Thompson · Aging 2018 · Comparative biomarker development and validation study in animal datasets · n=1,189 mouse samples

A multi-tissue full lifespan epigenetic clock for mice.

Cited 242 times in the scientific literature.

Level 5 - mechanism / opinion, no new human data

Preclinical animal and computational modeling study without human clinical data.

PubMed 30348905 · doi:10.18632/aging.101590 · record verified 2026-08-30

What was done

Researchers pooled DNA methylation data from previous publications and in-house datasets totaling 1,189 mouse tissue samples across 193,651 CpG sites. They developed four epigenetic clocks comparing two machine learning methods (elastic net versus ridge regression) and two sets of features (all CpGs versus highly conserved CpGs). Clocks were evaluated for their ability to predict chronological age and detect the anti-aging effects of calorie restriction and growth hormone receptor knock-out (dwarf mice).

What was found

Accurate age estimators were developed from conserved CpGs, but the most accurate chronological age predictor resulted from elastic net regression across all CpGs. All clock models detected the anti-aging effect of calorie restriction; however, only ridge regression-based clocks detected the slowed epigenetic aging effect in dwarf mice. Specific numerical performance metrics were not reported in the abstract.

Why it matters

This work establishes multi-tissue epigenetic clocks across the mouse lifespan and reveals that optimizing purely for chronological age accuracy can compromise a clock's sensitivity to specific anti-aging interventions.

Limits

The study is restricted to mouse models. The abstract does not provide exact numerical accuracy metrics (such as median absolute error or correlation coefficients) or detail the breakdown of specific tissue types analyzed.

Cited by