Functionally relevant variations of the interleukin-10 gene associated with antineutrophil cytoplasmic antibody–negative Churg-Strauss syndrome, but not with Wegener's granulomatosis

Authors


Abstract

Objective

Wegener's granulomatosis (WG) and Churg-Strauss syndrome (CSS) belong to the heterogeneous group of antineutrophil cytoplasmic antibody (ANCA)–associated vasculitides. Current understanding of their pathogenesis and genetic background is limited. Expression levels of interleukin-10 (IL-10), a potent and pleiotropic cytokine, are largely determined by variations in the gene encoding the IL-10 precursor. This study was undertaken to determine the impact of IL10 polymorphisms on the pathogenesis of both WG and CSS in large cohorts.

Methods

Three single-nucleotide polymorphisms (SNPs) tagging the promoter haplotypes of the IL10 gene (IL10 −3575, IL10 −1082, and IL10 −592) were analyzed in 403 patients with WG and 103 patients with CSS as well as 507 matched control subjects from Germany. In addition, 3 informative SNPs in other parts of IL10 were genotyped.

Results

None of the markers or their haplotypes was associated with WG or any of its subgroups classified according to ANCA status, sex, or presence of further WG genetic risk factors. In contrast, the IL10 −3575/−1082/−592 TAC haplotype, part of the extended ancient haplotype IL10.2, was highly significantly associated with ANCA-negative CSS (χ2 = 19.14, P = 0.000012, corrected P = 0.0003, odds ratio 2.16, 95% confidence interval 1.52–3.06).

Conclusion

These findings challenge those from previous studies of IL10 in WG and provide further evidence that CSS and WG have distinct genetic backgrounds. Because the IL10.2 haplotype has been correlated reproducibly with increased IL10 expression, the possible role of IL-10 in the pathogenesis of ANCA-negative CSS needs to be further elucidated.

Wegener's granulomatosis (WG) and Churg-Strauss syndrome (CSS) belong to the group of systemic vasculitides associated with the presence of antineutrophil cytoplasmic antibodies (ANCAs) in the serum (referred to as ANCA-associated vasculitides [AAVs]). Both disorders are rare, with prevalences of 2.4/100,000 person-years for WG and only 1/100,000 person-years for CSS in Western European populations (1, 2). Although both diseases share some clinical manifestations, and sometimes WG can also be associated with mild eosinophilia, CSS and WG differ clinically and apparently pathogenetically in many aspects, e.g., preferential involvement of the cardiac system in CSS and of the renal system in generalized WG. The cytokine profile of each disease is distinct, with a predominant Th2 (interleukin-4 [IL-4], IL-5) profile in CSS (3, 4) and a predominant Th1 (interferon-γ [IFNγ]) and Th17 profile in WG (5, 6).

ANCAs have been reported to be present in 10–50% of patients with CSS. More recently, 2 large studies demonstrated a prevalence of ANCAs of ∼40% in patients with active CSS (7, 8), which is a much lower ANCA prevalence compared with that in patients with active generalized WG (prevalence of 90%). Moreover, ANCA-positive patients with CSS usually have myeloperoxidase ANCAs, whereas patients with WG have proteinase 3 ANCAs. These clinical and pathophysiologic aspects challenge the hypothesis that CSS and WG represent phenotypic variants of the same entity (small- vessel vasculitis) (9) and support the notion that they may be unique diseases that share ANCA positivity to some degree (10).

Both disorders, WG and CSS, appear to be multifactorial diseases that are associated with a complex interaction of extrinsic and genetic factors. In view of the concept of AAVs, the question is whether CSS and WG could share a similar genetic background that results in ANCA formation. Whereas variations in numerous candidate genes have been tested for associations with WG (for review, see refs. 11 and12), only a few genetic studies have been carried out in patients with CSS. Accordingly, some well-defined genetic risk factors for WG have been identified, e.g., HLA–DPB1*0401 and the PTPN22 allele 620W (13, 14). Two recent studies showed that certain alleles of the HLA–DRB1 gene are significantly associated with CSS (15,16), an association that has not been identified in patients with WG. In contrast, frequencies of alleles in polymorphisms of the tumor necrosis factor α (TNFα) promoter did not differ between patients with CSS and controls (15). Given the distinct cytokine profiles of both diseases, we hypothesized that CSS and WG may have distinct genetic variations in cytokines other than TNFα.

