HANNA SÄILÄ, JANNE PITKÄNIEMI, JAAKKO TUOMILEHTO, ANNELI SAVOLAINEN, NOORA ALAKULPPI, EVA TUOMILEHTO-WOLF, MARJATTA LEIRISALO-REPO, and KIMMO AHO

ABSTRACT.

Objective. To determine the effects of class I (A, B, and C) and II (DRB1 and DQB1) HLA loci alleles and DRB1-DQB1 haplotypes on genetic susceptibility to juvenile idiopathic arthritis (JIA) in families with 2 or more affected siblings.

Methods. A total of 83 affected siblings belonging to 38 families and corresponding to 50 affected sibpairs, their parents, and 45 healthy sibs were typed for HLA in A, C, B, DRB1, and DQB1 loci. Two study designs were used to explore linkage and association: a case-population control design and a family design using the linkage method: identical-by-descent (IBD) allele-sharing and the association analysis methods. Associations in family data were analyzed using the independent transmission disequilibrium test (TDT) for linkage in the presence of association. This was supplemented by the family-based association test (FBAT) to look for association in the presence of linkage, and is robust for population stratification and phenotype-based selection of data.

Results. Significantly increased HLA allele frequencies among the affected siblings compared to Finnish bone marrow donors were observed for HLA alleles Cw4 (odds ratio, OR, 1.7), B27 (1.8), B35 (1.7), and DR8 (3.7). The observed ratio of sharing 0, 1, and 2 HLA haplotypes (A, C, B, DRB1, and DQB1) among affected sibpairs (ASP) was 10:23:17, significantly different from expected (p < 0.001), using a formula that takes into account disease prevalence and the sibling recurrence risk. In the univariate association analysis, both independent TDT and FBAT found significantly increased transmission of the DRB1*0801 and DQB1*0402 alleles and Cw*0401. Independent positive allele effects of Cw*0401, DRB1*0801, and DQB1*0402 as well as negative effects of Cw*0701 and DQB1*0302 were shown by the family-based association analysis of the joint allele main effects. Multi-allelic test for association of each locus confirmed significant associations of the DRB1 and DQB1 loci in the risk of JIA. We found DRB1*0801/DQB1*0402 haplotype to be strongly associated (p < 0.001) with JIA, supporting findings of the haplotype associations-based ASP design.

Conclusion. Both linkage analysis of the affected sibpairs and association analysis of nuclear families with JIA provided overwhelming evidence of the major contribution of HLA to genetic susceptibility to JIA. The association analysis of HLA-A, C, B, DRB1, and DQB1 alleles by both TDT and FBAT tests confirmed in the Finnish population that the most significant associations prevailed for DRB1*0801, DQB1*0402, as expected from previous observations, and supported the independent role of Cw*0401. (J Rheumatol 2004;31:2281-5)

Key Indexing Terms:

JUVENILE IDIOPATHIC ARTHRITIS
SIBLINGS
HLA
GENETICS

From the Rheumatism Foundation Hospital, Heinola; Diabetes and Genetic Epidemiology Unit, National Public Health Institute and Department of Public Health, University of Helsinki, Helsinki; University Central Hospital, Department of Medicine, Division of Rheumatology, Helsinki; and the National Public Health Institute, Helsinki, Finland.

Supported in part by National Institutes of Health grant AR44422, Academy of Finland grant 46558, Päivikki and Sakari Sohlberg Foundation, and by a grant from Helsinki University Central Hospital Research Funds.

H. Säilä, MD, Pediatric Rheumatology Fellow, Rheumatism Foundation Hospital; J. Pitkäniemi, MSc (Biometry), Diabetes and Genetic Epidemiology Unit, Department of Epidemiology; J. Tuomilehto, MD, MPolSc, Professor, Diabetes and Genetic Epidemiology Unit, National Public Health Institute and Department of Public Health, University of Helsinki; A. Savolainen, MD, Chief Specialist in Pediatric Rheumatology, Rheumatism Foundation Hospital; N. Alakulppi, MSc, Diabetes and Genetic Epidemiology Unit, National Public Health Institute, currently at the Finnish Red Cross Blood Transfusion Service; E. Tuomilehto-Wolf, MD, FRCPath, Senior Researcher, Diabetes and Genetic Epidemiology Unit, National Public Health Institute; M. Leirisalo-Repo, MD, Professor, Department of Medicine, Division of Rheumatology, Helsinki University Central Hospital; K. Aho, MD, Professor, National Public Health Institute.

