CLEAR – clozapine in early psychosis: study protocol for a multi-centre, randomised controlled trial of clozapine vs other antipsychotics for young people with treatment resistant schizophrenia in real world settings

The purpose of the trial is to assess whether clozapine is more effective than TAU (standard antipsychotics) in young people with TRS in real-world settings, over a 12-week period, at the level of clinical symptoms, patient-rated outcomes, quality of life and cost-effectiveness. The primary objective is to compare the treatments on the change in total Positive and Negative Syndrome Scale (PANSS) [44] score from baseline to 12 weeks. The secondary objectives are to compare the treatments on function, adverse effects, quality of life, subjective improvement, and cost-effectiveness.

In the CLEAR trial the following hypotheses will be tested:

Figure 1 shows the trial flowchart. During the clinical trial (treatment period), participants will be allocated to either clozapine or TAU. After 12 weeks of treatment, participants will be able to continue or change treatment according to clinical needs and existing guidelines, and they will be followed up via phone or videocall whenever possible, otherwise case notes will be reviewed to assess longer-term outcomes. Schedule of enrolment, interventions and assessments are shown in the SPIRIT flow diagram (Fig. 2).

We will recruit from NHS-funded secondary care, both inpatient and community settings, using a hub and spoke model, with 6 academic centres (King’s College London, University of Birmingham, Cardiff University, The University of Manchester, University of Oxford, University College London) coordinating approximately 30 NHS recruitment sites.

We will recruit 260 children and young people from 12 to 24 years old with TRS as defined by NICE (CG178, Sect. 1.5.7.2). Inclusion and exclusion criteria are specified in greater detail in Table 1.

During the 12 weeks of the trial, we will compare clozapine vs. TAU.

Intervention arm: Clozapine, oral, flexible dose within dose range defined by British National Formulary [47], at the discretion of the prescriber, for a minimum of 12 weeks. As clozapine has unpredictable pharmacokinetics with high heterogeneity in plasma concentration, depending on age, sex, smoking status, and genetics of the liver enzymes that metabolise it (especially CYP-1A2 and-2D6), enforcing a fixed dose would reduce the acceptability of the trial to patients and clinicians. It also requires titration over the first 2 weeks up to therapeutic doses to minimise postural hypotension, and the optimal balance between efficacy and adverse effects can only be achieved on an individual basis. The previously prescribed antipsychotic can be titrated down during the first 2 weeks of clozapine treatment and must be stopped within the first two weeks.

Control arm: Any oral antipsychotic other than clozapine, within licensed dose range defined by BNF [47], for a minimum of 12 weeks. The choice of antipsychotic will be agreed by the clinical team in collaboration with the participant, and the dose titrated to achieve the best balance between response and adverse effects.

The primary outcome is the change in total blind rated PANSS total scores at 12 weeks from baseline.

Secondary outcomes include: blind rated change in overall Clinical Global Impression Scale (CGI) [48], clinician rated level of adherence (CRS) [49], adverse effects (GASS-C) [50], health-related quality of life (EQ-5D-Y) [51], subjective experience (DAI-10) [52], readmission rate, (EI-AD-SUS) [53], and change in PANSS sub-scale (positive, negative and general) [44]. We will also combine these outcomes (EQ-5D-Y and total PANSS score) with service use data (EI-AD-SUS) to compare treatments on cost-effectiveness.

Patients will be followed up for 12 months with the potential for further follow-up using clinical records for longer-term evaluations.

For the mechanistic study, outcomes include the change in brain glutamate, regional cerebral blood flow (rCBF) and GSH, and peripheral cytokines and GSH from baseline to 12 weeks.

The standardised mean difference for clozapine versus other antipsychotics in adults based on PANSS total score at 12 weeks is 0.39 [55]. The standard deviation of the PANSS score at 12 weeks in our similar sample of treatment resistant patients being started on clozapine [43] is 14.2 (60% CI 12.96–16.20). Using an upper limit of the 60% confidence interval [56], an effect size of 0.39 corresponds to a between group difference of 6.4 points. In our previous research [43], the 95% CI for the baseline-12-week correlation in the sample was 0.553 to 0.880, and we use a value of 0.5. This is conservative but realistic, since the time points are only 12 weeks apart. Under these assumptions, and assuming 20% attrition to 12-week follow-up, the target sample size to recruit at baseline is 260 participants.

For the mechanistic study we estimate that, of all participants recruited at MRI sites, 50% will agree to participate in MRI. This results in a sample for the mechanistic study of 90 participants at baseline and, assuming 20% attrition, complete mechanistic data in 70 participants. This sample size has 80% power to detect a minimum between group effect size of 0.7SD on a standardised scale and 90% power for 0.8SD. In our previous observational study in TRS [43], we found a within-group effect after 12 weeks of clozapine of 0.7. Assuming there is no further change from prior treatment in the standard antipsychotic treatment arm on these measures over 12 weeks, we would have sufficient power to detect between-group effects of these magnitudes and expect the effect sizes to be greater in the younger population.

Participants will be randomised 1:1 using an online system hosted by the King’s Clinical Trials Unit, using computer generated blocks of random sizes. The sample will be stratified by sex and age group (< 18 years, >  = 18 years).

