Investigation of an FGFR-Signaling-Related Prognostic Model and Immune Landscape in Head and Neck Squamous Cell Carcinoma

Background: There is accumulating evidence on the clinical importance of the fibroblast growth factor receptor (FGFR) signal, hypoxia, and glycolysis in the immune microenvironment of head and neck squamous cell carcinoma (HNSCC), yet reliable prognostic signatures based on the combination of the fibrosis signal, hypoxia, and glycolysis have not been systematically investigated. Herein, we are committed to establish a fibrosis–hypoxia–glycolysis–related prediction model for the prognosis and related immune infiltration of HNSCC. Methods: Fibrotic signal status was estimated with microarray data of a discovery cohort from the TCGA database using the UMAP algorithm. Hypoxia, glycolysis, and immune-cell infiltration scores were imputed using the ssGSEA algorithm. Cox regression with the LASSO method was applied to define prognostic genes and develop a fibrosis–hypoxia–glycolysis–related gene signature. Immunohistochemistry (IHC) was conducted to identify the expression of specific genes in the prognostic model. Protein expression of several signature genes was evaluated in HPA. An independent cohort from the GEO database was used for external validation. Another scRNA-seq data set was used to clarify the related immune infiltration of HNSCC. Results: Six genes, including AREG, THBS1, SEMA3C, ANO1, IGHG2, and EPHX3, were identified to construct a prognostic model for risk stratification, which was mostly validated in the independent cohort. Multivariate analysis revealed that risk score calculated by our prognostic model was identified as an independent adverse prognostic factor (p < .001). Activated B cells, immature B cells, activated CD4(+) T cells, activated CD8(+) T cells, effector memory CD8(+) T cells, MDSCs, and mast cells were identified as key immune cells between high- and low-risk groups. IHC results showed that the expression of SEMA3C, IGHG2 were slightly higher in HNSCC tissue than normal head and neck squamous cell tissue. THBS1, ANO1, and EPHX3 were verified by IHC in HPA. By using single-cell analysis, FGFR-related genes and highly expressed DEGs in low-survival patients were more active in monocytes than in other immune cells. Conclusion: A fibrosis–hypoxia–glycolysis–related prediction model provides risk estimation for better prognoses to patients diagnosed with HNSCC.