Adjuvant nivolumab plus chemotherapy versus placebo plus chemotherapy for stage III gastric or gastro-oesophageal junction cancer after gastrectomy with D2 or more extensive lymph-node dissection (ATTRACTION-5): a randomised, multicentre, double-blind, placebo-controlled, phase 3 trial

Background In Asia, adjuvant chemotherapy after gastrectomy with D2 or more extensive lymph-node dissection is standard treatment for people with pathological stage III gastric or gastro-oesophageal junction (GEJ) cancer. We aimed to assess the efficacy and safety of adjuvant nivolumab plus chemotherapy versus placebo plus chemotherapy administered in this setting. Methods ATTRACTION-5 was a randomised, multicentre, double-blind, placebo-controlled, phase 3 trial conducted at 96 hospitals in Japan, South Korea, Taiwan, and China. Eligible patients were aged between 20 years and 80 years with histologically confirmed pathological stage IIIA–C gastric or GEJ adenocarcinoma after gastrectomy with D2 or more extensive lymph-node dissection, with an Eastern Cooperative Oncology Group (ECOG) performance status score of 0 or 1 and available tumour tissue for PD-L1 expression analysis. Patients were randomly assigned (1:1) to receive either nivolumab plus chemotherapy or placebo plus chemotherapy via an interactive web-response system with block sizes of four. Investigational treatment, either nivolumab 360 mg or placebo, was administered intravenously for 30 min once every 3 weeks. Adjuvant chemotherapy was administered as either tegafur–gimeracil–oteracil (S-1) at an initial dose of 40 mg/m2 per dose orally twice per day for 28 consecutive days, followed by 14 days off per cycle, or capecitabine plus oxaliplatin consisting of an initial dose of intravenous oxaliplatin 130 mg/m2 for 2 h every 21 days and capecitabine 1000 mg/m2 per dose orally twice per day for 14 consecutive days, followed by 7 days off treatment. The primary endpoint was relapse-free survival by central assessment. The intention-to-treat population, consisting of all randomly assigned patients, was used for analysis of efficacy endpoints. The safety population, defined as patients who received at least one dose of trial drug, was used for analysis of safety endpoints. This trial is registered with ClinicalTrials.gov (NCT03006705) and is closed. Findings Between Feb 1, 2017, and Aug 15, 2019, 755 patients were randomly assigned to receive either adjuvant nivolumab plus chemotherapy (n=377) or adjuvant placebo plus chemotherapy (n=378). 267 (71%) of 377 patients in the nivolumab group and 263 (70%) of 378 patients in the placebo group were male; 110 (29%) of 377 patients in the nivolumab group and 115 (31%) of 378 patients in the placebo group were female. 745 patients received assigned treatment (371 in the nivolumab plus chemotherapy group; 374 in the placebo plus chemotherapy group), which was the safety population. Median time from first dose to data cutoff was 49·1 months (IQR 43·1–56·7). 3-year relapse-free survival was 68·4% (95% CI 63·0–73·2) in the nivolumab plus chemotherapy group and 65·3% (59·9–70·2) in the placebo plus chemotherapy group; the hazard ratio for relapse-free survival was 0·90 (95·72% CI 0·69–1·18; p=0·44). Treatment-related adverse events occurred in 366 (99%) of 371 patients in the nivolumab plus chemotherapy group and 364 (98%) of 374 patients in the placebo plus chemotherapy group. Discontinuation due to adverse events was more frequent in the nivolumab plus chemotherapy group (34 [9%] of 371 patients) than the placebo plus chemotherapy group (13 [4%] of 374 patients). The most common treatment-related adverse events were decreased appetite, nausea, diarrhoea, neutrophil count decreased, and peripheral sensory neuropathy. Interpretation The results of this trial do not support the addition of nivolumab to postoperative adjuvant therapy for patients with untreated, locally advanced, resectable gastric or GEJ cancer. Funding Ono Pharmaceutical and Bristol Myers Squibb.