Meta-Analysis Conundrums

For patients whose psychosis doesn’t adequately respond to antipsychotics, especially clozapine, treatment options include electroconvulsive therapy, cognitive-behavior therapy, and augmentation with another medication. Investigators have done cotreatment trials with other antipsychotics, anticonvulsant/mood stabilizers, and even the antibiotic minocycline and anti-inflammatory agents such as aspirin. The trials are small in many cases, so meta-analyses allow various trials to be pooled. Two recent publications give some new guidance but also may sow confusion.

One team examined clinical trials of the anticonvulsants lamotrigine, topirimate and valproate added to clozapine. They included 22 randomized controlled trials (RCTs) comprising 1227 subjects; 613 received clozapine alone, whereas the others received, in order of frequency, valproate, lamotrigine or topirimate. The primary outcome was change in Positive and Negative Syndrome Scale (PANSS) or Brief Psychiatric Rating Scale (BPRS) total score. The analysis showed no significant effect for lamotrigine, confirming a previous meta-analysis, but also confirming a prior analysis, topirimate was superior to clozapine alone for reduction in total, positive and negative psychotic symptoms ratings. Sodium valproate, but not magnesium valproate, was superior to clozapine alone in reducing total and positive psychotic symptom scores; it was not effective for negative symptoms. In terms of tolerability, topirimate but not valproate had a significant all-cause discontinuation rate compared to clozapine monotherapy.

One of the limitations of this pooled data set was that many of the patients were not clearly designated as having clozapine resistance; for instance, clozapine serum levels were not recorded. Another concern was that all the trials for valproate augmentation were done in China, and the generalizability to other settings and ethnic groups is uncertain.

The anticonvulsant meta-analysis is one of many that has examined antipsychotic augmentation, which prompted a group led by Christoph Correll to do a meta-meta-analysis. Summarizing this article will not do it justice, so I recommend reading it in its entirety. The investigators looked at 29 existing meta-analyses of trials of augmentation of any antipsychotic with any of 42 medications including, mood stabilizers, anticonvulsants, antidepressants, minocycline, a second antipsychotic, or various hormones such as estrogenic agents. They applied a novel method to assess the quality of the meta-analyses, AMSTAR-Plus Content. Five of the meta-analyses looked at augmentation of clozapine.

In combination with clozapine for positive symptoms of psychosis, only glycine, an amino acid which modulates the NMDA glutamate receptor, had a significant effect size. No treatments showed efficacy in combination with clozapine for total psychopathology or negative symptom scores. In combination with non-clozapine antipsychotics, lamotrigine, estrogenic agents, mirtazapine and a few others showed efficacy.

An important finding is buried in the discussion: “When all this metanalytic literature was compared regarding the quality of its meta-analyzed content, the effect sizes were inversely correlated with the study quality, reducing confidence in these affirmative recommendations.” In other words, many of the studies in this uber-study were small or contained biases, and those studies tended to overrate the effects of the adjunctive treatments. They point out that individual patients may benefit from specific interventions, but the evidence to guide treatment selection is lacking. This leaves the clinician without clear direction for the most difficult-to-treat patients. In an accompanying editorial, however, Wolfgang Fleishhacker suggests that this meta-analysis does not necessarily invalidate all preceding analyses.


Zheng W, Xiang YT, Yang XH, Xiang YQ, de Leon J. Clozapine Augmentation with Antiepileptic Drugs for Treatment-Resistant Schizophrenia: A Meta-Analysis of Randomized Controlled Trials. J Clin Psychiatry. 2017;78(5):e498-e505. Abstract

Correll CU, Rubio JM, Inczedy-Farkas G, Birnbaum ML, Kane JM, Leucht S. Efficacy of 42 Pharmacologic Cotreatment Strategies Added to Antipsychotic Monotherapy in Schizophrenia: Systematic Overview and Quality Appraisal of the Meta-Analytic Evidence. JAMA Psychiatry. 2017;74(7):675-684. Astract

Riverview Refractory Psychosis Data Presented at International Psychiatry Congress



View the poster in a PDF file here


Treatment-resistant psychosis is a challenge to psychiatry and a substantial burden to health-care systems. The province of British Columbia in Canada has publicly funded, universal health care, and patients with treatment-resistant psychosis may receive care in a specialized residential program. Between 1993 and 2011, 663 patients were admitted to this program; this cohort contains one of the largest known series of patients with treatment-resistant schizoaffective disorder.

