Construction of prognostic model of gastric cancer and prediction of drug sensitivity based on multi-omics analysis

WANG Yi; JIN Haijie; WANG Wenling; CHEN Weiwei

doi:10.3969/j.issn.1009-0460.2025.10.004

Abstract：Objective To identify causal proteins and develop a prognostic model for gastric cancer,with the ultimate goal of unraveling the biological basis of risk stratification and its implications for drug sensitivity.Methods Utilizing an integrative multi-omics strategy,this study systematically combined cross-cohort gastric cancer genome-wide association study and protein quantitative trait loci(pQTL)data for Mendelian randomization analysis.This approach was integrated with Kyoto Encyclopedia of Genes and Genomes and Gene Ontology functional annotations to identify key biological pathways.A prognostic model was constructed using Least Absolute Shrinkage and Selection Operator(LASSO)-Cox and multivariate Cox regression analyses,and a risk score was calculated.All patients were divided into high-risk and low-risk cohorts based on the median risk score.The model was trained and validated using the GSE62254,GSE15459,and TCGA-STAD datasets.Gene Set Enrichment Analysis,immune microenvironment analysis,and drug sensitivity prediction were used to reveal the molecular characteristics of the high-and low-risk groups.Results Gastric cancer risk-associated pQTLs were significantly enriched in immune response(cytokine receptor interactions,complement cascade)and metabolic pathways(phosphatidylinositol 3-kinase-protein kinase B,mitogen-activated protein kinase).LASSO-Cox and multivariate Cox regression analyses,based on causally associated pQTLs and prognostic genes,revealed that seven genes including inhibin subunit beta B,matrilin 3,TATA-box binding protein like 1,calmegin,chondroitin sulfate proteoglycan 4,lipopolysaccharide binding protein and ST3 beta-galactoside alpha-2,3-sialyltransferase 6 were included in the prognostic model.Kaplan-Meier analysis and area under the receiver operating characteristic curve analysis validated the model's robust predictive performance in both the training and validation sets.The high-risk group exhibited a proinflammatory microenvironment.Furthermore,this group exhibited significantly higher sensitivity to 10 drugs,including 5-fluorouracil,compared with the low-risk group.Conclusion The 7-gene prognostic model established in this study achieves survival risk stratification for gastric cancer,reveals that immune-metabolic imbalance is the core mechanism of risk stratification,and provides a potential biomarker basis for the selection of personalized drug treatment options for high-risk gastric cancer patients.