IL-10 is a pleiotropic cytokine with complex and multiple effects in immune modulation. It is produced by numerous immune cells, including B lymphocytes, Th2 cells, monocytes, macrophages, and dendritic cells. The properties of IL-10 comprise enhancing effects on B lymphocyte activation and stimulation, which has previously been associated with autoimmune disorders such as systemic lupus erythematosus (SLE) (17, 18). Conversely, IL-10 has antiinflammatory effects that are mediated by down-regulation of the Th1 cytokines TNFα, IL-1, IL-8, and IFNγ (19, 20). IL-10 production is largely (∼50–70%) determined by genetic factors (21, 22). Elevated levels of IL-10 in plasma have been found in CSS, but not in WG (23, 24). Moreover, there is evidence, which has been reproduced, to indicate that IL10 expression is influenced by cis regulatory elements of the gene for IL-10 precursor messenger RNA on chromosome 1q31–q32, i.e., by certain configurations of promoter variations (25–28). Since polymorphisms of the IL10 promoter have been shown to be in strong linkage disequilibrium (LD), it is possible to characterize the architecture of IL10 by analyzing exclusively an informative subset of markers.

In studies demonstrating their clinical relevance, haplotypes of IL10 have been shown to be associated with different immune system disorders, such as SLE (29), rheumatoid arthritis (27), and giant cell arteritis (30), and have also been associated with the outcome of various infections (for review, see ref. 31) and graft-versus-host disease (32). Variations in IL10 have already been analyzed in patients with WG, but not in patients with CSS, and yet only preliminary results have been obtained in WG, because of the limited size of the cohorts analyzed (33–36). It was therefore the aim of the present study to evaluate the impact of functionally relevant IL10 polymorphisms on the pathogenetic mechanisms of WG and CSS in well-characterized patient cohorts, with sufficiently large sample sizes to adjust for the low prevalence of each disorder.

PATIENTS AND METHODS

Subjects

One hundred three patients with CSS (45 male, 58 female) and 403 patients with WG (201 male, 202 female) whose diagnoses met the American College of Rheumatology criteria (37, 38) and the Chapel Hill consensus conference criteria (39) were included in this study. All patients were recruited from the interdisciplinary vasculitis center at the University of Luebeck/Rheumaklinik Bad Bramstedt in Germany. The patients were asked to report their ancestry, and all reported having German descent for at least 2 generations. Other potential genetic risk factors have already been analyzed by our group in large parts of the WG cohort (13, 14, 40, 41). Most of the patients with WG (392 [97.3%] of 403) had been genotyped for the well-defined risk allele of HLA–DPB1 (HLA–DPB1*0401), as has been previously described (13).

Because some studies have shown that the genetic background of WG may be different between ANCA-positive and ANCA-negative patients (14, 42), the patients were also analyzed separately by ANCA status. The presence of ANCAs was tested as described previously (40). Among the 399 patients with WG tested, 364 (91.2%) were found to have positive titers of classic ANCA (cANCA)/perinuclear ANCA (pANCA). Among the patients with CSS, the ANCA status was known for 100 (97.1%), of whom 25 (25.0%) had detectable titers of pANCA or cANCA, respectively. A clinical summary of the patients in the WG cohort has been provided previously (14), and a brief characterization of the CSS cohort is given in Table 1.

Table 1. Clinical features of the cohort of patients with Churg-Strauss syndrome (n = 103)*
  • *

    Except where indicated otherwise, values are the number (%). ANCAs = antineutrophil cytoplasmic antibodies.

  • Defined as the presence of at least 1 of the following symptoms: (night) sweats, fever, weight loss.

