A methodological study of genome-wide DNA methylation analyses using matched archival formalin-fixed paraffin embedded and fresh frozen breast tumors.
Author(s): Espinal AC, Wang D, Yan L, Liu S, Tang L, Hu Q, Morrison CD, Ambrosone CB, Higgins MJ, Sucheston-Campbell LE
Publication: Oncotarget, 2017, Vol. 8, Page 14821-14829
PubMed ID: 28118602 PubMed Review Paper? No
Purpose of Paper
This paper compared methylation profiles obtained with formalin-fixed paraffin-embedded (FFPE) sections, curls and punches and case-matched frozen specimens obtained from ten breast tumor specimens.
Conclusion of Paper
Correlations in methylation status obtained with frozen and FFPE specimens were very strong for CpG islands and shores, but were lower for shelves. Correlations in methylation status were stronger when FFPE sections, punches and curls were compared to one another than when they were compared to frozen specimens for CpG islands and shores and much higher for the CpG shelves. Hierarchical clustering of all CpG sites revealed a single branch per site that included both frozen and FFPE specimens, and the three FFPE specimen types clustered by patient for 6 of the 10 patients. Although more loci were shown to be differentially methylated in estrogen receptor (ER) positive versus ER negative breast tumors when frozen specimens were used compared to FFPE specimens, 73.2%, 58.3% and 65.5% of the differentially methylated loci (DML) identified in frozen specimens were also identified using FFPE sections, punches and curls, respectively. Importantly, all FFPE specimens were properly segregated by ER status using 100 loci that were previously identified as DML with frozen specimens, as 84 or more were identified as ER status DML in the three FFPE specimen types. When DML identified in frozen specimens were designated as true values, the proportion of ER-status associated DMLs detected in FFPE specimens had a positive predictive value of 0.87. Weighted Correlation Gene Network Analyses (WCGNA) revealed that the modules with the greatest number of genes had the strongest correlations between frozen and FFPE specimens, with preservation observed for 2246 of the 2664 loci.
Studies
-
Study Purpose
This study compared methylation profiles obtained with FFPE sections, curls and punches and case-matched frozen specimens from ten breast tumor specimens. Part of each surgically resected breast tumor specimen was snap-frozen in liquid nitrogen and stored at -80˚C while the remainder was fixed in 10% formalin for an unspecified duration and paraffin-embedded. DNA was extracted from frozen specimens using Puregene kit. DNA was extracted from FFPE specimens in the form of 10 mm punches, 20 µm thick curls or four 10 µm thick sections by deparaffinization in xylene and proteinase K digestion to completion at 56˚C, after which isolated DNA was cleaned with the DNA Clean & Concentrator-5 kit. DNA was quantified using the Quant-iT Picogreen dsDNA assay kit. Methylation was determined using bisulfate-converted DNA and the Illumina 450K Beadchip array after DNA repair (for FFPE specimens) with the Infinium HD DNA Restoration Kit.
Summary of Findings:
Correlations in methylation status between frozen specimens and either FFPE sections, curls, or punches were very strong for CpG islands (r=0.95-0.96, all) and shores (r=0.94-0.95, all), but correlations between frozen and either FFPE sections, curls, or punches were lower for CpG shelves (r=0.87-0.88). Correlations in methylation status between FFPE specimen types were slightly stronger than those observed between FFPE and frozen specimens for CpG islands (r=0.97-0.98 versus r=0.95-0.96) and shores (r=0.97-0.98 versus r=0.94-0.95) and much stronger for shelves (r=0.94-0.96 versus r=0.87-0.88). Hierarchical clustering of all CpG sites grouped specimens by patient with all FFPE specimens forming a single branch with only one exception; further, the frozen specimen and FFPE specimen was in the same branch for 6 of 10 patients. In the one exception, the frozen and FFPE specimens clustered separately from one another but were still within the same branch. SNP analysis with the 450K array showed all specimens were from the same patient. More loci were differentially methylated in ER positive versus ER negative breast tumors when frozen specimens were analyzed than when FFPE specimens were used (21,925 versus 13,594). Of the differentially methylated loci (DML) identified in frozen specimens 73.2%, 58.3% and 65.5% were also identified using FFPE sections, punches and curls, respectively; and frozen specimens identified 45.4%, 31.3% and 35.7% of the DML identified in FFPE sections, punches and curls, respectively. Importantly, all FFPE specimens were properly segregated by ER status using 100 loci previously identified by DML between ER positive and negative tumors, with 84 or more of the 100 DML shown to be differentially methylated in the three FFPE specimen types as well. Further, for those ER dependent 100 loci the correlation between frozen and FFPE sections, punches and curls were r=0.75-0.76. When the DML identified in frozen specimens were used as true values, the proportion of ER-status associated DMLs detected in case-matched FFPE specimens had a positive predictive value of 0.87 for ER status. Weighted Correlation Gene Network Analyses (WCGNA) revealed that the modules with greatest number of genes had the strongest correlations between frozen and FFPE specimens, with preservation observed for 2246 of the 2664 loci.
Biospecimens
Preservative Types
- Frozen
- Formalin
Diagnoses:
- Neoplastic - Carcinoma
Platform:
Analyte Technology Platform DNA DNA microarray DNA Bisulfite conversion assay Pre-analytical Factors:
Classification Pre-analytical Factor Value(s) Preaquisition Diagnosis/ patient condition ER-positive breast tumor
ER-negative breast tumor
Biospecimen Preservation Type of fixation/preservation Formalin (buffered)
Snap frozen
Biospecimen Aliquots and Components Aliquot size/volume Four 10µm sections
20µm curl
10 mm punch