Noopept (Omberacetam)

Chemical identity, and a suffix the structure does not support

Noopept carries more names than most compounds its size, and all of them resolve to the same entry. PubChem holds it as CID 180496, CAS 157115-85-0, formula C17H22N2O4, molecular weight 318.4, InChIKey PJNSMUBMSNAEEN-AWEZNQCLSA-N, UNII 4QBJ98683M, ChEMBL4303687. The IUPAC name given there is ethyl 2-[[(2S)-1-(2-phenylacetyl)pyrrolidine-2-carbonyl]amino]acetate, the systematic form of the shorthand N-phenylacetyl-L-prolylglycine ethyl ester. The FDA substance register carries the same CAS and UNII under the International Nonproprietary Name omberacetam, INN code 10682. The development codes GVS-111, SGS-111 and DVD-111 sit on that single record as well, so the four designations denote one molecule.

That name is the first thing on the record worth arguing with. The stem -racetam is an International Nonproprietary Name convention marking the piracetam pharmacological group; it is not a structural statement. This molecule carries no 2-oxopyrrolidine acetamide core. It is a proline-glycine dipeptide with a phenylacetyl group on the ring nitrogen and an ethyl ester at the C-terminus, designed in Moscow in the 1980s as a short-peptide imitation of the piracetam pharmacophore and not as a member of its chemical series. Araj and colleagues, publishing the first thorough physicochemical characterisation in 2025, placed it in the racetam group while recording that its structure is quite different from the rest of that group (J Pharm Biomed Anal 2025, PMID 39298839).

Regulatory position differs sharply by jurisdiction. In Russia the compound is a registered prescription nootropic, entered in the State Register of Medicines under number LS-001577 as 10 mg tablets with omberacetam named as the active substance. In the United States it is not approved for human use. Cohen and colleagues in 2021 searched supplement databases for products labelled with four drugs unapproved for human use in the United States, omberacetam among them, and detected five unapproved drugs in the ten products tested (PMID 34484905). A 2025 market-surveillance report by twelve official medicines control laboratories across Europe, with Australia, classified noopept among substances available on prescription in Russia and not authorised in the participating territories (Vanhee 2025, PMID 40558871).

The prodrug argument, and where the originating laboratory disagreed with itself

Gudasheva and colleagues reported in 1997 that one hour after 5 mg/kg intraperitoneally, the parent compound was not detectable in rat brain down to the limit of detection of their HPLC method (PMID 9358206). Three candidate metabolites, phenylacetic acid, prolylglycine and cyclo-prolylglycine, were present in brain samples from treated and control animals alike; the report gives no explanation for their presence in controls. Only cyclo-prolylglycine rose, by a factor of 2.5. Conversion from the parent was then demonstrated in vitro with plasma and brain enzymes. The authors concluded that the compound is a prodrug converting to a cyclic dipeptide identical to an endogenous neuropeptide, and that the endogenous peptide produces the activity.

Boiko and colleagues reported in 2000, by HPLC in rats, that the dipeptide is absorbed from the gastrointestinal tract, enters the circulation and crosses the blood-brain barrier in an unmodified state after oral administration (PMID 10977920). The retrieved abstract states no dose, no sampling time, no measured concentration and no group size. Read together the two reports do not sit comfortably: one found no parent compound in brain an hour after a parenteral dose, the other described intact parent crossing after an oral one. Neither indexed abstract carries enough experimental detail to identify which conditions account for the difference.

Four years after the prodrug paper, the same laboratory published an active-avoidance experiment in which the parent stimulated one-session learning after a single administration and, on repeated administration, raised the number of successful learners among animals that had failed initial training (PMID 11782792). In that respect, the authors wrote, it differs in principle from its main metabolite and from piracetam. Boyko and colleagues restated the position in 2018: the metabolite's pharmacokinetic parameters differ significantly from the parent's, while its pharmacological spectrum is similar (PMID 30378564). The prodrug question is therefore open in the originating group's own record rather than settled by it.