Age at onset, years 
 Mean ± SD45.32 ± 15.81
 Median (range)45 (15–86)
Male/female45/58 (43.7/56.3)
Positive for ANCAs25/100 (25.0)
Clinical manifestation 
 Lung involvement75 (72.8)
 Constitutional symptoms70 (68.0)
 Ear/nose/throat involvement68 (66.0)
 Peripheral neuropathy57 (55.3)
 Arthralgia/arthritis50 (48.5)
 Cardiac involvement41 (39.8)
 Skin involvement33 (32.0)
 Gastrointestinal involvement25 (24.3)
 Skeletal muscle involvement17 (16.5)
 Renal involvement15 (14.6)
 Ocular involvement13 (12.6)
 Central nervous system involvement12 (11.7)

Five hundred seven healthy blood donors from Germany (274 male, 233 female) were used as controls. This control group was previously described in an earlier genetic study (43), and ancestry of the control subjects was evaluated in the same manner as for the patient group. This control group has already been used in numerous genetic association studies of (auto)immune disorders, such as multiple sclerosis (40) and WG (13, 14, 40, 41). For the present study, the principles of ethical medical research involving human subjects were followed, as defined in the Declaration of Helsinki. The study design was approved by the local ethics committee at the University of Luebeck (AZ 05-219 and AZ 06-087).

Selection of single-nucleotide polymorphisms (SNPs)

A detailed description of the architecture of the IL10 promoter haplotypes has been provided previously (29, 44). In order to determine the IL10 promoter haplotypes, 3 informative SNPs were genotyped: rs1800872, rs1800896, and rs1800890. The SNP rs1800872 (herein designated IL10 −592, although the nucleotide position differs slightly from that reported in the current genome database assembly as of March 2006) and the SNP rs1800896 (herein designated IL10 −1082) were chosen to identify the proximal promoter haplotype (consisting of promoter SNPs IL10 −592, −819, −1082, and −1330). In addition, the SNP rs1800890 (IL10 −3575) was chosen to define the distal haplotype (consisting of IL10 −2763, −2849, and −3575).

To evaluate the genetic variability of the other parts of the IL10 gene (i.e., exons, introns, and 3′ sequences), 3 further SNPs were analyzed: rs1518111 (intron 2), rs3024496 (exon 5′/3′-untranslated region), and rs6673928 (3′ of IL10). These SNPs were chosen using the tagging algorithm implemented in Haploview software (version 4.0) (45), and all 3 served as tagSNPs for all database SNPs within a ∼12-kilobase genomic region around IL10 and having a minor allele frequency of ≥0.1. Pairwise tagging was performed using an r2 threshold of 0.8. With the use of these selection criteria, 26 SNP alleles of the IL10 genomic region were tagged with a mean r2 of 0.987.

Genotyping

IL10 −592, IL10 −1082, IL10 −3575, and rs6673928 were genotyped using polymerase chain reaction (PCR) and differential enzymatic analysis with the restriction fragment length polymorphism method (primer sequences and PCR conditions are shown in Table 2). All PCRs were run on Robocycler 96-well thermal cyclers (Stratagene, La Jolla, CA) using HotStar polymerase (Qiagen, Hilden, Germany) according to the manufacturer's instructions, in the following steps: 15 minutes of initial denaturation at 95°C, 35 cycles of 1 minute at 95°C, 1 minute at annealing temperature, and 1 minute at 72°C, and a final extension period of 10 minutes. Enzymatic digestions were performed as recommended by the supplier (Qiagen). Fragments were separated on 2.5% agarose gels, stained with ethidium bromide, and visualized under ultraviolet light.

Table 2. Primer sequences and restriction enzymes used for polymerase chain reaction (PCR) and restriction fragment length polymorphism analyses
IL10 SNPForward primer (5′–3′)Reverse primer (5′–3′)Annealing temperature, °CRestriction enzymeLength of PCR product, bpAllele/fragments
  • *

    A 17-mer oligonucleotide tail from M13 phage sequence (5′-GTAAAACGACGGCCAGT-3′) was added to the 5′ in order to increase the length difference between the restriction fragments. Since no a priori restriction site was present at the single-nucleotide (SNP) position, a mismatch primer containing a false nucleotide was used to create a potential restriction site. The mismatch nucleotide is underlined.

