Recent advances in the immunogenetics of idiopathic inflammatory myopathy

This review summarizes the previous and current literature on the immunogenetics of idiopathic inflammatory myopathy (IIM) and updates the research progress that has been made over the past decade. A substantial part of the genetic risk for developing adult- and juvenile-onset IIM lies within the major histocompatibility complex (MHC), and a tight relationship exists between individual human leukocyte antigen alleles and specific serological subtypes, which in turn dictate clinical disease phenotypes. Multiple genetic regions outside of the MHC are increasingly being identified in conferring IIM disease susceptibility. We are still challenged with the task of studying a serologically and clinically heterogeneous disorder that is rarer by orders of magnitude than the likes of rheumatoid arthritis. An ongoing and internationally coordinated IIM genome-wide association study may provide further insights into IIM immunogenetics.


Introduction
Th e idiopathic infl ammatory myopathies (IIMs) repre sent a group of rare and heterogeneous 'orphan' auto immune diseases characterized by infl ammation of skele tal muscle and other organ systems, potentially leading to irreversible damage and resulting disability. Th e etio pathogenesis of IIM is likely to result from an interaction of genetic and environmental factors, which together are required to initiate the onset of a clinical disease phenotype [1]. IIMs traditionally have been classifi ed broadly according to a limited set of subtypes: polymyositis (PM), dermatomyositis (DM), myositis overlapping with another connective tissue disease (myositis-CTD/overlap), inclusion body myositis (IBM), and juvenile dermatomyositis (JDM). However, serological status according to circulating myositis-specifi c antibodies (MSAs) or myositisassociated antibodies (MAAs) is proving to be increasingly useful in the classifi cation of IIM and often correlates with defi ned IIM clinical phenotypes.
To date, our understanding of IIM immunogenetics has increasingly been facilitated by candidate gene studies examining the frequency of selected polymorphisms in IIM cases compared with controls. Historically, these studies have often included considerably heterogeneous IIM populations in order to increase statistical power. To facilitate meaningful research in rare diseases such as IIM, present and future approaches must include careful recruitment of confi rmed ethnically homogeneous cohorts, and this requires collaboration across national and international recruitment centers. Already, new tech nologies mean that genome-wide association scans are now the norm in the genetic investigation of complex genetic diseases. Th e remit of this article is to discuss and summarize what is currently known about the immunogenetics of IIM and to concentrate on candidate gene studies that currently provide the best evidence for a genetic basis in IIM.

The evidence for a genetic basis in myositis
Th ere is increasing evidence to suggest a genetic basis in IIM, but owing to the scarcity of aff ected sibling pairs and twins, this evidence currently comes from anecdotal familial aggregation and candidate gene studies only [2,3]. To date, no IIM familial linkage or twin studies have been published, and the 'heritability' of the disease (λs) is unknown. Th e scarcity of familial IIM cases suggests that, when encountering two or more fi rst-degree family members with symptoms and signs in keeping with possible IIM, attending physicians should be highly suspicious of making such diagnoses without careful clinicopathological correlation. If necessary, DNA or further biochemical analysis or both should also be conducted at a specialized neuromuscular center to exclude non-infl ammatory forms of neuromuscular disease such as dystrophies, mitochondrial, or metabolic myopathies.

Abstract
This review summarizes the previous and current literature on the immunogenetics of idiopathic infl ammatory myopathy (IIM) and updates the research progress that has been made over the past decade. A substantial part of the genetic risk for developing adult-and juvenile-onset IIM lies within the major histocompatibility complex (MHC), and a tight relationship exists between individual human leukocyte antigen alleles and specifi c serological subtypes, which in turn dictate clinical disease phenotypes. Multiple genetic regions outside of the MHC are increasingly being identifi ed in conferring IIM disease susceptibility. We are still challenged with the task of studying a serologically and clinically heterogeneous disorder that is rarer by orders of magnitude than the likes of rheumatoid arthritis. An ongoing and internationally coordinated IIM genomewide association study may provide further insights into IIM immunogenetics.

