COMPARISON · ALL THREE
The Substitution Ledger
Same theme, three different distances from the claim. What each peptide was measured in, and what has to be assumed to get from there to a nerve in a person.
The honest summary
This page lines up BPC-157, GHK-Cu and TB-500 on the things that decide how much a study is worth to a nerve question: which molecule was tested, which species it was tested in, which tissue was measured, how it was delivered, and what exists in humans.
The short answer is that the three fail the nerve question in three different ways. BPC-157 has the deepest animal file and the thinnest human one. GHK-Cu has the best human data and the least relevant tissue. TB-500 has the only neurological result and did not use the molecule people actually buy.
Read across the table below and the pattern is easier to see than to argue with: every strong result in this hub belongs to a species, a tissue or a molecule other than the one a nerve-repair claim needs. Nothing here is medical advice and no dose for a person appears anywhere on this site.
The ledger
| Dimension | BPC-157 | GHK-Cu | TB-500 |
|---|---|---|---|
| What it is | Synthetic 15-amino-acid peptide from a gastric juice protein | Tripeptide Gly-His-Lys bound 1:1 to copper(II) | 7-amino-acid fragment (Ac-LKKTETQ) of thymosin beta-4 |
| Molecule actually tested | The same peptide throughout | GHK-Cu topically; one key trial used a 5-ALA combination [10] | Mostly full-length thymosin beta-4, not the fragment [14][16] |
| Species behind the main claims | Rat, plus chick membrane and cell culture [6][5][4] | Human, in small topical trials [11][8][10][12] | Rat for efficacy [14]; human for safety only [16] |
| Tissue measured | Tendon, gastric mucosa, hindlimb vasculature [6][5][4] | Skin and hair [11][8][10][12] | Brain after experimental stroke; muscle; actin in vitro [14][17] |
| Route studied | Intramuscular, intragastric, intravenous [5][3][1] | Topical, with penetration measured ex vivo [12][8] | Intraperitoneal in rats; intravenous in the human safety study [14][16] |
| Human data in this index | Two-person IV safety pilot; review counting three pilots total [1][2] | Four human sources incl. one randomised trial [11][8][10][12] | One Phase 1 safety study of the parent protein, n=40 [16] |
| Neural evidence in this index | One review of rodent brain-gut work [7] | None; neurons appear on a target list only | One rat stroke study, parent protein [14]; CNS repair named as a trial rationale [15] |
| Peripheral nerve evidence | None | None | None |
| Regulatory status | Not approved anywhere; FDA flagged it as not eligible for 503A compounding in 2023 | Topical Copper Tripeptide-1 is a legal cosmetic; injectable is unapproved | No approved indication; research and veterinary contexts |
| Anti-doping status | Prohibited at all times (non-approved substances) | Not on the 2024–2025 Prohibited Lists; catch-all category can still apply | Prohibited (peptide and growth-factor categories), with published detection methods |
| The caution that matters most | Human evidence is extremely thin and largely single-group [2] | Injectable use is unapproved and unstudied in people | The fragment is not the protein that produced the results [14][16] |
Three different bets on the same problem
Mechanistically these are not variations on one idea. They are three separate wagers about what limits repair.
BPC-157's wager is blood supply. Its best-characterised action is angiogenesis through up-regulation and internalisation of VEGFR2 with downstream Akt-eNOS signalling, shown in a chick chorioallantoic membrane, cultured human endothelial cells and a rat hindlimb with its circulation cut off [4]. The implicit claim is that tissue rebuilds when it is perfused.
GHK-Cu's wager is the matrix and the copper needed to build it. It stimulates fibroblast synthesis of collagen, elastin, glycosaminoglycans and decorin while rebalancing matrix metalloproteinases against their inhibitors, and the bound copper enables lysyl oxidase cross-linking and superoxide-dismutase-like antioxidant activity [11]. The implicit claim is that tissue rebuilds when the scaffold and its cofactors are present.
TB-500's wager — strictly, its parent protein's wager — is cell movement. Thymosin beta-4 sequesters monomeric actin one-to-one by capping both ends of the monomer [17], buffering the pool that cells draw on to migrate; the downstream account adds anti-scarring, anti-apoptotic and pro-angiogenic effects [15]. The implicit claim is that tissue rebuilds when the right cells can reach the injury.
All three wagers are plausible for nerve. None has been tested there in this index. And the three do not converge on a shared mechanism the way a marketing page tends to imply — they are competing emphases with different evidence behind each.
