Identifying DNA methylation signatures in high-grade serous ovarian cancer: Results vary by control tissue type.

e17559 Background: High grade serous ovarian cancer (HGSC) is the most common and fatal epithelial ovarian cancer and is often diagnosed in late stages. DNA methylation has emerged as a potential biomarker for the early detection of cancer, including ovarian cancer. Studies examining DNA methylation in ovarian tumor tissue have used adjacent non-tumor tissues or tissues from unaffected women as the control; however, few have used paired tissue and research is sparse on how results may vary by the control non-tumor tissue type. Therefore, we examined DNA methylation signatures in HGSC tumor tissue using different control non-tumor tissue types. Methods: We examined DNA methylation signatures using the Illumina Infinium MethylationEPIC BeadChip (850k) in formalin-fixed paraffin-embedded tissue. In women with HGSC, we compared DNA methylation patterns between paired adjacent non-tumor ovary (n = 74) and fallopian tube (n = 80) tissues. We compared DNA methylation patterns in these adjacent non-tumor tissues with non-tumor ovary (n = 8) and fallopian tube (n = 8) tissues from unaffected women without ovarian cancer carrying a pathogenic variant in BRCA1/2 ( BRCA1/2+). Lastly, we compared the overlap in differentially methylated CpGs identified in HGSC tumor tissue compared to paired adjacent non-tumor ovary (n = 50) and fallopian tube (n = 52) tissues. We processed the methylation data and used principal components analysis (PCA) and the adjusted Rand Index to compare the non-tumor tissue type methylation patterns. We used a paired t-test between individual-matched tumor and adjacent non-tumor tissue for each CpG site and then applied Bonferroni’s method to adjust the obtained p-values. Results: PCA comparing methylation patterns between paired adjacent non-tumor ovary and fallopian tube tissues from women with HGSC showed an adjusted Rand Index of 0.78, revealing separate clusters that cannot be combined. PCA comparing methylation patterns from the adjacent non-tumor tissues from women with HGSC and the non-tumor tissues from unaffected BRCA1/2+ women showed an adjusted Rand Index of -0.10 and 0.07 for the ovary and fallopian tube tissues, respectively, revealing overlapping clusters that can be combined. Lastly, comparison of the top differentially methylated CpGs identified in HGSC tumor tissue when using paired adjacent non-tumor ovary versus fallopian tube tissues as the control showed minimal overlap (6.8% and 4.0% for hypermethylated and hypomethylated CpGs, respectively). Conclusions: These results suggest that paired adjacent non-tumor ovary and fallopian tube tissues from women with HGSC have different DNA methylation patterns and result in different methylation signatures identified in HGSC tumor tissue. When comparing results across studies, including for validation, the type of non-tumor tissue control must be considered.