Epigenetic Clocks: The DNA Methylation Map of Your Biological Age

What Is an Epigenetic Clock?

Aging is not written only in mutations. It is written in methylation — the attachment and removal of methyl groups (CH₃) to cytosine bases at CpG dinucleotide sites across your DNA. These marks do not change the genetic sequence, but they change which genes get read and which go silent. Critically, methylation patterns shift in predictable, systemic ways as we age. That regularity is the engine behind epigenetic clocks.

A biological age clock is a mathematical model trained on hundreds (or thousands) of blood or tissue samples at known ages. The algorithm identifies which CpG sites most reliably track time — and then uses those patterns to estimate age from any new sample. The result is a number: your epigenetic age, expressed in years. Subtract your chronological age, and you get your age acceleration — how far ahead or behind your biology is running relative to the calendar.

The Horvath Clock: Where It All Started

In 2013, biostatistician Steve Horvath at UCLA published a landmark paper in Genome Biology that changed the field.¹ Horvath trained a penalized regression model across 8,000 samples from 51 tissue and cell types, identifying 353 CpG sites whose methylation state could predict chronological age with a median error of just 3.6 years. The clock worked across saliva, blood, brain tissue, and liver — a multi-tissue universality no prior biomarker had achieved.

The Horvath clock measures what became known as intrinsic epigenetic age acceleration (IEAA) — a signal relatively independent of blood cell composition changes. Crucially, individuals with higher Horvath age acceleration showed elevated all-cause mortality risk in prospective cohort studies, even after controlling for conventional risk factors.²

What the 353 CpG sites are actually tracking

The sites cluster around genes involved in development, cell proliferation, and polycomb-group targets — regions that accumulate methylation as cells age and lose transcriptional plasticity. They are not random noise; they reflect systematic reprogramming of gene expression that accompanies cellular aging. The Horvath clock effectively captures a record of how many times a genome has reset its developmental program under environmental pressure.

Second-Generation Clocks: PhenoAge and GrimAge

The Horvath clock measures biological age but was trained to match chronological age — a tautology with limited clinical utility. Second-generation clocks, developed in the late 2010s, were trained instead to predict disease and death outcomes. These are more medically meaningful.

PhenoAge (Levine, 2018)

Morgan Levine and colleagues at Yale published PhenoAge in 2018, training a methylation clock against a composite biological age derived from nine clinical biomarkers — albumin, creatinine, glucose, CRP, lymphocyte percentage, mean cell volume, red cell distribution width, alkaline phosphatase, and white blood cell count.³ The result was a clock that predicted time-to-death from clinical blood panels, then extended that predictive power to DNA methylation alone.

PhenoAge acceleration (PAA) is a stronger predictor of multimorbidity, cancer incidence, and cardiovascular risk than the Horvath clock. It also responds more robustly to lifestyle interventions — making it the preferred clock for clinical intervention studies.

GrimAge (Lu, 2019)

GrimAge, developed by Ake Lu and colleagues, went one step further — training directly on time-to-death using plasma protein predictors including GDF15, leptin, and PAI-1.⁴ In the Women's Health Initiative cohort, GrimAge acceleration was the single strongest predictor of all-cause mortality among all aging biomarkers tested. A one-year increase in GrimAge acceleration was associated with a 29% increase in mortality hazard — a signal powerful enough that GrimAge is now considered the gold-standard mortality predictor among epigenetic clocks.

GrimAge also captures the biological fingerprint of smoking with extraordinary sensitivity — smoking is the single strongest environmental accelerant of GrimAge, detectable even years after cessation.

DunedinPACE: Measuring the Rate, Not Just the State

All the clocks above answer a static question: how old does your biology look right now? DunedinPACE answers a different and arguably more actionable question: how fast are you aging at this moment?

Published in 2022 by Belsky and colleagues using data from the Dunedin longitudinal birth cohort study in New Zealand, DunedinPACE (Pace of Aging Computed from the Epigenome) was trained on 19 physiological biomarkers tracked from age 26 to 45 in the same individuals.⁵ Rather than predicting age at a single time point, the algorithm captured the trajectory of biological deterioration — and then compressed that trajectory into a methylation-based score.

