Adalimumab in combination with methotrexate for refractory uveitis associated with juvenile idiopathic arthritis: a RCT

Article history

The research reported in this issue of the journal was funded by the HTA programme as project number 09/51/01. The contractual start date was in August 2011. The draft report began editorial review in December 2017 and was accepted for publication in June 2018. The authors have been wholly responsible for all data collection, analysis and interpretation, and for writing up their work. The HTA editors and publisher have tried to ensure the accuracy of the authors’ report and would like to thank the reviewers for their constructive comments on the draft document. However, they do not accept liability for damages or losses arising from material published in this report.

Declared competing interests of authors

Athimalaipet V Ramanan reports grants from the National Institute for Health Research (NIHR) Health Technology Assessment (HTA) programme and Arthritis Research UK during the conduct of the study and others from AbbVie Inc. (Ludwigshafen, Germany) and the University Hospitals Bristol NHS Foundation Trust, outside the submitted work. He has received speaker fees from AbbVie Inc. and lectured in symposia, sponsored by AbbVie Inc.. He has also been on an Advisory Board organised by AbbVie Inc. and has been supported by AbbVie Inc. to attend European and American Rheumatology Society meetings. Andrew D Dick reports other from data and revenue sharing outside the submitted work for consultancy work, paid to University of Bristol by AbbVie Inc., Ashley P Jones, Andrew McKay, Anna Rosala-Hallas, Ben Hardwick, Helen Hickey, Naomi Rainford, Graeme Hickey, Ruwanthi Kolamunnage-Dona and Paula R Williamson report grants from the NIHR HTA programme and Arthritis Research UK during the conduct of the study, other from AbbVie Inc. and University Hospitals Bristol NHS Foundation Trust and personal fees from University of Liverpool, outside the submitted work. In addition, Paula R Williamson is the Director of the Clinical Trials Research Centre, which is the Clinical Trials Unit that managed the day-to-day running of this trial. Dyfrig Hughes and Patricia Woo report grants from the NIHR HTA programme and Arthritis Research UK during the conduct of the study and other from AbbVie Inc., outside the submitted work. Giovanna Culeddu and Eifiona Wood report grants from Arthritis Research UK during the conduct of the study. Sandrine Compeyrot-Lacassagne reports grants from the NIHR HTA programme and Arthritis Research UK during the conduct of the study, and grants and others from AbbVie Inc., outside the submitted work. Clive Edelsten reports grants from the NIHR HTA programme and Arthritis Research UK during the conduct of the study, and others and personal fees from AbbVie Inc., outside the submitted work. Michael W Beresford reports grants from the NIHR HTA programme and Arthritis Research UK during the conduct of the study and others from AbbVie Inc. and the University Hospitals Bristol NHS Foundation Trust, outside the submitted work.

Copyright statement

© Queen’s Printer and Controller of HMSO 2019. This work was produced by Ramanan et al. under the terms of a commissioning contract issued by the Secretary of State for Health and Social Care. This issue may be freely reproduced for the purposes of private research and study and extracts (or indeed, the full report) may be included in professional journals provided that suitable acknowledgement is made and the reproduction is not associated with any form of advertising. Applications for commercial reproduction should be addressed to: NIHR Journals Library, National Institute for Health Research, Evaluation, Trials and Studies Coordinating Centre, Alpha House, University of Southampton Science Park, Southampton SO16 7NS, UK.

2019 Queen’s Printer and Controller of HMSO

Scientific background

Children with juvenile idiopathic arthritis (JIA), the most common rheumatic disease in children, are at risk of inflammation of the uvea in the eye (uveitis). Overall, 20–25% of all paediatric uveitis is associated with JIA. 1,2 Several major risk factors are known for the development of uveitis in JIA, including oligoarticular pattern of arthritis, onset of arthritis at < 7 years of age and antinuclear antibody positivity. 3 Generally, in the initial stages of mild to moderate inflammation, the uveitis is entirely asymptomatic; therefore, current practice is to screen all children with JIA regularly for uveitis. Between 12% and 38% of patients with JIA will develop uveitis in the initial 7 years following the onset of arthritis. 4,5 Structural complications are present in 30–50% of children with JIA-associated uveitis at diagnosis. 6 Importantly, 50–75% of those children with severe uveitis will eventually develop visual impairment secondary to ocular complications including cataracts, glaucoma, band keratopathy and macular pathology. 7–9 Defining the severity of inflammation and structural complications in uveitis patients can now be more consistently described following Standardisation of Uveitis Nomenclature (SUN) guidelines. 10 These guidelines allow incorporation into randomised controlled trials and cohort studies. 10

Poor prognosticators of poor visual acuity include structural changes at presentation, need for intraocular surgery, posterior segment inflammation, abnormal intraocular pressure (IOP) and the failure to maintain long-term disease control as marked by persistent anterior chamber (AC) cell scores of ≥ 1+. 6–8,11 Despite current screening and therapeutic options (pre biologics), some 10–15% of children with JIA-associated uveitis may eventually develop bilateral visual impairment and will be certified legally blind. 12,13 It is, therefore, critical to find more effective therapeutic interventions.

Rationale for research

Methotrexate (MTX) is well established as the first-line disease-modifying agent in the management of JIA. 14,15 Topical corticosteroids are among the current approaches to treatment of mild JIA-associated uveitis. In children with moderate to severe JIA-associated uveitis, MTX is also effective. 16–18 However, there have been no prospective randomised placebo-controlled trials of MTX or corticosteroid regimens for JIA-associated uveitis.

A systematic review of the evidence of the use of MTX in JIA is restricted to joint involvement14 and does not include paediatric uveitis. Despite the scarce evidence, MTX has become the mainstay of treatment for JIA-uveitis. 19 However, up to 15–50% of children will have refractory uveitis in spite of optimal therapy with MTX. 16–18 In a small study,20 some 30% of patients started on MTX for JIA-associated uveitis did not achieve disease control during the first year of therapy and, even when remission was achieved with MTX, nearly 70% of patients will later relapse, suggesting that only 4 out of 22 (18%) patients achieved total remission. 13 In a Dutch study, only 12% were found to be in total remission 5 years after starting MTX. 20 In small, retrospective case series, other agents including ciclosporin and mycophenolate mofetil have been shown to be of partial benefit in controlling JIA-uveitis. 21,22 However, there is little evidence that they rescue MTX-refractory patients and their use is restricted because of intolerability and adverse reactions. Neither ciclosporin nor mycophenolate mofetil is very effective in controlling joint manifestations of JIA in children. 19

More recently, animal models and corroborative human evidence23 support the role of tumour necrosis factor alpha (TNF-α) in the aetiopathogenesis of uveitis and, moreover, the potential value of its inhibition as a therapeutic intervention. 24 Studies on experimental models of autoimmune uveitis have demonstrated that TNF-α plays a pivotal role in pathogenesis of intraocular inflammation,23 which has been borne out in treatment of adult uveitis. 24 In mouse models of anterior uveitis, deleting the TNF p55 receptor alone is as effective as combined tumour necrosis factor p55 and p75 knockout animal in demonstrating reduced ocular inflammation,25 equivalent to the effect of tumour necrosis factor p55 fusion protein. 26 In an animal model of uveitis, infliximab reduced disease severity,27 albeit at doses of 20 mg/kg.

Translating this to humans, several case series have been published demonstrating the efficacy of infliximab and adalimumab (Humira^®; AbbVie Inc., Ludwigshafen, Germany) in the treatment of severe refractory uveitis in adults and children. 28–33 In contrast, etanercept has been reported not to halt the onset of uveitis or be more effective than placebo,34,35 and is less effective than infliximab in treating JIA-uveitis. 31,36,37 There are a number of reports of new-onset uveitis associated with etanercept use in JIA. 38 An adverse events (AEs) register-based study examining these cases determined that, although the frequency was greater for etanercept than for infliximab or adalimumab (n = 20, 4 and 2 cases, respectively), causality could not be established. 39 Etanercept is not considered to be effective in treating intraocular inflammation. 31

Intervention

Adalimumab is a fully human monoclonal antibody, engineered by gene technology that uses site-directed mutagenesis to enhance its binding efficiency to TNF-α. It does not contain non-human or artificial protein sequences. Adalimumab binds only to TNF-α and has an elimination half-life of approximately 2 weeks. The antibody has been studied extensively in vitro as well as in vivo and in animal toxicology experiments. A clinical trial of adalimumab as monotherapy or in combination with MTX in adult subjects with rheumatoid arthritis showed a significant clinical response. 40 In children with JIA, a multicentre randomised, double-blind stratified parallel-group trial has shown a significant benefit in children with active arthritis: disease flares (the primary end point) occurred in a significantly lower percentage of those receiving adalimumab than of those receiving placebo [13 of 30 (43%) vs. 20 of 28 (71%); p = 0.03]. 41

Studies in paediatric non-infectious uveitis have shown very promising results with adalimumab, with 21 out of 26 eyes from 14 children with JIA- or idiopathic-uveitis showing improvement in inflammation. 42 In another retrospective case series of 18 paediatric patients with uveitis, 88% had a substantial decrease in ocular inflammation and adalimumab showed corticosteroid-sparing potential. 28

