Genetic Study in a Large Cohort Supported Different Pathogenesis of Graves’ Disease and Hashimoto’s Hypothyroidism Free

Abstract

Graves’ disease (GD) and Hashimoto’s thyroiditis (HT) are the 2 main forms of autoimmune thyroid diseases (AITDs). Hashimoto’s thyroiditis is the most common cause of hypothyroidism, with about 11% and 4.6% of positive TPOAb and hypothyroidism (0.3% clinical and 4.3% subclinical), respectively, in the U.S. population (1). In classical HT, there is a diffuse lymphocytic infiltration with germinal center formation and the destruction of thyroid follicle cells by widespread apoptosis, accompanied by a variable degree of fibrosis (2). In GD patients, there is hyperthyroidism, diffuse goiter, and even exophthalmos with autoantibodies to thyroid-stimulating hormone receptor (TSHR). In GD patients, the thyroid gland is characterized by nonhomogeneous lymphocytic infiltration with an absence of easily found follicular destruction, and the thyroid follicles were small and lined with hyperplastic columnar epithelium (2). The thyroid follicular epithelial cell size correlates with the intensity of the local infiltration. Thus, HT and GD used to be considered as distinct entities. However, both GD and HT could occur within the same families, and they may share some susceptibility genes (3). Interestingly, both GD and HT patients could have autoantibodies to thyroglobulin (Tg) and thyroid peroxidase (TPO). TPOAb and TgAb present in some GD patients and may contribute to an increased risk of developing hypothyroidism (4). In fact, patients with both GD and HT have thyroiditis with immune cell infiltration, suggesting a simple classification for AITD into destructive and nondestructive pathologies (5). Therefore, in this respect, HT and GD must be closely related.

Autoimmune thyroid disease was caused by the interaction of genetic and environmental factors. In recent years, our group (6–9) and others (10, 11) have identified and confirmed several GD susceptibility genes through genome-wide association studies (GWAS) or positional candidate cloning in large-scale, distinct ethnic populations. Most of the novel susceptibility genes identified from the Chinese Han cohort have been confirmed in other populations; at least one single-nucleotide polymorphism (SNP) in each of the 17 regions, including FCRL3, SLAMF6, TRIB2, CTLA4/CD28, RHOH/CHRNA9, HLA-B, HLA-DPB1, BACH2, RNASET2, TG, ABO, FAM76B/SESN3, TSHR, C14orf177/VRK1, CD40, C1QTNF6, and GPR174/ITM2A, reached genome-wide significance of 5 × 10^–8. Autoimmune diseases share many susceptibility genes, and for GD and HT, both common and different genetic susceptibilities have been implicated in previous studies (10, 12, 13). Thus, in this study, the 17 SNPs with genome-wide significance difference in 17 well-confirmed susceptibility regions of GD as well as 1 SNP in LPP, which was reported to be a HT risk locus in the European population (P = 7.09 × 10^–3, OR = 1.20), were genotyped in 2682 patients with HT, 4980 patients with GD, and 3892 controls to identify the susceptibility genes for HT and analyze the heterogeneity between GD and HT. These findings will provide important evidence to reveal the difference in the pathogenic mechanisms between HT and GD.

Subjects and Methods

Subjects and samples collection

With cooperation from multiple hospitals, we recruited patients and controls from the Chinaese Han populations in Shandong, Jiangsu, Anhui, and Shanghai (6–9). All the patients provided informed written consent and the project was conducted with approval from the ethics committee of the Shanghai Ninth People’s Hospital, Linyi People’s Hospital, Hospital Affiliated to Jiangsu University, and the Central Hospital of Xuzhou.

