REVIEW A review of the MHC genetics of rheumatoid arthritis

(2004) 5, 151 157 & 2004 Nature Publishing Group All rights reserved 1466-4879/04 $25.00 www.nature.com/gene REVIEW A review of the MHC genetics of rheumatoid arthritis JL Newton 1, SMJ Harney 1, BP Wordsworth 1 and MA Brown 1 1 Institute of Musculoskeletal Sciences, University of Oxford, The Botnar Research Centre, Nuffield Orthopaedic Centre, Headington, Oxford, UK Rheumatoid arthritis is a common complex genetic disease, and, despite a significant genetic element, no gene other than HLA-DRB1 has been clearly demonstrated to be involved in the disease. However, this association accounts for less than half the overall genetic susceptibility. Investigation of other candidate genes, in particular those that reside within the major histocompatibility complex, are hampered by the presence of strong linkage disequilibrium and problems with study design. (2004) 5, 151 157. doi:10.1038/sj.gene.6364045 Published online 29 January 2004 Keywords: major histocompatibility complex; rheumatoid arthritis, genetics The genetic contribution to rheumatoid arthritis (RA) RA has a strong genetic component but the exact heritability is uncertain. The best estimates of heritability come from twin studies. The reported concordance rates have varied between studies, most likely because of differences in the severity of the cases studied. The MZ concordance rate for RA is four times greater than the dizygotic (DZ) twin concordance rate, indicating a heritability of 40 60%. 1 3 The overall MZ twin concordance rate is 12 15%. These twin studies provide an upper limit to the genetic contribution to RA. Studies including twins with milder disease have found lower MZ concordance rates, most likely because HLA-DRB1, clearly the major gene in RA, significantly affects disease severity. 4 Major histocompatibility complex (MHC) The only region of the genome that has been consistently shown to be associated with disease is the MHC. The MHC, situated on chromosome 6 (6p21.3), extends over 3.6 Mb (Figure 1). It is divided into three regions, classes I, II and III. The class I region, at the telomeric end of the MHC, contains the HLA class 1 genes, HLA-A, -B and C, and extends over 2000 kb. In the HLA class II region are the HLA-DR, -DP and -DQ loci, encoding the a and b chains of the various HLA class II molecules. The class III region lies between the class I and II regions. The MHC is a highly gene dense region containing about 220 genes, many of which have immunoregulatory functions. 5 Correspondence: MA Brown, Institute of Musculoskeletal Sciences, University of Oxford, The Botnar Research Centre, Nuffield Orthopaedic Centre, Windmill Road, Headington, Oxford, UK. E-mail: mbrown@well.ox.ac.uk Received 08 September 2003; revised 03 November 2003; accepted 04 November 2003 The shared epitope (SE) hypothesis The association of RA with HLA-Dw4 was first reported by Stastny in 1976. 6 The development of higher resolution HLA-DRB1 genotyping led to the demonstration that different HLA-DR4 alleles are not equally associated with RA. Further, studies in different populations demonstrated that non-dr4 HLA-DRB1 alleles were also associated with disease. These findings indicated that the complexity of the HLA-DRB1 association was greater than supposed previously. Gregerson and co-workers first reported a unifying hypothesis for the association of different HLA-DRB1 specificities associated with RA, termed the SE hypothesis. They demonstrated that RA is associated with specific HLA-DRB1 (DRB1) alleles that encode a conserved sequence of amino acids, ( 70 QRRAA 74, 70 RRRAA 74 or 70 QKRAA 74 ) comprising residues 70 74 in the third hyper variable region (HVR3) of the DRb1 chain. 7 These residues constitute an a helical domain forming one side of the antigen binding site, a site likely to affect antigen presentation. The alleles carrying this nucleotide sequence are DRB1*0401, *0404, *0405, *0408, *0101, *0102, *1402, *09 and *1001. In different ethnic groups the predominant RA-associated alleles vary considerably: *0401 and *0404 are the predominant RA associated alleles in Caucasians, *0405 in Japanese and *0101 in Israeli Jews. Association with DRB1*09 has been described in Chilean RA patients, Japanese and more recently UK Caucasians. 8 In contrast, there are other alleles that are negatively associated with RA and therefore provide a protective role (DRB1*0103, *0402, *0802, *1302). 8 The SE hypothesis assumes that these particular class II molecules are directly involved in the pathogenesis of RA, but the exact mechanism remains unknown. If the pathogenic process is the simple presentation of an arthritogenic peptide, a dominant mode of inheritance would be likely. Recurrence risk modelling has rejected dominant models of inheritance and suggests the presence of more than one MHC susceptibility gene. 9 11