−3575 (rs1800890)TGGGCTTCTTGATGAGTGAGACCTCAAGCCCAGATAGT55Apo I251A/251 bp
      T/188 bp + 63 bp
−1082 (rs1800896)AACACTACTAAGGCTCCTTTGGGA*CAAGGAAAAGAAGTCAGGATTCCATGGA55Xag I119A/119 bp
      G/37 bp + 82 bp
−592 (rs1800872)AAGCTTTCAGCAAGTGCAGAATACCCAAGACTTCTCCTT49Rsa I301C/301 bp
      A/153 bp + 148 bp
+8545 (rs6673928)GATTTCATTTGGCAGTTATAGCAAAACGTCTCCCCAGGTGC60Hinf I451T/451 bp
      G/164 bp + 287 bp

The remaining SNPs (rs1518111 and rs3024496) were analyzed using commercially available TaqMan genotyping assays on 50 ng genomic DNA with 1× TaqMan Genotyping Mastermix (Applied Biosystems, Foster City, CA) in accordance with the protocol recommended by the manufacturer, in the following steps: 10 minutes of initial denaturation, followed by 40 cycles of 15 seconds at 95°C and 1 minute at 60°C. Fluorescence signals were measured on an iCycler device (Bio-Rad, Munich, Germany) after each PCR cycle. Results were analyzed with iCycler IQ software (version 3.1; Bio-Rad).

Statistical analysis.

Genotypes were recorded in Linkage format. Each SNP was analyzed separately for statistical associations of allele and genotype distributions, using the chi-square test on contingency tables and a significance threshold of 0.05. Subsequently, haplotype distribution was evaluated using the maximum-likelihood method and chi-square test, implemented in Haploview. P values were corrected for multiple comparisons using the Bonferroni method, as appropriate.

In addition, Hardy-Weinberg equilibrium was checked for each cohort and each marker using the chi-square test on contingency tables, with 1 degree of freedom. LD between each marker pair was determined on the basis of r2 values, using Haploview. Patients were also analyzed according to stratified subgroups of sex (in both the WG and CSS groups), HLA–DPB1*0401 carriership (in the WG group only), and ANCA status (i.e., ANCA-positive versus ANCA-negative, in both the WG and CSS groups). Power analyses were performed using Quanto (version 1.1.1) (46).

RESULTS

In our detailed analyses of IL10 SNPs in CSS patients, WG patients, and controls, no statistically significant deviation from Hardy-Weinberg equilibrium was detectable for the distribution of any of the SNPs in any of the cohorts (except for rs1800890 in ANCA-positive patients with CSS [see Appendix A, available on the Arthritis & Rheumatism Web site at http://www.mrw.interscience.wiley.com/suppmat/0004-3591/suppmat/]). There was no significant deviation of allele and genotype frequencies for any of the SNPs studied in the control group as compared with the corresponding data from the HapMap database (see Appendix B, on the Arthritis & Rheumatism Web site).

Whereas analysis of the distribution of alleles and genotypes of IL10 −592 did not yield any significant differences between the WG patients, CSS patients, and controls, analysis of the distribution of IL10 −1082 alleles and genotypes revealed significant differences in the CSS cohort, but not in the WG cohort (Figure 1) (see also Appendix C, on the Arthritis & Rheumatism Web site). Similar, but less significant findings were obtained for IL10 −3575, located upstream on chromosome 1q31–q32 (see Appendix C, on the Arthritis & Rheumatism Web site). For each marker, associations were also tested according to carriership of a certain allele, i.e., carriership compared with noncarriership in patients and controls (or 1 homozygous genotype versus the sum of the other 2), but these analyses did not reveal any additional major information (see Appendix C, on the Arthritis & Rheumatism Web site). Moreover, allele, genotype, and haplotype distributions did not differ significantly between male and female subjects in any of the cohorts (results not shown).

Figure 1.

Distribution of alleles and genotypes of the IL10 −1082 (rs1800896) promoter polymorphism. Frequencies of alleles (G, A) and genotypes (GG, GA, AA) did not differ significantly between patients with Wegener's granulomatosis (WG) and controls, whereas the overall genotype distribution was significantly different between antineutrophil cytoplasmic antibody (ANCA)–negative patients with Churg-Strauss syndrome (CSS) and controls (χ2 = 9.17, degrees of freedom [df] 2, P = 0.010, corrected P [Pcorr] = 0.02); accordingly, allele frequencies also differed significantly (χ2 = 9.67, 1 df, P = 0.002, Pcorr = 0.004, odds ratio 1.75, 95% confidence interval 1.23–2.50). In ANCA-positive patients with CSS, there was an overrepresentation of heterozygous individuals, and despite the small sample size of this group (n = 25), the increased frequency of heterozygotes was a significant deviation from Hardy-Weinberg equilibrium (P = 0.02).