At the time of starting the randomised controlled trial of the clinical effectiveness, SafetY and Cost-effectiveness of Adalimumab in Combination with MethOtRExate for the treatment of juvenile idiopathic arthritis associated uveitis (SYCAMORE), there were no prospective studies of efficacy and safety of anti-TNF-α agents in JIA-associated uveitis, or of their cost-effectiveness. In the randomised controlled trial of adalimumab in JIA that demonstrated efficacy and supported its safety, the most commonly reported AEs were infections and injection-site reactions. 41 Serious adverse events (SAEs) that were considered to be possibly related to the study drug by the investigator occurred in 14 patients. Seven of these included one case of bronchopneumonia, one of herpes simplex infection, one of pharyngitis and one of pneumonia, and two cases of herpes zoster infection. In this trial, there were no deaths, malignant conditions, opportunistic infections, cases of tuberculosis, demyelinating diseases or lupus-like reactions. 41 The fixed-dose model of fortnightly 20 mg of adalimumab (for 16 weeks) for children weighing < 30 kg and 40 mg for children weighing ≥ 30 kg selected for this trial is based on the data generated in the previously mentioned trial using the same dosing regimen. 41

Although there are no published economic evaluations of JIA-associated uveitis, there are a number of economic evaluations of anti-TNF-α agents (including adalimumab) in JIA. These are of interest but limited applicability, because they are not directly transferable for estimating the cost-effectiveness of treatment in the context of uveitis management. The only study to adopt a costing perspective of the NHS in the UK is Shepherd et al.,43 who constructed a cost–utility Markov model to compare abatacept, adalimumab, etanercept and tocilizumab44–47 in JIA using disease flare as the measure of efficacy. The analysis was based on four economic evaluations of biological disease-modifying antirheumatic drugs (DMARDs) in JIA. Utility values were sourced from the Prince et al. study. 47 The incremental cost-effectiveness ratios (ICERs) for adalimumab, etanercept, tocilizumab and abatacept, versus MTX, were £38,127, £32,256, £38,656 and £39,536 per quality-adjusted life-year (QALY), respectively. The model results were found to be most sensitive to changes in utility values and the differences in cost-effectiveness of the biological DMARDs were primarily due to differences in drug acquisition cost. A limitation common to economic analyses in JIA is the challenge of obtaining valid utility scores and extrapolation of effects over a longer time period, both of which can significantly influence cost-effectiveness. A recent economic evaluation of adalimumab and dexamethasone intravitreal implant (Ozurdex^®; Allergan Ltd, Marlow, UK) for treating non-infectious intermediate uveitis, posterior uveitis or panuveitis in adults indicated that adalimumab was not cost-effective at £94,523 per QALY gained in active uveitis,48 but these findings may not be generalisable to children with active JIA-associated uveitis. The aim of the economic evaluation as part of the SYCAMORE trial was to assess the cost-effectiveness of adalimumab, based on utility and cost data acquired directly within the trial, and extrapolated using data on representative patients from routine care.

Objectives

The primary objective of the trial was to compare the clinical effectiveness of adalimumab in combination with MTX versus placebo with MTX alone, with regard to controlling disease activity in refractory uveitis associated with JIA.

The secondary objectives of the trial were to:

• evaluate short-term safety and tolerability of adalimumab in combination with MTX versus MTX alone, with regard to ocular complications of treatment, AEs and laboratory assessments

• determine quality of life and cost-effectiveness of adalimumab in combination with MTX versus MTX alone in severe uveitis associated with JIA

• determine the clinical effectiveness of adalimumab in combination with MTX versus MTX alone, with regard to underlying JIA disease activity

• determine the durability and magnitude of adalimumab efficacy response in sustaining inactive disease and achieving complete clinical remission

• determine the long-term safety of adalimumab in combination with MTX versus MTX alone

• assess the efficacy of treatment with adalimumab to permit concomitant medication reduction, in particular regional and parenteral steroids

• develop a fully consented, trial-related tissue bank for subsequent investigation.

Study design

This was a randomised, parallel-group, double-blind, placebo-controlled, multicentre clinical trial that compared the effects of adalimumab in combination with MTX versus placebo in combination with MTX in participants with active uveitis in association with JIA refractory to MTX monotherapy. Participants were randomised applying a ratio of 2 : 1 (in favour of adalimumab), stratified by centre.

Patients with persistently active JIA-associated uveitis (despite optimised MTX treatment for at least 12 weeks) were recruited from tertiary care centres throughout the UK.

A schematic of the study design can be seen in Figure 1.

FIGURE 1.

The SYCAMORE study design. OPD, outpatient department; QoL, quality of life.

The trial protocol has previously been published in an open access journal. 49

Participant inclusion and exclusion criteria

Patients who met the following eligibility criteria were considered for the trial.

Recruitment

Recruitment took place in 14 tertiary care centres throughout the UK. Participants were identified through rheumatology and ophthalmology outpatient clinics. Most tertiary care centres also set up referral links with local district general hospitals.

Informed consent

This trial recruited minors and young people aged < 16 years. Informed consent procedures reflected the legal and ethics requirements to obtain valid informed consent for this population. Prior written informed consent was required for all trial participants. In obtaining and documenting informed consent, the investigator complied with applicable regulatory requirements and adhered to good clinical practice and to the ethics principles that have their origin in the Declaration of Helsinki. 50

Information was provided to potential participants and their families verbally and in writing. All participants had the opportunity to discuss the project with the responsible investigator at site and/or a designated member of the research team. Discussions were supported with detailed written, ethics-approved patient information sheets and consent forms, provided directly to young people able to consent for themselves (defined in statutory instrument 2004 number 1031 as aged ≥ 16 years51) and parents/legal guardian of minors (aged < 16 years). Information leaflets appropriate to age and stage of development were provided to minors and their assent was obtained, when appropriate. Careful presentation was made of the known risks of the disease and trial medications and the possible benefits, as well as a detailed explanation of the trial procedures and protocol.

All participants were given the opportunity to ask any questions that may arise and to discuss the study with their surrogates and were given the time to consider the information prior to agreeing to participate.

All of the recruiting investigators were experienced rheumatologists and/or ophthalmologists familiar with imparting information to families and young people. All investigators obtaining consent had attended good clinical practice courses. When potentially eligible minors and young people were identified, they/their parent/the person with parental responsibility were approached by the investigator, or a designated member of the investigating team, and an opportunity was given to understand the objectives of the trial. The treatment schedule and trial visits were in line with standard clinical care, although they were made aware that additional travel may be needed if the trial assessments required they be reviewed at their tertiary centre rather than their local hospital. The potential risks and benefits of the anti-TNF-α agent were discussed, as were treatment failure criteria and what would happen if they chose not to enter the trial or had to withdraw from the trial for any reason. In addition, the rationale for the use of a placebo and the applied randomisation ratio was explained.

The right of the patient (non-minors) or parent/legal guardian (for minors) to refuse consent to participate in the trial without giving reasons was respected. After the patient had entered the trial, the clinician remained free to give alternative treatment to that specified in the protocol, at any stage, if they felt that it was in the best interest of the patient. However, the reason for doing so was recorded and the patient remained within the trial for the purpose of follow-up and data analysis in accordance with the treatment option to which they had been allocated. Similarly, the patient remained free to withdraw at any time from the protocol treatment and trial follow-up without giving reasons and without prejudicing their further treatment.

Adequate time to consider trial entry (generally 24 hours, although it was acknowledged that some patients/families came to a decision sooner) was allowed before written consent of the participant/parent/legal representative was obtained by the responsible clinician or other designated team member (recorded on the signature and delegation log).

Randomisation

Randomisation was undertaken during normal working hours (Monday to Friday, 09.00 to 17.00) by the pharmacy departments of participating centres on receipt of a randomisation request form and prescription from authorised clinicians. Pharmacy personnel verified that these documents were appropriately completed before proceeding.

It was the responsibility of the principal investigator or delegated research staff to:

• notify the pharmacy of potential randomisations so that the pharmacy could ensure that adequate drug supplies were at site

• complete the appropriate trial documents and deliver these to the pharmacy department at their centre in order that the pharmacy could proceed with a randomisation.

Participants were randomised using a secure (24-hour) web-based randomisation system, which was controlled centrally by the Clinical Trials Research Centre (CTRC). Randomisation lists were generated in a 2 : 1 ratio in favour of the active therapy. The lists were produced by an independent statistician (based at CTRC, but otherwise not involved in the trial) in Stata^® version 9.2 (StataCorp LP, College Station, TX, USA) using the ralloc command. The randomisation lists were stratified by centre but in order to reduce the predictability of the randomisation sequence, the randomisation numbers were sequential across all sites (rather than within each site) to make it appear that there was no stratification by centre. For smaller sites (i.e. expected recruitment of < 10), a block size of three was used. For larger sites (i.e. expected recruitment of at least 10), random block sizes of three and six were used.

Participant treatment allocation was displayed on a secure webpage and an automated e-mail confirmation was sent to the authorised randomiser.

Description of interventions

All participants received a stable dose of MTX and either adalimumab (20 mg/0.8 ml for participants weighing < 30 kg or 40 mg/0.8 ml for participants weighing ≥ 30 kg) or placebo (0.8 ml, based on body weight) via a subcutaneous injection every 2 weeks for a maximum period of 18 months.

All participants in both arms continued to receive a stable dose of MTX at a minimum dose of 10–20 mg/m² and a maximum dose of 25 mg as a non-IMP throughout the 18-month treatment period.

Clinical trial supplies were to be delivered to an investigator site only once the site had been initiated by the CTRC, acting on behalf of the sponsor to ensure full ethics and regulatory approvals had been granted. The size of the shipments to each site were predetermined, based on the participant recruitment target for that individual site. Recruitment was monitored centrally and drug shipment dates tailored accordingly, to ensure that pharmacies held adequate supplies of trial treatment. Pharmacies documented all shipment receipts and provided copies of this documentation to the CTRC. IMP stock was to be received by a designated member of the pharmacy department and stored at 2–8 °C, with temperature monitoring and in accordance with IMP regulations. Records of all shipments were to be kept in the pharmacy site file. All temperature excursions or damage to stock was reported to CTRC, which liaised with AbbVie Inc. for assessment.

