Articles

Maternal thyroid function during pregnancy and child brain

morphology: a time window-specific analysis of a
prospective cohort
Toyah A Jansen, Tim I M Korevaar, Tessa A Mulder, Tonya White, Ryan L Muetzel, Robin P Peeters, Henning Tiemeier

Summary
Background Adequate thyroid hormone availability during pregnancy is necessary for optimal fetal brain development. Lancet Diabetes Endocrinol
During the first 18–20 weeks of gestation, fetal thyroid hormone availability largely depends on the placental transfer 2019; 7: 629–37

of maternal thyroxine. Although various studies have shown that maternal thyroid dysfunction is associated with Published Online
June 28, 2019
suboptimal child neurodevelopmental outcomes, the most vulnerable time window remains to be identified. The aim
http://dx.doi.org/10.1016/
of this study is to examine the association of maternal thyroid function with child brain morphology and to study S2213-8587(19)30153-6
whether any association depends on the timing of thyroid assessment. See Comment page 589
Generation R Study Group
Methods This prospective cohort study was part of the Generation R Study in Rotterdam, Netherlands, with a (T A Jansen MSc,
prospective population-based birth cohort. Pregnant women living in Rotterdam with an expected delivery date T I M Korevaar MD,
between April 1, 2002, and Jan 1, 2006, were eligible. Other inclusion criteria were maternal serum thyroid-stimulating T A Mulder BSc);
and Department of
hormone (TSH) and free thyroxine (FT4) measurement in early or mid-pregnancy (≤18 weeks) and available brain Epidemiology
MRI data for child at age 10 years. Exclusion criteria were pre-existing thyroid disorder, thyroid disorder treatment, (R L Muetzel PhD), Erasmus
twin pregnancy, in-vitro fertilisation-induced pregnancy, and suboptimal-quality MRI data or major incidental finding University Medical Center,
Rotterdam, Netherlands;
on MRI. The main outcome was the association between maternal TSH and FT4 concentrations with brain MRI
Netherlands Institute for
outcomes of children. Regression analyses accounted for gestational age at blood sampling, maternal age, ethnicity, Health Sciences, Erasmus
education level, smoking, thyroid peroxidase antibody positivity, child sex, age at MRI, and total intracranial volume. University, Rotterdam,
Effect modification by gestational age at blood sampling was also investigated. Netherlands (T A Jansen,
T I M Korevaar); Department of
Internal Medicine
Findings Between Dec 1, 2001, and June 30, 2005, 7069 women were enrolled during early or mid-pregnancy (T I M Korevaar, T A Mulder,
(≤18 weeks of gestation), of whom 5088 were not included because they did not have available data on maternal R P Peeters MD), Academic
serum TSH or FT4 concentrations (n=1175), their child did not have brain MRI done (n=3377), or they met Center for Thyroid Diseases
(T I M Korevaar, R P Peeters),
exclusion criteria (n=536). Thus, 1981 mother–child pairs were included in the study, with TSH and FT4
Department of Child and
concentrations measured during pregnancy at a median of 13·1 weeks of gestation (IQR 12·1–14·5) and offspring Adolescent Psychiatry
brain morphology assessed by MRI at a median age of 9·9 years (9·7–10·2). Maternal TSH had an inverted (T A Mulder, T White MD,
U-shaped association with offspring total grey matter volume (p=0·007) and with cortical grey matter volume R L Muetzel,
Prof H Tiemeier MD),
(p=0·022). The association of maternal TSH with child total grey matter volume (pinteraction=0·053) and cortical
and Department of Radiology
volume (pinteraction=0·086) differed by the duration of gestation. Analyses stratified for gestational age at blood (T White), Erasmus Medical
sampling showed an inverted U-shaped association of maternal TSH with child total grey matter volume and Center, Rotterdam,
cortical grey matter volume, which was most evident at 8 weeks gestation. After about 14 weeks of gestation, TSH Netherlands; Department of
Social and Behavioral Sciences,
was no longer associated with child brain morphology. Maternal FT4 concentrations were not associated with child Harvard T H Chan School of
total grey matter volume after adjusting for total intracranial volume (p=0·75). Public Health (Prof H Tiemeier)
Correspondence to:
Interpretation Here, we show that both low and high maternal thyroid function are associated with smaller child total Prof Henning Tiemeier,
grey matter and cortical volume. To the best of our knowledge, this study is the first to show that an association with Department of Child and
Adolescent Psychiatry, Erasmus
a neurodevelopmental outcome is most evident when maternal thyroid function is measured early in pregnancy.
Medical Center, 3000 CA,
These novel findings suggest that embryonic brain development is particularly vulnerable to altered maternal thyroid Rotterdam, Netherlands
function. h.tiemeier@erasmusmc.nl

Funding Netherlands Organisation for Health Research and Development and the Sophia Children’s Hospital
Foundation.

Copyright © 2019 Elsevier Ltd. All rights reserved.

