Introduction

Clozapine—considered the most effective antipsychotic—was introduced in the early 1970s for the treatment of schizophrenia. First, clozapine was believed to have not only superior efficacy but also to have overall better tolerability compared to first-generation antipsychotics (FGA) due to a low risk for extrapyramidal symptoms (EPS). However, in 1975, clozapine was voluntarily withdrawn since 17 out of 2660 (0.7%) patients treated with clozapine in Finland developed agranulocytosis and eight patients subsequently died [1]. In 1988, Kane et al. confirmed clozapine’s safety and superiority vs. chlorpromazine in treatment-resistant schizophrenia (TRS) [2], and subsequently, the Federal Drug Agency (FDA) and other health authorities approved its re-introduction for the indication of TRS with regular hematological monitoring.

Evidence-based treatment guidelines for the management of difficult-to-treat schizophrenia currently recommend clozapine [3,4,5]. Nevertheless, definitions of TRS, typically involving two failed trials of different non-clozapine antipsychotics, differ significantly across guidelines [6] as do criteria for TRS in clinical trials: if TRS is operationalized at all, it differs in up to 95% of trials [7]. A lack of consensus is also represented in the extent and frequency of mandatory safety monitoring procedures beyond hematological monitoring during clozapine treatment according to the respective national regulations [6]. Further indications or recommendations, when clozapine can be applied in clinical practice, are poorly harmonized: in certain European countries, (e.g. Germany, the Netherlands) clozapine is indicated for the treatment of Parkinson’s disease psychosis (PDP), whereas in the US it was given a Level B recommendation by the American Academy of Neurology (AAN) for this indication. Furthermore, the FDA approved clozapine as the first agent indicated for suicidality in people with schizophrenia and schizoaffective disorder. Furthermore, the American Psychiatric Association (APA) recommends (1B) that patients with TRS be treated with clozapine and recommends (1B) patients with schizophrenia be treated with clozapine if the risk for suicide attempts or suicide remains substantial despite other treatments and suggests (2C) that patients with schizophrenia be treated with clozapine if the risk for aggressive behavior remains substantial despite other treatments [8]. Of note, clozapine is recommended in some clinical guidelines for treatment-refractory bipolar disorder [9] with an uncertain body of evidence suggesting beneficial effects on e.g. mania, depression, rapid cycling and psychotic symptoms [10].

Even though clozapine is considered one of the most effective medications and is listed in the WHO Model List of Essential Medicines [11], there is frequently a delay in clozapine initiation, leading to poorer mental health and functional outcomes [10, 12], preceded by attempts of polypharmacy treatment without evidence for effectiveness [13].

The scientific literature regarding clozapine is vastly increasing and evidence-based psychiatry might help clinicians to judge the best evidence and decision-makers and clinicians are overstrained by the number of individual studies, reviews and meta-analyses [14].

Thus, with our quantitative meta-review of meta-analyses we aimed at elucidating the state-of-the-art of efficacy, effectiveness, tolerability and management of clozapine and clozapine-related adverse-events in order to synthesize evidence, provide orientation for decision-makers and clinicians and identify treatment gaps for future research.

Methods

Information sources and search

This meta-review was pre-registered on PROSPERO (CRD42020164135). Following the structure of the International Statistical Classification of Diseases and Related Health Problems 10th Revision (ICD-10 WHO Version, 2015), we searched the PubMed/MEDLINE database and the EMBASE databases using the following search terms with limitation to systematic reviews and/or meta-analyses: “clozapine” OR “leponex” OR “clozaril”. The literature searches and selection were independently performed by EW and PiyF and validated by AH. The titles and the abstracts of each citation were screened manually, and the full text of each potentially relevant citation was retrieved for detailed review. Pharmacological or non-pharmacological clozapine augmentation/combination strategies with the purpose of clinical improvement were excluded a priori since evidence in this field was already meta-reviewed by members of our group [15]. Furthermore, studies focusing on genetics and/or pharmacogenetics, brain-imaging studies, cost-effectiveness studies, and animal studies were excluded. Three publications [16,17,18] were added by hand since two were published after the search period [16, 18] and one included sub-analyses for a new domain [17] (see Fig. 1).