152 telomeric centromeric Figure 1 FLOT BAT5 DDAH 2 CLIC HSPAIL RAGE NOTCH4 DRB1 TAP2 HLA - E Transcriptional Orientation The MHC region. HLA- C HLA-B MICA MICB NFKBIL1 TNF BAT2 HSP CLUSTER DQA1 DQB1 LMP TAP1 DMA 300kb 1930kb 1110kb Class I region Class III region Class II region A further complexity is that there is a hierarchy of strength of the association of the different SE-positive HLA-DRB1 alleles and RA, and that some combinations of SE alleles carry greater risk than homozygosity for those alleles. For example, the DRB1*0401/*0404 compound heterozygote genotype is strongly associated with early disease onset and a more severe disease phenotype than either DRB1*0401 or DRB1*0404 homozygosity. 12 14 Carriage of DRB1*0401/DRX, where X is a non- SE-encoding allele, has a relative risk (RR) for developing RA of 4.7, for DRB1*0404/DRX the RR ¼ 5.0, *0401/*0401 the RR ¼ 18.8 and for DRB1*0401/DRB1*0404 heterozygotes the RR ¼ 31. 14,15 The effect size of the association of DRB1*0404/*0404 homozygosity has, however, not yet been formally compared with DRB1*0401/DRB1*0404 heterozygotes. It is possible that the heterozygote association may prove intermediate between the association of DRB1*0401 and *0404 homozygotes, contrary to the prediction of the compound heterozygosity theory. In this regard, it is important to note that the combination *0404,5,8/*0404,5,8 has RR ¼ 36.2, greater than the magnitude of the association of DRB1*0401/DRB1*0404 heterozygosity with RA. 15 Calculations from the extent of sharing of MHC alleles identical by descent within families suggest that the contribution from the MHC is B30% of the total genetic effect. 16 This may be an underestimate, as in many families the parents may carry more than one MHC haplotype bearing SE-positive HLA-DRB1 alleles. While the DRB1 association with RA is robust, the penetrance of the genotype is low, with B30% of the normal UK Caucasian population being HLA-DRB1*04 positive. Further, in Caucasians, B30% of cases do not carry an SE-encoding allele. Indeed, in some studies of patients with mild disease, no association with HLA-DRB1 is apparent. The absolute risk for developing RA if carrying an SE-encoding allele is one in 35 for *0401, one in 20 for *0404, one in 80 for *0101 and one in 7 for *0401/*0404 compound heterozygotes. 17 Thus the RR is high but the absolute risk is relatively low, and although the associations with the MHC are strong, the presence of the SE is neither necessary nor sufficient for disease to occur. HLA-DRB1*04 or HLA-DQB1*03? As with many diseases with MHC associations, the extent and complexity of LD within the MHC has complicated attempts to define precisely the diseasecausing genes. The serologically determined specificities, DQ3 and DR4, encoded at the HLA-DQB1 (DQB1) and - DRB1 loci, respectively, are both strongly associated with RA in Caucasians. It has been proposed that DQB1*03 and *05 alleles are in fact the true disease-susceptibility alleles in RA, and that the DRB1 associations with disease are merely the result of linkage disequilibrium (LD) with these alleles. Support for this hypothesis comes from studies in collagen-induced arthritis in HLA- DQ transgenic mice. Transfection of susceptible mice with HLA-DQ8 (DQB1*0302 is the human equivalent) resulted in the development of an inflammatory arthritis similar to RA after immunization with type II collagen. In contrast, transfection with a non-ra-associated gene, HLA-DQ6, created a phenotype resistant to the development of inflammatory arthritis. 18 22 Some association studies in humans have suggested a direct role for DQ alleles in RA. 23 However, further larger studies have not supported this hypothesis. 8,24 Milicic and co-workers studied 685 RA patients and a total of B14 000 healthy bone marrow donors, and found no role for DQB1 alleles independent of the SE. Several studies have, however, demonstrated the association of the HLA-DRB1*0401-DQB1*0301 haplotype with greater disease severity, positive IgM rheumatoid factor and greater degrees of joint deformity. 25 Therefore, it has been suggested that DR/DQ complementarity may affect the clinical expression of RA in humans as well as mouse models. However, SE-encoding alleles are also increased in these clinical subgroups, quite likely contributing to their more severe phenotype. Carefully designed studies, particularly in Felty s syndrome, also strongly suggest that the primary association is with DRB1*04 alleles. 26 In conclusion, there is no definitive evidence that DQB1 alleles influence susceptibility to RA, but there may be an influence on the clinical expression of the disease from DR DQ complementarity. Other MHC candidate genes Considerable research has gone into trying to identify the further MHC genes, which are likely to be involved in RA (see Table 1 for a summary). The most widely studied locus is TNF, which lies 1000 kb from HLA-DRB1 within

