Oral squamous cell carcinoma (OSCC) remains a clinical challenge with limited prognostic tools. This study aimed to identify a novel autophagy-related gene signature for precise prognosis prediction and to elucidate its potential clinical utility in guiding risk stratification and personalized treatment strategies for OSCC patients.

A prognostic risk-score model was established based on autophagy-related differentially expressed genes (ARDEGs) in The Cancer Genome Atlas(TCGA)cohort and further validated in the independent GSE41613 cohort from the Gene Expression Omnibus (GEO) . Clinical relevance was further evaluated using multivariate Cox regression and functional enrichment. In addition, the expression levels of Glyceraldehyde-3-Phosphate Dehydrogenase(GAPDH), Fas-Associated protein with Death Domain(FADD), and Autophagy Related 5(ATG5) in OSCC tissues and normal control tissues were validated using immunohistochemistry (IHC).

A three-gene signature comprising GAPDH, FADD, and ATG5 was significantly associated with OSCC outcomes. High-risk patients demonstrated markedly reduced survival, and the risk score served as an independent prognostic factor.

A prognostic model for OSCC, comprising GAPDH, FADD and ATG5, was established in this study. The model effectively discriminated among patients with different risk levels, and its performance was preliminarily validated in an external cohort. As a simplified scoring tool derived from differentially expressed autophagy‑related genes, this model holds potential value for prognostic assessment and risk stratification in OSCC and warrants further investigation.