This highly provocative and substantive paper in Nature Medicine identifies putative epigenetic biorhythmic clock pathways for ageing which may be nutritionally responsive. The RCT study design for a large multicultural population adds plausibility to possible effects of cocoa flavanols, other phytonutrients such as the carotenoid lutein and of micronutrients such as folic acid and B-12, although these are small.

Nevertheless, cumulatively, these effects and others likely to be identified by cognate methodology may provide new scope for greater longevity through nutritional biology. Caution is required in any kind of nutritional supplement approach, however. The present study does take background diet into account by way of a healthy pattern index whose excursions will have attributions beyond those recognised in the study, but will matter. For example, we now know that a myriad of foods contain factors like melatonin with its unique biorhythmic effects. This study from Harvard will be a catalyst to future food and nutrition and ageing science.

The report showing that a daily multivitamin supplement can slow biological markers of ageing is exciting.

It is important to note that more is probably not better - for some vitamins, excess doses are harmful. An example of this is vitamin B6, which at high doses causes nerve damage.

In people who are concerned that their diet does not contain enough of the range of vitamins that it could, this data supports taking 1 multivitamin supplement per day.

"This is a well-conducted trial, but the interpretation of the findings needs to be cautious. The study reports that a daily multivitamin slightly slowed two of five DNA-based ‘epigenetic clocks’ used as markers of biological ageing in older adults. However, the magnitude of the effect is extremely small, and the effect was not consistent across all the ageing measures tested.

Another important issue is that several ageing markers were analysed and the statistical tests were not adjusted for multiple comparisons. When many outcomes are tested simultaneously, it becomes more likely that some apparently ‘positive’ results will appear simply by chance.

It is also important to remember that epigenetic clocks are still research tools rather than validated clinical endpoints. While they can respond to lifestyle or pharmacological interventions, we do not yet know whether small changes in these biomarkers translate into meaningful reductions in disease risk, disability or mortality.

Participants in this study were generally healthy older adults rather than people with clear nutritional deficiencies. If the effect is real, it may simply reflect correction of mild micronutrient insufficiencies in some individuals rather than a fundamental slowing of the biological ageing process.

Overall, the findings are scientifically interesting, but they do not provide convincing evidence that taking a daily multivitamin meaningfully slows human ageing.

-----------------------------------

Below are my extended comments:

1. The data show a statistically significant but very small slowing of two of the five epigenetic ageing measures tested (PCGrimAge and PCPhenoAge) with daily multivitamin–multimineral (MVM) supplementation (Centrum Silver) over two years in a COSMOS ancillary subcohort. The reported between-group differences in yearly change were −0.113 years for PCGrimAge and −0.214 years for PCPhenoAge, corresponding to extremely small standardized effects (Cohen’s d ≈ −0.03 to −0.04).

   However, the framing of the findings risks overinterpretation for two main reasons.

   First, multiple testing. Five epigenetic clocks were analysed, and the paper notes that the P values were assessed without adjustment for multiple comparisons. When multiple outcomes are tested, the likelihood that some statistically significant findings occur by chance increases, particularly when the effect sizes are extremely small.

   Second, clinical relevance remains uncertain. The manuscript itself acknowledges that further studies are needed to determine the clinical significance of these findings. Importantly, a statistically detectable change in a methylation-derived biomarker does not constitute validated evidence of slowed ageing, reduced disease risk, or improved survival.

2. At most, the study provides a modest signal that daily multivitamin supplementation might slightly influence certain methylation-derived ageing biomarkers in older adults. However, the effects are small in absolute terms and not consistent across the ageing measures examined. There was no effect on the first-generation epigenetic clocks, no statistically significant change in DunedinPACE, and cocoa extract did not slow any of the clocks; if anything, it was associated with a small increase in PCPhenoAge.
   

   From a geroscience perspective, the key question is whether such biomarker changes translate into meaningful health benefits. Current evidence does not support this assumption. As emphasized in a recent consensus framework published in Cell, demonstrating that a biomarker responds to an intervention is only an early step. Establishing a surrogate endpoint requires robust evidence that biomarker changes reliably predict improvements in clinical outcomes. This has not yet been demonstrated for epigenetic clocks. For these reasons, the findings are best interpreted as a biomarker response of uncertain biological and clinical significance, rather than evidence that multivitamins slow human ageing.

3. If the effect is real, the most plausible explanation is not a fundamental anti-ageing mechanism but rather the correction of mild micronutrient insufficiencies that could influence inflammatory or metabolic biomarkers embedded within second-generation clocks.
   
   The manuscript reports changes in components of PCGrimAge such as DNAm-derived surrogates for cystatin C, GDF-15, beta-2 microglobulin, and DNAm telomere length, suggesting that the observed shifts may reflect modest alterations in circulating biomarker profiles rather than a true slowing of the underlying ageing process.
   
   Importantly, the study population was not obviously malnourished: participants had a mean BMI of ~27, an average AHEI diet quality score of ~43, and only about 3.5% were current smokers. Therefore, if the effect is real, it may reflect correction of subclinical nutrient insufficiencies in a subset of participants, regression to the mean, or subtle changes in blood cell composition or methylation variability, rather than a robust systemic effect on biological ageing.
4. The study population consisted of older adults (mean age ~70 years), and the analysis was restricted to a selected “healthy survivor” subcohort—participants free of major chronic disease at baseline and without incident cardiovascular disease or cancer during the two-year follow-up period.
   
   These findings therefore, cannot be extrapolated to younger populations, where nutritional status, physiology, and ageing dynamics differ substantially.

Future research should focus on:

- replication in independent cohorts with pre-specified primary epigenetic endpoints and appropriate control for multiple testing

- establishing whether changes in epigenetic clocks predict clinically meaningful outcomes, such as reduced frailty, improved function, or lower disease incidence

- evaluating whether potential effects are confined to individuals with biochemically confirmed micronutrient insufficiencies, rather than the general population.

5. To place the magnitude of these effects in context, lifestyle interventions such as long-term calorie restriction and endurance exercise are associated with consistent biological age differences across multiple molecular and physiological layers, including epigenomic, transcriptomic, metabolomic, and microbiome measures.
   
   In our recent Aging Cell study, individuals practicing long-term calorie restriction with optimal nutrition or endurance exercise showed broadly lower multi-omic biological age measures compared with Western-diet controls across several biological systems. While that study is observational and cannot prove causality, it illustrates that meaningful ageing biology signals tend to appear coherently across multiple biological domains, rather than as extremely small changes in a subset of epigenetic clocks.
   
   In conclusion, the COSMOS trial itself is rigorous, but this ancillary analysis provides weak evidence that multivitamin supplementation alters biological ageing in a meaningful way. The observed effects are extremely small, limited to a subset of epigenetic clocks, uncorrected for multiple testing, and not linked to clear clinical outcomes.

A more balanced interpretation would be that daily multivitamin supplementation produced minor statistical changes in certain epigenetic ageing biomarkers of uncertain biological and clinical relevance. Presenting these findings as evidence that multivitamins “slow ageing” risks overstating what the data actually demonstrate.