Recent genome-wide association studies have identified 1q31 (RGS1), 2q11–12 (IL18RAP), 3p21 (CCR1/CCR3/CCR2), 3q25–26 (IL12A/SCHIP1), 3q28 (LPP), 4q27 (IL2/IL21), 6q25 (TAGAP) and 12q24 (SH2B3) as susceptibility regions for coeliac disease (CD). We have earlier replicated association with the IL2/IL21 region. This study aimed at replicating the remaining regions in a family cohort using the transmission disequilibrium test, which is not prone to population stratification as a source of false-positive results. Nine single nucleotide polymorphisms (SNPs) within these regions were genotyped in 325 Swedish–Norwegian CD families. We found significant associations with the same alleles in the regions 1q31 (rs2816316; Pnc=0.0060), 3p21 (rs6441961; Pnc=0.0006), 3q25–26 (rs17810564; Pnc=0.0316 and rs9811792; Pnc=0.0434) and 3q28 (rs1464510; Pnc=0.0037). Borderline, but non-significant, associations were found for rs917997 (IL18RAP), whereas no evidence for association could be obtained for rs13015714 (IL18RAP) or rs1738074 (TAGAP). The lack of replication of the latter SNPs could be because of limited power. rs3184504 (SH2B3) was not analysed because of assay failure. The most significantly associated region, 3p21 (CCR1/CCR3/CCR2), was further analysed by typing of 30 SNPs, with the aim of identifying the causal variant responsible for the initial association. Several SNPs showed association with CD, but none displayed associations stronger than rs6441961, nor did any of them add to the effect initially marked by rs6441961 in a conditional analysis. However, differential effects of rs6441961*C carrying haplotypes were indicated, and we thus cannot exclude the possibility that our inability to obtain evidence for multiple independent effects in the CCR1/CCR3/CCR2 gene region was related to a power issue.
Coeliac disease (CD) displays complex aetiology and develops in genetically susceptible individuals as a result of an inappropriate immune response to dietary gluten proteins. The lesion is localized in the proximal small intestine and is characterized by the absence of villi, crypt hyperplasia and infiltration of inflammatory cells in the epithelium and in the lamina propria. A strong genetic component in CD is well documented by familial clustering and a high concordance rate in monozygotic twins.1 CD is strongly associated with the human leukocyte antigen (HLA) class II molecules DQ2.5 (DQA1*05/DQB1*02) and DQ8 (DQA1*03/DQB1*0302) as 90–95% of CD patients express DQ2.5 and most of the remaining patients express DQ8.2 The role of the DQ2.5 and DQ8 molecules in CD pathogenesis is now fairly well established. These DQ molecules preferentially bind gluten peptide and present them on the surface of antigen-presenting cells to T cells in the gut mucosa. The T cells that recognize the gluten peptides complexed with DQ2.5 or DQ8 initiate a cascade of events that eventually lead to the formation of the CD lesion. Although HLA-DQ alleles are major genetic risk factors, the great majority of individuals who carry DQ2.5 or DQ8 never develop CD, and there is a large discrepancy in the concordance rates of monozygotic twins and HLA-identical dizygotic twins.3 Collectively, this indicates that HLA is necessary but not sufficient for CD development.
Genome-wide association studies (GWASs) represent a major breakthrough in the genetics of complex disorders. Many novel susceptibility genes have been identified in GWASs of chronic immune-related diseases.4 The first GWAS of CD identified, besides the HLA association, association with single nucleotide polymorphisms (SNPs) within a linkage disequilibrium (LD) block on chromosome 4q27 harbouring the KIAA1109, TENR, IL2 and IL21 genes.5 This region has also shown association with type I diabetes and rheumatoid arthritis,6 and the immune functions of interleukin-2 and interleukin-21 make the IL2 and IL21 the prime suspects to explain this association. Association between the 4q27 region and CD has been confirmed in independent CD populations, including our own.7, 8 In a follow-up study of the GWAS, 1020 SNPs with the lowest P-values were genotyped in a larger CD cohort, and significant associations were found for seven novel regions.9 In the same study, the genetic CD risk accounted for by the eight non-HLA chromosomal regions was estimated to be ∼3–4%, whereas the risk attributed to DQ2 and DQ8 could explain ∼35% of the heritability.9 All the seven novel regions contain genes that are related to immune response. Association with six of these eight novel regions was recently replicated in an Italian case–control study.8
In this study, we aim at replicating the association with the seven novel candidate regions in a family setting using the transmission disequilibrium test (TDT). The TDT is robust for population stratification10 and replication of the GWAS findings in a family material would strengthen the evidence that these are true susceptibility regions for CD. Moreover, we conducted an extensive SNP association screen of the 3p21 region (CCR1/CCR3/CCR2) with the aim of identifying the factor(s) responsible for the association of this region with CD.