The dose of adalimumab or placebo remained the same as at trial entry, regardless of minor fluctuations in weight that may cause a participant to cross the 30-kg threshold for the upper and lower doses.

The first dose was administered by the research/clinical team looking after the participant. Participants, or a family member, were invited to self-administer the study treatment after the first dose and taught how to do this, following the procedures in place within each participating centre. The first self-administered dose was carried out under the supervision of the clinical team, which would ensure that the participant was confident and able to carry out all parts of the procedure appropriately and accurately. The trial provided validated cooler bags for participants to transport their trial medication home. If participants did not want to self-administer the trial medication, then arrangements were put in place, on an individual basis, to ensure that trial medication was administered as prescribed.

Investigational medicinal products were labelled in accordance with regulation 46 SI2004/1031 and the detailed guidance provided in annex 13 of the EU Good Manufacturing Practice (GMP) guide. 51

Blinding

Participants, investigators, study personnel, the trial co-ordinator, statisticians and data management personnel were all blinded to the trial medication that the participant received. Pharmacy department staff were not blinded to the trial medication that the participant received.

This trial was placebo controlled and all trial assessments were carried out by health professionals, parents/carers and participants without knowledge of treatment allocation. The placebo solution for subcutaneous injection was a clear, colourless solution (matching the adalimumab vial) presented in a single-use vial for subcutaneous injection in volumes of 0.8 ml.

The packaging of the kit for adalimumab and placebo was identical. Each kit consisted of two vials of adalimumab or placebo in an outer carton. The vials of adalimumab and placebo were also identical in appearance.

Treatment allocation was concealed unless knowledge was essential for the ongoing care of the participant. If knowledge of treatment allocation was required by the responsible investigator, the process was to obtain it via the pharmacy department at the respective hospital, which would then complete an unblinding case report form (CRF) and submit this to the CTRC.

All children participating in this study during the active treatment phase of the study were immunosuppressed, in view of their concomitant MTX therapy and/or potential corticosteroid therapy, irrespective of whether they were on adalimumab or placebo. In addition, for the purpose of out-of-hours management of the patient, all participants were presumed to be on anti-TNF-α therapy and managed as such. In this way, in the event of an AE or SAE, such as an intercurrent infection, the treating clinician managed patients presuming them to be on anti-TNF-α therapy. For this reason, if unblinding was deemed necessary, this was carried out via the local pharmacy department following the procedure described below. If out-of-hours unblinding was required, this was accessed via the local pharmacy department’s on-call service.

Data collection and management

For SYCAMORE, a paper CRF was used to collect participant data at each study visit. The paper CRF was designed by the Trial Management Group (TMG) and CTRC specifically for the study, in line with the trial protocol. Completed paper CRFs were transferred from the trial sites to the CTRC; data were then entered in to a Good Clinical Practice-compliant database (MACRO; Elsevier, Amsterdam, the Netherlands) by trial staff.

The configuration of the database was specific to SYCAMORE: there were built-in validations on certain aspects of the trial data. Any missing or inconsistent data were queried with the site using paper data query forms: query responses were completed by site staff and returned to the CTRC for entry into the database. A full audit trail of changes to the data was maintained.

Outcome measures

Data collection tools

Sample size

Sample size calculations were undertaken using nQuery Advisor software version 4.0 (Statistical Solutions, Saugus, MA, USA).

Details of the original and revised sample size calculations are given below. Sample size revisions were necessary as a result of lower patient availability than expected.

Tissue bank

A blood sample collection system was developed, in accordance with Arthritis Research UK’s guidelines61 on detailed clinical and related material banks.

Written information was provided to families for this part of the study and written informed consent (with assent when appropriate) obtained for those who wished to provide blood samples. Participants who did not give consent to provide samples were eligible to take part in the main part of the trial.

Samples have been collected for future analysis and could be used as a resource to investigate the pharmacogenetics, aetiopathogenesis and identification of biomarkers of JIA-associated uveitis, for example. Understanding the genetic basis of age-specific disease processes allows consideration of the unique and rapid period of human development through to adulthood. Pharmacologic modulation of developing gene networks may have unintended and unanticipated consequences that do not become apparent or relevant until later in life. Early predictors of response allow future personalised treatment prescription in children.

Blood samples were collected pre treatment (0 months) and at two time points post treatment (at 3 months and 18 months).

Patient and public involvement

Patient and public involvement (PPI) representatives have been an integral part of SYCAMORE since the initial prioritisation, design stage and funding applications. PPI representatives provided detailed input into all aspects of the protocol design and all subsequent amendments and patient information sheets and consent forms and amendments. The patient information sheets, consent forms and assent forms were reviewed and feedback was given by the Medicine for Children Research Network young person’s advisory group. PPI representatives also provided input into any information sent to patients, such as letters to explain the closure of recruitment and the frequently asked questions section on the SYCAMORE website.

Changes to the protocol

Over the course of the whole trial, eight amendments were made to the protocol. These consisted of six non-substantial and 15 substantial amendments. A summary of amendments follows and the full list is reported in the trial protocol.

Compliance with intervention

Participant diaries and dosing records determined tolerability and compliance throughout the trial period. The parent/guardian of a participant maintained a diary for all trial and other medications that were administered outside the trial visit (i.e. at home). In the diary, the date and time that the drug was administered were recorded. The dosing records were reviewed and verified for compliance at each visit by the research personnel at the trial centre.

Trial management and oversight

Interim analysis

Interim monitoring reports of the accumulating data were performed at regular intervals (at least annually) for review by the IDSMC. In addition to the interim monitoring reports, the IDSMC requested that an interim analysis of the primary outcome was to be undertaken. The Peto–Heybittle stopping rule was applied to both interim analyses of the primary outcome. This required an extreme p-value of p < 0.001 as evidence to indicate potentially stopping for benefit. This approach was used to allow the IDSMC flexibility with the number and timing of further analyses, based on current safety and efficacy data, as it has the added benefit of preserving an overall two-sided type I error of 0.05 for the final analysis62 (see Chapter 4).

Final analysis

The results of this report are based on the data collected during three phases of the trial. Chapter 5 presents the results from the double-blind phase of the trial, Chapter 6 presents the results from the integrated analysis of the double-blind phase and the open-label phase, and Chapter 7 presents the results of the integrated analysis of the double-blind phase, open-label phase and the follow-up phase of the trial.

The data presented in this report are based on data at the final database lock that occurred on 2 August 2017.

The results from the blinded phase of the trial that were presented in the published manuscript of SYCAMORE were based on data snapshots taken on 11 September 2015 (primary outcome and AE data) and 24 May 2016 (secondary outcomes).

Because data were still being received from sites during this period, there were minor ongoing updates and changes to the database. The differences between the results in the New England Journal of Medicine manuscript63 and Chapter 5 in this report are documented in Appendix 2.

There were three statistical analysis plans (SAPs) written for the final analysis of study results. The first SAP was written by the trial statistician and contained detail only of the analyses for the primary outcome and the safety data of the blinded phase of the trial. The second and third SAPs were written after the completion of the primary analysis (after the blind had been broken to treatment allocation) and, therefore, written by independent statisticians who were blinded to the allocation of the trial. The second SAP described the analysis that was conducted on the secondary outcomes for the blinded phase of the trial and the third SAP described the analyses that were conducted for the open-label and follow-up phases of the trial. All three SAPs are available at www.journalslibrary.nihr.ac.uk/programmes/hta/095101 (accessed 4 February 2019).

General statistical considerations

The primary and secondary outcomes were all analysed using the (two-sided) 5% level of significance and 95% confidence intervals (CIs) are presented throughout. There were no adjustments for multiple testing; rather, all secondary analyses were treated as hypothesis generating.

The primary and secondary analyses used the principle of intention to treat (ITT) based on all the randomised participants, meaning that participants who consented and were randomised were analysed on the basis of the treatment they were randomised to, regardless of whether or not they received it. If consent for treatment was withdrawn but the participant was happy to remain in the study for follow-up, they were followed up until completion. However, if they decided to withdraw consent completely, then the reasons for withdrawal of consent were collected (if possible) and reported for both groups.

All statistical analyses were conducted using SAS^® V9.3 (SAS Institute, Cary, NC, USA), except for the joint modelling and competing risk analyses, which were conducted in R (R Foundation for Statistical Computing, Vienna, Austria).

Analysis of baseline data

Demographic and baseline characteristics were summarised for each treatment group and overall, using descriptive statistics. No formal statistical testing was performed on these data. Descriptive statistics, including the number of observations; mean; standard deviation (SD); median, minimum and maximum for continuous variables; and counts and percentages for discrete variables, are presented as appropriate.

Analysis of the primary outcome

The primary outcome of ‘treatment failure’ was a time-to-event outcome. For those patients who entered the trial and whose eyes (both) met the entry criteria, the time to the first eye to fail treatment was used. This was not observed in all participants; those participants who did not experience an event were classed as censored. The event time or censoring was calculated by subtracting the randomisation date from one of the following scenarios:

• Participants who failed treatment – date of the visit at which they failed treatment.

• Participants who completed the trial treatment phase without failing treatment – censored at date of 18-month treatment visit.

