Quick preview
In 2026 a research group published the first human topical dataset for urolithin A: two randomised, placebo-controlled trials plus a patch test, all built on a simple moisturising cream carrying 0.5% or 1% of the compound. The headline result is real but narrow — at 1% the periorbital wrinkle depth separated from placebo at eight weeks, while 0.5% never did. For anyone formulating with the ingredient, the more useful information is the awkward part: the study reports no ex vivo skin penetration data, the vehicle is a base cream rather than a serum, and urolithin A is practically insoluble in water with a calculated partition coefficient of 2.3. This article separates what the trial measured from what a formulator still has to prove, and sets out the dispersion, stability and document work that comes before any claim.
What the published topical evidence actually shows
The paper appeared in iScience in 2026 and reports three studies run in sequence: an aging-efficacy trial, a UV-induced photodamage trial, and a human repeat insult patch test (HRIPT) for irritation and sensitisation [1]. The design details matter more than the abstract, because they decide whether the result transfers to the product you intend to make.
In the aging-efficacy trial, 55 candidates were screened and 48 middle-aged women were randomised — mean age 58.6 ± 6.6 years, all of Caucasian ethnicity — in a split-face and split-arm design. One forearm and one side of the face carried the lower dose against placebo, the other side carried the higher dose against placebo: 0.5% and 1% urolithin A in a moisturising base cream, applied twice daily for eight weeks. No participant withdrew. Forearm biopsies were taken at baseline and at week eight.
The photodamage trial enrolled 22 participants (10 male, 11 female; mean age 44.0 ± 14.1 years) and 21 completed. Test sites on the back were irradiated once with 1.6 MED of UVB from a 300 W solar simulator, and erythema was read 24 hours later on the a* axis. The safety study used 112 healthy volunteers aged 19 to 65, both genders and multiple ethnicities, with nine occlusive induction patches over three weeks and challenge patches on days 38 and 40.
| Endpoint | What was reported | What it does not establish |
|---|---|---|
| Periorbital wrinkle depth, 1% versus placebo | Week 2: +5.105 ± 12.40% (placebo) versus −2.972 ± 14.09% (1%), p = 0.0632, n = 23 per arm. Week 8: +4.027 ± 10.88% versus −3.405 ± 11.06%, p = 0.0437, n = 22 per arm. | That the change is visible to a consumer, or that it would repeat in a different vehicle, panel or climate. |
| Periorbital wrinkle depth, 0.5% versus placebo | No statistically significant difference at any assessment time point; the authors state the data are not shown. | That a lower use level is equivalent or ineffective — an underpowered null is not a dose-response curve. |
| UVB-induced erythema, 1% versus untreated site | 24 hours after 1.6 MED: a* value 13% lower than the untreated control site, p = 0.02, n = 21 completers. | Photoprotection. A single erythema reading is not an SPF or a UVA rating, and no claim of that kind follows from it. |
| Irritation and sensitisation, HRIPT | No adverse reactions during the study and no skin reactions at the day 38 and day 40 challenges in 112 volunteers. | Long-term use in a different formula, on compromised skin, or at a higher use level. |
| Biopsy transcriptomics | Gene set enrichment showed collagen fibril organisation as the most upregulated pathway, with epidermal development also up and inflammatory and senescence signatures down. Data deposited as GSE264314. | A mechanism for the wrinkle result. Transcript changes are not a clinical endpoint and were not dose-matched to the visual outcome. |
Two facts about the study are easy to skim past and expensive to ignore. First, the vehicle was a moisturising skin base cream with standard excipients, and the active arms were that same cream with 0.5% or 1% urolithin A added. It was not a serum, not a gel, not an anhydrous balm. Second, the sponsor is a supplier of the ingredient and several authors are employees of the sponsor and of a major cosmetics group, with patents declared. That does not make the numbers wrong — the trials were registered (NCT05079607, NCT05300984, NCT05300542), randomised, double-blind and placebo-controlled — but it does mean the dataset should be read as one well-run study of one vehicle, not as a specification for your product.
