A four-gene amino acid metabolism signature predicts prognosis, immune infiltration, and chemotherapy response in oral squamous cell carcinoma

Oral squamous cell carcinoma (OSCC) represents a prevailing oral malignancy worldwide. Amino acid metabolism exerts crucial functions in immune inhibition and drug resistance. This study aimed to identify and validate a clinically applicable AMRG-based prognostic signature for OSCC that integrates survival prediction with tumor immune microenvironment characterization and chemotherapy sensitivity stratification. Methods: RNA-seq and clinical data were retrieved from online databases to identify amino acid metabolism-related genes (AMRGs). Differentially expressed AMRGs (DEAMRGs) were identified through differential expression and weighted gene co-expression network analyses (WGCNA). A Least Absolute Shrinkage and Selection Operator (LASSO)-Cox model was then used to pinpoint DEAMRGs with prognostic potential, leading to the development of a prognostic risk model. The model was verified using Receiver Operating Characteristic (ROC), Kaplan-Meier (K-M), and risk curves. Validation of prognostic gene expression in OSCC was further conducted using immunohistochemical (IHC) staining and reverse transcription quantitative PCR (RT-qPCR), incorporating data from TCGA, GEO databases, and wet-lab experiments. Firstly, we identified 428 DEAMRGs, among which four candidates ( IMMP2L, C21orf33, PADI3, and ACAA1 ) with prognostic significance were utilized in risk model establishment. The model's accuracy in forecasting OSCC patient prognosis was confirmed through ROC and K-M curves. IHC confirmed differential protein expression of PADI3, C21orf33, and ACAA1 in OSCC tissues, while RT-qPCR validated mRNA dysregulation of C21orf33 and ACAA1. IMMP2L showed no significant change at mRNA or protein level in our validation cohort, suggesting context-dependent regulation. Our study identifies a novel 4-gene AMRG signature that robustly predicts OSCC prognosis, correlates with tumor immune microenvironment composition, and stratifies patients for differential chemotherapy sensitivity — offering a multi-functional tool for personalized risk assessment and treatment selection.