Association of BTG2,CYR61,ZFP36,and SCD Gene Polymorphisms with Graves' Disease and Ophthalmopathy

Abstract

Background:

Environmental and genetic factors predispose an individual to the development of Graves' disease (GD). In an expression study of intraorbital tissue, adipocyte-related immediate early genes (IEGs) and immunomodulatory genes were found to be overexpressed in patients with Graves' ophthalmopathy (GO). We hypothesized that genetic variations in these genes could be associated with GD and/or GO.

Methods:

A total of 98 single nucleotide polymorphisms (SNPs) in 12 genes were genotyped in 594 GD patients with (n=267) or without (n=327) GO and 1147 sex- and ethnicity-matched controls from Malmö, Sweden.

Results:

Ten SNPs in four genes (BTG family, member 2 [BTG2], cysteine-rich, angiogenic inducer 61 [CYR61], zinc finger protein 36, C3H type, homolog mouse [ZFP36], and stearoyl-coenzyme A desaturase [SCD]) showed an association with GD and/or GO. SNPs rs12136280 (odds ratio [OR] 1.29, p=0.002), rs6663606 (OR 1.26, p=0.004), and rs17534202 (OR 1.21, p=0.02) in BTG2 and rs3753793 (OR 1.21, p=0.03) in CYR61 were associated with GD. An association with GO was shown for SNPs rs3753793 (OR 1.45, p=0.008), rs6682848 (OR 1.55, p=0.03), rs12756618 (OR 1.77, p=0.049), and rs1378228 (OR 1.29, p=0.049) in CYR61, rs1057745 (OR 1.56, p=0.03) and rs11083522 (OR 1.32, p=0.04) in ZFP36, and rs1393491 (OR 1.38, p=0,048) in SCD. Smoking and CYR61 rs12756618 interacted to increase the risk of GO.

Conclusions:

We found associations of SNPs in IEGs and SCD with GD and/or GO; however, confirmation in a different population is required.

Introduction

Graves' disease (GD) is an autoimmune disorder that develops as a result of the interplay between endogenous, genetic, and environmental factors. The two main groups of autoimmune thyroid disease (AITD)—GD and Hashimoto thyroiditis—share part of the genetic background, and the genetic pattern is complex (1). Although the genetics of GD has been extensively studied, only a few positive results from linkage and candidate gene studies have been replicated. Cytotoxic T-lymphocyte-associated protein 4 (CTLA4), a key negative regulator of the T cell–mediated immune response, is linked to and associated with all AITD phenotypes (including GD) in different populations and also with other autoimmune disorders (1).

Virtually all patients with GD display retrobulbar morphological changes, though only approximately one-third of GD patients develop Graves' ophthalmopathy (GO) (2); furthermore, the factors leading to the development of GO in an individual patient are not entirely known. Therefore, identifying parameters that help to predict GO is of great importance. Nonetheless, genetic studies of GO have thus far given conflicting results. The most promising candidate, CTLA4, was associated with GO in some studies (3,4), whereas others found no evidence for a specific risk of GO beyond that conferred for GD (5 –8). Known environmental risk factors include smoking (9,10), untreated hyperthyroidism and hypothyroidism (11), and treatment with radioiodine (12 –14).

In expression studies, we have shown that a number of genes were upregulated in intraorbital adipose/connective tissue from patients with severe active GO compared to thyroid-healthy controls (15). The major finding involved the overexpression of 14 adipocyte-related immediate early genes (IEGs) in the patient group, and these genes were also overexpressed, though to a lower degree, in chronic ophthalmopathy (16). We hypothesized that genes showing abnormal expression in intraorbital tissue from patients with GO might also play a role in the genetic susceptibility to GD and GO. The majority of the genes with a confirmed association with GD are involved in the immune response. Moreover, previous studies have demonstrated abnormalities in monocytes and lymphocytes from patients with AITD (17), and lymphocyte and monocyte infiltration is shared by the thyroid and intraorbital tissue in GD. Therefore, among the genes overexpressed in active GO, IEGs and genes with functions in lymphocytes/monocytes were considered plausible candidates for an association with GO and/or GD.