Fig. 1: Flowchart for literature search and study selection process [19].
figure 1

Study selection flow diagram of meta-analyses providing quantitative data. Three meta-analyses were manually identified.

Full size image

Eligibility criteria

The inclusion criteria were all meta-analyses published in English between January 1, 1970, and December 19, 2019 (PubMed) and 1970–2019 (EMBASE) with quantitative data of people treated with clozapine alone or clozapine vs any control (clozapine, placebo, or non-clozapine antipsychotics). The major exclusion criteria were the absence of clozapine-specific meta-analytic data. We extracted clozapine-specific meta-analytic data on effectiveness, efficacy, and tolerability of clozapine, management of clozapine, and clozapine-related adverse events. The applied search strategy according to The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [19] is shown in Fig. 1.

Data collection process

After full-text review, one researcher (EW) extracted quantitative data from pairwise meta-analyses with validation by a second researcher (PiyF). Network meta-analytic data was extracted if pairwise analyses were presented. If standardized mean difference (SMD), mean difference (MD), risk difference (RD) > 0 demonstrated a beneficial outcome for clozapine (e.g. more response or less adverse events) then the direction ‘clozapine’, was extracted, however, if <0 then the direction “control” was extracted. If RR, odds ratio (OR), hazard ratio (HR) > 1 meant a beneficial outcome for clozapine (e.g. more response or less adverse events/dropouts) then the direction ‘clozapine’, otherwise ‘control’, was extracted. Furthermore, we grouped outcomes into short-term (up to 12 weeks), medium-term (13–26 weeks), and long-term (over 26 weeks).

Data transformation

The data transformation process was conducted by two authors (EW and SS) with validation by a third author (SL) using R statistical software version 4.0.3 [20] and the package tidyverse version 1.1.3 [21] OR and RD were transformed to RR [22] while HR and incidence rate ratio (IRR) was used as RR. MD was transformed into SMD [23], and in case the total number of participants in the control and experimental group were not given, equal groups were assumed. A beneficial outcome for the experimental intervention was represented with SMD > 0 or OR > 1, and minus or inverse transformations were applied whenever the opposite direction was reported. Due to limited data, adverse events of clozapine add-on strategies were not able to be included in the analyses.

Endpoints

Endpoints were defined as (1) efficacy of clozapine (SMD and RR), (2) tolerability/adverse events of clozapine (SMD and RR), and (3) efficacy of add-on strategies to improve clozapine-related adverse events (SMD and RR).

Methodological quality assessment of included meta-analyses

The Assessing the Methodological Quality of Systematic Reviews 2 (AMSTAR-2) checklist [24] was used independently by two reviewers (EW, PiyF). Disagreements were solved by consensus with a third reviewer (AH). Then, meta-analyses were categorized into different domains according to their objectives, taking into consideration participant characteristics, comparisons, and outcomes. In case of an overlap of two domains within one meta-analysis, categorization was performed with a primary focus on population characteristics (e.g. first-episode schizophrenia) before outcomes (e.g. metabolic outcomes) (see Table 1).

Table 1 Description of included meta-analyses.
Full size table

Results

1078 records were identified and the publications were added manually [16,17,18]. After the removal of duplicates, 959 records remained. A total of 767 records were excluded on the title/abstract level. The remaining 192 publications were retrieved as full texts and were further assessed for eligibility. From these, 112 records were included in this meta-review. 80 records were excluded as they met at least one of the exclusion criteria on full text-level (see Fig. 1). Since no evidence is considered an important finding according to the Cochrane Handbook [25], two clozapine-specific Cochrane Database reviews/meta-analyses that yielded no quantitative data due to a lack of relevant studies [26, 27], were included in our umbrella review.

Study characteristics/AMSTAR ratings

From the 112 included meta-analyses [10, 16,17,18, 26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59,60,61,62,63,64,65,66,67,68,69,70,71,72,73,74,75,76,77,78,79,80,81,82,83,84,85,86,87,88,89,90,91,92,93,94,95,96,97,98,99,100,101,102,103,104,105,106,107,108,109,110,111,112,113,114,115,116,117,118,119,120,121,122,123,124,125,126,127,128,129,130,131] a majority reported data on clozapine as subgroup or sensitivity analysis, whereas 34 exclusively targeted populations of clozapine users (see Table 1). According to AMSTAR-2, 61 (54.5%) meta-analyses were rated as high-quality. A description of the results of each meta-analysis along with their overall quality is presented in the Supplementary Tables (see Supplementary Tables S1 and S2, https://github.com/sksiafis/clozapine_meta_review).