• Participants who discontinued treatment early and agreed to follow-up –

  • if they were assessed to be a treatment failure during a follow-up visit within 18 months following randomisation: date of follow-up visit

  • if they were not assessed to be a treatment failure during their follow-up visits: censored at date of last follow-up visit within 18 months following randomisation.

• Participants that were lost to follow-up – censored date of last treatment visit.

Survival estimates were calculated using the method of Kaplan and Meier, with curves for each treatment group presented graphically with numbers at risk.

The p-value obtained from the log-rank test and the hazard ratios (HRs) with 95% CIs were used to assess differences in failure estimates across treatment groups. The statistical test for the primary end point was performed at a two-sided significance level of 0.05.

Participants who withdrew from trial treatment (providing they did not withdraw from the entire study or withdraw consent) moved to the follow-up phase of the trial, were assessed for the primary outcome and could still contribute to the ITT analysis.

Any participants who withdrew from follow-up contributed primary outcome data until the point at which they withdrew from the trial.

Missing data were monitored and strategies developed to minimise this occurrence. Missing data were handled by considering the robustness of the complete-case analysis to sensitivity analyses using various imputation assumptions; this was informed by data collected on the reasons for missing data.

Nine sensitivity analyses were carried out to determine the robustness of the results from the primary analysis:

1. Best case – all participants who withdrew from treatment were treated as censored at time of treatment withdrawal.

2. Worst case – all participants who withdrew from treatment were treated as treatment failures (i.e. events at time of treatment withdrawal).

3. Methotrexate – any participants who withdrew from treatment because of MTX intolerance were classified as treatment failures at the time of treatment withdrawal.

4. Component 1 of primary outcome – all participants who failed for component 1 at a treatment failure assessment had their event date as the mid-point between this visit and the previous visit instead of the date of this visit.

5. Component 2 of primary outcome – all participants who failed for component 2 at a treatment failure assessment had their event date as the date that they commenced concomitant medications (a) used against predefined acceptable criteria (see SYCAMORE protocol49) or (b) any of the concomitant medications not allowed. The event date was determined by the co-chief investigators making a clinical decision following review of the participants’ concomitant medications taken since their previous visit.

6. Component 3 – all participants who failed for component 3 at a treatment failure assessment had their event date as the exact date that they qualified as ‘intermittent or continuous suspension of study treatment (adalimumab/placebo) for a cumulative period longer than 4 weeks’. 28 The event date was determined by the chief investigator making a clinical decision following review of a participant’s trial treatment dose recordings in the treatment diaries.

7. Any missing primary outcome data – any cases of missing data for any of the primary outcome components (except for unscheduled visits) had data imputed on a worst-case basis, because the missing data could have meant that a participant failed earlier than recorded. All participants were treated the same, regardless of whether or not they had a treatment failure.

8. Loss to follow-up – in the primary analysis of the primary outcome, participants who were lost to follow-up were treated as withdrawals, assuming that they were non-informative. The reasons for loss to follow-up, when available, were blindly reviewed by Michael W Beresford (co-chief investigator) and Andrew Dick (ophthalmology expert on the TMG) to see whether or not any might be related to the prognosis. If any were deemed to be related, a sensitivity analysis would be undertaken, assuming these participants to be a treatment failure at the time of last recorded visit.

9. Incorrectly identified to be a treatment failure – once a participant was deemed to have failed treatment, treatment was stopped and they entered the follow-up phase of the study, providing that they still wished to be followed up. Any participants wrongly identified as treatment failures by the assessing physician would be classed as a withdrawal at their time of ‘treatment failure’.

Analysis of secondary outcomes

The secondary continuous outcomes were analysed using the following methods:

• Chi-squared test, relative risk (RR) and 95% CI for the number of participants –

  • failing treatment

  • needing pulsed corticosteroids

  • having uveitis disease flares

  • having resolution of associated optic nerve or macular oedema

  • with uveitis disease control

  • entering disease remission for uveitis

  • undergoing JIA disease flare

  • with minimum disease activity of JIA

  • having remission of their JIA on and off medication

  • requiring a change in biologics due to failure to respond from arthritis.

• Change from baseline –

  • laboratory parameters (haematological and biochemical assessments).

• Poisson regression –

  • total oral corticosteroid dose

  • systemic corticosteroid dose.

• Competing risks –

  • reduction in systemic corticosteroid dose

  • topical corticosteroid use.

• Joint modelling of longitudinal and time-to-treatment failure data –

  • visual acuity

  • CHAQ and CHQ

  • ACR 30, 50, 70, 90 and 100

  • JADAS.

• Random intercept model –

  • duration of sustaining inactive disease.

No between-group statistical analyses were conducted for compliance data or urinalysis data; instead, summary data are presented for these outcomes.

Adverse events were tabulated by the Medical Dictionary for Regulatory Activities (MedDRA) version 18.0 system by organ class and preferred term. In addition, summaries by severity and relationship to study drug were completed. SAEs and events that led to premature withdrawal were listed in detail. No formal testing of AE or SAE data was planned.

Initial meeting

The first meeting of the IDSMC took place on 2 August 2011. During this meeting, the IDSMC reviewed the study protocol, was updated on study progress and agreed the IDSMC charter and also the expedited safety reports that it wished to receive.

Future meetings

The IDSMC met at regular intervals following the initial meeting (11 April 2012, 18 December 2012, 3 July 2013 and 18 February 2014). During these meetings, the committee was updated on the progress of the trial and was also presented with a report that contained data on recruitment, data completeness and safety.

There were two sessions held during the meetings, an open and closed session. Present at the open session were the independent members of the IDSMC and also relevant members of the TMG (i.e. co-chief investigators, lead trial ophthalmologist, trial co-ordinator, statistical team). During the open session, data were presented overall and not split by treatment group. The open session was then followed by a closed session, which was attended only by the independent members of the IDSMC and the statistical team responsible for the production of the report.

On 18 February 2014, the IDSMC decided that the recruited sample was approaching a size at which an interim analysis was needed in order to allow them to protect participants from potential harm or to stop the trial early if there was a clear benefit. This allowed for avoidance of delay by bringing the benefits to patients once the question that had been posed in the trial had been answered with sufficient statistical certainty.

The IDSMC, therefore, requested that an interim analysis be conducted and the results be presented at the next IDSMC meeting, which was to be held on 29 September 2015.

Statistical methods

A full SAP was written prior to conducting the interim analysis.

The Peto–Haybittle stopping rule was applied to the interim analysis of the primary outcome. This required an extreme p-value of < 0.001 as evidence to stop for benefit. This approach was used to allow the IDSMC flexibility with the number and timings of further analyses, based on current safety and efficacy data, as it had the added benefit of preserving an overall two-sided type I error of 0.05 for the final analysis.

Survival estimates were calculated using the method of Kaplan and Meier, with curves for each treatment group presented graphically along with numbers at risk.

The log-rank test was used to assess differences in failure estimates across treatment groups.

Interim analysis 1

The first interim analysis report was based on follow-up data from 71 participants who had been randomised (adalimumab, n = 48; placebo, n = 23). Safety data (SAE/AE) were based on data from 64 participants (adalimumab, n = 43; placebo, n = 21).

There was a total of nine treatment failures recorded in 48 participants in the adalimumab group (19%) and 13 treatment failures recorded in 21 participants in the placebo group (62%), and five withdrawals from 48 participants recorded in the adalimumab group (10%) and four withdrawals from 23 participants recorded in the placebo group (17%).

The Kaplan–Meier plot is shown in Figure 2. The log-rank chi-squared statistic was 15.77 and the associated log-rank p-value was < 0.0001.

FIGURE 2.

Interim analysis 1: Kaplan–Meier plot.

At the time of the first interim analysis, 353 AEs had been reported in 34 of the 43 participants in the adalimumab group who qualified for the safety analysis set (i.e. received at least one dose of treatment). There had been 89 AEs reported in 14 out of the 21 participants in the placebo group who qualified for the safety analysis set (i.e. received at least one dose of treatment).

There had been 10 SAEs reported in 9 participants from a total of 43 participants in the adalimumab group (21%). There had been one SAE reported from a total of 21 participants in the placebo group (5%).

The IDSMC considered the report and made the recommendation that the trial should continue with recruitment. However, it did make three observations with regard to the trial based on the report that it received:

1. We (the IDSMC) have been much reassured over recent months, and in the current report, that the recruitment rate is now running just ahead of the amended estimated rate. We would like to commend the TMG on overcoming significant difficulties to achieve this favourable situation.

2. We do, however, remain concerned that some centres are performing exceptionally badly; there is, for example, a marked contrast between the performance of Bristol, where ≈50% of those screened are recruited, and Birmingham, where < 1% of those screened are being recruited. While these two are at the extreme ends, there are several other poorly recruiting centres – this appears as a weakness with potential implications for the representativeness of the trial sample. An investigation into the problems in these centres, with a view to resolving their issues, would be time well spent.

3. We are also concerned about the levels of missing data, which seem high for certain variables. For example, in table 10.2 [referring to the meeting report], in 14% the relationship between the AE and the trial medication is missing and in 5% the severity of the AE is missing. Whilst we appreciate that you are taking steps to reduce these rates of missing data, we would like to register our concern in regard to this problem of missing data, which is evident in other parts of the report. Please take all steps to recover missing data and reduce the levels going forward.

The TMG provided further information with regards to missing data for both the primary outcome and the safety data. The response contained a more detailed breakdown of missing and unobtainable data for each component of the primary outcome and safety data, by site and overall.

