CMTM3 links inflammation-related transcriptional dysregulation to immune and therapeutic stratification in gastric cancer.

Gastric cancer (GC) shows strong biological heterogeneity and frequent disruption of inflammatory and metabolic programs, which affect tumor progression, immune escape, and treatment response. To identify biomarkers related to this inflammation-metabolism axis, we developed an integrative framework combining single-cell RNA sequencing, bulk transcriptomic cohorts, machine learning, immune and drug-response prediction, and in vitro validation. Inflammation-related gene modules in malignant cells were identified using hdWGCNA, and candidate prognostic genes were screened by CoxBoost and Random Survival Forest models. CMTM3 was prioritized as a candidate inflammation-associated gene and was further evaluated across independent GC cohorts. High CMTM3 expression was associated with poor survival, immune cell infiltration, increased immune checkpoint expression, and enrichment of multiple immunotherapy-related signatures, indicating an immune-infiltrated but functionally suppressed tumor state. Pathway analyses linked CMTM3 to inflammatory signaling and metabolic regulation, suggesting a role in coordinating tumor inflammatory and metabolic programs. Drug sensitivity prediction showed that low-CMTM3 tumors may be more responsive to selected chemotherapies and kinase inhibitors, whereas high-CMTM3 tumors may be more suitable for immune checkpoint-based treatment. Functional assays further showed that CMTM3 knockdown reduced GC cell proliferation and weakened macrophage recruitment. Overall, this study identifies CMTM3 as a prognostic and predictive biomarker in GC and provides a practical strategy for translating inflammation-metabolism-related molecular dysregulation into precision therapeutic stratification.