Multicenter Phase I/II Study of Nivolumab Combined with Paclitaxel Plus Ramucirumab as Second-line Treatment in Patients with Advanced Gastric Cancer Free

Gastric cancer reportedly harbors the fifth highest rate of somatic mutations among major cancer types (1). Programmed death-ligand 1 (PD-L1) and its receptor, programmed cell death-1 (PD-1), is overexpressed in gastric cancer (2, 3), and on T cells in gastric cancer, respectively (4). However, the efficacy of anti–PD-1/PD-L1 antibodies in treating advanced gastric cancer (AGC) is limited. Although nivolumab, a human monoclonal IgG4 antibody targeting PD-1, confers a survival benefit as salvage therapy for patients with AGC and is an established standard treatment, the overall response rate (ORR) was only 11% in the ATTRACTION-2 trial (5). Even as second-line treatment, pembrolizumab targeting PD-1 was not more efficacious than paclitaxel alone in the KEYNOTE-061 trial (6), median progression-free survival (PFS) was 1.5 months [95% confidence interval (CI), 1.4–2.0] for pembrolizumab and 4.1 months (95% CI, 3.1–4.2) for paclitaxel (HR, 1.27; 95% CI, 1.03–1.57). The standard second-line regimen is paclitaxel plus ramucirumab, an IgG1 anti–VEGFR-2 antibody that demonstrated superiority to paclitaxel alone in PFS and overall survival (OS) rates in the RAINBOW trial (7).

As first-line treatment, pembrolizumab showed noninferiority in OS to chemotherapy of cisplatin and fluoropyrimidines for combined positive score (CPS) ≥1 AGC tumors in the KEYNOTE-062 trial (8). However, survival in the pembrolizumab arm was lower than that in the chemotherapy-only arm at approximately 1 year from the start of treatment, indicating that pembrolizumab is not optimal first-line treatment for all patients. However, the survival curve of the combination of pembrolizumab and chemotherapy in the KEYNOTE-062 trial almost overlapped in the early period and was superior in the later period to that in the pembrolizumab arm for CPS ≥ 1 tumors, although statistical superiority was not demonstrated. Conversely, nivolumab combined with chemotherapy in the first-line setting recently showed significant superiority to chemotherapy alone in both OS and PFS rates in patients with CPS ≥ 5 tumors (CheckMate 649 study; ref. 9). However, in the ATTRACTION-4 (ONO-4538-37) trial conducted in Japan, South Korea, and Taiwan, nivolumab combined with chemotherapy in the first-line setting could not show superiority to chemotherapy alone in OS rates in all patients (10). These findings suggest that chemotherapy might overcome resistance to pembrolizumab or can only be given to patients who respond to chemotherapy but not to pembrolizumab.

A synergistic antitumor effect of simultaneous blockade of PD-1 and taxanes, such as paclitaxel, has been reported. Low-dose paclitaxel promotes Toll-like receptor 4–dependent maturation of dendritic cells (DCs) and enhances antigen-specific, IFNγ-secreting CD8⁺ T cells in vivo (11). On the other hand, simultaneous blockade of PD-1 and VEGFR-2 enhanced T-cell recruitment, activated local immune status, and induced synergistic antitumor effects (12). These findings support the development of a combination regimen of nivolumab and paclitaxel plus ramucirumab, the standard second-line treatment for AGC (7).

We conducted a multicenter phase I/II study of nivolumab and paclitaxel plus ramucirumab, registered as UMIN-CTR (UMIN000025947).

