Rituximab compared to intravenous cyclophosphamide in adults with connective tissue disease-associated interstitial lung disease: the RECITAL RCT

Rituximab

Scleroderma-associated ILD remains a major cause of morbidity and mortality despite the widespread use of CP and MMF. 21 The same is true for MCTD and the idiopathic inflammatory myositidies where the lack of evidence-based therapies further complicates treatment choice. Rituximab (RT), a chimeric (human/mouse) monoclonal antibody with a high affinity for the CD20 surface antigen expressed on pre-B and B-lymphocytes, results in rapid depletion of B cells from the peripheral circulation for 6–9 months. 22,23 Evidence for the efficacy of B cell depletion exists in a number of immune-mediated conditions, including rheumatoid arthritis,24–26 antineutrophil cytoplasmic antibody (ANCA)-associated vasculitis,27,28 immune thrombocytopenic purpura (TTP),29 membranous nephropathy and pemphigus. In these conditions RT tends to be better tolerated than previously used immunosuppressant regimens and, in many cases, was associated with reductions in corticosteroid usage.

Several case series suggest RT may also be effective in ILD occurring in the context of immunological over-activity, with favourable responses reported in antisynthetase (ASS) associated ILD30 and SSc-ILD. 31,32 Our own experience has demonstrated RT to be an effective, potentially life-saving therapeutic intervention in the treatment of very severe, progressive CTD-ILD unresponsive to conventional immunosuppression. 33 In a retrospective analysis34 of 50 subjects with ILD due to a variety of aetiologies all of whom had failed multiple prior therapies we demonstrated that RT therapy was associated with an average improvement in FVC and stabilisation of DLco in the subsequent 6–12 months following treatment. In an open-label prospective study31 of 14 individuals with scleroderma ILD, 8 were randomised to receive two cycles (each cycle being 4, weekly doses of RT at a dose of 375 mg/m²) while the remaining 6 subjects received standard care. At 1 year FVC improved from 68.1 ± 19.7% predicted to 75.6 ± 19.7% predicted in the RT group compared with a drop from 86.0 ± 19.5% predicted to 81.7 ± 20.7% predicted in the standard of care group. Although these data suggest a benefit of RT it should be noted that in addition to being a small open-label study there were clear differences in baseline disease severity between the RT and the standard of care arms. Another recently published open-label trial35 compared RT (two times 1000 mg) to intravenous CP (500 mg/m²) in 60 subjects with scleroderma ILD with the primary outcome being change in FVC at 24 weeks. RT was associated with an improvement in FVC (from 61.3 ± 11.3 to 67.5 ± 13.6% predicted) while the CP arm saw a loss of FVC over 6 months (from 59.3 ± 13.0 to 58.1 ± 11.2% predicted). RT, but not CP, was associated with an improvement in skin score. There were fewer side effects in the RT arm. The major limitations of the study were the open-label design and the exclusion of subjects aged over 60.

Potential risks of rituximab

Rituximab was first approved for clinical use in 1997 for the treatment of relapsed or refractory CD20 positive B-cell low grade or follicular non-Hodgkin’s lymphoma. 36 Since then, it has been widely used as a treatment for a broad range of conditions including rheumatoid arthritis, granulomatosis with angiitis, idiopathic thrombocytopaenia purpura, pemphigus vulgaris and myasthenia gravis. The safety profile of RT is therefore well-established. Common side effects include rash, itchiness, hypotension and dyspnoea at, or close to, the time of infusion and are usually associated with hypersensitivity to the drug (this is more likely in those who have had occupational or domestic exposure to mice). RT is also associated with an increased risk of infection. Severe but very rare side effects include reactivation of hepatitis B, progressive multifocal leucoencephalopathy (PML) and toxic epidermal necrolysis syndrome. The risk of hepatitis B can be mitigated by pre-treatment testing. The potential adverse effects of RT are comparable, or even less frequent, than those associated with other commonly used treatments for CTD-ILD, especially CP.

Rationale for primary outcome

The primary outcome for this study was rate of change in FVC over 24 weeks. FVC provides a measure of lung volume. With worsening pulmonary fibrosis FVC tends to diminish. In trials of patients with a range of fibrotic ILDs (including IPF, scleroderma ILD and progressive fibrotic ILD of any cause) change in FVC has been accepted by the Federal Drugs Administration (FDA) in the USA and the European Medicines Agency as a surrogate for survival. 37 As such, FVC has been the regulatory end point of choice for the approval of medications targeting ILD. Additionally, in scleroderma ILD both baseline FVC and change in FVC over time have been linked to longer-term prognosis and survival. 33,38 Twenty-four weeks was chosen for the primary end point as this was felt to be the time-point at which the effect of RT and CP alone could best be measured. Beyond 24 weeks subjects were permitted to receive other immunosuppressants according to the recommendation of their treating clinician (and as tends to happen in routine clinical practice). FVC change at 48 weeks was captured as a secondary end point with the prior recognition that this might also be influenced by treatments received beyond 24 weeks.

Rationale for biomarkers

There are currently no available biomarkers for assessing response to therapy or risk of disease progression in CTD-ILD. By closely studying patients in each treatment arm and undertaking exploratory biomarker analysis it was hoped that we might identify potential disease and therapy-specific biomarkers for future development and use in clinical practice. Serial samples of whole blood [for ribonucleic acid (RNA) and DNA analysis], serum and plasma were collected to explore changes in the levels of a range of candidate biomarkers. The choice of biomarkers was driven by insights derived from other fibrotic and inflammatory ILDs and included measures of extra cellular matrix turnover (collagen neoepitopes, matrix metalloprotease-7),39 epithelial damage [cancer antigen 125 (CA125), cytokeratin 19 fragment, Krebs von den lungen-6]40 and inflammatory cell activation (IL-6, monocyte chemotractant protein-1, Chitinase-3-like protein-1, CC motif chemokine ligand 18). 40 The aim of this part of the study was to identify a candidate biomarker panel for use in future studies and for development as a clinical tool for guiding therapeutic decision making.

Primary objective

• To demonstrate that intravenous RT has superior efficacy compared to current best treatment (intravenous CP) for CTD-ILD as measured by change in FVC at 24 weeks.

