IL23R in the Swedish, Finnish, Hungarian and Italian populations: association with IBD and psoriasis, and linkage to celiac disease

Background Association of the interleukin-23 receptor (IL23R) with inflammatory bowel disease (IBD) has been confirmed in several populations. IL23R also associates with psoriasis, suggesting that the gene may be an important candidate for many chronic inflammatory diseases. Methods We studied association of single-nucleotide variants in IL23R with IBD in Swedish patients, in both Crohn's disease (CD) and ulcerative colitis (UC) subsets. The same genetic variants were also studied in Finnish patients with psoriasis or celiac disease, and in Hungarian and Italian patients with celiac disease. Results Association of IL23R with IBD was replicated in our Swedish patients, and linkage and association of the IL23R region with psoriasis was found in the Finnish population. The IL23R region was also linked to celiac disease in Finnish families, but no association of IL23R variants with celiac disease was found in the Finnish, Hungarian or Italian samples. Conclusion Our study is the first to demonstrate association of IL23R with CD and UC in Swedish patients with IBD. It is also the first study to report linkage and association of the IL23R region with psoriasis in the Finnish population. Importantly, this is the first report of linkage of the IL23R region to celiac disease, a chronic inflammatory condition in which IL23R has not been previously implicated.


Methods:
We studied association of single-nucleotide variants in IL23R with IBD in Swedish patients, in both Crohn's disease (CD) and ulcerative colitis (UC) subsets. The same genetic variants were also studied in Finnish patients with psoriasis or celiac disease, and in Hungarian and Italian patients with celiac disease.
Results: Association of IL23R with IBD was replicated in our Swedish patients, and linkage and association of the IL23R region with psoriasis was found in the Finnish population. The IL23R region was also linked to celiac disease in Finnish families, but no association of IL23R variants with celiac disease was found in the Finnish, Hungarian or Italian samples.

Conclusion:
Our study is the first to demonstrate association of IL23R with CD and UC in Swedish patients with IBD. It is also the first study to report linkage and association of the IL23R region with psoriasis in the Finnish population. Importantly, this is the first report of linkage of the IL23R region to celiac disease, a chronic inflammatory condition in which IL23R has not been previously implicated.
Our study set out to test the reported association of IL23R with IBD and psoriasis, using Swedish and Finnish patient materials. Given the chronic inflammatory nature of both conditions, as well as the tissue targeted by the immune response, we hypothesised that these same SNPs in the IL23R region might also play a role in celiac disease-a chronic inflammatory enteropathy caused by exposure to dietary gluten that can also manifest as a blistering cutaneous disease [24]. Association of IL23R with celiac disease was studied in three populations (Finnish, Italian and Hungarian) due to heterogeneous allele and/or haplotype frequencies in different population groups, where a functional variant may explain a significant part of disease risk in one population but not another.

Subject materials
Inflammatory bowel disease Diagnosis of IBD (CD or UC) was assessed according to established clinical criteria, including endoscopic, radiology and histopathology data. The dataset consisted of 280 controls and 803 cases, of which 455 were UC cases and 348 were patients with CD. All patients were recruited at the Karolinska University Hospital in Stockholm; controls consisted of a random sample of ethnically-matched unselected individuals.

Celiac disease
The material consisted of 3 population datasets. From Finland, 260 families with celiac disease (185 multiplex families, 75 trios) and 165 single cases were collected at the University of Tampere. In most patients, celiac disease was diagnosed according to ESPGHAN criteria [25]. However, in a small number of cases, diagnosis was based only on the presence of disease-specific anti-endomysial antibodies (ca. 1% of cases). The 292 Finnish controls have been previously described [26]. The Hungarian dataset consisted of 400 families (204 multiplex families, 196 trios), 270 cases and 270 controls; they have been previously described [27]. The Italian dataset consisted of 139 cases and 198 controls. Untreated celiac patients were diagnosed in accordance with ESPGHAN criteria [25] and intestinal biopsies were analysed using Oberhuber's classification [28]. A subset of Finnish (10%) and Hungarian (12%) cases were also afflicted with dermatitis herpetiformis (DH), the blistering cutaneous manifestation of celiac disease [29]. Given that the genetic risk factors underlying the gut and skin pathologies of celiac disease are speculated to be distinct [27], linkage analyses were performed on the total patient cohort as well as stratifying the Finnish and Hungarian pedigrees by the presence or absence of DH. Information regarding DH was not available for Italian patients with celiac disease or for a minor subset of the Finnish families (27 families).

Psoriasis
The dataset consisted of 255 Finnish families (64 multiplex families, 191 trios), including 385 affected individuals, and has been described previously [30]. One psoriasis patient from each family was randomly chosen and analysed against the Finnish control material described previously [26]. Table 1 gives a summary of all sample datasets in this study. Written informed consent was obtained from all participants in this study and local ethics committees approved the study protocols.

