Quantitative profiling of CpG island methylation in human stool for colorectal cancer detection.

PURPOSE: The aims of this study were to investigate the use of quantitative CGI methylation data from stool DNA to classify colon cancer patients and to relate stool CGI methylation levels to those found in corresponding tissue samples. METHODS: We applied a quantitative methylation-specific PCR assay to determine CGI methylation levels of six genes, previously shown to be aberrantly methylated during colorectal carcinogenesis. Assays were performed on DNA from biopsies of "normal" mucosa and stool samples from 57 patients classified as disease-free, adenoma, or cancer by endoscopy, and in tumour tissue from cancer patients. Additionally, CGI methylation was analysed in stool DNA from an asymptomatic population of individuals covering a broad age range (mean = 47 ± 24 years) RESULTS: CGI methylation levels in stool DNA were significantly higher than in DNA from macroscopically normal mucosa, and a significant correlation between stool and mucosa was observed for ESR1 only. Multivariate statistical analyses using the methylation levels of each CGI in stool DNA as a continuous variable revealed a highly significant (p = 0.003) classification of cancer vs. non-cancer (adenoma + disease-free) patients (sensitivity = 65 %, specificity = 81 %). CONCLUSION: CGI methylation profiling of stool DNA successfully identified patients with cancer despite the methylation status of CGIs in stool DNA not generally reflecting those in DNA from the colonic mucosa.