Urolithin A: what the human trials show, who paid for them and what the newest independent trial found

Urolithin A is a compound that gut bacteria make from ellagitannins, which are found in pomegranate, walnuts and some berries. It is sold as a supplement, including under the Mitopure brand. In October 2026, JAMA Network Open published a randomised trial of 1,000 mg a day in adults with obesity. It found no improvement in its primary measure of blood-vessel function.

The wider picture is similar. A 2026 systematic review found five small randomised trials of urolithin A on muscle outcomes, with only one outcome that could be pooled, and low certainty in that result. Four of the seven human trials described here were paid for by Amazentis SA, the company that developed the ingredient.

The short answer: urolithin A was tolerated in the trials described here, which ran from four weeks to four months, and some biomarker and secondary results moved. The two largest trials, both funded by Amazentis, missed the primary endpoints meant to show a walking or exercise-performance benefit. Smaller trials reported better strength, endurance or performance measures, but they are small, and the newest trial, funded by the American Heart Association, found no effect on its blood-vessel primary measure.

What urolithin A is, and why not everyone makes it

Urolithin A is not found in the ellagitannin-rich foods it comes from, such as pomegranate, berries and walnuts. Gut bacteria produce it, and not everyone's microbiome does. In an Amazentis-funded study of 100 healthy adults, 12% had detectable urolithin A at baseline, and after pomegranate juice about 40% converted its precursors into it. A food product containing 500 mg gave more than six times the exposure of the juice. That difference in conversion is the reason supplement makers give for taking it directly.

The trials, and who paid for them

• Older adults, 4 months (Liu 2022, JAMA Network Open): 66 adults aged 65 to 90 took 1,000 mg a day or placebo. The two primary endpoints, the change in six-minute walk distance and maximal ATP production in a hand muscle, showed no significant difference from placebo. Muscle endurance, a secondary endpoint, improved more with urolithin A at 2 months, with no significant effect at 4 months, and several blood markers shifted. Amazentis funded the trial, the paper states that the funder had a role in its design, conduct, analysis and manuscript, and Amazentis employees are among the authors.
• Healthy, overweight, middle-aged adults, 4 months (Singh 2022, Cell Reports Medicine): 88 adults took 500 mg, 1,000 mg or placebo. The primary endpoint, peak power output, did not differ significantly from placebo. Hamstring average peak torque rose 12% on 500 mg (P = 0.027 versus placebo) and 9.8% on 1,000 mg (P = 0.029), while placebo fell 9.8%. At 1,000 mg, peak oxygen consumption (P = 0.058) and six-minute walk distance (P = 0.098) favoured urolithin A without reaching significance versus placebo. The authors call the 33.43 m walk-distance gain within that group clinically meaningful. Amazentis funded the study, and several authors are its employees or board members.
• Resistance-trained men, 8 weeks (Zhao 2024, Journal of the International Society of Sports Nutrition): 20 men took 1 g a day or placebo. Maximum voluntary isometric contraction (P = 0.048) and repetitions to failure (P = 0.011) were higher than on placebo, while bench-press and squat one-repetition maximum did not differ significantly (P = 0.462 and P = 0.710). The funder was a Jiangsu Province education grant, not Amazentis.
• Highly trained male distance runners, 4 weeks (Whitfield 2025, Sports Medicine): 42 runners took 1,000 mg a day or placebo during an altitude camp. In a subset of 11 per group, the 3,000 m time trial, a co-primary endpoint, did not improve significantly in either group. Creatine kinase, a muscle-damage marker and the other co-primary endpoint, showed a time-by-treatment interaction after the time trial (P = 0.0489). After downhill running in all 42 runners, creatine kinase showed no treatment effect (P = 0.137) and no time-by-treatment interaction (P = 0.891). Maximal aerobic capacity did not differ between groups (time-by-treatment P = 0.138). Amazentis funded the trial, three authors are its employees, and the first author has received travel support from Amazentis.
• Academy soccer players, 6 weeks (Monsalve Acevedo 2025, Frontiers in Nutrition): 20 players with a mean age of 17.5, including players under 18, took 1,000 mg a day or placebo with a post-training drink. The trial was single-blind: the players were blinded and the investigators were not. The primary outcome, the Yo-Yo Intermittent Recovery Test Level 1 (an aerobic endurance test), was 239 m better than placebo (95% confidence interval 20 to 454 m, P = 0.048). The authors call it a pilot and caution against over-interpretation, because adherence, training load and baseline values differed between groups. Two authors are funded by a Memphasys Ltd grant to their university, a third author is a Memphasys consultant and scientific director, and Timeline Nutrition, the source of the Mitopure powder, paid one author's conference travel and accommodation. The paper says those parties had no role in design, conduct or analysis.
• Healthy adults aged 45 to 70, 4 weeks (Denk 2025, Nature Aging): 50 adults took 1,000 mg a day or placebo. The reported primary immune-cell result was 0.50 percentage points more naive-like CD8 T cells (95% confidence interval 0.16 to 0.83, P = 0.0437), without adjustment for multiple comparisons. A co-primary metabolic result, CD8 fatty-acid oxidation capacity, differed by 14.72 percentage points (95% confidence interval 6.46 to 22.99, P = 0.0061). Amazentis funded the trial, supplied the product and co-designed it, and had no role in data collection, analysis or preparation of the manuscript. Three authors are Amazentis employees and one is its president and a board delegate. The senior author has received consulting fees from Amazentis, and the first author has received travel compensation from Amazentis.
• Adults with obesity, 4 weeks (da Costa Pinaffi-Langley 2026, JAMA Network Open): 50 adults aged 40 to 64 with a BMI of 30 or more took 1,000 mg a day or placebo. The primary outcome was change in brachial artery flow-mediated dilation. It did not differ between groups (adjusted difference 1.74 percentage points, 95% confidence interval -1.33 to 4.81, P = .33), and neither did the three other measures of peripheral blood-vessel function. The funders were the American Heart Association and the Reynolds Foundation, and the paper reports no conflicts of interest. The authors also report a larger task-evoked blood-flow response in one frontal region of the brain during a working-memory task (measured by near-infrared spectroscopy), with unchanged task performance, and describe it as warranting further investigation.

