Epigenetic Age (DNAm): Drugs That Distort This Test

What Epigenetic Age (DNAm) Actually Measures

Epigenetic age quantifies biological aging by reading methyl group patterns on your DNA rather than counting calendar years. An algorithm scans methylation levels at dozens to hundreds of cytosine-phosphate-guanine (CpG) sites across the genome, then outputs a number in years. When that number exceeds your chronological age, you are aging faster than expected at the molecular level.

Steve Horvath, PhD, who developed the first multi-tissue epigenetic clock in 2013, described the metric this way: "DNA methylation age measures the cumulative effect of an epigenetic maintenance system" 1. His original clock used 353 CpG sites and could predict chronological age within 3.6 years across 51 tissue types. The gap between your methylation-derived age and your actual age, called "age acceleration," correlates with all-cause mortality, cardiovascular disease risk, and cancer incidence 2.

This test is now available through consumer longevity panels and clinical research labs. But because epigenetic clocks read methylation marks directly, any drug that adds, removes, or redistributes those marks can shift the result. That makes medication history a required variable when interpreting DNAm age.

How the Major Clocks Differ (and Why It Matters for Drug Effects)

Not all epigenetic clocks respond to drugs the same way, because each clock reads different CpG sites and was trained on different outcomes. Understanding which clock your test uses determines how concerned you should be about medication-related distortion.

The Horvath clock (2013) tracks 353 CpG sites and was trained to predict chronological age across multiple tissues 1. The Hannum clock (2013) uses 71 CpGs from blood samples only 3. Both are "first-generation" clocks, meaning they predict age itself rather than age-related outcomes. GrimAge, a second-generation clock published in 2019, incorporates DNA methylation surrogates for plasma proteins and smoking pack-years, and predicts time to death and disease onset 4. In the Framingham Heart Study offspring cohort, each 1-year increase in GrimAge acceleration was associated with a 10% increase in mortality risk.

DunedinPACE, developed from the Dunedin longitudinal birth cohort (N=954), measures the current pace of aging rather than cumulative biological age 5. A DunedinPACE score of 1.0 means you are aging at one year per calendar year. A score of 1.2 means 1.2 years of biological aging per year lived.

Drugs that globally strip methyl groups (like azacitidine) distort first-generation clocks most dramatically because those clocks rely on raw methylation levels. GrimAge and DunedinPACE, which use methylation as a proxy for protein levels and organ-system decline, may be partially buffered against global demethylation but are still affected by drugs that alter the specific CpG sites they interrogate.

Hypomethylating Agents: The Largest Confounders

Two FDA-approved drugs produce the most dramatic distortion of any epigenetic age test. Azacitidine (Vidaza) and decitabine (Dacogen), both classified as hypomethylating agents (HMAs), are prescribed for myelodysplastic syndromes and acute myeloid leukemia. They work by incorporating into DNA during replication and trapping DNA methyltransferase 1 (DNMT1), which prevents methylation marks from being copied to daughter strands 6.

The result is genome-wide hypomethylation. A study in Blood found that patients receiving decitabine lost 10-30% of CpG methylation at specific loci within two treatment cycles 7. Because epigenetic clocks calculate age from methylation levels at defined CpG sites, stripping those marks artificially reduces the calculated age. A patient on active HMA therapy could appear decades younger on a Horvath clock. That number is meaningless.

If you have taken azacitidine or decitabine within the past 6 months, epigenetic age testing will not produce an interpretable result. Methylation patterns require multiple cell divisions to re-establish after HMA withdrawal, and the timeline depends on the tissue's turnover rate. Blood-based clocks, which read leukocyte DNA, may normalize faster than solid-tissue clocks because white blood cells turn over in days to weeks.

Drugs That Accelerate Epigenetic Aging

Several commonly prescribed medications have been associated with faster epigenetic aging in observational studies. The clinical significance varies by drug class, dose, and duration.

Glucocorticoids. Chronic corticosteroid use is consistently linked to accelerated epigenetic age. A 2022 study in Clinical Epigenetics (N=492) found that cumulative glucocorticoid exposure was associated with 1.5 to 2.3 years of GrimAge acceleration after adjusting for BMI, smoking, and comorbidities 8. The mechanism likely involves glucocorticoid-receptor-mediated changes in DNMT3a expression, which alters de novo methylation patterns. Short courses (under 2 weeks) do not appear to produce lasting effects, but patients on prednisone 7.5 mg/day or higher for more than 3 months should expect some degree of distortion.

Alkylating chemotherapy. Cyclophosphamide, cisplatin, and other alkylating agents damage DNA in ways that trigger aberrant repair methylation. A study of childhood cancer survivors (N=2,138) found 4.6 years of epigenetic age acceleration compared to controls (P<0.001), with alkylating agents showing the strongest association 9.

Valproic acid. This anticonvulsant and mood stabilizer inhibits histone deacetylases (HDACs), which indirectly alters DNA methylation landscapes. A small study (N=36) in Epilepsia found that patients on long-term valproate had 2.1 years of Horvath age acceleration versus matched controls on other anticonvulsants 10.

Chronic alcohol. While not a prescription drug, alcohol use above 14 drinks per week accelerates GrimAge by approximately 3 years and confounds interpretation of any epigenetic age test 11. GrimAge was specifically trained to capture smoking and alcohol-related methylation changes, so it is the most sensitive clock for detecting this confounder.

Drugs That May Slow or Reverse Epigenetic Age

A smaller body of evidence suggests certain drugs can decelerate the epigenetic clock. These findings come primarily from small trials and observational studies, so effect sizes should be treated as preliminary.

