GWAS of thyroid stimulating hormone highlights pleiotropic effects and inverse association with thyroid cancer

Abstract

Thyroid stimulating hormone (TSH) is critical for normal development and metabolism. To better understand the genetic contribution to TSH levels, we conduct a GWAS meta-analysis at 22.4 million genetic markers in up to 119,715 individuals and identify 74 genome-wide significant loci for TSH, of which 28 are previously unreported. Functional experiments show that the thyroglobulin protein-altering variants P118L and G67S impact thyroglobulin secretion. Phenome-wide association analysis in the UK Biobank demonstrates the pleiotropic effects of TSH-associated variants and a polygenic score for higher TSH levels is associated with a reduced risk of thyroid cancer in the UK Biobank and three other independent studies. Two-sample Mendelian randomization using TSH index variants as instrumental variables suggests a protective effect of higher TSH levels (indicating lower thyroid function) on risk of thyroid cancer and goiter. Our findings highlight the pleiotropic effects of TSH-associated variants on thyroid function and growth of malignant and benign thyroid tumors.

Fig. 1. Both the TG-P118L and TG-G67S point mutants exhibit a secretion defect.

a–c Three independent replicate experiments. Western blotting of TG in 293T cells that were either untransfected (a, no detectable bands) or transfected with constructs encoding mouse TG wild type (WT) or P118L or G67S point mutants (in the pcDNA3.1 background, in which the CMV promoter drives the respective cDNA expression). Serum-free media (M) were collected overnight and the cells (C) were lysed. Equal volumes of media and cells were analyzed by SDS-PAGE, electrotransfer to nitrocellulose, and immunoblotting with anti-Tg-specific antibodies. Full scans of western blotting are presented in Supplementary Fig. 14. From scanning densitometry, d shows the content of thyroglobulin (and its variants) intracellularly and in the secretion. e The extracellular : intracellular (M/C) ratio of each construct. d, e Three independent replicate experiments. All boxplots in d and e indicate median (center line), 25th and 75th percentiles (bounds of box), and minimum and maximum (whiskers).

Fig. 2. The risk of thyroid cancer is lower for individuals with genetically predicted higher TSH levels.

Plots of thyroid cancer prevalence by quintiles of TSH PGS (left) and odds ratio of thyroid cancer in relation to the lowest quintile (right) in data sets UKBB (a; N case = 358, N control = 407,399), deCODE (b; N case = 1003, N control = 278,991), and FinnGen (c; N case = 501, N control = 135,137). N: sample size. N case: sample size of cases. N control: sample size of controls. Error bars represent 95% confidence intervals.

Fig. 3. Two-sample Mendelian randomization analysis for casual associations between TSH and thyroid cancer and between TSH and goiter.

a. Two-sample MR between TSH and thyroid cancer based on summary statistics from the meta-analysis of HUNT and ThyroidOmics (n = 106,360) for TSH and from the meta-analysis of UKBB^,, MGI, deCODE and four other case–control data sets with European ancestry as reported in Gudmundsson et al. for thyroid cancer association (cases/controls = 4,146/731,903). b TSH and goiter based on summary statistics from the same TSH meta-analysis as above and from the GWAS of UKBB^, for goiter association (N case = 1,143, N control = 391,429) The crosshairs on the plots represent the 95% confidence intervals for each SNP-TSH or SNP-outcome association. The variant on the top left corner is the rare nonsynonymous variant B4GALNT3 p.R724W (rs145153320). N: sample size. N case: sample size of cases. N control: sample size of controls.