Clinical trial methods

Sven Plein; Bara Erhayiem; Graham Fent; Jacqueline Andrews; John Greenwood; Paul Baxter; Elizabeth M Hensor; Sue Pavitt; Maya H Buch

NCBI Bookshelf. A service of the National Library of Medicine, National Institutes of Health.

Plein S, Erhayiem B, Fent G, et al. Using cardiovascular magnetic resonance to define mechanisms of comorbidity and to measure the effect of biological therapy: the CADERA observational study. Southampton (UK): NIHR Journals Library; 2021 Mar. (Efficacy and Mechanism Evaluation, No. 8.4.)

Cover of Using cardiovascular magnetic resonance to define mechanisms of comorbidity and to measure the effect of biological therapy: the CADERA observational study

Using cardiovascular magnetic resonance to define mechanisms of comorbidity and to measure the effect of biological therapy: the CADERA observational study.

Show details

Contents

< Prev Next >

Chapter 2Clinical trial methods

Objectives

Primary objectives

To determine if CV abnormalities, as assessed by CMR imaging, are present in a treatment-naive inception cohort of early RA patients compared with control subjects.
To establish whether or not (any) RA DMARD strategy is associated with improvement in CV abnormalities, as measured by CMR imaging, in a treatment-naive inception cohort of early RA patients over a 1-year period.
To establish whether or not TNFi therapy confers a quantitative difference in the incidence of CVD compared with standard therapy, as measured by CMR imaging, in a treatment-naive inception cohort of early RA patients over a 1-year period.

Secondary objectives

To determine changes in CV abnormalities, as assessed by CMR imaging, in a treatment-naive inception cohort of early RA patients over a 2-year period.
Improvement in CV abnormalities relative to RA disease control (based on response status) and also each treatment strategy (initial TNFi/MTX vs initial csDMARD including MTX). This was an additional analysis undertaken as part of a protocol amendment.

Exploratory objectives (not included in this report and subject to separate funding)

To undertake bolt-on studies on biological samples (blood) collected from all patients recruited into the study to identify RA patients with a high-risk profile for preclinical CVD and identify candidate markers and shared mechanisms.
To link CMR imaging findings to clinical outcomes through long-term follow-up in a longitudinal observational registry.

Design

The study was a longitudinal, observational CMR imaging study that bolted onto the parent VEDERA trial, a phase IV, pragmatic, single-centre, open-label randomised controlled trial of new-onset, treatment-naive early RA comparing the clinical effectiveness of two different first-line therapeutic strategies.²³ Participants were randomised to one of two first-line therapeutic strategies: (1) immediate TNFi therapy (ETN) and MTX or (2) MTX ± additional csDMARD therapy in a treat-to-target approach, with a switch to delayed ETN and MTX in the event of failure to achieve clinical remission at 6 months. The primary end point was the comparison of the proportion of patients in each treatment arm to achieve DAS28-ESR remission. Secondary end points included evaluation of the response to first-line ETN compared with delayed ETN (following MTX).

The VEDERA trial participants who also consented to the CADERA substudy underwent CMR scans at baseline (prior to the commencement of randomised RA treatment) and at 1 and 2 years after enrolment.

The study was reviewed and approved by the National Research Ethics Service Research Ethics Committee Leeds (West) (reference 10/H1307/138).

Patient and public involvement

Mrs Ailsa Bosworth, Chief Executive and founder of the National Rheumatoid Arthritis Society (NRAS), was the patient and public involvement (PPI) contributor on the Trial Steering Group. Through membership of the Trial Steering Committee, the PPI contributor also provided input into the conduct of the trial as needed; this ensured a patient-centred approach to maximise the chance of obtaining patient therapeutic benefits.