Secondary objectives

• To compare the safety profile of RT to intravenous CP in individuals with CTD-ILD.

• To assess the health economic benefits of RT compared to current standard of care for CTD-ILD – including measurements of healthcare utilisation, quality of life (QoL) and carer burden.

• To evaluate a range of exploratory biomarkers for disease severity, prognosis and treatment response in CTD-ILD.

Study setting

RECITAL study was conducted at 11 UK sites, all of which were accredited by NHS England as specialist ILD or Rheumatology centres. The investigators were qualified by education, training and experience to assume responsibility for the proper conduct of the trial and provided evidence of such qualifications through up-to-date curriculum vitae and GCP training certificates. The local principal investigator (PI), who could be either a respiratory physician or rheumatologist, was responsible for the proper conduct of the study, for the safety and welfare of study participants and being thoroughly familiar with the appropriate use of the investigational products. They were responsible for ensuring that all persons supporting the trial were appropriately qualified and trained, adequately informed about the protocol, the investigational product(s), and their trial-related duties and functions. Both the Sponsor and local trust legal representatives signed the site agreements. GCP training was required for all staff responsible for trial activities. The frequency of repeat training was dictated by the requirements of their employing institution, or two yearly where the institution had no policy. The PI or delegate was required to document and explain any deviation from the approved protocol and to communicate this to the Sponsor.

Participant inclusion criteria

• A diagnosis of CTD, based on internationally accepted criteria, in one of the following categories:41–44

  • systemic sclerosis

  • idiopathic interstitial myopathy (including polymyositis/dermatomyositis)

  • MCTD.

• Severe and/or progressive ILD associated with the underlying CTD.

• Chest high-resolution computer tomography performed within 12 months of randomisation.

• Intention of the caring physician to treat the ILD with intravenous CP (with treatment indications including; deteriorating symptoms attributable to ILD, deteriorating lung function tests, worsening gas exchange or extent of ILD at first presentation) and where there is a reasonable expectation that immunosuppressive treatment will stabilise or improve CTD-ILD. In individuals with scleroderma it is anticipated that patients will fulfil the criteria for extensive disease defined by Goh et al. 33

• Written informed consent.

Participant exclusion criteria

• Age <18 or >80 years.

• Previous treatment with RT and/or intravenous CP.

• Known hypersensitivity to RT or CP or their components.

• Significant (in the opinion of the investigator) other organ comorbidity including cardiac, hepatic or renal impairment.

• Coexistent obstructive pulmonary disease (e.g. asthma, chronic obstructive pulmonary disease (COPD), emphysema) with pre-bronchodilator FEV₁/FVC < 70%.

• Patients at significant risk for infectious complications following immunosuppression including human immunodeficiency virus (HIV) positive or other immunodeficiency syndromes such as hypogammaglobulinaemia.

• Suspected or proven untreated tuberculosis.

• Viral hepatitis.

• Infection requiring antibiotic treatment in the preceding 4 weeks.

• Unexplained neurological symptoms [which may be suggestive of progressive multifocal leucoencephalopathy (PML)]. Neurological symptoms arising as a consequence of the underlying CTD do not necessitate exclusion.

• Other investigational therapy (participation in research trial) received within 8 weeks of randomisation.

• Immunosuppressive or CTD disease-modifying therapy (other than corticosteroids) received within 2 weeks of the first intravenous treatment.

• Pregnant or breastfeeding women, or women of child-bearing potential, not using a reliable contraceptive method for up to 12 months following investigational medicinal product (IMP).

• Unexplained haematuria or previous bladder carcinoma.

• CT scan > 12 months from randomisation.

• Unable to provide informed written consent.

Drug preparation and supply

Bath ASU was responsible for manufacture of the IMPs and matched-placebo to standards compliant with good medical practice (GMP) requirements.

The IMPs used in this study were infusions of either CP, RT or matching placebo. All products were compounded using licensed starting materials as per the relevant summary of product characteristics (SmPC). All active drugs were prepared by the addition of drug solution to a 250 ml NaCl 0.9% Freeflex^® infusion bag following the withdrawal of a comparable volume of solution to ensure the final bag volume remained at 250 ml. The matching placebo was a suitability labelled 250 ml NaCl 0.9% Freeflex bag.

Starting materials: RT (Roche EU/1/98/067/001 EU/1/98/067/002).

Cyclophosphamide (Baxter PL 001 16/0388).

Sodium chloride infusion (Fresenius Kabi PL 08828/0084).

Following manufacturing, the fully reconstituted study IMP was transported, in a temperature-monitored environment (by Movianto UK limited) to study sites. Each infusion bag arrived together with the qualified person (QP) batch release certificate. A copy of the QP release was filed in the site pharmacy file along with the shipping paperwork. Temperature monitoring during transit occurred via a temperature monitoring device within the shipment. Any identified temperature excursions during transit to site was reported by pharmacy to the Trial Manager, using the Temperature Excursion Report Form. Site pharmacies documented receipt of IMP using the IMP accountability Log.

Randomisation allocation was performed using an Interactive Web-based Randomisation System (InForm). Randomised patients were given a study specific 24 hours emergency contact immediately following randomisation. The card included: study title, details of the IMPs, patient trial number, chief investigator (CI)/PI’s contact details along with out-of-hours contact details in case of emergency.

A patient was withdrawn if their treatment code was unblinded by the investigator or treating physician. The primary reason for discontinuation (the event or condition which led to the unblinding) was recorded in the electronic case report form.

Dispensing procedure

It was the responsibility of the site pharmacy to dispense IMP in accordance with the study protocol and against a prescription provided by the medical team. The study-specific prescription and accountability logs were used by all sites.

Initial infusion

Site pharmacies documented the dispensing details on the prescription form. Details were also recorded on a subject-specific IMP accountability log (one log used per patient). Entries on the accountability log were signed by the dispenser and checked by a second member of staff. The IMP accountability log was updated with the date of dispensing.

Ongoing infusions

A new prescription was completed for each dose required for each subject recruited. Site pharmacies checked the details on the prescription against the corresponding subject-specific IMP accountability log to ensure the correct patient number and visit number. Site pharmacies completed the next entry on the accountability log for each ongoing prescription received. The IMP accountability log was updated with the date of dispensing.

