Systematic analysis and prediction model construction of alternative splicing events in hepatocellular carcinoma: a study on the basis of large-scale spliceseq data from The Cancer Genome Atlas.

Growing evidence showed that alternative splicing (AS) event is significantly related to tumor occurrence and progress. This study was performed to make a systematic analysis of AS events and constructed a robust prediction model of hepatocellular carcinoma (HCC). The clinical information and the genes expression profile data of 335 HCC patients were collected from The Cancer Genome Atlas (TCGA). Information of seven types AS events were collected from the TCGA SpliceSeq database. Overall survival (OS) related AS events and splicing factors (SFs) were identified using univariate Cox regression analysis. The corresponding genes of OS-related AS events were sent for gene network analysis and functional enrichment analysis. Optimal OS-related AS events were selected by LASSO regression to construct prediction model using multivariate Cox regression analysis. Prognostic value of the prediction models were assessed by receiver operating characteristic (ROC) curve and KaplanMeir survival analysis. The relationship between the Percent Spliced In (PSI) value of OS-related AS events and SFs expression were analyzed using Spearman correlation analysis. And the regulation network was generated by Cytoscape. A total of 34,163 AS events were identified, which consist of 3,482 OS-related AS events. UBB, UBE2D3, SF3A1 were the hub genes in the gene network of the top 800 OS-related AS events. The area under the curve (AUC) of the final prediction model based on seven types OS-related AS events was 0.878, 0.843, 0.821 in 1, 3, 5 years, respectively. Upon multivariate analysis, risk score (All) served as the risk factor to independently predict OS for HCC patients. SFs HNRNPH3 and HNRNPL were overexpressed in tumor samples and were signifcantly associated with the OS of HCC patients. The regulation network showed prominent correlation between the expression of SFs and OS-related AS events in HCC patients. The final prediction model performs well in predicting the prognosis of HCC patients. And the findings in this study improve our understanding of the association between AS events and HCC.