japanese population of Study 120. Although cross-trial comparisons should be made with caution, the results from Study 120 indicate that E7389-LF in combination with nivolumab exhibited promising antitumor activity relative to nivolumab monotherapy for ORR, median PFS, and median OS. This result may be due to OS improvement through successful posttreatment after third-line E7389-LF in combination with nivolumab, which was supported by the biomarker analysis. The ORR from Study 120 also indicate that E7389-LF in combination with nivolumab exhibited numerically improved antitumor activity compared with E7389-LF monotherapy for ORR (17.6%; 95% CI, 6.8–34.5) in Study 114 (11), acknowledging the limitations of cross-study comparisons. Overall, the combination appeared to show promising efficacy compared with either nivolumab or E7389-LF monotherapies, and CPS-independent antitumor activity was possibly driven by E7389-LF-induced immune-modulation (discussed further below). Considering the limitations in comparing these studies, further evaluation would be necessary to determine whether the combination is superior to either. No new safety signals were observed compared with the safety profiles of E7389-LF alone or nivolumab monotherapy. The most common treatment-related TEAE were neutropenia, leukopenia, and decreased appetite. These treatment-related TEAE are consistent with those seen in other studies of E7389-LF (16, 17). Although 51.6% of patients had a treatment-related TEAE that led to dose reduction of E7389-LF, PR were observed after dose reductions, suggesting that judicious dose reductions can allow patients to remain on treatment and optimize efficacy.

Biomarker changes suggested vascular remodeling activity (specifically vessel density increases), induction of tumor-infiltrating lymphocytes, and enhancement of antitumor immunity via IFNγ signaling. Treatment with E7389-LF in combination with nivolumab led to increases from baseline in microvessel density [as measured by PECAM1 (CD31) IHC] and in tumor-infiltrating lymphocytes (as measured by the number of CD8-positive cells). Because of the limited number of patients who had tumor tissues available for biomarker analysis, associations between these biomarker changes and clinical outcomes were not able to be assessed. Moreover, changes observed using IHC were supported by changes in plasma markers. Although the number of patients in our study is small, these results are particularly interesting because they mirror those seen in preclinical studies of E7389-LF alone and in combination with an anti–PD-1 antibody (6). From baseline to C1D8 and/or C2D8, six of seven vascular markers increased [endoglin, PECAM1, ICAM1, TIE2 (TEK), collagen IV, and VEGFR3]. All IFN-related markers increased [IFNγ, IP10 (CXCL10), MIG (CXCL9), and ITAC]. Although clear associations with most IFN-related markers were not observed at baseline, patients with higher plasma levels of vascular and IFNγ markers after treatment with E7389-LF in combination with nivolumab showed longer PFS than patients with lower levels of these biomarkers. The association of these markers with PFS indicates that this subset of plasma markers are suitable as surrogate markers of potential efficacy for the combination of E7389-LF and nivolumab.

On the basis of results from the CheckMate-649 (18) and ATTRACTION-4 trials (19), nivolumab in combination with chemotherapy has recently been approved as a first-line therapy for gastric cancer in the United States, Europe, and Japan. Patients from these or other studies of ICI were included in Study 120; however, it was not known if these patients had received ICI or control treatment. Study 120 included one patient who had confirmed prior ICI treatment and 10 patients who had received either an ICI study drug or control. This lack of data on prior ICI treatment is a limitation for this study, and impact of prior ICI treatment remains an avenue for future study. In addition, Study 120 is limited by the small number of patients overall and by the fact that all patients were Japanese, which may impact the relevance to a more global population. The study was also limited in that ORR and PFS data were not confirmed by a blinded independent central review. Finally, the lack of comparable biomarker data available for patients treated with nivolumab monotherapy limits the ability to compare the biomarkers results of patients in this study who received E7389-LF in combination with nivolumab with other biomarker analyses.

The availability of therapies after the failure of ICI is an unmet need in the gastric cancer treatment landscape. These results from Study 120 demonstrate promising antitumor activity of E7389-LF in combination with nivolumab in patients with gastric cancer, with no new or unexpected safety signals. This combination of the liposomal formulation of eribulin plus immune checkpoint inhibition could represent an addition to the treatment landscape for patients with pretreated gastric cancer.