Introduction of neuronal cells that ultimately develop into the grey

Adequate maternal thyroid function is required for matter of the brain.1–3
optimal fetal brain development. Thyroid hormone Neurogenesis starts around the fifth week of gestation
regulates fundamental neurodevelopmental processes, and thyroid hormone receptors are present in the fetal
including the proliferation, migration, and differentiation brain from 8 weeks of gestation onwards.4 Fetal thyroid

www.thelancet.com/diabetes-endocrinology Vol 7 August 2019 629

 Articles
Research in context
Evidence before this study
In the past two decades, epidemiological studies have shown
that overt maternal hypothyroidism and hypothyroxinaemia
is negatively associated with adverse neurodevelopmental
outcomes such as low IQ, autism, schizophrenia, and
attention-deficit hyperactivity disorder. Furthermore, clinical
studies suggest that maternal thyroid dysfunction might
affect brain morphology, particularly grey matter. However,
the most vulnerable time window during which maternal
thyroid function is associated with offspring
neurodevelopmental outcomes remains to be identified.
Added value of this study
In our study we identified an inverted U-shaped association of
maternal thyroid-stimulating hormone with total grey matter
and cortical grey matter volume as assessed by MRI scanning in
hormone production starts around the 14th week, but
the fetal thyroid is not fully functional until week 18–20.5
Therefore, fetal thyroid hormone availability during
crucial phases of early brain development predominantly
depends on the placental transfer of maternal thyroxine.
Maternal hypothyroidism or hypothyroxinaemia during
early pregnancy is associated with various child adverse
neurodevelopmental outcomes, including a lower
intelligence quotient (IQ) and a higher risk of autism,
schizophrenia, and attention-deficit hyperactivity
disorder.6–13 Although it is generally accepted that
maternal thyroid dysfunction during pregnancy is
associated with suboptimal child neurodevelopmental
outcomes, the most crucial period for maternal thyroid
hormone to ensure optimal fetal brain development
remains to be elucidated. Identification of this crucial
period can have direct clinical implications related to
risk assessment and the time window for potential
treatment.
Although of clinical relevance, clinical neuro­
developmental outcomes are less precise and more
indirect measures of brain development. By contrast,
brain imaging by MRI is an objective reproducible
measure of brain development with detailed information
on different brain structures. Imaging data can provide an
indication of which neurogenesis processes depend on
maternal thyroid function during specific stages of fetal
brain development. Results of brain imaging studies
suggest that maternal thyroid function might have a
different effect on grey matter than on white matter: initial
studies demonstrated that maternal hypothyroidism
during pregnancy is associated with altered child brain
morphology, including suboptimal hippocampal and
cortical development.14,15 Our group have previously shown
an inverted U-shaped association of maternal free
thyroxine (FT4) with total grey matter volume and cortical
grey matter volume using data of 646 mother–child pairs.8
630 www.thelancet.com/diabetes-endocrinology Vol 7 August 2019
the offspring, which was strongest if thyroid function was
assessed in early pregnancy (<14 weeks). To the best of our
knowledge, this study is the first to show that an association
with a neurodevelopmental outcome is more evident if thyroid
function is measured before the 14th week of pregnancy.
Implications of all the available evidence
In line with previous studies, our study shows that both
exposure to a low and a high maternal thyroid function is
negatively associated with child total grey matter volume and
cortical grey matter volume as assessed by MRI. The absence of
association from the 14th week onwards is an important novel
finding that should be considered during clinical risk assessment
and when timing interventions during clinical practice, as well as
in the design of future studies of the effects of levothyroxine
treatment for mild thyroid disease in pregnancy.
This result suggests that both a low and a high maternal
thyroid function might have an adverse effect on fetal grey
matter development. Although thyroid hormone also
regulates white matter formation by stimulating
myelination of neuronal cells, evidence from human
imaging studies for adverse effects of maternal thyroid
dysfunction on the development of white matter tracts is
scarce.2,16 The difference in the association of maternal
thyroid function on grey matter versus white matter might
be because myelination starts in the late second trimester,
when the fetus has become increasingly dependent on its
own thyroid hormone production.2,17 Moreover, the effect
of thyroid hormone on neuronal migration is largely
mediated via the α-thyroid hormone receptor, which is
expressed throughout fetal develop­ ment, whereas
myelination is largely regulated via the β-thyroid hormone
receptor, which is expressed much later in fetal
development.18
Based on models of fetal thyroid physiology and neuro­
developmental processes, maternal thyroid dysfunction
might have a more deleterious effect in early gestation,
yet no study has been able to demonstrate the crucial
period. The aim of our study is to examine the association
of maternal thyroid function with child brain morphology
as a detailed and objective marker of brain development
and to study whether this association with specific brain
structures differs according to gestational age at blood
sampling. We hypothesised that low maternal thyroid
function, particularly in early pregnancy (first and early
second trimester), is associated with a smaller grey
matter volume in pre-adolescence.
Methods
Study design and participants
This prospective cohort study was part of the Generation
R Study, a population-based prospective cohort from early
fetal life onwards in Rotterdam, Netherlands.19 Pregnant

