https://orcid.org/0000-0002-3436-422X
Abstract
Thyroid dysfunction in pregnancy is associated with adverse offspring outcomes and recent birth-cohort studies suggest that even mild degrees of thyroid dysfunction may be linked with a range of late cognitive and behavioural effects in childhood and adolescence.
This review summarizes recent literature of observational studies and critically appraises randomized controlled trials (RCTs) of antenatal thyroid screening and Levothyroxine intervention.
Overt hypothyroidism and hyperthyroidism carry significant risks for unfavourable offspring outcomes and should be appropriately corrected in pregnancy.
The significance of subclinical hypothyroidism and hypothyroxinaemia is still unclear. Meta-analyses of birth-cohort studies show associations of maternal subclinical hypothyroidism and hypothyroxinaemia with intellectual deficits, attention deficit hyperactivity disorder (ADHD) and autism spectrum disorders, while hyperthyroidism and high-normal FT4 were linked with ADHD. RCTs have shown no benefits of screening on neurodevelopmental outcomes although Levothyroxine could have been initiated too late in pregnancy in these trials.
A small number of studies have shown inconsistent associations of maternal thyroid dysfunction with offspring cardiometabolic indices including blood pressure and body weight. Correction of maternal thyroid dysfunction was, however, associated with favourable long-term metabolic profiles in mothers, including lipid profiles, fat mass and body mass index. Antenatal thyroid screening may therefore present opportunities for optimizing a wider range of outcomes than envisaged.
Future trials with early antenatal thyroid screening and intervention are necessary to clarify the impact of screening on late offspring and maternal effects.
Introduction
Thyroid hormones are essential for optimal foetal growth and development.1 The developing foetus is wholly dependent on maternal thyroid hormone sources up until about 18–20 weeks’ gestation when its own thyroid gland becomes functional.1,2 Thus, in early embryonic life adequate transplacental delivery of maternal thyroid hormones is necessary for critical neuro-developmental processes such as neuronal migration, proliferation and neural tube formation.2,3 In the foetal brain, the active thyroid hormone, triiodothyronine (T3) binds to its nuclear receptor through which it regulates gene expression in the brain.3 In addition, normal pregnancy demands a series of maternal adaptations designed to ensure adequate foetal thyroid hormone delivery in the face of increased iodine requirements and peripheral thyroid hormone metabolism.4 While these adaptations occur seamlessly in euthyroid pregnancies, women with thyroid dysfunction may exhibit blunted adaptive responses that ultimately impair foetal thyroid hormone delivery.4 Inadequate thyroid hormone delivery may have permanent sequelae for the offspring ranging from mild cognitive deficits to more severe intellectual impairment as seen in children with undiagnosed congenital hypothyroidism5 or severe iodine deficiency.6
The aetiology, prevalence and laboratory diagnoses of the various forms of thyroid dysfunction in pregnancy are summarized in Table 1. The laboratory definitions of thyroid dysfunction in pregnancy are ideally based on trimester-specific, population-specific and assay-specific reference ranges for FT4 and TSH.7 Maternal hypothyroidism is associated with an increased risk of foetal loss, preterm birth and offspring cognitive impairment while hyperthyroidism is linked with preeclampsia, low birth weight and foetal morbidity and mortality.8 These associations are well established for overt disease and are also reported with borderline states of thyroid dysfunction namely subclinical hypothyroidism (elevated TSH and normal FT4) and isolated hypothyroxinaemia (FT4 in the lowest 2.5th percentile of the pregnancy reference range).8,9 Subclinical hyperthyroidism on the other hand is not typically associated with adverse outcomes and does not warrant specific treatment as in most cases it represents a transient physiological state or gestational thyrotoxicosis due to the thyroid stimulatory properties of human chorionic gonadotrophin.10 There is a long-held consensus that overt hypothyroidism and hyperthyroidism should be corrected in pregnancy and therefore outcome studies in untreated overt maternal disease are lacking as it would be unethical not to correct these conditions in the light of current knowledge.4,11 Thus, our understanding of the adverse effects of maternal thyroid dysfunction is largely based on cohorts with subclinical hypothyroidism or hypothyroxinaemia, conditions for which there still exists uncertainty regarding treatment and outcomes.
