Lead Investigator: Stefan Leucht, Technical University of Munich
Title of Proposal Research: Individual-patient-data (IPD) meta-analysis of the efficacy of clozapine versus second-generation antipsychotic drugs in patients with treatment-resistant schizophrenia
Vivli Data Request: 6881, 5484
Funding Source: Government Funding (German Ministry of Education and Research)
Potential Conflicts of Interest: In the last 3 years Stefan Leucht has received honoraria as a consultant or for lectures for LB Pharma, Otsuka, Lundbeck, Boehringer Ingelheim, LTS Lohmann, Janssen, Johnson&Johnson, TEVA, MSD, Sandoz, SanofiAventis, Angelini, Sunovion, Recordati and Geodon Richter.
None of these honoraria was related to the proposed research project. There is no other financial relationship (grant support, royalties, pending patents) with any of these or other companies. No conflict of interest results in this regard.
Stefan Leucht is the head of a section for evidence based medicine in psychiatry. His work group received research grants from the German Ministry of Education and Research to conduct the proposed project and other systematic reviews and meta-analyses in the field of psychiatry. No conflict of interest arises from this government-funding. The funding agency will not have any role in analysing or interpreting the data and it will not be involved in writing the report.
Stefan Leucht is the last author of a scientific publication on a network-meta-analysis in treatment resistant schizophrenia, which showed no significant difference in efficacy between clozapine and other second-generation antipsychotics, whereas another analysis found significant differences. With the proposed project, clozapine should be compared to other second-generation antipsychotics in more detail using IPD data. There is an academic interest to solve this important clinical question but no conflict of interest for the conduct of this project.
Summary of the Proposed Research:
Schizophrenia is a debilitating disorder affecting approximately 1% of the public (McGrath et al., 2008), and up to one third of the patients can be considered treatment-resistant (Lally et al, 2016). Clozapine is the only antipsychotic with approved superiority, yet it is reserved as last resort due to its poor tolerability (Hasan et al., 2012). Nevertheless, current syntheses of evidence based on meta-analyses of aggregated data (Samara et al., 2016; Asenjo Lobos et al., 2010; Siskind et al., 2017) show conflicting results and challenge the superiority of clozapine against the other so-called second-generation antipsychotics, which have a better safety profile.
Therefore, we will re-analyze randomized evidence of the comparison between clozapine and other second-generation antipsychotics in treatment-resistant schizophrenia using individual-participant-data (IPD) meta-analysis, which is the “gold standard” of systematic reviews. We will be able to produce more precise results and identify which patients might benefit from clozapine. Our methods will follow the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) statement for IPD (Stewart et al., 2015), and the highest Cochrane standards. Statistical analyses will be conducted in collaboration with Professor Georgia Salanti, one of the leading statisticians in meta-analysis. Our team includes experts in schizophrenia and IPD meta-analysis.
Results from this IPD analysis could be utilized in the production of clinical guidelines and in the design of future trials. Hence, they are very likely to have a real impact in the treatment of this critical population.
Statistical Analysis Plan:
SEARCH AND SELECTION OF ELIGIBLE STUDIES (Criteria for selecting the requested study)
We will conduct a systematic literature search to identify eligible studies in collaboration with Farhad Shokraneh, information specialist of the Cochrane Schizophrenia Group. Study selection will be done independently by two reviewers.
In a preliminary search we already identified 17 eligible studies (we doubt that there are more) and started to ask for data sharing, including the requested study here (see below for studies from other sources). This approach was requested by the funding agency, as it is known that these processes can take long.
Selection criteria are:
1.Population: We will include patients with a treatment-resistant form of schizophrenia, schizophreniform disorder, or schizoaffective disorder (no restriction in setting, age, gender, and ethnicity). There is no clear evidence that the latter schizophrenia-like psychoses are caused by fundamentally different disease processes or require different treatment approaches. We will include trials irrespective of the diagnostic criteria used. It is a general strategy of the Cochrane Schizophrenia Group (CSG) to include not only studies that used specific diagnostic criteria such as ICD-10 or DSM-5, because these criteria are not meticulously used in clinical routine either. Nevertheless, we will exclude studies that did not use operationalized criteria in a sensitivity analysis. Participants within a trial had to be classified by inclusion criteria as “treatment-resistant” or “non-responsive” to previous treatment with antipsychotics. Criteria of treatment-resistant schizophrenia were standardized in the recent consensus of Howes et al. (Am J Psychiatry 2017; 174(3):216–29) defining minimum and optimal criteria based on current symptoms and previous adequate treatment. However, previous definitions varied across studies. Therefore, we will accept any definition of treatment-resistance used in the studies, but we will consider different definitions in a subgroup analysis, i.e. meeting at least the minimum criteria of Howes et al 2017 or meeting only study-defined criteria. These decisions should increase generalizability and representativeness. All participants within the included trials will be eligible in the IPD-analysis.
