Alternative Splicing in Glioblastoma and its Clinical Implication in Outcome Prediction.

Background and objective: Alternative splicing (AS) offers an important mechanism to form protein polymorphism. A growing body of evidence indicates the correlation between splicing abnormality and carcinoma. Nevertheless, an overall analysis of AS signatures in glioblastoma (GBM) is absent and urgently needed.

Methods: TCGA SpliceSea data was used to evaluate the AS profiles and further classified into different AS events. The survival analysis was based on these AS events, and AS-related genes were identified and performed with enrichment analysis. At last, the splicing factor-AS regulatory network was established in Cytoscape.

Results: Eight hundred forty-two splicing events were confirmed as prognostic molecular events in GBM. Furthermore, the final prognostic signature constructed by seven AS events gave good result with an area under the curve (AUC) of receiver operating characteristic (ROC) curve up to 0.935 for five years, showing high potency in predicting patients' outcome. We built the splicing regulatory network to show the internal relationship of splicing events in GBM. PC4 and SFRS1 interacting protein 1 (PSIP1) and histone H4 acetylation may play a significant part in the prognosis induced by splicing events.

Conclusion: In our study, a high-efficiency prognostic prediction model was built for GBM patients based on AS events, which could become potential prognostic biomarkers for GBM. Meanwhile, PSIP1 may be a critical target for pharmaceutical treatment.