DNA methylation ratio variability may impede clinical application of cancer diagnostic markers

Hypermethylation at promoter regions of tumour suppressor genes is diagnostic for many cancers. Many genomic regions that may be the targets for clinical diagnostic assays have been identified through use of measuring systems reliant on bisulphite conversion, but few of these promising markers are in clinical use. The comparability of a widely used DNA methylation measuring system involving bisulphite conversion was evaluated by supplying three experienced centres with methylated DNA reference material mixtures that were independently prepared and characterised by mass spectrometry and high-pressure liquid chromatography. A replication scheme was designed to evaluate reproducibility of key analytical steps within and between laboratories by regression analysis. In general, methylation was underestimated and methylation ratio values were highly variable. The difference in methylation ratio between CpG sites was the key contributor to variable results. The CpG site effect followed a similar pattern at all centres and at all methylation levels examined indicating that sequence context had a major effect on methylation ratio measurement using the bisulphite conversion process. The magnitude of underestimation combined with the variability of measurements between CpG sites compromises the concept of measuring genomic regional methylation by averaging the methylation ratios of many CpG sites. There were no significant differences in replicate bisulphite conversions or sample work-up and instrument analysis at each centre thus making this technique suitable for comparative intralaboratory investigations. However, it may not be suitable for a routine diagnostic assay without extensive standardisation efforts.