Eligibility criteria included age ≥20 years; Eastern Cooperative Oncology Group (ECOG) performance status (PS) of 0 or 1; histologically verified gastric adenocarcinoma (papillary, tubular, or poorly differentiated), signet-ring cell carcinoma, mucinous adenocarcinoma, or hepatoid adenocarcinoma; recurrence more than 6 months after the completion of postoperative adjuvant chemotherapy or patients with stage IV disease who received one prior systemic chemotherapy of platinum and fluoropyrimidine doublet chemotherapy (patients with recurrence within 6 months after completion of postoperative adjuvant chemotherapy of platinum and fluoropyrimidine doublet chemotherapy were eligible); at least one measurable lesion; preserved organ function, including neutrophil count ≥ 1,500/mm³, platelet count ≥ 100,000/mm³, hemoglobin ≥ 8.0 g/dL, aspartate aminotransferase and alanine aminotransferase ≤100 U/L, total bilirubin ≤ 1.5 mg/dL and serum creatinine ≤ 1.5 mg/dL, and prothrombin time-international normalized ratio ≤ 1.5; one of the following conditions: (i) urinary protein with a negative (−) or 1+ result or (ii) where urinary protein was 2+ or higher, 24-hour urine protein must be ≤1 g/24 hours; and adequate blood pressure control (≤two antihypertensive agents and systolic and diastolic blood pressures ≤150 and 90 mm Hg, respectively). Exclusion criteria included previous administration of anti–PD-1 antibodies, anti–PD-L1 antibodies, anti–CTLA-4 antibodies, or other T-cell suppression therapy; anticancer treatment, such as chemotherapy, molecular-targeted therapy, immunotherapy, and radiotherapy, administered within 14 days before enrollment; systemic corticosteroids at prednisolone-equivalent doses of >10 mg/day (except when administered temporarily) or immunosuppressive agents administered within 14 days before enrollment; active multiple cancers; active infection; uncontrolled complications, such as heart disease, pulmonary fibrosis, or active pneumonitis; and pregnancy or lactation. Written informed consent was obtained from each patient before the initiation of study procedures. The institutional review boards of all participating institutions approved the study protocol, which was conducted according to the Declaration of Helsinki and Japanese Good Clinical Practice guidelines.

This was a multicenter, open-label, nonrandomized phase I/II study with dose deescalation in the phase I part and cohort expansion in the phase II part. In the phase I part, we assessed dose-limiting toxicity (DLT) and determined the recommended dose (RD) of nivolumab combined with paclitaxel plus ramucirumab. The phase I part, starting from level 1, followed a modified 3+3 design and included two dose-level cohorts (Fig. 1). Toxicity profiles of nivolumab and paclitaxel plus ramucirumab did not generally overlap. We, therefore, selected a dose deescalation design starting from a standard dose and schedule as described previously for nivolumab (5), combined with the clinically established fixed dose and schedule of paclitaxel plus ramucirumab (Fig. 1; ref. 10). If DLT was observed in zero to two patients in level 1, it was determined as the RD for phase II. If DLTs were observed in ≥3 patients of a total of six patients in level 1, level 0 would be evaluated. If DLTs were observed in zero to two patients of the total six patients in level 0, it was determined as the RD for phase II. If DLTs were observed in ≥3 patients in level 0, the study would be discontinued. Dose deescalation decisions were made by the primary investigator and an Independent Data Monitoring Committee, based on safety and other parameters in phase I. Eligible patients received protocol treatment on day 1 of a 28-day cycle and continued until disease progression or unacceptable toxicity.

DLTs were assessed in cycle 1 with the following definitions: (i) grade 4 neutropenia, maintained for at least 8 days; (ii) febrile neutropenia; (iii) grade 4 thrombocytopenia; (iv) grade 2 or higher pneumonitis; (v) uncontrolled, grade 2 or higher uveitis, eye pain, or optic nerve disorder; (vi) grade 3 or higher nausea, vomiting, anorexia, or diarrhea, uncontrolled by supporting treatment; (vii) grade 3 nonhematotoxicity other than (v), (vi), or electrolyte abnormalities; (viii) one or more of the drugs included in combination therapy not meeting the cycle initiation criteria, and ≥28 days elapsed since the scheduled initiation date for the second cycle without initiation being possible; and (xi) per-protocol treatment discontinued because of adverse reactions other than (i) to (viii), above.

The primary objective of the second stage was to assess the clinical efficacy of nivolumab combined with paclitaxel plus ramucirumab as second-line treatment for AGC. Key secondary objectives included safety assessment at the RD level.

Adverse events (AEs) were classified according to Common Terminology Criteria for Adverse Events version 4.0. The AE reporting period was from day 1 of cycle 1 until 30 days after the last dose of any protocol drugs. CT scans with ≤5-mm-thick sections were performed for tumor assessment every 4 weeks for 12 weeks from day 1 of cycle 1 and every 8 weeks thereafter, evaluated on the basis of RECIST version 1.1 (13). Physical examinations and laboratory tests were performed on days 1, 8, and 15. Serious AEs were death, life-threatening AEs, AEs requiring hospitalization or prolongation of hospitalization for treatment, AEs leading to permanent or major disability or dysfunction, AEs leading to congenital abnormalities for later generations, and AEs judged as the result of another medically important condition.

