To the Editor:

A recent report by Vaglio et al published in Arthritis & Rheumatism described one of the first genetic association studies of Churg-Strauss syndrome (CSS) (1). Associations of selected HLA loci were investigated in 48 Italian CSS patients and 350 healthy controls. While no association with TNFA (the gene for tumor necrosis factor α) or with HLA–A or HLA–B alleles or genotypes was identified, significant associations of CSS with HLA–DRB1*07 (corrected P [Pcorr] = 0.0042) and the linked HLA–DRB4 gene (Pcorr = 0.0002) encoding the HLA–DR53 antigen supertype were detected. Conversely, HLA–DRB3, the gene for the HLA–DR52 antigen, was underrepresented in CSS patients compared with controls (Pcorr = 0.0069). For all association studies, confirmation of the results in independent cohorts is mandatory in order to evaluate the true significance of any finding. This requirement applies especially to rare diseases, for which only small sample sizes are usually available, possibly increasing the likelihood of stochastic errors. We therefore investigated HLA–DRB1 associations in an independent CSS cohort comprising more than twice as many patients as those studied by Vaglio and colleagues. Low-resolution typing of HLA–DRB1 was performed as described elsewhere (2). Subsequently, further subtyping was performed using additional primer systems established earlier (3). We studied 102 German patients with CSS diagnosed according to the criteria of the American College of Rheumatology (4) and 341 healthy German controls.

In general, our findings were similar to those of Vaglio et al. While the overall allele distribution was significantly different between patients and controls (P = 0.0036), the HLA–DRB1*07 allele was only moderately more frequent in the CSS cohort (P = 0.046). However, we found a stronger association with the HLA–DRB1*04 allele (P = 0.0028). On the other hand, DRB1*13 was underrepresented in the CSS cohort (P = 0.019) (Figure 1A). As in the study of Italian subjects (1), we calculated the frequency of HLA–DRB4 (from DRB1*04, *07, and *09 data) and HLA–DRB3 (from DRB1*03 and *11–*14 data). While HLA–DRB4 was more common in CSS patients (36.8% versus 23.7%; odds ratio [OR] 1.87, 95% confidence interval [95% CI] 1.34–2.62, P = 0.0002), DRB3 was more frequent in controls (41.0% versus 29.9%; OR 0.61, 95% CI 0.44–0.86, P = 0.004). As can be deduced from our subtyping data, these effects were mainly due to the risk alleles DRB1*04 and *07 and the “protective” alleles DRB1*03 and *13.

thumbnail image

Figure 1. HLA–DRB1 and HLA–DPB1 allele frequencies in patients with Churg-Strauss syndrome (CSS) compared with healthy controls. Allele frequencies were analyzed using the chi-square statistic and contingency tables. A, HLA–DRB1. The overall distribution of DRB1 alleles differed significantly between CSS patients (n = 102) and controls (n = 341) (P = 0.0036). This was mainly due to significant differences in the distribution of the alleles DRB1*04 (20.1% in CSS patients versus 11.9% in controls; odds ratio [OR] 1.86, 95% confidence interval [95% CI] 1.23–2.82, P = 0.0028), DRB1*07 (15.7% in patients versus 10.6% in controls; OR 1.57, 95% CI 1.00–2.46, P = 0.046), and DRB1*13 (6.9% in patients versus 12.9% in controls; OR 0.50, 95% CI 0.28–0.90, P = 0.019). B. HLA–DPB1. Neither the overall distribution nor the frequency of any specific allele differed significantly between CSS patients (n = 102) and controls (n = 369). All alleles with a frequency of <5% were combined into one category.

Download figure to PowerPoint

Some authors have proposed that CSS and Wegener's granulomatosis (WG) represent different forms of a pathogenetically very similar disease group termed “antineutrophil cytoplasmic antibody (ANCA)–associated small vessel vasculitides” (5). However, recent observations that the association of CSS with ANCA is much weaker than the association of WG with ANCA (5) have raised concerns about the use of this nomenclature. Our data further challenge this view, indicating that CSS and WG have distinct genetic backgrounds. A striking association of WG with HLA–DPB1 has been reproducibly demonstrated (6). Although HLA–DPB1 and HLA–DRB1 are not in very strong linkage disequilibrium, we additionally investigated HLA–DPB1 alleles in our CSS cohort, in order to exclude the possibility that this locus has an impact on CSS manifestation. Nineteen DPB1 alleles were genotyped in our CSS cohort and in 369 controls, using methods described previously (6). Allele frequencies in the 2 groups were virtually identical (Figure 1B). Thus, the HLA association in CSS is not due to linkage disequilibrium with the HLA–DPB1 locus. Nonetheless, the HLA region contains numerous suitable candidate genes for CSS, and the question of whether one or several of these are the basis for the findings reported by Vaglio et al and confirmed by us requires further investigation.

S. Wieczorek MD*, B. Hellmich MD†, W. L. Gross MD†, J. T. Epplen MD‡, * Ruhr University, Bochum, Germany, † University Hospital of Schleswig-Holstein, and Rheumaklinik Bad Bramstedt Luebeck, Germany, ‡ Ruhr University, Bochum, Germany.