Oxiracetam (ISF 2522)

Chemical identity and the enantiomer split

Oxiracetam is piracetam with a hydroxyl group added at ring position four. Identity is otherwise unambiguous. PubChem holds a single compound record: CID 4626, CAS 62613-82-5, formula C6H10N2O3, molecular weight 158.16, InChIKey IHLAQQPQKRMGSS-UHFFFAOYSA-N, IUPAC name 2-(4-hydroxy-2-oxopyrrolidin-1-yl)acetamide, XLogP minus 2.2. The FDA substance register gives the same CAS and formula under UNII P7U817352G, status approved, molecular weight 158.1554, and hangs the rest of the cross-references off that one record: WHO ATC code N06BX07, INN code 4762, ChEMBL36633, MeSH C040619, DrugBank DB13601, DrugCentral 2021 and Merck Index m8311. A first-hand ChEMBL fetch returned CHEMBL36633 with maximum phase 3 and a chirality flag of 0.

That hydroxyl creates a stereocentre, and everything since 2016 has turned on it. The registers describe the substance as racemic: GSRS records stereochemistry as unknown with optical activity plus or minus, and PubChem carries a separate record for the S-enantiomer at CID 6603951, same formula and mass, InChIKey IHLAQQPQKRMGSS-BYPYZUCNSA-N. Li and colleagues reported in 2017 that in rats subjected to permanent bilateral carotid occlusion, the S-enantiomer and not the R-enantiomer produced the measured effects on spatial learning, cerebral blood flow and white matter (PMID 28855592). Every human trial registered since then has tested the single enantiomer against the racemate rather than against placebo alone.

Regulatory position differs by jurisdiction and has moved in one direction. Cohen and colleagues in 2021 listed the compound among four drugs not approved for human use in the United States. A 2025 market-surveillance study by twelve official medicines control laboratories with Australia records it in appendix Table A1 as a piracetam homologue not authorised for human use by any health authority in the EU or Australia, detected once by one laboratory, with the seized form recorded as unknown. South Korea's medicines regulator issued a drug safety communication in January 2023 recommending suspension of oxiracetam prescriptions; the market action reported to have followed it is recorded in the untraced claims rather than here. China approved the phase 3 trial of the S-enantiomer under registration 2016L03521, with the racemate serving as its active comparator.

What reaches the blood, and what reaches the brain

Absolute availability was measured once. Perucca and colleagues gave 2000 mg intravenously and orally to four healthy volunteers and reported absolute availability of the oral dose at 75 plus or minus 7 per cent, mean residence times of 3.9 to 6.5 hours, clearance of 100 to 119 millilitres per hour per kilogram, and more than 90 per cent of the intravenous dose recovered unchanged in urine within 48 hours (PMID 6519128). Gschwind and colleagues at Ciba-Geigy, working with a carbon-14 labelled preparation, put peroral absorption at 28 to 42 per cent in rats, 81 to 90 per cent in dogs and about 56 per cent in two healthy men given 800 mg. The two figures measure different quantities: parent availability from paired intravenous and oral curves, against absorption of total radioactivity.

Human half-life figures span an order of magnitude. Perucca followed six elderly female patients aged 69 to 96 after 800 mg orally: peak serum 25 plus or minus 6 micrograms per millilitre within one to three hours, half-life three to six hours, 84 per cent of the dose recovered unchanged in urine within 24 hours (PMID 3691580). Lecaillon and colleagues gave the same dose to eighteen elderly patients and six healthy non-geriatric adults and reported a mean terminal half-life of 12.3 hours in the elderly against 7.7 hours in the healthy group, with area under the curve doubled in the elderly. Two studies in elderly people, three years apart, disagree by a factor of two to four.

Renal function governs that spread. In twenty patients with creatinine clearance from 9 to 95 millilitres per minute, terminal half-life ran from 10.6 to 68.1 hours and the fraction excreted unchanged over 48 hours from 8.3 to 82.6 per cent, with renal clearance correlating against creatinine clearance at 0.985; volume of distribution at steady state was 48.3 plus or minus 21.5 litres and did not vary with the degree of impairment (PMID 2253654). Since the compound leaves the body essentially unmetabolised through the kidney, every half-life figure in this literature is really a statement about the population's renal function rather than about the molecule.