Materials and methods
This study included DNA from 325 CD families: 100 families with two or more affected children (multiplex families) and 225 single-child families (simplex families) earlier described.11, 12 All CD patients fulfilled the European Society of Paediatric Gastroenterology and Nutrition diagnostic criteria,13 and all family members were of Swedish or Norwegian origin.
Single nucleotide polymorphism selection for confirmation of CD associations proposed by the GWAS
Nine SNPs reported to tag the seven regions identified by Hunt et al.9 were selected and subjected to genotyping: rs2816316 (1q31, RGS1), rs13015714 and rs917997 (2q11–12, IL18RAP), rs6441961 (3p21, CCR1/CCR3/CCR2), rs17810546 and rs9811792 (3q25–26, IL12A/SCHIP1), rs1464510 (3q28, LPP), rs1738074 (6q25, TAGAP) and rs3184504 (12q24, SH2B3). As rs17810546 and rs9811792 were reported to be independently associated, both were included.
Single nucleotide polymorphism selection for the follow-up screen of 3p21 (CCR1/CCR3/CCR2)
Genotype data pertaining to individuals from the Centre d’Etude du Polymorphisme Humain (Utah residents with ancestry from northern and western Europe (CEU)) available at the HapMap public release #23 (which included SNP genotypes from the Affymetrix 6.0 array) were used for SNP selection (www.hapmap.org) using Haploview v4.0.14 We selected SNPs with pairwise r2 value in the range 0.5<r2<1.0 with rs6441961 within a region defined as the last SNP with D′=1 with rs6441961 in each direction (that is, corresponding to chromosomal positions 46 153 542–46 458 540 according to the NCBI genome build 36.3). We based this selection on the rationale that the causal variant must be in LD with rs6441961 as the association was identified with this SNP, but to give a stronger disease association and to be a better candidate for the causal SNP, the test SNPs should not be fully correlated (that is r2<1) with rs6441961. In addition, to obtain a better genetic coverage and to allow for capturing possible additional risk factors, tag-SNPs were selected within the LD block (defined by Gabriel et al.15) surrounding rs6441961 (position 46 320 417–46 377 022). This LD block had distinct borders in the centromeric direction. However, in the telomeric direction the pairwise LD was less confined within the block. Therefore, the region was extended telomerically, and tag-SNPs were selected between 46 311 405 and 46 377 022. Further, haplotype tag-SNPs described by Clark and Dean,16, 17 as well as five SNPs selected from the ABI SNPbrowser version 3.5 because of insufficient coverage in HapMap of the most centromeric region, were included.
Single nucleotide polymorphism genotyping
The 44 SNPs were genotyped by iPLEX MassArray technology on a Maldi-TOF mass spectrometer integrated in the Sequenom instrument (Sequenom, San Diego, CA, USA), or by TaqMan chemistry on an ABI Prism 7900HT instrument (Applied Biosystems, Foster City, CA, USA). The SPECTRO ACQUIRE v.188.8.131.52 or SDS v2.3 software was used for genotype calling. The 32-bp insertion–deletion of the CCR5 gene (rs333, also denoted as CCR5-wt/Δ32) was genotyped by PCR (using primer sequences and PCR conditions as described in Melum et al.18), followed by fragment length separation on an ABI Prism 3730 Sequence Analyser (Applied Biosystems) using standard conditions. Allele calling was determined using GeneMapper v3.7.
