Abstract
Gastric cancer (GC) remains a leading cause of cancer mortality, yet the causal roles of microRNAs (miRNAs) in its pathogenesis are poorly characterized. While observational studies implicate miRNAs in GC progression, confounding biases and tissue-specific limitations hinder causal inference and clinical translation. We conducted a 2-sample Mendelian randomization (MR) analysis using genetic instruments derived from plasma miRNA expression quantitative trait loci (eQTLs). Summary-level data for miRNA-eQTLs were obtained from a study by Huan et al. (involving 5239 individuals and 280 miRNAs), while genetic associations with GC were sourced from 3 independent genome-wide association studies (ebi-a-GCST90018849, ebi-a-GCST90018629, and bbj-a-119) accessed via the IEU OpenGWAS Project. Instrumental variables were constructed using miRNA-eQTLs that reached significance at a false discovery rate (FDR < 0.1. Causal estimates were primarily generated using inverse-variance weighted regression, supplemented by MR-Egger regression to assess and adjust for potential pleiotropy. Sensitivity analyses, including leave-one-out validation, were performed to evaluate the robustness of the findings. Experimentally validated targets were analyzed for differential expression, prognostic relevance, and somatic mutations. Functional enrichment and pan-cancer analyses were conducted to delineate oncogenic mechanisms. MR analysis revealed 5 plasma miRNAs with consistent causal effects on GC risk: hsa-miR-127-3p, hsa-miR-370-3p, hsa-miR-382-5p, hsa-miR-409-3p, and hsa-miR-654-5p. All 5 miRNAs conferred increased risk (ORs 1.021-1.037, all ≤0.0025) across the 3 cohorts (ebi-a-GCST90018849, ebi-a-GCST90018629, bbj-a-119). These miRNAs collectively targeted 549 genes, of which 76 were differentially expressed in GC tissues. Seventeen dysregulated targets showed prognostic significance, with enrichment in immune regulation (T/B cell receptor signaling) and cancer pathways. In GC, miR-409-3p overexpression independently predicted poor survival (H = 1.55, P = .0098) and inversely correlated with multiple targets (XKR4, F2, ATAD5, GNAL, GDNF, UNC13A, and ELL2). Pan-cancer analysis revealed oncogenic roles for causal miRNAs in 16 malignancies, with miR-409-3p showing GC-specific prognostic significance. This MR study establishes plasma miRNAs as causal mediators of gastric carcinogenesis, with miR-409-3p emerging as a key prognostic biomarker. The identified miRNA-target networks highlight actionable pathways for therapeutic intervention, bridging genetic epidemiology with functional genomics in GC precision oncology.