All patients were evaluated by a psychiatrist, social worker, pharmacist, nurse, general physician, and neuropsychologist. Records from previous hospital admissions were reviewed and all information was presented at a multidisciplinary conference. This resulted in a consensus DSM-III or -IV multiaxial diagnosis and a detailed treatment plan. Ratings of symptoms and functioning at admission and discharge included the Positive and Negative Syndrome Scale (PANSS), the Global Assessment of Functioning Scale, the Social and Occupational Functioning Scale, and the Clinical Global Impression of Severity. A research psychologist compiled all data at the time of each patient’s hospitalization.

Patients who did not complete treatment or had a diagnosis other than schizophrenia (SZ), schizoaffective (SZA) or mood disorder (MD) were excluded; the following describes 551 included patients (SZ = 63%, SZA = 29%, MD = 8%). More than half were male (59%), and the mean duration of hospitalization was 30 weeks. The proportion receiving clozapine increased from 21% at admission to 61% at discharge. Those with a MD were less likely to receive clozapine than either SZ or SZA (SZ = 64%, SZA = 61%, MD = 41%). In each diagnostic group, both antipsychotic polypharmacy and the ratio of prescribed daily dose to defined daily dose (PDD/DDD) of antipsychotic medication decreased during hospital stay (polypharmacy: SZ: 52% to 16%, SZA: 52% to 14%, MD: 43% to 0%; PDD/DDD: SZ: 2.1 to 1.6, SZA: 2.1 to 1.4, MD: 1.6 to 1.1). The use of mood stabilizers declined in all groups, but antidepressant use declined only in SZ and SZA. Mean total PANSS score declined in all diagnostic groups, but most in MD, least in SZ, and intermediate in SZA.

In an intensive inpatient program for treatment-resistant psychosis, aggregate improvement occurred despite global reduction in medications while clozapine use nearly tripled. Lower total antipsychotic dose correlated with greater improvement at discharge.

Polypharmacy increases risk of NMS

A study from the UK examined 67 cases of neuroleptic malignant syndrome and 254 controls (1). The researchers made several interesting findings, in particular that patients on 3 or more antipsychotics had an odds ratio of 5.4 for NMS. This, along with evidence for increased risk of diabetes in patients on multiple antipsychotics (2), give reason to be cautious about antipsychotic polypharmacy.


1. Retrospective chart review on exposure to psychotropic medications associated with neuroleptic malignant syndrome. Su Y-P, Chang C-K, Hayes RD, et al. Acta Psychiatr Scand. Published online 15 November 2013. Abstract

2. Treatment with antipsychotics and the risk of diabetes in clinical practice. Kessing LV, Thomsen AF, Mogensen UB, Andersen PK. Br J Psychiatry. 2010;197:266–271. Full text

12-month trial of clozapine augmentation with aripiprazole or haloperidol

The latest trial for clozapine-resistant patients comes from Italy (1). The Clozapine Haloperidol Aripiprazole Trial (CHAT) is a multicentre, randomized, 12-month trial of patients with schizophrenia and persistent positive symptoms despite at least six months of clozapine therapy at a minimum dose of 400 mg. Outcome measures included the Brief Psychiatric Rating Scale and, for subjective rating of adverse effects, the Liverpool University Neuroleptic Side Effect Rating Scale (LUNSERS). The primary outcome was treatment discontinuation, “a pragmatic and reliable estimate of treatment efficacy and tolerability” according to the authors.