Address reprint requests to Dr. H. Säilä, Rheumatism Foundation Hospital, Fin-18120, Heinola, Finland. E-mail: Hanna.Saila@reuma.fi

Submitted December 1, 2003; revision accepted April 7, 2004.

Juvenile idiopathic arthritis (JIA) consists of a group of childhood-onset rheumatic diseases having in common chronic inflammation of one or more joints¹. The subtypes of JIA differ in clinical manifestations, prognosis, and specific immunological features. Genetic and environmentally associated determinants interact, contributing to the development and pathogenesis of JIA^2,3. JIA is a complex polygenic disease and genetic determinants may also vary among the subtypes^4,5. The degree of familial clustering is probably the best measure of the strength of the genetic component in a disease. It is estimated from the risk of the disease in 2 comparable populations, most often siblings of affected individuals and control subjects representing the general population⁶. Since JIA is a relatively rare disease, twin and family studies are difficult to carry out. Such studies have been few and often based on a small number of cases. Recent data suggest that the contribution of genetic factors to JIA may be considerable^4,7. We previously collected a set of siblings from a known JIA patient population⁸. However, those data did not allow accurate computation of sibling recurrence risk. Our aim was to assess the contribution of HLA alleles and haplotypes to the genetic susceptibility of JIA in a set of sibling series collected from the isolated Finnish population.

MATERIALS AND METHODS

During the last 2 decades, juvenile patients attending the Rheumatism Foundation Hospital who fulfil the Durban criteria for JIA¹, together with their parents, have been systematically asked about their family history of rheumatic diseases. Altogether 41 affected sibpair (ASP) families were identified; HLA genotyping was performed in 38 nuclear families. Among the 83 JIA patients in this study, the onset type was oligoarticular in 55 patients (66%). The disease course type progressed into extended oligoarthritis in 16% of the patients, was persistently oligoarticular in 42 (51%) of the patients, and in most instances, ran a mild course. The mean age of disease onset was 4.5 years.

The study series comprised 31 ASP and 7 sets of triples (3 affected siblings in a family) with JIA. However, one triple set included a monozygotic twin pair, which in genetic analyses could be regarded as 2 genetically independent siblings only; the triplet including monozygotic twins was thus considered as one sibpair. Each other triplet set was counted for the sake of simplicity as 3 sibpairs; the series thus comprised a total of 50 ASP. Clinical data on the patients were recorded and the type of disease classified according to the Durban criteria during the first 6 months from the diagnosis (onset type) and in addition, during the subsequent followup (course type). Data on HLA allele frequencies among Finnish bone marrow donors (14,752 people) were used to represent frequencies in the Finnish background population. HLA-A, C, B, DRB1, and DQB1 typing was performed in all JIA patients, their parents, and 45 healthy siblings. This family-based approach enabled unambiguous determination of haplotypes for each sibling. HLA polymerase chain reactions were performed according to the Dynal^® Allset^TM SSP kit protocol (Dynal, Oslo, Norway) and serologic typing or sequence-based typing by Visible Genetics^® GeneKit^TM (Visible Genetics, Toronto, ON, Canada).

Odds ratios (OR) and 95% confidence intervals (CI) were applied to compare the HLA allele frequencies between JIA patients and bone marrow donors. The affected sibpair method compares the observed sharing of parental alleles or haplotypes to that expected. Expected sharing is calculated using Risch⁹ formulation of the sibling relative risk (l_S) as a ratio of the prevalence of JIA in siblings of the affected probands (estimated to be 15)⁴ and the prevalence of JIA in the Finnish population (estimated to be 0.1%)¹⁰. Standard chi-square goodness of fit test, with 2 degrees of freedom, was used to obtain a p value.

Two association-based methods were used: the independent transmission disequilibrium test (TDT) association and the family-based association test (FBAT). The classical TDT compares the transmission of alleles from heterozygous parents to the affected offspring with the number of times transmission does not occur. The JIA-affected sibpairs were treated as independent and the p value was adapted using a within-family Monte Carlo permutation z¢ score procedure with continuity correction¹¹. The p value for z¢ score was obtained using standard normal distribution. Multiple comparisons corrections to p values were made by the Bonferroni method.