For biomarker analyses, tumor tissues were obtained before treatment initiation (either the archival or taken immediately just before the study enrollment) for IHC analysis of HER2, mismatch repair (MMR) proteins of MutS Homolog 6 (MSH6) and PMS1 Homolog 2 (PMS2), and PD-L1. PD-L1 tumor expression was examined with the Dako PD-L1 IHC 22C3 pharmDx. The tumor proportion score (TPS) was defined as the percentage of viable tumor cells showing partial or complete membrane staining, and positivity was defined as staining in ≥1% of tumor cells. CPS was the number of PD-L1–positive cells (tumor cells, macrophages, and lymphocytes) divided by the total number of viable tumor cells multiplied by 100. EB-virus encoded small RNAs (EBER) were analyzed using ISH.

The primary endpoint of the phase I part was RD determination based on DLT evaluation. The DLT analysis population included all patients in phase I who completed the DLT evaluation period of 28 days after study treatment initiation. The secondary endpoint of phase I was AE rate. The safety population included all patients who received any dose of protocol drugs. The primary endpoint of the phase II part was 6-month PFS rate with two-sided 80% CI, which was estimated for the efficacy analysis populations [full analysis set (FAS) and per-protocol set] by the Kaplan–Meier method. Greenwood formula for variance was used to establish the CI. In addition, an appropriate test statistical value was established, and the P value was calculated for a test relating to the null hypothesis, which was the 6-month PFS rate of 35%, it had been reported 36% in paclitaxel + ramucirumab arm in the RAINBOW trial (7).

Secondary endpoints in the phase II part were ORR, disease control rate (DCR), PFS, OS, and AE rate. DCR was the proportion of patients who experienced complete response, partial response, or stable disease. An accurate CI based on the binomial distribution was used for the two-sided 95% CI of ORR and DCR. PFS and OS were estimated for the efficacy analysis populations by the Kaplan–Meier method. PFS was the time from enrollment to the first documentation of disease progression or death. For surviving patients without documented disease progression, data on PFS were censored on the date the absence of progression was confirmed. PFS in patients who discontinued protocol treatment due to toxicity was the time to disease progression in subsequent therapies or to death. OS was the duration from enrollment to death from any cause. All statistical analyses were performed using SAS Software version 9.4 (TS1M3; SAS Institute Inc.).

A total of 43 patients (six patients in phase I and 37 patients in phase II) were enrolled between February 2017 and July 2018, and all patients were included in the FAS. Table 1 summarizes the baseline characteristics in the FAS. The majority of patients were male, with an age range of 38–78 years. Twenty-two patients (51.2%) had PS of 0, TPS was positive in six patients (14.0%), and CPS was ≥1 in 26 patients (60.5%) and ≥10 in seven patients (16.3%). MMR and EBER were positive in zero and 4 patients (9.3%), respectively.

Median treatment duration was 4.6 months (95% CI, 2.8–6.2). Median number of administrations of paclitaxel, ramucirumab, and nivolumab was seven (range, 1–58), nine (range, 1–49), and nine (range, 1–44), respectively. Five patients (11.6%) discontinued protocol treatment because of toxicities: two paclitaxel-related toxicities and four nivolumab-related toxicities (one patient had both paclitaxel- and nivolumab-related toxicities). By final data cutoff of January 2019, the median follow-up was 16.8 months and protocol treatment was ongoing in one patient. Postprotocol treatment was administered to 31 patients (72.1%). Regimens containing irinotecan were administered to 26 patients (60.5%; Supplementary Table S1).

All six patients in the phase I part completed the DLT evaluation period of 28 days after study treatment initiation. DLTs (febrile neutropenia and neutropenia over a period of 8 days) were observed in two of six patients in level 1, and the RD was determined at level 1. No patients were enrolled at level 0, and all patients included in the phase II part were treated at the RD. Thirty-nine (90.7%) patients experienced grade ≥3 treatment-related AEs (Table 2). The most common grade ≥3 AE was neutrophil count decrease (33 patients, 76.7%), while febrile neutropenia was observed in seven patients (16.3%). Fourteen (32.6%) patients experienced grade ≥3 immune-related AEs, with all frequencies ≤5%. There was one treatment-related death from thrombocytopenia.