This thorough investigation of the missing and unobtainable data reassured the IDSMC that the monitoring procedures being implemented by the TMG and the engagement of sites with the data manager were ensuring that these figures were kept to a minimum.

The IDSMC requested that a further interim analysis be conducted at their next meeting in March 2015.

Interim analysis 2

The IDSMC met on the 25 March 2015 to discuss the results of the second interim analysis of the SYCAMORE data.

At the time of the report, there were a total of 85 participants who had been randomised (adalimumab, n = 57; placebo, n = 28). From these 85 participants:

• 80 were included in the analysis (adalimumab, n = 54; placebo, n = 26).

• Five were excluded –

  • four had no randomisation CRF input on the trial database

  • one had their randomisation CRF input on the trial database but had not had any further follow-up visits input to the trial database.

There were a total of 10 withdrawals from the trial: five withdrawals from the adalimumab group (n = 57) (9%) and five withdrawals from the placebo group (n = 28) (18%). There were a total of 27 treatment failures in the trial: 12 treatment failures on adalimumab (21%) and 15 on placebo (54%).

The Kaplan–Meier plot is shown in Figure 3. The log-rank chi-square statistic was 14.63 and the associated log-rank p-value was 0.0001.

FIGURE 3.

Interim analysis 2: Kaplan–Meier plot.

The safety data set was based on 80 participants (adalimumab, n = 54; placebo, n = 26).

There were a total of 10 withdrawals from the trial: five withdrawals from the adalimumab (n = 57) (9%) and five withdrawals from the placebo group (n = 28) (18%). There were a total of 27 treatment failures in the trial: 12 treatment failures on adalimumab (21%) and 15 on placebo (54%).

Following the review of the second interim analysis, the IDSMC carefully considered evidence from the first interim analysis, the subsequent satisfactory conclusion of the vast majority of missing data, and also the continued presence of such a strong treatment effect. The IDSMC subsequently agreed unanimously that:

• The levels of missing data were acceptable with all reasonable efforts being made to collect outstanding missing data items where feasible, or to confirm the fact that the data were unobtainable.

• The AEs and SAEs were in keeping with expectations for the medications in use, based on clinical experience and previous published reports.

• The statistical significance of the beneficial effect of the investigational medicinal product (IMP) in the interim analysis substantially exceeded the predetermined requirement for consideration of stopping the trial on the basis of a powerful positive treatment effect.

• The IDSMC should obtain guidance on procedure prior to making any stopping recommendation. Options for stopping included either to immediately stop and fully close the trial or to stop recruitment and continue collecting data until all current participants had completed their passage through the protocol.

It was decided by the IDSMC that no immediate recommendation should be made, but that the IDSMC chairperson should take advice in a timely manner, following which the IDSMC would communicate further among themselves before arriving at a final recommendation.

Summary of Independent Data and Safety Monitoring Committee and Trial Steering Committee meetings following interim analysis 2

The IDSMC and TSC met on 8 April 2015 in a combined meeting that involved all of the independent members of both committees, as well as both co-chief investigators and the lead statistician (non-independent members). The IDSMC advised the TSC that the trial should stop recruiting with immediate effect. They further recommended that all participants currently in the trial should continue in their randomly allocated treatment regimen, blinded, and follow treatment scheduling as per protocol. The IDSMC did not unblind the TSC (or co-chief investigators) to the results of the trial during the meeting. The TSC decided to consider the recommendations of the IDSMC overnight and meet again on 9 April 2015.

The independent members of the TSC met with the co-chief investigators and lead statistician (non-independent members of the TSC) again on 9 April 2015. The chairperson of the TSC contacted the chairperson of the IDSMC, requesting that the IDSMC make formal recommendations to the TSC in a formal document; they agreed to meet again on 10 April 2015.

The same independent members of the TSC, the co-chief investigators and lead statistician met on 10 April 2015 and discussed the formal recommendations that had been made by the IDSMC.

The voting members of the TSC (the three independent members and one non-independent member) then voted unanimously in favour of the following specific IDSMC recommendations:

• Recruitment to SYCAMORE should not be reinstated. This was based on a positive signal of efficacy of the IMP (adalimumab) versus placebo, which exceeded the prespecified level.

• All participants in the trial should be invited to attend a final blinded assessment visit. On completion of assessments, their treatment allocation should then be unblinded.

• All participants in the active arm of the trial (adalimumab) should continue follow-up in an open-label fashion, as this long-term follow-up would contribute important additional data on quality of life, utilities and long-term efficacy. Subsequent changes in therapy would be at the discretion of the treating clinical team.

Participant recruitment

The first participant was randomised into SYCAMORE on 27 October 2011 and the last participant was randomised on 31 March 2015. The last blinded visit took place on 16 June 2015.

Fourteen of the 17 sites randomised at least one participant and five centres randomised five or more participants. The flow of participants through the trial is represented in the Consolidated Standards of Reporting Trials (CONSORT)64 flow diagram in Figure 4.

FIGURE 4.

The CONSORT flow diagram for all trial participants.

A total of 332 patients were assessed for eligibility from 519 screenings (patients could be screened on multiple occasions).

The number of participants enrolled overall by month is provided in Appendix 3 (see Figure 13); the screening overview by centre is also given in Appendix 3 (see Table 40). The main reasons for ineligibility included the following: the patient did not have active uveitis (49%), the patient had not failed on MTX or been on MTX for at least 12 weeks with a stable dose (19%), the patient had been on another biological agent within the previous five half-lives of the agent (6%), the patient had more than six topical steroid eye drops per day at randomisation (6%) or other reasons (11%). For five patients, it was not clear whether or not they were eligible.

A total of 130 patients were eligible to participate in the trial from 139 screening visits; 90 patients were randomised. The reasons for the 45 patients (49 screening visits) not providing consent are given in Appendix 3 (see Table 41).

Recruitment rates

The original target sample size of 154 participants was expected to be achieved within 30 months of recruitment. This original target was based on feasibility data provided by each of the centres that took part in the trial. The actual rates of recruitment during the trial were lower than anticipated (see Appendix 3, Figure 13) and, therefore, the recruitment period of the trial was extended by 36 months on 16 June 2014, to allow for additional time to recruit the necessary number of participants.

The sample size of the trial was also revised, which meant that the target sample size was reduced from 154 to 114 participants. Other strategies were used to improve recruitment, including the aligning of the rheumatology and ophthalmology clinics in any sites that did not already have this service and holding a series of regional investigator meetings and local investigator meetings each month via teleconference, which included members of the CTRC and the co-chief investigators and lead SYCAMORE ophthalmologists. Regular newsletters were sent to sites to keep sites aware of recruitment and to maintain interest in the trial. All sites were also encouraged to build referral links with district general hospitals to allow referrals of potential patients for screening.

Comparison of interventions

The ITT analysis population included all 90 participants who were randomised (see Figure 4). There were no exclusions from either the safety or the ITT population.

All participants who withdrew consent for trial continuation contributed outcome data up until the point of withdrawal.

The memberships of the analysis set for the primary outcome and safety data set were determined and documented prior to the blinding being broken and the treatment allocations being requested.

Trial completion and trial exit

There were 19 participants who withdrew prematurely during the blinded treatment phase of the trial (Table 1). There were a total of 11 (18%) withdrawals in the adalimumab group (nine continued into the follow-up phase of the trial and two withdrew complete consent) and eight (23%) in the placebo group (seven continued into the follow-up phase of the trial and one withdrew complete consent). The majority of premature withdrawals were for non-safety reasons (see Table 1). The most common reason for withdrawal in the adalimumab group was because of MTX intolerance and in the placebo group it was the worsening of uveitis (that did not meet the exit criteria).

Baseline characteristics

The demographic baseline data of the 90 participants randomised across all centres were comparable between the two groups (Table 2). The mean age in the placebo group was slightly lower than that in the adalimumab group and the proportion of females and males was approximately the same in the two treatment groups, with more females than males.

The distribution in weight was similar in both groups, as was physical global assessment of disease activity, antinuclear antibody and double-stranded deoxyribonucleic acid.

A total of 65 (72%) participants entered the trial with one eye that was eligible for evaluation (i.e. they met the inclusion criteria for active uveitis in one eye only) and 25 (28%) participants entered the study with two eyes that were eligible (i.e. they met the inclusion criteria for active uveitis in both eyes). Therefore, a total of 115 eligible eyes entered into the study.

Table 3 shows ocular data collected at baseline, presented at the eye level rather than the individual level for ease of reading. The overall mean for logMAR score was 0.05; this was slightly higher in the placebo group than in the treatment group. Of the eligible eyes, 76 (66%) had a score of 1+ for the AC score; the numbers in each of these categories were similar in both groups, as were the flare score and vitreous haze grading. The mean IOP in each group was similar, with an overall mean of 14.54 mmHg.

When the best and worst scores were entered for those with two eligible eyes, the baseline results were very similar to those when the eyes were looked at independently.

Table 4 shows the baseline data describing the classification by disease subtype of the participants’ JIA status. A total of 53 (59%) participants had persistent oligoarticular JIA and 21 (23%) had extended oligoarticular JIA. The proportions for each of the subcategories were similar for both groups. The mean overall JIA disease duration was 5.33 years, with the duration being slightly longer in the treatment group. A total of 66 (73%) participants had a negative rheumatoid factor (RF); this was slightly higher in the adalimumab group.

Unblinding of randomised treatment

There were 57 (63%) participants who were unblinded during the course of the trial: 38 (67%) in the adalimumab group and 19 (33%) in the placebo group. A breakdown of the stage of the trial that the participants were in at the time of unblinding is given in Table 5.

