Tier 2 — strong

Supplements and Drugs Are Associated With Biological Age in a Cohort of Exceptionally Healthy Individuals

Kamil Pabis, Weilan Wang, Kumar Selvarajoo, Yelena V. Budovskaya, Vincenzo Sorrentino, Jan Gruber, Brian K. Kennedy

Aging Cell 2026 Volume 25, issue 6, article e70517

Bibliography

DOI: 10.1111/acel.70517
PubMed: PMID 42166733
PubMed Central: PMC13239866
Funding: V.S. is supported by the NUHS Internal Grant Funding under NUS Start-up grant NUHSRO/2022/047/Startup/11, and an MOE Tier 2 grant T2EP30124-0014. B.K.K., V.S., and K.P. are supported by a Healthy Longevity Catalyst grant (HLCA25Jan-0012). TruMe Labs provided access to proprietary data but had no role in the interpretation of the data or the final decision to publish the manuscript. The authors otherwise state that they have nothing to report regarding funding.
Competing interests: The published Conflicts of Interest statement reads: "The authors declare no conflicts of interest." However, the Acknowledgments section discloses that B.K.K. (Brian K. Kennedy, senior author) is a scientific advisor and board member of PDL Health, the company that produces the Rejuvant delayed-release calcium-α-ketoglutarate supplement analysed as the primary finding of this paper. Co-author Y.V. Budovskaya is affiliated with TruMe Inc., the company that manufactures the saliva-based epigenetic clock used as the primary biomarker in the study, and TruMe Labs provided access to the proprietary participant dataset. Participants recruited via PDL Health received free epigenetic test vouchers upon Rejuvant purchase, which the authors acknowledge as a potential source of recruitment bias.

Study snapshot

Design	Cross-sectional observational cohort with longitudinal subset analysis; no randomisation, no placebo control
Model	Human — convenience sample of 4260 "health enthusiast" adults (mean age 53.5 ± 13.2 years; 59% male; 96% non-smokers; predominantly White/Caucasian), recruited between 2020 and 2025 via supplement companies and telehealth clinics; unusually low prevalence of smoking (4% vs 18% in NHANES) and poor/fair self-rated health (12% vs 25% in NHANES)
Sample	4260 unique participants provided a valid date of birth and epigenetic test; 3628 provided complete questionnaire data for multivariate analysis; 755 provided ≥2 tests for the longitudinal subset; 143 self-declared dAKG users, 294 dAKG subscribers identified via order history
Intervention	Self-reported use of 84 supplement categories and 22 medication classes; primary supplement of interest was delayed-release calcium-α-ketoglutarate (dAKG, Rejuvant, containing added vitamin A for men or vitamin D3 for women). Additional named supplement analyses: any AKG (dAKG plus non-delayed formulations), coenzyme Q10, vitamin D, NAD+ precursors (NMN, NR, niacin), carotenoids, calcium, curcumin, and others. Dose, frequency, and duration were not recorded.
Duration	Cross-sectional snapshot; longitudinal subset had mean 367 days (range 0–1741 days) between epigenetic tests
Endpoints	Age Residual (residual from linear regression of Biological Age on Chronological Age); Age Delta (Biological Age minus Chronological Age); Odds ratio of improvement in Age Residual over time in the longitudinal subset; Biological Age measured via proprietary 9-CpG saliva-based TruAge epigenetic clock (mean absolute error 5.4 years)

What the study showed, in plain terms

This 2026 paper from Brian Kennedy's group at the National University of Singapore looked at whether people who take specific supplements or medications have a younger biological age than people who don't. Biological age was measured with a saliva-based DNA methylation test called TruAge — a nine-marker epigenetic clock that estimates how old your body appears at a molecular level, regardless of your chronological age.

The study analysed 4260 "health enthusiasts" who purchased at least one TruAge test between 2020 and 2025. Because 71% of them were already taking supplements — well above the general-population rate — the researchers had unusually good power to detect associations between specific supplements and biological age. Of 84 supplement categories tested, one stood out: delayed-release calcium-α-ketoglutarate (dAKG, sold as Rejuvant), which was associated with an average 1.8-year lower Age Residual after adjusting for age, sex, smoking, and health status. Plain (non-delayed-release) AKG showed a smaller, non-significant benefit. Coenzyme Q10 (CoQ10) users at baseline were roughly twice as likely to show improvement in biological age at follow-up in the smaller longitudinal analysis (n=755).

For the Ca-AKG cluster this paper matters because it is the largest observational dataset to date suggesting a biological-age benefit from the Rejuvant formulation, following the smaller 42-person Demidenko 2021 study from the same product's manufacturer. However, the paper has important caveats — the senior author is a board member of the company that produces Rejuvant, the epigenetic test used is a first-generation clock less sensitive to interventions than newer clocks, and the longitudinal dAKG signal did not survive adjustment for baseline age and regression to the mean. The findings should be read as hypothesis-generating and pending randomised confirmation from trials such as ABLE (NCT05706389) and the Chinese Ca-AKG RCT (NCT07114536).

