Memantine in OCD: A Meta-analysis

A meta-analysis pools six trials and finds mostly noise, with one hopeful subgroup

STUDY: Lyndon S et al, CNS Drugs 2026 Jul 22

STUDY TYPE: Systematic review and meta-analysis

FUNDING: Independent

Background

Many clinicians have written me about what they find useful for augmentation in OCD. Ondansetron and memantine stand out, with a few votes for n-acetylcysteine. This analysis explores memantine, an NMDA receptor antagonist approved for Alzheimer’s that targets glutamate signaling in the brain circuits thought to drive obsessions and compulsions.

Earlier meta-analyses found it effective, but those pooled a small handful of mostly positive trials.

The Study

• Six double-blind, placebo-controlled trials, 288 adults with OCD total
• Adjunctive memantine (10 or 20 mg/day) versus placebo, added to an SSRI or clomipramine
• 8 to 16 weeks of treatment, with OCD severity (Y-BOCS) as the main outcome

Results

Overall, memantine didn’t beat placebo. The pooled difference in Y-BOCS scores was 3.74 points in memantine’s favor, but the confidence interval crossed zero (-9.95 to 2.47), and heterogeneity between trials was enormous.

That heterogeneity traced back to who was studied. In two trials that enrolled patients already resistant to adequate serotonergic treatment, memantine showed a much larger effect, roughly 9 points on the Y-BOCS. In the four trials that started patients on a new SSRI alongside memantine or placebo, the drug added almost nothing, about 1 point. The resistant-patient signal didn’t reach statistical significance, but that’s not surprising with only two trials behind it.

Response rates (at least 35% symptom reduction) numerically favored memantine, but the estimate was wildly unstable, with a risk ratio of 3.62 and a confidence interval spanning 0.16 to 80.71, essentially uninformative.

Side Effects

Tolerability looked good. It was comparable to placebo in terms of adverse events, discontinuation for adverse events, and all-cause dropout.

Limitations

Only six trials, all conducted in Iran, several in the same city. Treatment-resistance criteria varied across trials and weren’t always reported. Certainty of evidence was rated very low for the main efficacy outcomes.

Practice Implications

Here’s the score so far:

1. Memantine may work when SSRIs fail, but is not for everyone with OCD.
2. NAC barely has a positive signal, depending on the meta-analysis.
3. Ondansetron has more robust data, but also has a few negative trials.
4. Antipsychotics are preferred by experts because their data is more robust, but only slightly, and their risks are many times greater.