Children Born to Women Treated for Hypothyroidism During Pregnancy Show Abnormal Corpus Callosum Development

Abstract

Background:

Thyroid hormone (TH) is essential for the developing brain, and because the fetal thyroid develops relatively late in gestation, the maternal TH supply is critical for fetal brain development. However, if the mother has hypothyroidism during pregnancy, fetal brain and neuropsychological development may be compromised. Rodents experiencing maternal TH insufficiency show abnormal corpus callosum (CC) morphology, but it is not known if children born to women treated for hypothyroidism (HYPO) show similar effects. The purpose of the current study was to investigate HYPO for CC morphology and morphometry and to determine whether any specific CC abnormalities were associated aspects of maternal hypothyroidism and were correlated with reduced neuropsychological functioning in the children.

Methods:

ANALYZE software was used to trace CCs in archived magnetic resonance imaging scans from 22 HYPO and 22 matched controls. Areas of two sub-regions and six segments and different shape metrics (angles, lengths, ratios) were determined. CC parameters were correlated with maternal thyrotropin (TSH) values and number of hypothyroid trimesters as well as the child's neuropsychological test performance.

Results:

HYPO showed a smaller anterior CC and genu and larger posterior CC and splenium areas than controls as well as shape abnormalities in genu and splenium. Results were correlated with the duration of maternal hypothyroidism. Executive function skills were positively associated with genu size in HYPO, while verbal comprehension skills were negatively associated with splenium and overall posterior CC sizes.

Conclusions:

Maternal hypothyroidism contributes to CC abnormalities in the offspring, and effects differ for anterior versus posterior CC regions.

Introduction

Thyroid hormone (TH), through its regulatory effect on gene expression (1), is essential for such key aspects of brain development as neurogenesis, gliogenesis, process growth, neuronal migration, and myelination (2 –5). Studies with rodents show exposure to insufficient TH gestationally and perinatally leads to myelin sheath malformations and reduced size of white-matter tracts (6,7). The corpus callosum (CC), which is the brain's largest white-matter tract connecting homologous regions of right and left hemispheres (8), is particularly vulnerable to early TH deficiency and has aberrant cortical projections (7).

In the human brain, the need for TH begins very early in gestation, as evidenced by the presence of TH and TH receptors in the first trimester (9,10). Since the fetal thyroid matures relatively late and does not achieve full function until near term, placentally transferred TH from the mother serves all fetal brain needs in early gestation and a supplementary role later. However, if the mother has hypothyroidism, hypothyroxinemia, or subclinical hypothyroidism, her TH contribution to the fetus will be insufficient and the child may have a suboptimal outcome. Indeed, studies of the offspring of these women show reduced IQs (11), memory and attention difficulties (12,13), behavioral and regulatory problems (14), and school underachievement (12). Recent magnetic resonance imaging (MRI) studies also indicate a smaller than normal hippocampus and abnormal cortical morphology (15,16), but their CC has not been studied.

The human CC is comprised of six segments (see Fig. 1), each with a distinct developmental trajectory that begins in early gestation (17). Anterior CC regions, which include the genu and anterior and posterior midbodies and connect with the superior frontal, supplementary premotor, and primary motor cortices, respectively, start to develop at 14–16 weeks of gestation (17). Next to develop at about 18–19 weeks of gestation are the posterior CC sub-regions, namely the isthmus and splenium, which connect sensory cortex and parietal, temporal, and occipital cortical regions (18,19). The rostrum, which is inferior to the genu and connects corresponding regions of the orbitofrontal lobes, forms last around midgestation (17). Notably, these times are when the fetal brain is under maternal TH control.

FIG. 1.

Corpus callosum (CC) segments on the mid-sagittal magnetic resonance imaging (MRI) view.

In light of recent studies showing CC abnormalities in pediatric conditions such as autism (19) and ADHD (20), which are both coincidentally increased in children exposed to maternal hypothyroidism (14,21), the current study investigated (a) whether CC morphology is abnormal in children born to women treated for hypothyroidism during pregnancy (HYPO); (b) if this reflects the severity and duration of their mothers' hypothyroidism; and (c) whether, in HYPO, CC abnormalities and specific neuropsychological weaknesses are linked.