Table 1 Summary of the previous MHC candidate gene and mapping studies in RA listed in chronological order 153 Chromosome Year Gene(s) Sample size Result Angelini et al 44 1992 HLA-DP, -DQ, -DR 48 cases, 109 controls Negative Perdriger et al 45 1992 HLA-DPB1 71 cases, 148 controls RR ¼ 2.74, Po0.03 Brinkman et al 46 1994 TNF 13 cases, 88 controls Negative Vandevyver et al 47 1994 TNF 77 cases, 58 controls Negative Singal et al 48 1994 TAP2 Suggest a role for TAP polymorphisms in DR4-positive patients Wilson et al 49 1995 TNF 147 cases, 135 controls Negative Vandevyver et al 50 1995 TAP 82 cases, 66 controls Negative Maksymowych et al 51 1995 LMP 2 168 cases, 210 controls Negative Wordsworth et al 52 1995 AP genes 60 cases, 60 controls Negative Mulcahy et al 53 1996 TNF 50 multiplex families Positive for a TNF effect independent of DRB1 Hillarby et al 54 1996 TAP2D 89 cases, 64 controls OR ¼ 2.6 (1.2 5.8) Field et al 55 1997 TNF 98 cases, 91 controls Negative Brinkman et al 56 1997 TNF 283 cases, 116 controls TNF-238GA OR ¼ 4.1 (1.0 17) Vinasco et al 57 1997 TNF 60 cases, 102 controls Negative Vinasco et a7 58 1997 HSP 70 Negative Singal et al 59 1998 MHC microsatellites 97 cases, 100 controls Positive for a second MHC susceptibility locus Tuokko et a8 60 1998 TAP genes 40 cases, 60 controls P ¼ 0.024 Vinasco et al 61 1998 TAP genes 60 cases, 102 controls Negative van Krugten et al 62 1999 TNF 112 cases, 138 controls RR ¼ 3.89 Waldron-Lynch et al 63 1999 TNF 23 cases, 10 controls Negative Seki et al 64 1999 TNF 387 cases, 575 controls Negative Bali et al 65 1999 HLA region 60 multiplex families Positive for an independent role of TNF-c alleles Singal et al 39 1999 MHC 97 cases, 95 controls Positive for a second MHC susceptibility locus Perdriger et al 66 1999 HLA-DMB1 163 cases, 146 controls Possible role for HLA-DM alleles Singal et al 67 1999 HLA-D region 20 cases, 20 controls DRB1and/or inappropriate expression of certain DR genes contributes to RA development Martinez et al 28 2000 TNF 52 families TNFa6b5 positively associated with disease Hajeer et al 27 2000 TNF 179 cases, 145 controls Negative Shibue et al 68 2000 TNF 545 cases, 265 controls Negative Singal et al 69 2000 HLA class III region 97 cases, 95 controls Positive for a second MHC susceptibility locus Vejbaesya et al 70 2000 TAP genes 82 cases, 100 controls Negative Jenkins et al 71 2000 HSP 70 60 families P ¼ 0.003 Castro et al 72 2001 TNF 79 cases, 69 controls TNFa6 Po0.0076 Waldron-Lynch et al 30 2001 TNF 33 multiplex families Positive for 308 and 857 depending on the SE status Hadj Kacem et al 73 2001 HLA-DQB1 CAR1/CAR2 60 cases, 150 controls Negative for TNF, positive for HLA-DQB1, CAR1/CAR2 Ota et al 74 2001 70kb region telomeric of TNF 120 cases, 248 controls Positive for a second susceptibility locus telomeric of TNF Tuokko et al 75 2001 HLA haplotypes in RA 67 cases, 77 controls Suggest loci outside DRB1 contribute to RA susceptibility Zanelli et al 35 2001 HLA region 132 cases, 254 controls Positive for a second susceptibility locus in the telomeric HLA region Singal et al 76 2001 MICA 90 cases, 85 controls Negative Martinez et al 77 2001 MICA 54 families, 211 cases, OR ¼ 0.39, P ¼ 0.007(TDT), P ¼ 0.0005 (case control) 200 controls Low et al 78 2002 TNF 238 cases, 217 controls Negative Udalova et al 79 2002 TNF 81 cases, 176 controls P ¼ 0.02 863, OR ¼ 1.89 (1.04 3.41) Cvetkovic et al 80 2002 TNF 154 cases and controls Allele 1 Po0.01 (OR ¼ 1.62) Jawaheer et al 38 2002 HLA complex 469 multi-case families Positive for two additional genetic effects to DRB1 Pascual et al 81 2002 HLA haplotypes 147 cases, 202 controls Positive for a second RA susceptibility factor Okamoto et al 37 2003 I kappa BL 116 cases, 100 controls P ¼ 0.0062 Newton et al 34 2003 TNF 300 cases, 300 controls P ¼ 0.007 with one haplotype the MHC class III region. These studies have reported conflicting findings, and very few of them have followed designs that would control fully for LD with DRB1. A common approach has been to match cases and controls either for the carriage of DRB1*04or for carriage of SEcontaining DRB1 alleles. The distribution of SE carrying DRB1 alleles, or DR4 alleles, may then differ between cases and controls, because of their differential strength of the association with RA. Thus, previous studies of other MHC markers in RA may have been affected by LD as well as by the effects of the disease association that they were designed to investigate. These two effects may be indistinguishable. 27 30 These problems are not unique to RA, affecting studies of any disease with strong MHC associations. Several different approaches have been taken to overcoming this problem. Using within-family association, studies have compared the transmission of haplotypes within families where the parents carry more than one copy of an HLA- DRB1 susceptibility allele (the Homozygous Parent Test 31 (Figure 2). Thus, for example, if there are two parental haplotypes carrying HLA-DRB1*0401, the

