Abstract
BACKGROUND: Previous studies have shown that alternative splicing (AS) plays a key role in carcinogenesis and prognosis of cancer. However, systematic profiles of AS signatures in head and neck cancer (HNC) have not yet been reported. METHODS: In this study, AS data, RNA-Seq data, and corresponding clinicopathological information of 489 HNC patients were downloaded from The Cancer Genome Atlas. Univariate and multivariate Cox regression analyses were performed to screen for survival-associated AS events. Functional and pathway enrichment analysis was also performed. The prognostic models and splicing networks were constructed using integrated bioinformatics analysis tools. RESULTS: Among the 42,849 alternating splicing events identified in 10,121 genes, 5,165 survival-associated AS events in 2,419 genes were observed in univariate Cox regression analysis. Among the seven types, alternate terminator events were the most powerful prognostic factors. Multivariate Cox analysis was then used to screen for the AS genes with prognostic value. Four candidate genes (TPM1, CLASRP, PRRC1, and DNASE1L1) were found to be independent prognostic factors for HNC patients. A prognostic prediction model was built based on the four genes. The area under the receiver operating characteristic risk score curve for predicting the survival status of HNC patients was 0.704. In addition, splicing interaction network indicated that the splicing factors have significant functions in HNC. CONCLUSION: A comprehensive analysis of AS events in HNC was performed. A powerful prognostic predictor for HNC patients was established based on AS events could.