Association of smoking with triple concordant seropositivity in RA patients, and with rheumatoid factor in the remaining patients

Abstract

Background: The contribution of cigarette smoking to the risk of rheumatoid arthritis (RA) is larger for seropositive than for seronegative patients. Current pathogenic models explain this difference via the production of anti-citrullinated protein antibodies (ACPA) induced by protein citrullination in the lungs as a consequence of smoking. However, there are inconsistent results with this model including the van Wesemael et al . report, which found an association of smoking with the concurrent presence of ACPA, rheumatoid factor (RF) and anti-carbamylated protein antibodies (ACarPA), rather than with ACPA.

Objectives: We aimed to explore the relationship between smoking and the presence of FR, ACPA and ACarPA and their combinations in patients with RA.

Methods: Six cohorts with 2253 RA patients were obtained to replicate the van Wesemael et al . results. After replication, they were combined with the three cohorts from that study comprising 2238 patients to a total of 4491 patients. The results from the different cohorts were combined by fixed-effects meta-analysis. Most analyses involved comparisons between two levels of smoking exposure, never and ever smokers, considering the RF-/ACPA-/ACarP-patients as the reference.

Results: Analysis of the replication set confirmed the predominant association of smoking with the concurrent presence of the three RA autoantibodies (OR = 1.99, P = 2.5 x 10-8), whereas the patients with one or two antibodies were not different from the patients without antibodies (OR = 1.22, p = 0.4 and OR = 1.22, p = 0.3, respectively). The combined meta-analysis with all the cohorts clarified the dominant association of smoking with the triple concordant seropositivity (OR = 2.00, p = 4.4 x 10-16). In contrast, there was no association with the patient bearing one autoantibody (OR = 1.12, p = 0.4), and an in-between association with the concurrent presence of two antibodies (OR = 1.26, p = 0.009). Moreover, the association with the presence of three antibodies was significantly larger than with two antibodies (OR = 1.54, p = 1.4 x 10-6), whereas the association with the presence of two antibodies was not larger than with one autoantibody (OR = 1.11, p = 0.3). In the patients remaining after exclusion of the triple seropositive, the smokers were exclusively associated with the RF+ (OR = 1.28, p = 0.03 and OR = 1.30, p = 0.004 in the double and single positive patients, respectively). This association was independent of the reference, either the triple seronegative (OR = 1.29, p = 0.001) or all the RF-patients (OR = 1.32, p = 0.0002).

Conclusion: Smoking increases RA susceptibility by promoting pathways leading to the concurrent presence of the three RA autoantibodies and, in its defect, to the production of RF. These actions are not covered by current pathogenic models and suggest that smoking accelerates epitope spreadi