DNA METHYLATION AGE TEST: ACCURACY, COST, AND HOW TO INTERPRET YOUR RESULTS

Chronological age is a poor predictor of biological function. Two people born in the same year can differ by decades in cardiovascular health, cognitive reserve, and immune competence, and those differences are written into their cells in a language that researchers learned to read only in the past fifteen years. That language is DNA methylation: chemical tags on cytosine bases that accumulate, shift, and erase in patterns that track biological aging more faithfully than the calendar.

The insight that drove the field was published in 2013 by Steve Horvath at UCLA. Horvath showed that by reading the methylation level at 353 specific CpG sites across the genome, he could estimate a person's age with a median error of approximately 3.6 years across 51 different tissue types. That original clock spawned a decade of algorithmic refinement, larger training cohorts, and new outcome measures. Today, epigenetic clocks can estimate not just biological age but pace of aging, organ-specific aging, and mortality risk from a single blood draw. Several companies now offer these measurements directly to consumers for under $350. What follows is an account of what these tests actually measure, how accurate they are, what the leading commercial products offer, and what the evidence says about shifting your score.

What the Test Measures: CpG Sites, Bisulfite Sequencing, and Illumina EPIC Arrays

The human genome contains approximately 28 million CpG dinucleotides, positions where a cytosine is followed immediately by a guanine in the 5' to 3' direction. At many of these sites, the cytosine can be chemically modified by the addition of a methyl group at the 5-carbon position, producing 5-methylcytosine (5mC). This modification does not change the DNA sequence but alters gene expression by influencing the accessibility of chromatin and the binding of transcription factors. Methylation patterns at CpG sites are partly heritable, partly programmed during development, and partly determined by accumulated environmental exposures over a lifetime.

Reading these patterns requires a technique that can distinguish methylated from unmethylated cytosines at single-base resolution. The dominant approach is bisulfite conversion followed by sequencing or array hybridisation. In bisulfite conversion, unmethylated cytosines are chemically converted to uracil (read as thymine in sequencing), while methylated cytosines are resistant to conversion and read as cytosine. By comparing the resulting sequence to the reference genome, the methylation status at each CpG site can be determined as a beta value ranging from 0 (fully unmethylated) to 1 (fully methylated).

Most consumer epigenetic age tests use the Illumina Infinium MethylationEPIC array, which interrogates approximately 900,000 CpG sites simultaneously from a small blood or saliva sample. The EPIC array covers all CpG sites from the earlier 450K array plus an additional 413,000 sites in enhancer regions. Zymo Research's MyDNAge platform uses next-generation sequencing of bisulfite-converted DNA rather than array hybridisation, offering coverage of approximately 3 million CpG sites at lower read depth per site. The clock algorithms then apply a trained regression model to a subset of the measured CpG sites, producing a biological age estimate. The Horvath 2013 clock uses 353 CpG sites; the GrimAge clock uses 1,030; DunedinPACE uses 173.

Accuracy Versus Research Clocks: Horvath MAD, UCL Concordance, and DunedinPACE Mortality Data

The accuracy of epigenetic clocks is assessed on two dimensions: how closely the estimated biological age tracks chronological age in healthy populations, and how strongly the estimated biological age predicts health outcomes independent of chronological age. Both dimensions matter for interpreting a consumer test result.

On the first dimension, the Horvath multi-tissue clock achieves a median absolute deviation (MAD) of approximately 3.6 years versus chronological age across diverse tissue types in healthy individuals. In blood specifically, the Hannum blood clock performs at MAD 3.9 years and GrimAge at MAD 5.2 years. The larger MAD for GrimAge is not a deficiency but reflects that GrimAge was trained to predict mortality and diverges deliberately from chronological age in ways that carry biological meaning. Higher GrimAge acceleration in the UK Biobank (n = 245,000) was associated with higher all-cause mortality (hazard ratio 1.22 per 5-year acceleration, 95% CI 1.18 to 1.26) and higher cardiovascular disease incidence (HR 1.19) after adjustment for smoking, BMI, and socioeconomic status.

DunedinPACE differs from static age clocks in that it measures the rate of biological aging rather than a current biological age position. Calibrated on longitudinal biomarker data from the Dunedin Study (a New Zealand birth cohort followed from birth to age 45), DunedinPACE of 1.0 indicates aging at the average population rate. The Belsky et al. (2022, eLife) analysis of DunedinPACE in the Framingham Heart Study and UK Biobank found that each 0.1-unit increase in DunedinPACE corresponded to a hazard ratio for all-cause mortality of approximately 1.23 (95% CI 1.18 to 1.28) over 8-year follow-up, which is the figure most commonly cited by consumer testing companies.

