Abstract
Gastric cancer (GC) is the fourth most common cancer type. "Disulfidptosis," a distinct form of cell death, is initiated through aberrant intracellular disulfide metabolism. Here, we identified various GC subtypes based on disulfidptosis-related genes (DRGs) and constructed a risk score model to identify relevant genes to help predict patient prognosis and guide treatment. We downloaded RNA sequencing (RNA-seq) data from the TCGA-STAD database, performed a difference analysis, and combined the data with GSE84437 to successfully perform an unsupervised clustering analysis based on DRGs and differentially expressed genes (DEGs). Risk-scoring models were established by screening prognosis-related DEGs. The GC samples were segregated into high-risk (HR) and low-risk (LR) groups according to their risk scores. We then evaluated the genes screened with the model in terms of prognosis, tumor, and immune cell infiltration. The response of patients with GC to immunological therapy was assessed using tumor mutational burden, microsatellite instability, and tumor immune dysfunction and exclusion scores. Using unsupervised cluster analysis, we identified two DRG clusters and two gene clusters that differed in prognosis and tumor microenvironment. A six-gene model was developed for risk score assessment. The LR group demonstrated superior performance compared to the HR group in terms of immunity, exhibiting greater sensitivity to immunotherapy. Thereafter, we selected the model gene GPC3 for single-gene analysis and verified it by experimental validation. The results demonstrated that GPC3 can serve as a standalone biomarker with promising clinical applicability in the prognostic prediction and clinical management of GC.