The grade in the study is not automatically the grade in your drum
Clinical material is normally a characterised, standardised lot. A commercial drum is a specification plus a certificate, and the two are only equivalent if the specification says so. Urolithin A is a defined compound — CAS 1143-70-0, C13H8O4, 228.20 g/mol, IUPAC name 3,8-dihydroxybenzo[c]chromen-6-one — so the specification can be written tightly, and the purchasing question becomes which of the tight parts are actually measured.
The urolithin A product page publishes the grade as a light yellow powder at not less than 98.0% by HPLC, supplied from 1 kg in a 20 kg fibre drum. That is a purity statement and a packaging statement. Neither of them tells you the particle size distribution, the polymorphic form, the residual solvent profile or the bulk density — and all four change how the powder behaves when you try to put it into an emulsion.
| Attribute | Why it changes the trial | What to ask for |
|---|---|---|
| Assay and method | An area percentage and a calibrated assay against a reference standard are different numbers. | The method, the reference standard and the acceptance range on the batch certificate. |
| Identity confirmation | A single retention time is a purity check, not an identity test. | A second orthogonal technique reported separately, for example 1H NMR or an accurate-mass determination. |
| Particle size | Decides whether the powder disperses or speckles, and how fast it settles in a thin phase. | A distribution (D10/D50/D90) on the lot, not a general statement. |
| Residual solvents | Route-dependent; some solvents are incompatible with a leave-on cosmetic brief. | Class 2 and class 3 results against USP <467> limits on the lot certificate. |
| Bulk and tapped density | Governs handling, dusting and — if you also fill capsules — metering. | Measured values, and a filling trial if the powder is going into a shell as well. |
| Solubility statement | Sets the incorporation route before you write a single line of the formula. | The supplier's descriptive solubility and any measured values in ethanol, propylene glycol or glycerin. |
If your brief also covers a capsule or a powder blend, the density and fill arithmetic belong in the same conversation; the bulk purchasing guide works that side of the calculation through, including what a per-kilogram quote actually covers. The document questions are the same either way: a specification states limits, a batch certificate states measured values for one lot, and the difference is set out in COA, SDS and TDS compared.
Why the molecule resists a simple water-phase addition
Two published numbers explain most of the difficulty. The PubChem record for CID 5488186 lists a computed XLogP of 2.3 [2], and the supplied grade is described as practically insoluble in water and soluble in DMSO and ethanol. A compound with that partition behaviour does not dissolve in the water phase of an emulsion; at best it disperses, and the difference between dispersing and dissolving is the difference between a formula that survives a stability chamber and one that sediments in a jar.
Put the use level on a molar basis and the practical scale becomes clearer. At 1% w/w the cream carries 10 mg of urolithin A per gram, which at 228.20 g/mol is about 43.8 µmol per gram of product. That is not a trace addition, and it is roughly two orders of magnitude above the micromolar concentrations used in most cell-culture work on the compound. Any argument that a formula "contains the studied concentration" therefore needs to state which concentration it means: mass fraction in the finished product, or molar concentration in the test medium. They are not interchangeable.
| Property | Published value | Consequence for the formula |
|---|---|---|
| Molecular formula and mass | C13H8O4; 228.20 g/mol [2] | Use-level arithmetic is straightforward; molar and mass bases must be stated explicitly. |
| Water solubility | Practically insoluble; soluble in DMSO and ethanol (supplied specification) | No direct water-phase dissolution. A pre-wet or carrier step is required. |
| Computed XLogP | 2.3 [2] | Lipophilic enough to favour the oil phase and interfaces, which is why emulsified carriers have been studied. |
| Appearance and form | Light yellow powder | Visible specking is a real risk in a white emulsion; dispersion quality is a visual acceptance criterion, not only an analytical one. |
| Stability in the vehicle | Not established by the clinical paper | Assay recovery over time has to be generated by the brand on its own production batches. Literature cannot substitute for it. |
The formulation literature has already taken two routes around the solubility limit. A 2023 study encapsulated urolithin A in liposomes prepared by a pH-driven method and reported smaller particle size, lower polydispersity and zeta potential, higher encapsulation efficiency, better thermal and storage stability, and a 1.91-fold improvement in rat absorption efficiency over liposomes made by thin-film dispersion [3]. A 2025 study screened nineteen submicron emulsion formulations and selected one carrying 0.2 mg/mL, which kept its physicochemical and process attributes for three months, released the compound in a biphasic profile roughly 2.35 times faster than a suspension, and reduced footpad swelling in a mouse model at 2 mg/kg and 4 mg/kg where the suspension and the blank emulsion did nothing [4].
