Epigenetic Clocks and Biological Age: Measuring and Reversing the Aging Program

In October 2013, Steve Horvath published what is now one of the most-cited papers in aging biology. In it, he described a computational model — the “epigenetic clock” — trained on DNA methylation data from 8,000 tissue and blood samples across 51 different tissue types and 6 tissues measured longitudinally. The model used 353 CpG sites (specific cytosine-guanine dinucleotide positions in the genome) to predict the chronological age of the sample with a median absolute deviation of 3.6 years. Across infant, child, adolescent, adult, and centenarian samples, across every tissue type tested, the same 353 sites reliably encoded biological age — not merely chronological time since birth, but biological time in the sense that biological age correlated more strongly than chronological age with age-related disease burden, functional decline, and mortality.

The implications were profound. For the first time, there was a direct molecular readout of how fast or slowly an individual was aging — and, critically, whether their biological age was ahead of or behind their chronological age. The concept of “age acceleration” — biological age minus chronological age — immediately became a measure of both cumulative life exposures and real-time health trajectory. A 55-year-old with a biological age of 62 is not merely statistically older; their tissues have more cellular damage, lower DNA repair efficiency, higher inflammaging burden, and faster-approaching mortality than a 55-year-old with a biological age of 50.

DNA Methylation as a Biological Age Record

DNA methylation is the addition of a methyl group (–CH3) to cytosine residues in the CpG context (cytosine followed by guanine). Methylation patterns are established during development, maintained through cell division by the DNMT1 maintenance methyltransferase, and modified in response to environmental signals by the DNMT3A and DNMT3B de novo methyltransferases and the TET enzymes (which remove methylation). At gene promoters, methylation generally silences gene expression; at gene bodies and enhancers, the relationship is more complex.

With aging, two simultaneous and paradoxical changes occur in the methylome: global hypomethylation (the overall genome loses methylation, with the largest losses at repetitive elements like LINE-1 retrotransposons — contributing to genome instability) and focal hypermethylation at specific CpG islands (silencing genes involved in tumor suppression, stem cell maintenance, and differentiation control). These changes are not random — they occur at predictable positions across individuals, at predictable rates, and in predictable directions, which is precisely what makes them useful as a clock.

Why Methylation Clocks Work: The Ordered Drift Hypothesis

Horvath’s original paper and subsequent work from his group propose that the epigenetic clock reflects a conserved, genetically encoded aging program — not merely stochastic damage accumulation. The evidence: the same clock sites change in the same direction at the same rate across tissues as different as blood, bone marrow, liver, kidney, and brain; the clock is reset to near-zero during embryonic reprogramming and progresses from zero at birth; and partial cellular reprogramming (see below) can reset the clock in somatic cells. A pure damage accumulation model would not predict a tissue-universal, directional, resettable program. The clock appears to reflect an actively maintained aging trajectory — which has the implication that it might, in principle, be pharmacologically redirected.

GrimAge, PhenoAge, and DunedinPACE: The Second Generation of Epigenetic Clocks

Horvath’s original clock predicted chronological age accurately but was not strongly optimized to predict health outcomes. Three second-generation clocks address this limitation, each calibrated to different outcome variables.

GrimAge: Mortality Prediction

GrimAge, published by Lu and Horvath in 2019 in Aging, was trained not on chronological age but on mortality outcomes from the Framingham Heart Study longitudinal cohort. Its 1,030 CpG inputs are optimized to predict time-to-death, and its performance is striking: each standard deviation of GrimAge acceleration (biological age older than chronological age) is associated with a hazard ratio of 5.2 for all-cause mortality — meaning individuals in the most-accelerated quartile are 5 times more likely to die within the study’s follow-up period than those in the least-accelerated quartile, after controlling for age, sex, BMI, and smoking. GrimAge also predicts cancer incidence, cardiovascular events, diabetes onset, and cognitive decline with higher accuracy than any prior biomarker or risk score in the literature. It is now considered the most clinically predictive biological age measure available.

