Abstract
Head and neck squamous cell carcinoma (HNSC) is a common cancer worldwide with poor prognosis. Current treatment methods have limited effect on improving the prognosis of patients with HNSC. Differentially expressed PANoptosis-related genes in HNSC were identified from the TCGA using limma and WGCNA. A prognostic model was established using univariate and multivariate Cox regression analyses and machine learning, and its performance was evaluated using Kaplan-Meier and receiver operating characteristic curves. SNP data was analyzed using maftools package. Immune analysis was performed using IOBR package and TIDE website. The scRNA data was analyzed using Seurat and cellchat package. The expression of hub genes was validated in vitro. The prognostic model comprising 5 hub PANoptosis-related genes (AIFM1, AKT3, CDKN2A, EGFR, IL1RAP) accurately predicted patient outcomes, with the high-risk group exhibiting poorer survival. mRNA expression levels of all 5 hub genes were elevated in the tumor cells, but only AIFM1, AKT3 and IL1RAP's protein expression were higher in tumor tissues. Additionally, high expression of AIFM1, AKT3, EGFR, IL1RAP and low expression of CDKN2A indicated poor prognosis of HNSC patients. The decreasing levels of CD4 T cells, CD8 T cells and M1 macrophages were observed in high-risk groups. There was a significant difference of 5-fluorouracil in low and high-risk groups. scRNA analysis exhibited that TNF pathway was important in the interaction between macrophages and T cells. We identified 5 hub genes and constructed a great risk model for the prognosis of HNSC. The immune cells may influence the HNSC malignant through TNF signal pathway.