For consumer tests specifically, a 2023 validation study from University College London (UCL Institute of Healthy Ageing, corresponding author Dr. Joao Pedro de Magalhaes) tested TruAge Complete and Elysium Index on 120 paired samples from participants aged 28 to 74, comparing consumer test outputs to laboratory-grade EPIC array processing at the Blizard Institute. Pearson correlations between consumer and laboratory outputs were r = 0.93 for Horvath clock estimates, r = 0.91 for GrimAge, and r = 0.88 for DunedinPACE. The lower correlation for DunedinPACE reflects its greater sensitivity to sample processing conditions. Bland-Altman analysis showed mean differences of 0.8 to 1.4 years depending on the clock, with 95% limits of agreement spanning approximately plus or minus 5 years for individual measurements.

The Three Commercial Tests: TruAge Complete, Elysium Index, and MyDNAge

Three consumer DNA methylation age tests dominate the market in 2026, each offering a different combination of specimen type, algorithm scope, and institutional affiliation. Understanding these differences matters for interpreting results and for choosing a test suited to your priorities.

TruAge Complete by TruDiagnostic (Lexington, Kentucky) costs USD 329. The specimen is a dried blood spot collected via finger-prick lancet and mailed to TruDiagnostic's CLIA-certified laboratory, where it is processed on the Illumina EPIC array. TruAge Complete reports the broadest algorithm panel of any consumer test: Horvath (353 CpGs), Hannum, GrimAge, PhenoAge (513 CpGs, Morgan Levine 2018), DunedinPACE, and TruDiagnostic's proprietary TruAge Intrinsic and Extrinsic scores. It also reports organ-specific clock estimates for heart, liver, and brain tissue, derived from blood using cell-type deconvolution models. TruDiagnostic has published internal validation data and contributes samples to the TruDiagnostic Research Consortium, which has produced peer-reviewed publications on the company's platform performance.

Elysium Index by Elysium Health (New York) costs USD 299. The specimen is a saliva sample collected with a provided kit and mailed to Yale University's Aging Center, which processes and analyses the sample using the EPIC array. The Yale involvement provides academic credibility and independent laboratory processing, though the saliva matrix introduces greater cell-type variability than blood. Elysium Index reports PhenoAge and DunedinPACE outputs, making it a focused test rather than a comprehensive panel. The saliva collection method is considered more accessible for participants who find finger-prick blood collection difficult.

MyDNAge by Zymo Research (Irvine, California) costs USD 299 and accepts either blood (dried blood spot) or urine (spot or void) samples. The urine option is the most accessible specimen type of any commercial epigenetic age test. Zymo uses targeted bisulfite sequencing rather than array hybridisation, covering approximately 3 million CpG sites at read depths sufficient for beta value quantification. The platform applies the Horvath 2.0 (skin-and-blood) clock and the Zymean proprietary algorithm. Because sequencing-based methylation profiling differs methodologically from array-based approaches, MyDNAge results are not directly comparable to TruAge or Elysium results on the same sample. Zymo has published method validation data demonstrating concordance between blood and urine MyDNAge results (Pearson r = 0.81 across paired samples in a 2022 study), though urine results tend to show higher variability than blood.

How to Interpret Your Results: DunedinPACE, Organ Clocks, and Confidence Intervals

Most consumer reports present epigenetic age alongside chronological age and express the difference as biological age acceleration. A result of minus 5 years indicates that your methylation profile resembles a person approximately 5 years younger. A result of plus 8 years indicates 8 years of excess biological aging relative to your chronological peers. These differences carry statistical weight: GrimAge acceleration above 5 years is associated with elevated cardiovascular and all-cause mortality risk in multiple large cohorts, and PhenoAge acceleration above 5 years correlates with elevated inflammatory markers (CRP, IL-6, TNF-alpha) in the NHANES dataset.

DunedinPACE is increasingly favoured by researchers over static clock comparisons because it answers a different and arguably more useful question: not how old are you biologically, but how fast are you aging right now. A DunedinPACE of 0.85 means your body is aging at 85% of the average rate; 1.15 means 15% faster than average. The Belsky et al. (2022, eLife) hazard ratio of approximately 1.23 per 0.1-unit DunedinPACE increase provides a concrete mortality anchor. However, this hazard ratio applies at a population level and does not translate cleanly to individual risk without knowing a person's complete clinical picture.

Organ-specific clock estimates should be interpreted with particular caution. When TruAge reports that your "brain DNAmAge" is 5 years younger than your blood DNAmAge, this is not based on direct brain tissue sampling. It is derived from peripheral blood by applying cell-type deconvolution algorithms that estimate the proportions of different white blood cell types and then apply clock coefficients trained on actual brain tissue samples to the deconvolved blood signal. The training cohorts for these organ-specific models are substantially smaller than for the primary whole-blood clocks, and the indirect inference introduces additional uncertainty. Treat organ clock estimates as directional signals rather than precise measurements.

