Risk stratification for bladder recurrence of upper urinary tract urothelial carcinoma after radical nephroureterectomy

Objectives To identify risk factors and develop a model for predicting recurrence of upper urinary tract urothelial carcinoma (UTUC) in the bladder in patients without a history of bladder cancer after radical nephroureterectomy (RNU). Patients and Methods We retrospectively reviewed 754 patients with UTUC without prior or concurrent bladder cancer or distant metastasis at 13 institutions in Japan. Univariate and multivariate Fine and Gray competing risks proportional hazards models were used to examine the cumulative incidence of bladder recurrence of UTUC. A risk stratification model and a nomogram were constructed. Two prediction models were compared using the concordance index (c-index) focusing on predictive accuracy and decision-curve analysis, which indicate whether a model is appropriate for decision-making and determining subsequent patient prognosis. Results The cumulative incidence rates of bladder UTUC recurrence at 1 and 5 years were 15 and 29%, respectively; the median time to bladder UTUC recurrence was 10 months. Multivariate analysis showed that papillary tumour architecture, absence of lymphovascular invasion and higher pathological T stage were both predictive factors for bladder cancer recurrence. The predictive accuracy of the risk stratification model and the nomogram for bladder cancer recurrence were not different (c-index: 0.60 and 0.62). According to the decision-curve analysis, the risk stratification was an acceptable model because the net benefit of the risk stratification was equivalent to that of the nomogram. The overall cumulative incidence rates of bladder cancer 5 years after RNU were 10, 26 and 44% in the low-, intermediate-and high-risk groups, respectively. Conclusions We identified risk factors and developed a risk stratification model for UTUC recurrence in the bladder after RNU. This model could be used to provide both an individualised strategy to prevent recurrence and a risk-stratified surveillance protocol.