Strong pairwise LD was observed for the 2 nonassociated SNPs, rs1518111 and IL10 −592 (r2 = 0.84), as well as for the associated SNPs, IL10 −1082 and rs3024496 (r2 = 0.93). Furthermore, the associated IL10 SNP −3575 was in moderate LD with both other associated SNPs, with an r2 value of 0.65 for LD with IL10 −1082 and an r2 value of 0.68 for LD with rs3024496. Pairwise LD for all other markers was considerably weaker (r2 ≤ 0.42) (Figure 2).

Figure 2.

Schematic illustration of the investigated IL10 genomic region. Nucleotide positions are given according to the database assembly as of March 2006. The direction of IL10 transcription is from telomere to centromere, i.e., from right to left. The linkage disequilibrium (LD) plot was created with Haploview software, version 4.0. The r2 values, indicating the pairwise LD between 2 single-nucleotide polymorphisms (SNPs), are given in each diamond, and the level of pairwise LD is intensity coded, with darker shades of gray corresponding to higher degrees of LD. The genomic structure of the IL10 gene is presented to scale, with thin lines representing introns, medium lines representing untranslated exon sequences, and thick lines representing translated exon sequences. Asterisks indicate the SNPs that had allele/genotype distributions that were significantly different (after Bonferroni correction) between antineutrophil cytoplasmic antibody–negative patients with Churg-Strauss syndrome and controls.

When the distribution of haplotypes of the 3 promoter SNPs was calculated for the 2 patient groups and for the control group, the haplotype TAC was highly significantly overrepresented in the CSS patients as compared with the WG patients (χ2 = 10.699, P = 0.0011, corrected P [Pcorr] = 0.0086 [n = 8 comparisons], odds ratio [OR] 1.70, 95% confidence interval [95% CI] 1.23–2.34) and as compared with the controls (χ2 = 12.115, P = 0.0005, Pcorr = 0.012 [n = 8 comparisons], OR 1.73, 95% CI 1.27–2.37) (Figure 3). Detailed results for all of the haplotype comparisons are given in Table 3. The associated TAC haplotype can be unambiguously assigned to the extended IL10.2 haplotype, which is 1 of 4 major ancient extended IL10 haplotypes (43).

Figure 3.

Distribution of the IL10 promoter haplotypes among patients with Wegener's granulomatosis (WG), patients with Churg-Strauss syndrome (CSS), and controls. Maximum-likelihood estimates of the haplotype frequency distributions were calculated from the 3 promoter single-nucleotide polymorphisms (−3575/−1082/−592) in WG patients, CSS patients, and controls. Promoter haplotypes have been assigned to the ancient extended IL10 haplotypes (IL10.1–IL10.4) according to the study by Eskdale et al (44). Distribution of the haplotypes did not differ significantly between WG patients and controls. The TAC (IL10.2) haplotype was highly significantly associated with antineutrophil cytoplasmic antibody (ANCA)–negative CSS (χ2 = 19.14, P = 0.000012, corrected P = 0.0003, odds ratio 2.16, 95% confidence interval 1.52–3.06).

Table 3. Distribution and association of IL10 promoter haplotypes in patients with Wegener's granulomatosis (WG) and patients with Churg-Strauss syndrome (CSS)*
 IL10 −3575/−1082/−592 haplotype
AGC (IL10.1)TAC (IL10.2)TGC (IL10.3)TAA (IL10.4)
  • *

    NS = not significant; OR = odds ratio; 95% CI = 95% confidence interval; ANCA = antineutrophil cytoplasmic antibody.

  • Corrected P (Pcorr) values are for comparisons of 4 haplotypes in 6 patient groups/subgroups (n = 24 comparisons). All comparisons are with controls.

