Vitamin B3: niacin and NMN, what their promises are worth

In 2011 a trial gave niacin, vitamin B3 at drug doses, to 1,718 patients already on a statin, against placebo. HDL rose from 35 to 42 mg/dL. Triglycerides fell from 164 to 122. LDL from 74 to 62. The entire lipid panel improved, exactly as expected.

Those three numbers deserve translating. LDL is the fraction of cholesterol that builds up as deposits in the arteries, and those deposits lead to heart attacks and strokes. HDL is the one that carries cholesterol and those deposits back to the liver to be flushed out. Triglycerides are the most common type of fat in the body, the one it makes out of surplus calories and stores in fat cells.

Lower the LDL, lower the triglycerides, raise the HDL: that is word for word what a cholesterol treatment is asked to do. Niacin did all three.

16.4% of events in the niacin group, 16.2% in the placebo group. The trial was stopped early, for lack of efficacy.

What vitamin B3 actually does

It is the starting point for two molecules, NAD and NADP, on which a reference review says a vast array of processes and enzymes depend, involved in every aspect of cellular function, in the body as in the brain: oxidative reactions, antioxidant protection, DNA repair, cellular signalling.

Its deficiency has a name, pellagra, and it is severe: skin damage, hair loss, muscle weakness, burning in the extremities, altered gait, diarrhoea. It also has a psychiatric side, from depression and anxiety through to memory loss, paranoia and psychotic symptoms. As with B12, a B vitamin deficiency can wear the mask of a cognitive disorder.

None of what follows disputes that. What follows concerns the next step, the one where the vitamin is given to people who are not short of it.

Round one: niacin against heart attacks

The hypothesis held up. If HDL clears cholesterol out of the arteries, having more of it should protect, particularly in patients already on a statin, where risk remains that the statin does not cover.

After the 2011 failure, the next trial was built to leave nothing to chance. HPS2-THRIVE enrolled 25,673 adults with vascular disease, all on a statin, randomly assigned to 2 g of extended-release niacin or placebo, over a median of 3.9 years.

The markers moved again: LDL lowered by 10 mg/dL on average, HDL raised by 6. The rate of major vascular events stayed put: 13.2% against 13.7%, rate ratio 0.96, with no statistical significance.

And that is not all: the niacin group showed an excess of serious disturbances of diabetes control, of 3.7 percentage points in absolute terms. The authors conclude that adding niacin did not significantly reduce the risk of major vascular events but did increase the risk of serious adverse events.

Two trials, 29,087 people in total, every number on the panel moved in the right direction, no benefit, and a safety signal on the second. One precision matters though: these trials used 1.5 to 2 g per day under medical supervision. That is a drug, not a supplement.

Round two: NAD+ and the anti-ageing promise

The current reasoning has three steps. NAD is essential. Its level is said to fall with age. So raising it should help. The authors of the trial below state the limit of that second step themselves: the decline is reported in preclinical models, and human data are sparse.

The trial gave 1 g of nicotinamide riboside per day to 12 aged men for 21 days, crossover and against placebo, with muscle biopsies to check. Muscle NAD+ did rise, which settles the bioavailability question. Mitochondrial bioenergetics, meaning the mitochondria's ability to produce energy, did not change. Inflammatory markers in blood, for their part, went down.

On the NMN side, a 2026 meta-analysis pooled 15 trials, at doses from 250 to 2,000 mg per day and over durations from 14 days to 24 weeks. The conclusion: good tolerability, no rise in liver enzymes, and no effect on weight, fasting glucose, HbA1c, lipid profile or systolic blood pressure. Diastolic pressure falls slightly, and insulin resistance shows a trend that does not reach significance.

Same shape, different decade. The marker rises, function does not follow. With one difference that carries its own irony: in 2011 the product sold was a drug meant to act on a marker, today the product sold is the marker itself.

What would settle it is no mystery either: longer trials, in larger groups, with endpoints that describe a person rather than a molecule. Grip strength, walking distance, events that happened. As long as the primary endpoint stays the NAD+ level, what is being measured is the product's bioavailability, not its usefulness.

And the brain?

This is the sector's best-selling promise and its least supported one: more NAD, so more cellular energy, so mental clarity and focus.

One thing has been demonstrated, though, and it is stronger than people assume. In a phase I trial, 30 newly diagnosed, treatment-naive Parkinson's patients received 1,000 mg of nicotinamide riboside or placebo for 30 days, with NAD measured directly in the brain by spectroscopy. Cerebral NAD rose, significantly but variably from one participant to another.

In those whose cerebral NAD rose, brain metabolism was altered on imaging, with mild clinical improvement. The authors see a lead there, to be confirmed in larger trials. A signal, in thirty people, in one specific disease, at the earliest stage of research.

What it establishes matters: the molecule crosses over and moves the marker. What it does not establish matters as much. I found no trial measuring cognition in healthy adults on a NAD+ precursor. The only solid link between B3 and the brain remains the deficiency one.

Which B3 to choose, and which to avoid

EFSA sets two very different upper limits under a single vitamin name. For nicotinic acid, 10 mg per day, on the flushing endpoint. For nicotinamide, 900 mg per day, on liver toxicity. A ratio of 90 between two forms of the same vitamin.

On labels this vitamin goes by at least four names. Nicotinic acid is also called niacin: that is the form that causes flushing. Nicotinamide is also called niacinamide, and does not. Then come nicotinamide riboside and NMN, which arrived long after those opinions. And there is a separate category, sold as «no-flush».

A word on that famous flush, since it is what organises the entire shelf. Nicotinic acid triggers redness and a sensation of heat rising in the face, the neck and the upper chest, often with tingling. It is not an allergy: the mechanism runs through prostaglandins released by immune cells in the skin, in response to a receptor specific to niacin.

It is dramatic, benign, and it does not last. In a tolerance study run over five days of treatment, the mediator responsible became undetectable in most participants, while blood niacin levels did not fall. In other words the body stops making the reaction, not absorbing the molecule.

That leaves the «no-flush» category, which deserves a stop. A team bought over-the-counter preparations across the three categories and measured their actual free nicotinic acid content by chromatography.

Three times the price for no active molecule. The authors conclude that these preparations should not be used to treat dyslipidaemia. As with magnesium and its salts, the name on the box and its contents are two different pieces of information.

What this article does not say

• That vitamin B3 is useless. Pellagra exists, and it is severe. Being short of a vitamin and taking extra of it are unrelated situations.
• That NAD+ is a dead end. The available trials run from 14 days to 24 weeks in small groups. An absence of demonstrated benefit is not a demonstrated absence of benefit.
• That the niacin results apply to supplements. The trials used 1.5 to 2 g per day, under medical supervision, in patients already on a statin.
• That the second trial condemns niacin on its own. It tested extended-release niacin combined with a second molecule, laropiprant, and it is that combination its adverse-event conclusion covers.
• That this replaces a doctor. High-dose niacin remains a treatment, and stopping it or adding it is discussed with whoever prescribed it.

« A marker on the move is a promise. A trial is an answer. »

What Helix does

The mistake described here is not a mistake about a molecule, it is a mistake in reasoning, and it repeats across every subject. A number improves, the person is assumed to be better off, and nobody checks the step in the middle.

That is why the engine shows, for every effect, what established it: a clinical trial, an intermediate marker, or an assumed mechanism. For B3, the distinction separates 29,087 people from a conclusion they did not validate.