Formalin-fixed, paraffin-embedded (FFPE) tissue epigenomics using Infinium HumanMethylation450 BeadChip assays

• Published in: Laboratory investigation Vol. 95; no. 7; pp. 833 - 842
• Main Authors: de Ruijter, Tim C, de Hoon, Joep PJ, Slaats, Jeroen, de Vries, Bart, Janssen, Marjolein JFW, van Wezel, Tom, Aarts, Maureen JB, van Engeland, Manon, Tjan-Heijnen, Vivianne CG, Van Neste, Leander, Veeck, Jürgen
• Format: Journal Article
• Language: English
• Published: New York Elsevier Inc 01.07.2015; Nature Publishing Group US; Nature Publishing Group
• Subjects: 45/61; 631/208/176/1988; Cluster Analysis; DNA Methylation; Epigenomics - methods; Fluorescence; Formaldehyde; Humans; Laboratory Medicine; Medicine; Medicine & Public Health; Oligonucleotide Array Sequence Analysis; Paraffin Embedding; Pathology; Reproducibility of Results; technical-report; Tissue Fixation
• ISSN: 0023-6837; 1530-0307
• DOI: 10.1038/labinvest.2015.53

Current genome-wide methods to detect DNA-methylation in healthy and diseased tissue require high-quality DNA from fresh-frozen (FF) samples. However, well-annotated clinical samples are mostly available as formalin-fixed, paraffin-embedded (FFPE) tissues containing poor-quality DNA. To overcome this limitation, we here aimed to evaluate a DNA restoration protocol for usage with the genome-wide Infinium HumanMethylation450 BeadChip assay (HM-450K). Sixty-six DNA samples from normal colon (n=9) and breast cancer (n=11) were interrogated separately using HM-450K. Analyses included matched FF/FFPE samples and technical duplicates. FFPE DNA was processed with (FFPEr) or without a DNA restoration protocol (Illumina). Differentially methylated genes were finally validated in 24 additional FFPE tissues using nested methylation-specific PCR (MSP). In summary, β-values correlation between FFPEr duplicates was high (ρ=0.9927 (s.d. ±0.0015)). Matched FF/FFPEr correlation was also high (ρ=0.9590 (s.d. ±0.0184)) compared with matched FF/FFPE (ρ=0.8051 (s.d. ±0.1028). Probe detection rate in FFPEr samples (98.37%, s.d. ±0.66) was comparable to FF samples (99.98%, s.d. ±0.019) and substantially lower in FFPE samples (82.31%, s.d. ±18.65). Assay robustness was not decreased by sample archival age up to 10 years. We could also demonstrate no decrease in assay robustness when using 100 ng of DNA input only. Four out of the five selected differentially methylated genes could be validated by MSP. The gene failing validation by PCR showed high variation of CpG β-values in primer-binding sites. In conclusion, by using the FFPE DNA restoration protocol, HM-450K assays provide robust, accurate and reproducible results with FFPE tissue-derived DNA, which are comparable to those obtained with FF tissue. Most importantly, differentially methylated genes can be validated using more sensitive techniques, such as nested MSP, altogether providing an epigenomics platform for molecular pathological epidemiology research on archived samples with limited tissue amount.