Memantine

Moderate to severe Alzheimer's disease

• The McShane 2019 Cochrane review covered 44 trials with almost 10,000 participants. In moderate to severe Alzheimer's disease (up to 14 studies, about 3,700 people), high-certainty evidence showed a small benefit over placebo at six to seven months: 0.21 points on the CIBIC+ global rating, 3.11 points on the Severe Impairment Battery, 1.09 points on a 19-item daily-living scale and 1.84 points on the Neuropsychiatric Inventory, with or without a cholinesterase inhibitor. Fewer people on memantine had agitation as an adverse event, but three trials found memantine of no benefit as a treatment for agitation (PMID: 30891742).
• Reisberg 2003 was a 28-week double-blind trial in 252 patients at 32 US centers. Patients received 20 mg a day or placebo. Of the two primary endpoints, the daily-living scale favored memantine (P=0.02), while the CIBIC-Plus global rating missed significance with the last observation carried forward (P=0.06) and reached it in observed cases (P=0.03); the Severe Impairment Battery favored memantine (P<0.001), and adverse events were not significantly more frequent (PMID: 12672860).
• Tariot 2004 was a 24-week trial at 37 US sites in 404 patients already taking donepezil. Memantine or placebo was added to donepezil. Both primary endpoints favored memantine: Severe Impairment Battery change of 0.9 against minus 2.5 points (P<.001) and daily-living change of minus 2.0 against minus 3.4 points (P=.03); 7.4 percent on memantine and 12.4 percent on placebo stopped because of adverse events (PMID: 14734594).
• Winblad and Poritis 1999 studied 166 nursing home patients with severe dementia for 12 weeks. About half had Alzheimer's disease and half vascular dementia. Patients received 10 mg a day or placebo, and both primary endpoints favored memantine: 73 against 45 percent had a positive response on the physician-rated global impression of change (p<0.001), and care dependence improved by 3.1 against 1.1 points (p=0.016) (PMID: 10885864). The label notes that this trial did not assess cognitive function as a planned endpoint (US prescribing information for memantine hydrochloride tablets).
• In Grossberg 2013, a 24-week trial in 677 patients on a cholinesterase inhibitor, a once-daily 28 mg extended-release capsule beat placebo on both co-primary endpoints, the Severe Impairment Battery (difference 2.6 points, p=0.001) and the CIBIC-Plus (p=0.008), but not on the secondary daily-living scale (p=0.177) (PMID: 23733403).
• DOMINO-AD (Howard 2012) followed 295 community-dwelling patients on donepezil for 52 weeks. Patients assigned memantine scored 1.2 points higher on the Standardized Mini-Mental State Examination (P<0.001) and 1.5 points better on the Bristol daily-living scale (P=0.02) than those assigned memantine placebo; both differences were smaller than the trial's minimum clinically important differences of 1.4 and 3.5 points. Adding memantine to donepezil gave no significant benefit over donepezil alone (PMID: 22397651).

Mild to moderate Alzheimer's disease

• In Peskind 2006, a 24-week trial in 403 outpatients at 42 US sites, memantine at 20 mg a day beat placebo on both primary endpoints, ADAS-cog cognition and the CIBIC-Plus global rating; 9.5 percent on memantine and 5.0 percent on placebo stopped because of adverse events (PMID: 16861375).
• In Porsteinsson 2008, a 24-week trial in 433 patients on a cholinesterase inhibitor (donepezil, rivastigmine or galantamine), memantine showed no significant difference from placebo on any primary or secondary measure (PMID: 18288936).
• TEAM-AD (Dysken 2014) followed 613 patients on a cholinesterase inhibitor at 14 Veterans Affairs medical centers for a mean of 2.27 years. Daily-living scores declined 1.98 units less with memantine than with placebo, a difference that was not significant (P=.40), while vitamin E slowed decline by 3.15 units (P=.03) (PMID: 24381967).
• The 2019 Cochrane review also covered mild Alzheimer's disease (MMSE 20 to 23): mainly moderate-certainty evidence from post hoc subgroups of up to four studies with about 600 people showed probably no difference from placebo in cognition, daily living or behavior (PMID: 30891742).

Vascular dementia

• In MMM 300 (Orgogozo 2002), a 28-week trial in 321 patients in France, ADAS-cog scores improved by 2.0 points relative to placebo, but the second primary endpoint, the CIBIC-plus global rating, did not differ significantly (60 against 52 percent improved or stable, P=0.227) (PMID: 12105362).
• MMM 500 (Wilcock 2002) was a 28-week trial in 579 patients in the UK. ADAS-cog favored memantine by 1.75 points, while global impression of change did not differ; adverse events affected 77 percent on memantine and 75 percent on placebo, most often dizziness (11 against 8 percent) (PMID: 12409683).