Causes of thyroid dysfunction in pregnancy
| Low TSH | Normal TSH | High TSH | |
|---|---|---|---|
| High FT4 | Overt hyperthyroidism (0.1–0.5%) • Gestational transient thyrotoxicosis • Graves’ disease • Toxic nodular goitre • Thyroiditis | High FT4, Normal TSH (rare) • Assay interference • TSH secreting adenoma • Thyroid hormone resistance | High FT4, high TSH (rare) • Assay interference • TSH secreting adenoma • Thyroid hormone resistance |
| Normal FT4 | Subclinical hyperthyroidism (1–4%) • Physiological • Gestational transient thyrotoxicosis • Graves’ disease • Toxic nodular goitre • Molar pregnancy | Normal thyroid function (85–95%) | Subclinical hypothyroidism (2–5%) • Hashimoto’s thyroiditis • Post-surgical • Post-radioiodine |
| Low FT4 | Low FT4, low TSH (rare) • Non-thyroidal illness • Central hypothyroidism • Assay interference | Hypothyroxinaemia (2.5%) • Iodine deficiency • Assay problems • Idiopathic | Overt hypothyroidism (0.2–0.5%) • Hashimoto’s thyroiditis • Post-surgical • Post-radioiodine • Iodine deficiency |
| Low TSH | Normal TSH | High TSH | |
|---|---|---|---|
| High FT4 | Overt hyperthyroidism (0.1–0.5%) • Gestational transient thyrotoxicosis • Graves’ disease • Toxic nodular goitre • Thyroiditis | High FT4, Normal TSH (rare) • Assay interference • TSH secreting adenoma • Thyroid hormone resistance | High FT4, high TSH (rare) • Assay interference • TSH secreting adenoma • Thyroid hormone resistance |
| Normal FT4 | Subclinical hyperthyroidism (1–4%) • Physiological • Gestational transient thyrotoxicosis • Graves’ disease • Toxic nodular goitre • Molar pregnancy | Normal thyroid function (85–95%) | Subclinical hypothyroidism (2–5%) • Hashimoto’s thyroiditis • Post-surgical • Post-radioiodine |
| Low FT4 | Low FT4, low TSH (rare) • Non-thyroidal illness • Central hypothyroidism • Assay interference | Hypothyroxinaemia (2.5%) • Iodine deficiency • Assay problems • Idiopathic | Overt hypothyroidism (0.2–0.5%) • Hashimoto’s thyroiditis • Post-surgical • Post-radioiodine • Iodine deficiency |
Causes of thyroid dysfunction in pregnancy
| Low TSH | Normal TSH | High TSH | |
|---|---|---|---|
| High FT4 | Overt hyperthyroidism (0.1–0.5%) • Gestational transient thyrotoxicosis • Graves’ disease • Toxic nodular goitre • Thyroiditis | High FT4, Normal TSH (rare) • Assay interference • TSH secreting adenoma • Thyroid hormone resistance | High FT4, high TSH (rare) • Assay interference • TSH secreting adenoma • Thyroid hormone resistance |
| Normal FT4 | Subclinical hyperthyroidism (1–4%) • Physiological • Gestational transient thyrotoxicosis • Graves’ disease • Toxic nodular goitre • Molar pregnancy | Normal thyroid function (85–95%) | Subclinical hypothyroidism (2–5%) • Hashimoto’s thyroiditis • Post-surgical • Post-radioiodine |
| Low FT4 | Low FT4, low TSH (rare) • Non-thyroidal illness • Central hypothyroidism • Assay interference | Hypothyroxinaemia (2.5%) • Iodine deficiency • Assay problems • Idiopathic | Overt hypothyroidism (0.2–0.5%) • Hashimoto’s thyroiditis • Post-surgical • Post-radioiodine • Iodine deficiency |
| Low TSH | Normal TSH | High TSH | |
|---|---|---|---|
| High FT4 | Overt hyperthyroidism (0.1–0.5%) • Gestational transient thyrotoxicosis • Graves’ disease • Toxic nodular goitre • Thyroiditis | High FT4, Normal TSH (rare) • Assay interference • TSH secreting adenoma • Thyroid hormone resistance | High FT4, high TSH (rare) • Assay interference • TSH secreting adenoma • Thyroid hormone resistance |
| Normal FT4 | Subclinical hyperthyroidism (1–4%) • Physiological • Gestational transient thyrotoxicosis • Graves’ disease • Toxic nodular goitre • Molar pregnancy | Normal thyroid function (85–95%) | Subclinical hypothyroidism (2–5%) • Hashimoto’s thyroiditis • Post-surgical • Post-radioiodine |
| Low FT4 | Low FT4, low TSH (rare) • Non-thyroidal illness • Central hypothyroidism • Assay interference | Hypothyroxinaemia (2.5%) • Iodine deficiency • Assay problems • Idiopathic | Overt hypothyroidism (0.2–0.5%) • Hashimoto’s thyroiditis • Post-surgical • Post-radioiodine • Iodine deficiency |
Antenatal thyroid screening represents an attractive public health intervention for improving foetal and maternal health given that thyroid dysfunction can be readily detected and safely corrected in pregnancy.12 However, a limited number of randomized controlled trials (RCTs) have failed to show benefits of universal screening on pregnancy and offspring outcomes.13,14 An important consideration for screening is the potential for late offspring effects of maternal thyroid dysfunction, i.e. effects extending beyond pregnancy into childhood. While the association between maternal thyroid function and offspring cognitive function has been recognized for decades,15 more recent data are now revealing intriguing associations with a broader range of offspring characteristics including cognition,16 neuropsychiatric behaviour17 and metabolic indices.18 The advent of large longitudinal birth cohorts, including observational and RCT cohorts, has provided fresh perspectives on offspring effects even into the late childhood and adolescence years. In this review, we examine current understanding of these late effects and address their implications for antenatal thyroid screening.
Methods
We conducted a PubMed search for articles published from database inception to December 2021, using various combinations of the keywords ‘hypothyroidism’, ‘subclinical hypothyroidism’, ‘isolated hypothyroxinaemia’, ‘levothyroxine’, ‘pregnancy’, ‘gestation’, ‘screening’, ‘hyperthyroidism’, ‘FT4’, ‘TSH OR thyrotropin’ and ‘offspring OR child OR progeny’. We did not address the obstetric effects of thyroid dysfunction as our focus in this review was on the association between maternal thyroid dysfunction and post-delivery or late outcomes as well as the potential impact of systematic thyroid screening on these outcomes. Our preference was for RCTs, well-designed observational studies and the most recent meta-analyses that addressed these topics.
