Epigenetic clocks measure 'biological age' by analyzing patterns in DNA methylation—chemical tags on DNA that change with age. Unlike chronological age, these clocks could theoretically reveal who is aging faster at the cellular level and who might be at risk for age-related diseases like cognitive decline. However, previous studies have yielded inconsistent results about whether epigenetic clocks actually predict brain aging. This study aimed to settle the question by systematically comparing 14 different epigenetic clocks in a well-characterized cohort of older German adults.
The researchers analyzed data from 1,014 participants in the Berlin Aging Study II (BASE-II), who underwent genetic testing and cognitive assessments at baseline and two follow-up visits. They calculated biological age using all 14 epigenetic clocks—ranging from first-generation clocks (Horvath, Hannum) developed around 2013 to newer fifth-generation frameworks (SystemsAge). They then tested whether each clock's estimate of 'age acceleration' (the difference between epigenetic age and actual age) was associated with cognitive performance across multiple domains and time points.
The standout finding: DunedinPACE, a third-generation clock, showed the strongest and most consistent associations with cognitive decline. This clock performed better across both cross-sectional comparisons (at single time points) and longitudinal tracking (over time). Interestingly, the fifth-generation SystemsAge framework also showed robust associations, while older clocks (Horvath v1, Hannum, GrimAge v2, PhenoAge) either showed weak or no consistent relationship with cognitive outcomes. Telomere length estimates derived from DNA methylation did not predict cognition well. The authors also noted that DunedinPACE was more informative than sex-specific patterns or comparisons with frailty measures.
Important limitations: This is a preprint (not yet peer-reviewed), so findings await validation by independent groups. The study is observational, meaning it shows association, not causation—we cannot conclude that biological age acceleration *causes* cognitive decline. The cohort is from a single geographic and ethnic background (Berlin, primarily European ancestry), which may limit generalizability. The authors do not report effect sizes, confidence intervals, or multiple-testing corrections clearly, making it difficult to assess statistical robustness. Finally, with zero citations so far, there is no external replication.
Why this matters: If DunedinPACE replicates in independent cohorts, it could become a practical biomarker for predicting cognitive aging in clinical settings—potentially allowing earlier identification of people at risk for dementia. However, biomarkers are only useful if they enable better prevention or treatment. This study does not show that targeting DunedinPACE acceleration slows cognitive decline. The findings also highlight an important principle in aging research: not all biological clocks are created equal, and newer does not always mean better—but systematic evaluation of multiple tools helps narrow the field.
0 Comments
Log in to join the discussion.