Controls, frequencies of haplotype (n = 507)0.3960.2720.0930.231
WG patients (n = 403)    
 All    
  Frequency of haplotype0.3780.2760.0960.238
  χ20.610.060.020.09
  PNSNSNSNS
  Pcorr
  OR (95% CI)0.93 (0.77–1.22)1.03 (0.84–1.26)1.04 (0.76–1.42)1.04 (0.83–1.29)
 ANCA-positive (n = 364)    
  Frequency of haplotype0.3760.2740.0970.244
  χ20.650.010.030.38
  PNSNSNSNS
  Pcorr
  OR (95% CI)0.92 (0.76–1.12)1.01 (0.82–1.25)1.03 (0.74–1.42)1.07 (0.86–1.34)
 ANCA-negative (n = 35)    
  Frequency of haplotype0.3850.3140.0860.171
  χ20.030.580.051.36
  PNSNSNSNS
  Pcorr
  OR (95% CI)0.96 (0.58–1.58)1.23 (0.73–2.07)0.91 (0.38–2.15)0.69 (0.36–1.30)
CSS patients (n = 103)    
 All    
  Frequency of haplotype0.3290.3930.0640.213
  χ23.2512.1151.790.32
  PNS0.0005NSNS
  Pcorr0.012
  OR (95% CI)0.75 (0.55–1.03)1.73 (1.27–2.37)0.65 (0.36–1.89)0.90 (0.63–1.30)
 ANCA-positive (n = 25)    
  Frequency of haplotype0.4390.2590.0610.240
  χ20.400.040.650.02
  PNSNSNSNS
  Pcorr
  OR (95% CI)1.20 (0.68–2.13)0.94 (0.49–1.79)0.62 (0.19–2.02)1.05 (0.54–2.04)
 ANCA-negative (n = 75)    
  Frequency of haplotype0.2860.4460.0670.200
  χ26.6519.141.160.75
  P0.010.000012NSNS
  PcorrNS0.0003
  OR (95% CI)0.61 (0.42–0.89)2.16 (1.52–3.06)0.69 (0.35–1.36)0.83 (0.54–1.27)

In the patients with WG, additional stratification by ANCA status or by HLA–DPB1*0401 carriership did not reveal any significant differences in haplotype associations (Table 3) (see also Appendices D and E on the Arthritis & Rheumatism Web site). In contrast, when analyzing the patients in the CSS cohort by ANCA status, we found that the association was restricted to the ANCA-negative subgroup (Table 3) (see also Appendix A on the Arthritis & Rheumatism Web site). Although the group of ANCA-negative patients with CSS was considerably smaller (n = 75) than the entire CSS cohort (n = 103), lower P values were obtained in comparing the ANCA-negative CSS group with controls than were obtained in comparing the entire CSS cohort with controls, even after quite conservative Bonferroni correction for multiple testing (χ2 = 19.14, P = 0.000012, Pcorr = 0.0003 [n = 24 comparisons], OR 2.16, 95% CI 1.52–3.06). Similarly, the association of the individual SNPs was only demonstrable in ANCA-negative patients with CSS, but not in ANCA-positive patients with CSS (see Appendix A on the Arthritis & Rheumatism Web site).

Several previous studies on IL10 promoter polymorphisms in WG revealed significant ORs, especially for the −1082AA genotype. We therefore performed a power analysis for this marker in our WG cohort. ORs for this marker have previously been reported to be 2.82 (35) and 0.5 (33). Even with the latter OR, a statistical power of 97.83% could be achieved in our WG sample, when the following parameters, in addition to an OR of 0.5, are used: WG prevalence of 2.4/100,000 person-years, sample size of 403 WG patients versus 507 controls, minor allele frequency of 0.49 (as found in controls), two-sided test, significance threshold of 0.05, and recessive risk model (reflecting the putatively associated −1082AA genotype). When an OR of 2.82 is used along with the other parameters, the statistical power would be virtually 100% in our complete WG sample. Analogous calculations for the ANCA-positive WG cohort (n = 364) would result in virtually identical power values from ORs of 2.82 and 0.5 (97.11% and 100%, respectively), whereas the corresponding power values for the smaller ANCA-negative WG cohort would be considerably lower (31.69% and 81.52%, respectively).

However, of greater interest in the present context is the power calculation for the association of the TAC haplotype with ANCA-negative CSS (OR 2.16). Because of the small sample size of the ANCA-positive patients with CSS (n = 25), power yields would be considerably smaller for this nonassociated subgroup; assuming the same OR of 2.16, power would only reach 72.79% in a multiplicative model and would be even smaller in a recessive risk model and dominant risk model (22.68% and 45.13%, respectively). For ANCA-negative WG (n = 35), power levels in multiplicative, recessive, and dominant risk models would be 85.18%, 29.26%, and 57.69%, respectively. The respective values for both the complete WG sample (n = 407) and for the ANCA-positive WG sample (n = 364) would again be close to 100%, regardless of the risk model used.

