ABSTRACT
Objective:To analyze the alternative splicing (AS) events of patients with thyroid carcinoma (THYC) and explore the correlation between AS events and the prognosis of THYC.Methods:The clinical data and the Percent Splice In (PSI) value of AS events of THYC were downloaded from The Cancer Gene Atlas (TCGA) database and the TCGA SpliceSeq database respectively. The occurrence of seven kinds of AS events including AA, AD, AF, AP, ME, ES and RI in THYC was investigated and the matrix of AS events and survival data was constructed. Univariate Cox regression analysis was used to screen AS events related to prognosis of THYC. To avoid over-fitting, the least absolute shrinkage and selection operator (LASSO) regression analysis was performed. Then Multivariate Cox regression analysis was used to construct prognosis model. Kaplan-Meier curve and receiver operating characteristic (ROC) curve were performed to evaluate the prognosis ability of the risk model. We also used Pearson correlation analysis to select splicing factors (SF) which were correlated with survival associated AS events. Above SF genes were enrolled to gene ontology (GO) enrichment and KEGG pathway analysis.Results:A total of 10 447 genes and 45 150 AS events in 507 THYC patients were found in the present study. Among them, ES was the main type (38.84%) and ME was the type with the least frequency (0.51%). Totally 1 842 AS events associated with prognosis of THYC patients were identified. Three AS events including USHBP1-48249-AA、CACNB1-40626-AT and BEX5-89679-AP were selected to construct the prognosis model. The risk score of 0.807 was indicated as the best cut-off value of prognosis model. The patients were divided into high-risk group (240 cases) and low-risk group (241 cases) based on the risk score. The results demonstrated that the risk model could be used as a valuable prognostic factor for THYC ( P<0.001, AUC=0.929). The SF-AS network was constructed and several SF genes, including CDK12, RBM25, DDX39B, SRRM2 and DDX46 were identified as hub genes. Conclusions:The risk model based on 3-AS events was valuable prognosis predictor of THYC. The SF-AS network provided new insight for the exploration of tumorigenesis and development of THYC.