Materials and Methods

Study population

The study consisted of 594 patients (109 male and 485 female) with GD with (n=267) or without (n=327) ophthalmopathy and 1147 controls (204 male and 943 female) from one town, Malmö, in southern Sweden (Table 1). A total of 477 patients (80%) were selected from the registry GD2002 in which clinical data on patients with GD have been collected since 2002 by a single endocrinologist. The remaining 117 patients were recruited from the studies MFM (Malmö Preventive Project) (18) and MKC (Malmö Diet and Cancer Study) (19) on the basis of an International Statistical Classification of Diseases and Related Health Problems, 10th revision, diagnosis of E05.0 (GD) and H06.2 (GO). These studies were performed in the 1990s as a primary preventive project (MFM) and a study investigating the association of diet and risk of cancer (MKC). As patients with thyrotoxicosis in Malmö are always referred to an endocrinologist, in all cases, diagnosis of GD was made by an endocrinologist based on clinical symptoms and signs and biochemical hyperthyroidism, in combination with the presence of thyrotropin receptor antibodies and/or diffuse uptake by technetium scintigraphy. GO diagnosis was made by an endocrinologist and/or ophthalmologist based on the presence of clinical signs, as previously described (14). Eyelid retraction alone was not classified as GO. The controls were recruited from the MFM and MKC databases. The case and control groups were matched for sex and group-matched for ethnicity. Regarding ethnicity, the origin of the individuals was assigned as Swedish, born in Europe other than Sweden, or born outside of Europe. For additional analysis, a total of 116 newly recruited patients with GD from GD2002 (n=44, not previously analyzed) and from the TT 96 trial [n=72, described in (14)] were used. For these newly recruited patients, data on clinical parameters such as age, ethnicity, smoking and presence of GO, were not available. All patients gave written informed consent, and the study was approved by the local research ethics committee.

Table 1.

Characteristics of the Study Population

	Cases	Controls
N	594	1147
Age at inclusion (years)^a	49±14	57±6
Sex
Male (%)	109 (18.4)	204 (17.8)
Female (%)	485 (81.6)	943 (82.2)
Ethnicity
SWE (%)	434 (73.1)	835 (72.8)
EUR (%)	100 (16.8)	177 (15.4)
OUT (%)	59 (9.9)	84 (7.3)
Missing (%)	1 (0.2)	51 (4.4)
Smoking
Yes (%)	229 (38.6)	345 (30.1)
No (%)	338 (56.9)	802 (69.9)
Missing (%)	27 (4.5)	0 (0)
Ophthalmopathy
Yes (%)	267 (44.9)	0 (0)
No (%)	327 (55.1)	1147 (100)

Data for age are presented as the mean±SD.

SWE, Swedish; EUR, born in Europe outside of Sweden; OUT, born outside of Europe.

Gene selection

Gene selection was based on the results of a previous expression study of active GO (15). Genes with high expression in intraorbital adipose/connective tissue from patients with severe active ophthalmopathy compared to thyroid-healthy controls were chosen, including BTG family, member 2 (BTG2), cysteine-rich, angiogenic inducer, 61 (CYR61), early growth response 1 (EGR1), dual specificity phosphatase 1 (DUSP1), zinc finger protein 36, C3H type, homolog [mouse] (ZFP36), and prostaglandin-endoperoxide synthase 2 (PTGS2), representing adipocyte-related IEGs with functions in lymphocytes and/or monocytes. In addition, the following genes with known functions in inflammation and/or adipogenesis were selected: nuclear receptor subfamily 4, group A, member 2 (NR4A2), chemokine [C-X-C motif ] ligand 2 (CXCL2), suppressor of cytokine signaling 3 (SOCS3), regulator of G-protein signaling 2, 24 kDa (RGS2), and stearoyl-coenzyme A desaturase (SCD). CTLA4 was chosen for confirmation in our population on the basis of previously reported associations with GD.

