Nefiracetam (DM-9384)

Chemical identity

Nefiracetam travels under four other names, and every one of them lands on the same substance: the development codes DM-9384 and DZL-221, and the trade designations Translon and Motiva. PubChem holds a single record for it — CID 71157, CAS 77191-36-7, formula C14H18N2O2, molecular weight 246.30, InChIKey NGHTXZCKLWZPGK-UHFFFAOYSA-N, IUPAC name N-(2,6-dimethylphenyl)-2-(2-oxopyrrolidin-1-yl)acetamide. The FDA substance register gives the same CAS and formula, files the substance as achiral under UNII 1JK12GX30N, and gathers the cross-references on that one entry: CHEMBL260829, DrugBank DB13082, MeSH C058876, NSC 759830 and EPA DTXSID2020923. That register also marks the record status as approved, which describes the curation state of the substance entry rather than any marketing authorisation for a medicine.

Structurally the molecule is piracetam with one substitution. Both carry a 2-pyrrolidinone ring bearing an acetamide on the ring nitrogen; in nefiracetam the amide nitrogen carries a 2,6-dimethylphenyl group in place of one of its two hydrogens, leaving a secondary anilide. That anilide governs the pharmacokinetics described further down, because the phenyl ring and the pyrrolidine ring are hydroxylated along separate routes and produce separate metabolites. PubChem gives a computed XLogP of 1.4 and a topological polar surface area of 49.4 square angstroms for the parent. Buol and colleagues, working in 2020, describe the compound as poorly water-soluble, which was the premise of their work (PMID 32660115).

Solid-state characterisation is recent and postdates the clinical programme by nearly two decades. Buol and colleagues reported in 2019 that three solid forms of the molecule had until then gone unidentified: a monohydrate and two polymorphic phases of the anhydrate, which they compared on dissolution rate and solubility (PMID 31348939). The same group published thirteen cocrystals in 2020, three of them formed with citric acid, oxalic acid and zinc chloride, and characterised those three structurally and physically against the parent, comparing solubility, dissolution rate and stability (PMID 32660115). Neither paper reports a shelf-life for the parent as supplied, and the 2020 stability comparison is reported without a storage interval.

Three published human trials, one author group

Filtering PubMed to clinical-trial publication types returns exactly three records for this compound. All three enrolled post-stroke patients and all three run through one continuous author group: the 2008 trial is from the University of Iowa, Starkstein and the Western Australian centres join from 2009, and the 2016 trial is Western Australian with Robinson as a co-author. The doses are not common to all three. Two used placebo, 600 mg and 900 mg daily; the third tested 900 mg daily against placebo alone. No other group has replicated the depression or apathy endpoints. The wider PubMed set is 144 records, of which 28 are indexed to humans and most of those are reviews, analytical-chemistry papers or metabolism studies rather than administration to patients.

Robinson and colleagues enrolled 159 patients within three months of a stroke who met criteria for major depression, and randomised them double-blind to placebo, 600 mg daily or 900 mg daily. Repeated-measures analysis of covariance failed to show a significant time-by-treatment interaction. Response rates exceeded 70 per cent and remission rates exceeded 40 per cent on both drug and placebo. A significant effect appeared only in the top quintile of Hamilton Depression Rating Scale scores at 900 mg. The authors' own stated conclusion is that the compound was not an effective treatment for post-stroke depression, with improvement confined to the most severely depressed patients.

The 2009 report by the same group examined apathy inside that depressed cohort. Of 137 stroke patients with major depression, 70 met diagnostic criteria for apathy; participants received placebo, 600 mg or 900 mg daily for at least four weeks. Analysis showed a significant time-by-treatment interaction, and the 900 mg group showed a significantly greater change in Apathy Scale scores than either the 600 mg group or placebo. PubMed types this record as a phase II randomised controlled trial. The population was selected for depression first and apathy second, which constrains what the apathy result describes.

Starkstein and colleagues then ran a dedicated test of that finding. Their two-centre trial randomised 13 stroke patients with apathy to twelve weeks of 900 mg daily or placebo, with the fourteen-item Apathy Scale as the primary measure. Scores fell by a mean of 7.0 points across the study, and there was no significant between-group difference at week 12. The authors report the compound as ineffective for post-stroke apathy and name the very small randomised sample as the main limitation, and no 600 mg arm was included in this trial.

