The One-Lab Problem

When 200 papers trace back to a single research group, you don't have a literature — you have a brochure. BPC-157 is the case study; the detection skill is the takeaway.

Here is a situation modern science keeps producing: a compound with hundreds of published papers, glowing results across dozens of conditions — and nearly all of it written by the same small group of people who discovered it. The papers are real. The journals are real. The peer review happened. What is missing is the one thing that turns a claim into knowledge: somebody else, somewhere else, getting the same result.

That is the one-lab problem. This page explains how to detect it, using the best-documented case in the peptide world — BPC-157 — as the worked example. The skill generalizes: apply it to any compound, supplement, or therapy whose evidence base turns out to have a single return address.

The case study: BPC-157’s literature, counted

BPC-157 is a synthetic peptide modeled on a protective protein found in human gastric juice. It has been credited, in print, with effects on gastric ulcers, heart damage, liver damage, bone fractures, burns, wounds, nerve injuries, and dozens more conditions. The volume is the first thing sellers cite: over 200 papers can’t be wrong.

Counted properly, the volume is the problem:

Authorship concentration. Józwiak and colleagues, in a 2025 literature and patent review in Pharmaceuticals, searched PubMed (May 2025) and retrieved more than 190 BPC-157 articles — over 80% of them listing Predrag Sikirić or Sven Seiwerth as first or senior author. Sikirić leads the University of Zagreb group that created the compound. An independent 2026 evidence review counted 227 PubMed-indexed papers, 175 of them (77%) sharing Sikirić as an author. The Zagreb group disputes the critics’ framing — the exchange ran in the same journal in 2025 — but the authorship arithmetic itself is not in dispute.
Animal, not human, evidence. Vasireddi and colleagues’ 2025 systematic review in the HSS Journal screened 544 articles; 36 met inclusion; 35 were preclinical and one was clinical — a retrospective, uncontrolled series of 12 knee-pain patients.
The whole human record. Three small uncontrolled reports, roughly 30 people in total: the knee series, a two-person IV safety pilot, and a 12-person bladder-pain series, all from private clinical practice (McGuire et al., 2025).
Independent replication: thin but real. The most-cited independent line comes from Chang Gung University in Taiwan — Chang and colleagues on tendon-healing pathways, and Hsieh and colleagues (2017) showing BPC-157 activates the VEGFR2 blood-vessel-growth pathway in endothelial cells. That replication matters; it is also a handful of papers against a Zagreb ocean.

As AsapSCIENCE put it in its June 2026 video — the sharpest line in mainstream coverage of this compound — when you ask an AI chatbot about BPC-157, all roads lead back to this one lab. AI summaries inherit the concentration. So do review articles, clinic websites, and this paragraph’s sources, which is why we publish the counts instead of the vibes.

The timeline — and the stakes nobody disclosed

Single-source science is a yellow flag on its own. What turned this case from a methodological caution into a story is the paper trail around it:

1989 — The Zagreb group, after a decade isolating the parent “body protection compound” from gastric juice, files the patent: priority date September 12, 1989; U.S. Patent 5,288,708, granted February 22, 1994.
1993 — The landmark paper: Sikirić et al., “A new gastric juice peptide, BPC,” Journal de Physiologie (Paris). Croatian pharma company Pliva licenses the compound and develops it as PL-14736 for ulcerative colitis.
2006–2009 — GSK acquires Pliva’s research arm and drops the program; rights revert.
~2011 — The molecule patent’s term runs out. With no exclusivity left, no sponsor will fund the large trials that could settle the question — a 2026 Pharmaceutics review names the missing exclusivity as a significant disincentive to industry-funded development.
2015 — The only registered human trial, NCT02637284, a Phase I safety study sponsored by PharmaCotherapia, begins with about 42 healthy volunteers. Its ClinicalTrials.gov record is never updated after December 2015. No results are ever posted.
February 2026 — STAT News and Undark report what the papers didn’t: Sikirić part-owns PharmaCotherapia — the trial’s sponsor — and is CEO of Diagen, a company holding a live formulation patent on a stable BPC-157 salt that it advertises for sale. The reporters found none of these interests disclosed in the papers they reviewed. They also reported that trial data was submitted and then withdrawn before outside review.

None of this proves the compound doesn’t work. It proves something more basic: the institution producing the evidence had undisclosed reasons to produce it, the one experiment that could have checked it independently went dark, and the discoverer’s financial story was told by journalists, not by the papers. That is the complete shape of a one-lab problem.

Why single-lab evidence is fragile

Nobody has failed to replicate it — because almost nobody has tried. A result replicated only by its originators has not been tested against different hands, different equipment, different animal colonies, different unconscious biases. Science’s error-correction machinery is replication; a literature without it is a first draft, however many pages long.
One lab’s methods become the field’s “facts.” Preparation, models, endpoints, and what counts as a positive result are all set by the same group. Even honest researchers drift toward the protocols that produce publishable outcomes.
Citation volume masquerades as consensus. Two hundred papers citing each other look, from a search engine or a chatbot, like two hundred confirmations. Authorship concentration converts one group’s output into the appearance of a field’s agreement.
Conflicts compound quietly. Undisclosed stakes don’t have to corrupt any single experiment. They only have to shape which experiments get run, which results get submitted, and which get withdrawn — decisions no reader can see from the published record.

The reader’s checklist

For any compound or therapy, before the paper count impresses you:

Who wrote it? Search the author names across the top results. If one surname appears on most of them, you are reading a group, not a field.
Who else got the result? Look for the flagship finding replicated by a group with no shared authors. One independent replication outweighs fifty same-lab sequels.
Where are the humans? Count human studies separately from animal studies, and controlled studies separately from case series. “36 studies” that are 35 animal experiments and one uncontrolled series is a different fact than the number suggests.
Is there a registered trial — and did it report? A trial that was registered, begun, and never reported is information. Absence of results is not absence of evidence; it is evidence of abandonment.
Who profits, and is it disclosed? Check whether the evidence-producers hold equity, patents, or executive roles connected to the compound — and whether the papers say so.
What expired? Patent status explains silences. A compound nobody can own is a compound nobody will pay to prove.

The bottom line

When every road leads back to one lab, you don’t have a literature — you have a brochure, peer-reviewed and paginated, but a brochure. BPC-157 may yet prove to be a real drug; the Taiwan replication line is genuinely interesting, and the FDA advisory committee’s 8–6 vote in July 2026 shows credentialed people on both sides. But as of November 2026, the human evidence for it would fit in a small room, and the animal evidence has one return address in Zagreb.

Hold every compound on this site to the same standard. We do.

The One-Lab Problem

The case study: BPC-157’s literature, counted

The timeline — and the stakes nobody disclosed

Why single-lab evidence is fragile

The reader’s checklist

The bottom line

Read next

Educational information only

Before you act on anything here

THE NETWORK — ONE PUBLISHER, EVERY FRONT