Diagnosis criteria of HT includes biochemical hypothyroidism (1) requiring thyroid hormone replacement or enlarged thyroid-positive autoantibodies (2) to TPO or Tg in at least 2 different laboratory tests (TPOAb > 100 IU/ml or TgAb > 300 IU/ml). Antithyroglobulin and anti-TPO antibodies were measured by specific radioimmunoassay (Roche, Mannheim, Germany). Diagnosis of GD was based on the principles as previously described: documented clinical and biochemical evidence of hyperthyroidism, diffuse goiter, and the presence of at least 1 of positive TSHR antibody tests, increased ¹³¹I (iodine^–131) uptake in the thyroid gland, or exophthalmos (6–8). All individuals classified as having GD or HT were interviewed and examined by experienced clinicians. All the controls were individuals over 35 years of age without AITD, a family history of AITD, or any other autoimmune disorders. Each participant provided written informed consent. All the pertinent clinical and laboratory data were recorded and stored in our database. At the time of the interview, blood was collected for DNA purification. Clinical characteristics of HT and GD patients and controls are summarized in Table1.

Genotyping and quality control

Genomic DNA of all subjects was extracted from 5 ml of blood from each participant with QuickGene-610 system (FUJIFILM, Tokyo, Japan). The SNPs were genotyped in HT patients and controls by 192.24 Dynamic Array IFC (Fluidigm, South San Francisco, CA) in BioMark HD System (Fluidigm, South San Francisco, CA) using the Taqman fast genotyping procedure. Reagents used were as recommended in the instruction of 192.24 Dynamic Array IFC. Single-nucleotide polymorphism genotyping in GD patients was performed either by 192.24 Dynamic Array IFC (Fluidigm) in BioMark HD System (Fluidigm) or TaqMan SNP Genotyping Assay on ABI Vii Real-Time PCR system. The concordance rate of the 2 methods was tested to be 100%. Although the genotypes of the 9529 Graves’ patients were reported in our previous study, patients with genotyping results of all the 18 SNPs were preferred in the current study. After genotyping, for all SNPs, the genotyping call rate was higher than 98% and minor allele frequency (MAF) was higher than 0.05.

Statistics

Association analysis of 18 SNPs in HT patients and controls were carried out on statistic package for social science (SPSS) using the logistic regression analysis adjusted for sex. Heterogeneity analysis between HT and GD patients was carried out on SPSS using the logistic regression analysis adjusted for sex and age. False-positive report probability (FPRP) analysis was calculated using the FPRP calculation spreadsheet provided by Wacholder et al (14).

Results

Genotyping of 18 GD-risk SNPs in HT and control to identify the susceptibility genes to HT

The 18 selected SNPs were genotyped in 2682 Chinese Han patients with HT and 3892 controls by 192.24 Dynamic Array IFC (Fluidigm) (Table 1). After quality control for the genotyping of 18 SNPs, no SNP was excluded and the logistic regression analyses adjusted for sex and age were applied. We identified 4 SNPs—rs1265883 in SLAMF6, rs1024161 in CTLA4, rs1521 in HLA-B region, and rs5912838 in GPR174/ ITM2A at X chromosome—that were significantly associated with HT at the level of genome-wide significant associations (5 × 10^–8) (Table 2). Among the 4 susceptibility loci for HT, the most significant association was detected at rs1521 in HLA-B locus (P = 7.94 × 10^–20, OR = 1.55; 95% CI, 1.41–1.71) (Table 2). MHC region contained many immune-related genes and has been confirmed to be associated with AITD as well as with many other autoimmune diseases. Previous studies reported HLA-DR3 as AITD susceptible loci in MHC, which have been proved to be associated with both HT and GD. In the current study, we found that rs1521 in HLA-B at 6p21 was significantly associated with HT (Table 2), which was also strongly associated with GD in the Chinese Han population (7). However, the association of rs1521 in HLA-B at 6p21 with HT was weaker than that with GD (Table 2). Another locus we found to be strongly associated with HT was rs1265883 in SLAMF6 at 1q23.2 (P = 3.67 × 10^–16, OR = 1.59; 95% CI, 1.42–1.78) (Table 2), which was never reported in previous studies. The 1q23.2 chromosomal region, which encodes for the SLAMF receptors, has also been identified as a susceptibility locus for systemic lupus erythematosus (SLE), rheumatoid arthritis, neuropsychiatric SLE, and lupus nephritis (15). Another new risk locus for HT, rs5912838, was identified in this study; it is located between the immune receptor G protein-coupled receptor 174 gene (GPR174) and integral membrane protein 2A gene (ITM2A) at Xq21.1 (P = 1.58 × 10^–10, OR = 1.26; 95% CI, 1.17–1.35) (Table 2). ITM2A was initially identified as a novel risk gene for GD in our previous GWAS study (7), but now, for the first time, it is proved to be associated with HT. ITM2A plays critical roles in the development of T-cell development, differentiation, and activation, indicating ITM2A is involved in the pathology of thyroid autoimmunity (16). However, the association of ITM2A with other autoimmune diseases has not been reported. A solid evidence for the association of rs1024161 in CTLA4 at 2q33 with HT was provided in the current study (P = 2.31 × 10^–11, OR = 1.31; 95% CI, 1.21–1.42) (Table 2). rs231775 in CTLA4 has been shown to confer risk to HT in a meta-analysis in both East Asian and white individuals and a small-dataset study in people in the UK (17, 18). In addition, evidence is accumulating that the CTLA-4 was associated with other autoimmune diseases, such as Addison’s disease, insulin-dependent diabetes mellitus, systemic lupus erythematosus, and systemic sclerosis.

