Thyroid-stimulating hormone (TSH) is a sensitive indicator of thyroid function. High and low TSH levels reflect hypothyroidism and hyperthyroidism, respectively. Even within the normal range, small differences in TSH levels, on the order of 0.5–1.0 mU/l, are associated with significant differences in blood pressure, BMI, dyslipidemia, risk of atrial fibrillation and atherosclerosis. Most of the variance in TSH levels is thought to be genetically influenced. We conducted a genome-wide association study of TSH levels in 1346 Chinese Han individuals. In the replication study, we genotyped four candidate SNPs with the top association signals in an independent isolated Chinese She cohort (n = 3235). We identified a novel serum TSH susceptibility locus within XKR4 at 8q12.1 (rs2622590, Pcombined = 2.21 × 10−10), and we confirmed two previously reported TSH susceptibility loci near FOXE1 at 9q22.33 and near CAPZB at 1p36.13, respectively. The rs2622590_T allele at XKR4 and the rs925489_C allele near FOXE1 were correlated with low TSH levels and were found to be nominally associated to patients with papillary thyroid carcinoma (PTC) (OR = 1.41, P= 0.014 for rs2622590_T, and OR = 1.61, P= 0.030 for rs925489_C). The rs2622590 and rs925489 genotypes were also correlated with the expression levels of FOXE1 and XKR4, respectively, in PTC tissues (P = 2.41 × 10−4 and P= 0.02). Our findings suggest that the SNPs in XKR4 and near FOXE1 are involved in the regulation of TSH levels.

INTRODUCTION

Impinging on virtually every tissue of the body, thyroid hormones affect a variety of metabolic and developmental processes in humans (1). They are vital to brain development and skeletal maturation in fetuses and infants, and they influence protein synthesis, fat and carbohydrate metabolism, bone deposition, metabolic rate, temperature regulation and blood pressure in adults (24).

Thyroid function is regulated by a homeostatic negative-feedback loop involving the pituitary–thyroid axis and is evaluated by measuring circulating concentrations of thyroid-stimulating hormone (TSH), free thyroxine (FT4), and free triiodothyronine (FT3) (5). Secreted by the pituitary gland, TSH interacts with the TSH receptor on thyroid cells to control the release of T4 and T3 through a series of signal transduction pathways (6). TSH and FT4 have an inverse, log-linear relationship such that small changes in FT4 levels result in dramatic changes in TSH secretion. Hence, the serum TSH concentration is a very sensitive indicator of the function of the thyroid gland (7).

The variability of serum TSH concentration is more dramatic among healthy individuals than within healthy individuals (measured multiple times in the same individual) (8), suggesting that the regulation of the thyroid-hormone axis is exact and varies among individuals. Evidence from studies of healthy Danish twins has shown that genetic factors can account for 64% (57–70%) of the variance in circulating TSH levels (9). Another heritability study of Mexican Americans showed that non-genetic covariates only accounted for 2–18% of the total phenotypic variation, whereas genetic factors accounted for 26–64% of the total phenotypic variation (10).

Previous studies have identified many putative genetic susceptibility variants for serum TSH levels, but unequivocal replication has been limited to a few loci (such as PDE8B at 5q13.3, CAPZB at 1p36.13, MAF at 16p23.2 and NR3C2 at 4q31.23) (1117). Moreover, more recent GWAS studies in large samples recruited from Caucasian by Gudmundssonn et al. (15) and Porcu et al. (17), identified 22 novel susceptibility loci for TSH level. All the 26 TSH risk loci only explained about 4.3–5.6% of the inter-individual variability in serum TSH levels, suggesting that as-yet-unknown variants may also be important.

To identify novel serum TSH susceptibility loci, we conducted a GWAS of 1346 healthy Chinese Han individuals and an independent replication study of 3235 healthy Chinese She individuals. We identified a new susceptibility locus (rs2622590, near the XKR4 region at 8q12.1; Pcombined = 2.21 × 10−10) and confirmed two previously reported loci (near FOXE1 at 9q22.33 and near CAPZB at 1p36.13).

RESULTS

Genome-wide association study of serum TSH levels

The clinical and demographic characteristics of the study participants are shown in Table 1. After stringent QC filters, a total of 483 947 genotyped SNPs and 8 019 905 imputed SNPs were analyzed for associations with serum TSH levels among 1346 individuals whose serum TSH levels ranged from 0.35 and 4.94 mU/l (Fig. 1). The results for the typed and imputed SNPs (P < 10−2) (http://www.dropbox.com/s/q32p5yrd1accd3a/Data%20for%20TSH.zip), organized by chromosome location in a Manhattan plot, are shown in Figure 2. A quantile–quantile plot of TSH levels is shown in the Supplementary Material (Supplementary Material, Fig. S1). The estimated inflation factor was modest (λ = 1.004), and thus the distribution of P-values for the association tests shows no evidence of systematic bias. A principal component analysis (PCA) and a multidimensional scaling (MDS) analysis, both described in our previous study (18), showed that all of the subjects clustered around Chinese and Japanese lines of descent.

Table 1.

Characteristics of samples used in the GWAS and replication analyses

Study population n Female, n (%) Age (years) (Mean ± SD) BMI (kg/m2) (Mean ± SD) TSH (mU/l) (Mean ± SD) Log10 TSH (Mean ± SD) 
GWAS 
 Xuzhou 713 545 (76.44) 47.71 ± 7.70 22.27 ± 2.19 1.81 ± 0.84 0.21 ± 0.20 
 Linyi 633 482 (76.15) 50.33 ± 9.90 NA 1.50 ± 0.73 0.13 ± 0.21 
Replication 
 Ningde 3235 1,767 (54.62) 50.89 ± 12.70 22.74 ± 3.11 1.61 ± 0.84 0.15 ± 0.23 
Study population n Female, n (%) Age (years) (Mean ± SD) BMI (kg/m2) (Mean ± SD) TSH (mU/l) (Mean ± SD) Log10 TSH (Mean ± SD) 
GWAS 
 Xuzhou 713 545 (76.44) 47.71 ± 7.70 22.27 ± 2.19 1.81 ± 0.84 0.21 ± 0.20 
 Linyi 633 482 (76.15) 50.33 ± 9.90 NA 1.50 ± 0.73 0.13 ± 0.21 
Replication 
 Ningde 3235 1,767 (54.62) 50.89 ± 12.70 22.74 ± 3.11 1.61 ± 0.84 0.15 ± 0.23 

Values are means and standard deviations (SD) unless otherwise specified.

BMI, body mass index; TSH, thyroid-stimulating hormone; NA, not assessed, SD, standard deviation.

Figure 1.

Flowchart for quality filtering in the two-stage GWAS for serum TSH levels. In the initial GWAS stage, quality filtering was performed on SNPs and samples before analysis to ensure robust association tests. Of the 655 214 markers assayed, the Y-chromosome and mitochondrial SNPs, CNV-related markers, and Illumina controls were excluded (n = 3185), leaving 652 029 SNPs for further analysis. We further excluded SNPs with a low call rate (< 98%), an MAF < 1% within the population, or a significant deviation from Hardy-Weinberg equilibrium (P ≤ 10−6). Samples were excluded if they had a high missing call rate (<98%), gender inconsistencies, cryptic relatedness, abnormal TSH (TSH ≥4.94 or ≤0.35 mU/l) or high TPO-Ab (TPO-Ab ≥5.61 U/ml). In the replication study, three SNPs associated with TSH (P < 10−6) were selected as candidate loci for replication. Additionally, one SNP that was previously reported to be associated with TSH was also chosen for genotyping in the replication samples.

Figure 1.

Flowchart for quality filtering in the two-stage GWAS for serum TSH levels. In the initial GWAS stage, quality filtering was performed on SNPs and samples before analysis to ensure robust association tests. Of the 655 214 markers assayed, the Y-chromosome and mitochondrial SNPs, CNV-related markers, and Illumina controls were excluded (n = 3185), leaving 652 029 SNPs for further analysis. We further excluded SNPs with a low call rate (< 98%), an MAF < 1% within the population, or a significant deviation from Hardy-Weinberg equilibrium (P ≤ 10−6). Samples were excluded if they had a high missing call rate (<98%), gender inconsistencies, cryptic relatedness, abnormal TSH (TSH ≥4.94 or ≤0.35 mU/l) or high TPO-Ab (TPO-Ab ≥5.61 U/ml). In the replication study, three SNPs associated with TSH (P < 10−6) were selected as candidate loci for replication. Additionally, one SNP that was previously reported to be associated with TSH was also chosen for genotyping in the replication samples.

Figure 2.

Manhattan plots of the SNPs associated with serum TSH levels in the GWAS data. Manhattan plots of associations between SNPs and TSH levels. Typed and imputed SNPs (P < 10−2) are represented on the X-axis, organized by chromosome. On the Y-axis, statistical significance is expressed as -log10P values. The red horizontal line represents the discovery cohorts (1346 subjects) threshold P-value of 10−6.

Figure 2.

Manhattan plots of the SNPs associated with serum TSH levels in the GWAS data. Manhattan plots of associations between SNPs and TSH levels. Typed and imputed SNPs (P < 10−2) are represented on the X-axis, organized by chromosome. On the Y-axis, statistical significance is expressed as -log10P values. The red horizontal line represents the discovery cohorts (1346 subjects) threshold P-value of 10−6.

In our initial GWAS, 10 genotyped SNPs located in five different chromosomal regions had P-values <1 × 10−5 (Table 2). From those 10 SNPs, we identified the four SNPs, located in three different chromosomal regions, with the strongest (P < 10−6) serum TSH associations: rs1348271 at 11q22.1 (P = 1.09 × 10−7), rs925489 and rs7850258 at 9q22.3 (P = 3.19 × 10−7 and P = 3.19 × 10−7, respectively), and rs2622590 at 8p12.1 (P = 6.49 × 10−7). The minor allele frequencies (MAFs) for the four SNPs ranged from 0.05 to 0.46.

Table 2.