Single nucleotide polymorphism selection

Single nucleotide polymorphisms (SNPs) were selected using data from the Hap Map consortium (20). Applying the tag SNP approach reduces the amount of SNPs that have to be genotyped to cover the genetic variation in a region. Therefore, the region surrounding each of the selected genes, including an extra 10 kb upstream and downstream, was analyzed by Tagger in the Haplowiew program (21), providing tag SNPs for each gene. Additional potentially important SNPs (nonsynonymous SNPs; SNPs that alter an intronic splice site; and SNPs that occur in a predicted promoter, in a region of predicted regulatory potential, or in a region with conservation scores ≥99th percentile genomewide) were chosen from TAMAL (http://neoref.ils.unc.edu/tamal/), as described previously (22), and from the National Center for Biotechnology Information SNP database (www.ncbi.nlm.nih.gov/SNP/). In total, 98 SNPs in twelve genes were selected for the analysis. The minor allele frequency (MAF) for all SNPs was >0.05.

Power calculations

As the allele frequencies in our population of most SNPs studied in this study were unknown prior to genotyping, we based our power calculations on the CTLA4 SNPs where the allele frequencies were known from other studies using the Genetic Power Calculator (23). Described MAF for the previously associated CTLA4 SNPs varied between 0.07 and 0.47. Given a control/case ratio of 2, MAF 0.30, and odds ratio (OR) 1.39, as presented in the recent GWAS for GD (24), the number of cases required for 80% power and significance 0.05 is 500. Given MAF 0.40 and OR 1.39, the number of cases required for 80% power and significance 0.05 is 637.

Genotyping

DNA was extracted from whole blood using the MaxiPrep Kit (QIAGEN, Sweden), and SNPs were genotyped using the Sequenom platform (MALDI-TOF) at the DNA/RNA Genotyping Lab, SWEGENE Resource Center for Profiling Polygenic Disease, Lund University, Skåne University Hospital, Malmö, Sweden, and in our laboratory. A total of 15 SNPs failed the genotyping (n=8) or Hardy Weinberg equilibrium (n=7) analysis. These SNPs were removed from the further analysis, leaving a total of 83 SNPs. The SNPs included in the further analysis had a genotyping success rate >95%.

Statistical analysis

Logistic regression with age and smoking as covariates and standard case/control association for the additional analysis including the 116 newly recruited cases were used for estimating SNP associations. Analyses of the combined effects of SNPs and smoking were performed using additive models of interaction within a logistic regression framework. The data are presented as odds ratios (OR) with 95% confidence intervals (CI). The p-values are based on additive models for the genetic variants. Correction for multiple testing was performed using permutations (maxT permutation method in PLINK). All statistical calculations were performed using PLINK version 1.07 (http://pngu.mgh.harvard.edu/∼purcell/plink/index.shtml) (25).

Results

SNP association with Graves' disease

Using logistic regression, a total of three tag SNPs in the BTG2 gene region showed an association with GD: rs12136280 (OR 1.29, p=0.002), rs6663606 (OR 1.26, p=0.004), and rs17534202 (OR 1.21, p=0.02). The promoter SNP rs3753793 in the CYR61 gene was also associated with GD (OR 1.21, p=0.03) (Table 2). A condition analysis revealed that the significant SNPs in BTG2 influence each other but have an effect that is independent of the CYR61 SNP rs3753793. None of the investigated SNPs in CTLA4 were associated with GD. However, including the 116 newly recruited patients with GD, the SNPs rs11571297 (OR 1.15, p=0.05) and rs3087243 (OR 1.15, p=0.05) in CTLA4 showed association with GD with borderline significance. The above described significantly associated SNPs (Table 2) remained significant even after inclusion of the new 116 GD cases. Correction for multiple testing was done using permutations, with the maxT method performing 1000 permutations. Without the additional 116 GD individuals, the association of rs12136280 with GD showed borderline significance (corrected empirical p-value of 0.06). Including the additional 116 GD individuals, the association of rs12136280 with GD remained significant (corrected empirical p-value of 0.047). The significance of the association between the remaining SNPs associated with GD (Table 2) disappeared after correction for multiple testing both without and with the additional 116 cases included. However, a new analysis with Haploview using most recent information on the presence of linkage disequilibrium (LD) revealed that many of the analyzed SNPs in BTG2, CYR61, ZFP36, SCD, PTGS2, EGR1, DUSP1, NR4A2, RGS2, CXCL2, and SOCS3 were not independent of each other due to the presence of LD; therefore, adjusting for the number of tested sites rather than the number of tested SNPs might be more appropriate, and with this approach, the reported associations remain significant (data not shown).

