Martin Jensen proposed validating epigenetic clocks (measures based on chemical marks on DNA) against health outcomes before using them to assess the effects of therapies
Martin Jensen proposed validating epigenetic clocks (measures based on chemical marks on DNA) against health outcomes before using them to assess the effects of therapies
On August 22, Martin Borch Jensen responded to TranslAGE, a new database that tracks how epigenetic clocks respond to interventions. He called for changes in these measures to be compared with changes in people’s health.
On August 21, Nature Medicine published a paper on TranslAGE. The authors compiled 51 longitudinal studies in which blood samples were collected from participants before and after an intervention. The database contains 3 128 samples. They applied the same method to every dataset to calculate 16 epigenetic clocks, which use chemical marks on DNA to estimate age-related changes or mortality risk. The paper makes it possible to compare which clocks change after drugs, diets, and other interventions.
The response map for the 16 clocks shows that different measures respond differently to drugs and lifestyle changes. Jensen proposes comparing these shifts with outcomes that matter to patients, including physical function, disease, and mortality. This could make the clocks useful measures for assessing the benefits of an intervention in a clinical trial.
In his thread about TranslAGE, he points to the COSMOS study. In a randomized trial involving 21 442 older participants, daily multivitamins did not produce a statistically significant reduction in the overall cardiovascular outcome or all-cause mortality over a median of 3.6 years. Jensen uses this example to make a specific point: a change in an individual biomarker must be compared with what happened to the participants’ health.
Jensen proposes an independent test for this purpose. Clock developers would submit predictions for a set of interventions in advance, without knowing the outcomes. The predictions would then be compared with the observed data. In a detailed essay, he describes an open, standardized validation process and clocks that can be applied to both humans and mice. Lifespan can be measured directly in mouse experiments, allowing it to be compared with clock measurements more quickly.
This test compares an intervention’s health outcome with how accurately the clock reflects that outcome.