Neither paper is a cosmetic formulation. Both are pharmaceutical delivery work. What they establish is narrower and still useful: the compound can be carried, the carrier changes the release and the exposure, and a simple suspension is the least effective option of those tested. They also show why the unit on a specification matters — 0.2 mg/mL in the emulsion study is about 880 µmol/L, while that study's cytocompatibility screen ran at 5 µmol/L and below. The two numbers are not on the same scale and should never be quoted side by side as if they were.
Four incorporation routes, and what each one costs you
There is no single correct route. There is a route that fits the product format, the use level, the claim and the factory that will run the batch. The four below are the ones a formulator should cost out before choosing.
| Route | How it works | What to test first | Trade-off |
|---|---|---|---|
| Powder dispersed in an oil-in-water cream | The powder is dispersed into the oil phase or into the finished emulsion under shear, with no attempt at molecular dissolution. | Dispersion uniformity, specking under a microscope, sedimentation on storage, and assay recovery across the batch. | Closest to the published vehicle and the simplest to scale. Depends entirely on particle size and shear. |
| Solvent or glycol pre-wet, then dispersed | The powder is wetted with ethanol, propylene glycol or glycerin before it meets water, which breaks surface tension and limits agglomeration. | How much pre-wet solvent the formula can carry, whether the solvent is declared, and whether the wetting step is reproducible at production scale. | Practically the standard trick for a poorly wetting powder. Solvent load and odour constrain the use level. |
| Liposomal or vesicular carrier | The compound is encapsulated in a lipid vesicle, as in the 2023 study [3]. | Encapsulation efficiency, vesicle size, physical stability, and whether the supplier can deliver the carrier as a reproducible intermediate. | Better dispersion and a marketing story. Adds a supplier, a specification and a cost line, and the cosmetic benefit is unproven. |
| Submicron emulsion concentrate | The compound is carried in a fine oil-in-water emulsion, as in the 2025 screening work [4]. | Loading capacity in your oil system, droplet size, and stability after you dilute or fold it into the base. | Fits a fluid product well. Loading is low (0.2 mg/mL in the published example), so a 1% use level is not reachable by that route. |
Two practical notes cut across all four. First, the emulsifier system you already use decides how much of the powder the interface will tolerate; an olive-derived emulsifier and a self-emulsifying glyceryl stearate will not behave the same way at the same load, and the screening logic is set out in selecting emulsifiers and thickeners. Second, a xanthan gum or similar yield-point builder will hold a dispersion in suspension and buy you bench time, but it will not make an undissolved powder dissolve; if the emulsion destabilises or the viscosity drifts, work through emulsion and viscosity troubleshooting before adding more thickener.