Offspring cognition
Observational studies
A possible association between maternal suboptimal thyroid function and offspring intelligence has been recognized for decades.15,19 A landmark publication in 1999 by Haddow et al. showed that offspring of women with hypothyroidism in early pregnancy had a 7-point deficit in intelligent quotient (IQ) scores at 8 years compared to the children of matched mothers without hypothyroidism.15 This study comprised women with overt and subclinical hypothyroidism with a mean TSH of 13.2 mU/L in the cohort. Subsequent studies also reported an increased risk of offspring neuro-intellectual impairment in women with either overt or subclinical hypothyroidism although an association was not confirmed in all studies.16 For example, an analysis of a large Finnish birth cohort by Pakkila et al. showed no association between thyroid function and intellectual performance even though children of women with thyroid dysfunction were more likely to repeat a class year in school.20 Another study of 4615 mother–child pairs from the UK Avon Longitudinal Study of Parents and Children (ALSPAC) by Nelson et al. also found no association between first trimester TSH and educational attainment up till age 15 years.21
With respect to isolated hypothyroxinaemia, early studies by Pop et al. in an iodine-sufficient area of the Netherlands reported an 8-point deficit in the cognitive scores of 3-year-old children of women with FT4 below the tenth percentile in early pregnancy.19 A more recent study from the Netherlands involving 3727 mother–child pairs from the Generation-R study cohort showed that 6-year-old offspring of women with isolated hypothyroxinaemia (FT4 < 5th percentile) exhibited non-verbal IQ scores that were 4.3 points lower than that of children of euthyroid mothers.22 A subsequent analysis of the same Generation-R cohort showed an inverted U-shaped association between FT4 and child IQ with both low and high maternal FT4 associated with lower child IQ as well as lower brain grey matter and cortical volume on MRI.23 A recent study in a Greek iodine-sufficient population showed that offspring exposed to maternal hypothyroxinaemia exhibited reduced perceptual performance and motor function at 4 years of age.24 In contrast to the above studies, no relationship between maternal hypothyroxinaemia in early pregnancy and child scholastic performance was demonstrated in the Finnish20 and UK ALSPAC cohort.21
The results of several meta-analyses, however, suggest an overall effect of thyroid dysfunction on offspring cognition. These studies are summarized in Figure 1.16,17,25 A meta-analysis of individual participant data from 9036 mother–child pairs from three birth cohorts from Spain, Netherlands and the UK showed that children exposed to gestational hypothyroxinaemia had lower non-verbal and verbal IQ scores.25 An association with TSH was, however, lacking in this study. In addition, an aggregate data meta-analysis of 11 cohort studies also showed an increased risk of neuro-intellectual impairment in the children of women with gestational subclinical hypothyroidism16 (Fig. 1). However, there was significant heterogeneity across studies with differences in the thresholds for thyroid dysfunction, timing of sampling in pregnancy, the age of child assessments and the tools used in evaluating cognitive function.
Association between maternal thyroid dysfunction and neurodevelopmental outcomes: data from published meta-analyses of observational studies. Legend: Forest plots indicate odds ratios and 95% confidence intervals from published meta-analyses.16,17,25 The most recent or comprehensive meta-analyses are presented.
RCTs
Despite the numerous observational studies, data from prospective interventional controlled trials are limited. Two major RCTs have evaluated the impact of antenatal thyroid screening on neurodevelopmental outcomes, resulting in several research outputs, as summarized in Table 2.13,14,26,27 In the first large RCT, the Controlled Antenatal Thyroid Screening (CATS) study in 2002, Lazarus et al. recruited almost 22 000 women with no previous history of thyroid disease at a median gestation of 12 weeks and 3 days from the UK and Italy.13 Participants were divided into screening and control groups, in which thyroid function was either assessed immediately or samples were stored and tested after delivery, respectively. Women in the screening group with TSH >97.5th percentile and/or FT4 < 2.5th percentile were started on 150 mcg of Levothyroxine daily. Data on offspring IQ at 3 years of age showed no significant difference in IQ scores or in the proportion of children with IQ < 85 between the treated and untreated groups. The second trial, a United States National Institutes of Health (NIH) study by Casey et al., was designed as two parallel trials for gestational subclinical hypothyroidism and isolated hypothyroxinaemia.14 Thyroid hormone replacement was initiated at a median gestational age of 16 weeks 4 days for the subclinical hypothyroidism trial and at 18 weeks for the isolated hypothyroxinaemia trial with developmental testing in children undertaken annually for 5 years post-delivery. Similar to the CATS study, Levothyroxine did not improve IQ or other neurodevelopmental indices in either of the cohorts compared to placebo at 3–5 years of age.