Methods

Participants

Participants were 44 individuals aged 9–14 years (M=10.6 years; SD=0.78 years; 68.2% male) derived from a cohort born between 1996 and 2001 and followed since infancy. The current sample represents approximately one third of subjects from the original cohort and did not differ from non-participants in maternal or other background factors.

The HYPO group included 22 children whose mothers were diagnosed de novo during pregnancy (17%) and so underwent a brief period of hypothyroidism or who were treated for hypothyroidism that predated pregnancy but had not had an increment in thyroxine dosage during pregnancy (83%). All women were recruited during or right after pregnancy by local endocrinologists and obstetricians or from the Motherisk program at The Hospital for Sick Children (SickKids), an antenatal counseling service for prenatal medication exposures including L-thyroxine (L-T4). Table 1, which presents trimester-specific thyrotropin (TSH) values, shows a broad range of values. Six women (27.3%) were hypothyroid for one trimester only, 10 (45.5%) for two trimesters, and four (18.2%) for all three trimesters, while two (9.1%) on L-T4 had normal TH levels throughout pregnancy. In these women, the mean daily pregnancy dose of L-T4 was 105±39 μg (range 50–200 μg).

Table 1.

Maternal Disease Information by Trimester for HYPO Group

	Trimester 1	Trimester 2	Trimester 3
Mean±SD TSH value (mIU/L)^a	9.85±8.44	6.39±7.87	3.27±3.50
TSH range	2.30–38.50	0.14–38.50	0.10–17.60
% above cutoff	90.9	63.6	18.2
fT4^b	11.15±2.90	12.77±1.96	12.27±2.09

TSH upper normal values were 2.5 mIU/L for the first trimester, 3.0 mIU/L for the second trimester, and 3.5 mIU/L for the third trimester.

10 individuals were missing values for first and second trimesters, and 13 were missing values for the third trimester.

HYPO, hypothyroidism; TSH, thyrotropin; fT4, free thyroxine.

Controls were 22 normally developing children also followed since infancy. Their mothers were recruited via posters displayed in offices of local endocrinologists and obstetricians or from the Motherisk database of women with no known teratogenic exposure. A small subset of women was also recruited directly by one obstetrician who measured her patients' TH levels at the first prenatal visit and determined these women were euthyroid. Even though TH levels in pregnancy were unavailable for the remaining women, all reported never being treated for hypothyroidism throughout the course of their child's follow-up. The current sample of controls was selected from a larger group and matched for age (within six months), sex, and socioeconomic status (SES) with individual HYPO cases.

Procedures

Data were acquired from a larger study of TH and memory (15). In this study, children on day 1 received a four-hour test battery assessing intelligence, reading, memory, attention, and aspects of executive functioning, and on day 2, approximately one week later, they underwent MRI scanning. Parents, while their children were being tested, concurrently completed questionnaires. The SickKids Research Ethics Board approved all procedures.

Neuropsychological tests

Intelligence was assessed using the Vocabulary and Matrix Reasoning subtests of the Wechsler Abbreviated Scale of Intelligence (WASI), and reading was assessed with the Wide Range Achievement Test-3 (WRAT-3) Reading subtest. The complete Test of Memory and Learning (TOMAL), copy and delayed recall conditions of the Rey–Osterrieth Complex Figure task (ROCF), and Dot Locations and Stories subtests of the Children's Memory Scale (CMS) were used to assess memory. Attention and executive functioning were assessed with the Map Mission and Score DT subtests of the Test of Everyday Attention for Children (TEACh), and Rapid Visual Information, Spatial Working Memory, and Stockings of Cambridge subtests of the CANTAB.

Parent questionnaires included: (a) a demographic history checklist on the child's birth and developmental history, current health, school status, family history, and family background (for computing SES); (b) the Behavioral Rating Inventory of Executive Function (BRIEF), an 86-item questionnaire evaluating eight aspects of executive function (e.g., shift, inhibit, working memory, planning); and (c) the Everyday Memory Questionnaire (EMQ), a 28-item rating scale of children's everyday memory difficulties.