Genotyping and analysis
Power calculations were performed using the online Genetic Power Calculator [31], with power estimated to replicate the results of Duerr et al. (2006) [1] in our TDT and case-control materials. The prevalence of IBD was set as 0.001, celiac disease set as 0.01 and psoriasis set as 0.02. The frequency of the rare protective allele was set as 0.07; the OR was set as 0.50 (and not 0.26 as in [1] because this risk effect is likely to be inflated).
SNP markers in the IL23R region were chosen from Duerr et al. (2006) [1] and were typed by MALDI-TOF analysis on the Massarray Analyzer platform (Sequenom, San Diego, USA). All genotyping was performed at the MAF core facility in Karolinska Institute, Stockholm, Sweden. Markers rs7517847 and rs1495965 were not typed due to technical difficulties. Genotyping results, pedigree structure and affection status were imported into a BC|GENE LIMS database (Biocomputing Platforms, Espoo, Finland).
Deviations from Mendelian inheritance in the family materials were determined with PEDCHECK 1.1 [32] and marker genotypes were discarded in families where Mendelian inheritance errors were detected. Mendelian inheritance error-rates for the pedigrees with celiac disease were 0.18% for the Finnish dataset and 0.15% for Hungarian; the error-rate for the Finnish pedigrees with psoriasis was 0.25%. Genotyping success rates and conformity to Hardy-Weinberg equilibrium (HWE) are shown in Additional file 1. MERLIN 1.0.1 [33] was used to perform non-parametric linkage analysis in the Finnish and Hungarian pedigrees with celiac disease, as well as the Finnish pedigrees with psoriasis. The -pairs and -exp functions were used. Haploview 4.0 [34] was used to study linkage disequilibrium (LD) patterns, estimate haplotypes and perform association analysis. Odds ratios (OR) were calculated and twosided p-values reported. Genehunter (version 2.1_r5 beta) [35] was used to perform transmission-disequilibrium tests (TDT) in the family materials. To extract information about association from the family materials without the presence of linkage, we utilised a script that randomly selected one affected individual from each family to be used as cases in a case-control type analysis (Biocomputing Platforms, Espoo, Finland).
The Cochran-Mantel-Haenszel (CMH) function in PLINK [36] was used to perform a meta-analysis of the three celiac disease datasets. The Breslow-Day test, estimating Datasets in this study. Multiplex families have more than one affected individual, trio families are comprised of father, mother, and an affected child. IBD indicates inflammatory bowel disease, UC indicates ulcerative colitis, and CD indicates Crohn's disease. homogeneity of odds ratios, was non-significant (p > 0.05).

Results
Additional file 2 shows the linkage disequilibrium (LD) patterns for the region surrounding IL23R and IL12RB2 in the HapMap dataset. The position of the single-nucleotide polymorphisms (SNPs) analysed in this study is also shown.  = 0.08, 132 informative families; Table 3A) nor significant association (data not shown).

IL23R and inflammatory bowel disease
The Finnish and Hungarian pedigrees with celiac disease were stratified on the basis of their co-presentation of dermatitis herpetiformis (i.e. families with DH or families without DH) and linkage was calculated. As shown in Table 3B, the Finnish families with DH (34 of whom were informative for linkage) yielded a non-parametric lodscore of 1.03 (p = 0.015), while the Finnish families without DH (74 were informative for linkage) yielded a nonparametric lodscore of 1.71 (p = 0.0013). The linkage signal (p = ns) in the Hungarian dataset came only from the families without DH (121 families informative for linkage). The Finnish pedigrees were also stratified into families with one or more individuals carrying the rare protective IBD allele (rs11209026*A) and families without this allele. Families with rs11209026*A comprised approximately 10% of the Finnish material and contributed to approximately 5% of the lodscore (p = ns, data not shown), suggesting that this rare allele does not explain the linkage in Finnish pedigrees with celiac disease.
One individual from each Finnish or Hungarian pedigree was added to the single case material to increase the power of the case-control analysis. The celiac disease casecontrol datasets showed no significant association to any of the markers at the IL23R locus, not as individual populations nor when combined for a meta-analysis (Table 2B and Additional file 3). The celiac disease material also demonstrated no association to the haplotypes of the SNPs in the IL23R region (Additional file 4). An association analysis was performed by comparing cases with psoriasis (selected from the family material) with controls from [26]; it found no association of the IL23R locus with psoriasis (Table 2C, and Additional files  3 and 4). Neither the rs11465804-rs11209026-rs1343151-rs10889677 TGCC risk haplotype (0.164 frequency in cases and 0.165 in controls, p = 0.96) nor the rs1343151-rs10889677-rs11209032 TCG protective haplotype (0.537 frequency in cases, 0.549 in controls, p = 0.67) were associated with psoriasis in the Finnish casecontrol material.

