Resveratrol

Resveratrol is the stilbene from grape skin and Japanese knotweed that launched the red-wine-molecule era, and it is sold on two overlapping stories: NF-κB and COX-2 inhibition, the anti-inflammatory case it shares with curcumin, and SIRT1 activation, the longevity case. Almost nobody takes it for a knee. This entry appraises the ageing literature first, because that is what the compound is taken for, then the joint trials, because that is what sets the tier.

The lifespan claim and what tested it

The claim began in mice, and the frame for it appears in one of those papers' own titles. A 2008 study found that resveratrol reproduced dietary-restriction gene-expression patterns and delayed a long list of ageing signs — less vascular inflammation, greater aortic elasticity, better motor coordination, fewer cataracts, preserved bone density — in mice on a standard diet who did not live any longer for it. Its albuminuria finding belongs to the high-calorie arms of the same experiment, which were spared a rise the standard-diet mice never had. Healthspan effects without a lifespan effect: that pattern has held every time somebody has looked harder.

The most rigorous rodent test came from the NIA Interventions Testing Program, which runs the same protocol at three independent sites in genetically heterogeneous mice. Rapamycin extended median lifespan by 10 percent in males and 18 percent in females at every site. Resveratrol, at 300 and 1,200 parts per million in food, had no significant effect on survival in either sex.

The question has since been put to a primate, which is the nearest thing to a direct test anyone is likely to run. Thirty-three male grey mouse lemurs were followed from adulthood to natural death, all on the same 105 kJ a day, with eighteen of them given 200 mg of resveratrol per kilogram daily. At middle age the supplemented animals did better on everything measured: fewer errors in a spatial memory task, less time to find the target, and a higher combined score for motor coordination, endurance and strength, with the gap visible from four years of age. Survival did not follow. Median lifespan was 7.9 years against 6.4, but neither overall nor age-related mortality separated, maximal lifespan was unchanged, and age-related disease accounted for much the same share of deaths in each group. Serial MRI over four years then found grey matter atrophy spread more widely in the supplemented animals than in the controls. The authors' reading is that long-term resveratrol buys health at middle age, does little for longevity, and may cost something in the brain later.

The SIRT1 activation dispute

The mechanism on the label is that resveratrol directly activates SIRT1. This has been argued at the bench for twenty years and both sides are worth stating precisely, because the disagreement is technical rather than rhetorical.

The original 2003 screen used a peptide substrate carrying a fluorescent tag. The most thorough challenge to it came from a pharmaceutical laboratory in 2010, and it tested the claim three ways rather than one. Against a p53-derived peptide with the fluorophore stripped off, and against purified full-length p53 and acetyl-CoA synthetase 1, neither resveratrol nor three synthetic sirtuin activators produced any apparent activation of SIRT1; against a peptide carrying a covalently attached fluorophore, all four did. Nuclear magnetic resonance, surface plasmon resonance and isothermal calorimetry then showed the compounds binding the tagged peptide directly, which supplies a mechanism for the discrepancy rather than merely reporting it. The same paper found that one of those synthetic activators neither lowered plasma glucose nor improved mitochondrial capacity in mice on a high-fat diet, and that all four compounds had off-target activity against receptors, enzymes, transporters and ion channels.

Resveratrol's promiscuity is not in dispute on either side. Reviews sympathetic to the pathway list AMPK, complex III of the mitochondrial electron transport chain, PARP1 and phosphodiesterase among its targets, and treat the search for cleaner activators as the reason the synthetic compounds were made at all.

The counter-argument arrived in 2013. Hydrophobic motifs present in genuine SIRT1 substrates such as PGC-1α and FOXO3a, that group showed, permit activation with no fluorophore involved; a single residue in SIRT1, Glu230, was required for activation by every chemically distinct activator class tested; and in primary cells reconstituted with an activation-defective SIRT1, the metabolic effects of those compounds disappeared. Several authors were affiliated with the company built on sirtuin activators, which is disclosed and belongs in the reading. Two years later a crystal structure resolved three resveratrol molecules bound to SIRT1, two of them bridging the enzyme's N-terminal domain to a coumarin-tagged peptide — a physical explanation for why activation is real, allosteric and substrate-dependent, obtained with a substrate that carries the very tag at issue.