2. Intervention: Clozapine at any dose, if used as monotherapy, will be examined.
3. Comparator(s): Any other second-generation antipsychotic will be used as a comparator. Second-generation antipsychotic drugs listed in WHO (amisulpride, aripiprazole, asenapine, brexpiprazole, cariprazine, clozapine, iloperidone, lurasidone, olanzapine, paliperidone, quetiapine, risperidone, sertindole, ziprasidone, zotepine) at any dose and in any form of administration that were compared with clozapine, will be included if the antipsychotics were used as monotherapy. The effects compared to specific second-generation antipsychotics will be examined in subgroup analyses.
4. Outcomes: See question on specific outcomes in Research Proposal section.
5. Design of primary studies: We will include randomised trials (RCTs) comparing clozapine with a second-generation antipsychotic in treatment-resistant schizophrenia as monotherapy. Trials that antipsychotic drugs were used as combination or augmentation will be excluded. We will only include blinded trials (at least single-blind) because we previously found that open trials favored the sponsored drug. In the case of crossover studies, we will only use the first crossover phase to avoid the problem of carryover effects. Furthermore, studies that demonstrated a high risk for bias arising from randomization process will be excluded. If a trial is described as double-blind but randomization is not explicitly mentioned, we will assume that study participants are randomized, and we will exclude the trial in a sensitivity analysis. Trials with an experimental focus and design component, like neuroimaging, will be excluded, because we anticipate that the included patients were highly selected. We will exclude studies from mainland China in order to avoid systematic bias, because there is evidence that many of these studies, including on newer-generation antipsychotics, do not use appropriate randomization procedures and do not report their methods. In addition, patterns of clozapine use are different in China; clozapine is one of the most frequently utilized antipsychotics and it could also be used as first-line treatment. The minimum duration of trials is set at six weeks, as shorter studies are unlikely to be sufficient in treatment-resistant schizophrenia.
STUDIES REQUESTED FROM OTHER SOURCES
In a preliminary search we identified 17 eligible studies (we doubt that there are more) and started to ask for data sharing (this was requested by the funding agency as it is know that these processes can take long).
In addition to the requested study at Vivli, we are aware of 16 other eligible studies. Also for those, we started the request for patient-level-data. These are:
1. Azorin JM, Spiegel R, Remington G, Vanelle JM, Pere JJ, Giguere M et al. A double-blind comparative study of clozapine and risperidone in the management of severe chronic schizophrenia. Am J Psychiatry 2001; 158(8):1305–13.
2. Bondolfi G, Dufour H, Patris M, May JP, Billeter U, Eap CB. Risperidone versus clozapine in treatment-resistant chronic schizophrenia: a randomised double-blind study. The Risperidone Study Group. American journal of psychiatry 1998; 155(4):499–504.
3. Schooler NR, Marder SR, Chengappa KN, Petrides G, Ames D, Wirshing WC et al. Clozapine and risperidone in moderately refractory schizophrenia: a 6-month randomized double-blind comparison. Journal of clinical psychiatry 2016; 77(5):628–34.
4. Breier AF, Malhotra AK, Su TP, Pinals DA, Elman I, Adler CM et al. Clozapine and risperidone in chronic schizophrenia: effects on symptoms, parkinsonian side effects, and neuroendocrine response. Am J Psychiatry 1999; 156(2):294–8.
5. Wahlbeck K, Cheine M, Tuisku K, Ahokas A, Joffe G, Rimón R. Risperidone versus clozapine in treatment-resistant schizophrenia: a randomized pilot study. Progress in neuro-psychopharmacology & biological psychiatry 2000; 24(6):911–22.
6. Daniel DG, Goldberg TE, Weinberger DR, Kleinman JE, Pickar D, Lubick LJ et al. Different side effect profiles of risperidone and clozapine in 20 outpatients with schizophrenia or schizoaffective disorder: a pilot study. Am J Psychiatry 1996; 153(3):417–9.
7. Bitter I, Dossenbach MR, Brook S, Feldman PD, Metcalfe S, Gagiano CA et al. Olanzapine versus clozapine in treatment-resistant or treatment-intolerant schizophrenia. Progress in neuro-psychopharmacology & biological psychiatry 2004; 28(1):173–80.