Family-based association analysis of the JIA and HLA alleles used disease versus healthy phenotypes¹². In this study, we applied family-based association tests assuming linkage using the FBAT program¹³. We tested the null hypothesis of no association in the presence of linkage between HLA markers and any gene influencing the JIA. The FBAT program calculates the score, which is a summation over all offspring and all families of phenotype and marker value products. Because we had more than one sibling per family and we assumed linkage is present, sibling genotypes in the data are correlated and therefore the empirical variance estimate score test statistic is the proper choice for testing the association. The Z score test statistic calculated by the FBAT program avoids biases due to population stratification, mis-specification of the trait distribution, and/or selection based on the phenotype (ascertainment). Each HLA locus was analyzed by univariate analysis (allelic effects and overall test) and also joint allelic main effects were analyzed to assess the independent allele effects. Then haplotype analysis was undertaken based on the DRB1 and DQB1 loci. The observed association between candidate allele and JIA was considered statistically significant if 0.01 ≤ p < 0.05, very significant if 0.001 ≤ p < 0.01, and highly significant if p < 0.001. Exact p values are not reported if p < 0.001.

RESULTS

The frequencies of HLA-A, Cw, B, and DRB1 alleles in the affected subjects were first compared with a large series of Finnish bone marrow donors. HLA results are shown at the serological level because the HLA results on bone marrow donors were available only at serological level (Table 1). In JIA patients frequencies of the following HLA alleles were significantly increased: Cw4, B27, B35, and DR8. HLA-A2 was of borderline significance (OR 1.4, 95% CI 1.0–1.9). The frequencies of the alleles Cw6, DR2, and DR4 in JIA patients were significantly lower than those in bone marrow donors. The prevalence of DR5 did not differ from that in controls. The sharing of parental haplotypes was studied among the affected sibpairs. The observed ratio of sharing 0, 1, or 2 HLA haplotypes was 10:23:17, which was statistically highly significant (p < 0.001), if the expected sharing (0.83:25:24.17) was calculated using the formula devised by Risch⁹ by taking into account the estimate of the relative risk for a sib, l_S = 15⁴.

Table 1. Odds ratios (OR) and their 95% confidence intervals (CI) of HLA-C, B, and DR alleles with significant difference in allele frequencies between the Finnish patients with JIA and bone marrow donors.

Family-based multi-allelic locus-specific associations of DRB1 and DQB1 were statistically highly significant (Table 2). In the locus-specific analysis using the independent TDT with Bonferroni correction, highly significantly increased transmissions were observed for DRB1*0801 and DQB1*0402 alleles, while HLA-Cw*0401 was significantly increased (Table 3). Transmission of Cw*0701 and DQB1*0302 alleles was significantly decreased. Results of the locus-specific univariate family-based association tests were comparable to those obtained with the independent TDT: highly significant positive associations were seen for DRB1*0801 and DQB1*0402. There was a very significant positive association for Cw*0401 and for DQB1*0501 and B*3501 (Table 3). To assess the independent allele effects of the HLA-A, C, B, DRB1, and DQB1 loci, we analyzed the joint allelic main effects using the FBAT (Table 4). At the C locus, Cw*0401 (positive association) and Cw*0701 (negative association) were very significant. No significant effects were detected at the A locus, and at the B locus B*0801 was the only associated allele (negative association). At the DRB1 locus, DRB1*0801 was strongly positively associated with JIA, while DRB1*0101 (positive association), DRB1*0301 (negative association), and DRB1*0401 (negative association) were also significant. At the DQB1 locus both DQB1*0302 (negative association) and DQB1*0402 (positive association) were highly significant, while DQB1*0201 (negative association) and DQB1*0501 (positive association) were significant, when looking for independent effects.

Table 2. Family-based multiallelic association test of HLA-A, C, B, DR, and DQ loci in Finnish families with JIA using chi-square analysis.

Table 3. Locus-specific univariate allelic associations using independent TDT and family-based association tests (FBAT) in Finnish families with JIA.

Table 4. Family-based association test of the joint allele main effects of the HLA-A, C, B, DR, and DQ loci in Finnish families with JIA.

