Identification the Tumor Mechanics-Related Biomarkers in Gastric Cancer Patients Based on Bioinformatics and Machine Learning.

Background: Gastric cancer (GC) remains a major cause of cancer related mortality worldwide. Tumor mechanics, reflecting the physical and mechanical properties that influence tumor cell behavior and the tumor microenvironment (TME), play important roles in cancer progression. However, the prognostic relevance of tumor mechanics-related genes (MRGs) in GC remains unclear. Methods: GC datasets from TCGA and GEO were analyzed to identify differentially expressed genes (DEGs). WGCNA was conducted to identify MRGs-related modules. Univariate Cox regression and three machine learning algorithms were applied to screen prognostic genes and construct a prognostic model. Pan-cancer analysis, immune infiltration, tumor mutation burden (TMB), immunophenotypic score (IPS), and somatic mutation analyses were performed to explore TME characteristics. Additionally, drug sensitivity and ceRNA network analyses were conducted. Finally, the prognostic genes were verified using RT-PCR. Results: Eight mechanics-related genes (SERPINE1, CYP1B1, LOX, HEYL, VCAN, IGFBP7, TWIST2, and ATP1B2) were identified through integrated computational analysis. The resulting model demonstrated prognostic potential for 2-, 3-, and 5-year survival prediction. High-risk patients exhibited increased immunosuppressive infiltration compared with low-risk patients. Drug sensitivity analysis revealed significant differences in therapeutic responses across risk groups. Finally, the differential expression of several prognostic genes was preliminarily confirmed by RT-PCR in limited tissue samples. Conclusion: This study identifies eight tumor mechanics-related genes as prognostic biomarkers for GC through comprehensive bioinformatic analyses. These findings may provide preliminary insights into prognostic assessment and targeted therapy for GC, although further validation with larger sample sizes is required to substantiate their clinical applicability.