Table 2.

Single Nucleotide Polymorphisms Associated with Graves' Disease

Gene	SNP	Selection criteria	Minor allele frequency	Associated allele	OR for associated allele (95% CI)	p-Value
BTG2	rs12136280	tag, missense	0.46 (T)	T	1.29 (1.10–1.52)	0.002
	rs6663606	tag	0.46 (T)	A	1.26 (1.08–1.48)	0.004
	rs17534202	tag	0.47 (G)	C	1.21 (1.03–1.41)	0.02
CYR61	rs3753793	tag, promoter	0.27 (G)	G	1.21 (1.02–1.45)	0.03

The p-values were calculated using logistic regression with age and smoking as covariates.

CI, confidence interval; OD, odds ratio; SNP, single nucleotide polymorphism.

SNP association with Graves' ophthalmopathy

Four SNPs in the CYR61 gene region were associated with GO: the promoter SNP rs3753793 (OR 1.45, p=0.008) and the tag SNPs rs6682848 (OR 1.55, p=0.03), rs12756618 (OR 1.77, p=0.049), and rs1378228 (OR 1.29, p=0.049) (Table 3). An association with GO was also shown for the ZFP36 tag SNPs rs1057745 (OR 1.56, p=0.03) and rs11083522 (OR 1.32, p=0.04) and for the SCD tag SNP rs1393491 (OR 1.38, p=0.048) (Table 3). The significance of these associations disappeared after correction for multiple testing. In the condition analysis, the significant SNPs in CYR61 influenced each other but were independent of the significant SNPs in ZFP36 and SCD. The same result was obtained for the significant SNPs in ZFP36, which influenced each other, but were independent of the significant SNPs in CYR61 and SCD. None of the investigated SNPs in CTLA4 were associated with GO.

Table 3.

Single Nucleotide Polymorphisms Associated with Graves' Ophthalmopathy

Gene	SNP	Selection criteria	Minor allele frequency	Associated allele	OR for associated allele (95% CI)	p-Value
CYR61	rs3753793	tag, promoter	0.27 (G)	T	1.45 (1.91–2.64)	0.008
	rs6682848	tag	0.10 (A)	A	1.55 (1.04–2.31)	0.03
	rs12756618	tag	0.06 (T)	C	1.77 (1.00–3.12)	0.049
	rs1378228	tag	0.36 (T)	G	1.29 (1.00–1.66)	0.049
ZFP36	rs11083522	tag	0.29 (C)	C	1.32 (1.02–1.71)	0.04
	rs1057745	tag	0.10 (C)	C	1.56 (1.05–2.33)	0.03
SCD	rs1393491	tag	0.17 (C)	T	1.38 (1.00–1.91)	0.048

The p-values were calculated using logistic regression with age and smoking as covariates.

Interactions between SNPs and smoking

Because smoking is a risk factor for GD and a strong risk factor for GO, we investigated the possible interactions between smoking and the significant SNPs in GD and GO. Although no interaction with smoking was observed for the SNPs that were associated with GD, an interaction between smoking and the CYR61 SNP rs12756618 was observed, further increasing the risk of GO (OR 4.75, CI 1.33-17.05, p=0.02).

Discussion

In this study, we used the functional candidate gene approach to examine possible associations with GD and GO. Except for CTLA4, the candidate genes were chosen on the basis of our earlier expression study in the same population (15), in which the principal finding was the overexpression of 14 adipocyte-related IEGs in active GO. Variants in three IEGs (BTG2, CYR61, and ZFP36) and in the proadipogenic gene SCD showed an association with GD and/or GO.