Dispersion, not dissolution: the trial you actually have to run
The clinical paper does not tell you whether the compound stayed in your cream, and no supplier document will either. That evidence is generated by the brand, on the batch it intends to sell, and it is the difference between a formula that can be defended and one that merely contains an ingredient on the label. The sequence below is deliberately front-loaded: the cheap tests come before the expensive ones.
| Stage | Test | What passing looks like |
|---|---|---|
| Pre-formulation | Wetting and dispersion screening in the intended phases | No visible agglomerates at the target load; a documented pre-wet step that a technician can repeat. |
| Bench batch | Microscopy at 100× and 400×; particle or droplet size where relevant | Uniform distribution, no crystalline clusters, no oil-phase crystallisation. |
| Bench batch | Assay recovery by HPLC, sampled top, middle and bottom | Recovery within the agreed range of the theoretical load, with no stratification between sampling points. |
| Stability | Centrifuge, then freeze-thaw cycles, then 4 °C, 25 °C and 40 °C storage | No separation, no sediment, no colour drift, and assay within limits at each condition. |
| Compatibility | pH, preservative efficacy in the presence of the powder, and packaging contact | Preservative performance retained; no interaction with the closure or the label adhesive. |
| Consumer use | Patch test or HRIPT on the finished formula, not on the raw material | No irritation or sensitisation signal in the panel used. The published HRIPT tested a base cream, not your formula. |
The assay recovery test is the one that gets skipped and matters most. A cream can look uniform and still hold half the declared active, because the powder clung to the mixing vessel or was screened out at the homogeniser. Sampling top, middle and bottom, and reporting the recovery against theory, turns that from a suspicion into a number. The general sequence for screening a poorly soluble material is covered in solubility and dispersion testing, and the phase logic behind it in water-soluble versus oil-soluble actives.
Storage deserves its own line. The powder keeps under the conditions on its certificate; the finished emulsion does not inherit that shelf life. Plan the stability programme on the finished product, and keep raw-material storage and finished-goods storage in separate records — the distinction is set out in ingredient storage and stability checks. If a similar active in a similar emulsion is already in your development file, the bakuchiol work in bakuchiol stability is a useful structural analogue for how to design the trial, not a substitute for running it.
The vehicle decides the result
Every number in the published study belongs to one base cream. Change the emulsifier, the oil phase, the humectant load, the pH or the preservative system and you have changed the experiment. Nothing in the paper tells you which of those variables mattered, because the vehicle was never varied — it was the constant against which the active was tested.
Three consequences follow. The oil phase is not neutral: a compound with XLogP 2.3 will partition toward it, so an emulsion with a large oil fraction may hold the active away from the interface and the skin surface, while a low-oil gel may leave it suspended. A plant-derived squalane and a lipophilic ascorbate ester do not present the same partitioning environment, and neither resembles the base cream in the trial. The pH is not neutral either: the compound carries two phenolic hydroxyls, so its ionisation and its reactivity both move with pH, and the storage data have to be generated rather than assumed.
Then there is the question the paper explicitly leaves open. The authors state that one limitation of the study is the absence of ex vivo skin penetration data for urolithin A, and they reason that the observed effects are likely to be penetration-dependent without having measured penetration. That is an honest sentence and a warning for formulators: the clinical result does not prove that the compound reaches the epidermis from an arbitrary vehicle, and a claim built on "it worked in a cream" is a claim built on an untested step. A simple hyaluronic acid or glycerin humectant system will not fix that, and neither will a higher use level.
What the evidence does not support
The gap between what was measured and what is often claimed is wide enough to be worth stating plainly, because claims work belongs to the finished-product file and reviewers ask for the study behind the sentence.
| Wording | Status | Reason |
|---|---|---|
| "Contains 1% urolithin A, the concentration evaluated in a published eight-week split-face study" | Defensible, with the citation | The concentration and the design are reported; the sentence is a statement about the study, not about your product. |
| "Reduces wrinkle depth in eight weeks" | Not defensible as written | It states an outcome for your product. Your formula, panel and vehicle were not tested. |
| "Boosts mitophagy" or "restores mitochondrial function in skin" | Not defensible | Mechanistic findings come from cell and explant models; they are not a consumer outcome and require a different evidence base. |
| "Protects against UV damage" or any SPF implication | Not defensible | The erythema result is a single-dose biomarker reading, not a photoprotection test. Sunscreen claims are regulated separately. |
| "Oral urolithin A benefits apply to skin" | Not defensible | Oral research does not transfer to a topical product without grade-specific identity, safety and regulatory review — the caution already stated on the product page. |
| "0.5% is the effective dose" or "1% is the minimum" | Not defensible | The lower dose did not separate from placebo, which does not establish a threshold; only two concentrations were tested. |
Two more boundaries belong in the development file. The panel in the aging study was 48 women, all of Caucasian ethnicity, with a mean age of 58.6; the photodamage panel was 22 people with a mean age of 44.0. Neither is a general population, and neither answers a question about a different skin type or a younger user. And the study is a single publication from a sponsor with a commercial interest in the result, with authors employed by the ingredient supplier and by a cosmetics group. That is normal for ingredient science and it is not a reason to dismiss the data; it is a reason to read the methods, keep the limitation sentence about penetration in your file, and treat the result as one input rather than as a proof.