Later syntheses place the compound at the bottom of its comparison set. Sun and colleagues built a multiple-treatments meta-analysis of ten antidepressants and placebo in post-stroke depression from twelve trials and 707 participants; all drugs separated significantly from placebo apart from sertraline, fluoxetine and this one, and the standardised mean differences ranged from minus 6.54 for the best-ranked drug to 0.51 for the worst-ranked, which was nefiracetam. A 2025 systematic review of post-stroke apathy treatment covering ten trials and 2,359 patients describes the results for this compound as mixed depending on dose and coexisting depression.

The trials that decided its fate were never published

Crespi's 2002 development review in Current Opinion in Investigational Drugs records the programme in some detail. By September 1999 the compound was in phase II in the United States for mental symptoms associated with the sequelae of stroke and for Alzheimer's-type dementia. Trials in China had been completed by August 2000 in preparation for a filing. Modified phase III trials were being repeated in Japan in preparation for a re-filing, with the protocol for the first Japanese retrial approved at the end of July 1999 and completion expected by the end of 2000. In February 2002 Daiichi withdrew the Japanese application, citing insufficient efficacy in the revised trial.

None of that appears in the indexed literature. A PubMed search for the compound combined with cerebral infarction or cerebrovascular terms returns eight records, none indexed as a clinical trial; seven are rodent studies, reviews or electrophysiology reports. The eighth is human administration data from an unrelated group: Hirata and colleagues at Dokkyo University School of Medicine recorded auditory oddball event-related potentials in fourteen chronic cerebral-thrombosis patients and fourteen normal subjects, and reported that P3 and N1 abnormalities improved after administration of the compound (PMID 8728419). That report is uncontrolled, states no dose, and is not indexed as a trial. The Japanese phase III programme, the completed Chinese trials and the United States phase II work in dementia have no located publications. What is readable is the sponsor's own stated reason for withdrawing its application, reported second-hand in a commercial development digest.

ClinicalTrials.gov returns a single registered study: NCT00001933, "Nefiracetam Therapy of Alzheimer's Type Dementia", a phase 2 study sponsored by the National Institute of Neurological Disorders and Stroke, enrolling 50 participants on three capsules twice daily for twenty weeks, running from July 1999 to January 2002 and marked completed. The registry field for posted results reads false, and no corresponding publication was located. Searching the registry across the compound name and both development codes returns that single record and nothing else. The compound is not approved for human use in any jurisdiction identified here, and a Drugs@FDA query returns no matching application.

Pharmacokinetics in people

Fujimaki and colleagues at Daiichi published the human disposition data in 1992, in healthy Japanese male volunteers given single oral doses and then 200 mg three times daily for seven days. Serum concentrations peaked within two hours in every dosage group and declined monophasically with half-lives of three to five hours. Area under the curve and peak concentration were linearly related to dose, apparent clearance ran 94.4 to 140.3 mL per minute, and less than 10 per cent of the dose was recovered in urine as the unchanged compound within 24 hours. Faecal recovery after a 300 mg dose was under 0.1 per cent at 24 hours. Food delayed absorption without significantly modifying the pharmacokinetics, and multiple dosing produced no clinically significant accumulation.

That abstract does not agree with itself on one point. It states that single doses of 10 to 200 mg were studied, then reports that renal clearance did not change significantly as dose increased from 10 to 1200 mg. The retrieved text does not reconcile the two ranges, and the number of volunteers is not stated in it. The figure most often quoted from this paper elsewhere, the three-to-five-hour half-life, is one of the few widely circulated numbers about this compound that traces cleanly to a primary measurement.

A companion paper in 1993 followed the parent and three metabolites. After a single 200 mg dose the peak serum level of the parent was 16.3 plus or minus 0.9 nmol/mL, against 0.96 to 4.89 nmol/mL for the metabolites; time to peak was 1.6 hours for the parent against 4.1 to 9.6 hours for the metabolites, and terminal half-life 3.9 hours against 7.8 to 21.9 hours. Urinary recovery of the parent over 24 hours was about 5 per cent of dose, while the major urinary metabolite, a pyrrolidine ring-scission product, reached 17.8 per cent; all four compounds together accounted for 43.4 per cent. Steady state on three daily 200 mg doses was reached within seven days.

