Skin has its own cannabinoid signalling machinery, and that single fact has sold an enormous amount of cream — so it is worth separating the receptor biology from what has actually been measured in people.
1. Why the Skin Became a Target
Applying something feels safer than swallowing it: no first-pass metabolism, a local dose, a visible area. Add the discovery that skin cells carry cannabinoid receptors and the commercial logic writes itself. The biology is interesting; the clinical evidence is thin. This article keeps them apart. Nothing below is a recipe or a recommendation: Thailand’s controlled-herb framework, revised in 2025, restricts cannabis products and prohibits therapeutic advertising.
2. The Skin’s Own Cannabinoid Signalling
Where the receptors are
The endocannabinoid system is not confined to the brain. In skin it is the cutaneous endocannabinoid system: receptors, the lipid messengers that activate them (chiefly anandamide and 2-arachidonoylglycerol), and the enzymes that make and clear them. A 2005 study in the Journal of Dermatological Science, on human tissue, mapped cannabinoid receptor type 1 and type 2 onto sensory nerve fibres and adnexal structures — hair follicles and sebaceous glands. A 2023 review in the International Journal of Molecular Sciences adds keratinocytes and mast cells.
How well characterised it is
Less well than the marketing implies. A 2009 review in Trends in Pharmacological Sciences framed this system as a therapeutic opportunity — a scientist’s way of saying the work was not yet done. Most of what followed is receptor mapping and cell-culture pharmacology, not physiology in living humans — and a receptor two hundred micrometres down helps only if the molecule reaches it.
3. Topical Versus Transdermal
Two different engineering goals
| Feature | Topical | Transdermal |
|---|---|---|
| Design intent | Act in or under treated skin | Cross skin, enter the bloodstream |
| Blood levels | Low to undetectable by design | Meaningful and intended |
| Typical form | Cream, balm, salve, lotion | Patch or rate-controlled gel |
| Regulatory status | Often cosmetic | Almost always a medicine |
Why most “transdermal” claims are wrong
A jar of oily balm with cannabidiol dissolved in it is a topical, whatever the label says. A real transdermal system is defined by delivery rate per hour and blood-level data; a product stating neither is using the word decoratively. For contrast: an open-label trial in Scientific Reports (2024), in fifteen completers, used a 4% transdermal gel at about 30 mg daily and found cannabidiol and its metabolites in every urine sample.
4. The Penetration Problem
The stratum corneum, the dead outermost layer, is the barrier. Molecules that cross it well are under about 500 daltons with a partition coefficient — oil-versus-water preference, written logP — of roughly 1 to 3. Cannabinoids are the wrong shape: a 2022 review in Pharmaceutics puts their logP at 6 to 7, with very low water solubility. They enter the stratum corneum readily and stay there, forming a reservoir in dead tissue. Extreme lipophilicity is not a bonus for skin delivery; it is the reason skin delivery is hard.
A 2024 review in Cannabis and Cannabinoid Research catalogues chemical enhancers, nanocarriers such as liposomes, microneedles and prodrugs, while noting that clinical trial literature here is limited. Magnitudes matter: a 2023 study in Pharmaceutics put cannabidiol in organosilane particles on excised human skin and moved 0.41% of the applied dose into the lower epidermis and dermis, against 0.27% for plain cannabidiol — better by half, still under one percent, and below therapeutic levels by the authors’ own account.
5. Eczema and Atopic Dermatitis
Much of the eczema evidence attributed to cannabis is not about plant cannabinoids at all. It concerns palmitoylethanolamide, an endogenous fatty acid amide that is cannabinoid-adjacent: it acts through nuclear receptors and endocannabinoid tone, not by activating cannabinoid receptors directly.
The most cited item is the ATOPA study, reported in 2008 in the Journal of the European Academy of Dermatology and Venereology: a multicentre, observational, uncontrolled assessment in 2,456 patients aged 2 to 70 using a barrier cream with physiological lipids and palmitoylethanolamide, reporting reductions in erythema, pruritus, dryness and steroid use. Large sample, weak design — emollients help eczema anyway, and it fluctuates.
A 2024 double-blind, comparator-controlled randomised trial in Skin Pharmacology and Physiology (Rao, Moussa, Erickson and Briskey) is stronger: 72 adults, 65 completing, palmitoylethanolamide cream against a moisturiser for four weeks, the active arm improving more on two severity measures. Cannabidiol-specific work is thinner: a 2024 review identifies a 14-patient study of a 1% cannabidiol gel and a 2019 report of 20 uncontrolled patients.