Interspecies work from the same pharmacokinetics laboratory reported in 2004 that the elimination rate falls on passing from rats to rabbits to humans (PMID 15079908). Rats showed intensive presystemic metabolism producing a metabolite hydroxylated at the phenyl ring; rabbits produced no analogous metabolite and cleared the parent more slowly; in human plasma no metabolites were found at all, which the authors attributed to a small dose and a low concentration, not to absence of metabolism. No half-life value appears in the retrievable abstract of that paper, or of any other in the series.

The rodent behavioural record

Oral activity was the finding that made a tablet possible, and it was established late. Ostrovskaya and colleagues reported in 2001 that in rats trained on passive avoidance, antiamnesic activity previously seen after parenteral administration was retained after oral administration, with effective doses of 0.5 to 10 mg/kg; the indexed abstract states no group size (PMID 11782792). The same paper reported the active-avoidance dissociation from the metabolite described above.

Cholinergic models supply the most-repeated animal result. Belnik and colleagues reported in BALB/c mice that scopolamine at 1 mg/kg impaired the long-term memory trace, that 0.5 mg/kg of the dipeptide alone had no significant effect, and that the combination prevented the deficit (PMID 18214292). Radionova and colleagues extended this in rats on a three-way passive-avoidance model, reporting that the compound abolished the amnesic effect of both scopolamine and mecamylamine, blockers of muscarinic and nicotinic receptors respectively (PMID 19145351). The indexed abstract of that second paper prints the phrase antiamnesic effect at that point, which its own surrounding sentences contradict; the wording is the source's. Neither indexed abstract states how many animals were in a group.

Disease-model work is dominated by one laboratory. Ostrovskaya and colleagues reported in 2007 that in NMRI mice after olfactory bulbectomy, 0.01 mg/kg given for 21 days and through a further five days of training restored spatial memory in the Morris water maze and raised serum antibody levels to oligomers of the amyloid beta 25-35 peptide, but not to equine lysozyme amyloid or to the S100b protein; the indexed abstract states neither route nor group size (PMID 17092975). Zarubina and Shabanov reported in Wistar rats with both common carotid arteries ligated, stratified beforehand by sensitivity to hypoxia, that 0.5 mg/kg injected for seven days after occlusion was followed by fewer neurological disorders and higher survival on the measured indices; group size is not stated in the indexed abstract (PMID 19529857).

The neurotrophin claim, and the rat study that did not reproduce it

Ostrovskaya and colleagues measured NGF and BDNF messenger RNA by Northern blot in rat cerebral cortex and hippocampus after single and 28-day administration (PMID 19240853). In hippocampus, expression of both increased after acute administration, and chronic treatment potentiated rather than blunted the effect. In cortex, expression after a single administration was below control, and chronic administration produced only a slight rise in BDNF. Direction of effect therefore depended on region and on acute versus chronic exposure. The summary that circulates from this paper, that the compound raises NGF and BDNF, discards both distinctions and the fact that the measurement is of transcript, not protein.

An independent group produced the only other rodent measurement of hippocampal NGF and BDNF located in PubMed, and it did not agree. Gurbuz and colleagues randomised 60 prepubertal male Sprague-Dawley rats into six groups of ten, induced type 1 diabetes with streptozotocin on postnatal day 28, and gave 0.5 mg/kg intraperitoneally for 14 days (PMID 31356906). Hippocampal NGF and BDNF were measured as protein by ELISA. The paper reports no difference between groups in either neurotrophin and no difference in spatial learning, alongside differences that did reach significance in pubertal timing, blood glucose and HOMA-IR.

Strain, model, duration and assay all differ between the two studies, and a transcript result and a protein result are not required to match. The narrower statement is the one that holds: the hippocampal neurotrophin finding has been measured once by an independent laboratory, in a different preparation, and was not reproduced there. A third rodent measurement exists, in whole brain. Dagda and colleagues, two of whom co-founded the company commercialising the formulation, reported raised brain BDNF and NGF in PINK1-knockout rats given an intranasal preparation containing forskolin at 10 microM and noopept at 20 nM (Int J Mol Sci 2022, PMID 36614135); a two-component formulation cannot apportion the effect between its components, and the measurement is not hippocampus-specific.

