Development of a predictive model for optimizing the selection of patients for second transurethral resection bladder

Abstract

Background: Transurethral resection of bladder (TURB) is considered the gold standard treatment of non-muscle invasive bladder tumor (NMIBT). The urological clinical practice guidelines recommend second-TURB (re-TURB), in certain situations, to achieve a complete resection and appropriately stratify the tumor. Objective: To design a predictive model of residual tumor in re-TURB, to optimize patient selection and avoid unnecessary surgeries. Methods: Retrospective analysis of 413 NMIBT with a macroscopically complete TURB and identification of the muscle layer, with subsequent re-TURB (2-6 weeks), according to EAU Clinical Guidelines criteria, from January 2010 to December 2021. We have identified predictive variables of residual tumor through univariate and multivariate analysis using logistic regression. The evaluation of the exactitude of the predictive model has been made using AUC (area under curve-ROC curve). Results: Median age in the series was 74 (38-91) years old, and 86.7% were male. Tumors that required second-TURB were mostly primary (64.6%), stage T1 (84.5%), high grade (82.6%), multifocal (54.8%) and with a size less than 3 cm (73.1%). Residual tumor was found in 28.1% of re-TURB. The independent predictors factors of residual tumor identified were: recurrence tumor (OR 1.87; IQ 1.14-3.06, p=0.01) and multifocality tumors (OR 2.11; IQ 1.31-3.4; p=0.002). High-grade shows a trend to statistical signification as risk factor (p=0.07) and early administration of mitomycin-C behaved as an independent protector factor (OR 0.40; IQ 0.3-0.81; p=0.006). The predictive model shows an AUC=0.7 (IQ 0.62-0.73; p=0.0001). Conclusion: Our predictive model evaluates the probability of finding residual tumor in second-TURB with a 70% accuracy. This way, we could optimize the selection of patients in low risk of residual tumor, thus avoiding unnecessary surgeries.