UROLITHIN A: THREE TRIALS, THREE MISSED PRIMARY ENDPOINTS

Three randomized trials have tested urolithin A in humans for muscle and endurance outcomes — in middle-aged adults, in adults 65 to 90, and in elite distance runners. All three missed their primary endpoint. All three found something in their secondary measures. Here is what each one actually measured.

What urolithin A is

Urolithin A is not a compound found in food. It's a metabolite your gut bacteria produce after you eat ellagitannin-rich foods — pomegranate, walnuts, certain berries. The compound reaching your bloodstream is the bacterial product, not the thing you ate.

That matters, because the conversion is not universal. Work on urolithin "metabotypes" has found that people fall into groups by which urolithins their microbiome produces, including a group producing essentially none. One small bioavailability study — Araujo and colleagues, Food Research International, 2021, in 16 healthy and 7 metabolic-syndrome subjects consuming an ellagitannin-containing fruit juice — found 19 and 29 percent of those two groups respectively in the non-producing metabotype. That's a small sample in one population and shouldn't be read as a general prevalence figure, but the underlying point holds: eating the food does not guarantee producing the metabolite.

The supplements in these trials sidestep that entirely by delivering synthesized urolithin A directly. The proposed mechanism is mitophagy — clearing damaged mitochondria so they can be replaced — which is why the outcomes tested are muscle and endurance ones.

The middle-aged trial

Singh and colleagues, Cell Reports Medicine, 2022 (NCT03464500). A randomized, placebo-controlled trial of urolithin A at two doses for four months in middle-aged adults. The study population was untrained adults aged 40 to 64 who were overweight and had low physical endurance.

The primary endpoint was peak power output, and it did not improve significantly. The authors state that directly.

What did move: muscle strength improved roughly 12 percent. The authors describe "clinically meaningful improvements" on aerobic endurance (peak oxygen consumption) and physical performance (six-minute walk test). Plasma acylcarnitines and C-reactive protein were significantly lower, which they read as higher mitochondrial efficiency and reduced inflammation. In skeletal muscle, expression of proteins linked to mitophagy and mitochondrial metabolism increased significantly.

The competing-interest statement is part of the record: six of the nine authors were employees of Amazentis SA, two were board members, and two sat on its scientific advisory board. Amazentis sponsored the study and sells the product.

The older-adult trial

Liu and colleagues, JAMA Network Open, 2022 (NCT03283462). A double-blind, placebo-controlled randomized trial in adults aged 65 to 90, run at a medical center and a cancer research center in Seattle between March 2018 and July 2020. Sixty-six participants, mean age 71.7, randomized to 1,000 mg urolithin A daily or placebo for four months. Analysis was intention-to-treat.

The co-primary endpoints were change in six-minute walk distance and change in maximal ATP production in hand skeletal muscle, measured by magnetic resonance spectroscopy.

Neither reached significance. Six-minute walk distance rose 60.8 m in the urolithin A group and 42.5 m in placebo — a difference, but not a significant one. Maximal ATP production in the hand muscle changed 0.07 mM/s versus 0.06; in the tibialis anterior it was −0.03 versus 0.03.

The secondary endpoint did move. Muscle endurance — contractions until fatigue — improved significantly in both tested muscles at two months. In the first dorsal interosseus, 95.3 additional contractions versus 11.6 for placebo; in the tibialis anterior, 41.4 versus 5.7. Plasma acylcarnitines, ceramides and C-reactive protein were lower at four months. No statistical difference in adverse events.

The authors' own conclusion is worth quoting rather than paraphrasing: supplementation "was safe and well tolerated," the primary endpoints were not significant, long-term supplementation "was beneficial for muscle endurance and plasma biomarkers, suggesting that urolithin A may counteract age-associated muscle decline; however, future work is needed to confirm this finding."

Composition of the sample: 75.8 percent women, and all participants were White individuals.

The trained-athlete trial

Whitfield and colleagues, Sports Medicine, 2025 (NCT04783207). This is the one closest to an endurance-athlete question, and the answer it produced is the most specific of the three. Forty-two competitive male distance runners, mean age 27.2, mean VO2max 66.4 mL/kg/min, randomized to 1,000 mg/day urolithin A or placebo for four weeks during an altitude training camp at roughly 1,700–2,200 m.

Running performance did not improve. In the 3,000 m time trial subset, neither group improved significantly (urolithin A p = 0.116, placebo p = 0.771).

Two recovery measures did move. Ratings of perceived exertion were significantly lower on urolithin A (p = 0.02). Creatine kinase — an indirect marker of muscle damage — was substantially lower by total area under the curve after the time trial (p < 0.0001).

Aerobic capacity is the finding most likely to get misreported, so the numbers matter. There was no statistically significant time × treatment interaction for VO2max (p = 0.138). Within the urolithin A group alone, VO2max rose 5.4 percent (66.4 to 70.0, p = 0.009); within placebo it rose 3.6 percent (66.4 to 68.7, p = 0.098). Both groups were at altitude camp, which is itself a VO2max stimulus. A within-group change is not a between-group effect, and the between-group comparison did not reach significance.

Muscle biopsies in a subset showed proteomic upregulation of mitochondria-associated pathways and downregulation of inflammatory pathways. Markers of mitophagy showed a medium effect size but were not statistically significant, and mitochondrial function itself did not change.

The authors' conclusion: four weeks of supplementation "facilitates recovery by downregulating inflammatory pathways and indirect markers of muscle damage. However, despite a reduction in rating of exertion and increased aerobic capacity, UA supplementation did not further enhance performance in highly trained male endurance athletes."

This trial was supported by research funding from Amazentis SA, and three of its authors are Amazentis employees, per the paper's declarations.

What these three studies have and haven't established

The consistent thread across all three is biochemical rather than performance-based: acylcarnitines down, C-reactive protein down, mitochondria-related protein expression or pathway activity up. That showed up in untrained middle-aged adults, in adults over 65, and in elite runners' muscle biopsies.

The performance picture is less uniform. Strength moved ~12 percent in the middle-aged untrained group. Muscle endurance moved in the over-65 group. Neither moved the headline outcome its own trial had pre-specified. In trained runners, race performance didn't move at all, while perceived exertion and muscle damage markers did.

What no trial here measured:

• Trained athletes over 60. The two aging trials used untrained or general-population participants; the athlete trial had a mean age of 27. I could not find a randomized trial of urolithin A in trained masters athletes.
• Hard outcomes. No falls, fractures, disability, or mortality endpoints. These are strength, endurance, walk-distance and biomarker studies over four months or less.
• Independence from the manufacturer. All three trials were sponsored by, funded by, or staffed with employees of Amazentis SA, as stated in each paper's own competing-interest section. That doesn't invalidate a randomized, placebo-controlled design — but it's part of the record, and the papers put it there themselves.

The protocol, as studied

If you're reading these papers to see what was actually administered:

• The dose in both the older-adult and the athlete trial was 1,000 mg/day. The middle-aged trial used two dose levels.
• Duration was four months in the two aging trials, four weeks in the athlete trial.
• Tolerability was reported as good — the JAMA Network Open trial found no statistical difference in adverse events versus placebo.
• Pomegranate and walnuts supply the precursor, not the compound. Whether you convert it depends on your own gut microbiome, which no food label can tell you.

That's what the evidence base looks like as of now: three randomized trials, three missed primary endpoints, a consistent biomarker signal, real but scattered secondary findings, and no data yet in the population reading this.