Based on a power calculation, the investigators intended to recruit 194 participants but randomized 106 patients, about 65% men and with a mean age of 41 years. In addition to clozapine, patients received either haloperidol or aripirpazole for 12-months. Baseline demographics and BPRS scores did not significantly differ between the groups. The dose of adjunctive antipsychotic was adjusted according to clinical response, and the patients could receive any necessary additional medications including mood stabilizers and antidepressants. Although the investigators did not report the mean dose of adjunctive antipsychotics in this publication, they reported in 2011 that at month 3, the mean aripiprazole dose was 11.8 mg and the mean haloperidol dose was 2.8 mg (2).

During the study, 19 aripirpazole and 15 haloperidol patients withdrew, which was not significantly different. At 6 months, the mean BPRS score in the aripirpazole group had decreased by 8.8, and in the haloperidol group, by 8.1; the improvement persisted at 12 months. By 3 months the mean LUNERS score decreased significantly in the aripirpazole group but not in the haloperidol group, which indicated that patients taking aripirpazole reported a decrease in adverse effects whereas those taking haloperidol did not.
Both aripirpazole and haloperidol showed effectiveness in augmenting clozapine, but aripirpazole was better tolerated by self-report. Despite this apparent preference for aripirpazole, the groups had a comparable rate of treatment discontinuation: about a third of patients dropped out, most during the first 6 months of augmentation. The findings of the study are limited by lack of statistical power and of a placebo arm. The researchers claim, however, that CHAT is the longest non-industry-sponsored RCT of treatment-resistant schizophrenia.


1. Cipriani A, Accordini S, Nose M, et al. Aripiprazole versus haloperidol in combination with clozapine for treatment-resistant schizophrenia. J Clin Psychopharmacol. 2013;33: 533-537. Abstract

2. Barbui C, Accordini S, Nose M, et al. Aripiprazole versus haloperidol in combination with clozapine for treatment-resistant schizophrenia in routine clinical care: a randomized, controlled trial. J Clin Psychopharmacol. 2011;31: 266-273. Abstract

Benzoate and not benzodiazepines for antipsychotic augmentation

Dr. Jari Tiihonen, Professor of Clinical Neuroscience at the Karolinska Institure in Stockholm, Sweden, discussed findings from his research on polypharmacy in people with schizophrenia on September 20 at the 7th Annual Pacific Psychopharmacology Conference. Using national databases in Finland, his team examined prescription medications and all-cause mortality in a cohort of 2,588 patients during 7 years. Antipsychotic polypharmacy was not associated with increased mortality, and antidepressant use was associated with a decreased risk of death by suicide. In contrast, benzodiazepine use was associated with an increase in suicide and all-cause mortality. Dr. Tiihonen speculated that the sedative effects of benzodiazepines may predispose to unintentional injuries, and withdrawal symptoms may increase agitation and dysphoria which could elevate the risk of suicide.

As for new findings in treatment-resistant schizophrenia, Dr. Tiihonen mentioned the recently published clinical trial of famotidine (see my blog post of July 14). His group is now attempting a replication for which they plan to recruit 140 patients. He also called attention to a randomized clinical trial of sodium benzoate at a dose of 1 gram daily which was presented at the December, 2012 meeting of the American College of Neuropsychopharmacology. The researchers added benzoate or placebo to treatment of 52 patients with chronic schizophrenia “stabilized with antipsychotics for 3 months or longer.” According to the abstract posted on the ACNP Web site, at 6 weeks of treatment, the benzoate group showed significant improvement in PANSS total, positive, and negative scales, as well as in overall neurocognition. The effect size on PANSS total was 1.76, which is very large. It’s not clear if these patients met the usual definition of treatment resistance, i.e. poor response to adequate trials of two antipsychotics, but we should know more soon as the study is in press.


