UCLA reviewed 565 peptide studies. Two-thirds were in animals.

A team at UCLA read 565 studies on the six compounds most heavily marketed for recovery and muscle: BPC-157, TB-500, CJC-1295, MK-677, ipamorelin and GHK-Cu. More than two-thirds of those studies were in animals only. The rest, the human studies, were mostly small, mostly uncontrolled, and in the review's words showed "modest improvements at best" outside orthopedics and "essentially no" significant clinical benefit inside it.

Six peptides, 565 papers, and the mice are still doing most of the work.

What they found

The review was published September 2, 2026 in the American Journal of Sports Medicine under the title "Peptide Supplements and Their Therapeutic Applications in Sports Medicine", by Kushagra Tewari, a third-year medical student, and Thomas Kremen, a sports medicine surgeon and associate professor of orthopedic surgery at the David Geffen School of Medicine, with colleagues. Kremen's one-line summary: "The evidence that these peptides are effective is just not there."

Four findings stand out.

The evidence base is mostly rodents. More than two-thirds of the 565 studies never involved a human. That matters because the peptides in question are marketed for human tendons, ligaments, joints and muscle, and none of those heal in a rat the way they heal in a person.

The human studies that exist are weak. A handful showed modest effects for non-orthopedic uses. For musculoskeletal conditions, the marketing's home turf, the reviewers found essentially nothing demonstrating significant clinical benefit. Studies with positive results, Kremen said, "were done very poorly," lacking controls and rigorous design.

The largest BPC-157 study had a conflict of interest. The group that ran it had a financial interest in the compound itself. That doesn't make its result wrong, but it does mean the single biggest data point in the compound's favor was generated by people who benefit from a favorable result.

One trial was stopped for safety. A clinical trial of MK-677 was halted after one group showed elevated rates of congestive heart failure. Kremen also described a patient who came in for a rotator cuff repair, disclosed peptide use, and developed dangerously low cardiac output. MK-677, incidentally, is not a peptide at all; it is a small molecule that stimulates growth hormone release and gets sold alongside peptides because it is marketed to the same buyers.

Why animal studies don't settle it

A rodent study can show that a molecule does something to tissue. It cannot show that the same molecule, in a person, at a concentration a person's body would see, produces a benefit that outweighs its risks. The gaps are structural. Rodent tendons and ligaments are loaded differently and heal on a different timeline. Many animal studies apply the compound directly to an injury site rather than systemically. Sample sizes are small, blinding is rare, and a positive result in a mouse model is far more likely to be published than a null one.

That is why medicine draws the line where it does: an approved drug has cleared randomized, controlled human trials with a pre-registered outcome. Stanford's Katrin Svensson and Jonathan Long, writing in August, put the count at roughly 100 peptide therapeutics approved by the FDA since insulin. None of the six compounds in the UCLA review is on that list. They are, quite literally, research compounds: the human question is still open.

What is known, and what isn't

Known: the size and shape of the literature, which is the review's real contribution. Anyone can now say, with a citation, that two-thirds of the evidence for these six compounds is preclinical and the clinical third is thin. Also known: at least one safety signal serious enough to stop a trial.

Unknown: whether well-designed human trials would find anything. Absence of good evidence is not evidence of absence, and the review does not claim otherwise. Also unknown, and arguably more urgent, is what is actually in the vials being sold. The UCLA authors note that unregulated products may contain contaminants such as arsenic or lead, are often produced outside the United States from unvetted ingredients, and have contents the buyer cannot verify. A study result is only as good as the identity and purity of the material it was run on, and for most of the 565 papers that is not something a reader can check.

What this means for researchers

The review is, read one way, a verdict on marketing. Read another way, it is a to-do list. The open questions are specific and answerable: controlled human data, defined outcomes, independent funding, and material whose identity and purity are documented rather than assumed. A field with 565 papers and almost no good human trials is a field with room in it.

It also raises the bar on the material itself. If the biggest unknown in a compound's literature is what was actually in the vial, then a lot-specific certificate of analysis from an independent laboratory is not paperwork; it is the first data point of the study. Begin there, and read it, rather than filing it.

And it sharpens a distinction worth keeping: what a compound has been studied for is a fact about the literature. What it does in people is a fact nobody has established yet. The journal will keep reporting the first without pretending it is the second.