BPC-157: A Deep Preclinical File and an Empty Clinical One

Evidence tier: Preclinical. No completed modern controlled human trial exists for any indication. Everything below describing efficacy comes from animal models or cell culture.

BPC-157 occupies an odd position in the peptide literature. It has more published preclinical work behind it than almost anything else in the recovery conversation — hundreds of papers, converging mechanisms, effects reproduced across independent laboratories over three decades. It also has essentially no human evidence. One small trial, two decades ago, in an indication most people discussing the compound have never heard of.

Both of those statements are true at once, and most writing about BPC-157 picks one and ignores the other. This article covers what the mechanism research shows, why the human file stayed empty, and how to weigh a compound where the science is genuinely interesting and the clinical validation genuinely absent.

What the molecule is

BPC stands for Body Protection Compound. BPC-157 is a synthetic 15-amino-acid fragment of a larger protein originally identified in the protective mucus lining of the human stomach. The sequence is Gly-Glu-Pro-Pro-Pro-Gly-Lys-Pro-Ala-Asp-Asp-Ala-Gly-Leu-Val. Predrag Sikirić's group at the University of Zagreb synthesised it in the 1990s and has driven most of the literature since.

It appears in some older papers as PL 14736 — the designation used when a pharmaceutical programme briefly pursued it for inflammatory bowel disease.

The property that makes it unusual is stability in gastric acid. Most peptides are cleaved by pepsin within minutes of reaching the stomach, which is why they're injected. BPC-157 retains activity after oral administration in animal work, which opens routes most peptides don't have.

Regulatory position: no marketing authorisation in the US, EU, UK, Canada or Australia for any indication. Following the FDA's 2023 review of nominated substances, it was placed in category 2 and not added to the 503A bulks list, which closed the legitimate compounding-pharmacy route in the US. The wider regulatory landscape is covered separately.

How it's proposed to work

The interesting thing about the mechanism literature is that it doesn't converge on one pathway. Several appear to operate in parallel, which is the most plausible explanation for why effects turn up across such unrelated tissue types.

Pathway	Proposed effect	Downstream consequence
VEGFR2 activation	Upregulates the receptor for vascular endothelial growth factor	Increased capillary density at injury sites
Nitric oxide modulation	Bidirectional — dampens excess, restores deficiency	Protection across opposite pathological states
Growth hormone receptor sensitivity	Upregulates GHR expression on tendon fibroblasts	Increased type I collagen deposition
Gut–brain axis	Modulates serotonergic and dopaminergic tone	Neuroprotective and behavioural effects in rodents

The angiogenesis arm

This is the most reproduced finding. Chang and colleagues showed BPC-157 activating the VEGFR2–Akt–eNOS loop in cultured endothelial cells without external VEGF ligand present — meaning the peptide appears to act at the receptor rather than by increasing the growth factor itself.

Why this matters for tendon specifically: tendons and ligaments heal slowly largely because they're poorly vascularised to begin with. Accelerating capillary ingrowth changes the substrate for everything downstream — oxygen delivery, nutrient turnover, waste clearance, immune cell trafficking. In tendon-injury models, capillary density rises within days, well before mechanical strength recovers. The sequence is consistent with vascularisation being upstream of the repair rather than a side effect of it.

The nitric oxide arm

BPC-157 behaves less like an agonist and more like a rheostat. In models of NO excess — endotoxin challenge, reperfusion injury — it dampens NO-mediated damage. In models of NO deficiency induced by L-NAME, it pushes signalling back toward baseline.

Bidirectional behaviour is unusual and slightly suspicious-sounding, but it's the most economical explanation for why protective effects show up in cardiovascular, gastrointestinal and musculoskeletal preparations that otherwise share no obvious mechanism.

The growth factor arm

Beyond VEGF signalling, the peptide increases growth hormone receptor expression on tendon fibroblasts, which potentiates the effect of the growth hormone already circulating. Related work describes upregulation of early growth response 1 (EGR-1) and its repressor NAB2 — transcription factors governing cytokine and extracellular matrix gene expression.