Where that leaves a reader: direct activation of SIRT1 by resveratrol is substrate-dependent and structurally explained, and whether physiological substrates in a living human cell behave like the tagged ones has not been settled. Indirect activation through upstream signalling remains a separate and live possibility.

The human trials outside the joint

Resveratrol has one of the larger supplement trial literatures, and reading it in order is the fastest way to understand the compound.

The metabolic trials. In 2011, eleven obese men given 150 mg a day for thirty days showed lower sleeping and resting metabolic rate, activated AMPK, raised muscle SIRT1 and PGC-1α protein, better mitochondrial respiration, less liver fat, and improvements in glucose, triglycerides, ALT, inflammatory markers, systolic pressure and HOMA index — the calorie restriction mimetic, apparently confirmed. What followed did not reproduce it. Twenty-nine non-obese postmenopausal women on 75 mg for twelve weeks showed no change in body composition, metabolic rate, lipids, inflammatory markers or clamp-measured insulin sensitivity in liver, muscle or fat — and none in AMPK, SIRT1, NAMPT or PPARGC1A in biopsied muscle and fat, meaning the pathway itself did not move. Twenty-four obese men on 1,500 mg a day for four weeks — ten times the 2011 dose — were null on clamp insulin sensitivity, glucose turnover, blood pressure, energy expenditure, ectopic fat and every biomarker measured. Seventy-four men with metabolic syndrome on 1,000 or 150 mg for sixteen weeks were null on inflammation, blood pressure, body composition, liver and muscle lipid and glucose metabolism; the 1,000 mg arm significantly raised total cholesterol, LDL cholesterol and fructosamine against placebo.

What the pooled data say. A meta-analysis of thirty randomised placebo-controlled trials found the split that explains the pattern. Pooled across everyone, glucose fell 5.24 mg/dL and insulin 1.23 mIU/L while HbA1c and HOMA-IR did not move. Stratified by health status, participants with type 2 diabetes improved on all four: glucose by 13.36 mg/dL, insulin by 0.94 mIU/L, HbA1c by 0.22 percentage points and HOMA-IR by 0.83. In participants without diabetes no glycaemic measure changed, which its authors read as resveratrol acting on a pathway that is only disturbed in the first group.

An umbrella review of forty-five systematic reviews then graded the whole field — 129 associations across 68 outcomes, every review rated moderate or high on AMSTAR. Of the thirty-five significant associations, four survived GRADE at high certainty, and all four are metabolic and small: waist circumference down 0.80 cm, total cholesterol in overweight adults down 0.19 mmol/L, and in type 2 diabetes diastolic pressure down 3.55 mmHg and systolic down 7.97. Twenty-three more sat at moderate certainty. The nulls were graded too, and some are held at high certainty: no effect on mean arterial pressure, pulse pressure, alkaline phosphatase, bilirubin or albumin, none on waist circumference or HDL in type 2 diabetes, none on HDL or triglycerides in overweight adults. Dropping the trials at high risk of bias left twenty-seven of the moderate-to-high associations standing and pushed eighteen down.

Cognition. The pooled signal is real and small: working memory improved by a standardised mean difference of 0.20, on moderate certainty, in the umbrella review above. Twenty-six weeks at 200 mg a day (with quercetin) in forty-six overweight adults aged fifty to seventy-five improved word retention over thirty minutes, lowered HbA1c from 5.83 to 5.70 percent and increased hippocampal functional connectivity — a small, publicly funded, unreplicated positive. In 119 patients with mild to moderate Alzheimer disease, fifty-two weeks escalating to 1,000 mg twice daily confirmed that resveratrol crosses into cerebrospinal fluid; CSF and plasma amyloid-β40 declined more on placebo than on drug, and brain volume loss was greater on resveratrol than placebo, a finding the authors could not explain and nobody has followed up.

Bone. In the same metabolic-syndrome cohort that was null on everything metabolic, bone alkaline phosphatase rose 15 to 16 percent above placebo at 1,000 mg a day and lumbar-spine trabecular bone mineral density rose 2.6 percent against placebo. Hip sites did not move. Sixteen weeks is short for bone, and bone is not cartilage — but this is the strongest positive structural finding in the compound's human record.