Familial aggregation in myositis
Th e evidence for familial aggregation in IIM arises from case reports, the fi rst of which was by Wedgwood and colleagues [4], who described JDM in twins, whose onset occurred within a year of each other. Other cases, in which two or more family members suff er from IIM (including DM, PM, IBM, and orbital and amyopathic myositis), have since been described [2].
Rider and colleagues [5] recruited 36 patients with PM, DM, or IBM from 16 unrelated families in which two or more individuals fulfi lled criteria for probable or defi nite myositis. A cohort of non-familial IIM cases (n = 181) was used as a comparison group. MSAs were more frequent in the non-familial compared with the familial myositis group. HLA-DRB1*0301 was a signifi cant risk factor in both familial and non-familial disease versus controls, but the genetic contribution in familial cases was less than that of non-familial cases (attributable risk of 0.35 versus 0.51). Homozygosity in a combined analysis of all studied HLA-DQA1 alleles combined proved a risk factor for familial disease. Th e modest diff erences between familial/non-familial IIM cases suggest that there may be further as-yet-unidentifi ed environmental and genetic factors in IIM [5].

HLA candidate gene studies
To date, the strongest immunogenetic associations found in IIM arise from the MHC region in a fashion analogous to that of other autoimmune diseases [6]. Given the rarity of IIM, early candidate gene studies examined only small numbers of patients and often grouped clinical IIM subgroups together (including PM, DM, IBM, and JDM) in an eff ort to increase statistical power. In some early studies, even patients of diff ering ethnic groups were exam ined together. Studies in the last 5 years have attemp ted to address these case-selection issues by stratify ing cases according to ethnic, clinical, and serological subtypes.
Th e HLA associations observed in IIM subgroups of Caucasian ethnicity are not generally found in other non-Caucasoid populations, although alleles of the 8.1 haplotype have been described in two studies of African-American IIM [11,18]. No HLA associations have been found in Mexican-American or Korean myositis popu-lations, illustrating the importance of stratifi cation by ethnicity in such studies [10,12].
Recent larger studies have enabled the identifi cation and characterization of HLA alleles that act as risk as well as protective factors in diff erent IIM phenotypes. For example, HLA-DRB1*0301 represents a strong PM risk factor whereas HLA-DQA1*0201 confers protection in the same clinical phenotype [12,13]. Th e concept of shared alleles within defi ned haplotypes may also contribute to this. For example, in PM, the high frequency of the 8.1 haplotype may be responsible, in part, for lowering the frequency of DRB1*07-DQA1*02-DQB1*02 and this is due to the shared DQB1*02 allele within both haplotypes. Th is combination of risk and protection may contribute to the stability and exclusivity of IIM phenotypes [23]. A primary amino acid sequence of HLA-DRB1 alleles known to share a peptide-binding motif has been described in IIM. Th is sequence is called 9 EYSTS 13 and refers to the fi rst hypervariable HLA-DRB1 region sequence, which is shared by the DRB1*03, 11, 13, and 1 4 alleles (Table 1). A strong asso ciation has been found for anti-Mi-2 positivity in a poly morphism that also incorpor ates the fi rst hypervariable region but that codes for an aromatic tryptophan residue at position 9 (HLA-DRB*01, 02, 07) [24]. Further HLA peptide-binding motifs and haplotypes that are both risk and protective factors in IIM have been described in larger-scale studies. Th ese fi ndings emphasize the importance of HLA haplotypes and peptide-binding motifs in discriminating between IIM clinical and serological subtypes [12,13,25].

