Phenibut

Identity, and the salts that complicate it

PubChem CID 14113 resolves phenibut as 4-amino-3-phenylbutanoic acid, CAS 1078-21-3, molecular formula C10H13NO2, molecular weight 179.22 g/mol, InChIKey DAFOCGYVTAOKAJ-UHFFFAOYSA-N. The FDA Global Substance Registration System carries it under UNII T2M58D6LA8. ChEMBL indexes it as CHEMBL315818, ChEBI as 136039. It is a small molecule rather than a peptide: gamma-aminobutyric acid with a phenyl ring hung on the beta carbon. Baclofen is the 4-chlorophenyl version of the same skeleton and gabapentin swaps the ring for a cyclohexane. Zvejniece and colleagues assayed all three against the same alpha-2-delta ligand (PMID 26234470).

The registry number above describes the racemate only. The World Health Organization's 2021 pre-review lists separate numbers for the two enantiomers, 35568-36-6 for the R form and 62596-63-8 for the S, alongside more than a dozen salts and co-crystals including the hydrochloride at 3060-41-1. It is marketed in Russia under several proprietary names, as the hydrochloride and as a citrate salt; the citrate carries its own registry number, CAS 697285-55-5. The distinction matters because the salts carry different registry numbers and different pharmacokinetic literature. A small Russian pharmacokinetic literature exists for the citrate salt under a separate drug name and was conducted in rats; those figures do not transfer automatically to the hydrochloride, and neither set describes a human.

Khaunina's report on the tranquillising effects of beta-phenyl-gamma-aminobutyric acid, published in Biulleten Eksperimentalnoi Biologii i Meditsiny in January 1964, is the earliest phenibut pharmacology paper indexed in PubMed (PMID 14162373); a second Khaunina report, on the compound's effect on centrally acting drugs, followed in Farmakologiia i Toksikologiia later the same year (PMID 14193102). Both are Russian-language with no abstract in the index, which is the characteristic shape of the compound's early record. Phenibut is a registered prescription medicine in Russia, marketed there under proprietary names that are not reproduced here. The WHO pre-review describes it as possibly approved in a handful of neighbouring states without confirming each one.

The enantiomers are not equivalent

Dambrova and colleagues at the Latvian Institute of Organic Synthesis separated the isomers and compared them directly (PMID 18275958). In radioligand binding against the selective GABA-B antagonist tritiated CGP54626, affinity constants were 177 plus or minus 2 micromolar for racemic phenibut, 92 plus or minus 3 micromolar for R-phenibut, and 6.0 plus or minus 1 micromolar for baclofen; the indexed abstract states neither the species nor the preparation. Across tests of locomotor activity, antidepressant effect and pain, S-phenibut was inactive at doses up to 500 mg/kg while R-phenibut was roughly twice as potent as the racemate on most measures. The GABA-B antagonist CGP35348 blocked the antidepressant, antinociceptive and locomotor effects of R-phenibut.

Seven years later the same group tested a second target and found the stronger interaction there (PMID 26234470). Using radiolabelled gabapentin as a subunit-selective ligand on rat brain membranes, they reported dissociation constants at the alpha-2-delta subunit of the voltage-dependent calcium channel of 23 micromolar for R-phenibut, 39 micromolar for S-phenibut, 156 micromolar for baclofen and 0.05 micromolar for gabapentin. The authors reported the affinity of R-phenibut for the alpha-2-delta subunit as about four times higher than for the GABA-B receptor. In the formalin paw-licking test its antinociceptive effect was not blocked by CGP35348, which the authors read as evidence that the analgesia ran through alpha-2-delta rather than through GABA-B. At doses up to 100 mg/kg, R-phenibut did not alter pentylenetetrazole-induced seizures.

One consequence of those two papers is that the shorthand describing S-phenibut as the inactive isomer holds for one target and not the other. No measurable S-phenibut binding to GABA-B was reported (PMID 26234470), but its alpha-2-delta constant of 39 micromolar sits within a factor of two of the R form's. Irie and colleagues supplied a potency comparison at GABA-B in a different preparation, recording outward current in mouse cerebellar Purkinje cells in slice (PMID 32735986). Half-maximal outward current density came at 1,362 micromolar for phenibut, 23.3 micromolar for F-phenibut and 6.0 micromolar for baclofen. On that readout the ratio between the phenibut and baclofen values was greater than two hundred.

The animal record, and what is absent from it

Vavers and colleagues gave R-phenibut to rats after middle cerebral artery occlusion induced two different ways, filament insertion in male Wistar rats and endothelin-1 microinjection in CD rats, dosing at 10 or 50 mg/kg for fourteen or seven days respectively (PMID 26621244). The 10 mg/kg group showed a difference in histological outcome at day seven in the endothelin model. The 50 mg/kg group showed reduced loss of brain volume in the damaged hemisphere in both models, alongside increased BDNF and VEGF gene expression. On the vibrissae-evoked forelimb-placing test the authors report a trend toward recovery rather than a separation. The indexed abstract does not state the route of administration.