DISCUSSION

IL-10 expression is, for the most part, determined genetically, as shown by analysis of promoter polymorphisms and the respective haplotypes. These polymorphisms are in strong LD. Only 4 ancient haplotypes (IL10.1–IL10.4) account for ∼75% of the genetic heterogeneity at this locus (44). According to our data, the IL10.2 haplotype is associated with ANCA-negative CSS, but not with WG. IL10.2 has been convincingly demonstrated to be associated with increased IL10 expression on a transcriptional level in lipopolysaccharide-stimulated whole-blood cultures, leading to higher IL-10 protein levels in IL10.2 carriers (28). Moreover, IL10.2 harbors an allele (−2849G) that is correlated with increased IL-10 production (47, 48). Independently, 2 further alleles of IL10.2 (−3575T/−2763G) have been linked to higher IL-10 production (29). The −3575 SNP was analyzed individually in the present study and showed a significant association with an increased risk of ANCA-negative CSS (allele-specific P = 0.006), but showed no association with WG.

IL10 promoter polymorphisms have been investigated previously in WG. Yet, the conclusions that could be drawn from the majority of those studies are limited, since only small sample sizes were included, which are, a priori, prone to stochastic errors and limited statistical power. Zhou et al described a statistical association of allele 8 of the IL10.G promoter microsatellite in patients with WG in Sweden (n = 36) (34). Because this allele is not covered by the IL10.2 haplotype, but may be part of the other 3 major ancient haplotypes (44), any conclusion about its effect on IL10 expression is rather difficult to draw. In a different study, Spriewald et al found that the IL10 −592/−819/−1082GCC/ACC haplotype was overrepresented in the subgroup of WG patients with end-stage renal disease (15 of 32 patients) (36). Muraközy et al described a significant shift (P < 0.05) toward the −1082AA genotype in 39 Caucasian patients with WG (33).

The only study of IL-10 that included a substantial number of Caucasian patients with WG (n = 125), reported by Bartfai et al (35), revealed a significant overrepresentation of the −1082AA genotype (P < 0.005). The −1082A allele may be a component of both the IL10.2 major ancient haplotype and the IL10.4 major ancient haplotype (in addition to the 4 less frequent haplotypes IL10.9–IL10.12 [28]). Since, in the study by Bartfai et al, no other IL10 promoter SNPs were genotyped, it is not possible to evaluate the exact distribution of the extended haplotypes in that cohort. Therefore, any legitimate conclusion from those findings is problematic. Moreover, it is noteworthy that Bartfai et al reported quite a low frequency of the −1082AA genotype (10.5%) in their Caucasian control cohort. At the same time, the frequency of this genotype in their Caucasian subjects with WG (0.25) was similar to the frequency in both our patients with WG (also Caucasian) (0.28) and our controls (0.28), which might suggest that a stochastic bias was present in their control group. This aspect is supported by the finding that the data from their control group significantly deviated from Hardy-Weinberg equilibrium (P = 0.0007).

The present study is by far the largest study of IL10 promoter polymorphisms in WG performed to date. The power calculations for the −1082AA genotype make it highly unlikely that our study findings constitute false-negative results by chance. The fact that neither one of the (promoter) SNPs nor the respective haplotypes were associated with WG in our cohort challenges the previous findings from considerably smaller studies. This finding of a lack of association with WG is true despite the fact that the −1082A allele has been previously demonstrated to be associated with decreased IL-10 levels (22, 49), and its overrepresentation could therefore fit into Th1 and Th17 models that have been discussed in the context of WG pathogenesis (5, 6).

Furthermore, IL-10 treatment of T cell cultures derived from WG patients has been shown to suppress the Th1 response (24). In this context, Kurreeman et al raised the question as to whether further common variations of IL10 may have an effect on the production of IL-10, since the IL10.4 haplotype was associated with both low and high IL10 transcription in that study (28). With the use of indirect analyses of the coding and intronic sequences of IL10 with a tagSNP-based approach, we have herein produced further evidence that genetically determined IL-10 production is not a major aspect of the pathogenesis of WG.