Endpoints

Efficacy of clozapine (SMD and RR)

Positive symptoms in schizophrenia

Clozapine appears to be superior to FGAs in RCTs (short, medium, and long-term) with small to medium effects sizes [48, 79, 125]. Clozapine appears to be superior to risperidone in Japanese populations with a medium effect size [29, 31]. For TRS, clozapine appears to be not significantly superior to pooled SGAs in observational studies [82], and not significantly superior to other single SGAs [100] in RCTs. When FGAs/SGAs are pooled, clozapine appears to be superior in improving positive symptoms in RCTs in TRS with a small effect size [106] see Fig. 2A).

Fig. 2: Quantitative meta-review of clozapine-specific meta-analytic data: Efficacy.
figure 2

A Positive symptoms. FGA first-generation antipsychotic, k number of studies, L long-term, M medium-term, n number of participants, RCT randomized controlled trial, S short-term, SGA second-generation antipsychotic. Abbreviated study descriptions: Leucht et al., 2009a [76], Leucht et al., 2009b [76], Leucht et al., 2009c [79]. For continuous outcomes, SMD > 0 means a beneficial outcome for clozapine (e.g. more response or less adverse events. B Negative symptoms. FGA first-generation antipsychotic, k number of studies, L long-term, M medium-term, n number of participants, RCT randomized controlled trial, S short-term, SGA second-generation antipsychotic. Abbreviated study descriptions: Leucht et al., 2009a [76], Leucht et al., 2009b [76], Leucht et al., 2009c [79]. For continuous outcomes, SMD > 0 means a beneficial outcome for clozapine (e.g. more response or less adverse events. C Overall symptoms. FGA first-generation antipsychotic, k number of studies, L long-term, M medium-term, n number of participants, RCT randomized controlled trial, S short-term, SGA second-generation antipsychotic. Abbreviated study descriptions: Leucht et al., 2009a [76], Leucht et al., 2009b [76], Leucht et al., 2009c [79]. For continuous outcomes, SMD > 0 means a beneficial outcome for clozapine (e.g. more response or less adverse events. D Global impression. FGA first-generation antipsychotic, k number of studies, L long-term, M medium-term, n number of participants, RCT randomized controlled trial, S short-term, SGA second-generation antipsychotic. For continuous outcomes, SMD > 0 means a beneficial outcome for clozapine (e.g. more response or less adverse events. E Quality of life and global functioning. CGAS Children’s Global Assessment Scale, FGA first-generation antipsychotic, k number of studies, L long-term, M medium-term, MLDL Münchner Lebensqualitäts–Dimensionen–Liste, n number of participants, RCT randomized controlled trial, S short-term, SGA second-generation antipsychotic, SWN Subjective Wellbeing under Neuroleptics Scale, WHO-QOL: WHO-Quality of life. For continuous outcomes, SMD > 0 means a beneficial outcome for clozapine (e.g. more response or less adverse events. F Other symptoms (cognition, hostility, depression) as a continuous outcome. BPRS Brief Psychiatric Rating Scale, CGI clinical global impressions, FGA first-generation antipsychotic, k number of studies, L long-term, M medium-term, n number of participants, PANSS Positive and Negative Syndrome Scale, RCT randomized controlled trial, S short-term, SGA second-generation antipsychotic. For continuous outcomes, SMD > 0 means a beneficial outcome for clozapine (e.g. more response or less adverse events. G Response to treatment. FGA first-generation antipsychotic, k number of studies, L long-term, M medium-term, n number of participants, RCT randomized controlled trial, S short-term, SGA second-generation antipsychotic. For dichotomous outcomes, RR > 1 means a beneficial outcome for clozapine (e.g. more response or less adverse events/dropouts). H Relapse. FGA first-generation antipsychotic, k number of studies, L long-term, M medium-term, n number of participants, RCT randomized controlled trial, S short-term, SGA second-generation antipsychotic. abbreviated study descriptions: Leucht et al., 2009a [76], Leucht et al., 2009b [76], Leucht et al., 2009c [79]. For dichotomous outcomes, RR > 1 means a beneficial outcome for clozapine (e.g. more response or less adverse events/dropouts). I Dropouts. FGA first-generation antipsychotic, k number of studies, L long-term, M medium-term, n number of participants, RCT randomized controlled trial, S short-term, SGA second-generation antipsychotic. For dichotomous outcomes, RR > 1 means a beneficial outcome for clozapine (e.g. more response or less adverse events/dropouts).

