Umbrella review of meta-analyses: perioperative immune checkpoint inhibitor therapy in resectable gastric and gastroesophageal junction adenocarcinoma

Background Immune checkpoint inhibitors (ICIs) have entered the perioperative treatment of resectable gastric and gastroesophageal junction (G/GEJ) adenocarcinoma, and perioperative durvalumab plus FLOT is now the standard of care. Numerous meta-analyses on this question have appeared. This umbrella review appraises their methodological quality, quantifies their overlap, grades the credibility of each pooled outcome and re-analyses the randomized evidence at the trial level. Methods PubMed/MEDLINE, Embase, Web of Science, Scopus and the Cochrane Library were searched from January 1, 2020 to July 31, 2026 for systematic reviews with meta-analysis of perioperative (neoadjuvant and/or adjuvant) ICI therapy in resectable G/GEJ adenocarcinoma. Results Nineteen meta-analyses met the eligibility criteria (six comparative RCT-only, seven comparative mixed-design, six single-arm syntheses). The comparative reviews re-analyze nine randomized trials comprising 2,974 unique randomized participants (2,718 in the six pivotal datasets); overlap was very high (CCA 44.4% across all randomized comparisons, 100% for EFS and OS). pCR was consistently increased (Rodrigues et al. RR 2.93, 95% CI 2.39–3.59; umbrella re-analysis of six randomized datasets, n=2,718: 17.7% vs 6.2%, Ioannidis Class II). Event-free survival improved (HR 0.76, 95% CI 0.66–0.86; Class III). OS improved in the RCT-only synthesis that incorporates the final MATTERHORN analysis (HR 0.82, 95% CI 0.71–0.94, P = .005; Class IV by the strict P-value threshold) and in MATTERHORN itself (HR 0.78, 95% CI 0.63–0.96, P = .021). Grade ≥3 treatment-related adverse events were not increased (RR 1.07, 95% CI 0.97–1.17), immune-related adverse events were (RR 3.18, 95% CI 2.48–4.08; Class II), and surgical morbidity, R0 resection and 30-day mortality were unchanged. MSI-H/dMMR status predicted the largest pCR gain (RR 5.18, 95% CI 2.43– 11.05); PD-L1 was an assay-dependent modifier and MATTERHORN’s benefit was independent of PD-L1. Conclusions Perioperative chemo-immunotherapy is a new standard of care for resectable G/GEJ adenocarcinoma with demonstrated pCR, EFS and OS benefit. The many meta-analyses are not independent corroboration but repeated re-analyses of a small shared set of trials, and their credibility is bounded by the primary trials, not by their number. Immune-related toxicity, not surgical morbidity, is the principal cost; MSI/MMR and assay-specified PD-L1 testing should inform, but not gate, patient selection. Systematic review registration https://www.crd.york.ac.uk/PROSPERO/view/CRD420261474260 , identifier CRD420261474260.