Comparing the Responsiveness of Epigenetic Aging Biomarkers

In a recent study, researchers have evaluated how responsive several DNA methylation (DNAm) biomarkers are to a range of longevity interventions [1].

Waiting too long

Testing lifespan-extending interventions in humans has a major problem: it can take many years, or even decades, to see results. Nobody wants to wait that long. Therefore, it is essential to find proxies, such as biomarkers, that accurately reflect whether an intervention is working. Some of the most popular biomarkers in the aging field are epigenetic aging clocks, DNAm biomarkers that are intended to reflect a person’s biological age or rate of aging.

While these clocks are becoming increasingly popular, they are still not sufficiently validated to serve as surrogate endpoints for human longevity clinical trials, and there isn’t enough data on their responsiveness to longevity-promoting interventions.

“If these new biomarkers are eventually validated to predict long-term health, scientists will be able to evaluate anti-aging therapies much faster,” said Raghav Sehgal, the lead author of this study. “Instead of waiting decades for evidence from clinical trials, we’ll be able to see which interventions are effective and in which people in a few years or even months.”

Multiple comparisons

To analyze how epigenetic aging clocks and other biomarkers respond to various aging interventions, the researchers used data from 51 existing longitudinal interventional studies. The researchers grouped the interventions in their dataset into four groups: “lifestyle (including diet and exercise), pharmacological (for example, metformin, rapamycin, semaglutide, ketamine and anti-TNF therapy), supplements (for example, omega-3 fatty acids and folate) and medical procedures (for example, hyperbaric oxygen therapy, organ transplants and gene therapy).”

Then, they calculated the effect of each intervention using 16 epigenetic clocks and 94 other DNA methylation biomarkers; they used ‘DNAm biomarkers’ as an umbrella term for both of those tools.

Using multiple biomarkers can serve two functions. First, surrogate biomarkers can help predict long-term outcomes. Second, discovery biomarkers can help explain which molecular mechanisms an intervention affects [2, 3].

“What we did was pretty unique,” Sehgal said. “We already know that certain things might prolong healthspan and lifespan. There are data from retrospective analyses as well as animal models. We took all that knowledge along with the real-world clinical studies to identify which interventions in humans were slowing down aging across the board in these known biomarkers.”

Comparing multiple interventions using multiple biomarkers allowed the authors to identify patterns in how different intervention types affect health and aging. For example, they noted that pharmacological interventions produced stronger DNAm biomarker responses than any other category and were the only ones with significantly larger effect sizes. They were also one of two categories, along with a lifestyle intervention, to show significantly reduced epigenetic age. The authors hypothesize that pharmacological interventions’ robust effects may stem from their ability to target inflammation and metabolic pathways such as TNF, AMPK, and mTOR.

Replications matter

The authors note that their compiled studies include replicated studies of the same intervention and use multiple clocks to measure similar outcomes, which allows them to test whether certain interventions have consistent effects.

To identify intervention types with a strong, consistent effect on biomarkers, the authors propose two requirements: “the intervention should modify DNAm biomarkers of a given generation to the same magnitude and direction in a particular study” and “a second study of the same intervention should modify the same biomarkers.”

Both requirements were met by therapies that reduce tumor necrosis factor (TNF) and are used to treat conditions such as inflammatory arthritis and inflammatory bowel disease, suggesting that such therapies may prevent pathological aging in patients with autoimmune disorders. The requirements were also met by two different types of Mediterranean diets in healthy populations.

Not all DNAm biomarkers are created equal

Tests of responsiveness and concordance of the DNAm biomarkers, defined as ”the likelihood that if a DNAm biomarker detected a significant effect, then others would agree on the effect”, showed the superiority of generation 2+ biomarkers.

