Predictive factors for final pathologic ureteral sections on 700 radical cystectomy specimens: Implications for intraoperative frozen section decision-making
Résumé
Purpose: To identify preoperative predictive factors for final ureteral section invasion after radical cystectomy (RC) and to validate significant factors on an external independent cohort. Material and methods: We retrospectively reviewed data of all consecutive RC performed for bladder cancer in 2 high-volume institutions. Clinical, pathological, and follow-up data were collected prospectively and reviewed retrospectively. Pathological evaluation was performed by 2 well-trained uropathologists in each center. Logistic regression analyses were performed to identify predictive factors for final ureteral sections involvement. Significant factors in cohort A were validated in cohort B. Receiver operating curve and area under curve were modeled to evaluate predictive accuracy of the markers. Results: A total of 441 RC were performed in center A and 307 RC were performed in center B. Mean follow-ups were 36.2 and 38.1 months, respectively. Invasion of the final ureteral section was observed on 5.5% of patients in cohort A and 4.8% of patients in cohort B. In cohort A, multivariable logistic regression identified preoperative hydronephrosis on computed tomography scan (odds ratio [OR] = 4.9, P = 0.004) and presence of Carcinoma in situ (CIS, OR = 3.9, P = 0.01) as the only factors associated with ureteral sections positivity. In cohort B, hydronephrosis and CIS were both associated with ureteral sections positivity in univariable analysis. In multivariable analysis, only hydronephrosis remained significant (OR = 5.9, P = 0.01). Predictive accuracy of hydronephrosis and CIS combined in 1 variable was 0.72. Conclusion: Hydronephrosis and bladder CIS have good accuracy in predicting ureteral sections positivity after RC. In the presence of those factors, ureteral frozen sections should be performed. (C) 2017 Elsevier Inc. All rights reserved.