154 a b DRB1*0401-TNF*2 DRB1*0403-TNF*3 DRB1*0401-TNF*3 DRB1*0401-TNF*2 DRB1*0401-TNF*2 DRB1*0404-TNF*2 DRB1*0401-TNF*3 DRB1*0403-TNF*4 DRB1*0404-TNF*2 DRB1*0701-TNF*4 DRB1*0403-TNF*4 DRB1*0403-TNF*3 Figure 2 TDT studies of the MHC. Standard TDT compares the transmitted and untransmitted allele frequencies. In (a), if looking at the TNF alleles only, there may appear to be an overtransmission of TNF*2, but when the DRB1 data are included it becomes clear that this is likely just to represent overtransmission of DRB1 SE carrying alleles. In (b), each parent is homozygous for the same DRB1 allele so the role of the TNF alleles can be specifically examined. preferential transmission of one of those haplotypes compared with the other may indicate the presence of a further susceptibility allele on the overtransmitted haplotype. This approach has been used with success in type I diabetes, but requires very large sample sizes to obtain sufficient informative families. A further approach is to pool on the basis of carriage of specific DRB1 alleles (ie antigen positivity ), but this assumes that both cases and controls are in Hardy Weinberg equilibrium. In RA cases, the DRB1 alleles are not in Hardy Weinberg equilibrium due to the strong disease association of this locus, and therefore such studies are still susceptible to bias due to LD with DRB1. Lastly, a form of logistic regression applied to the transmission disequilibrium test has been developed, which allows comparison of transmission rates of haplotypes identical at the candidate locus, but different at the secondary locus. This method loses power rapidly with increasing LD such as one would expect across the MHC, and is therefore not ideal for this application. 32 Full matching of haplotypes in cases and controls at HLA-DRB1 does effectively control for LD. Until recently, such haplotypic studies have required the use of families for both cases and controls. Although the nontransmitted parental haplotypes could be used as controls, this method requires large data sets to obtain sufficient power. Further, as described below, the nontransmitted maternal SE-bearing haplotypes may not be randomly distributed, and may influence disease susceptibility themselves, thereby invalidating their use as a control group. The development of accurate Bayesian approaches to determining haplotype frequencies in unrelated cases and controls has permitted a novel powerful and efficient method that overcomes this problem. 33 Using this method, we have recently studied the role of the TNF gene in RA. Studying only DR4-positive cases and controls, the frequencies of different LTA-TNF haplotypes were compared between cases and control haplotypes carrying the same DRB1*04 allele. A single LTA-TNF haplotype (termed LTA-TNF2) was identified as modifying the DRB1 associations with RA; being overrepresented on case *0404 haplotypes (P ¼ 0.007) and under-represented on case *0401 haplotypes (P ¼ 0.007). These results strongly support the presence of an additional MHC susceptibility locus or loci outside of the TNF region. 34 However, as the knowledge of the haplotype structure of this area is currently limited further, large studies are necessary to rule out the LT- TNF region definitively. There is an increasing body of evidence implicating the region telomeric to TNF of harboring additional RAsusceptibility elements. 35,36 Ota et al, using 18 microsatellite markers across a 3.6 Mb region of the MHC and five TNF SNPs, in 120 patients and 248 controls, identified a 70 kb region telomeric of TNF that demonstrated association with RA. No LD was seen with DRB1*0405, but this was only assessed on the basis of carriage of *0405 rather than at a haplotypic level, and no assessment was carried out of other RA-associated DRB1 alleles. Thus this study may still be affected by LD with DRB1. A subsequent paper by Okamoto et al concluded that the true disease-associated variant was an SNP (SNP 96452) within the IkBL gene. 37 No LD studies between the SNPs and DRB1 were reported. It appears that the results from the preceding paper, showing that the microsatellites were not in LD with DRB1, were extrapolated to conclude that the SNPs in this region were also not in LD. In British Caucasians, this SNP is in significant LD with the DRB1 locus (unpublished data). Okamota and co-workers also demonstrated stronger association of a haplotype of three markers than with SNP 96452. If the conclusion is that this SNP is directly involved in the susceptibility to RA, then the strongest association should have been with this SNP alone. The results do provide evidence of an additional susceptibility region for RA, but do not necessarily implicate SNP 96452 specifically. Further evidence as to the complexity of the MHC associations of RA came from the work of Jawaheer et al 38 who genotyped