Abstract

Unlocking the genetic contribution to autoimmune thyroid disease (AITD) will hold one of the keys to understanding disease pathogenesis and developing improved treatments. Significant increases in our understanding of the human genome combined with methodological advances in our ability to search for genetic variation have transformed the way in which we screen the genome for susceptibility loci. From early linkage analysis through to candidate gene studies and most recently genome-wide association screening, each methodology has revealed important insights into not just the heritability of AITD but also the best way of identifying disease causing DNA variants. This review will examine each of the different genome screening techniques, highlighting the successes and failures of each methodology and the lessons learnt which have helped inform the next phase of the disease-gene identification process. We will also look to see where we should be focusing our research efforts in the future.

HOW DO YOU SOLVE A PROBLEM LIKE AUTOIMMUNE THYROID DISEASE?

Unlocking the genetic contribution to complex diseases such as autoimmune thyroid disease (AITD), including Graves’ disease (GD) and Hashimoto's thyroiditis (HT), is vital to our understanding of disease pathogenesis and the development of improved therapeutic interventions. Although the desire to identify disease causing DNA variants has not changed in over 50 years, the ways in which we go about screening for susceptibility loci has evolved (Figure 1 timeline of genome-wide screening technologies), with breakthroughs in our understanding of the human genome and improved genotyping technologies helping to inform future screening methodologies. We will focus on the successes and failures of genome-wide screening over the years, what they have taught us about AITD heritability and where to focus future efforts into finding the remainder of disease susceptibility loci.

Figure 1:

Timeline of the different techniques used to screen for the genetic heritability to AITD. Since the 1970s different techniques have been used to screen for the genetic contribution to AITD. This figure shows the different techniques used since the 1970s through to the present time, the size of data sets in which these techniques were applied and how much genetic variation was captured. PCR-RFLP = polymerase chain reaction-restriction fragment length polymorphism.

Figure 1:

Timeline of the different techniques used to screen for the genetic heritability to AITD. Since the 1970s different techniques have been used to screen for the genetic contribution to AITD. This figure shows the different techniques used since the 1970s through to the present time, the size of data sets in which these techniques were applied and how much genetic variation was captured. PCR-RFLP = polymerase chain reaction-restriction fragment length polymorphism.

ONCE UPON A TIME … LINKAGE STUDIES IN AITD

The first approach employed to screen the genome for the genetic contribution to AITD was that of genome-wide linkage analysis. Linkage analysis involved the use of a series of microsatellite markers (short segments of DNA anywhere from 1–6 bp repeated in tandem) spread across the genome looking for co-segregation with disease, in family based data sets. Once a region of linkage was detected a ‘reverse genetics’ approach was employed to work out which gene within the region was linked with AITD, before determining the functional consequences [1]. Linkage analysis had previously proved highly successful in locating rare variants with large effects responsible for monogenic diseases, where the disease is caused by one gene. It was hoped that these successes in monogenic disease could be replicated in complex disease such as AITD where disease was thought to be the result of many genetic factors.

The first genome linkage scan in AITD was performed in 1999 in 56 multiplex families (containing unaffected and GD or HT affected family members). Three areas of linkage to GD, designated GD-1, GD-2 and GD-3 on chromosomes 14q31, 20q11.2 and Xq21, respectively, two areas of linkage to HT, designated HT-1 and HT-2 on chromosome 13q32 and 12q22, respectively, and linkage to AITD in general, AITD-1 on chromosome 6p11, distinct from the previously associated HLA affects within the region, were reported [2]. Increasing the number of families screened to 102 rather than providing support for these regions, produced no support for GD-3 or HT-1, with only weak evidence for GD-2 found after removing a subset of 29 Italian families [3]. Some evidence for association of AITD-1 was reported however, the position of this association was put into doubt as a stronger association was detected ∼30 cm away from the original signal [3]. At the same time, new areas of linkage were also detected including linkage of GD to chromosome 7q and linkage of chromosome 6p, 8q and 10q to AITD in general [3]. Lack of replication between data sets however raised concerns over the reproducibility and indeed the robustness of linkage studies. Independent studies in a small White Caucasian family cohort showed some support for GD-3, albeit with association now at Xp11 [4], and detected linkage to chromosome 18q21 [4, 5]. Results from the largest AITD genome linkage screen conducted in 1119 Caucasian AITD affected-sib-pairs although identifying three new regions of linkage to GD on chromosomes 18p11, 2q36 and 11p15, and suggestive linkage to GD and HT at several other regions, failed to replicate any previously detected linkage results [6]. More interestingly this screen also failed to detect association of the HLA region, CTLA-4 and PTPN22, which have been unequivocally associated with GD by candidate gene studies [7–9], suggesting that linkage may not be suitable for finding susceptibility loci for AITD.

Genome-wide linkage studies in Oriental AITD populations showed a similar story to that seen in Caucasians. A genome-wide screen in 123 affected Japanese sib pairs revealed that chromosome 5q31-33 produced their strongest linkage to AITD in general, with evidence of strong linkage also reported to chromosome 8q23-24 in HT, similar to that detected in a previous Caucasian linkage screen to AITD [3, 10]. Disappointingly, no other previously detected regions showed any linkage to disease in this Japanese cohort [10]. Some support for linkage to chromosome 5q31 was obtained from a genome linkage screen of 54 Chinese Han multiplex families, which also reported linkage to chromosome 1p33 and 1q42 [11]. Beyond replication of chromosome 5q31 within two Oriental populations and potentially some evidence for linkage of chromosome 8q23-24 to AITD in general, no other regions of linkage have been consistently replicated [3, 10, 11].

Even for replicated regions of linkage, such as 5q31-q33, these regions cover many hundreds of possible candidate genes making it difficult, if not impossible, to localize individual susceptibility loci. To narrow down association, positional cloning techniques have been employed whereby additional microsatellite markers and single nucleotide polymorphisms (SNPs) have been used to try to localize the signal to a specific gene/s. Only after nearly 10 years and numerous failed attempts was a GD susceptibility locus, SCGB3A2, located within the 5q31 region [12, 13]. Additional screening of the thyroglobulin gene within the 8q24 region of linkage showed some association with GD, although further replication is still required. Whether either of these associations accounts for the linkage signals within these regions is still open to debate. Even for the established AITD genes, TSHR, located in the GD-1 region of linkage and CD40, located in the GD-2 region, it is not clear whether it is these genes driving the respective linkage signal [9].

Several technical issues were proposed to explain why linkage failed in AITD. When linkage was first undertaken no dense microsatellite maps of the human genome were available to inform linkage study design, affecting genome coverage and mapping resolution [14, 15]. In contrast to monogenic disease, where disease is the result of single rare variants of large effect size it became apparent that in complex diseases, such as AITD, we were looking for a series of genes with much smaller effect size. Many of the first linkage studies were conducted on a small number of families which can be significantly affected by population substructure within the data set. This was clearly evident with GD-2 which only showed association in a cohort of 102 families when 29 Italian families were removed [3]. Although positional cloning techniques were employed to try and narrow down linkage signals to a specific gene/s, many signals covered regions encompassing hundreds of genes making it difficult to further localize the association detected.

Lack of success of linkage studies has been seen for almost all complex diseases, suggesting that although successful for monogenic diseases, linkage analysis has major limitations in complex disease. Although linkage has not delivered novel susceptibility loci for AITD, some conclusions, into the genetic contribution to AITD, could be drawn:

  • No large genetic effects were discovered for AITD, confirming that disease susceptibility was more likely to be due to several genes of smaller effect;

  • There are differences between the genetic contribution to Oriental and Caucasian subjects, suggesting that a different set of genes may explain disease susceptibility in different environments;

  • Greater understanding of the genome and new technological breakthroughs were required to perform comprehensive screening.

CANDIDATE GENE STUDIES TO THE RESCUE

After the failure of linkage to detect replicable susceptibility loci, candidate gene studies, which were used in parallel with linkage, were more heavily relied upon to detect AITD genetic effects. Candidate gene studies involve investigating specific genes, which are selected based on a prior hypothesis about their role in disease. They can be conducted in either family cohorts by comparing transmission of ‘disease’ alleles from parents to affected and un-affected offspring or in case–control cohorts of non-related individuals with disease and controls without disease to look for differences in allele frequency between cohorts.