A development sequence that survives review
The order below keeps the expensive work until the cheap questions are answered, and it produces the paperwork a quality reviewer will ask for at each gate.
- Fix the grade in writing: assay and method, identity by a second technique, particle size, residual solvents, density. Request the current lot certificate before the sample, not after.
- Choose the route against the product format. A cream can carry a dispersion; a clear gel cannot. Do not pick a carrier because it reads well on a slide deck.
- Run the pre-wet and dispersion screen at the target load, and record the step in the batch record, not in a notebook.
- Build the bench batch and test recovery top, middle and bottom before testing anything else. If recovery fails, nothing downstream is worth running.
- Start stability at the same time as the sensory work, not after it. Three months of real-time data cannot be compressed later.
- Test the finished formula for irritation and sensitisation. The published patch test covered a base cream; your formula is a different article.
- Write the claim against the evidence you hold, and cite the study for the study, not for your product. Keep the marketing file and the technical file separate.
None of this requires a novel ingredient. It requires the discipline of treating a well-publicised active as an ordinary raw material with an awkward solubility profile — which is exactly what it is. If the brief is still at the selection stage, how to select cosmetic active ingredients sets out the comparison logic, and the actives family page lists the grades that are supplied alongside it.
Summary and next step
The 2026 topical dataset is worth having: at 1% in a base cream, urolithin A separated from placebo on periorbital wrinkle depth at eight weeks (p = 0.0437) and reduced UVB erythema by 13% at 24 hours (p = 0.02), with a clean 112-person patch test. It is also narrow — one vehicle, two concentrations, one panel, no penetration data — and the lower dose did not separate from placebo at all. For a formulator, the practical conclusion is that the ingredient is a dispersion problem before it is a claims opportunity: fix the grade, pre-wet the powder, prove recovery across the batch, and generate your own stability and tolerability evidence before any sentence goes on a pack.
Download the urolithin A product guide →
Working through the supply side at the same time? The bulk purchasing guide covers order minimums, fill arithmetic and the documents to request before a deposit.
References and data summary
Numerical research data, catalog specifications and development recommendations are separated below. No customer survey or RAWESS clinical trial is claimed.
Read and checked: 13 November 2026
[1] D'Amico D, Faitg J, Amaral F, et al. Topical application of urolithin A activates mitophagy and enhances mitochondrial function to promote skin longevity. iScience. 2026;29(10):117646. doi:10.1016/j.isci.2026.117646. PMID 42866687; PMCID PMC13657230. doi.org/10.1016/j.isci.2026.117646. Source version / publication: peer-reviewed article, open access under CC BY-NC-ND.
Randomised, double-blind, placebo-controlled aging-efficacy trial (n = 48 randomised, 48 completed) and UV-induced photodamage trial (n = 22 enrolled, 21 completed), plus an HRIPT safety study (n = 112). Vehicle: moisturising skin base cream with 0.5% or 1% urolithin A. Wrinkle depth week 8: p = 0.0437 for 1% versus placebo; 0.5% not significant at any time point. Erythema a* at 24 h after 1.6 MED UVB: −13% versus untreated, p = 0.02. Authors report the absence of ex vivo skin penetration data as a limitation and declare employment and shareholding with the ingredient sponsor and with a cosmetics group, plus related patents.
