Research
How to read a study cited on a peptide page
A citation under a product claim is doing one of two jobs: supporting it, or decorating it. Six checks separate them, and each one takes under a minute on the abstract alone.
First check: was the compound given, or measured?
This single question disqualifies more citations in the peptide market than anything else, and almost nobody asks it.
A study can administer a peptide to a subject and report what happened. Or it can measure how much of that peptide a subject's own body produced, inside a study of something else entirely. Both papers show up under the same search term, and only the first one supports a product.
MOTS-c is the clean example. A September 2026 search returns four records tagged as randomized controlled trials. Reading all four shows what they studied: repeated heat stress during calf muscle immobilization, circulating levels in breast cancer patients treated with metformin, aerobic and resistance exercise in breast cancer survivors, and acute endurance exercise.
Every one measures circulating MOTS-c. None gives it to anyone. A seller citing "four randomized controlled trials" would be quoting a real count from a real database. The count is right and the implication is false.
Second check: which species, and beware the animal filter
The abstract usually says. Look for mice, rats, cells, cell lines, or a named tissue preparation. Words like "in vitro" mean a dish, and "in vivo" only means a living organism, not a human one.
The database filters are less helpful than they look, and one of them is actively misleading. In the National Library of Medicine's subject vocabulary, the term Humans sits underneath Animals in the hierarchy. Its tree position runs from Animals through Chordata, Vertebrates, Mammals, Primates and Hominidae down to Humans.
Because the search explodes a term to include everything beneath it, filtering for animals returns human papers too. Measured in one September 2026 session, aspirin returns 52,378 records under the animal tag and only 6,301 once human-tagged records are excluded. Almost nine in ten hits on the animal filter were also tagged human.
The fix is to exclude the human tag explicitly rather than trusting the animal one. For BPC-157, the same session returns 217 records, 151 under the animal tag, and 102 once human-tagged records are removed.
Third check: the human tag is not evidence of human dosing
The subject tag for humans marks a paper as involving human subjects or human material. Cell lines derived from people carry it. So do tissue samples, blood samples and post-mortem specimens.
The gap this creates is large. In the same September 2026 session, BPC-157 returns 49 records carrying the human tag and zero records tagged as a randomized controlled trial or as a clinical trial of any kind.
Forty-nine human-tagged papers and no trials is not a contradiction. It is a description of a literature made of laboratory work on human material.
Fourth check: publication type is an index label, not a quality score
Records carry publication-type tags assigned during indexing. Randomized controlled trial, clinical trial, review, case report, comparative study.
The tag describes what kind of article it is. It says nothing about how large the study was, how well it was run, whether it was blinded, or whether its result was positive. A four-person crossover and a ten-thousand-person trial carry the same tag.
The tag is also assigned to the paper, not to the intervention you care about, which is how the MOTS-c records above came to be tagged as randomized controlled trials while giving nobody any MOTS-c.
Use the tag to narrow a search. Do not use it to end an argument.
Fifth check: the route and the form have to match
A study of a topical cream does not support an injection. A study of an oral capsule does not support either. Peptides in particular behave completely differently by route, because absorption and degradation are route-specific problems.
The regulatory record for this market makes the point repeatedly. When KPV was evaluated in July 2026, the product form put forward was a topical cream and gel, while the market sells oral, injectable, topical and nasal versions. Injection is where the immunogenicity question lives, and it was not the form assessed.
The form of the substance matters too. Many peptides are sold either as a free base or as an acetate salt, and those are different active ingredients with different chemical registry numbers and different molecular weights. A paper on one is not automatically a paper on the other.
Sixth check: the molecule cited has to be the molecule sold
This is the failure that hides best, because the names look related and often are.
Thymosin beta-4 is a protein of 43 amino acids that has been through human trials. TB-500 is a seven-amino-acid fragment of it. They are cited interchangeably on product pages and they are not the same molecule.
Epitalon has the same problem from the other direction. Searching the name surfaces controlled trials of Epithalamin, a pineal gland extract, rather than the four-amino-acid synthetic sold under the Epitalon name.