a total of 54 microsatellite markers across the MHC in 469 multicase RA families. There was significant association with one microsatellite haplotype on the background of the ancestral A1-B8-DRB1*03 haplotype, a non-se-encoding haplotype. The region of association covered a B500 kb segment of the central MHC, which did not include DRB1. There was also evidence of an additional susceptibility element on the background of DRB1*0404 in the class I region of the MHC. These findings are in keeping with our results showing different disease-modifying elements on the background of DRB1*0401 and DRB1*0404. Singal et al 39 provided evidence for an additional susceptibility region in the Bat2-Hsp region of the class III MHC in SEnegative patients. This study using B100 cases and controls confirmed LD between these microsatellites and DRB1. There was not sufficient matching to control for this confounding effect when analyzing the SE-positive group, but analyses on the SE-negative group demonstrated association with the Bat2 138 allele and the

D6S273 138 allele. Zanelli et al genotyped six microsatellites residing in the class III region and the centromeric portion of the class I region in 132 RA patients and 254 controls. There was evidence for a susceptibility factor in the region telomeric to TNF independent of DRB1. 35 In conclusion, there is compelling evidence against a direct role for polymorphisms in the TNF gene contributing directly to RA susceptibility and an increasing body of support for additional susceptibility elements elsewhere in the MHC, in particular, the regions telomeric to the TNF gene. This is a highly gene dense region and well-designed haplotype studies that adequately control for the strong LD will be required to finely localize the actual susceptibility elements. Is there a role for noninherited maternal alleles (NIMAs)? As well as the previously discussed associations of HLA genotype and RA, there has also been interest in the role of the NIMAs in RA susceptibility. Even in hospitalbased studies of RA no more than 87% of patients are positive for the SE. A possible explanation for the occurrence of SE-negative cases is the exposure to noninherited maternal SE-positive HLA antigens during fetal development, or even persistent exposure due to microchimerism. This has been investigated by comparing the occurrence of SE-positive or DRB1*04-positive NIMA and NIPA in DRB1*04-negative cases. Initial Dutch studies suggested that there was an excess of HLA-DRB1*04 NIMA in SE-negative RA patients. 40 Studies since then have been divided and there is no clear consensus. 41 43 Combining all five reported data sets, an increased prevalence of DRB1*04- and SEpositive NIMA has been reported in SE-negative cases compared with the prevalence of DRB1*04- and SEpositive NIPA (odds ratio 2.1, P ¼ 0.003; odds ratio 2.0, P ¼ 0.04, respectively). 43 Marked heterogeneity between data sets was observed, and there is a clear need for sufficiently large study to address this question definitively. Conclusion The study of complex diseases is currently the biggest challenge in genetics. For complex heterogenous diseases, like RA, the accuracy and sensitivity of association studies is impeded by genetic epistasis, where the genotype at one locus influences the phenotypic expression of the genotype at another locus, and genetic heterogeneity, where different combinations of genes may produce the RA phenotype. Small sample sizes, subgroup analysis with multiple testing and poorly matched control groups are all potential design pitfalls. With the completion of the human genome sequence, the daily increasing number of publicly available SNPs and the production of high-throughput genotyping methods, the resources available for identifying the genetic susceptibility elements are improving. In parallel with these developments there is also likely to be a flurry of association studies increasing the need for well-characterized patient and family populations and careful study design to eliminate as many confounding factors as possible. References 1 Silman AJ, MacGregor AJ, Thomson W et al. Twin concordance rates for rheumatoid arthritis: results from a nationwide study. Br J Rheumatol 1993; 32: 903 907. 2 Aho K, Koskenvuo M, Tuominen J, Kaprio J. Occurrence of rheumatoid arthritis in a nationwide series of twins. J Rheumatol 1986; 13: 899 902. 3 Lawrence JS. The epidemiology and genetics of rheumatoid arthritis. Rheumatology 1969; 2: 1 36. 4 Silman AJ. Problems complicating the genetic epidemiology of rheumatoid arthritis. J Rheumatol 1997; 24: 194 196. 5 Milner CM, Campbell RD. Genetic organization of the human MHC class III region. Front Biosci 2001; 6: D914 D926. 6 Stastny P. Mixed lymphocyte cultures in rheumatoid arthritis. J Clin Invest 1976; 57: 1148 1157. 7 Gregersen PK, Silver J, Winchester RJ. The shared epitope hypothesis. An approach to understanding the molecular genetics of susceptibility to rheumatoid arthritis. Arthritis Rheum 1987; 30: 1205 1213. 