Power calculations were conducted before the study using the TDT power calculator located at http://pngu.mgh.harvard.edu/~purcell/gpc/ (at a significance level of 0.05). We calculated the expected allelic odds ratio (OR) that is to detect with our sample size (n=325) given a power of 80% using allele frequencies in the range 0.10–0.50. All SNP genotypes were checked for deviation from Hardy–Weinberg equilibrium (P<0.001), and Pedcheck19 was used to evaluate the Mendelian error rate. Families showing Mendelian inconsistencies were excluded before statistical analysis, and SNPs showing inconsistencies within more than five families were left unanalysed. The genotype call rate was calculated for each SNP to ensure sufficient statistical power. LD analysis was conducted using Haploview v4.0.14 Both single and multipoint association analyses were conducted by the TDT10 using the TDTPHASE application implemented in the UNPHASEDv2.403 package.20 In the multiplex families, only transmissions to the probands were counted. Allelic ORs were calculated using transmitted and non-transmitted alleles from the parents.
For the SNP screen of 3p21 (CCR1/CCR3/CCR2), we conducted the initial haplotype association analysis with the eight associated SNPs using a cutoff of 10% (that is, only haplotypes found with a frequency of at least 10% in the parents were included in the statistical analysis). Estimations of haplotype population frequencies were conducted using the expectation–maximization algorithm implemented in the UNPHASEDv2.403 using simplex families and a 5% cutoff. A main effect test (logistic regression) and the homozygous parent TDT21 were conducted, using UNPHASED v2.403, to test one SNP while conditioning on another. An uncorrected P-value (Pnc) below 0.05 (two-sided test) was considered statistically significant.
Single nucleotide polymorphism quality control
All nine SNPs located within the seven novel risk regions were genotyped, but rs3184504 was left unanalysed because of difficulty in interpreting the raw data. The remaining SNPs showed the following genotype call rates: rs2816316 and rs9811792 (97%), rs13015714, rs917997 and rs6441961 (95%), rs1464510 (98%), rs17810546 and rs1738074 (99%). Of the 30 additional genotyped SNPs in the 3p21 (CCR1/CCR3/CCR2) region, one SNP was excluded because of more than five Mendelian inconsistencies (rs6441957), one SNP showed to be non-polymorphic in our population (rs9875668) and four were excluded because of deviation from Hardy–Weinberg equilibrium (rs4683211, rs4683213, rs4683215, rs7374671). The 24 remaining SNPs showed >95% genotype success rates.
In this study, all the 32 successfully analysed SNPs were in Hardy–Weinberg equilibrium (P>0.23). Pedcheck revealed Mendelian inconsistencies for rs6441950, rs9990343, rs6441959, rs17141079 and rs12494200 within one family each. These families were excluded from the analysis of the SNP in question. Power calculations showed that we had 80% power to detect effects with a minimum OR in the range 1.91–1.56 with our sample size for SNPs with minor allele frequencies in the range 0.10–0.50.
Initial association analysis of single nucleotide polymorphisms within seven novel susceptibility regions
Of the eight successfully genotyped SNPs tagging the seven candidate regions, statistically significant associations with the same allele as originally reported were found for the SNPs rs2816316 (RGS1), Pnc=0.0060, rs6441961 (CCR1/CCR3/CCR2), Pnc=0.0006, rs9811792 (IL12A), Pnc=0.0434, rs17810546 (IL12A/SCHIPI), Pnc=0.0316 and rs1464510 (LPP), Pnc=0.0037. Borderline, but non-significant associations (0.05<Pnc<0.1) were seen for rs917997 located at 2q12 (IL18RAP), Pnc=0.0778. We were unable to replicate association with rs13015714 (IL18RAP), Pnc=0.1315, and rs1738074 (TAGAP), Pnc=0.1446 (Table 1).
Single nucleotide polymorphism association screen of 3p21 (CCR1/CCR3/CCR2)
rs6441961*T at 3p21 showed a transmission skewness of 61%, and this SNP gave the strongest association signal of the eight analysed SNPs in terms of both statistical P-value and relative risk (Pnc=0.0006 and relative risk=1.59) (Table 1). Thus, we chose to follow up this region by performing a SNP association screen, selecting tagging SNPs as well as SNPs with moderate-to-high correlation with rs6441961 (see Materials and methods), to search for a putative causal variant and narrow the region of interest, which, through LD, could explain the association initially detected by rs6441961. Seven of the SNPs genotyped in this screen showed statistically significant association with Pnc in the range 0.0292–0.0019, with the percentage of transmission (%T) ranging from 58 to 61 (Table 2).