BTG2 is a member of the BTG/Tob gene family with antiproliferative effects and belongs to a group of genes differentially expressed between resting and activated mouse T cells (26). Because BTG2 is a downstream target of CD28, it is tempting to speculate that BTG2 could be involved in the inhibition of T cell activation, together with other negative regulators of activated T cells associated with GD, such as CTLA4 and protein tyrosine phosphatase, non-receptor type 22 (PTPN22). Associations of three SNPs in the BTG2 gene region with GD were shown in this study. However, the causative variant is not known, and all SNPs tested were tag SNPs, leaving the possibility of the causative SNP being in LD with any of the associated SNPs. Nonetheless, rs12136280 is a coding SNP leading to an amino acid alteration (Val→Met at position 153; V153M), making it the most promising candidate.

Seven SNPs in three genes, CYR61, ZFP36, and SCD, were associated with GO in this study. CYR61 is a multifunctional gene with roles in adipogenesis, inflammation, cell proliferation, extracellular matrix production, and fibrosis (27,28). Previous studies have implicated CYR61 in the pathogenesis of inflammatory diseases, including rheumatoid arthritis (RA), and a microarray analysis of B cells from monozygotic twins revealed a significantly higher expression of CYR61 in the twin with RA compared to the healthy twin (29). CYR61 is also overexpressed in RA synovial tissue and plays a critical role in the proliferation of rheumatoid arthritis synovial fibroblasts (30). CYR61 expression is induced in human skin dermal fibroblasts after exposure to cigarette smoke (31). In our study, four SNPs in CYR61 were associated with GO, with one, the promoter SNP rs3753793, even being associated with GD. As smoking is a strong risk factor for GO and even a risk factor for GD, one hypothesis is that the associated risk variants could interact with smoking to increase the risk for GO and GD. In our study, an interaction between rs12756618 in CYR61 and smoking was observed. However, our study most likely does not have sufficient power to detect all possible gene-environment interactions.

ZFP36 codes for the protein tristetraprolin (TTP), which promotes the destruction of the tumor necrosis factor-alpha mRNA. (32). TTP-knockout mice exhibit a severe inflammatory phenotype (33), which makes TTP an interesting candidate for associations with autoimmunity. Indeed, the SNP ZFP36*8 was found to be associated with RA in African Americans in a genetic study of autoimmune diseases (34). In the present study, we demonstrate an association of two SNPs with GO.

IEGs are a group of genes known to be expressed in differentiating preadipocytes and mitogen-stimulated fibroblasts (35,36). The expression of IEGs during adipogenesis initiates the mitotic clonal expansion phase of growth-arrested preadipocytes, which is followed by a second growth-arrest phase and then the terminal differentiation phase, leading to the mature adipocyte phenotype (37). In our expression studies of active (15) and chronic (16) ophthalmopathy, IEGs showed markedly increased expression in patients with active GO compared to controls and moderately increased expression in patients with chronic GO compared to controls. However, it remains unknown whether the overexpression of an entire group of adipocyte-related IEGs contributes to the etiology of GO or is rather a consequence of the disease. The finding of multiple SNPs in the IEGs CYR61, ZFP36, and BTG2 as being associated with GO supports the hypothesis that genetic variation in this group of genes could contribute to the development of GO.

SCD is an enzyme that catalyzes a rate-limiting step in the synthesis of unsaturated fatty acids. It is a proadipogenic gene, and genetic variations in the SCD gene have been associated with body fat distribution and insulin sensitivity in Swedish men (38). In the present study, we showed a borderline association of one SCD SNP with GO but not with GD. Thus, it could be speculated that the genetic variant associated with GO promotes adipogenesis, a key pathogenetic process in GO.