Mortality. The InCHIANTI cohort measured twenty-four-hour urinary resveratrol metabolites in 783 Italians aged sixty-five and over and followed them nine years. Mortality across quartiles from lowest to highest intake was 34.4, 31.6, 33.5 and 37.4 percent. Resveratrol levels were unrelated to CRP, IL-6, IL-1β, TNF, or cardiovascular disease or cancer. Dietary intake is far below supplement doses, so this tests the red-wine claim rather than the capsule — but it tests it cleanly.

Exercise. One result deserves separate weight for anyone reading this site. Twenty-seven inactive men averaging sixty-five years old did eight weeks of high-intensity training on 250 mg of trans-resveratrol a day or placebo. Maximal oxygen uptake rose 444 ml/min on placebo and 308 on resveratrol — a 45 percent larger gain without the supplement. Mean arterial pressure fell only in the placebo group, and the training-induced improvements in LDL, cholesterol ratio and triglycerides were abolished by resveratrol. It is one trial, publicly funded, not cleanly replicated, and it points the opposite way to everything on the label.

Bioavailability

The classic pharmacokinetic study gave volunteers radiolabelled resveratrol and found a paradox: at least seventy percent is absorbed, but free resveratrol in blood stays at trace levels — under five nanograms per millilitre after 25 mg — because the gut and liver conjugate it almost instantly, by glucuronidation, sulfation and hydrogenation by gut bacteria. The plasma half-life runs four to ten hours whatever the dose, and neither repeated dosing nor dose escalation meaningfully raises the amount of unmetabolised compound in blood. At gram doses the sulfate and glucuronide metabolites reach areas under the curve up to twenty times those of the parent compound; the most abundant of them stays detectable beyond ten hours. That leaves a gap nothing in this literature has closed: the concentrations that produce effects in chondrocyte experiments are micromolar, and human blood levels are nanomolar. Whether the conjugates retain activity, or regenerate free resveratrol inside tissue, is unresolved.

The trial built to settle the knee question

ARTHROL, published in 2024, is the trial the joint field had been waiting for: a phase 3, multicenter, double-blind, placebo-controlled trial in 142 adults with painful knee osteoarthritis, run across three French tertiary care centers and funded by the French Ministry of Health rather than a supplement maker. It used a soluble caplet formulation designed to defeat the absorption problem above — 40 mg twice daily for a week, then 20 mg twice daily, a dose whose plasma peak matched several hundred milligrams of ordinary powder — for six months.

The result: knee pain fell 15.7 points on a 100-point scale with resveratrol and 15.2 points with placebo. The difference — six tenths of a point, with a confidence interval from minus eight to plus seven — excludes any effect a patient could feel. At six months the difference was four tenths of a point in the other direction, and about half of each group met formal responder criteria. The trial stopped at 142 of a planned 164 participants and was powered against an optimistic effect size; even so, the interval it produced leaves no room for a clinically important benefit at this dose and duration.

The positive knee trials, and their pattern

Two lines of positive human evidence exist, and both repay a close look.

An Iraqi trial randomized 110 patients with knee osteoarthritis to resveratrol 500 mg per day or placebo for 90 days, with both arms taking meloxicam, a prescription NSAID. The reported result: total WOMAC scores (0–96, lower is better) fell from about sixty to eight in the resveratrol group while the meloxicam-plus-placebo group moved from about sixty-four to fifty-six. That between-group effect is larger than anything recorded for any osteoarthritis intervention, including NSAIDs themselves, and a placebo arm that scarcely improves on an active NSAID over three months is itself unusual. The same cohort's own follow-up analysis found weak, nonsignificant correlations between its falling inflammatory biomarkers and its improving clinical scores.

More recently, a Sharjah–Peshawar group published three positive trials of resveratrol 500 mg per day in knee osteoarthritis within about a year — 12, 16, and 18 weeks, each around 120 to 140 patients, each single-center, each reporting improved WOMAC scores and raised plasma SIRT1. All three were registered only after completion, and whether they are three independent cohorts cannot be determined from the papers. None of this proves the results wrong; it does mean they cannot currently be checked against the field's one checkable trial.

