Identification of Prognostic Alternative Splicing Signature in Gastric Cancer
Abstract Aberrant alternative splicing (AS) events serve as prognostic indicators in a large number of malignancies, whereas comprehensive analysis of prognostic AS in gastric cancer (GC) has not yet been understood. To identify prognostic AS events and clarify the function of the splicing variants in GC. RNA-Seq data, clinical information and percent spliced in (PSI) values were available in the cancer genome atlas (TCGA) and TCGA SpliceSeq data portal. A three-step regression method was conducted to screen prognostic AS events and construct multi-AS-based signatures. The associations between prognostic AS events and splicing factors were also investigated. We identified a total of 1318 survival related AS events in GC, Parent genes of which were implicated in numerous oncogenic pathways. The final prognostic signatures stratified by seven types of AS events or not stratified performed well in risk prediction for GC patients. Moreover, five signatures base on AA, AD, AT, ES and RI events function as independent prognostic indicators after multivariate adjustment of clinicalpathological variables. Splicing network also showed marked correlation between the expression of splicing factors and PSI value of AS events in GC patients. The AS derived signatures allow to predict GC prognosis and might serve as potential therapeutic target for GC.