Abstract
BACKGROUND: Laryngeal cancer is a prevalent and aggressive type of head and neck malignancy that often presents with a poor prognosis. Its high incidence and significant mortality rate pose a substantial challenge in clinical management, but the prognostic biomarkers and the mechanisms of progression are poorly understood. We sought to identify the prognostic biomarkers and their effects in the progression of laryngeal cancer. METHODS: The gene expression profiles and clinic-pathological data of laryngeal cancer patients were retrieved from the The Cancer Genome Atlas (TCGA) database (N=128) and Gene Expression Omnibus (GEO) database (GSE27020, N=109; GSE65858, N=48). Differential expression analysis was conducted to evaluate the differentially expressed genes between laryngeal tumors and normal laryngeal tissues, and the significantly differentially expressed genes were selected to explore the altered biological processes. Machine learning methods [least absolute shrinkage and selection operator (LASSO) and random forest] were performed to identify the significant prognostic genes in laryngeal cancer, and Kaplan-Meier survival analysis and receiver operating characteristic (ROC) curves were applied to validate their prognostic values. Their correlation with oncogenic pathways and immune infiltrations was enquired using gene set variation analysis (GSVA) analysis. The colony-forming, 5-ethynyl-2'-deoxyuridine (EdU) staining, and transwell assays were applied to evaluate its role in affecting progression of laryngeal cancer. RESULTS: Functional enrichment analysis revealed up-regulated Epithelial-to-mesenchymal transition, integrin signaling, and lipid metabolism in laryngeal cancer. Machine learning methods identified the significant prognostic genes in laryngeal cancer, including FRMD5, CLDN23, PLIN5, and THBS1. The prognostic significance of THBS1 was further validated in independent datasets. Bioinformatics analyses suggested that the high expression of THBS1 shapes an immune suppressive tumor microenvironment in laryngeal cancer. The oncogenic role of THBS1 and its therapeutic vulnerability were validated by in vitro experiments using human laryngeal cancer cell lines. CONCLUSIONS: Our study supports THBS1 as a potential prognostic predictor with therapeutic vulnerability for laryngeal cancer, warranting further clinical validation.