Counting the steps to a human nerve
Using the four substitutions from the home page — species, molecule, tissue, route — the three sort cleanly.
GHK-Cu: one substitution for skin claims, four for nerve claims. For a topical skin outcome it barely substitutes at all, which is why its evidence is the sturdiest here [11][10]. For a nerve claim it substitutes tissue completely, substitutes route (topical to systemic), and adds the delivery problem that a molecule with a clogP of -2.24 barely crosses intact skin [8].
BPC-157: three substitutions. Species throughout, tissue whenever tendon or ulcer results are read as neural ones, route because no human pharmacokinetic data exist. Molecule is the one axis it does not substitute, and its neural signal is a rodent neurotransmitter finding in a single-group review [7] rather than a repair outcome.
TB-500: four substitutions, all at once. Species, molecule, tissue and route are each substituted in the stroke study that anchors its neural reputation [14].
The ranking is uncomfortable and worth stating plainly: the compound with the most neural-sounding evidence is the one whose evidence transfers least, and the compound with the best human evidence has nothing to say about nerve at all.
Evidence maturity: one Phase 1, one small randomised trial, and a great many rats
Adding up every human dataset in this hub gives a short list. One randomised placebo-controlled Phase 1 safety study of full-length thymosin beta-4 in 40 healthy volunteers [16]. One six-month randomised hair-growth trial in 45 men of a 5-ALA and GHK combination [10]. A set of small topical dermatology studies and reviews behind the 70% procollagen figure [11][8]. An ex vivo human skin penetration study [12]. A two-person intravenous safety pilot of BPC157 [1]. And a 2025 review reporting that only three pilot studies of BPC-157 in humans exist at all, with rigorous large-scale trials lacking [2].
Not one of those measured a neurological outcome. The Phase 1 study measured tolerability and pharmacokinetics; the rest measured skin, hair or safety biomarkers.
Against that, the preclinical file is broad — rodent tendon, rodent ulcer, rodent ischaemia, rodent stroke, cell culture, structural biology. A 2026 sports-medicine review of unapproved peptides for musculoskeletal injury describes exactly this shape: favourable tissue-repair outcomes in animal models, scarce human safety data, real potential for harm, and little regulatory oversight [13]. That sentence covers this hub better than any single study in it.
Where the regulatory and anti-doping lines fall
The three diverge sharply here, and the divergence is often flattened in general "peptide" writing.
BPC-157 is not approved as a medicine in any jurisdiction, and in 2023 the FDA placed it among bulk substances identified as not eligible for pharmacy compounding under 503A pending further evaluation. It is prohibited in sport at all times under WADA's non-approved-substances category.
GHK-Cu is the split case. Topical Copper Tripeptide-1 is a legal cosmetic ingredient in the US, EU and UK with a long consumer safety record. Injectable or oral systemic GHK-Cu is an unapproved research chemical with no regulatory pathway and no validated human pharmacokinetics. It was not listed on the 2024–2025 WADA Prohibited Lists, though the catch-all category for non-approved pharmacological substances can still reach it, which is worth verifying against the current list rather than assuming.
TB-500 has no approved therapeutic indication, is handled as a research chemical with a veterinary history, and is prohibited by WADA under the peptide and growth-factor categories, with published laboratory detection methods for it and its breakdown products. It is a prescription medicine in some jurisdictions.
One consequence is shared: for the two unregulated compounds, identity and purity are unverified outside formal studies, which is an uncertainty stacked on top of every scientific uncertainty already listed.
What would actually change the picture
It is worth being concrete about what evidence would move any of these three from inference to demonstration on the nerve question, since none of it exists yet.
For TB-500, the first requirement is not a human trial at all — it is a direct comparison of the seven-amino-acid fragment against full-length thymosin beta-4 in the same model, which would settle whether the borrowed results are transferable in principle. Without that, every result attributed to TB-500 remains attributed to a different molecule.
For BPC-157, the missing piece is a peripheral nerve injury model with a measured regeneration outcome, plus independent replication outside the originating group — the replication problem is flagged in the literature itself [2].
For GHK-Cu, the question does not really begin until there is human pharmacokinetic data for a non-topical route, since the compound's documented behaviour is dominated by a delivery barrier [8][12].
And for all three, the honest sequence runs animal nerve model, then human safety, then human efficacy — a sequence none of them has started. Until then, what this hub contains is a well-evidenced skin compound, a well-evidenced rodent repair compound, and a fragment of a protein with an interesting stroke result attached to the wrong molecule.