A new ultrasensitive assay for detection of hypermethylated tumor DNA in liquid biopsies

The methylation pattern of cancer DNA differs from that of normal DNA, wherein some loci are hypermethylated while others are hypomethylated. The ability to detect tumor-derived DNA in body fluids such as urine and blood plasma can open the door to development of non-invasive, `liquid biopsy` type of assays, for early detection and companion diagnostics. However, the small amounts of total cell free DNA, and the small fraction of tumor-derived DNA within the sample are both potential limiting factors for an effective liquid biopsy assay, especially for early stage detection. We developed EpiCheck – a simple, cost effective, and highly sensitive sodium bisulfite-free assay for detection of hypermethylated genomic loci associated with cancer. The assay is based on methylation-sensitive enzymatic digestion of DNA followed by concurrent real time PCR amplification of target and internal reference loci and software analysis of the output file. We tested the analytical sensitivity of the assay on mixtures of DNA from unmethylated plasmid molecules and a methylated human cell line. The assay detected 10 methylated molecules on a background of 2,000,000 unmethylated molecules, demonstrating an analytical sensitivity of 1:200,000. Seeking to develop the clinical potential of the assay, we developed new methylation biomarkers and created two clinical assays. “Bladder EpiCheck” is a 15-marker assay for detection of bladder cancer recurrence in urine, and “Lung EpiCheck” is a 6-marker assay for detection of lung cancer in plasma. Bladder EpiCheck was tested in a blinded, prospective, multicenter study on 353 patients undergoing surveillance, and achieved 68.2% sensitivity at 88% specificity for detection of any bladder cancer recurrence, and 91.7% sensitivity and 99.3% NPV for non-Ta-low grade tumors. Lung EpiCheck was tested on a validation cohort consisting of 181 newly-diagnosed lung cancer patients and 141 high-risk controls and achieved 74% sensitivity at 91% specificity.