DNA Methylation Concordance Between Fresh-Frozen and Formalin-Fixed Paraffin-Embedded Human Brain Tissue in Neurodegenerative Diseases
DNA Methylation Concordance Between Fresh-Frozen and Formalin-Fixed Paraffin-Embedded Human Brain Tissue in Neurodegenerative Diseases
Publication status: preprint
A plain-language reading has not been prepared for this paper yet.
Original abstract
Formalin-fixed, paraffin-embedded (FFPE) brain tissue is a major archival resource, yet its suitability for high-density DNA methylation profiling is limited by fixation-induced DNA fragmentation and crosslinking. Fresh-frozen (FF) tissue is the gold standard, but its availability is limited, especially for neurodegenerative diseases. DNA methylation was profiled using the Illumina Infinium MethylationEPIC v2 array on paired FFPE and FF brain tissue from 26 donors with neurodegenerative disease (Alzheimer’s disease, Parkinson’s disease, dementia with Lewy bodies, or mixed pathology). After quality control, 50 samples remained, including 24 FFPE–FF pairs (6 matched cortical regions, 18 unmatched regions). Concordance was assessed using Pearson correlations of M-values. Variation across samples was evaluated by Principal Component Analysis (PCA) of the 10,000 most variable CpG sites. Cell-type compositions were estimated using the HiBED R package to deconvolute bulk methylation profiles. FFPE and FF samples showed high concordance (median Pearson r = 0.86). Matched-region pairs had higher correlations (median Pearson r = 0.93) than unmatched-region pairs (p < 0.001. PCA suggested that the preservation method was a major driver of variation, with PC1 explaining 24.2–48.5% of the variance, while PC2 captured additional variation (6.7–14.3%). FFPE samples had higher QC failure rates (26.5% vs. 0%) and elevated probe dropout. Preservation significantly affected cell-type composition, including astrocytes, GABAergic neurons, endothelial cells, and oligodendrocytes (FDR < 0.05), whereas diagnosis had no effect (FDR 0.81–0.98). Overall, FFPE brain tissue showed concordance with FF tissue on the EPIC v2 array. These findings support the use of FFPE brain samples for epigenome-wide association studies, particularly when fresh-frozen cohorts are unavailable. Investigators should account for preservation-related variance, and anatomical matching is recommended.