Most significant associations for TSH levels from the GWAS cohorts

Chr. SNP Position A1/A2 MAF βb t P-valuea Annotated gene Distance 
1p36.13 rs6683419 19827780 G/A 0.257 0.039 4.371 1.33 × 10−5 CAPZB 15.7 kb 
1q44 rs12033048 246429681 T/C 0.436 −0.034 −4.347 1.49 × 10−5 SMYD3 Intron 
2p14 rs11681944 68322579 T/C 0.437 −0.035 −4.451 9.24 × 10−6 C1D 32.4 kb 
7q21.12 rs43123 87876090 G/A 0.171 0.051 4.412 1.11 × 10−5 SRI 19.8 kb 
8q12.1 rs2622590 56358274 A/G 0.406 −0.039 −4.999 6.54 × 10−7 XKR4 Intron 
8q12.1 rs2929029 56387105 G/A 0.247 −0.037 −4.823 1.58 × 10−6 XKR4 Intron 
9q22.33 rs925489 100546600 C/T 0.461 −0.074 −5.138 3.19 × 10−7 FOXE1 68.9 kb 
9q22.33 rs7850258 100549013 G/A 0.125 −0.074 −5.138 3.19 × 10−7 FOXE1 66.5 kb 
9q22.33 rs965513 100556109 C/T 0.077 −0.070 −4.868 1.26 × 10−6 FOXE1 59.4 kb 
9q22.33 rs10759944 100556972 A/G 0.077 −0.070 −4.868 1.26 × 10−6 FOXE1 58.6 kb 
9q22.33 rs7870926 100796945 A/G 0.076 0.041 4.497 7.49 × 10−6 ANP32B 18.7 kb 
11q22.1 rs11223396 100118859 A/G 0.076 −0.074 −4.366 1.36 × 10−5 CNTN5 Intron 
11q22.1 rs1348271 100137346 T/C 0.053 −0.087 −5.014 6.04 × 10−7 CNTN5 Intron 
16q21 rs2582597 60125785 G/T 0.051 −0.046 −4.565 5.44 × 10−6 NA Intergenic 
16q23.2 rs11866972 79738425 A/G 0.233 0.034 4.423 1.05 × 10−5 MAF 103.8 kb 
17q25.1 rs4789661 72425159 T/C 0.397 −0.039 −4.411 1.11 × 10−5 GPR142 2.5 kb 
Chr. SNP Position A1/A2 MAF βb t P-valuea Annotated gene Distance 
1p36.13 rs6683419 19827780 G/A 0.257 0.039 4.371 1.33 × 10−5 CAPZB 15.7 kb 
1q44 rs12033048 246429681 T/C 0.436 −0.034 −4.347 1.49 × 10−5 SMYD3 Intron 
2p14 rs11681944 68322579 T/C 0.437 −0.035 −4.451 9.24 × 10−6 C1D 32.4 kb 
7q21.12 rs43123 87876090 G/A 0.171 0.051 4.412 1.11 × 10−5 SRI 19.8 kb 
8q12.1 rs2622590 56358274 A/G 0.406 −0.039 −4.999 6.54 × 10−7 XKR4 Intron 
8q12.1 rs2929029 56387105 G/A 0.247 −0.037 −4.823 1.58 × 10−6 XKR4 Intron 
9q22.33 rs925489 100546600 C/T 0.461 −0.074 −5.138 3.19 × 10−7 FOXE1 68.9 kb 
9q22.33 rs7850258 100549013 G/A 0.125 −0.074 −5.138 3.19 × 10−7 FOXE1 66.5 kb 
9q22.33 rs965513 100556109 C/T 0.077 −0.070 −4.868 1.26 × 10−6 FOXE1 59.4 kb 
9q22.33 rs10759944 100556972 A/G 0.077 −0.070 −4.868 1.26 × 10−6 FOXE1 58.6 kb 
9q22.33 rs7870926 100796945 A/G 0.076 0.041 4.497 7.49 × 10−6 ANP32B 18.7 kb 
11q22.1 rs11223396 100118859 A/G 0.076 −0.074 −4.366 1.36 × 10−5 CNTN5 Intron 
11q22.1 rs1348271 100137346 T/C 0.053 −0.087 −5.014 6.04 × 10−7 CNTN5 Intron 
16q21 rs2582597 60125785 G/T 0.051 −0.046 −4.565 5.44 × 10−6 NA Intergenic 
16q23.2 rs11866972 79738425 A/G 0.233 0.034 4.423 1.05 × 10−5 MAF 103.8 kb 
17q25.1 rs4789661 72425159 T/C 0.397 −0.039 −4.411 1.11 × 10−5 GPR142 2.5 kb 

SNP, single-nucleotide polymorphism; Chr., chromosome; A1/A2, Min-allele/Max-allele; MAF, minor allele frequency; NA, not assessed.

aP-values are from linear regression models adjusted for age, gender, and geographic region.

bβ-values estimate the difference in TSH level and are obtained from linear regression models in PLINK.

Analyses of replication and combined data

Given that the sample size of the initial GWAS data set in the current study is underpowered to detect common variants associated with TSH levels, the three SNPs (rs1348271, rs925489 and rs2622590) with the strongest serum TSH associations, and one SNP (rs6683419, PGWAS = 1.33 × 10−5) in 1p36.13 region previously reported to be associated with serum TSH levels (13), were selected and genotyped in an independent sample of the Southern Chinese She population, who is one of the important ethnic minorities, amounts to 0.7 million people in China (in the 2010 census). They mainly work in agriculture, forestry, animal husbandry, fishing and water industry, and have their own living customs. The marriage between Han and She were prohibited 60 years ago. The replication sample was recruited from the She population residing in Ningde city of Fujian Province and consisted of 3235 individuals with serum TSH levels between 0.35 and 4.94 mU/l (19). Three of the four selected SNPs (rs925489, P = 9.13 × 10−9; rs2622590, P = 5.62 × 10−6; and rs6683419, P = 5.86 × 10−4) showed evidence of replication in the tested data set (Table 3).

Table 3.

Association results for the selected SNPs in the GWAS and replication phases

SNP Chr. Position A1/A2 Nearest gene GWAS (n = 1346)
 
Replication (n = 3235)
 
Combined (n = 4581)
 
MAF βa P-valueb n MAF βa P-valueb n MAF βa P-valueb 
Novel loci 
 rs2622590 8q12.1 56358274 A/G XKR4 0.436 −0.039 6.49 × 10−7 3,077 0.449 −0.026 5.62 × 10−6 4,423 0.445 −0.030 2.21 × 10−10 
 rs1348271 11q22.1 100137346 T/C CNTN5 0.051 −0.092 1.09 × 10−7 3,120 0.027 −0.012 4.89 × 10−1 4,466 0.034 −0.046 4.14 × 10−4 
Known loci 
 rs6683419 1p36.13 19827780 G/A CAPZB 0.257 0.039 1.33 × 10−5 3,159 0.268 0.022 5.86 × 10−4 4,505 0.265 0.027 2.90 × 10−7 
 rs925489 9q22.33 100546600 C/T FOXE1 0.077 −0.074 3.19 × 10−7 3,027 0.122 −0.053 9.13 × 10−9 4,373 0.109 −0.058 1.02 × 10−13 
SNP Chr. Position A1/A2 Nearest gene GWAS (n = 1346)
 
Replication (n = 3235)
 
Combined (n = 4581)
 
MAF βa P-valueb n MAF βa P-valueb n MAF βa P-valueb 
Novel loci 
 rs2622590 8q12.1 56358274 A/G XKR4 0.436 −0.039 6.49 × 10−7 3,077 0.449 −0.026 5.62 × 10−6 4,423 0.445 −0.030 2.21 × 10−10 
 rs1348271 11q22.1 100137346 T/C CNTN5 0.051 −0.092 1.09 × 10−7 3,120 0.027 −0.012 4.89 × 10−1 4,466 0.034 −0.046 4.14 × 10−4 
Known loci 
 rs6683419 1p36.13 19827780 G/A CAPZB 0.257 0.039 1.33 × 10−5 3,159 0.268 0.022 5.86 × 10−4 4,505 0.265 0.027 2.90 × 10−7 
 rs925489 9q22.33 100546600 C/T FOXE1 0.077 −0.074 3.19 × 10−7 3,027 0.122 −0.053 9.13 × 10−9 4,373 0.109 −0.058 1.02 × 10−13 

SNP, single-nucleotide polymorphism; Chr, chromosome; A1/A2, Min-allele/Max-allele; MAF, minor allele frequency.

aβ-values estimate the difference in TSH level and are obtained from linear regression models in PLINK.

bP-values are from linear regression models adjusted for age, gender, geographic region and race.

In the combined samples, two SNPs (rs925489 at 9q22.33, P = 1.02 × 10−13; and rs2622590 at 8q12.1, P = 2.21 × 10−10) showed unequivocal evidence of association with serum TSH levels, with a genome-wide significance threshold of P = 5 × 10−8; and one more SNP (rs6683419 at 1p36.13) nearly met the genome-wide significance level for association with serum TSH levels (P = 2.90 × 10−7) (Table 3). These three SNPs included one new serum TSH susceptibility locus (the XKR4 region at 8q12.1) and two previously reported TSH susceptibility loci (Table 3).