Key findings

Cohort characteristics. 4260 participants; mean chronological age 53.5 years; mean biological age 51.4 years (2.1-year average reduction versus chronological age). Smoking rate 4% versus 18% in NHANES; poor/fair self-rated health 12% versus 25% in NHANES.
dAKG primary finding. In the cross-sectional analysis, dAKG users had an average 1.8-year lower Age Residual than non-users. The association remained significant (p<0.01) after adjustment for age, sex, smoking, health status, and additional covariates (weight, height, physical activity, alcohol, mental health, sleep quality, stress, ethnicity, recruiting company, country). Matching analysis showed a 1.27-year Age Residual reduction versus matched controls.
Non-delayed AKG. Participants taking plain (non-delayed-release) AKG showed a smaller, statistically non-significant Age Residual benefit, suggesting the delayed-release formulation or its added vitamins may drive the signal.
CoQ10 longitudinal signal. In the longitudinal subset (n=755, mean 367 days between tests), CoQ10 users at baseline were approximately twice as likely to show improvement in Age Residual over time (OR 1.87, 95% CI 1.01–3.50, p=0.048, n=56) — a nominally significant unadjusted finding.
dAKG longitudinal signal was not significant. Only 26 self-declared dAKG users had multiple tests; the analysis was underpowered. Adding the larger dAKG-subscriber group (n=294) produced an unadjusted odds ratio of 1.37 (95% CI 1.01–1.89, p=0.045) for improvement, but the effect became non-significant after adjusting for baseline Age Residual and regression to the mean.
Vitamin D and NAD⁺ boosters were not significant. After multiple-testing correction, neither vitamin D subtypes, NAD⁺ precursors (NMN, NR, niacin), omega-3, nor most other supplements showed a significant Age Residual benefit. Branded Tru Niagen (nicotinamide riboside) was nominally significant before correction only.
Medications. Antihistamines ranked among the top for lower Age Residual (non-significant); statins, antithrombotic agents, and antihypertensives showed lower Age Delta but the signals attenuated substantially after adjustment for age.
Health behaviours are weak but consistent predictors. Smoking, poor sleep quality, and low self-rated health together explained only 2–3% of the variance in Age Residual; participants in the top decile of a composite health score had a 4.4-year younger Age Delta than the bottom decile.
Test–retest reliability. The intraclass correlation coefficient for the TruAge saliva clock over short intervals is 0.45–0.50, lower than for blood-based epigenetic clocks such as DunedinPACE — reflecting the trade-off for a cheap, non-invasive assay.

What this study can and cannot tell us

Observational, not randomised. The study cannot establish causation. All findings are consistent with confounding by unmeasured health-conscious behaviours in dAKG users versus non-users.

Senior-author financial interest in the primary intervention. Brian K. Kennedy (senior author) is a scientific advisor and board member of PDL Health, the manufacturer of Rejuvant. Kamil Pabis (first author) and Vincenzo Sorrentino receive Healthy Longevity Catalyst grant funding co-supporting the work. Co-author Y.V. Budovskaya is affiliated with TruMe Inc., the manufacturer of the epigenetic clock used as the primary readout. The published Conflicts of Interest statement declares "no conflicts of interest," which materially understates the disclosure surfaced in the Acknowledgments.

Recruitment bias. A large subset of dAKG users were recruited via PDL Health and received free epigenetic test vouchers upon Rejuvant purchase. The authors acknowledge this may bias the sample toward more health-seeking individuals within the dAKG-using group. Sensitivity analyses stratified by recruiter (TruMe vs PDL Health) showed dAKG associated with lower Age Residual in both sub-cohorts, but the effect was larger in the PDL Health sub-cohort (unadjusted p=0.00072) than the TruMe sub-cohort (p=0.036).

Longitudinal signal did not survive full adjustment. The longitudinal analysis is the closer analogue of a controlled trial. There, dAKG subscribers showed a nominal improvement in Age Residual (OR 1.37, p=0.045) that became non-significant after adjusting for baseline Age Residual and regression to the mean.

First-generation epigenetic clock, low test–retest reliability. The 9-CpG TruAge clock was trained on chronological age (first-generation design), which the authors themselves note is less sensitive to interventions than second- and third-generation clocks such as PhenoAge, GrimAge, or DunedinPACE. Test–retest ICC of 0.45–0.50 means individual-level change scores over months carry substantial noise.

Self-reported supplement use. Frequency, dose, brand (except where dAKG subscription data was available), and duration of supplement use were not recorded. Supplement lists were cleaned partially by ChatGPT-assisted fuzzy matching. Users who stopped a supplement between tests may be misclassified as active users.

Cohort demographics limit generalisability. Predominantly White/Caucasian; only heterogeneous self-reported ethnicity data available. Findings may not extend to other ancestries, non-health-enthusiast populations, or people with active chronic disease.

dAKG is a mixed product. Rejuvant contains delayed-release calcium α-ketoglutarate plus vitamin A (men's formula) or vitamin D3 (women's formula). The paper cannot separate the AKG contribution from the vitamin contribution, and plain calcium α-ketoglutarate showed a smaller non-significant effect — leaving the mechanism of the dAKG-specific signal unresolved.

Multiple testing. 84 supplement categories and 22 medication classes were tested. Benjamini–Hochberg correction was applied to univariate comparisons, but the substantial number of downstream subgroup, sensitivity, and longitudinal analyses raises the risk of chance findings surviving.

Contradicts the Demidenko 2021 effect size. Demidenko et al. (the retrospective 42-person Rejuvant study) reported an 8-year biological-age reduction after ~7 months of use. The Pabis 2026 effect is 1.8 years in the far larger cross-sectional cohort. If both are real, the true effect is likely closer to the smaller Pabis estimate, given the much larger sample and adjustment for confounders. The gap between marketing claims ("8 years younger") and the current best observational estimate is substantial and should be flagged in cluster articles that cite either study.