Background: The aim of this study is to establish new risk tables for the current clinical setting, enabling short- and long-term risk stratification for recurrence, progression, and cancer-specific death after transurethral resection in non-muscle invasive bladder cancer (NMIBC). Currently available risk tables lack input from the 2004 World Health Organization grading system and risk prediction for cancer-specific death.
Methods: This was a multi-institutional database study of 1490 patients diagnosed with NMIBC (the development cohort). A multivariate Fine and Gray subdistribution hazard model was used to assess the prognostic impact of various factors. Patients were classified into low-, intermediate-, and high-risk groups according to a sum of the weight of selected factors, and predicted cumulative rates were calculated. Internal validation was conducted using 200 bootstrap resamples to assess the optimism for the c-index and estimate a bias-corrected c-index. External validation of the developed risk table was performed on an independent dataset of 91 patients.
Results: The Japanese NIshinihon uro-onCology Extensive collaboration group (J-NICE) risk stratification table was derived from six, five, and two factors for recurrence, progression, and cancer-specific death, respectively. The internal validation bias-corrected c-index values were 0.619, 0.621, and 0.705, respectively. The application of the J-NICE table to an external dataset resulted in c-indices for recurrence, progression, and cancer-specific death of 0.527, 0.691, and 0.603, respectively.
Conclusions: We propose a novel risk stratification model that predicts outcomes of treated NMIBC and may overcome the shortcomings of existing risk models. Further external validation is required to strengthen its clinical impact.
Keywords: Prediction; Prognosis; Progression; Recurrence; Risk; Urinary bladder neoplasms.