OP0022 Is a primary good response to nsaids predictive of the subsequent response to the first tnf inhibitor in patients with recent axial spondyloarthritis

Background Good response to NSAIDs is a SpA feature included in classification criteria for axial spondyloarthritis (axSpA). Among patients eligible for a TNF inhibitor (TNFi), some patients may have never responded to NSAIDs (non-responders to NSAIDs) while others initially responded (responders to NSAIDs) but have secondly escaped and need to be treated with biologics. Objectives Our aim was to determine if the initial response to NSAIDs is an independent predictive factor of a subsequent good response to the first TNFi in axSpA. Methods: Patients Patients from the prospective observational DESIR cohort of early axSpA cohort who started a TNFi over the 5 years of follow-up. NSAIDs response and TNFi response definitions NSAIDs response was defined by the item ‘good response to NSAIDs according to Amor’s criteria’ at the inclusion visit. TNFi response was defined by the BASDAI50 response between the ‘baseline’ visit (last cohort visit before TNFi initiation) and the ‘follow-up’ visit (visit taking place after at least 8 weeks of TNFi treatment). Analysis We compared the characteristics of the NSAIDs responder to the non-responders and their response to the first TNFi. We performed a multivariate logistic regression modelling the impact of an NSAID response to the TNFi response. We included known predictive factors of TNFi response in this model (age, gender, HLAB-B27, activity of the disease [ASDAS-CRP], CRP, X-ray and MRI sacroiliitis). To account for selection bias and for confirmation purpose, we applied a propensity score with Inverse Probability Weighting (IPW) method to predict TNFi response (SAS, version 9.2). Results Among the 708 patients of the cohort, 236 were included in the analysis. At the inclusion, the main characteristics were the following: 106 (44.7%) males, mean age 33.8±3.9 years, mean BASDAI 54.5±17.3 and 202 (85.6%) were NSAIDs responders. The NSAIDs responder and non-responder groups were comparable at M0 except for HLA-B27 positive status: 59.9% vs 40.1%, p=0,041, CRP level: 13.4±20.3 mg/L vs 6.3±6.6 mg/L, p=0,027, history of psoriasis: 17.8% vs 35.3%, p=0.001 and BASDAI: 53.0±18.1 vs 61.8±13.2, p=0,001, in responder and non-responder patients, respectively. The percentage of TNFi responders was 32.2% (65/202) and 23.5% (8/34) in the NSAIDs responder and non-responder groups, respectively (univariate analysis [OR 1.54 [95% CI: 0.7 to 3.6], p=0.313. The multivariate logistic regression found the following independent factors of the TNFi response: gender [adjusted OR (aOR)=2.9 [CI95%: 1.4–6.0], p=0.004], age [aOR=0.9 [95% CI: 0.91 to 0.99], p=0.026], HLA-B27 status [aOR=2.5 [CI95%: 1.2–5.3], p=0.02], ASDAS-CRP score [aOR=1.6 [95% CI: 1.1 to 2.4], p=0.016], and MRI sacroiliitis [aOR=2.0 [CI95%: 1.0–4.2], p=0.054]. Response to NSAIDs was not significantly associated to the response to the TNFi [aOR=1.93 [95% CI: 0.6 to 6.3], p=0.275]. The IPW aOR confirmed the non-association between NSAIDs good response and TNFi good response: 1.60 [CI95%: 0.7–3.3], p=0.20. Conclusions The good response to NSAIDs according to the Amor’s criteria does not seem to be an independent predictive factor of a good response to the first TNFi in early axSpA patients. Disclosure of Interest None declared