A novel 16-gene alternative mRNA splicing signature predicts tumor relapse and indicates immune activity in stage I–III hepatocellular carcinoma

Abstract

Background: Hepatocellular carcinoma (HCC) is a lethal disease with high relapse and dismal survival rates. Alternative splicing (AS) plays a crucial role in tumor progression. Herein, we aim to integratedly analyze the relapse-associated AS events and construct a signature predicting tumor relapse in stage I–III HCC. Methods: AS events of stage I–III HCC with tumor relapse or long-term relapse-free survival were profiled to identify the relapse-associated AS events. A splicing network was set up to analyze the correlation between the relapse-associated AS events and splicing factors. Cox regression analysis and receiver operating characteristic curve were performed to develop and validate the relapse-predictive AS signature. Single-sample gene set enrichment analysis (ssGSEA) and the ESTIMATE algorithm were used to assess the immune infiltration status of the HCC microenvironment between different risk subgroups. Unsupervised cluster analysis was conducted to assess the relationship between molecular subtypes and local immune status and clinicopathological features. Results: In total, 2441 ASs derived from 1634 mRNA were identified as relapse-associated AS events. By analyzing the proteins involved in the relapse-associated AS events, 1573 proteins with 11590 interactions were included in the protein–protein interaction (PPI) network. In total, 16 splicing factors and 61 relapse-associated AS events with 85 interactions were involved in the splicing network. The relevant genes involved in the PPI network and splicing network were also analyzed by Gene Ontology enrichment analysis. Finally, we established a robust 16-gene AS signature for predicting tumor relapse in stage I–III HCC with considerable AUC values in all of the training cohort, testing cohort, and entire cohort. The ssGSEA and ESTIMATE analyses showed that the AS signature was significantly associated with the immune status of the HCC microenvironment. Moreover, four molecular subgroups with distinguishing tumor relapse modes and local immune status were also revealed. Conclusion: Our study built a novel 16-gene AS signature that robustly predicts tumor relapse and indicates immune activity in stage I–III HCC, which may facilitate the deep mining of the mechanisms associated with tumor relapse and tumor immunity and the development of novel individualized treatment targets for HCC.