DNA hypomethylation in plasma as a cancer biomarker: when less is more?

Abstract

Early cancer detection strategies are required to identify patients at initial stages, when curative-intended treatment is more effective. Assessment of genome-wide hypomethylation might allow for early cancer detection in bodily fluids, but requires high throughput technologies, such as next generation sequencing. In the study under evaluation, performance of a hypomethylation and copy number aberration (CNA) assay for detection of hepatocellular carcinoma (HCC), based on massive parallel bisulfite sequencing of plasma DNA was assessed. Sensitive (92/69%) and specific (88/94%) HCC detection (using 'OR'/'AND' algorithms) was achieved using a mean sequencing depth of 93 million reads in one lane. Using the 'AND' or the 'OR' algorithms, other cancer types were detected with 60% sensitivity and 94% specificity, or 85% sensitivity and 88% specificity, respectively. Lowering sequencing depth to 10 million reads (decreasing analysis time and cost) did not alter hypomethylation performance but impaired that of CNA.