The raters for primary outcome (PANSS) and CGI-S and CGI-I will be centralised and blinded to minimise observer bias. The participants, the treating team, and the rest of the trial team will not be blinded. There will be four blinded members of the study team: two research psychiatrists, one for adult participants and the other for child participants (each with appropriate training for that age group), the Chief Investigator (CI) and Co-Chief Investigator. They will not be informed which treatment arm the participants are in and will not enquire about the participant’s treatment or side effects.

A web based electronic data capture (EDC) system will be designed, using the InferMed Macro 4 system. The EDC will be created in collaboration with the trial analyst and the CI and maintained by the King’s Clinical Trials Unit for the duration of the project. It will be hosted on a dedicated server within King’s College London. The system is compliant with FDA 21 CFR part 11 and Good Clinical Practice (GCP). The CI will act as custodian for the trial data. Should a participant decide to withdraw from the study, all efforts will be made to report the reason for withdrawal as thoroughly as possible. Should a participant withdraw from study drug only, efforts will be made to continue to obtain follow-up data, with the permission of the participant.

The trial will be conducted in compliance with the principles of the Declaration of Helsinki (1996), the principles of GCP and in accordance with all applicable regulatory requirements including but not limited to the Research Governance Framework and the Medicines for Human Use (Clinical Trial) Regulations 2004, as amended in 2006 and any subsequent amendments. The trial has been approved by London—Dulwich Research Ethics Committee (REC reference: 22/LO/0723) and obtained Clinical Trial Authorisation (CTA) from the Medicines & Healthcare products Regulatory Agency (MHRA; CTA 22926/0007/001–0003).

The Statistical Analysis Plan will be drafted within 6 months after recruitment starts and will be approved by the Data Monitoring Committee (DMC) and Trial Steering Committee (TSC) before the junior statistician sees any outcome data split by arm. The senior statistician will remain fully blind throughout the study.

We will report data in line with the CONSORT 2010 statement [57] showing attrition rates and loss to follow-up. The target estimand is the treatment policy estimand, and all primary and secondary analyses will be carried out following the intention to treat principle, incorporating data from all participants including those who do not complete treatment. Every effort will be made to follow up all participants in both arms for research assessments. The trial will be blind rated to minimize observer bias. The primary analysis will be conducted after the final 12-week follow-up assessment is completed for the final patient recruited into the study. The study team including CI and trial statisticians will be aware of the results of the primary analysis from this point. The remaining 24-and 52-week assessments will continue to be conducted blind to individual patient’s allocation. We will analyse the 24- and 52-week outcomes in a secondary analysis.

Analyses will be conducted in Stata version 17or later. Descriptive statistics within each randomized group will be presented for baseline values. For the primary analysis, the treatment effects on primary and secondary outcomes will be estimated using linear mixed models fitted to all outcome variables up to and including the 12-week assessment. Fixed effects will be sex, age (< 18, >  = 18) and duration of previous treatment (± 3 years), baseline assessment for the outcome under investigation, treatment, time, and time*treatment interactions. Participant will be included as a random intercept to account for repeated measures. Marginal treatment effects will be estimated for primary outcome (PANSS score at 12 weeks), and for PANSS scores at each other time point (6 weeks, 12 weeks), and reported separately as adjusted mean differences in scores between the groups with 95% confidence intervals and 2-sided p-values. For secondary outcomes the same approach will be followed using linear mixed models to estimate and report the treatment effect at each time point. Cohen’s D effect sizes will be calculated as the adjusted mean difference of the outcome divided by the sample standard deviation of the outcome at baseline. These will be displayed in a forest plot showing the treatment effects on the primary and the secondary outcomes at 12 weeks. For the secondary analysis including 24-and 52-week time points, we will repeat the linear mixed model approach with these additional outcomes in the response vector. We will report the estimated marginal treatment effects at 24-and 52-weeks.

Missing data on individual measures will be pro-rated if more than 80–90% (depending on questionnaire) of the items are completed; otherwise, the measure will be considered as missing. We will check for differential predictors of missing outcomes by comparing responders to non-responders on key baseline variables. Any significant predictors will be included in the analysis models in a sensitivity analysis. This accounts for missing outcome data under a missing at random assumption, conditional on the covariates included in the model. As a sensitivity analysis, we will assess whether treatment adherence is associated with missing data, and if it is associated, use inverse probability weights or multiple imputation to compare results.

A pre-specified subgroup analysis will test the treatment effect in children (age < 18 years) by estimating the effect in each group separately. If the final proportion of under 18 s is low, we may use a Bayesian subgroup analysis to increase precision of the effect in the groups. There are no planned interim analyses.

For the mechanistic study, descriptive statistics for biomarkers at baseline and 12 weeks will be presented by treatment group. We will test for between-group differences on peripheral cytokines, GSH, brain glutamate and rCBF at 12 weeks using appropriate generalised linear models. Mechanistic analyses will be based on causal mediation analysis and the primary outcome variable will be the PANSS total score at 12 weeks. We will estimate the direct and indirect effects of clozapine on total PANSS via the mechanistic measures by fitting models for the outcome with and without mediators using the difference in coefficients approach, extended for multiple mediators. We will adjust for baseline values of the mediators [58] baseline PANSS total score, and fixed effects of sex, age, duration and number of previous treatment trials [59].