Blinding

The study was double-blind. Medication was identified by a unique trial identifier (randomisation code). Patients and investigators remained blinded until after database lock and only the DMC had access to unblinded information.

The investigator was encouraged to maintain the blind as far as possible. Investigators were able to unblind individual patient’s treatment via the eCRF in the case of an emergency, when knowledge of the study treatment was essential for the appropriate clinical management or welfare of the patient. The PI was asked to provide a reason for unblinding and was required to enter their password as confirmation. The PI was then shown which treatment arm the patient was allocated to. The investigator referred to the unblinding instructions stored securely in the investigator site file (ISF). An email notification stating that a patient was unblinded was sent out to local investigator, pharmacy and the Trial Manager.

A decision to unblind was made in three cases prior to week 24 where information of the treatment allocation was deemed essential (to allow the treatment of disease progression for two patients and because of a SAE in one case). There were five other instances where patients were unblinded; this occurred after they had received all doses of IMP, either during the follow-up period or following study completion and was performed to enable appropriate choice of therapy for ongoing disease worsening. All details of the eight instances of unblinding were recorded in the eCRF, the ISF/trial master file (TMF) (or pharmacy file when manual unblinding was used).

Backup manual unblinding

When the investigator needed to unblind the treatment for the patient in case of electronic failure of the database (InForm), a manual backup system was also provided. A randomisation list was provided to the Sponsor pharmacy department. During normal hours the investigator was able to contact the Sponsor pharmacy department via switchboard on 020 7352 8121. Outside of normal hours, the Sponsor pharmacy on-call pharmacist could be contacted via the same number.

Storage and administration

The investigational products were stored securely in a temperature-controlled unit between 2 and 8°C within a pharmacy department. Temperature logs were kept and made available for monitoring and audit purposes. Specific instructions for storage and administration were provided in a study-specific IMP handling manual which was provided in the ISF and pharmacy file.

Accountability procedures for the investigational medicinal product(s)

The site hospital pharmacy department was responsible for maintaining and updating the study pharmacy IMP Accountability Log, filed in the hospital pharmacy file. A study prescription form documented the dispensing of IMP to the research nurse or doctor; this was retained in the pharmacy file. The research nurse or doctor was responsible for completing an IMP Administration Record Form at the point of administration; the completed form was filed in the ISF.

Any IMP dispensed to the research nurse or doctor that was not administered, for example due to cancellation, was returned to pharmacy for destruction. All unused IMP, including unused returns, were destroyed by the site pharmacy in accordance with local pharmacy practice, once agreed by the local PI and Sponsor. Destruction of unused IMP was documented on the IMP Destruction Form and filed in the hospital pharmacy file.

All used or partially used IMP was discarded in accordance with local cytotoxic waste disposal procedures; no used IMP was returned to pharmacy. A record of discarded used IMP was made on the IMP Administration Form and filed in the ISF.

Prohibited concomitant medication

• Pre-existing immunosuppression (including azathioprine, MMF, methotrexate, and ciclosporin) was stopped, following signed informed consent and during the screening period, at least 14 days prior to the first intravenous treatment. Hydroxychloroquine was on the list of prohibited medications until the last amendment of the study protocol. Following identification of a number of protocol deviations, due to administration of hydroxychloroquine, the CI confirmed that inclusion of hydroxychloroquine on the proscribed list was an oversight by the protocol design team. The topic was discussed by the DMC who agreed that hydroxychloroquine posed no risks to patients enrolled in the study and receiving IMP. Consequently, hydroxychloroquine was removed from the list of withheld concomitant medications with study amendment 13, approved in April 2018.

• Between weeks 0 and 24, patients were not permitted to receive additional immunosuppression (including oral agents, intravenous immunoglobulins or other monoclonal antibody therapies) other than corticosteroids.

Primary outcome

The primary outcome for the study was change in FVC measured in ml over 24 weeks. Spirometry was measured at baseline and at every subsequent visit (weeks 2, 4, 8, 12, 16, 20 and 24) according to the standards outlined in the guidelines from the American Thoracic Society (ATS) and European Respiratory Society. 45 Sites were encouraged to undertake spirometry at the same time (plus or minus 1 hour) at each visit to minimise the effects of diurnal variation on FVC measurements. The Global Lung Function Initiative (GLI) predictive equations were used to calculate the predicted normal and percentage predicted values for FVC. 46

Twenty-four weeks was chosen as the time point for the primary outcome measure on the basis that this was the point in time at which the effect of the investigational drugs alone could best be determined. After 24 weeks, subjects were permitted to start other immunosuppressant agents and thus interpretation of later time points is confounded by this variable use of immunosuppressants other than CP or RT.

Secondary outcomes

A range of physiological, QoL and biomarker measures were assessed as secondary end points for the study.

Questionnaires

Physiological measures

Safety

Vital signs (temperature, weight, pulse, blood pressure, oxygen saturations, oxygen status, respiratory rate) and the findings from physical examination were recorded at each study visit. Blood samples for laboratory tests (full blood count, urea, electrolytes, liver function tests) and urine for urinalysis were collected at every infusion visit (baseline and weeks 2, 4, 8, 12, 16 and 20)

Statistical analysis

A statistical analysis plan (SAP) was produced and agreed with the TSC and DMC prior to analysis (see the project webpage www.fundingawards.nihr.ac.uk/award/11/116/03). Deviations from the SAP were recorded in the SAP deviation document. A summary of the planned analysis strategy for the primary and secondary outcomes can be found below.

Primary efficacy analysis

• Analysis of the primary outcome was by modified intention to treat. In other words, data were included in respect of all subjects who met all the entry criteria for the trial and had been randomised and received at least one dose of study drug. The data was analysed according to the initial randomisation groups with no changes made in respect of subsequent withdrawals or crossovers.

• The hypothesis to be tested was that RT is superior to CP. The study was to be considered positive if statistical significance at the level of 0.05 (two-tailed) was achieved.

