Augmentation With Memantine in Obsessive-Compulsive Disorder

Obsessive-compulsive disorder (OCD) is not easy to treat. Whereas serotonin reuptake inhibitor (SRI) drugs are first-line agents for OCD, many patients require SRI augmentation, usually with an atypical antipsychotic drug, in addition to cognitive behavioral therapy. However, antipsychotic drugs are associated with a range of adverse effects. So other augmentation agents have also been studied. Memantine is one such agent that has been examined in uncontrolled and controlled trials. In this connection, Modarresi et al1 described a systematic review and meta-analysis of memantine augmentation in moderate to severe OCD. They found that memantine augmentation was associated with very large benefits. If their findings are valid, then the approach to the management of OCD could change.

This article therefore presents a critical examination of the meta-analysis1 and its findings. Readers are also referred to earlier articles in this column on how to critically read a meta-analysis.2-5

Meta-Analysis: Memantine Augmentation in Obsessive-Compulsive Disorder

Modarresi et al1 searched scientific electronic databases, clinical trial registries, and reference lists for trials of the use of memantine as an augmentation agent in adults with moderate to severe OCD. They identified 3 small, open-label, uncontrolled trials; 1 small, single-blind, nonrandomized controlled trial; and 4 small-to-medium, double-blind, randomized controlled trials (RCTs).

The primary outcome was the mean reduction, across trials, in memantine-treated patients, in the Yale-Brown Obsessive Compulsive Scale (Y-BOCS) total score. Memantine augmentation was associated with a very large benefit: the mean improvement after at least 8 weeks of augmentation was 11.7 (95% CI, 8.3-15.1) Y-BOCS points, equivalent to a nearly 40% mean reduction in illness severity. This analysis was based on data from 125 patients in 8 trials. Visual inspection of the funnel plot suggested publication bias.

There were several secondary analyses, such as the examination of outcomes after stratification for SRI refractoriness, after stratification for memantine dose, and after stratification for comorbid symptoms. Meta-regression examined outcomes as a function of treatment duration. The most important analysis, the pooling of data from RCTs, was presented as the final secondary analysis.

There were 4 RCTs6-9 in which 68 patients received placebo and 67 patients received memantine in the target dose of 5-10 mg/d (1 RCT) or 20 mg/d (3 RCTs) for 8 weeks (1 RCT), 12 weeks (2 RCTs), or 16 weeks (1 RCT). Only 1 of these trials9 recruited only SRI-refractory patients; 2 trials7,8 placed no restrictions on prior treatment exposure and response thereto, and 1 trial,6 in fact, specifically recruited patients who had not received psychotropic medications during the previous 6 weeks. Very unusually, one trial8 was conducted specifically during the manic phase of bipolar type I disorder.

A random effects meta-analysis of findings from these 4 RCTs showed that memantine augmentation was substantially superior to placebo by a mean of 7.8 (95% CI, 2.6-13.0) Y-BOCS points. Heterogeneity was high (I2 = 84%). Treatment response was declared in 54 (81%) of 67 memantine patients as compared with only 13 (19%) of 68 placebo patients. This translates to a number needed to treat (NNT) value of 1.6 (stated as 1.5 by the authors); that is, fewer than 2 patients with moderate to severe OCD must receive memantine augmentation for 1 extra patient to respond.

Critical Comments

Modarresi et al1 did not present a pooled effect size; nevertheless, the pooled response rate of 81% vs 19% for memantine vs placebo, with a corresponding NNT of 1.6, is so striking that the prudent reader would want to take a closer look at the data. There are two important concerns.

The first concern is obvious, and the authors of the meta-analysis1 themselves commented on it in their discussion: all 4 RCTs were conducted in Iran. Whereas the authors1 stated that this limits the generalizability of the findings, there is a deeper issue at stake. During the past 10-15 years, well over a hundred psychopharmacology RCTs have been published from this part of the world. These RCTs have examined a wide range of drugs for a wide range of indications, and despite small sample sizes (such as n = 30), which usually indicates inadequate statistical power, the results of the RCTs have almost always statistically significantly favored the intervention. This is particularly striking when one observes that a substantial number of the RCTs examined unusual rather than conventional interventions for psychiatric disorders. The most parsimonious interpretation of these results, therefore, is that one needs to view them with caution and seek replication from other parts of the world.

The second concern is not at all obvious, and the reader needs to be diligent in examining the RCTs6-9 on which the meta-analysis1 was based. To cut a long story short, each RCT presented a completer analysis; that is, they omitted the data of patients who dropped out of treatment. In the Ghaleiha et al6 RCT, 4 out of 42 patients dropped out; in the Haghighi et al7 RCT, 11 out of 40 patients dropped out; in the Sahraian et al8 RCT, 20 out of 58 patients dropped out; and in the Modarresi et al9 RCT, 2 out of 32 patients dropped out.

As a moderating note to the previous paragraph, in 1 RCT,8 the authors stated that they performed an "intention-to-treat analysis with at least 1 assessment after the baseline assessment." This effectively eliminated 15 out of 20 dropouts, with the remaining 5 dropouts presumably included in the intent-to-treat sample. Thus, the analysis was very close to a completer analysis. The study, however, was additionally problematic because the authors did not state how missing values were imputed. Furthermore, nowhere in their text and tables did they state the sample sizes on which specific numbers (eg, 4-weekly ratings, response percentages) were based. Therefore, extraction of data for meta-analysis would have involved assumptions based on the CONSORT diagram.

The internal validity of a study is compromised when data from dropouts are ignored in the analysis.10 So a meta-analysis that is based on compromised data will yield results of questionable validity. In a nutshell, we can no longer confidently consider memantine as a potential augmentation agent in OCD.