We were initially not able to confirm the previously reported associations of SNPs in CTLA4 and GD or GO, including 594 cases and 1147 controls. The allele frequencies in our population were similar to those reported in the other association studies and the diagnostic criteria for GD and GO the same. For these allele frequencies (MAF around 0.4), the sample size might be slightly small, as demonstrated in the power calculation. Indeed, including additional 116 cases with GD in the analysis resulted in association of two CTLA4 SNPs with GD (with borderline significance), suggesting that the lack of association between CTLA4 and GD can probably at least partly be explained by the sample size. However, the Wellcome Trust Case Control Consortium recently performed a large study with a genome-wide set of nonsynonymous coding variants and provided evidence that three previously reported loci (major histocompatibility complex [MHC], thyroid stimulating hormone receptor [TSHR], and Fc receptor-like 3 [FCRL3]) were associated with GD in individuals of European ancestry (39). In this study, associations with the CTLA4 locus were not shown, suggesting that diversity may exist between different populations.

Lastly, several methodological issues need to be discussed. It is well known that genetic diversity exists between different ethnical groups, and we attempted to overcome this problem by group-matching the cases and controls in three groups, Swedish, European outside of Sweden, and non-European. We then tested the effect of ethnicity in the logistic regression analysis and could not observe any effect. We therefore assumed that the group-matching in our population was sufficient. In contrast, age and smoking, factors known to influence the etiology of GD, were significantly associated with both GD and GO in our population and were therefore used as covariates in the logistic regression analysis to minimize the effect of confounding factors. The number of subjects in the study is in line with most association studies for GD; regardless, recent research has shown that larger numbers of individuals are needed to avoid false-positive results. Our cohort, however, has the advantages of being phenotypically well defined and consisting of a population from a single town. Moreover, the choice of the candidate genes was based on expression data in the same population and the associated genes are biologically relevant to the disease. Furthermore, we have controlled for three known confounding factors in our analyses, which has not been the case in most association studies for GD/GO.

In conclusion, the results of this study suggest associations of SNPs in BTG2, CYR61, ZFP36, and SCD with GD and/or GO. Nonetheless, replication in a different population is required.

Footnotes

Acknowledgments

The study was supported by grants from Region Skåne, Svenska Läkarsällskapet, and Svenska Endokrinologföreningen. We are grateful to Anders Dahlin for help with the choice and matching of the controls and Peter Almgren and Claes Ladenvall for support with the statistical issues.

Author Disclosure Statement

No competing financial interests exist.

References

Tomer

. 2010. Genetic susceptibility to autoimmune thyroid disease: past, present, and future. Thyroid, 20:715–725.

Bahn

. 2010. Graves' ophthalmopathy. New Engl J Med, 362:726–738.

Vaidya

, Imrie

, Perros

, Dickinson

, McCarthy

, Kendall-Taylor

, Pearce

. 1999. Cytotoxic T lymphocyte antigen-4 (CTLA-4) gene polymorphism confers susceptibility to thyroid associated orbitopathy. Lancet, 354:743–744.

Park

, Chung

, Park

, Kim

, Koh

, Cho

. 2000. Polymorphism in the promoter and exon 1 of the cytotoxic T lymphocyte antigen-4 gene associated with autoimmune thyroid disease in Koreans. Thyroid, 10:453–459.

Villanueva

, Inzerillo

, Tomer

, Barbesino

, Meltzer

, Concepcion

, Greenberg

, MacLaren

, Sun

, Zhang

, Tucci

, Davies

. 2000. Limited genetic susceptibility to severe Graves' ophthalmopathy: no role for CTLA-4 but evidence for an environmental etiology. Thyroid, 10:791–798.

Allahabadia

, Heward

, Nithiyananthan

, Gibson

, Reuser

, Dodson

, Franklyn

, Gough

. 2001. MHC class II region, CTLA4 gene, and ophthalmopathy in patients with Graves' disease. Lancet, 358:984–985.

Zaletel

, Krhin

, Gaberscek

, Pirnat

, Hojker

. 2002. The influence of the exon 1 polymorphism of the cytotoxic T lymphocyte antigen 4 gene on thyroid antibody production in patients with newly diagnosed Graves' disease. Thyroid, 12:373–376.