In vitro mechanism, and the one retracted paper

The most specific mechanistic result comes from a reporter screen. Vakhitova and colleagues transfected HEK293 cells with luciferase constructs for nine transcription factors: CREB, NFAT, NF-kappaB, p53, STAT1, GAS, VDR, HSF1 and HIF-1 (Acta Naturae 2016, PMID 27099787). At 10 microM, DNA-binding activity rose for HIF-1 alone and for none of the other eight; piracetam at 1 mM moved none of the nine. Molecular docking placed the L-isomer, and its phenylacetylprolyl metabolite, in the active site of prolyl hydroxylase 2, and did not place the pharmacologically inactive D-isomer there. Zainullina and colleagues reported the same HIF-1 direction in SH-SY5Y cells in 2020, a cell line printed as SH-SH5Y in the indexed abstract (Dokl Biochem Biophys 2020, PMID 33119829).

Independent cell work predates all of that. Pelsman and colleagues, working at the University of Connecticut with the Moscow group, exposed cultured human cortical neurons to 50 microM hydrogen peroxide for one hour, which degenerated more than 60 per cent of the neurons present (Int J Dev Neurosci 2003, PMID 12711349). GVS-111 protected across a range from 10 nM to 100 microM with an IC50 of 1.21 plus or minus 0.07 microM, and outperformed piracetam, vitamin E, propyl gallate and s-PBN in the same comparison. The indexed abstract states no number of replicates. Cortical cultures from Down's syndrome tissue showed fewer degenerative changes under chronic treatment.

Electrophysiology points at a nicotinic site. Kondratenko and colleagues recorded from rat hippocampal slices and reported that 5 microM raised the firing frequency of GABAergic interneurons in stratum radiatum; that firing-frequency increase was almost completely abolished, and the increase in spontaneous inhibitory postsynaptic current frequency in CA1 pyramidal cells eliminated, by alpha-bungarotoxin at 6 nM and methyllycaconitine at 20 nM (Neurosci Lett 2022, PMID 36195298). Separately, Jia and colleagues found by two-dimensional heteronuclear NMR that the compound binds no sterically specific site on alpha-synuclein, and attributed its effect on oligomer cytotoxicity to hydrophobic sequestration of toxic oligomers into larger aggregates (J Mol Biol 2011, PMID 21986202).

One paper in the corpus carries a formal integrity flag. Taghizadeh and colleagues published in Heliyon in February 2021 on hyperalgesia, spinal microglia and BDNF expression in a complete Freund's adjuvant rat model, and the paper was retracted in May 2021 (Heliyon 2021;7(2):e06219, PMID 33644478). The notice records that concerns were raised on PubPeer by Elisabeth Bik, that western blot panels reported for different treatment conditions looked astonishingly similar, that images supplied by the authors in response had the same underlying problem, and that this cast doubt on all the data and the conclusions drawn from it. PubMed types the article as a Retracted Publication; a publication-type query across the whole corpus returns that record and no other.

The human record, and one paper nobody can read

Counts first. A PubMed search across noopept, omberacetam and GVS-111 returns 115 records. Three carry the publication type Clinical Trial. None carries the type Randomized Controlled Trial, there is no systematic review indexed, and there is no meta-analysis. A ClinicalTrials.gov search on each of the five designations that denote this molecule, noopept, omberacetam, GVS-111, SGS-111 and DVD-111, returns zero registered studies, which means no trial of this compound has been prospectively registered anywhere that registry indexes.

The study most often cited as the human evidence is the comparison against piracetam by Neznamov and Teleshova, published in Russian in 2008 and in English translation in 2009, in patients with mild cognitive disorders of vascular and traumatic origin (PMID 19234797). PubMed types it Comparative Study, not Clinical Trial, and indexes it with no abstract. Europe PMC returns the record with an empty abstract field. Crossref holds no abstract for the DOI. The ProQuest preview page returns bibliographic metadata and nothing else. Sample size, allocation method and blinding cannot be established from any retrievable record, which means the figures attributed to it on secondary pages cannot be checked against anything.

