SAT0098 ANTIBODIES AGAINST CITRULLINATED ANTIGENS ARE ASSOCIATED WITH TREATMENT RESPONSE IN ACPA-NEGATIVE RHEUMATOID ARTHRITIS PATIENTS

Background RA patients who are ACPA-positive (ACPA + ) are known to have worse prognosis. Less is known regarding predictors of treatment outcomes in ACPA-negative (ACPA - ) patients. Objectives We investigated whether autoantibodiescaptured on a custom array were associated with response to therapy in ACPA - RA patients. Methods RA patients were recruited to either the Biologics in RA Genetics and Genomics Study Syndicate(BRAGGSS, starting adalimumab, established disease) or the RA Medication Study (RAMS, starting methotrexate, early disease). Serum samples were collected at pre-treatment and 3/6 months in BRAGGSS/RAMS, respectively. Treatment groups were pooled for analysis. ACPA was measured using a commercially available ELISA (Axis-Shield Diagnostics Ltd, Dundee, UK). Pre-treatment RA and healthy blood donor control (HC) serum samples were incubated on a bead-based assay (Luminex FlexMap 3D) containing 376 human protein antigens associated with autoimmune disease (39 in citrullinated (cit) form) to detect autoantibodies. Median fluorescence intensity (MFI) values were calculated for each autoantibodyfor RA and HC, then normalised and log2-transformed. The 95 th percentile for each autoantibodyin HC was used to determine whether an RA sample was positive/negative for that autoantibody. Proteins with - patients. Results 168 patients with RA were included in analysis (mean age 59.6 years, mean disease duration 14.3 years, 126 (75%) female patients, 90 (53.6%) ACPA + patients). 34 autoantibodieswere differentially expressed in RA patients, only one of which (TNF ligand superfamily member 13, TNFSF13) was lower than in HC. 29/34 autoantigens were in cit-form. In multivariate models of all differentially expressed autoantibodies, heterogeneous nuclear ribonucleoproteinA1 (HNRNPA1) was significantly associated with EULAR response (coefficient (coef) 0.7, 95% CI 0.1-1.3), and cit-vimentin (VIM) was significantly associated with poor EULAR response (OR adj 4.2, 95% CI 1.1-18.3) and reduced odds of good EULAR response (OR adj 0.2, 95% CI 0.1-0.8). ACPA remained the best predictor of treatment response. In a subanalysis of the 78 ACPA - patients, cit- cleavageand polyadenylation specificity factorsubunit 6 (CPSF6) was significantly associated with worse DAS28 (coef (-1.8), 95% CI (-3.6)-(-0.1)). cit-DnaJ homolog subfamily B member 1 ( DNAJB1) was significantly associated with DAS28 improvement (coef 2.2, 95% CI 0.6-3.8). No autoantibodywas associated with EULAR response. Conclusion A subset of ACPA - patients have ACPA fine specificities not seen by a commercial assay. Larger ACPA profiling may provide additional information on treatment response. This requires validation with larger sample sizes and replication in an independent cohort. Reference [1] Ann Rheum Dis2013;72:844. 2. Arthritis Res Ther2016;18:235. Disclosure of Interests Stephanie Ling: None declared, Nisha Nair: None declared, Darren Plant: None declared, Hans-Dieter Zucht Employee of: Hans-Dieter Zucht is an employee of Protagen AG, Petra Budde Employee of: Petra Budde is an employee of Protagen AG, Peter Schulz-Knappe Shareholder of: Peter Schulz-Knappe is a shareholder of Protagen AG, Consultant for: Peter Schulz-Knappe is a consultant to Protagen AG, Employee of: Peter Schulz-Knappe was an employee of Protagen AG, MATURA Consortium: None declared, Anne Barton: None declared