rs6441961*T explains the main susceptible effect in a logistic regression analysis
To analyse whether the associated SNPs represent the same causal variant or independent risk effects, we conducted a logistic regression test adjusting for each of the associated SNPs. None of the associated SNPs showed significant association when adjusting for rs6802288, rs13096142 or rs6441961, indicating that either of these SNPs could explain the association we observed in the region (Supplementary Table 1). In contrast, several SNPs remained significantly associated when adjusting for any of the five other associated SNPs. In line with this, strong LD was observed between rs6441961 and rs6802288 (D′=0.963, r2=0.867) and between rs6441961 and rs13096142 (D′=0.962, r2=0.865) (Table 2). Thus, these SNPs seem to mark the same effect, rather than independent risk effects. Specifically, the main effect test found no evidence for rs333 or any of the tested haplotype tag-SNPs adding to the effect of rs6441961.
Linkage disequilibrium across the CCR1/CCR3/CCR2 region
The genotyped SNPs showed strong LD when measured by D′ (Supplementary Figure 1a), as expected because the SNPs were mainly selected to partly correlate with the CD-associated rs6441961. Apart from the SNPs that showed r2 values in the range 0.5<r2<1.0 with rs6441961, all the additional tag-SNPs showed pairwise r2 values below 0.550 (Table 2 and Supplementary Figure 1b). Interestingly, we further found limited correlations between the significantly associated SNPs (that is, low pairwise r2 values; Figure 1). In contrast, strong LD, as measured by D′, was found between the eight associated SNPs (Figure 1). These LD observations indicate that the associated SNPs may occur together on a disease-associated haplotype.
Haplotype analysis reveals both predisposing and protective haplotypes
We further went on to analyse haplotypes. Initially, we analysed haplotypes comprising the five haplotype tag-SNPs (rs4987053, rs3918357, rs1799865, rs1800023 and rs333), which has been reported to tag the eight most frequent haplotypes found across the CCR1, CCR3, CCR2 and CCR5 gene regions in the European–American population.15, 16 Among our simplex families, seven common haplotypes with an estimated population frequency (based on the non-transmitted haplotypes) of at least 5% were seen in our Scandinavian population. The observed haplotype frequencies in our population were similar to the frequencies earlier described for the European–American population (Table 3). When testing for haplotype association with these haplotype tag-SNPs by TDT, one protective haplotype was evident (38%T and Pnc=0.039). The 32-bp deletion, CCR5-Δ32 (rs333), within the CCR5 gene was present on this protective haplotype (No. 6; Table 3).
As the next step, we aimed to explore the association between CD and haplotypes of the eight associated SNPs, motivated by the strong LD (D′>0.88) pointing towards the possibility of them being inherited together on one disease-associated haplotype. Five common haplotypes were present, of which borderline statistical significance with one over-transmitted haplotype as well as one protective haplotype was seen (Table 4a). The two under-transmitted haplotypes, haplotypes IV and V (both displaying 40%T), were identical except for a deviation at the last position (that is, the 32-bp insertion–deletion), which means that these haplotypes were split by rs333. Thus, removal of rs333 from the analysis merged these two under-transmitted haplotypes and revealed one negatively associated haplotype: haplotype IV in Table 4b (38%T and Pnc=0.006). A statistically significant over-transmitted haplotype, haplotype I, was also evident (63%T and Pnc=0.04) (Table 4b), which, as hypothesized, carried all the disease-associated SNP alleles from the single-point analysis (Table 2).
Indications of differential effects of rs6441961*C-carrying haplotypes
If rs6441961 is the sole risk factor in the CCR1/CCR3/CCR2 region, as suggested by the logistic regression analysis, one should expect that all haplotypes carrying rs6441961*T should have equally increased risk and all haplotypes carrying rs6441961*C should have equally reduced risk. To explore whether this was the case, we first selected three SNPs that were necessary to capture the four common haplotypes: rs4987053 (*C being unique for I), rs6441961 (representing the common risk alleles) and rs7641466 (*G unique for IV). rs6441961*T was found on two haplotypes (I and II), both of which appeared to be over-transmitted (Table 4c). These two haplotypes were split by rs4987053, but rs4987053*C (unique on haplotype I) did not show a differential transmission when fixing for the shared rs6441961*T by the homozygous parent TDT (52%T; Pnc=0.9). Hence, rs4987053 did not appear to be an independent risk factor and removal of rs4987053 merged the over-transmitted haplotypes into one haplotype, rs741466*A–rs6441961*T (65%T; 95% CI (57–72); Pnc=0.001). The negatively associated rs6441961*C was found on two haplotypes (III and IV), of which haplotype IV was significantly under-transmitted although haplotype III appeared to be neutral (Table 4c). A skewed, but insignificant, transmission distortion (41%T; Pnc=0.1) of the rs741466*G allele was observed in transmissions from parents homozygous for rs6441961*C. Reduction of the dataset after stratification leading to small numbers (38T vs 54NT) could be the reason for the lack of significance.