Kupats and colleagues ran a lateral fluid percussion injury model in male Swiss-Webster mice and, in the same paper, measured where the compound went (PMID 33274011). After 50 mg/kg, R-phenibut reached brain tissue fifteen minutes after both intraperitoneal and peroral administration, with maximal brain concentrations of 0.6 micrograms per gram and 0.2 micrograms per gram respectively. Those two numbers are the most concrete distribution data in the retrieved record, and they describe mice given a single enantiomer rather than people given the racemate.

Absent from the animal literature is any study of the property the compound is best known for outside Russia. The WHO pre-review states that no studies in animals have examined the dependence potential of phenibut and that no controlled animal or human study has examined its abuse potential. A PubMed search combining phenibut with self-administration, drug discrimination and conditioned place preference returned six records, none of them an abuse-liability experiment and three of them synthetic chemistry papers. A 1989 rat study of fetal development expressed its 50 mg/kg dose as one twentieth of the LD50 without printing the LD50 (PMID 2806525); two Soviet-era cat cardiovascular papers from the same period use the same device, but the LD50 they divide is that of phenibut esters rather than of phenibut.

The human record rests on one review

Lapin's 2001 article in CNS Drug Reviews is the source that nearly every English-language account of phenibut leads back to (PMID 11830761). It describes the compound as GABA-mimetic at GABA-B and to a lesser extent GABA-A, reports that it stimulates dopamine receptors and antagonises beta-phenethylamine, and lists the indications for which it was used in Russian practice. It is a review rather than a study, and it carries the whole weight of the human account.

The World Health Organization's Expert Committee on Drug Dependence examined that article when it pre-reviewed phenibut in 2021. The pre-review states that the article is flawed, that in many places references are not provided and in others they are inaccurate, and that of three references in it which could potentially be considered clinical trials, one is a small open-label trial, one is cited for preclinical results, and one is listed as published in an English-indexed journal but does not come up in a search and is not mentioned in the body. On pharmacokinetics the Committee wrote that none of the statements includes references, preventing examination of the primary studies, and that all other literature discussing those effects cites Lapin 2001.

Kupats and colleagues assembled the clinical literature in 2020 (PMID 32340063). Across eleven trials totalling 583 patients they recorded adverse events in 5.66 per cent of participants, most often somnolence at 1.89 per cent, and set that against fourteen case reports covering sixteen patients in which the amounts involved ranged from 0.5 to 100 g per day. That review notes that none of the eleven trial reports it included is available in English and that none of their abstracts carries a statistical analysis.

Four Russian-language papers are typed by PubMed as randomised controlled trials, covering chronic fatigue with cerebrovascular insufficiency in 53 patients, paediatric tension-type headache in 30 against 30 controls, anxiety with chronic cerebral ischaemia in 62, and attention deficit hyperactivity disorder in 50, while the WHO document states that there are no randomised controlled trials of phenibut for any indication. Both positions are recorded here as they stand. One of those four papers does print a statistical comparison in its indexed English abstract (PMID 28300804), so the absence of statistics is a finding about the eleven trials Kupats reviewed rather than a general property of the Russian-language record.

ClinicalTrials.gov returns nothing at all. Registry searches for phenibut, fenibut and aminophenylbutyric acid each came back with zero studies. The closest thing to a human pharmacokinetic study in the indexed record is a Russian open randomised crossover comparison in eighteen healthy volunteers that measured unchanged aminophenylbutyric acid in plasma by HPLC with ultraviolet detection and concluded that two formulations were bioequivalent (PMID 21809698). Its indexed abstract reports no half-life, no clearance and no bioavailability figure.

What poison centres recorded

The largest dataset is a CDC analysis of calls to United States poison centres between 2009 and 2019 (PMID 32881852). It counted 1,320 phenibut exposures across all fifty states and the District of Columbia. Men accounted for 75.5 per cent, adults aged 18 to 34 for 58.4 per cent, and ingestion for 93.2 per cent of routes. Agitation was recorded in 30.4 per cent of cases, drowsiness or lethargy in 29.0 per cent, tachycardia in 21.9 per cent and confusion in 21.3 per cent. Coma appeared in 6.2 per cent, or 80 cases. Outcomes were coded moderate in 49.6 per cent and major in 12.6 per cent, and three deaths were reported, one of them in an exposure involving phenibut alone.