With approximately 20 000–25 000 genes in the human genome, many of whose function is still unknown, where do you start looking? As disruption to the immune system is key to AITD onset, this provided an obvious starting point, with a role for several immune response genes subsequently identified using this approach, including the HLA class II and class I region, CTLA-4, and PTPN22 [odds ratio (OR) ≥1.50] (Table 1 GD associations detected by candidate gene studies) [7, 8].

Table 1:

Genes shown to be associated with GD using candidate gene studies

Associated gene References 
Consistently replicated using candidate gene studies alone 
    HLA class II Ban et al. [56], Chen et al. [57], Simmonds et al. [58], Wongsurawat et al. [59
    CTLA-4 Furugaki et al. [60], Ueda et al. [61], Vaidya et al. [62], Yanagawa et al. [63
    PTPN22 Heward et al. [64], Smyth et al. [65], Velaga et al. [66
    TSHR Brand et al. [17], Dechairo et al. [67], Hiratani et al. [68
    IL-2RA/CD25 Brand et al. [19
    CD40 Simmonds et al. [23], Houston et al. [69], Jacobson et al. [70, 71], Tomer et al. [72
    SCGB3A2 Song et al. [13], Simmonds et al. [12
Genes detected using candidate gene studies awaiting confirmation 
    FCRL3* Kochi et al. [20], Simmonds et al. [21], Owen et al. [73
    HLA Class I* Farid et al. [74], Grumet et al. [75], Huang et al. [76], Mather et al. [77] Simmonds et al. [78
    IFIH1* Sutherland et al. [79
    ADRB2 Chu et al. [80
    DIO2 Chistiakov et al. [81
    FCGRIIa Yesmin et al. [82
    IL-23 Huber et al. [83
    IL-3 Chu et al. [84
    IRF1 Yang et al. [85
    LMP7 Heward et al. [86], Vives-Pi et al. [87
    LMP2 Heward et al. [86], Vives-Pi et al. [87
    PDCD1 Newby et al. [88
    TNFα Nakkuntod et al. [89], Simmonds et al. [90
    TCR-β Demaine et al. [91], Pickerill et al. [92], Zhang et al. [93
    TPO Kotani et al. [94, 95], Ludgate et al. [96], Pirro et al. [97
    UGRP1 Yang et al. [85
Inconsistent findings  
    FOXP3 Ban et al. [98], Owen et al. [et al. [99
    IL-13 Bednarczuk et al. [100], Hiromatsu et al. [101], Simmonds et al. [102
    IL-4 Heward et al. [103], Yang et al. [85
    IL-1RN Blakemore et al. [104], Muhlberg et al. [105
    Tg Collins et al. [106, 107], Tomer et al. [108, 109
    TNFβ Nakkuntod et al. [89
    Vitamin D receptor Ban et al. [110, 111], Collins et al. [112], Ramos-Lopez et al. [113], Stefanic et al. [114
Associated gene References 
Consistently replicated using candidate gene studies alone 
    HLA class II Ban et al. [56], Chen et al. [57], Simmonds et al. [58], Wongsurawat et al. [59
    CTLA-4 Furugaki et al. [60], Ueda et al. [61], Vaidya et al. [62], Yanagawa et al. [63
    PTPN22 Heward et al. [64], Smyth et al. [65], Velaga et al. [66
    TSHR Brand et al. [17], Dechairo et al. [67], Hiratani et al. [68
    IL-2RA/CD25 Brand et al. [19
    CD40 Simmonds et al. [23], Houston et al. [69], Jacobson et al. [70, 71], Tomer et al. [72
    SCGB3A2 Song et al. [13], Simmonds et al. [12
Genes detected using candidate gene studies awaiting confirmation 
    FCRL3* Kochi et al. [20], Simmonds et al. [21], Owen et al. [73
    HLA Class I* Farid et al. [74], Grumet et al. [75], Huang et al. [76], Mather et al. [77] Simmonds et al. [78
    IFIH1* Sutherland et al. [79
    ADRB2 Chu et al. [80
    DIO2 Chistiakov et al. [81
    FCGRIIa Yesmin et al. [82
    IL-23 Huber et al. [83
    IL-3 Chu et al. [84
    IRF1 Yang et al. [85
    LMP7 Heward et al. [86], Vives-Pi et al. [87
    LMP2 Heward et al. [86], Vives-Pi et al. [87
    PDCD1 Newby et al. [88
    TNFα Nakkuntod et al. [89], Simmonds et al. [90
    TCR-β Demaine et al. [91], Pickerill et al. [92], Zhang et al. [93
    TPO Kotani et al. [94, 95], Ludgate et al. [96], Pirro et al. [97
    UGRP1 Yang et al. [85
Inconsistent findings  
    FOXP3 Ban et al. [98], Owen et al. [et al. [99
    IL-13 Bednarczuk et al. [100], Hiromatsu et al. [101], Simmonds et al. [102
    IL-4 Heward et al. [103], Yang et al. [85
    IL-1RN Blakemore et al. [104], Muhlberg et al. [105
    Tg Collins et al. [106, 107], Tomer et al. [108, 109
    TNFβ Nakkuntod et al. [89
    Vitamin D receptor Ban et al. [110, 111], Collins et al. [112], Ramos-Lopez et al. [113], Stefanic et al. [114

This table shows the susceptibility loci detected for GD via candidate gene studies split into consistent findings, findings awaiting confirmation and inconsistent finding. Asterisk indicates the findings which were later confirmed by GWAS. Adapted from Zeitlin, et al. [18].

Candidate gene studies were, however, not without their limitations. Most researchers employed the use of polymerase chain reaction-restriction fragment length polymorphism (RFLP-PCR) technology to type for specific DNA variants, often only looking at SNPs in coding regions and 3′- and 5′-ends of genes. These incomplete approaches lead to some genes being excluded as being associated with GD in early candidate gene studies. A good example of this is the TSHR, which was later shown to be strongly associated with GD, once a more comprehensive strategy was embarked upon [16, 17].

An appreciation for the need to screen all common variation within a gene combined with increases in our understanding of the linkage disequilibrium (LD) between SNPs enabled us to use specific tag SNPs to capture a series of SNPs in strong LD without having to individually screen each variant [18]. The replacement of PCR-RFLP with fluorescent based genotyping platforms also increased the speed at which a specific gene region could be screened. This strategy allowed us to generate evidence of association of CD25, FCRL3 and CD40 with GD (OR ≤ 1.30) [19–23]. As time progressed it became obvious that the size of effect on the development of disease, or OR, for susceptibility loci now being detected was becoming smaller and smaller with gene effects of OR ≥ 1.50 giving way to effects of OR < 1.30. This necessitated the collection of larger data sets to provide sufficient power to detect, and exclude, effects of this magnitude (Figure 1).

Although the usefulness of candidate gene studies to follow-up on associated gene regions was clear, it was questionable whether as an efficient method of genome-wide screening it could ever deliver. This was because for every successful gene detected by this approach, tens or hundreds of other genes investigated failed to show replicable association with disease [18]. A novel genome-wide screening method was, therefore, required and it was this idea that lead to the development of genome-wide association studies (GWAS) in the mid-2000s that have revolutionized the field of complex disease.

THE NEW KID ON THE BLOCK—GWAS IN AITD

Before GWAS could be contemplated, the limitations encountered with both linkage and candidate gene studies, including, genome coverage, data set size and high throughput genome-wide screening technology needed to be addressed.

Due to the results obtained from linkage and candidate gene studies, it was believed that complex disease heritability was due to numerous common variants [minor allele frequency (MAF) >5%], known as the ‘common variant, common disease’ hypothesis [14, 24]. The Hapmap project was set up with the aim of cataloguing all common variation in the human genome, including non-coding regions, in four different ethnic populations; Caucasians, Africans, Chinese and Japanese, to test this hypothesis [25, 26]. Three million common SNPs were identified, on which LD analysis was performed, to determine how many SNPs would need to be directly screened to capture all common variation in the human genome. It was calculated that >80% of the three million common variants could be captured by only 500 000 SNPs [27, 28]. Advances in screening methodologies also meant that specially designed SNP genotyping chips could be used to perform dense genotyping on 500 000 or more variants in a fast and relatively cost efficient manner [1]. Candidate gene studies had demonstrated that the size of gene effects likely to be detected for complex disease were low (OR < 1.30) highlighting the requirement for large samples sizes (>2000 cases and controls) and replication data sets of equal or greater size to confirm such effects. As it was assumed that any variant of interest detected would be consistent across the whole population investigated, careful matching for geographical variation and ethnicity would also be required to reduce the chances of false positive results [14]. As GWAS of 500 000 SNPs has the potential to generate a large number of false positives, power calculations suggested that only variants producing P ≤ 10−7 should be considered significant at a genome-wide level [29, 30].

