Four things get sold under this heading: copper peptides for skin and hair, melanotan for tanning, oral collagen for wrinkles, and injectable “glow” blends. Three have a gap between claim and data. The fourth has a safety signal, so it goes first.
The melanotan case reports
The compound sold online as Melanotan II has a case-report record nothing else here matches. On melanoma, the indexed cases include a 20-year-old Fitzpatrick II woman with a gluteal melanoma excised three months after a self-injected course of three to four weeks alongside sunbed use (Dermatology 2014, PMID 24355990), a melanoma in situ (PMID 22724573), five more in situ melanomas in one patient who also reported tanning bed and anabolic hormone exposure (JAAD Case Rep 2026, PMID 42328529), and a 22-year-old woman with an oral mucosal melanoma of the anterior maxilla after a Melanotan II nasal spray (Int J Oral Maxillofac Surg 2025, PMID 40210573). That last one matters because mucosal melanoma isn’t ultraviolet driven, so sunbed co-exposure doesn’t explain it. A fifth report (Br J Dermatol 2011, PMID 21564053) carries two erratum notices that aren’t indexed or retrievable, so what was corrected is unknown and it shouldn’t be cited until that’s resolved.
The naevus reports are more numerous and more informative. The strongest describes a 40-year-old man with a prior melanoma and multiple dysplastic nevi who developed crops of new nevi with atypical clinical and histopathologic features, while pre-existing nevi darkened and acquired growth features. After he stopped, the nevi progressively lightened and lost those features (Arch Dermatol 2009, PMID 19380666). That dechallenge is the closest thing to causal evidence in the file. Others report eruptive naevi 24 hours after a single injection (PMID 24334249), new atypical naevi within a week of two injections (PMID 23914578), new nevi and lentigines after more than 150 self-injected doses (PMID 21453640), severe dysplasia on histopathology after a four-week course (PMID 22425244), and a 16-year-old girl with FAMMM syndrome (PMID 23785612). One reported harm isn’t cancer at all: videodermoscopy during use showed pigment changes that made nevus and melanoma harder to tell apart (PMID 23052015).
Two syntheses bound it. A 2017 Int J Dermatol review (PMID 28266027) counted four melanoma reports, noted that several national health organisations have issued warnings, and stated that conclusive evidence linking these phenomena is lacking. A 2019 systematic review of eruptive melanocytic nevi covering 93 articles and 179 patients attributed 41% of cases to immunosuppressants, chemotherapy or melanotan, found 16% had a histologically confirmed dysplastic nevus and five had associated melanoma, and concluded that the data precluded assessing risk of malignant transformation (PMID 31119650). Preclinical work points the other way: topical Melanotan II suppressed B16-F10 melanoma in mice via PTEN upregulation (PMID 31968661).
So this is a case-report-level signal with a positive dechallenge, a plausible mechanism (alpha-MSH analogues driving proliferation of neoplastic melanocytes in predisposed people, per PMID 19380666), and replication across more than fifteen reports in at least five countries. It isn’t trial-grade and shouldn’t be presented as such. The reports cluster in young users with dysplastic naevus syndrome, FAMMM, prior melanoma or concurrent sunbed use.
The rest of the melanotan record
The non-dermatologic cases are all Melanotan II and all single patients: rhabdomyolysis in a 39-year-old man who injected 6 mg subcutaneously, with creatine kinase rising from 1,760 to 17,773 IU/L at 12 hours and three days in intensive care, the substance confirmed by mass spectrometry (Clin Toxicol 2012, PMID 23121206), two priapism cases, one with erectile function not recovered at four weeks (PMID 30796078, PMID 23537392), renal infarction (PMID 31953620), and sympathomimetic toxicity two hours post-injection (PMID 35736369).
Then there’s what’s in the vials. Validated LC-UV-MS/MS on material from three online shops found 4.32 to 8.84 mg of content against a labelled 10 mg, with unknown impurities at 4.1 to 5.9% in two of the three (PMID 24771717). Seized vials came back at 30% purity (PMID 39302005). The MHRA has raised blood-borne virus transmission through needle sharing on top of impurity (PMID 20545686), and an analysis of 623 forum entries from 205 participants found misinformation, sunbed use and needle sharing as recurring themes (PMID 34464955).