A 2026 systematic review combined the five randomised muscle trials it found, with 236 participants: Liu 2022, Singh 2022, Zhao 2024, Whitfield 2025 and Monsalve Acevedo 2025. Only the six-minute walk test could be pooled, from two trials. After the two urolithin A arms in Singh were combined, the difference versus placebo was +17.03 m with a 95% confidence interval of -5.33 to 39.40 m (P = 0.135), and the reviewers rated the certainty of that evidence as low. They describe the other muscle and mitochondrial findings as exploratory signals, not reproducible effects, and ask for larger, longer, standardised trials.

How to read this

Two things are true at once. Urolithin A has been studied in several small human trials, with short-term tolerability and some biomarker changes. The two largest trials, Singh 2022 (88 adults) and Liu 2022 (66 adults), were both sponsored by Amazentis. Singh missed its peak-power primary endpoint. Liu missed both of its primary endpoints, the six-minute walk and hand-muscle ATP production. Smaller trials, including two not funded by Amazentis, reported better strength, endurance or performance measures, and the Amazentis-funded immune trial reported significant primary results without adjusting for multiple comparisons. The American Heart Association-funded trial of blood-vessel function found nothing on its primary measure.

That does not show urolithin A does nothing. The trials are small, the longest ran four months, and the populations differ widely, from academy soccer players with a mean age of about 17.5 and competitive runners with a mean age of 27, to adults with a mean age of about 72. It does mean that the strong claims about mitochondria, muscle and ageing outrun the human evidence, and that an honest label would say the evidence is early. Longer trials that are not company-funded, with pre-registered primary endpoints, are what would change that.

Safety

Liu 2022 reported no statistical difference in adverse events between urolithin A and placebo. In the 2026 JAMA Network Open trial, 22 of 50 participants reported 43 events in total, and headache was the most common (6 participants on placebo and 3 on urolithin A). All of these trials ran for four months or less. Most enrolled adults. The academy-soccer trial had a mean age of 17.5 and included players under 18. Nobody can say from these trials what long-term daily use does.

South African rules

SAHPRA's public information on peptide products states that, under the Medicines and Related Substances Act 101 of 1965, any product intended to treat, prevent or alter bodily functions must be registered with SAHPRA before it can be sold in South Africa. The page gives weight loss, muscle growth and anti-ageing as examples of how unregistered peptide products are promoted. It does not mention urolithin A, and it does not say that every food on a shelf needs a SAHPRA registration.

Side effects can be reported to SAHPRA at adr@sahpra.org.za, the address on that peptide page.

FAQ

Does urolithin A improve muscle strength?

The evidence is mixed and early. In a 4-month Amazentis-funded trial, hamstring torque rose about 10% to 12% depending on dose, the primary peak-power endpoint was missed, and walking distance at 1,000 mg did not differ significantly from placebo. A 4-month trial in older adults missed both primary endpoints and found better muscle endurance at 2 months as a secondary result. A 2026 systematic review of five trials found a pooled six-minute walk difference of +17.03 m that was not statistically significant, with low certainty.

Did the American Heart Association-funded trial find any benefit?

On its primary outcome, no. The 2026 trial in 50 adults with obesity found no difference in flow-mediated dilation after four weeks of 1,000 mg a day. It reported a larger task-evoked blood-flow response in one frontal region of the brain during a working-memory task and no change in task performance, and the authors call for further research.

Can I get urolithin A from pomegranate?

Only some people can. In an Amazentis-funded study, about 40% of adults converted pomegranate juice into urolithin A, and only 12% had detectable levels before the juice. A food product containing 500 mg of urolithin A produced more than six times the exposure of the juice.

Who funded the trials?

Amazentis SA, which developed the ingredient, funded Liu 2022, Singh 2022, Denk 2025 and Whitfield 2025, and its employees are among the authors of each paper. In Denk 2025 the company also co-designed the trial. The first author has received travel compensation from Amazentis, and the senior author has received consulting fees from Amazentis. Zhao 2024 was funded by a Jiangsu Province education grant, not by Amazentis. Monsalve Acevedo 2025 was not funded by Amazentis: two authors are supported by a Memphasys Ltd grant to their university, and one author is a Memphasys consultant and scientific director. The 2026 blood-vessel trial was funded by the American Heart Association and the Reynolds Foundation.

Where can I report a suspicious product in South Africa?

SAHPRA's 11 November 2025 warning asks the public to report suspicious medicine sales or false claims of SAHPRA approval on 0800 204 307. Side effects can be reported to adr@sahpra.org.za.