Yin

, Latif

, Bahn

, Davies

. 2012. Genetic profiling in Graves' disease: further evidence for lack of a distinct genetic contribution to Graves' ophthalmopathy. Thyroid, 22:730–736.

Hagg

, Asplund

. 1987. Is endocrine ophthalmopathy related to smoking?. Br Med J (Clin Res Ed), 295:634–635.

10.

Thornton

, Kelly

, Harrison

, Edwards

. 2007. Cigarette smoking and thyroid eye disease: a systematic review. Eye (Lond), 21:1135–1145.

11.

Wiersinga

, Bartalena

. 2002. Epidemiology and prevention of Graves' ophthalmopathy. Thyroid, 12:855–860.

12.

Bartalena

, Marcocci

, Bogazzi

, Manetti

, Tanda

, Dell'Unto

, Bruno-Bossio

, Nardi

, Bartolomei

, Lepri

, Rossi

, Martino

, Pinchera

. 1998. Relation between therapy for hyperthyroidism and the course of Graves' ophthalmopathy. New Engl J Med, 338:73–78.

13.

Tallstedt

, Lundell

, Torring

, Wallin

, Ljunggren

, Blomgren

, Taube

. 1992. Occurrence of ophthalmopathy after treatment for Graves' hyperthyroidism. The Thyroid Study Group. New Engl J Med, 326:1733–1738.

14.

Traisk

, Tallstedt

, Abraham-Nordling

, Andersson

, Berg

, Calissendorff

, Hallengren

, Hedner

, Lantz

, Nystrom

, Ponjavic

, Taube

, Torring

, Wallin

, Asman

, Lundell

. 2009. Thyroid-associated ophthalmopathy after treatment for Graves' hyperthyroidism with antithyroid drugs or iodine-131. J Clin Endocrinol Metab, 94:3700–3707.

15.

Lantz

, Vondrichova

, Parikh

, Frenander

, Ridderstrale

, Asman

, Aberg

, Groop

, Hallengren

. 2005. Overexpression of immediate early genes in active Graves' ophthalmopathy. J Clin Endocrinol Metab, 90:4784–4791.

16.

Planck

, Parikh

, Brorson

, Martensson

, Asman

, Groop

, Hallengren

, Lantz

. Gene expression in Graves' ophthalmopathy and arm lymphedema: similarities and differences. Thyroid, 21:663–674.

17.

Canning

, Ruwhof

, Drexhage

. 2003. Aberrancies in antigen-presenting cells and T cells in autoimmune thyroid disease. A role in faulty tolerance induction. Autoimmunity, 36:429–442.

18.

Nilsson

, Nilsson

, Berglund

. 2006. Population-attributable risk of coronary heart disease risk factors during long-term follow-up: the Malmo Preventive Project. J Intern Med, 260:134–141.

19.

Manjer

, Carlsson

, Elmstahl

, Gullberg

, Janzon

, Lindstrom

, Mattisson

, Berglund

. 2001. The Malmo Diet and Cancer Study: representativity, cancer incidence and mortality in participants and non-participants. Eur J Cancer Prev, 10:489–499.

20.

International HapMap Consortium. 2003. The International HapMap Project. Nature, 426:789–796.

21.

Barrett

, Fry

, Maller

, Daly

. 2005. Haploview: analysis and visualization of LD and haplotype maps. Bioinformatics, 21:263–265.

22.

Hemminger

, Saelim

, Sullivan

. 2006. TAMAL: an integrated approach to choosing SNPs for genetic studies of human complex traits. Bioinformatics, 22:626–627.

23.

Purcell

, Cherny

, Sham

. 2003. Genetic Power Calculator: design of linkage and association genetic mapping studies of complex traits. Bioinformatics, 19:149–150.

24.

Wellcome Trust Case Control Consortium, Maller

, McVean

, Byrnes

, Vukcevic

, Palin

, Su

, Howson

et al. 2012. Bayesian refinement of association signals for 14 loci in 3 common diseases. Nat Genet, 44:1294–1301.

25.

