Background Antibodies against citrullinated peptide antigens (ACPAs) strongly predict the risk of developing RA. ACPA also seem to predict a more severe RA phenotype.

Objectives Little is, however, known about the predictive value of specific ACPAs, or other autoantibodies (AAbs), in terms of response to methotrexate treatment.

Methods A total of 20 AAbs were measured by a custom-made microarray assay based on the ImmunoCAP ISAC system (Phadia AB, Uppsala, Sweden), in baseline samples from patients participating in two randomized controlled trials (RCTs): Swefot, n=340; Improved, n=450). Validation was performed in two large inception cohorts (EIRA linked to the SRQ register, n=1481; Glasgow, n=384). All patients started methotrexate in monotherapy as their first DMARD. Response to methotrexate was defined as EULAR response at first follow-up visit (3–6 months). The likelihood of EULAR good or moderate (vs. no) response was calculated by logistic regression adjusted for potential confounders, and the proportion of patients positive for each AAb is provided (adj. OR, 95%CI; % positive).

Results For the RCTs, 5 out of 20 AAbs were significantly associated with response in Improved, and 1 out of 17 in Swefot (fewer peptides on an earlier version of ISAC). Fib36_52_cit (citrullinated fibrinogen) was significantly associated with good/moderate response in both RCTs (Improved: adj.OR=2.27 (1.19–4.31), 32% positive; Swefot: adj.OR=1.87 (1.06–3.29), 44% positive). Fib591_cit (also citrullinated fibrinogen) was associated with good/moderate response in Improved (adj.OR=3.70 (1.27–10.76), 15% positive) and a non-significant trend was observed in Swefot (adj.OR=1.65 (0.89–3.06), 30% positive). Vim60_75_cit (citrullinated vimentin) was associated with response in Improved (adj.OR=1.85 (1.06–3.21), 46% positive) but not in Swefot (adj.OR=1.06, 63% positive). For CEP_1, a borderline significant association was observed in Improved (adj.OR=1.70 (0.96–3.00), 40% positive) but not in Swefot (adj.OR=0.90, 51% positive). Further, AAbs against citrullinated peptides named pept_Z2 and pept_5 were associated with response in Improved (adj.OR=1.81 (1.02–3.23), 41% positive; and 2.25 (1.27–4.01), 44% positive, respectively), while no association was observed in Swefot (adj.OR=0.98 and 1.17). Interestingly, none of the above associations were replicated in the inception cohorts (EIRA-SRQ, Glasgow), nor did we observe any trend for a dose-response association with response for the total number of positive AAbs.

Conclusions Whilst two AAbs against citrullinated fibrinogen predict response to methotrexate in both early RA RCTs, neither these nor any other of the AAbs tested predicted response in the real-life population-based inception cohort setting. The reasons for the discrepant results are not yet clear, but our results highlight the importance of testing in multiple study populations and settings when identifying predictors to be used for clinical prediction.

Disclosure of Interest None declared