The purpose of this paper was to determine if cancer subgroups could be accurately identified based upon methylation levels with formalin-fixed, paraffin-embedded (FFPE) colon cancer specimens.
Genome-wide DNA methylation analysis of formalin-fixed paraffin embedded colorectal cancer tissue.
Dumenil TD, Wockner LF, Bettington M, McKeone DM, Klein K, Bowdler LM, Montgomery GW, Leggett BA, Whitehall VL
Purpose
Conclusion
Unsupervised cluster analysis of data from the Infinium HumanMethylation450 BeadChip array resulted in four distinct clusters which were segregated by preservation method (frozen, FFPE) and then by CpG island methylator phenotype (CIMP) status (CIMP-high, CIMP-negative). CIMP-dependent clustering was similar within the FFPE and frozen groups, including for specimens with KRAS mutations. Differential methylation analysis between CIMP-high and CIMP-negative specimens revealed fewer unique genes for FFPE specimens (4,455) compared to frozen specimens (5,915), and commonality in differentially methylated genes between FFPE and frozen specimens ranged between 61-91%; however, similar results were seen between fresh and frozen specimens when batch effect was assessed (68-85%). Differential methylation analysis revealed that FFPE block storage adversely affected the level of similarity with frozen specimens. While sample-independent quality controls performed well with samples from both frozen and FFPE specimens, sample-dependent controls performed poorly for samples from FFPE specimens, indicating a lower bisulfite conversion efficiency. However, methylation status of a panel of five CIMP markers assessed by the Methyl Light assay exhibited perfect concordance between frozen and FFPE specimens with the exception of a single FFPE specimen outlier.
Studies (1)
Purpose
The purpose of this study was to compare CIMP status and methlyation levels in frozen and FFPE colon cancer specimens. Twenty-one matched frozen and FFPE colon cancer specimens, and 18 additional fresh specimens were subjected to bisulfite conversion, whole genome amplification, and detection by an Infinium HumanMethylation450 BeadChip array. Fresh, frozen and FFPE specimens were extracted by different methods, and FFPE specimens were also subjected to a DNA restoration procedure with ZR-96 DNA Clean and Concentrator-5 plates. Samples were classified with a CIMP-high status if three or more genes were methylated.
Summary of Findings
Unsupervised cluster analysis of case-matched frozen and FFPE colon cancer specimens resulted in four distinct clusters which were segregated first by preservation method (frozen, FFPE) and then by CIMPstatus (CIMP-high, CIMP-negative), with the exception of a single sample from an FFPE specimen. This sample was believed to be contaminated with normal adjacent mucosa tissue. CIMP-dependent clustering was similar within the FFPE and frozen groups. Differential methylation analysis between CIMP-high and CIMP-negative specimens revealed 5,915 unique genes in frozen specimens, only 61% of which were also differentially methylated in case-matched FFPE specimens. Similarly, 4,455 unique genes were differentially methylated between CIMP-high and CIMP-negative FFPE specimens, 91% of which were also differentially methylated in case-matched frozen specimens. However, a batch effect was identified by comparing frozen specimens with a second cohort of fresh specimens, as commonality among differentially methylated genes ranged between 68-85%. Differential methylation analysis also revealed that FFPE block storage adversely affected the level of similarity with frozen specimens, as Euclidean distance between FFPE and frozen cohorts was weekly correlated to FFPE block age (r2=0.57). While sample-independent quality controls performed well with samples from both frozen and FFPE specimens, sample-dependent controls performed poorly for samples from FFPE specimens, indicating a lower bisulfite conversion efficiency. When methylation status of a panel of five CIMP markers were assessed by the Methyl Light assay in samples obtained from both frozen and FFPE specimens, perfect concordance was achieved with the exception of the single specimen outlier discussed previously.
Biospecimens
TissueColorectalPreservation Types
FormalinFrozenNone (Fresh)Diagnoses
NeoplasticCarcinomaPlatforms
DNABisulfite conversion assayDNAWhole genome amplificationDNADNA microarrayDNAReal-time qPCRPre-analytical Factors
Classification Pre-Analytical Factor Value(s) Biospecimen Preservation Type of fixation/preservation Formalin (buffered)FrozenNone (fresh)Storage Storage duration 4-19 y