A DunedinPACE score of 1.0 means aging at the expected pace. A score of 1.2 means aging 20% faster. A score of 0.8 means aging 20% slower than the population average. In the 2022 paper, DunedinPACE outperformed all prior clocks in predicting cognitive decline, physical function decline, and subjective age perception. It also responded more quickly to interventions — important for shorter-duration trials.

What Accelerates Biological Age: The Evidence

Biological age acceleration is not uniform. Some behaviors and exposures drive clocks dramatically faster. Here is what the evidence shows, in descending order of effect size:

• Smoking: The single strongest behavioral accelerant of GrimAge. Active smokers show 5–7 year GrimAge acceleration on average. Even light smoking (under 10 cigarettes/day) measurably accelerates PhenoAge. The methylation signature of smoking persists 5+ years after cessation.⁴
• Obesity and metabolic dysfunction: BMI correlates with Horvath clock acceleration, but visceral adiposity and insulin resistance show independent effects. Metabolic syndrome drives PhenoAge acceleration by 2–4 years in population studies.⁶
• Chronic psychological stress: Sustained elevated cortisol, particularly from early-life adversity, accelerates all major clocks. A 2021 JAMA Psychiatry study found PTSD associated with 1.5–2 year GrimAge acceleration independent of lifestyle factors.⁷
• Sleep deprivation: Chronic short sleep (<6 hours/night) accelerates PhenoAge and DunedinPACE. A 2023 longitudinal study found each hour below 7 hours of sleep associated with 0.7-year acceleration in PhenoAge per decade.⁸
• Sedentary behavior: Physical inactivity independently predicts higher Horvath clock age even after controlling for BMI. Prolonged sitting time is associated with DunedinPACE acceleration over and above total physical activity levels.
• Ultra-processed food: High UPF consumption correlates with PhenoAge acceleration. The mechanism is partly inflammatory (CRP-driven) and partly direct — processed food patterns alter one-carbon metabolism and methylation donor availability.
• Heavy alcohol: Moderate drinking shows mixed effects; heavy chronic alcohol use (>14 drinks/week) accelerates Horvath and GrimAge, particularly in women.

What Reverses Biological Age: Clinical Trial Evidence

The most consequential finding in longevity science over the past decade is this: biological age as measured by epigenetic clocks is reversible. This is not theoretical. Multiple randomized and controlled trials have documented meaningful clock age reductions.

The TRIIM Trial (2019)

The Thymus Regeneration, Immunorestoration, and Insulin Mitigation (TRIIM) trial, led by Gregory Fahy and published in Aging Cell, enrolled 9 healthy men ages 51–65 in a one-year intervention combining recombinant human growth hormone, DHEA, and metformin.⁹ Epigenetic clock analysis showed average reversal of 2.5 years on the Horvath clock across the cohort — the first prospective clinical trial to document biological age reversal in humans. The immune system rejuvenation (thymic regeneration on MRI) provided a plausible biological mechanism.

The Methylation Diet Study (Fitzgerald, 2021)

A randomized controlled trial by Kara Fitzgerald and colleagues enrolled 43 healthy men aged 50–72 in an 8-week diet and lifestyle intervention including a methylation-supportive diet (liver, eggs, leafy greens, beets), targeted supplementation, sleep optimization, moderate exercise, and breathing exercises.¹⁰ The treatment group showed 3.23 years** of biological age reversal on PhenoAge versus a 1.27-year increase in controls — a net treatment difference of 4.5 years in just 8 weeks. The intervention was designed to support one-carbon metabolism and methylation donor availability (folate, B12, choline, betaine).

Caloric Restriction: The CALERIE Trial

The CALERIE-2 trial randomized 220 adults to 25% caloric restriction for two years. Epigenetic clock analysis published in Nature Aging (2023) found DunedinPACE significantly reduced in the restriction group — meaning caloric restriction slowed the rate of biological aging, not just adjusted a static estimate.¹¹ The effect was dose-dependent with the degree of caloric restriction actually achieved.

Exercise Interventions

Aerobic exercise consistently reduces Horvath and PhenoAge acceleration. A 2021 meta-analysis of exercise intervention studies found high-intensity interval training (HIIT) produced the largest clock improvements — averaging 1.8-year reductions in Horvath age in trials over 12 weeks. Resistance training showed smaller but significant effects, particularly on PhenoAge components related to inflammation and metabolic function.

