Study design
COAST-V (NCT02696785) and COAST-W (NCT02696798) were phase 3, multicentre, randomised, double-blind, active (COAST-V only) and placebo-controlled, 52-week trials. The trial designs for COAST-V and COAST-W have been previously published [6, 7, 9]. Briefly, patients in both studies who were assigned to ixekizumab were randomised on a 1:1 ratio to receive a starting dose of either 80 mg or 160 mg. The efficacy and safety of ixekizumab was assessed for 80 mg every 2 weeks (Q2W) and every 4 weeks (Q4W) compared with placebo during a 16-week placebo-controlled blinded treatment dosing period, followed by a dose double-blind extended treatment period (up to 52 weeks). COAST-V contained an active-reference arm including patients receiving 40 mg of adalimumab Q2W up to week 16. The trial protocols were approved by the ethics review boards at each study site. The master ethics committee was Schulman Associates IRB, Cincinnati, OH, USA (IRB # 201506061 and IRB # 201506079 for COAST-V and COAST-W, respectively), and full listings of investigators and sites are available in previously published manuscript supplements [6, 7]. Both trials were conducted in accordance with the ethical principles of the Declaration of Helsinki. All patients gave written informed consent prior to the start of trial-related procedures.
Patients
Inclusion criteria for COAST-V and COAST-W have been previously described [6, 7, 9]. Eligible adult patients fulfilled the ASAS criteria for axSpA, including radiographic evidence of sacroiliitis according to modified New York criteria associated with ≥1 SpA feature. In addition, patients had a Bath Ankylosing Spondylitis Disease Activity Index (BASDAI) ≥4 and total back pain score ≥ 4 on a numeric rating scale (NRS) and inadequate response or intolerance to non-steroidal anti-inflammation drug therapy. COAST-V included only patients with no prior or current use of bDMARDs. COAST-W included only patients with prior treatment with 1 or 2 TNFis.
Treatment protocol
The dosing period regimens for COAST-V and COAST-W have been published [6, 7, 9, 10]. Briefly, 341 patients in COAST-V and 316 in COAST-W were randomized to receive subcutaneous injections of 80-mg ixekizumab Q2W or Q4W, 40-mg adalimumab Q2W (COAST-V only), or placebo Q2W for 16 weeks. In the dose double-blind extended treatment period (weeks 16 to 52), patients who originally received adalimumab or placebo were rerandomised 1:1 to ixekizumab Q2W or Q4W, while patients originally randomised to receive ixekizumab Q2W or Q4W continued those treatments.
Assessments
The primary endpoint for both COAST-V and COAST-W was the proportion of patients achieving an ASAS40 response at week 16. Secondary efficacy endpoints included ASAS20 and ASAS partial remission, and changes from baseline in ASAS treatment response domains (PtGA, spinal pain, function, and stiffness), in addition to the clinically impactful symptoms of fatigue, spinal pain at night, and sleep quality.
Function was evaluated as the average score of Bath Ankylosing Spondylitis Functional Index (BASFI) responses [8]. Stiffness was evaluated as the average score from responses to BASDAI questions 5 and 6 (BASDAI Q5 and Q6) [8]. Fatigue was evaluated by the Fatigue NRS [11] and by BASDAI question 1 (BASDAI Q1) [8]. Each ASAS response domain was scored on a scale from 0 to 10. Sleep quality was assessed by the Jenkins Sleep Evaluation Questionnaire (JSEQ) and scored on a scale from 0 to 20, with higher scores indicating greater sleep disturbances [12]. Assessments were made at weeks 0, 1, 2, 4, 8, 12, and 16 during the placebo-controlled blinded treatment dosing period [10], and weeks 20, 24, 28, 32, 36, 44, and 52 during the dose double-blind extended treatment period [9], except for Fatigue NRS and JSEQ where assessments were made at week 8, 16, 36, and 52.
Week 52 data were pooled across treatment groups to assess the relationship between improvements in individual PROs and the achievement of different levels of ASAS response. These pooled data were then stratified by ASAS response level: ASAS20 non-responders, ASAS20 but not ASAS40 responders, and ASAS40 responders.
Statistical analyses
Efficacy and health outcome analyses for the placebo-controlled blinded treatment (weeks 0 to 16) and the dose double-blind extended treatment period (weeks 16 to 52) were conducted on all patients randomized to treatment groups in COAST-V and COAST-W. Comparisons between treatment groups, and least square mean changes from baseline for the ASAS treatment response domains and other PROs up to week 16 were analyzed using a mixed-effects model for repeated measures, as described previously [10]. The mixed-effects model included treatment, geographic region, baseline CRP status, baseline value, visit, baseline value-by-visit, and treatment-by-visit interaction as fixed factors, with variance-covariance structure set to unstructured. Number of prior TNFi was an additional fixed factor for the COAST-W model. After week 16, changes from baseline were summarized as mean and standard deviation for the ASAS treatment response domains and other PROs.
Post hoc association analyses were performed using data from week 52 between ASAS response levels and changes from baseline in the four ASAS response domains (PtGA, spinal pain, function, stiffness) and the additional PROs fatigue, spinal pain at night, and sleep quality. From weeks 0 to 52, changes from baseline in these PROs were compared between the ASAS responder groups listed above using analysis of covariance (ANCOVA) models after adjusting for baseline values, age, and gender. Standard error and standard deviation were used as measures of dispersion for ANCOVA. Post hoc comparisons were conducted with Scheffé’s correction, which was based on multiple comparisons of outcome measures. Fold increases were determined by the following calculations: (ASAS40 responder/ASAS20 non-responder) -1 or (ASAS40 responder/ASAS20 but not ASAS40 responder) -1. Modified baseline observation carried forward (mBOCF) was used for imputation of missing data [13].
All analyses were conducted using SAS 9.4. Residual plots were generated via SAS to confirm normality of residuals.