Numerous ≥500 000 SNP GWAS have been completed for several autoimmune disease. To date, however, no such screens have been completed in AITD. Smaller screens using many of the concepts of larger GWAS studies have been completed for GD (Figure 2). As part of the Wellcome Trust Case Control Consortium (WTCCC) our group investigated a genome-wide set of 14 436 non-synonymous (nsSNPs) comprising all experimentally validated nsSNPs with a MAF >1% in western European samples in a data set of 1000 UK Caucasian GD cases, as well as 1000 breast cancer, ankylosing spondylitis (AS) and multiple sclerosis (MS) cases and 1500 controls [31]. Due to the reduced number of tests being performed a P < 10−5 was considered to be significant at a genome-wide level [32]. Unsurprisingly the largest association detected with GD was the HLA region P < 10−20, with the maximal signal concentrated around previously detected GD HLA class I and class II associations [8, 31]. Due to such high P-values within this region even very distant SNPs in modest LD showed indirect evidence of association, making it difficult to determine additional HLA loci. Analysis of data from this region is ongoing to disentangle any additional associations.

Figure 2:

Genes associated with GD detected either through candidate gene studies and/or GWAS. This figure shows a diagrammatical representation of the chromosomes within the human genome and on them highlighted the name and location of GD susceptibility loci confirmed by candidate gene studies alone, candidate gene studies and GWAS and loci found by GWAS in other autoimmune diseases, including MS and T1D and confirmed in GD by candidate gene studies.

Figure 2:

Genes associated with GD detected either through candidate gene studies and/or GWAS. This figure shows a diagrammatical representation of the chromosomes within the human genome and on them highlighted the name and location of GD susceptibility loci confirmed by candidate gene studies alone, candidate gene studies and GWAS and loci found by GWAS in other autoimmune diseases, including MS and T1D and confirmed in GD by candidate gene studies.

None of the other nsSNPs reached a P < 10−5 but several variants showed association at P < 10−4, which just fell short of this cut off value. Interestingly association of a nsSNP in FCRL3 (P = 2.1 × 10−4), which has previously been proposed to be associated with GD in Oriental and Caucasian populations, was detected [20, 21, 31]. Association of this nsSNP was believed to be the result of LD with the previously screened FCRL3 associations [20, 21]. To determine if larger GD associations existed within FCRL3, tag SNP screening was performed using 7 tag SNPs in an extended Caucasian data set, which revealed a stronger association of another SNP within this region, rs11264798 (P = 1.6 × 10−5, OR = 1.22). Additional nsSNP associations were also found within neighboring FCRL5 (P = 1.3 × 10−4–5.0 × 10−5), with tag SNP analysis of FCRL5 and logistic regression analysis with previously typed FCRL3 tag SNPs, revealing that any associations at either gene was due to LD with FCRL3 rs11264798 and rs10489678 SNPs [33]. Ten other novel GD associations were detected at P < 10−4 but as these regions showed no prior evidence of association the results were viewed with caution [31]. A replication study of these loci in an independent collection of 1578 UK GD patients and 1946 matched controls showed no evidence of association with GD, although when combined with the original WTCCC study group, minor associations (P ≥ 10−3) remained for four of the nsSNPs, present within HDLBP, TEKT1, JSRP1 and UTX. Tag SNP screening of these regions in a UK Caucasian GD collection of 2478 samples and 2690 controls, only found minor additional associations within HDLBP and TEKT1 (P = 0.042–0.002) [34]. With the exception of minor associations at HDLBP and TEKT1, the remainder of the novel loci detected failed to be replicated, suggesting that these nsSNPs do not have a significant effect on the risk to GD in the UK [34]. A second analysis of the nsSNP GWAS data undertaken comparing the 1000 GD against all non-AITD diseases cases and controls (4500 samples), detected genome-wide association of a nsSNP within TSHR (P = 2.1 × 10−5) [31]. This provided further evidence for association of this region with GD, which has now been refined to two variants, rs179247 and rs12101255 (P ≤ 10−7, OR = 1.53−1.55) present within a 40 kb region of TSHR intron 1 [17].

The main limitation of the nsSNP GD screen was that of genome coverage. Two of the most well known GD associations, CTLA-4 and PTPN22, were not detected due to lack of representation on the genotyping chip [31]. Similarly for the other established GD susceptibility loci, CD40 and CD25 had only limited SNP coverage on this chip and SCGB3A2 had no coverage at all, explaining why these associated variants were not detected. This may also explain why other biologically plausible candidates were not picked up in this screen. This suggests that much of the remainder of the genetic contribution to GD may lie outside coding regions and may instead be due to common variation in non-coding regions as has been seen for many of the susceptibility loci detected to date. This serves to highlight the need for further larger GWAS in AITD encompassing common variation in non-coding regions. Whilst data from such screens is awaited, AITD has been shown to co-cluster with other autoimmune diseases, including T1D, MS and rheumatoid arthritis (RA), suggesting the sharing of susceptibility loci between autoimmune diseases [35]. It has been well documented that although each autoimmune disease has a set of disease specific susceptibility loci, such as the TSHR in GD, they also share a series of common autoimmunity susceptibility loci, such as the HLA region, CTLA-4 and PTPN22 [7, 36]. With data now generated from several GWAS studies, including MS, AS, T1D, RA and the aforementioned nsSNP GWAS in GD, it is now possible to determine the similarities and differences in the genetic architecture between these diseases [37]. After analyzing a set of 15 SNPs associated with all five autoimmune diseases, the diseases were shown to cluster within two disease groups with MS and AITD having similar but distinct genetic profiles from RA and AS [37]. T1D also showed strong similarity to AITD [37]. This suggests that data obtained for other GWAS studies, in particular T1D and MS, may also provide potential insights into the aetiology of GD and could help explain why different autoimmune diseases co-occur in the same individual.

INSIGHTS FROM GWAS IN OTHER AUTOIMMUNE DISEASES

Due to the long established link between T1D and GD many of the T1D GWAS loci detected have also been studied in GD. Two ≥500 000 SNP screens have been performed for T1D [38, 39]. The first was performed on 550 000 SNPs in a cohort of 563 European T1Ds and 1146 controls and 483 family trios [39] and the second, as part of the WTCCC, investigated seven different complex disease cohorts of 2000 cases each and 3000 shared controls [38]. Interestingly both screens identified association of previously detected T1D loci, with the 550 000 SNP screen identifying association of several SNPs within the HLA region, PTPN22 and INS (P < 10−8) and the WTCCC detecting association of the HLA region and PTPN22 (P < 10−26) as well as providing some evidence of association of CTLA4 (P = 10−5), CD25 (P = 10−5) and IFIH1 (P = 10−3) [38, 39]. Both scans also detected novel susceptibility loci. The 550 000 SNP scan found strong replicable association of three variants within KIAA0350 on chromosome 16p13 (P = 6.7 × 10−11) [39]. The WTCCC also reported genome-wide significance for seven novel T1D signals (P < 5 × 10−7) throughout the genome including further independent replication of KIAA0350 and associations at chromosomal locations 12q13, 12q24, 4q27 12p13, 18p11 and 10p15 [38]. Replication of the T1D WTCCC GWAS and a previous 13 378 nsSNP GWAS performed on 3400 cases and 3300 controls, in an additional 4000 cases and 5000 controls and a family data set of 2997 parent child trios was performed [40]. Eleven SNPs showing association with T1D (P ≤ 5 × 10−7) within the 500 000 WTCCC screen were also screened, with four of these regions showing convincing replication (P < 1.15 × 10−14) including KIAA0350 on chromosome 16p13 and C12orf30 on 12q24, ERBB3 on 12q13, and PTPN2 on 18p11 (P = 1.14 × 10−14–1.52 × 10−20) [40]. Fourteen nsSNPs from the T1D nsSNP screen that showed a trend towards association (P > 1 × 10−3) were also followed up and in addition to confirming previously reported associations at known T1D susceptibility loci IFIH1 and PTPN22, associations were also detected for CD226 on chromosome 18q22 (P = 1.38 × 10−8) and some evidence for nsSNPs within CAPSL, C20orf168, IL7R and MED12L, although they did not reach genome-wide significance (P < 8.25 × 10−4) [40].

