Multi-omics and Mendelian randomization reveal ABC transporter alternative splicing as a pan-cancer prognostic marker and therapeutic target

Abstract Purpose The ABC transporter gene family, pivotal in multidrug resistance and tumor progression, exhibits extensive alternative splicing (AS) events with potential clinical implications in pan-cancer prognosis and therapy. This study investigates the causal relationship between ABC transporter AS events and cancer risk, integrating multi-omics analysis with Mendelian randomization (MR) and Bayesian Weighted MR (BWMR) to explore their role in tumor immunology and prognosis. Methods Transcriptomic data from TCGA and GWAS were analyzed to identify AS events in 114 ABC transporter genes across 33 cancers. A multifactorial Cox prognostic model was constructed, validated via LASSO regression, and evaluated for immune infiltration using ssGSEA and TIMER. SF-AS regulatory networks were mapped via Cytoscape, and causal inferences were tested using MR and BWMR. Results Eleven AS genes (e.g., PSMA4, PSMD7) were linked to pan-cancer prognosis, with high-risk patients showing elevated immune infiltration (CD8 + T cells, B cells) and checkpoint gene expression (CD274, CTLA4). The SF-AS network revealed 50 splicing factors regulating 17 AS events. MR confirmed UBA52 (protective) and ABCB4 (risk) as causally associated with pan-cancer, corroborated by BWMR. The nomogram model predicted survival with moderate accuracy (AUC: 0.695–0.756). Conclusion This study establishes a causal link between ABC transporter AS events and pan-cancer risk, highlights their interplay with immune checkpoints, and proposes novel biomarkers for prognosis and immunotherapy. MR/BWMR validation underscores the potential of AS signals in guiding targeted therapies.