H LA associations with myositis autoantibodies
HLA alleles are known to be strongly associated with the likelihood of developing MSAs/MAAs in IIM, in which the strength of association is considerably greater than that in association with stratifi cation by traditional clinical subtype (Table 2) [9,11,13,16,18,[24][25][26][27][28]. Th is suggests that stratifi cation by serological, rather than traditional clinical, subtype represents a more homogeneous form of IIM classifi cation [13,29].
Th e association of anti-Jo-1 antibodies and alleles comprising the 8.1 haplotype has been confi rmed in several studies [11,13,[29][30][31]. Th is strong relationship between HLA and serological subtype is also present for other MSAs/MAAs, in which antibodies other than anti-Jo-1 are associated with other haplotypes. However, HLA alleles comprising the 8.1 haplotype are also strongly associated with the presence of anti-PM-Scl antibodies, an MAA often associated with myositis/sclerodema overlap syndrome [28]. Th is antibody/phenotype association has been confi rmed in a UK JDM population [15]. Mierau and colleagues [24] demonstrated that HLA-DRB1*0701 represented a strong risk factor in German Caucasian anti-Mi-2 antibody-positive patients versus controls. Th is association has since been confi rmed in larger studies [13,25,26]. HLA-DQA1*0201, known to share strong linkage disequilibrium with DRB1*0701, also confers risk in anti-Mi-2 antibody-positive cases. Th e strong relationship between HLA and these MSAs/ MAAs may partly explain the exclusivity that these antibodies possess, as illustrated by the extreme rarity of multiple MSAs in the same patient. It is currently unclear whether strong associations exist with HLA alleles among rarer MSAs, although a recent US study has identifi ed potential genetic risk factors for SRP and the less frequent anti-synthetases [25].
Th ese HLA-related risk factors among serological subgroups clearly diff er according to ethnicity (Tables 1 and 2HLA associations in ethnically diff erent myositis populations). For example, African-Americans with IIM do not share the 8.1 haplotype-related risk in the same way as Caucasians. However, a recent large-scale US study showed that African-Americans with either DM or anti-Jo-1 antibodies do appear to share the risk of HLA-DRB1*0301 with Caucasians [18]. Furthermore, HLA-DRB1*0302 is a signifi cant risk factor for anti-Mi-2 antibody-positive African-American cases. Th is latter allele shares amino acid side chains coded for by the Caucasian anti-Mi-2 risk factor DRB1*0701, and further comparative analyses suggested identical orientations within the peptide-binding groove. In a recent smaller study of UK-based non-Caucasians, HLA-DRB1*03 was also detected as a risk factor in anti-Mi-2 and anti-PM-Scl antibody-positive cases [32]. Th e issue of shared HLA susceptibility risk factors across ethnic groups may suggest affi nity for common antigenic peptides. However, the results of these non-Caucasoid studies also highlight that the between-ethnic diff erences demonstrated for myositis genotype, serotype, and phenotype need to be taken into account when making future case-case or case-control comparisons.