Collagen type I deposition, the load-bearing element of tendon and ligament matrix, is consistently elevated in treated animals against controls. Since functional tendon adaptation ultimately depends on collagen synthesis and cross-linking, this is the arm most directly relevant to the recovery use case.

The gut–brain arm

Because the fragment survives gastric acid, oral dosing produces measurable systemic effects — genuinely unusual for a peptide. Rodent work describes downstream modulation of serotonergic and dopaminergic systems, including counter-regulation of neuroleptic-induced catalepsy and reduction of amphetamine-induced hyperactivity.

Whether that reflects direct CNS activity, vagal signalling from the gut, or indirect effects via microbiome changes is unresolved.

What the studies actually tested

Nearly every BPC-157 paper shares a shape: controlled surgical injury in a rat or mouse, saline versus peptide, endpoints at fixed post-injury timepoints.

Tissue	Typical model	Endpoint improvement	Evidence strength
Tendon	Rat Achilles transection	Load-to-failure, capillary density	Strong preclinical
Ligament	Medial collateral transection	Functional recovery, collagen deposition	Moderate preclinical
Muscle	Crush injury, denervation	Reduced fibrosis, faster recovery	Moderate preclinical
Bone	Segmental defect	Accelerated osteogenesis	Limited preclinical
Gastrointestinal	Ulcer, colitis, anastomosis	Mucosal healing, lesion area	Strong preclinical + minimal human
Neural	TBI, stroke, spinal cord	Lesion volume, functional score	Early preclinical

On reading the effect sizes

Rat Achilles work reports recovery times roughly halved. That number gets quoted constantly and it deserves a heavy asterisk.

These are surgically transected tendons in young, genetically homogeneous, healthy rats. A clean cut in healthy tissue is a fundamentally different biological problem from a human chronic tendinopathy — which is degenerative rather than traumatic, has been accumulating for months or years, sits in a middle-aged body with worse baseline perfusion, and involves disorganised collagen rather than severed but otherwise normal tissue.

"Fifty per cent faster in a rat transection model" and "fifty per cent faster in a human with a nagging Achilles" are not the same claim, and the literature only supports the first.

The gastrointestinal exception

GI is where the file is deepest and where the only human data lives. The compound accelerates healing in gastric and duodenal ulcer models, TNBS- and DSS-induced colitis, and intestinal anastomosis. It's the original indication and the longest-running body of work.

It's also the one indication that reached humans: a small Phase 2 study of PL 14736 as an enema in mild-to-moderate ulcerative colitis, published in 2005. Results were modest. The programme did not advance.

Where the evidence stops

Claim	Preclinical	Human	Confidence
Accelerates tendon/ligament repair	Extensive, multi-lab	None	Moderate mechanistically, low translationally
Angiogenic via VEGFR2	Multiple in vitro and in vivo	None	High for mechanism, unknown for human relevance
Heals GI ulcers and colitis	Strong	One small Phase 2, modest	Low to moderate
Neuroprotective in TBI/stroke	Early, single-lab dominated	None	Very low
Safe for long-term human use	Short-duration animal data reassuring	Not established	Unknown

Two things worth pulling out of that table.

Mechanism confidence and efficacy confidence are separate questions. A well-characterised pathway is a hypothesis about what might happen in humans, not evidence that it does. The VEGFR2 work is genuinely solid as receptor biology. It says nothing on its own about whether a middle-aged person's shoulder gets better faster.

Single-lab dominance is a real limitation. A large fraction of the BPC-157 literature comes from one research group. That isn't an accusation — Sikirić's group has been rigorous and has published in credible journals for thirty years. But independent replication across labs is how findings become robust, and the neural work in particular hasn't had much of it.

Why the human file stayed empty

This is the part most coverage skips, and it's more interesting than the pharmacology.

BPC-157 is a naturally occurring sequence. In its native form it is essentially unpatentable. Phase 2 and Phase 3 trials cost tens to hundreds of millions of pounds, and that money comes from companies expecting exclusivity at the end of it. No exclusivity, no funding, no trials.