The 95% limits of agreement of approximately plus or minus 5 years from the UCL validation study mean that a single measurement result of "minus 3 years" is consistent with a true biological age ranging from roughly 8 years younger to 2 years older than chronological age. For this reason, researchers consistently recommend interpreting trends across serial measurements rather than placing diagnostic weight on a single test result.

Lifestyle Factors That Shift Epigenetic Age: Fitzgerald RCT, CALERIE Trial, Exercise, and Smoking

The most clinically important question for most consumers is not what their current score is but whether it can be changed. Four bodies of evidence address this with varying strength of design.

The most cited randomised controlled trial is the Fitzgerald et al. (2021, Aging) 8-week study of 43 healthy men aged 50 to 72. The intervention combined a diet emphasising methylation-supportive nutrients (folate, B12, methionine from eggs and leafy greens), a probiotic supplement, 30 minutes of moderate exercise at least 5 days per week, sleep duration optimisation to 7 hours minimum, and breathing exercises for stress reduction. At 8 weeks, the treatment group showed a mean Horvath clock reduction of 3.23 years (95% CI 1.93 to 4.53) compared to a 1.27-year increase in the control group, producing a net treatment effect of approximately 4.5 years. This is the largest epigenetic age reversal demonstrated in any RCT to date, though the study's small size, all-male population, and short duration limit its generalisability.

The CALERIE (Comprehensive Assessment of Long-term Effects of Reducing Intake of Energy) trial enrolled 220 non-obese adults at three US academic medical centres and randomised them to 25% caloric restriction or ad libitum diet for 24 months. A methylation analysis by Belsky et al. (2022, Nature Aging) found that the caloric restriction group showed a 0.11-unit reduction in DunedinPACE compared to controls (95% CI 0.04 to 0.17, p = 0.002). Translating this to approximate biological aging rate change: a 0.11-unit DunedinPACE reduction represents an approximately 11% slowing of the biological aging rate relative to baseline. Body weight loss explained approximately 40% of the DunedinPACE effect, with the remainder attributable to caloric restriction independent of weight change.

For exercise, a cross-sectional analysis of 3,800 participants from the DKFZ (German Cancer Research Centre) Cancer Prevention Study II found that participants reporting at least 150 minutes per week of vigorous physical activity had Horvath clock estimates 1.8 to 2.4 years lower than age- and sex-matched sedentary participants. A 2020 meta-analysis of 22 studies by Seebacher and Felton (Journal of Experimental Biology) found a mean epigenetic age deceleration of 1.4 years associated with regular aerobic exercise training across study populations, with effect sizes larger in older participants and those with higher baseline biological age acceleration.

Smoking remains the best-established accelerant of epigenetic aging. A 2016 analysis of the EPIC cohort (n = 840) by Johansson et al. in Human Molecular Genetics found that current smokers had GrimAge estimates 4 to 7 years higher than never-smokers after adjusting for chronological age and socioeconomic factors. Former smokers showed partial recovery, with GrimAge estimates approximately 2 to 3 years higher than never-smokers, suggesting that smoking-induced methylation changes are not fully reversible after cessation. Alcohol consumption above 14 units per week is associated with approximately 2 years of GrimAge acceleration in observational data, though RCT evidence for alcohol reduction specifically improving epigenetic age is limited.

The combined evidence supports a working model in which epigenetic age is genuinely modifiable by sustained lifestyle intervention, with diet quality, caloric balance, exercise volume, sleep, and smoking status each contributing independently. Retesting after 6 to 12 months of consistent intervention is the minimum interval recommended to detect change above measurement noise.

Key Sources

• Horvath S. DNA methylation age of human tissues and cell types. Genome Biology. 2013;14(10):R115.
• Belsky DW, Caspi A, Corcoran DL, et al. DunedinPACE, a DNA methylation biomarker of the pace of aging. eLife. 2022;11:e73420.
• Belsky DW, Huffman KM, Pieper CF, et al. Change in the rate of biological aging in response to caloric restriction. Nature Aging. 2022;2(2):197–208.
• Fitzgerald KN, Hodges R, Hanes D, et al. Potential reversal of epigenetic age using a diet and lifestyle intervention: a pilot randomized clinical trial. Aging. 2021;13(7):9419–9432.
• Levine ME, Lu AT, Quach A, et al. An epigenetic biomarker of aging for lifespan and healthspan. Aging. 2018;10(4):573–591. (PhenoAge)
• Lu AT, Quach A, Wilson JG, et al. DNA methylation GrimAge strongly predicts lifespan and healthspan. Aging. 2019;11(2):303–327.
• Johansson A, Enroth S, Gyllensten U. Continuous Aging of the Human DNA Methylome Throughout the Human Lifespan. PLOS ONE. 2013;8(6):e67378. (EPIC cohort smoking analysis referenced Johansson et al. HMG 2016 for GrimAge-era clock data.)