Image acquisition, preprocessing, and morphological analysis of CC

Structural images were obtained on a 1.5 Tesla GE Signa scanner using a 3D-FSPGR (IR prep Zip X2) T1-weighted series in the axial plane (TR=10.37 ms, TE=4.26 ms, TI=400 ms, flip angle=20°, field of view=240 mm, slice thickness=1.50 mm, no gap). Preprocessing involved extracting brain from skull and cerebral spinal fluid and then normalizing the extracted brain in standard space using anterior and posterior commissure alignment, as is standard in neuroimaging analyses.

Scans were examined in ANALYZE 9.0, and a subtle increase in contrast was used to facilitate white-matter recognition. The following landmarks were identified in the mid-sagittal slice: aqueduct of sylvius, third ventricle, and interhemispheric fissure. ANALYZE's Region of Interest (ROI) tool was used to manually trace the CC in the sagittal view; coronal and axial views were used for confirmation. The intracranial area (ICA), defined as the cross-sectional area of the brain in the mid-sagittal slice excluding the cerebellum and brainstem, was also measured.

The morphometric analysis was based on a modification of Hofer and Frahm's system (18), originally derived from their diffusion tensor imaging (DTI) fiber tractography findings (Fig. 2A). This system identifies five CC segments that project to distinct cortical areas; the current modification involved also including the rostrum. Initially, a horizontal line was extended between vertical tangents at anterior-most and posterior-most ends, and the resulting region was partitioned into halves, thirds, and quarters (Fig. 2B). Within the posterior half, the posterior-most quarter signified the splenium, while the region within the posterior-most third minus the splenium signified the isthmus; the remaining region after removing both splenium and isthmus was the posterior midbody. Within the anterior half, a vertical line drawn at the anterior border of the septum pellucidum (SP, ∼1/6 CC) served to designate the genu as the part anterior to the line and the rostrum as the part posterior to the line and inferior to the SP; the anterior midbody was the remainder posterior to the line and superior to the SP. Based on knowledge about the developmental timing of different CC segments (22), the genu, anterior midbody, and posterior midbody were assigned to the anterior CC subregion, and the isthmus and splenium to the posterior CC subregion; the rostrum was not included in this designation. Areas of the six individual segments and the anterior/posterior sub-regions were measured within ANALYZE. A second individual also performed tracings on a subset of scans and obtained an inter-rater reliability of 0.97 with the main analyst. Both analysts were masked to subject group, age, and sex.

FIG. 2.

CC connections and measurement methods. (A) Cortical projections of CC determined using diffusion tensor imaging tractography. Figure adapted from Hofer and Frahm (18). (B) Tangents at anterior-most (a) and posterior-most (b) used to divide CC into halves (c), thirds (d), and quarters (e) in order to define posterior midbody (III), isthmus (IV), and splenium (V). Anterior border of the septum pellucidum (f) used to define genu (I), anterior midbody (II), and rostrum (VI). (C) Tangent lines at the anteroinferior surface of the genu going through the anterior commissure, anterosuperior surface of the anterior midbody, posteriosuperior surface of isthmus, and floor of the fourth ventricle define a, δ, and ɛ angles.

The method of Gabrielli et al. (23) was used to characterize aspects of CC shape. First, four tangent lines were drawn (Fig. 2C): one from the anteroinferior surface of the genu through the anterior commissure; a second at the anterosuperior surface of the anterior midbody; a third at the posterosuperior surface of the isthmus; and a fourth along the floor of the fourth ventricle. Different line pairs were connected to form the following three angles: (a) “α” between the first two lines reflecting genu bulbosity; (b) “δ” between the second and third lines reflecting CC curvature; and (c) “ɛ” between the third and fourth lines reflecting splenium orientation. Genu and splenium plumpness/skinniness were also determined by computing the ratios between their respective heights (lengths) and widths.