Haplotype analysis of the DRB1 and DQB1 loci was also performed, since it is known that the DRB1 locus, with linkage disequilibrium with DQB1, may be the major HLA locus determining susceptibility to JIA. The haplotype DRB1*0801, DQB1*0402 was highly significantly positively associated (p = 0.0001) and was also the most common haplotype among our JIA patients. Haplotypes DRB1*0101, DQB1*0501 (positive association) and DRB1*0401, DQB1*0302 (negative association) were statistically suggestive (Table 5).

Table 5. Family-based association tests of DR and DQ haplotypes in Finnish families with JIA.

DISCUSSION

Familial clustering is best studied in relatively isolated populations such as the Finns. To date, only one study has been published dealing with the HLA allele distribution in Finnish patients with JIA¹⁴. Ours is the first study describing whole HLA haplotypes in Finnish patients with JIA. We first looked at the HLA allele distribution in our sibling series. The study sample was too small to carry out separate analyses of JIA subgroups. Based on the entire study material, significantly increased frequencies among the affected siblings compared to control cases were found for the HLA-Cw4, B27, B35, and DR8 alleles; DR8 was the strongest. These HLA alleles also have been found to be increased in some earlier studies^5,15,16.

Univariate locus-specific multi-allelic tests for association showed significant associations for the DRB1 and DQB1 loci. Univariate locus-specific and joint allele main effect analyses supported the important role of DRB1 and DQB1 loci as well as some contribution of C locus. Association studies using the FBAT and TDT analysis found similar statistically significant associations. Transmissions of the HLA-Cw*0401 allele, and especially DRB1*0801 and DQ B1*0402 alleles, were markedly increased. In another TDT-based study dealing with nuclear Finnish families with a single affected offspring, in addition to DRB1*0801, DRB1*1101 (DR5 specificity) also showed significantly increased transmission¹⁷. In our study DR5 evinced neither increased frequency in JIA patients nor increased transmission in TDT analysis (DRB1*1101). In the FBAT we were not able to assess the contribution of DR5 to the risk of JIA due to low number of informative families. Moreover, the distribution of the HLA alleles in the Finnish JIA patients did not differ from those observed in JIA patients in other European populations. The amount of data did not allow extensive analysis of haplotype effects other than DRB1 and DQB1, but we plan to investigate this in more detail in future.

Sibling series can be used to study allele and haplotype sharing. Along these lines, 2 small series detailing haplotype sharing^18,19 and a larger one of allele sharing²⁰ have been published. A marked excess of shared haplotypes was recorded in the 2 first series. Prahalad, et al²⁰ focused their study on the DR locus and reported that, as affected siblings only had been tested and not other family members, information on identity by descent could not be obtained²¹. The authors showed a significant excess of allele sharing and computed a sibling recurrence risk of 2.5 for the DR locus. In addition to the affected siblings, in this study we typed all the parents and most of the unaffected sibs. This family-based approach enabled an unambiguous determination of haplotype and allele-sharing by descent in each instance.

A characteristic feature of JIA is the interaction of HLA genes within the same locus or between different loci that increases the risk of the disease⁵. Some HLA associations are due to linkage disequilibrium between markers involved, but accumulating evidence indicates that several different genes in the HLA region are implicated in the pathogenesis of JIA²². It is therefore difficult to assess the independent role of different HLA loci and alleles contributing to the genetic component associated with the HLA region.

Families comprising multiple siblings with the disease may differ genetically from families where only a single offspring is affected. For instance, in families with multiple affected siblings both parents may carry disease susceptibility genes or one parent may have inherited such genes from both of his/her parents. As the penetrance of the JIA susceptibility genes is not particularly high, as shown by twin studies (the concordance rate is about 25% in monozygotic twins⁷), the likelihood is that in families with multiple cases of JIA, parents possess more than one JIA susceptibility haplotype.

ACKNOWLEDGMENT

Data on HLA allele frequencies among Finnish bone marrow donors were kindly provided by Dr. Jukka Partanen from the Finnish Red Cross Blood Service.

REFERENCES

Search PubMed for:

2. Murray K, Thompson SD, Glass DN. Pathogenesis of juvenile chronic arthritis: genetic and environmental factors. Arch Dis Child 1997;77:530-4. [MEDLINE]

3. Thomson W, Silman AJ. Juvenile idiopathic arthritis. In: Silman AJ, Hochberg MC, editors. Epidemiology of the rheumatic diseases. 2nd ed. Oxford: Oxford University Press; 2001:72-84.