Open the paper and read what it says it studied, in its own words, in the methods. A title is a summary and summaries lose exactly this distinction.
When a page cites a count instead of a study
Sometimes the claim is a number: dozens of studies, hundreds of publications, a large body of research. Counts are checkable, and the check has one rule.
A search that returns zero has to be run alongside a term you know returns a large number, through the identical filter, in the same session. Without that, a zero is ambiguous between "nothing has been published" and "the search was broken".
That failure is not hypothetical. A commonly copied filter string for clinical trials is invalid and silently returns zero for every term entered, aspirin included. A count published without a control beside it has not been checked, it has been performed.
Every census on this site prints its control next to its zero for that reason.
What a well-supported claim looks like
It names a study that gave the compound to people, by the route being sold, in the form being sold, and reported an outcome someone cares about rather than a laboratory marker.
It links the record rather than describing it, so the design, the size and the result can be read directly.
It does not lean on a count, a database tag, or a paper about a different molecule with a similar name.
Most claims in this market fail that test, and failing it does not make a compound harmful. It makes the claim unsupported, which is a smaller and more useful conclusion.
Key takeaways
- Ask first whether a study administered the compound or merely measured the body's own levels of it.
- All four randomized-trial records for MOTS-c measure endogenous levels inside trials of something else.
- The human subject tag covers cell lines and tissue, so it is not evidence that anyone was dosed.
- Humans is indexed beneath Animals, so an animal filter returns human papers unless they are excluded.
- Publication type is an index label describing article kind, not a measure of size, rigor or result.
- A zero result means nothing without a known-positive control run through the identical filter.
Frequently asked questions
How can a peptide have randomized controlled trials and no evidence behind it?
Because a trial can measure a peptide without administering it. All four randomized-controlled-trial records returned for MOTS-c in a September 2026 search study heat stress, metformin or exercise, and measure how much MOTS-c the participants' own bodies released. Reading the four records shows it immediately; counting them does not. Ask what was given to whom before asking how many studies exist.
Does a paper tagged "humans" mean the compound was given to people?
No. That subject tag marks a record as involving human subjects or human material, which includes cell lines, tissue samples and blood samples. In one September 2026 session, BPC-157 returned 49 human-tagged records and zero records tagged as any kind of clinical trial. The tag describes what the work involved, not that anyone was dosed.
Is filtering a search for animal studies a reliable way to exclude them?
No, and it fails in the surprising direction. In the National Library of Medicine's vocabulary, Humans is indexed underneath Animals, so the animal filter returns human papers as well. Aspirin returned 52,378 animal-tagged records in one September 2026 session and 6,301 once human-tagged records were excluded. Exclude the human tag explicitly rather than trusting the animal one.
What does "in vitro" actually rule out?
It means the work was done outside a living organism, typically in cultured cells or isolated tissue. Such a study can establish that a molecule binds a receptor or changes a signal in a dish. It cannot establish that an injection reaches the tissue, survives long enough to act, does anything a person would notice, or is safe. Those are separate questions and each needs its own study.
Why does the free base or acetate distinction matter for a citation?
Because they are different active ingredients. A free base and an acetate salt of the same peptide carry different chemical registry numbers and different molecular weights, and regulators evaluate them separately. A study on one form is evidence about that form. It becomes evidence about the other only if someone shows the two behave the same way, which is itself a study.
Sources
Each document below is named as it names itself, with the date printed on that document rather than the day it was read.
- MeSH descriptor record for Humans (D006801), showing its tree position beneath Animals — National Library of Medicine, September 2026
- PubMed literature census for MOTS-c, BPC-157 and aspirin, run through identical filters in one session — National Library of Medicine, September 2026
- Record summaries for the four randomized-controlled-trial records returned for MOTS-c — National Library of Medicine, September 2026
- FDA Briefing Document for KPV-Related Bulk Drug Substances, Pharmacy Compounding Advisory Committee — U.S. Food and Drug Administration, May 2026
- FDA Briefing Document for TB-500-Related Bulk Drug Substances, Pharmacy Compounding Advisory Committee — U.S. Food and Drug Administration, May 2026