Cognitive and sleep peptides

Semax

Also known as N-acetyl semax amidate, NASA, ACTH(4-10) analog

Plain-English version: A seven amino acid peptide built from a piece of a stress hormone, with the hormonal activity engineered out.

Strongest evidence, and what it covers [Animal] Focus and memory, the use it is sold for. Two published human trials cover stroke recovery instead, and are described below [1] [2] [3]

Semax takes ACTH(4-10), a short stretch of adrenocorticotropic hormone, and attaches a proline-glycine-proline tail to it. The change removes the hormonal activity the parent carries and slows the breakdown of the peptide in blood. It is sold for focus, memory and recovery after a brain injury.

A ClinicalTrials.gov search run for this page on 2 August 2026 returned no registered study of semax, and none under N-acetyl semax either. The human record is a Russian-language literature from the institute that developed the peptide, and two of those reports are indexed in PubMed: a 1997 clinical and electrophysiological study in the acute period of hemispheric ischaemic stroke [1], and a 2018 study of patients at different stages of ischaemic stroke [2]. Both appear in the same journal, both are in a stroke population, and neither one tested the focus-and-memory use the peptide is actually sold for. The work behind that use is in rodents, where semax was reported to change BDNF and trkB expression in the rat hippocampus [3].

Where it stands Semax is a registered medicine in Russia and is not an approved drug in the United States. FDA put semax-related bulk drug substances to its Pharmacy Compounding Advisory Committee on 24 July 2026, as two separate votes covering the free base and the acetate [19]. A committee recommendation is advisory in either direction, FDA has published no vote record, and nothing about the substance's status changed on that date. Reporting from the law firms following the docket says the committee recommended inclusion, against FDA staff's own written proposal, and the 503A record on this site sets out what that did and did not change.

Our takeTwo indexed human studies, in a language most readers cannot check, from the laboratory that made the peptide, in a population nobody buys it for. The rodent work is the actual basis of everything written about it in English.

• Semax and selank: two peptides, one shared abbreviation

Selank

Also known as N-acetyl selank amidate, NAS, TP-7

Plain-English version: A seven amino acid peptide based on tuftsin, an immune-signalling fragment, sold for anxiety.

Strongest evidence, and what it covers [Animal] Anxiety and stress [4] [5] [6]

Selank is built on tuftsin, a four amino acid fragment released from the heavy chain of immunoglobulin G that signals to immune cells, with the same proline-glycine-proline tail semax carries. It comes from the same Moscow institute and it is almost always sold beside semax.

ClinicalTrials.gov returned 10 registrations on 2 August 2026. They are largely non-interventional or Russian-sponsored, and none of them tests the anxiety use the peptide is marketed for. The published work is rodent behaviour: long-term treatment with the tuftsin analog TP-7 and its effect on anxiety-phobic behaviour and body weight in rats [4], a review of selank and the tuftsin family in stress-related behaviour [5], and a rat study reporting changes in hippocampal and prefrontal BDNF content after ethanol exposure [6].

The comparison people make for selank is with the benzodiazepines, and the specific claim is an absence of any reported dependence signal. That claim rests on rodent behavioural work and on Russian-language reports rather than on a controlled human comparison, and an absent report is not the same thing as a measured finding.

Where it stands Selank is a registered medicine in Russia and is not an approved drug in the United States. It was not among the seven substances FDA put to its Pharmacy Compounding Advisory Committee on 23 and 24 July 2026 [19].

Our takeTen registrations reads like a real programme until the registrations are opened. Almost none of them gives anyone the peptide, and none of them addresses anxiety.

• Semax and selank: two peptides, one shared abbreviation

Cerebrolysin

Also known as FPF-1070

Plain-English version: Not a defined peptide. A mixture of small peptides and amino acids prepared from pig brain tissue.