Summary of RCTs of antenatal thyroid screening on neurodevelopmental outcomes
| Author, year, country | Intervention and control groups | Gestational age at intervention | Levothyroxine, dose per day | Assessment, tools | Age at evaluation | Conclusion |
|---|---|---|---|---|---|---|
| Offspring IQ | ||||||
| Lazarus, 2012, UK and Italy13 | SCH and/or hypothyroxinaemia treated (N = 303) vs. control (N = 306) | 12.4 weeks | 150 mcg, target TSH 0.1–1.0 mU/L | WPPSI-III | 3 years | No difference in IQ scores or % with IQ < 85 |
| Casey, 2017, USA14 | SCH treated (N = 323) vs. control, (N = 326) | 16.6 weeks | 100 mcg, target TSH 0.1–2.5 mU/L | WPPSI-III, DAS-II | 3–5 years | No difference in IQ scores between treated and control |
| Casey, 2017, USA14 | Hypothyroxinaemia treated (N = 254) vs. control (N = 253) | 18 weeks | 50 mcg, target FT4 0.86–1.9 ng/dl | WPPSI-III, DAS-II | 3–5 years | No difference in IQ scores in treated vs. control groups |
| Hales, 2018, UK26 | RCT follow-up: SCH and/or hypothyroxinaemia treated (N = 119), untreated (N = 98), normal gestational thyroid function (GTF) (N = 232) | 12.4 weeks | 150 mcg, target TSH 0.1–1.0 mU/L | WISC | 9.5 years | No difference in % with IQ <85 in normal GTF vs. merged treated and untreated |
| Offspring behaviour | ||||||
| Lazarus, 2012, UK and Italy13 | SCH and/or hypothyroxinaemia treated (N = 303) vs. control (N = 306) | 12.4 weeks | 150 mcg, target TSH 0.1–1.0 mU/L | CBCL, Brief-P | 3 years | No difference in behaviour scores between groups |
| Casey, 2017, USA14 | SCH treated (N = 323) vs. control, (N = 326) | 16.6 weeks | 100 mcg, target TSH 0.1–2.5 mU/L | CBCL | 3–5 years | No difference in behaviour scores between groups |
| Casey, 2017, USA14 | Hypothyroxinaemia treated (N = 254) vs. control (N = 253) | 18 weeks | 50 mcg, target FT4 0.86–1.9 ng/dl | CBCL | 3–5 years | No difference in behaviour scores between groups |
| Hales, 2020, UK27 | RCT follow-up: SCH and/or hypothyroxinaemia treated (N = 125), untreated (N = 104), normal gestational thyroid function (GTF, N = 246) | 12.4 weeks | 150 mcg, target TSH 0.1–1.0 mU/L | SDQ, SCQ, ADHDQ | 9.5 years | No difference in behaviour scores between groups; higher scores in over-treated vs. normal GTF |
| Author, year, country | Intervention and control groups | Gestational age at intervention | Levothyroxine, dose per day | Assessment, tools | Age at evaluation | Conclusion |
|---|---|---|---|---|---|---|
| Offspring IQ | ||||||
| Lazarus, 2012, UK and Italy13 | SCH and/or hypothyroxinaemia treated (N = 303) vs. control (N = 306) | 12.4 weeks | 150 mcg, target TSH 0.1–1.0 mU/L | WPPSI-III | 3 years | No difference in IQ scores or % with IQ < 85 |
| Casey, 2017, USA14 | SCH treated (N = 323) vs. control, (N = 326) | 16.6 weeks | 100 mcg, target TSH 0.1–2.5 mU/L | WPPSI-III, DAS-II | 3–5 years | No difference in IQ scores between treated and control |
| Casey, 2017, USA14 | Hypothyroxinaemia treated (N = 254) vs. control (N = 253) | 18 weeks | 50 mcg, target FT4 0.86–1.9 ng/dl | WPPSI-III, DAS-II | 3–5 years | No difference in IQ scores in treated vs. control groups |
| Hales, 2018, UK26 | RCT follow-up: SCH and/or hypothyroxinaemia treated (N = 119), untreated (N = 98), normal gestational thyroid function (GTF) (N = 232) | 12.4 weeks | 150 mcg, target TSH 0.1–1.0 mU/L | WISC | 9.5 years | No difference in % with IQ <85 in normal GTF vs. merged treated and untreated |
| Offspring behaviour | ||||||
| Lazarus, 2012, UK and Italy13 | SCH and/or hypothyroxinaemia treated (N = 303) vs. control (N = 306) | 12.4 weeks | 150 mcg, target TSH 0.1–1.0 mU/L | CBCL, Brief-P | 3 years | No difference in behaviour scores between groups |
| Casey, 2017, USA14 | SCH treated (N = 323) vs. control, (N = 326) | 16.6 weeks | 100 mcg, target TSH 0.1–2.5 mU/L | CBCL | 3–5 years | No difference in behaviour scores between groups |
| Casey, 2017, USA14 | Hypothyroxinaemia treated (N = 254) vs. control (N = 253) | 18 weeks | 50 mcg, target FT4 0.86–1.9 ng/dl | CBCL | 3–5 years | No difference in behaviour scores between groups |
| Hales, 2020, UK27 | RCT follow-up: SCH and/or hypothyroxinaemia treated (N = 125), untreated (N = 104), normal gestational thyroid function (GTF, N = 246) | 12.4 weeks | 150 mcg, target TSH 0.1–1.0 mU/L | SDQ, SCQ, ADHDQ | 9.5 years | No difference in behaviour scores between groups; higher scores in over-treated vs. normal GTF |
SCH, subclinical hypothyroidism; WPPSI-III, Wechsler Preschool and Primary Scale of Intelligence, 3rd edition; DAS-II, Differential Ability Scales–II; WISC, Wechsler Intelligence Scale for Children, 4th edition UK; CBCL, Child Behaviour Checklist 2000; Brief-P, Behaviour Rating Inventory of Executive Function, preschool version; SDQ, Strengths and Difficulties Questionnaire; SCQ, Social Communication Questionnaire (SCQ); ADHDQ, Child Attention Deficit Hyperactivity Disorder Questionnaire; mcg, micrograms.