Brain penetration is where the record contradicts itself outright, and both sides are company work. Ponzio and colleagues at ISF Laboratories in Milan gave rats carbon-14 oxiracetam at 200 mg/kg orally or 100 mg/kg intra-arterially and recovered unmetabolised drug in brain, largest in septum, then hippocampus, then cerebral cortex and striatum; delivering 1.9 to 19 nanomoles per rat directly into the lateral ventricles antagonised scopolamine-induced amnesia dose-dependently. Gschwind and colleagues at Ciba-Geigy, using the same isotope in the same species, reported very low brain levels and described slow diffusion across the blood-brain barrier. No paper located here addresses the discrepancy.

A receptor result obtained at micromolar concentrations

Copani and colleagues reported in 1992 that micromolar concentrations of piracetam, aniracetam and oxiracetam enhanced AMPA-stimulated calcium-45 influx in primary cultures of rat cerebellar granule cells, raising the efficacy of AMPA without changing its potency, with the effect surviving the calcium-channel blocker nifedipine. Potentiation by oxiracetam did not extend to kainate, to N-methyl-D-aspartate, or to phospholipid hydrolysis driven by quisqualate. All three drugs increased the maximal density of specific binding sites for tritiated AMPA in rat cortical synaptic membranes. That paper is the origin of most secondary statements that this compound modulates AMPA receptors, and it is a cell-culture result in rat neurons.

One binding study points the other way. Fallarino and colleagues characterised tritiated aniracetam recognition sites in rat brain membranes with an apparent dissociation constant near 70 nanomolar and a maximal density of 3.5 picomoles per milligram of protein, and found that specifically bound tritiated aniracetam was displaced neither by aniracetam metabolites nor by piracetam or oxiracetam (PMID 7616253). Gouliaev and Senning, surveying 407 references across the class in 1994, recorded no affinity for glutamate receptors among the racetams they reviewed, and in the same abstract proposed that the class effect arises from potentiated sodium influx through AMPA-receptor-gated channels (PMID 8061686). That review therefore holds both measurements at once: absent receptor binding, and modulation of the ion flux those receptors gate.

Electrophysiology on the parent compound arrived thirty-four years after the culture work. Guo and colleagues reported in 2026 that oxiracetam increased the frequency and amplitude of spontaneous excitatory postsynaptic currents, AMPA-receptor-dependent evoked currents and miniature currents in hippocampal neurons from APP/PS1 transgenic mice, slowed the desensitisation rate of the GluA1 and GluA2 subunits, and promoted recovery from desensitisation for GluA2 but not GluA1 (PMID 41739317). The indexed abstract states neither dose, route nor group size, and an elink query returns no PMC record for the paper, so the abstract is the only retrievable text.

Concentration arithmetic runs the opposite way here to most compounds in this category. Copani's potentiation appeared at micromolar concentrations; peak serum after 800 mg orally in six elderly women was 25 plus or minus 6 micrograms per millilitre, which at a molecular weight of 158.16 is roughly 158 micromolar. Plasma exposure in a person therefore sits inside the range used in the culture work rather than three orders of magnitude below it. What remains unresolved is the step after that one: the two carbon-14 studies disagree about how much of the compound reaches brain tissue at all.

Cholinergic work always requires a deficit

Spignoli and colleagues reported in 1986 that oxiracetam at 100 and 300 mg/kg intraperitoneally increased acetylcholine utilisation in rat cerebral cortex and hippocampus, measured as the fall in acetylcholine after intracerebroventricular hemicholinium-3, while steady-state acetylcholine levels were unchanged. Repeated daily administration at 100 mg/kg raised high-affinity choline uptake in hippocampus by 31 per cent. A single 300 mg/kg dose raised uptake for both compounds, but piracetam's effect had ended within three hours where oxiracetam still produced a 40 per cent increase at three hours (PMID 3594455). Group sizes are not stated in the indexed abstract. One dose pair, one measure, one time point is the closest thing in the indexed literature to a potency comparison against piracetam.

Dose-response in the behavioural work is not monotonic. Spignoli and Pepeu gave rats scopolamine at 0.63 mg/kg intraperitoneally, which prevented acquisition of a passive avoidance response and lowered acetylcholine by 64, 56 and 42 per cent in cortex, hippocampus and striatum; oxiracetam at 50 and 100 mg/kg intraperitoneally reduced both the amnesia and the cortical and hippocampal acetylcholine fall, while lower and higher doses were ineffective (PMID 3659072). Hlinak and Krejci reported in 2002 that 3, 10 and 30 mg/kg given to mice immediately after acquisition prevented the scopolamine-induced but not the diazepam-induced prolongation of transfer latency in an elevated plus maze (PMID 12110475).