Two qualifications sit on that dataset. Its own footnote records that the terms used to find cases entered the database at different points: one term predates the study window, a second was added in 2012 and a third in 2015, so part of the rise reflects case-finding rather than incidence. None of the exposures was biochemically confirmed. A single-state series from Minnesota covering 2000 to 2018 found 56 exposure calls, 48 of them in the final five years, with 19.6 per cent of patients intubated and no deaths (PMID 30878413).

Feldman and colleagues systematically reviewed the withdrawal literature and found 25 case reports and conference abstracts, all describing male patients (PMID 38112312). Median age was 30 years across a range from four days to 68. The median daily amount preceding withdrawal was 10 g, interquartile range 4.75 to 21.5 g and outer range 1 to 200 g. The shortest recorded period of use before withdrawal was one week at 2 to 3 g daily, and symptoms began as early as two hours after the last dose. Seizures were reported in 8 per cent of patients, intubation in 24 per cent and intensive care admission in 44 per cent; 68 per cent received a benzodiazepine and 76 per cent required at least two drug therapies.

A German review reaches a different picture from a different denominator (PMID 38377332). Over 2008 to 2022 the Goettingen poison information centre received 17 enquiries about phenibut, against 55 for gabapentin, 126 for pregabalin and 4,324 for benzodiazepines. Overdoses of between 2 and 100 g were reported in 15 of the 17 cases and eight of those patients were somnolent. Bonnet and colleagues record no case of respiratory depression or coma, including in the patient who had taken 100 g, and state that no death from phenibut intoxication has been published in Germany or elsewhere in Western Europe while allowing that such cases may have been missed. The American and German records describe severity differently and are left here as two separate observations rather than averaged.

Labels, and the regulatory position

On 10 April 2019 the FDA issued twelve warning letters, three of them concerning products marketed as dietary supplements and labelled to contain phenibut. The agency's stated position, published on its own constituent update, is that phenibut does not meet the definition of a dietary ingredient under the Federal Food, Drug, and Cosmetic Act, and that products listing it as one are misbranded. Phenibut is not an approved drug in the United States and is not federally scheduled, which leaves it in the unusual position of being unlawful to sell in a supplement without being controlled to possess.

Cohen and colleagues measured what happened next (PMID 34550038). Four brands met their inclusion criteria of being labelled as containing phenibut and sold both before and after the warnings. Beforehand, two of the four actually contained it, at 484 and 487 mg per serving. Afterwards all four did, at between 21 mg and 1,164 mg per serving, with the quantity higher in three of the four; quantities detected per dose ran as much as 450 per cent greater than a typical 250 mg pharmaceutical tablet manufactured in Russia. A companion analysis of ten products marketed for cognitive enhancement and labelled for other unapproved drugs detected phenibut as an undeclared ingredient at up to 15.4 plus or minus 0.3 mg, and found nine of twelve declared quantities on those labels to be inaccurate (PMID 34484905).

The WHO Expert Committee recommended in October 2021 that phenibut should not proceed to critical review but should be kept under surveillance by the Secretariat, citing limited information on withdrawal, on abuse liability, on the magnitude of misuse and on similarity to internationally controlled substances. It has been on the Committee's surveillance list since 2017. Phenibut is not controlled under the 1961, 1971 or 1988 United Nations conventions. Australia moved it to Schedule 9, prohibited substance, in 2018, and the WHO pre-review records the four reasons the Australian regulator gave, among them rapid development of tolerance and dependence. Hungary, Italy, Lithuania and France appear in the same document as the other countries controlling it.

What is not known

No human pharmacokinetic study of phenibut was located, so absorption, elimination half-life, clearance, bioavailability and interactions in a human body are unmeasured in the retrievable record; the figures in wide circulation descend from an unreferenced review. Dependence and abuse potential have never been examined experimentally in any species, which means the property the compound is most discussed for outside Russia is documented only through case reports and forum text. The clinical trial literature exists but is Russian-language, unregistered and small; Kupats 2020 records that none of the eleven trials it reviewed carries a statistical analysis in its abstract, and whether any of the wider Russian-language literature qualifies as a randomised controlled trial is itself disputed between PubMed's indexing and the WHO pre-review. Nothing has been established about prolonged exposure: no chronic toxicology, no carcinogenicity work and no reproductive study beyond a single 1989 rat fetal-development paper were found, and the phenibut LD50 that that paper divides by is never printed. The enantiomers have been separated pharmacologically in rodents, but all human use and all clinical data concern the racemate, so the rodent enantiomer findings cannot be mapped onto the human reports. Poison centre records from the United States and Germany describe severity differently and cannot be reconciled from the published material. Phenibut holds no approval in the United States, the European Union or the United Kingdom, is a prohibited substance in Australia, and is not controlled under any United Nations convention.

Questions