8 Milicic A, Lee D, Brown MA, Darke C, Wordsworth BP. HLA-DR/DQ haplotype in rheumatoid arthritis: novel allelic associations in UK Caucasians. J Rheumatol 2002; 29: 1821 1826. 9 Rigby AS, Silman AJ, Voelm L et al. Investigating the HLA component in rheumatoid arthritis: an additive (dominant) mode of inheritance is rejected, a recessive mode is preferred. Genet Epidemiol 1991; 8: 153 175. 10 Rigby AS, Voelm L, Silman AJ. Epistatic modeling in rheumatoid arthritis: an application of the Risch theory. Genet Epidemiol 1993; 10: 311 320. 11 Genin E, Babron MC, McDermott MF et al. Modelling the major histocompatibility complex susceptibility to RA using the MASC method. Genet Epidemiol 1998; 15: 419 430. 12 Gough A, Faint J, Salmon M et al. Genetic typing of patients with inflammatory arthritis at presentation can be used to predict outcome. Arthritis Rheum 1994; 37: 1166 1170. 13 MacGregor A, Ollier W, Thomson W, Jawaheer D, Silman A. HLA-DRB1*0401/0404 genotype and rheumatoid arthritis: increased association in men, young age at onset, and disease severity. J Rheumatol 1995; 22: 1032 1036. 14 Wordsworth P, Pile KD, Buckely JD et al. HLA heterozygosity contributes to susceptibility to rheumatoid arthritis. Am J Hum Genet 1992; 51: 585 591. 15 Hall FC, Weeks DE, Camilleri JP et al. Influence of the HLA- DRB1 locus on susceptibility and severity in rheumatoid arthritis. QJM 1996; 89: 821 829. 16 Deighton CM, Walker DJ, Griffiths ID, Roberts DF. The contribution of HLA to rheumatoid arthritis. Clin Genet 1989; 36: 178 182. 17 Nepom GT, Nepom BS. Prediction of susceptibility to rheumatoid arthritis by human leukocyte antigen genotyping. Rheum Dis Clin N Am 1992; 18: 785 792. 18 Nabozny GH, Baisch JM, Cheng S et al. HLA-DQ8 transgenic mice are highly susceptible to collagen-induced arthritis: a novel model for human polyarthritis. J Exp Med 1996; 183: 27 37. 19 Bradley DS, Nabozny GH, Cheng S et al. HLA-DQB1 polymorphism determines incidence, onset, and severity of collagen-induced arthritis in transgenic mice. Implications in human rheumatoid arthritis. J Clin Invest 1997; 100: 2227 2234. 20 Taneja V, Griffiths MM, Luthra H, David CS. Modulation of HLA-DQ-restricted collagen-induced arthritis by HLA-DRB1 polymorphism. Int Immunol 1998; 10: 1449 1457. 155

156 21 Zanelli E, Krco CJ, David CS. Critical residues on HLA- DRB1*0402 HV3 peptide for HLA-DQ8-restricted immunogenicity: implications for rheumatoid arthritis predisposition. J Immunol 1997; 158: 3545 3551. 22 Zanelli E, Breedveld FC, de Vries RR. HLA association with autoimmune disease: a failure to protect? Rheumatology (Oxford) 2000; 39: 1060 1066. 23 van der Horst-Bruinsma IE, Visser H, Hazes JM et al. HLA-DQ-associated predisposition to and dominant HLA- DR-associated protection against rheumatoid arthritis. Hum Immunol 1999; 60: 152 158. 24 de Vries N, van Elderen C, Tijssen H, van Riel PL, van de Putte LB. No support for HLA-DQ encoded susceptibility in rheumatoid arthritis. Arthritis Rheum 1999; 42: 1621 1627. 25 Cranney A, Goldstein R, Pham B, Newkirk MM, Karsh J. A measure of limited joint motion and deformity correlates with HLA-DRB1 and DQB1 alleles in patients with rheumatoid arthritis. Ann Rheum Dis 1999; 58: 703 708. 26 Lanchbury JS, Jaeger EE, Sansom DM et al. Strong primary selection for the Dw4 subtype of DR4 accounts for the HLA- DQw7 association with Felty s syndrome. Hum Immunol 1991; 32: 56 64. 27 Hajeer AH, Dababneh A, Makki RF et al. Different gene loci within the HLA-DR and TNF regions are independently associated with susceptibility and severity in Spanish rheumatoid arthritis patients. Tissue Antigens 2000; 55: 319 325. 28 Martinez A, Fernandez-Arquero M, Pascual-Salcedo D et al. Primary association of tumor necrosis factor-region genetic markers with susceptibility to rheumatoid arthritis. Arthritis Rheum 2000; 43: 1366 1370. 29 Plevy SE, Targan SR, Yang H, Fernandez D, Rotter JI, Toyoda H. Tumor necrosis factor microsatellites define a Crohn s disease-associated haplotype on chromosome 6. Gastroenterology 1996; 110: 1053 1060. 30 Waldron-Lynch F, Adams C, Amos C et al. Tumour necrosis factor 5 0 promoter single nucleotide polymorphisms influence susceptibility to rheumatoid arthritis (RA) in immunogenetically defined multiplex RA families. Genes Immun 2001; 2: 82 87. 31 Lie BA, Ronningen KS, Akselsen HE, Thorsby E, Undlien DE. Application and interpretation of transmission/disequilibrium tests: transmission of HLA-DQ haplotypes to unaffected siblings in 526 families with type 1 diabetes. Am J Hum Genet 2000; 66: 740 743. 32 Koeleman BP, Dudbridge F, Cordell HJ, Todd JA. Adaptation of the extended transmission/disequilibrium test to distinguish disease associations of multiple loci: the conditional extended transmission/disequilibrium test. Ann Hum Genet 2000; 64: 207 213. 