Full size image

Negative symptoms in schizophrenia

Clozapine is not superior to SGAs in observational studies [82], but to most FGAs in RCTs with both small and large effect sizes [48, 79, 125]—except short-term data vs chlorpromazine [100, 128]. There is conflicting evidence regarding the superiority of clozapine vs. pooled SGAs in TRS [100, 106] and clozapine appears inferior to quetiapine (short-term, only 2 studies with n total = 142) with medium effect sizes [29, 30, 76] and aripiprazole medium-term in RCTs with a small effect size [61] (see Fig. 2B).

Overall symptoms in schizophrenia

Clozapine appears to be superior to placebo in short-term RCTs with large effect sizes [76, 78], superior to FGAs in RCTs with small to medium effect sizes [44, 48, 79, 99, 125] and to SGAs in observational studies with a small effect size [82] and quetiapine in long-term RCTs with a large effect size [65]. For TRS, clozapine appears to be superior vs. CPZ with a medium effect size [100], superior vs. mixed FGAs/SGAs in RCTs with small effect sizes [85, 106], but the evidence is suggestive that clozapine is not superior vs. other antipsychotics in long-term RCTs [100, 106]. (see Fig. 2C).

Other efficacy measures in schizophrenia

Clozapine has a favorable profile in terms of dropout due to inefficacy compared to placebo with a large effect size [57] and to CPZ with a medium effect size [99] and SGAs, namely risperidone with medium effect sizes [29, 65, 67, 70, 76] and in terms of ACD rates compared to FGAs with small effect sizes [48, 99, 125], grouped SGAs in observational studies with a small effect size [82] and some single SGAs (e.g. risperidone and quetiapine) with small effect sizes [65] (see Fig. 2I).

With regard to relapse, clozapine appears to be superior to FGAs long-term [79, 125], but evidence from meta-analyses is inconsistent [64] (see Fig. 2H). With regard to response, clozapine appears to be superior to placebo with large effect sizes [57, 66], superior to FGAs short-term with small effect sizes [48, 99, 106, 125], but not superior to single SGAs (e.g. quetiapine, risperidone, olanzapine) [29, 30, 61, 67, 70, 122] (see Fig. 2G). As a second-line agent, clozapine appears to be superior to risperidone and other antipsychotics with small effect sizes (see Fig. 2D) [90]. Evidence does not support superiority of clozapine for hospitalization rate vs. SGAs (see Supplementary Fig. 16) or reduction of suicide/self-injurious behavior vs. SGAs in observational studies [82] (see Supplementary Fig. 17), and does not support superiority for anti-suicidal effects in long-term RCTs vs. olanzapine [29], but meta-analytic evidence from one long-term trial (n = 980) showed superior effects of clozapine vs. olanzapine [67] (see Supplementary Fig. 17). Meta-analytic evidence suggests superior effects of clozapine on hostility compared to FGAs in RCTs in mixed short-, medium-, and long-term RCTs with a medium effect size [50] (see Fig. 2F) and on cognition vs. SGAs in TRS in observational studies with a small effect size [82] (see Fig. 2F), whereas mostly nonsignificant effects on cognition compared to FGAs and SGAs [119] were observed in RCTs and even inferior effects vs. single FGAs, e.g. sertindole [89] (see Fig. 2F). With regard to psychosocial functioning, clozapine appears not to have significantly more beneficial effects compared to SGAs [92] (see Fig. 2E). For quality of life, available data is scarce (see Fig. 2E). A detailed report with regard to different disease entities and levels is presented in the Supplementary Results S1 (https://github.com/sksiafis/clozapine_meta_review). For additional outcomes, please see Supplementary Figs. S1–S19.

