PO-342 Integrating copy number analysis and tumour DNA methylation profiling

Abstract

IntroductionThe cancer epigenome is a subject of intense research to identify new therapeutic targets and markers for diagnosis. Independent reports have showed that the cancer methylome is heterogeneous and that high levels of intra-tumour heterogeneity (ITH) correlate with poor prognosis. In this project, we aim to perform copy number profiling from bisulfite sequencing data and integrate allele specific copy number and purity estimates to deconvolve the pure tumour methylome and better our understanding of ITH and tumour evolution.Material and methodsAs part of the TRACERx study (TRAcking Cancer Evolution through therapy (Rx)), we analyse 38 cases of non-small cell lung cancer (NSCLC) with multi-region whole exome (WES) and reduced-representation bisulfite (RRBS) sequencing of the primary tumour for a total of 169 samples.We developed a novel bisulfite sequencing module for ASCAT (allele-specific copy number analysis of tumours), a method originally designed for SNP array data which computes the copy number profile of tumour cells, accounting for normal cell admixture and tumour aneuploidy.Results and discussionsTumour purity and copy number estimates were obtained using RRBS and validated using WES. We show that local copy number affects methylation rates extracted from the RRBS data in a way that can be modelled. If we have a CpG that is methylated in the tumour and unmethylated in the normal contaminating cells, the observed fraction of methylated reads in the bulk increases as a function of copy number given sample purity. Vice versa, the observed methylation rate will decrease with increasing tumour copy number when the tumour is unmethylated but the normal is methylated.Leveraging the observed relationship between methylation rate, copy number and tumour purity, we can extract the pure tumour methylation profiles from the RRBS data. The pure profiles allow for identification of differentially methylated regions (DMRs) between cancer and normal cells as well as between cancer subclones. We then reconstruct the evolutionary history of the tumour by hierarchal clustering of the DMRs. The obtained phylogenetic trees are validated by performing SNV clustering on the WES data.ConclusionWe show that epigenetic ITH is present in NSCLC and integrate the cancer epigenome with other layers of ‘omics data to deepen our understanding of tumour evolution. Next, we aim to provide timing estimates for DMRs and identify potential drivers based on event recurrence and impact on gene expression from RNA-Seq data.