33 Stephens M, Smith NJ, Donnelly P. A new statistical method for haplotype reconstruction from population data. Am J Hum Genet 2001; 68: 978 989. 34 Newton J, Brown MA, Milicic A et al. The effect of HLA-DR on susceptibility to rheumatoid arthritis is influenced by the associated lymphotoxin alpha-tumor necrosis factor haplotype. Arthritis Rheum 2003; 48: 90 96. 35 Zanelli E, Jones G, Pascual M et al. The telomeric part of the HLA region predisposes to rheumatoid arthritis independently of the class II loci. Hum Immunol 2001; 62: 75 84. 36 Jawaheer D, Gregersen PK. Rheumatoid arthritis. The genetic components. Rheum Dis Clin North Am 2002; 28: 1 15. 37 Okamoto K, Makino K, Yoshikawa Y et al. Identification of I kappa BL as the second major histocompatibility complexlinked susceptibility locus for rheumatoid arthritis. Am J Hum Genet 2003; 72: 303 312. 38 Jawaheer D, Li W, Graham RR et al. Dissecting the genetic complexity of the association between human leukocyte antigens and rheumatoid arthritis. Am J Hum Genet 2002; 71: 585 594. 39 Singal DP, Li J, Lei K. Genetics of rheumatoid arthritis (RA): two separate regions in the major histocompatibility complex contribute to susceptibility to RA. Immunol Lett 1999; 69: 301 306. 40 ten Wolde S, Breedveld FC, de Vries RR et al. Influence of noninherited maternal HLA antigens on occurrence of rheumatoid arthritis. Lancet 1993; 341: 200 202. 41 Barrera P, Balsa A, Alves H et al. Noninherited maternal antigens do not play a role in rheumatoid arthritis susceptibility in Europe. European Consortium on Rheumatoid Arthritis Families. Arthritis Rheum 2000; 43: 758 764. 42 Barrera P, Balsa A, Alves H et al. Noninherited maternal antigens do not increase the susceptibility for familial rheumatoid arthritis. European Consortium on Rheumatoid Arthritis Families (ECRAF). J Rheumatol 2001; 28: 968 974. 43 Harney S, Newton J, Milicic A, Brown MA, Wordsworth BP. Non-inherited maternal HLA alleles are associated with rheumatoid arthritis. Rheumatology (Oxford) 2003; 42: 171 174. 44 Angelini G, Morozzi G, Delfino L et al. Analysis of HLA DP, DQ, and DR alleles in adult Italian rheumatoid arthritis patients. Hum Immunol 1992; 34: 135 141. 45 Perdriger A, Semana G, Quillivic F et al. DPB1 polymorphism in rheumatoid arthritis: evidence of an association with allele DPB1 0401. Tissue Antigens 1992; 39: 14 18. 46 Brinkman BM, Giphart MJ, Verhoef A et al. Tumor necrosis factor alpha-308 gene variants in relation to major histocompatibility complex alleles and Felty s syndrome. Hum Immunol 1994; 41: 259 266. 47 Vandevyver C, Raus P, Stinissen P, Philippaerts L, Cassiman JJ, Raus J. Polymorphism of the tumour necrosis factor beta gene in multiple sclerosis and rheumatoid arthritis. Eur J Immunogenet 1994; 21: 377 382. 48 Singal DP, Ye M, Qiu X, D Souza M. Polymorphisms in the TAP2 gene and their association with rheumatoid arthritis. Clin Exp Rheumatol 1994; 12: 29 33. 49 Wilson AG, de Vries N, van de Putte LB, Duff GW. A tumour necrosis factor alpha polymorphism is not associated with rheumatoid arthritis. Ann Rheum Dis 1995; 54: 601 603. 50 Vandevyver C, Geusens P, Cassiman JJ, Raus J. Peptide transporter genes (TAP) polymorphisms and genetic susceptibility to rheumatoid arthritis. Br J Rheumatol 1995; 34: 207 214. 51 Maksymowych WP, Tao S, Luong M et al. Polymorphism in the LMP2 and LMP7 genes and adult rheumatoid arthritis: no relationship with disease susceptibility or outcome. Tissue Antigens 1995; 46: 136 139. 52 Wordsworth BP, Pile KD, Gibson K, Burney RO, Mockridge I, Powis SH. Analysis of the MHC-encoded transporters TAP1 and TAP2 in rheumatoid arthritis: linkage with DR4 accounts for the association with a minor TAP2 allele. Tissue Antigens 1993; 42: 153 155. 53 Mulcahy B, Waldron-Lynch F, McDermott MF et al. Genetic variability in the tumor necrosis factor-lymphotoxin region influences susceptibility to rheumatoid arthritis. Am J Hum Genet 1996; 59: 676 683. 54 Hillarby MC, Davies EJ, Donn RP, Grennan DM, Ollier WE. TAP2D is associated with HLA-B44 and DR4 and may contribute to rheumatoid arthritis and Felty s syndrome susceptibility. Clin Exp Rheumatol 1996; 14: 67 70. 55 Field M, Gallagher G, Eskdale J et al. Tumor necrosis factor locus polymorphisms in rheumatoid arthritis. Tissue Antigens 1997; 50: 303 307. 56 Brinkman BM, Huizinga TW, Kurban SS et al. Tumour necrosis factor alpha gene polymorphisms in rheumatoid arthritis: association with susceptibility to, or severity of, disease? Br J Rheumatol 1997; 36: 516 521. 57 Vinasco J, Beraun Y, Nieto A et al. Polymorphism at the TNF loci in rheumatoid arthritis. Tissue Antigens 1997; 49: 74 78. 58 Vinasco J, Beraun Y, Nieto A et al. Heat shock protein 70 gene polymorphisms in rheumatoid arthritis. Tissue Antigens 1997; 50: 71 73. 59 Singal DP, Li J, Ye M, Lei K. D6S273 microsatellite polymorphism and susceptibility to rheumatoid arthritis. Tissue Antigens 1998; 52: 353 358.