Afamelanotide isn’t the same molecule
This distinction gets blurred, sometimes deliberately. Afamelanotide is [Nle4-D-Phe7]-alpha-MSH, a linear peptide selective for MC1R. Melanotan II is a cyclic lactam that acts across the melanocortin receptor family rather than selectively at MC1R. That receptor promiscuity is why the priapism and sympathomimetic cases attach to Melanotan II and not to afamelanotide.
Afamelanotide is approved. The EMA cleared it in 2014 and the FDA on 8 October 2019, as a 16 mg subcutaneous implant every 60 days, with a label restricted to adults with erythropoietic protoporphyria. The approval rests on two multicentre double-blind placebo-controlled trials published together in N Engl J Med 2015 (PMID 26132941), NCT01605136 with 94 US patients and NCT00979745 with 74 in Europe. Over six months the US trial reported a median 69.4 hours of pain-free direct sunlight exposure against 40.8 (P = 0.04). Over nine months the European trial reported 6.0 hours against 0.8 (P = 0.005), with 77 phototoxic reactions against 146 (P = 0.04). Clinuvel Pharmaceuticals funded the work. Read the European figures as absolutes and the effect is small, which a 2014 review concedes (PMID 25470471). A 2017 review reported no late effects in volunteers 25 years on, continuous use up to eight years, and headache, fatigue, nausea and implant-site reactions as the common adverse events (PMID 28063031).
One registered trial tests Melanotan II under oversight: NCT07437560, a Phase 2 randomised placebo-controlled study of 60 participants alongside narrowband UVB in vitiligo, recruiting with no results. A registered medical trial isn’t a validation of cosmetic tanning use.
GHK-Cu: the topical trials are negative
The highest-grade human trial of topical copper tripeptide is Arch Facial Plast Surg 2006 (PMID 16847171), a randomised controlled trial with 13 completers after circumoral CO2 laser resurfacing. Erythema by computer analysis and by blinded evaluators showed no significant difference, and wrinkles and overall skin quality assessed objectively showed no difference between groups. The one positive result came from the patient self-report questionnaire (P = .04). The authors wrote that objective evaluation found no significant improvement in wrinkles or overall skin quality.
The other human trial rarely appears in copper-peptide marketing. J Vasc Surg 1992 (PMID 1495150) randomised 86 evaluable patients with venous stasis ulcers to silver sulfadiazine 1%, tripeptide-copper complex 0.4% cream, or inert vehicle. Silver sulfadiazine beat both, and the paper states there was no difference between the latter two. The copper peptide matched placebo, at a concentration inside the commercial cosmetic range.
That’s the controlled human base. A PubMed query for GHK-Cu, copper tripeptide, copper peptide or glycyl-histidyl-lysine, filtered to trial publication types, returns three records, the third a positive 2006 open-wound study in 18 rabbits (PMID 17083573). The other 236 records in the corpus are in vitro, rodent, zebrafish, chemistry or review.
A 2026 split-face study does exist (J Cosmet Dermatol, doi 10.1111/jocd.71129), running GHK-bound microneedle spicules over five weekly sessions in 20 Asian women of mean age 55.20, with nasolabial wrinkle depth at -0.0227 mm against -0.0173 mm (p = 0.016) and dermal density at +7.27 against +4.49 (p < 0.001). It was active-controlled rather than randomised, and both arms got the GHK regimen, with an added glutathione booster the only difference. It isn’t GHK against vehicle, the between-arm wrinkle gap is roughly five microns, and the funding source isn’t stated.
Injected GHK-Cu has no human evidence at all
ClinicalTrials.gov holds no registered trial of injectable GHK-Cu for any skin, hair or aesthetic endpoint, and PubMed holds no controlled human trial of it either. The only registered GHK-Cu drug trial is NCT07437586, a Phase 2 vehicle-controlled split-wound study of a topical gel in 60 participants, still recruiting.
