Testing Narrow Epigenetic Clocks in Centenarians
Many different epigenetic clocks have been proposed and tested in recent years, all using different weighted combinations of DNA methylation status at various CpG sites on the genome, some using fewer than ten sites, others using hundreds of sites. DNA methylation is in constant flux, regulating gene expression in cells, but some changes are characteristic of age, and machine learning approaches have produced clocks with strong correlations to chronological age. Where clock age is higher than chronological age, individuals have been shown to have greater incidence and risk of age-related disease and mortality.
Researchers still, however, do not have more than the rudimentary beginnings of a map to link methylation at specific CpG sites to the underlying damage and dysfunction of aging. Thus it is hard to treat epigenetic clock data as actionable for any given individual and their treatments. The clocks are quite good good for unmodified aging, but what we really want is a way to cost-effectively, rapidly assess the outcome of potential rejuvenation therapies, each of which will tend to only directly affect one of the many mechanisms of aging, without undertaking the time and expense of life span studies.
Given this, it is hard to trust narrow epigenetic clocks that use few CpG sites. They seem very unlikely to accurately reflect all of the processes of aging, and thus even trying to calibrate them against specific therapies seems likely to produce poor results. Nonetheless, since such narrow clocks are cheaper than broad clocks using hundreds of CpG sites, many research groups are working in this direction.
The study of DNA methylation in human aging has revealed the occurrence of two types of age-related DNA methylation changes. The first, known as epigenetic drift, is characterized by the progressive divergence of the methylome of individuals acquired environmentally and stochastically across their lifespan, which even affects monozygotic twins. The second type of DNA methylation changes is called the epigenetic clock and refers to all age-related DNA methylation variations that consistently increase or decrease in every individual, thereby correlating to their chronological age.
The latter type of epigenetic modifications has been widely used as biomarkers of aging in several age-prediction models to estimate the chronological and biological age of individuals, mainly from blood DNA samples. These models are based on multiple regression, machine learning, and deep learning approaches using either a large number of CpGs requiring high-throughput technologies such as genome-wide epigenotyping array or a smaller number of CpGs requiring high resolution locus-specific methods such as pyrosequencing. DNA methylation-based age (DNAmage) prediction has proven to be of great interest in several bio-medical applications. It could notably give a better estimation of the biological age than chronological age and could also be a good indicator or predicator of different risks, health conditions and age-related diseases when compared to the chronological age.
In the present study, we investigated the DNAmage of French long-lived individuals (LLI) including centenarians and semi-supercentenarians (n = 214), as well as nonagenarian's and centenarian's offspring (n = 143) of the CEPH aging cohort using blood extracted DNA and four epigenetic clocks based on a small number of CpGs and locus-specific pyrosequencing. These clocks, known as Bekaert, Thong, Garali MQR and Garali GBR clocks, were developed from 2 to 4 CpGs located in the promoters of 1 to 4 genes (ASPA, EDARADD, ELOVL2, KLF14, PDE4C, and TRIM59).
Compared to their chronological age, DNAmage of centenarians and semi-supercentenarians was strongly underestimated (15 to 28.5 years in average), which was still strongly significantly underestimated when compared to control group DNAmage (10.8 to 21 years in average). This might indicate that the epigenetic clock and potentially aging were decelerated in exceptionally long-lived individuals, who presented younger DNAmage and potentially also younger biological age.