Protocol deviations

Protocol deviations were monitored centrally via evaluation of inclusion/exclusion criteria at trial entry and throughout the course of the trial. During the course of the blinded phase of the trial, a total of 74 (82%) participants had at least one major protocol deviation (Table 6). All protocol deviations were agreed with the co-chief investigators before the final treatment allocations were requested on 11 September 2015.

Primary outcome

Adalimumab plus MTX significantly delayed the time to treatment failure compared with placebo and MTX. There were a total of 14 (23%) treatment failures for the 60 participants in the adalimumab group and 17 (57%) failures for the 30 participants in the placebo group. The median time to treatment failure was 24.10 weeks (95% CI 14.70 weeks to 81.00 weeks) in the placebo group and was not reached in the adalimumab group within the 18-month treatment period because fewer than half of the subjects experienced treatment failure at the conclusion of the trial (Figure 5).

FIGURE 5.

Primary outcome ITT Kaplan–Meier plot.

Reasons for the treatment failure of each participant can be found in Appendix 3 (see Table 42).

In the adalimumab group, five participants (trial numbers 0114011, 0116006, 0116026, 0249024 and 0030073) were classified as treatment failures because they had taken permitted concomitant medications against the acceptable criteria, three participants (trial numbers 0243023, 0249043 and 0133058) were given non-permitted concomitant medications, three participants (trial numbers 0069039, 0116062 and 0116066) had missed doses that met the failure criteria, two participants (trial numbers 0246055 and 0069076) had sustained SUN scores (as recorded at entry grade) that were still present after 6 months of therapy and one participant (trial number 0030035) had both taken permitted concomitant medications against the acceptable criteria and missed doses that met the failure criteria.

In the placebo group, two participants (trial numbers 0114033 and 0116067) received permitted concomitant medications against the acceptable criteria, seven participants (trial numbers 0249019, 0249025, 0246030, 0243032, 0540037, 0393075 and 0249086) were given non-permitted concomitant medications, one participant (trial number 0116003) missed doses that met the failure criteria and seven participants (trial numbers 0116008, 0036018, 0116005, 0249029, 0249047, 0116051 and 0249059) had sustained SUN scores (as recorded at entry grade) that were still present after 6 months of therapy.

Four participants who were classified as treatment failures did not subsequently enter the follow-up phase of the trial and withdrew completely from the trial with no further follow-up. One of these was in the adalimumab group (2%) and three were in the placebo group (10%).

The HR indicated that treatment with adalimumab significantly decreased the hazard of treatment failure by 75% (HR 0.25, 95% CI 0.12 to 0.51). The results of the log-rank test offered strong statistical evidence that the placebo group and adalimumab group differed with respect to time to treatment failure (p < 0.0001), relative to placebo. The HR was derived using Statistical Analysis Systems Procedure (SAS PROC) proportional hazards regression methods with no stratification factors.

Additional analyses

Post hoc analyses

Secondary outcomes

Use of corticosteroids over duration of study period

Topical corticosteroid use (frequency) compared with entry use

Time to reduction to fewer than two drops in topical corticosteroid

The outcome was time to reduction to fewer than two drops per day for those participants already at more than two drops per day at baseline. There were 63 participants who were on more than two drops per day at baseline [18 (60%) in the placebo group and 45 (75%) in the adalimumab group] and who were, therefore, included in the analysis.

Twenty-four (53.3%) of the 45 participants on adalimumab and three (16.70%) of the 18 participants on placebo reached fewer than two drops per day before treatment failure (or the 18-month treatment visit). Five participants (11.10%) on adalimumab and one (5.60%) participants on placebo reached the 18-month visit before reaching fewer than two drops per day and 16 (35.6%) of the adalimumab group and 14 (77.8%) of the placebo group had a treatment failure/withdrawal before reaching fewer than two drops per day.

The time to reduction to fewer than two drops per day was statistically significant in favour of adalimumab (HR 3.99, 95% CI 1.18 to 13.48; p = 0.03); the incidence plot is shown in Figure 6.

FIGURE 6.

Incidence plot of time to reduction to fewer than two drops per day.

Time to reduction to zero drops in topical steroid (post hoc analysis)

The outcome was time to reduction to zero drops for those participants already at more than zero drops at randomisation. There were 74 participants [25 (34%) on placebo and 49 (66%) on adalimumab] who were on more than zero drops at randomisation and who were, therefore, included in the analysis.

Twenty-five (51%) of the 49 participants on adalimumab and four (16%) of the 25 participants on placebo reached zero drops before treatment failure (or the 18-month treatment visit). Six participants (12%) on adalimumab and two participants (8%) on placebo reached the 18-month visit before reaching zero drops and 18 (37%) of the adalimumab group and 19 (76%) of the placebo group had a treatment failure/withdrawal before reaching zero drops.

The time to reduction to zero drops was statistically significant in favour of adalimumab (HR 4.01, 95% CI 1.40 to 11.51; p = 0.01); the incidence plot is shown in Figure 7.

FIGURE 7.

Incidence plot for time to reduction to zero drops.

Optic and ocular

Sensitivity analysis

The residuals from the separate fitted linear missed models (LMMs) for the logMAR indicated slight departures from the normality assumption. In general, fixed-effects estimates are robust to non-normal errors in LMMs. 65 The histograms of baseline logMAR scores (for analyses 1 and 2) appeared approximately normal (not shown) and, therefore, no further analysis from log-transformed data was considered.

In the primary analysis, a random-intercepts model for the longitudinal submodel was fitted. The second sensitivity analysis investigated fitting a random-intercepts and random-slopes model. This showed that inferences remained similar. For analyses 1 and 2, the treatment effect on the longitudinal outcome was –0.01 (95% CI –0.05 to 0.03) and for the random-intercepts and random-slopes model the treatment effect was –0.01 (95% CI –0.05 to 0.03); these were not statistically significant.

Quality-of-life assessment

Number of participants undergoing disease flare, in remission on and/or off medication for their juvenile idiopathic arthritis, and with minimum disease activity

Primary outcome

During the course of the open-label phase of the trial, there were three additional treatment failures that occurred in the adalimumab arm. One participant was classified as having a treatment failure because they had taken permitted concomitant medications against the acceptable criteria and two participants had sustained scores (as recorded at entry grade) that were still present after 6 months of therapy (see Appendix 3, Table 50).

There were a total of 17 (28.3%) treatment failures for the 60 participants in the adalimumab group and 17 (56.7%) failures for the 30 participants in the placebo group. Median time to treatment failure was 24.1 weeks (95% CI 14.7 weeks to 81 weeks) in the placebo group and not reached in the adalimumab group within the 18-month treatment period because fewer than half of the subjects experienced treatment failure at the conclusion of the study (Figure 8).

FIGURE 8.

Primary outcome Kaplan–Meier plot for blinded and open-label phase.

The results of the log-rank test from SAS PROC LIFETEST (SAS Institute In, Cary, NC, USA) offered strong statistical evidence that the placebo and adalimumab groups differed with respect to time to treatment failure.

The HR indicated that treatment with adalimumab significantly decreased the hazard of treatment failure by 74% (HR 0.26, 95% CI 0.13 to 0.51; p < 0.0001), relative to placebo.

Additional analyses

Post hoc analyses

Secondary outcomes

Sensitivity analysis

The inferences of the sensitivity analyses for the data from the blinded phase of the trial combined with the open-label data were the same as those from the blinded phase alone.

Quality-of-life assessment

Number of participants undergoing disease flare, in remission on and/or off medication for their juvenile idiopathic arthritis, and with minimum disease activity

Number of participants requiring change in biological or disease-modifying antirheumatic drug therapy as a result of failure to respond from arthritis

There were no further cases of participants requiring a change in biological or DMARD therapy as a result of failure to respond from arthritis in the open-label phase.

Juvenile Arthritis Disease Activity Score

The parameter estimates for each of the joint models of JADAS 10, 27 and 71 can be seen in Table 25. The treatment effect on the longitudinal JADAS 10, 27 and 71 was non-significant at the 5% level, implying that there is no difference between the treatments on the scores. However, all three p-values are close to the margin of statistical significance.

Laboratory parameters (haematological, biochemical analysis and urinalysis)

The reduced follow-up period meant that only a small proportion of participants had data at six time points. Therefore, only data from the first two follow-up visits have been presented in this report.

Use of corticosteroids over duration of study period (blinded, open-label and follow-up phases)

Methods

The economic analysis adopted the perspective of the NHS and Personal Social Services providers in England. A trial-based evaluation was extrapolated by 10 years using a Markov model in order to assess the costs and consequences of adalimumab treatment over an appropriate analytical time horizon. The primary outcome of the economic evaluation is the incremental cost per QALY with adalimumab in addition to MTX versus MTX alone.

Resource use and costs

Within-trial costs were estimated by measuring health-care resource use associated with both arms of the trial during the study period, including (1) adalimumab, MTX and other concomitant medication costs; (2) outpatient and accident and emergency (A&E) visits and contact with health-care professionals, including general practitioners (GPs) and school nurses; (3) hospitalisations; and (4) management of AEs.

The measurement of resource use required complementary approaches using data collected as part of the trial and as part of routine care. Trial participants’ use of health-care services was obtained from:

• Medication forms. All medication use from 3 months before randomisation was recorded by trial physicians at each trial visit and was supplemented by participant diary records.

• Baseline forms. Research nurses completed the relevant sections of the baseline forms to identify participant contact with hospital and health-care professionals in the 3 months before randomisation.