The animal joint literature is more consistent: pooled across fifteen studies, resveratrol lowered rodent cartilage damage scores, though much of that work used intra-articular injection, a delivery no oral supplement replicates.

The NAD+ precursor pairing

Resveratrol is often taken alongside NMN or nicotinamide riboside on a tidy-sounding rationale: SIRT1 consumes NAD+ as substrate, the precursor restores NAD+, and resveratrol activates the enzyme — supply the fuel and press the accelerator. Two parts are solid, and the accelerator half is the contested one appraised above.

There is now one randomised human test. A six-month trial in ninety people with peripheral artery disease compared nicotinamide riboside at 1,000 mg a day, the same NR plus 125 mg of trans-resveratrol, and placebo. NR alone improved six-minute walk distance by 17.6 metres over placebo at six months. NR plus resveratrol improved it by 3.7 metres and was not significant. At three months both active arms beat placebo, by 22.4 and 20.6 metres. Two things complicate the reading: the combination arm reported nausea or vomiting in 36 percent against 14 percent on NR alone, and only 52 percent of that arm took at least three-quarters of their pills against 75 and 76 percent elsewhere. Resveratrol may have added nothing, or may not have been swallowed. It is the only randomised evidence on the combination in people, and no joint endpoint has been tested in any of it.

The pairing has also been tested directly, in mice, and by a study asking a narrower question. A single oral dose of NMN with resveratrol left NAD+ 1.59-fold higher in heart and 1.72-fold higher in skeletal muscle than the same NMN dose alone, measured across the six hours after dosing in three animals per group at each timepoint; the authors name that sample size as a limit on their own statistics. Six tissues were sampled and none of them was a joint. So resveratrol does change where an NMN dose ends up. Whether that matters to cartilage is untested in any species.

Safety

The trial-level picture is reassuring at supplement doses and unreassuring at the top of the range. ARTHROL recorded five serious adverse events across both arms over six months, none related to the intervention, and the Iraqi cohort monitored liver enzymes, kidney markers, lipids and blood counts at 500 mg per day for 90 days with no adverse signal. In healthy volunteers taking 0.5 to 5 g a day for twenty-nine days, the 2.5 and 5 g doses caused mild to moderate gastrointestinal symptoms, and circulating IGF-1 and IGFBP-3 fell against the volunteers' own pre-dosing levels, most markedly at 2.5 grams — that study had no placebo arm and assigned its doses in sequence. A gram a day for sixteen weeks raised total cholesterol, LDL and fructosamine.

Pooled, the liver picture is quieter than any single trial makes it look. Across thirty-one randomised trials in metabolic disease, alanine aminotransferase did not change and neither did aspartate aminotransferase. The one enzyme that moved was gamma-glutamyl transferase, up 1.76 U/L, and it is the estimate with the least disagreement between trials behind it. Total cholesterol fell 7.65 mg/dL. What that adds up to is a small signal in one liver enzyme rather than the transaminase rise individual trials have occasionally reported.

The interaction question is not theoretical. At a gram a day and above, resveratrol inhibits hepatic cytochrome P450, and exposure to drugs cleared that way — cisapride, cyclosporine, felodipine and midazolam among them — rises accordingly. Anyone on a prescription metabolised by CYP3A4, or on an anticoagulant, should raise it with a prescriber. Nothing concerning has surfaced at the 500 mg daily dose common in supplements.

Where the evidence stands

Resveratrol is not a compound with nothing behind it. It has real biology, one of the best-documented mechanism disputes in pharmacology, a bone finding worth following, a clean cognitive signal in one small trial, and a consistent metabolic effect in people who already have metabolic disease. It also has a bioavailability ceiling that no oral formulation has convincingly cleared, a set of large independent nulls in healthy people, and a randomised result suggesting it can work against exercise adaptation in older men.

For cartilage specifically, the question has been asked properly once and the answer was no. What would move that is a well-powered, independently funded, prospectively registered trial of supplement-typical resveratrol in knee osteoarthritis finding what the single-center trials report — or a trial of the intra-articular route the animal work actually used, which nobody has run in people.