Alternative splicing (AS) is a molecular event that drives protein diversity through the generation of multiple mRNA isoforms. Growing evidence demonstrates that dysregulation of AS is associated with tumorigenesis. However, an integrated analysis in identifying the AS biomarkers attributed to esophageal carcinoma (ESCA) is largely unexplored.
AS percent-splice-in (PSI) data were obtained from the TCGA SpliceSeq database. Univariate and multivariate Cox regression analysis was successively performed to identify the overall survival (OS)-associated AS events, followed by the construction of AS predictor through different splicing patterns. Then, a nomogram that combines the final AS predictor and clinicopathological characteristics was established. Finally, a splicing regulatory network was created according to the correlation between the AS events and the splicing factors (SF).
We identified a total of 2389 AS events with the potential to be used as prognostic markers that are associated with the OS of ESCA patients. Based on splicing patterns, we then built eight AS predictors that are highly capable in distinguishing high- and low-risk patients, and in predicting ESCA prognosis. Notably, the area under curve (AUC) value for the exon skip (ES) prognostic predictor was shown to reach a score of 0.885, indicating that ES has the highest prediction strength in predicting ESCA prognosis. In addition, a nomogram that comprises the pathological stage and risk group was shown to be highly efficient in predicting the survival possibility of ESCA patients. Lastly, the splicing correlation network analysis revealed the opposite roles of splicing factors (SFs) in ESCA.
In this study, the AS events may provide reliable biomarkers for the prognosis of ESCA. The splicing correlation networks could provide new insights in the identification of potential regulatory mechanisms during the ESCA development.