We also found 7 SNPs in distinct chromosome regions associated with HT at the level of Bonferroni-corrected P value (P < 2.78 × 10^–3). Among them, rs1881145 in TRIB2 at 2p24.3 (P = 2.00 × 10^–6, OR = 1.19; 95% CI, 1.11–1.28), rs13093110 in LPP at 3q28 (P = 5.89 × 10^–4, OR = 1.14), rs2474619 in BACH2 at 6q15 (P = 2.75 × 10^–3, OR = 1.12; 95% CI, 1.04–1.21), and rs12575636 in intergenic of FAM76B/SESN3 at 11q21 (P = 3.59 × 10^–5, OR = 1.28; 95% CI, 1.14–1.45) or their closely linked SNPs have been reported to be associated with HT in 462 HT patients and 9364 controls in the UK population (10). However, the other 3 SNPs, rs6832151 in intergenic of RHOH/CHRNA9 at 4p14 (P = 1.06 × 10^–4, OR = 1.16; 95% CI, 1.07–1.24), rs9355610 in RNASET2 at 6q27 (P = 1.34 × 10^–4, OR = 1.15; 95% CI, 1.07–1.23), and rs229527 in C1QTNF6 at 22q12.3 (P = 3.00 × 10^–6, OR = 1.22; 95% CI, 1.12–1.32) were not reported to be associated with autoimmune diseases other than GD.

Notably, the association of these 2 SNPs, rs7528684 in FCRL3 at 1q23.1 (P = 2.98 × 10^–3, OR = 1.11; 95% CI, 1.04–1.20) and rs505922 in ABO at 9q34.2 (P = 4.07 × 10^–3, OR = 1.11; 95% CI, 1.03–1.19), with HT did not reach the level of Bonferroni-corrected P value but was close to the Bonferroni-corrected level (Table 2), which indicated that the 3 SNPs were associated with HT at the level of suggestive significance. In fact, BACH2 and FCRL3 have been reported to be associated with several autoimmune diseases, such as BACH2 with multiple sclerosis, Crohn’s disease, type 1 diabetes, vitiligo, and celiac disease (19), while FCRL3 was associated with rheumatoid arthritis, systemic lupus erythematosus, and thyroid-peroxidase autoantibody levels in HT patients (20).

Interestingly, 5 GD susceptibility SNPs, including rs2281388 in HLA class II (HLA–DPB1), rs1883832 in CD40, rs12101261 in TSHR, rs2294025 in TG, and rs1456988 in intergenic of C14orf177/VRK1, were proved to have no association with HT in the current sample size (P > 0.01), and more than 18 000 case-control pairs were needed to detect the genome-wide significant association of the 5 SNPs with HT, according to the results calculated by Quanto software version 1.2.4 (University of Southern California, Los Angeles, CA) (Table 2).

