Article first published online: 27 JUL 2012
Copyright © 2012 by the American College of Rheumatology
Arthritis & Rheumatism
Volume 64, Issue 8, pages 2808–2809, August 2012
How to Cite
Otomo, K. and Atsumi, T. (2012), Reply. Arthritis & Rheumatism, 64: 2808–2809. doi: 10.1002/art.34548
- Issue published online: 27 JUL 2012
- Article first published online: 27 JUL 2012
- Accepted manuscript online: 21 MAY 2012 10:54AM EST
To the Editor:
We thank Drs. Devilliers and Roggenbuck and their colleagues for their comments regarding our article on the aPL-S.
Devilliers et al had several concerns regarding the method we used to calculate the aPL-S. They first note that the meaning of 5 × exp([OR] − 5)/4, the formula for determining the aPL-S, is unclear. This formula was regressively designed after drawing the most suitable curve for the definition of aPL-S. For this reason, the formula itself was meaningless.
Second, the authors claim that setting the upper limit of the score to 20 is unreasonable. Our study included a small number of patients with high-titer IgG aPL (aCL, anti-β2GPI, and anti-PS/PT) as determined by enzyme-linked immunosorbent assay (ELISA). Consequently, the 95% confidence interval of the OR for these assays was very large. Thus, the OR value had little significance, and we decided to optimize the point up to 20.
Third, Devilliers and coauthors argue that each OR of a single aPL assay should be adjusted in relation to the others through a multivariate regression model. The calculation of the OR in this study was for the purpose of referring the value for distributing points, but not for evaluating the correlation between each aPL and the clinical manifestations of APS. Ultimately, we succeeded in establishing a correlation between the aPL-S as we defined it and the occurrence of thrombotic/obstetric events.
Finally, the authors suggest that the analysis should be limited to patients with aPL. We mentioned that the rate of thrombosis among patients with an aPL-S of ≥30 was significantly higher than that in patients with an aPL-S of 1–29 (P < 0.05 by log rank test) (Figure 2B in our article). Therefore, among aPL-positive patients, the positive correlation between aPL-S and thrombotic events remained significant.
In their editorial, Andreoli and Tincani noted that “an underlying systemic autoimmune disease may also increase the risk of thrombosis” (Andreoli L, Tincani A. Beyond the “syndrome”: antiphospholipid antibodies as risk factors [editorial]. Arthritis Rheum 2012;64:342–5). We believe that risk analysis of thrombosis, not only in patients with APS, but also in patients with systemic autoimmune disease, would be helpful in the management of these conditions. In any case, our aPL-S may not be the only significant score. The aPL-S should be further refined in future studies for better recognition of high-risk patients.
We also greatly appreciate the letter by Roggenbuck et al wherein the authors discuss their original study concerning the association of the aPL profile with the clinical phenotype of APS. They found that IgM aPL occurred more frequently in APS with arterial thrombosis. Accordingly, they concluded that aPL profiling may be a useful tool not only for prediction of risk of thrombosis but for the differentiation of clinical phenotypes of APS.
Unlike the data from the study by Roggenbuck and colleagues, our data did not show a relationship between the aPL-S and specific manifestations of APS, and we cannot exactly compare our results with theirs because patient backgrounds likely differed between the 2 studies. Our study was performed only in patients with autoimmune disease, which is an additional risk factor for thrombosis. In our cohort, levels of IgM aPL (aCL, anti-β2GPI, and anti-PS/PT) determined by ELISA were not related to arterial thrombosis. In addition to the aPL-S and the aPL profile, patient backgrounds, including information on each patient's underlying risk of thrombosis, would highly affect the phenotype of APS.
Further studies of aPL should be considered in order to establish a more robust scoring of the aPL profile by performing various aPL tests in patients with different backgrounds, rather than attempting to standardize each aPL test.
Kotaro Otomo MD*, Tatsuya Atsumi MD, PhD*, * Hokkaido University Graduate School of Medicine, Sapporo, Japan.