Evidence of DNA methylation heterogeneity and epipolymorphism in kidney cancer tissue samples

Clear cell renal cell carcinoma (ccRCC) is characterised by significant genetic heterogeneity, which has diagnostic and prognostic implications. Very limited evidence is available regarding DNA methylation heterogeneity. We therefore generate sequence level DNA methylation data on 136 multi-region tumour and normal kidney tissue from 18 ccRCC patients, along with matched whole exome sequencing (85 samples) and gene expression (47 samples) data on a subset of samples. We perform a comprehensive systematic analysis of heterogeneity between patients, within a patient and within a sample. We demonstrate that bulk methylation data may be deconvoluted into cell-type-specific latent methylation components (LMCs), and that LMC1, which is likely to represent T cells, is associated with prognostic parameters. Differential epipolymorphism was noted between ccRCC and normal tissue in the promoter region of genes which are known to be associated with kidney cancer. This was externally validated in an independent cohort of 71 ccRCC and normal kidney tissues. Differential epipolymorphism in the gene promoter was a predictor of gene expression, after adjusting for average methylation. This represents the first evaluation of epipolymorphism in ccRCC and suggests that gains and losses in methylation disorder may have a functional relevance, gleaning important information on tumourigenesis.