6. Psoriasis
The laboratory case is clean. Plaques involve excessive keratinocyte proliferation, and Wilkinson and Williamson reported in the Journal of Dermatological Science in 2007 that tetrahydrocannabinol, cannabidiol, cannabinol and cannabigerol all inhibited cultured human keratinocyte proliferation at low micromolar concentrations — half-maximal inhibition averaging about 2.3 micromolar — through a mechanism not explained by type 1 or type 2 receptors alone.
Clinically, almost nothing followed for a decade. The most substantial human study is Thai: a dual-centre randomised placebo-controlled study reported in 2022 by Puaratanaarunkon, Sittisaksomjai and colleagues, in 51 patients with mild plaque psoriasis, using 2.5% cannabidiol ointment in a split-body design over 12 weeks. Treated areas scored better than placebo areas with no adverse effects; the authors described a favourable trend and framed the work as paving the way for future studies.
7. Acne and the Sebaceous Gland Study
What the laboratory work showed
The study everyone cites is Oláh and colleagues, Journal of Clinical Investigation, 2014. Cannabidiol was applied to immortalised cultured human sebaceous gland cells (the SZ95 line) and to human skin in organ culture, mostly at 1 to 10 micromolar. It suppressed lipid production driven by acne-promoting stimuli, suppressed proliferation without killing cells, and blunted inflammatory cytokines, acting through the TRPV4 channel and an adenosine A2a pathway.
What it was not: a study in people. Cultured cells and excised tissue in a dish — no barrier to cross, no blood supply, no immune system, no follicular bacteria. A rationale for a clinical trial, and nothing more.
What happened in humans
Botanix Pharmaceuticals ran a 12-week randomised, double-blind, vehicle-controlled Phase 2 study of BTX 1503, a synthetic cannabidiol solution, in 368 patients with moderate-to-severe acne, reporting in October 2019. The pooled analysis did not reach significance on the primary endpoint, change in inflammatory lesion count; the best dose reduced lesions by 11.8 against 8.6 for vehicle.
8. Localised Pain and Inflammation
The most advertised use, and one of the weakest evidence bases.
For joints, the 2024 Scientific Reports open-label trial — 15 completers, 4% transdermal gel, four weeks — reported roughly two-point reductions on a ten-point pain scale, better grip strength, and reversal of both during washout, which expectation alone explains poorly. The authors listed the limits themselves: no placebo control, small sample, open label. A 2022 randomised trial in the Journal of Hand Surgery, in thumb basal joint arthritis, also reported benefit on a similarly small scale.
For nerve pain, a randomised placebo-controlled trial by Xu, Cullen, Tang and Fang in Current Pharmaceutical Biotechnology (2020) gave 29 people with lower-limb peripheral neuropathy (15 active, 14 placebo) a topical cannabidiol oil for four weeks, finding significant reductions in intense pain, sharp pain, and cold and itchy sensations.
The honest verdict: favourable direction, benign safety profile, samples far too small — and carrier oils, massage and counterirritants lower pain scores on their own.
9. Itch
Itch is where the mechanistic argument is strongest. It is signalled by unmyelinated C-fibre endings terminating in the upper dermis and reaching into the epidermis — the shallowest target in skin, and the one a poorly penetrating molecule has the best chance of reaching. Cannabinoid receptors sit on those fibres and on the mast cells and keratinocytes around them.
A 2006 report in Der Hautarzt by Ständer, Reinhardt and Luger described open, uncontrolled use of a palmitoylethanolamide cream in 22 patients with prurigo and chronic pruritus: 14 of 22 responded, mean itch reduction 86.4%. A 2005 study by Szepietowski and colleagues used a similar cream in 21 haemodialysis patients with uraemic pruritus, resolving itch completely in 38%.
Then the controlled trial. Viße, Blome, Phan, Augustin and Ständer, in Acta Dermato-Venereologica, randomised 100 people with chronic pruritus due to dry skin to a palmitoylethanolamide lotion or the vehicle alone. Both improved substantially; the active group did slightly better (30.1% versus 24.2%), but the difference was not significant. That is the pattern in miniature: uncontrolled studies produce spectacular numbers, and once a vehicle control is added, most of the effect belongs to the base.
10. Intoxication and Drug Testing
Can a topical make you high? For an ordinary cream, balm or salve on intact skin — no. A 2017 study in Forensic Science International put its conclusion in its title: topical application of tetrahydrocannabinol-containing products cannot cause a positive cannabinoid finding in blood or urine.
The exceptions are real. Genuine transdermal systems cross — the 2024 hand osteoarthritis trial found cannabidiol and its metabolites in every participant’s urine. That is not the same as failing a cannabis test, since standard screens target the tetrahydrocannabinol metabolite THC-COOH, not cannabidiol; but an absorbing formulation that also contains tetrahydrocannabinol is a different calculation. Mucosal surfaces have no stratum corneum, so oral, vaginal or rectal application bypasses the barrier entirely. Compromised skin matters: broken or inflamed skin has a defective barrier, and those treating the largest areas are exactly the people whose barrier is impaired.
