Clinical utility of urine DNA for noninvasive detection and minimal residual disease monitoring in urothelial carcinoma.

Abstract

4583 Background: Current methods for the early detection and minimal residual disease (MRD) of urothelial carcinoma (UC) are often invasive and/or possess sub-optimal sensitivity, especially in upper tract urothelial carcinoma (UTUC). Methods: We developed an efficient workflow named the urine tumor DNA multidimensional bioinformatic predictor, utLIFE, by using low-coverage whole-genome sequencing and targeted deep sequencing. We identified the UC specific mutations and large CNV in the discovery cohort. The utLIFE-UC model was trained in a bladder cancer retrospective cohort (n = 150) and validated in an bladder cohort(n = 674) and UTUC cohort(n = 22). 31 patients established diagnosis of BC who are on neoadjuvant were also enrolled, including a MRD training cohort (n = 16) with serial urine samples at baseline, during treatment and before surgery; a independent MRD validation cohort (n = 15) with urine samples before surgery. Results: The utLIFE-UC model could discriminate UC with high accuracy (94.3%), sensitivity (92.8%), and specificity (96.0%). Furthermore, compared to cytology, the assay achieved a great improvement in sensitivity in the detection of non–muscle-invasive bladder cancer (NMIBC, 94.7% vs. 31.6%,p = 0.0002), and muscle-invasive bladder cancer(MIBC, 82.6% vs. 69.6%, p = 0.4894). The utLIFE-UC model was also validated in independent BC cohort (sensitivity 94.3%, specificity 92.6%) and UTUC cohort (sensitivity 90.9%, specificity 90.9%). utLIFE-UC score also showed outstanding potential on dynamic surveillance of residual disease in UC as the score showed dramatically decreased in pCR patients. As expected, utLIFE-UC score could classify pCR and non-pCR(PR, SD, PD) with NPV 100% in MRD training and validation cohorts, which showed superior sensitivity over that of urine cytology (100% vs.37.5%, p = 0.0070) and FISH (100 vs. 42.9%, p = 0.0125). Conclusions: utLIFE-UC diagnostic model for early diagnosis, residual disease detection in UC is a cost-effective, rapid, high-throughput, noninvasive, and promising approach, which may reduce the burden of cystoscopy and blind surgery.