In this study, the authors of evaluate the real-world utility of the risk calculator in a United States multi-institutional cohort, specifically evaluating its effectiveness at predicting outcomes.
Utilizing a cohort of 1498 patients treated in a healthcare system with vertical integration (median follow-up of 2.1 years for recurrence, 4.1 years for progression), the authors calculated the prognostic index (PI) for recurrence and progression using the published EORTC weights and classification of cases into four risk categories. Specifically this was done at the Kaiser Permanente Northwest system over a 24 year period. Pathology reports (n=4678) were reviewed and recorded by a fellowship trained urologic oncologist (author MN). Patients were primarily white and male. 88.6% of patient were assessed for primary occurrence, while 11% had had a recurrence in the prior year from the start point. 73% were pTa, 24% pT1, 0.7% were pure Carcinoma in situ (CIS) while 10% had concomitant CIS.
In terms of recurrence, the discriminatory ability was smaller in their validation sample (0.66) compared to discriminatory ability in the original paper (0.61), whereas for progression it was larger in the validation sample (0.78 vs 0.75). Comparing the survival curves for their validation cohort compared to the discovery cohort for the four risk groups, calibration and discrimination were adequate for all groups except the highest risk group. For progression, the cumulative incidence was smaller for the three highest risk groups in the validation cohort compared to the discovery sample, suggesting some miscalibration.
Unfortunately, it would appear that the EORTC risk calculator had limited discrimination and calibration for US patients treated in a real-world setting, particularly for the large group with lower-risk NMIBC. Future work will develop novel predictive models evaluating more granular intermediate NMIBC outcomes to inform implementation of risk-stratified care pathways – indeed, future risk calculators will likely incorporate molecular testing to help risk stratify patients.
1. As this was done over a 24 year period, the staging of NMIBC changed during that time frame multiple times. How was this accounted for?
1. Sylvester RJ, van der Meijden AP, Oosterlinck W, Witjes JA, Bouffioux C, Denis L, Newling DW, Kurth K. Predicting recurrence and progression in individual patients with stage Ta T1 bladder cancer using EORTC risk tables: a combined analysis of 2596 patients from seven EORTC trials. Eur Urol. 2006 Mar;49(3):466-5; discussion 475-7. Epub 2006 Jan 17.
Presented by: Tullika Garg, MD, MPH
Co-Authors: Tullika Garg MD, MPH¹, Carmit McMullen PhD², Michael Leo PhD², Maureen O'Keefe-Rosetti PhD², Sheila Weinmann PhD², Matthew Wagner MD² and Matthew Nielsen MD, MS³
Affiliation: ¹Geisinger Health System, Danville, PA; ²Kaiser Permanente Northwest, Portland, OR; ³UNC Lineberger Cancer Center, Chapel Hill, NC
Written by: Thenappan Chandrasekar, MD, Clinical Fellow, University of Toronto, twitter: @tchandra_uromd at the 18th Annual Meeting of the Society of Urologic Oncology, November 20-December 1, 2017 – Washington, DC