Clinical courses of all patients are shown in Fig. 2. At the primary analysis (January 2019, median follow-up time of 8.2 months), 23 PFS events (53.5%) had occurred. Six-month PFS was 46.5% (80% CI, 36.4–55.8; P = 0.067), which means primary endpoint was met (Fig. 3A). At final analysis with a median follow-up time of 23.2 months (95% CI, 17.4–28.0), OS at 12 and 18 months was 55.8% (95% CI, 39.8–69.1) and 32.1% (95% CI, 18.2–46.8), respectively (Fig. 3B). Median PFS was 5.1 months (95% CI, 4.5–6.5 months): 6.4 months (95% CI, 4.2–7.9 months) in patients with CPS ≥ 1 and 5.1 months (95% CI, 2.6–6.7 months) in patients with CPS < 1 (Fig. 4A); 6.4 months (95% CI, 1.0–6.9 months) in patients with CPS ≥ 5 and 5.9 months (95% CI, 4.6–6.9 months) in patients with CPS < 5 (Supplementary Fig. S1A); 6.7 months (95% CI, 1.0–8.8 months) in patients with CPS ≥ 10 and 5.5 months (95% CI, 4.2–6.7 months) in patients with CPS < 10 (Supplementary Fig. S1B); and 5.8 months (95% CI, 4.2–7.9 months) in patients with PS 0 and 4.9 months (95% CI, 3.2–6.4 months) in patients with PS 1 (Fig. 4B). Thirty-two OS events (74.4%) occurred and median survival time (MST) was 13.1 months (95% CI, 8.0–16.6 months): 13.8 months (95% CI, 8.0–19.5 months) in patients with CPS ≥ 1 and 8.0 months (95% CI, 4.8–24.1 months) in patients with CPS < 1 (Fig. 4C); 13.1 months (95% CI, 5.1 months–NA) in patients with CPS ≥ 5 and 14.9 months (95% CI, 7.4–19.5 months) in patients with CPS < 5 (Supplementary Fig. S1C); 13.8 months (95% CI, 8.0 months–NA) in patients with CPS ≥ 10 and 13.0 months (95% CI, 7.1–18.6 months) in patients with CPS < 10 (Supplementary Fig. S1D); and 14.4 months (95% CI, 7.7–28.0 months) in patients with PS 0 and 8.0 months (95% CI, 4.8–16.6 months) in patients with PS 1 (Fig. 4D).

DCR was 83.7% (95% CI, 69.3–93.2). ORR was 37.2% (95% CI, 23.0–53.5): 46.2% (95% CI, 26.6–66.6) in patients with CPS ≥ 1 and 30.8% (95% CI, 9.1–61.4) in patients with CPS < 1; and 45.5% (95% CI, 24.4–67.8) in patients with PS 0 and 28.6% (95% CI, 11.3–52.5) in patients with PS 1.

Subset analyses of OS are described in Supplementary Table S2. Low albumin, high lactate dehydrogenase, and peritoneal metastasis may represent markers of poor prognosis with the study treatment regimen. Prognostic and predictive effects of PD-L1 expression by CPS and TPS were not identified.

Here, we report the first study showing promising efficacy of nivolumab combined with paclitaxel plus ramucirumab as second-line treatment for AGC, in which the primary endpoint was met. RD was determined at the standard doses and schedules of paclitaxel plus ramucirumab and nivolumab. The most common AEs were hematotoxicities, and the frequency and grade of immune-related AEs were manageable.

VEGF-A impairs DC maturation to induce PD-L1 expression on DCs and activate regulatory T cells (Tregs) via neuropilin-1 (NRP-1; ref. 14). Moreover, VEGF-A induces the accumulation of myeloid-derived suppressor cells, immature DCs, Tregs, and tumor-associated macrophages (15), which are potential mechanisms for treatment failure of PD-1 blockade (16–21). Immunosuppressive cell–related markers, such as forkhead box P3 (Foxp3) or colony-stimulating factor 1 receptor, are more highly expressed in the PD-L1–positive than the PD-L1–negative population (22). Therefore, synergistic antitumor effects induced by simultaneous blockade of VEGF-A and PD-1 have been investigated. Concomitant administration of a VEGF-A inhibitor and an anti–PD-1 antibody produced strong synergistic antitumor activity against tumors showing high VEGF-A production (23). In addition, when VEGF was inhibited in renal cell carcinoma using a tyrosine kinase inhibitor, intratumor PD-L1 and Foxp3 expression was markedly reduced (24). Combination therapy of antiangiogenesis treatment and PD-1/PD-L1 blockade has subsequently become standard therapy in non–small cell and renal cell carcinoma (25–28).

Ramucirumab has been shown to reduce Tregs in local tumors of patients with gastric cancer (29), and showed promising activity in combination with pembrolizumab in patients with gastric or gastro-esophageal junction adenocarcinoma whose disease had progressed on one or two lines of previous therapy (n = 41); MST was 5.9 months, 18-month survival was 22%, median PFS was 2.5 months, and ORR was 7% (30). The combination of regorafenib 80 mg plus nivolumab had a manageable safety profile and encouraging antitumor activity in patients with gastric cancer who had received ≥2 previous lines of chemotherapy (n = 25); median PFS was 5.6 months and ORR was 44% (31).