Although large parts of the pathogenesis of CSS are still to be elucidated, current hypotheses regarding its pathogenetic mechanisms include hyperresponsiveness to antigenic stimuli such as parasites or allergens, as well as phenomena related to withdrawal of antiinflammatory drugs in a subset of patients with atopic disease, particularly patients with asthma (for review, see ref. 50). Consistent with the observation that plasma IL-10 levels are elevated in CSS, but not in WG (23), our finding of a genetically primed increase in IL-10 production in patients with CSS yields interesting insights into the potential mechanisms of CSS pathogenesis.

Current models of CSS pathogenesis consistently describe Th2-related inflammation patterns, which are thought to be a result of initially physiologic, but subsequently dysregulated, Th2 responses (51). T cells predominantly secrete Th2-type cytokines (IL-4, IL-5) in CSS (3, 4). IL-10 from activated macrophages exhibits properties of a negative feedback regulator, by inhibiting Th1-type (IL-12) production in activated macrophages and dendritic cells and subsequently inhibiting Th1-type T cell responses (19).

Although plasma IL-10 levels are known to be elevated in CSS (23, 24), none of the studies of T cell cytokine production and plasma IL-10 levels have made a distinction between ANCA-negative and ANCA-positive CSS. Our present data indicating an association of the −3575/−1082/−592 TAC haplotype, part of the extended IL10.2 ancient haplotype, with ANCA-negative CSS suggest a role of IL-10 in the ANCA-negative CSS subgroup. Intriguingly, differences between ANCA-negative and ANCA-positive CSS have been reported, with ANCA-negative CSS being characterized by a higher frequency of cardiac involvement and fever, compared with more prevalent renal involvement, peripheral neuropathy, and biopsy-proven vasculitis in ANCA-positive CSS (7). Whether these differences are related to differences in the genetic background of the 2 subsets has to be further investigated, but our data suggest that there are differences between ANCA-negative and ANCA-positive CSS and also differences between CSS itself and WG.

Interestingly, in another disorder involving organ damage associated with eosinophils, the hypereosinophilic syndrome (HES), different subtypes, e.g., the FIP1L1-PDGFRA fusion gene mutation generating tyrosine kinase activation in myeloproliferative HES, clonal T cell proliferation in T cell–associated HES, and various other mutations in HES associated with chronic myeloproliferative disorders, have been associated with distinct genetic abnormalities (52). We suggest that further genetic studies of Th2-related cytokines (e.g., IL-4, IL-5, and IL-13) may produce valuable additional insights into the pathogenesis of CSS.

Confirmation of our results in larger and/or independent cohorts of patients with CSS and matched control groups is desirable. This is especially true because our results from the ANCA-positive CSS subgroup and the ANCA-negative WG subgroup were limited by reduced statistical power (attributable to small sample sizes). Moreover, as we observed for the IL10 −3575 SNP in ANCA-positive CSS, in which the genotype distribution (but not the allele frequency) differed significantly in comparison with controls, such small sample sizes may be prone to random errors. This supposition was supported in our study by the statistically significant deviation from Hardy-Weinberg equilibrium.

Two recent studies have independently shown an increased frequency of the HLA–DRB1*07 allele in patients with CSS compared with controls (15, 16), an association that has not been found in patients with WG. The finding that ANCA-negative CSS, but not WG, is associated with the IL10.2 haplotype (and that the frequency of IL10.2 is significantly different between patients with WG and patients with CSS) is consistent with the aforementioned possibility that CSS and WG are not based on the same genetic background.

Taken together, the findings presented here provide further evidence of the existence of distinct genetic backgrounds in CSS and WG, and, in fact, seem to indicate that CSS and WG are pathogenetically distinct diseases. Furthermore, our data suggest that IL-10 should be considered an important cytokine in the pathogenesis of CSS, because of its potential role in influencing the hyperergic Th2 inflammatory response in this disease.

AUTHOR CONTRIBUTIONS

Dr. Wieczorek had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.

Study design. Wieczorek, Hellmich, Epplen.

Acquisition of data. Wieczorek, Hellmich, Moosig.

Analysis and interpretation of data. Wieczorek, Arning, Moosig, Lamprecht, Gross, Epplen.

Manuscript preparation. Wieczorek, Hellmich, Moosig, Lamprecht, Gross, Epplen.

Statistical analysis. Wieczorek, Arning.

Acknowledgements

We thank I. Alheite and N. M. Gruening for excellent technical assistance, as well as an anonymous reviewer for valuable suggestions.

Ancillary