Notably, we identified a novel serum TSH susceptibility locus, rs2622590 (P = 2.21 × 10−10) in the 8p12.1 region. In the initial GWAS, we analyzed 2966 imputed SNPs and 164 genotyped SNPs in a ∼900 kb linkage disequilibrium (LD) block containing XKR4 and SBF1P1, TMEM68, TGS1 and LYN located at 8p12.1 (Fig. 3A). We found that all of the SNPs with P < 10−5 within the LD block were located within a 35 kb region of intron 2 of XKR4 (Fig. 3A; Supplementary Material, Tables S1 and S2). After conditioning on rs2622590 (PGWAS = 6.49 × 10−7; Table 3), none of the other SNPs at 8p12.1 remained significantly (P < 0.01) associated with serum TSH levels, suggesting that rs2622590 might be an independent SNP associated with serum TSH levels. Therefore, we selected rs2622590 for further genotyping in the replication sample. The replication study confirmed the serum TSH association of rs2622590 (Preplication = 5.62 × 10−6; Table 3), and the association reached the genome-wide significance level in the combined analysis (Pcombined = 2.21 × 10−10; Table 3). After adjustment for the relevant covariates, the serum TSH levels among individuals with the rs2622590 CC genotype were higher than those among individuals with the homozygous-susceptible TT genotype in the GWAS cohort (1.79 ± 0.82 for the CC genotype, 1.52 ± 0.75 for the TT genotype; P = 6.31 × 10−6; Supplementary Material, Table S3) and in the replication cohort (1.67 ± 0.85 for the CC genotype, 1.47 ± 0.79 for the TT genotype; P = 3.44 × 10−6; Supplementary Material, Table S3). Rs2622590 is located in intron 2 of XKR4, 3.5 kb downstream of the pseudogene SBF1P1 (Fig. 3A). XKR4 is expressed abundantly in the tissues of the brain, esophagus and stomach and relatively sparsely in the tissues of the thyroid, kidney and pancreas (Fig. 4C). To further test whether the TSH associated SNPs regulate the expression of XKR4, we inspected three cis-gene expression quantitative trait loci (cis-eQTL) databases from European Caucasian population (2022) and found that out of the 327 TSH associated SNPs (P < 0.05) in a ∼900 kb region on 8q12.1 (Supplementary Material, Table S2), no SNPs were correlated with the expression of XKR4. Furthermore, a ∼40 kb region surrounding the rs2622590 on 8q12.1 was analyzed to identify regulatory elements by inspection the UCSC Genome Browser. Interestingly, a ∼5 kb fragment near to rs2622590, which contained four TSH strongly associated SNPs (rs2975987, rs2939632, rs2975986 and rs2975985), was conserved from zebrafish to human (Fig. 5). These four SNPs were in complete LD with rs2622590 (r2 > 0.97 in the 1000 Genomes Asian (ASN) samples). The ∼5 kb fragment near to rs2622590 was a DNase I hypersensitive site based on searching the ENCODE database (http://genome.ucsc.edu/ENCODE/) (23), suggesting that the fragment containing TSH associated SNPs might be a transcript regulatory element (Fig. 5). Moreover, non-synonymous SNPs in XKR4 were in complete LD with the strong associated SNPs in the TSH risk locus on 8q12.1.

Figure 3.

Regional plots of association results at 8p12.1, 9q22.33 and 1p36.13. Regional plots of association results in the GWAS samples at 8p12.1 (A), 9q22.33 (B) and 1p36.13 (C). The color of each SNP spot, ranging from red to blue, reflects the magnitude of the r2 between the SNP spot and the candidate SNP (purple circle) within each candidate locus for serum TSH levels. Genetic recombination rates, estimated using the 1000 Genomes ASN samples, are plotted in the background in cyan. The vertical axis represents the -log10P-value. Physical positions are based on NCBI build 36/hg19 of the human genome.

Figure 3.

Regional plots of association results at 8p12.1, 9q22.33 and 1p36.13. Regional plots of association results in the GWAS samples at 8p12.1 (A), 9q22.33 (B) and 1p36.13 (C). The color of each SNP spot, ranging from red to blue, reflects the magnitude of the r2 between the SNP spot and the candidate SNP (purple circle) within each candidate locus for serum TSH levels. Genetic recombination rates, estimated using the 1000 Genomes ASN samples, are plotted in the background in cyan. The vertical axis represents the -log10P-value. Physical positions are based on NCBI build 36/hg19 of the human genome.

Figure 4.

The expression analyses of the candidate genes XKR4, FOXE1 and CAPZB at 1p36.13, 8p12.1 and 9q22.33. (A) The mRNA levels of XKR4, FOXE1 and CAPZB in the tumor tissues (n = 66) and adjacent normal thyroid tissues (n = 40) of patients with PTC. Data are shown as means ± SD. The mRNA levels of XKR4, FOXE1 and CAPZB are normalized to that of GAPDH. (B) The correlation of XKR4, FOXE1 and CAPZB mRNA levels in PTC tumor tissues with the genotypes of rs2622590 in XKR4 (CC, n = 14; TC, n = 34; TT, n = 18), rs925489 near FOXE1 (TT, n = 59; CT, n = 7) and rs6683419 near CAPZB (AA, n = 28; GA, n = 30; GG, n = 8), respectively (normalized to GAPDH). (C) The expression patterns of CAPZB, XKR4 and FOXE1 in human tissues resolved by real-time PCR. *P < 0.05; **P < 0.01; ***P < 0.001. Error bars, ±SD.

Figure 4.

The expression analyses of the candidate genes XKR4, FOXE1 and CAPZB at 1p36.13, 8p12.1 and 9q22.33. (A) The mRNA levels of XKR4, FOXE1 and CAPZB in the tumor tissues (n = 66) and adjacent normal thyroid tissues (n = 40) of patients with PTC. Data are shown as means ± SD. The mRNA levels of XKR4, FOXE1 and CAPZB are normalized to that of GAPDH. (B) The correlation of XKR4, FOXE1 and CAPZB mRNA levels in PTC tumor tissues with the genotypes of rs2622590 in XKR4 (CC, n = 14; TC, n = 34; TT, n = 18), rs925489 near FOXE1 (TT, n = 59; CT, n = 7) and rs6683419 near CAPZB (AA, n = 28; GA, n = 30; GG, n = 8), respectively (normalized to GAPDH). (C) The expression patterns of CAPZB, XKR4 and FOXE1 in human tissues resolved by real-time PCR. *P < 0.05; **P < 0.01; ***P < 0.001. Error bars, ±SD.

Figure 5.

The epigenetic analysis of XKR4 at 8q12.1 from the ENCODE database. Rs2622590 shows no DNase I hypersensitivity and does not bind any transcription factors. However, a ∼5 kb fragment near to rs2622590, which contained four TSH strongly associated SNPs (rs2975987, rs2939632, rs2975986 and rs2975985), was conserved from zebrafish to human. The data from ENCODE manifested a ∼35 kb LD block in the 8q12.1 region containing rs2975987, rs2939632, rs2975986 and rs29759854 (blue SNPs) were located in a DNaseI hypersensitivity clusters, which were in complete LD with rs2622590 (r2 > 0.97 in the 1000 Genomes ASN samples). SNPs in red font are candidate SNPs in our study. SNPs in black font are in high LD with candidate SNPs (r2 > 0.8 in the 1000 Genomes ASN samples).

Figure 5.

The epigenetic analysis of XKR4 at 8q12.1 from the ENCODE database. Rs2622590 shows no DNase I hypersensitivity and does not bind any transcription factors. However, a ∼5 kb fragment near to rs2622590, which contained four TSH strongly associated SNPs (rs2975987, rs2939632, rs2975986 and rs2975985), was conserved from zebrafish to human. The data from ENCODE manifested a ∼35 kb LD block in the 8q12.1 region containing rs2975987, rs2939632, rs2975986 and rs29759854 (blue SNPs) were located in a DNaseI hypersensitivity clusters, which were in complete LD with rs2622590 (r2 > 0.97 in the 1000 Genomes ASN samples). SNPs in red font are candidate SNPs in our study. SNPs in black font are in high LD with candidate SNPs (r2 > 0.8 in the 1000 Genomes ASN samples).

Rs965513 is located at 9q22.33 and was associated with serum TSH levels and thyroid cancer in previous studies (12,15). The 9q22.33 region contains one thyroid-specific transcription factor, FOXE1 (or thyroid transcription factor 2, TTF2). In our initial GWAS, we obtained the genotypes of 2944 imputed SNPs and 186 genotyped SNPs in an ∼800 kb LD block at 9q22.33. We found that the SNPs with the strongest serum TSH associations (P < 1 × 10−6) within the ∼30 kb region were in a complete LD block (r2 > 0.8 in the 1000 Genomes ASN samples) located between XPA and FOXE1 (Fig. 3B, Supplementary Material, Table S4). In our replication study, we analyzed a previously genotyped SNP, rs925489, which was in complete LD with rs965513 (r2 = 0.97 in our Chinese Han samples). We confirmed the serum TSH association of rs925489 in the replication study, and we found that rs925489 reached the genome-wide significance level in the combined cohorts (PGWAS = 3.19 × 10−7; Preplication = 9.13 × 10−9; Pcombined = 1.02 × 10−13; Table 3). Rs925489 is located about 69 kb upstream of FOXE1, and 87 kb upstream of XPA (Fig. 3B). Notably, FOXE1 is specifically expressed in human thyroid tissues (Fig. 4C).

Rs6683419, located in the 1p36.13 region harboring CAPZB, MINOS1, NBL1 and other genes, was previously reported to be associated with serum TSH levels in European population (13,1517). Our initial GWAS provided the genotypes of 1300 imputed SNPs and 86 genotyped SNPs in a ∼400 kb LD block in the 1p36.13 region. Among all genotyped SNPs, rs6683419 had the most significant association with TSH levels (PGWAS = 1.33 × 10−5) (Supplementary Material, Table S5). We therefore analyzed rs6683419 in the replication study (Fig. 3C and Table 3). After the results of the discovery and replication studies were combined, rs6683419 showed a suggestive serum TSH association that nearly met the genome-wide significance level (Preplication = 5.86 × 10−4; Pcombined = 2.90 × 10−7; Table 3). Rs10917469, rs10799824 and rs10917477 in the 1p36.13 region, which were previously reported as the best SNPs associated with serum TSH levels in European cohort, respectively (13,1517), were further analyzed in our GWAS. Although rs10917469 and rs10799824 were only nominally associated with TSH levels (both PGWAS = 0.015), we did find rs10917477 was strongly associated with TSH levels in our Chinese Han population (PGWAS = 4.19 × 10−4) (Supplementary Material, Table S5). Moreover, we found that rs10917469 and rs10799824 were not in complete LD with rs6683419 (both r2 = 0.056 in the 1000 Genomes ASN samples), otherwise rs10917477 was strongly correlated with rs6683419 (r2 = 0.744 in the 1000 Genomes ASN samples) (Table 4).

