93P - UriMee: A novel non-invasive test for diagnosis of urothelial carcinoma by detection of methylation markers in urinary sediment DNA

Background

Urothelial carcinoma (UC) is a prevalent malignancy in the urinary system, which consists of approximately 82% of bladder cancer and 18% upper tract urothelial carcinomas (UTUCs) in China. The diagnosis of UC poses significant challenges due to the invasiveness of cystoscopy and the limited sensitivity of cytology. Therefore, there is a pressing clinical demand for developing more sensitive and non-invasive diagnostic techniques to complement the existing approaches.

Methods

In this study, we conducted a comprehensive analysis utilizing methylation data from The Cancer Genome Atlas (TCGA), the Gene Expression Omnibus (GEO) database, and a cohort from Renji Hospital. These datasets served as discovery cohorts for the identification of UC-specific methylation markers. Urine samples from 180 cases were collected as a training cohort. Quantitative methylation-specific PCR was performed on these samples, followed by logistic regression for constructing a diagnostic model. The model's performance was further validated in a prospective cohort comprising 508 cases.

Results

Our analysis of the discovery cohort identified three significant methylation markers. UriMee, a diagnostic model, based on two urine-based methylation markers in the training cohort (65 UC cases vs. 115 non-UC cases), exhibited remarkable performance with an area under the curve (AUC) of 0.96. In the independent validation cohort (154 UC cases vs. 354 non-UC cases), UriMee demonstrated a sensitivity of 91% and a specificity of 93% (AUC=0.94). Notably, for early-stage (Ta-1) cases, the sensitivity reached 87%. Moreover, a significant correlation was observed between the UriMee detection values and tumor grade, with an 81% sensitivity for detecting low grade cases. UriMee also demonstrated high accuracy, reaching 91%, in diagnosing UTUCs.

Conclusions

UriMee holds significant promise as a non-invasive urine-based testing method, which has the potential to greatly enhance the early diagnosis of UC.

Clinical trial identification

Editorial acknowledgement

Legal entity responsible for the study

The authors.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.