women living in the municipality of Rotterdam with an
expected delivery date between April 1, 2002, and
Jan 1, 2006, were eligible and enrolled during visits to a
midwife or the hospital.19 We included mother–child pairs
if maternal thyroid-stimulating hormone (TSH) or FT4
serum measurements had been assessed during early or
mid pregnancy (≤18 weeks of gestation). Only mother–
child pairs with children who had brain MRI data at the
age of 10 years were included. We excluded pairs if the
mother had a pre-existing thyroid disorder, had treatment
for a thyroid disorder, had a twin pregnancy or in-vitro
fertilisation (IVF)-induced pregnancy, or if the child MRI
data was of suboptimal quality or showed a major
incidental finding. The study design, research aims, and
the specific measurements in the Generation R Study
have been approved by the Medical Ethical Committee of
the Erasmus Medical Center, Rotterdam, Netherlands.
Written informed consent was obtained from all the
participants or the children’s parents or guardians.
Procedures
Maternal serum thyroid parameters were obtained in
early or mid-pregnancy (≤18 weeks). Up to 3 h after
blood sampling, plain tubes were centrifuged and serum
was stored at –80°C. TSH and FT4 were measured in
maternal blood using chemiluminescence assays (Vitros
ECiQ Immunodiagnostic System; Ortho Clinical
Diagnostics, Rochester, NY, USA). The reference range
for TSH concentration was 0·03–4·04 mU/L and for FT4
it was 10·4–22·0 pmol/L. Maternal thyroid peroxidase
(TPO) antibody was measured using the Phadia
250 Immunoassay (Phadia AB, Uppsala, Sweden) and
considered positive when more than 60 IU/mL.20
When the children reached age 9 years, they were
invited to visit the research centre at the Erasmus MC-
Sophia Children’s Hospital in Rotterdam for brain MRI.21
Before scanning, all children had a 30-min mock
scanning session to familiarise them with the scanner
environment. MRI scans were obtained with a Discovery
MR750w 3.0T scanner (GE Healthcare, Milwaukee, WI,
USA) using an eight-channel head coil. T1-weighted
structural images were acquired with an inversion
recovery prepared fast spoiled gradient recalled sequence.
Structural MRI data were processed through the
FreeSurfer analysis suite, version 6.0.22 Global metrics of
volume were extracted (eg, total grey matter volume and
cortical grey matter volume). Image quality was assured
by visual inspection of FreeSurfer reconstructions using
a protocol similar to previously defined methods.23
Details on child IQ and maternal urinary iodine
assessment are in the web appendix.
The choice of potential confounders was made on the
basis of background knowledge about the study
question.6,8,24 Information on maternal age at enrolment,
ethnicity, parity, education level, marital status, family
income, and smoking behaviour was obtained through
postal questionnaires completed during pregnancy.19
www.thelancet.com/diabetes-endocrinology Vol 7 August 2019 631
Articles
Maternal total human chorionic gonadotropin concen­
trations were analysed in the same serum sample as
maternal thyroid parameters. Information on child sex,
gestational age at birth, and birthweight was obtained
from midwives and hospital registries. Child TSH and
FT4 concentrations were measured in serum samples
obtained at the research centre when children were aged
about 6 years.
Outcomes
We assessed the association between maternal TSH and
FT4 concentrations with brain MRI outcomes of children.
We also analysed and quantified whether gestational age
at blood sampling or human chorionic gonadotropin
concentrations modified this effect. MRI outcomes used
in this study were total brain volume, total grey matter
volume, cortical grey matter volume, subcortical grey
matter volume, and cerebral white matter volume. We
also did subgroup analyses excluding participants with
overt thyroid disease.
Statistical analyses
We used multivariable linear regression models to study
the association of maternal TSH and FT4 with brain MRI
outcomes. We tested for effect modification of the
association of TSH and FT4 with brain MRI outcomes by
gestational age at blood sampling as a continuous variable
using a product interaction term of gestational age at
blood sampling and maternal TSH or FT4 concentrations.
We further quantified potential relevant differences by
performing stratified analyses by gestational age at blood
sampling if there was any indication of effect modification.
A p value for interaction of less than 0·15 was used as a
cutoff to further explore potential relevant effect
modification by stratification.25 In a similar manner, we
tested for effect modification by human chorionic
gonadotropin concentrations. Models were built using a
non-automated stepwise approach in which covariates
were included on the basis of biological confounding
plausibility, change in effect estimate of the variable of
interest, or change in the residual variance of the model.
We tested the non-linearity assumption by adding
quadratic terms to the models. When the quadratic term
showed non-linearity, we investigated the nature of the
exact association using ordinary least-squares regression
models with restricted cubic splines with three knots at
the tenth, 50th, and 90th percentiles. In all statistical
models, maternal age, ethnicity, education level, smoking,
TPO antibody positivity (>60 IU/mL) and gestational age
at blood sampling were selected as potential confounders.
Child sex and age at MRI were included as independent See Online for appendix
predictors of the outcome. With total intracranial volume
in the model, coefficients can be interpreted as the
association of exposure and outcome with intracranial
volume held constant. We checked the assumption of
normally distributed residuals by visual inspection of the
residuals. Missing values of covariates (maternal
 Articles

7069 mothers enrolled during early pregnancy (≤18 weeks)

