BPC-157

BPC-157 is a synthetic 15-amino-acid fragment — GEPPPGKPADDAGLV — of a protein isolated from human gastric juice. It is unusual among peptides in being stable in gastric acid, which is why the animal work keeps reporting that swallowing it does what injecting it does. Its original indication was the gut, and it reached company-run human trials for inflammatory bowel disease in the 2000s under the drug code PL 14736. Almost nobody takes it for that now. It is taken for healing.

What BPC-157 is taken for

The healing literature is not one rat study. It is roughly 170 primary animal and in-vitro papers spanning tendon, ligament, muscle, the junctions between them, bone, peripheral nerve, spinal cord, skin, and the entire gastrointestinal tract, beginning in 1993 and still growing. The consistency of that record is both its strength and the thing most worth understanding about it, and both are covered below.

Four mechanisms carry the healing claims, and three were characterized by groups with no connection to the laboratory that discovered the peptide. The most interesting for anyone taking it for tendon is the growth hormone receptor: in rat Achilles tendon fibroblasts, a gene-expression screen found the growth hormone receptor among the most strongly upregulated genes after BPC-157 exposure, dose- and time-dependently at both message and protein level. Adding growth hormone to those cells then increased proliferation with the downstream JAK2 pathway switching on. That is a plausible route from a small peptide to tendon anabolism, and it came out of Chang Gung University in Taiwan.

The second is vascular. BPC-157 raises VEGFR2 — the receptor, not the growth factor — in human endothelial cells, drives it inside the cell, and activates the VEGFR2–Akt–eNOS cascade; block the internalization and the whole effect disappears, tube formation included. In isolated rat aorta the resulting vasodilation is endothelium-dependent and abolished by nitric-oxide synthase inhibition. In 2026 a Turkish group reproduced the pharmacology in human tissue for the first time, using artery segments left over from twelve coronary bypass operations.

The third is that the angiogenic effect is conditional on injury. Applied to cells in a dish, BPC-157 does not sprout vessels. Applied to crushed muscle or a transected tendon, it raises VEGF expression and vessel density. It appears to amplify a repair signal rather than generate one.

Tendon, ligament and the junctions

Transected rat Achilles tendon heals faster with BPC-157 given by injection, by mouth, or applied topically. Detached Achilles tendon reattaches to bone, and does so even when the animal is on corticosteroids. The cleanest of the tendon experiments looks at the first four days after tendon-to-bone transection — before collagen healing has started — and finds the Achilles functional index raised at every time point, with myeloperoxidase down and new vessel formation up. The comparison arm is what makes it useful: methylprednisolone matched the anti-inflammatory effect exactly, suppressed new vessels, and left function unchanged. Damping inflammation and restoring function came apart, and only one of the two agents did both.

Ligament work is thinner but points the same way. Rat medial collateral ligament transection healed better under BPC-157 given intraperitoneally, in drinking water, or applied to the skin over the joint, followed out to 90 days.

The junctions are where the peptide's claims are most distinctive, because growth factors delivered locally have generally failed there. A myotendinous junction surgically disabled in rats — a defect that does not close on its own — closed under either injected or drinking-water dosing, with eNOS and COX-2 expression and oxidative stress returning toward normal. Quadriceps detached from bone reattached under oral dosing alone; at three months the reattachment fibers could be traced into newly formed cortical bone at the surface, in animals whose untreated counterparts had permanent knee flexure.

Muscle

The strongest numbers in the whole corpus are in muscle. A rat quadriceps transected completely, one centimeter above the patella, leaves a defect that does not compensate. Under daily BPC-157 the healing improved across a full 72-day follow-up on four independent axes at once: load to failure rose, walking recovery and extensor postural thrust returned toward healthy values, desmin-positive fibers bridged the stumps with myofibril diameters reaching those of uninjured animals, and the post-surgical leg contracture that controls develop never appeared.