Other efficacy measures in BP and PDP

No superior efficacy of clozapine vs. other antipsychotics could be shown for mania in bipolar disorder short-term [45] (see Fig. 2F). For PDP, clozapine seems to be superior vs. quetiapine short-term in terms of clinical global impression with large effect sizes [51] (see Fig. 2D). A detailed report with regard to different disease entities and levels is presented in the Supplementary Results S1 (https://github.com/sksiafis/clozapine_meta_review).

Tolerability of clozapine (SMD and RR)

Clozapine is equivocally associated with a significantly higher risk for weight gain with small to medium effect sizes (see Fig. 3A) and an increased risk to develop type 2 diabetes compared to most other antipsychotics [93] and with significantly fewer EPS or use of antiparkinson medication compared to FGAs with small effect sizes [48, 81, 125], SGAs [121] with large effect size and especially risperidone with a medium effect size [29, 96] (see Fig. 3E and F). Despite an unfavorable profile regarding sedation/dizziness, anticholinergic, hematological, and cardiac events, different metabolic outcomes and dropouts due to adverse events compared to both FGAs and SGAs with small to large effect sizes (see Fig. 2B, C, G, H, J, K) clozapine is associated with a significantly lower mortality [124]. A detailed report with regard to different diseases entities and levels is presented in the Supplementary Results S1 (https://github.com/sksiafis/clozapine_meta_review). For additional outcomes, please see Supplementary Figs. S1–S19.

Fig. 3: Quantitative meta-review of clozapine-specific meta-analytic data: adverse-events.
figure 3

A Weight as continuous outcome. BMI body-mass-index, k number of studies, kg kilogram, L long-term, M medium-term, n number of participants, RCT randomized-controlled trial, S short-term, SGA second-generation antipsychotic. For continuous outcomes, SMD > 0 means a beneficial outcome for clozapine (e.g. more response or less adverse events. B Lipid levels. k number of studies, L long-term, M medium-term, n number of participants, RCT randomized-controlled trial, S short-term, SGA second-generation antipsychotic. C Glucose, insulin and inulin resistance levels. FGA first-generation antipsychotic, k number of studies, L long-term, M medium-term, n number of participants, RCT randomized-controlled trial, S short-term, SGA second-generation antipsychotic. For continuous outcomes, SMD > 0 means a beneficial outcome for clozapine (e.g. more response or less adverse events. D Weight as dichotomous outcome. FGA first-generation antipsychotic, k number of studies, L long-term, M medium-term, n number of participants, RCT randomized-controlled trial, S short-term, SGA second-generation antipsychotic. For dichotomous outcomes, RR > 1 means a beneficial outcome for clozapine (e.g. more response or less adverse events/dropouts). E Extrapyramidal symptoms as measured by scales. k number of studies, L long-term, M medium-term, n number of participants, RCT randomized-controlled trial, S short-term, SAS Simpson–Angus Scale, SGA second-generation antipsychotic, TD tardive dyskinesia. For continuous outcomes, SMD > 0 means a beneficial outcome for clozapine (e.g. more response or less adverse events. F Extrapyramidal symptoms as dichotomous outcome. EPS extrapyramidal symptoms, FGA first-generation antipsychotic, k number of studies, L long-term, M medium-term, n number of participants, RCT randomized-controlled trial, S short-term, SGA second-generation antipsychotic. For dichotomous outcomes, RR > 1 means a beneficial outcome for clozapine (e.g. more response or less adverse events/dropouts). G (Anti-)cholinergic symptoms. FGA first-generation antipsychotic, k number of studies, L long-term, M medium-term, n number of participants, RCT randomized-controlled trial, S short-term, SGA second-generation antipsychotic. For dichotomous outcomes, RR > 1 means a beneficial outcome for clozapine (e.g. more response or less adverse events/dropouts). H Sedation and dizziness. FGA first-generation antipsychotic, k number of studies, L long-term, M medium-term, n number of participants, RCT randomized-controlled trial, S short-term, SGA second-generation antipsychotic. For dichotomous outcomes, RR > 1 means a beneficial outcome for clozapine (e.g. more response or less adverse events/dropouts). I Other CNS symptoms (insomnia, headache, seizures, anxiety). FGA first-generation antipsychotic, k number of studies, L long-term, M medium-term, n number of participants, RCT randomized-controlled trial, S short-term, SGA second-generation antipsychotic. For dichotomous outcomes, RR > 1 means a beneficial outcome for clozapine (e.g. more response or less adverse events/dropouts). J WBC abnormalities. FGA first-generation antipsychotic, k number of studies, L long-term, M medium-term, n number of participants, RCT randomized-controlled trial, S short-term, WBC white blood count. For dichotomous outcomes, RR > 1 means a beneficial outcome for clozapine (e.g. more response or less adverse events/dropouts). K Any adverse event and dropouts due to adverse events. FGA first-generation antipsychotic, k number of studies, L long-term, M medium-term, n number of participants, RCT randomized-controlled trial, S short-term. For dichotomous outcomes, RR > 1 means a beneficial outcome for clozapine (e.g. more response or less adverse events/dropouts).