Ten urinary metabolites were isolated and identified from human volunteers in 1990, the principal routes being hydroxylation of the pyrrolidine ring at C5 followed by oxidative carbon-nitrogen cleavage, and hydroxylation of the phenyl ring followed by sulphate conjugation. Work in expressed human cytochromes in 1996 assigned formation of the 5-hydroxy metabolite principally to CYP3A4, and a human liver microsome study in 2001 assigned that formation principally to CYP3A4 with a relatively minor contribution from CYP1A2, and found CYP2C19 capable of the reaction but judged its contribution negligible (PMID 11428655). Rat liver microsomes distribute the same reactions across CYP3A2, CYP2C11, CYP2B1 and CYP2E1, so the rodent and human metabolic maps are not interchangeable.

Nanomolar potentiation, a bell-shaped curve, and a species discrepancy

Zhao and colleagues recorded from rat cortical neurons in long-term primary culture using whole-cell patch clamp. Alpha-bungarotoxin-sensitive alpha7 currents were only weakly inhibited, while alpha4beta2-type currents were potentiated to 200 to 300 per cent of control at 1 nM. Concentrations near 10 micromolar also potentiated, but less, giving a bell-shaped concentration-response. Protein kinase A and protein kinase C inhibitors did not prevent the effect, pertussis toxin pretreatment did not affect it, and cholera toxin abolished it, which the authors read as involvement of Gs proteins rather than Gi or Go.

The same paper carries a result that is absent from every secondary summary of it reviewed here. Human alpha4beta2 subunits expressed in human embryonic kidney cells were inhibited by the compound rather than potentiated. The nanomolar potentiation figure that circulates as a description of what this molecule does was obtained in rat neurons, and the one human-subunit preparation in the paper went the other way. Whether that reflects subunit species differences, the expression system, or something else is not settled by the paper, and no follow-up resolving it was located.

Moriguchi and colleagues extended the same approach to glutamate receptors. NMDA-evoked currents in rat cortical neurons were potentiated with a minimum effective concentration of 1 nM and a maximum of 170 per cent of control at 10 nM, again bell-shaped; 7-chlorokynurenic acid prevented the potentiation, and AMPA and kainate currents were unaffected at 10 nM, which the authors attributed to an interaction with the glycine binding site. A 2007 paper from the same group added that the potentiation was blocked by the protein kinase C inhibitor chelerythrine but not by the protein kinase A inhibitor H-89, that PKCalpha activity rose with a bell-shaped dose-response peaking at 10 nM, and that 10 nM largely eliminated voltage-dependent magnesium block.

Watabe and colleagues reported that 10 mg/kg orally for seven days in adult rats raised GABA turnover and glutamic acid decarboxylase activity in cortex and hippocampus and stimulated sodium-dependent high-affinity GABA uptake in cortical synaptosomes; potassium-evoked release of labelled GABA from cortical slices rose at 10 nM and 1 nM, and GABA-A and benzodiazepine receptor binding were unaffected from 10^-10 to 10^-3 M (PMID 8405098). Luthman and colleagues gave male Sprague-Dawley rats single doses of 1 to 100 mg/kg and daily doses of 1 to 30 mg/kg for fourteen days and found minor, regionally restricted, non-dose-dependent changes in tissue monoamines. Microdialysis showed no significant effect on striatal HVA, DOPAC or 5-HIAA, or on hippocampal HVA, DOPAC or noradrenaline; extracellular hippocampal 5-HIAA fell by 20 per cent at the 1 and 3 mg/kg doses. Synaptosomal uptake of labelled noradrenaline, serotonin and dopamine was unaffected from 1 nM to 10 micromolar.

The 1994 toxicology package

Twelve papers published together in a February 1994 supplement of Arzneimittelforschung constitute the regulatory toxicology programme, run partly at Daiichi's own safety laboratories and partly at a contract laboratory in Cambridgeshire. Coverage is broad for a compound in this category: single-dose studies in mice, rats and dogs, thirteen-week and fifty-two-week oral studies in rats and dogs, reproductive toxicity in rats and rabbits, drug dependence in rats, antigenicity, mutagenicity and two-year oncogenicity in mice and rats. No repeat of any study in that package was located; PubMed was searched for the compound combined with each study type and species, and the most recent toxicology located is Shimomura 2004 (PMID 15082078).

In the thirteen-week rat study at 30, 120 or 480 mg/kg, the 480 mg/kg group showed salivation, prone position, raised serum cholesterol, protein, albumin and bilirubin, increased liver weight and hepatocyte hypertrophy, with hypertrophy also present in males at 120 mg/kg; the stated non-toxic dose was 30 mg/kg (PMID 8018092). In the thirteen-week Beagle dog study at 20, 60 or 180 mg/kg, hypospermatogenesis and a slight increase in splenic haemosiderin deposition were seen in male dogs at 60 mg/kg and above, and 180 mg/kg produced reduced food consumption, raised urinary volume and protein, and renal papillary necrosis in both sexes; the stated non-toxic dose was 20 mg/kg (PMID 8018093).