Table 4.

The association of the top SNPs in 26 previously reported TSH risk loci with TSH in our GWAS data

Chr. Marker name Annotated Gene Tested SNP P-valuea r2b Top SNPc P-valuea 
1p36.13 rs10799824 CAPZB rs10799824 1.51 × 10−2 0.056 rs2314146 5.80 × 10−6 
rs10917469 CAPZB rs10917469 1.49 × 10−2 0.056 
rs10917477 CAPZB rs10917477 4.19 × 10−4 0.744 
1p31.3 rs334725 NFIA rs334725 1.78 × 10−2 0.005 rs334707 2.86 × 10−2 
rs334699 NFIA rs334699 2.75 × 10−2 0.004 
1p13.3 rs17020124 VAV3 rs17020124 3.41 × 10−3 0.734 rs9787296 4.66 × 10−4 
2p25.3 rs11694732 TPO rs11694732 7.10 × 10−1 0.098 rs4927648 2.74 × 10−2 
2q35 rs737308 IGFBP5 rs737308 4.30 × 10−2 0.702 rs2712172 1.25 × 10−2 
rs13015993 IGFBP5 rs13015993 3.21 × 10−2 0.727 
4q31.23 rs10028213 NR3C2 rs10028213 7.48 × 10−3 0.266 rs9968300 1.77 × 10−5 
rs10030849 NR3C2 rs10030849 7.48 × 10−3 N/A 
rs10032216 NR3C2 rs10032216 7.48 × 10−3 N/A 
5q13.3 rs4704397 PDE8B rs4704397 2.14 × 10−2 0.001 rs10055027 1.36 × 10−3 
rs6885099 PDE8B rs6885099 7.55 × 10−3 
rs2046045 PDE8B rs2046045 8.13 × 10−3 
6p21.1 rs729761 VEGFA NA NA N/A rs10948107 9.57 × 10−4 
rs9472138 VEGFA rs9472138 3.39 × 10−2 
rs6923866 VEGFA rs6923866 1.85 × 10−3 0.915 
rs11755845 VEGFA rs11755845 1.60 × 10−3 0.831 
6q24.3 rs9497965 SASH1 rs9497965 1.75 × 10−1 0.031 rs58488869 8.03 × 10−3 
6q27 rs753760 PDE10A rs753760 7.39 × 10−2 0.825 rs58885758 7.15 × 10−3 
rs3008043 PDE10A rs3008043 2.42 × 10−2 0.866 
8p12 rs7825175 NRG1 rs7825175 1.68 × 10−2 0.685 rs11991469 6.56 × 10−3 
rs2439302 NRG1 rs2439302 6.89 × 10−3 0.947 
9p24.2 rs1571583 GLIS3 rs1571583 4.17 × 10−2 0.064 rs11999791 2.62 × 10−2 
9q22.33 rs965513 FOXE1 rs965513 1.26 × 10−6 0.777 rs10818050 1.17 × 10−7 
9q34.2 rs657152 ABO rs657152 8.81 × 10−1 0.091 rs10901263 6.60 × 10−4 
10q24.2 rs7913135 NKX2-3 rs7913135 2.27 × 10−1 1.000 rs7913750 1.97 × 10−1 
11p11.2 rs17723470 PRDM11 rs17723470 2.66 × 10−2 N/A rs72907072 2.57 × 10−4 
rs7128207 PRDM11 rs7128207 2.83 × 10−1 N/A 
12q23.1 rs61938844 ELK3 NA NA N/A rs12814360 2.16 × 10−2 
14q13.3 rs1537424 MBIP rs1537424 2.48 × 10−1 0.217 rs2780310 5.64 × 10−4 
rs116909374 MBIP NA NA N/A 
14q32.12 rs34269820 ITPK1 rs34269820 3.79 × 10−2 0.211 rs1263428 5.62 × 10−3 
rs11624776 ITPK1 rs11624776 1.04 × 10−1 0.166 
14q32.33 rs73362602 SIVA1 rs73362602 5.47 × 10−1 N/A rs12896579 1.28 × 10−2 
15q21.2 rs73398284 FGF7 rs73398284 8.00 × 10−2 N/A rs74406736 5.55 × 10−3 
rs10519227 FGF7 rs10519227 1.76 × 10−1 N/A 
15q26.1 rs17776563 MIR1179 rs8030854 (r2 = 0.94) 2.49 × 10−1 0.007 rs74027996 6.89 × 10−3 
16p23.2 rs7190187 MAF rs7190187 1.18 × 10−4 0.71 rs4575545 1.91 × 10−6 
rs3813582 MAF rs3813582 1.50 × 10−5 0.955 
17q24.3 rs9915657 SOX9 rs9915657 2.54 × 10−1 0.021 rs16977126 1.20 × 10−2 
19p13.2 rs4804416 INSR rs4804416 3.74 × 10−2 0.108 rs4804433 1.47 × 10−2 
rs10420008 INSR rs10420008 6.52 × 10−2 0.332 
20p11.21 rs6082762 FOXA2 rs6082762 2.91 × 10−1 0.609 rs4499503 3.26 × 10−2 
Chr. Marker name Annotated Gene Tested SNP P-valuea r2b Top SNPc P-valuea 
1p36.13 rs10799824 CAPZB rs10799824 1.51 × 10−2 0.056 rs2314146 5.80 × 10−6 
rs10917469 CAPZB rs10917469 1.49 × 10−2 0.056 
rs10917477 CAPZB rs10917477 4.19 × 10−4 0.744 
1p31.3 rs334725 NFIA rs334725 1.78 × 10−2 0.005 rs334707 2.86 × 10−2 
rs334699 NFIA rs334699 2.75 × 10−2 0.004 
1p13.3 rs17020124 VAV3 rs17020124 3.41 × 10−3 0.734 rs9787296 4.66 × 10−4 
2p25.3 rs11694732 TPO rs11694732 7.10 × 10−1 0.098 rs4927648 2.74 × 10−2 
2q35 rs737308 IGFBP5 rs737308 4.30 × 10−2 0.702 rs2712172 1.25 × 10−2 
rs13015993 IGFBP5 rs13015993 3.21 × 10−2 0.727 
4q31.23 rs10028213 NR3C2 rs10028213 7.48 × 10−3 0.266 rs9968300 1.77 × 10−5 
rs10030849 NR3C2 rs10030849 7.48 × 10−3 N/A 
rs10032216 NR3C2 rs10032216 7.48 × 10−3 N/A 
5q13.3 rs4704397 PDE8B rs4704397 2.14 × 10−2 0.001 rs10055027 1.36 × 10−3 
rs6885099 PDE8B rs6885099 7.55 × 10−3 
rs2046045 PDE8B rs2046045 8.13 × 10−3 
6p21.1 rs729761 VEGFA NA NA N/A rs10948107 9.57 × 10−4 
rs9472138 VEGFA rs9472138 3.39 × 10−2 
rs6923866 VEGFA rs6923866 1.85 × 10−3 0.915 
rs11755845 VEGFA rs11755845 1.60 × 10−3 0.831 
6q24.3 rs9497965 SASH1 rs9497965 1.75 × 10−1 0.031 rs58488869 8.03 × 10−3 
6q27 rs753760 PDE10A rs753760 7.39 × 10−2 0.825 rs58885758 7.15 × 10−3 
rs3008043 PDE10A rs3008043 2.42 × 10−2 0.866 
8p12 rs7825175 NRG1 rs7825175 1.68 × 10−2 0.685 rs11991469 6.56 × 10−3 
rs2439302 NRG1 rs2439302 6.89 × 10−3 0.947 
9p24.2 rs1571583 GLIS3 rs1571583 4.17 × 10−2 0.064 rs11999791 2.62 × 10−2 
9q22.33 rs965513 FOXE1 rs965513 1.26 × 10−6 0.777 rs10818050 1.17 × 10−7 
9q34.2 rs657152 ABO rs657152 8.81 × 10−1 0.091 rs10901263 6.60 × 10−4 
10q24.2 rs7913135 NKX2-3 rs7913135 2.27 × 10−1 1.000 rs7913750 1.97 × 10−1 
11p11.2 rs17723470 PRDM11 rs17723470 2.66 × 10−2 N/A rs72907072 2.57 × 10−4 
rs7128207 PRDM11 rs7128207 2.83 × 10−1 N/A 
12q23.1 rs61938844 ELK3 NA NA N/A rs12814360 2.16 × 10−2 
14q13.3 rs1537424 MBIP rs1537424 2.48 × 10−1 0.217 rs2780310 5.64 × 10−4 
rs116909374 MBIP NA NA N/A 
14q32.12 rs34269820 ITPK1 rs34269820 3.79 × 10−2 0.211 rs1263428 5.62 × 10−3 
rs11624776 ITPK1 rs11624776 1.04 × 10−1 0.166 
14q32.33 rs73362602 SIVA1 rs73362602 5.47 × 10−1 N/A rs12896579 1.28 × 10−2 
15q21.2 rs73398284 FGF7 rs73398284 8.00 × 10−2 N/A rs74406736 5.55 × 10−3 
rs10519227 FGF7 rs10519227 1.76 × 10−1 N/A 
15q26.1 rs17776563 MIR1179 rs8030854 (r2 = 0.94) 2.49 × 10−1 0.007 rs74027996 6.89 × 10−3 
16p23.2 rs7190187 MAF rs7190187 1.18 × 10−4 0.71 rs4575545 1.91 × 10−6 
rs3813582 MAF rs3813582 1.50 × 10−5 0.955 
17q24.3 rs9915657 SOX9 rs9915657 2.54 × 10−1 0.021 rs16977126 1.20 × 10−2 
19p13.2 rs4804416 INSR rs4804416 3.74 × 10−2 0.108 rs4804433 1.47 × 10−2 
rs10420008 INSR rs10420008 6.52 × 10−2 0.332 
20p11.21 rs6082762 FOXA2 rs6082762 2.91 × 10−1 0.609 rs4499503 3.26 × 10−2 

Chr., chromosome; SNP, single-nucleotide polymorphism; NA, not assessed.

aP-values were from linear regression models adjusted for age, gender and geographic region.

br2 in the 1000 Genomes ASN samples.

cTop SNP; the top association signal in each TSH risk locus in our data (recombination rate < 0.3 in the 1000 Genomes ASN samples).

