We appreciate the interest in our article shown by van der Woude and colleagues. Their letter deals with the specific association we found between anti–Cit-vimentin antibodies and the presence of joint erosion in patients with rheumatoid arthritis (RA). This finding is at odds with the results obtained by van der Woude et al in a previous study (1). The letter includes several questions and a recommendation about aspects of the design and interpretation of our study that we would like to address.
We were aware of the conflicting results, and we cited and discussed them in our article. Our group interpreted the contradictory findings as resulting from the clear differences between the disease evolution (early versus established RA), the time of antibody assessment (baseline or after a long disease duration), and the genetic and environmental factors pertaining to the patients in the 2 studies. Any of these differences could justify the contradictory results, as we indicated in our article. This interpretation was reflected in our qualification of the association between anti–Cit-vimentin and joint erosions as observed “in Spanish patients with longstanding disease.”
With respect to the questions posed by van der Woude et al, we appreciate the opportunity to provide pertinent details of our methods and analysis of our findings. Assessment of joint erosions was performed by a single reader (an experienced rheumatologist who was a coauthor) following only a qualitative distinction between presence/absence of definite bone erosions. The radiographs we examined were the most recent available radiographs of the hands and feet of each patient. Our results correspond to a cross-sectional evaluation of patients with established disease. Consistent with these methods, no comment on quantitative scores or on the progression of radiographic damage was included in the text. In addition, we stated that other studies, including the study by van der Woude and colleagues (1), “are more valid than ours for assessing the predictive value of [anti–citrullinated protein antibodies] ACPAs.” Finally, the association of anti–Cit-vimentin antibodies, including the time since disease onset among the covariates, was determined by multivariate analysis as we described in the article. In this way, variable disease duration was accounted for in the analysis. Therefore, confounding by this factor was excluded in the association between anti–Cit-vimentin antibodies and joint erosions.
Van der Woude and colleagues recommend caution before drawing conclusions about pathogenic mechanisms based on antibody determinations. We agree with this recommendation, and we applied it throughout our study. We did not make inferences about mechanisms or about the pathogenic relevance of our findings; only the possibility that the findings could play a pathogenic role was signaled. In addition, we further applied caution by reporting only the findings of subanalyses that were supported by data from previous studies, including the association between anti–Cit-vimentin and joint erosions (2). Only studies that directly relate antibodies to disease mechanisms, as has recently been reported (3), or studies of animal models will permit conclusions about pathogenesis.
We appreciated the opportunity to review our methods and data analysis, and we think our findings are unchanged by this additional assessment. We hope to have resolved any doubts about the specific associations of anti–citrullinated α-enolase peptide and anti–Cit-vimentin antibodies with the clinical and genetic features of patients with RA. We further hope to have demonstrated the value of pursuing this analysis for a better understanding of RA. In this difficult task, independent and collaborative studies from multiple teams will be needed.