Your Cells Have a Different Age Than Your Passport. You Can Now Find Out Which One Is True.
What epigenetic clocks are, why a 5-year increase in biological age acceleration raises all-cause mortality risk by 50%, and how the number can be changed.
You’ve had a birthday every year for your entire life, so you know your chronological age to the day.
However, this number, used to calculate your cardiovascular risk on every clinical decision support tool your doctor uses, is the least accurate measure of how your body is actually aging.
Two people born in the same year can walk into a clinic today with biological ages a decade apart. One set of cells is running on a 50-year-old timeline. The other is running on a 40-year-old timeline. Neither the patients nor their doctors can tell the difference from the outside. The cholesterol panels, the blood pressure readings, the BMI… None of these measures distinguish between the two trajectories until something goes wrong.
Until recently, we couldn’t tell the difference either. Now we can.
What DNA Methylation Actually Is
Your DNA is fixed. Every cell in your body carries the same genetic sequence you were born with, and it doesn’t change. But layered on top of that fixed sequence is a second layer of information — a pattern of chemical marks attached to your DNA that control which genes are turned on and which are silenced, without altering the sequence itself. This is epigenetics.
The most studied of these marks is DNA methylation: the addition of a methyl group to specific sites (called CpG sites) along the genome. Methylation at a given site typically silences the gene below it. The pattern of methylation across thousands of these sites changes systematically with age, with lifestyle, with disease, and with environmental exposure.
In 2013, biostatistician Steve Horvath made a discovery that changed the field: the pattern of DNA methylation across 353 specific CpG sites predicted a person’s chronological age with remarkable accuracy. A blood sample from a 55-year-old looked different from a sample from a 35-year-old, even before any clinical disease appeared. The methylation pattern was, in effect, a biological timestamp.
He called it an epigenetic clock. The timestamp it kept was not identical to the calendar.
The Clocks That Followed — and Why the Generation Matters
Horvath’s first clock was a proof of concept: yes, methylation tracks age. But it predicted chronological age, not health. It told you how old your cells looked. It didn’t tell you how fast they were aging or how likely you were to get sick.
The second and third generation clocks fixed this by training on outcomes that actually matter — mortality, disease incidence, functional decline — rather than simply the passage of time. The result is a set of biological age measures with meaningfully different predictive power.¹
PhenoAge (Levine, 2018) was trained on a composite of nine clinical biomarkers — albumin, creatinine, glucose, CRP, lymphocyte percentage, red cell volume, red blood cell distribution, alkaline phosphatase, and white blood cell count — that together predict biological fitness. PhenoAge predicts all-cause mortality better than the original Horvath clock and shows meaningful associations with lifestyle factors.
GrimAge (Lu, 2019) was the leap. Rather than using clinical biomarkers, GrimAge used DNA methylation-based surrogates for plasma proteins — including GDF-15, adrenomedullin, and the cystatin C signaling associated with kidney aging — combined with a DNA methylation surrogate for lifetime smoking pack-years. Trained specifically to predict time-to-death, GrimAge is the most consistently validated mortality predictor in the epigenetic clock literature across populations, tissues, and follow-up periods.¹
The mortality data on GrimAge is precise and striking.
A 2025 retrospective cohort study of 1,942 NHANES participants found that GrimAge and GrimAge2 age acceleration were significantly and positively associated with all-cause, cancer-specific, and cardiac mortality — the only clocks among eleven tested to show these associations across all three outcome categories.²
In studies that have calculated hazard ratios by year of biological age acceleration, GrimAge EAA per 5-year increase predicts roughly a 50% increase in all-cause mortality and a 55% increase in cardiovascular mortality, after adjustment for conventional risk factors.²
Independent of blood pressure, cholesterol, BMI, and smoking history.
DunedinPACE (Belsky, 2022) is the most conceptually distinct of the major clocks, and arguably the most useful for the person trying to change something.³ Where GrimAge estimates your cumulative biological age, DunedinPACE estimates your pace of aging: how many biological years you are accumulating per calendar year right now. The scale is intuitive:
A DunedinPACE of 1.0 means you are aging at the average rate for your chronological age.
A pace of 0.8 means you are aging more slowly — roughly 80% as fast as average.
A pace of 1.2 means you are aging 20% faster than average.
DunedinPACE responds more rapidly to lifestyle changes than GrimAge because it measures current pace rather than cumulative history. You can’t undo 30 years of accelerated aging in eight weeks. But you can slow the current pace, and DunedinPACE captures that change faster.
The Finding That Changes What You Think Is Possible
Here is the result that most people in this field find most significant:
Biological age as measured by epigenetic clocks is modifiable, and it can move in both directions.
A randomized controlled trial led by Kara Fitzgerald published in Aging in 2021 enrolled 43 healthy men aged 50-72 and randomized them to either an 8-week diet and lifestyle intervention (high-phytonutrient and methyl-donor-rich diet, exercise guidance, sleep guidance, breathing-based stress reduction, and targeted supplementation) or a no-intervention control.⁴ Genome-wide DNA methylation analysis was conducted before and after the 8-week period.
The treatment group showed a mean biological age decrease of 3.23 years compared to controls on the Horvath DNAmAge clock. Within the treatment group, participants scored a mean of 2.04 years younger at the end of the program than at the start.⁴
This was the first randomized controlled trial to demonstrate a potential reversal of biological age. Eight weeks. An average of three years younger on the epigenetic clock. In healthy, middle-aged men with no specific disease process to reverse.
This result is preliminary, but the direction of the effect, and the magnitude, are consistent with what the mechanistic evidence would predict: epigenetic marks respond to the inputs you provide.
Change the inputs significantly enough, consistently enough, and the marks shift.
Why This Is Not on Your Annual Blood Report
The gap between what this technology can measure and what your annual physical captures is one of the most significant mismatches in contemporary preventive medicine.
Biological age testing via epigenetic clocks is not covered by standard insurance panels. It’s not yet integrated into primary care electronic health records. Most general practitioners have not ordered one, and most clinical guidelines have not yet incorporated it.
This is partly a matter of clinical translation timing — the evidence base is robust but relatively young, the standardization between commercial platforms is still developing, and the regulatory and reimbursement infrastructure hasn’t caught up.
A 2026 review published in eBioMedicine specifically noted that “important questions remain about the reliability of these measures in commercial wellness settings and how they should be used in clinical practice,” while simultaneously documenting that the evidence for their predictive validity continues to strengthen rapidly.⁵
Below, you have a complete guide to what the tests actually measure, which commercial formats have the most robust evidence, what to do with the result once you have it, the specific lifestyle factors that move the number fastest and by how much, the women-specific considerations, and what the Fitzgerald intervention program actually containe, in enough detail to replicate the inputs that produced a 3.23-year improvement.