8. Conley RR, Kelly DL, Richardson CM, Tamminga CA, Carpenter WT. The efficacy of high-dose olanzapine versus clozapine in treatment-resistant schizophrenia: a double-blind crossover study. Journal of clinical psychopharmacology 2003; 23(6):668–71.
9. Meltzer HY, Bobo WV, Roy A, Jayathilake K, Chen Y, Ertugrul A et al. A randomized, double-blind comparison of clozapine and high-dose olanzapine in treatment-resistant patients with schizophrenia. Journal of clinical psychiatry 2008; 69(2):274–85.
10. Naber D, Riedel M, Klimke A, Vorbach E-U, Lambert M, Kuhn K-U et al. Randomized double blind comparison of olanzapine vs. clozapine on subjective well-being and clinical outcome in patients with schizophrenia. Acta psychiatrica scandinavica 2005; 111(2):106–15.
11. Tollefson GD, Birkett MA, Kiesler GM, Wood AJ. Double-blind comparison of olanzapine versus clozapine in schizophrenic patients clinically eligible for treatment with clozapine. Biological psychiatry 2001; 49(1):52–63.
12. Volavka J, Czobor P, Sheitman B, Lindenmayer J-P, Citrome L, McEvoy JP et al. Clozapine, olanzapine, risperidone, and haloperidol in the treatment of patients with chronic schizophrenia and schizoaffective disorder. Am J Psychiatry 2002; 159(2):255–62.
13. Lewis SW, Barnes TR, Davies L, Murray RM, Dunn G, Hayhurst KP et al. Randomized controlled trial of effect of prescription of clozapine versus other second-generation antipsychotic drugs in resistant schizophrenia. Schizo-phrenia bulletin 2006; 32(4):715–23.
14. Meyer-Lindenberg A, Gruppe H, Bauer U, Lis S, Krieger S, Gallhofer B. Improvement of cognitive function in schizophrenic patients receiving clozapine or zotepine: results from a double-blind study. Pharmacopsychiatry 1997; 30(2):35–42.
15. Kumra S, Kranzler H, Gerbino-Rosen G, Kester HM, DeThomas C. Clozapine and “high-dose” olanzapine in refractory early-onset schizophrenia: a 12-week randomized and double-blind comparison. Biological Psychiatry 2008: 63(5) 524-529.
16. Shaw P, Sporn A, Gogtay N, Overman GP, Greenstein D. Childhood-onset schizophrenia: A double-blind, randomized clozapine-olanzapine comparison. Archives of general psychiatry 2006: 63(7) 721-730.
CHECKING, HARMONIZATION AND MERGING OF DATA
Ideally, IPD-datasets of each study will be transferred to the research team (using secure data transfer technologies and stored at a secure computer server) and a masterfile including data of all studies (after conversion of different formats to a common format and recoding of variables to documented common variable names) will be created for the purpose of the analysis.
If transfer of the IPD-datasets is not possible, analyses in secure computer environments might be performed (this approach may be preferred by some pharmaceutical companies). In this case, we will use a two-step method to integrate the data in the overall meta-analysis: First, the individual study IPD-data provided in the secure environment will be analyzed according to the strategy outlined below and aggregate summary statistics (unadjusted and adjusted for potential effect modifiers) will be calculated. In a second step, the aggregate summary statistics of this study will be included in the overall IPD-meta-analysis (combing IPD and aggregate data from different studies).
IPD integrity of each study will be evaluated, including checking for missing data, for duplicates, for extreme outliers or unusual values, and if randomization was appropriate. For internal consistency, obtained data of each study will be cross-checked against summary statistics as reported in the available publications by two independent reviewers: This will assure that IPD are reasonable and any issue will be resolved by discussion with the trialists.
In case of studies whose IPD will not be available at all (neither transfer of data, nor analysis of data in secure computer environments), two independent reviewers will extract aggregated data based on published documents on specifically developed forms in our Microsoft Access database, which is a tailor-made database for schizophrenia trials. Discrepancies in double data extraction will be identified by an algorithm, and they will be resolved by discussion with a third senior reviewer. Where this is not possible, we will contact the study authors. Intention-to-treat data will be used whenever available, change scores will be preferred to endpoint scores, and missing standard deviations will be derived as described in the Cochrane Handbook.
Two independent reviewers will evaluate risk of bias of individual studies using the Cochrane Risk of Bias tool version 2. This tool evaluates biases arising from randomization process, deviations of indented interventions, missing outcome data, measurement of the outcome and selection of the reported result. In case of studies with IPD, missing outcome bias will be evaluated according to the provided dataset after missing data are imputed and not according to the publications. Studies could still be rated with high risk of bias, when a large number and unbalanced dropouts exist in the obtained raw data. In case of studies with aggregated data only, missing outcome bias will be rated according to the publications.
