Five epigenetic clocks reveal shared and distinct signals of aging

A large multi-omics study maps why epigenetic clocks capture different aspects of aging and shows how gene-expression scores may sharpen their interpretation.

Study: How epigenetic clocks tick: unpacking the black box by deciphering biological pathways and transcriptomic signatures of accelerated aging. Image Credit: Lightspring / Shutterstock

In a recent study published in the journal npj Aging, a group of researchers identified the biological pathways and gene expression signatures associated with widely used epigenetic clocks and evaluated how transcriptomic aging gene scores (TAGS) were associated with aging-related health outcomes.

Background

Why do some people remain healthier than others despite being the same age? The difference may be due to biological aging rather than chronological age.

Scientists use epigenetic clocks to measure how fast the human body ages. Epigenetic clocks can better predict health risks than chronological age, but the biological processes underlying their predictions remain unclear. These clocks measure age-related DNA methylation patterns, which may be associated with gene activity.

Examining associations between epigenetic clocks and gene activity could improve understanding of their biological interpretation.

About the study

The researchers analyzed data from participants enrolled in the HRS Venous Blood Study, focusing on those with both DNAm and RNA-seq data available. Among the 4,018 participants with DNAm results, 3,227 were studied due to the availability of complete transcriptomic and covariate data.

To develop and internally assess the transcriptomic scores, the analytic sample was randomly divided into two sets: a training set comprising 80% of the total sample and a hold-out test set comprising 20% of the total sample. The study analyzed the five epigenetic clocks: Horvath, Hannum, PhenoAge, GrimAge, and DunedinPACE clocks.

Using the training dataset, the investigators performed differential gene expression (DGE) analyses to identify genes whose expression levels were associated with each epigenetic age acceleration measure.

The resulting gene lists were analyzed using GSEA to identify biological pathways associated with each clock. The analysis also addressed the similarities and differences between the shared and specific pathways among the five clocks.

Finally, the researchers developed TAGS from the differentially expressed genes (DEGs) and evaluated their relationships with epigenetic age acceleration measures and multiple aging-related health outcomes in the hold-out test dataset.

The research assessed the transferability of the transcriptomic scores by comparing them with their parent epigenetic clocks and available disease-related outcomes across three external datasets.

Study results

The analysis revealed substantial differences in the molecular signatures represented by the five epigenetic clocks. The number of DEGs varied considerably, ranging from 49 for the Horvath clock to 3,204 for DunedinPACE. These differences did not match the number of CpGs in each clock. This shows that larger clocks did not necessarily capture more gene expression changes.

No DEGs were shared across all five clocks, and DunedinPACE had the highest proportion of unique DEGs, suggesting that each clock captures distinct processes in biological aging. The greatest DEG overlap was observed among the second- and third-generation clocks, especially GrimAge, PhenoAge, and DunedinPACE.

Further analyses showed that the biological pathways associated with the clocks also differed markedly. DunedinPACE had the most associated pathways, while Horvath had the fewest. No Reactome biological pathway was common to all clocks, although several immune-related pathways, including neutrophil degranulation, innate immune system signaling, and immune system signaling, were shared by Hannum, PhenoAge, GrimAge, and DunedinPACE.

In addition, pathways associated with antimicrobial actions were observed in subsets of the clocks. The analysis of each clock revealed distinct biological pathways enriched for genes: the Horvath clock was associated with pathways involved in metabolism and signal transduction, while the Hannum clock was associated with homeostasis and vascular wall processes.

The GrimAge clock was linked to interferon signaling, whereas the PhenoAge clock was linked to cellular senescence pathways, and the DunedinPACE clock was linked to a variety of biological pathways, including protein metabolism, immune signaling, nervous system development, and cellular respiration.

When broader Gene Ontology (GO) biological processes were examined, greater overlap among the clocks emerged than in individual pathways. The analysis produced four principal functional themes: metabolic and macromolecular processes, developmental processes, immune system functions, and regulatory and signaling pathways. This shows that although the molecular signatures of the clocks are different, they share several broad biological processes associated with aging.

The researchers then developed TAGS for each clock and tested their associations with aging-related outcomes. Each TAGS score showed positive correlations with its corresponding epigenetic clock, with the strongest correlation observed for DunedinPACE. In several cases, TAGS showed larger associations with aging-related measures than the original epigenetic clocks.

Stronger associations were seen for mortality, frailty, ADLs, walking speed, heart problems, diabetes, lung problems, telomere length, and interleukin-6. TAGS were not significant predictors of psychological problems, while associations with grip strength, cognitive ability, and interleukin-10 were mixed.

Validation in external datasets produced comparable but variable correlations with parent epigenetic clocks and associations with disease-related outcomes, providing preliminary support for the transferability of the transcriptomic scores.

Conclusion

The research showed that the known epigenetic clocks were associated with different biological pathways, rather than a single universal aging mechanism, and that epigenetic age-acceleration measures were associated with blood gene-expression patterns, helping clarify the clocks’ biological interpretation.

TAGS complemented existing epigenetic clocks and, in several cases, showed stronger associations with aging-related diseases, physical function, and mortality. Because the analyses used blood samples from predominantly White older adults, and TAGS testing used a hold-out sample from the same cohort, further validation in diverse independent populations is needed before clinical use.

The results make the epigenetic clocks more biologically interpretable and support their applications in aging research and future validation studies.