The number is p = 5.48e−14. That is how confidently the RENEWAL-NAD+ trial can say that its intracellular NAD signal is real: a 53% rise in whole-blood icNAD versus placebo at Day 6, with an effect size of Hedges' g = 3.66. In a field where interventions routinely produce effect sizes near zero, that is a mountain.
And here is the twist that makes it worth arguing about: the same trial reports that plasma NAD — reportedly the number consumer supplements have been selling — did not move at all. Circulating NAD, p = 0.60. Flat.
The trial declares this a feature, not a failure. It titled itself around the finding: intracellular NAD rises "without elevating plasma NAD."
To understand why that framing is clever, look at what landed weeks earlier. In June 2026, a Nature Metabolism study out of Vrije Universiteit Amsterdam measured whole-blood NAD across more than 300 people — young and old, elite athletes, frail elders on diet and exercise programs, people taking nicotinamide riboside.
Their conclusion, from lead authors Maria Trętowicz and Riekelt Houtkooper: whole-blood NAD "cannot be used to track the aging process in humans." It doesn't fall with age. It barely moves with lifestyle. It sits inside "a narrow physiological range" that seems actively buffered.
So the field spent a month declaring plasma NAD a dead-end biomarker — and then a company-funded RCT arrived claiming victory precisely by not touching it.
That is the paradox worth sitting with. The molecule is being abandoned and rehabilitated at the same time, and the only thing that determines which is which compartment you decide to sample.
The molecule is being abandoned and rehabilitated at the same time, and the only thing that determines which is which compartment you decide to sample.
Start with what the trial gets right, because it gets the science right. The core critique is legitimate: plasma NAD is the wrong yardstick. Trętowicz and Houtkooper showed it doesn't reflect tissue-level changes — "whole-blood NAD+ levels do not reflect tissue-specific changes, such as those observed in aging muscle."
Mark Tarnopolsky, the McMaster mitochondrial physiologist, put the mechanistic version bluntly. The biologically relevant quantity isn't the concentration of circulating NAD but the redox state inside cells and organelles — the mitochondrial NAD+/NADH ratio that actually governs how efficiently a cell makes energy.
Those measurements are hard to get in humans. "That convenience," he said of blood sampling, "has been overinterpreted."
So a trial that deliberately measures the intracellular compartment is measuring a better thing. The 53% icNAD rise is a real signal in a real compartment, and the catabolite data reinforce it: 1-methyl-nicotinamide (p = 5.39e−13) and 2PY (p = 2.95e−16) both climbed sharply, which is what you'd expect if the delivered NAD were actually being metabolized downstream rather than just sitting inert.
That is genuine pharmacodynamic evidence. The compound gets in, and something happens to it. On the narrow question of cellular delivery, the trial clears its bar.
But here is where the argument turns, and it turns on the exact word Tarnopolsky chose.
He said increases in NAD metabolites "should be regarded as pharmacodynamic target engagement rather than evidence of improved mitochondrial function, biological age or healthspan." Target engagement. The drug reaches its target.
That is precisely — and only — what the RENEWAL trial demonstrates. It has moved the biomarker from "wrong compartment" to "right compartment," but it has not moved from "biomarker" to "benefit." A better biomarker is not proof it works.
And this is where the same argument that rescues intracellular NAD as a measurement becomes a tool that can be turned against the reader. A company can now say: plasma NAD was always the wrong yardstick, we hit the right one, therefore we have "target engagement." All true. All insufficient.
Because the honest reading of this trial sits in one sentence buried in its own abstract: "no secondary clinical, vital-sign, wellbeing, or wearable-derived endpoint survived multiplicity correction."
Read that slowly. The trial measured secondary outcomes — clinical labs, vitals, self-reported wellbeing, and data from a wrist-worn wearable tracking sleep and activity.
None of it held up once you correct for testing many things at once. Not one functional endpoint touching how a person feels, sleeps, moves, or metabolizes survived the statistical honesty of multiplicity correction.
So what the trial actually establishes is this: a molecule went into cells, and its breakdown products showed up in plasma. That is the entire durable finding. Everything downstream — everything a buyer of a longevity supplement actually cares about — did not reach significance.
Now the disclosures, which are not incidental. Read the competing-interests statement on the preprint.
The CEO of the sponsor, BioNADRx (doing business as Bryleos), is an author and holds stock options and shares. The lead author and a co-author hold stock options and took consultancy fees during manuscript preparation. Another author is a company advisor with options. Others are investors holding shares. The assay was reportedly run by a lab whose founders include two of the authors.
Every author has financial exposure to the result. That doesn't make the p-value fake — 5.48e−14 is 5.48e−14. But it shapes which true sentence gets promoted to the headline and which true sentence gets buried in a subordinate clause.
The headline sentence is the 53% rise. The buried sentence is that nothing functional survived. Both are honest. Only one sells product.
There is also the matter of provenance, worth noting without overreaching. The trial ran in early 2022; it was registered on ClinicalTrials.gov retrospectively, in January 2026, years after completion, and the product was "commercially available prior to study initiation." The science can still be sound. But a trial registered after the fact, on a product already on shelves, by a team that owns the product, is a document to read with both eyes open.
So here is the rule, and it cuts two ways.
For trial designers: choosing the better compartment raises your evidentiary bar, it does not lower it. If you're going to argue that plasma NAD was always meaningless and intracellular is what matters, then intracellular delivery is now merely your entry ticket — the thing you must show before you're allowed to ask the real question. Target engagement earns you the right to run the functional trial. It is not a substitute for one.
For consumers: when a company tells you its NAD product is different because it raises NAD inside your cells while others only raise it in your blood, that claim may well be true — and it still tells you nothing about whether the product does anything for you. "Gets into cells" is a chemistry result. "Makes you healthier" is a clinical one. This trial delivered the first and, by its own admission, not the second.
The Amsterdam study did the field a real service by killing a bad biomarker. The danger is that killing the bad biomarker becomes a license to declare victory on a better one — as if the compartment were ever the point.
The compartment was never the point. Healthspan was. And on that question, a 53% intracellular rise with p = 5.48e−14, real catabolite flux, and not a single surviving functional endpoint says exactly one thing, clearly: the drug arrives. Whether it helps remains unmeasured.
The evidence here is a single Phase 0/1b trial, n = 50 in the primary analysis, five days of dosing, industry-funded, retrospectively registered, with target engagement as its only durable result. That is the level. Treat any claim beyond it as a hypothesis wearing a p-value.