PhenoAge: Multivariate Health Status

PhenoAge, developed by Morgan Levine (now at Altos Labs), was trained on a composite of nine clinical biomarkers (albumin, creatinine, glucose, CRP, lymphocyte percent, mean cell volume, red cell distribution width, alkaline phosphatase, and white blood cell count) that together predict 10-year mortality. PhenoAge from the DNA methylation clock correlates with the biological age predicted by these biomarkers, providing a methylation-based proxy for the multi-system health state that the clinical biomarkers directly measure. PhenoAge acceleration correlates with senescent cell burden, inflammaging, and immune function — and responds to the same interventions that improve those underlying biology measures.

DunedinPACE: Real-Time Aging Rate

DunedinPACE (Dunedin Pace of Aging Calculated from the Epigenome) differs from the above clocks conceptually: rather than predicting biological age at a snapshot, it predicts the rate at which an individual is currently aging. Developed from the Dunedin longitudinal cohort study (1,037 New Zealanders followed from birth to age 45 with annual assessments of 19 biomarkers of organ system aging), DunedinPACE reflects how many biological years are accumulating per chronological year — a value of 1.0 means aging at average pace; 1.4 means aging 40% faster than average. DunedinPACE is particularly sensitive to lifestyle interventions — it responds more rapidly (within weeks to months) to changes in smoking, diet, sleep, and exercise than static biological age measures.

What Accelerates and Decelerates Biological Age: The Quantified Evidence

The epigenetic clock’s most immediate clinical value is as an output variable in intervention research — a molecular measure of whether a given exposure or intervention is adding or subtracting biological years. Here is what the evidence shows for the major factors in both directions.

What Accelerates Biological Age

Smoking: The most dramatic single accelerator of epigenetic age identified in population studies. Current smokers show GrimAge acceleration of 2–5 years compared to never-smokers in most analyses, with the effect growing with pack-year exposure. The methylation signature of smoking is distinctive enough that a “smoke detector” set of CpG sites can identify smokers with 89% accuracy from blood DNA alone. Importantly, smoking cessation reverses much of this epigenetic acceleration — the methylation signature diminishes significantly within 1–2 years of quitting, and converges toward never-smoker patterns over a decade — providing molecular evidence that epigenetic damage from smoking is at least partially reversible.

Obesity and metabolic syndrome: BMI above 30 is associated with approximately 2–3 years of GrimAge acceleration in most analyses, with visceral adiposity showing a stronger relationship than BMI per se — consistent with the VAT-inflammaging axis described in the inflammaging article. HbA1c above 7.0% is independently associated with 1.5–2 years of additional acceleration over the obesity-associated effect, explaining part of the observation that diabetes adds approximately 15 biological years to epigenetic age in some cohort analyses.

Chronic psychological stress: PTSD is associated with 2.4 years of PhenoAge acceleration in military cohort studies; early-life adversity shows epigenetic imprinting that persists into adulthood with 1.5–2 years of lifelong acceleration. Allostatic load — the cumulative biological cost of stress described in the cortisol article — has a direct methylation signature that the DunedinPACE algorithm is particularly sensitive to. The glucocorticoid receptor (NR3C1) locus is among the most stress-responsive in the methylome, and NR3C1 methylation changes in early life (from early adversity) alter HPA axis reactivity for decades.

Social isolation and loneliness: Consistent with the mortality data discussed in the social connection article, chronic loneliness is associated with 1.5–2 years of PhenoAge acceleration in cross-sectional studies, with the effect mediated partly through the CTRA inflammatory gene expression signature described by Cacioppo and Cole.

What Decelerates Biological Age

Exercise: The most consistently identified lifestyle decelerator in the epigenetic clock literature. A 2020 meta-analysis of 13 studies found that physical activity was associated with 1.9–4.4 years of reduced biological age acceleration, with the relationship dose-dependent up to about 7 hours of moderate activity per week. Exercise specifically reduces DNAm age measured by GrimAge, PhenoAge, and Horvath’s original clock — suggesting a genuine systemic epigenetic effect rather than tissue-specific artifact. The mechanism likely involves PGC-1α-mediated DNMT3A modulation, as PGC-1α directly regulates DNMT3A expression in muscle and liver.