All participants (n=1981)
Maternal characteristics
Thyroid-stimulating hormone, mU/L* 1·33 (0·85–2·01)
1175 excluded because no
thyroid-stimulating hormone or Free thyroxine, pmol/L* 14·8 (13·2–16·6)
free thyroxine measurement available Thyroid peroxidase antibody positive 110 (6%)
Gestational age at blood sampling, weeks 13·1 (12·1–14·5)
5894 mothers with thyroid-stimulating hormone or Urinary iodine-to-creatine ratio, μg/g† 212·3 (142·1–308·9)
free thyroxine measurement available Urinary iodine-to-creatine ratio less than 242 (28%)
150 μg/g†
3377 mother–child pairs excluded Age at intake, years 31·2 (27·9–33·9)
because children did not have BMI 23·3 (21·4–26·0)
MRI scan
Parity
0 1219 (62%)
2517 mother–child pairs in which children had MRI scan 1 560 (28%)
2 155 (8%)
≥3 47 (2%)
536 mother–child pairs excluded
452 poor imaging quality Smoking status
10 major incidental findings No 1505 (76%)
35 thyroid disorder or treatment
for thyroid disorder Until pregnancy was known 192 (10%)
25 twin pregnancies Yes 284 (14%)
14 IVF treatment
Education level
None or primary only 128 (6%)
1981 mother–child pairs included in analysis Secondary phase 1 (3–4 years) 208 (10%)
Secondary phase 2 (4–5 years) 589 (30%)

Figure 1: Flowchart of participants through the study Higher phase 1 (6–8 years) 484 (24%)
Higher phase 2 (>8 years) 572 (29%)
Ethnicity
smoking 10·1%, TPO antibody 6·6%, education Dutch 1162 (59%)
level 4·2%, and ethnicity 1·3%) were imputed using Indonesian 75 (4%)
multiple imputation and variables without missing data Cape Verdian 91 (5%)
such as gestational age at blood sampling, maternal age, Moroccan 94 (5%)
child sex, child age at MRI, and total intracranial volume Dutch Antilles 39 (2%)
were included in the model as predictors.26 The imputed Surinamese 155 (8%)
variables across the different imputed and non-imputed Turkish 111 (6%)
dataset did not differ. We pooled ten imputed datasets at Other European, North American, or 172 (9%)
the end to obtain the final estimates of interest using Oceanian
Rubin’s rules to estimate the variance. All statistical Other Asian, African, or South American 82 (4%)
analyses were done with R statistical software, version 3.3.2 Child characteristics
or SPSS, version 21.0 for Windows. Gestational age at birth, weeks 40·3 (39·3–41·0)
Birthweight, g 3460 (3100–3803)
Role of the funding source Age at MRI, years 9·9 (9·7–10·2)
The funders of this study had no role in study design, Sex
data collection, data analysis, data interpretation, or Female 1003 (51%)
writing of the report. The corresponding author had full Male 978 (49%)
access to all data in the study and had final responsibility
Data are median (IQR) or n (%). Data shown are after imputation of missing data.
for the decision to submit for publication. *Thyroid hormone measurements are available for n=1961 (TSH) and n=1971
(FT4). †Urinary iodine-to-creatinine ratio available for 867 participants.
Results Table: Descriptive characteristics of the study population
Between Dec 1, 2001, and June 30, 2005, 7069 women
were enrolled during early or mid-pregnancy (≤18 weeks
of gestation). Data on TSH or FT4 concentrations were maternal TSH concentrations, FT4 concentrations, and
available for 5894 mothers, and of these, 2517 had children TPO antibody positivity did not differ between mother–
who had brain MRI scanning. 536 mother–child pairs child pairs grouped on by MRI data availability. However,
were excluded because they met exclusion criteria, thus, the group of mother–child pairs with available MRI data
the final study population comprised 1981 mother–child had a higher education level, a lower mean BMI, and were
pairs (figure 1). Non-response analyses showed that older (appendix).

632 www.thelancet.com/diabetes-endocrinology Vol 7 August 2019

 Articles

770
p=0·007 p=0·75

765
Mean total grey matter volume (cm³)

760

755

750

590
p=0·022 p=0·33

585
Mean cortex volume (cm³)

580

575

0
0 1 2 3 4 5 0 10 15 20 25 30
Thyroid-stimulating hormone (mU/L) Free thyroxine (pmol/L)

Figure 2: Association of maternal thyroid stimulating hormone and free thyroxine with offspring brain morphology
Analyses were adjusted for gestational age at blood sampling, maternal age, ethnicity, education level, smoking, thyroid peroxidase antibody positivity (>60 IU/mL),
child sex, age at MRI, and total intracranial volume.

The study population was mainly of Dutch origin, with a matter volume (p=0·007) and cortical grey matter volume
median maternal TSH concentration of 1·33 mU/L (p=0·022; figure 2; appendix). Associations remained
(IQR 0·85–2·01) and a median maternal FT4 concentration after exclusion of women with overt thyroid disease
of 14·8 pmol/L (13·2–16·6). The median gestational age at (appendix). Evidence showed that the association of TSH
blood and urinary sampling and was 13·1 weeks with child total grey matter volume (pinteraction=0·053) and
(12·1–14·5). In the subset of women with urinary iodine cortical grey matter volume (pinteraction=0·086) differed
data, the median urinary iodine-to-creatinine ratio was according to the gestational age at blood sampling
212·3 μg/g (142·1–308·9) and most of these women were (figure 3). Subsequent stratification showed that the
iodine sufficient according to WHO classification association of TSH with child total grey matter and
(242 [28%] had a ratio of 150 μg/g).27 Children had brain cortical grey matter volume was less evident if blood was
MRI at a median age of 9·9 years (9·7–10·2; table). sampled after the first 14 weeks of pregnancy (figure 3;
Descriptive characteristics of the unimputed dataset are in appendix). Maternal FT4 concentrations had an inverted
the appendix. U-shaped association with child total grey matter volume
No linear association of maternal TSH concentrations (p=0·028) and cortical grey matter volume (p=0·018;
with total grey matter or cortical grey matter volume appendix), which disappeared when adjusting for total
were noted. However, maternal TSH concentrations had intracranial volume (p=0·75 for total grey matter and
an inverted U-shaped association with child total grey p=0·33 for cortical grey matter; figure 2; appendix).