Participants

Patients were recruited between February 2012 and November 2015 from a single tertiary centre, an early RA rheumatology outpatient clinic (Chapel Allerton Hospital, Leeds, UK). Consecutive patients diagnosed with new-onset RA according to American College of Rheumatology (ACR)/EULAR 2010 criteria²⁷ were invited to enrol into the VEDERA randomised controlled trial²³ and the parallel CADERA CVD substudy.²⁴

Thirty control subjects (free of CVD) were recruited through local advertisements, e-mail shots and word of mouth. From the responders, those of similar age and sex distribution to the study subjects were selected. Exclusion criteria were any history of rheumatological and/or CVD.

Eligibility

The VEDERA inclusion criteria

The VEDERA inclusion criteria were no previous use of DMARD therapy, a symptom duration of ≤ 1 year, a DAS28-ESR of ≥ 3.2 and at least one poor prognostic factor [positive for anti-citrullinated peptide antibody (ACPA) ± rheumatoid factor (RF) or, in ACPA- and RF-negative individuals, the presence of abnormal power Doppler in any joint].

The VEDERA exclusion criteria

The VEDERA exclusion criteria were previous use of DMARD therapy, contraindications to TNFi therapy or patients considered clinically unsuitable for TNFi therapy by the treating physician.

Additional CADERA exclusion criteria

Additional CADERA exclusion criteria were renal failure (estimated glomerular filtration rate < 30 ml/minute/1.73 m²), contraindications to intravenous adenosine (asthma or high-grade atrioventricular block), a known allergy to gadolinium-based contrast agent, claustrophobia, a weight of ≥ 120 kg and a metallic implant/foreign body deemed unsafe for CMR imaging.

Interventions

This observational study bolted onto the parent VEDERA trial, which included the following 1-year duration randomised intervention treatment arms:

experimental treatment arm (group 1) – ETN group: immediate ETN and MTX treatment
control treatment arm (group 2) – standard treatment group: initial MTX monotherapy with a treat-to-target regimen – escalation to combination conventional synthetic DMARD therapy at or after 8 weeks (up to week 24) if failing to meet the predefined target at 4-weekly assessments, and if failing to meet the predefined target of clinical remission at week 24, step up to bDMARD, ETN and MTX.

Etanercept (Enbrel^®, Pfizer Inc., New York, NY, USA) is a human tumour necrosis factor (TNF) receptor p75Fc fusion protein produced by recombinant deoxyribonucleic acid (DNA) technology. MTX and/or additional csDMARDs include sulfasalazine and hydroxychloroquine.

After week 48, participants on ETN stopped therapy and standard of care treatment was maintained for the year 2 observational period.

Study procedures

Following written informed consent (to participate in the VEDERA and the CADERA studies) and prior to any trial-related procedures for the CADERA study, patients were registered into the CADERA study. All patients underwent screening within the 4 weeks prior to the baseline visit for the VEDERA trial. At the baseline visit eligibility for the study was confirmed, patients were randomised to one of the treatment groups and study treatment was initiated. Further visits, which included assessment of disease activity, took place at weeks 4, 12 and every 12 weeks thereafter, up to week 96, for both treatment arms, with additional visits (for group 2) at weeks 8, 16 and 20 for safety and efficacy evaluation within a treat-to-target protocol.

At baseline and at 1 and 2 years following initiation of therapy, all patients recruited to the CADERA study also underwent comprehensive CMR imaging.

Cardiovascular magnetic resonance imaging protocol

Each CMR imaging study was conducted using the same protocol. Imaging took approximately 1 hour. The CMR imaging protocol included:

cine imaging in a stack of LV short-axis sections covering the entire heart for measurement of ventricular volumes and LV mass (LVM)
MR tagging in three LV short-axis sections to derive LV longitudinal strain and twist
cine imaging of the thoracic aorta to allow measurement of aortic distensibility (AD)
adenosine stress and rest first-pass perfusion imaging using a dual-bolus technique, with intravenous infusion of 0.01 and 0.1 mmol/kg gadopentate dimeglumine (Magnevist^®, Bayer, Berlin, Germany) – this part of the protocol was optional for patients and the results are not reported here
T1 mapping (modified look-locker inversion recovery method) pre and 10 minutes post contrast to derive the myocardial extracellular volume (MECV) fraction
late gadolinium enhancement (LGE) imaging.

