Urinary DNA methylation biomarkers for bladder cancer diagnosis: a single center case-control study

Background Noninvasive detection of bladder cancer remains challenging because many urine tests trade sensitivity for specificity. We evaluated whether a biologically grounded, complementary three−gene methylation panel measured in urine could improve diagnostic performance. Methods We conducted a single-center 1:1 matched case-control study using clinically relevant urine samples. Urine DNA was subjected to bisulfite conversion and quantitative methylation analysis targeting three gene loci ( TRPS1 , HAND2 , ZNF154 ). We evaluated the performance of individual genes and a combined panel, reporting sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), area under the receiver operating characteristic curve (AUC and ROC), and 95% confidence intervals (CI). Results One hundred and seventy-two subjects (86 cases and 86 controls) with a mean age of 60.24 ± 12.54 and 59.70 ± 12.75 years old were included. TRPS1 showed the highest sensitivity at 0.97 with specificity 0.791, yielding PPV 0.82 and NPV 0.97; HAND2 maintained strong sensitivity (0.95) but lower specificity (0.56), with PPV 0.68 and NPV 0.92; and ZNF154 had sensitivity 0.93 with the lowest specificity (0.35), corresponding to PPV 0.59 and NPV 0.83. A combined methylation assay where all three genes are methylation-positive significantly improved specificity to 0.95 while maintaining acceptable sensitivity (0.86), resulting in the highest positive predictive value (0.95) and a robust negative predictive value (0.87). The ROC analysis showed strong diagnostic performance for all three urinary methylation markers: TRPS1 had the highest AUC at 0.98 (95% CI: 0.96–0.99), followed by HAND2 at 0.90 (95% CI: 0.85–0.95) and ZNF154 at 0.82 (95% CI: 0.76–0.89). Conclusions A urine−based three−gene methylation panel ( TRPS1 , HAND2 , ZNF154 ) demonstrates promising accuracy for bladder cancer detection, with complementary signals improving specificity. Findings are constrained by a single–center cohort and moderate size and require external, multi−center validation in the future.