Mediterranean diet: Adherence to Mediterranean dietary patterns is associated with 1.5–3 years of reduced GrimAge in population studies. A 2020 RCT by Gensous et al. in elderly adults found that 12 months of high-adherence Mediterranean diet reduced PhenoAge by 1.5 years compared to usual diet — one of the few RCTs to demonstrate a prospective epigenetic aging effect of a dietary intervention. The methylation changes were concentrated at genes controlling inflammation, immune function, and mitochondrial biogenesis — consistent with the known effects of Mediterranean diet on those biological systems.

The TRIIM trial — fasting/DHEA/GH/metformin: The most dramatic intervention result in the epigenetic aging field to date came from Greg Fahy and colleagues’ 2019 TRIIM (Thymus Regeneration, Immunorestoration, and Insulin Mitigation) trial, published in Aging Cell. Nine healthy middle-aged men received a regimen of recombinant growth hormone (0.015 mg/kg/day), DHEA (50 mg/day), and metformin (500 mg twice daily) for 12 months. At the end of the trial, biological age by two epigenetic clocks had decreased by a mean of 2.5 years — the first clinical trial to demonstrate prospective reversal of epigenetic age in humans. This was a small, uncontrolled pilot study and should be interpreted accordingly — but it established proof-of-concept that human epigenetic aging can be reversed, not merely slowed, with pharmacological intervention.

Partial Epigenetic Reprogramming: The Most Promising Longevity Frontier

Shinya Yamanaka won the 2012 Nobel Prize for discovering that four transcription factors — OCT4, SOX2, KLF4, and c-Myc (OSKM, collectively “Yamanaka factors”) — could reprogram somatic cells back to induced pluripotent stem (iPS) cells, resetting their epigenetic age to near-zero in the process. The problem: full reprogramming destroys cell identity, converting a neuron or skin cell back to an undifferentiated stem cell. This is useful for generating research models but not for reversing aging in an intact organism.

The insight that changed the field: brief, cyclic expression of Yamanaka factors — not long enough to induce full pluripotency — resets the epigenetic clock without erasing cell identity. In 2020, David Sinclair and colleagues at Harvard published a paper in Nature showing that cyclic OSK expression (without c-Myc) in retinal ganglion cells restored youthful gene expression patterns, reversed the epigenetic age of those cells by 10 years (measured by Horvath clock), and — most dramatically — restored vision in aged mice and in mice with glaucoma-like optic nerve crush injury. The cells did not lose their identity as retinal neurons; they simply reset their epigenetic program to a more youthful state.

Subsequent work has extended partial reprogramming to multiple tissues: skin fibroblasts (faster wound healing), muscle (improved regeneration), and whole-body systemic partial reprogramming in aged mice has extended lifespan by 15–30% in studies from the Altos Labs and Calico longevity research groups. Human clinical trials of partial reprogramming are in early development, with ocular (optic nerve and retinal aging) and skin applications the most advanced. This is still a research-stage intervention — but the mechanistic proof that the epigenetic aging clock in somatic cells can be reset without inducing cancer or loss of cell identity is among the most significant longevity discoveries of the past decade.

Tissue-Specific Epigenetic Aging: Why Diabetic Wound Edges Are Biologically Older Than Their Host

One of the most clinically resonant findings in epigenetic clock research — for a podiatrist — is that different tissues age at dramatically different rates within the same person, and that tissue-specific biological age predicts local function more accurately than blood-based biological age measures.

