External Validation of A Decision Tree Model forPredicting Cancer-Specific Mortality After RadicalCystectomy: A Multi-Center Study in South-East Spain

Background: Accurate prognostic tools are essential for guiding therapeutic decisions and follow-up strategies in patients with urothelial bladder cancer undergoing radical cystectomy. Although decision tree models have shown promise in predicting cancer-specific outcomes, their clinical utility depends on external validation across diverse populations. This study aimed to externally validate a previously published decision tree model for predicting cancer-specific mortality using an independent multi-centre cohort from south-east Spain. Methods: A multi-centre retrospective cohort study was conducted across five university hospitals in south-east Spain. A total of 553 patients with histologically confirmed urothelial bladder cancer who underwent standard radical cystectomy between 2004 and 2025 were included. A decision tree model originally developed using the Surveillance, Epidemiology, and End Results (SEER) database to predict cancer-specific survival was externally validated. The development and validation cohorts were comparatively analysed using chi-square tests and analysis of variance. Model discrimination was assessed using the concordance index (C-index) and calibration plots. Clinical utility was evaluated using decision curve analysis (DCA). Results: The C-index was 0.733 in the Spanish validation cohort, indicating good discriminative performance. DCA demonstrated a net benefit of using the model over a “treat-all” strategy at threshold probabilities >25%. Despite demographic and staging differences between the validation and original cohorts, the model maintained its predictive accuracy. Conclusions: This study provides formal external validation of a decision tree model for predicting cancer-specific mortality after radical cystectomy, demonstrating robust performance and supporting its use in clinical settings in south-east Spain.