Memantine shows therapeutic potential in children with ASD

Treatment with memantine may improve social functioning in children with autism spectrum disorder (ASD), as shown in a small study.

After 12 weeks, a significantly greater proportion of memantine-treated vs placebo-treated participants achieved treatment response (56.2 percent vs 21 percent; odds ratio [OR], 4.8, 95 percent confidence interval [CI], 1.1–21.2; p=0.03). [JAMA Netw Open 2025;8:e2534927]

Response was defined as a ≥25-percent reduction in total scores on the Social Responsiveness Scale–Second Edition, which was completed by the parent or caregiver, and a score of ≤2 on the Clinical Global Impression–Improvement subscale anchored for ASD, which was completed by a clinician.

“Nearly all of the memantine treatment responders met the a priori response criteria by week 6 of the trial,” the authors noted.

The number needed to treat (NNT) was 3, indicating that one in three memantine-treated youths with ASD would respond to treatment. This very low number “is particularly noteworthy in the context of ASD, in which effective treatments for core symptoms can be challenging to identify,” they said.

Glutamate as predictor of response

Response to treatment differed based on glutamate levels in the pregenual anterior cingulate cortex (pgACC). In the subset of participants with abnormally high pgACC glutamate levels, memantine was associated with 16-fold greater odds of achieving response compared with placebo (80 percent vs 20 percent; OR, 16.0, 95 percent CI, 1.8–143.2; p=0.007). This association was not observed in the subset of participants with average glutamate levels.

The NNT decreased from 3 in the overall population to 2 in the high-glutamate subset.

On receiver operating characteristic (ROC) curve analysis, pgACC glutamate concentration demonstrated high efficiency at identifying responders to memantine treatment. A pgACC glutamate level of 99 IU emerged as the optimal threshold for identifying memantine responders. It had a corresponding sensitivity of 88 percent, specificity of 86 percent, positive predictive value of 88 percent, and negative predictive value of 86 percent.

“These findings suggest a selective response to memantine based on pgACC glutamate levels in youths with ASD,” and point to pgACC glutamate levels as a promising biomarker for identifying youths who may respond to memantine treatment, according to the authors.

Memantine well tolerated

In terms of safety, memantine was well tolerated. The percentages of participants who experienced adverse events (AEs) and those who required rescue medication were similar between the memantine and placebo groups. Treatment-emergent AEs were mild or moderate in severity, with the most common being headache, upper respiratory conditions (eg, cold, infection, or allergies), increased or decreased appetite, suicidal ideation, sedation, and abdominal discomfort.

Severe AEs occurred in two memantine-treated participants, including one with transient severe headaches at weeks 2 and 6 and another with severe worsening of irritability and agitation in the first week at 5 mg/d, leading to early trial termination. Three additional participants in the memantine group withdrew due to worsening anxiety during the titration phase. One placebo-treated participant withdrew at week 2 due to worsening irritability. Notably, one memantine-treated participant successfully achieved treatment response, even when the dose was limited to 15 mg/d due to dizziness.

The study included 42 children ages 8–17 years who had ASD (mean age 13.2 years, 76.2 percent male). These participants were randomly assigned to receive treatment with memantine (n=21) or placebo (n=21). Treatment was taken orally, once daily for 12 weeks. Memantine was titrated to a maximum daily dose of 20 mg over the first 4 weeks and then maintained at the maximum achieved dose until the end of the trial.

Of the participants, 35 were included in the intention-to-treat efficacy analysis (n=16 in the memantine arm, n=19 in the placebo arm), and 33 completed the trial (n=16 and n=17, respectively). 

More studies needed

In an accompanying editorial, Dr Daniel Felsky from the University of Toronto in Toronto, Ontario, Canada, pointed out that the current trial, despite its promising findings, is not definitive and should be replicated in larger groups. [JAMA Netw Open 2025;8:e2534938]

A similar study involving 121 youths ages 6–12 years with ASD showed no significant difference in social functioning between the placebo and memantine (3–15 mg/d) groups over 12 weeks, Felsky noted. [J Child Adolesc Psychopharmacol 2017;27:403-412]

Additionally, “the findings of pgACC glutamate as a predictor of treatment response is complicated by previous work showing potential age-dependent relationships between glutamatergic metabolites and ASD. As such, the current findings … may not generalize to older adult populations with ASD,” he said. [Neurosci Biobehav Rev 2015;52:74-88]

“Generalizability … is a limitation in other respects as well. The recruited population was predominantly male (consistent with prevalence estimates), was mostly White, and excluded youths with intellectual disability,” he added.

Felsky highlighted that the participants in the trial had high levels of comorbid lifetime diagnoses. Specifically, among the 33 who completed the trial, 79 percent had ADHD, 73 percent had ≥2 anxiety disorders, and 67 percent had major depression. Furthermore, 79 percent of the participants had received concomitant pharmacological treatment, with a mean number of 2.3 medications (mostly selective serotonin reuptake inhibitors [65 percent], stimulants [46 percent], and α-2 adrenergic agonists [35 percent]).

“Despite randomization, there remained small treatment group differences in baseline diagnoses and medication use. Given the small sample size, models could not covary for differing psychopathological or medication profiles, and so it remains possible that the findings were influenced by these factors,” he said.