Which claims for seven FDA-reviewed peptides were tested in people?

Peptide evidence often arrives as a stack of references. The stack looks substantial. But the decisive question is not how many citations sit under a name. It is whether any of them tested the claim a reader is being asked to believe.

We audited seven peptide identities discussed at one public FDA advisory meeting in July 2026. Four had been given to people somewhere in the record. For two, we found no study in which researchers gave the peptide to a person. One had entered a recruiting trial but had no posted result.

None of the seven broad promises cleared the complete path in our audit.

That is not a verdict that every peptide is ineffective. It is a more useful finding: the name of a peptide is not the unit of evidence. The study must use the same substance and form, give it in the same way, study the relevant people and measure the result being claimed.

Why these seven—and no others?

This is not a popularity ranking. FDA reviewed all seven peptide pairs—free base and acetate—at the same July 2026 Pharmacy Compounding Advisory Committee meeting. That gives the comparison a fixed starting point without pretending that every peptide is the same kind of treatment.

The meeting did not approve a drug or add clinical evidence. It asked a compounding-policy question and produced non-binding advice. A separate Edge analysis explains what the 2026 FDA peptide hearing actually changed.

A peptide name can hide several different experiments

Consider what sits under the single label BPC-157. The human record includes a rectal enema in ulcerative colitis, injections into painful knees, an injection into the bladder, two short intravenous exposures and a commercial gummy study listed in a registry. A recruiting trial plans subcutaneous administration for acute hamstring strain.

Those are not seven confirmations of one idea. They involve different or uncertain forms, ways of giving the peptide, groups of people and outcomes.

The same fracture appears elsewhere. Human thymosin beta-4 trials do not become TB-500 evidence merely because TB-500 is a seven-amino-acid fragment of the larger peptide. Studies of human cells or cadaver skin do not become KPV administration in people. Human mortality findings for the pineal preparation epithalamin do not become evidence that synthetic Epitalon extends life.

Identity is not a footnote. It is the first gate.

Four peptides reached people—but not the advertised destination

BPC-157 has the most visibly active new program in this set: a randomized 120-person hamstring trial is recruiting. But a registry entry is a plan, not a result. The older human record contains two uncontrolled studies based on symptoms people reported themselves. A randomized ulcerative-colitis abstract reported an estimate too uncertain to rule out no difference between groups. None establishes general tissue repair or general safety.

Emideltide, or DSIP, was given intravenously in several tiny historical sleep studies. Some produced signals. The better placebo comparisons did not converge on a reliable, clinically meaningful insomnia benefit. No study tested the subcutaneous route FDA was asked to evaluate.

Semax has been studied in more people, largely as a nasal spray. But studies in healthy volunteers measured changes in brain networks—not better memory, attention or daily function. Stroke and pain reports ask different questions. They cannot supply the missing repeat of a healthy-cognition test.

The most direct Epitalon sleep study changed urinary 6-sulfatoxymelatonin—a melatonin metabolite—in a selected group over 20 days. It did not measure insomnia, sleep duration, awakenings, sleep quality or daytime function. A biomarker can sharpen a question about how something might work. It cannot replace the benefit being claimed.

Human administration matters. Matching human administration matters more.

Human biology is not automatically a human treatment result

MOTS-c creates the cleanest example. Studies have measured the body’s own MOTS-c in muscle or blood around exercise and metabolic health. That shows the peptide is part of human biology. It does not show what happens when researchers give someone a synthetic product.

A 120-person Phase 2a study in adults with prediabetes and overweight or obesity is now registered as recruiting. It plans to compare MOTS-c with placebo on insulin sensitivity and safety outcomes. Until results are posted, the trial tells us what researchers intend to test—not whether MOTS-c improves human metabolism.

KPV presents the inverse illusion. Its literature includes human-derived cell lines and excised human cadaver skin. The word human is accurate for the material. It is inaccurate for the intervention. FDA located no clinical trial, pharmacokinetic study or case report in which KPV was administered to a person.

Twenty-one records still do not make twenty-one confirmations

For the BPC-157 pilot, we split five reported programs in people and three current registry records into 21 exact claim-and-outcome rows. That prevents a study with several planned outcomes from masquerading as several positive findings.

The work comes from fewer independent groups than the paper count suggests. Three of the five reported programs include the same recurring author; the other two come from one older development group. We found no result that a separate team had repeated for the same claim, delivery method, group and outcome.

That does not erase an observation. It changes what the next study must do. Independent replication is not another paper from the same scientific network. It is a fresh test by another team—one that is capable of failing.

Registry status needs the same discipline. None of the three current BPC-157 registry records had posted results at the review date. One completed commercial gummy study was registered after completion; the other records were recruiting or of unknown status. All remain visible in the audit, and none is counted as an efficacy result.

What evidence would change the answer?

The audit is not designed to make uncertainty permanent. It makes the missing test explicit.

• Verify the exact peptide identity, form, formulation and finished product.
• Register a controlled study before outcomes are known.
• Test the route and population attached to the public claim.
• Measure a clinical outcome rather than an adjacent biomarker alone.
• Post complete results, including nulls, harms and protocol changes.
• Repeat the exact claim outside the originating research network.

A positive result that survives those steps should move the conclusion. So should a null result or a safety signal. The point is not to protect a skeptical position. It is to stop evidence from changing identity as it travels.

Sources and independence

The comparison uses FDA’s seven substance-specific briefing packages, original papers and current ClinicalTrials.gov records. Each peptide name above links to its complete molecule-specific LifeScore analysis; those pages retain their own question and source ledger rather than being duplicated here.

The public comparison and BPC-157 claim records are versioned downloads in the inspection layer below. LifeScore performed the synthesis independently. FDA, the cited researchers, sponsors, authors and publishers did not participate in or endorse this work.