#0094

Prediction Nomograms of Prognosis in Bladder Urothelial Carcinoma Patients Receiving Radical Cystectomy-A Two-center Retrospective Research

J. Ji¹, Z. Wen², Y. Yao¹, L. Jiang¹, Q. Yang³, G. Zhang¹

¹The Affiliated Hospital of Qingdao University, Urology, Qingdao, China
²Peking University Cancer Hospital & Institute, Thoracic Surgery II, Beijing, China
³Qilu Hospital (Qingdao), Cheeloo College of Medicine, Shandong University, Urology, Qingdao, China

Introduction:

This study aims to develop and validate nomograms for predicting overall survival (OS), cancer-specific survival (CSS), and disease-free survival (DFS) in patients with resectable bladder urothelial carcinoma (BUC) after radical cystectomy (RC).

Material and methods:

Patients with BUC who received RC from the Affiliated Hospital of Qingdao University between January 2018 to December 2021 were assigned to the training cohort, while those from the Affiliated Hospital of Qingdao University between January 2016 to December 2017 and those from Qilu Hospital of Shandong University were assigned to the testing cohort. Demographic, pathological, imaging, and laboratory information were retrospectively collected. Univariate and multivariate COX regression analyses were employed to identify independent predictors for OS, CSS, and DFS in training cohort. The concordance index (C-index), area under the receiver operating characteristic (ROC) curve (AUC), corrected AUC after 1000-Bootstrap resampling with calibration curve, and DCA curve were used to assess the performance of nomograms in training cohort and testing cohort.

Results:

393 and 156 patients were enrolled in training cohort and testing cohort, respectively. Multivariate analyses identified age, tumor size, lymph node metastasis (LNM), lymphovascular invasion (LVI), urea nitrogen, creatinine, and albumin/fibrinogen ratio (AFR) as independent predictors for OS; tumor size, LNM, LVI, urea nitrogen, and AFR as independent predictors for CSS; and LNM with LVI as independent predictors for DFS. The OS and CSS nomograms showed high prediction accuracy in C-indexes and ROC curves, reliability in calibration curves with corrected AUCs, and clinical application value in both cohorts. The DFS nomogram showed high accuracy with little changed corrected AUCs but limited stability in calibration curves in both cohorts.