• To test the hypothesis above and estimate the difference in FVC at week 24 (together with 95% confidence interval), a three-level hierarchical (mixed/multilevel) model with unstructured correlation matrix and adjusted for baseline FVC and CTD type (stratification factor) treating the primary outcome as an interaction term between treatment and visits, including at week 24 was used.

Representing FVC in mL as FVC_iv (in ml) for patient I at visit v with treatment (RT or CP) t(i), the FVC_iv is modelled as the sum of four components:

• baseline FVC: represents the estimated FVC at baseline. This term comprised an individual level random effect which was drawn from a distribution parameterised using the associated centre-level random effect. Hence unexplained variations in the FVC measures were split into three components corresponding to the three levels of the model, that is the variation attributable to the centre (between centre variation) and the individual (between individual variation), as well as the residual variation (within individual variation).

• CTD diagnosis stratum (categorical) used for randomisation was added as a covariate.

• visit: represents the visit number as a categorical variable.

• treatment: a binary variable representing treatment arm.

• Interaction term: this represents a coefficient that captures the changes in FVC at each visit between the arms. The magnitude of change at 24 weeks and its 95% confidence interval will be calculated to answer the research question.

• No other baseline covariates were added as no substantial imbalances were revealed between the arms.

The Stata code used for the primary outcome was:

mixed changeFVC i.CTD i.week i.Arm#i.week || SITENAME: || SUBJECTNUMBERSTR:

Secondary efficacy analysis

• Analysis of secondary efficacy outcomes was also performed by modified intention to treat.

• Change in continuous physiological variables between baseline and 48 weeks were assessed by similar multilevel model as described for the primary outcome.

• Categorical change in physiological variables was measured using chi-squared tests under the null hypothesis of no difference between the treatment groups.

Mortality, treatment failure and progression-free survival were measured using Kaplan–Meier estimates. A log-rank test was used to compare treatment groups and a Cox proportional hazards model was used to determine hazard ratios (HRs) for survival analyses.

Sample size

Previous studies of intravenous CP in SSc demonstrated a 1% decline in FVC at 12 months, with a coefficient of variation of 7.8%. Our observational data and a previous non-randomised study of RT (used as rescue therapy in those failing treatment with CP) suggested improvements in FVC at 6–12 months of between 9.5% and 20% compared to baseline. Using 1 : 1 randomisation, a sample size of 52 patients in each group gives a 90% power to detect a 5% difference (approximately 140 ml) between groups at 24 weeks in the change in FVC (as measured in ml) with a significance level (alpha) of 0.05 (two-tailed). Anticipating a dropout rate of 10% our target recruitment was 58 patients in each arm of the study. On the basis of data derived in other interstitial lung diseases 5% change in FVC is associated with change in long-term prognosis and can therefore be considered a clinically meaningful difference between the two groups. Given the number of individuals treated with CP at units participating in the study, this number was considered feasible to deliver within the planned trial timelines.

Primary efficacy variable

• Change in FVC (expressed in ml) at week 24.

Secondary efficacy variables

• Change in DLco at 24 weeks.

• Change in health-related QoL scores (SGRQ, SF-36, K-BILD).

• Change in Global Disease Activity Score (GDAS).

• Change in 6MWD over 48 weeks.

• Change in FVC and DLco at 48 weeks.

• Absolute categorical change of %FVC at 24 and 48 weeks (decrease by >5%, increase by >5% and change within <5%).

• Absolute categorical change of %FVC at 24 and 48 weeks (decrease by >10%, increase by >10% and change within <10%).

• 48-week change in FVC.

• Disease-related mortality (adjudicated by steering committee at close of study).

• Overall survival.

• Progression-free survival [i.e. avoiding any of the following: mortality, transplant, treatment failure (see below) or decline in FVC > 10% compared to baseline].

• Treatment failure (as determined by need for transplant or rescue therapy with either open-label CP or RT at any point until 48 weeks).

• Total corticosteroid requirement over 48 weeks.

• Change from baseline in SpO₂ at 24 and 48 weeks.

• Healthcare utilisation during study period (visits to primary care, unscheduled hospital visits, emergency admissions).

• Scleroderma-specific end points (change in scleroderma HAQ, mRSS).

Safety variables

• Vital signs (temperature, weight, pulse, blood pressure, oxygen saturations, oxygen status, respiratory rate).

• Physical examination (skin, lungs, cardiovascular, abdomen).

• Laboratory tests (blood counts, urea, electrolytes, liver function tests, urinalysis).

• Adverse and SAEs.

• Discontinuation of RT or CP due to intolerance or side effects.

Exploratory variables

• Change in lymphocyte subsets in relation to efficacy outcomes.

• Change in plasma protein levels following therapy and in relation to markers of disease activity (FVC, DLco, QoL, global disease scores).

• Outcome in relation to underlying CTD.

Cost-effectiveness analysis

A cost-effectiveness analysis was conducted to estimate the costs and outcomes associated with using RT versus CP to treat patients with CTD-ILD in the RECITAL study. Data on health resources used, such as primary care service, hospitalisation, outpatient, oxygen, their frequencies and durations were collected from patient diaries as reported during the RECITAL clinical trial. The average frequencies of resources used were calculated by dividing the number of visits by the number of patients who used the service, as reported in the RECITAL diary. The proportion of patients who used a specific healthcare service was calculated in an intention-to-treat analysis, dividing the number of patients who used a particular healthcare service by the total number of patients randomised for each treatment arm.

We calculated the incremental cost-effectiveness ratio (ICER) to compare the costs and effectiveness of RT versus CP. This was calculated by the difference in costs divided by the difference in effects of both technologies. A willingness to pay (WTP) threshold of £30,000 per quality-adjusted life-year (QALY) gained was assumed in this analysis (based on the commonly accepted thresholds used by the UK NICE). A discount rate on the cost and effects was not applied due to the temporal framework as our analysis did not exceed 1 year. The uncertainties around the input parameters were assessed by carrying out probabilistic sensitivity analysis (PSA), incremental net monetary benefit (INMB) and cost-effectiveness acceptability curve (CEAC).

