Purpose: Urothelial bladder cancer (UBC) presents high recurrence rates, mandating continuous monitoring via invasive cystoscopy. The development of non-invasive tests for disease diagnosis and surveillance remains an unmet clinical need. In this study, validation of two urine-based biomarker panels for detecting primary and recurrent UBC was conducted. Experimental Design: Two studies (total n=1357) were performed for detecting primary (n=721) and relapsed UBC (n=636). Cystoscopy was applied for detecting UBC, while patients negative for recurrence had follow-up for at least one year to exclude presence of an undetected tumor at the time of sampling. Capillary electrophoresis coupled to mass spectrometry (CE-MS) was employed for the identification of urinary peptide biomarkers. The candidate urine-based peptide biomarker panels were derived from nested cross-sectional studies in primary (n=451) and recurrent (n=425) UBC. Results: Two biomarker panels were developed based on 116 and 106 peptide biomarkers using support vector machine algorithms. Validation of the urine-based biomarker panels in independent validation sets, resulted in AUC values of 0.87 and 0.75 for detecting primary (n=270) and recurrent UBC (n=211), respectively. At the optimal threshold, the classifier for detecting primary UBC exhibited 91% sensitivity and 68% specificity, while the classifier for recurrence demonstrated 87% sensitivity and 51% specificity. Particularly for patients undergoing surveillance, improved performance was achieved when combining the urine-based panel with cytology (AUC of 0.87). Conclusions: The developed urine-based peptide biomarker panel for detecting primary UBC exhibits good performance. Combination of the urine-based panel and cytology resulted in improved performance for detecting disease recurrence.
- Received November 10, 2015.
- Revision received February 22, 2016.
- Accepted March 11, 2016.
- Copyright ©2016, American Association for Cancer Research.