The nearest human data can’t support attribution. A 2018 paper reported intradermal scalp injections in 1,000 patients with a significant reduction in hair fall on pull test in 83% of patients (PMID 29482481), but it’s open-label, single-arm and uncontrolled, and the formulation also holds VEGF, bFGF, IGF, KGF and thymosin beta-4. A retrospective series of 18,918 men with androgenetic alopecia is likewise uncontrolled, with copper tripeptide mixed into an injectable of minoxidil, caffeine and vitamins, given alongside oral finasteride and minoxidil (PMID 30057663).
Where the copper-peptide claims come from
The sentence that circulates everywhere, that GHK-Cu tightens skin, improves elasticity and firmness and reduces fine lines, wrinkles, photodamage and hyperpigmentation, is verbatim from the abstract of J Biomater Sci Polym Ed 2008 (PMID 18644225). That paper and the other heavily cited GHK reviews (PMID 29986520, 26236730, 25302294, 22666519, 28212278) share an author whose listed affiliation is R&D Skin Biology, a copper-peptide skincare seller. The 2018 Int J Mol Sci review declares no conflict of interest while listing that company as its affiliation. The claims are attributed to controlled studies on aged skin, and no such indexed trial exists.
Several specific citations don’t survive checking. A “split-face controlled trial, 67 subjects, 2% GHK-Cu cream, 12 weeks, 23% reduction in fine wrinkle depth” appears on seller blogs, and a PubMed search for GHK or copper peptide with split-face returns zero records. “Watson et al. 2009, British Journal of Dermatology, GHK-Cu double-blind RCT” is a misattribution: PMID 19438432 is a 60-patient randomised trial of an unnamed over-the-counter cosmetic product, it doesn’t test GHK-Cu, and it was itself contested in print (PMID 19678818, PMID 20479305). The load-bearing source for the claim that copper peptides outperform retinoic acid, Abdulghani et al. 1998 in the defunct Disease Management and Clinical Outcomes, isn’t PubMed indexed and its numbers all trace to secondary sources.
Oral collagen: positive until you sort by who paid
The collagen literature is large and almost entirely positive. Meta-analyses have pooled 19 RCTs and 1,125 participants (PMID 33742704) and 26 RCTs and 1,721 participants, with hydration at Z = 4.94 and elasticity at Z = 4.49, both p < 0.00001 (PMID 37432180). A 2020 review of ten publications noted that almost all reported benefit and that no inconsistencies were seen (PMID 32436266), a uniformity worth checking for sponsorship bias.
That check now exists. Am J Med 2025 (PMID 40324552) pooled 23 RCTs and 1,474 participants and found significant improvement in hydration, elasticity and wrinkles overall. Stratified by funding, trials without pharmaceutical or industry funding showed no effect on hydration, no effect on elasticity and no effect on wrinkles. Stratified by quality, high-quality studies showed no significant effect in any category, and only low-quality studies showed an elasticity benefit. The authors, based in Korea and Australia with no industry affiliation, concluded that there’s currently no clinical evidence supporting collagen supplements for skin ageing. The full text is paywalled, so the pooled effect sizes aren’t reproduced here.
The individual trials explain that result. The two most-cited are both Skin Pharmacol Physiol 2014. One enrolled 114 women aged 45 to 65 on 2.5 g daily for eight weeks and reported roughly 20% eye wrinkle volume reduction against placebo (p < 0.05), with procollagen I up 65% and elastin up 18% in a biopsy subgroup, while fibrillin rose 6% and missed significance (PMID 24401291). The other ran 69 women across 2.5 g, 5.0 g and placebo arms of 23 each, found elasticity improved at both amounts, and missed significance on skin moisture and water loss (PMID 23949208). Both state in methods that the product was supplied by GELITA AG, neither published PDF contains any conflict of interest, disclosure or acknowledgement section, and co-authors sit at the Collagen Research Institute in Kiel and at a contract testing house in Hamburg. In that three-arm design 5 g gave no clear advantage over 2.5 g, and an absent dose-response is a negative signal the abstract doesn’t foreground.