The survival data in the pembrolizumab arm of the KEYNOTE-062 trial of first-line treatment for AGC showed that the antitumor effect of cytotoxic agents remains important even in CPS ≥ 1 AGC (7). Most previous reports on concomitant treatment with cytotoxic anticancer agents and nivolumab have been in lung cancer. In the Impower150 trial (25), addition of atezolizumab, an anti–PD-L1 antibody, to carboplatin plus paclitaxel and bevacizumab as first-line treatment for metastatic nonsquamous non–small cell lung cancer resulted in significant improvements in PFS and OS regardless of PD-L1 expression, indicating a favorable association of paclitaxel and VEGF-A inhibition with anti–PD-L1 blockade. Of note, Kaplan–Meier curves of PFS and OS in the arms with concomitant chemotherapy and PD-1/PD-L1 antibodies in the KEYNOTE-062 and Impower150 trials showed that treatment benefits were delayed and increased substantially after the median was reached. In our study, MST was 13.1 months, comparable with 11.4 months in Japanese patients receiving paclitaxel plus ramucirumab in the RAINBOW trial (32). Although cross-trial comparisons require careful interpretation, 12- and 18-month OS rates in our trial were 55.8% (95% CI, 39.8–69.1) and 32.1% (95% CI, 18.2–46.8), respectively, which were better than that among Japanese patients receiving paclitaxel plus ramucirumab in the RAINBOW trial. The numbers of patients were small in the CPS subgroups, PD-L1 expression may have had some influence on clinical outcome in this combination regimen.

Very recently, nivolumab combined with chemotherapy in the first-line setting showed significant superiority to chemotherapy alone in OS rates both in CPS ≥ 5 patients with AGC and in CPS ≥ 1 patients with AGC (CheckMate 649 study; ref. 9). Although the treatment benefit of the combination arm compared with the chemotherapy-alone arm increased gradually in both CPS ≥ 5 AGC and CPS ≥ 1 AGC, the Kaplan–Meier curves were more differentiated in CPS ≥ 5 AGC than in CPS ≥ 1 AGC. Furthermore, the HR was lower in CPS ≥ 5 AGC [0.71 (98.4% CI, 0.59–0.86)] than in CPS ≥ 1 AGC [0.77 (98.4% CI, 0.64–0.92)]. These data may indicate that CPS has some impact on clinical outcome in AGC. However, in the ATTRACTION-4 (ONO-4538-37) trial conducted in Asia, the Kaplan–Meier curves of OS in both arms including CPS < 1 AGC showed no significant difference (10). Data on clinical outcome according to CPS, proportion of postprotocol treatment, and proportion of microsatellite instability status, among others, have not yet been reported. Taken together, these data indicate that the standard regimen and optimal use of PD-1 antibody for AGC have not yet been determined in Asian countries.

Concerning safety, the frequencies of AEs of neutrophil count decrease and febrile neutropenia were higher in this study than in the comparable arm of the RAINBOW trial (32), potentially attributable to factors such as high proportion of patients with PS 1, stage IV disease, and metastatic sites ≥3 in this study, indicative of poor patient condition. However, AEs in this study were evaluated as tolerable.

This study had some limitations. The small sample size and single-arm design mean that this regimen requires further evaluation. Only four EBER-positive patients were included (Fig. 2), and efficacy of the regimen thus requires further investigation in this subgroup. Although HER2-positive patients receiving nivolumab in the ATTRACTION-2 trial (33) and HER2-positive patients receiving cabazitaxel, a novel next-generation taxane, in the phase II study recently reported (34) tended to have better prognosis, HER2 positivity was not a prognostic marker in subset analyses of OS in this study (Supplementary Table S2; Supplementary Fig. S2).

Nivolumab with paclitaxel plus ramucirumab demonstrated promising antitumor activity with manageable toxicities as second-line treatment for patients with AGC. Further biomarker results from this study will be reported at a later date and may clarify the characteristics of patients suitable for nivolumab with paclitaxel plus ramucirumab.

