Nomogram for predicting survival of postcystectomy recurrent urothelial carcinoma of the bladder.

ABSTRACT

PURPOSE: We aimed to identify prognostic clinicopathological factors and to create a nomogram able to predict overall survival (OS) in recurrent urothelial carcinoma of the bladder (UCB) after radical cystectomy (RC). MATERIALS AND METHODS: Among 1,087 patients with UCB who had undergone RC at our 11 institutions between 1990 and 2010, 306 patients who subsequently developed distant metastasis or local recurrence or both were identified. Clinical data were collected with medical record review. Univariate and multivariate Cox regression models addressed OS after recurrence. A nomogram predicting postrecurrence OS was constructed based on Cox proportional hazards model, without using postrecurrence factors (systemic chemotherapy and resection of metastasis). The performance of the nomogram was internally validated by assessing concordance index and calibration plots. RESULTS: Of the 306 patients, 268 died during follow-up with a median survival of 7 months (95% CI 5.8-8.5). Postrecurrence chemotherapy was administered in 119 patients (38.9%). Multivariable analysis identified 9 independent predictors for OS; period of time from RC to recurrence (time-to-recurrence), symptomatic recurrence, liver metastasis, hemoglobin level, serum alkaline phosphatase level, serum lactate dehydrogenase level, serum C-reactive protein level, postrecurrence chemotherapy, and resection of metastasis. A nomogram was formed with the following 5 variables to predict OS time-to-recurrence, symptomatic recurrence, liver metastasis, albumin level, and alkaline phosphatase level. Concordance index rate was 0.75 (95% CI 0.72-0.78) by internal validation using Bootstraps with 1,000 resamples. Calibration plots showed that the nomogram fitted well. CONCLUSIONS: We identified 9 clinicopathological factors as independent OS predictors in postcystectomy recurrence of UCB. We also created a validated nomogram with 5 variables that efficiently stratified those patients regardless of eligibility for chemotherapy. The nomogram would be useful for acquiring relevant prognostic information and for stratifying patients for clinical trials.