www.thelancet.com/diabetes-endocrinology Vol 7 August 2019 633

 Articles

and cortex volume at the median age of 10 years, and vice

A
versa (appendix).
Week 8 Week 10 Week 12
780 In a previous study using a smaller subset of the
Mean total grey matter

770 Generation R cohort, we showed that, in contrast to the

volume (cm³)

760 results of this study, there was an inverted U-shaped

750
association of maternal FT4 but not TSH concentrations
with child total grey matter and cortical grey matter
740
volume.8 To investigate whether this discrepancy could
730
0
be due to selection bias by MRI data availability we did
several sensitivity analyses. In mother–child pairs with
Week 14 Week 16 Week 18
780 MRI data available for both measurements (median age
Mean total grey matter

770 8 years [IQR 7·1–8·6] and median age 10 years

volume (cm³)

760 [9·7–10·2), there was an inverted U-shaped association

750 of maternal TSH concentrations but not FT4
concentrations with child total grey matter (p=0·016)
740
and cortical volume (p=0·032; appendix). In addition,
730
0
we studied differences in the associations of TSH and
FT4 with child IQ at the median age of 6 years because
this measure is correlated with grey matter and was
B
available in a larger sample (n=3668).8 In line with our
Week 8 Week 10 Week 12
590 previous data, maternal FT4 showed an inverted
U-shaped association with child IQ. No significant
Mean cortex volume (cm³)

590
difference was detected in the association of maternal
580
FT4 concentrations with IQ between those mother–child
570 pairs with or without MRI data at the median age of
560 10 years (pinteraction=0·90; appendix).
0
Discussion
Week 14 Week 16 Week 18
590 In this study, we show that both low and high maternal
TSH concentrations, particularly in early pregnancy, are
Mean cortex volume (cm³)

590
580
570
560
0 pregnancy onwards, suggesting that this is the most
0 1 2 3 4 0 1 2 3 4 0 1
Thyroid-stimulating hormone Thyroid-stimulating hormone Thyroid-stimulating hormone
(mU/L) (mU/L)
Figure 3: The association of maternal thyroid stimulating hormone with child grey matter volume (A) or
cortical grey matter volume (B) stratified by gestational age at blood sampling
Analyses were adjusted for gestational age at blood sampling, maternal age, ethnicity, education level, smoking,
thyroid peroxidase antibody positivity (>60 IU/mL), child sex, age at MRI, and total intracranial volume.
Maternal FT4 concentrations (p=0·023), but not TSH
concen­trations (p=0·45), had an inverted U-shaped
association with total brain volume (appendix).
Neither TSH nor FT4 concentrations were associated
with cerebral white matter volume or subcortical grey
matter volume (appendix). None of the results
differed according to human chorionic gonadotropin
concentrations or after additional adjustment for child
TSH or FT4 concentrations measured in cord blood at
birth or at the median age of 6 years. In the subset of
women with urinary iodine data, additionally adjusting
for iodine-to-creatinine ratio did not change our results
(data not shown). Child IQ at the median age of 6 years
was positively associated with total grey matter volume
associated with a smaller volume of total grey matter and
cortical grey matter in children aged 9–12 years. To the
best of our knowledge, this study is the first to show that
the association of maternal thyroid function with
offspring neurodevelopment attenuates from early
2 3 4
vulnerable period of the fetus for low or high maternal
(mU/L) thyroid function. In contrast to previous findings, the
inverted U-shaped associations of maternal FT4 with total
grey matter volume and cortex volume were not present
after adjustment for total intracranial volume.
The identification of the most crucial gestational time
window for maternal thyroid dysfunction to be associated
with suboptimal neurodevelopment is relevant for various
reasons. First, identification of this period improves
understanding of thyroid physiology in pregnancy and the
ability to translate animal studies to human physiology.
Second, these data can inform future observational and
intervention studies to investigate the role of maternal
thyroid function in fetal neurodevelopment. Our results
strongly indicate that future studies will benefit from
inclusion of participants during early pregnancy, preferably
during the first trimester, and from stratified analysis
depending on gestational age at blood sampling. Our
finding that the association of maternal thyroid function
with fetal neurodevelopment is dependent on gestational
age at blood sampling even offers an explanation for why