Thirteen of the associated replicated T1D loci were also screened in 2200 UK Caucasian GD and 3600 controls with some evidence of association of IFIH1, PTPN2, CAPSL, CD226 and loci at chromosome 4q27 and 2q11 (Figure 2). Additional meta-analysis of the T1D GWAS data are helping to identify numerous additional susceptibility loci that will in the future also be screened in GD. Like T1D, several MS GWAS have been performed and a series of susceptibility loci identified. Unlike T1D there has been less of a drive to replicate these effects in AITD, but some MS loci have previously been detected through GD candidate gene studies (Figure 2). Determining which susceptibility loci cross over between each autoimmune disease will provide important insights into disease genetic susceptibility. The WTCCC Immunochip project is currently underway, using a 200 000 SNP custom array to provide dense SNP mapping within all significant GWAS loci from a series of autoimmune disease, including GD, T1D, AS, MS, RA and systemic lupus erythematosus (SLE), and will provide previously unimaginable insights into the genetics of not just each disease itself but to the autoimmune disease process in general.

THE MORAL OF THE GWAS STORY IN AITD

Although complete GWAS screening has not been performed in AITD, insights into AITD genetic susceptibility have been achieved through the nsSNP study and studies in other related autoimmune diseases. At this stage, therefore, we can conclude that:

  • GWAS has not identified any large effects (OR > 1.40) not previously detected by candidate gene studies;

  • Newly detected associations are modest (OR < 1.30) and are likely to only contribute to a fraction of the genetic heritability;

  • Many GWAS associations, although better than linkage studies, do not survive replication, demonstrating the need to screen large replication cohorts;

  • Susceptibility loci detected for AITD are not all in coding gene regions, suggesting that focusing on coding SNPs may miss true disease causing DNA variants.

Even though GWAS signals have been defined to specific genes we cannot be certain that the gene named actually contains the aetiological variant/s attributable to these regions of association. As such determining the aetiological variants represented by these associations remains an important challenge.

THE SEQUEL: SEARCHING FOR THE ‘MISSING’ HERITABILITY

In T1D, GWAS has increased the number of confirmed susceptibility loci from between 5 and 6 to over 40 [41]. However, even with the large number of new susceptibility loci identified it is believed that only a proportion of T1D genetic heritability is accounted for. T1D is not unique in this regard, with similar findings being reported in most complex diseases [42]. It has previously been suggested, by some, that genetic heritability in complex disease had been overestimated and that the ‘missing’ heritability does not exist. Although new methods based on GWAS results are being developed to better estimate heritability, it remains the belief that current heritability measures are more or less accurate [42], which leaves us with the question of where is the ‘missing’ heritability? Whilst GWAS have screened the majority of common variation (MAF >5%) within the human genome, other types of variants may hold the key to finding the missing heritability, with other approaches required to identify these effects including:

Trans-ethnic mapping

GWAS in data sets of different ethnicity may detect regions of association that have only appeared weakly associated in a specific population but by exploiting the differences in SNP allele frequencies and LD patterns between different ethnic groups, may be easier to detect in another ethnic group [42, 43].

Copy number variants

Copy number variants (CNV) are structural variations within the genome including deletions, insertions, inversions and translocations [1]. Depending on their location CNVs may influence gene number and if located in a significant part of the gene or its regulatory region may also alter the gene product quantity. CNVs were not routinely captured during the first phase of GWAS. Improved cataloguing and CNV screening methodology have enabled more extensive screening of CNVs to be carried out [44]. Recently 3432 non-duplicated, multi class CNVs were screened in 2000 cases from eight common complex diseases, including T1D and RA and 3000 shared controls [45]. Three replicable CNV loci were identified including a CNV within the HLA region associated with T1D (P = 8 × 10−153, OR = 0.20) and RA (P = 1.4 × 10−39, OR = 1.77). However, no evidence could be found for association of this CNV independent of previously detected strongly associated common variants within this region [45]. Although not all known CNVs, in particular rare CNVs, could be captured in this study, the results suggest that common CNVs are unlikely to play a major role in disease either through a specific CNV having a large effect (OR > 1.30) or through numerous CNVs having a small effect (OR < 1.20) and are unlikely to explain the missing heritability [45].

Rare and low minor allele frequency variants

Low minor allele frequency variants (MAF = 0.5–5.0%) or rare variants (MAF < 0.5%) are common in the human genome and have been shown to influence complex diseases including autism, epilepsy and schizophrenia [42, 46]. It has been proposed that rare variants with large effects, similar to those previously found to explain many monogenic diseases and which were originally discounted for complex disease, may also play a role in complex disease [42, 46, 47]. A lack of information on low minor allele/rare variants within the genome and high throughput technologies to screen such variants has hampered previous efforts to look at the role of these variants. Insights from the 1000 genomes project, which was established with the aim of sequencing 1500 individuals of different ethnicity, is extending our knowledge of human variation within the genome from all known common variation down to all low and rare variants in the genome with an MAF = 1–5% (http://www.1000genomes.org). So far the pilot study has sequenced 172 individuals at low depth and identified 11 million new low and rare variant SNPs [14]. Combined with the development of high throughput next generation sequencing tools able to screen 55 billion bases in 10 days, the search for rare variants in complex disease is becoming a real possibility [48]. Interestingly some evidence for a role of rare variants has already been found in the T1D susceptibility loci, IFHI1, where four additional rare variants were found independent of the common GWAS signal originally detected [49]. This supports the idea that rare variants are likely to occur in regions already associated with disease and are proposed to either enhance/modulate previous detected associations or through LD could be accounting for the previous associations within the region [42, 46, 47]. It has also been suggested that linkage results in complex disease, which showed great success at finding rare variants in monogenic disease, may in fact be able to inform the search for rare variants in complex disease [15]. While still very much in its infancy it will be interesting to see the impact of rare variants upon complex disease susceptibility.

COULD GENE INTERACTION EXPLAIN THE MISSING HERITABILITY?

Screening further variation in the genome could produce important insights into locating the missing heritability for complex disease. However, it has been suggested that how susceptibility loci interact with each other, the environment, regulatory elements within the genome and parent of origin/trans-generational effects may also be accounting for some of the missing heritability.

Interaction with regulatory elements within the genome

Variation in how genes interact within the genome and in particular how they are regulated will undoubtedly add another level of complexity to the genetic architecture of disease. It has been suggested that multiple common variants could affect protein–protein or protein–DNA interactions, potentially altering gene regulation and protein production [47]. These effects are not as simple as one effect causing a change in expression and will be more difficult to access. It is envisaged that advances in our ability to screen for non-coding DNA with regulatory potential including sequence conservation, histone modifications, DNAse I hypersensitivity and cis-regulatory mutations, can, in the future, be combined with GWAS data to uncover if interaction of multiple variants will help explain disease [47, 50]. It is also believed that this approach could inform fine mapping of already identified genetic associations. Combining fine mapping of a specific association with information on non-coding DNA with regulatory potential within the region may help refine association and identify the etiological variants present.

Gene–gene and gene–environment interaction

Gene–gene interaction or epistasis has been proposed to occur between genes that cluster within specific immune pathways enhancing their effect on disease susceptibility [47]. Some examples already exist in GD including a possible interaction between HLA class II molecules and thyroglobulin [51]. This suggests that once additional GWAS have been performed in AITD, gene–gene interaction studies will be vital to determine the true association of different susceptibility loci with disease. Similarly, interaction of specific environmental factors with specific genes has also been proposed to modulate or enhance their effect on disease but has been difficult to take into account in GWAS. This is due, in part, to relatively little being known about the environmental factors linked to complex disease and the difficulties associated with relying on patients retrospective re-call. Future advances will hopefully make it easier to quantify these effects [47].

Different genetic effects for different clinical phenotypes

The use of large data sets comprising more than 2000 cases and 2000 controls and in many cases data sets of more than 10 000 cases and matched controls to perform GWAS and replication studies have afforded the power to detect a plethora of small effect common variants (OR = 1.10–1.20) in complex disease [38]. As complex diseases tend to consist of heterogeneous combinations of different sub-phenotypes, which can vary from patient to patient, one potential downside of using such large data sets is that they could be diluting down genetic effects which only predispose to a specific sub-phenotype [42]. Screening large data sets of ‘super’ cases, consisting of a data set enriched for a specific clinical sub-phenotype, will hopefully reveal whether any of the missing heritability is due to sub-phenotype effects.