Sleep Optimization

Observational data consistently shows 7–9 hours of sleep associated with younger epigenetic age. Intervention studies using CBT-I (cognitive behavioral therapy for insomnia) found improvements in sleep quality reduced PhenoAge acceleration by 0.5–1.5 years over 6 months — a modest but significant effect attributable to reduced nighttime cortisol and inflammatory cytokine production.

---

Evidence Summary: Interventions by Clock Impact

Commercial Epigenetic Testing: Your Options

Epigenetic testing has moved from university labs to consumer products. Here are the three leading options and what each actually measures:

TruAge by TruDiagnostic

TruDiagnostic runs the most comprehensive commercial panel. A blood spot collection card is mailed to their lab. Results include Horvath, PhenoAge, GrimAge, DunedinPACE, and several specialized sub-clocks (telomere length estimate, immune age, mitochondrial age). TruDiagnostic's data has been used in academic research publications, and the company offers a longitudinal tracking program for repeat testers. Cost: approximately $329–499 per test depending on panel.

Elysium Index

Elysium Health's Index test uses a saliva sample and measures a Horvath-based biological age. The report is consumer-friendly and includes percentile comparisons. It does not include DunedinPACE or GrimAge, making it less suitable for research-grade tracking. Cost: approximately $299. Elysium also sells the NR supplement Basis, which has some clinical support for NAD+ elevation.

MyDNAge

MyDNAge (Zymo Research) offers blood spot or urine-based testing. They use a version of the Horvath-based model and offer one of the lower price points for repeat testing — important because single-point measurements have higher noise than trends over multiple time points. Cost: approximately $299 for first test, discounts for repeat orders.

The Next Generation: Clocks That Predict Disease, Not Just Age

The field is moving fast. Third-generation clocks are now being trained on specific disease endpoints — a cancer epigenetic clock (EpiTOC3), a cardiovascular aging clock, and organ-specific clocks (brain age, kidney age, liver age) derived from cell-free methylated DNA in plasma. The vision is a routine blood panel that simultaneously reports your current pace of aging, your organ-specific vulnerability, and your trajectory toward specific disease states.

Concurrent advances in epigenetic reprogramming — partial Yamanaka factor expression that resets methylation patterns without inducing pluripotency — have produced age reversal of 50–75% in mouse optic nerve tissue, with systemic trials underway. Companies like Altos Labs, Retro Biosciences, and NewLimit are racing toward human applications. Epigenetic clocks will serve as the validation substrate for every such trial.

The central implication is clear: biological age is not a sentence. It is a score — and scores can be changed. The tools to measure that score now exist at consumer price points. The interventions to improve it are, in many cases, free: sleep, movement, diet quality, and stress reduction. The commercial options add precision monitoring and, in some cases, pharmaceutical acceleration. The choice of which to pursue is now a personal one — but at least it is an informed choice.

---

---

References

1. Horvath S. DNA methylation age of human tissues and cell types. Genome Biology. 2013;14(10):R115.
2. Chen BH, et al. DNA methylation-based measures of biological age: meta-analysis predicting time to death. Aging. 2016;8(9):1844–1865.
3. Levine ME, et al. An epigenetic biomarker of aging for lifespan and healthspan. Aging. 2018;10(4):573–591.
4. Lu AT, et al. DNA methylation GrimAge strongly predicts lifespan and healthspan. Aging. 2019;11(2):303–327.
5. Belsky DW, et al. DunedinPACE, a DNA methylation biomarker of the pace of aging. eLife. 2022;11:e73420.
6. Quach A, et al. Epigenetic clock analysis of diet, exercise, education, and lifestyle factors. Aging. 2017;9(2):419–446.
7. Wolf EJ, et al. PTSD and the pace of epigenetic aging. JAMA Psychiatry. 2021;78(9):1021–1030.
8. Carroll JE, et al. Epigenetic aging and immune senescence in women with insomnia symptoms. Biological Psychiatry. 2022;92(5):460–468.
9. Fahy GM, et al. Reversal of epigenetic aging and immunosenescent trends in humans. Aging Cell. 2019;18(6):e13028.
10. Fitzgerald KN, et al. Potential reversal of epigenetic age using a diet and lifestyle intervention. Aging. 2021;13(7):9419–9432.
11. Waziry R, et al. Effect of long-term caloric restriction on DNA methylation measures of biological aging in healthy adults from the CALERIE trial. Nature Aging. 2023;3:248–257.