Two smaller studies can at least be read. Amelin and colleagues described an open prospective study of 60 patients after stroke given 20 mg daily, reporting improvement against controls on the Mini-Mental State Examination and on lateral and categorical association tests at two months; the same abstract also describes treatment over 12 months, and does not reconcile the two durations (PMID 22500312). Mordyk and colleagues studied 60 patients with newly diagnosed respiratory tuberculosis; the 30 patients in group 1 received 10 mg twice daily during the first month, and the report describes fewer neurotoxic and cardiotoxic responses to the antituberculous regimen than in the comparison group (Klin Med (Mosk) 2009, PMID 19565831).

The limits run across the whole human set. The studies that can be read are Russian-language and unregistered. Two of the three typed Clinical Trial come from the same institutional group; the third is from a separate laboratory and is typed Controlled Clinical Trial, though its indexed abstract describes neither blinding nor allocation. None of the set has been replicated outside Russia and Ukraine. A 2019 paper in Georgian Medical News is titled as the management of amnestic disorders after ketamine anaesthesia and states an objective of proving clinical use, but its reported results describe animals (PMID 31687967).

What has actually been measured in products sold outside Russia

Cohen and colleagues bought ten cognitive-enhancement supplements labelled as containing omberacetam, aniracetam, phenylpiracetam or oxiracetam and analysed them by non-targeted liquid chromatography quadrupole time-of-flight mass spectrometry (PMID 34484905). Omberacetam and aniracetam were found, along with three further unapproved drugs that were not the object of the search: phenibut, vinpocetine and picamilon. The highest omberacetam content in a recommended serving was 40.6 plus or minus 0.4 mg, which the authors characterised as roughly four times the pharmaceutical reference amount. Several detected drugs were absent from the labels, several labelled drugs were absent from the products, and of the declared quantities that could be checked, nine of twelve were inaccurate.

The European and Australian picture comes from a network of laboratories, not a single one. Twelve official medicines control laboratories documented 159 samples between January 2020 and September 2024, yielding 166 molecular identification entries covering 34 distinct molecules (J Xenobiot 2025, PMID 40558871). Forty-nine per cent of samples were presented as dietary supplements and 32 per cent as medicines; 69 per cent were collected from the illegal market. Noopept, phenylpiracetam and phenibut were among the substances intercepted as bulk quantities of raw material rather than as finished products.

Neither study measured an effect in a person. Both measured content, and for that reason both belong in a separate register from the pharmacology above. In the sampled sets the label predicted the contents poorly in both directions: drugs were detected that no label declared, and drugs were declared that no analysis found. Of the declared quantities that could be checked in the ten-product set, the printed figure was wrong more often than it was right. Nothing in either report speaks to what the molecule does; both speak to what was in the container.

What is not known

No trial of this compound has been prospectively registered: ClinicalTrials.gov returns zero studies under noopept, omberacetam, GVS-111, SGS-111 and DVD-111 alike, and no systematic review or meta-analysis is indexed in PubMed. Nothing in the human record is typed as a randomised controlled trial, and none of the retrievable human records describes blinding, allocation concealment or a placebo arm. The comparison against piracetam that carries most of the compound's reputation is indexed with no abstract in PubMed, Europe PMC, Crossref or the ProQuest preview, so its sample size, allocation and outcome definitions cannot be established at all. No pharmacokinetic parameter for people has been located in a retrievable primary report: no half-life, no bioavailability figure, no clearance value, and no metabolite was detected in human plasma in the one interspecies study that looked. Longest human observation in the retrievable set is two months of assessed outcome; the only chronic toxicology is a six-month rabbit study whose group sizes are not stated. Nothing characterises effects in people under 18 or over the ages studied, in pregnancy, or on abrupt discontinuation. The prodrug question is unresolved in the originating laboratory's own publications, and the proposed HIF-1 mechanism has been measured only in transfected and immortalised cell lines.

Questions