Outcome measures

Primary outcome measure

The primary outcome measure for the study was AD, a measure of aortic stiffness. This was chosen because of its dynamic nature, high reproducibility and ability to predict major CV events independently of traditional clinical risk scoring models in patients with no known CVD.³² The normality of data were tested using a Shapiro–Wilk test.

Secondary outcome measures

Left ventricular ejection fraction (LVEF).
Left ventricular longitudinal strain (LVLS).
Left ventricular twist.
Left ventricular mass.
Myocardial perfusion reserve (not reported here).

Exploratory outcome measures

Myocardial extracellular volume fraction from T1 mapping.
Biomarkers of CVD (as part of future biological studies; not reported here).

Patient withdrawal

Patients could withdraw from the study at any time without explanation and continue to receive treatment as per standard clinical practice. Patient withdrawal was categorised as withdrawing consent for further study follow-up evaluation only, but willing to have CMR imaging data to date kept and used.

Sample size and power calculation

The sample size estimation for the primary end point AD was based on previous work by Ikonomidis et al.,²³ which showed improved AD in RA patients in response to interleukin-1 therapy. We calculated that 33 patients in each treatment group would equate to an 80% power to detect a difference in AD between the two treatment groups at a 5% significance level.

Blinding

The CMR fellows analysing the CMR scans were blinded to the VEDERA treatment allocation.

Analysis

Statistical analysis

The first 30 CADERA patients (cases) were approximately matched by age and sex to control subjects, who were invited for a CMR scan. Matched pairs analysis was conducted on the primary outcome (AD), secondary outcomes [LVEF, LVLS, peak left ventricular twist (PLVTw)] and exploratory outcome (MECV) at baseline. To reflect the approximate matching, additional linear regression analyses are presented for each of the primary, secondary and exploratory outcomes at baseline including the 30 control subjects and all cases when data are available. Each analysis is presented unadjusted (with a single independent variable case/control) and adjusted (with dependent variables of case/control, age, sex, systolic blood pressure and pack-years smoked). Outcomes are natural log transformed to achieve normality assumptions (an approach favoured over non-parametric analysis to allow straightforward adjustment in the additional regression analyses and consistency with the analyses at the 1-year and 2-year follow-ups. Back-transformation is performed when presenting results [thus, geometric mean ratios and associated 95% confidence intervals (CIs) are reported]. To assess the balance of demographic characteristics across all cases and the 30 control subjects, summary statistics are presented for age, sex, systolic blood pressure and pack-years smoked.

For analysis of the primary, secondary and exploratory outcomes in the CADERA patients at the 1-year and 2-year follow-ups, an analysis of covariance (ANCOVA) method has been adopted. The outcome at follow-up is linearly regressed on a grouping variable (defined below) as the dependent variable, first as an unadjusted analysis (exploring differences in the outcome by group at 1 year or 2 years), second, additionally adjusted for baseline outcome (an ANCOVA approach to analysis of change, replacing the differences from baseline method originally proposed) and, third, additionally adjusted for baseline outcome, age, sex, systolic blood pressure, pack-years smoked and, where appropriate (see below), baseline DAS28-ESR (again, an ANCOVA approach to analysis of change, replacing the differences from baseline method originally proposed). The change of approach compared with the CADERA study protocol, from differences from baseline to an ANCOVA approach, was undertaken to enhance the power, given the additional grouped analyses to be undertaken (see below). Outcomes are naturally log transformed to achieve normality assumptions. Back-transformation is performed when presenting results and thus estimates are geometric means or ratios across groups.

