Steve Horvath, a German-American aging researcher, geneticist, and biostatistician at UCLA, has developed a series of “epigenetic clocks” — algorithms that read DNA methylation patterns to predict biological age and mortality risk with remarkable accuracy. The episode explores how these clocks work, why they matter for longevity research, and what open questions they raise about the nature of aging itself.
What epigenetic clocks are and how they work
DNA methylation is a chemical process in which a methyl group attaches to a specific genetic sequence, effectively turning that gene off. This is what allows different cell types (neurons, liver cells, blood cells) to develop from the same underlying DNA.
Over time, methylation patterns change in a remarkably consistent, linear fashion across tissues — especially after about age 20 in humans. This regularity is what makes it possible to build a clock.
An epigenetic clock is a prediction model: you take a blood or saliva sample, measure methylation at specific sites in the genome, and the algorithm outputs a predicted age. The pan-tissue clock (2013) was Horvath’s breakthrough — the first clock that works across virtually all human tissues and cell types.
The GrimAge clock goes further: it was designed specifically to predict mortality risk and time to death, named after the Grim Reaper for that reason. It is currently the best mortality predictor among Horvath’s clocks.
A newer development is the universal mammalian clock, a preprint that extends epigenetic age prediction across almost all mammalian species, suggesting the underlying methylation process is deeply conserved.
Why epigenetic clocks matter for longevity research
The core problem in aging research has been that clinical trials for longevity are extraordinarily expensive and slow — you would need to follow a 50-year-old subject for decades to see if an intervention delayed death. This is unlike diabetes trials, where blood glucose gives you a quick readout.
Epigenetic clocks offer a surrogate endpoint: a blood draw that can reveal whether an intervention (better sleep, a drug, a lifestyle change) has shifted someone’s biological age in a short time frame — potentially one to two years.
This could dramatically lower the cost and duration of longevity trials and open the field to more researchers and “crazy ideas” that currently can’t get funded or approved through IRBs.
Horvath notes that many diseases (cancer, Alzheimer’s, Parkinson’s) are downstream of aging. If you’re running a trial for any drug, it would be valuable to also measure epigenetic age to see whether the treatment has an unexpected anti-aging side effect.
Open questions about methylation and aging
It is still unknown whether DNA methylation causes aging or merely correlates with it. If it does cause aging, it raises a puzzle: why would a process essential for cell differentiation and development also be the thing that kills us?
The rate of methylation change is remarkably constant after development — Horvath calls this “the clock ticks at a constant rate.” The underlying mechanism is not understood. Speculations include maintenance processes or circadian rhythm, but Horvath’s own circadian rhythm studies do not support that link. He plans to study astronauts to see whether space travel (cosmic radiation, microgravity) affects the methylome, though his existing data suggest X-rays and similar radiation do not meaningfully alter methylation.
Lifestyle factors (diet, exercise, sleep) appear to have a surprisingly weak effect on epigenetic age. Horvath acknowledges this is shocking and explains why some people with unhealthy habits live long while some health-conscious people die early. He personally eats a lot of salami and is candid that the field does not yet have strong evidence that specific lifestyle changes reliably reverse epigenetic age.
Promising interventions and future directions
Young plasma treatment: A pilot study in rats (collaboration with Harold Katcher and Akshay Sanghavi) found that young plasma-based treatments reduced epigenetic age in liver, blood, heart, and brain by over 50% in some tissues. Horvath finds the results so dramatic that he is nervous and is awaiting a replication dataset before drawing firm conclusions.
Epigenetic reprogramming: A small number of genes (three to five) can, when modified, potentially lead to dramatic age reversal. This idea has spawned companies, including one Horvath is involved with founded by David Sinclair of Harvard, which has used the approach to regenerate damaged optic nerves in the eye via viral delivery. The hope is to extend this to other organs.
Histone modifications vs. methylation: Some researchers argue histone modifications would make better clocks because they are more dynamic than cytosine methylation. Horvath chose methylation because the technology was mature and affordable; he acknowledges this is a missed opportunity and encourages young researchers to pursue histone-based clocks.
Bioinformatics opportunities: Horvath believes the field is rich with low-hanging fruit for data scientists — public datasets can be mined to find genes and proteins related to maximum lifespan or responses to anti-aging interventions. His personal standard is finding a strong signal in datasets of ~100 people, not the massive genome-wide association studies requiring millions of participants that have yielded disappointing results in traditional statistical genetics.
Social and psychological factors: Loneliness and social isolation appear to be major mortality risks, and Horvath agrees it would be valuable to study how social connectedness affects methylation, though he is not aware of such studies yet.