Tumor necrosis factor-alpha
TNF-α is a proinfl ammatory cytokine with a diverse range of activities, playing a major role in immune response regulation. Th e TNF-α gene is encoded within the MHC class III region, and associated SNPs have been impli cated in many autoimmune diseases, including dermatitis herpetiformis, systemic lupus erythematosus, ankylosing spondylitis, and Crohn's disease, and also in various infectious diseases, including malaria, leprosy, and hepatitis B/C [7].
A TNF-α promoter SNP at position -308 (rs1800629), resulting from a G-to-A substitution has been associated with IIM in a number of candidate gene studies [33][34][35][36]. In a recent juvenile DM study, homozygosity for the A allele of the TNFα-308 SNP was shown to confer risk for soft-tissue calcinosis (9% calci nosis versus 1% no calcinosis, corrected P value (P corr ) = 0.045, odds ratio (OR) 7.3, 95% confi dence interval (CI) 1.4 to 37.2) [36],   [25] fi ndings consistent with those from a previous study [33]. However, in both adult and juvenile IIM, other alleles forming part of the 8.1 haplotype have been reported to confer the primary association over and above that resulting from TNFα-308A [36,37]. It is known that subjects with the 8.1 haplotype have an altered immune response and thus immune stress may produce an imbalanced array of cytokines [38]. In a study of normal subjects, stratifi ed by the presence or absence of the 8.1 haplotype, TNF-α levels in both serum and stimulated mononuclear cells were signifi cantly higher in 8.1 haplotype-positive versus -negative carriers (P <0.0005) [39]. Studies have suggested a genetic contribution to TNF-α regulation, in which TNFα-308A and other TNF-α gene polymorphisms are associated with higher-circulating levels of TNF-α [40]. Th ese fi ndings have also been suggested from IIM studies [33]. B-cell activating factor (BAFF) and a proliferation-inducing ligand (APRIL) are members of the TNF superfamily and have key functions in both B-and T-cell homeostasis, and both ligands are signifi cantly increased in the serum of IIM cases versus controls [41]. Th e TNFα-308A polymorphism therefore may have functional signifi cance in driving an altered immune response. Supporting evidence for this comes from a gene expression study in juvenile DM, the results of which suggested that TNF-α was a key molecule within a pathogenesis model that included antiviral, ischemic, and degeneration/regeneration processes [42]. However, identifi cation of the relevant functional variant at the TNF-α locus is complicated by the presence of extensive linkage disequilibrium within this region.

HLA-DPB1 associations
Genetic variations outside of the traditional 8.1 haplotype region may be involved in determining serotype/phenotype diff erences. Th e HLA-DPB1 gene lies at the centromeric end of MHC and is separated from other HLA class II loci by one or more genetic recombination hotspots, which may weaken the degree of any linkage disequilibrium with other class II loci [43].
In a UK-based cohort of 311 adult-and juvenile-onset PM, DM, or myositis/CTD-overlap cases, a strong association was observed between HLA-DPB1*0101 and the presence of anti-Jo-1 antibodies (Table 1) [44]. Despite the strong association of HLA-DRB1*03 with both anti-Jo-1 and anti-PM-Scl antibodies, no signifi cant diff erence was noted in the frequency of HLA-DPB1*0101 between anti-PM-Scl antibody-positive cases (15%) and controls (13%). Furthermore, a signifi cant diff erence at HLA-DPB1*0101 was noted when anti-Jo-1-and PM-Sclpositive cases were compared with each other, and this allele was much more common in anti-Jo-1 cases (37% Jo-1 versus 15% PM-Scl, uncorrected P = 0.03, OR 3.3, 95% CI 1.01 to 12.7). Th us, additional disease susceptibility signals for anti-Jo-1 may be present in the centromeric end of the MHC region. Th ese additional signals genetically discriminate risk for anti-Jo-1 from that of anti-PM-Scl antibodies. Fine mapping and further detailed HLA analyses in large-scale colla borations are now required within the MHC to carefully investigate these diff erences.

Mannose-binding lectin 2 polymorphisms
Mannose-binding lectin (MBL2) is a protein that may play a role in reducing photosensitive autoimmunity by altering the clearance of apoptotic cells [45]. Several polymorphisms have been described in the MBL gene and can aff ect serum MBL concen trations. One such SNP (GlyAsp, codon 42, exon 1, rs1800450), known to be associated with low serum MBL concentrations, was studied in a small US study com par ing DM cases with cutaneous lupus erythematosus cases and controls [45]. Th is MBL polymorphism was signifi cantly increased in DM (P = 0.0004) but not cutaneous lupus erythematosus versus controls. Combinations of low-producing MBL variants were also associated with DM, but again not in subacute cutaneous lupus erythematosus. It was thus hypothesized that these MBL polymorphisms could infl uence the pathogenesis of DM by infl uencing overproduction of apoptotic keratinocytes and impairing the clearance of such apoptotic cells.

