Identification of tumour immune microenvironment-related alternative splicing events for the prognostication of pancreatic adenocarcinoma

Abstract

Abstract Purpose Pancreatic adenocarcinoma (PAAD) is characterized by low antitumour immune cell infiltration in an immunosuppressive microenvironment. This study aimed to systematically explore the impact on prognostic alternative splicing events (ASs) of tumour immune microenvironment (TIME) in PAAD. Methods The ESTIMATE algorithm was implemented to compute the stromal/immune-related scores of each PAAD patient, followed by Kaplan–Meier (KM) survival analysis of patients with different scores grouped by X-tile software. TIME-related differentially expressed ASs (DEASs) were determined and evaluated through functional annotation analysis. In addition, Cox analyses were implemented to construct a TIME-related signature and an AS clinical nomogram. Moreover, comprehensive analyses, including gene set enrichment analysis (GSEA), immune infiltration, immune checkpoint gene expression, and tumour mutation were performed between the two risk groups to understand the potential mechanisms. Finally, Cytoscape was implemented to illuminate the AS-splicing factor (SF) regulatory network. Results A total of 437 TIME-related DEASs significantly related to PAAD tumorigenesis and the formation of the TIME were identified. Additionally, a robust TIME-related prognostic signature based on seven DEASs was generated, and an AS clinical nomogram combining the signature and four clinical predictors also exhibited prominent discrimination by ROC (0.762 ~ 0.804) and calibration curves. More importantly, the fractions of CD8 T cells, regulatory T cells and activated memory CD4 T cells were lower, and the expression of four immune checkpoints—PD-L1, CD47, CD276, and PVR—was obviously higher in high-risk patients. Finally, functional analysis and tumour mutations revealed that aberrant immune signatures and activated carcinogenic pathways in high-risk patients may be the cause of the poor prognosis. Conclusion We extracted a list of DEASs associated with the TIME through the ESTIMATE algorithm and constructed a prognostic signature on the basis of seven DEASs to predict the prognosis of PAAD patients, which may guide advanced decision-making for personalized precision intervention.