Trans-generational and parent of origin effects

Studies on mice have suggested a role for trans-generational genetic effects in which phenotypic variation in one generation results from genetic variation in inherited epigenetic markers, such as small RNA, RNA binding proteins, micro RNA and DNA methylation, that are passed down through several generations [52]. These effects can alter genotype–phenotype correlations in offspring, suggesting that the genotype of individuals in previous generations is a better predictor of a disease phenotype than those in the present generation [47, 52]. It is unknown if trans-generational genetic memory occurs in humans. However, a recently identified variant in KCNQ1 on chromosome 11p15 has been shown to either confer or reduce risk to type 2 diabetes depending on whether the allele is inherited paternally or maternally, respectively [53]. These types of variants are difficult to detect, with large family studies required to make this possible. It has also been suggested that variation in the microbiota within the human gut, which is believed to interact with the immune system to help prevent infection, could also play a role in autoimmune disease onset [54, 55]. The microbiota is laid down during the first few years of life and can be influenced by the mother's microbiota, passed on to the offspring through breast feeding [54, 55]. Such studies could be of particular interest in AITD where there is a strong female preponderance and where family studies have shown that AITD patients are more likely to have a female relative with AITD than a male relative [35]. Such studies may not only explain some of the missing heritability but also the strong female preponderance seen in AITD.

CONCLUSION

The different approaches employed to screen the genome for susceptibility loci have revealed key insights into the role played by genetic heritability in AITD. Greater understanding of the type and size of associations likely to be detected combined with increases in our understanding of and ability to screen the human genome has enabled the development of new and improved genome-wide screening methodologies. Although the most recent genome screening approach of GWAS is still in its infancy in AITD, some key insights have already been achieved in our understanding of the genetic susceptibility to AITD. Combined with replication of GWAS performed in other autoimmune diseases, several novel susceptibility loci have been identified for AITD. The challenge that now faces us is the need to refine association of the susceptibility loci already identified, including performing fine mapping of GWAS signals to determine the aetiological variants driving these effects, as well as undertaking additional screens to detect other possible associated variants. Lessons from other complex diseases inform us that the search for genetic heritability is far from over and that many potential avenues still need to be explored to help us fully understand the genetic contribution to AITD pathogenesis.

Funding

This work was supported by the Wellcome Trust grant 084702.

Key Points

  • Linkage failed to detect susceptibility loci for complex disease such as AITD.

  • Candidate gene studies successfully detected association of several AITD susceptibility loci, revealing associations with OR > 1.50 for the HLA region, CTLA-4, PTPN22 and TSHR, later giving way to effects of OR < 1.30 for CD40, FCRL3 and CD25. This suggests that much of the remainder of the genetic contribution to AITD is likely to be due to small effects of this order of magnitude.

  • Knowledge from both linkage and candidate gene studies provided insights into disease susceptibility and has helped inform the design and implementation of genome wide screening.

  • GWAS have revealed additional susceptibility loci through replication of gene effects found in other autoimmune diseases, such as T1D.

  • Further screening for different types of DNA variants and methods to account for gene-gene/environmental interaction, genomic interaction and/or trans-generational/parent of origin effects will also need to be undertaken to determine the complete genetic contribution to AITD.

References

1
Nakamura
Y
DNA variations in human and medical genetics: 25 years of my experience
J Hum Genet
 , 
2009
, vol. 
54
 (pg. 
1
-
8
)
2
Tomer
Y
Barbesino
G
Greenberg
DA
, et al.  . 
Mapping the major susceptibility loci for familial Graves' and Hashimoto's diseases: evidence for genetic heterogeneity and gene interactions
J Clin Endocrinol Metab
 , 
1999
, vol. 
84
 (pg. 
4656
-
64
)
3
Tomer
Y
Ban
Y
Concepcion
E
, et al.  . 
Common and unique susceptibility loci in Graves and Hashimoto diseases: results of whole-genome screening in a data set of 102 multiplex families
Am J Hum Genet
 , 
2003
, vol. 
73
 (pg. 
736
-
47
)
4
Imrie
H
Vaidya
B
Perros
P
, et al.  . 
Evidence for a Graves' disease susceptibility locus at chromosome Xp11 in a United Kingdom population
J Clin Endocrinol Metab
 , 
2001
, vol. 
86
 (pg. 
626
-
30
)
5
Vaidya
B
Imrie
H
Perros
P
, et al.  . 
Evidence for a new Graves disease susceptibility locus at chromosome 18q21
Am J Hum Genet
 , 
2000
, vol. 
66
 (pg. 
1710
-
4
)
6
Taylor
JC
Gough
SC
Hunt
PJ
, et al.  . 
A genome-wide screen in 1119 relative pairs with autoimmune thyroid disease
J Clin Endocrinol Metab
 , 
2006
, vol. 
91
 (pg. 
646
-
53
)
7
Simmonds
MJ
Gough
SCL
Endocrine Autoimmunity
Oxford textbook of Endocrinology and Diabetes
 , 
2011
Oxford, UK
Oxford University Press
8
Gough
SCL
Simmonds
MJ
The HLA region and autoimmune disease: Associations and mechanisms of action
Current Genomics
 , 
2007
, vol. 
8
 (pg. 
453
-
65
)
9
Zeitlin
AA
Simmonds
MJ
Gough
SC
Genetic developments in autoimmune thyroid disease: an evolutionary process
Clin Endocrinol
 , 
2008
, vol. 
68
 (pg. 
671
-
82
)
10
Sakai
K
Shirasawa
S
Ishikawa
N
, et al.  . 
Identification of susceptibility loci for autoimmune thyroid disease to 5q31-q33 and Hashimoto's thyroiditis to 8q23-q24 by multipoint affected sib-pair linkage analysis in Japanese
Hum Mol Genet
 , 
2001
, vol. 
10
 (pg. 
1379
-
86
)
11
Jin
Y
Teng
W
Ben
S
, et al.  . 
Genome-wide scan of Graves' disease: evidence for linkage on chromosome 5q31 in Chinese Han pedigrees
J Clin Endocrinol Metab
 , 
2003
, vol. 
88
 (pg. 
1798
-
1803
)
12
Simmonds
MJ
Yesmin
K
Newby
PR
, et al.  . 
Confirmation of association of chromosome 5q31-33 with United Kingdom Caucasian Graves' disease
Thyroid
 , 
2010
, vol. 
20
 (pg. 
413
-
7
)
13
Song
HD
Liang
J
Shi
JY
, et al.  . 
Functional SNPs in the SCGB3A2 promoter are associated with susceptibility to Graves' disease
Hum Mol Genet
 , 
2009
, vol. 
18
 (pg. 
1156
-
70
)
14
Manolio
TA
Collins
FS
The HapMap and genome-wide association studies in diagnosis and therapy
Annu Rev Med
 , 
2009
, vol. 
60
 (pg. 
443
-
56
)
15
Almasy
L
Blangero
J
Human QTL linkage mapping
Genetica
 , 
2009
, vol. 
136
 (pg. 
333
-
40
)
16
Brand
OJ
Gough
SC
Genetics of thyroid autoimmunity and the role of the TSHR
Mol Cell Endocrinol
 , 
2010
, vol. 
322
 (pg. 
135
-
43
)
17
Brand
OJ
Barrett
J
Simmonds
MJ
, et al.  . 
Association of the thyroid stimulating hormone receptor gene (TSHR) with Graves’ disease (GD)
Hum Mol Genet
 , 
2009
, vol. 
18
 (pg. 
1704
-
13
)
18
Zeitlin
AA
Heward
JM
Newby
PR
, et al.  . 
Analysis of HLA class II genes in Hashimoto's thyroiditis reveals differences compared to Graves' disease
Genes Immun
 , 
2008
, vol. 
9
 (pg. 
358
-
63
)
19
Brand
OJ
Lowe
CE
Heward
JM
, et al.  . 
Association of the interleukin-2 receptor alpha (IL-2Ralpha)/CD25 gene region with Graves' disease using a multilocus test and tag SNPs
Clin Endocrinol
 , 
2007
, vol. 
66
 (pg. 
508
-
12
)
20
Kochi
Y
Yamada
R
Suzuki
A
, et al.  . 
A functional variant in FCRL3, encoding Fc receptor-like 3, is associated with rheumatoid arthritis and several autoimmunities
Nat Genet
 , 
2005
, vol. 
37
 (pg. 
478
-
85
)
21
Simmonds
MJ
Heward
JM
Carr-Smith
J
, et al.  . 
Contribution of single nucleotide polymorphisms within FCRL3 and MAP3K7IP2 to the pathogenesis of Graves' disease
J Clin Endocrinol Metab
 , 
2006
, vol. 
91
 (pg. 
1056
-
61
)
22
Tomer
Y
Davies
TF
Greenberg
DA
What is the contribution of a Kozak SNP in the CD40 gene to Graves' disease?
Clin Endocrinol
 , 
2005
, vol. 
62
 pg. 
258
 
