AUTHOR=Grodstein Francine , Lemos Bernardo , Yu Lei , Iatrou Artemis , De Jager Philip L. , Bennett David A. 

TITLE=Characteristics of Epigenetic Clocks Across Blood and Brain Tissue in Older Women and Men

JOURNAL=Frontiers in Neuroscience

VOLUME=14

YEAR=2021

URL=https://www.frontiersin.org/journals/neuroscience/articles/10.3389/fnins.2020.555307

DOI=10.3389/fnins.2020.555307

ISSN=1662-453X

ABSTRACT=<p>Epigenetic clocks are among the most promising biomarkers of aging. It is particularly important to establish biomarkers of brain aging to better understand neurodegenerative diseases. To advance application of epigenetic clocks—which were largely created with DNA methylation levels in blood samples—for use in brain, we need clearer evaluation of epigenetic clock behavior in brain, including direct comparisons of brain specimens with blood, a more accessible tissue for research. We leveraged data from the Religious Orders Study and Rush Memory and Aging Project to examine three established epigenetic clocks (Horvath, Hannum, PhenoAge clocks) and a newer clock, trained in cortical tissue. We calculated each clock in three different specimens: (1) antemortem CD4+ cells derived from blood (<italic>n</italic> = 41); (2) postmortem dorsolateral prefrontal cortex (DLPFC, <italic>n</italic> = 730); and (3) postmortem posterior cingulate cortex (PCC, <italic>n</italic> = 186), among older women and men, age 66–108 years at death. Across all clocks, epigenetic age calculated from blood and brain specimens was generally lower than chronologic age, although differences were smallest for the Cortical clock when calculated in the brain specimens. Nonetheless, we found that Pearson correlations of epigenetic to chronologic ages in brain specimens were generally reasonable for all clocks; correlations for the Horvath, Hannum, and PhenoAge clocks largely ranged from 0.5 to 0.7 (all <italic>p</italic> &lt; 0.0001). The Cortical clock outperformed the other clocks, reaching a correlation of 0.83 in the DLFPC (<italic>p</italic> &lt; 0.0001) for epigenetic vs. chronologic age. Nonetheless, epigenetic age was quite modestly correlated across blood and DLPFC in 41 participants with paired samples [Pearson r from 0.21 (<italic>p</italic> = 0.2) to 0.32 (<italic>p</italic> = 0.05)], indicating that broader research in neurodegeneration may benefit from clocks using CpG sites better conserved across blood and brain. Finally, in analyses stratified by sex, by pathologic diagnosis of Alzheimer disease, and by clinical diagnosis of Alzheimer dementia, correlations of epigenetic to chronologic age remained consistently high across all groups. Future research in brain aging will benefit from epigenetic clocks constructed in brain specimens, including exploration of any advantages of focusing on CpG sites conserved across brain and other tissue types.</p>