634 www.thelancet.com/diabetes-endocrinology Vol 7 August 2019


some randomised controlled trials examining the effect of
levothyroxine treatment of maternal subclinical hypo­
thyroidism or hypothyroxinaemia in pregnancy on
offspring IQ showed no beneficial effect.28,29 The initiation
of treatment in these trials, with thyroid hormone
replacement therapy starting between 13 and 18 weeks of
gestation, was later than our estimated optimal timing.30,31
Our data could aid clinicians in optimising risk assess­
ment strategies related to a timely thyroid function
assessment and potential replacement therapy.
Well replicated associations of brain imaging parameters
with neurodevelopmental outcomes, including our results
with child IQ, suggest that the associations of maternal
thyroid dysfunction with offspring total grey matter and
cortex volume might be particularly relevant for child
cognitive development.8,32 In line with previous findings,
not only a low but also a high maternal thyroid function
was associated with suboptimal child neurodevelopmental
outcomes.8,33,34 The mechanism through which a high
maternal thyroid function leads to suboptimal brain
development remains to be elucidated. Possibly, high
maternal thyroid hormone availability could affect the
process of neuronal migration, as has been described in
rodent studies.35–37 This hypothesis is supported by studies
showing that a high maternal thyroid function is
associated with autism, because defects of neuronal
migration are known to play a role in the neuropathology
of autism.33,34,38 Further studies are required to elucidate
potential underlying mechanisms through which a high
maternal thyroid function leads to suboptimal brain
development. The possible clinical implications require
further studies because thyroid hormone replacement
therapy during pregnancy comes with the potential risk of
overtreatment.
Furthermore, our results provide insight on the specific
brain regions that might be affected in the offspring by
maternal thyroid dysfunction. We identified an association
of maternal TSH with grey matter and the cerebral cortex
but not with other morphological outcomes. These results
are in line with earlier findings of imaging studies.8,15
Since we did not identify any significant association with
subcortical grey matter volume, any effects of maternal
thyroid function on total grey matter volume are most
likely driven by the effect on cortical grey matter volume.
As we speculated earlier, a high maternal thyroid function
might affect neuronal migration and this theory could
explain why we found an association with cortical grey
matter but not with subcortical grey matter volume.
Animal and human studies have shown that thyroid
hormone exerts different effects in different brain areas
throughout fetal development.2,39 It is therefore likely that
an inadequate maternal thyroid function would mainly
affect the neurodevelopmental processes that take place
before week 14, when fetal thyroid hormone availability
largely depends on the placental transfer of maternal
thyroid hormone. Since proliferation and differentiation
of neuronal precursors and neuronal migration mainly
www.thelancet.com/diabetes-endocrinology Vol 7 August 2019 635
Articles
take place during the first trimester, it is biologically
plausible that maternal thyroid function affects grey
matter development. By contrast, myelination becomes
increasingly prominent during the second and third
trimester where fetal thyroid hormone availability is less
influenced by maternal thyroid function. This might
explain why we do not find evidence of an association of
maternal thyroid function with white matter volume.2
In this study, we were able to investigate the association
of maternal thyroid function with child brain morphology
using a large dataset with unique brain imaging data and
detailed data on various maternal and child covariates.
An important limitation of our study is that we did not
have longitudinal thyroid function measurements over
the course of pregnancy of mothers participating in the
Generation R cohort. Therefore, we need to be
particularly cautious in deducting any causal effect of
gestational age on the association of maternal TSH and
brain morphology. Additionally, the interaction with
gestational age at blood sampling was not significant.
Ideally, future studies will use repeated thyroid function
measurements throughout pregnancy to fully assess the
effect on child neurodevelopment and to identify the
most crucial timeframe.
Another potential limitation is that the gestational age-
dependent effect of TSH on brain morphology could
partly be attributed to selection bias, meaning that
women who came to the research centre earlier in
pregnancy have different characteristics to those who
visited the research centre later in gestation. We adjusted
for multiple socioeconomic variables, but we cannot
exclude any residual confounding or selection effects.
With regard to the interpretation of our results, our study
is limited by the scarcity of data on pubertal status, which
might influence brain morphology.40
In contrast to a previous study from our group, there was
an association of maternal TSH but not of FT4
concentrations with pre-adolescent brain morphology.
Sensitivity analyses suggested some potential selection
effects in our previous sample that might have influenced
the results. Notably, sensitivity analyses in this study with a
larger dataset showed that maternal FT4 was associated
with total brain volume, total grey matter volume, and
cortical grey matter volume if not adjusted for total
intracranial volume. Our results imply that TSH might be
a more specific predictor for total grey matter or cortical
grey matter development than FT4. However, taken
together, the results of both studies support the concept
that both low and high maternal thyroid function might
adversely affect fetal brain development. Our results
cannot distinguish whether either TSH or FT4 is a better
marker for neuro­ developmental processes.41 Subclinical
hypo­thyroidism has been associated with various adverse
pregnancy outcomes, but the relationship with child
neurocognitive development has been less consistent.6,8,42
The negative results of previous studies of subclinical
hypothyroidism might be due to the curvilinear
 Articles
association—ie, that a low TSH concentration is also
associated with suboptimal child neurocognitive outcomes,
but this is often not accounted for. Future studies are
needed to identify the best maternal markers of fetal
thyroid hormone availability.
In conclusion, our study shows that both a low and a
high maternal thyroid function are negatively associated
with child total grey matter volume and cortical grey matter
volume as assessed by MRI. We show that these effects are
more evident in early pregnancy with a suggested effect
threshold around the 14th week of pregnancy. This time
window is an important finding that should be considered
in the design of future clinical studies of levothyroxine