To evaluate the reliability of the association about these 13 SNPs with significant association signals with HT, we further analyzed the false-positive report probability (FPRP) of these SNPs, except for rs1521 since its strong significance is not applicable for the calculating method. The FPRP value was calculated with the reported P values and ORs under assigned probabilities ranging from 0.25 to 0.00001, and the statistical power was 99–100% to detect SNPs at levels equal to their reported P values (Table 3). The FPRP values of the 4 SNPs with an association signal at the level of genome-wide association significance remained below the level of 2 × 10^–5, even at the prior probability of 0.00001 (Table 3). The FPRP values for rs1881145 in TRIB2 and rs229527 in C1QTNF6 remained below the level of 0.03, even at the very low prior probability of 0.0001, which proved the high credibility of reported P values and ORs. The FPRP values of rs12575636 in intergenic of FAM76B/SESN3, rs6832151 in intergenic of RHOH/CHRNA9, and rs9355610 in RNASET2 were less than 0.01 from the prior probability of 0.25 to 0.01, which was a relatively high prior probability range (Table 3). The FPRP values for rs13093110 in LPP and rs2474619 in BACH2 were more than 0.05 at the prior probability of 0.01(Table 3).

Heterogeneity analysis for 18 GD susceptibility SNPs between HT and GD

To further clarify the heterogeneity of susceptibility genes between HT and GD, as well as to find specific susceptibility loci for GD rather than HT, the frequency difference of these 18 GD-risk SNPs was analyzed by logistic regression analysis adjusted for age and sex in 2682 HT patients and 4980 GD patients. As a result, the association of 9 SNPs with disease states showed significant difference between GD and HT patients. Out of them, the association difference of 6 SNPs reached the level of the Bonferroni-corrected P value (P_{heterogeneity} < 2.78 × 10^–3), including 2 HLA loci (HLA-B and HLA-DPB1), 2 thyroid-specific genes (TSHR and TG), and immune-related genes CD40 and SLAMF6 (Table 2). All the associations of the 6 SNPs with GD were stronger than that with HT, except for rs1265883 in SLAMF6, as indicated by the higher risk allele frequency in GD compared with HT (Table 2). Notably, HLA-DPB1, TG, TSHR, and CD40, which were not associated with HT, were specific susceptible genes for GD (Table 2). Interestingly, the association of SLAMF6 with HT was stronger than that with GD, as indicated by the higher risk allele frequency of rs1265883 in HT (P_{heterogeneity} = 1.00 × 10^–3) (Table 2).

Discussion

In the current study, an association study of 2682 Chinese Han patients with HT and 3892 controls was performed to identify risk loci for HT from 18 previous, well-confirmed GD-susceptibility SNPs. We found 4 susceptibility loci for HT at the level of genome-wide significant associations: rs1265883 in SLAMF6, rs1024161 in CTLA4, rs1521 in HLA-B, and rs5912838 in GPR174/ ITM2A at X chromosome (P < 5 × 10^–8). We found 7 susceptibility loci for HT at the level of Bonferroni-corrected P value: rs1881145 in TRIB2 at 2p24.3, rs13093110 in LPP at 3q28, rs2474619 in BACH2 at 6q15, rs12575636 in intergenic of FAM76B/SESN3 at 11q21, rs6832151 in intergenic of RHOH/CHRNA9 at 4p14, rs9355610 in RNASET2 at 6q27, and rs229527 in C1QTNF6 at 22q12.3 (P < 2.78 × 10^–3). Out of the 11 susceptibility loci for HT, 5 loci were reported to be associated with HT for the first time: rs1265883 in SLAMF6, rs5912838 in GPR174/ ITM2A on the X chromosome, rs6832151 in intergenic of RHOH/CHRNA9 at 4p14, rs9355610 in RNASET2 at 6q27, and rs229527 in C1QTNF6 at 22q12.3. Recently, a GWAS meta-analysis study reported that rs11675434 in TPO, -rs653178 in ATXN2, and -rs10944479 in BACH2 were associated with TPOAb-positivity. While the MAGI3 variant was also associated with an increased risk of hypothyroidism (21), the association of BATH2 with HT was confirmed in this current study, while the other risk genes were not investigated. Interestingly, we conducted a heterogeneity analysis for these 18 GD-susceptibility SNPs in 2682 HT patients and 4980 GD patients and found 8 loci with significant heterogeneity between the 2 kinds of AITDs. Out of them, the association difference of 6 SNPs with GD and HT reached the level of Bonferroni-corrected P value, including 2 HLA loci (HLA-B and HLA-DPB1), 2 thyroid-specific genes (TSHR and TG), and 2 immune-related genes (CD40 and SLAMF6), which suggested that GD and HT were probably distinct entities with differences in pathogenic mechanisms.

