We have attempted to validate differences in DNA methylation levels of 30 candidate genes reported to discriminate between normal colon mucosa of colon cancer patients and normal colon mucosa of individuals without cancer. We found that CpG sites in 16 of the 30 candidate genes show significant differences in mean methylation level in normal colon mucosa in an independent cohort of 24 cancer patients and 24 controls. A support vector machine trained on these data and data for an additional 66 CpGs set yielded an 18-gene signature, composed of 10 of the validated candidate genes plus 8 additional candidates. This model exhibited 96% sensitivity and 100% specificity in a 40-sample training set and classified all eight samples in the test set correctly. Moreover, we found a moderate-strong correlation (Pearson coefficients r=0.253-0.722) between methylation levels in colon mucosa and methylation levels in peripheral blood for seven of the 18 genes in the support vector model. These 7 CpGs, alone, classified 44 of the 48 patients in the validation set correctly and five CpGs selected from only two of the seven genes classified 41 of the 48 patients in the discovery set correctly. These results suggest that methylation biomarkers may be developed that will, at minimum, serve as useful objective and quantitative diagnostic complements to colonoscopy as a cancer screening tool. These data also suggest that it may be possible to monitor biomarker methylation levels in tissues collected much less invasively than by colonoscopy.

Citation Format: Matteo Cesaroni. Validation of methylation biomarkers that distinguish normal colon mucosa from cancer patients from normal colon mucosa of patients without cancer. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 413. doi:10.1158/1538-7445.AM2014-413