Model input parameters

Model assumptions

In the formulation of the key assumptions, we made conservative estimates to avoid favouring intervention. Key assumptions of the model were: we assumed only grade 3 and 4 AEs (moderate and severe) and all hospitalisation required to treat AEs were non-elective short stay (day-case) to account for the total cost of the AE; we included the main treatment options reported in the RECITAL diary as alternatives to treat the reported AEs; we assumed that AEs occurred once; CP dosage was calculated assuming an idealised patient with a weight of 70 kg; one-way sensitivity analysis was carried out assuming either a 10% increase or decrease in input parameters. We examined the main assumptions in a sensitivity analysis.

Sensitivity analysis

The robustness of the outcome was assessed by sensitivity analyses. One-way sensitivity analyses were performed based on the parameters with the highest level of uncertainty (including all costs and utility values). These analyses consisted of varying each key parameter, assuming either a 10% increase or decrease in input parameters. Besides that, the impact of the uncertainties around the key parameters on the ICER was assessed through 1000 PSA using Monte Carlo simulation. PSA was carried out with all parameters being simultaneously varied in pre-specified ranges using pre-set distributions (gamma for cost and beta for utility values) (see Table 4).

Trial oversight committees

An independent DMC undertook interim review of the trial’s progress including updated figures on recruitment, data quality and main outcomes and safety data. More specifically the DMC:

• assessed data quality, including completeness (and by doing so encouraged collection of high-quality data)

• monitored recruitment figures and losses to follow-up

• monitored compliance with the protocol by participants and investigators

• monitored evidence for treatment differences in the main efficacy outcome measures

• monitored evidence for treatment harm (e.g. toxicity data, SAEs, deaths)

• decided whether to recommend that the trial continues to recruit participants or whether recruitment should be terminated either for everyone or for some treatment groups

• suggested additional data analyses for the monitoring purposes of DMC

• advised on protocol modifications suggested by investigators or the Sponsor (e.g. to inclusion criteria, trial end points or sample size)

• monitored compliance with previous DMC recommendations

• considered the ethical implications of any recommendations made by the DMC

• assessed the impact and relevance of external evidence.

The committee met by teleconference every 6 months or at least annually for the duration of the study and comprised:

• Dr Andrew Wilson – Chair – University of East Anglia

• Dr Clive Kelly – Queen Elizabeth Hospital, Gateshead (to May 2019 – retired)

• Dr Harsha Gunawardena – North Bristol NHS Trust, Southmead

• Dr Ashley Jones – Centre for Medical Statistics and Health Evaluation, University of Liverpool.

The TSC provided overall supervision for the trial and considered reports from the TMG, the trial team and the DMC. The roles and responsibilities were set out in the TSC Charter.

Independent members

• Dr Tim Harrison – Chair, University of Nottingham.

• Dr Bridget Griffiths – Freeman Hospital, Newcastle.

• Dr Nik Hirani – University of Edinburgh.

• Mrs Kim Fliglestone – Consumer representative.

• Mrs Alexandra Marler – Consumer representative.

Non-independent members

• Mrs Ira Jakupovic – Royal Brompton and Harefield NHS Foundation Trust.

• Mrs Vicky Taylor – Efficacy and Mechanism Evaluation (EME) (until 15 September 2015).

• Ms Maggie Shergill – EME (15 September 2015–07 September 2018).

• Anna Williams – EME (7 March 2018 until the end of the study).

• Dr Rachel Hoyles – Oxford Radcliffe Hospitals NHS Trust.

• Dr Helen Parfrey – Cambridge University Hospitals NHS Foundation Trust.

• Prof Deborah Ashby – Imperial College London.

All TSC and DMC members were required to complete the TSC/DMC Charter and declare any potential competing interests.

There were four substantial and one non-substantial protocol amendments, which are summarised in Table 5.

Breaches and protocol deviations

There were no serious breaches reported in this trial. Protocol deviations were recorded on the electronic Protocol deviation log in the database and were included in TMG, DMC and TSC reports.

A cumulative summary of all the protocol deviations recorded in the database is given in Report Supplementary Material 1, Table S1. The protocol deviations were included and discussed in DMC reports.

Screening and recruitment

Screening for participants started in December 2014 and ended February 2020. The last visit of the last patient was in January 2021. A total of 11 sites as seen in Table 6 screened participants and Royal Brompton Hospital was the largest recruitment site. A graph of recruitment against projected recruitment is given in Figure 2. The graph demonstrates that the study had a slower than anticipated recruitment rate and this was determined mainly by issues related to excess treatment costs for RT (these proved to be a major problem with respect to initiating centres and delivering the study on time and target). Another factor that increasingly affected the recruitment rate towards the end of the study was the availability of a generic alternative to RT, which cost significantly less and meant that some eligible patients received treatment with RT outside the setting of the trial thus reducing the pool of patients available to be recruited.

FIGURE 2.

Graph of projected and actual recruitment to RECITAL.

In response to COVID-19 pandemic the trial was kept open, however, patients active in the trial at the time were reluctant to attend study or follow-up visits. Some of these visits were conducted over telephone or video calls. Given these factors, and especially the unprecedented impact of the COVID-19 pandemic on trial recruitment and conduct, it was decided that any further recruitment would be almost impossible and would unacceptably extend trial timelines. Recruitment was therefore halted in April 2020 (with the agreement of the Sponsor, Data safety monitoring committee, TSC and funder), however, the study continued until the final patient completed their week 48 visit.

Participant flow

The Consolidated Standard of Reporting Trials (CONSORT) flow chart for RECITAL is detailed in Figure 3. In summary, a total of 145 patients were assessed for eligibility. Of these 104 were recruited and 101 randomised. Of the 101 randomised patients, 50 were randomised to CP and 51 were randomised to RT. Of these, four did not receive at least one treatment for the reasons given in the withdrawals table and Consort diagram. The number of patients who therefore received at least one treatment and were part of the modified intention to treat statistical analysis was 48 in the CP arm and 49 in the RT arm.

FIGURE 3.

The CONSORT flow chart.

