Men aged 50 to 72 who followed a structured 8-week diet and lifestyle program shaved an average of 3.23 years off their biological age, according to a small pilot randomized trial published in the journal Aging in April 2021 by Kara Fitzgerald and colleagues.1 The control group, who made no changes, did not see the same shift.
That number is striking, but it deserves the asterisks the authors themselves attached to it. Eighteen men completed the intervention. None of them lost weight on purpose. None of them took an experimental drug. They ate a particular pattern of food, slept on a schedule, exercised most days, did breathing practice, and took a small stack of supplements. Their DNA, measured before and after, looked younger by a clock the field has been refining since 2013.2
What does “biological age” actually mean here?
The age on your driver’s license is fixed. The age your cells are behaving at is not. Researchers measure the second one with what is called an epigenetic clock, and the most influential version was developed by UCLA biostatistician Steve Horvath in 2013.2 Horvath’s clock looks at hundreds of specific spots on your DNA where small chemical tags called methyl groups attach and detach over a lifetime. Those tags do not change the genetic code itself. They change how loudly each gene is read.
Think of methyl groups as little bookmarks. As you accumulate years, the pattern of bookmarks shifts in a predictable way across thousands of people. By comparing your pattern to that average curve, the clock estimates how old your cells “think” they are. That estimate is called your DNA methylation age, or DNAmAge.
The Fitzgerald team used the original Horvath clock as their primary outcome. The men in the treatment arm dropped 3.23 years on that clock over the 8 weeks of the program, while the controls did not, and the difference between groups reached statistical significance.1 Newer clocks like PhenoAge3 and GrimAge4, which are tuned to predict things like lifespan and disease risk rather than chronological age, were not the primary endpoint.
Inside the 8-week protocol
The intervention was deliberately layered. Fitzgerald and her co-authors did not isolate one variable. They asked participants to do a lot of small things at once, then measured the combined effect.
The food piece was a plant-forward, lower-glycemic plan that emphasized leafy greens (a target of two cups per day), cruciferous vegetables, beets, pumpkin seeds, sunflower seeds, methyl-donor foods like eggs and liver, fresh herbs and garlic, and clean animal protein. Wild salmon and other fatty fish appeared three times a week, and a serving of berries plus six walnuts per day were standard. Most starches, dairy, legumes, grains, and added sugar were off the menu for the eight weeks.1
The men also followed a 12-hour overnight fast, which gave them roughly a 12-hour eating window each day. Sleep was specifically prescribed: a minimum of 7 hours a night. Exercise was 30 minutes a day, at least five days a week, at moderate intensity. A daily 10-minute breathing or relaxation practice was added on top.
Two supplements were included. The first was a probiotic called Lactobacillus plantarum 299v. The second was a phytonutrient powder containing fruit and vegetable concentrates. The protocol was built to support what biochemists call one-carbon metabolism, the cellular pathway that produces methyl groups in the first place. The whole thing reads more like a doctor’s recommendation list than a magic bullet.
Why so much focus on methyl donors? Because every methylation event in the body, including the ones that label your DNA with age-related marks, draws on a pool of one-carbon units supplied largely by folate, vitamin B12, choline, betaine, and methionine. Eggs, liver, leafy greens, beets, and seeds carry those nutrients in concentrated form. Run that pool low, and methylation patterns drift in noisy, unhelpful ways. Keep it well stocked, and the cell has the raw material to apply, remove, and maintain methyl groups where they are supposed to sit. The diet was not aiming to add tags. It was aiming to give the cell what it needs to put the right tags in the right places.
Why three years off, and not thirty?
A common misread of the headline is that the participants got younger. They did not. Their epigenetic clock reading went down. Their bodies still had whatever wear, scarring, and accumulated damage they walked in with.
What the result suggests is that some of what we call aging is more like a setting than a wound. Methylation patterns can be nudged. They are not carved into the genome. The Fitzgerald paper estimates a 1.96-year decrease in DNAmAge within the treatment group from baseline to end, plus a 3.23-year difference between treatment and control at the end of the study.1 Those are different framings of the same underlying signal, and both are modest in absolute terms when you remember the men spanned more than two decades of age between them.
This is also where pilot status matters. Eighteen finishers is a small number. The trial was not blinded in the way a drug trial would be. Participants knew they were doing the program, and the placebo effect on something as systemic as cellular methylation is genuinely unknown. The authors flagged all of this themselves and called for larger replications.
What the salmon, walnuts, and dark chocolate are doing
The image circulating with this story features fish, walnuts, and dark chocolate, and that shorthand is not wrong, but it is not the whole answer either. Each of those foods earns its place for slightly different reasons inside the protocol.
Wild salmon brings long-chain omega-3 fatty acids, mainly EPA and DHA, which are precursors to anti-inflammatory signaling molecules. Chronic low-grade inflammation correlates with faster epigenetic aging on clocks like PhenoAge and GrimAge.3,4 Reducing the inflammatory tone of the body is a plausible mechanism for the methylation pattern shifting in a younger direction, although the Fitzgerald trial did not isolate inflammation as the cause.
Walnuts were specified at six per day. They contribute alpha-linolenic acid, a plant omega-3, plus polyphenols and a meaningful dose of magnesium. They do not appear in the protocol because they are a superfood. They appear because they are a calorie-dense, easy-to-tolerate way to layer healthy fats and polyphenols into a daily routine for eight weeks.