Full size image

Efficacy of pharmacological add-on strategies to improve clozapine-related adverse-events

Metformin and GLP1-RA as add-on strategies seem promising for improving metabolic outcomes short-term with mostly small effect sizes [108, 131], but also aripiprazole appears effective in terms of short-term weight reduction and reduction of lipid levels with small effect sizes [112]. Limited evidence is available for the efficacy of topiramate for weight reduction [43]. Evidence is scarce for clozapine-related hypersalivation and constipation treatment [41, 49, 115] (see Fig. 4A, B). A detailed report with regard to different disease entities and levels is presented in the Supplementary Results S1 (https://github.com/sksiafis/clozapine_meta_review) and in the Supplementary Figs. S18, 19.

Fig. 4: Add-on strategies for adverse-event management.
figure 4

A Treatment options for clozapine-associated metabolic dysfunctions. BMI body-mass index, cm centimeter, GLP-1RA GLP-1 receptor agonist, HDL high-density lipoprotein, HOMA homeostatic model assessment for insulin resistance, k number of studies, kg kilogram, L long-term, LDL low-density lipoprotein, M medium-term, n number of participants, RCT randomized-controlled trial, S short-term. For continuous outcomes, SMD > 0 means a beneficial outcome for clozapine (e.g. more response or less adverse events. B Treatment options for clozapine-associated hypersalivation. k number of studies, L long-term, M medium-term, n number of participants, RCT randomized-controlled trial, S short-term. For dichotomous outcomes, RR > 1 means a beneficial outcome for clozapine (e.g. more response or less adverse events/dropouts).

Full size image

Discussion

In our meta-review, we aimed at synthesizing all available evidence for clozapine’s efficacy and safety across all medical conditions where clozapine is used. We were able to give a systematic overview of all relevant clozapine indications and clozapine-associated endpoints derived from a total of 112 meta-analyses. Based on this overview and the methodological evaluation of all included meta-analyses, guideline developers and clinicians are now able to provide a strict risk-benefit evaluation taking into consideration all dimensions of clozapine treatment.

Symptomatic endpoints

Clozapine is significantly superior to placebo and superior to FGAs with regard to overall and positive symptoms according to high-quality meta-analytic evidence from RCTs [48, 76]. Meta-analytic evidence suggests significant superiority of clozapine in terms of efficacy on overall and positive symptoms compared to most SGAs [29, 85, 121, 122, 125] even though results are inconsistent [79].

With regard to evidence for clozapine’s effectiveness derived from observational studies, clozapine is associated with significantly lower hospitalization and ACD rate compared with other SGAs [65, 82]. For multi-episode schizophrenia and TRS, the superiority of clozapine compared to other SGAs is challenged according to meta-analytic evidence derived from RCTs: specifically for multi-episode schizophrenia (excluding TRS), clozapine appears to be not significantly different from e.g. amisulpride, olanzapine, zotepine and risperidone in terms of overall symptoms [57]. For TRS, clozapine is presumed to be not more efficacious than olanzapine, risperidone or ziprasidone in the subanalyses including only TRS trials in overall symptoms in the meta-analysis from Leucht et al. [79] being in line with the evidence from the meta-analysis from Samara et al. [100], where also only blinded RCTs were included and clozapine was not significantly superior to most other APs with regard to overall symptom reduction [100].