A standardized gastrocnemius crush injury improved by injection or by cream, with less hematoma and edema, no contracture, and lower creatine kinase, LDH, AST and ALT. When systemic methylprednisolone was added — which markedly worsened healing on its own — BPC-157 by either route restored function and histology at every interval measured.

Outside muscle proper, transected sciatic nerve regenerated faster on histomorphometry, on EMG, and on the sciatic functional index, with autotomy absent. A 60-second spinal-cord compression injury treated with a single intraperitoneal dose ten minutes later produced clinical improvement in every treated animal, spasticity resolved by day 15, and reduced tissue damage that held out to 360 days. Started four days late, on an established lesion, oral dosing still recovered tail function.

The gut, and NSAID and corticosteroid damage

This is the oldest and largest part of the file, and the part where independent laboratories are most present. The earliest of them was Parke-Davis in Ann Arbor in 1995, which found that intraperitoneal peptide reduced chemically induced colonic necrosis and neutrophil infiltration dose-dependently across five orders of magnitude — while the same dose delivered directly into the colon did nothing at all. Independent Chinese work later found 46 to 66 percent inhibition of gastric ulcer formation, with the intramuscular route effective at a lower dose than the swallowed one — though every comparison that paper tested was against a control rather than against its own famotidine arm or its other route, so the two head-to-heads its abstract claims were never run. Famotidine, dosed at fifty times the peptide, did better than every BPC-157 arm by the swallowed route in one of the three models.

Rats with most of the small bowel removed gained weight above their pre-operative values immediately under either oral or injected dosing, with taller villi, deeper crypts and stronger anastomoses, against controls that lost weight throughout.

The counteraction studies are the reason the peptide has a following among people who take anti-inflammatories. Diclofenac at high dose for three days produces gastric, intestinal and liver injury plus frank hepatic encephalopathy in rats; BPC-157 by injection or in drinking water antagonized the whole cluster. Celecoxib at 1 g/kg produced gastric, liver and brain lesions that BPC-157 counteracted at 24 and 48 hours whether or not nitric-oxide synthase was blocked — the point of that arm being that the peptide appears to modulate the nitric-oxide system rather than simply feed it. Corticosteroid-impaired burn healing in mice improved under topical peptide on both histology and tensiometry. These are toxicity models at doses far above anything a person takes, which is worth holding onto: they show the peptide protects against a chemical insult, not that it makes a daily ibuprofen safe.

Dose, route and how long it lasts

Almost the whole animal record uses 10 µg/kg, 10 ng/kg and sometimes 10 pg/kg intraperitoneally, once daily. Those three doses span six orders of magnitude and usually behave identically, which is a genuinely odd feature of the literature and one reason true dose-response curves are worth singling out. They exist in the colitis model, the gastric ulcer model — where the curve flattens above the middle dose — and a diabetic wound model, where topical peptide accelerated healing dose-dependently and matched the approved growth-factor product at the top doses — though in that same study, healing in non-diabetic animals was already too fast for either agent to improve.

Route has been tested directly and repeatedly, which is unusual for a gray-market compound. Drinking-water dosing reproduces injected effects in muscle-to-bone reattachment, the myotendinous junction, spinal cord, short bowel, and NSAID toxicity. Topical application works in muscle crush, corticosteroid-impaired muscle, burns and skin wounds. Where routes have failed, they failed cleanly: intracolonic delivery did nothing where intraperitoneal worked.

Pharmacokinetics are the weakest part of the file. The only formal study of absorption, distribution and elimination was run in rats and beagle dogs, and found the intact peptide's elimination half-life under 30 minutes with linear kinetics, intramuscular bioavailability of roughly 14 to 19 percent in rats and 45 to 51 percent in dogs, urinary and biliary excretion, and rapid breakdown into short fragments and free amino acids. A half-life measured in minutes against effects reported to last days is the central unexplained feature of the compound's pharmacology, and no human study of comparable design exists. What people actually inject is not derived from any of it: a 2026 orthopaedic review of the peptide literature reports BPC-157 given at 200 to 1000 mcg a day across regimens with no consensus on a therapeutic threshold, and asks for a minimum reporting standard before the field can compare one protocol with another.

