Daily Multivitamin Slows Epigenetic Aging Clocks in COSMOS Trial

SNIPPET: A large randomized controlled trial (COSMOS, n=958) published in Nature Medicine found that daily multivitamin-multimineral supplementation modestly slowed epigenetic aging clocks over two years — reducing PCGrimAge progression by −0.113 years/year and PCPhenoAge by −0.214 years/year versus placebo. Effects were strongest in those with accelerated biological aging at baseline. Cocoa extract showed no effect.

THE PROTOHUMAN PERSPECTIVE#

Here is something that should make you stop and think. The cheapest, most accessible supplement on the planet — a daily multivitamin — just demonstrated a statistically significant effect on the most sophisticated biomarkers of biological aging we have. Not a proprietary longevity compound. Not a $500-a-month NAD+ precursor. A multivitamin.

Published March 9, 2026 in Nature Medicine, the COSMOS trial data from Li, Sesso, Hamaya et al. represents the first large-scale RCT evidence that basic nutrient supplementation can modify DNA methylation-based aging clocks in humans^[2]. For the biohacking and longevity community, this is a moment that demands careful interpretation — not hype, but genuine reckoning with what it means when a $15 bottle of Centrum Silver moves the same biomarkers that elite longevity clinics charge thousands to track.

The implications cut both ways. Either multivitamins are quietly doing more than anyone credited them for, or epigenetic clocks are more sensitive to nutritional inputs than we assumed. Both possibilities matter.

THE SCIENCE#

What Are Epigenetic Clocks, Exactly?#

Epigenetic clocks are algorithms that estimate biological age by measuring DNA methylation patterns at specific CpG sites across the genome. They are not measuring how old you are — they're measuring how fast you're deteriorating. First-generation clocks (Horvath, Hannum) were trained to predict chronological age. Second-generation clocks like PCPhenoAge and PCGrimAge were trained against mortality and disease outcomes, making them considerably more informative as health biomarkers^[3]. DunedinPACE, developed by Belsky et al., goes further — it estimates the pace of aging rather than a static age estimate, calibrated against longitudinal data from the Dunedin birth cohort^[4].

The distinction matters here. The COSMOS trial tested five clocks, and the signal appeared specifically in the second-generation clocks. That's important.

The COSMOS Trial Design#

The COcoa Supplement and Multivitamin Outcomes Study enrolled 21,442 participants originally, with this epigenetic aging ancillary study analyzing a subset of 958 participants (482 women, 476 men)^[2]. It was a 2×2 factorial RCT testing daily multivitamin-multimineral supplementation (Centrum Silver) and cocoa extract (500 mg cocoa flavanols, including 80 mg (−)-epicatechin) against placebo over two years.

DNA methylation was measured at baseline, year one, and year two. The statistical models used linear mixed-effects approaches, adjusting for chronological age, sex, baseline biological aging, recruitment source, and the other randomized arm^[2].

The sample size for a methylation study is solid. Not massive, but sufficient to detect the effect sizes they reported.

The Key Findings#

Daily multivitamin supplementation modestly but significantly slowed second-generation epigenetic clocks. The between-group difference in yearly change was −0.113 years (95% CI: −0.205 to −0.020; P = 0.017) for PCGrimAge and −0.214 years (95% CI: −0.410 to −0.019; P = 0.032) for PCPhenoAge^[2].

Let me be direct about the magnitude. A reduction of 0.113 years per year on PCGrimAge means the multivitamin group aged roughly 41 days less per calendar year on this clock. For PCPhenoAge, it's closer to 78 days. These are not dramatic numbers in isolation. Over a decade, you're looking at roughly one to two years of biological age savings — if the effect sustains, which we don't know.

But here's where the data told me something genuinely interesting. The effect was substantially stronger in participants who were already biologically accelerated at baseline. Among those with accelerated PCGrimAge, the between-group difference was −0.236 years (95% CI: −0.380 to −0.091) — more than double the overall effect. Among those with normal or decelerated aging, the effect was essentially null: −0.013 (95% CI: −0.130 to 0.104; P = 0.018 for interaction)^[2].

That interaction finding is the real story. It suggests the multivitamin isn't pushing already-healthy methylation patterns further — it's correcting deficiency-driven epigenetic drift in people who are aging faster than they should be.

Cocoa Extract: A Null Result Worth Noting#

Cocoa extract supplementation (500 mg flavanols daily) had zero detectable effect on any of the five epigenetic clocks tested.^[2] This matters because cocoa flavanols had shown promise in other COSMOS outcomes — cardiovascular markers, cognition. The fact that they didn't register on methylation clocks suggests either that their benefits operate through non-epigenetic pathways or that epigenetic clocks are not capturing the mechanisms flavanols influence.

I'm less inclined to dismiss cocoa flavanols based on this result than some might be. The clocks measure specific methylation sites, not all cellular health. Still, it's an honest null finding and deserves reporting.

The Skepticism Section#

Let me push back on the excitement. There are real issues.

First, Belsky and Ryan's companion editorial in Nature Medicine explicitly flags that modifying an epigenetic clock is not the same as slowing aging^[1]. Clocks are predictive biomarkers, not causal mechanisms. Changing the reading on a thermometer doesn't change the temperature of the room. Whether these methylation shifts translate to actual healthspan extension remains, in their words, "an open question."

Second, the P-values (0.017 and 0.032) are significant but not overwhelming, particularly given that five clocks were tested. The first-generation clocks (PCHannum, PCHorvath) and DunedinPACE showed no significant effects. Multiple comparisons weren't formally adjusted for. The honest answer is this could be a real but small signal — or it could be partially inflated by testing multiple endpoints.