A detailed health economic analysis plan (HEAP) will be prepared by the trial health economists and approved by the Data Monitoring Committee (DMC) and Trial Steering Committee (TSC) before the trial health economist sees any data split by arm. The HEAP will be drafted before recruitment starts and will be approved before the junior health economist sees any outcome data split by arm. The senior health economist will remain fully blind until review of the first draft of the final economic analysis reports for checking when they will become fully unblinded and any future amendments to the HEAP will be made by them.

A within-trial cost-effectiveness analysis will be carried out, taking the NHS and social services perspective preferred by NICE, including relevant education-based health and social care services, given the age group. Service use will be collected in interview at baseline (covering the previous 3 months) and at the 6-, 12-, 24- and 52-week follow-up assessments (covering the period since last interview) using the Early Intervention Adult Service Use Schedule (EI-AD-SUS) [53]. The EI-AD-SUS was originally designed and successfully applied in populations of young people and young adults at risk of or with psychosis. Nationally applicable unit costs will be applied to all services (for example, NHS Reference Costs for hospital contacts, British National Formulary for medications, PSSRU Unit Costs of Health & Social Care for community-based services etc.) to estimate total costs per participant.

The primary economic evaluation will be a cost-utility analysis carried out at the 12-week follow-up, in line with the primary clinical analysis, with outcomes expressed in terms of quality adjusted life years (QALYs) calculated from the EQ-5D-Y [51], as preferred by NICE [60], and using the recommended area under the curve approach [61]. Secondary analyses will explore a) cost-effectiveness using the primary clinical measure of outcome (PANSS total score) at 12-weeks and b) cost-utility using QALYs at the 52-week follow-up. Given evidence to suggest the EQ-5D may not be particularly sensitive in psychosis populations [62], we will additionally include the Recovering Quality of Life-10items measure (ReQoL-10), a new generic self-reported outcome measure for use with people experiencing mental health difficulties, which may be more sensitive to change than the EQ-5D. The ReQoL is not appropriate as the main measure of effectiveness for the economic evaluation because it is not yet associated with preference weights to generate QALYs for use in cost-effectiveness analyses and it is currently considered suitable for people aged 16 and over. However, the inclusion of this brief measure will support exploration of the sensitivity of the EQ-5D in comparison to the ReQoL and the validity of the measure in young people under the age of 16.

Costs and QALYs will be presented as mean values by trial arm with standard deviations. Mean differences in costs and 95% confidence intervals will be obtained by non-parametric bootstrap regressions to account for the non-normal distribution commonly found in economic data [63]. Cost-effectiveness will be assessed using the net benefit approach and following standard approaches [64]. A joint distribution of incremental mean costs and effects for the two groups will be generated using bootstrapping to explore the probability that clozapine is the optimal choice compared to TAU, subject to a range of possible maximum values (ceiling ratio) that a decision-maker might be willing to pay for unit improvements in outcomes. Cost-effectiveness acceptability curves will be presented by plotting these probabilities for a range of possible values of the ceiling ratio [65]. These curves are the recommended decision-making approach to dealing with the uncertainty that exists around the estimates of expected costs and expected effects associated with the interventions under investigation and uncertainty regarding the maximum cost-effectiveness ratio that a decision-maker would consider acceptable. To provide more relevant treatment-effect estimates, all economic analyses will include adjustment for the variable(s) of interest and baseline covariates [66], which will be prespecified and in line with the clinical analyses. Complete case analyses will be carried out with the impact of missing data explored in sensitivity analyses. The pattern of missing data and the plausibility of assuming the data are Missing at Random will be examined for missing cost and EQ-5D-Y tariff values. Any variables predictive of missingness or predictive of response (at P < 0.1) will be included in the equation to impute missing values.

Participants will be asked at each visit from consent onwards to report any suspected adverse reactions. Any suspected adverse events will be recorded from consent visit to end of trial. Any suspected adverse events recorded will be explored again at each visit thereafter. Events or reactions listed in the Summary of Product Characteristics (SmPC) do not need to be reported unless they fulfil seriousness criteria.

All Serious adverse Event (SAE), Serious Adverse Reaction (SAR) or Suspected Unexpected Serious Adverse Reaction (SUSAR), excepting those specified in the protocol as not requiring reporting, will be reported immediately by the CI (and certainly no later than 24 h) to the KHP-CTO in accordance with the current Pharmacovigilance Policy. The KHP-CTO will report SUSARs to the regulatory authority (MHRA).

A DMC will be established to review accruing data and safety information, reporting to the TSC. Independent membership will include an adult psychiatrist, a child and adolescent psychiatrist, and a clinical trial statistician. The Investigator will permit trial-related monitoring, audits, REC review, and regulatory inspections by providing the Sponsor, Regulators and REC direct access to source data and other documents.

The datasets used and/or analysed during the current study will be available from the corresponding author on reasonable request.