Six weeks after researchers stopped injecting GHK-Cu into rat knees, the measured advantage was gone.
The study was an ACL reconstruction model at the Chinese University of Hong Kong (n=72, J Orthop Res 2015, PMID 25731775). Rats received saline, 0.3 mg/mL GHK-Cu, or 3 mg/mL, injected into the joint once a week for four weeks starting at post-op week two. At six weeks the GHK-Cu groups showed a smaller side-to-side laxity difference than saline (p=0.009), and the 0.3 mg/mL group specifically showed a stiffer graft complex (p=0.026). Ultimate load, gait parameters and histology scores never separated between groups at any timepoint. By twelve weeks, six weeks after the last injection, the early laxity advantage had closed. The authors’ own conclusion: “the beneficial effects could not last as treatment discontinued.”
There are two honest readings of that result and the study can’t distinguish them. One is that the effect faded when dosing stopped. The other is that treatment accelerated early graft maturation and the saline controls simply caught up, which is the expected trajectory for a self-limiting rat ACL graft model. Both produce identical twelve-week data. Anyone citing this paper as proof that peptide effects decay on withdrawal is choosing one reading and not telling you the other exists.
The signal moves cells rather than building tissue
The most-cited mechanism paper on BPC-157 came from Chang Gung University in Taiwan, an independent lab rather than the Zagreb group that produced most of this field (J Appl Physiol 2011, PMID 21030672). The researchers used rat Achilles tendon explants and isolated tendon fibroblasts. Cell proliferation on MTT assay was not affected at all. What changed was fibroblast migration and spreading, survival under hydrogen peroxide stress, F-actin formation, and dose-dependent phosphorylation of FAK and paxillin with total protein unchanged. The authors assigned the mechanism to FAK-paxillin activation, which is the motility machinery a cell uses to crawl.
In that model BPC-157 moved cells and supported their survival. It didn’t make more of them. Marketing copy built on “regrows tendon” is not describing what the primary paper measured.
Thymosin beta-4 approaches the same theme from a different angle. It binds actin, the internal scaffolding cells use to move, and in wound models fibroblasts and vessel cells reach the site faster. TB-500 is not thymosin beta-4. It is the seven amino acid actin-binding fragment (Ac-LKKTETQ, about 889 Da) cut out of the 43 amino acid, roughly 4,921 Da protein. It carries the movement function and none of the rest. GHK-Cu is a copper shuttle that hands copper to collagen-remodeling enzymes and switches on a broad remodeling gene program.
Each of these is a logistics signal rather than a building material. If that is what the mechanism amounts to, then a plateau once cells are in position and the remodeling window closes is what the mechanism predicts. That is an inference drawn from the mechanism papers rather than a demonstrated finding, and it is worth holding loosely.
The one time they were tested against each other
A Turkish orthopaedic group at Bagcilar Training and Research Hospital ran the only published head-to-head (Jt Dis Relat Surg 2026, PMID 42542926). The design used 32 male Sprague-Dawley rats in an Achilles transection and repair model, four groups of eight, with intraperitoneal dosing daily for four weeks.
The TB-500 arm at 60 µg/kg/day reached significance on maximum load to failure (p<0.05) and on both total Bonar (p=0.016) and total Movin (p=0.017) histology scores. The BPC-157 arm at 10 µg/kg/day was numerically better than control and did not reach significance on any total score. That inverts the widespread assumption that BPC-157 is the tendon compound.
The combination arm produced no additional benefit over either agent alone, which is worth weighing against the stacking protocols that circulate online. Collagen I expression didn’t differ between any group. Only collagen III moved.
The authors describe their own work as exploratory and preliminary, pending dose optimization and longer-term study. Eight animals per group, one timepoint, one laboratory. It should not be over-read in either direction, including the direction that flatters TB-500.
The compound with the most human data is the one that failed at scale
Thymosin beta-4 is the only one of these four to reach large randomized human trials, and the large trials were negative.