For treatment-resistant positive symptoms, clozapine seems to have significantly superior beneficial effects compared to quetiapine and haloperidol on single-substance level, but not compared to olanzapine [100]. When comparators are pooled as a group (FGA + SGA) clozapine was shown to have superior effects for treatment-resistant overall and positive symptoms [85, 106]. Nevertheless, for overall and positive symptoms in TRS, inconsistent evidence is reported in meta-analyses due to differences in study selections, study populations, in the handling of study characteristics, and in methodological approaches [100, 106].

For treatment-resistant negative symptoms, clozapine was shown to be slightly superior to FGAs [48] despite inconsistent results [73], but—according to a large body of evidence—not significantly superior in comparison to SGAs [29, 85, 121], and if, then only on short-term [106]. Nevertheless, negative symptom data did not include a separation of primary from secondary negative symptoms, which hampers interpretability of the results.

For cognition and psychosocial functioning, clozapine is not presumed to be significantly superior compared to other SGAs [89, 92]. While evidence for the efficacy of clozapine for first-episode psychosis is scarce [128], limited evidence suggests superior effects for clozapine as a second-line agent compared to other antipsychotics, such as, e.g. risperidone [90].

Clozapine shows beneficial effects on psychosocial function but without superiority to other antipsychotics [92]. Inconclusive results are available for pro-cognitive effects of clozapine vs. FGAs and SGAs [89, 119, 126]. For children with schizophrenia and childhood-onset schizophrenia, clozapine seems to have superior efficacy compared with FGAs [60, 74]. Limited evidence is available for schizophrenia and comorbid depression or comorbid substance abuse, but when clozapine was compared with any other antipsychotic drug plus an antidepressant or placebo, patients treated with clozapine constantly scored better on Hamilton scores [52], and clozapine was superior to other antipsychotics in substance use [71] and to risperidone in reducing craving for cannabis [118]. Furthermore, clozapine is likely to have some beneficial effects on hostility [50], suicidal behavior [56]—and maybe aggression versus others in schizophrenia, at least when compared with FGAs [62]. Nevertheless, negative evidence for suicidal behavior and self-injurious behavior for clozapine vs. SGA in observational studies was also reported [82]. Of note, meta-analytic evidence for the efficacy of clozapine in suicidal symptoms is mainly from registry data and non-randomized trials, whereas to our knowledge, only one high-quality RCT [132] fosters the evidence and contributes to long-term RCT data [29]. With regard to dosing, there is only little meta-analytic evidence that in studies with mean clozapine dosages above 400 mg/day, clozapine was superior to risperidone, but not olanzapine [79] and evidence of effects between clozapine standard, low and very low dose regimes on overall outcome in schizophrenia is sparse [114]. For bipolar disorder, the efficacy of clozapine seems to be similar to other antipsychotics in manic episodes [45]. For neurological disorders, the largest body of evidence is available for PDP, where low-dose clozapine (range from 12.5 to 50 mg) showed beneficial effects on psychotic symptoms) [51, 58] even though negative results are reported [127].

Non-symptomatic efficacy/effectiveness endpoints

Limited evidence hints at superior effects vs. SGAs in reducing drug abuse in schizophrenia short and medium-term [71, 118]. With regard to relapse prevention, clozapine is superior to FGAs [48, 77] and SGAs [125], even though results in the latter are inconsistent [64]. Mortality rate ratios seem to be lower in patients continuously treated with clozapine compared to patients on non-clozapine antipsychotics [82, 124]. Clozapine significantly reduces hospitalization rates compared to non-clozapine SGAs [75, 82] and all-cause discontinuation rates [65, 82].