It is unexpected to find that immune-related genes rs2281388 in HLA-DPB1 and rs1883832 in CD40 were associated with GD (P_{heterogeneity} = 5.34 × 10^–58 and 6.57 × 10^–7, respectively) (Table 2) but not HT, and the association of rs1521 in HLA-B with GD was stronger than that with HT (P_{heterogeneity} = 3.78 × 10^–8), while the association of SLAMF6 with HT was reversed (P_{heterogeneity} = 1.00 × 10^–3) (Table 2). However, we cannot exclude that other SNPs in the HLA class II region may be associated with HT, because there was a lot of linkage disequilibrium (LD) in the HLA region. Further research is needed to reveal the associations of the HLA region with HT. The interaction of the CD40 ligand (CD154) on T-cells with CD40 on antigen-presenting cells (APCs) may upregulate the expression of costimulatory molecules on all APCs, including B-cells, macrophages, and, most importantly, dendritic cells (DC), which are involved in the initiation of the immune response (22). It was reported that the CC genotype of rs1883832 induced a 15–32% increase in CD40 protein expression (23). Moreover, thyroidal CD40 overexpression augmented the production of thyroid-specific Abs, resulting in more severe experimental autoimmune Graves’ disease, whereas deletion of thyroidal CD40 suppressed the disease, which is mediated by activating downstream cytokines and chemokines of the CD40–CD154 pathway, such as IL-6 (24). This suggests that CD40 may play an essential role in GD rather than in HT.

Another interesting finding in the present study was that there was no association of thyroid-specific genes TSHR and TG with HT (P_{heterogeneity} = 3.70 × 10^–14 and 5.29 × 10^–4, respectively). The SNPs in TSHR associated with GD have been refined to within a 40 Kb region of TSHR intron 1 (25), with SNPs rs12101261, rs179247, and rs12101255 showing the strongest evidence of association with GD (25). Furthermore, absolute levels of 2 truncated mRNA transcripts, previously termed ST4 and ST5, were shown to be significantly increased compared to levels of full-length TSHR mRNA in thyroids carrying the GD susceptibility genotypes of rs179247 and rs12101255 (25). The truncated mRNA transcripts ST4 and ST5 encode the majority of soluble A subunit directly. This increased expression by the GD susceptibility allele may trigger the autoimmune response and the production of TSHR autoantibodies (25). Most recently, the rs12101261 site has been found as a regulatory element that acts through the binding of the transcriptional repressor promyelocytic leukemia zinc finger protein (PLZF) (26). The disease-associated variant in rs12101261 interacted with PLZF and reduced TSHR expression in thymic, leading to an escape from autoimmune tolerance to the TSHR (26). The fact that the immune-related genes HLA-DPB1 and thyroid-specific gene TSHR were associated with GD rather than HT could partially explain why the TSHR autoantibodies (TRAb) appear in most of the GD patients but were rarely detected in HT patients, and our previous study has found that the combination of the susceptible alleles of TSHR and HLA-DPB1 might determine the persistence of TRAb in GD patients (6). Rs2294025 in TG has been confirmed to be associated with GD by our GWAS study (6). The susceptibility allele of rs2294025 was correlated with the expression of the non–E46 TG isoform rather than with the E46 TG isoform (6). Interestingly, previous studies reported that a interaction between an amino acid variant, W1999R in TG and HLA-DRB1-Arg74, conferred a very high-odds ratio for GD (27). Using recombinant HLA-DRB1-Arg74 protein, 1 study reported 4 Tg peptides bound specifically to the HLA-DRB1-Arg74 pocket but not to the Gln-74 pocket (28). Unexpectedly, TG was not associated with HT in the current study, suggesting that elevated TgAb levels in HT may be caused by different mechanisms that are independent of the abnormity of TG, as discussed later.