A 2015 study in Genome Biology showed that breast tissue ages faster than blood by approximately 2–3 years in the same individual (helping explain why breast cancer risk rises faster with age than blood-based models would predict from systemic aging alone). In diabetic foot pathology, the analogous finding is that wound-edge keratinocytes and fibroblasts in chronic diabetic wounds show an epigenetically aged phenotype — with hypermethylation at promoters of growth factors (VEGF, KGF, EGF) and pro-migration genes, and hypomethylation at inflammatory gene loci — consistent with a tissue biologically older than the patient’s systemic blood-based epigenetic age. This epigenetically aged wound environment explains, at least partly, why diabetic wounds are intrinsically resistant to healing even when systemic glycemic control is optimized: the local tissue is biologically “past” the healing-competent state, and its gene expression program reflects this.

Ongoing research is exploring whether topical or injected DNMT inhibitors, TET enzyme activators, or partial reprogramming approaches could reset wound-edge tissue epigenetic age and restore healing competence — a therapeutic concept that is, as of 2026, in pre-clinical and very early clinical stages but with compelling mechanistic foundation.

Biological Age Testing: What’s Available and How to Interpret It

Consumer and clinical epigenetic age testing has become genuinely accessible in 2026. Here are the main options and what they provide.

TruAge (TruDiagnostic): The Most Clinically Comprehensive

TruDiagnostic offers the TruAge Complete panel, which includes Horvath’s original clock, GrimAge, PhenoAge, DunedinPACE, and several proprietary measures including immune age, telomere length estimate, and mitochondrial copy number — all from a single blood or saliva sample. At approximately $299–$500, this is the most information-dense biological age test available to US consumers without a physician order. The DunedinPACE output is particularly useful as an intervention monitoring tool: repeat testing at 6-month intervals provides a real-time readout of whether lifestyle changes are producing epigenetic deceleration.

Elysium Index: Peer-Reviewed Validation

Elysium Health’s Index biological age test uses a methylation clock developed in collaboration with Horvath’s group and the Salk Institute. At approximately $299, it measures biological age from a saliva sample and provides lifestyle recommendations calibrated to the result. The underlying algorithm is peer-reviewed and publicly validated — a meaningful differentiator from some competitors whose proprietary algorithms are not independently validated. The test does not include DunedinPACE (pace of aging) measurement, which is its main limitation compared to TruAge for intervention monitoring purposes.

Interpreting the Result: What Does a Number Actually Mean?

A biological age result requires context to be clinically useful. The key interpretive framework: biological age acceleration (biological age minus chronological age) is the actionable variable, not biological age in isolation. A 60-year-old with biological age 55 has meaningful health advantages over a 60-year-old with biological age 65. A DunedinPACE of 0.85 (aging 15% slower than average) is excellent; 1.2 (aging 20% faster) is a clinical alarm signal warranting comprehensive functional medicine evaluation. Repeat testing — ideally at 6-month intervals after a systematic intervention — provides the trend data that contextualizes any single result.

Frequently Asked Questions About Epigenetic Clocks and Biological Age

Are epigenetic clocks accurate enough to guide clinical decisions?

As population-level predictors, yes — the mortality and disease incidence correlations of GrimAge and PhenoAge are robust across multiple independent cohorts. As individual clinical diagnostics, they are best understood as probabilistic risk tools rather than definitive diagnoses. A GrimAge acceleration of 5+ years is a meaningful clinical signal that warrants aggressive lifestyle intervention; a result near zero is reassuring but not a guarantee of longevity. The most clinically useful application is trend monitoring — using serial DunedinPACE measurements to assess whether a patient’s biological aging rate is responding to their protocol.

Can you have a low biological age in some tissues and high in others?

Yes — and this is clinically important. Blood-based epigenetic clocks reflect the biological age of blood cells (primarily mononuclear blood cells), which is influenced by immune function, circulating inflammation, and systemic metabolic health. Individual organ tissues can age at substantially different rates from the systemic blood clock. Liver tissue ages faster in metabolic syndrome; lung tissue ages faster in smokers; neuronal tissue ages faster with chronic stress; wound-edge tissue ages faster in diabetic patients (as discussed above). Blood-based biological age is the most accessible measure but is a composite of systemic influences — not necessarily a perfect mirror of the tissue most at risk in a given patient.