60 Tuokko J, Pushnova E, Yli-Kerttula U, Toivanen A, Ilonen J. TAP2 alleles in inflammatory arthritis. Scand J Rheumatol 1998; 27: 225 229. 61 Vinasco J, Fraile A, Nieto A et al. Analysis of LMP and TAP polymorphisms by polymerase chain reaction-restriction fragment length polymorphism in rheumatoid arthritis. Ann Rheum Dis 1998; 57: 33 37. 62 van Krugten MV, Huizinga TW, Kaijzel EL et al. Association of the TNF +489 polymorphism with susceptibility and radiographic damage in rheumatoid arthritis. Genes Immun 1999; 1: 91 96. 63 Waldron-Lynch F, Adams C, Shanahan F, Molloy MG, O Gara F. Genetic analysis of the 3 0 untranslated region of the tumour necrosis factor shows a highly conserved region in rheumatoid arthritis affected and unaffected subjects. J Med Genet 1999; 36: 214 216. 64 Seki N, Kamizono S, Yamad A et al. Polymorphisms in the 5 0 - flanking region of tumor necrosis factor-alpha gene in patients with rheumatoid arthritis. Tissue Antigens 1999; 54: 194 197. 65 Bali D, Gourley S, Kostyu DD et al. Genetic analysis of multiplex rheumatoid arthritis families. Genes Immun 1999; 1: 28 36. 66 Perdriger A, Guggenbuhl P, Chales G et al. Positive association of the HLA DMB1*0101-0101 genotype with rheumatoid arthritis. Rheumatology (Oxford) 1999; 38: 448 452. 67 Singal DP, Ye M, Buchanan WW, Qiu X. HLA-D region genes and susceptibility to rheumatoid arthritis. Transplant Proc 1997; 29: 1126. 68 Shibue T, Tsuchiya N, Komata T et al. Tumor necrosis factor alpha 5 -flanking region, tumor necrosis factor receptor II, and HLA-DRB1 polymorphisms in Japanese patients with rheumatoid arthritis. Arthritis Rheum 2000; 43: 753 757. 69 Singal DP, Li J, Zhu Y. HLA class III region and susceptibility to rheumatoid arthritis. Clin Exp Rheumatol 2000; 18: 485 491. 70 Vejbaesya S, Luangtrakool P, Luangtrakool K, Sermduangprateep C, Parivisutt L. Analysis of TAP and HLA-DM polymorphism in Thai rheumatoid arthritis. Hum Immunol 2000; 61: 309 313. 71 Jenkins SC, March RE, Campbell RD, Milner CM. A novel variant of the MHC-linked hsp70, hsp70-hom, is associated with rheumatoid arthritis. Tissue Antigens 2000; 56: 38 44. 72 Castro F, Acevedo E, Ciusani E et al. Tumour necrosis factor microsatellites and HLA-DRB1*, HLA-DQA1*, and HLA- DQB1* alleles in Peruvian patients with rheumatoid arthritis. Ann Rheum Dis 2001; 60: 791 795. 73 Hadj Kacem H, Kaddour N, Adyel FZ, Bahloul Z, Ayadi H. HLA-DQB1 CAR1/CAR2, TNFa IR2/IR4 and CTLA-4 polymorphisms in Tunisian patients with rheumatoid arthritis and Sjogren s syndrome. Rheumatology (Oxford) 2001; 40: 1370 1374. 74 Ota M, Katsuyamana Y, Kimura A et al. A second susceptibility gene for developing rheumatoid arthritis in the human MHC is localized within a 70-kb interval telomeric of the TNF genes in the HLA class III region. Genomics 2001; 71: 263 270. 75 Tuokko J, Nejentsev S, Luukkainen R, Toivanen A, Ilonen J. HLA haplotype analysis in Finnish patients with rheumatoid arthritis. Arthritis Rheum 2001; 44: 315 322. 76 Singal DP, Li J, Zhang G. Microsatellite polymorphism of the MICA gene and susceptibility to rheumatoid arthritis. Clin Exp Rheumatol 2001; 19: 451 452. 77 Martinez A, Fernandez-Arquero M, Balsa A et al. Primary association of a MICA allele with protection against rheumatoid arthritis. Arthritis Rheum 2001; 44: 1261 1265. 78 Low AS, Gonzalez-Gay MA, Akil M et al. TNF +489 polymorphism does not contribute to susceptibility to rheumatoid arthritis. Clin Exp Rheumatol 2002; 20: 829 832. 79 Udalova IA, Richardson A, Ackerman H, Wordsworth P, Kwiatkowski D. Association of accelerated erosive rheumatoid arthritis with a polymorphism that alters NF-kappaB binding to the TNF promoter region. Rheumatology (Oxford) 2002; 41: 830 831. 80 Cvetkovic JT, Wallberg-Jonsson S, Stegmayr B, Rantapaa- Dahlqvist S, Lefvert AK. Susceptibility for and clinical manifestations of rheumatoid arthritis are associated with polymorphisms of the TNF-alpha, IL-1beta, and IL-1Ra genes. J Rheumatol 2002; 29: 212 219. 81 Pascual M, Mataran L, Jones G et al. HLA haplotypes and susceptibility to rheumatoid arthritis. More than class II genes. Scand J Rheumatol 2002; 31: 275 278. 157