Our findings suggest that the pathogenesis of HT did not require the abnormity of 2 immune-related genes (HLA-DPB1 and CD40) and the 2 thyroid-specific genes (TSHR and TG). In fact, different from GD, thyrocytes were destructed in HT, which was mediated by either the mechanism of apoptosis or as a result of the activity of T-cells. Interleukin-1β (IL-1β), abundantly produced by infiltrated lymphocytes in HT glands, induced the aberrant Fas expression in normal thyrocytes. Fas will bind to FasL, which is constitutively expressed in normal and HT thyrocytes or infiltrated lymphocytes, and results in massive thyrocyte apoptosis. Moreover, previous studies have shown that molecules such as fragmented genomic DNA released from dying or dead cells promoted the development of a sterile inflammatory milieu and activated the immune response through a damage-associated molecular patterns (DAMPs) pathway (29).Therefore, we hypothesized that the amount of thyrocyte destruction in HT, which released a high concentration of fragment genomic DNA and thyroid-specific autoantigens, such as TG and TPO, activated the immune response, leading to the production of autoantibodies, such as TPOAb and TGAb. The hypothesis may help explain the fact that positive TSHR antibody (TRAb) is rare in HT because the expression of TG and TPO are much higher than TSHR in thyrocytes.

Our study has several limitations. For example, only 1 GD-risk SNP was investigated in each susceptibility region, but in some cases the association can be driven by different SNPs of the region in different diseases or even different ethnicities, especially for MHC regions. Moreover, a larger sample size may be required to prove significant associations with HT for those weakly associated SNPs in the current study. Besides, more work will be needed in the future to help explain why TG and CD40 is not associated with HT and to help find HT-specific risk genes that contribute to HT pathogenesis.

In summary, we identified 11 HT susceptibility loci from established GD susceptibility loci in the Chinese Han population, with 4 loci reaching the genome-wide significance of 5 × 10^–8, and 5 loci were reported to be associated with HT for the first time. We also found 2 immune-related genes and 2 thyroid-specific genes that were associated with GD but not with HT. Our findings suggest that the pathogenesis of HT and GD was different.

Abbreviations

Abbreviations
 
• APC

  antigen presenting cells

 
• FPRP

  false positive report probability

 
• GD

  Graves’ disease

 
• HT

  Hashimoto’s thyroiditis

 
• Tg

  Thyroglobulin

 
• TRAb

  TSHR antibody

 
• TSHR

  thyroid-stimulating hormone receptor

Acknowledgments

We thank all members of the core laboratory in the medical center of clinical research of the Shanghai Ninth People’s Hospital and the China Consortium for the Genetics of Autoimmune Thyroid Disease for their valuable support and discussion.

Financial Support: The work was supported in part by the National Natural Science Foundation of China (81430019, 81870540, 81870537, 81800749, 81770786, 31571296, 31501015, 81661168016), National Key R&D Program of China (2017YFC1001801), Shanghai Municipal Education Commission-Gaofeng Clinical Medicine Grant Support (20161318).

Author Contributions: H.D.S. conceived and designed the project. Q.Y.Z., W.L., C.L.Z., and W.H.D. contributed to the project management and genotyping. Q.Y.Z. and L.L. did the statistical analysis. J.L., G.Q.G., S.X.Z., X.P.Y., Z.Z., F.F.Y., Y.R.M., S.S.Y., H.J.X., F.S., C.R.Z., Y.X.Y., and G.Y..Y took part in the collection of clinical samples, DNA extraction, and sample quality control. H.D.S., J.L., G.Q.G., and S.X.Z. interpreted and analyzed the data. Q.Y.Z. and H.D.S. wrote and revised the paper. All authors read and approved the final manuscript.

Ethics Approval and Consent to Participate: This study was approved by the local ethics committee of Shanghai Ninth People’s Hospital, Linyi People’s Hospital, the Hospital Affiliated to Jiangsu University, and the Central Hospital of Xuzhou. All subjects in this study provided written informed consent using protocols approved by the local ethics committee.

Additional Information

Disclosure: The authors have nothing to disclose.

Data Availability: The datasets generated during and/or analyzed during the current study are not publicly available but are available from the corresponding author on reasonable request.

References

Author notes