Strongest evidence, and what it covers [Human RCT] Stroke recovery and dementia [7] [8] [9]

Cerebrolysin is a hydrolysate: pig brain tissue broken down into low-molecular-weight peptides and free amino acids. There is no single molecule to name and no molecular weight to quote, which changes what the evidence can be about. Every trial of it is a trial of a preparation.

It also carries by far the largest trial record on this page. ClinicalTrials.gov returned 42 registrations on 2 August 2026, and the randomised results have been pooled by Cochrane repeatedly rather than once. The 2023 review of acute ischaemic stroke is the seventh version of that review, and it found no reduction in death or dependency while recording an excess of non-fatal serious adverse events in the arms that received the porcine brain hydrolysate [7]. The 2020 version of the same review put that excess at a risk ratio of 2.15 [8]. Cochrane covers vascular dementia in a separate review [9]. Across those updates the raised rate of non-fatal serious adverse events is the finding that has repeated.

The preparation is also porcine-derived and injectable, which is a different risk conversation from the one that applies to a synthetic peptide, and it is the reason the trial reports pay attention to reactions rather than only to outcome scales.

Where it stands The preparation holds no FDA approval and no central European Medicines Agency authorisation, and it is marketed in a number of other countries. It was not among the seven substances FDA put to its Pharmacy Compounding Advisory Committee on 23 and 24 July 2026 [19].

Our takeThe only compound in this group whose evidence base is large enough to be conclusive, and what it concluded is the negative. That is worth more to a reader than everything else on this page put together.

Dihexa

Also known as PNB-0408, N-hexanoic-Tyr-Ile-(6) aminohexanoic amide

Plain-English version: A small analog of angiotensin IV, a fragment of a blood-pressure hormone, sold for cognition.

Strongest evidence, and what it covers [Animal] Memory and cognition [10] [11] [12] [13]

Dihexa is a short peptide with fatty acid groups added at both ends, which is what the chemical name describes. Those groups let it survive in blood and reach the brain, and make it behave more like a small drug molecule than a peptide.

ClinicalTrials.gov returned no registered study of dihexa on 2 August 2026, and no human study of any design has been published. It came from one laboratory at Washington State University, and the 2014 paper establishing its proposed mechanism, hepatocyte growth factor signalling through the c-Met receptor, was retracted in 2025 [10] [11]: PubMed flags the original as a retracted publication and carries the notice as its own record, so both are cited. Independent rodent work does exist, a 2021 study in APP/PS1 mice reporting differences in memory testing and in PI3K and AKT signalling [12].

The same mechanism has been tested in people, on a different molecule. Fosgonimeton, a characterised drug on that target, ran a phase 2 and 3 trial in mild to moderate Alzheimer's disease, NCT04488419, with 554 participants enrolled [13]. That trial sets a tier for fosgonimeton and none for dihexa. c-Met is also a proto-oncogene.

Where it stands Dihexa is not an approved drug anywhere and it was not among the seven substances FDA put to its Pharmacy Compounding Advisory Committee on 23 and 24 July 2026 [19].

Our takeA retracted mechanism paper, zero human studies, and a properly characterised drug on the identical target that ran a 554-person trial and published the result. Every one of those is a dated, checkable document, which makes this the best-documented compound on the page and one of the thinnest.

• Dihexa and fosgonimeton: one mechanism, two very different records

P21

Also known as P021, Cnt-2-Ada

Plain-English version: A four amino acid derivative of a fragment of a nerve growth factor, studied only in rodents.

Strongest evidence, and what it covers [Animal] Memory, neurogenesis and neuroprotection [14] [15]

P21 is a tetrapeptide taken from a fragment of ciliary neurotrophic factor, a protein that keeps nerve cells alive. The published form, written P021 or Cnt-2-Ada, carries an adamantane group, a rigid hydrocarbon cage, attached to make the peptide stable enough to survive being swallowed. It is sold for memory and neuroprotection.

