Conditional survival following radical cystectomy for urothelial carcinoma of the bladder.

INTRODUCTION

Traditional surveillance protocols do not adequately account for the decreasing risk of mortality over time in aggressive malignancies, such as bladder cancer. Rather, the risk of death depends on both the baseline risk of mortality and the time survived since treatment. We therefore evaluated the conditional survival of patients diagnosed with urothelial carcinoma of the bladder (UCB) following radical cystectomy (RC).

PATIENTS AND METHODS

We identified patients aged 18 to 75 with Charlson 0-1 and pTany pN0-3 cM0 UCB diagnosed from 2006 to 2015 in the National Cancer Database and treated with RC. The 2- and 5-year conditional overall survival (COS)-i.e., the probability of surviving an additional 2- or 5-years given a specified time survived since treatment-was estimated using the Kaplan-Meier method. Multivariable Cox regression models with landmark time analysis were used to evaluate the associations of baseline characteristics with OS over time.

RESULTS

A total of 15,594 patients were included in the study. Median follow-up was 27.8 months. The 2- and 5-year COS for the overall cohort increased through 36 months follow-up and then plateaued. When stratified by pT and pN stage, the COS gain increased with higher pT and pN stage, demonstrating the greatest increase over time for patients with pTany N1-3 disease (5-year COS of 23% at baseline, 58% at 36-months, and 71% at 60-months). In multivariable Cox regression modeling, pT and pN stage were significantly associated with higher all-cause mortality at baseline (HR 3.27 for pT4, HR 2.57 for pT3 vs. ≤pT2; HR 2.26 for pN2-3, HR 1.77 for pN1 vs. pN0), but these associations were attenuated in magnitude with increasing landmark times of 36- and 60-months (HR 1.63 for pT4, HR 1.35 for pT3 vs. ≤pT2; HR 1.34 for pN2-3, HR 1.27 for pN1 vs. pN0). Our study is limited by the retrospective design and the lack of cancer-specific survival data.

CONCLUSIONS

Risk of death after RC varies with time elapsed since treatment and disease stage. Accordingly, stage-specific COS may be used to improve prognostication and surveillance protocols.