23
Simmonds
MJ
Heward
JM
Franklyn
JA
, et al.  . 
The CD40 Kozak SNP: a new susceptibility loci for Graves' disease?
Clin Endocrinol
 , 
2005
, vol. 
63
 (pg. 
232
-
3
)
24
Collins
FS
Guyer
MS
Charkravarti
A
Variations on a theme: cataloging human DNA sequence variation
Science
 , 
1997
, vol. 
278
 (pg. 
1580
-
1
)
25
Consortium
IH
A haplotype map of the human genome
Nature
 , 
2005
, vol. 
437
 (pg. 
1299
-
320
)
26
Consortium
IH
The International HapMap Project
Nature
 , 
2003
, vol. 
426
 (pg. 
789
-
96
)
27
Barrett
JC
Cardon
LR
Evaluating coverage of genome-wide association studies
Nat Genet
 , 
2006
, vol. 
38
 (pg. 
659
-
62
)
28
Pe'er
I
de Bakker
PI
Maller
J
, et al.  . 
Evaluating and improving power in whole-genome association studies using fixed marker sets
Nat Genet
 , 
2006
, vol. 
38
 (pg. 
663
-
67
)
29
Pearson
TA
Manolio
TA
How to interpret a genome-wide association study
Jama
 , 
2008
, vol. 
299
 (pg. 
1335
-
44
)
30
Hunter
DJ
Kraft
P
Drinking from the fire hose–statistical issues in genomewide association studies
N Engl J Med
 , 
2007
, vol. 
357
 (pg. 
436
-
9
)
31
WTCCC, TACS
Association study of 14,500 nsSNPs in four common diseases identifies variants involved in autoimmunity
Nat Genet
 , 
2007
, vol. 
39
 (pg. 
1329
-
37
)
32
Thomas
DC
Clayton
DG
Betting odds and genetic associations
J Natl Cancer Inst
 , 
2004
, vol. 
96
 (pg. 
421
-
3
)
33
Simmonds
MJ
Brand
OJ
Barrett
JC
, et al.  . 
Association of Fc receptor-like 5 (FCRL5) with Graves' disease is secondary to the effect of FCRL3
Clin Endocrinol
 , 
2010
, vol. 
73
 (pg. 
654
-
60
)
34
Newby
PR
Pickles
OJ
Mazumdar
S
, et al.  . 
Follow-up of potential novel Graves' disease susceptibility loci, identified in the UK WTCCC genome-wide nonsynonymous SNP study
Eur J Hum Genet
 , 
2010
, vol. 
18
 (pg. 
1021
-
6
)
35
Manji
N
Carr-Smith
JD
Boelaert
K
, et al.  . 
Influences of age, gender, smoking, and family history on autoimmune thyroid disease phenotype
J Clin Endocrinol Metab
 , 
2006
, vol. 
91
 (pg. 
4873
-
80
)
36
Simmonds
MJ
Gough
SC
Genetic insights into disease mechanisms of autoimmunity
Br Med Bull
 , 
2004
, vol. 
71
 (pg. 
93
-
113
)
37
Sirota
M
Schaub
MA
Batzoglou
S
, et al.  . 
Autoimmune disease classification by inverse association with SNP alleles
Plos Genetics
 , 
2009
, vol. 
5
 pg. 
e1000792
 
