A 24-week NMN trial raised blood NAD—not hearing or physical function

The biochemical result was large—and narrow

After 24 weeks, median whole-blood NAD rose from 14.08 to 51.17 micromolar in the NMN arm. It moved from 12.02 to 9.37 in placebo. The median changes were +38.63 and −3.14 micromolar, respectively; the baseline-adjusted comparison was highly significant. The paper describes the NMN arm’s value as about 3.6 times baseline and reports coordinated changes in several related metabolites.

That is evidence of biochemical target engagement in whole blood. It is not a clinical outcome, a measurement of NAD in untested tissues, a biological-age result or evidence that participants aged more slowly. A compound can move a molecule without producing a meaningful change in how people feel, function or live.

The registered primary outcome was hearing

The researchers randomized 62 healthy Japanese adults age 65 or older to 1,250 mg of NMN per day or a matched placebo. Both the initial 2022 registry snapshot and the current record identify change in hearing ability at 24 weeks as the primary outcome. The manuscript specifies a pure-tone average across 500, 1,000, 2,000 and 4,000 hertz.

The authors’ efficacy analysis included 60 treated participants without major protocol deviation: 31 assigned NMN and 29 placebo. This was not a conventional all-randomized intention-to-treat analysis. Fifty-eight completed 24 weeks—30 and 28—after one NMN and three placebo discontinuations or exclusions.

Median hearing change was +1.25 decibels in each arm, with p=0.95 and reported q=0.999. The result does not prove the treatments were equivalent: the planned calculation targeted a 5 dB difference, and the authors note that subtle auditory change may require a larger or longer study.

Physical function did not follow the blood marker

Grip strength, gait speed, chair stands, six-minute walk and cardiopulmonary measures showed no significant attributable improvement. Nominal findings in soft lean mass, lipids and dietary behavior did not remain significant after the paper’s false-discovery-rate correction. Subgroup and exploratory analyses were small, not multiplicity-adjusted and underpowered.

That pattern matters more than selecting one favorable p-value. A 2024 synthesis of eight short randomized trials likewise found no significant glucose-control or lipid-profile benefit. The new study extends exposure to 24 weeks and documents strong whole-blood target engagement, but it does not overturn the broader absence of established metabolic or functional benefit.

How much confidence should readers place in it?

Randomization, placebo control and blinding strengthen the comparison. Limits remain substantial: one center, a selected population, a small analyzed set and 24 weeks of follow-up. The eligibility rules excluded several chronic conditions, medication or supplement use, severe hearing loss or hearing aids, and sustained daily exercise. Results should not be generalized to younger, sicker, medicated or more active people.

A later observational registry record, jRCT1030240368, reuses trial data and samples for body-composition, skin and NAD-related analyses. It is not another randomized trial or an independent replication. Those secondary-use analyses cannot be added to the randomized evidence as though a second experiment confirmed the first.

Safety evidence is still short-term

Adverse events occurred in 7 of 31 NMN participants and 6 of 29 placebo participants; the small comparison was not significant. One placebo participant had serious pancytopenia judged unrelated to NMN. One NMN participant who already had mildly abnormal liver enzymes had further asymptomatic elevation at week 6 and stopped. Investigators judged a causal link unlikely but could not exclude it.

A 2026 review of 15 randomized trials found no clear pooled increase in short-term adverse events or liver enzymes and no broad metabolic benefit. The longest trials lasted 24 weeks. Small, heterogeneous studies cannot rule out uncommon harms or establish long-term safety.

Funding and regulation belong beside the result

MIRAILAB BIOSCIENCE funded the trial, supplied the NMN and supported ELISA measurements. Three authors were employees, including one who was CEO, CTO and representative director. The paper says the company did not participate in study design, clinical data collection, statistics, interpretation, writing or the submission decision. Both the relationship and that reported role boundary matter.

FDA’s 2025 petition response said NMN was not excluded from the dietary-supplement definition under one statutory clause because the agency found earlier U.S. supplement marketing. That is not product approval or a finding of safety or efficacy. FDA does not approve dietary supplements for safety and effectiveness before sale, and an NDI notification is not approval. A separate FDA warning about food-grade NAD+ in sterile compounding concerns injectable products, not this oral trial.

What the next trial would need to add

A useful follow-up would prespecify a clinically meaningful functional outcome, use an all-randomized analysis, enroll a larger and more representative population, and last long enough to assess durability and less common harms. Replication should test whether biochemical target engagement predicts any patient-relevant outcome rather than assume it does.

The practical reading is therefore two-part: the regimen clearly changed a measured blood metabolite under these conditions, while this experiment did not demonstrate better hearing or physical function. Keeping those statements separate is the difference between reporting target engagement and advertising an anti-aging effect.