This study is, to our knowledge, the first family-based study aimed at replicating the novel CD susceptibility regions identified by a case–control GWAS. Compared with case–control studies, which will always render the problem of producing false-positive findings because of population stratification and inadequately matched controls, analysis of families using the TDT is highly robust in this matter. Hence, successful replication in families substantiates the notion of true initial findings.
We have earlier replicated the association with 4q27 (IL2/IL21),7 and in this study we were able to replicate association with SNPs tagging four of the additional disease regions: 1q31 (RGS1), 3p21 (CCR1/CCR3/CCR2), 3q25–26 (IL12A/SCHIP1) and 3q28 (LPP). We were unable to analyse rs3184504 at 12q24 (SH2B3) because of assay failure. Neither rs1738074 (TAGAP) nor rs13015714 and rs917997 (IL18RAP) were replicated in our population, most likely because of limited power to detect the low-risk effects given the size of our study cohort. Our sample size was found to provide 80% power to detect effects with OR in the range 1.56–1.91. The effect size of the SNPs within the candidate regions is lower, within the range OR=0.70–1.39,9 which implicates the need for substantial sample sizes to achieve 80% power. Hence, limited power consequently only allows confirmation, and not rejection, of suggested risk loci.
Association with six of the eight novel regions was recently established in an Italian cohort.8 In this population, association with the IL18RAP and CCR regions was not replicated. Further, a recent study of IL18RAP in three European populations reported inability to replicate association with this region in Italian and Finnish cohorts, whereas association was confirmed in a Hungarian population.22 Apart from the limited power, which is a probable explanation, replication failure might be due to population heterogeneity, with IL18RAP not being associated with CD in the Italian, Swedish–Norwegian and Finnish populations. It is noteworthy that the risk estimates in our study (OR=1.27, 95% CI (0.97–1.66) and 1.23, 95% CI (0.94–1.61)) were of the same magnitude as the risk initially observed (OR=1.27, 95% CI (1.17–1.36)).9 Accordingly, IL18RAP cannot be discarded as a false-positive finding on the basis of the present replication efforts.
One of the regions that were convincingly replicated in our family-based association analysis was the CCR1/CCR3/CCR2 region. Our screen of this region included rs333 (CCR5-wt/Δ32), which is a 32-bp insertion–deletion that has been intensively analysed in autoimmune diseases and in HIV. The CCR5-Δ32 results in a complete lack of surface expression of CCR5, and individuals who are homozygous for CCR5-Δ32/32 have been shown to be protected against HIV because CCR5 is a co-receptor for this virus.23 A protective effect of CCR5-Δ32 has also been reported for inflammatory diseases such as rheumatoid arthritis24 although conflicting results exist.25 A correlation between the CCR5-Δ32 and copy number variation within the CCL3L1 gene has been found to enhance HIV/AIDS susceptibility,26 as well as increase the risk of developing systemic lupus erythematosus.27 Our results did not support rs333 as the primary causal polymorphism within the CCR1/CCR3/CCR2 gene region. In fact, CCR5-Δ32 was shown to split the protective haplotype and the haplotype association analysis further showed that rs7641466*G was the unique variant, among the analysed polymorphisms, on this haplotype for the protective effect and not CCR5-Δ32. Thus, it is possible that the protective signal earlier established by CCR5-Δ32 reflects a true disease protective variant located proximal to CCR5-Δ32 and that this signal is best marked by rs7641466*G in our population.