Though our GWAS implicated XKR4, FOXE1 and CAPZB gene with a significant effect on variation of serum TSH levels, other variants also contribute to this quantitative trait. Therefore, we further analyzed the 45 SNPs in the 26 loci, which were reported to be associated with TSH levels in previous studies, and found that out of the 45 SNPs, 27 in 15 risk loci were associated with TSH levels in our GWAS data (P < 0.05) (Table 4). Interestingly, there were 36 loci harbored more than one SNPs associated with TSH level at the P < 1 × 10−4 in our GWAS data, which included four loci reported to be associated with TSH in previous studies. However, only nine out of 280 loci associated with TSH level at P < 1 × 10−3 in our GWAS data , were confirmed to be the susceptibility loci for TSH in previous studies (Table 4) (1117).

Association of the TSH candidate SNPs with thyroid diseases and expression patterns in target tissues

Serum TSH levels were previously reported to be correlated with the risk of thyroid cancer and could be a marker to predict thyroid cancer relapse (2426). Hence, we investigated whether the four serum-TSH candidate SNPs identified in our initial GWAS were associated with papillary thyroid carcinoma (PTC) in a sample of 108 patients with PTC and 1490 control individuals from the Chinese Han population. Two of the SNPs, rs2622590 and rs925489, were nominally associated with PTC (rs2622590, P = 0.014, OR = 1.41, 95% CI = 1.07–1.87; rs925489, P = 0.030, OR = 1.61, 95% CI = 1.04–2.47; Table 5).

Table 5.

Association of the selected SNPs with PTC risk and GD

SNP PTC association analysis (108 versus 1490)
 
GD association analysis (1442 versus 1468)
 
 Risk allele Case MAF Control MAF OR (95% CI) P-value Risk allele Case MAF Control MAF OR (95% CI) P-value 
rs2622590 0.519 0.432 1.41 (1.07–1.87) 0.014 0.426 0.439 1.03 (0.93–1.14) 0.567 
rs1348271 0.046 0.051 1.11 (0.58–2.14) 0.757 0.046 0.050 1.10 (0.86–1.40) 0.445 
rs6683419 0.315 0.262 1.29 (0.96–1.74) 0.092 0.265 0.264 1.00 (0.89–1.12) 0.935 
rs925489 0.120 0.079 1.61 (1.04–2.47) 0.030 0.089 0.078 1.16 (0.72–1.04) 0.125 
SNP PTC association analysis (108 versus 1490)
 
GD association analysis (1442 versus 1468)
 
 Risk allele Case MAF Control MAF OR (95% CI) P-value Risk allele Case MAF Control MAF OR (95% CI) P-value 
rs2622590 0.519 0.432 1.41 (1.07–1.87) 0.014 0.426 0.439 1.03 (0.93–1.14) 0.567 
rs1348271 0.046 0.051 1.11 (0.58–2.14) 0.757 0.046 0.050 1.10 (0.86–1.40) 0.445 
rs6683419 0.315 0.262 1.29 (0.96–1.74) 0.092 0.265 0.264 1.00 (0.89–1.12) 0.935 
rs925489 0.120 0.079 1.61 (1.04–2.47) 0.030 0.089 0.078 1.16 (0.72–1.04) 0.125 

SNP, single-nucleotide polymorphism; PTC, papillary thyroid carcinoma; GD, Graves’ disease; OR, odds ratio for the risk allele; CI, confidence interval.

Although we apply a Bonferroni correction for multiple testing of association between SNPs (rs2622590, rs1348271, rs6683419 and rs925489) and PTC (P 0.05/4 = 0.0125), rs965513 near FOXE1 on 9q22.33 (P = 0.030) and rs2622590 in XKR4 on 8p12.1 (P = 0.014) were no longer associated with PTC. However, we found that rs965513 in FOXE1 was significantly associated with PTC in Chinese population in a previous report (P = 1.18 × 10−4, OR = 1.53, 95% CI = 1.23–1.90; Supplementary Material, Table S6) (27). Because rs965513 was in complete LD with rs925489 (r2 = 0.97 in our Chinese Han samples), we performed a meta-analysis and found that rs925489 or rs965513 was significantly associated with PTC (PMeta = 1.28 × 10−5, OR = 1.54, 95% CI = 1.27–1.88) in combined Chinese PTC population (Supplementary Material, Table S6). We have neither more PTC samples nor reported data about the genotypes of SNPs in XKR4 region to further confirmed the association of rs2622590 with PTC. Therefore, the relation of the XKR4 SNP to PTC remains to be determined, and further studies based on larger PTC samples to confirm these two findings were needed.

The PTC risk alleles of the two SNPs (the TT genotype for rs2622590 and the CC genotype for rs925489) were also associated with low serum TSH levels (Table 5 and Supplementary Material, Table S3). None of the four candidate SNPs were associated with Graves' disease (GD), however, in a sample of 1442 patients with GD and 1468 control individuals collected from the Chinese Han population in our previous GWAS (18) (Table 5).

We used quantitative real-time PCR to measure the expression of XKR4, FOXE1 and CAPZB in 66 PTC tissues and 40 normal thyroid tissues adjacent to PTC tissues. We found that the expression levels of XKR4, FOXE1 and CAPZB in the PTC tissues were significantly higher than those in the normal thyroid tissues adjacent to PTC tissues (P = 3.85 × 10−7, 3.24 × 10−4, and 2.67 × 10−15, respectively; Fig. 4A). Moreover, the rs2622590_T and rs925489_C alleles, which were the risk alleles for PTC and were correlated with lower serum TSH levels, were also correlated with high expression levels of XKR4 and FOXE1, respectively, in the thyroid tissues of patients with PTC (PANOVA = 2.41 × 10−4 for rs2622590; PANOVA = 0.02 for rs925489; Fig. 4B). The rs6683419 genotypes, however, were not correlated with the expression of CAPZB in the PTC tumor tissues (PANOVA = 0.65; Fig. 4B).

DISCUSSION

The identification of susceptibility loci and genes related to serum TSH levels may provide important insight into the regulation of thyroid hormones, and may also be valuable for designing new preventative and therapeutic strategies for the disorders influenced by TSH, such as hypothyroidism, dyslipidemia, cardiovascular disease, myocardial infarction and atrial fibrillation (2834). Through a two-stage GWAS of circulating TSH levels in individuals from the Chinese populations, we identified a novel susceptibility locus in XKR4 at 8p12.1, and we confirmed two previously reported susceptibility loci near FOXE1 at 9q22.33 and near CAPZB at 1p36.13 (1217).

This study is the first to associate rs2622590, located in intron 2 of XKR4 at 8p12.1, with serum TSH levels. The XKR4 gene encodes a member of the XK, Kell blood group complex subunit-related protein family. Kell and XK are two distinct red blood cell-membrane proteins that form the Kell blood-group complex. XK is considered to be a membrane transport protein, and the absence of XK can lead to McLeod syndrome, a rare X-linked hereditary disease characterized by red blood cell acanthocytosis and late-onset central nervous system and neuromuscular abnormalities (35). More recently, a few SNPs near XKR4 at the 8p12.1 locus have been reported to be associated with attention deficit/hyperactivity disorders, responses to iloperidone and risperidone, and paclitaxel-induced peripheral sensory neuropathy (3639). XKR4 is abundantly expressed in the tissues of the brain, esophagus and stomach, but is relatively weakly expressed in the tissues of the thyroid, kidney and pancreas. We found that the expression level of XKR4 was significantly higher in the tumor tissues of patients with PTC than in the adjacent normal thyroid tissues. The rs2622590 genotypes were correlated with the expression of XKR4 in the tumor tissues of patients with PTC. Determining whether the rs2622590 risk allele mediates the regulation of serum TSH levels by XKR4, however, will require further study.

We found that the rs2622590_T and rs925489_C alleles, which were associated with lower serum TSH levels, might be risk alleles for patients with PTC. These alleles were significantly correlated with the enhanced expression of XKR4 and FOXE1, respectively, in the PTC tumor tissues. FOXE1, located about 69 kb downstream of rs925489, is important for both pituitary-gland development and thyroid-gland development (4042). FOXE1 is also necessary for the synthesis of the thyroid hormones T3 and T4, because it regulates the transcription of the TG (thyroglobulin) and TPO (thyroperoxidase) genes by binding to response elements in the promoter regions (43,44).