DETAILS OF THE STATISTICAL ANALYSIS
1. Effect size: For continuous outcomes, it will be the mean difference (e.g. of total PANSS scores). If we are unable to convert the outcome of all studies into a single scale then we will transform the outcome using z-scores and then synthesize the data to obtain standardized mean difference between interventions. For dichotomous outcomes, it will be the relative risk supplemented by numbers needed to treat to benefit/harm (NNTB/NNTH). Relative risks are preferred to odds ratios, since the former can be interpreted more intuitively by clinicians. Effect sizes will be presented with their 95% CIs.
2. Synthesis of the results: Primary analyses will be conducted according to the intention-to-treat principle, i.e. as randomized. We will synthesize study results using random-effects IPD meta-analysis and meta-regression models as described by Burke et al. (Stat Med 2017; 36(5):855–75). We will opt for IPD data from all included studies; however, if there are studies for which only aggregated data are available, we will combine all eligible studies (IPD and aggregated data when the former are not available) by developing a Bayesian hierarchical model as described by Sutton et al. (Stat Med 2008; 27 (5): 651-669). Such a flexible model distinguishes between two levels, first, the patient-level data for IPD, and second, the study-level information which enables the IPD and the aggregated data to be incorporated in the overall estimated effects. Our primary time-point will be recorded as close to 6-8 weeks as possible. If the study outcome is not reported in this range we will record the outcome closest to 8 weeks and we will explore the timing of outcome recorded as an effect modifier.
The model will be fitted in R using self-programmed routines in a Bayesian context using the JAGS package. Vague priors will be assumed for all location parameters (effect sizes and regression coeffi-cients). For the heterogeneity, we will use a half-normal prior on the heterogeneity standard deviation. We will explore heterogeneity by including in the model effect modifiers from the list of prognostic factors above (see predictors-section in Research Proposal part), if data are available. Residual heterogeneity will be measured by monitoring the common heterogeneity parameter τ2 (the relative change after accounting for effect modifiers) and by comparing it to its empirical distribution.
Missing outcome and co-variate data will be naturally imputed in the Bayesian model, assuming a missing at random (MAR) mechanism.
3. Subgroup analyses: For the primary outcome, the following subgroup analyses will be performed using the meta-regression model a) definition of treatment resistance, b) sponsoring of pharmaceutical industry c) second-generation antipsychotic drug used as comparator, d) follow-up duration.
4. Sensitivity analyses: The following sensitivity analyses of the primary outcome are planned a priori a) exclusion of single-blind studies, b) exclusion of studies with implied randomization, c) exclusion of studies that did not use operationalised criteria to diagnose schizophrenia, d) exclusion of studies with high risk of bias in deviations of indented interventions, missing outcome data, measurement of the outcome and selection of the reported result.
5. Publication bias: For the primary outcome, small study effects and the potential publication bias will be examined by contour-enhanced funnel plot if more than 10 studies are available. We will also fit a meta-regression model by including the study precision as a covariate.
6. Data availability bias: We anticipate that some of the eligible studies will not provide IPD data. Differences between studies providing IPD and studies whose IPD were not available will be tested by examining subgroup heterogeneity using meta-analysis of aggregated data.
7. Quality of evidence for the following outcomes will be evaluated using the GRADE framework: overall, positive, negative symptoms, dropouts due to any cause and inefficacy, quality of life and overall functioning. This approach considers study limitations, inconsistency of the results, indirect-ness of evidence, impression and publication bias.
Requested Studies:
DOUBLE BLIND, DOUBLE DUMMY MULTICENTER, PARALLEL GROUP COMPARISON OF THE TOLERABILITY OF ZIPRASIDONE VS CLOZAPINE IN SCHIZOPHRENIC PATIENTS WHO ARE INTOLERANT TO ANTIPSYCHOTIC THERAPY
Sponsor: Pfizer
Study ID: NCT00649844
Sponsor ID: A1281039
(Note: Additional studies added as part of data request 6881)
Olanzapine Versus Clozapine in Refractory Schizophrenic Patients
Data Contributor: Lilly
Study ID: F1D-EW-HGCF
Public Disclosures:
Schneider-Thoma, J., Hamza, T., Chalkou, K., Siafis, S., Dong, S., Bighelli, I., Hansen, W.P., Scheuring, E., Davis, J.M., Priller, J. and Baumann, P., 2025. Efficacy of clozapine versus second-generation antipsychotics in people with treatment-resistant schizophrenia: a systematic review and individual patient data meta-analysis. The Lancet Psychiatry. Doi : 10.1016/S2215-0366(25)00001-X