Researchers have developed many aging clocks in recent years. Machine learning techniques are applied to large sets of biological data assessed at various ages in order to identify algorithms that predict chronological age, mortality risk, disease risk, or other measures of interest. If an individual has a predicted age higher than chronological age, this is known as accelerated biological age. A major challenge facing the use of clocks at present is that most clocks are inscrutable, in the sense that little to nothing is known of how the individual values making up the clock algorithm are linked to specific underlying mechanisms of aging or specific outcomes of age-related dysfunction and disease.
One path to making this situation better is to run as many clocks as possible in as many human studies as possible: accumulate as much data as possible and see what emerges from that data. Thus we should expect to see many publications that are similar to today's open access paper in the years ahead, in which researchers apply one or more clocks to a specific study and health context. Here, the context is age-related cognitive decline, assessed in a test conducted seven years after the collection of initial clock data. As we might expect for a good clock, accelerated biological age correlated with a greater loss of cognitive function.
Neuropsychological (NP) tests are typically used to measure cognitive functions for individuals, focusing on one or several specific cognitive domains. For example, the long-used Clock Drawing Test (CDT) evaluates executive functioning and spatial skills. The digital Clock Drawing Test (dCDT), a digital version of the standardized CDT done with pen and paper, provides a much more robust assessment of cognitive functioning. We used linear mixed regression to evaluate the associations between epigenetic aging metrics (Horvath, Hannum, GrimAge, PhenoAge, DunedinPACE) and digital Clock Drawing Test (dCDT) scores in the Framingham Heart Study (FHS), adjusting for covariates.
Among the 1,789 FHS participants (mean age 65 ± 13, 53% women), higher epigenetic age acceleration metrics at baseline predicted lower dCDT scores approximately seven years later. The magnitude of these associations was greater in older participants (≥65 years, n = 985). The strongest association was observed between the dCDT total score and DunedinPACE in the full sample (beta = -2.1), the younger (<65 years; beta = -1.9), and older (beta = -2.2) age groups. Additionally, the dCDT total score was associated with age acceleration estimated by Horvath (beta = -1.9) and PhenoAge (beta = -2.5) in older participants, while not in the full sample or younger participants. Furthermore, higher levels of DNAm-based PAI1 (beta = -0.9) and ADM (beta = -2.9), components of GrimAge, were significantly associated with lower dCDT total scores. In analyses of cognitive subdomains, simple motor function was significantly associated with DunedinPACEin both age groups, and with GrimAge in the older age group, suggesting that deterioration in various organ systems may particularly impact this domain.
Our findings suggest that advanced biological aging, particularly as captured by DunedinPACE and GrimAge components, is significantly associated with poorer cognitive performance measured by dCDT, especially in older adults, highlighting a potential link between systemic aging processes and cognitive decline.
View the full article at FightAging