The pattern repeats. Authors employed by the manufacturer or sponsor appear on PMID 26362110, PMID 26840887, PMID 30513923 (which discloses that sponsor employees designed the study and wrote the paper) and PMID 29949889, whose conflict statement says the authors and the company Newtree declare no conflict while two authors work for Newtree. Multi-ingredient products that make collagen’s own contribution unidentifiable cover PMID 31627309 and PMID 30122200, the latter reporting 40% elasticity improvement from a blend including chondroitin, glucosamine and L-carnitine. A trial funded by the collagen supplier reported a 35% wrinkle reduction from baseline (p = .035) that was only 24% greater than placebo, and only on the right side of the face (PMID 32799362). The one trial with no identifiable industry tie measured elasticity only (PMID 32009486).
Three things survive. Safety looks clean, with no serious adverse events across the 1,474 to 1,721 pooled participants, unsurprising for a food-grade protein. Mechanistic plausibility is real, since Pro-Hyp and Hyp-Gly dipeptides are detectable in human blood after ingestion (PMID 26840887). And the pressure-ulcer question is separate and better controlled: a multicentre RCT across 23 long-term care facilities completed 71 of 89 and reported better PUSH scores at eight weeks (P < .05) in Adv Skin Wound Care 2006 (PMID 16557055), though randomisation was unequal at 56 against 33 and the means of 3.55 and 3.22 carry standard deviations of 4.66 and 4.11, large enough that the paper’s “twice the rate” conclusion isn’t supported by its own numbers.
“Glow” blends
The circulating blend is GHK-Cu with BPC-157 and TB-500, injected subcutaneously. Controlled human trials of that combination number zero, in PubMed and on ClinicalTrials.gov alike.
Component by component, injectable GHK-Cu has nothing. The only human BPC-157 data is a retrospective chart review of 17 patients, 16 reachable, with no control group, no validated instrument, and telephone recall six to twelve months after intra-articular injection, in which 11 of 12 reported improvement (PMID 34324435). The paper states that no specific tools were used to measure improvement in function, quality of life, stiffness or daily activities. A 2026 Am J Sports Med paper calls it methodologically flawed and uncontrolled, and states that human orthopaedic data for TB-500 are lacking and that both compounds remain banned in sport (PMID 41476424). BPC-157 isn’t FDA approved, and its safety profile is under live dispute in the literature (PMID 40005999, PMID 41155565).
Independent 2026 reviews land in the same place. Sports Med (PMID 41966639) describes a gray market of unapproved compounds outside regulatory oversight, says rigorous human safety data are scarce with potential for serious harm, and names the placebo effect, amplified by social media, as a mediator of apparent peptide efficacy. A PRISMA scoping review found 67% of identified publications were preclinical animal models, with human studies limited to a handful, most lacking adequate controls (PMID 42578445).
What’s established and what’s marketing
Established: afamelanotide performs on its narrow approved indication across two randomised placebo-controlled trials, with modest absolute effects and a long safety record. Collagen peptides showed no serious adverse events across the pooled trials and have a plausible mechanism.
Not established: that topical GHK-Cu improves wrinkles or skin quality on any blinded or instrumented endpoint, that injected GHK-Cu does anything in humans, that oral collagen changes skin once funding and study quality are accounted for, or that any “glow” blend has been tested in humans. Several of the trials cited for copper peptides don’t exist in the indexed literature.
And outside the efficacy question sits the melanotan file: melanoma, eruptive dysplastic naevi, rhabdomyolysis, priapism and renal infarction, in people injecting vials that assay at 30% purity and 43 to 88% of the labelled dose.
This article is a review of published research. It is not medical advice, and nothing here is a recommendation to use any compound. The compounds discussed are research-use-only materials, not approved for human consumption and not endorsed here for use in people.
