Prospective external validation of a bladder cancer detection model.

Abstract

Few studies have combined clinical prognostic factors with urinary biomarkers into risk profiles that can be used to predict the likelihood of bladder cancer. We previously developed and internally validated a bladder cancer detection nomogram that combines clinical features with the NMP22® BladderChek® test. To consider extensive use of the model the nomogram was tested in a prospective cohort of patients who presented with hematuria. Patients referred for hematuria evaluation were prospectively enrolled at 3 centers. Each patient underwent complete urological evaluation, including history, examination, cystoscopy, cytology and NMP22. A logistic regression model to predict urothelial bladder carcinoma was also developed to compare the performance of clinical data with and without adding NMP22 and urinary cytology. The study included 381 patients (50.7% women) with a median age of 58 years. Urothelial bladder carcinoma was detected in 23 patients (6%). It was associated with age greater than 65 (11.1% vs 4% of patients, p = 0.012), male gender (10.1% vs 2%, p = 0.003), white ethnicity (9.2% vs 3.1%, p = 0.016), gross hematuria (9.9% vs 2.5%, p = 0.005), positive NMP22 (37% vs 3.7%, p <0.001) and positive cytology (83.3% vs 3.9%, p <0.001). Predictive accuracy of the bladder cancer detection nomogram was 80.2%. The calibration plot indicated that the previously published nomogram was well calibrated in patients with a less than 15% predicted probability of urothelial bladder carcinoma. We prospectively validated a highly accurate tool that combines clinical factors and a urinary biomarker to detect bladder cancer. This tool can help prioritize urological referrals for patients with hematuria.