A urine-based DNA methylation assay to facilitate early detection and risk stratification of bladder cancer

Abstract

Background: Current non-invasive tests have limited sensitivities and lack capabilities of pre-operative risk stratification for bladder cancer (BC) diagnosis. We aimed to develop and validate a urine-based DNA methylation assay as a clinically feasible test for improving BC detection and enabling pre-operative risk stratifications. Methods: A urine-based DNA methylation assay was developed and validated by retrospective single-center studies in patients of suspected BC in Cohort 1 (n = 192) and Cohort 2 (n = 98), respectively. In addition, a prospective single-center study in hematuria patient group (Cohort 3, n = 174) was used as a second validation of the model. Results: The assay with a dual-marker detection model showed 88.1% and 91.2% sensitivities, 89.7% and 85.7% specificities in validation Cohort 2 (patients of suspected BC) and Cohort 3 (patients of hematuria), respectively. Furthermore, this assay showed improved sensitivities over cytology and FISH on detecting low-grade tumor (66.7–77.8% vs. 0.0–22.2%, 0.0–22.2%), Ta tumor (83.3% vs. 22.2–41.2%, 44.4–52.9%) and non-muscle invasive BC (NMIBC) (80.0–89.7% vs. 51.5–52.0%, 59.4–72.0%) in both cohorts. The assay also had higher accuracies (88.9–95.8%) in diagnosing cases with concurrent genitourinary disorders as compared to cytology (55.6–70.8%) and FISH (72.2–77.8%). Meanwhile, the assay with a five-marker stratification model identified high-risk NMIBC and muscle invasive BC with 90.5% sensitivity and 86.8% specificity in Cohort 2. Conclusions: The urine-based DNA methylation assay represents a highly sensitive and specific approach for BC early-stage detection and risk stratification. It has a potential to be used as a routine test to improve diagnosis and prognosis of BC in clinic.