Third — and this is the conflict of interest I want named — co-author Daniel Belsky is the inventor of DunedinPACE, which is licensed to TruDiagnostic^[1]. He declared this. I appreciate the transparency. But DunedinPACE showed no effect here, and the clocks that did (PCGrimAge, PCPhenoAge) aren't his. So the conflict, in this particular case, doesn't obviously bias the positive findings.

COMPARISON TABLE#

THE PROTOCOL#

Based on the COSMOS trial data, here is a practical protocol for anyone interested in using daily multivitamin supplementation to potentially slow epigenetic aging. This is grounded in what the study actually tested — not extrapolation.

Step 1: Establish your baseline biological age. Before starting any intervention, get a DNA methylation test through services like TruDiagnostic (TruAge) or Elysium (Index). Request PCGrimAge and PCPhenoAge specifically. Without a baseline, you cannot measure whether anything changes. Cost runs $200–$400 per test.

Step 2: Choose a comprehensive MVM supplement. The trial used Centrum Silver, a widely available and inexpensive multivitamin-multimineral formulation. The specific brand matters less than completeness — ensure it covers B vitamins (B6, B12, folate), vitamin D, zinc, magnesium, and selenium, all of which are implicated in methylation support and NAD+ synthesis cofactor pathways. Take one tablet daily with a meal containing fat for optimal absorption.

Step 3: Prioritize consistency over complexity. The COSMOS effect emerged over two years of daily use^[2]. This is not a protocol where you cycle on and off. The data suggests cumulative methylation modification, not an acute effect. Set a daily reminder. Compliance drives the outcome.

Step 4: Assess your starting risk profile. The strongest effects appeared in participants with accelerated biological aging at baseline^[2]. If your epigenetic testing shows you are aging faster than your chronological age, the evidence suggests you may benefit more from MVM supplementation than someone already aging normally. If your clocks show decelerated aging, expect minimal additional benefit from an MVM alone.

Step 5: Retest at 12 and 24 months. The COSMOS trial measured DNA methylation at baseline, year one, and year two. Mirror this cadence. Retesting sooner than 12 months is unlikely to show meaningful change and wastes money. Compare your PCGrimAge and PCPhenoAge trajectories against baseline.

Step 6: Stack with foundational interventions, not instead of them. An MVM does not replace sleep optimization (targeting 7–9 hours, high HRV overnight), regular exercise (150+ minutes moderate or 75+ vigorous per week), or dietary quality. The COSMOS participants were older adults already living normal lives — the multivitamin effect sits on top of, not in place of, a functional baseline.

Step 7: Do not add cocoa extract expecting epigenetic benefits. The trial explicitly showed no effect of 500 mg daily cocoa flavanols on any of the five clocks tested^[2]. Cocoa may have cardiovascular benefits through other mechanisms, but if your goal is methylation clock modification, the data doesn't support it.

What is an epigenetic clock and how does it measure aging?#

An epigenetic clock is an algorithm that reads DNA methylation patterns at hundreds of specific genomic sites to estimate biological age — how fast your body is deteriorating, as opposed to how many years you've been alive. Second-generation clocks like PCGrimAge and PCPhenoAge are trained against mortality and disease data, making them more clinically relevant than earlier versions^[3]. They are currently the most validated molecular biomarkers of aging we have.

How much did the multivitamin actually slow aging in this study?#

The between-group difference was −0.113 years per year on PCGrimAge and −0.214 years per year on PCPhenoAge, meaning participants taking the multivitamin aged roughly 41 to 78 fewer days per calendar year compared to placebo^[2]. These are modest but statistically significant effects. For people with already accelerated biological aging, the PCGrimAge effect more than doubled to −0.236 years per year.

Why didn't cocoa flavanols work on epigenetic clocks?#

The COSMOS trial found zero effect of 500 mg daily cocoa flavanols on any of the five epigenetic aging clocks despite prior evidence of cardiovascular and cognitive benefits^[2]. The most likely explanation is that flavanols operate through pathways — vascular function, nitric oxide signaling — that these specific methylation-based biomarkers don't capture. It's not that cocoa is useless; it's that these clocks aren't measuring what cocoa does.

Who would benefit most from this protocol?#

Based on the subgroup analysis, individuals with accelerated biological aging at baseline saw the strongest effects from multivitamin supplementation^[2]. If you are aging faster than your chronological age — which can be determined through a DNA methylation test — you appear to be the most likely responder. Those already aging normally or slowly showed essentially no benefit on epigenetic clocks from MVM use alone.

Does slowing an epigenetic clock actually mean you'll live longer?#

Honestly, we don't know yet. Epigenetic clocks predict mortality and disease risk in observational studies, but no trial has yet proven that an intervention which slows a clock also extends lifespan or healthspan^[1]. As Belsky and Ryan note in their editorial, modifying a biomarker is not the same as modifying the underlying process. This is the central open question in the field, and it would be dishonest to pretend otherwise.

VERDICT#

Score: 7/10

This is a well-designed RCT published in the highest-tier journal for clinical medicine. The data is real, the effect is statistically significant, and the subgroup finding — that accelerated agers benefit most — is biologically plausible and clinically useful. I take it seriously.

But the effect sizes are small. Very small. And five clocks were tested with only two reaching significance, without formal multiple comparison adjustment. The leap from "modifies a methylation biomarker" to "slows aging" is exactly that — a leap. Belsky himself acknowledges this in the editorial, which I respect.

What moved me most is the interaction finding. If your body is already under methylation stress from micronutrient deficiency, a multivitamin corrects that drift. That's not magic. That's biochemistry filling gaps. For people who eat well, sleep well, and aren't deficient in anything — this probably does very little.

A daily multivitamin won't make you immortal. But for $15 a month, in people who need it, the data now says it measurably slows a validated biomarker of biological aging. That's not nothing.