It is well known that the product of TG is the precursor of T3 and T4, the synthesis of which is catalyzed by TPO. Therefore, it is tempting to presume that the C allele of rs925489 increases the expression of FOXE1 in thyroid tissues, up-regulating TG and TPO transcription and enhancing the synthesis of T3 and T4; the increased synthesis of thyroid hormones acts in a negative-feedback loop, further decreasing the serum concentration of TSH, leading to the association of the C allele of rs925489 with lower TSH levels. Indeed, all the four PTC risk alleles except rs965513 near to FOXE1 were associated with the lower TSH levels and trended to increased FT4 levels. Moreover, for rs966423 on 2q35 and rs116909374 on 14q13.3, the lower-TSH alleles were significantly associated with increase FT4 at the P < 0.05 level (15). Although the PTC risk allele of rs965513 in FOXE1 was associated with lower TSH and FT4 levels, the risk allele appeared to be associated with the increase FT3 level (P < 0.05) (15). Therefore, all the data suggested that the PTC risk allele trended to lower the TSH and increase the physiological function of thyroid as well as supported our hypothesis. With regard to the SNPs near to FOXE1 at 9q22.33, ours and previous reports supported that rs925489, a proxy rs965513 (r2 = 0.97 in our Chinese Han samples), was associated with TSH concentration at the GWAS significant level (12,15). However, the more recent study from Porcu et al. (17) found that rs7045138 and rs965513 near FOXE1 were associated with FT4 level (P = 1.5 × 10−11 and 3.45 × 10−8, respectively), but not associated with the TSH level. Interestingly, the rs965513 was weakly linked with rs7045138 in Caucasian, but not linked with rs7045138 in Chinese Han population (r2 = 0.409 in the 1000 Genomes CEU samples and 0.001 in the 1000 Genomes ASN samples, respectively). In our GWAS data, rs965513, but not rs7045138 was significantly associated with TSH level (P = 1.26 × 10−6 and 0.70, respectively). These data suggested that two SNPs near FOXE1 region were independently associated with the TSH or FT4 levels, respectively. Regretfully, the FT4 levels have not been collected in our current study, the association study of the SNPs in FOXE1 region with FT4 and TSH levels in the Chinese population in further to unlock the perplexed was expected.

The rs2622590_T allele, which was associated with lower serum TSH levels, was also correlated with the increased expression of XKR4 in thyroid tumor tissues; however, the relationship between serum TSH levels and XKR4 expression in thyroid tissues remains unknown. Consistent with the results of our study, the alleles of four other SNPs associated with lower serum TSH levels were found to increase the risk of PTC in previous studies (12,15). All of the available data suggest that low serum TSH levels are correlated with an increased risk of PTC.

In summary, we identified a new serum TSH susceptibility locus in the XKR4 region at 8q12.1 and confirmed two previously reported susceptibility loci near FOXE1 at 9q22.33 and CAPZB at 1p36.13, respectively. The rs2622590_T allele in XKR4 and the rs925489_C allele near FOXE1, both associated with lower serum TSH levels, were correlated with the increased expression of XKR4 and FOXE1, respectively, in the thyroid tumor tissues, and possibly conferred a predisposition to PTC. These findings provided new insights into the mechanisms regulating serum TSH levels and suggested that low serum TSH levels were correlated with susceptibility to PTC.

MATERIALS AND METHODS

GWAS and replication cohorts

The GWAS cohorts included in this study were collected from the Chinese Han population through collaborations with two hospitals in northern China (Xuzhou and Linyi) (18,45,46). The replication cohort was collected from the Chinese She population, who is one of the important ethnic minorities, residing in the Ningde City, Fujian Province in southern China. The marriage between Han and She were prohibited 60 years ago, suggesting the She ethnic is an independent isolated population in China (19). Subjects were excluded if they had thyroid nodules, autoimmune thyroid disease (AITD) or a family history of AITD or other autoimmune disorders.

During a physical examination, 5 ml blood samples were collected from each participant and stored at −80°C for DNA extraction and measurements of serum TSH sensitivity and TPO-Ab levels. TSH and TPO-Ab were measured using a chemiluminescence immunoassay (CLIA) according to the manufacturer's instructions in our laboratory (Siemens ADVIA Centaur CP). To exclude subjects with clinical or subclinical AITD and subjects taking medications likely to influence serum thyroid function, we measured the serum levels of TPO-Ab among the 1816 GWAS subjects and the 4233 replication subjects. Based on those measurements, we excluded 1420 individuals with TPO-Ab levels <5.60 U/ml and TSH levels <4.93 mU/l or <0.36 mU/l. Ultimately, we included 1394 and 3235 subjects for the GWAS and replication studies, respectively (Fig. 1). The median TSH level was 1.62 mU/l and the interquartile range was 1.02–2.06 mU/l. We also included 108 patients with PTC recruited from the First hospital affiliated to Bengbu Medical College, in northern China to evaluate whether selected loci were associated with PTC. The sample sizes, sex distributions, mean ages, BMIs and serum TSH levels of the GWAS and replication participants are summarized in Table 1.

All of the participating cohorts were granted approval by the local Ethics Committee from Ruijin Hospital, the Central Hospital of Xuzhou, Linyi People's Hospital, the Fujian Province Hospital Affiliated to Fujian Medical University and the First Hospital Affiliated to Bengbu Medical College. All of the subjects in this study provided written informed consent using protocols approved by the local Ethics Committees.

GWAS genotyping and initial quality control

DNA was extracted from the blood samples using a FUJIFILM QuickGene-610L system. Genome-wide genotyping was performed using the Illumina Human Omni-Express BeadChip platform on the Illumina Human660-Quad BeadChips at the Chinese National Human Genome Center in Shanghai, China (18). Genotype clustering was conducted using Illumina GenomeStudio V2011.1 software based on the 660W-Quad_v1_H manifest files, which converted the fluorescence intensities into SNP genotypes. The mean call rate across all 1394 genotyped samples was 99.8%. No individuals were excluded from the GWAS analysis. Quality control was performed on the SNPs and samples before analysis using the PLINK software to ensure robust association tests (47).

Of the 655 214 markers assayed, 3185 were removed because they were from the Y or mitochondrial chromosomes or were CNV-related. We discarded 168 082 markers because of deviation from Hardy–Weinberg equilibrium P ≤ 10−6, genotype call rates below 98%, or MAFs < 0.01, leaving 483 947 SNPs for subsequent analyses. Next, we excluded 48 samples with call rates <98%, gender inconsistencies or cryptic relatedness. Ultimately, 1346 samples were available for the GWAS analysis (Fig. 1).

Genotype imputation

To infer the genotypes of SNPs in the GWAS cohort that were not previously genotyped, we used the IMPUTE2 software (48) with the 1000G phase-1 interim impute data (March 2012) as a reference. We used an estimated probability >0.90 to call imputed genotypes. Taking imputation uncertainty into account, the association analysis was carried out using the SNPTESTv2 software (frequentist) association tests with score method (49).

SNP selection and genotyping in the replication study

We selected three SNPs (rs1348271, rs925489 and rs2622590) that were strongly associated with serum TSH levels (P < 1 × 10−6) as candidate loci for the replication study (rs7850258 was in complete LD with rs925489, r2 = 1.00 in our Chinese Han samples). We also selected SNP rs6683419 (P = 1.33 × 10−5) at 1p36.13, which was previously reported (13) to be associated with serum TSH levels. In the replication study, the DNA concentrations of all samples were standardized to 50 ng/μl in 96-well plates, and one negative control (DNase-free and RNase-free water) was included in one of the 96 wells at random. Four SNPs were genotyped using the ABI ViiA™ 7 Real-Time PCR System according to the manufacturer's protocol by technicians who were blinded to the sample status. The average call rate was >95%.

Statistical analysis

To reduce potential bias caused by deviation of TSH levels from a normal distribution, we applied the natural-log transformation to normalize the distributions of all phenotypic data before testing for associations. A standard linear regression model was used to analyze the association of each SNP with quantitative traits, assuming an additive genetic model, which was implemented in the PLINK software package (47). For typed or imputed genotypes, the allelic dosage at each SNP was the independent variable, adjusted for primary covariates of age, gender, geographic region and race. A P-value of 5 × 10−8 was used as the cutoff for genome-wide significance. The effects of selected SNPs on PTC risk were further evaluated for the case–control samples using logistic regression models.

PCA and MDS analysis were implemented using the EIGENSTRAT (50) and PLINK (47) software, respectively, to evaluate population stratification in the GWAS samples. The Haploview software version 4.2 was used to generate the genome-wide plot of P-values, and regional plots were generated using the LocusZoom Version 1.1 software (51). Quantile–quantile plots were constructed by plotting the observed distribution of the P-values for the given SNPs against the theoretical distribution of the expected P-values. The calculations and the plots were implemented using the PLINK and R statistical packages.

mRNA expression analyses

In the mRNA expression analysis, 66 PTC samples and 40 paraneoplastic tissue samples were surgically collected from patients. Total RNA was isolated using Trizol reagent (Invitrogen) according to the manufacturer's instructions and then treated with DNase I at room temperature for 10 min to degrade possible contaminating genomic DNA. cDNAs were made from 1 μg RNA templates using reverse transcriptase and oligo(dT) primer (Takara). Quantitative real-time PCR for a series of genes was performed in triplicate using the SYBR Green and ABI ViiA™ 7 Real-Time PCR System. Data were analyzed and presented relative to the expression of the GAPDH housekeeping gene. The primer sequences used for real-time PCR are shown in the Supplementary Material, Table S7. We performed statistical analysis of the expression data using an ANOVA and an unpaired Student's t-test (the two-tail P-values are indicated on the figures).

URLs

PLINK v1.07, http://pngu.mgh.harvard.edu/~purcell/plink/

R statistical environment version 2.10.0, http://www.r-project.org/

LocusZoom Version 1.1, http://csg.sph.umich.edu/locuszoom/

IMPUTE2, http://mathgen.stats.ox.ac.uk/impute/impute_v2.html/

SNPTEST v2 software, https://mathgen.stats.ox.ac.uk/genetics_software/snptest/snptest.html/

ENCODE database, http://genome.ucsc.edu/ENCODE/

AUTHOR CONTRIBUTIONS

H.-D.S. was responsible for the coordination of the project. H.-D.S., J.-L.C, S.-J.C, G.N., C.-M.P., S.-X.Z. and M.Z. contributed to study design. The writing team consisted of H.-D.S., M.Z. and S.-X.Z. H.-D.S., M.Z., C.-M.P. and S.-X.Z. contributed to the project management. G.C., H.-D.S., L.J., G.-Q.G., X.-M.Z., C.-L.Z., B.S., C.-M.P., G.-Y.Y., W.L., S.-X.Z. and M.Z. took part in the collection of clinical samples, DNA and sample QC. W.H. contributed to whole genome genotyping. C.-M.P., M.Z., X.-P.Y., S.-S.Y. and H.-J.X. took part in replication genotyping. C.-M.P., M.Z. and X.-P.Y. contributed to CAPZB, XKR4 and FOXE1 real-time RT-PCR. Z.-H.G., S.-X.Z., H.-N.W. and M.Z. performed imputation analysis and took part in statistical analysis.