Discussion
Pathologies that are apparently unrelated have increasingly been shown to share common genetic risk factors, such as celiac disease and schizophrenia in terms of MYO9B [37,38]. Conditions that are driven by an overactive/inappropriate immune response can be more easily envisaged to share common factors that contribute to their chronic inflammatory state. It is for this reason that we have examined IBD, psoriasis and celiac disease in this study.
Our study is the first to show genetic association of IL23R with IBD in the Swedish population. In particular, we found strong association of the marker rs1004819 with Crohn's disease, thereby replicating the association reported in the German population [4]. In the German study, association was only demonstrated with this particular SNP, which is located at the junction of two linkage disequilibrium (LD) blocks (Additional file 2). These findings raise the possibility of another polymorphism in the LD block upstream of the rs1004819 marker that affects susceptibility to IBD. Such a polymorphism is envisaged to act independently of rs11209026 or the variants immediately surrounding it.
The risk and protective haplotypes associated with Crohn's disease and ulcerative colitis in the Swedish population were identical to those found in the first reported association of IL23R with IBD [1]. However, IL23R was not involved in susceptibility to ulcerative colitis in the Finnish population [26]. These findings are interesting given that the Swedish and Finnish are neighbouring populations and they also share association of Crohn's disease to the same genetic variants and risk/protective haplotypes at this locus. The differences in Swedish and Finnish cases with ulcerative colitis may reflect genetic variation in the two populations at other IBD risk loci.
The IL23R region demonstrated linkage to psoriasis in the Finnish pedigrees, suggesting that the locus is involved in this inflammatory skin disorder. Linkage of this gene region has not previously been shown to psoriasis in the Finnish population. Importantly, in support of our linkage results, association of IL23R with psoriasis was confirmed in the Finnish case-controls, demonstrating under-transmission of previously reported protective variants and over-transmission of the corresponding risk variants.
Other studies have reported association to the rs11209026 (Arg381Gln) polymorphism in IL23R, as well as to polymorphisms in IL12B, which encodes the βchain of the IL23 cytokine [14,15,39]. Taken together, these collective findings suggest that polymorphisms in several components of the IL23/IL23R-pathway may be important for susceptibility to psoriasis. Indeed, anti-IL12/IL23 antibody treatment ameliorates the symptoms of psoriasis [40], further supporting the importance of multiple components this pathway in the disease.
The gene region containing IL23R was strongly linked to celiac disease in the Finnish families, but did not reach statistical significance in the Hungarian pedigrees-a result that perhaps reflects the heterogeneity of the two populations. The discrepancy between the populations could raise questions about the biological relevance of the linkage findings, but large differences in allele and haplotype frequencies between populations have been reported by others, suggesting that the effect of the same genetic risk variant may vary considerably across different populations. Indeed, it is thought that some of the celiac disease risk loci for the Finnish and Hungarian populations are distinct [27]. Therefore, it is possible that the IL23R region is linked to celiac disease in the Finnish population but not the Hungarian. Future replication studies may help to determine whether the region is truly linked to celiac disease.
Interestingly, stratification of the celiac disease dataset by the co-occurrence of dermatitis herpetiformis indicated that linkage was independent of celiac disease sub-phenotype. Despite the linkage results in the Finnish families, the case-control material of all three populations (Finnish, Hungarian and Italian) showed no association of IL23R with celiac disease, and family-based association tests supported the lack of association of these markers.
The linkage of celiac disease to the IL23R region may be due to other genetic factors than the ones reported by Duerr et al. (2006) [1]. In support of this, stratification of the celiac disease dataset by the A-allele of rs11209026 indicated that linkage in the Finnish families with celiac disease was independent of this variant. Intriguingly, this allele A was recently found to be associated with risk instead of protection from both celiac disease and multiple sclerosis in the Spanish population [41]. These findings in the Spanish population suggest that, if rs11209026 is the primary functional variant, IL23R would play an opposite role in celiac disease and multiple sclerosis compared to IBD and psoriasis. Perhaps more likely, is that these opposite alleles are simply tagging distinct haplotypes carrying the real primary risk variants. It is apparent that the variants in the IL23R region examined in this study are not associated with celiac disease. However, it is equally clear that some factor in this gene region affects susceptibility to celiac disease. The region needs to be investigated further to identify the precise genetic variants that contribute to celiac disease. Future studies employing a tagging strategy of the whole IL12RB2 gene may be useful in IBD, psoriasis and celiac disease. Such a strategy may be able to elucidate the disease-causing variants in celiac disease, as well as identifying additional variants that modify the risk of IBD and psoriasis.

Conclusion
Our study is the first to report association of IL23R with Crohn's disease and ulcerative colitis in Swedish patients with IBD. It is also the first study to demonstrate linkage and association of the IL23R region with psoriasis in the Finnish population. But perhaps the most interesting novel finding is that of linkage of the IL23R region to celiac disease. It remains to be elucidated whether IL23R, IL12RB2 or another gene at 1p31 confers risk to celiac disease.

Competing interests
The authors declare that they have no competing interests. Additional material