Development and Validation of the UriMee Model: A Methylation-based Diagnostic Tool for Early Diagnosis of Urothelial Carcinoma

Background and objective

Urothelial carcinoma (UC) is a common malignancy that imposes a significant health care burden. Current diagnostic methods are limited by their invasiveness and low sensitivity, particularly for detecting low-grade tumors. Noninvasive, accurate, and reliable diagnostic tests for an early diagnosis of UC are urgently needed.

Methods

UC-specific DNA methylation biomarkers were identified by combining public datasets from The Cancer Genome Atlas and Gene Expression Omnibus with a cohort from Renji Hospital (n = 50). Using the Least Absolute Shrinkage and Selection Operator regression, we developed a diagnostic model, termed the UriMee model, by selecting key biomarkers from a model cohort (n = 322) and subsequently validating it in an independent cohort (n = 131). The diagnostic performance of the assay was evaluated and compared with that of urine cytology.

Key findings and limitations

At 30% threshold probability, the UriMee model demonstrated high sensitivity (92%) and specificity (92%) in distinguishing UC cases, with particularly strong performance in early-stage tumors (83% sensitivity for Ta, 93% for T1, and 100% for Tis). It significantly outperformed urine cytology, offering greater sensitivity (90% vs 25%, p < 0.001) while maintaining comparable specificity. Additionally, the model was highly effective in identifying upper tract urothelial carcinoma (UTUC), achieving sensitivity of 96%. The study’s limitations include the necessity for larger multicenter studies and long-term follow-up to validate the findings and assess the test’s effectiveness across diverse populations, as well as its utility in monitoring disease progression and recurrence.

Conclusions

The UriMee test demonstrated high sensitivity and specificity, particularly in detecting early-stage tumors and UTUC, significantly outperforming traditional methods.