38
WTCCC
Genome-wide association study of 14,000 cases of seven common diseases and 3,000 shared controls
Nature
 , 
2007
, vol. 
447
 (pg. 
661
-
78
)
39
Hakonarson
H
Grant
SF
Bradfield
JP
, et al.  . 
A genome-wide association study identifies KIAA0350 as a type 1 diabetes gene
Nature
 , 
2007
, vol. 
448
 (pg. 
591
-
4
)
40
Todd
JA
Walker
NM
Cooper
JD
, et al.  . 
Robust associations of four new chromosome regions from genome-wide analyses of type 1 diabetes
Nat Genet
 , 
2007
, vol. 
39
 (pg. 
857
-
64
)
41
Barrett
JC
Clayton
DG
Concannon
P
, et al.  . 
Genome-wide association study and meta-analysis find that over 40 loci affect risk of type 1 diabetes
Nat Genet
 , 
2009
, vol. 
41
 (pg. 
703
-
7
)
42
Manolio
TA
Collins
FS
Cox
NJ
, et al.  . 
Finding the missing heritability of complex diseases
Nature
 , 
2009
, vol. 
461
 (pg. 
747
-
53
)
43
Prokopenko
I
McCarthy
MI
Lindgren
CM
Type 2 diabetes: new genes, new understanding
Trends Genet
 , 
2008
, vol. 
24
 (pg. 
613
-
21
)
44
Conrad
DF
Pinto
D
Redon
R
, et al.  . 
Origins and functional impact of copy number variation in the human genome
Nature
 , 
2010
, vol. 
464
 (pg. 
704
-
12
)
45
Craddock
N
Hurles
ME
Cardin
N
, et al.  . 
Genome-wide association study of CNVs in 16,000 cases of eight common diseases and 3,000 shared controls
Nature
 , 
2010
, vol. 
464
 (pg. 
713
-
20
)
46
Cirulli
ET
Goldstein
DB
Uncovering the roles of rare variants in common disease through whole-genome sequencing
Nat Rev Genet
 , 
2010
, vol. 
11
 (pg. 
415
-
25
)
47
Eichler
EE
Flint
J
Gibson
G
, et al.  . 
Missing heritability and strategies for finding the underlying causes of complex disease
Nat Rev Genet
 , 
2010
, vol. 
11
 (pg. 
446
-
50
)
48
Metzker
ML
Sequencing technologies - the next generation
Nat Rev Genet
 , 
2010
, vol. 
11
 (pg. 
31
-
46
)
49
Nejentsev
S
Walker
N
Riches
D
, et al.  . 
Rare variants of IFIH1, a gene implicated in antiviral responses, protect against type 1 diabetes
Science
 , 
2009
, vol. 
324
 (pg. 
387
-
9
)
50
Epstein
DJ
Cis-regulatory mutations in human disease
Brief Funct Genomic Proteomic
 , 
2009
, vol. 
8
 (pg. 
310
-
6
)
51
Hodge
SE
Ban
Y
Strug
LJ
, et al.  . 
Possible interaction between HLA-DRbeta1 and thyroglobulin variants in Graves' disease
Thyroid
 , 
2006
, vol. 
16
 (pg. 
351
-
5
)
52
Nadeau
JH
Transgenerational genetic effects on phenotypic variation and disease risk
Hum Mol Genet
 , 
2009
, vol. 
18
 (pg. 
R202
-
10
)
53
Kong
A
Steinthorsdottir
V
Masson
G
, et al.  . 
Parental origin of sequence variants associated with complex diseases
Nature
 , 
2009
, vol. 
462
 (pg. 
868
-
74
)
54
Round
JL
Mazmanian
SK
The gut microbiota shapes intestinal immune responses during health and disease
Nat Rev Immunol
 , 
2009
, vol. 
9
 (pg. 
313
-
23
)
55
Cerf-Bensussan
N
Gaboriau-Routhiau
V
The immune system and the gut microbiota: friends or foes?
Nat Rev Immunol
 , 
2010
, vol. 
10
 (pg. 
735
-
44
)
56
Ban
Y
Davies
TF
Greenberg
DA
, et al.  . 
Arginine at position 74 of the HLA-DR beta1 chain is associated with Graves' disease
Genes Immun
 , 
2004
, vol. 
5
 (pg. 
203
-
8
)
57
Chen
QY
Nadell
D
Zhang
XY
, et al.  . 
The human leukocyte antigen HLA DRB3*020/DQA1*0501 haplotype is associated with Graves' disease in African Americans
J Clin Endocrinol Metab
 , 
2000
, vol. 
85
 (pg. 
1545
-
9
)
58
Simmonds
MJ
Howson
JM
Heward
JM
, et al.  . 
Regression mapping of association between the human leukocyte antigen region and Graves disease
Am J Hum Genet
 , 
2005
, vol. 
76
 (pg. 
157
-
63
)
59
Wongsurawat
T
Nakkuntod
J
Charoenwongse
P
, et al.  . 
The association between HLA class II haplotype with Graves' disease in Thai population
Tissue Antigens
 , 
2006
, vol. 
67
 (pg. 
79
-
83
)
60
Furugaki
K
Shirasawa
S
Ishikawa
N
, et al.  . 
Association of the T-cell regulatory gene CTLA4 with Graves' disease and autoimmune thyroid disease in the Japanese
J Hum Genet
 , 
2004
, vol. 
49
 (pg. 
166
-
8
)
61
Ueda
H
Howson
JM
Esposito
L
, et al.  . 
Association of the T-cell regulatory gene CTLA4 with susceptibility to autoimmune disease
Nature
 , 
2003
, vol. 
423
 (pg. 
506
-
11
)
62
Vaidya
B
Imrie
H
Perros
P
, et al.  . 
The cytotoxic T lymphocyte antigen-4 is a major Graves' disease locus
Hum Mol Genet
 , 
1999
, vol. 
8
 (pg. 
1195
-
9
)
63
Yanagawa
T
Hidaka
Y
Guimaraes
V
, et al.  . 
CTLA-4 gene polymorphism associated with Graves' disease in a Caucasian population
J Clin Endocrinol Metab
 , 
1995
, vol. 
80
 (pg. 
41
-
5
)
64
Heward
JM
Brand
OJ
Barrett
JC
, et al.  . 
Association of PTPN22 haplotypes with Graves' disease
J Clin Endocrinol Metab
 , 
2007
, vol. 
92
 (pg. 
685
-
90
)
65
Smyth
D
Cooper
JD
Collins
JE
, et al.  . 
Replication of an association between the lymphoid tyrosine phosphatase locus (LYP/PTPN22) with type 1 diabetes, and evidence for its role as a general autoimmunity locus
Diabetes
 , 
2004
, vol. 
53
 (pg. 
3020
-
3
)
66
Velaga
MR
Wilson
V
Jennings
CE
, et al.  . 
The codon 620 tryptophan allele of the lymphoid tyrosine phosphatase (LYP) gene is a major determinant of Graves' disease
J Clin Endocrinol Metab
 , 
2004
, vol. 
89
 (pg. 
5862
-
5
)
67
Dechairo
BM
Zabaneh
D
Collins
J
, et al.  . 
Association of the TSHR gene with Graves' disease: the first disease specific locus
Eur J Hum Genet
 , 
2005
, vol. 
13
 (pg. 
1223
-
30
)
68
Hiratani
H
Bowden
DW
Ikegami
S
, et al.  . 
Multiple SNPs in intron 7 of thyrotropin receptor are associated with Graves' disease
J Clin Endocrinol Metab
 , 
2005
, vol. 
90
 (pg. 
2898
-
903
)
69
Houston
FA
Wilson
V
Jennings
CE
, et al.  . 
Role of the CD40 locus in Graves' disease
Thyroid
 , 
2004
, vol. 
14
 (pg. 
506
-
9
)
70
Jacobson
EM
Concepcion
E
Oashi
T
, et al.  . 
A Graves' Disease Associated Kozak Sequence Single Nucleotide Polymorphism Enhances the Efficiency of CD40 Gene Translation: A Case for Translational Pathophysiology
Endocrinology
 , 
2005
, vol. 
146
 (pg. 
2684
-
91
)
71
Jacobson
EM
Huber
AK
Akeno
N
, et al.  . 
A CD40 Kozak sequence polymorphism and susceptibility to antibody-mediated autoimmune conditions: the role of CD40 tissue-specific expression
Genes Immun
 , 
2007
, vol. 
8
 (pg. 
205
-
14
)
72
Tomer
Y
Concepcion
E
Greenberg
DA
A C/T single-nucleotide polymorphism in the region of the CD40 gene is associated with Graves' disease
Thyroid
 , 
2002
, vol. 
12
 (pg. 
1129
-
35
)
73
Owen
CJ
Kelly
H
Eden
JA
, et al.  . 
Analysis of the Fc receptor-like-3 (FCRL3) locus in Caucasians with autoimmune disorders suggests a complex pattern of disease association
J Clin Endocrinol Metab
 , 
2007
, vol. 
92
 (pg. 
1106
-
11
)
74
Farid
NR
Barnard
JM
Marshall
WH
The association of HLA with autoimmune thyroid disease in Newfoundland. The influence of HLA homozygosity in Graves' disease
Tissue Antigens
 , 
1976
, vol. 
8
 (pg. 
181
-
9
)
75
Grumet
FC
Payne
RO
Konishi
J
, et al.  . 
HL-A antigens as markers for disease susceptibility and autoimmunity in Graves' disease
J Clin Endocrinol Metab
 , 
1974
, vol. 
39
 (pg. 
1115
-
9
)
76
Huang
SM
Wu
TJ
Lee
TD
, et al.  . 
The association of HLA -A, -B, and -DRB1 genotypes with Graves' disease in Taiwanese people
Tissue Antigens
 , 
2003
, vol. 
61
 (pg. 
154
-
8
)
77
Mather
BA
Roberts
DF
Scanlon
MF
, et al.  . 
HLA antigens and thyroid autoantibodies in patients with Graves' disease and their first degree relatives
Clin Endocrinol
 , 
1980
, vol. 
12
 (pg. 
155
-
63
)
78
Simmonds
MJ
Howson
JM
Heward
JM
, et al.  . 
A novel and major association of HLA-C in Graves' disease that eclipses the classical HLA-DRB1 effect
Hum Mol Genet
 , 
2007
, vol. 
16
 (pg. 
2149
-
53
)
79
Sutherland
A
Davies
J
Owen
CJ
, et al.  . 
Genomic polymorphism at the interferon-induced helicase (IFIH1) locus contributes to Graves' disease susceptibility
J Clin Endocrinol Metab
 , 
2007
, vol. 
92
 (pg. 
3338
-
41
)
80
Chu
X
Dong
Y
Shen
M
, et al.  . 
Polymorphisms in the ADRB2 gene and Graves disease: a case-control study and a meta-analysis of available evidence
BMC Med Genet
 , 
2009
, vol. 
10
 pg. 
26
 
