Clinical performance and utility of a DNA methylation urine test for bladder cancer

Abnormal gene methylation has been observed in several cancers. A prior study reported methylation of TWIST1 and NID2 as a quantitative biomarker for urothelial carcinoma, but external validation has yet to be performed. We sought to externally validate a urine-based methylation assay combining TWIST1 and NID2 and assess its clinical utility. A prospective trial was conducted comparing the methylation assay to cystoscopy and biopsy in patients with hematuria or prior non–muscle invasive bladder cancer. Sensitivity, specificity, negative and positive predictive values, and likelihood ratios of the methylation assay were calculated. Area under the receiver operating characteristic curves for each gene and the combined assay were computed. Bayesian analyses were performed to assess utility of the assay for a variety of clinical scenarios. Complete data were available for 111 patients. In validating the prior assay definition in the current cohort, sensitivity and specificity were 79% and 63%, respectively, and when optimized for the current cohort were 75% and 71%, respectively. The area under the curve for the assay was 0.73 compared with biopsy and 0.71 compared with cystoscopy. We failed to replicate the excellent performance of the methylation assay in this external validation; however, this assay may have utility for screening or surveillance for non–muscle invasive bladder cancer.