Where the evidence comes from

A PubMed search on 15 August 2026 for BPC-157 and its synonyms returns 228 records. Predrag Sikirić or Sven Seiwerth of the University of Zagreb are co-authors on 176 of them, or 77 percent. Of the 173 records that are not reviews, 140 — 81 percent — carry one of those two names. That is the shape of the field, and it is a fact about who has done the work rather than a verdict on the work itself.

The remaining 33 primary records come from roughly a dozen groups: Chang Gung University in Taiwan, which produced the tendon-fibroblast and VEGFR2 mechanism work; the Fourth Military Medical and Air Force Medical Universities in Xi'an, which produced the gastric-ulcer, burn-wound and pharmacokinetic studies; Gazi University and Gülhane in Ankara, which produced the ischemia and human-artery work; Southeast University in Nanjing; Pusan National University; the University of Pécs; Parke-Davis in Ann Arbor; PLIVA Research Institute, the company that developed it; and the anti-doping laboratories in Salt Lake City, Shanghai and Ghent. Those groups cluster in mechanism, pharmacokinetics, gastroenterology and analytical chemistry.

They do not cluster in musculoskeletal healing. The tendon, muscle, junction and nerve models — the reason the compound is taken — have been attempted by exactly one group outside the originating network, and that attempt came out short of significance on load to failure. An independent Turkish orthopedic team ran BPC-157 against TB-500 and against both combined in rat Achilles healing: TB-500 significantly improved load to failure and total histology scores, BPC-157 fell short of significance on those totals while reaching it on a few individual subscores, the two peptides never differed from each other, and the combination added nothing over either alone. That is one small study, and one small study does not settle anything. It is also, at present, the entire independent in-vivo replication record for musculoskeletal BPC-157.

The cartilage question

Searching the same database for BPC-157 alongside cartilage, chondrocyte, osteoarthritis, chondral or meniscus returns three records, and all three are reviews or the knee case series. There is no primary study. No articular-cartilage repair model, no osteoarthritis model, no chondral defect model, and no rotator cuff or ACL model of this peptide has been published in thirty years. The one joint-disease animal model is adjuvant arthritis — an inflammatory model, not a degenerative one — in which BPC-157 attenuated both the development and the course of the disease.

The distinction that matters here is that this is unmeasured rather than measured and null. Nobody has run the experiment. A peptide that closes myotendinous junctions and reattaches muscle to bone is not obviously irrelevant to a joint, and it is not obviously relevant either; cartilage is avascular, and an angiogenesis-dependent mechanism has a specific reason to behave differently there. Both statements are speculation until somebody puts BPC-157 into a chondral defect.

The human record

Three uncontrolled pilot studies exist, together enrolling fewer than thirty people, and all three come from one US clinical group. The joint-relevant one is a retrospective series of sixteen patients given intra-articular BPC-157 for knee pain, twelve of them on the peptide alone. Its own authors report subjective improvement in eleven of those twelve. An independent systematic review that appraised the same paper summarizes it as seven of twelve reporting relief beyond six months, and a 2026 orthopaedic review reads it as seventeen patients with reduced symptoms in more than 90 percent. Three readings of one small series is a fact about the series: none is anchored to a validated outcome instrument, and there were no controls and no blinding. The second pilot injected 10 mg around the bladder wall in twelve women with interstitial cystitis who had failed pentosan polysulfate, with ten reporting complete symptom resolution on a single-item questionnaire. The third gave intravenous peptide to two people over two days and measured biomarkers.

Behind those sits an older and stranger fact. BPC-157 did go into company-run human trials as PL 14736, at Pliva in Croatia, for ulcerative colitis. A first-in-human safety and pharmacokinetic study in healthy volunteers was presented at a gastroenterology meeting, and a phase II enema trial in mild-to-moderate ulcerative colitis was run and reported in trade press and conference abstracts. Neither has ever been published in a peer-reviewed journal. Human trials were conducted and never reported, which is a different situation from human trials never having been conducted, and it is the reason careful reviews say "no completed phase II" while the developers' own papers say the peptide was "tested in clinical phase II".

