Risk stratification and conditional recurrence after radical cystectomy: toward adaptive follow-up.

Objective To develop a data-driven risk-stratification model to identify high-risk patients following radical cystectomy (RC) for bladder cancer and propose a risk-adapted follow-up (FU) schedule.Patients And Methods We performed a retrospective analysis of an individual patient data registry comprising 3196 patients with clinical T stage (cT)2-T4 N0M0 bladder cancer who underwent RC at 16 European centres (1990-2024). All treatment decisions, including the use of neoadjuvant chemotherapy, adjuvant therapy, and the FU schedule, were made at the discretion of the treating physician in accordance with the patient's preference. A Classification and Regression Tree (CART) analysis, incorporating pathological T and N stages, lymphovascular invasion (LVI), and other features, was used to stratify patients into Low-Risk and High-Risk groups for recurrence. The primary endpoint was recurrence-free survival (RFS). We used a landmark analysis to evaluate the conditional risk of recurrence at 1, 2, 3, 4, and 5 years after RC.Results At a median FU of 81.8 months, 891 patients recurred. CART analysis identified a High-Risk group (pathological T stage pT3-pT4, node-positive disease, or pT2 with LVI) with significantly worse 5-year RFS than the Low-Risk group (37.8% vs 76.2%; P < 0.001). This stratification was strongly prognostic for recurrence (hazard ratio HR 4.29; 95% confidence interval 3.63-5.00), cancer-specific survival (subdistribution HR 5.80), and overall survival (HR 3.04) (all P < 0.001). Landmark analysis confirmed that the elevated risk persisted up to 4 years; however, the conditional risk for event-free patients converged after 5 years (HR 1.37; P = 0.3).Conclusions This study establishes a simple, pathologically derived model (pT3-4/pN+/pT2 + LVI) that effectively stratifies post-RC patients, enabling a risk-adapted FU strategy. Prospective evaluation of this framework is required to confirm its clinical utility, safety, and cost-effectiveness.