122 Chapter 5 studies (8, 10, 12). Yet, the current assays may not reach the limit of detection needed for the low tumor-derived methylation signals. Nucleic acids that are released from the bladder epithelium may further dilute the ovarian cancer signal in urine. Another explanation for the absence of tumor-derived methylation signals of some genes in the urine could be linked to the origin of urinary cfDNA. Urine cfDNA is described to be even shorter as compared to plasma cfDNA (modal size of 82 vs. 167 basepairs) (30). Differences in fragmentation patterns between plasma and urine are likely caused by Dnase1 cleavage activity in the urine and high concentrations of urea and salt that affect histone-DNA binding (31). Histone-bound DNA is more protected against degradation as compared to DNA that is not histone-bound (32). Hypothetically, hypermethylated regions of interest that are not histone-bound could be further degraded and become unmeasurable. We partly accounted for this by including methylation markers with proven diagnostic value in plasma in our selection (i.e. C2CD4D(21, 22), CDO1(22)), which both appeared suitable for ovarian cancer detection in urine. Clear SCNA profiles harboring common chromosomal gains (e.g. 1q, 3q, 7q, 8q) and losses (e.g. 17p, 19q, 22q) could be obtained from four urine supernatant samples of ovarian cancer patients, verifying the presence of tumor-derived DNA in the urine (33). Furthermore, a focal amplification at chromosome 19 was identified in the urine of one patient with stage IIIA serous carcinoma, which is a clinically relevant alteration that has previously been described in a subgroup of serous ovarian cancers (34). Aneuploidy was detected previously in cervical scrape samples of ovarian cancer patients using the PapSEEK test (16). We also observed shorter fragment sizes in urine supernatant samples with a high tumor fraction, which is another indication for the presence of tumor-derived DNA in the urine, as shown previously in urine samples of glioma patients (30). Given the feasibility of ovarian cancer detection in cervical scrapes by DNA methylation analysis (14, 17), similar findings were expected for self-collected cervicovaginal samples. While C2CD4D and CDO1 distinguished healthy versus malignant in cervical scrapes, none of the markers showed elevated methylation levels in cervicovaginal self-samples. Our findings are in line with those of van Bommel et al. who reported that mutation analysis in cervicovaginal self-samples of ovarian cancer patients was not feasible (35). None of the pathogenic mutations found in surgical specimens could be detected in cervicovaginal self-samples. Ovarian cancer signals might be more diluted in cytological specimens collected from areas further away from the ovaries. This was also observed for the PapSEEK test, which detected 45% of ovarian cancers when using intrauterine sampling (Tao brush) as compared to 17% when using endocervical sampling (Pap brush) (16).
RkJQdWJsaXNoZXIy MjY0ODMw