A supplement stack built on NMN was linked to a roughly 4-year drop in an epigenetic aging clock in about six weeks. That line is made for a product page. The study behind it is more modest, and the authors say so themselves.
The paper, published in Nutrients (2026; DOI 10.3390/nu18213490), tested NMN-based formulations in middle-aged Japanese adults. Here is what it found, what it didn't, and how to read similar claims when you see them on a supplement label.
The Headline Number: ~4.1 Years on DNAm PhenoAge
In a separate, uncontrolled molecular pilot of eight participants, DNAm PhenoAge relative to chronological age decreased by approximately 4.1 years. The pilot also showed nominal transcriptomic changes.
The authors then add a caveat: these exploratory age-related measures "do not establish biological age reversal." The pilot had no control group and only eight people. The abstract also doesn't break down per-participant results or say how much of the shift could be measurement noise.
A drop of about four years on a methylation-based clock is an observation. It is not proof that anyone became biologically younger.
What Was Actually Tested
The main trial was a 42-day, open-label, four-arm exploratory study of 40 Japanese adults aged 45–65 in the final analysis:
- Placebo: n = 10
- NMN alone: n = 9
- Core combination: n = 10
- Expanded eight-component formulation: n = 11
The combinations built on NMN (a nicotinamide adenine dinucleotide, or NAD+, precursor) with β-hydroxybutyrate (BHB), ginsenosides, and what the authors call "complementary metabolic modulators." The abstract doesn't list all eight components or their doses, so this article doesn't either.
Participants had cardiopulmonary exercise testing (CPET) and computerized cognitive assessments (CogniTrax) at baseline and week 6. Exploratory work added immune and inflammatory biomarkers and model-derived functional-age estimates for a placebo-versus-expanded-formulation subset.
The Fine Print That Matters
No significant between-group wins on the main outcomes
This is the most important sentence in the abstract. Between-group comparisons did not demonstrate statistically significant improvements in the selected cardiopulmonary or cognitive outcomes. The authors say the work does not establish clinical efficacy or ingredient-specific effects.
Descriptive cognitive differences, not proof
The expanded-formulation group did show favorable descriptive changes compared with placebo:
- Neurocognitive Index: +3.09 vs. −1.90
- Composite memory: +5.82 vs. −5.30
- Verbal memory: +9.00 vs. −2.50
With about 10 people per arm, and no significance in the formal comparisons, these numbers can swing on a few individuals. In an open-label design, participants know what they are taking, and expectation can influence cognitive test performance.
Biomarkers with unadjusted p-values
Nominal between-group differences appeared in CTACK (p = 0.015) and GDF-15 (p = 0.022). Neither was adjusted for multiple testing, and the authors state that the direction of change does not establish clinical benefit. When many markers are checked, a few will cross p < 0.05 by chance.
The functional-age composite
A model-derived functional-age composite fell 2.79 years in the expanded-formulation group versus 0.63 years in placebo. It is a modeled estimate built from other measurements, not a direct measure of aging, and it is also labeled exploratory.
Safety and adherence
On the positive side, the intervention was associated with high recorded adherence and no reported serious adverse events. That supports the feasibility the authors claim, but a six-week study with 40 people can't answer long-term safety questions.
Who Funded and Formulated This?
The disclosures are worth knowing. Abe Yoando Pharmaceutical Co., Ltd. developed and provided the study formulations. The conflict-of-interest statement also notes that supplied project materials state a patent application (No. 2025-079089) has been filed concerning NMN+BHB-related muscle, cognitive, and anti-inflammatory effects. One listed author was employed by SOUKEN Co., Ltd.
This doesn't invalidate the work. The authors are open about the study's limits. But a company with a patent filing and a product stake is a reason to ask for independent replication.
Practical Takeaways
1. Read "open-label" as "everyone knew." Participants and investigators weren't blinded, so expectation effects and subjective bias can't be separated from any real effect.
2. Small arms limit conclusions. With 9–11 people per group, one unusual responder can change an average. Descriptive differences like +9.00 vs. −2.50 are hypothesis fodder, not results.
3. An uncontrolled clock pilot can't show cause. Without a comparison group, a 4.1-year PhenoAge shift can't be separated from test-retest variability, lifestyle changes, or the placebo effect. Eight participants is too few to say what a typical person would see.
4. Know what an epigenetic clock is. DNAm PhenoAge is a methylation-based estimate trained to predict health-related outcomes. Moving the number is not the same as changing your actual risk of disease or death. Marketing that says "X years younger" skips that gap.
5. Spot the red flags on supplement pages. Look for:
- a clock result with no control group
- an n that isn't stated
- "reverse aging" language
- p-values cited without mention of multiple testing
- a stack of eight ingredients presented as if each one were proven
The authors themselves say the study cannot establish ingredient-specific effects. Any marketing that credits NMN, BHB, or ginsenosides individually goes beyond what this paper shows.
6. Who might reasonably experiment? This study doesn't make a recommendation. Healthy adults curious about NAD+-related supplements may reasonably wait for randomized, double-blind data. Anyone with a medical condition or on medication should talk to a clinician before adding a multi-ingredient stack, especially one whose components and doses aren't fully disclosed. If you do try something, track objective measures with a clinician's help rather than relying on a single clock number.
Bottom Line
This is a feasibility-stage study, not a breakthrough. It suggests that six weeks of an NMN-based stack was tolerable and generated some interesting signals in cognition, inflammation markers, and methylation. It does not show that the stack works, that any one ingredient is responsible, or that anyone's biological age was reversed. The authors call for larger randomized, double-blind trials with prespecified outcomes. Until those arrive, treat the four-year headline as a question to test, not a result.