Immunoglobulin gene polymorphisms
Immunoglobulin gene polymorphisms encoding constant regions of immunoglobulin gamma heavy (GM, 14q32.33) and kappa light (KM, 2p12) chains have been examined in a series of US Caucasian and African-American IIM cohorts [10,26,46]. A number of studies have examined individuals with allelic variants of specifi c GM/KM genes, in which raised titres of specifi c IgG antibody subclasses were present, against various antigenic epitopes of infectious disease agents or self-proteins [46]. Th e most recent study of GM/KM allotypes, in 514 US Caucasian and 123 African-American adult/juvenile IIM cases [46], adds to previous studies of Mesoamerican and Korean populations [10,26]. Th e GM 13 allotype showed the strongest association in Caucasian juvenile DM versus controls (P corr <0.0001, OR 3.9, 95% CI 2.26 to 6.76). When analyzed as a paired combination with either the KM 1 or 3 allotype, the GM marker also conferred risk in Caucasian adult IIM (P corr <0.004 for each pair). In adults, the GM 3 23 5,13 phenotype was signifi cantly increased in anti-Jo-1 antibody-positive adult DM cases versus controls (P corr = 0.0031, OR 3.4, 95% CI 1.68 to 7.12). Other GM/KM markers also diff erentiated between the Caucasian and African-American cohorts.
Th ese results indicate mechanistic interactions between two indepen dent loci outside of the HLA system. It would be interesting to examine these markers in geographically diverse populations to confi rm the fi ndings and to further test the apparent ethnic heterogeneity with these loci.

Protein tyrosine phosphatase N22
Previous large-scale genetic association studies have confi rmed a missense SNP in the protein tyrosine phosphatase N22 (PTPN22) gene in association with a variety of autoimmune diseases [47]. At position 1858 of the PTPN22 gene, a CT change (rs2476601) leads to an amino acid substitution (argininetryptophan, R620W) in the lymphocyte phosphatase (LYP) protein, thought to be a 'gain-of-function mutation' . Th e LYP*W620 is able to dephosphorylate signaling proteins more effi ciently than LYP*R620, leading to increased T-and B-lymphocyte inhibition, thymic hypo-responsiveness, and an increase in circulating autoreactive T cells [48]. Th e PTPN22 gene has been investigated in a large, adult and juvenile UK IIM cohort [49]. Th e results showed that the R620W polymorphism was associated with the combined adult/ juvenile IIM group (13.6% IIM versus 8.2% controls, P corr <0.0009, OR 1.8, 95% CI 1.3 to 2.4) and also within the adult PM (16.4% PM, P corr = 0.003, OR 2.2, 95% CI 1.4 to 3.3) and juvenile DM (15.9% juvenile DM, P corr = 0.009, OR 2.1, 95% CI 1.3 to 3.3) subgroups. Importantly, this IIM/PTPN22 association was shown to be independent of the 8.1 haplotype. Other PTPN22 SNPs were examined, but only a haplotype incorporating the minor T allele of R620W conferred any signifi cant degree of suscep tibility for IIM. Unlike the described HLA associations with IIM, no associations were observed within serological subgroups after corrections for multiple comparisons. Th us, the R620W variant appears to be a risk factor for IIM regardless of MSA/MAA status, implying a risk for general autoimmunity rather than conferring risk for a specifi c serological subtype (cf HLA) Th e eff ect size of this SNP in conferring risk for IIM is in keeping with observations for other autoimmune diseases such as rheumatoid arthritis.

