Preoperative Multivariable Prognostic Model for Prediction of Nonorgan Confined Urothelial Carcinoma of the Upper Urinary Tract

Abstract

We created a prognostic tool for the accurate preoperative prediction of nonorgan confined upper tract urothelial carcinoma. A computerized data bank containing comprehensive information on 1,453 patients who underwent radical nephroureterectomy at 13 academic institutions was generated and continuously updated. This study comprised a subset of 659 patients in whom all appropriate preoperative prognostic variables (age, gender, race, symptoms, Eastern Cooperative Oncology Group performance status, primary tumor location, tumor architecture, tumor grade and history of previous bladder cancer) were available for statistical analysis. A multivariable logistic regression model containing relevant clinicopathological variables addressed the prediction of nonorgan confined stage disease (T3-4 and/or N+) at radical nephroureterectomy. A backward step-down selection process was applied to achieve the most informative and parsimonious model. Internal validation was performed using 200 bootstrap resamples. Pathological nonorgan confined urothelial carcinoma was found in 40% of patients. Grade, architecture and location of the tumor were independently associated with nonorgan confined disease. A nomogram including these 3 variables achieved 76.6% accuracy in predicting nonorgan confined upper tract urothelial cancer. We developed a simple and accurate prognostic tool for the prediction of locally advanced upper tract urothelial cancer. This preoperative prediction model can be used for designing clinical trials, selecting patients for preoperative systemic therapy and guiding the extent of concomitant lymph node dissection at nephroureterectomy.