The entire published record is rodent work, and it comes from one group. A 2014 paper in Neurobiology of Disease gave the compound in the diet to 3xTg-AD mice, a strain engineered to develop both of the protein deposits found in Alzheimer's disease, and reported differences in behavioural testing and in markers of amyloid and tau against untreated animals [14]. A second 2014 paper worked in aged rats rather than in a disease model, and reported differences in measures of new nerve cell formation and in cognitive testing [15]. Khalid Iqbal is an author on both, at the New York State Institute for Basic Research in Developmental Disabilities, and no laboratory outside that group has published a replication of either result.

A registry search adds nothing, and the reason is worth showing. ClinicalTrials.gov returns five records for the string P021, and every one of them is an unrelated study whose own internal protocol identifier happens to contain it: a telepsychiatry evaluation in children, a sickle-cell cohort, a hip-fracture care model, a rare-neurological-disease registry and a pharmacist-led quality improvement study, each one opened and checked on 2 August 2026. There is no registered trial of the peptide, no open-label series, no case report, and no measurement of what happens to it in a person.

Sixteen years, one laboratory, no human data. That is unusual even in this catalog, where most compounds have at least a second group that tried to reproduce the headline result and reported what it got. The rodent work is a reason to run a human study. Nothing on the record says one was ever run.

Where it stands P21 is not an approved drug anywhere, it holds no registered trial under either of its written forms, and it was not among the seven substances FDA put to its Pharmacy Compounding Advisory Committee on 23 and 24 July 2026 [19].

Our takeTwo papers from one laboratory, sixteen years apart from any human work, is not an evidence base. It is a research programme that stopped, and the market picked the compound up anyway.

DSIP

Also known as Delta sleep-inducing peptide, Emideltide

Plain-English version: A nine amino acid peptide isolated from rabbit blood in 1977 and sold for sleep.

Strongest evidence, and what it covers [Human RCT] Sleep [16] [17] [18]

DSIP was isolated in 1977 from the blood of rabbits during sleep experiments, which is where the name comes from. Its regulatory alias, the one FDA uses, is emideltide. It is sold for sleep, and unlike most of this page it has been given to people.

The human record is old and small. A 1981 study in Experientia gave synthetic DSIP to people with disturbed sleep and reported changes in sleep measures [17]. A 1992 study in Neuropsychobiology, double-blind and indexed by PubMed as a randomised controlled trial, tested it in patients with chronic insomnia [18]. Those two are the studies the marketing rests on, and they are small, decades old, and were not followed by anything modern: a ClinicalTrials.gov search on 2 August 2026 returned no registered study of DSIP and none of emideltide.

The most useful summary of the position is the title of a 2006 review in the Journal of Neurochemistry by Kovalzon and Strekalova, which calls DSIP a still unresolved riddle [16]. That review sets out two separate problems. The peptide's own job in the body has never been pinned down, and the sleep results published across the 1970s and 1980s did not agree with each other, so there is no consistent finding for a later study to confirm or overturn.

DSIP carries the most recent agency record in this group. FDA put emideltide-related bulk drug substances to its Pharmacy Compounding Advisory Committee on 24 July 2026, as separate votes on the free base and the acetate [19]. The same document lists the full set the committee worked through over the two days: BPC-157, KPV, TB-500, MOTS-c, emideltide, epitalon and semax. Press coverage reports that emideltide was the one substance of those seven the committee declined to recommend. FDA has published no vote record, so that outcome rests on the coverage rather than on an agency document. The rest of the meeting, including what FDA itself proposed on all fourteen substances, is on the 503A record page.

Where it stands DSIP is not an approved drug in the United States. FDA put emideltide-related bulk drug substances to its Pharmacy Compounding Advisory Committee on 24 July 2026, in two votes covering the free base and the acetate [19]. A committee recommendation is advisory: the agency must still decide, publish a proposed rule, take comment and issue a final rule before anything on the 503A Bulks List changes.

Our takeThe only compound on this page whose human studies actually tested the thing it is sold for. They were small, they disagreed with each other, and the last one was published in 1992. Thirty-four years of silence is itself a result.