Dynamic prognostication of non-muscle invasive bladder cancer using conditional recurrence- and progression-free...

Accurate prediction of recurrence and progression risk in non-muscle invasive bladder cancer (NMIBC) is essential for patient counseling and risk-adapted management. However, conventional models fail to account for the decrease in baseline risk over time. We therefore examined the conditional survival free of recurrence and progression in older adults with NMIBC to develop a dynamic risk prediction model.

We identified patients 66 to 89 years with Ta/Tis/T1 cN0 cM0 urothelial bladder cancer treated with transurethral resection of bladder tumor (TURBT) between 2000 and 2017 in SEER-Medicare. Conditional recurrence-free (RFS) and progression-free (PFS) survival were estimated using the Kaplan-Meier method. The associations of baseline characteristics with RFS and PFS at prespecified landmark times were evaluated using Cox-proportional hazards models.

A total of 39,862 patients were included. Of these, 26,339 (66%) had Ta, 11,758 (29%) had T1, and 1,765 (4%) had Tis-disease. Median follow-up was 65 months. The 60-month RFS and PFS increased from 0.39 and 0.85 at baseline to 0.73 and 0.89 at 24-months event-free survival. Conditional RFS rapidly improved within the first 24 months before plateauing. Patients with T1-disease demonstrated the greatest improvement in conditional RFS. On multivariable analyses, T stage and tumor grade were less predictive of RFS with longer landmark times. Limitations include measurement error and risk heterogeneity within grade and stage subgroups.

Among patients with NMIBC, recurrence and progression risks decrease with longer event-free intervals, particularly among patients at highest risk of each event as reflected by tumor stage and grade. A dynamic risk prediction model can improve patient counseling and support risk-adapted management during follow-up.