Summary of RCTs of antenatal thyroid screening on neurodevelopmental outcomes
| Author, year, country | Intervention and control groups | Gestational age at intervention | Levothyroxine, dose per day | Assessment, tools | Age at evaluation | Conclusion |
|---|---|---|---|---|---|---|
| Offspring IQ | ||||||
| Lazarus, 2012, UK and Italy13 | SCH and/or hypothyroxinaemia treated (N = 303) vs. control (N = 306) | 12.4 weeks | 150 mcg, target TSH 0.1–1.0 mU/L | WPPSI-III | 3 years | No difference in IQ scores or % with IQ < 85 |
| Casey, 2017, USA14 | SCH treated (N = 323) vs. control, (N = 326) | 16.6 weeks | 100 mcg, target TSH 0.1–2.5 mU/L | WPPSI-III, DAS-II | 3–5 years | No difference in IQ scores between treated and control |
| Casey, 2017, USA14 | Hypothyroxinaemia treated (N = 254) vs. control (N = 253) | 18 weeks | 50 mcg, target FT4 0.86–1.9 ng/dl | WPPSI-III, DAS-II | 3–5 years | No difference in IQ scores in treated vs. control groups |
| Hales, 2018, UK26 | RCT follow-up: SCH and/or hypothyroxinaemia treated (N = 119), untreated (N = 98), normal gestational thyroid function (GTF) (N = 232) | 12.4 weeks | 150 mcg, target TSH 0.1–1.0 mU/L | WISC | 9.5 years | No difference in % with IQ <85 in normal GTF vs. merged treated and untreated |
| Offspring behaviour | ||||||
| Lazarus, 2012, UK and Italy13 | SCH and/or hypothyroxinaemia treated (N = 303) vs. control (N = 306) | 12.4 weeks | 150 mcg, target TSH 0.1–1.0 mU/L | CBCL, Brief-P | 3 years | No difference in behaviour scores between groups |
| Casey, 2017, USA14 | SCH treated (N = 323) vs. control, (N = 326) | 16.6 weeks | 100 mcg, target TSH 0.1–2.5 mU/L | CBCL | 3–5 years | No difference in behaviour scores between groups |
| Casey, 2017, USA14 | Hypothyroxinaemia treated (N = 254) vs. control (N = 253) | 18 weeks | 50 mcg, target FT4 0.86–1.9 ng/dl | CBCL | 3–5 years | No difference in behaviour scores between groups |
| Hales, 2020, UK27 | RCT follow-up: SCH and/or hypothyroxinaemia treated (N = 125), untreated (N = 104), normal gestational thyroid function (GTF, N = 246) | 12.4 weeks | 150 mcg, target TSH 0.1–1.0 mU/L | SDQ, SCQ, ADHDQ | 9.5 years | No difference in behaviour scores between groups; higher scores in over-treated vs. normal GTF |
| Author, year, country | Intervention and control groups | Gestational age at intervention | Levothyroxine, dose per day | Assessment, tools | Age at evaluation | Conclusion |
|---|---|---|---|---|---|---|
| Offspring IQ | ||||||
| Lazarus, 2012, UK and Italy13 | SCH and/or hypothyroxinaemia treated (N = 303) vs. control (N = 306) | 12.4 weeks | 150 mcg, target TSH 0.1–1.0 mU/L | WPPSI-III | 3 years | No difference in IQ scores or % with IQ < 85 |
| Casey, 2017, USA14 | SCH treated (N = 323) vs. control, (N = 326) | 16.6 weeks | 100 mcg, target TSH 0.1–2.5 mU/L | WPPSI-III, DAS-II | 3–5 years | No difference in IQ scores between treated and control |
| Casey, 2017, USA14 | Hypothyroxinaemia treated (N = 254) vs. control (N = 253) | 18 weeks | 50 mcg, target FT4 0.86–1.9 ng/dl | WPPSI-III, DAS-II | 3–5 years | No difference in IQ scores in treated vs. control groups |
| Hales, 2018, UK26 | RCT follow-up: SCH and/or hypothyroxinaemia treated (N = 119), untreated (N = 98), normal gestational thyroid function (GTF) (N = 232) | 12.4 weeks | 150 mcg, target TSH 0.1–1.0 mU/L | WISC | 9.5 years | No difference in % with IQ <85 in normal GTF vs. merged treated and untreated |
| Offspring behaviour | ||||||
| Lazarus, 2012, UK and Italy13 | SCH and/or hypothyroxinaemia treated (N = 303) vs. control (N = 306) | 12.4 weeks | 150 mcg, target TSH 0.1–1.0 mU/L | CBCL, Brief-P | 3 years | No difference in behaviour scores between groups |
| Casey, 2017, USA14 | SCH treated (N = 323) vs. control, (N = 326) | 16.6 weeks | 100 mcg, target TSH 0.1–2.5 mU/L | CBCL | 3–5 years | No difference in behaviour scores between groups |
| Casey, 2017, USA14 | Hypothyroxinaemia treated (N = 254) vs. control (N = 253) | 18 weeks | 50 mcg, target FT4 0.86–1.9 ng/dl | CBCL | 3–5 years | No difference in behaviour scores between groups |
| Hales, 2020, UK27 | RCT follow-up: SCH and/or hypothyroxinaemia treated (N = 125), untreated (N = 104), normal gestational thyroid function (GTF, N = 246) | 12.4 weeks | 150 mcg, target TSH 0.1–1.0 mU/L | SDQ, SCQ, ADHDQ | 9.5 years | No difference in behaviour scores between groups; higher scores in over-treated vs. normal GTF |
SCH, subclinical hypothyroidism; WPPSI-III, Wechsler Preschool and Primary Scale of Intelligence, 3rd edition; DAS-II, Differential Ability Scales–II; WISC, Wechsler Intelligence Scale for Children, 4th edition UK; CBCL, Child Behaviour Checklist 2000; Brief-P, Behaviour Rating Inventory of Executive Function, preschool version; SDQ, Strengths and Difficulties Questionnaire; SCQ, Social Communication Questionnaire (SCQ); ADHDQ, Child Attention Deficit Hyperactivity Disorder Questionnaire; mcg, micrograms.
