Repeat blood draws revealed variability in most epigenetic clocks
Repeat blood draws revealed variability in most epigenetic clocks
On 29 July, Raghav Sehgal and colleagues compared 18 epigenetic clocks. Most models produced similar results when the same DNA sample was analyzed repeatedly, but were much less consistent across blood draws taken after meals, stress, and other short-term exposures.
Epigenetic clocks use chemical marks on DNA to estimate an age-related measure. Researchers use them in studies of diets and drugs by collecting blood before and after an intervention and checking whether the value has changed. This design requires the difference between the two samples to reflect the intervention rather than the conditions under which the blood was collected.
In an Aging Cell paper, the authors separated two types of repeatability tests. In the first, the laboratory processed the same DNA sample twice. In the second, blood was drawn from the same person at different times, after the body had responded to food, stress, air pollution, or a change in altitude.
For repeated processing of the same DNA sample, the agreement coefficient exceeded 0.90 for most clocks. Across four datasets with repeat blood draws, the pooled estimate for most models was approximately 0.4–0.7. The food and social stress experiments each included 34 participants, with blood collected from each participant four times during one day. After adjustment for immune cell proportions inferred from methylation, most clocks became even less consistent.
PCGrimAge and SystemsAge were more consistent than the other clocks. In PC-based variants, the original DNA marks are first combined into shared components, which are then used to calculate the age-related measure. A clock's technical precision was almost unrelated to its consistency across blood draws. A successful laboratory replicate therefore does not establish that a model is suitable for tracking changes within a person over time.
In the authors' subsequent analyses, the more consistent clocks also showed a more stable association with future cognitive decline. When intervention responses were analyzed, estimates from the less consistent models changed more when a different technical replicate was selected. In anti-aging experiments, this means that collection time, food intake, and the person's condition before the blood draw become part of the measurement itself.