Glucocorticoid-sparing effect of first-year anti-TNFa treatment in rheumatoid arthritis (CORPUS cohort)

Objective Anti-TNFa agents are indicated in selected patients with rheumatoid arthritis (RA) who respond inadequately to methotrexate and particularly when glucocorticoids are mandatory. We evaluated whether a glucocorticoid-sparing effect occurred during the first year of anti-TNF-alpha therapy. Methods Between 2007 and 2009, the French multicentre, longitudinal, prospective, observational, population-based CORPUS cohort included biologic-naive patients with inflammatory joint disease. Patients with active RA treated with glucocorticoids were included. Patients who received at least one anti-TNF-alpha injection during follow-up were compared to anti-TNF-a non-users. Results Among the 205 patients, 76.1% were women, mean disease duration was 7.7 +/- 8.3 years, mean DAS28 was 5.2 +/- 1.3, mean follow-up was 13.1 +/- 2.8 months, and mean prednisone dose was 9.9 +/- 9.6 mg/day. The 75 (36.6%) anti-TNF-alpha recipients were younger, had a longer RA duration, more often tested positive for rheumatoid factor and anti-citrullinated peptide antibody, more often received previous DMARDs, received a higher methotrexate dosage, had fewer intra-articular glucocorticoid injections at baseline and were more often followed by hospital practitioners than non-recipients. Mean prednisone dosage decreased from 11.8 +/- 12.7 to 5.9 +/- 9.7 mg/day in recipients and from 8.7 +/- 7.1 to 5.0 +/- 4.4 mg/day in non-recipients. Prednisone was stopped more often among recipients (21/59, 35.6%) than among non-recipients (16/94, 17.0%) (p=0.01). By multivariate analysis, factors independently associated with lower prednisone requirements were baseline daily prednisone dosage, a CRP > 10 mg/l and not to be followed by an office-based practitioner. Conclusion This study showed a significantly higher glucocorticoid discontinuation rate among anti-TNF-a recipients than among non-recipients. However, the glucocorticoid-sparing effect was small and not observed by multivarite analysis.