Inclusion body myositis
Th e recent larger-scale IIM candidate gene studies already cited have not concentrated on IBM, and, consequently, little progress has been made in the study of IBM immunogenetics. Previous IBM genetic studies have contained only small numbers of patients and have concentrated on the HLA region [3,52]. Consequently, global collaborative eff orts between neurology and rheumatology are now required to readdress this recruitment issue.
A recent study investigated HLA class I and II associations in a cohort of 80 Australian sporadic IBM cases compared with 190 controls [53]. Th e association with HLA-DR3 and IBM was confi rmed (75% IBM versus 23% controls, P <0.01, OR 9.56, 95% CI 5.12 to 18.96). In a case-control subanalysis of DR3-positive carriers, the frequency of HLA-DR3/DR1 heterozygotes was significantly increased in IBM compared with controls (P <0.003). Within this subgroup, the mean age of onset of disease was signifi cantly lower when compared with the remaining patients (55.8 years versus 62.3 years, P = 0.006). Th e average quadriceps femoris muscle strength was lower in HLA-DR3-positive compared with HLA-DR3-negative cases after disease duration and treatment were adjusted for (P = 0.01), confi rming the fi ndings of a previous study [54]. Such a result may suggest that the 8.1 haplotype not only infl uences susceptibility for IBM but also may aff ect clinical disease expression.
Similarities have previously been noted in accumulated proteins from sporadic IBM muscle biopsies and brain tissue plaques from Alzheimer's disease cases, including amyloid-β precursor protein, amyloid-β, and apolipoprotein E (apoE) [55]. Th e APOE gene was previously investigated in sporadic IBM [52]. A gene expression mRNA profi ling study has demonstrated the increased expression of amyloid and apoE in IBM, but signifi cantly elevated levels of the same genes were also found in the biopsies of PM and DM cases (in which the expression of these proteins is not typically found). Th ese fi ndings suggest that the observed accumulation of these proteins in IBM may be due to post-transcriptional events downstream of mRNA expression [56].

Dissecting genetic associations to identify the functional variants
As indicated above, it is becoming increasingly clear from published research that multiple genes and genetic variants, acting in a variety of biological pathways, are involved in susceptibility to autoimmune disorders. Identi fi cation of the precise role of specifi c HLA alleles is com plicated by the presence of extensive linkage disequili brium across common autoimmune risk haplotypes; analyses conditional on known genetic risk variants have been carried out in an attempt to alleviate this problem and to identify independent signals. Further more, the majority of research to date has been published on Caucasian populations, and it is not yet clear how much variability in disease susceptibility there is between diff erent ethnic groups. Further multiethnic comparisons may help to identify the functional genetic variants and to interpret diff erences in clinical presentation and disease severity between diff erent populations.

Conclusions
Th is review illustrates the progress that has been made in IIM immunogenetic research over the past decade. A substantial part of the genetic risk for developing adultand juvenile-onset IIM lies within the MHC, but multiple genetic regions outside of the MHC are increasingly being identifi ed as conferring disease susceptibility risk. Th ere is still the challenge of studying a serologically heterogeneous disorder that is rarer by orders of magnitude than the likes of rheumatoid arthritis. Nevertheless, fi ndings increasingly appear to demonstrate the strong relationship between HLA genotype and IIM serological/clinical phenotype. Further and larger collabora tive studies will help to discern whether an individual's genotype will assist the attending physician in the assessment of disease severity and outcome. Th rough MYOGEN, a fully international Myositis Genetics Consortium, a genome-wide association scan is already being undertaken and analyzed. It is hoped that the results will repeat the successes already demonstrated in other autoimmune diseases and that post-genomic work in IIM will result in the identifi cation of novel disease susceptibility variants. Exciting times beckon in the diffi cult fi eld of IIM immunogenetics! Abbreviations apoE, apolipoprotein E; CI, confi dence interval; DM, dermatomyositis; IBM, inclusion body myositis; IIM, idiopathic infl ammatory myopathy; IL, interleukin; JDM, juvenile dermatomyositis; LYP, lymphocyte phosphatase; MAA, myositis-associated antibody; MBL, mannose-binding lectin; MHC, major histocompatibility complex; MSA, myositis-specifi c antibody; OR, odds ratio; P corr , corrected P value; PM, polymyositis; PTPN22, protein tyrosine phosphatase N22; SNP, single-nucleotide polymorphism; TNF, tumor necrosis factor.