ARISE-2 (NCT02974907) enrolled 601 people with dry eye disease using RGN-259 0.1% eye drops. Both co-primary endpoints came out null and numerically favored placebo: ocular discomfort change of +0.07 for the peptide against −0.04 for placebo, and corneal fluorescein staining change of +0.07 against −0.01. ARISE-3 (NCT03937882, n=700) also missed its primary outcomes, with only the secondary symptom of ocular grittiness reaching significance.
The earlier figures that built enthusiasm for the program, a 35.1% discomfort reduction (p=0.0141) and a 59.1% staining reduction (p=0.0108), come from a company press release describing a nine-patient pilot rather than from a peer-reviewed report, and should be weighted as such. An intermediate Phase 2 with 72 patients (Clin Ophthalmol 2015, PMID 26056426) missed both co-primaries and salvaged only secondary endpoints. The 2010 venous stasis ulcer trial (n=73, Ann NY Acad Sci, PMID 20536470) was a dose-escalation safety study reporting that the 0.03% dose “may have the potential” to accelerate closure, which is not a powered efficacy result.
Small-n results in this field have a track record of shrinking when someone finally powers the trial properly. That pattern is the most useful thing to carry into every rodent study above.
Route, stability and the half-life problem
Nearly all the rodent tendon work used intraperitoneal injection, a route with no human equivalent in this context. The only registered orthopaedic trial uses subcutaneous administration, so the human data being generated will not be testing the route that produced the animal literature. That trial (NCT07437547, n=120 hamstring strain, started February 2026, estimated completion 2028) specifies once-daily subcutaneous dosing for 14 days and doesn’t state the dose in its public registry summary.
BPC-157 plasma half-life is under 30 minutes, reported in a two-species preclinical ADME study and a two-subject human pilot, with linear kinetics and intramuscular bioavailability of 14 to 51% by species (Pharmaceutics 2026;18:625, PMID 42198317). No published work reconciles a sub-30-minute half-life with effects claimed to persist for days.
Formulation is a larger problem than the dosing debate it usually gets buried under. The 2026 biopharmaceutics review puts total human BPC-157 exposure at fewer than 30 subjects across three uncontrolled pilots, “none of which employed standardized pharmaceutical preparations,” and notes that no pharmaceutical-grade formulation exists. FDA placed BPC-157 in Category 2 of the 503A bulks list in 2023, citing immunogenicity and peptide impurity concerns. Secondary sources report a change to that listing in April 2026, which isn’t confirmable against a primary FDA document and shouldn’t be relied on. Material sold for research use carries no established purity or content standard in any case.
KPV has a distinct stability problem. It degrades in solution, to the point that a 2021 group built a hydrogel delivery system specifically because “KPV solution is very unstable when rectally administered” (PMID 34547895). Encapsulated KPV in 400 nm particles matched free KPV at a concentration 12,000-fold lower (PMID 19909746), which indicates free peptide in solution is a poor delivery form.
What the studies actually dosed
Reported research parameters, none of which are human dosing recommendations and none of which come from a completed human efficacy trial in this indication:
BPC-157 in rats was reported effective at 10 µg/kg, 10 ng/kg and 10 pg/kg per day. The TB-500 fragment appears at 60 µg/kg/day intraperitoneally in a single rat study, with no published human dosing study of the fragment at any dose by any route. Full-length thymosin beta-4 in humans has been given at 0.05 to 25 µg/kg as a single IV dose (NCT04555824, n=54) and 0.5 to 5.0 µg/kg/day for 10 days (NCT04555850, n=30). GHK-Cu injectable data is rat intra-articular only. KPV appears as mouse oral administration in drinking water, commonly cited at 100 µM, with no human data at any dose.
