Semax: An ACTH(4–7) Fragment, BDNF, and What the Russian Literature Does and Does Not Show

Semax is proposed to act by raising brain-derived neurotrophic factor (BDNF) and activating its receptor, TrkB, in the hippocampus and basal forebrain. That proposal rests on rodent experiments from one closely connected group of Russian laboratories 123. Semax is a synthetic heptapeptide, Met-Glu-His-Phe-Pro-Gly-Pro. Its first four residues are positions 4 to 7 of adrenocorticotropic hormone (ACTH). The last three are a proline-rich tail added to slow enzymatic breakdown. It is registered as a medicine in Russia, where its clinical literature concerns ischaemic stroke. It is approved nowhere else.

This article is not a second Selank article. The two peptides share a country of origin, a research programme and the same Pro-Gly-Pro tail, and the network's Selank review already covers that tail and the general problem of appraising Russian-language literature. Semax differs in the three places that matter. Its parent molecule is a pituitary hormone, not an immune fragment. Its central mechanistic claim is neurotrophic, not GABAergic. And its human evidence is in stroke, not anxiety. Those differences are the subject here.

A hormone fragment with the hormone removed

ACTH is a thirty-nine-residue pituitary peptide whose endocrine job is to make the adrenal cortex produce cortisol. That job depends on residues well beyond position 10, including the basic cluster around positions 15 to 18, which the adrenal ACTH receptor requires. Short N-terminal fragments were studied for decades because they kept behavioural effects in animals while losing steroidogenic activity. ACTH(4–10) was the classic example. The literature calls Semax an ACTH(4–10) analogue because it keeps positions 4 to 7 and replaces Arg-Trp-Gly at positions 8 to 10 with Pro-Gly-Pro 23.

The substitution has two consequences. The first is stability. Unmodified ACTH fragments are cleared by peptidases within minutes, and a proline-rich C-terminus resists exopeptidase attack. The second is often overlooked. Removing Arg-Trp also removes part of the His-Phe-Arg-Trp core through which melanocortin peptides engage their receptors. So Semax cannot be assumed to be a melanocortin agonist simply because it descends from ACTH. Whether it binds melanocortin receptors at all, and whether it does so as an agonist or an antagonist, is not settled in the indexed literature.

The BDNF and TrkB claim

The neurotrophic proposal is built from three rodent studies. The first reported that Semax raised BDNF expression in several rat brain regions 1. The second, in rat hippocampus after intranasal administration, reported a 1.4-fold increase in BDNF protein, a 1.6-fold increase in TrkB phosphorylation and a threefold increase in exon III BDNF messenger RNA, alongside improved conditioned avoidance learning 2. The third reported specific binding of labelled Semax in rat basal forebrain membranes and a rapid rise in BDNF protein there 3.

Taken together, this is a coherent chain. It runs from a binding site in a brain region, to a rise in a neurotrophin, to phosphorylation of that neurotrophin's receptor, to a behavioural change. It is also incomplete in the way that matters for a mechanism. The binding site has not been identified as a known receptor. The TrkB step is correlational: no study in this literature blocks TrkB, or deletes BDNF, and shows that the behavioural effect disappears. Without that test, BDNF is a readout that moves with Semax, not a demonstrated route by which Semax acts.

BDNF is also one of the most commonly measured endpoints in neuropharmacology. Antidepressants, exercise, environmental enrichment and many experimental compounds raise it in rodent hippocampus. A compound that joins that list has shown something real but not specific. It is the same caution the Selank review applies to the BDNF findings for that peptide, and it applies here with more force, because for Semax the neurotrophic strand is the main claim, not one of three.

Monoamines and the melanocortin question

A second strand concerns monoamine systems. In rodents, Semax raised striatal tissue levels of the serotonin metabolite 5-hydroxyindoleacetic acid at two hours, and extracellular levels measured by microdialysis rose to about 180 percent of baseline over one to four hours. The same work reported that Semax given shortly before D-amphetamine markedly enhanced the psychostimulant's effect on extracellular dopamine and on locomotor activity 4. Antagonism at melanocortin receptors has been proposed elsewhere as a possible link to these monoamine changes, but no study in the indexed record demonstrates it for Semax. The work measured neurotransmitter turnover, not receptor occupancy.