In the fifty-two-week rat study at 10, 30, 100 and 300 mg/kg/day the kidney was the main target, with papillary and collecting-duct epithelial hyperplasia, cortical scarring and occasional papillary mineralisation, and a no-effect level of 10 mg/kg/day (PMID 8018094). Oncogenicity studies dosed B6C3F1 mice at 30, 90 and 270 mg/kg/day and F344 rats at 200, 600 and 1800 ppm in diet for 104 weeks and reported no evidence of an oncogenic effect, with slight high-dose male body-weight reductions the only finding (PMID 8018100). Reproductive studies found no effect on fertility up to 480 mg/kg/day and no teratogenicity up to 1000 mg/kg/day in rats or 270 mg/kg/day in rabbits, though rat fetuses at 1000 mg/kg showed lower body weights, delayed ossification and more skeletal variations alongside reduced maternal weight gain and food intake (PMID 8018096).

Renal papillary necrosis in dogs, testicular lesions in dogs and rats

Tsuchiya and colleagues went back to the renal result a decade later with rats, dogs and monkeys in the same investigation. Only the dog developed renal papillary necrosis. Beagles given 300 mg/kg/day over eleven weeks showed decreased urinary osmotic pressure from week five, increased urine volume and urinary lactate dehydrogenase from week eight, and necrosis of papillary ductal epithelium first seen at week eight. Concentrations of the metabolite designated M-18 in the renal papilla of dogs sat between those in rats and monkeys, but the papilla-to-cortex and papilla-to-medulla ratios were markedly higher, and in canine renal papillary slices only M-18, and not the parent, reduced synthesis of prostaglandin E2 and of the stable prostacyclin metabolite. Basal prostaglandin synthesis in those canine slices was extremely low relative to rat and monkey.

Shimomura and colleagues dosed male beagles at 180 or 300 mg/kg/day and found testicular testosterone reduced four hours after a single 300 mg/kg dose, serum testosterone reduced after single, one-week and two-week treatment, serum oestradiol raised from one to four weeks, and luteinising hormone, follicle-stimulating hormone and inhibin B unchanged throughout. Reduced sperm motility and increased malformed sperm appeared first at four weeks, and histopathology showed moderate to severe seminiferous atrophy with multinucleated giant cell formation at four weeks but not at one week. The authors attribute the effect to impaired conversion of progesterone to testosterone in Leydig cells (PMID 15082078).

The rat data are not silent on the testis. Shimada and colleagues gave male Slc:SD rats 1500 mg/kg/day orally for four weeks and killed animals sequentially. Retention of step 19 spermatids, sporadic degeneration of pachytene spermatocytes and step 7 spermatids in stage VII seminiferous tubules, and a fall in sperm head counts were the earliest changes at one week, advancing to atrophy of seminiferous tubules with multinucleated giant cell formation by four weeks; serum and testicular testosterone fell and recovered within a day after a single dose, and fell again with repeated dosing (PMID 12849691). That dose is roughly five times the highest used in the dog work.

What is not known

The human record is truncated at the point where it would have mattered. Three randomised trials are indexed, all in post-stroke patients, all from one continuous author group, and the largest enrolled 159 people; one further human report, an uncontrolled event-related-potential study in fourteen chronic cerebral-thrombosis patients by an unrelated group, states no dose (PMID 8728419). The phase III programme in Japan, the trials completed in China by August 2000, and the United States phase II work in Alzheimer's-type dementia produced no located publications, so the data that ended the development programme cannot be read or checked; what survives is a one-line summary in a commercial development digest, reporting the sponsor's own stated reason for withdrawing its application. No human pharmacokinetic study in older people, in renal or hepatic impairment, or in women was located, and the two disposition papers do not state how many volunteers were dosed. No human adverse-event tabulation was located at all, which means the kidney and testis findings in the animal package have no human counterpart to be compared against in either direction. The species question raised by the receptor work is unresolved: the potentiation figures come from rat neurons, and the one human-subunit preparation in the same paper was inhibited. Nothing establishes what any of the nanomolar electrophysiology corresponds to at the serum concentrations measured in people, since no study located here relates the two.

Questions