Every positive in vivo result located here required an induced or intrinsic deficit. Paoli and colleagues used the NMDA antagonist AP-5 at 6 micrograms intracerebroventricularly in rats and reported antagonism by oxiracetam from 50 to 500 mg/kg subcutaneously (PMID 2153307). Nardella and colleagues measured choline incorporation into choline phosphoglycerides in hippocampal and cortical slices from spontaneously hypertensive rats and reported restoration towards Wistar-Kyoto values (PMID 1807290). Li and colleagues gave male Wistar rats 50, 100 and 200 mg/kg intravenously for six weeks after bilateral carotid occlusion, seven to nine animals per group. No study located here tested unimpaired animals across a behavioural battery.

A 1987 result complicates the whole account. Mondadori and Petschke, also at Ciba-Geigy, reported that the memory-enhancing action of piracetam-like compounds depended on the presence of the adrenals, and proposed that the central effects might be mediated peripherally (PMID 3427459). The indexed abstract carries no numbers, no species detail and no group sizes. So far as the searches here reach, the finding has not been followed up for this compound in the thirty-nine years since.

Ten placebo-controlled reports

Counts first. A PubMed search on oxiracetam returned 220 records as of 18 August 2026. Twenty carry the publication type Randomized Controlled Trial, twenty-nine carry Clinical Trial, two carry Meta-Analysis, three carry Systematic Review and fifteen carry Review. No record in the corpus carries Retracted Publication, Retraction of Publication, Expression of Concern or Published Erratum. Those twenty do not contain twenty trials of this compound: two are acupuncture trials in which oxiracetam is the control arm, one is a nicergoline combination study, one is a ginkgo and tacrine electroencephalography study that mentions the compound only as a database entry, one is a study protocol carrying no results, and three test the single enantiomer rather than the racemate.

What survives is ten placebo-controlled reports of the racemate, nine of them published between 1985 and 1993. Saletu randomised 40 patients with organic brain syndrome of late life to 2400 mg daily or placebo for four weeks and reported that between-group differences on the Sandoz clinical assessment geriatric scale failed to reach statistical significance (PMID 3897895). Hjorther compared drug and placebo over twelve weeks in 106 middle-aged patients with organic brain syndrome after prolonged solvent exposure and found no difference on any measure. Aldenkamp gave 800 mg three times daily or placebo to 30 patients with epilepsy for twelve weeks and reported no meaningful change in memory function, matching the subjective reports and the P300 measures.

The positive reports are Italian, small, and partly overlapping. Villardita published in 1987 on 40 outpatients, twenty per arm, at 800 mg twice daily for 90 days (PMID 3479527), and again in 1992 on 60 outpatients at the same dose, the same double-blind duration, the same department in Catania and an overlapping test battery; neither paper states the relationship between the two cohorts. The 1992 report carried 29 of its 30 drug-arm patients into an uncontrolled open follow-up to one year, with worsening against baseline on Digit Span Backward and Gibson's Spiral in the late phase. Rozzini, Zanetti and Bianchetti published the same 96-outpatient cohort at 1600 mg daily twice, in 1992 covering 26 double-blind weeks and in 1993 covering those weeks plus 26 open weeks, without the second report describing itself as an extension of the first. Bottini enrolled 65 patients across multiple centres and reported a difference on a quality-of-life scale at p below 0.01.

Then thirty-three years, and a null result at ten times the scale. Lim and colleagues randomised 500 patients at 30 Korean hospitals, mean age 68.9 and a median of 32 months after stroke, to 800 mg twice daily or placebo for 36 weeks, tracking physical activity by wrist actigraphy and imaging resting-state networks at baseline and week 36. Neither coprimary endpoint separated: Mini-Mental State Examination changed by plus 0.13 plus or minus 2.27 against plus 0.27 plus or minus 2.09 (p equals 0.49), and Clinical Dementia Rating-Sum of Boxes by minus 0.14 plus or minus 0.70 against minus 0.08 plus or minus 0.80 (p equals 0.38). Adverse events occurred in 41.0 per cent on drug against 34.9 per cent on placebo (p equals 0.16).

Sequence matters here as much as the result. South Korea's Ministry of Food and Drug Safety commissioned that trial; its funding statement names Korean Drug Company, Ltd and states that the funder had no role in design, conduct, analysis or the decision to submit. In January 2023 the ministry issued a drug safety communication recommending suspension of oxiracetam prescriptions (PMID 41614470). The trial's design paper appeared in 2023 carrying no results. The results themselves appeared in 2026. For three years the only citable record of the largest placebo-controlled trial of the racemate was a protocol. The market action reported to have followed that communication, and the assertion that the funder manufactured part of what was withdrawn, are recorded in the untraced claims.

