A six-gene prognostic model based on radioresistance-related genes in head and neck squamous cell carcinoma

Background Head and neck squamous cell carcinoma (HNSC) is a frequent malignant tumor. Identifying molecular biomarkers associated with radiosensitivity and prognosis is crucial to guide personalized treatment strategies for HNSC patients. Objectives This study aimed to identify radioresistance-related genes and establish a gene-based prognostic model for HNSC patients. Methods Differentially expressed genes (DEGs) between radiosensitive and radioresistant samples were identified based on the GEO and TCGA-HNSC datasets, and functional annotation was performed on these DEGs. From the DEGs, six key genes (JAK2, FAM172A, TXLNA, PLAA, MED18, MYO5A) were selected via univariate Cox, LASSO, and stepwise regression to construct a prognostic risk model. The model was evaluated using Kaplan-Meier (KM) analysis, ROC curves, repeated training-validation splits, and multivariate Cox regression analyses. Based on the risk model, patients were divided into high- and low-risk groups, and CIBERSORT, ESTIMATE, TIDE, and pRRophetic algorithms were used to compare immune microenvironment characteristics, immunotherapy-related features, and predicted drug sensitivity between the two risk groups. In vitro experiments were performed to further assess the effects of MYO5A silencing on HNSC cell phenotypes. Results DEGs were enriched in cell cycle and mitosis pathways. Patients in the high-risk group had significantly worse PFS and OS (p < 0.05), and repeated internal validation showed stable prognostic performance of the model. High-risk patients showed lower ImmuneScore and reduced infiltration of CD8 + T-cells and B-cells (p < 0.05). TIDE analysis demonstrated differences in immune exclusion and dysfunction scores between the two risk groups (p < 0.05), whereas the overall TIDE score did not differ significantly. Drug sensitivity prediction showed that low-risk patients had lower IC50 values for rapamycin (p < 0.05). MYO5A silencing markedly suppressed the malignant phenotype of HNSC cells (p < 0.05). Conclusions This study established a six-gene prognostic model based on genes related to radioresistance in HNSC. The model was associated with patient prognosis, immune microenvironment characteristics, and predicted drug sensitivity. These findings may provide a potential framework for prognostic stratification in HNSC, although further external validation and radiotherapy-focused functional studies are required.