Our SNP screen of 3p21 was designed to enable identification of possible stronger disease associations, as well as to narrow the associated region by haplotype analysis and in that way identify better candidates for the causal variant(s). Recently, a study by Smyth et al.28 claimed independent association of rs333 (CCR5) and rs6441961 (CCR1/CCR3/CCR2), arguing for two or more causal variants or genes within this region. We were unable to obtain evidence for multiple independent effects by logistic regression analysis and by homozygous parent TDT. However, the skewed, transmission distortion of rs741466*G and differential risk of the rs6441961*C-carrying haplotypes indicate a possible additional protective effect within this region, an effect that because of small numbers did not reach statistical significance. Our failure to detect multiple independent effects in the CCR1/CCR3/CCR2 region may thus well relate to a power issue.
In conclusion, our family-based association analysis confirms that 1q31 (RGS1), 3p21 (CCR1/CCR3/CCR2), 3q25–26 (IL12A/SCHIP1) and 3q28 (LPP) are risk regions for CD. To clarify whether rs6441961 and rs7641466 are functional relevant disease variants within the CCR1/CCR3/CCR2 region, further investigation is needed.
Conflict of interest
The authors declare no conflict of interest.
Nistico L, Fagnani C, Coto I, Percopo S, Cotichini R, Limongelli MG et al. Concordance, disease progression, and heritability of coeliac disease in Italian twins. Gut 2006; 55: 803–808.
Sollid LM . Coeliac disease: dissecting a complex inflammatory disorder. Nat Rev Immunol 2002; 2: 647–655.
Greco L, Romino R, Coto I, Di CN, Percopo S, Maglio M et al. The first large population based twin study of coeliac disease. Gut 2002; 50: 624–628.
Lettre G, Rioux JD . Autoimmune diseases: insights from genome-wide association studies. Hum Mol Genet 2008; 17: R116–R121.
van Heel DA, Franke L, Hunt KA, Gwilliam R, Zhernakova A, Inouye M et al. A genome-wide association study for celiac disease identifies risk variants in the region harboring IL2 and IL21. Nat Genet 2007; 39: 827–829.
Zhernakova A, Alizadeh BZ, Bevova M, van Leeuwen MA, Coenen MJ, Franke B et al. Novel association in chromosome 4q27 region with rheumatoid arthritis and confirmation of type 1 diabetes point to a general risk locus for autoimmune diseases. Am J Hum Genet 2007; 81: 1284–1288.
Adamovic S, Amundsen SS, Lie BA, Gudjonsdottir AH, Ascher H, Ek J et al. Association study of IL2/IL21 and FcgRIIa: significant association with the IL2/IL21 region in Scandinavian coeliac disease families. Genes Immun 2008; 9: 364–367.
Romanos J, Barisani D, Trynka G, Zhernakova A, Bardella MT, Wijmenga C . Six new coeliac disease loci replicated in an Italian population confirm association with coeliac disease. J Med Genet 2009; 46: 60–63.
Hunt KA, Zhernakova A, Turner G, Heap GA, Franke L, Bruinenberg M et al. Newly identified genetic risk variants for celiac disease related to the immune response. Nat Genet 2008; 40: 395–402.
Spielman RS, McGinnis RE, Ewens WJ . Transmission test for linkage disequilibrium: the insulin gene region and insulin-dependent diabetes mellitus (IDDM). Am J Hum Genet 1993; 52: 506–516.
Gudjonsdottir AH, Nilsson S, Ek J, Kristiansson B, Ascher H . The risk of celiac disease in 107 families with at least two affected siblings. J Pediatr Gastroenterol Nutr 2004; 38: 338–342.
Louka AS, Nilsson S, Olsson M, Talseth B, Lie BA, Ek J et al. HLA in coeliac disease families: a novel test of risk modification by the ‘other’ haplotype when at least one DQA1*05-DQB1*02 haplotype is carried. Tissue Antigens 2002; 60: 147–154.
Walker-Smith J, Guandalini S, Schmitz J, Schmerling D, Visakorpi J . Report of working group of European Society of Paediatric Gastroenterology and Nutrition: revised criteria for diagnosis of coeliac disease. Arch Dis Child 1990; 65: 909–911.
Barrett JC, Fry B, Maller J, Daly MJ . Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics 2005; 21: 263–265.
Gabriel SB, Schaffner SF, Nguyen H, Moore JM, Roy J, Blumenstiel B et al. The structure of haplotype blocks in the human genome. Science 2002; 296: 2225–2229.