The enantiomer trial never tested the racemate against placebo

The LOCATE trial randomised 590 patients with traumatic brain injury and Glasgow Coma Scale scores of 10 to 15 across 51 Chinese hospitals between 2019 and 2024, in a 2:2:1 ratio, to intravenous L-oxiracetam at 4 g per day, racemic oxiracetam at 6 g per day, or placebo, for fourteen consecutive days. Mean age was 50.9 years and 421 of 590 patients were male. On the primary endpoint, change in Loewenstein Occupational Therapy Cognitive Assessment score at 90 days, least-squares mean change was 20.45 (95 per cent confidence interval 17.23 to 23.66) for L-oxiracetam, 15.90 (12.71 to 19.10) for the racemate and 11.47 (7.75 to 15.20) for placebo.

Two comparisons are reported and the third is not. L-oxiracetam against placebo gives a difference of 8.97 (5.69 to 12.26), Cohen's d 0.48, p below 0.001; L-oxiracetam against the racemate gives 4.54 (1.85 to 7.23). No comparison of racemic oxiracetam against placebo appears anywhere in the paper, in its table, its text or its abstract. This is the largest trial in which the compound has been administered, 590 randomised with 236 on the racemate, and it tested the racemate as an active comparator without ever formally testing it against its own placebo arm. A footnote to that same table states that the confidence intervals were not adjusted for multiplicity and cannot be used to infer treatment effects, which is the comparison the abstract's efficacy claim rests on.

Secondary outcomes were largely flat even for the enantiomer. Against placebo at 90 days, Mini-Mental State Examination differed by 0.71 (minus 0.16 to 1.57, p equals 0.108), activities of daily living by minus 0.19 (minus 1.33 to 0.96, p equals 0.747), and Glasgow Coma Scale change at 14 days by minus 0.01 (p equals 0.872); the proportion reaching good recovery (extended Glasgow Outcome Scale 7 to 8) at 90 days differed by minus 1.50 percentage points (minus 4.61 to 1.51, p equals 0.667), a proportion rather than a score change. Montreal Cognitive Assessment at 90 days reached 1.11 (0.08 to 2.14, p equals 0.036), unadjusted for multiplicity. The positive finding is confined to the primary instrument.

Adverse-event rates ran in an order worth stating plainly. Any adverse event occurred in 65.96, 66.53 and 67.23 per cent of the L-oxiracetam, racemate and placebo arms respectively (p equals 0.980), and serious adverse events in 5.53, 7.63 and 12.60 per cent (p equals 0.100), highest on placebo. Treatment-related adverse events ran 9.36, 17.37 and 9.24 per cent (p equals 0.049), highest on the racemate. Discontinuation for adverse events ran 0.43, 3.81 and 5.88 per cent (p equals 0.004). No treatment-related serious adverse event occurred in any arm.

What is not known

Absolute bioavailability rests on four healthy volunteers in one 1984 study, and the only other human absorption figure, from two men given a carbon-14 preparation, is roughly twenty percentage points lower and measures a different quantity. Brain penetration is unresolved and the two carbon-14 studies that address it, one from the originating laboratory and one from the licensee, reach opposite conclusions in the same species; no human imaging or cerebrospinal fluid measurement of this compound was located. Which chemical species carries any effect is now openly in question, since the development programme has moved to the S-enantiomer on the strength of one 2017 rat study and no published head-to-head in people establishes that the R-enantiomer is inert rather than merely unmeasured. Nothing in the human record characterises effects in people under 18, in pregnancy, beyond twelve months of exposure, or on discontinuation, and the twelve-month data come from two uncontrolled open-label datasets rather than one: the 96-patient Rozzini cohort, whose open second half had no control arm, and the 29 of 30 drug-arm patients Villardita carried from a 90-day placebo-controlled trial into an open follow-up to one year, during whose late phase worsening against baseline was recorded on Digit Span Backward and Gibson's Spiral. Of the nine placebo-controlled reports from 1985 to 1993, none reports blinding, allocation concealment and outcome definitions in enough detail to be appraised from the retrievable text, two pairs appear to describe overlapping cohorts without saying so, and none was prospectively registered because prospective registration did not yet exist. The 2025 phase 3, which is the largest trial in which this compound has been administered, never compared racemic oxiracetam against its own placebo arm, so the largest modern dataset contains no placebo-controlled estimate of the racemate's effect. No formal toxicology beyond a 13-week dog study was located, and no adverse-event dataset for the racemate traces to a primary source rather than to a secondary review. Whether a live marketing authorisation exists in any named European country was not confirmed here against a national register, and neither the Korean medicines register nor the Korean reimbursement register was reached.

Questions