SUPPLEMENTARY MATERIAL

Supplementary Material is available at HMG online.

FUNDING

This work was supported by National Basic Research Program of China (973) (2010CB529204 and 2012CB517604), National Natural Science Foundation of China (81370888, 81100553, 31171127, 81270863, 81200568, 81370965, 81270864, 81272748 and 81101444), Program for Graves′ Disease Innovative Research Team of Shanghai Municipal Education Commission, National Natural Science Foundation of Jiangsu Province (BK2009208), National Natural Science Foundation of Shandong Province (ZH2011HM067), Natural Science Foundation of Anhui Province (1208085MH140) and Dr Innovation Fund of SJTU School of Medicine (BXJ201208).

ACKNOWLEDGEMENTS

We are indebted to the staff, patients and all other individuals involved in this study for their dedication and contributions.

Conflict of Interest statement. None declared.

REFERENCES

1
Porterfield
S.P.
Hendrich
C.E.
The role of thyroid hormones in prenatal and neonatal neurological development--current perspectives
Endocr. Rev.
 
1993
14
94
106
2
Asvold
B.O.
Bjoro
T.
Nilsen
T.I.
Vatten
L.J.
Association between blood pressure and serum thyroid-stimulating hormone concentration within the reference range: a population-based study
J. Clin. Endocrinol. Metab.
 
2007
92
841
845
3
Asvold
B.O.
Vatten
L.J.
Nilsen
T.I.
Bjoro
T.
The association between TSH within the reference range and serum lipid concentrations in a population-based study
The HUNT Study. Eur. J. Endocrinol.
 
2007
156
181
186
4
Oppenheimer
J.H.
Schwartz
H.L.
Mariash
C.N.
Kinlaw
W.B.
Wong
N.C.
Freake
H.C.
Advances in our understanding of thyroid hormone action at the cellular level
Endocr. Rev.
 
1987
8
288
308
5
Larsen
P.R.
Thyroid-pituitary interaction: feedback regulation of thyrotropin secretion by thyroid hormones
N. Engl. J. Med.
 
1982
306
23
32
6
Vassart
G.
Dumont
J.E.
The thyrotropin receptor and the regulation of thyrocyte function and growth
Endocr. Rev.
 
1992
13
596
611
7
Keffer
J.H.
Preanalytical considerations in testing thyroid function
Clin. Chem.
 
1996
42
125
134
8
Andersen
S.
Pedersen
K.M.
Bruun
N.H.
Laurberg
P.
Narrow individual variations in serum T(4) and T(3) in normal subjects: a clue to the understanding of subclinical thyroid disease
J. Clin. Endocrinol. Metab.
 
2002
87
1068
1072
9
Hansen
P.S.
Brix
T.H.
Sorensen
T.I.
Kyvik
K.O.
Hegedus
L.
Major genetic influence on the regulation of the pituitary-thyroid axis: a study of healthy Danish twins
J. Clin. Endocrinol. Metab.
 
2004
89
1181
1187
10
Samollow
P.B.
Perez
G.
Kammerer
C.M.
Finegold
D.
Zwartjes
P.W.
Havill
L.M.
Comuzzie
A.G.
Mahaney
M.C.
Goring
H.H.
Blangero
J.
et al.  
Genetic and environmental influences on thyroid hormone variation in Mexican Americans
J. Clin. Endocrinol. Metab.
 
2004
89
3276
3284
11
Arnaud-Lopez
L.
Usala
G.
Ceresini
G.
Mitchell
B.D.
Pilia
M.G.
Piras
M.G.
Sestu
N.
Maschio
A.
Busonero
F.
Albai
G.
et al.  
Phosphodiesterase 8B gene variants are associated with serum TSH levels and thyroid function
Am. J. Hum. Genet.
 
2008
82
1270
1280
12
Gudmundsson
J.
Sulem
P.
Gudbjartsson
D.F.
Jonasson
J.G.
Sigurdsson
A.
Bergthorsson
J.T.
He
H.
Blondal
T.
Geller
F.
Jakobsdottir
M.
et al.  
Common variants on 9q22.33 and 14q13.3 predispose to thyroid cancer in European populations
Nat. Genet.
 
2009
41
460
464
13
Panicker
V.
Wilson
S.G.
Walsh
J.P.
Richards
J.B.
Brown
S.J.
Beilby
J.P.
Bremner
A.P.
Surdulescu
G.L.
Qweitin
E.
Gillham-Nasenya
I.
et al.  
A locus on chromosome 1p36 is associated with thyrotropin and thyroid function as identified by genome-wide association study
Am. J. Hum. Genet.
 
2010
87
430
435
14
Medici
M.
van der Deure
W.M.
Verbiest
M.
Vermeulen
S.H.
Hansen
P.S.
Kiemeney
L.A.
Hermus
A.R.
Breteler
M.M.
Hofman
A.
Hegedus
L.
et al.  
A large-scale association analysis of 68 thyroid hormone pathway genes with serum TSH and FT4 levels
Eur. J. Endocrinol.
 
2011
164
781
788
15
Gudmundsson
J.
Sulem
P.
Gudbjartsson
D.F.
Jonasson
J.G.
Masson
G.
He
H.
Jonasdottir
A.
Sigurdsson
A.
Stacey
S.N.
Johannsdottir
H.
et al.  
Discovery of common variants associated with low TSH levels and thyroid cancer risk
Nat. Genet.
 
2012
44
319
322
16
Rawal
R.
Teumer
A.
Volzke
H.
Wallaschofski
H.
Ittermann
T.
Asvold
B.O.
Bjoro
T.
Greiser
K.H.
Tiller
D.
Werdan
K.
et al.  
Meta-analysis of two genome-wide association studies identifies four genetic loci associated with thyroid function
Hum. Mol. Genet.
 
2012
21
3275
3282
17
Porcu
E.
Medici
M.
Pistis
G.
Volpato
C.B.
Wilson
S.G.
Cappola
A.R.
Bos
S.D.
Deelen
J.
den Heijer
M.
Freathy
R.M.
et al.  
A meta-analysis of thyroid-related traits reveals novel loci and gender-specific differences in the regulation of thyroid function
PloS Genet.
 
2013
9
e1003266
18
Chu
X.
Pan
C.M.
Zhao
S.X.
Liang
J.
Gao
G.Q.
Zhang
X.M.
Yuan
G.Y.
Li
C.G.
Xue
L.Q.
Shen
M.
et al.  
A genome-wide association study identifies two new risk loci for Graves’ disease
Nat. Genet.
 
2011
43
897
901
19
Lin
Y.
Lai
X.
Chen
B.
Xu
Y.
Huang
B.
Chen
Z.
Zhu
S.
Yao
J.
Jiang
Q.
Huang
H.
et al.  
Genetic variations in CYP17A1, CACNB2 and PLEKHA7 are associated with blood pressure and/or hypertension in She ethnic minority of China
Atherosclerosis
 
2011
219
709
714
20
Dixon
A.L.
Liang
L.
Moffatt
M.F.
Chen
W.
Heath
S.
Wong
K.C.
Taylor
J.
Burnett
E.
Gut
I.
Farrall
M.
et al.  
A genome-wide association study of global gene expression
Nat. Genet.
 
2007
39
1202
1207
21
Zeller
T.
Wild
P.
Szymczak
S.
Rotival
M.
Schillert
A.
Castagne
R.
Maouche
S.
Germain
M.
Lackner
K.
Rossmann
H.
et al.  
Genetics and beyond--the transcriptome of human monocytes and disease susceptibility
PLoS One
 
2010
5
e10693
22
Fairfax
B.P.
Makino
S.
Radhakrishnan
J.
Plant
K.
Leslie
S.
Dilthey
A.
Ellis
P.
Langford
C.
Vannberg
F.O.
Knight
J.C.
Genetics of gene expression in primary immune cells identifies cell type-specific master regulators and roles of HLA alleles
Nat. Genet.
 
2012
44
502
510
23
Consortium
E.P.
Birney
E.
Stamatoyannopoulos
J.A.
Dutta
A.
Guigo
R.
Gingeras
T.R.
Margulies
E.H.
Weng
Z.
Snyder
M.
Dermitzakis
E.T.
et al.  
Identification and analysis of functional elements in 1% of the human genome by the ENCODE pilot project
Nature
 
2007
447
799
816
24
Boelaert
K.
Horacek
J.
Holder
R.L.
Watkinson
J.C.
Sheppard
M.C.
Franklyn
J.A.
Serum thyrotropin concentration as a novel predictor of malignancy in thyroid nodules investigated by fine-needle aspiration
J. Clin. Endocrinol. Metab.
 
2006
91
4295
4301
25
Haymart
M.R.
Repplinger
D.J.
Leverson
G.E.
Elson
D.F.
Sippel
R.S.
Jaume
J.C.
Chen
H.
Higher serum thyroid stimulating hormone level in thyroid nodule patients is associated with greater risks of differentiated thyroid cancer and advanced tumor stage
J. Clin. Endocrinol. Metab.
 
2008
93
809
814
26
McLeod
D.S.
Watters
K.F.
Carpenter
A.D.
Ladenson
P.W.
Cooper
D.S.
Ding
E.L.
Thyrotropin and thyroid cancer diagnosis: a systematic review and dose-response meta-analysis
J. Clin. Endocrinol. Metab.
 
2012
97
2682
2692
27
Wang
Y.L.
Feng
S.H.
Guo
S.C.
Wei
W.J.
Li
D.S.
Wang
Y.
Wang
X.
Wang
Z.Y.
Ma
Y.Y.
Jin
L.
et al.  
Confirmation of papillary thyroid cancer susceptibility loci identified by genome-wide association studies of chromosomes 14q13, 9q22, 2q35 and 8p12 in a Chinese population
J. Med. Genet.
 
