121 Molecular analysis for ovarian cancer detection in patient-friendly samples masses. Only GHSR demonstrated slightly increased methylation levels in the urine sediment. Benign ovarian masses included in this study were highly suspicious for malignancy according to current triage methods (>40% risk of malignancy using the IOTA adnex model) as samples were collected in a tertiary oncology unit. Half of the included patients in our cohort were ultimately diagnosed with a benign ovarian mass, underlining that current triage for referral to tertiary oncology care is suboptimal. The majority of previous studies only included benign controls for methylation marker discovery in tissue but not during marker validation in plasma, as recently reviewed by Terp et al. (15), or benign controls were not age-matched to cancers (21). Similarly, studies on ovarian cancer detection in cervical scrapes did not include benign controls (16, 17). The inclusion of age-matched patients diagnosed with benign and malignant ovarian masses is essential to accurately assess the clinical value of DNA methylation testing for ovarian cancer detection. The presence of ovarian cancer-derived DNA in the urine is currently underexplored. So far, only Valle et al. reported on the detection of somatic mutation profiles and HIST1H2BB/MAGI2 promoter methylation in a small paired series of ascites, blood, tissue, urine, and vaginal swabs of HGSOC patients (28). Their data on two patients revealed that methylation levels in urinary cfDNA correlated stronger with tissue than with blood, indicating the potential of urine-based ovarian cancer detection. Unfortunately, the diagnostic potential of ovarian cancer detection in urine could not be determined in the study of Valle et al. as no control samples were included. In our study, different urine fractions were systematically compared to explore whether a preferred urine sample type for ovarian cancer detection exists. Full void urine most likely contains both genomic and cfDNA, whereas the urine sediment is enriched for genomic DNA and the urine supernatant for transrenally excreted cfDNA (29). This assumption is confirmed by the strong correlation for CDO1 between cervical scrapes and urine sediment, while cervical scrapes and urine supernatant correlated weakly to moderately. Most methylation markers significantly differentiated between healthy controls and ovarian cancer patients in the full void urine (3/12), followed by urine supernatant (1/12), and the urine sediment (1/12). These outcomes suggest that tumorderived methylation signals can originate from genomic DNA as well as transrenally excreted cfDNA. Yet, larger samples sizes are needed to determine whether a preferred urine sample type for methylation analysis exists. In the present study, genes with elevated methylation levels in HGSOC tissue, were not always measurable in urine. Our qMSP assays were designed to facilitate the detection of methylation in small DNA fragments present in the urine as shown in our previous 5
RkJQdWJsaXNoZXIy MjY0ODMw