Establishment of a prognostic prediction model for prostate cancer patients with androgen deprivation therapy based on alternative splicing events and splice factors / 肿瘤研究与临床

ABSTRACT

Objective:To explore the predictive effect of alternative splicing events and splicing factors on the prognosis of prostate cancer patients with androgen deprivation therapy (ADT).Methods:The relevant clinical and transcriptome information of 60 prostate cancer patients with ADT were downloaded from TCGA database, and the alternative splicing events information was obtained. LASSO and Cox regression models were used to screen the relevant alternative splicing events and splice factors through R language package. The related model was established, and the independent prognostic analysis and related regulatory network analysis were performed.Results:A total of 44 070 associated alternate splicing events were identified in prostate cancer patients, including 3 525 alternate acceptor sites (AA), 3 101 alternate donor sites (AD), 9 035 alternate promoters (AP), 8 663 alternate terminators (AT), 16 772 exon skipping (ES), 228 exon mutual exclusion (ME), and 2 747 retained intron (RI). Through univariate Cox proportional hazard regression model, 1 349 alternate splicing events were identified as disease-free survival-related splicing events (DFS-SE) related to ADT, including 145 AA events, 102 AD events, 243 AP events, 189 AT events, 557 ES events, 6 ME events, and 107 RI events. Multivariate Cox regression analysis show that after adding clinical related information, ME events (ANK3|11852|ME, TCF7L2|151705|ME, UBR2|127390|ME and SLC39A14|140283|ME) may be independent prognostic factors for predicting ADT resistance ( HR = 1.398, 95% CI 1.156-1.689, P<0.01). Conclusion:The combination of ME-type alternate splicing events and related clinical data has certain significances in predicting the effectiveness of ADT in prostate cancer patients, and plays an important role in the research on the resistance of prostate cancer patients to ADT.