To elucidate any late effects of Levothyroxine on cognition, a follow-up evaluation was conducted in children of the UK participants of the CATS study, median age 9.5 years. In this study, offspring of the treated or untreated mothers with suboptimal gestational thyroid function were additionally compared to children with normal thyroid function during pregnancy.26 Regression models showed no differences for odds of full-scale IQ < 85 points between the normal and the suboptimal gestational thyroid function groups. In addition, there was no interactive effect of treatment with Levothyroxine on IQ. Additionally, maternal TSH was not associated with IQ indices, further supporting the absence of any treatment benefits on cognition. Collectively, the major limitation of these two trial cohorts is that thyroid hormone replacement was introduced relatively late in pregnancy, i.e. after the early stages of foetal brain development should have occurred, and so treatment could have been too late to influence any adverse effects of thyroid dysfunction on offspring neurodevelopment.
Offspring behaviour
Observational studies
Beyond the associations with cognition, thyroid dysfunction during pregnancy has also been associated with a range of neurological, behavioural and psychiatric effects on the offspring, including epilepsy,28 autism spectrum disorder (ASD), attention-deficit hyperactivity disorder (ADHD), as well as effects on social and communication competencies.29,30 However, data from observational studies have given variable results, with some studies reporting a positive association between FT4 or TSH and ADHD symptoms28 and others showing no association31 or even a negative association for FT4 with ADHD symptoms.32 However, these studies had methodological differences with respect to the timing and tools used to assess ADHD, the gestational age at sampling in pregnancy, correction for confounders, iodine nutrition of the population and the analytical approach used to determine the association. A 2020 meta-analysis of 29 studies by Ge et al. found modest associations between maternal hypothyroidism and ADHD, ASD and epilepsy while hyperthyroidism was associated with ADHD and epilepsy (Fig. 1).17 In addition, the individual participant meta-analyses by Levie et al. reported an increased risk of autistic traits in the children of mothers with hypothyroxinaemia or hyperthyroxinaemia.17
RCTs
To address the impact of maternal Levothyroxine on offspring behaviour, Lazarus and colleagues evaluated the CATS RCT cohort using validated behavioural rating questionnaires. They found no significant benefit of Levothyroxine replacement at 3 years of age, for either of these tests13 (Table 2). Likewise, in their study, Casey et al. also examined behavioural and attention outcomes and in line with previous reports, found no difference between the offspring of the treated versus untreated groups at age 3 and 5 years for any of these indices.14 However, it is important to note that assessing behavioural parameters was not the primary aim of these studies and therefore both studies were underpowered for these outcomes. More recently, the extended follow-up CATS-II trial with child evaluation at 9 years of age also showed no beneficial effect of Levothyroxine on behavioural aspects.27 In this study, behaviour scores at ages 3 and 9 years were highly correlated, with the 3-year scores predicting 30% of the variation at 9 years. Furthermore, utilizing an expanded repertoire of validated questionnaire screening tools for ADHD, ASD and social competencies, did not reveal any significant differences between the study groups. A key finding of this study, however, was that offspring of over-treated mothers were more likely to exhibit ADHD traits compared to children of mothers with normal gestational thyroid function, emphasizing the risks of over-treatment and the need for close monitoring of Levothyroxine treatment during pregnancy.27
Cardiometabolic effects
Observational studies
Thyroid hormones are important physiological regulators of metabolism and energy balance in humans.33 The effects of thyroid dysfunction on various cardiometabolic parameters including lipid profile, blood pressure and cardiovascular function are well recognized.34 Furthermore, adverse metabolic and cardiovascular effects are seen with variations of thyroid hormones even within the reference range.35 Thyroid dysfunction also exerts well-documented effects on energy expenditure, and resting metabolic rate36 and higher baseline levels of thyroid hormones predict a favourable response to weight loss programmes in euthyroid individuals.37 In pregnancy, maternal subclinical hypothyroidism is associated with foetal growth restriction and low birth weight and some studies have shown associations of maternal thyroid dysfunction with pregnancy-induced hypertension and preeclampsia.1,8 Thus, given the diverse effects of thyroid hormones on various cardiovascular and metabolic processes it is plausible that the intrauterine thyroid hormone status can affect cardiovascular and metabolic outcomes in later life. Yet only a handful of studies to date have investigated the potential late effects of suboptimal gestational thyroid function on offspring cardiometabolic outcomes.