N. Yamamoto reports grants from Eisai during the conduct of the study; grants from Astellas, Chugai, Eisai, Taiho, BMS, Pfizer, Novartis, Eli Lilly, AbbVie, Daiichi-Sankyo, Bayer, Boehringer Ingelheim, Kyowa Kirin, Takeda, ONO, Janssen Pharma, MSD, MERCK, GSK, Sumitomo Pharma, Chiome Bioscience, Otsuka, Carna Biosciences, Genmab, Shionogi, TORAY, KAKEN, AstraZeneca, Cmic, InventisBio, and Rakuten Medical; and personal fees from Eisai, Takeda, Boehringer Ingelheim, Cmic, Chugai, MERCK, Healios, and Chugai outside the submitted work. N. Sugimoto reports grants from MSD, Ono Pharmaceutical, Taiho Pharmaceutical, Lilly Japan, Daiichi Sankyo, Sumitomo Pharma, Chugai Pharma, BeiGene, Solasia Pharma, Astellas Pharma, and Eisai outside the submitted work. H. Kawakami reports grants and nonfinancial support from Eisai Co. Ltd. during the conduct of the study; grants and personal fees from Taiho Pharmaceutical Co. Ltd., Bristol Myers Squibb Co. Ltd., and Daiichi Sankyo Co. Ltd.; grants from Kobayashi Pharmaceutical. Co., Ltd.; personal fees from Astellas Pharma Inc., Eli Lilly Japan K.K., MSD K.K., Ono Pharmaceutical Co. Ltd., Chugai Pharmaceutical Co. Ltd., Merck Biopharma Co., Ltd., Takeda Pharmaceutical Co. Ltd., Yakult Pharmaceutical Industry, Teijin Pharma Ltd., Otsuka Pharmaceutical Co., Ltd., Nippon Kayaku Co. Ltd., and GlaxoSmithKline K.K. outside the submitted work. K. Yamaguchi reports personal fees from Daiichi Sankyo Co., Ltd., Bristol Myers Squibb K.K., Eli Lilly Japan K.K., and Ono Pharmaceutical Co., Ltd.; and grants from Taiho Pharmaceutical Co., Ltd. outside the submitted work. Y. Kurokawa reports grants and personal fees from Taiho Pharmaceutical; personal fees from Medtronic, Johnson & Johnson, Stryker, Ono Pharmaceutical, Daiichi Sankyo, Lilly, Nippon Kayaku, MC Medical, and Kaken Pharmaceutical; and grants from Yakult Honsha and AstraZeneca outside the submitted work. T. Kawakami reports other support from Ono Pharmaceutical and Bristol Myers Squibb during the conduct of the study. H. Hara reports grants from Eisai during the conduct of the study; grants from AstraZeneca, Merck Serono, MSD, Ono Pharmaceutical, Taiho Pharmaceutical, Boehringer Ingelheim, Daiichi Sankyo, BeiGene, Astellas Pharma, Bayer, Amgen, Chugai Pharma, Janssen Oncology, ALX Oncology, Bristol Myers Squibb, and Jazz Pharmaceuticals outside the submitted work. S. Kaname reports personal fees from Ono Pharmaceutical Co., Ltd. during the conduct of the study. D. Matsuoka reports personal fees from Eisai Co., Ltd. during the conduct of the study and personal fees from Eisai Co., Ltd. outside the submitted work. T. Takase reports personal fees from Eisai Co., Ltd. during the conduct of the study. T. Semba reports personal fees from Eisai Co., Ltd. during the conduct of the study; and personal fees from Eisai Co., Ltd. outside the submitted work; T. Semba also has a patent for US11083705B2 issued, US20220117933A1 pending, WO2021/020336 pending, and TW202118486 pending. K. Muro reports grants from Eisai during the conduct of the study; grants from Astellas, Amgen, Sanofi, PRA Health Sciences, PAREXEL International, Taiho, MSD, Novartis, Ono, and Chugai; personal fees from Taiho, Takeda, Daiichi Sankyo, Ono, Bristol Myers Squibb, Eli Lilly, Amgen, AstraZeneca, Ono, Chugai, and Astellas outside the submitted work. No disclosures were reported by the other authors.

A. Kawazoe: Resources, supervision, investigation, writing–original draft, writing–review and editing. N. Yamamoto: Resources, investigation, writing–original draft, writing–review and editing. N. Sugimoto: Resources, investigation, writing–original draft, writing–review and editing. H. Kawakami: Resources, investigation, writing–original draft, writing–review and editing. T. Oshima: Resources, investigation, writing–original draft, writing–review and editing. K. Yamaguchi: Resources, investigation, writing–original draft, writing–review and editing. K. Hino: Resources, investigation, writing–original draft, writing–review and editing. M. Hirao: Resources, investigation, writing–original draft, writing–review and editing. Y. Kurokawa: Resources, investigation, writing–original draft, writing–review and editing. T. Kawakami: Resources, investigation, writing–original draft, writing–review and editing. M. Tsuda: Resources, investigation, writing–original draft, writing–review and editing. H. Hara: Resources, investigation, writing–original draft, writing–review and editing. S. Kaname: Resources, writing–original draft, writing–review and editing. D. Matsuoka: Conceptualization, resources, formal analysis, methodology, writing–original draft, project administration, writing–review and editing. Y. Otake: Conceptualization, resources, formal analysis, investigation, methodology, writing–original draft, writing–review and editing. K. Yasuda: Resources, formal analysis, methodology, writing–original draft, project administration, writing–review and editing. T. Takase: Resources, formal analysis, methodology, writing–original draft, writing–review and editing. S. Takashima: Resources, data curation, formal analysis, validation, visualization, methodology, writing–original draft, writing–review and editing. T. Semba: Resources, data curation, formal analysis, supervision, validation, visualization, methodology, writing–original draft, writing–review and editing. K. Muro: Resources, supervision, investigation, writing–original draft, writing–review and editing.

This trial was sponsored by Eisai Co., Ltd. Nivolumab was provided by Ono Pharmaceutical Co., Ltd. Medical writing was provided by Heather A. Mitchell, PhD, of Oxford PharmaGenesis Inc. and was funded by Eisai Inc.