Urolithin A for muscle: five small trials, one inconclusive result

Urolithin A, a compound the gut makes from the ellagitannins in pomegranates and nuts and now sold as a supplement, has been marketed on the promise that it improves muscle and mitochondrial health. The first systematic review to pool the randomised human trials, published in August 2026, found the pooled evidence rests on five small, short studies, with only a single outcome poolable from two of them — and that result did not reach statistical significance [s1].

That is the finding to carry away. Urolithin A is a genuine biological molecule with a plausible mechanism, but the claim that supplementing it measurably improves muscle function in people is, on the current randomised evidence, unproven rather than established.

What was pooled, and what was not

The review searched PubMed, Embase, Web of Science and Scopus for randomised controlled trials of oral urolithin A against placebo published up to December 2025, and identified five, together enrolling 236 participants [s1]. Risk of bias was rated with the Cochrane RoB 2 tool and the certainty of the pooled outcome with GRADE [s1].

Crucially, the trials were too different from one another to combine on most measures. Quantitative meta-analysis was feasible for only one outcome, the six-minute walk test (6MWT), and only two trials reported it in a poolable form [s1]. Everything else — muscle strength, endurance, aerobic capacity, and biochemical and mitochondrial biomarkers — varied so much across populations, doses and measurement methods that the authors declined to pool it, reporting those results narratively as "exploratory signals rather than reproducible effects" [s1].

That distinction matters. A supplement's marketing typically leans on the exploratory signals; the meta-analysis is explicit that they are hypothesis-generating, not evidence of efficacy [s1].

The one number that could be pooled

Combining the 500 mg and 1,000 mg urolithin A arms of the trial by Singh and colleagues against their shared placebo group, the pooled mean difference on the six-minute walk test was +17.03 metres, with a 95% confidence interval running from −5.33 to 39.40 metres (p = 0.135) [s1]. Because the interval crosses zero, the result is compatible with no benefit at all. Heterogeneity between the two trials was nil (I² = 0%), and a sensitivity analysis that split the placebo group differently gave a comparable estimate of +18.80 metres (95% CI −3.24 to 40.85; p = 0.095) [s1].

The certainty of this single pooled outcome was rated low on GRADE [s1]. A low GRADE rating means the true effect could be substantially different from the estimate — the opposite of a settled finding.

A directionally favourable but non-significant walk-test difference from two trials is a thin foundation for a health claim. It is the kind of result that later, larger trials frequently flatten toward zero.

Why the biology outruns the trials

A companion review in the same journal maps why urolithin A has been hard to translate from the laboratory to the clinic [s2]. Not everyone's gut bacteria can make urolithin A from dietary precursors at all — people fall into distinct "metabotypes" — so the internal exposure achieved from either food or a supplement varies widely between individuals [s2]. The review flags responder heterogeneity, uncertainty over which molecular form (free versus conjugated) is actually active in the body, and inconsistent trial design as the central reasons experimental promise has not yet produced dependable clinical outcomes [s2].

Those are not reasons to dismiss the molecule. They are reasons the human evidence is immature: the field has not yet standardised who to give it to, how much, or what to measure. That is precisely the gap the supplement market fills with confident language the trials do not support.

How this sits with the wider longevity market

Urolithin A joins a lengthening list of compounds sold for "healthy ageing" whose human trial base is smaller than the marketing implies. The PEARL trial of low-dose rapamycin missed its own primary endpoint; senolytic drugs remain in early randomised testing; the case for metformin as a geroprotector still awaits its definitive trial; and taurine failed to replicate as an ageing biomarker. The pattern is consistent: strong preclinical or mechanistic data, weak or inconclusive randomised human data, and a supplement or off-label market that runs ahead of both.

Urolithin A is also a reminder that "gut-derived" and "natural" are not evidence. The same microbiome science that makes the compound interesting is what makes its effects so variable between people.

What would change the picture

The review's own bottom line is that larger, longer and methodologically standardised trials in better-defined populations are required before firm clinical recommendations can be made [s1]. What that means concretely: trials that enrol people by metabotype or baseline function, use consistent, meaningful endpoints, and run long enough to detect a real change in muscle performance rather than a biomarker blip.

Until then, the honest summary is the meta-analysis's own. Urolithin A supplementation has a plausible mechanism, a benign short-term safety record in small studies, and a single poolable efficacy outcome that did not reach significance. Anyone told it is proven to build or preserve muscle is being sold ahead of the evidence.

Nothing here is a recommendation to take or avoid any supplement. Supplements are not assessed by regulators the way medicines are, and the contents and doses of urolithin A products are not standardised.

Sources

1. [s1] Effects of Urolithin A supplementation on muscle health outcomes in humans from randomized controlled trials. Frontiers in Nutrition, published online July 22, 2026. https://doi.org/10.3389/fnut.2026.1834344
2. [s2] Urolithins in clinical translation: from gut microbial metabolites to precision interventions. Frontiers in Nutrition, published online August 6, 2026. https://doi.org/10.3389/fnut.2026.1854240