11. What Is Actually in the Jar
Label accuracy and hemp seed oil
The most rigorous data comes from Spindle and colleagues, JAMA Network Open, 2022 — a Johns Hopkins-led analysis of 105 hemp-derived topical products bought online and in retail stores. Of those declaring a cannabidiol amount, 18% were overlabelled (more than 10% less than advertised), 58% underlabelled, and only 24% accurate within 10%. Tetrahydrocannabinol was found in 35%; of those, 11% were labelled tetrahydrocannabinol-free and 51% said nothing about it.
The commonest consumer error is an ingredient-list problem. Hemp seed oil — listed as Cannabis sativa seed oil in cosmetic nomenclature — is pressed from the seed. It is a good emollient containing essentially no cannabinoids, because cannabinoids form in the glandular trichomes of the flower, not the seed; any trace is processing residue. Cannabidiol extract comes from the aerial parts and is a different, costlier ingredient. A jar can carry leaf imagery, be sold as a cannabis cream and hold nothing but seed oil.
| What you see | What it means | What to check |
|---|---|---|
| Cannabis sativa seed oil | Seed emollient; no meaningful cannabinoids | Whether a cannabinoid is listed separately |
| Hemp extract, full-spectrum | Vague; may not contain cannabidiol | A milligram amount and a laboratory report |
| “500 mg” with no context | Total in the container, not a concentration | 500 mg in 500 g is 0.1% |
| “Tetrahydrocannabinol-free” | An assertion, not a measurement | The detection limit |
Reading a certificate of analysis
A certificate of analysis is a third-party laboratory report on one production batch. Useful ones name an independent laboratory, not the brand; carry a batch number matching the product in your hand; are dated; give a full cannabinoid panel with measured values and detection limits; and cover heavy metals, pesticides, solvents and microbes. Undated, unbatched or seller-issued ones are decoration, not data.
12. Irritation, Allergy and Patch Testing
Cannabinoid topicals were generally well tolerated in the trials above. Most reactions that occur come from everything else in the jar, chiefly fragrance terpenes. Limonene and linalool are common in botanical formulations and in cannabis-derived material itself, and both oxidise in air into hydroperoxides that are far more potent sensitisers. Patch-test data summarised by DermNet indicates around 10% of people investigated for suspected contact dermatitis react to these hydroperoxides — so a product tolerated when new can become a problem months later.
Testing a small area first is standard for any new topical on sensitive skin, and suspected allergy is confirmed by formal patch testing — both matters for a clinician, not a forum. Caution applies to broken or weeping skin, known fragrance or plant allergy, active infection, children, pregnancy and breastfeeding, workplace drug testing, and anyone whose eczema or psoriasis is under prescribed treatment — substituting an unproven cream for effective therapy is the real harm here.
13. Evidence at a Glance
| Topic | Best evidence type | Scale | Where it sits |
|---|---|---|---|
| Acne, laboratory | Cultured sebocytes and organ culture | Oláh 2014 | Rationale only |
| Acne, clinical | Human randomised, vehicle-controlled | 368 patients | Endpoint not met |
| Psoriasis, laboratory | Cultured keratinocytes | Wilkinson 2007 | Rationale only |
| Psoriasis, clinical | Human randomised, split-body | 51 | Favourable trend |
| Eczema, palmitoylethanolamide | Human randomised, controlled | 72 enrolled | Modest but real |
| Eczema, cannabidiol | Human, uncontrolled or pilot | 14 and 20 | Preliminary |
| Itch, controlled trial | Human randomised, vehicle-controlled | 100 | Not significant |
| Joint and nerve pain | Human, small randomised and open | 15 and 29 | Underpowered |
14. Questions Worth Asking
Five questions separate signal from marketing. Does the product state a cannabinoid amount in milligrams alongside container size? Is there a batch-matched, dated, third-party certificate of analysis? Does the ingredient list show an actual cannabinoid, or only Cannabis sativa seed oil? If it claims to be transdermal, is there delivery-rate or blood-level data behind the word? And has the condition been discussed with a clinician, given how much larger the evidence base is for standard treatments?
What has your experience been? If you have used a topical preparation for a skin or joint complaint, we want the unglamorous details: what was actually in it, whether a batch-matched certificate of analysis existed, and whether any effect outlasted two weeks. Community reports are not evidence, but honest ones — including the ones where nothing happened — beat another round of product claims. Post your account below, with the ingredient list.
This article is part of the Asiannabis Community educational series on medical cannabis. It is general information, not medical advice — treatment decisions belong with a licensed practitioner. Content is for educational purposes within jurisdictions where medical cannabis is legally permitted.