4. Glass DN, Giannini EH. Juvenile rheumatoid arthritis as a complex genetic trait. Arthritis Rheum 1999;42:2261-8. [MEDLINE]

5. Forre O, Smerdel A. Genetic epidemiology of juvenile idiopathic arthritis. Scand J Rheumatol 2002;31:123-8. [MEDLINE]

6. Risch N. Assessing the role of HLA-linked and unlinked determinants of disease. Am J Hum Genet 1987;40:1-14. [MEDLINE]

7. Savolainen A, Saila H, Kotaniemi K, Kaipiainen-Seppänen O, Leirisalo-Repo M, Aho K. Magnitude of the genetic component in juvenile idiopathic arthritis. Ann Rheum Dis 2000;59:1001-8. [MEDLINE]

8. Saila HM, Savolainen HA, Kotaniemi KM, Kaipiainen-Seppänen OA, Leirisalo-Repo MT, Aho KV. Juvenile idiopathic arthritis in multicase families. Clin Exp Rheumatol 2001;19:218-20. [MEDLINE]

9. Risch N. Linkage strategies for genetically complex traits. I. Multilocus models. Am J Hum Genet 1990;46:222-8. [MEDLINE]

10. Andersson Gare B. Juvenile arthritis: Who gets it, where and when? A review of current data on incidence and prevalence. Clin Exp Rheumatol 1999;17:367-74. [MEDLINE]

11. Spielman RS, Ewens WJ. A sibship test for linkage in the presence of association: the sib transmission/disequilibrium test. Am J Hum Genet 1998;62:450-8. [MEDLINE]

12. Laird N, Hovarth S, Xu X. Implementing a unified approach to family based tests of association. Genet Epidemiol 2000;19 Suppl 1:S36-S42.

13. Program for Population Genetics FBAT Web Page [Internet]. Departments of Biostatistics and Environmental Health, Harvard School of Public Health. [cited July 6, 2004]. Available from: http://www.biostat.harvard.edu/~fbat/default.html

14. Makela A-L, Tiilikainen A. HLA genetics in children with juvenile rheumatoid arthritis. Laeknabladid 1978;64 Suppl 3:113-5.

15. Howard JF, Sigsbee A, Glass DN. HLA genetic and predisposition to JRA. J Rheumatol 1985;12:7-12. [MEDLINE]

16. Murray KJ, Moroldo MB, Donnelly P, et al. Age specific effects of juvenile rheumatoid arthritis-associated HLA alleles. Arthritis Rheum 1999;42:1843-53. [MEDLINE]

17. Rundstadler JA, Säilä H, Savolainen A, et al. Analysis of MHC region genetics in Finnish patients with juvenile idiopathic arthritis: evidence for different locus specific effects in polyarticular vs. pauciarticular subsets and a shared DRB1 epitope. Genes Immun 2003;4:326-35. [MEDLINE]

18. Suciu-Foca N, Godfrey M, Jacobs J, et al. Increased frequency of DRw5 in pauciarticular JRA. In: Terasaki P, editor. Histocompatibility testing. Los Angeles: UCLA Tissue Typing Laboratory; 1980:953.

19. Clemens LE, Albert E, Ansell BM. Sibling pairs affected by chronic arthritis of childhood: evidence for a genetic predisposition. J Rheumatol 1985;12:108-13. [MEDLINE]

20. Prahalad S, Ryan MH, Shear ES, Thomson SD, Giannini EH, Glass DN. Juvenile rheumatoid arthritis. Linkage to HLA demonstrated by allele sharing of affected sibpairs. Arthritis Rheum 2000;43:2335-8. [MEDLINE]

21. Holmans P. Affected sib-pair methods for detecting linkage to dichotomous traits: review of the methodology. Hum Biol 1998;70:1025-40. [MEDLINE]

22. Zeggini E, Donn RP, Ollier WE, Thomson W. Evidence for linkage of HLA loci in juvenile idiopathic oligoarthritis. Independent effects of HLA-A and HLA-DRB1. Arthritis Rheum 2002;46:2716-20. [MEDLINE]