Abstract
BACKGROUND: Programmed death-ligand 1 (PD-L1) is widely used as a predictive biomarker for response to immune checkpoint inhibitors in gastric cancer (GC); however, its independent prognostic value for overall survival remains controversial. Differences in PD-L1 detection methods, scoring systems, and tumor immune microenvironment characteristics may contribute to inconsistent findings across studies. Therefore, this systematic review and meta-analysis aimed to evaluate the association between PD-L1 expression and overall survival in patients with GC. METHODS: A systematic search of PubMed, Scopus, MEDLINE, and Web of Science was conducted to identify eligible studies published between January 2018 and December 2025. Cohort studies reporting hazard ratios (HRs) for overall survival in relation to PD-L1 expression assessed by immunohistochemistry were included. A random-effects meta-analysis was performed to estimate pooled effect sizes. Heterogeneity was assessed using the I(2) statistic, and subgroup analyses were conducted according to PD-L1 assessment methods. Meta-regression analyses were performed to explore potential sources of heterogeneity. Study quality was evaluated using the Newcastle–Ottawa Scale, and the certainty of evidence was assessed using the GRADE approach. RESULTS: Seven retrospective cohort studies comprising 1,993 patients with GC were included in the analysis. PD-L1 expression was significantly associated with overall survival (HR = 1.70, 95% CI: 1.13–2.55), although the direction of the prognostic effect varied across studies (I(2) = 69.1%). Studies using the combined positive score (CPS ≥ 1) demonstrated more consistent prognostic associations compared with those employing semi-quantitative scoring systems. Meta-regression analyses did not identify significant modifying effects of publication year, mean age, or sex distribution on the pooled estimates. The overall certainty of evidence was rated as moderate. CONCLUSION: PD-L1 expression is associated with overall survival in GC; however, its prognostic implications appear to be context-dependent and influenced by methodological differences in PD-L1 assessment and tumor immune microenvironment characteristics. Standardization of PD-L1 evaluation methods and integration with molecular tumor subtypes may improve its prognostic utility in GC. TRIAL REGISTRATION: https://www.crd.york.ac.uk/PROSPERO/view/CRD420251249939, identifier CRD420251249939.