Analysis of relapse-associated alternative mRNA splicing and construction of a prognostic signature predicting relapse in I–III colon cancer
Abstract Background : The literature depicting the effects of alternative splicing (AS) events on relapse of colon cancer is little and there is no signature based on the alternative splicing. Methods : The bioinformatic analysis was performed based on data of The Cancer Genome Atlas (TCGA) to identify the relapse-associated ASs, the potential interactions were further analyzed and a robust signature was built after univariate Cox regression, LASSO Cox regression, and multivariate Cox regression analysis to predict the relapse in I–III colon cancer. Molecular subtypes was identified based on the signature. Results : We identified 1912 ASs of 1384 mRNA, based on the relapse-associated ASs, we constructed the network of protein-protein interactions (PPI) and ASs-splicing factors (SF) interactions. 1294 of proteins with 7396 interactions were included in the PPI network. 14 SFs combined with 78 relapse-associated ASs were included in the AS-SF network. We finally built a robust signature to predict the relapse of I–III colon cancer with a considerable AUC value in both the training group and the test group (0.857,0.839). Based on the ASs involved in the signature, samples were classified into 4 molecular subgroups distinguishing the relapse rate in diverse groups. Conclusion : Our study provides a profile of relapse-associated ASs in I–III colon cancer and build a robust signature to predict the relapse of I–III colon cancer patients and further classify the patients into 4 molecular subtypes.