treatment for mild thyroid disease in pregnancy.
Contributors
TAJ did the data analyses and was involved in writing the report. TIMK
and TAM contributed to data analyses and writing of the report. RLM and
TW contributed to data collection, data analyses, and writing of the report.
RPP supervised data analyses and contributed to data interpretation and
writing of the report. HT supervised data analyses, contributed to data
interpretation and writing of the report, and directed the project.
Declaration of interests
We declare no competing interests.
Acknowledgments
This work was supported by a clinical fellowship from The Netherlands
Organisation for Health Research and Development (ZonMw), VICI
Grant (project number 016.VICI.170.200 awarded to HT), VIDI Grant
(project number 91717331, awarded to RPP) and the Sophia Children’s
Hospital Foundation (SSWO, grant S17–19, awarded to TIMK, HT,
and RPP). The MRI component of the study was funded by ZonMw
(TOP project 91211021, awarded to TW). The contributions of the
collaborators and technicians at the diagnostic and immunology
laboratories are highly appreciated. The Generation R study is run by the
Erasmus Medical Center (Rotterdam, Netherlands) in close collaboration
with the School of Law and Faculty of Social Sciences of the Erasmus
University Rotterdam, the Municipal Health Articles Service for the
Rotterdam area, the Rotterdam Homecare Foundation, and the Stichting
Trombosedienst and Artsenlaboratorium Rijnmond (Rotterdam). We
gratefully acknowledge the contribution of children and parents, general
practitioners, hospitals, midwives, and pharmacies in Rotterdam. The
general design of the Generation R Study is made possible by financial
support from the Erasmus Medical Center, Erasmus University
Rotterdam, ZonMw, The Netherlands Organisation for Scientific
Research, the Netherlands Ministry of Health, Welfare, and Sport,
and the Netherlands Ministry of Youth and Families.
References
1 de Escobar GM, Obregón MJ, del Rey FE. Maternal thyroid
hormones early in pregnancy and fetal brain development.
Best Pract Res Clin Endocrinol Metab 2004; 18: 225–48.
2 Bernal J. Thyroid hormones and brain development. Vitam Horm
2005; 71: 95–122.
3 Tau GZ, Peterson BS. Normal development of brain circuits.
Neuropsychopharmacology 2010; 35: 147–68.
4 Iskaros J, Pickard M, Evans I, Sinha A, Hardiman P, Ekins R.
Thyroid hormone receptor gene expression in first trimester
human fetal brain. J Clin Endocrinol Metab 2000; 85: 2620–23.
5 Thorpe-Beeston JG, Nicolaides KH, Felton CV, Butler J,
McGregor AM. Maturation of the secretion of thyroid hormone and
thyroid-stimulating hormone in the fetus. N Engl J Med 1991;
324: 532–36.
6 Henrichs J, Ghassabian A, Peeters RP, Tiemeier H. Maternal
hypothyroxinemia and effects on cognitive functioning in
childhood: how and why? Clin Endocrinol 2013; 79: 152–62.
7 Haddow JE, Palomaki GE, Allan WC, et al. Maternal thyroid
deficiency during pregnancy and subsequent neuropsychological
development of the child. N Engl J Med 1999; 341: 549–55.
636 www.thelancet.com/diabetes-endocrinology Vol 7 August 2019
8 Korevaar TI, Muetzel R, Medici M, et al. Association of maternal
thyroid function during early pregnancy with offspring IQ and
brain morphology in childhood: a population-based prospective
cohort study. Lancet Diabetes Endocrinol 2016; 4: 35–43.
9 Roman GC, Ghassabian A, Bongers-Schokking JJ, et al. Association
of gestational maternal hypothyroxinemia and increased autism
risk. Ann Neurol 2013; 74: 733–42.
10 Modesto T, Tiemeier H, Peeters RP, et al. Maternal mild thyroid
hormone insufficiency in early pregnancy and
attention-deficit/hyperactivity disorder symptoms in children.
JAMA Pediatr 2015; 169: 838–45.
11 Ghassabian A, El Marroun H, Peeters RP, et al. Downstream effects
of maternal hypothyroxinemia in early pregnancy: nonverbal IQ
and brain morphology in school-age children.
J Clin Endocrinol Metab 2014; 99: 2383–90.
12 Vermiglio F, Lo Presti VP, Moleti M, et al. Attention deficit and
hyperactivity disorders in the offspring of mothers exposed to
mild-moderate iodine deficiency: a possible novel iodine deficiency
disorder in developed countries. J Clin Endocrinol Metab 2004;
89: 6054–60.
13 Gyllenberg D, Sourander A, Surcel HM, Hinkka-Yli-Salomaki S,
McKeague IW, Brown AS. Hypothyroxinemia during gestation and
offspring schizophrenia in a national birth cohort. Biol Psychiatry
2016; 79: 962–70.
14 Willoughby KA, McAndrews MP, Rovet JF. Effects of maternal
hypothyroidism on offspring hippocampus and memory. Thyroid
2014; 24: 576–84.
15 Lischinsky JE, Skocic J, Clairman H, Rovet J. Preliminary findings
show maternal hypothyroidism may contribute to abnormal cortical
morphology in offspring. Front Endocrinol 2016; 7: 16.
16 Samadi A, Skocic J, Rovet JF. Children born to women treated for
hypothyroidism during pregnancy show abnormal corpus callosum
development. Thyroid 2015; 25: 494–502.
17 Morte B, Manzano J, Scanlan T, Vennstrom B, Bernal J. Deletion of
the thyroid hormone receptor alpha 1 prevents the structural
alterations of the cerebellum induced by hypothyroidism.
Proc Natl Acad Sci USA 2002; 99: 3985–89.
18 Forrest D, Hallbook F, Persson H, Vennstrom B. Distinct functions
for thyroid hormone receptors alpha and beta in brain development
indicated by differential expression of receptor genes. Embo J 1991;
10: 269–75.
19 Kooijman MN, Kruithof CJ, van Duijn CM, et al. The Generation R
Study: design and cohort update 2017. Eur J Epidemiol 2016;
31: 1243–64.
20 Medici M, de Rijke YB, Peeters RP, et al. Maternal early pregnancy
and newborn thyroid hormone parameters: the Generation R study.
J Clin Endocrinol Metab 2012; 97: 646–52.
21 White T, Muetzel RL, El Marroun H, et al. Paediatric population
neuroimaging and the Generation R Study: the second wave.
Eur J Epidemiol 2018; 33: 99–125.
22 Fischl B, van der Kouwe A, Destrieux C, et al. Automatically
parcellating the human cerebral cortex. Cereb Cortex 2004;
14: 11–22.
23 Hibar DP, Stein JL, Renteria ME, et al. Common genetic variants
influence human subcortical brain structures. Nature 2015;
520: 224–29.
24 Ghassabian A, Henrichs J, Tiemeier H. Impact of mild thyroid
hormone deficiency in pregnancy on cognitive function in children:
lessons from the Generation R Study.
Best Pract Res Clin Endocrinol Metab 2014; 28: 221–32.
25 Selvin S. Statistical analysis of epidemiologic data. New York, NY:
Oxford University Press; 1996: 213–14.
26 Sterne JA, White IR, Carlin JB, et al. Multiple imputation for
missing data in epidemiological and clinical research: potential
and pitfalls. BMJ 2009; 338: b2393.
27 WHO. Urinary iodine concentrations for determining iodine status
deficiency in populations. Vitamin and mineral nutrition
information system. Geneva: World Health Organization; 2013.
https://www.who.int/vmnis/indicators/urinaryiodine/en/ (accessed
April 13, 2019).
28 Lazarus JH, Bestwick JP, Channon S, et al. Antenatal thyroid
screening and childhood cognitive function. N Engl J Med 2012;
366: 493–501.

