Research Article: A genome-wide association study of thyroid stimulating hormone and free thyroxine in Danish children and adolescents

Date Published: March 23, 2017

Publisher: Public Library of Science

Author(s): Tenna Ruest Haarmark Nielsen, Emil Vincent Rosenbaum Appel, Mathilde Svendstrup, Johanne Dam Ohrt, Maria Dahl, Cilius Esmann Fonvig, Mette Hollensted, Christian Theil Have, Haja N. Kadarmideen, Oluf Pedersen, Torben Hansen, Jens-Christian Holm, Niels Grarup, David Meyre.


Hypothyroidism is associated with obesity, and thyroid hormones are involved in the regulation of body composition, including fat mass. Genome-wide association studies (GWAS) in adults have identified 19 and 6 loci associated with plasma concentrations of thyroid stimulating hormone (TSH) and free thyroxine (fT4), respectively.

This study aimed to identify and characterize genetic variants associated with circulating TSH and fT4 in Danish children and adolescents and to examine whether these variants associate with obesity.

Genome-wide association analyses of imputed genotype data with fasting plasma concentrations of TSH and fT4 from a population-based sample of Danish children, adolescents, and young adults, and a group of children, adolescents, and young adults with overweight and obesity were performed (N = 1,764, mean age = 12.0 years [range 2.5−24.7]). Replication was performed in additional comparable samples (N = 2,097, mean age = 11.8 years [1.2−22.8]). Meta-analyses, using linear additive fixed-effect models, were performed on the results of the discovery and replication analyses.

No novel loci associated with TSH or fT4 were identified. Four loci previously associated with TSH in adults were confirmed in this study population (PDE10A (rs2983511: β = 0.112SD, p = 4.8 ∙ 10−16), FOXE1 (rs7847663: β = 0.223SD, p = 1.5 ∙ 10−20), NR3C2 (rs9968300: β = 0.194SD), p = 2.4 ∙ 10−11), VEGFA (rs2396083: β = 0.088SD, p = 2.2 ∙ 10−10)). Effect sizes of variants known to associate with TSH or fT4 in adults showed a similar direction of effect in our cohort of children and adolescents, 11 of which were associated with TSH or fT4 in our study (p<0.0002). None of the TSH or fT4 associated SNPs were associated with obesity in our cohort, indicating no pleiotropic effects of these variants on obesity. In a group of Danish children and adolescents, four loci previously associated with plasma TSH concentrations in adults, were associated with plasma TSH concentrations in children, suggesting comparable genetic determinants of thyroid function in adults and children.

Partial Text

Thyroid hormone concentrations exhibit a physiological narrow variability within the individual, but exhibit a considerable inter-individual variability [1]. This could be explained by a genetic influence, which has been shown to explain 65% of the variation in concentrations of thyroid stimulating hormone (TSH) and thyroxine (T4) in healthy adults [2]. Nineteen loci have been found to be associated with circulating TSH and six with free T4 (fT4) concentrations in genome-wide association studies (GWAS) in adults [3–12]. Despite a tight physiological regulation between TSH and fT4, these two hormones do not seem to share genetic influences [13]. A recent study showed that all common variants could explain ~20% of the variation in TSH and fT4 concentrations [12]. However, the identified loci specific to TSH only explained a small part of the variation in TSH concentration, suggesting that a part of the missing heritability in relation to thyroid hormones might be explained by larger effects of rare or low-frequency variants [12]. Only a few GWAS on TSH and fT4 have included children [9,10]. Only one study in neonates (N = 1,583) has focused exclusively on children, replicating two associations previously identified in adults; one in PDE8B and one upstream of FOXE1 [14]. The scarcity of studies in children makes it difficult to determine, whether the loci identified to be important for TSH and fT4 in adults, are also important during childhood and adolescence. Although associations with certain phenotypes, such as BMI and vitamin D, identified in adults have been replicated in children [15,16], results from GWAS of BMI in children and adolescents have also contributed with additional genetic associations [17]. Even during childhood the effects of genetic variants may act in opposite directions, as exemplified by a negative influence of some FTO variants in early childhood compared to a positive effect in later childhood [18], indicating that investigations in children may help explain more of the time-dependent influence of genetic variation on a given phenotype. The age-dependent effects of genetic variations on TSH and fT4 were illustrated in a recent study, in which a gene risk score (GRS) based on studies in adults explained considerably less of the variation in TSH and fT4 in newborns compared to children at six years of age (TSH 0.8−1.0%, fT4 0.2−0.3% vs. TSH 5.3−5.5%, fT4 1.9−3.6%) [19]. This difference may be biased by the maternal thyroid concentrations, which are known to pass the placental barrier. Thus the thyroid function of the newborn is not established and stabilized until several days after birth [19,20]. In adults, up to 7.1% of the variation in serum TSH concentrations and 1.9% of the variation in serum fT4 concentrations have been explained by gene risk scores for the given phenotype [12], thus there is still unexplained variation in children, not accounted for by the variation described in studies in adults.

A total of 1,680 individuals were available for the TSH discovery analysis of 8,508,717 genetic markers. Of these, 183 markers in seven loci reached the replication threshold of p < 1 ∙ 10−6 in the discovery analyses (Table 1 and Fig 1). When we adjusted for BMI SDS, we observed comparable results for the seven identified loci and no additional associated loci were detected. We performed the analyses stratified for sex, which did not reveal additional associated loci (data not shown). In a GWAS of 1,680 children and adolescents and replication in an additional set of 2,129 children and adolescents, four SNPs reached genome-wide significance for their associations with TSH concentrations (PDE10A (rs2983511), FOXE1 (rs7847663), NR3C2 (rs9968300), VEGFA (rs2396083)). These four loci have all previously been associated with serum TSH concentrations in adults [3].   Source:


0 0 vote
Article Rating
Notify of
Inline Feedbacks
View all comments