What the human evidence actually shows
An advisory committee recommended six of these peptides for compounding in July 2026. That vote was not a finding that any of them work. Here is what has actually been published in humans, drug by drug.
This is not an argument that they are dangerous. It is the more basic point that nobody knows, including the people selling them.
The summary table
| Peptide | Published human RCTs | Total published human exposure | Registered trials |
|---|---|---|---|
| BPC-157 | None | About 31 subjects across published reports, all uncontrolled | 4 registered, one dormant since 2015 |
| KPV | None | Zero — no human study of any design | None registered |
| TB-500 | None | Zero — never administered to a human in the published literature | None registered |
| MOTS-c | None | Zero administration studies; biomarker observation only | 1 recruiting |
| Epitalon | None with a clinical endpoint | One biomarker trial, 75 enrolled; one open cohort of 79 | None registered |
| Semax | None that could be verified | Two small open-label studies, both negative | None registered |
| Emideltide (DSIP) | Yes — three to four, from 1981 to 1992 | About 40 subjects, every one intravenous | None registered |
For scale: a generic tablet reaching a US pharmacy shelf has cleared bioequivalence testing. A new medicine typically carries several thousand subjects across its trial programme. The largest number in the exposure column above is about forty.
Drug by drug
BPC-157
Sold for: Tendon, ligament, muscle and gut healing; post-surgical recovery
The ulcerative colitis trial everyone cites is a 2005 conference abstract, never published in full. FDA reported its numbers: the between-group difference was 1.6 with a confidence interval of -4.84 to 1.62. That interval crosses zero, so the trial did not show a statistically significant benefit, and the development programme was abandoned. What is actually indexed is three papers in one journal, all first-authored by the same clinician, none randomised, controlled or blinded — including a safety pilot with two subjects. An independent 2025 systematic review screened 544 papers, included 36, and found 35 preclinical, 1 clinical, and zero randomised controlled trials.
Safety picture: No carcinogenicity studies exist at all. The proposed mechanism is pro-angiogenic, which is an unresolved proliferation concern no human data can currently exclude.
KPV
Sold for: Gut inflammation, IBD, acne and skin, wound healing
The emptiest file of the seven. FDA states plainly that it identified no human exposure data. The entire literature is in vitro, rodent, nanoparticle-delivery or analytical chemistry. A paper often cited as a human skin study used cadaver skin outside the body.
Safety picture: No human safety data of any kind, because there is no human data of any kind.
TB-500
Sold for: Injury and tendon repair
The critical distinction: TB-500 is a synthetic 7-amino-acid fragment. Thymosin beta-4 is the full-length 43-amino-acid protein. They are different molecules, and the real Thymosin beta-4 trials that exist were eye drops for dry eye and neurotrophic keratopathy — a different molecule, a different route and a different indication. None supports injected TB-500 for injury repair. A 2023 analysis of commercial TB-500 products found content not systematically consistent with the labelling.
Safety picture: No human safety data. Purity of commercial product is documented as unreliable.
MOTS-c
Sold for: Metabolic health, fat loss, exercise mimetic, longevity
The trap here catches a lot of people. Searching the literature returns roughly 142 records tagged human, but essentially all of them measure naturally circulating MOTS-c levels and correlate them with exercise, ageing or diabetes. That is not evidence that injecting MOTS-c does anything. The closest human dosing data involves a modified analogue, not MOTS-c itself, whose results were never posted to the trial registry and whose programme was not advanced.
Safety picture: No administration safety data. Injection-site reactions were common in the analogue trial.
Epitalon
Sold for: Longevity, telomeres, sleep
Two distinctions vendors collapse. Epithalamin is a crude bovine pineal extract from 1980s Soviet work; Epitalon is a synthetic tetrapeptide. Mortality data from the former is routinely used to sell the latter. The mortality study itself was open-label and unblinded with no placebo, and essentially the whole body of work comes from a single laboratory with no independent replication outside it. The one randomised placebo-controlled study measured urinary melatonin metabolites and clock-gene expression — biomarkers, not clinical outcomes.
Safety picture: No long-term safety data and no carcinogenicity data, for a compound marketed specifically as a telomerase activator — a theoretically pro-neoplastic mechanism. The telomere claim itself rests on a single cell-culture study.
Semax
Sold for: Cognition, stroke recovery, anxiety, ADHD — US vendors also claim ALS, Parkinson's and Alzheimer's
Semax is genuinely registered as a medicine in Russia and has been since 1994. It has never been submitted to the FDA or the EMA. Of the two usable human references, one was a meeting abstract with no clinical outcomes and the other found no change in pain, sensation or evoked potentials — its own authors concluded the drug does not exhibit analgesic activity by itself. A Russian clinical literature does exist, but FDA excluded it procedurally because no verified English translations were submitted. So neither 'proven in Russian trials' nor 'no Russian literature exists' is accurate.
Safety picture: Long Russian marketing history, but no Western safety review.
Emideltide (DSIP)
Sold for: Sleep, opioid withdrawal
The only one of the seven with real controlled trials, and they largely failed. The two most rigorous were negative: one concluded the effect was of little clinical significance, the other that it was not likely to be of major therapeutic benefit. Development was abandoned in the 1990s. The route problem is decisive on its own — every human study used intravenous administration, and it is sold today for subcutaneous injection, for which there is no human data at all.
Safety picture: This is the one the advisory committee voted down.
How to read a peptide claim without a pharmacology degree
Most of the confusion above follows four repeatable patterns. Once you can spot them, you can check almost any claim yourself:
- A fragment is not the parent molecule. TB-500 is not Thymosin beta-4; Epitalon is not Epithalamin. Trials of the parent get cited to sell the fragment.
- Measuring a substance is not administering it. MOTS-c has a large "human" literature that observes natural levels. That tells you nothing about injecting it.
- A conference abstract is not a published trial. The single most-cited BPC-157 result was never published in full, and its own numbers were not statistically significant.
- Approval elsewhere is not evidence here. Semax really is a registered medicine in Russia. It has never been submitted to the FDA or the EMA, and the two studies that could be read were negative.
And the meta-point: when a systematic review screens 544 papers and finds one clinical study, the volume of literature is not the same as the weight of evidence. A large preclinical corpus concentrated in a single research group is a reason for more caution, not less.
Evidence journalism, not medical advice, and not a recommendation for or against any compound. Every figure here is drawn from the FDA staff reviews prepared for the July 2026 advisory committee, which are public documents. None of these peptides is an FDA-approved drug. If you are considering any of them, the conversation to have is with a clinician who can see your history — and the question to ask is what evidence exists, not what the marketing says.