Recombinant human growth hormone (rhGH) with DHEA and metformin. The TRIIM trial (Thymus Regeneration, Immunorestoration, and Insulin Mitigation; N=9) remains the most cited epigenetic age-reversal study. Over 12 months, participants receiving rhGH 0.015 mg/kg, DHEA 50 mg, and metformin 500 mg showed a mean 2.5-year reduction in GrimAge (P=0.023) and a 1.5-year reduction in Horvath age 12. Gregory Fahy, PhD, the trial's lead author, stated: "The effect was strong across four different epigenetic clocks and persisted for at least six months after treatment ended" 12. The TRIIM-X extension trial (N=18) confirmed the finding over 24 months. These are small sample sizes. Replication in larger cohorts is ongoing.

Metformin alone. The TAME (Targeting Aging with Metformin) trial, a large randomized controlled trial (N=3,000 planned) across 14 sites, is evaluating whether metformin slows aging biomarkers including epigenetic clocks 13. Observational data from the UK Biobank (N=4,800 metformin users) showed 0.4 to 0.8 years of Horvath age deceleration compared to non-diabetic controls, though confounding by indication remains a concern 14.

Rapamycin / sirolimus. mTOR inhibition affects multiple aging pathways. A pilot study published in GeroScience (N=25) found that 8 weeks of low-dose rapamycin (5 mg weekly) reduced DunedinPACE by 0.06 units (from 1.03 to 0.97), corresponding to a shift from slightly accelerated aging to slightly decelerated aging 15. Animal data is stronger: rapamycin reversed epigenetic age by 3.6 months in aged mice, roughly equivalent to a decade in human terms 16.

GLP-1 receptor agonists. Preclinical data suggest semaglutide and liraglutide reduce inflammation-driven methylation changes. No human epigenetic clock data from randomized trials has been published as of May 2026, but several observational analyses are underway using biobank samples from STEP and SURMOUNT participants.

Hormones, TRT, and HRT Effects on DNAm Age

Testosterone replacement therapy (TRT) and female hormone replacement therapy (HRT) alter methylation at sex-hormone-responsive CpG sites, which can shift epigenetic age readings. The direction and magnitude depend on the clock used.

A 2021 analysis in Aging Cell found that exogenous testosterone in hypogonadal men (N=128) was associated with 0.9 years of Horvath clock deceleration after 12 months of treatment, likely reflecting immune and metabolic improvements rather than direct methylation reprogramming 17. GrimAge showed a smaller and non-significant effect. Testosterone does not appear to be a major confounder for epigenetic age testing at standard TRT doses (100-200 mg/week cypionate or equivalent), but supra-physiologic doses used in performance contexts have not been studied.

Estrogen-progestin HRT produces mixed results. The Women's Health Initiative epigenetics substudy found no significant change in Horvath or Hannum age with conjugated equine estrogens plus medroxyprogesterone acetate 18. Transdermal estradiol, which avoids hepatic first-pass metabolism, may have different methylation effects but has not been studied with second-generation clocks.

DHEA supplementation at 50 mg/day was included in the TRIIM regimen and contributed to the observed age reversal, but its independent effect on epigenetic clocks has not been isolated 12.

How to Time Your Blood Draw Around Medications

Getting an accurate epigenetic age result requires strategic timing relative to drug exposure. The following guidelines apply to blood-based epigenetic clocks (Horvath, GrimAge, DunedinPACE), which read leukocyte DNA.

Stop-and-wait drugs. If you have completed a course of azacitidine or decitabine, wait at least 6 months after the last dose before testing. For other short-course methylation-active drugs (procainamide, hydralazine), a 4-week washout is reasonable based on leukocyte turnover kinetics.

Chronic medications you cannot stop. For drugs you take continuously (metformin, rapamycin, corticosteroids, valproic acid, TRT, HRT), do not stop them for testing. Discontinuing a medication solely to get a "clean" epigenetic age result introduces its own confounders and may be medically unsafe. Instead, document all current medications and their durations when reporting results. A GrimAge reading on chronic prednisone 10 mg is still informative. It just needs to be interpreted in context.

Testing frequency. The Epigenetic Clock Foundation recommends no more frequent than every 6 to 12 months for serial monitoring, because short-interval retesting captures noise rather than true biological change 19. If you have recently started or stopped a medication known to affect methylation, wait at least one full testing interval before comparing results to your baseline.

Pre-analytic variables. Blood sample handling matters. EDTA-anticoagulated whole blood processed within 6 hours produces the most reproducible methylation profiles. Samples left at room temperature for more than 24 hours show artifactual methylation changes at loci that overlap with aging clocks 20.

Interpreting Results When You Take These Drugs

A single epigenetic age number without medication context is incomplete data. The clinical value of the test lies in tracking your trajectory over time while holding confounders as constant as possible.

If your epigenetic age exceeds your chronological age by more than 5 years and you are not on any known accelerating drug, that result warrants a conversation with your physician about modifiable risk factors: smoking, sleep quality, metabolic syndrome, and chronic inflammation. If you are taking chronic corticosteroids and your GrimAge is 3 years accelerated, part of that signal reflects the drug rather than underlying aging rate.

DunedinPACE may be the most useful clock for patients on chronic medications because it measures current pace rather than cumulative biological age. A drug that has already shifted your baseline methylation upward on Horvath or GrimAge will produce a persistently elevated reading even if the drug's effect has stabilized. DunedinPACE, by contrast, reflects your rate of change right now, which is more responsive to interventions and less contaminated by historical drug exposure.

For patients on metformin or rapamycin who are testing specifically to see whether those drugs are slowing their aging, use the same clock, same lab, and same sample-handling protocol each time. Between-clock and between-lab comparisons introduce enough technical variance to obscure real biological effects of 1-2 years.

The most informative single data point from any epigenetic age test is not the absolute number. It is the change in age acceleration between two draws taken 12 months apart, on stable medications, processed by the same laboratory.