Tsai GE, Lane H-Y, Green MF. A randomized, double-blind, placebo-controlled add-on treatment of benzoate, a D-amino acid oxidase Inhibitor, for schizophrenia. Neuropsychopharmacology. 2012; 38: S198-S313. Available at: (poster T162)

Pimozide is ineffective for clozapine augmentation

Partial or non-response to clozapine is a challenging clinical situation. For these patients, who are considered refractory or ultra-resistant, we have limited options. Available evidence for augmenting clozapine is discouraging, but even negative trial results are valuable as a guide for what not to do. Exposing patients to ineffective treatments increases both costs and risk of adverse effects.

Pimozide, a potent D2 receptor antagonist, was found to be effective in a 1997 open-label clinical trial in partial clozapine responders. In a 2011 double-blind, placebo-controlled, 12-week trial in patients with partial or non-response to clozapine, pimozide at a mean dose of 6.5 mg daily was ineffective (1). A different U.S. group just published another randomized, double-blind, placebo-controlled trial of pimozide at 4 mg daily in patients with partial clozapine response (2). Using the BPRS, the Schedule for the Assessment of Negative Symptoms, and evaluations of verbal memory, working memory and executive function, the investigators found no significant differences between the groups at 12 weeks, although both showed improvement in the BPRS over time.

With two negative trials, it seems that pimozide as a clozapine augmentation agent can be put to rest. In fact, the entire strategy of adding first-generation D2 antagonists to clozapine for partial or non-repsonse is dubious.


1. Friedman JI, Lindenmayer JP, Alcantara F, et al. Pimozide augmentation of clozapine in patients with schizophrenia and schizoaffective disorder unresponsive to clozapine monotherapy. Neuropsychopharmacology. 2011;36:1289-1295. Full text

2. Gunduz-Bruce H, Oliver S, Gueorguieva R, et al. Efficacy of pimozide augmentation for clozapine partial responders with schizophrenia. Schizophr Res. 2013;143:344-347. Abstract

A Canadian cohort of patients with treatment-resistant schizophrenia

Researchers at the Centre for Addiction and Mental Health in Toronto characterized a cohort of people with schizophrenia who had enrolled in a genetics study. In this convenience sample of 478 subjects, 156 were considered treatment resistant (TR) according to American Psychiatric Association guidelines. The APA guidelines define treatment resistance as “little or no symptomatic response to multiple (at least two) antipsychotic trials of an adequate duration (at least 6 weeks) and dose (therapeutic range).”

The investigators found no correlation between treatment resistance and sex; family history of psychosis; schizophrenia subtype; cannabis, alcohol or drug use; or number of cigarettes consumed daily. However, the TR patients had been ill for a mean of 21 years compared with 15 years for the non-TR group (P < .001). Among patients identified as having white European ancestry, 37% were TR, whereas 18% of nonwhites were TR (P =.03). Several treatment factors were significantly correlated with treatment resistance. In the TR group, 33% were on clozapine compared with 13.3% in the non-TR group, and 25% of TR patients were on more than one antipsychotic, double the rate in the non TR group. Ten percent of the TR patients were on clozapine and at least one other antipsychotic. Furthermore, the TR patients had a mean of 3 failed medication trials, whereas the non-TR patients had a mean of 0.5 failed trials. This nonrandom sample is not necessarily representative of all TR patients, so the significance of the lower rate among non-white patients is unclear. The study corroborates previous research indicating that treatment resistance occurs in chronic patients, and that polypharmacy is used possibly at the expense of clozapine. References Teo C, Borlido C, Kennedy JL, De Luca V. The role of ethnicity in treatment refractory schizophrenia. Compr Psychiatry. 2013;54(2):167-172. Link to abstract.

Practice guideline for the treatment of patients with schizophrenia, second edition. Am J Psychiatry. 2004;161(Suppl):1-56.