So the absence of human data isn't evidence that the compound failed in humans. Nobody has properly tried. That's a different situation from a drug that entered trials and disappointed, and it cuts both ways: there's no negative result to point at, and no positive one either.

It also means the gap is unlikely to close through normal commercial channels. If it closes at all, it'll be through academic or state-funded work.

Safety, and the honest unknowns

Across the preclinical literature the acute safety profile is favourable — no reproducible toxicity at doses well above those producing effects, no carcinogenicity signal in the short studies conducted. That is not a human safety profile, and it stands in sharp contrast to the extensive human trial data behind the GLP-1 class.

The unknowns worth naming:

Long-term human exposure is entirely undocumented. There is no dataset.

The angiogenesis question. A compound that promotes new blood vessel growth is, in principle, promoting something tumours also need. No preclinical work has demonstrated tumour promotion — but the question hasn't been rigorously tested in tumour-bearing models with long follow-up either. It's an open question, not a resolved concern in either direction, and anyone who tells you it's settled is overstating.

Interaction with common recovery drugs. Anticoagulants, corticosteroids and NSAIDs are exactly what people managing an injury are already taking. Interaction data doesn't exist.

Product identity. This is a supply-chain problem rather than a molecule problem, but it's the one most likely to bite. Independent testing across the research-peptide market has repeatedly found label mismatches — wrong purity, wrong concentration, occasionally wrong compound. Contradictory results in the informal literature are as likely to reflect what was actually in the vial as anything about the pharmacology.

Dosing conventions are extrapolations, not protocols

Figures circulating online typically sit in the 200–500 microgram per day range, sometimes split, sometimes injected near the injury site.

These come from rodent studies converted by body-surface-area scaling. That method produces rough allometric estimates for designing a first-in-human dose-escalation study. It does not produce a validated human dose, and it was never intended to.

No regulatory body anywhere has published a human dosing guideline for BPC-157. Route affects bioavailability in ways still being characterised. Anything presented online as a "protocol" is extrapolation dressed as prescription.

Where this sits in a recovery strategy

Recovery is a systems problem, and the variables with the deepest human evidence are also the ones most consistently neglected: sleep architecture, load management, protein intake, and progressive loading of the injured tissue.

That isn't a rhetorical hedge to seem balanced. It's a statement about effect sizes. Sleep restriction measurably impairs tissue repair in controlled human studies. Progressive loading is the only intervention with strong human evidence for tendon remodelling. These have larger, better-documented effects than anything in the peptide literature, and they're free.

A compound with promising preclinical data doesn't substitute for the inputs that are known to work. If the fundamentals aren't in place, the peptide question is premature regardless of what the rat data says.

The honest summary

The BPC-157 file is unusual. Most compounds with this little human evidence have little preclinical evidence either — they're simply undercharacterised. This one is the opposite: mechanistically well described, reproducible in animals, converging across pathways, and completely untested in a modern human trial.

Dismissing it as "just animal data" understates how much has been characterised. Treating it as a proven recovery therapy overstates it by considerably more. The gap between those positions is where the actual evidence sits, and it's uncomfortable precisely because it doesn't resolve into a clean verdict.

The useful posture is engaged scepticism: take the mechanism seriously, take the absence of human data equally seriously, treat online dosing protocols as extrapolation, and recognise that the reason there's no clinical file has more to do with patent law than with pharmacology.

What would change the picture

Five studies would move this from interesting to actionable:

Human pharmacokinetics. A basic PK study in healthy volunteers — half-life, distribution, metabolism — would close the largest single gap. Rat data suggests plasma clearance on the order of minutes, which raises the question of how durable tissue effects arise at all.

A dose-escalation study. Every dosing convention in use is scaled from rodents. A Phase 1 would establish an actual therapeutic window.

A chronic tendinopathy trial. The obvious indication given the preclinical footprint, in a condition where conventional treatment outcomes are genuinely poor.

Long-duration studies in tumour-bearing models. The angiogenesis question needs an answer rather than a caveat.

Independent replication of the neural work. The TBI and stroke findings are the most striking and the least replicated.

Frequently asked

---