[2] National Center for Biotechnology Information. PubChem Compound Summary for CID 5488186, Urolithin A. pubchem.ncbi.nlm.nih.gov/compound/5488186. Accessed 13 November 2026.
Molecular formula C13H8O4; molecular weight 228.20 g/mol; computed XLogP 2.3; IUPAC name 3,8-dihydroxybenzo[c]chromen-6-one; synonyms include 1143-70-0 and Uro-A.
[3] Hu Y, et al. pH-driven preparation of urolithin A-loaded liposomes: characterisation, stability and bioavailability. International Journal of Biological Macromolecules. 2023;253(Pt 7):127554. doi:10.1016/j.ijbiomac.2023.127554. PMID 37865359. doi.org/10.1016/j.ijbiomac.2023.127554. Source version / publication: peer-reviewed article.
Describes urolithin A as poorly water-soluble with low bioavailability. The pH-driven liposome route gave smaller particle size, lower polydispersity and zeta potential, higher encapsulation efficiency, and better thermal, storage and in vitro digestion stability and bioaccessibility than the thin-film dispersion route; rat pharmacokinetics showed a 1.91-fold higher absorption efficiency.
[4] Esposito E, et al. Submicron emulsions for the delivery of urolithin A: formulation screening and in vivo evaluation. Journal of Pharmacy & Pharmaceutical Sciences. 2025;28:15553. doi:10.3389/jpps.2025.15553. PMCID PMC12634460. doi.org/10.3389/jpps.2025.15553. Source version / publication: peer-reviewed article, open access.
States that the high lipophilicity of urolithin A hampers its pharmaceutical application. Nineteen submicron emulsion formulations were screened; the selected formulation carried 0.2 mg/mL, preserved its physicochemical and process attributes for three months, released the compound in a biphasic profile about 2.35 times faster than a suspension, showed no cytotoxicity up to 5 µmol/L, and reduced mouse footpad oedema at 2 mg/kg (24 h) and 4 mg/kg (all time points) where the suspension and blank emulsion were inactive.
[5] ClinicalTrials.gov registrations for the safety and efficacy studies reported in [1]: NCT05079607 (HRIPT), NCT05300984, NCT05300542. clinicaltrials.gov/study/NCT05079607. Accessed 13 November 2026.
Registration records used to confirm the study designs cited above. Trial registry entries summarise planned endpoints and do not replace the peer-reviewed results.
Data and calculation notes
Wrinkle-depth figures are percentage change from baseline, reported as mean ± standard deviation in the source. The 1% arm at week 2 is p = 0.0632 (n = 23 per arm) and at week 8 is p = 0.0437 (n = 22 per arm); the paper also quotes p = 0.04 for the week 8 comparison. The 0.5% arm showed no statistically significant difference from placebo at any time point and the underlying values are not tabulated in the article. The erythema result is an a* colour value read 24 hours after a single 1.6 MED UVB exposure against an untreated control site, with p = 0.02 in 21 completers; it is not an SPF, a UVA rating or a photoprotection measurement.
Use-level arithmetic: 1% w/w equals 10 mg per gram of product; dividing by 228.20 g/mol gives approximately 43.8 µmol per gram. The 0.2 mg/mL loading in [4] equals 0.2 g/L, or approximately 880 µmol/L; the cytocompatibility screen in the same paper ran up to 5 µmol/L, so the two figures are on different scales and are not compared in this article.
Product evidence used in this article
- Urolithin A — catalog reference for identity, assay basis, packaging and document request; request current lot documents.
- Olive Squalane — oil-phase reference for partitioning and sensory screening.
- Olivem 1000 Type — emulsifier reference for the carrier comparison.
- Glyceryl Stearate SE — self-emulsifying reference for the carrier comparison.
- Xanthan Gum Clear — rheology reference for holding a dispersion in suspension.