Other Comments

Modarresi et al1 included only those RCTs that recruited patients with a Y-BOCS score of 16 or higher. This excluded the large 8-week RCT (n = 99) of Farnia et al,11 which recruited patients with Y-BOCS scores of 15 and higher and in which memantine augmentation of fluoxetine was found to be no better than placebo augmentation.

In the Modarresi meta-analysis,1 the primary outcome was the mean reduction, across all trials, of the Y-BOCS total score in patients who had received memantine. This is an outcome of questionable value because it includes all components of a placebo response, including a true placebo response, regression to the mean, ratings contaminated by Hawthorne and expectancy effects, and other contaminants.12 The meta-analysis1 should have restricted itself to the examination of RCT data. In this context, as observed in an earlier article in this column,13 readers may sometimes need to consider whether the truly relevant clinical outcome is different from that set as the primary outcome by the authors of a meta-analysis.

One RCT7 stated in its title that it had been conducted in refractory OCD. However, nowhere in the text was this claim substantiated, and, in fact, in the discussion, the authors stated that the patients had suffered from severe OCD "for the first time in their lives" and that they were therefore unclear about whether memantine augmentation might be helpful for "therapy-resistant and refractory OCD."

Finally, one wonders whether the RCT8 that was conducted during the manic phase of bipolar I disorder belonged in the meta-analysis.1 In this RCT,8 memantine was used as an augmentation treatment with lithium, olanzapine, and clonazepam as primary treatments. SRIs were not prescribed. So, if the positive findings of the 4 RCTs6-9 are valid, a conclusion should be that memantine is effective in OCD regardless of what drug it augments; or, by extension, memantine may benefit OCD in monotherapy!

Take-Home Message

A meta-analysis is only as good as the studies on which it is based. There are concerns about the data and the analyses in all the source RCTs on which the meta-analysis of Modarresi et al1 was based. Therefore, the findings of this meta-analysis, impressive though they are, cannot guide clinical practice. The case for memantine as an augmentation agent in OCD remains unestablished.

Published online: December 3, 2019.

Each month in his online column, Dr Andrade considers theoretical and practical ideas in clinical psychopharmacology with a view to update the knowledge and skills of medical practitioners who treat patients with psychiatric conditions.

Department of Clinical Psychopharmacology and Neurotoxicology, National Institute of Mental Health and Neurosciences, Bangalore, India (candrade@psychiatrist.com).
Financial disclosure and more about Dr Andrade.

REFERENCES

1.Modarresi A, Chaibakhsh S, Koulaeinejad N, et al. A systematic review and meta-analysis: memantine augmentation in moderate to severe obsessive-compulsive disorder. Psychiatry Res. 2019;282:112602. PubMed CrossRef

2.Andrade C. Memantine as an augmentation treatment for schizophrenia: limitations of meta-analysis for evidence-based evaluation of research. J Clin Psychiatry. 2017;78(9):e1307-e1309. PubMed CrossRef

3.Andrade C. The use of statins for antipsychotic augmentation in schizophrenia: examination of meta-analyses with flawed methods and conclusions. J Clin Psychiatry. 2018;79(5):18f12562. PubMed CrossRef

4.Andrade C. Antidepressants, mood stabilizers, antipsychotics, and the risk of cataract. J Clin Psychiatry. 2019;80(1):19f12744. PubMed CrossRef

5.Andrade C. Anti-inflammatory treatments for depression: perspectives on how to read a meta-analysis critically. J Clin Psychiatry. 2019;80(3):19f12907. PubMed CrossRef

6.Ghaleiha A, Entezari N, Modabbernia A, et al. Memantine add-on in moderate to severe obsessive-compulsive disorder: randomized double-blind placebo-controlled study. J Psychiatr Res. 2013;47(2):175-180. PubMed CrossRef

7.Haghighi M, Jahangard L, Mohammad-Beigi H, et al. In a double-blind, randomized and placebo-controlled trial, adjuvant memantine improved symptoms in inpatients suffering from refractory obsessive-compulsive disorders (OCD). Psychopharmacology (Berl). 2013;228(4):633-640. PubMed CrossRef

8.Sahraian A, Jahromi LR, Ghanizadeh A, et al. Memantine as an adjuvant treatment for obsessive compulsive symptoms in manic phase of bipolar disorder: a randomized, double-blind, placebo-controlled clinical trial. J Clin Psychopharmacol. 2017;37(2):246-249. PubMed CrossRef

9.Modarresi A, Sayyah M, Razooghi S, et al. Memantine augmentation improves symptoms in serotonin reuptake inhibitor-refractory obsessive-compulsive disorder: a randomized controlled trial. Pharmacopsychiatry. 2018;51(6):263-269. PubMed CrossRef

10.Streiner DL. The case of the missing data: methods of dealing with dropouts and other research vagaries. Can J Psychiatry. 2002;47(1):68-75. PubMed CrossRef

11.Farnia V, Gharehbaghi H, Alikhani M, et al. Efficacy and tolerability of adjunctive gabapentin and memantine in obsessive compulsive disorder: double-blind, randomized, placebo-controlled trial. J Psychiatr Res. 2018;104:137-143. PubMed CrossRef

12.Andrade C. There’s more to placebo-related improvement than the placebo effect alone. J Clin Psychiatry. 2012;73(10):1322-1325. PubMed CrossRef

13.Andrade C. Gestational exposure to benzodiazepines, 2: the risk of congenital malformations examined through the prism of compatibility intervals. J Clin Psychiatry. 2019;80(5):19f13081. PubMed