81
Chistiakov
DA
Savost'anov
KV
Turakulov
RI
Screening of SNPs at 18 positional candidate genes, located within the GD-1 locus on chromosome 14q23-q32, for susceptibility to Graves' disease: a TDT study
Mol Genet Metab
 , 
2004
, vol. 
83
 (pg. 
264
-
70
)
82
Yesmin
K
Hargreaves
C
Newby
PR
, et al.  . 
Association of FcGRIIa with Graves' disease: a potential role for dysregulated autoantibody clearance in disease onset/progression
Clin Endocrinol
 , 
2010
, vol. 
73
 (pg. 
119
-
25
)
83
Huber
AK
Jacobson
EM
Jazdzewski
K
, et al.  . 
Interleukin (IL)-23 receptor is a major susceptibility gene for Graves' ophthalmopathy: the IL-23/T-helper 17 axis extends to thyroid autoimmunity
J Clin Endocrinol Metab
 , 
2008
, vol. 
93
 (pg. 
1077
-
81
)
84
Chu
X
Dong
C
Lei
R
, et al.  . 
Polymorphisms in the interleukin 3 gene show strong association with susceptibility to Graves' disease in Chinese population
Genes Immun
 , 
2009
, vol. 
10
 (pg. 
260
-
6
)
85
Yang
Y
Lingling
S
Ying
J
, et al.  . 
Association study between the IL4, IL13, IRF1 and UGRP1 genes in chromosomal 5q31 region and Chinese Graves' disease
J Hum Genet
 , 
2005
, vol. 
50
 (pg. 
574
-
82
)
86
Heward
JM
Allahabadia
A
Sheppard
MC
, et al.  . 
Association of the large multifunctional proteasome (LMP2) gene with Graves' disease is a result of linkage disequilibrium with the HLA haplotype DRB1*0304-DQB1*02-DQA1*0501
Clin Endocrinol
 , 
1999
, vol. 
51
 (pg. 
115
-
8
)
87
Vives-Pi
M
Vargas
F
James
RF
, et al.  . 
Proteasome subunits, low-molecular-mass polypeptides 2 and 7 are hyperexpressed by target cells in autoimmune thyroid disease but not in insulin-dependent diabetes mellitus: implications for autoimmunity
Tissue Antigens
 , 
1997
, vol. 
50
 (pg. 
153
-
63
)
88
Newby
PR
Roberts-Davies
EL
Brand
OJ
, et al.  . 
Tag SNP screening of the PDCD1 gene for association with Graves' disease
Clin Endocrinol
 , 
2007
, vol. 
67
 (pg. 
125
-
8
)
89
Nakkuntod
J
Wongsurawat
T
Charoenwongse
P
, et al.  . 
Association of TNF-alpha, TNF-beta, IFN-gamma and IL-1Ra gene polymorphisms with Graves' disease in the Thai population
Asian Pac J Allergy Immunol
 , 
2006
, vol. 
24
 (pg. 
207
-
11
)
90
Simmonds
MJ
Heward
JM
Howson
JM
, et al.  . 
A systematic approach to the assessment of known TNF-alpha polymorphisms in Graves' disease
Genes Immun
 , 
2004
, vol. 
5
 (pg. 
267
-
73
)
91
Demaine
AG
Ratanachaiyavong
S
Pope
R
, et al.  . 
Thyroglobulin antibodies in Graves' disease are associated with T-cell receptor beta chain and major histocompatibility complex loci
Clin Exp Immunol
 , 
1989
, vol. 
77
 (pg. 
21
-
4
)
92
Pickerill
AP
Watson
PF
Tandon
N
, et al.  . 
T cell receptor beta chain gene polymorphisms in Graves' disease
Acta Endocrinol
 , 
1993
, vol. 
128
 (pg. 
499
-
502
)
93
Zhang
J
Hu
S
Wang
H
Expression of T cell receptor V(alpha) gene families in intrathyroidal T cells of Chinese patients with Graves' disease
Chin Med Sci J
 , 
2000
, vol. 
15
 (pg. 
175
-
8
)
94
Kotani
T
Kato
E
Hirai
K
, et al.  . 
Immunoglobulin G subclasses of anti-thyroid peroxidase autoantibodies in human autoimmune thyroid diseases
Endocrinol Jpn
 , 
1986
, vol. 
33
 (pg. 
505
-
10
)
95
Kotani
T
Umeki
K
Matsunaga
S
, et al.  . 
Detection of autoantibodies to thyroid peroxidase in autoimmune thyroid diseases by micro-ELISA and immunoblotting
J Clin Endocrinol Metab
 , 
1986
, vol. 
62
 (pg. 
928
-
33
)
96
Ludgate
M
Vassart
G
The molecular genetics of three thyroid autoantigens: thyroglobulin, thyroid peroxidase and the thyrotropin receptor
Autoimmunity
 , 
1990
, vol. 
7
 (pg. 
201
-
11
)
97
Pirro
MT
De Filippis
V
Di Cerbo
A
, et al.  . 
Thyroperoxidase microsatellite polymorphism in thyroid diseases
Thyroid
 , 
1995
, vol. 
5
 (pg. 
461
-
4
)
98
Ban
Y
Tozaki
T
Tobe
T
, et al.  . 
The regulatory T cell gene FOXP3 and genetic susceptibility to thyroid autoimmunity: an association analysis in Caucasian and Japanese cohorts
J Autoimmun
 , 
2007
, vol. 
28
 (pg. 
201
-
7
)
99
Owen
CJ
Eden
JA
Jennings
CE
, et al.  . 
Genetic association studies of the FOXP3 gene in Graves' disease and autoimmune Addison's disease in the United Kingdom population
J Mol Endocrinol
 , 
2006
, vol. 
37
 (pg. 
97
-
104
)
100
Bednarczuk
T
Placha
G
Jazdzewski
K
, et al.  . 
Interleukin-13 gene polymorphisms in patients with Graves' disease
Clin Endocrinol
 , 
2003
, vol. 
59
 (pg. 
519
-
25
)
101
Hiromatsu
Y
Fukutani
T
Ichimura
M
, et al.  . 
Interleukin-13 gene polymorphisms confer the susceptibility of Japanese populations to Graves' disease
J Clin Endocrinol Metab
 , 
2005
, vol. 
90
 (pg. 
296
-
301
)
102
Simmonds
MJ
Heward
JM
Franklyn
JA
, et al.  . 
IL-13 and chromosome 5q31-q33: problems of identifying association within regions of linkage to Graves' disease
Clin Endocrinol
 , 
2005
, vol. 
63
 (pg. 
695
-
7
)
103
Heward
JM
Nithiyananthan
R
Allahabadia
A
, et al.  . 
No association of an interleukin 4 gene promoter polymorphism with Graves' disease in the United Kingdom
J Clin Endocrinol Metab
 , 
2001
, vol. 
86
 (pg. 
3861
-
3
)
104
Blakemore
AI
Watson
PF
Weetman
AP
, et al.  . 
Association of Graves' disease with an allele of the interleukin-1 receptor antagonist gene
J Clin Endocrinol Metab
 , 
1995
, vol. 
80
 (pg. 
111
-
5
)
105
Muhlberg
T
Kirchberger
M
Spitzweg
C
, et al.  . 
Lack of association of Graves' disease with the A2 allele of the interleukin-1 receptor antagonist gene in a white European population
Eur J Endocrinol
 , 
1998
, vol. 
138
 (pg. 
686
-
90
)
106
Collins
JE
Heward
JM
Howson
JM
, et al.  . 
Common allelic variants of exons 10,12, and 33 of the thyroglobulin gene are not associated with autoimmune thyroid disease in the United Kingdom
J Clin Endocrinol Metab
 , 
2004
, vol. 
89
 (pg. 
6336
-
9
)
107
Collins
JE
Heward
JM
Carr-Smith
J
, et al.  . 
Association of a rare thyroglobulin gene microsatellite variant with autoimmune thyroid disease
J Clin Endocrinol Metab
 , 
2003
, vol. 
88
 (pg. 
5039
-
42
)
108
Tomer
Y
Greenberg
D
The thyroglobulin gene as the first thyroid-specific susceptibility gene for autoimmune thyroid disease
Trends Mol Med
 , 
2004
, vol. 
10
 (pg. 
306
-
8
)
109
Tomer
Y
Greenberg
DA
Concepcion
E
, et al.  . 
Thyroglobulin is a thyroid specific gene for the familial autoimmune thyroid diseases
J Clin Endocrinol Metab
 , 
2002
, vol. 
87
 (pg. 
404
-
7
)
110
Ban
Y
Ban
Y
Taniyama
M
, et al.  . 
Vitamin D receptor initiation codon polymorphism in Japanese patients with Graves' disease
Thyroid
 , 
2000
, vol. 
10
 (pg. 
475
-
80
)
111
Ban
Y
Taniyama
M
Ban
Y
Vitamin D receptor gene polymorphism is associated with Graves' disease in the Japanese population
J Clin Endocrinol Metab
 , 
2000
, vol. 
85
 (pg. 
4639
-
43
)
112
Collins
JE
Heward
JM
Nithiyananthan
R
, et al.  . 
Lack of association of the vitamin D receptor gene with Graves' disease in UK Caucasians
Clin Endocrinol
 , 
2004
, vol. 
60
 (pg. 
618
-
24
)
113
Ramos-Lopez
E
Kurylowicz
A
Bednarczuk
T
, et al.  . 
Vitamin D receptor polymorphisms are associated with Graves' disease in German and Polish but not in Serbian patients
Thyroid
 , 
2005
, vol. 
15
 (pg. 
1125
-
30
)
114
Stefanic
M
Karner
I
Glavas-Obrovac
L
, et al.  . 
Association of vitamin D receptor gene polymorphism with susceptibility to Graves' disease in Eastern Croatian population: case-control study
Croat Med J
 , 
2005
, vol. 
46
 (pg. 
639
-
46
)