Clozapine-related adverse-events and complications

There is a strong body of meta-analytic evidence for especially unfavorable metabolic outcomes (e.g. weight gain) [78, 93], also for first-episode schizophrenia patients [128]. In line with meta-analytic evidence for weight gain and the increased risk for the onset of metabolic syndrome, treatment guidelines for adult patients with schizophrenia have previously suggested not to use clozapine as a first-line agent [3]. The application among elderly patients with schizophrenia remains to be further investigated [17]. Meta-analytic evidence unequivocally suggests that clozapine is associated with a lower risk for EPS and/or tardive dyskinesia compared to other FGAs and SGAs [38, 81]. Of note, meta-analytic evidence suggests clozapine as favorable therapeutic antipsychotic agent for the event of TD [83]. Clozapine use significantly increases the risk for gastrointestinal hypomotility/constipation compared to other APs [104], but no meta-analytic data is available for the prevalence of clozapine-related (sub-) ileus.

Clozapine appears to be the most unfavorable antipsychotic for sedation compared to FGAs and other SGAs [29, 78]. With regard to pneumonia, the only available meta-analytic evidence suggests that clozapine significantly increases pneumonia risk compared to no antipsychotic use [47], but in general, evidence suggests that clozapine-related pneumonia [47, 133] might be overseen.

The incidence for clozapine-associated neutropenia is presumed to be 3.8% and severe neutropenia (agranulocytosis) between 0.4% [18] and 0.9% [88], respectively according to two meta-analyses of observational studies and—according to another meta-analysis—the relative risk for neutropenia is not significantly associated with any individual clozapine add-on antipsychotic medication [87]. Death caused by clozapine-related agranulocytosis appears to be at 0.05% [18]. Meta-analytic evidence suggests a low event rate of both clozapine-related myocarditis (0.7%) and cardiomyopathia (0.6%) [16]. Nevertheless, clozapine’s potential effect to cause arrhythmia [28] might be overseen, as reflected in a low amount of evidence. For PDP, low-dose clozapine appears to be relatively safe compared to placebo with mixed results for the effects on motor symptoms [51, 58].

Treatment of clozapine-related adverse events and complications

Metformin [108], GLP-1RAs [105] and to a lesser extent aripiprazole [112] seem to be beneficial add-on-agents for the management of clozapine-related weight gain. Metformin was superior to placebo in terms of weight loss and BMI [108]. GLP-1RAs led to a significantly higher weight loss compared to control (placebo or usual care) [105] and aripiprazole was superior with regard to weight change and LDL-cholesterol compared to placebo [112]. In all scenarios, a close risk-benefit evaluation has to be performed, since e.g. the add-on use of aripiprazole was significantly associated with agitation/akathisia and anxiety [112].

For the treatment of clozapine-related constipation, there is not enough evidence from clinical trials to inform clinical practice [49], as it is the case for clozapine-related sinustachycardia, where no data for specific clinical interventions, e.g. the use of beta-blockers is available from clinical trials [26].

The results of this meta-review should be interpreted with caution due to the inherent limitations of the meta-analyses and their included studies. The quality of meta-analyses was evaluated using the AMSTAR-2 tool, which includes items for heterogeneity and publication bias, yet further exploration of their impact on meta-analytic estimates is out of the scope of this manuscript. In addition, overlapping meta-analyses on the same topic may have different results due to different eligibility criteria and statistical methods [134], such as differences about the efficacy of clozapine for treatment-resistance schizophrenia [100, 106]. Limitations of the included studies could also impact meta-analytic estimates, i.e. a meta-analysis of observational studies investigating mortality during treatment with clozapine [135]. The potential impact of study-level (e.g. rating scale used to measure symptom improvement), and participant-level factors (such as race/ethnicity) or other confounding factors specifically in observational studies (such as concomitant medications) could not be easily addressed at the level of an umbrella review. Our meta-review represents the first comprehensive quantitative analysis of clozapine with regard to its efficacy and safety in schizophrenia, schizoaffective and bipolar disorder and PDP. Our meta-review outlines the superior efficacy of clozapine compared to FGAs and most other SGAs in schizophrenia and suggests beneficial outcomes in bipolar disorder and PDP. Nevertheless, evidence to manage clozapine-related adverse-events is sparse. In addition, more studies are needed regarding the safety of clozapine beyond the scope of schizophrenia-spectrum disorders. Our quantitative meta-review suggests that if routine hematological monitoring and screening for the early detection of myocarditis are performed, a close and continuous risk-benefit evaluation with regard to cardiovascular risk factors is key to improve clozapine-related outcomes.