Primary outcome

Of the 101 randomised subjects, 49 in the RT and 48 in the CP arm received at least one dose of therapy and were included in the modified intention-to-treat analysis. At 24 weeks, both RT and CP resulted in an improvement in FVC from baseline (97 ± 234 ml and 99 ± 329 ml, respectively) (see Table 10 and Figure 4). In the primary end-point analysis, applying an adjusted mixed-effects model corrected for diagnosis and baseline FVC the difference in FVC between RT and CP at week 24 was −40 ml (95% CI −153 to 74 ml) p = 0.493. Table 11 and Figure 5 show the adjusted difference between groups at each time point.

FIGURE 4.

Observed change in FVC from baseline over time. Bars represent the standard deviation.

FIGURE 5.

Adjusted model of change in FVC (and 95% CI) at each visit.

The difference between groups persisted in an unadjusted model and in adjusted and unadjusted fixed-effects models (see Report Supplementary Material 1, Tables S2, S3 and S4). Visual inspection of the FVC data disclosed one extreme outlying FVC value for a single subject (who received CP) at week 4 (see Report Supplementary Material 1, Figure S1). A sensitivity analysis was performed to confirm that this value was not impacting interpretation of the primary end point (see Report Supplementary Material 1, Table S5).

Primary outcome by connective tissue disease subtype

Randomisation between treatment arms was stratified by underlying CTD diagnosis. In a pre-planned analysis response to treatment was assessed in each of the three CTD subgroups. The effect of both CP and RT was consistent across all three groups with the largest magnitude of benefit being seen in subjects with IIM and the least benefit being seen in the scleroderma subgroup (see Figure 6).

FIGURE 6.

Change in FVC (and 95% CI) by CTD subgroup at each visit.

Secondary outcomes

Quality of life outcomes

Corticosteroid exposure

The hydrocortisone equivalent of each steroid therapy administered to subjects (whether oral, intravenous, inhaled or topical) during the trial was calculated on blinded data by the study team and confirmed by the study CI. When the steroid treatment started before the trial only the portion of therapy taken after day 0 (start of study medication) was included in the calculations. The daily dosage per day per patient was calculated for each day for the whole study group and for individual CTD subtypes (see Figures 7–10). For the calculation the denominator was not the number of patients randomised but the number of patients currently followed (subjects were censored at death or following withdrawal from the trial). The mean follow-up was 311.7 (±94.7) days from day 0. The mean per-subject total steroid exposure during the study was 13,291 (±14,657) mg/hydrocortisone in the CP and 11,469 (±10,041) mg/hydrocortisone in the RT group. The daily mean dose per patient was 42.89 (±47.30) mg/hydrocortisone/day in the CP and 37.61 (± 32.93) mg/hydrocortisone/day in the RT group.

FIGURE 7.

Daily steroid exposure across whole study group.

FIGURE 8.

Daily steroid exposure in the idiopathic inflammatory myopathy subgroup.

FIGURE 9.

Daily steroid exposure in the systemic sclerosis subgroup.

FIGURE 10.

Daily steroid exposure in the MCTD subgroup.

Survival analyses

During the course of the study there were five deaths with a sixth death occurring shortly after a week 48 visit and therefore being captured within the trial database (see Appendix 1, Table 34 for details). All were deemed to be due to complications of either CTD or interstitial lung disease. Three (plus the additional late death) occurred in subjects receiving RT and two in subjects receiving CP. There was no difference between groups in time to death as assessed by an adjusted Cox proportional hazards model [HR 1.72 (95% CI 0.311 to –9.56, p = 0.534)] (see Figure 11 and Table 26). The rates of progression-free survival (see Figure 12 and Table 27) [HR 1.11 (95% CI 0.625 to 1.99, p = 0.715)], and time to treatment failure (see Figure 13 and Table 28), [HR 1.25 (95% CI 0.34 to 4.65, p = 0.742)] did not differ between treatment arms.

FIGURE 11.

Overall survival.

TABLE 26 - Results of the overall survival model

	Haz. ratio	Std. err.	z	p-value	[95% CI]
Overall survival	1.723	1.507	0.622	0.534	0.311	9.563

FIGURE 12.

Progression-free survival.

TABLE 27 - Results of the progression-free survival model

	Haz. ratio	Std. err.	z	p-value	[95% CI]
Progression-free survival	1.114	0.329	0.365	0.715	0.625	1.987

FIGURE 13.

Time to treatment failure.

TABLE 28 - Results of the time to treatment failure model

	Haz. ratio	Std. err.	z	p-value	[95% CI]
Time to treatment failure	1.247	0.837	0.329	0.742	0.335	4.645

Cost-effectiveness analysis

At the end of 12 months, RT (cost: £93,227.14; QALYs: 0.330) was a dominant strategy over CP (cost: £94,338.13; QALYs: 0.308) for treating patients with CTD-ILD, generating a cost savings of £1110.99; and an increase in QoL of 0.022 QALYs per patient (see Table 29). RT treatment provides an incremental monetary benefit compared to CP.

Probabilistic sensitivity analysis

Using 1000 Monte Carlo simulations, PSA revealed that at a WTP of £30,000, RT treatment is dominant over CP to treat patients with CDT-ILD. Although the scatter plot shows some uncertainty around the outcomes, most of ICERs of RT versus CP are within quadrant 2. Therefore, they are less costly and more effective, showing themselves to be dominant (see Figure 14). The CEAC (see Figure 15) indicates the probability of the intervention being cost-effective when compared to the alternatives, according to the different thresholds or values per QALYs. RT proved to be cost-effective in 81.4% of the simulations and the CP in 18.6% of the simulations, assuming a £30,000 WTP limit. Thus, there is a strong probability of RT being the cost-effective therapy.

FIGURE 14.

Scatter plot of ICER of RT vs. CP.

FIGURE 15.

Cost-effectiveness acceptability curve of RT vs. CP.

One-way sensitivity analysis

The one-way sensitivity analysis showed that the three main parameters that most influence the ICER values are the outpatient costs and the treatment of AEs in the CP group, alongside the RT cost. RT only becomes not cost-effective relative to CP if the price increases by over 5%, to £366.71 (see Figure 16).

FIGURE 16.

One-way sensitivity analyses of significant parameters that impact cost-effectiveness – Tornado diagram (ICER).