A study of a Danish birth cohort (n = 965) showed that offspring of mothers with subclinical hypothyroidism during gestation had higher systolic blood pressure at age of 20 years compared to the progeny of euthyroid mothers.38 However, they did not show any associations between maternal thyroid function and self-reported anthropometric parameters such as waist circumference and body mass index (BMI). Another study embedded in the Dutch generation-R birth cohort (n = 5646) examined associations of maternal TSH and FT4 levels in early pregnancy with offspring body weight and composition at 6 years of age.18 Although the authors did not show any association of dysfunction in pregnancy with cardiometabolic outcomes they found a positive association between maternal TSH levels and childhood BMI, total fat mass, abdominal subcutaneous fat mass and diastolic blood pressure while a negative association was seen with maternal FT4 and childhood BMI and abdominal fat mass. A recent analysis of the Danish National Birth Cohort (n = 7624) also found no association between maternal thyroid dysfunction and child weight at 7 years. An interesting finding of this study, however, was that mothers with hypothyroidism in pregnancy had a higher risk of subsequently being overweight or obese while those with hyperthyroidism were less likely to be obese or overweight.39
RCTs
The late cardiometabolic effects of suboptimal thyroid function during gestation was examined in the CATS follow-up study.40 In this study, offspring were examined at 9 years of age for BMI. In addition, biochemical and cardiovascular parameters, including thyroid function, thyroid antibodies, lipid profile, adiponectin levels, as well as aortic pulse wave velocity, peripheral and central blood pressure and bone densitometry assessments were obtained. Interestingly, there was no significant difference for any of these outcomes, except for marginally low HDL cholesterol levels in the treated suboptimal gestational thyroid function vs. the untreated group (P = 0.048).40 The study also examined the maternal cardiometabolic outcomes of suboptimal thyroid function during pregnancy ~9 years after delivery. In the study, thyroid dysfunction during pregnancy was not associated with cardiovascular outcomes or differences in bone density between study groups. However, mothers with untreated suboptimal gestational thyroid function exhibited higher BMI than mothers who received Levothyroxine replacement during their pregnancy. Body composition analysis suggested that this difference was primarily due to differences in fat mass and not lean mass. In addition, treated women had a number of more favourable metabolic parameters such as lower triglyceride levels and lower insulin levels presumably due to lower BMI in this group.40 However, the lack of physical activity and lifestyle information is a limitation of the data and adjustments for these parameters would be important to fully explore whether the difference in maternal BMI was subject to confounding.
Discussion
Observational studies
The above observational data must be interpreted in the context of several limitations. One consideration is that the data may depict association rather than causation and that thyroid function may simply serve as a proxy for other unexplored factors associated with poor pregnancy outcomes such as obesity, socioeconomic status or iodine nutrition. For example, an analysis of the NIH RCT cohort showed an association between economic vulnerability during pregnancy and increased risk of adverse neurodevelopmental outcomes in children at 2 and 5 years of life.41 Obesity displays a bi-directional relationship with thyroid function and maternal obesity is independently associated with adverse pregnancy outcomes,42 yet body weight is not routinely addressed in cohort studies. Similarly, iodine sufficiency is assumed in most studies based on the general population iodine nutrition. However, the population iodine status may not reflect the iodine status in pregnancy due to the increased vulnerability of the pregnant state to iodine deficiency.4 Thus, studies undertaken in iodine-deficient pregnant populations, whether recognized or not, will be difficult to interpret due to the compounding neurodevelopmental effects of iodine deficiency.
Other relevant factors include variations in diagnostic thresholds, timing of sampling in pregnancy and outcome evaluation measures. Inconsistent FT4 thresholds for hypothyroxinaemia have yielded contrasting prevalence and association with outcomes even within the same cohort.43 For subclinical hypothyroidism, several studies lacked information on thyroid antibody status or FT4 levels thereby blurring the distinction between overt, subclinical or autoimmune disease in the analysis. In addition, sampling time for thyroid hormone measurements varied from as early as 10 weeks gestation20 to the late third trimester.44 For the assessment of cognitive function, although most studies used standard measures such as the Bayley or Wechsler intelligence scales, others have evaluated overall educational attainment as an intuitively more meaningful endpoint.21 However, a direct correlation between educational attainment and cognitive development cannot be assumed as shown in the study by Pakkila et al., which reported differential effects of maternal thyroid function on psychometric evaluations and scholastic performance.20
These limitations notwithstanding, it is difficult to dismiss the findings from observational studies given the compelling biological plausibility of these associations and the wealth of animal and experimental evidence for a critical role of thyroid hormones in vital neurodevelopmental processes.2,3 These effects are exemplified in the extreme instances of cognitive impairment in individuals with untreated congenital hypothyroidism5 or severe iodine deficiency.6 In addition, an increased risk of ADHD is observed in individuals with excess thyroid hormone due to the syndrome of resistance to thyroid hormones arising from mutations in the thyroid hormone receptor ß-gene.45 Furthermore, as shown in the CATS follow-up analysis, children of mothers who were over-treated with Levothyroxine had higher ADHD scores than those with normal thyroid function.27 Thus, taken together, the balance of evidence favours associations of maternal subclinical hypothyroidism and hypothyroxinaemia with intellectual deficits, ADHD and ASD, while hyperthyroidism and high normal FT4 levels are linked with ADHD. More studies are needed to clarify the relationship of thyroid function with less well-studied outcomes such as epilepsy, schizophrenia or cerebral palsy.