2013
50
689
695
28
Persani
L.
Clinical review: central hypothyroidism: pathogenic, diagnostic, and therapeutic challenges
J. Clin. Endocrinol. Metab.
 
2012
97
3068
3078
29
Asvold
B.O.
Bjoro
T.
Vatten
L.J.
Associations of TSH levels within the reference range with future blood pressure and lipid concentrations: 11-year follow-up of the HUNT study
Eur. J. Endocrinol.
 
2013
169
73
82
30
Pasqualetti
G.
Tognini
S.
Polini
A.
Caraccio
N.
Monzani
F.
Is subclinical hypothyroidism a cardiovascular risk factor in the elderly?
J. Clin. Endocrinol. Metab.
 
2013
98
2256
2266
31
Nanchen
D.
Gussekloo
J.
Westendorp
R.G.
Stott
D.J.
Jukema
J.W.
Trompet
S.
Ford
I.
Welsh
P.
Sattar
N.
Macfarlane
P.W.
et al.  
Subclinical thyroid dysfunction and the risk of heart failure in older persons at high cardiovascular risk
J. Clin. Endocrinol. Metab.
 
2012
97
852
861
32
Hak
A.E.
Pols
H.A.
Visser
T.J.
Drexhage
H.A.
Hofman
A.
Witteman
J.C.
Subclinical hypothyroidism is an independent risk factor for atherosclerosis and myocardial infarction in elderly women: the Rotterdam Study
Ann. Intern. Med.
 
2000
132
270
278
33
Bielecka-Dabrowa
A.
Mikhailidis
D.P.
Rysz
J.
Banach
M.
The mechanisms of atrial fibrillation in hyperthyroidism
Thyroid Res.
 
2009
2
4
34
Sawin
C.T.
Geller
A.
Wolf
P.A.
Belanger
A.J.
Baker
E.
Bacharach
P.
Wilson
P.W.
Benjamin
E.J.
D'Agostino
R.B.
Low serum thyrotropin concentrations as a risk factor for atrial fibrillation in older persons
N. Engl. J. Med.
 
1994
331
1249
1252
35
Calenda
G.
Peng
J.
Redman
C.M.
Sha
Q.
Wu
X.
Lee
S.
Identification of two new members, XPLAC and XTES, of the XK family
Gene
 
2006
370
6
16
36
Lantieri
F.
Glessner
J.T.
Hakonarson
H.
Elia
J.
Devoto
M.
Analysis of GWAS top hits in ADHD suggests association to two polymorphisms located in genes expressed in the cerebellum
Am. J. Med. Genet. B Neuropsychiatr. Genet.,
 
2010
153B
1127
1133
37
Fijal
B.A.
Stauffer
V.L.
Kinon
B.J.
Conley
R.R.
Hoffmann
V.P.
Witte
M.M.
Zhao
F.
Houston
J.P.
Analysis of gene variants previously associated with iloperidone response in patients with schizophrenia who are treated with risperidone
J. Clin. Psychiatry
 
2012
73
367
371
38
Lavedan
C.
Licamele
L.
Volpi
S.
Hamilton
J.
Heaton
C.
Mack
K.
Lannan
R.
Thompson
A.
Wolfgang
C.D.
Polymeropoulos
M.H.
Association of the NPAS3 gene and five other loci with response to the antipsychotic iloperidone identified in a whole genome association study
Mol. Psychiatry
 
2009
14
804
819
39
Leandro-Garcia
L.J.
Inglada-Perez
L.
Pita
G.
Hjerpe
E.
Leskela
S.
Jara
C.
Mielgo
X.
Gonzalez-Neira
A.
Robledo
M.
Avall-Lundqvist
E.
et al.  
Genome-wide association study identifies ephrin type A receptors implicated in paclitaxel induced peripheral sensory neuropathy
J. Med. Genet.
 
2013
50
599
605
40
De Felice
M.
Ovitt
C.
Biffali
E.
Rodriguez-Mallon
A.
Arra
C.
Anastassiadis
K.
Macchia
P.E.
Mattei
M.G.
Mariano
A.
Scholer
H.
et al.  
A mouse model for hereditary thyroid dysgenesis and cleft palate
Nat. Genet.
 
1998
19
395
398
41
Fagman
H.
Nilsson
M.
Morphogenetics of early thyroid development
J. Mol. Endocrinol.
 
2011
46
R33
R42
42
Parlato
R.
Rosica
A.
Rodriguez-Mallon
A.
Affuso
A.
Postiglione
M.P.
Arra
C.
Mansouri
A.
Kimura
S.
Di Lauro
R.
De Felice
M.
An integrated regulatory network controlling survival and migration in thyroid organogenesis
Dev. Biol.
 
2004
276
464
475
43
Zannini
M.
Avantaggiato
V.
Biffali
E.
Arnone
M.I.
Sato
K.
Pischetola
M.
Taylor
B.A.
Phillips
S.J.
Simeone
A.
Di Lauro
R.
TTF-2, a new forkhead protein, shows a temporal expression in the developing thyroid which is consistent with a role in controlling the onset of differentiation
EMBO J.
 
1997
16
3185
3197
44
Ortiz
L.
Aza-Blanc
P.
Zannini
M.
Cato
A.C.
Santisteban
P.
The interaction between the forkhead thyroid transcription factor TTF-2 and the constitutive factor CTF/NF-1 is required for efficient hormonal regulation of the thyroperoxidase gene transcription
J. Biol. Chem.
 
1999
274
15213
15221
45
Song
H.D.
Liang
J.
Shi
J.Y.
Zhao
S.X.
Liu
Z.
Zhao
J.J.
Peng
Y.D.
Gao
G.Q.
Tao
J.
Pan
C.M.
et al.  
Functional SNPs in the SCGB3A2 promoter are associated with susceptibility to Graves’ disease
Hum. Mol. Genet.
 
2009
18
1156
1170
46
Zhao
S.X.
Xue
L.Q.
Liu
W.
Gu
Z.H.
Pan
C.M.
Yang
S.Y.
Zhan
M.
Wang
H.N.
Liang
J.
Gao
G.Q.
et al.  
Robust evidence for five new Graves’ disease risk loci from a staged genome-wide association analysis
Hum. Mol. Genet.
 
2013
22
3347
3362
47
Purcell
S.
Neale
B.
Todd-Brown
K.
Thomas
L.
Ferreira
M.A.
Bender
D.
Maller
J.
Sklar
P.
de Bakker
P.I.
Daly
M.J.
et al.  
PLINK: a tool set for whole-genome association and population-based linkage analyses
Am. J. Hum. Genet.
 
2007
81
559
575
48
Howie
B.N.
Donnelly
P.
Marchini
J.
A flexible and accurate genotype imputation method for the next generation of genome-wide association studies
PLoS Genet
 
2009
5
e1000529
49
Marchini
J.
Howie
B.
Myers
S.
McVean
G.
Donnelly
P.
A new multipoint method for genome-wide association studies by imputation of genotypes
Nat. Genet.
 
2007
39
906
913
50
Price
A.L.
Patterson
N.J.
Plenge
R.M.
Weinblatt
M.E.
Shadick
N.A.
Reich
D.
Principal components analysis corrects for stratification in genome-wide association studies
Nat. Genet.
 
2006
38
904
909
51
Pruim
R.J.
Welch
R.P.
Sanna
S.
Teslovich
T.M.
Chines
P.S.
Gliedt
T.P.
Boehnke
M.
Abecasis
G.R.
Willer
C.J.
LocusZoom: regional visualization of genome-wide association scan results
Bioinformatics
 
2010
26
2336
2337

APPENDIX

The china consortium for the genetics of autoimmune thyroid disease members

Huai-Dong Song [State Key Laboratory of Medical Genomics, Molecular Medicine Center, Shanghai Institute of Endocrinology and Metabolism, Ruijin Hospital Affiliated to Shanghai Jiaotong University (SJTU) School of Medicine, Shanghai, China], Shuang-Xia Zhao (State Key Laboratory of Medical Genomics, Molecular Medicine Center, Shanghai Institute of Endocrinology and Metabolism, Ruijin Hospital Affiliated to SJTU School of Medicine, Shanghai, China), Chun-Ming Pan (State Key Laboratory of Medical Genomics, Ruijin Hospital Affiliated to SJTU School of Medicine, Shanghai, China), Jun Liang (Department of Endocrinology, The Central Hospital of Xuzhou Affiliated to Xuzhou Medical College, Xuzhou, Jiangsu Province, China), Xiao-Mei Zhang (Department of Endocrinology, The First Hospital Affiliated to Bengbu Medical College, Bengbu, Anhui Province, China), Guo-Yue Yuan (Department of Endocrinology, The Hospital Affiliated to Jiangsu University, Zhenjiang, Jiangsu Province, China), Chang-Gui Li (Department of Endocrinology and Gout Laboratory, Medical School Hospital of Qingdao University, Qingdao, Shandong Province, China), Jia-Lun Chen (Shanghai Institute of Endocrinology and Metabolism, Department of Endocrinology, Ruijin Hospital Affiliated to SJTU School of Medicine, Shanghai, China), Guan-Qi Gao (Department of Endocrinology, Linyi People's Hospital, Linyi, Shandong Province, China), Li-Bin Liu (Department of Endocrinology, Xiehe Hospital Affiliated to Fujian Medical University, Fuzhou, Fujian Province, China), Gang Chen (Department of Endocrinology, Fujian Province Hospital, Fuzhou, Fujian Province, China), Qing Su (Department of Endocrinology, Xin-Hua Hospital Affiliated to SJTU School of Medicine, Shanghai 200092, China), Yong-De Peng (Department of Endocrinology, The First People's Hospital Affiliated to SJTU School of Medicine, Shanghai, China), Jia-Jun Zhao (Department of Endocrinology, Shandong Province Hospital, Shandong University, Jinan, China).

Author notes

Joint first authors.
Joint senior authors.
A full list of The China Consortium for the Genetics of Autoimmune Thyroid Disease members and affiliations appears at the end of this manuscript.