Safety

Laboratory safety results

Laboratory blood results were flagged up if they fell outside pre-specified values (see Table 33). For listings of patients with measurements outside cut-off values (see Report Supplementary Material 1, Tables S18, S19, S20, S21, S22 and S23). A new, transient leucopoenia was seen in only one patient receiving CP at week 24 and this subsequently resolved. All other documented episodes of leucopoenia predated the initiation of therapy. Similarly, the only documented episode of neutropenia was present at baseline in a single individual receiving RT. Similarly, there were no sustained elevations of either creatinine or liver enzymes in either treatment arm and no changes in blood parameters which were deemed clinically significant.

Rituximab

To our knowledge there have not been any previous randomised controlled trials of RT in patients specifically with CTD-ILD. The majority of prior evidence has been generated in retrospective cohorts and open-label studies. In a retrospective analysis of 50 individuals (33 of whom had CTD-ILD) treated with RT as rescue therapy following failure of other treatments including CP, Keir et al. 34 reported a median improvement in FVC between 6 and 12 months of 6.7% from pre-treatment baseline and stabilisation (0% change) in DLco. This contrasted with median declines in FVC and DLco of 14.3% and 18.8%, respectively in the 6–12 months prior to treatment. On average, these patients had much more severe disease than the subjects enrolled in RECITAL with a mean baseline FVC of 44% predicted and DLco 24.5% predicted.

In a 14-patient open-label proof-of-concept study Daoussis et al. 31 assessed the role of RT in patients with SSc-ILD. The authors randomly assigned patients (based on year of birth, with those born in an even year assigned to RT) to receive either two cycles of RT (consisting 375 mg of RT per week for 4 weeks) at baseline and again at 6 months with (undefined) standard of care. Eight of the 14 subjects were assigned to RT. There was a marked imbalance between groups in disease severity. In the RT group baseline FVC was 68.1 ± 19.7% predicted and DLco 52.3 ± 20.7% predicted compared with FVC 86 ± 19.6% predicted and DLco 65.3 ± 21.4% predicted in the standard of care group. At 1 year after study enrolment the RT-treated subjects saw an absolute improvement in FVC of over 7% with mean FVC rising to 75.6 ± 19.7%. In the standard of care group there was an absolute drop in FVC of over 4% down to 81.7 ± 20.7%. RT treatment was also associated with improvements in DLco and skin thickening over 12 months.

In a retrospective study conducted across six Portuguese rheumatology units the effect of RT was assessed in 49 patients with a range of autoimmune and CTD-associated ILDs (with the commonest ILD, accounting for almost two-thirds of cases, being rheumatoid-associated ILD). 52 The authors divided patients based on whether their imaging was most consistent with UIP or NSIP. In both groups there was an improvement in FVC within the first year (4.3% in the NSIP group and 4.2% in the UIP group).

A recent randomised open-label controlled trial35 of 60 subjects with diffuse cutaneous systemic sclerosis and lung involvement compared RT (1000 mg at baseline and 2 weeks) to CP (500 mg/m² BSA every month for 6 months). At 6 months the authors saw an improvement in FVC in the RT group from 61.3% ± 11.3% predicted to 67.5 ± 13.6% predicted (an absolute change of 140 ml) while in the CP arm it declined from 59.3 ± 13.0 to 58.1 ± 11.2%. The mRSS improved in the RT but not the CP arm and there were fewer AEs associated with RT.

The DESIRES study53 was a randomised placebo-controlled study of 56 individuals with a mRSS score > 10 which compared RT (375 mg/m² per week for 4 weeks) to placebo. The primary end point of the study was change in mRSS at 24 weeks. The study met its primary end point. Just under 90% of patients in each arm had interstitial lung disease although average lung function was well preserved at entry into the study; FVC was 87.9 ± 15.8% predicted in the RT arm and 89.4 ± 17.9% predicted in the placebo arm. DLco was 84.1 ± 19.3% predicted and 80.6 ± 16.6% predicted in the RT and placebo arms, respectively. At 6 months there was a small improvement in FVC compared with baseline in the RT group of 0.09% compared to a decline of 2.87% in the placebo group; a difference of 2.96% (95% CI 0.08 to 5.84%, p = 0.044). The data generated by this trial have led to the approval of RT as a treatment for systemic sclerosis in Japan.

Overall, the results of RECITAL are comparable to these prior studies. The effect of RT was bigger in retrospective and open-label studies but this is in keeping with the general observation that non-blinded and especially retrospective cohort analyses tend to overestimate treatment efficacy. 54 This issue notwithstanding we observed positive effects of RT on all measures of lung physiology, QoL and, in the scleroderma subset, skin thickening which were comparable in magnitude to these prior observations. The effect of RT in the systemic sclerosis ILD subset of RECITAL mirror the effects of RT reported in the DESIRES study.

Cyclophosphamide

The role of CP has been studied in three randomised controlled trials (RCTs). The first two of these, SLS-1¹² and -2¹³ were, as the names of the trials suggest, only conducted in individuals with scleroderma-associated ILD. In both studies, CP was administered orally on a daily basis for 12 months (dosing CP beyond 12 months is precluded by the drug’s cumulative toxicity). SLS1, conducted in 158 subjects, provided a comparison of CP with placebo. SLS2, which randomised 142 subjects, compared CP to MMF. In SLS-1, CP resulted in 2.53% (95% CI 0.28 to 4.79%) improvement in FVC compared to baseline with a mean adjusted difference compared with placebo at 12 months of 2.97% (95% CI 0.75 to 5.19%). There was no significant difference between groups in DLco at 12 months. There was a clinically meaningful improvement in dyspnoea score (measured using the transitional dyspnoea index) of 1.4 ± 0.23 in the CP group compared to a worsening of −1.5 ± 0.45 in the placebo group. The beneficial effect on FVC of CP was lost by 24 months with no difference between groups in this outcome at this timepoint. Side effects were more common in the CP arm; those most frequently reported included haematuria, leukopenia, neutropenia and anaemia.

