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 [pT]3–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.
Risk stratification and conditional recurrence after radical cystectomy: toward adaptive follow-up / Contieri, Roberto; Martini, Alberto; Furrer, Marc; D'Andrea, David; Claps, Francesco; Von Deimling, Markus; Soria, Francesco; Mari, Andrea; Bianchi, Lorenzo; Tonin, Elena; Flippot, Ronan; Teoh, Jeremy Y C; Marcq, Gauthier; Desprez, Pierre-Emmanuel; Pichler, Renate; Ślusarczyk, Aleksander; Subiela, José Daniel; Del Giudice, Francesco; Santarelli, Valerio; Porreca, Angelo; Amodeo, Antonio; Pavan, Nicola; Simonato, Alchiede; Scilipoti, Pietro; Perdonà, Sisto; Gontero, Paolo; Van Rhijn, Bas Wg; Hurle, Rodolfo; Moschini, Marco; Pradere, Benjamin; Mertens, Laura S. - In: BJU INTERNATIONAL. - ISSN 1464-410X. - (2026). [10.1111/bju.70232]
Risk stratification and conditional recurrence after radical cystectomy: toward adaptive follow-up
D'Andrea, David;Mari, Andrea;Del Giudice, Francesco;Santarelli, Valerio;
2026
Abstract
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 [pT]3–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.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