Clark VJ, Dean M . Haplotype structure and linkage disequilibrium in chemokine and chemokine receptor genes. Hum Genomics 2004; 1: 255–273.
Clark VJ, Dean M . Characterisation of SNP haplotype structure in chemokine and chemokine receptor genes using CEPH pedigrees and statistical estimation. Hum Genomics 2004; 1: 195–207.
Melum E, Karlsen TH, Broome U, Broome U, Thorsby E, Schrumpf E et al. The 32-base pair deletion of the chemokine receptor 5 gene (CCR5-Delta32) is not associated with primary sclerosing cholangitis in 363 Scandinavian patients. Tissue Antigens 2006; 68: 78–81.
O’Connell JR, Weeks DE . PedCheck: a program for identification of genotype incompatibilities in linkage analysis. Am J Hum Genet 1998; 63: 259–266.
Dudbridge F . Pedigree disequilibrium tests for multilocus haplotypes. Genet Epidemiol 2003; 25: 115–121.
Lie BA, Todd JA, Pociot F, Nerup J, Akselsen HE, Joner G et al. The predisposition to type 1 diabetes linked to the human leukocyte antigen complex includes at least one non-class II gene. Am J Hum Genet 1999; 64: 793–800.
Koskinen LL, Einarsdottir E, Dukes E, Heap GA, Dubois P, Korponay-Szabo IR et al. Association study of the IL18RAP locus in three European populations with coeliac disease. Hum Mol Genet 2009; 18: 1148–1155.
Benkirane M, Jin DY, Chun RF, Koup RA, Jeang KT . Mechanism of transdominant inhibition of CCR5-mediated HIV-1 infection by ccr5delta32. J Biol Chem 1997; 272: 30603–30606.
Prahalad S . Negative association between the chemokine receptor CCR5-Delta32 polymorphism and rheumatoid arthritis: a meta-analysis. Genes Immun 2006; 7: 264–268.
Lindner E, Nordang GB, Melum E, Flato B, Selvaag AM, Thorsby E et al. Lack of association between the chemokine receptor 5 polymorphism CCR5delta32 in rheumatoid arthritis and juvenile idiopathic arthritis. BMC Med Genet 2007; 8: 33–38.
Gonzalez E, Kulkarni H, Bolivar H, Mangano A, Sanchez R, Catano G et al. The influence of CCL3L1 gene-containing segmental duplications on HIV-1/AIDS susceptibility. Science 2005; 307: 1434–1440.
Mamtani M, Rovin B, Brey R, Camargo JF, Kulkarni H, Herrera M et al. CCL3L1 gene-containing segmental duplications and polymorphisms in CCR5 affect risk of systemic lupus erythaematosus. Ann Rheum Dis 2008; 67: 1076–1083.
Smyth DJ, Plagnol V, Walker NM, Cooper JD, Downes K, Yang JH et al. Shared and distinct genetic variants in type 1 diabetes and celiac disease. N Engl J Med 2008; 359: 2767–2777.
This work was supported by grants from the Research Council of Norway, the Swedish Medical Research Council and the Swedish Research Council. The Sequenom genotyping service was provided by the National Technology Platform Centre for Integrative Genetics (CIGENE) supported by the functional genomics program (FUGE) of the Research Council of Norway. The TaqMan genotyping was performed at Swegene Genomics and Bioinformatics Core Facilities in Göteborg. We would like to thank all the families who participated in the study, Britt-Marie Käck, and the Celiac Society in Sweden for help with collecting families and blood samples, as well as David A van Heel for communicating unpublished results.
About this article
Cite this article
Amundsen, S., Rundberg, J., Adamovic, S. et al. Four novel coeliac disease regions replicated in an association study of a Swedish–Norwegian family cohort. Genes Immun 11, 79–86 (2010). https://doi.org/10.1038/gene.2009.67
- coeliac disease
- genetic association
Genome-wide association study identified the human leukocyte antigen region as a novel locus for plasma beta-2 microglobulin
Human Genetics (2013)
European Journal of Human Genetics (2012)
Two-stage association study in Chinese Han identifies two independent associations in CCR1/CCR3 locus as candidate for Behçet’s disease susceptibility
Human Genetics (2012)