Surrogate endpoints for survival in KEYNOTE-585: neoadjuvant/adjuvant pembrolizumab plus chemotherapy versus placebo plus chemotherapy for gastric or gastroesophageal junction adenocarcinoma

Abstract

Background: In the randomized phase III KEYNOTE-585 trial, neoadjuvant/adjuvant pembrolizumab plus chemotherapy was not superior to placebo plus chemotherapy for event-free survival (EFS) in participants with locally advanced gastric or gastroesophageal (GEJ) adenocarcinoma. However, pembrolizumab plus chemotherapy significantly improved pathologic complete response (pCR) versus placebo plus chemotherapy (difference 10.9%, 95% confidence interval (CI) 7.5% to 14.8%, P < 0.00001). This post hoc analysis evaluated whether pCR, major pathologic response (mPR), and pathologic downstaging (pDS) after neoadjuvant/adjuvant pembrolizumab plus chemotherapy are associated with improved survival outcomes. Patients and methods: Eligible participants had untreated, locally advanced gastric or GEJ adenocarcinoma (including Siewert type 2 or 3) and were scheduled for surgery after preoperative chemotherapy. The main cohort received pembrolizumab or placebo plus chemotherapy [cisplatin plus capecitabine (XP) or cisplatin plus 5-fluorouracil (FP)], whereas the safety cohort received pembrolizumab or placebo plus docetaxel, oxaliplatin, 5-fluorouracil, and leucovorin (FLOT). The outcomes for this post hoc analysis were the relationship between pCR, mPR, pDS to N0, or any pDS with EFS per RECIST v1.1 (by investigator) and overall survival (OS). We report outcomes in the main and FLOT cohorts combined. The data cut-off date was 9 February 2023. Results: A total of 1007 participants were enrolled and randomly assigned (n = 502, pembrolizumab plus chemotherapy; n = 505, placebo plus chemotherapy); 221 (44.0%) and 172 (34.1%) participants, respectively, had pathologic nodal stage N0. The pCR rate was 13.9% with pembrolizumab plus chemotherapy and 2.8% with placebo plus chemotherapy; mPR rates were 31.5% and 22.2%, respectively. Among participants who experienced mPR (<= 10% residual viable tumor), the hazard ratios for EFS and OS were 0.6 (95% CI 0.4-1.0) and 0.7 (95% CI 0.4-1.2), respectively, for pembrolizumab plus chemotherapy compared with placebo plus chemotherapy. Conclusion: These findings suggest a potential association between pCR, mPR, or pDS and survival in patients with locally advanced gastric or GEJ adenocarcinoma, although further validation is needed.