We read with interest the article by Montes et al (1). The authors describe an independent association between antibodies directed against citrullinated vimentin (anti–Cit-vimentin) and joint erosion prevalence, which they conclude may be of pathogenic relevance. As is mentioned in the Discussion of their article, our group previously did not find an association between anti–Cit-vimentin antibodies and joint destruction in a large study of patients with early rheumatoid arthritis (2). In an effort to reconcile these seemingly conflicting results, we attempted to compare the methods used in both studies. However, the method used to determine the presence of joint erosions in their study was not described. Was a specific validated radiographic score, such as the Larsen score or the Sharp/van der Heijde score used? Neither the method, nor the reliability of the method (whether one or two readers were used and what the interobserver correlation was), nor the cutoff that was applied, is described in the article, making the relevance of this erosive phenotype unclear (3).
Furthermore, it is unclear at what point during followup the radiographs were taken. Do the results represent a cross-sectional study of patients with variable disease duration, and might the increased prevalence of erosive arthritis in anti–Cit-vimentin–positive patients therefore have been due to the longer disease duration? Was there a longitudinal followup that would allow conclusions about the progression of radiographic damage?
Despite the many interesting aspects of antibodies directed against specific citrullinated proteins, such as anti–Cit-vimentin, we feel it is important to answer these questions. Bearing in mind that the in vitro system in which these autoantibodies are measured cannot mimic the way in which antigens are presented in vivo, we would like to urge caution before drawing conclusions about the possible pathogenic relevance of these autoantibodies.