Thyroid function, physical activity and sedentary behaviour: A bidirectional two-sample Mendelian randomisation study

Abstract

Background: The interinfluence of thyroid function and daily physical activity (PA) remains unclear. We examined the causal relationship between genetically proxied thyroid-related traits; hypothyroidism, hyperthyroidism, thyroid stimulating hormone (TSH) and free thyroxine (FT4), and daily PA measures; leisure screen time (LST) and moderate-to-vigorous physical activity (MVPA), using Mendelian randomisation (MR) analysis.

Methods: We used genome-wide association study (GWAS) data from the ThyroidOmics Consortium and the most comprehensive meta-analysis on PA, comprising data on hypothyroidism (n = 53 423), hyperthyroidism (n = 51 823), TSH within the reference range (n = 54 288), fT4 within the reference range (n = 49 269), LST (n = 526 725), and MVPA (n = 608 595) to conduct a bidirectional two-sample MR analysis. The inverse variance weighted (IVW) method was employed as the primary result. Sensitivity analyses included MR-Egger, weighted median, and MR pleiotropy residual sum and outlier (MR-PRESSO) regression. Similar investigations were conducted in the reverse direction. Finally, we analysed a multivariable MR using body mass index (BMI)-related traits GWAS data.

Results: In the primary IVW analysis, an increase in genetically proxied TSH levels significantly increased LST (correlation coefficient (β) = 0.040; 95% confidence interval (CI) = 0.020-0.061, P = 9.776 × 10^-5). The multivariable MR analysis indicated that the positive causal effect still existed when considering the influence of BMI (MVMR-IVW: β = 0.042; 95% CI = 0.011-0.073, P = 0.007). Conversely, there was no evidence to suggest that PA impacts thyroid function.

Conclusions: The results of this MR analysis suggest that thyroid function influences daily PA. The positive association between TSH and LST is not confounded or mediated by BMI.

Figure 1

Study design and hypothesis overview. FT4 – free thyroxine, MR – Mendelian randomisation, MR-PRESSO – Mendelian randomisation pleiotropy residual sum and outlier; SNP – single nucleotide polymorphism, TSH – thyrotropin.

Figure 2

Flowchart of the selection process for instrumental variables and step-by-step execution analysis. Step 1: MR analysis with the complete selected SNPs. Step 2: MR analysis after removing the SNPs with FALSE in the MR Steiger test of directionality. Step 3: MR analysis after removing the SNPs with P-value less than threshold (0.05) in the MR-PRESSO outlier test. Step 4: MR analysis after removing all the SNPs with P-value less than 1 in the MR-PRESSO outlier test. MR – Mendelian randomisation, MR-PRESSO – Mendelian randomisation pleiotropy residual sum and outlier, SNP – single nucleotide polymorphism.

Figure 3

Univariate MR analysis evaluating the causal effects of thyroid-related traits on LST and MVPA. Step 1: MR analysis with the complete selected SNPs. Step 2: MR analysis after removing the SNPs with P value<0.05 in the MR-PRESSO outlier test. Step 4: MR analysis after removing all the SNPs with P-value <1 in the MR-PRESSO outlier test. FT4 – free thyroxine, LST – leisure screen time, IVW – Inverse variance weighted, MR – Mendelian randomisation, MR-PRESSO – Mendelian randomisation pleiotropy residual sum and outlier, MVPA – moderate-to-vigorous intensity physical activity during leisure time, SNPs – single nucleotide polymorphisms, TSH – thyrotropin. *P < 0.05. †P < 0.01. ‡P < 0.001.

Figure 4

Scatter plots, funnel plots, and plots of leave-one-out analysis for MR analysis of the causal effect between TSH and LST. Panel A. Scatter plots for MR analysis of the causal effect between TSH and LST. Panel B. Funnel plots for MR analysis of the causal effect between TSH and LST. Panel C. Plots of leave-one-out analysis for MR analysis of the causal effect between TSH and LST. LST – leisure screen time, MR – Mendelian randomisation, TSH – thyrotropin.

Figure 5

Univariate and multivariable MR analyses evaluating the causal effects of TSH and BMI on LST. BMI – body mass index, LST – leisure screen time, IVW – Inverse variance weighted, MR – Mendelian randomisation, MR-PRESSO – Mendelian randomisation pleiotropy residual sum and outlier, TSH – thyrotropin. *P < 0.05. †P < 0.01. ‡P < 0.001.