Article — "NAD⁺ Precursor Supplements Keep Failing: Latest NR Trial Shows No Cognitive or Vascular Benefit": Article — "NAD⁺ Precursor Supplements Keep Failing: Latest NR Trial Shows No Cognitive or Vascular Benefit": A 2.6-fold increase in blood NAD⁺, reportedly with a p-value under 0.001, and, per the trial's reported statistics, a 95% confidence interval on the NAD⁺ increase that never touched zero. That is what nicotinamide riboside did in a San Antonio trial of older adults with mild cognitive impairment. On the drug's own terms, this is a clean hit.
And then, in the same paper: MoCA scores and every other neurocognitive and psychometric metric "remained stable throughout the study." Cerebral blood flow in the default mode network actually went down.
That split-screen — the biomarker soaring, the patient unchanged — is now the signature of the entire NAD⁺ precursor field. It just happened again.
The biomarker soaring, the patient unchanged — that split-screen is now the signature of the entire NAD⁺ precursor field. It just happened again.
On July 21, 2026, the University of Delaware published a 12-week clinical trial of NR in older adults with mild cognitive impairment. NR "effectively increased blood NAD+ levels." It did not significantly improve cognitive function, cardiovascular function, or whole-brain blood flow.
By this article's count, this is reportedly the third independent lab in roughly three calendar years (2024, 2025, 2026) to publish that same self-contradicting result. Not one disappointing study.
The trial's leader, Christopher Martens, has studied nicotinamide riboside for more than a decade. His summary of where the science stands is not the language of a hype man.
"There's a possibility that this stuff just doesn't work for healthy aging," Martens said. "We may find that it's beneficial to some and neutral for most."
That is a striking thing to hear from someone who has spent ten years chasing this molecule. But it fits the data he's now holding.
Look at the pattern across the three trials, because the pattern is the whole story.
The San Antonio pilot, per GeroScience, published in 2024: twenty MCI subjects, NR titrated to 1 g/day over ten weeks. The target dose was reached — proven by that 2.6-fold NAD⁺ jump. Cognition didn't move. CBF in the default mode network dropped, with the largest change in the left inferior parietal lobe.
The authors were careful to note that the CBF reduction wouldn't have survived correction for multiple comparisons. Fair. But note what that concession means: the one signal that reached nominal significance in the brain pointed the wrong way.
Next, the Massachusetts General Hospital crossover, published in Alzheimer's & Dementia: Translational Research & Clinical Interventions in January 2025. Forty-six participants over 55 with subjective cognitive decline or mild cognitive impairment, randomized to NR-then-placebo or placebo-then-NR, 1 g/day for eight weeks.
The primary endpoint was RBANS — a standard neuropsychological battery. It did not move. Neither did Lumosity gameplay scores, the "novel digital assessment" built in precisely to catch subtle effects conventional tests might miss.
The conclusion, in the authors' own words: NR "is safe but did not alter cognition as measured by conventional or novel digital assessments."
The one positive to emerge from MGH was a 7% reduction in plasma pTau217 on NR versus an 18% rise on placebo — a between-group difference at p = 0.02. That is worth flagging honestly. pTau217 is a credible Alzheimer's biomarker.
But it is a secondary, and functionally an exploratory, biomarker — not a functional outcome. Nobody feels their pTau217. It is another lab number, moved modestly, while the thing that battery was designed to measure — cognition — stayed flat.
So stack them up. Three groups. Three MCI or SCD populations. Three confirmations that NR raises NAD⁺. Three nulls on the outcomes that matter — cognition and blood flow.
The consistency is the point. When one lab returns a null, you blame the sample size or the dose or the duration. When three independent teams, using different designs and different cognitive instruments, all land on the same null while all confirming the drug engaged its target, the parsimonious reading is not "we keep getting unlucky."
The parsimonious reading is the one Martens is willing to say out loud: maybe the target engagement doesn't cash out into benefit.
Here is the translational trap the field walked into. NR was designed to raise blood NAD⁺, on the premise that age-related NAD⁺ decline drives metabolic, cardiovascular, and neurodegenerative dysfunction. Raise the NAD⁺, reverse the dysfunction — that was the theory.
The trials confirm the first half repeatedly and gorgeously. Blood NAD⁺ goes up, multi-fold, with vanishingly small p-values. That is real pharmacology. The drug does what it says.
But hitting the biomarker was only ever proof of engagement — proof the pill reached the machinery it was aimed at. It was never proof of benefit. The field has been treating a surrogate as if it were the endpoint.
And a whole supplement industry has been built on exactly that conflation. Influencers sell NAD⁺ precursors as anti-aging miracles for brain fog, energy, memory, athletic performance, recovery. The pitch rests on the number that reliably moves — and stays silent on the outcomes that reliably don't.
To be precise about what these trials do and do not show: absence of benefit in these endpoints is not evidence of harm. NR was well tolerated across all three trials, with no between-group differences in adverse events. The reduced default-mode CBF was nominal and would not survive correction. Nobody here got hurt.
What is growing is the gap between what's marketed and what's shown. The marketing promises outcomes. The randomized data delivers a biomarker and a shrug.
There are honest caveats, and Martens names them. These were small, short trials — twenty subjects here, forty-six there, eight to twelve weeks. He was genuinely intrigued by a hint of increased blood flow to the hippocampus in his secondary measures, and he wants a bigger, longer, multi-site trial to chase it.
Maybe that trial finds something. Maybe NR is, as he put it, beneficial to some and neutral for most. That's a legitimate hypothesis, and it deserves a properly powered test rather than a marketing campaign.
But notice the asymmetry. The positive signals are always the secondary, the exploratory, the "hint" — hippocampal blood flow here, a pTau217 dip there, an epigenetic-age nudge somewhere else. The primary endpoints, the ones the studies were powered and pre-registered to measure, keep coming back null.
That is what a field looks like when enthusiasm has outrun the randomized evidence.
The discipline is simple and unglamorous. Judge an intervention by what it does to how people function, not by what it does to a number in a tube.
By that standard, the NAD⁺ precursor field has, quietly and repeatedly, produced its own disconfirmation. It proved it can raise NAD⁺. It has not yet proved that raising NAD⁺ does anything a patient can feel.
The interventions that still move real outcomes in older adults remain the free ones — exercise, sleep, not overeating. They are undefeated. NR keeps hitting its biomarker and losing on the scoreboard that counts.