Purcell

, Neale

, Todd-Brown

, Thomas

, Ferreira

, Bender

, Maller

, Sklar

, de Bakker

, Daly

, Sham

. 2007. PLINK: a tool set for whole-genome association and population-based linkage analyses. Am J Hum Genet, 81:559–575.

26.

Terra

, Luo

, Qiao

, Wu

. 2008. Tissue-specific expression of B-cell translocation gene 2 (BTG2) and its function in T-cell immune responses in a transgenic mouse model. Int Immunol, 20:317–326.

27.

Kireeva

, Mo

, Yang

, Lau

. 1996. Cyr61, a product of a growth factor-inducible immediate-early gene, promotes cell proliferation, migration, and adhesion. Mol Cell Biol, 16:1326–1334.

28.

Yeger

, Perbal

. 2007. The CCN family of genes: a perspective on CCN biology and therapeutic potential. J Cell Commun Signal, 1:159–164.

29.

Haas

, Creighton

, Pi

, Maine

, Koch

, Haines

, Ling

, Chinnaiyan

, Holoshitz

. 2006. Identification of genes modulated in rheumatoid arthritis using complementary DNA microarray analysis of lymphoblastoid B cell lines from disease-discordant monozygotic twins. Arthritis Rheum, 54:2047–2060.

30.

Zhang

, Wu

, Cao

, Xiao

, Wang

, He

, Ouyang

, Lin

, Shen

, Shi

, Zhang

, Li

. 2009. A critical role of Cyr61 in interleukin-17-dependent proliferation of fibroblast-like synoviocytes in rheumatoid arthritis. Arthritis Rheum, 60:3602–3612.

31.

Kim

, Kim

, Jeong

, Kye

, Son

. 2011. Cigarette smoke-induced early growth response-1 regulates the expression of the cysteine-rich 61 in human skin dermal fibroblasts. Exp Dermatol, 20:992–997.

32.

Lai

, Carballo

, Strum

, Kennington

, Phillips

, Blackshear

. 1999. Evidence that tristetraprolin binds to AU-rich elements and promotes the deadenylation and destabilization of tumor necrosis factor alpha mRNA. Mol Cell Biol, 19:4311–4323.

33.

Taylor

, Carballo

, Lee

, Lai

, Thompson

, Patel

, Schenkman

, Gilkeson

, Broxmeyer

, Haynes

, Blackshear

. 1996. A pathogenetic role for TNF alpha in the syndrome of cachexia, arthritis, and autoimmunity resulting from tristetraprolin (TTP) deficiency. Immunity, 4:445–454.

34.

Carrick

, Chulada

, Donn

, Fabris

, McNicholl

, Whitworth

, Blackshear

. 2006. Genetic variations in ZFP36 and their possible relationship to autoimmune diseases. J Autoimmun, 26:182–196.

35.

Almendral

, Sommer

, Macdonald-Bravo

, Burckhardt

, Perera

, Bravo

. 1988. Complexity of the early genetic response to growth factors in mouse fibroblasts. Mol Cell Biol, 8:2140–2148.

36.

Inuzuka

, Nanbu-Wakao

, Masuho

, Muramatsu

, Tojo

, Wakao

. 1999. Differential regulation of immediate early gene expression in preadipocyte cells through multiple signaling pathways. Biochem Biophys Res Commun, 265:664–668.

37.

Rosen

. 2002. The molecular control of adipogenesis, with special reference to lymphatic pathology. Ann NY Acad Sci, 979:143–158; discussion 188–196.

38.

Warensjo

, Ingelsson

, Lundmark

, Lannfelt

, Syvanen

, Vessby

, Riserus

. 2007. Polymorphisms in the SCD1 gene: associations with body fat distribution and insulin sensitivity. Obesity (Silver Spring), 15:1732–1740.

39.

Wellcome Trust Case Control Consortium, Australo-Anglo-American Spondylitis Consortium, Burton

, Clayton

, Cardon

, Craddock

, Deloukas

, Duncanson

, Kwiatkowski

et al. 2007. Association scan of 14,500 nonsynonymous SNPs in four diseases identifies autoimmunity variants. Nat Genet, 39:1329–1337.