Groups were defined as follows, in which group 1 patients received first-line ETN/MTX and group 2 patients received first-line MTX ± additional csDMARD therapy that either continued (2a) or escalated to delayed ETN/MTX (2b) at week 24. The intention of these subgroup analyses was to determine any differences between patients who were treated with first-line ETN and those who never received ETN:

group 1 (all patients) compared with group 2 (all patients)
combined group 1 and group 2 responders compared with combined group 1 and group 2 non-responders, adjusting for baseline DAS28-ESR
group 1 responders compared with group 1 non-responders, adjusting for baseline DAS28-ESR
group 1 responders compared with group 2a responders, adjusting for baseline DAS28-ESR.

Patients with a DAS28-ESR of ≥ 2.6 at 48 weeks were considered non-responders. Patients with DAS28-ESR of < 2.6 at week 48 for reasons other than withdrawal for lack of efficacy were assumed to be responders in the initial analysis. The data were then reanalysed assuming that they were non-responders.

For an intention-to-treat analysis, multiple imputation by chained equations was used to create 50 complete data sets per outcome analysed, the results from which were combined in accordance with Rubin’s rules. The imputation model included the following variables at weeks 0, 12, 24, 36 and 48: Swollen Joint Count-28 joints (SJC44), Ritchie Articular Index (RAI), C-reactive protein (CRP), ESR, physician visual analogue scale (VAS), patient general health VAS, patient pain VAS, patient disease activity VAS and early morning stiffness. Outcome values at 1 or 2 years were included (dependent on outcome analysed) in addition to baseline values. The model also included age, sex, systolic blood pressure, pack-years smoked, symptom duration, treatment and an indicator for whether or not a patient was a responder (as defined above). Multiple imputation chain mixing and distribution diagnostics were checked for plausibility of imputations.

For the primary outcome (AD) at 1 year only, an analysis adjusted for area under the curve (AUC) disease activity is reported. A trapezoid rule is used based on DAS28-ESR at weeks 0, 12, 24, 36 and 48 to calculate the AUC. The imputation model for this analysis excluded the indicator variable for whether or not a patient was a responder.

For comparison, demographics are presented for those CADERA patients who have at least one primary, secondary or exploratory outcome missing at the 1-year or 2-year follow-up and those CADERA patients who have complete data for all outcomes at 1 year and 2 years.

Where relevant, protocol amendments (see Appendix 4) informed any change and/or additional analyses.

Cardiovascular magnetic resonance analysis

All post-processing analysis of CMR scans was performed using CVI 42 software (Circle Cardiovascular Imaging, Calgary, Canada) except for tagging analysis by assessors with > 2 years’ experience in CMR imaging reporting blinded to all patient details.

AD was calculated from cross-sectional cine images of the thoracic aorta at the level of the pulmonary artery bifurcation. The endocardial aortic border was delineated in both diastole and systole. AD was calculated as follows: Δ relative aortic cross-sectional area/Δ systolic and diastolic blood pressure.
The LV endocardial and epicardial borders were manually traced from the short-axis LV cine stack at both end-systole and end-diastole. The LV end-systolic and end-diastolic volumes were generated by summation of discs and the LVEF calculated. The LVM was calculated from the end-diastolic myocardial volume, with trabeculation and papillary muscles excluded.
Tagged images were analysed with inTag software version 1.0 (Creatis, Lyon, France). Endocardial and epicardial borders were outlined. Peak circumferential strain and rotation were calculated for three slices. PLVTw was calculated by subtracting basal from apical rotation. LVLS was defined as peak LV systolic annular velocity ('Mid LV S prime').
Pre-contrast and post-contrast T1 was measured in a region of interest in the interventricular septum of the midventricular slice. The venous haematocrit (Hct) was used to calculate the MECV fraction according to the following equation:

MECV = (1 - Hct) \times ((1 / myocardial post contrast T 1 (milliseconds) - (1 / myocardial native T 1 (milliseconds)) / ((1 / blood post - contrast T 1 (milliseconds)) - (1 / blood pre - contrast T 1 (milliseconds)) .

(1)

Copyright © Queen’s Printer and Controller of HMSO 2021. This work was produced by Plein 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.

Bookshelf ID: NBK568924

Contents