How To Slow Biological Aging With a Multivitamin, Vegetables, & Omega-3 | Dr. Steve Horvath

The COSMOS trial was a large, rigorous randomized controlled trial in older adults. The cognitive arm showed that the multivitamin group had brain aging slowed by 2.1 years globally and episodic memory aging slowed by almost 5 years compared to placebo. The epigenetic sub-study (2 years) showed GrimAge and PhenoAge slowed by 2.7-5 months. DunedinPACE moved in the right direction but wasn't statistically significant. Horvath calls this a 'triumph for epigenetic clocks' because the clocks detected a signal that hard endpoints (mortality, CVD, cancer) did not—likely because those endpoints require even larger samples and longer follow-up. He frames it as 'testing the test': an intervention that should work based on biology (correcting deficiencies) did move the clocks, validating their sensitivity. Rhonda Patrick emphasizes that 90% of Americans don't get enough essential nutrients from diet, so a multivitamin acts as nutritional insurance.

The DO-HEALTH trial, led by Heike Bischoff-Ferrari in Switzerland, was a rigorous placebo-controlled RCT with 780 participants aged 71+. It tested omega-3 (1g), vitamin D (2000 IU vs 800 IU), and a home exercise intervention, alone and in combination. Omega-3 alone showed the most credible effect on epigenetic clocks. The combination of all three interventions produced the best results: PhenoAge was delayed by 3.8 months over 3 years, and this was associated with a 61% reduced chance of metastatic cancer and a 20% reduction in pre-frailty. There was a dose-dependent effect: omega-3 alone < omega-3 + vitamin D < all three combined. This is notable because the population was already relatively healthy (88% physically active at baseline), suggesting even healthy people can benefit.

Horvath has been deeply dissatisfied with the weak exercise signal in large observational studies (correlation ~0.1 with GrimAge). Those studies used step counts or mild home exercise in older adults. The 2025 cycling study was different: it imposed a real physiological challenge (4.5 hrs/week) that produced a 20% VO2 max improvement. Only then did the clocks respond robustly—7.4 months reduction in PC GrimAge over just 6 months. This dwarfs the 3-5 months from multivitamins over 2 years. Rhonda Patrick contextualizes this with Ben Levine's work showing that 5 hours/week of mixed aerobic exercise over 2 years reversed structural heart aging by ~20 years in 50-year-olds. The key insight: there's likely a threshold effect where exercise must be intense enough to improve VO2 max before epigenetic clocks register it.

Horvath and Patrick discuss the nuance: the DO-HEALTH trial showed no effect of vitamin D on epigenetic clocks, but that trial compared 2000 IU to 800 IU in a population that was already vitamin D sufficient. The Berlin BASE-II study, by contrast, studied a population in Berlin (limited sunshine) that was deficient at baseline, and correcting that deficiency over 7 years slowed epigenetic aging. Patrick emphasizes that many studies fail to measure baseline levels or only measure a subset, leading to misleading null results. She also notes that genetic SNPs affect vitamin D metabolism, and magnesium is a required cofactor for converting vitamin D to its active form. The take-home: test, don't guess. Correct a deficiency, but don't expect supranormal levels to provide additional anti-aging benefits.

The DO-HEALTH trial's factorial design allowed testing of individual and combined effects. The triple combination arm showed the best results on PhenoAge (3.8 months delay over 3 years) and on clinical outcomes. This is particularly impressive because the population was already relatively healthy (88% physically active at baseline, vitamin D sufficient). The dose-response pattern (one intervention < two < three) suggests genuine synergy rather than a threshold effect. Patrick notes that even in an already healthy population, adding these interventions produced measurable benefits, and the effects would likely compound over decades.

The study put participants on a bicycle regimen of 4.5 hours per week for 6 months. This is not an extreme amount for athletes, but Horvath notes it would be daunting for most busy people. The participants achieved a 20% improvement in VO2 max and many other physiological benefits, confirming they didn't fake the effort. PC GrimAge (a principal-component-based version of GrimAge with better test-retest reliability) showed a 7.4-month reduction. This effect size dwarfs the 3-5 months seen with multivitamins over 2 years. The study lacked a control arm, which Horvath acknowledges, but the physiological improvements lend credibility. The implication is that there may be a threshold effect: walking and light activity don't register on these clocks, but vigorous exercise that meaningfully improves cardiorespiratory fitness does.

The study, led by Sinisa Vratin at Harvard, used optogenetic or chemogenetic stimulation of pre-optic neurons to lower mouse body temperature by about 3°C. When they then measured methylation clocks in multiple organs, they found a very strong slowing of epigenetic aging compared to control mice. Horvath finds this fascinating because it suggests that metabolic rate and body temperature are directly coupled to the pace of epigenetic aging. He notes that while sauna and heat exposure have undisputed benefits, this study raises the possibility that lowering core body temperature by even a degree could be beneficial. Rhonda Patrick adds that during hibernation, everything slows down—metabolism, neural firing, neurotransmitter activity—and all of that normal activity produces damage, so slowing it reduces damage accumulation.

Horvath specifically mentions that he has never seen evidence that short-term psychological stress has a strong effect on epigenetic clocks. He frames this as a hopeful message for people who worry about their busy, stressful lives. The clocks appear to integrate long-term, severe stressors rather than transient ones. This aligns with his conceptualization of the epigenome as a memory of prolonged stressors that prime the cell for future challenges, not a real-time stress meter.