Predictors of long-term clinical response in patients with non-radiographic axial spondyloarthritis receiving certolizumab pegol

Background: Identification of predictive clinical factors of long-term treatment response may contribute to improved management of non-radiographic axSpA (nr-axSpA) patients. This analysis aims to identify whether any baseline characteristics or Week 12 clinical outcomes in nr-axSpA patients with elevated C-reactive protein (CRP) and/or sacroiliitis on magnetic resonance imaging (MRI) enrolled in the C-axSpAnd study are predictive of achieving clinical response after 1 year of certolizumab pegol (CZP). Methods: C-axSpAnd (NCT02552212) was a phase 3, multicentre study, including a 52-Week double-blind, placebo-controlled period. Enrolled patients were randomised to CZP 200 mg Q2W or placebo. Predictors of Week 12 (CZP group only) and Week 52 clinical response were identified using a multivariate stepwise logistic regression analysis. Response variables included Ankylosing Spondylitis Disease Activity Score major improvement (ASDAS-MI), Assessment of SpondyloArthritis International Society 40% response (ASAS40), Bath Ankylosing Spondylitis Disease Activity Index 50% response (BASDAI50) and ASDAS inactive disease (ASDAS-ID). Predictive factors assessed included demographic and baseline characteristics and clinical outcomes at Week 12. A p-value <0.05 was required for forward selection into the model and p >= 0.1 for backward elimination. Missing data or values collected after switching to openlabel treatment were accounted for using non-responder imputation. Sensitivity analyses accounted for patients with changes in non-biologic background medication. Results: Of 317 enrolled patients, 159 and 158 were randomised to CZP and placebo, respectively. Younger age and male sex were identified as predictors of Week 12 response across all assessed efficacy outcomes in CZP-treated patients. Consistent predictors of Week 52 response, measured by ASDAS-MI, ASAS40 and BASDAI50, included human leukocyte antigen (HLA)-B27 positivity and sacroiliitis on MRI at baseline. MRI positivity was also predictive of achieving ASDAS-ID at Week 52. Sensitivity analyses were generally consistent with the primary analysis. In placebo-treated patients, no meaningful predictors of Week 52 response were identified. Conclusions: In this 52-Week, placebo-controlled study in nr-axSpA patients with elevated CRP and/or active sacroiliitis on MRI at baseline, MRI sacroiliitis and HLA-B27 positivity, but not elevated CRP or responses at Week 12, were predictive of long-term clinical response to CZP. Findings may support rheumatologists to identify patients suitable for TNFi treatment.