DNA methylation biomarkers are increasingly utilized for the detection, prognosis and monitoring of cancer. Here we use publicly-available whole genome bisulfite sequencing data to identify differentially methylated regions (cDMRs) in diverse tumor types and further define a set of genomic target regions that have optimal characteristics for Methylation Sensitive Restriction Enzyme-PCR (MSRE-PCR)-based detection: conserved hypermethylation in tumors, abundant MSRE sites and low methylation levels in normal tissues. The identified MSRE-PCR target regions (n = 1,294) were primarily encompassed within CpG islands (97%) and promoters (81%) with 39% of the target regions overlapping the transcription start site. Gene set enrichment analysis of the target regions identified significant enrichment of genes involved in neuronal development. A multiplexed MSRE-PCR assay was developed interrogating 47 target regions and was tested on a set of genomic DNAs (n = 100) from diverse tumor and normal tissue types including colon, breast, lung, stomach and blood. A logistic regression model containing seven target region amplicons distinguished between tumor and normal tissue in the training (n = 50) with a ROC AUC of 0.97 (95% CI [0.92, 1]) and independent test set (n = 50) with an AUC of 0.93 (95% CI [0.84, 1]). These findings show that genomic regions with conserved hypermethylation across diverse tumor types, abundant MSRE sites and low methylation levels in normal tissues provide target regions for the detection of tumor DNA via MSRE-PCR. The selective amplification of tumor-derived DNA via MSRE-PCR may have utility in the development of non-invasive cancer detection and surveillance strategies.
Keywords: DNA methylation biomarkers; MSRE-PCR; cDMR; cancer; epigenetics.
Copyright: © 2020 Chapman et al.