T.E. Nakajima reports grants from Ono Pharmaceutical Co. during the conduct of the study and grants from Astellas Pharma Inc., Sumitomo Dainippon Pharma Co., Eisai Co, and Solasia Pharma K.K., personal fees from Mochida Pharmaceutical, Celltrion Healthcare Japan, grants and personal fees from Taiho Pharmaceutical Co., Chugai Pharmaceutical Co., Takeda Pharmaceutical Co., Sanofi K.K., Daiichi Sankyo Co., Eli Lilly Japan K.K., Nippon Kayaku Co., Ono Pharmaceutical Co., and MSD K.K., and personal fees from Merck Serono Co., Sawai Pharmaceutical Co., Bayer Yakuhin, Bristol-Myers Squibb, Teijin Pharma, Pfizer Japan Inc., Novartis Japan, Yakult Honsha Co., and Nipro Co. outside the submitted work. S. Kadowaki reports grants and personal fees from Ono Pharmaceutical Co. during the conduct of the study and grants and personal fees from Lilly outside the submitted work. K. Minashi reports grants from Ono Pharmaceutical Co., Ltd, during the conduct of the study, grants from MSD K.K., Mediscience Planning Inc, Merck Biopharma Co., Ltd, Astellas Pharma Inc., Taiho Pharmaceutical Co., LTD, and Daiichi Sankyo Co., Ltd, outside the submitted work. T. Nishina reports grants and personal fees from Ono Pharmaceutical Co. during the conduct of the study, grants and personal fees Taiho Pharmaceutical Co., Chugai Pharmaceutical Co., Ono Pharmaceutical Co., Bristol-Myers Squibb, and Lilly Pharma and grants from Daiichi Sankyo, MSD, Sumitomo Dainippon Pharma Co. outside the submitted work. T. Yamanaka reports grants and personal fees from Chugai and Bayer, grants from Taiho Pharmaceuticals, and personal fees from Takeda and AstraZeneca outside the submitted work. N. Izawa reports grants from Ono Pharmaceutical Co., Ltd during the conduct of the study and other from Ono Pharmaceutical Co., Ltd outside the submitted work. K. Muro reports grants from MSD, Solasia Pharma, Sanofi, Astellas Amgen Biopharma, Daiichi Sankyo, Parexel International, Taiho Pharmaceutical, Merck Serono, Pfizer, and Ono Pharmaceutical Co. and personal fees from Ono Pharmaceutical Co., Amgen, AstraZeneca, Taiho, Chugai, Takeda, Eli Lilly, Sanofi, Bristol-Myers Squibb, and Bayer during the conduct of the study. S. Hironaka reports personal fees from Bristol-Myers Squibb Japan, Ono Pharma, Taiho Pharmaceutical, Yakult Honsha, Daiichi Sankyo, Lilly, Chugai Pharmaceutical Co, Nippon Kayaku, Tsumura & Co, Sanofi, Merck, AstraZeneca, and MSD K.K outside the submitted work. T. Kajiwara reports grants from Ono Pharmaceutical Co. during the conduct of the study, grants from Chugai Pharmaceutical Co., Taiho Pharmaceutical Co., Bristol-Myers Squibb, Merck Biopharma Co., and Kyowa Hakko Kirin Co. outside the submitted work. Y. Kawakami reports grants from Ono Pharmaceutical Co during the conduct of the study, and grants from Bristol Myers Squibb, JSR, and CarnaBioSciences and personal fees from Bristol-Myers Squibb, MSD, AstraZeneca, Chugai, and Taiho Pharmaceutical outside the submitted work. No disclosures were reported by the other author.

T.E. Nakajima: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing-original draft, project administration, writing-review and editing. S. Kadowaki: Investigation, writing-review and editing. K. Minashi: Investigation, writing-review and editing. T. Nishina: Investigation, writing-review and editing. T. Yamanaka: Conceptualization, formal analysis, validation, methodology, writing-original draft, project administration, writing-review and editing. Y. Hayashi: Conceptualization, formal analysis, methodology, writing-review and editing. N. Izawa: Investigation, writing-review and editing. K. Muro: Investigation, writing-review and editing. S. Hironaka: Investigation, writing-review and editing. T. Kajiwara: Investigation, writing-review and editing. Y. Kawakami: Conceptualization, investigation, methodology, writing-review and editing.

We thank the participants and their families and our collaborators who contributed to the study: Dr. Toshiyuki Misumi for assistance with data analysis; the members of the Independent Data Monitoring Committee: Drs. Ichinosuke Hyodo, Chigusa Morizane, and Kan Yonemori; the members of St. Marianna University Data Center; and ASCA Corporation for editing a draft of this article. This work was supported by Ono Pharmaceutical Co., Ltd.

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.