Poor prognostic factors in predicting abatacept response in a phase III randomized controlled trial in psoriatic arthritis

Abstract

In ASTRAEA (NCT01860976), abatacept significantly increased American College of Rheumatology criteria 20% (ACR20) responses at Week 24 versus placebo in patients with psoriatic arthritis (PsA). This post hoc analysis explored relationships between prospectively identified baseline characteristics [poor prognostic factors (PPFs) ] and response to abatacept. Patients were randomized (1:1) to receive subcutaneous abatacept 125 mg weekly or placebo for 24 weeks; those without ≥ 20% improvement in joint counts at Week 16 switched to open-label abatacept. Potential predictors of ACR20 response were identified by treatment arm using multivariate analyses. Likelihood of ACR20 response to abatacept versus placebo was compared in univariate and multivariate analyses in subgroups stratified by the PPF, as defined by EULAR and/or GRAPPA treatment guidelines. Odds ratios (ORs) were generated using logistic regression to identify meaningful differences (OR cut-off: 1.2). 424 patients were randomized and treated (abatacept n = 213; placebo n = 211). In abatacept-treated patients, elevated C-reactive protein (CRP), high Disease Activity Score based on 28 joints (CRP), presence of dactylitis, and ≥ 3 joint erosions were identified as predictors of response (OR > 1.2). In placebo-treated patients, only dactylitis was a potential predictor of response. In the univariate analysis stratified by PPF, ACR20 response was more likely (OR > 1.2) with abatacept versus placebo in patients with baseline PPFs than in those without; multivariate analysis confirmed this finding. Response to abatacept versus placebo is more likely in patients with features indicative of high disease activity and progressive disease; these characteristics are recognized as PPFs in treatment guidelines for PsA.