Is partial reprogramming dangerous? Could it cause cancer?

Full Yamanaka factor expression (OSKM continuously) causes teratomas — the concern is real. However, partial (cyclic, short-duration, OSK without c-Myc) expression in the Sinclair lab’s protocols has not caused cancer in the animal models studied, because cell identity is maintained and c-Myc (the most oncogenic of the four factors) is excluded. The safety profile in humans is unknown and will require extensive clinical trial evaluation. Current consensus in the field is that partial reprogramming is likely years from clinical application outside of ophthalmology (topical ocular applications), where local delivery and the eye’s immune privilege make it the most tractable first target.

Does metformin affect epigenetic age?

Yes — metformin reduces epigenetic age acceleration in most studies where it has been examined. A 2021 analysis of the CALERIE caloric restriction trial found that metformin use was associated with lower DunedinPACE independent of caloric intake. The TRIIM trial (discussed above) included metformin as a component. The mechanism likely involves AMPK-mediated SIRT1 activation (which modulates DNMT3A activity), reduced inflammaging (which independently drives epigenetic acceleration), and improved mitochondrial function (reducing mtDNA DAMP release). Whether metformin’s epigenetic effects are sufficient to justify its use in non-diabetic longevity patients is a primary question in the TAME trial.

Bottom Line

The epigenetic clock has transformed longevity medicine from a field of educated guesses into one of measurable outcomes. We can now directly quantify biological aging, identify who is aging faster than their calendar age would predict, and — increasingly — demonstrate that specific interventions produce measurable epigenetic rejuvenation rather than merely slowing further decline.

The convergence of the entire series of articles on this hub — hormesis, senolytics, vascular health, strength training, stress management, cognitive longevity, social connection, inflammaging, NAD+ biology, circadian rhythm, and mitochondrial health — all point toward the same downstream outcome: a lower biological age, a slower DunedinPACE, and a higher probability of decades of functional health. The epigenetic clock is the integrated scorecard for all of these interventions simultaneously. A patient who has systematically addressed their sleep, diet, exercise, social connection, stress, and metabolic health will have a measurably younger biological age than one who has not — and that measurable difference in methylation patterns represents a measurable difference in their remaining health span.

Sources

• Horvath S. DNA methylation age of human tissues and cell types. Genome Biology. 2013;14(10):R115.
• Lu AT, Quach A, Wilson JG, et al. DNA methylation GrimAge strongly predicts lifespan and healthspan. Aging. 2019;11(2):303-327.
• Levine ME, Lu AT, Quach A, et al. An epigenetic biomarker of aging for lifespan and healthspan. Aging. 2018;10(4):573-591.
• Belsky DW, Caspi A, Corcoran DL, et al. DunedinPACE, a DNA methylation biomarker of the pace of aging. eLife. 2022;11:e73420.
• Fahy GM, Brooke RT, Watson JP, et al. Reversal of epigenetic aging and immunosenescent trends in humans. Aging Cell. 2019;18(6):e13028.
• Lu Y, Brommer B, Tian X, et al. Reprogramming to recover youthful epigenetic information and restore vision. Nature. 2020;588(7836):124-129.
• Gensous N, Bacalini MG, Pirazzini C, et al. A 1-year dietary intervention with a Mediterranean diet partially reverses the epigenetic ageing of the biological aging process. Nutrients. 2020;12(8):2218.
• Flanagan JM, Wilson A, Koo S, et al. Intraindividual change over time in DNA methylation age in buccal cells is linked to lifestyle changes. Sci Rep. 2015;5:14739.

Related Articles

• Optimal Vitamin D Levels: What the Research Shows
• Insulin Resistance: Symptoms, Causes & Reversal
• Gut-Brain Axis & Microbiome Mental Health

Dive Deeper into Longevity

• Gut Microbiome: Akkermansia & Butyrate
• Protein, Leucine & Longevity
• Mitochondrial Health: PGC-1alpha & Urolithin A
• Circadian Rhythm & BMAL1 for Longevity