And here is a summary of the research paper that was covered in that last podcast. Of course, the issue again is how reliable these epigenetic age tests are across all the different types of aging, and different avenues of mitigation via different supplements and medications.
Key point in the summary below is this:
Do not buy epigenetic age tests to audit your own stack. Test-retest reliability here was 0.45 to 0.50 over a few weeks. That puts roughly 8 years of uncertainty around any single reading and about 11 years around the difference between two readings. Any change smaller than that is noise, and every effect in this paper is smaller than that.
Eight Years Became Two: What 4260 Spit Tests Say About Whether Supplements Work
Researchers at the National University of Singapore analysed saliva-based epigenetic age tests from 4260 self-selected “health enthusiasts,” 71% of whom took supplements, and cross-referenced the results against questionnaires covering 84 supplements and 22 medication classes. Delayed-release calcium alpha-ketoglutarate (the branded product Rejuvant) came out on top, associated with an Age Residual roughly 1.8 years lower than the cohort average, falling to 1.27 years after covariate matching. Coenzyme Q10 showed a weaker signal that appeared only in a small longitudinal subgroup. Almost every other supplement, including NAD+ precursors, omega-3, vitamin D and multivitamins, showed nothing once chronological age was properly accounted for. The same was true of rapamycin, metformin and GLP-1 agonists. Ordinary behaviours, chiefly not smoking and self-rated health, moved the clock two to three times harder than any pill.
The longevity supplement market runs on a simple promise: take the compound, watch your biological age number drop. In 2021 a small study reported that Rejuvant, a delayed-release calcium alpha-ketoglutarate formulation, was associated with an eight-year reduction in biological age after about seven months of use. That figure has been quoted ever since. This new analysis, drawn from a cohort more than forty times larger, puts the number closer to 1.8 years, and closer still to 1.3 once you compare users against non-users matched for age, sex, smoking, health status, exercise and the rest.
That shrinkage is the real story. It is what usually happens when a small, enthusiastic first study meets a bigger sample and better controls, and it is worth more than the headline finding itself.
The cohort is unusual and that is deliberate. These are people who paid for a saliva DNA methylation test out of pocket. Only 4% smoke, against 18% in the general US survey population, and 87% rate their own health as good or excellent. Studying the exceptionally well means you are watching aging biology rather than the tail end of untreated disease. It also means everyone is bunched together, which makes small differences hard to detect. Across every health behaviour the team measured, from exercise to alcohol to stress, the total explained variation in age residual came to between 2% and 3%.
Two findings deserve more attention than they got. First, the clock itself is a nine-site methylation panel trained to predict calendar age, with an average error of 5.4 years. The claimed alpha-ketoglutarate benefit is about a third of the instrument’s own margin of error. Second, and more awkwardly, when the researchers identified alpha-ketoglutarate users from actual subscription purchase records rather than from what people wrote on a survey, the benefit disappeared. Only the self-reported users looked younger. People who tell you they take a longevity supplement differ from people who merely buy one, and they differ in ways a questionnaire cannot capture.
There is also a chemistry problem nobody in the paper raises. Alpha-ketoglutarate is the required co-substrate for the family of enzymes that strip methyl groups off DNA. The test being used reads methylation at nine sites. Feeding people the fuel for the erasing enzymes and then measuring how much has been erased is a route by which the supplement could move the number without touching the underlying biology. That does not make the result wrong, but it is a possibility the study cannot rule out.
The medication results are similarly instructive. Rapamycin, metformin and GLP-1 agonists, the three pharmaceutical agents with the strongest preclinical geroprotective case, all sat flat. Either these drugs do not move a first-generation methylation clock, or the sample sizes of eight to fifty-one people per drug class were never going to show it.
Put in perspective, the statistical model built around alpha-ketoglutarate accounted for about 1% of the differences between people. Whatever governs how fast this clock runs, it is overwhelmingly something else.
The paper is honest about most of this. The senior author, however, sits on the board of the company that makes the winning supplement, and a co-author works for the company that sells the test. The conflict of interest statement says there are none.
Actionable Insights
The largest effects here had nothing to do with pills. Not smoking was worth about 3.5 years on this clock, a Cohen’s d of 0.52. Rating your own health as excellent rather than poor was worth 5.2 years, d of 0.77. Sitting in the top tenth of a composite health score rather than the bottom tenth was worth 4.4 years, d of 0.65. Cohen’s d is simply how many standard deviations apart two groups sit; 0.2 is small, 0.5 moderate, 0.8 large.
The best supplement signal, delayed-release alpha-ketoglutarate, was 1.27 to 1.81 years, a d of 0.22 to 0.32. In plain terms, if you picked one user and one non-user at random, the user would look younger about 56 to 59 times out of 100. A coin flip is 50.
Nothing else in the cabinet moved the needle. NAD+ precursors, omega-3, vitamin D, magnesium, creatine, multivitamins, curcumin and quercetin were all statistically indistinguishable from taking nothing, as were rapamycin and metformin.
Do not buy epigenetic age tests to audit your own stack. Test-retest reliability here was 0.45 to 0.50 over a few weeks. That puts roughly 8 years of uncertainty around any single reading and about 11 years around the difference between two readings. Any change smaller than that is noise, and every effect in this paper is smaller than that.
Context and Source
-
Open Access Paper: Supplements and Drugs Are Associated With Biological Age in a Cohort of Exceptionally Healthy Individuals
-
Authors: Pabis K, Wang W, Selvarajoo K, Budovskaya YV, Sorrentino V, Gruber J, Kennedy BK
-
Institution: Healthy Longevity Translational Research Programme, Yong Loo Lin School of Medicine, National University of Singapore. Additional affiliations at A*STAR Bioinformatics Institute, Nanyang Technological University, and TruMe Inc. (Alameda, California, USA).
-
Country: Singapore, with a US industry co-author.
-
Journal: Aging Cell (Wiley, on behalf of the Anatomical Society), Accepted 16 April 2026
-
Impact evaluation: The impact score of this journal is 7.7 (2025 Journal Impact Factor, released in the 2026 JCR; five-year JIF 8.9; Scopus CiteScore approximately 13.8), evaluated against a typical high-end range of 0 to 60+ for top general science, therefore this is a Medium impact journal. Within its own field the picture is stronger: Aging Cell ranks Q1, fifth of 73 journals in its primary JCR category, and is a leading specialist venue for biogerontology.