Parkinson's disease dementia and dementia with Lewy bodies

• Aarsland 2009 was a 24-week trial in 72 patients in Norway, Sweden and the UK. The primary endpoint, clinical global impression of change, favored memantine (mean difference 0.7, p=0.03); among secondary measures only attention speed differed (PMID: 19520613).
• Emre 2010 ran for 24 weeks in 199 patients in eight countries and defined no primary endpoint. Global impression and behavior scores improved more with memantine in dementia with Lewy bodies (p=0.023 and p=0.041) but not in Parkinson's disease dementia or in the whole group, and most cognitive tests did not differ (PMID: 20729148).

Other conditions studied

• RTOG 0614 (Brown 2013) enrolled 508 adults with brain metastases having whole-brain radiotherapy. Memantine or placebo was started within 3 days of the start of radiotherapy and given for 24 weeks. The primary endpoint, decline in delayed recall at 24 weeks, was smaller with memantine but not significantly so (P=.059, with only 149 patients analyzable), while time to cognitive decline was longer (hazard ratio 0.78, P=.01) (PMID: 23956241).
• MEADOWS (Hanney 2012) enrolled 173 adults over 40 with Down syndrome. Over 52 weeks, cognition and function declined equally on memantine and placebo, and the authors concluded that memantine is not an effective treatment in this group (PMID: 22236802).
• For autism spectrum disorders in children, the US label reports that memantine failed to show efficacy in two 12-week controlled studies of 578 patients aged 6 to 12 (US prescribing information for memantine hydrochloride tablets).
• In an 8-week trial in 32 patients with major depression (Zarate 2006), memantine had no effect on depression ratings compared with placebo (PMID: 16390905).

Healthy people and nootropic use

• A systematic review (Repantis 2010) found only seven small single-dose studies of memantine in healthy people, and the reviewers judged them inadequate to answer whether memantine enhances cognition in healthy individuals (PMID: 20193764).
• Chan Kwong 2020 tested memantine in a sleep deprivation model, a double-blind crossover study in 36 healthy volunteers. After 15 days of memantine at 10 mg a day, N-back accuracy fell less after 24 hours without sleep than on placebo (minus 1.4 against minus 3.0 percent, P=0.027), and reaction time on a visual processing task rose less (16.1 against 41.3 milliseconds, P=0.034) (PMID: 32332458).
• Natter and Michel 2020 analyzed online self-reports. Of 136 memantine experiences posted on reddit.com, 42 described nootropic use and 39 recreational use, and the authors wrote that memantine misuse seems to be a growing phenomenon (PMID: 32602152).

Stopping memantine

A 2021 Cochrane review of withdrawing dementia drugs found no trial that tested stopping memantine alone and no evidence to guide decisions about discontinuing memantine (PMID: 35608903).

Heart rate and rhythm

The French pharmacovigilance database held 36 reports of heart-related adverse reactions to memantine between its launch in France in July 2003 and May 2007, 18 of them slow heart rate (bradycardia); in 7 of these memantine was the only suspected drug, and all but one resolved after memantine was stopped (PMID: 18500725). In 57 healthy Korean subjects, two weeks of memantine at 20 mg a day caused no clinically significant change in the QT interval (PMID: 33761171).

Seizures and postmarketing reports

In clinical trials, seizures occurred in 0.2 percent of memantine-treated and 0.5 percent of placebo-treated patients, and memantine has not been systematically studied in people with seizure disorders (US prescribing information for memantine hydrochloride tablets). Postmarketing reports listed in the label include agranulocytosis, leukopenia, pancytopenia, thrombocytopenia, thrombotic thrombocytopenic purpura, congestive heart failure, pancreatitis, hepatitis, suicidal ideation, acute renal failure and Stevens-Johnson syndrome, without an established causal link (US prescribing information for memantine hydrochloride tablets).

Overdose

The largest known memantine ingestion in the label was 2.0 grams, taken with antidiabetic medicines, and the patient had coma, double vision and agitation and then recovered; reported overdose signs include agitation, confusion, slow heart rate, ECG changes, raised blood pressure, psychosis and visual hallucinations (US prescribing information for memantine hydrochloride tablets).

Animal findings

In rats, memantine caused nerve cell vacuolation and death in two areas of the cerebral cortex, as other NMDA receptor antagonists do, and memantine combined with donepezil caused more neurodegeneration in female rats than memantine alone; the label states that the relevance to humans is unknown (US prescribing information for memantine hydrochloride tablets). Rats exposed during pregnancy had offspring with lower body weight and less bone formation at doses with minimal maternal toxicity, and no adequate data exist in pregnant women (US prescribing information for memantine hydrochloride tablets).

Self-reported use outside medical care

An analysis of 136 memantine experiences posted on the forum reddit.com from 2010 to November 2019 classed 87 as self-medication of off-label conditions, 42 as nootropic use and 39 as recreational use, and 77 of 100 chronic users reported at least one side effect (PMID: 32602152).