Abstract 4529: Tailoring approaches for global epigenome analysis from archival formalin-fixed paraffin-embedded tissue samples

Novel DNA extraction methodologies allow the use of archival material from formalin fixed paraffin embedded (FFPE) tissue samples in genomic studies. A major limitation of this source of DNA is the fragmented nature and low overall yield generally obtained from clinical materials, making downstream applications such as epigenetic analysis challenging. Previous published attempts have focussed on smaller regions of CpG islands, such as the use of Illumina's 450k arrays (which measure methylation at approximately 485,000 sites). The objective of this study was to optimize experimental and analytical workflows that allow effective interrogation of global DNA methylation profiles from FFPE samples. Methylation capture was conducted on DNA from matched FFPE and fresh frozen samples from the same metastatic colorectal cancer patient as well as two colorectal cancer cell lines, using the SeqCap Epi (Roche) methyl capture system. The custom capture designed includes 5.5 million CpG sites across the genome, a greater than 10-fold increase compared to previously published studies. The wet-lab protocol was robustly optimized for several parameters, such as overall yield and bisulphite conversion efficiency (measured by shallow-read next generation sequencing). A data analysis pipeline composed of the bisulfite-converted DNA aligner BWA-meth, as well as in-house Perl and R scripts, was used to generate detailed methylation maps for individual sample types in order to identify differentially methylated regions (DMRs), which were further validated using targeted bisulphite sequencing for selected loci. Preliminary analysis of the data revealed 98% bisulphite conversion efficiency and low PCR duplicate rate (4-5%) across both sample types. Intriguingly, we observe an 80% concordance between the overall DNA methylation profiles between FFPE and fresh frozen samples, with a 60% overlap between the FFPE and fresh frozen samples in terms of direction of methylation i.e. hyper or hypomethylation. Mapping overall methylation levels to CpG ‘resorts’ (+/- 4kb of CpG islands) indicated ∼10% of methylation occurred in these regions totalling 245MB. Known genomic features, including exons, promoters of coding/non-coding genes and enhancers, contained ∼30% (1.5 million) of the methylated positions. In addition, this methodology also allowed us to identify C to T transitions across all samples, a common artefact of formalin fixed samples. The analysis revealed that a significantly lo