Responsiveness of epigenetic aging biomarkers to longevity interventions in humans

Raghav Sehgal, Daniel Borrus, Albert Higgins-Chen

1. 01

   Mortality- and pace-of-aging-trained clocks (generation 2+, reliability-optimized) showed the highest responsiveness and the most agreement with each other

2. 02

   First-generation chronological-age predictors were substantially less responsive to interventions

3. 03

   Pharmacological and lifestyle interventions drove the strongest DNAm biomarker responses

4. 04

   Study population characteristics and study duration were key determinants of whether any response was detectable

5. 05

   Explainable multi-subscore clocks provided mechanistic specificity that single-score clocks did not

6. 06

   Implication: clock choice is not interchangeable and a single composite biological-age number is the wrong unit of inference

Study Design

Meta-Analysis

Sample Size

Not reported

Study Duration

Varied; pooled across 51 longitudinal interventional studies

Methodology

Curation of TranslAGE, a harmonized database of 51 public and private longitudinal interventional studies with DNA methylation data; consistent computation of 16 prominent epigenetic clocks plus 94 additional DNAm biomarkers across all studies; systematic comparison of responsiveness patterns by clock generation, intervention class, population and duration.

Limitations

Aggregates heterogeneous studies with differing designs, durations and populations; responsiveness is not the same as validated surrogacy for clinical outcomes; no clock met full surrogate-endpoint criteria; accompanying News & Views author discloses GrimAge/PhenoAge patent interests and Altos Labs equity.