Antenatal thyroid screening RCTs
The existing antenatal thyroid screening RCTs have been well conducted with robust approaches for screening, intervention and outcome evaluation. However, a few methodological considerations are critical to their interpretation. First, it is likely that the studies were not adequately powered for the outcomes examined. The original power calculation for the CATS study13 was based on the seminal study by Haddow et al. which reported a 7-point IQ difference between children of hypothyroid mothers and controls.15 However, the median TSH in the CATS screening cohort was much lower than in the Haddow study (3.8 vs. 13.2 mU/L) meaning that hypothyroidism was less severe in the CATS study. Accordingly, a smaller IQ deficit would be expected in the CATS cohort and would necessitate a larger sample size to demonstrate. This lack of power becomes even more relevant for the follow-up CATS evaluations which comprised only a proportion of the original cohort. Thus, any expected advantages of the follow-up evaluations would have been offset by diminished power. The study by Casey et al. is similarly afflicted, although to a lesser extent, as the power calculation was based on a 5-point IQ difference.14 As some experts have argued, a smaller IQ difference, e.g. 3-points, would still be clinically relevant but would require a larger sample size to demonstrate.46
A second consideration is that women in both trials had milder degrees of hypothyroidism based on the TSH 97.5th percentile of the population. This is not surprising as routine thyroid testing, albeit non-systematic testing, has become more frequent in the population as a whole.47 Accordingly, women with more marked degrees of hypothyroidism are more likely to have already been identified through prior testing and would be ineligible for thyroid screening trials. This could effectively have diluted the effects of hypothyroidism in the trial cohorts. Lastly and perhaps most importantly, both studies had a relatively late onset of Levothyroxine initiation, ~12 weeks in the CATS trial and 17 weeks in the Casey study. As shown in Figure 2, early developmental events, namely neuronal migration and proliferation, precede foetal thyroid hormone secretion.2,3 These key events are dependent on maternal thyroxine sources and determine subsequent development of neural pathways (Fig. 2). Thus, interventions in these trials would have been too late to influence these early events, a critical constraint that ultimately applies to the entire CATS follow-up series. Although timing of interventions was addressed in sensitivity analysis in both the CATS trial and the study by Casey et al., such analyses were clearly under-powered.
Timeline for foetal thyroid and brain development. Legend: Horizontal bars indicate timelines for various thyroid and neurodevelopmental events. Dashed lines/arrows indicate periods of reduced rates of development and faded bars indicate events that continue after birth.
Universal thyroid screening
Future studies addressing the above limitations will therefore be needed to determine the utility of antenatal thyroid screening. Meanwhile, a systematic approach to thyroid screening is warranted. Two key approaches have been addressed by international society guidelines, namely a risk-based or targeted approach that restricts screening of women with established risk factors for thyroid disease11 or a universal approach that screens all pregnant women for thyroid dysfunction.48 The well-rehearsed arguments for and against universal screening have been well documented.49 Briefly, the case against universal screening is that a case-finding approach should identify those women at greatest risk of thyroid dysfunction. Besides, most cases of screening-detected thyroid dysfunction are borderline states of questionable significance and existing RCTs have failed to show that correcting these abnormalities yield any gains for pregnancy or offspring outcomes. Instead, universal screening may promote over-diagnosis and generate anxiety by medicalizing laboratory findings of doubtful significance. This may lead to over-treatment which may in turn have adverse effects as seen in the association of over-treatment with offspring ADHD in the CATs study cohort.
The argument in favour of universal screening centres on the important public health significance of thyroid dysfunction in pregnancy and the potential for adverse outcomes even for mild degrees of thyroid dysfunction in pregnancy.49 As we highlight in this review, these effects extend beyond pregnancy and may have lifelong health and socioeconomic consequences. Cognitive deficits in children lead to lifetime reductions in academic achievement,50 impaired work performance,51 reduction in lifetime earnings and reduced overall survival.52 ADHD is associated with serious antisocial behaviour, substance misuse and increased adult mortality53 while children with ASD are at risk of developing epilepsy obesity and a host of psychiatric comorbidities.54 In addition, thyroid dysfunction in pregnancy may also have adverse consequences for the long-term cardiometabolic health of the mother as shown in the studies by Muller40 and Andersen.39 Lastly, cost-effective analyses have shown superiority of universal screening over targeted screening with incremental cost effectiveness ratios per quality adjusted life year of $7258.00 (universal vs. targeted screening) or £374.28 (screening vs. no screening at all) in published studies.55,56 Importantly, screening was shown to be worthwhile even for the detection of overt disease alone.39 Thus, universal antenatal thyroid screening stands to improve the health of the offspring in the short- and long term while also presenting opportunities for optimizing maternal health.
Conclusions
The last decade has witnessed a proliferation of observational studies addressing late offspring effects of gestational thyroid dysfunction. Data in this field have been enriched by the availability of longitudinal birth cohorts that have highlighted a range of plausible cognitive and neurobehavioural effects in the late childhood years. However, while these studies add to the wealth of observational data on the adverse effects of thyroid disease in pregnancy, they are unlikely to significantly alter the current narrative on antenatal thyroid screening without additional RCTs conducted with study designs that address the limitations of existing RCTs. Future trials should evaluate the impact of universal screening on late offspring and maternal effects using designs that address study power and early intervention on clearly defined disorders of thyroid dysfunction.
Funding statement
No funding was received for this study.
Conflict of interest statement
We declare that no conflict of interest exists with respect to this study.
Data availability
No new data were generated in the production of this review.
