The DEpth of Endoscopic Perforation scale to assess intraoperative perforations during transurethral resection of bladder tumor: subgroup analysis of a randomized controlled trial

Abstract

Purpose: Bladder perforation (BP) is the most important intraoperative adverse event of transurethral resection of bladder tumor (TURBT). It is frequently underreported despite its impact on the postoperative course. There is no standardized classification of BP. The study aims to develop a classification of the depth of endoscopic bladder perforation during TURBT. Methods: This is a sub-analysis of a prospective randomized trial enrolling 248 patients submitted to en-bloc vs conventional TURBT from 03/2018 to 06/2021. The DEpth of Endoscopic Perforation (DEEP) scale is as follows: “0” visible muscular layer with no perivesical fat; “1” visible muscle fibers with spotted perivesical fat; “2” exposition of perivesical fat; “3” intraperitoneal perforation. Logistic and linear regression models were used to investigate predictors of high-grade perforations (DEEP 2–3) and to assess whether the DEEP scale independently predicted patients' postoperative outcomes. Results: A total of 146/248 (58.9%), 56/248 (22.6%), 41/248 (16.5%), 5/248 (2.0%) patients presented DEEP grade 0, 1, 2, and 3, respectively. Female gender [B coeff. 0.255 (95% CI 0.001–0.513); p = 0.05], tumor location [B coeff. 0.188 (0.026–0.339); p = 0.015], and obturator-nerve reflex [B coeff. 0.503 (0.148–0.857); p = 0.006] were independent predictors of DEEP. The scale predicted independently major complications [Odd Ratio (OR) 2.221 (1.098–4.495); p = 0.026], no post-operative chemotherapy intravesical instillation [OR 9.387 (2.434–36.200); p = 0.001], longer irrigation time [B coeff. 0.299 (0.166–0.441); p < 0.001] and hospital stay [B coeff. 0.315 (0.111–0.519); p = 0.003]. Conclusion: The DEEP scale provides a visual tool for grading bladder perforation during TURBT, which can help physicians standardize complication reporting and plan postoperative management accordingly.