• Three-monthly patient questionnaires. Research nurses completed the relevant sections of the patient questionnaires during face-to-face interviews with trial participants to identify overnight hospital stays, number of nights, reason for admission and type of ward. The patient questionnaires were also used to identify contacts with health-care professionals including GPs, consultants, nurses, psychologists and rheumatologists, as well as the places and/or means of contact (i.e. A&E, outpatient, GP practice, home visits, telephone, text and e-mail).

• Electronic PLICS data and/or patient administration systems (PAS) data. These were accessed via the participating hospitals’ finance departments to identify inpatient stays, use of intensive care or high-dependency units, and outpatient visits.

• AE or SAE forms. Research nurses completed the AE and SAE forms when trial participants were admitted to hospital for events considered possibly related to the study drug, which included bronchopneumonia, herpes simplex infection, pharyngitis and pneumonia.

• Participant diaries recorded GP, social worker, district nurse and hospital visits.

In addition, for the estimation of long-term costs, longitudinal data for patients with JIA-associated and idiopathic uveitis were obtained from the Bristol Regional Tertiary Paediatric Uveitis clinic. This cohort provided data on the number and nature of surgeries performed from diagnosis with follow-up at 1, 3, 5 and 10 years.

Outcomes

The health outcome for the economic evaluation was the QALY, calculated from utilities measured from responses to the Health Utilities Index (HUI) questionnaire administered at baseline and at 3, 6, 9, 12 and 18 months. This was selected in preference to the EuroQol-5 Dimensions (EQ-5D) for its validity in paediatric populations. The HUI is a 40-item questionnaire that assesses health-related quality of life on eight single attributes: vision, hearing, speech, ambulation, dexterity, emotion, cognition and pain. Each HUI level code is estimated from responses to single questions or from a pattern of responses to a specific series of questions. Single attribute utility functions supply scores that express the morbidity for a person for each attribute. The health-related quality of life for a subject was determined by applying a multiattribute utility function to estimate HUI3 scores. 73

Modelled extrapolation

The Markov model was informed by a patient-level longitudinal data set of patients with idiopathic and JIA-associated uveitis. The only outcome recorded in both SYCAMORE and the Bristol data set74 (see Resource use and costs), and which is expected to directly affect patients’ health-related quality of life, was visual acuity, based on logMAR scores. To align with the model framework, the Bristol data74 were stratified as no visual impairment (VI) (logMAR < 0.3) and VI (logMAR ≥ 0.3). Health states were defined by patients’ vision in the worst eye as this was deemed the most clinically relevant. For trial participants, visual logMAR scores were recorded during all protocol-based visits and at unscheduled clinic visits. Data were ordered chronologically for each participant and time in each visual state (no VI, VI) was interpolated. The proportion of time in each state (no VI, VI) for each arm of the trial was used to determine the initial distribution of participants across the states of the Markov model (Figure 9). The cycle length was specified as 1 year, and a half-cycle correction applied.

FIGURE 9.

Markov model structure.

The Bristol cohort provided logMAR scores at diagnosis and at 1, 3, 5 and 10 years. This provided four time periods over which transitions among states may occur. Transitions between visual states, or to the same visual state, were defined as being either with or without surgery, which, along with transitions to the death state, resulted in 11 possible transitions. Annual transition probabilities were estimated by converting pooled transition probabilities over the 10-year time horizon into rates and back into transition probabilities.

The cost of surgery was taken as being the mean cost of a transition with surgery according to the longitudinal data set, as some transitions were associated with multiple surgery costs.

Published all-cause mortality data75 were adjusted for age but not sex because of the mixed cohort. A standardised mortality ratio of 3.9 (95% CI 0.8 to 11.3) for non-systematic JIA was applied. 76

Incremental analysis

Results

Outcomes

Visual acuity

Quality-adjusted life-year scores were calculated for both the complete-case and imputed analyses. Owing to missing data, QALY values were calculated only for three participants randomised to placebo and 25 participants randomised to adalimumab. In the complete-case analysis, QALY scores were lower in the adalimumab group [1.40 (95% CI 1.35 to 1.45)] than in the placebo group [1.45 (95% CI 1.41 to 1.5)], although the difference was not significant. After imputation, the QALY scores were higher for adalimumab than placebo [1.35 (95% CI 1.30 to 1.41) and 1.28 (95% CI 1.15 to 1.41), respectively] but, again, not significantly different.

Figure 10 illustrates the distribution of HUI3 level scores by treatment arm, attributes and time. Fewer participants in the placebo arm than in the adalimumab arm completed the HUI3; completion rates further reduced in both arms over the trial period. At baseline, 82% of participants in the adalimumab group and 72% of participants in the placebo group reported level 1 vision (no visual impairment); by 18 months, 76% of participants in the adalimumab group and 75% of participants in the placebo group were in the level 1 vision category. However, at 18 months, responses to this attribute were reported for only four participants in the placebo group, compared with 34 participants in the adalimumab group. For pain, at baseline, 49% (29/59) of participants in the adalimumab group and 58% (17/29) of participants in the placebo group reported level 1 pain score (no pain). At 18 months, 82% (27/33) of participants in the adalimumab group and 100% (4/4) of participants in the placebo group reported level 1 pain score, but, again, interpretation is hampered by low reporting rates at 18 months, especially in the placebo arm. No participants reported hearing-related problems for the duration of the trial.

FIGURE 10.

Distribution of participants’ responses to each HUI3 attribute, by treatment allocated and time. Levels range from 1 to 6, with 6 representing the most severe problem. The numbers of completed responses are reported by treatment arm. (a) Vision; (b) hearing; (c) speech; (d) ambulation; (e) dexterity; (f) emotions; (g) cognition; and (h) pain.

Visual acuity (logMAR) scores were available for 52 participants with complete-case data: 43 participants in the adalimumab group and nine participants in the placebo group. Baseline logMAR scores were complete and indicated the proportions of participants in the VI health state at baseline as 11.67% for adalimumab and 6.67% for placebo, which was anticipated to introduce bias in the ‘time in visual impairment’ outcome. At 18 months, 36 participants in the adalimumab arm of the trial and eight participants in the placebo arm had no time in the VI state. Mean proportion of time in the VI health state was 0.03 for participants in the adalimumab group and 0.02 for participants in the placebo group (difference 0.01, 95% CI –0.04 to 0.06) (see Table 36).

Following imputation, participants randomised to adalimumab spent 5.3% of time in VI during the 18-month analysis, whereas participants randomised to placebo spent 11.1% of time in VI. By including VI at baseline and time in VI, alongside demographics, costs and utility outcomes, the imputation model corrected for the imbalance in VI at baseline. In the imputed analyses, the rate of VI is higher in the placebo arm than in the adalimumab arm.

Data on logMAR from the Bristol cohort were available for 126, 117, 93, 75 and 22 patients at baseline and at 1, 3, 5 and 10 years, respectively. Based on the stratification of logMAR, 69 patients were recorded as having VI at one or more time point, whereas 88 patients were never recorded as having VI. There were 268 observed transitions over the four periods when transition was possible.

Analysis

Sensitivity analyses

Univariate sensitivity analysis

The results of the sensitivity analyses are presented in Table 38. These demonstrate that the model is most sensitive to adalimumab usage. The most cost-effective scenario is based on the assumption of adherence to MTX and adalimumab as recorded in the participant diaries over the 18-month time horizon (incremental cost of £117,514 per QALY gained) and the 18-month plus 10-year time horizon (incremental cost of £115,708 per QALY gained). However, participant diary recording of doses can be subject to unreliability and does not reflect the costs to the NHS of prescriptions issued. The least cost-effective analysis relates to adalimumab use being based on the number of vials issued, which was greater than the base-case assumption based on time in the study (110.55%), resulting in ICERs of £149,040 and £140,576 per QALY gained over 18 months and 18-month plus 10-year time horizons, respectively.

TABLE 38 - Results of the sensitivity analyses

H : L, high to low; L : H, low to high.

Based on a 7-year-old female (median demographic of SYCAMORE).

This is the observed percentage of participants randomised to the adalimumab arm who were administered adalimumab beyond 18 months.

Sensitivity analysis	Costs (£)			QALY			ICER (£)
	Treatment group		Incremental	Treatment group		Incremental
	Adalimumab	Placebo		Adalimumab	Placebo
Base casea	70,719	31,403	39,316	8.60	8.29	0.30	129,025
Time horizon of analysis
18 months
Adalimumab adherence based on vials issued	17,399	1734	15,666	1.36	1.25	0.11	149,040
Adalimumab adherence based on accountability logs	15,716	2095	13,621	1.36	1.25	0.11	129,587
Adalimumab and MTX adherence based on participant diaries	14,345	1993	12,352	1.36	1.25	0.11	117,514
18 months + 10 years
Adalimumab treatment for 18 months only	45,504	31,403	14,101	8.40	8.29	0.11	133,656
Adalimumab treatment for 18 months + 10 years	120,262	31,403	88,858	8.99	8.29	0.70	127,646
Adalimumab adherence based on vials issued	74,011	31,175	42,835	8.60	8.29	0.30	140,576
Adalimumab adherence based on accountability logs	68,800	31,536	37,264	8.60	8.29	0.30	122,291
Adalimumab and MTX adherence based on participant diaries	59,896	24,638	36,258	8.60	8.29	0.30	115,708
No discounting	77,634	36,743	40,621	9.88	9.57	0.32	128,886
23% of participants who were administered adalimumab beyond 18 monthsb	51,304	31,403	19,900	8.44	8.29	0.15	131,511
Adalimumab : placebo VI proportions H : L	70,864	31,147	39,716	8.60	8.29	0.30	130,586
Adalimumab : placebo VI proportions L : H	70,574	31,659	38,915	8.60	8.29	0.31	127,471

The analysis in which the duration of adalimumab treatment was increased resulted in a lower ICER (£127,646 per QALY gained). Conversely, a shorter duration of treatment of 18 months, compared with 3 years in the base case, reduces the incremental QALY gain and raises the ICER to £133,656 per QALY gained.

