DNA methylation is one of the most extensively characterized epigenetic alterations in cancer and provides a biologically stable, analytically tractable foundation for biomarker development. In genitourinary (GU) oncology, the early occurrence, clonal maintenance, and chemical stability of aberrant cytosine-phosphate-guanine methylation enable reliable detection in tissue, urine, and plasma, supporting its use in liquid biopsybased diagnostics and surveillance. Recent advances have enabled clinical evaluation of DNA methylation assays across multiple GU malignancies. Among these, urothelial carcinoma has progressed furthest toward clinical translation: urine-based methylation tests have demonstrated high diagnostic accuracy and reproducibility and, in selected settings, have achieved regulatory approval or guideline incorporation for surveillance. In renal cell carcinoma, genome-wide plasma and urine methylome profiling has shown feasibility for early detection and prognostic stratification, although clinical implementation remains investigational. In prostate cancer, tissue-based methylation assays support repeat-biopsy decision-making, whereas urine- and plasma-based tests contribute to detection of clinically significant disease, molecular subtyping, and treatment-response monitoring. Despite strong analytical performance, broad clinical adoption of DNA methylation biomarkers remains constrained by limited prospective validation, interlaboratory variability, and incomplete evidence of cost-effectiveness in real-world care pathways. Economic modeling suggests that these assays could reduce invasive procedures and surveillance costs, but these findings require confirmation in prospective studies. Given the accessibility of urine as a noninvasive biospecimen and ongoing integration with machine-learning and multiomic frameworks, DNA methylation profiling is well positioned to complement conventional diagnostics and advance precision detection, surveillance, and prognostication in GU cancers.
Lee et al. (Thu,) studied this question.