Background

Maternal subclinical hypothyroidism in pregnancy is associated with impaired neuropsychological development of the child (1) and a number of other adverse outcomes for both mother and offspring (2, 3). The Endocrine Society's Clinical Practice Guidelines (4) recommend treatment with T4 for women found to have subclinical hypothyroidism, defined by a serum TSH concentration above the upper limit of the trimester-specific and assay-specific reference range, with a normal free T4 (FT4). This treatment has not been proven to improve neurological development of the child. However, the potential benefits are considered to outweigh the potential risks.

Using data on healthy pregnant women from the Exeter Family Study (EFSOCH), we recently showed that the PDE8B rs4704397 polymorphism, known to strongly associate with TSH levels in the general population, is associated with subclinical hypothyroidism in pregnancy(5). More recently, the international thyroid consortium has used meta-analysis of genome-wide association studies to identify a total of 10 genetic loci that are robustly associated with TSH levels, 9 of which are not associated with other phenotypes. Only 3 of these loci have been published. The manuscript of the latest meta-analysis is in preparation. Pre-publication analysis of these 9 SNPs in the EFSOCH pregnant women shows that the SNPs together explain 4.4% of the variation in maternal TSH levels at 28 weeks of gestation. Each additional unit of an allele score composed of the 9 SNPs was associated with a 0.1SD higher TSH level in pregnancy (P=8e-12; Figure 1; Table 1). The result was essentially unchanged when excluding women who tested +ve for TPO antibodies or when adjusting for child's genotype at the 9 loci.

Table 1: Association between allele score composed of 9 TSH-altering SNPs and maternal TSH levels in pregnancy (unpublished EFSOCH data).

Model*

N

Beta (per TSH-raising allele), SD units

SE

P-value

Allele count

813

0.121

0.018

4E-11

Allele score**

858

0.104

0.015

8E-12

Allele score TPO-Ab negative women only

794

0.113

0.015

1E-13

Allele score adjusted child genotypes

561

0.104

0.023

8E-06

*Linear regression with outcome =TSH level in pregnancy (inverse-normal transformation), with age and age-squared as covariables, excluding women on TF medication and non-Eur.

**Allele_score=Weighted_score*N_available_SNPs/Sum_weights_of_available_SNPs, where Weighted_score=W1*SNP1+W2*SNP2+W3*SNP3...etc. Weighted by effect size in EFSOCH study. Included women with up to 2/9 SNPs missing.

MAIN RESEARCH QUESTION:

Are SNPs which explain 4.4% of the variance in TSH levels in pregnancy associated with offspring birth weight and gestational age?

Aim and hypothesis

We propose to use this allele score and available phenotypes in ALSPAC to test the hypothesis that exposure to higher TSH levels in utero is associated with offspring birth weight and length of gestation. Rare, monogenic nonautoimmune hypothyroidism is associated with preterm birth and low birth weight (6).

To maximise statistical power, we will meta-analyse the ALSPAC data with data from the EFSOCH and HAPO studies. Using the measured TSH levels in the EFSOCH study, we will use the triangulation method to approximate a Mendelian randomisation analysis.