Transcriptomics after experimental stroke

The third strand links the rodent work to the clinical indication. In rats with permanent occlusion of the middle cerebral artery, genome-wide expression profiling of ischaemic cortex compared Semax-treated animals with untreated ischaemic controls at three and twenty-four hours 5. More than half of the genes Semax altered were immune-related, mainly immunoglobulin and chemokine genes, and the immune effect grew larger by twenty-four hours. Genes involved in vascular processes, including endothelial development and smooth muscle migration, were also affected.

This is a broader and more modern data set than the BDNF work, and it points somewhere different. If Semax is neuroprotective after ischaemia, these data suggest immune and vascular modulation contributes as much as neurotrophin signalling. Expression profiling shows what changed, not what caused the change. It also shows transcripts, not proteins or function.

The human evidence is stroke, not cognition

The compound is described in its source literature as a nootropic, and that label dominates how it is discussed in English. The indexed human literature does not match the label. It concerns ischaemic stroke. A 1997 clinical and electrophysiological study compared 30 patients with acute hemispheric stroke who received Semax on top of intensive therapy with 80 patients of similar severity who received conventional therapy alone. It reported faster neurological recovery, particularly of motor deficits, supported by EEG mapping and evoked potentials 6.

Three features of that study limit what it can show. The groups are unequal in size, which is unusual for a randomised design, and the abstract does not say how patients were assigned. Nothing is reported about blinding of the outcome assessors. The sample is small. A later study of 110 patients at different stages after ischaemic stroke measured plasma BDNF, motor performance and a standard disability index against the timing of rehabilitation 7. It is valuable because it measures BDNF in people. But it is a Russian-language report from the same clinical school, and its available abstract does not establish a placebo-controlled, blinded design.

The honest summary is narrow. No randomised, placebo-controlled, blinded trial of Semax with a registered primary endpoint appears in the internationally indexed literature, for stroke or for any cognitive endpoint in healthy people. The cognitive claims made for the compound in English rest on rodent learning tasks 2, not on human data.

Regulatory position in 2026

Semax is not approved by the regulators of the United States, the United Kingdom or the European Union. In the United States its 2026 status concerns compounding, not approval. The FDA sorts bulk substances nominated for compounding under section 503A into categories. Category 2 holds substances for which the agency has identified significant safety risks 8. On 15 April 2026 the FDA announced that twelve peptides, Semax among them, would be removed from Category 2 after a seven-day notice period, because their nominations had been withdrawn. Removal did not place them in Category 1 and does not in itself make them eligible for compounding 9.

Semax was one of seven of those peptides reviewed by the Pharmacy Compounding Advisory Committee on 23 and 24 July 2026. The committee voted 8–5–1 to recommend adding it to the 503A bulks list. That vote is advice. The FDA must still complete notice-and-comment rulemaking before anything is added 10. None of this is a finding about efficacy. An advisory vote on compounding eligibility is not a marketing authorisation and does not evaluate whether the compound works for any indication.

What remains untested

1. Identity of the binding site reported in basal forebrain, and whether it is a melanocortin receptor, a known neurotrophin-related protein or something else.
2. A causal test of the BDNF route, using TrkB blockade or BDNF-deficient animals, showing whether behavioural or neuroprotective effects depend on it.
3. Independent replication of the BDNF and transcriptomic findings by a laboratory outside the originating programme.
4. Human pharmacokinetics adequate to show whether the intact heptapeptide, or a fragment, reaches the brain after intranasal delivery.
5. A randomised, blinded, placebo-controlled stroke trial with a registered primary endpoint, reported to international standards.

The fair summary

Semax is a well-defined molecule with a clear design rationale: keep a behaviourally active hormone fragment, remove its endocrine activity, and protect it from peptidases. The rodent evidence for a BDNF and TrkB effect is consistent across three reports from one group 123. The monoaminergic and transcriptomic data widen the picture without narrowing it to a mechanism 45. The human evidence is real, limited to stroke, and not appraisable to current standards from what is available internationally 67.

That leaves Semax in the same place as most compounds on the April 2026 list. It has been studied enough to generate a plausible mechanism, but not enough to confirm it. It has been used clinically in one country without the trials another regulator would require. And it now sits in a regulatory process that says nothing about the underlying biology. The gap is specific and could be tested, and it has not been.