29 Casey BM, Thom EA, Peaceman AM, et al. Treatment of subclinical
hypothyroidism or hypothyroxinemia in pregnancy. N Engl J Med
2017; 376: 815–25.
30 Cooper DS, Pearce EN. Subclinical hypothyroidism and
hypothyroxinemia in pregnancy - still no answers. N Engl J Med
2017; 376: 876–77.
31 Korevaar TIM, Chaker L, Peeters RP. Improving the clinical impact
of randomised trials in thyroidology. Lancet Diabetes Endocrinol
2018; 6: 523–25.
32 Reiss AL, Abrams MT, Singer HS, Ross JL, Denckla MB.
Brain development, gender and IQ in children: a volumetric
imaging study. Brain 1996; 119: 1763–74.
33 Andersen SL, Andersen S, Vestergaard P, Olsen J. Maternal thyroid
function in early pregnancy and child neurodevelopmental
disorders: a Danish nationwide case-cohort study. Thyroid 2018;
28: 537–46.
34 Levie D, Korevaar TIM, Bath SC, et al. Thyroid function in early
pregnancy, child IQ, and autistic traits: a meta-analysis of individual
participant data. J Clin Endocrinol Metab 2018; 103: 2967–79.
35 Balázs R, Kovács S, Cocks WA, Johnson AL, Eayrs JT. Effect of
thyroid hormone on the biochemical maturation of rat brain:
postnatal cell formation. Brain Research 1971; 25: 555–70.
36 Lauder JM. Granule cell migration in developing rat cerebellum:
Influence of neonatal hypo- and hyperthyroidism.
Developmental Biology 1979; 70: 105–15.
www.thelancet.com/diabetes-endocrinology Vol 7 August 2019 637
Articles
37 Cuevas E, Ausó E, Telefont M, De Escobar GM, Sotelo C, Berbel P.
Transient maternal hypothyroxinemia at onset of corticogenesis
alters tangential migration of medial ganglionic eminence-derived
neurons. Eur J Neurosci 2005; 22: 541–51.
38 Wegiel J, Kuchna I, Nowicki K, et al. The neuropathology of autism:
defects of neurogenesis and neuronal migration, and dysplastic
changes. Acta Neuropathol 2010; 119: 755–70.
39 Zoeller RT, Rovet J. Timing of thyroid hormone action in the
developing brain: clinical observations and experimental findings.
J Neuroendocrinol 2004; 16: 809–18.
40 Herting MM, Sowell ER. Puberty and structural brain development
in humans. Front Neuroendocrinol 2017; 44: 122–37.
41 Korevaar TIM, Tiemeier H, Peeters RP. Clinical associations of
maternal thyroid function with foetal brain development:
epidemiological interpretation and overview of available evidence.
Clin Endocrinol 2018; published online April 24.
DOI:10.1111/cen.13724.
42 Julvez J, Alvarez-Pedrerol M, Rebagliato M, et al. Thyroxine levels
during pregnancy in healthy women and early child
neurodevelopment. Epidemiology 2013; 24: 150–57.