BPC-157 Was Built From a Protein Found in Human Stomach Fluid
The line you'll see repeated almost everywhere is that BPC-157 comes from human gastric juice. That's close to right. The way it isn't right is the interesting part, and it changes how you read almost every sentence written about the compound.
What the originating literature actually described
The claim starts with a protein, not a peptide. The originating literature describes a protective protein reported in human gastric juice, and calls it a body-protection compound. That phrase is where the initials come from.
BPC-157 is described as a partial sequence of that protein. Fifteen amino acids taken from a longer chain.
Both of those hedges are load-bearing, and our own write-up keeps them. The parent protein is what the originating literature reported. The relationship between it and the fifteen-residue peptide is how that same literature describes it. This is a compound whose origin story comes from one source rather than from a stack of independent ones.
The detail that gets skipped
Here's the part that almost never survives retelling.
The fifteen-residue molecule does not occur in nature in that form. It isn't extracted from anyone's stomach and it isn't purified from tissue. Every study ever run on it used synthetic material, made to order.
So the accurate description isn't "a peptide found in gastric juice." It's a designed fragment, cut from a protein that the originating literature reported finding in gastric juice. Those are meaningfully different statements, and only one of them is true.
That distinction matters most when you meet the phrase "naturally occurring," which follows this compound around. The parent protein is the natural part of the story. The molecule people actually study is not.
The nickname is the whole biography in three words
The compound is often called a stable gastric pentadecapeptide, and it's worth unpacking that, because each word is doing something.
Pentadecapeptide is the plain fact: fifteen amino acids. Gastric carries the origin claim. Stable carries an implicit claim about how it behaves.
It's an unusually efficient nickname, and like a lot of efficient nicknames it hardened into common knowledge before most people ever checked what it was asserting. Its identity record also carries other designations, PL-14736 and Bepecin, which is a reminder that this molecule has had more than one life as a named thing.
Where the research came from
The other half of the biography is geographical. A substantial part of the in vivo tissue-repair work traces to linked studies by researchers at the University of Zagreb, and it built up over roughly a decade.
Staresinic and colleagues started the thread in 2003 with a transected rat Achilles tendon model, reporting improved healing. Krivic and colleagues followed in 2006 with a different question in the same tissue, looking at tendon-to-bone healing across functional, biomechanical and histological measures. Cerovecki and colleagues extended the line again in 2010, this time to ligament, following medial collateral ligament healing in rats over ninety days.
Read as a research program rather than a pile of citations, that's a coherent decade of work. One tissue, then a harder problem in the same tissue, then a different tissue entirely.
The Krivic study is the one I'd point at. Alongside the healing question, it examined whether the peptide offset the healing impairment that a corticosteroid causes. Testing something against a known obstacle is a better-designed question than testing it against nothing, and it's the sort of choice that doesn't show up in a summary.
Our own review notes the flip side plainly: when foundational work comes from a small number of affiliated laboratories, independent replication becomes the open question rather than a formality.
Then other groups came asking why
The second phase of the biography is different in kind. Instead of more injury models, groups outside that circle started asking what the molecule was doing.
Chang and colleagues worked on rat tendon fibroblasts, reporting in 2011 on cell outgrowth, survival under oxidative stress and migration. They returned in 2014 to look at receptor expression at both the message and protein level. Hsieh and colleagues published in 2017 on a blood-vessel-formation pathway, combining a rat model with cultured human cells.
Whether-it-works studies and why-it-works studies are different projects, and a field that produces both is in better shape than one producing only the first. The full picture of that literature sets out what each of those studies used and what they reported.
Where the biography stands
A 2025 systematic review by Vasireddi and colleagues took stock of the orthopedic literature and counted thirty-six studies. Thirty-five were preclinical. The one clinical entry was a retrospective case series of twelve patients.
The compound has no approved indication anywhere, and it sits in the World Anti-Doping Agency's section for substances not approved for human therapeutic use, which is prohibited at all times.
What the field says it needs next
The literature is reasonably clear-eyed about its own gaps. The directions named in it are independent replication of the blood-vessel mechanism, better characterization of how the molecule holds up and how it is metabolized, and properly controlled clinical studies.
A field able to state its own agenda that specifically is in better shape than one that can't, even when the agenda is a long one. BPC-157 is also covered alongside TB-500 in a wider write-up of that regenerative class, where the two evidence bases sit next to each other.
That's a compound with two decades of animal research behind it and a name that sounds like it was discovered rather than assembled. Knowing which of those is which is most of what the origin story is good for. It sits in a regeneration range on that basis: as a research compound with a substantial preclinical file, and an origin more precisely described than the usual one-liner suggests.
BPC-157 is a research compound, supplied for laboratory research use only.