Dark chocolate did not appear in the published protocol’s named food list, but cocoa polyphenols broadly fit the same flavonoid family as berries and the phytonutrient concentrate. The story image folds it in for visual appeal. If you eat a square or two of 80 percent dark chocolate as part of a varied diet, you are within the spirit of the intervention. If you eat a bar a day in addition to everything else, you are not.
It is not just one study
The Fitzgerald pilot is the headline, but it sits on top of a decade of methylation-clock research. Horvath’s 2013 paper showed that DNA methylation age tracked chronological age across more than 50 tissue types in over 8,000 samples, with a median error of less than four years.2 That made the clock usable as a research tool, not just a curiosity.
Levine and colleagues extended the work in 2018 with PhenoAge, a clock trained not on chronological age but on nine clinical biomarkers tied to mortality. PhenoAge was a stronger predictor of all-cause death and disease than chronological age alone in their dataset.3 The next year, Lu and colleagues published GrimAge, which was trained on plasma protein levels and smoking history, and predicted time-to-death and time-to-coronary-heart-disease better than its predecessors.4
None of those papers ran a diet trial. They built the rulers. The Fitzgerald work is one of the early attempts to use those rulers to score a lifestyle intervention. More trials, with bigger samples, longer follow-up, and women as well as men, are still needed before a clinician can say with confidence that a specific protocol delivers a specific number of “epigenetic years” reliably.
How long does the effect last?
This is the question the trial cannot answer. Follow-up ended at 8 weeks. We do not know whether the men who returned to their previous diets bounced back to their starting methylation pattern within a month, held the gains for a year, or slowly drifted back over time.
The biology hints that habits matter more than heroics. Methylation responds to ongoing inputs: what you eat this week, how much you slept last night, whether you moved your body today. The interventions stacked in the Fitzgerald protocol overlap heavily with what longevity researchers have recommended for decades. Sleep more than 7 hours. Eat plants. Move every day. Manage stress. Eat fewer processed carbohydrates. The novel part is the measurement, not the advice.
That said, durability is a real open question. Studies on related markers, like inflammatory cytokines and fasting insulin, generally show that gains start to fade within weeks of stopping the behavior that produced them. There is no biological reason to expect epigenetic age reduction to behave differently. If anything, methylation is a sensitive readout precisely because it tracks current cellular conditions. The honest answer is that nobody yet has a published 12-month follow-up on these participants, and consumers should not assume an 8-week sprint will hold for a year of regression to old habits.
Common questions about reversing epigenetic age
Does this work for women?
The pilot only enrolled men aged 50 to 72. The same broad mechanisms (methylation, inflammation, sleep, methyl-donor nutrition) apply to women, but Fitzgerald and colleagues have since started extending the work to women, and confirmation requires those data. Acting on the lifestyle pattern is reasonable; expecting the exact same 3.23-year number is not.
Can I just take a supplement instead?
The trial used probiotic and phytonutrient supplements as part of a larger package, not as standalone treatments. There is no published evidence that any single pill produces a comparable shift on the Horvath clock.1,2
How is biological age measured outside of a research lab?
Several consumer companies now sell saliva-based methylation age tests. The accuracy depends on which clock they use, the quality of the lab, and how they handle individual variation. Treat any single result as a noisy snapshot, not a verdict.
Is the 3-year reduction permanent?
Unknown. The trial measured the men at 8 weeks and stopped. Methylation responds to current inputs, so a return to the prior diet and sleep pattern would likely move the clock again, in the other direction.
Should I try the exact protocol?
The published protocol is restrictive (no dairy, no legumes, no grains, daily breathing practice, two named supplements) and was designed under clinical supervision. Most people will get most of the value by adopting the easier pieces: more leafy greens, fatty fish a few times a week, a 12-hour overnight fast, consistent sleep, and daily moderate exercise. Talk to your own physician before adding supplements, especially if you take medications.
What this changes, and what it does not
The honest reading of the Fitzgerald pilot is that lifestyle has a measurable, biological signature, and that signature can move within weeks, not years. That alone is worth the headline. It pushes back against the fatalistic version of aging in which decline is a one-way road and only pharmaceuticals can intervene.
It does not change the underlying advice. It does not unlock a magic combination of foods. The men in the trial did everything together: ate well, slept well, moved daily, breathed, supplemented, fasted overnight. Pulling out salmon or walnuts or dark chocolate as the active ingredient is the kind of simplification that sells well on social media but misses the design. The intervention worked because the pieces stacked. If you want to borrow from it, borrow the stack, not the souvenir.
Sources
- 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 (Albany NY). 2021;13(7):9419–9432. PubMed: 33844651
- Horvath S. DNA methylation age of human tissues and cell types. Genome Biology. 2013;14(10):R115. PubMed: 24138928
- Levine ME, Lu AT, Quach A, et al. An epigenetic biomarker of aging for lifespan and healthspan. Aging (Albany NY). 2018;10(4):573–591. PubMed: 29676998
- Lu AT, Quach A, Wilson JG, et al. DNA methylation GrimAge strongly predicts lifespan and healthspan. Aging (Albany NY). 2019;11(2):303–327. PubMed: 30669119