Similarly, biomarkers differed in sensitivity. Focusing on lifestyle and pharmacological interventions, the researchers noted intervention-type-specific sensitivity across multiple biomarkers and differences in effect-size magnitude between categories: DunedinPACE was highly responsive in the lifestyle category, but in the pharmacological category, it was affected by fewer interventions than other second-generation biomarkers.

Study population health status also affected DNAm biomarker responses, with several biomarkers significantly more responsive in disease groups than in healthy populations, similar to previous studies suggesting that people with health problems may have more room for improvement than healthy people [4, 5]. Only DunedinPACE showed similar response levels in both populations, suggesting that it can be applied in both contexts.

The authors suggest that the training population on which those biomarkers were developed partly explains these observations, since biomarkers trained on populations with more diverse health statuses are more sensitive to changes in populations with disease, while those trained on healthier groups have higher sensitivity in healthier cohorts.

Understanding biology

Those researchers also went a step further, using Generation X (GenX) DNAm biomarkers to gain mechanistic insights into biological aging, which can later support more precise, targeted aging therapies. With this technique, the authors observed which organs are affected by different interventions, such as a reduction in lung system scores upon smoking cessation or the impact of metformin on inflammatory, brain-related, and metabolic pathways, as well as the effect of different diets on various biomarkers and scores, which can aid in measuring diet-specific impacts on different aspects of aging.

Such analyses let researchers focus on biomarkers specific to certain pathways, capturing earlier changes in those systems and organs that composite biomarkers can miss by “averaging out” effects across components.

Narrowing down the choices

In summary, understanding which DNAm biomarkers work best and provide the most useful information across different interventions helps researchers choose the best tools for future clinical trials, minimizing time and cost while maximizing meaningful results.

This analysis of DNAm biomarkers suggests that clinical trials should prioritize generation 2+ clocks, mainly DunedinPACE and PCGrimAge, since their responses were the strongest and the most consistent. However, they still need validation as surrogate endpoints in human clinical trials.

“As researchers, we have a long way to go in understanding the aging process and whether the steps we take to manage it really work,” said Sehgal.

Literature

[1] Sehgal, R., Borrus, D., Armstrong, J. F., Gonzalez, J., Kasamoto, J., Markov, Y., Priyanka, A., Smith, R., Carreras-Gallo, N., Lasky-Su, J., Dwaraka, V. B., Corley, M. J., & Higgins-Chen, A. (2026). Responsiveness of epigenetic aging biomarkers to longevity interventions in humans. Nature medicine, 10.1038/s41591-026-04562-9. Advance online publication.

[2] Moqri, M., Herzog, C., Poganik, J. R., Biomarkers of Aging Consortium, Justice, J., Belsky, D. W., Higgins-Chen, A., Moskalev, A., Fuellen, G., Cohen, A. A., Bautmans, I., Widschwendter, M., Ding, J., Fleming, A., Mannick, J., Han, J. J., Zhavoronkov, A., Barzilai, N., Kaeberlein, M., Cummings, S., … Gladyshev, V. N. (2023). Biomarkers of aging for the identification and evaluation of longevity interventions. Cell, 186(18), 3758–3775.

[3] Fleming, T. R., & Powers, J. H. (2012). Biomarkers and surrogate endpoints in clinical trials. Statistics in medicine, 31(25), 2973–2984.

[4] Yao, A., Gao, L., Zhang, J., Cheng, J. M., & Kim, D. H. (2024). Frailty as an Effect Modifier in Randomized Controlled Trials: A Systematic Review. Journal of general internal medicine, 39(8), 1452–1473.

[5] Pandey, A., Kitzman, D. W., Nelson, M. B., Pastva, A. M., Duncan, P., Whellan, D. J., Mentz, R. J., Chen, H., Upadhya, B., & Reeves, G. R. (2023). Frailty and Effects of a Multidomain Physical Rehabilitation Intervention Among Older Patients Hospitalized for Acute Heart Failure: A Secondary Analysis of a Randomized Clinical Trial. JAMA cardiology, 8(2), 167–176.