Comprehensive identification and experimental validation of programmed cell death-anoikis genes in oral squamous cell carcinoma through machine learning approaches

This study aimed to identify anoikis-related genes (ARGs) associated with Oral squamous cell carcinoma (OSCC) progression and prognosis. We obtained RNA-sequencing data from TCGA (268 OSCC samples and 19 normal samples) and GEO validation cohort (GSE41613, 97 patients). Candidate anoikis-related genes were identified at the intersection of differentially expressed genes, WGCNA modules, and 338 known ARGs from literature. A comprehensive machine learning framework incorporating 117 algorithm combinations was developed to construct a prognostic risk model, with performance evaluated using concordance index and time-dependent ROC curves. Immune infiltration analysis was performed using the xCell algorithm. Functional validation was conducted through siRNA-mediated knockdown of risk genes and overexpression of protective genes in CAL27 cells, with transfection efficiency confirmed by qRT-PCR and cell viability assessed using CCK-8 assays. Through integrated bioinformatic analysis of TCGA and GEO datasets, we identified 53 candidate genes at the intersection of differentially expressed genes, weighted gene co-expression network analysis modules, and known ARGs. Functional enrichment analysis revealed significant involvement in biological processes related to anoikis regulation, cell-matrix adhesion, and cancer-related pathways. Using 117 machine learning algorithm combinations, we established a prognostic risk model based on eight ARGs (CTTN, PLAU, ITGA5, MET, TNFRSF12A, PLK1, PDK4, and SESN1), which effectively stratified OSCC patients into high and low-risk groups with significantly different survival outcomes in both training and validation cohorts. Multivariate Cox regression confirmed the risk score as an independent prognostic factor alongside T and N stages. In vitro experiments demonstrated that knockdown of risk genes and overexpression of protective genes significantly reduced cell viability in CAL27 cells. This study provides novel insights into the molecular mechanisms of anoikis resistance in OSCC and potential therapeutic targets for improving patient outcomes.