Three of the eighteen NMN bottles a Singapore lab opened in 2024 contained no NMN.
The lab was Andrea Maier’s group at NUS Medicine’s Healthy Longevity Translational Research Programme. They bought eighteen nicotinamide mononucleotide products from online stores, from pharmacies and direct from manufacturers, then measured what was inside each one. Against the printed label, the deviations ran from minus 100 percent to plus 11.2.
Minus 100 percent means the capsule was a capsule.
Set the empty ones aside, because the harder problem is the honest bottle. Take one holding exactly what its label promises.
As of early 2026, no human trial has shown that NMN extends life or reverses ageing, for the plain reason that no trial has measured it. The human work runs short, runs small, and reports surrogate markers.
A 2024 meta-analysis in Current Diabetes Reports pooled eight randomised controlled trials and 342 middle-aged and older adults, and found no significant effect on fasting glucose, fasting insulin, HbA1c or the lipid panel.1 Insulin resistance drifted toward significance, then failed the sensitivity analysis.
So the honest bottle buys a compound that raises NAD+ in your blood and has not yet moved anything you would feel.
Absence of evidence is not proof of absence, and that distinction is why the field deserves watching rather than dismissal. Cheap mouse data is giving way to expensive human trials, which is the right sequence, arriving late.
One supplement trial did move a clock. TALENTs, run in Chengdu and published in Advanced Science last May, randomised 121 adults aged 60 to 70 to nucleotides or placebo for nineteen weeks and reported methylation age 3.08 years below control. Telomere length in the same participants held still.
One clock moved, one did not, and a clock is not a life.
One clock moved, one did not, and a clock is not a life.
The interventions carrying real momentum now arrive with a prescriber attached, which is itself the tell. Urolithin A is the instructive case. In JAMA Network Open, 66 adults aged 65 to 90 took it for four months and missed both primary endpoints, the six-minute walk and hand-muscle ATP production, while muscle endurance, a secondary measure, improved.
That is the shape of credible longevity data in 2026.
A secondary endpoint moves, the primary ones hold still, and the result earns another trial rather than a subscription.
A secondary endpoint moves, the primary ones hold still, and the result earns another trial rather than a subscription.
Take what you swallow to your own doctor before you change any of it. Nothing on this page is a protocol, and journalism has never seen your bloodwork.
In April, the Lancet Healthy Longevity published the other side of the ledger. Forty-four studies, 145,465 people: higher physical activity ran with lower biological age on the Horvath and GrimAge clocks. The authors flag their evidence as mostly cross-sectional, which caps what anyone may claim about cause.
Maier’s name is on that paper too.
The same Singapore researcher stands behind both halves: the bottles that did not contain what they sold, and the thing nobody bills for that keeps turning up in the clocks. Both papers argue the same point, which is that you should know which of your longevity spending is a bet and which is a finding.
The three empty capsules could not be told from the fifteen full ones without a mass spectrometer. The stairwell in your building needs no such check.
Footnotes
-
The pooled trials ran short and enrolled adults who were largely healthy to begin with, which is exactly the population in which a metabolic marker has least room to move. A null there is weaker news than it sounds, and still the best evidence anyone has. ↩