GHK-Cu in Humans: What Clinical Studies Exist

Short answer

Human evidence exists, but it is limited. There is at least one topical randomized trial citable as a primary source, a systematic review that found only 2 randomized trials among 20 studies, and a Phase 2 topical registry record with no results.

That is different from BPC-157, where the published human evidence we use as an anchor is extremely sparse. It is also different from a drug with a mature Phase 3 program. GHK-Cu occupies neither extreme.

This page inventories. It does not recommend a serum, does not specify a technique, and does not turn a registry record into a treatment. The general map is in the GHK-Cu hub.

Inventory also means saying what does not belong. Cultured fibroblasts do not belong. Reviews that recount animal papers as if they were clinical do not belong. A commercial serum does not belong because it says copper. A registry record without results does not belong as if an outcome already existed.

Controlled studies

Only trials that can be traced belong here. A review that says “there were trials” does not, by itself, create a trial citation if the primary publication is not identified.

In this system, the controlled trial with a verifiable primary publication is Miller and colleagues, 2006. The 2026 review reports a second randomized trial among its 20 included studies; that second trial is not identified here as an independent primary publication.^1,2

Older narrative reviews are not counted as trials. They may mention human observations; that does not turn them into an RCT inventory.⁵

Miller 2006

Objective. Computer analysis and blinded evaluation did not find statistically significant advantages for earlier erythema resolution, objective wrinkle improvement, or overall skin quality.¹

Subjective. Patient-reported overall skin-quality satisfaction did favor the GHK-Cu regimen.¹

Route: topical, in a post-laser setting. That fact is not used to justify an injectable use, and the regimen is not reproduced as an instruction.

The skin detail — why the paper is not summarized as wrinkle proof — is in what the clinical evidence actually shows for skin.

2026 systematic-review evidence inventory

Reading “2 RCTs” as if a solid clinical body already existed would be a mistake. Two is more than zero, and it is still few. The authors ask for larger trials, less methodological variability, and standardized formulations.

On the second RCT: if the primary publication cannot be verified in this source system, that is stated. The reference is not fabricated. Miller 2006 is the primary trial that is anchored.

A separate topical review reaches the same practical conclusion: despite cosmetic use, published clinical information is insufficient.⁴

Active Phase 2

A recruiting trial is not a demonstrated treatment. It belongs on the map of research underway, not in the efficacy column. The design is discussed in tissue-repair research.

Route: topical gel. That registry record does not report an injectable compounding trial and is not read as one.

What better evidence would look like

• A sample larger than 13 participants.¹
• A standardized, declared formulation, not a generic “copper peptide” name.²
• Prespecified objective endpoints, not merged with reported satisfaction.¹
• Replication in more than one center and more than one trial.
• Prospective registration and posted results, not only a recruiting record.³

Until then, the exact sentence is: limited human evidence exists, not a clinical base that can carry the stronger cosmetic claims.

A short list helps keep the inventory from inflating. Verifiable primary publication: Miller 2006, topical, n = 13, objective and subjective outcomes kept apart. Synthesis: 2026 systematic review, 20 studies, 2 randomized trials, the second without a primary publication anchored here. Active registry: NCT07437586, topical Phase 2, recruiting, no results. Supporting topical review: Mortazavi and colleagues, published clinical information insufficient.

If another primary publication appears later, it is added with its size, route, and endpoints. Until it exists, a blank is not reserved under an invented name.