In SLS-2 treatment with 12 months of oral CP was associated with a 2.88% (95% CI 1.19 to 4.58) improvement in FVC at 24 months compared with baseline. Mycophenolate mofetil, administered at a dose of 1500 mg twice daily for 2 years was associated with a similar magnitude of benefit at 24 months with an improvement of 2.19% (95% CI 1.53 to 3.84%) compared to baseline. DLco remained stable in the mycophenolate arm and decreased slightly in the CP arm. Again, the commonest side effects were leukopenia and anaemia which occurred in 30 subjects on CP and 4 on mycophenolate. Overall, by the end of the trial, a higher proportion of subjects remained on treatment with mycophenolate.

In a small, 45-patient trial of monthly intravenous CP for 6 months followed by oral azathioprine compared with placebo in individuals with scleroderma-associated ILD, Hoyles et al. observed a 1.4% improvement in FVC compared to baseline in the treatment group; this compared with a 3% loss of FVC in the placebo arm. Because of recruitment challenges the study was underpowered and the differences between arms failed to reach statistical significance. AEs resulting in withdrawal only occurred in two subjects on CP (these were intractable nausea and abnormal liver function tests). The commonest side effects in the active treatment arm were nausea, mood disturbance, mouth ulceration, rash, diarrhoea and altered liver function tests.

Overall, both the efficacy and tolerability of CP seen in RECITAL were in keeping with that observed in these previous trials (with intravenous CP being better tolerated than orally dosed CP). The lack of a bone marrow effect in RECITAL compared with SLS-1 and SLS-2 likely relates to the differences in administration route and the lower cumulative exposure achieved by intravenous dosing. The bigger treatment effect observed with CP in the RECITAL study likely reflects the inclusion of a sizable proportion of individuals with myositis. As shown in subgroup analysis of RECITAL the largest treatment effect appeared to occur in individuals with myositis while the effect in the scleroderma ILD subgroup was more in keeping with that seen in SLS-1 and SLS-2.

Mycophenolate

Mycophenolate mofetil, an immunosuppressant medication originally developed to prevent organ rejection following transplant, is widely used in the treatment of CTD-ILD despite only limited evidence to support its use. In a retrospective study of 125 individuals with a diverse range of CTD-ILDs treated at a single centre, Fischer et al. reported a one-year improvement in FVC of 4.9% ± 1.9% compared with FVC prior to treatment initiation. 55 The improvement was most striking in the individuals with a non-UIP pattern on CT scan. As mentioned above, MMF at a dose of 1.5 g twice daily for 2 years was compared to 12 months of oral CP in the SLS-2 study. MMF failed to show a benefit compared with CP but as noted, both drugs resulted in improved FVC compared with baseline. 13

In the SENSCIS trial,17 which compared nintedanib to placebo in individuals with SSc-ILD, subjects were permitted to be on mycophenolate as long as they had been on a stable dose for at least 6 months prior to study enrolment. Approximately half of subjects in the study were taking MMF at baseline. In a post hoc analysis the greatest rate of annual decline in FVC was observed in patients who were neither receiving nintedanib nor MMF. 18 Mycophenolate approximately halved the rate of decline while the group of patients with the greatest preservation of lung function were those taking both nintedanib and mycophenolate.

Nintedanib

Nintedanib is a multityrosine kinase inhibitor which was first developed as an antifibrotic therapy for patients with IPF. In trials performed in individuals with IPF nintedanib approximately halves the loss of FVC over 52 weeks. 16,56 In the SENSCIS trial nintedanib was compared with placebo in 580 individuals with SSc-ILD and a minimum of 10% disease extent on CT. 18 Over 52 weeks patients receiving nintedanib lost 52.4 ml of FVC compared with a loss of 93.3 ml in the placebo group; a difference of 41 ml (95% CI 2.9 to 79.0 ml). There was no difference in change of QoL over 12 months between treatment groups. On the back of this study nintedanib was approved in Europe and the USA as a treatment for SSc-ILD.

In the INBUILD trial, nintedanib was compared to placebo in a basket trial incorporating patients with any form of progressive fibrotic lung disease (variously referred to as progressive pulmonary fibrosis, progressive fibrosing-ILD or chronic fibrotic ILD with a progressive phenotype). 57 To be included in the trial patients had to have evidence of pulmonary fibrosis with >10% disease extent on CT scan and evidence of disease progression in the prior 2 years as characterised by either: (1) >10% loss of FVC, (2) >5% but <10% loss of FVC combined with worsening symptoms, (3) >5% but <10% loss of FVC combined with evidence of worsening fibrosis on serial CT scans or, (4) worsening symptoms combined with evidence of worsening fibrosis on serial CT scans. The study enrolled 663 subjects of whom 170 had various forms of autoimmune-related ILD; of these 39 had SSc-ILD and 19 had MCTD-ILD. Overall the effect of nintedanib was similar to that seen in IPF trials with an approximate halving in the rate of FVC decline over 52 weeks [a difference between nintedanib and placebo arms of 107 ml (95% CI 65.4 to 148.5 ml)]. The effect of nintedanib was consistent across disease subgroups including in those patients with CTD-ILD.

Tocilizumab

Tocilizumab is a humanised monoclonal antibody targeted against the IL-6 receptor. It has recently been approved as a treatment for SSc-ILD following two clinical trials which were designed to test the drug’s efficacy in treating the skin thickening associated with scleroderma. Both trials recruited individuals with diffuse cutaneous systemic sclerosis which had been rapidly deteriorating in the prior year. Furthermore, patients had to have evidence of active inflammation characterised by one of arthritis, a raised C reactive protein level or a raised platelet count. It was not necessary for subjects to have ILD to be enrolled in the studies and overall only about two-thirds of participants in either trial had evidence on CT of ILD. The first study, the faSScinate trial, enrolled 87 patients and showed a trend towards improved skin thickening. 14 An unexpected finding in the study was the rapid rate of FVC decline in the placebo group and the striking attenuation of this decline in the active treatment arm. This study was followed by a 210-patient phase 3 study (the focuSSed trial) that used the same inclusion and exclusion criteria. 15 Again, the primary end point for the study was change in skin score measured using mRSS and again this showed a trend towards (but not a statistically significant) benefit with active treatment. As with the faSScinate trial, the placebo group lost >200 ml of FVC in 1 year and this loss was attenuated by treatment with tocilizumab. In general, tocilizumab was safe and well tolerated in both trials.