Two things follow from that list. Figures circulating in consumer forums for TB-500 sit orders of magnitude above the highest dose ever administered to a human in a registered trial, and describe a different molecule from the one those trials used. And rodent studies report equivalent BPC-157 effects across a millionfold range, from 10 pg/kg to 10 µg/kg. A compound showing no dose gradient across six orders of magnitude either has biology nobody has characterized, or is producing measurements that are not reliable. In pharmacology that pattern is normally read as a warning rather than as evidence of a wide therapeutic window.
The studies that measured anything dosed through the active repair window and then stopped. Four weeks was the dosing period in the head-to-head. Endpoints in the 2006 rat Achilles work were read at days 1, 4, 7, 10, 14 and 21. The GHK-Cu protocol was weekly injections for four weeks beginning two weeks after surgery. Nothing in this literature examines open-ended continuous administration, and the one study that looked past its dosing window found the groups had converged.
Where the evidence is genuinely thin
BPC-157 has no completed, published, controlled human efficacy trial for any indication. The 2025 systematic review from Case Western and University Hospitals Cleveland (HSS J 21:485-495, PMID 40756949) screened 544 articles and included 36: thirty-five preclinical and one clinical. The single clinical entry was a retrospective, uncontrolled, unblinded case series of 12 patients with no comparison group. The review states verbatim: “No clinical safety data were found.”
The corpus is also concentrated. A PubMed search for “BPC-157” on 9 September 2026 returns 228 records, of which 175 list Sikiric at the University of Zagreb as an author, about 77%. Independent replication exists as a small minority, and in the independent 2026 tendon study BPC-157 was the arm that missed significance.
Nobody has tested whether BPC-157 affects tumor growth. It is a confirmed pro-angiogenic agent acting through VEGFR2 (J Mol Med 2017, PMID 27847966). Pairing BPC-157 with carcinogen, neoplasm and tumor terms in PubMed on 9 September 2026 returns 17 records, none of which is a tumor-growth or carcinogenicity experiment. The closest, PMID 29898649, is a cancer-cachexia proposal co-authored by Sikiric. Absence of reported harm in that set reflects absence of investigation rather than evidence of safety.
GHK-Cu human trial data is sparse and mixed. A 2006 facial study (n=13, Arch Facial Plast Surg, PMID 16847171) was negative on every objective endpoint, with only the patients’ own satisfaction questionnaire favoring it (p=0.04). An earlier and larger trial in venous stasis ulcers (n=86, J Vasc Surg 1992, PMID 1495150) compared copper tripeptide cream against placebo and silver sulfadiazine. Sources claiming GHK-Cu has only one human RCT haven’t found the 1992 work. The widely repeated claim that plasma GHK falls from about 200 ng/mL at age 20 to 80 ng/mL at 60 traces to review articles by Pickart, who founded a copper-peptide company. The primary measurement behind that figure is not locatable in the indexed literature.
KPV has no registered human study of any kind.
None of these four compounds has a completed human trial in tendon, ligament or muscle injury.
What to watch next
NCT07437547, the Phase 2 hamstring strain trial that began February 2026 with estimated completion in 2028, is the first properly controlled human orthopaedic test in this class. A rotator cuff pilot at the University of Arkansas (NCT07803250, n=30) is not yet recruiting, has an estimated 2027 start, and its registry record describes a no-treatment sham rather than a true placebo in one field while another field of the same record conflicts, so read the protocol rather than the summary.
Read the registry itself with care. NCT07487363 is titled as a Phase 1/2 placebo-controlled dose escalation of TB-500 in cardiovascular disease, and its own brief summary contains the line “This fictional study is an example of a ClinicalTrials.gov-style record.” The search listing looks entirely legitimate. Any source citing it as human TB-500 evidence hasn’t opened the record.
On competition status, BPC-157 was added to Section S0 of the WADA Prohibited List effective 2022, and no therapeutic use exemption is available for it.
This article is a review of published research. It is not medical advice, and nothing here is a recommendation to use any compound. These materials are sold for laboratory research use only and are not approved for human consumption.
