Multi-trait analysis characterizes the genetics of thyroid function and identifies causal associations with clinical implications

Abstract

To date only a fraction of the genetic footprint of thyroid function has been clarified. We report a genome-wide association study meta-analysis of thyroid function in up to 271,040 individuals of European ancestry, including reference range thyrotropin (TSH), free thyroxine (FT4), free and total triiodothyronine (T3), proxies for metabolism (T3/FT4 ratio) as well as dichotomized high and low TSH levels. We revealed 259 independent significant associations for TSH (61% novel), 85 for FT4 (67% novel), and 62 novel signals for the T3 related traits. The loci explained 14.1%, 6.0%, 9.5% and 1.1% of the total variation in TSH, FT4, total T3 and free T3 concentrations, respectively. Genetic correlations indicate that TSH associated loci reflect the thyroid function determined by free T3, whereas the FT4 associations represent the thyroid hormone metabolism. Polygenic risk score and Mendelian randomization analyses showed the effects of genetically determined variation in thyroid function on various clinical outcomes, including cardiovascular risk factors and diseases, autoimmune diseases, and cancer. In conclusion, our results improve the understanding of thyroid hormone physiology and highlight the pleiotropic effects of thyroid function on various diseases.

Fig. 1. Schematic design of the project and analyses.

a The hypothalamic-pituitary-thyroid axis is characterized by a negative feedback loop. The hypothalamus produces thyrotropin releasing hormone (TRH), which stimulates the pituitary to produce thyroxine-stimulating hormone (TSH). TSH stimulates the thyroid to produce thyroxine (T4) and triiodothyronine (T3), the active thyroid hormone affecting transcription in target cells. The majority of circulating T3 is produced by the liver and kidney by T4 to T3 conversion. b Step 1 represents the meta-analysis of 46 different European ancestry cohorts for eight thyroid function traits: TSH, FT4, FT3, TT3, FT3/FT4 ratio, TT3/FT4 ratio, high and low TSH. Step 2 shows the different secondary analyses performed using the meta-analyses results to identify the underlying mechanisms of the specific genome-wide variants and the translation to clinical diagnoses.

Fig. 2. Genome wide association results for TSH, FT4, FT3 and FT3/FT4 ratio.

The circos plot depicts the association results for TSH, FT4, FT3 and the FT3/FT4 ratio combined: red band: –log₁₀(p) for association in the meta-analysis of TSH, ordered by chromosomal position. The blue line indicates genome-wide significance (p = 5 × 10⁻⁸). Blue band: –log₁₀(p) for association with FT4, ordered by chromosomal position. The red line indicates genome-wide significance (p = 5 × 10⁻⁸). Purple band: –log₁₀(p) for association with FT3, ordered by chromosomal position. The blue line indicates genome-wide significance (p = 5 × 10⁻⁸). Green band: –log₁₀(p) for association with the FT3/FT4 ratio, ordered by chromosomal position. The red line indicates genome-wide significance (p = 5 × 10⁻⁸). The outer band indicates the positions of the associated loci as defined in Methods. Adjacent loci for a trait with the same gene names are merged. The color follows the same pattern as the association plots of the four traits. All p-values were obtained from two-sided association tests (z-statistics), where correction for multiple testing is indicated by the level of genome-wide significance.

Fig. 3. Zoomed Manhattan plot for TSH and FT4.

Zoomed Manhattan plot of the GWAS meta-analysis results for TSH (panel a) and FT4 (panel b). Variants are plotted on the x-axis according to their position on each chromosome with the -log₁₀(p-value) of the association test on the y-axis. The horizontal line indicates the threshold for genome-wide significance, (p = 5 × 10⁻⁸). All p-values were obtained from two-sided association tests (z-statistics), where correction for multiple testing is indicated by the level of genome-wide significance. Novel loci are colored in orange, and novel independent associations within known loci are colored in light blue. Genetic variants were assigned to the nearest gene. Variants were considered known when they are in linkage disequilibrium with a previously identified variant (see Methods).

Fig. 4. Genetic correlations of thyroid hormone parameters.

Pairwise genetic correlations were estimated via bivariate LD score regression. In the upper part, positive genetic correlations are shown in blue, and negative correlations are depicted in red as indicated by the legend. The lower part shows the genetic correlation values. FDR was calculated via the Benjamini–Hochberg method to correct for multiple testing of all 15 correlations. Larger squares correspond to stronger genetic correlation. Significant correlations are indicated by asterisks (FDR: * <0.05, ** <0.001, *** <0.0001).

Fig. 5. Colocalization of associations for thyroid function parameters and gene expression.

In panel a, the different thyroid function traits are shown on the y-axis and the tested tissues (n = 49) derived from the GTEx database are shown on the x-axis. The number of significant colocalizations between thyroid function genome-wide significant variants and gene expression in the different tissues are shown in each box using a probability (P12) > 0.85 to confirm the H4 hypothesis (same shared causal variant). Tissue names are similarly colored when present in the same organ or belonging to the same group of tissues (blue or black). Detailed results of the colocalizations of TSH (panel b) and FT4 (panel c) are shown in the tissues of the hypothalamus-pituitary-thyroid axis.

Fig. 6. Effects of genetic variants on different thyroid-related outcomes using Mendelian Randomization.

Twenty-four clinical outcomes were tested against the thyroid-related parameters. Significance thresholds are depicted either with a closed bubble (Bonferroni corrected) or an open bubble (nominal significance at p < 0.05). The direction of effect is depicted in color: blue shows a negative beta and red is a positive beta. The size of the bubble indicates the degree of the association in terms of -log₁₀(p-value). All p-values were obtained from two-sided weighted median Mendelian Randomization tests. Clinical outcomes with at least one association with thyroid function parameters are depicted. Bone mineral density, anxiety, and major depressive disorder with zero associations are therefore not shown.

Fig. 7. Polygenic risk score association results.

Significant associations between thyroid function parameter polygenic scores (PGSs) and diseases in the UKBB. The data points are color-coded by trait (legend) by phenotype groups (x-axis) and -log₁₀(p-value) obtained from two-sided association tests (z-statistics) together with the direction of the effect (y-axis). All results shown passed the Bonferroni significance threshold (p < 0.05/1460).

Fig. 8. Thyroid function polygenic risk scores and the risk of thyroid cancer.

Plots of the associations between thyroid parameter polygenic scores TSH (panel a) and FT3 (panel b) and thyroid cancer in deCODE (a; N_case = 620, N_control = 106,168, b; N_case = 686, N_control = 119,187). The y-axis shows the probability of thyroid cancer. The x-axis shows the percentage of risk alleles carried out based on a weighted polygenic score. The histogram shows the distribution of the polygenic score in the study sample per trait. N_case: sample size of cases. N_control: sample size of controls.