Probabilistic sensitivity analysis

Based on the results of the probabilistic sensitivity analysis applied to the base case, the mean total costs of the adalimumab arm were £70,951 [95% credible range (CR) £45,204 to £123,764]. The mean total costs of the placebo arm were £31,587 (95% CR £5308 to £83,320). Mean QALYs were 8.60 (95% CR 8.00 to 9.19) for the adalimumab arm and 8.29 (95% CR 7.42 to 9.17) for the placebo arm.

The mean incremental costs and QALYs for adalimumab were £39,364 (95% CR £24,728 to £58,235) and 0.31 (95% CR –0.04 to 0.66), respectively.

The cost-effectiveness plane (Figure 11) and the cost-effectiveness acceptability curve (Figure 12) indicate that adalimumab is highly unlikely to be cost-effective in the £20,000- to £30,000-per-QALY threshold range. The probability of being cost-effective at the £30,000-per-QALY threshold is < 1%. In 96% of simulations, adalimumab was both more costly and more effective; however, in 4% of simulations, adalimumab was seen to be less effective than placebo and remained more costly.

FIGURE 11.

Cost-effectiveness plane.

FIGURE 12.

Cost-effectiveness acceptability curve.

Implications for practice

This trial provides robust evidence regarding the use of adalimumab in the management of children and adolescents with JIA-associated uveitis refractory to MTX. The results show that adalimumab therapy in combination with MTX controlled inflammation and had a lower rate of treatment failure than placebo in this patient group. This finding, taken alongside the expected increased incidence of AEs in the adalimumab group and the economic analysis, can potentially inform current clinical practice with the aim of reducing uveitis-related ocular morbidity, including VI.

Recommendations for research

This trial answered some key questions for the management of JIA-associated uveitis refractory to MTX. However, it also identified a number of crucial next-step research challenges. These include the following:

1. defining the optimum time after control of uveitis to withdraw/stop adalimumab

2. defining the development and clinical utility of testing for antidrug antibodies in patients with waning of response to adalimumab

3. consideration of the option of increasing the frequency of administration of adalimumab for those with a suboptimal response

4. developing biomarkers for predicting early and complete response to adalimumab so that patients can be stratified more quickly to the appropriate treatment.

Overall conclusion

Adalimumab in combination with MTX is safe and effective in the management of JIA-associated uveitis. However, the likelihood of cost-effectiveness is < 1% at the £30,000-per-QALY threshold. Several limitations of the trial are noted. However, overall, this trial is, to our knowledge, the first randomised, double-blind, placebo-controlled, multicentre trial with an integrated economic evaluation comparing the efficacy, safety and cost-effectiveness of adalimumab in combination with MTX versus placebo with MTX alone, with regard to controlling disease activity in refractory uveitis associated with JIA. It also notes important future research priorities, including a future clinical trial to define the most effective time to stop therapy and to identify prognostic biomarkers of early and complete response.

Contributions of authors

Athimalaipet V Ramanan and Michael W Beresford were chief investigators of the trial and provided rheumatology expertise for the trial design and throughout the trial period, with input from Sandrine Compeyrot-Lacassagne and Patricia Woo.

Athimalaipet V Ramanan, Ashley P Jones, Dyfrig A Hughes and Michael W Beresford wrote early versions of the manuscript with significant input from all co-authors. The manuscript was finalised and revisions addressed by Athimalaipet V Ramanan and Michael W Beresford with all co-authors’ consent.

Andrew D Dick and Clive Edelsten provided ophthalmology support and expertise during the design of the trial and throughout the trial period.

The study was conceived by Athimalaipet V Ramanan and Michael W Beresford, and the protocol and design were by Athimalaipet V Ramanan, Clive Edelsten, Dyfrig A Hughes, Ashley P Jones, Ben Hardwick, Helen Hickey and Michael W Beresford with support from all co-authors.

The day-to-day management of the trial was undertaken by Ashley P Jones, Andrew McKay, Paula R Williamson, Ben Hardwick and Helen Hickey, with ongoing support from Athimalaipet V Ramanan, Michael W Beresford and all co-authors.

Data were gathered and analysed by Athimalaipet V Ramanan, Ashley P Jones, Dyfrig A Hughes, Andrew McKay, Anna Rosala-Hallas, Paula R Williamson, Naomi Rainford, Graeme Hickey, Ruwanthi Kolamunnage-Dona, Giovanna Culeddu, Catrin Plumpton, Eifiona Wood and Michael W Beresford, and who vouch for the data and the analyses.

The authors assume responsibility for the accuracy and completeness of the data and vouch for the fidelity of the trial to the protocol.

Publications

Ramanan AV, Dick AD, Jones AP, McKay A, Williamson PR, Compeyrot-Lacassagne S, et al. Adalimumab plus methotrexate for uveitis in juvenile idiopathic arthritis. N Eng J Med 2017;376:1637–46.

Hughes DA, Culeddu G, Plumpton CO, Wood E, Dick AD, Jones AP, et al. Cost-effectiveness analysis of adalimumab for the treatment of uveitis associated with juvenile idiopathic arthritis [published online ahead of print 15 October 2018]. Ophthalmology 2018.

Data-sharing statement

All data requests should be submitted to the corresponding author for consideration. Access to available anonymised data may be granted following review.

University Hospitals Bristol NHS Foundation Trust, as the sponsor of this trial, has a data-sharing agreement with AbbVie Inc. in support of regulatory purposes.

Patient data

This work uses data provided by patients and collected by the NHS as part of their care and support. Using patient data is vital to improve health and care for everyone. There is huge potential to make better use of information from people’s patient records, to understand more about disease, develop new treatments, monitor safety, and plan NHS services. Patient data should be kept safe and secure, to protect everyone’s privacy, and it’s important that there are safeguards to make sure that it is stored and used responsibly. Everyone should be able to find out about how patient data are used. #datasaveslives You can find out more about the background to this citation here: https://understandingpatientdata.org.uk/data-citation.

Disclaimers

This report presents independent research funded by the National Institute for Health Research (NIHR). The views and opinions expressed by authors in this publication are those of the authors and do not necessarily reflect those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care. If there are verbatim quotations included in this publication the views and opinions expressed by the interviewees are those of the interviewees and do not necessarily reflect those of the authors, those of the NHS, the NIHR, NETSCC, the HTA programme or the Department of Health and Social Care.

Trial Steering Committee

Independent Data and Safety Monitoring Committee

Professor John Sparrow (Chairperson), Consultant Ophthalmologist, Bristol Eye Hospital, Bristol, UK.

Professor Justine Smith, Consultant Ophthalmologist, School of Medicine, Adelaide, SA, Australia.

Dr Nico Wulffraat, Paediatric Rheumatologist, UMC Utrecht, Department of Paediatrics, Utrecht, Netherlands.

Professor Steff Lewis, Personal Chair in Medical Statistics, Public Health Sciences section, Centre for Population Health Sciences, The University of Edinburgh, Medical School, Edinburgh, UK.

Trial Management Group

Professor Athimalaipet Vaidyanathan Ramanan (Co-Chief Investigator), Consultant Paediatric Rheumatologist, Bristol Royal Hospital for Children & Royal National Hospital for Rheumatic Diseases, UK.

Professor Michael W Beresford (Co-Chief Investigator), Professor of Child Health, University of Liverpool, Liverpool, UK.

Diana Benton, Research & Innovation, University Hospitals Bristol NHS Foundation Trust, Bristol, UK.

Jessica Bisset, Research & Innovation, University Hospitals Bristol NHS Foundation Trust, Bristol, UK.

Professor Andrew Dick, Professor of Ophthalmology, Department of Clinical Sciences, University of Bristol, Bristol, UK.

Dr Clive Edelsten, Consultant Ophthalmologist, Great Ormond Street Hospital for Children, London, UK.

Mr Ben Hardwick, Trial Co-ordinator, CTRC, University of Liverpool, Liverpool, UK.

Ms Helen Hickey, Head of Trial Management, CTRC, University of Liverpool, Liverpool, UK.

Professor Dyfrig Hughes, Professor of Pharmaeconomics, Centre for Health Economics and Medicines Evaluation, Bangor University, Bangor, UK.

Debbie Janson, patient representative.

Dr Ashley P Jones, Senior Statistician, CTRC, University of Liverpool, Liverpool, UK.

Dr Sandrine Lacassagne, Paediatric Rheumatologist, Great Ormond Street Hospital for Children NHS Trust, London, UK.

Mr Andrew McKay, Trial Statistician, CTRC, University of Liverpool, Liverpool, UK.

Jennifer Oliver, Bristol Children’s Vaccine Centre, Bristol, UK.

Emma Stoica, Research & Innovation, University Hospitals Bristol NHS Foundation Trust, Bristol, UK.

Ms Giovanna Culeddu, Research Project Support Officer, Centre for Health Economics and Medicines Evaluation, Bangor University, Bangor, UK.

Professor Paula Williamson, Director, CTRC, University of Liverpool, Liverpool, UK.

Professor Patricia Woo, Emeritus Professor of Paediatric Rheumatology, University College London.