Three trials are now on ClinicalTrials.gov. The one that matters is NCT07437547: a randomized, quadruple-blind, placebo-controlled phase 2 trial of daily subcutaneous BPC-157 for fourteen days in 120 people with an MRI-confirmed grade II hamstring strain, on top of standardized rehabilitation, with co-primary endpoints of time to unrestricted sport at eight weeks and change in MRI-assessed injury volume at day 14 under blinded central radiology review. It began enrolling on 2 February 2026 and estimates primary completion in February 2027. The other two are a 2015 phase 1 oral safety study in 42 volunteers that has sat at "unknown" status for a decade without reporting, and a manufacturer's single-arm study of BPC-157 gummies with no results posted.

Regulatory and anti-doping status

WADA prohibits BPC-157 at all times under S0, the non-approved substances category, and USADA states there is no realistic route to a therapeutic use exemption. Detection is not theoretical: a validated urine method reaches 0.1 ng/mL, the peptide forms a stable urinary metabolite, and it stays detectable for at least four days.

The FDA picture moved twice in 2026. The agency had placed BPC-157 in Category 2 of its 503A interim bulk-substances list in 2023, citing immunogenicity, impurity and characterization concerns. On 15 April 2026 it announced removal from Category 2 within seven days because the nominations had been withdrawn — a procedural step, not an authorization. Then on 23 July 2026 the FDA's Pharmacy Compounding Advisory Committee voted 8 to 6, with one abstention, to recommend adding BPC-157 to the 503A Bulk Drug Substances List. It did so against the written recommendation of FDA review staff, who had found insufficient evidence of effectiveness for the indication under review, described the peptide as not well characterized, and identified three adverse-event reports of unclear causality. The vote is non-binding and would require formal rulemaking to take effect. None of this makes BPC-157 an approved drug.

Safety

Human safety data amount to fewer than thirty subjects across three uncontrolled pilots, none using a standardized pharmaceutical preparation, all from one clinical group. The 2025 systematic review that screened 544 records found no clinical safety data at all. Preclinical toxicology reports no adverse effects across organ systems and the lowest lethal dose has not been reached, but long-term repeated-dose human safety is genuinely unknown.

Two specific signals belong on the record. The first is a drug interaction: in rats, BPC-157 restored platelet aggregation that aspirin, clopidogrel and cilostazol had suppressed, across three different aggregation agonists. The originating group frames this as rescuing platelet function. For anyone taking aspirin or clopidogrel for cardiovascular reasons it reads as reversal of the effect they are taking the drug for, and it has never been tested in humans. The second is the angiogenic mechanism and occult malignancy — a theoretical concern raised in a 2025 independent review, answered in a formal published comment by the originating group citing anti-tumor findings of its own, with no human data on either side. The 2026 orthopaedic review raises the same point for pentadecapeptide arginate, the stabilised salt form sold for shelf life and bioavailability, and calls it speculative for the same reason: the compound has also inhibited tumour growth in cell lines.

Gray-market vials carry the usual unregulated-product risks. Confiscated vials have been analyzed and did contain the peptide; identity in one seizure says nothing about identity, sterility or dose accuracy in any other. Animal studies used microgram-per-kilogram doses and no validated human dose exists. This entry is an appraisal of published research, not dosing guidance.

What would change the tier

The hamstring trial reports in 2027. A positive result on blinded, centrally read MRI injury volume would be the first controlled human evidence this compound has ever had, and it would move the healing case decisively — though it would still be a muscle result, not a cartilage one.

For the cartilage tier specifically, the shortest path is also the most obvious one: a chondral defect model. The peptide has been put into tendon, ligament, muscle, bone, nerve, spinal cord, cornea, skin, bladder and most of the digestive tract. Putting it into articular cartilage is a single experiment nobody has yet run, and it would settle in one study a question that thirty years of adjacent work has left open.