Study Design
The CZP-SPEED study (NCT01443364 and EudraCT 2011-000385-35) was a 52-week, open-label, prospective, interventional, multicenter study which evaluated the predictability of disease control at week 52 based on early response to CZP (in combination with MTX) in Italian patients with RA. CZP was supplied by UCB Pharma. Concomitant use of oral corticosteroids (≤ 10 mg/day prednisone, or equivalent) and non-steroidal anti-inflammatory drugs (NSAIDs) was permitted at stable doses during the course of the study. Analgesics without anti-inflammatory action and oral opioid analgesics were also permitted during the study but could not be used prior to clinical assessments on the day of a scheduled visit. Assessments were carried out at weeks 0, 1, 2, 4, 6, 8, 12, 24, 36, and 52.
Patients
Patients enrolled in the trial were biologic-naïve, had moderate-to-severe RA (defined as at least six swollen and at least six tender joints) and had failed treatment with at least one disease-modifying anti-rheumatic drug (DMARD); which is in line with the indication of CZP stated in the approved label [Summary of product Characteristics (SmPC)] [
20]. Patients were treated with CZP (400 mg at weeks 0, 2 and 4, followed by 200 mg every 2 weeks) in combination with MTX (≥ 15 or ≤ 25 mg per week, unless limited by subject intolerance or toxicity). The doses of CZP used in the study were in agreement with the approved label (SmPC) for treatment of RA with CZP [
20]. The safety set (SS) consisted of all patients who received at least one dose of CZP. The full analysis set (FAS) consisted of all patients in the SS who had a valid baseline and post-baseline efficacy measurement for the primary objective. A subgroup of these patients (the PDUS subgroup) provided additional ultrasound and radiographic imaging evaluations if they had been diagnosed with RA for at most 2 years and they presented with at least two of the following PDUS findings: synovial proliferation, synovial vascularization, or presence of fluid in at least one metacarpophalangeal (MCP) joint.
Study Procedures and Evaluations
The primary objective was to detect the time point of clinical response with the highest positive predictive value (PPV). PPV was defined as the percentage of patients with a clinical response [reduction from baseline in DAS28(ESR) of at least 1.2] up to week 12, who also had clinical response [reduction from baseline in DAS28(ESR) of at least 1.2] at week 52.
Additional clinical exploratory variables included American College of Rheumatology ACR20/50/70 response [
21], DAS28(ESR), DAS28(CRP), Patient Global Assessment of Disease Activity (PtGADA) [
22], Physician Global Assessment of Disease Activity (PhGADA), Patient Assessment of Arthritis Pain (PtAAP), and Health Assessment Questionnaire Disability Index (HAQ-DI) [
21].
Imaging Evaluations
The secondary objective of the study was to observe the effectiveness of PDUS in assessing both disease activity and impact of treatment in the subgroup of subjects undergoing PDUS. Additionally, the degrees of joint damage of the hands, wrists, and feet were assessed radiographically using the van der Heijde modified total Sharp score (mTSS) in the PDUS subgroup of patients [
23,
24].
The second and third MCP joints, bilaterally together with another MCP joint (the most clinically involved), were examined by ultrasound measurements at weeks 0, 1, 2, 4, 6, 8, 12, 24, 36, and 52. Ultrasound variables under evaluation were joint cavity widening, synovial proliferation, Power Doppler (PD) signal (synovial vascularization), bone erosion, and cartilage damage score measurements according to international guidelines. Joint cavity widening and PD signal were scored from 0 to 3 depending on the extent of synovial effusion and/or proliferation [
25,
26] and the level of PD signal at the synovial tissue [
27,
28], respectively. Cartilage damage and bone erosions were scored from 0 to 4 on the basis of metacarpal cartilage thickness at the MCP joint level [
29,
30] and diameter of bone erosion per joint [
26,
31], respectively. PDUS sum scores were the sum of the scores generated using these two semiquantitative scoring systems [
25,
27].
A standardized instrumentation (MyLab ™70 XVG or Mylab ™Twice-Esaote Biomedica, or GE Logiq 9, equipped with a broadband linear probe of 6–18 MHz) and operational technique was adopted with multiplanar scanning technique on dorsal, lateral, and volar aspects of the MCP joints. All sonographers were trained before the start of the study. All images were re-assessed by an assigned ultrasound expert reader and analyzed for inter-reader reliability.
Safety Measurements
Safety measurements assessed included adverse events (AEs) classified by system organ class (SOC) and preferred term (PT) using the Medical Dictionary for Regulatory Activities (MedDRA) version 18.0. Only treatment-emergent AEs (TEAEs) were included in the summary tables.
Statistical Analysis
The SS consisted of all enrolled patients who received at least one dose of study medication. Clinical outcomes and PROs were assessed in the full analysis set (FAS), defined as all patients who received at least one dose of CZP and had at least one baseline and at least one post-baseline DAS28(ESR) assessment (the primary clinical variable). PDUS outcomes were assessed in a subgroup of enrolled patients with RA disease duration less than 2 years and proof of at least two of the following: synovial proliferation, synovial vascularization, and presence of fluid in at least one selected joint.
A sample size of 120 patients was calculated to be appropriate to produce 95% confidence intervals (CIs) with sufficient precision (± 0.069) to estimate predictability for the base case of a 60% response rate. Precision would be higher if the proportion of responders was greater than 60%. At least 40 (33.3%) of the 120 included patients were planned to be part of the PDUS subgroup of patients.
In order to analyze the primary objective, DAS28(ESR) was measured at weeks 1, 2, 4, 6, 8, and 12 and was used to predict maintenance of long-term efficacy in CZP-treated patients, using the PPV and negative predictive values (NPV). Logistic regression was performed post hoc to evaluate the predictive ability of early responders, up to week 12, and long-term efficacy at week 52, after controlling for the prognostic factors [baseline DAS28(ESR), gender, age, and duration of RA]. Area under curve (AUC) was obtained from fitting the logistic regression model with week 52 response as the dependent variable and early response (weeks 1–12) as the independent variable. The model with the maximum AUC was considered the better predictor.
PPV was defined using the equation: [
32]
\( \frac{n}{N} \times 100, \) where
n is the number of patients with a clinical response at week 52 and a clinical response at a specified early time point, and
N is all patients with a clinical response at a specified early time point. Clinical response was defined as a reduction in DAS28(ESR) of at least 1.2. For NPV,
n is the number of patients who did not have a clinical response at week 52 nor a clinical response at a specified early time point.
A mixed effects linear model for repeated measures was used to examine the effect of each of the prognostic factors on the sum scores describing disease activity (joint cavity widening, synovial proliferation, and PD signal/blood flow) and disease severity (progression in cartilage damage and bone erosions). Within this model, “week” was a repeated measure and the prognostic factors [baseline DAS28(ESR), gender, age, duration of RA] were fixed effects. The regression coefficient, standard error (SE), and p values associated with each prognostic factor are presented. During a post hoc analysis, patient duration of RA and the presence or absence of bone erosion were assessed for PDUS outcomes. Correlation analysis was conducted between PDUS and mTSS at each visit using the Spearman’s correlation coefficient. Inter-reader reliability of all six joints was assessed in a subgroup of the PDUS patients at baseline and weeks 6, 12, and 36 using a blinded expert to review the images for PD signal/blood flow and cartilage damage. The first and second readings of the images were analyzed for inter-reader reliability using Cohen’s kappa statistics.
Missing data were imputed using non-responder imputation (NRI) for categorical data and last observation carried forward (LOCF) for continuous data.