Statistical analysis

All data were analyzed using IBM SPSS Statistics for Windows v21 (IBM Corp., Armonk, NY). t-Tests and chi-square tests were used to compare groups for demographics and intracranial and total CC areas. Separate multivariate analyses of covariance (MANCOVA) with ICA as the covariate were used to examine group differences in areas of CC subregions or segments as well as linear measurements and angles. Proportional CC areas determined by calculating ratios between segment area and total CC area were analyzed via multivariate analyses of variance (MANOVA). Similarly, neuropsychological test data were analyzed by separate MANOVAs for each test. For MANCOVA and MANOVA, both omnibus (across all effects) and univariate results are reported. Effect sizes (η²) were determined for each analysis. Only cases whose mothers had at least one abnormal TSH value in pregnancy were included in this set of analyses, since the two women who were euthyroid throughout gestation despite no increase in treatment were not technically hypothyroid in pregnancy.

Within the HYPO group, trimester-specific maternal TSH values were correlated with CC measures. The entire HYPO sample was included in these analyses (including those with normal TSH throughout pregnancy) in order to examine the full range of effects. To assess duration of hypothyroidism, the HYPO group was subclassified by the number of trimesters involving an elevated maternal TSH level based on trimester-specific reference ranges; for example, mothers first diagnosed in the second trimester of this pregnancy were considered to have two trimesters with TSH elevations, and mothers who were euthyroid throughout pregnancy were assigned a score of zero. These values were entered into linear regression analyses to predict size of each CC sub-region and segment. To determine structure–function relationships, correlations were performed (within the HYPO group only) between specific sub-region or segment ratio values and neuropsychological test scores.

The p-value was set at 0.05 for group comparisons and at 0.01 for correlations with TSH levels. Because of the large number of neuropsychological test indices, the critical p-value for these correlations was set at 0.001. Given the exploratory nature of the study, a few near-significant trends are also mentioned (e.g., p<0.10 for group comparisons and p<0.05 for correlations).

Results

Table 2 presents the groups' demographic data. They did not differ in age, sex, SES, handedness, IQ, or ICA.

Table 2.

Demographic Data for HYPO and Control Groups

	Hypo (n=22)	Control (n=22)
Sex: males, n (%)	15 (68.2%)	15 (68.2%)
Age	10.25±0.59	10.86±0.83
SES	1.86±0.77	1.77±0.75
Right-handed, n (%)	21 (95.5%)	21 (95.5%)
Full-scale IQ	113.1±12.5	115.7±9.9
ICA (cm²)	117.2±8.4	117.2±10.8

Results expressed as mean±standard deviation; groups do not differ on any variable.

Table 3 provides area and ratio measurements for CC subregions and segments. These data were analyzed via MANCOVAs (area measurements) and MANOVAs (ratio measurements). Although groups did not differ in ICA, this was included as a covariate in the MANCOVA given its strong correlation with total CC area (p=0.004). Although HYPO and control groups did not differ in their area values, it was found that when ratio data were used, significant differences were seen for both sub-regions and segment values. For sub-regions, a near-significant omnibus group effect (p=0.059, η²=0.135) reflected significantly smaller anterior (p=0.019, η²=0.130) and larger posterior sub-regions (p=0.018, η²=0.132) in HYPO. For individual CC segments, a significantly larger splenium ratio was observed (p=0.045, η²=0.097) in HYPO who also had a smaller genu, at a trend level (p=0.059, η²=0.086), than controls. Groups did not differ in size of the other segments or in their linear or angle measurements (Table 4).

Table 3.

Mean±Standard Deviation Area and Proportion Values of Total CC and CC Sub-Regions and Segments for HYPO and Control Groups

	Hypo^a	Control	p	η²
Total CC
Subarea in mm²
Anterior CC	285.6±61.7	300.3±44.1	ns	0.020
Posterior CC	190.1±33.7	180.1±28.7	ns	0.026
Genu	108.7±28.9	119.2±25.5	ns	0.038
Anterior midbody	133.6±26.0	135.8±20.6	ns	0.002
Posterior midbody	43.3±11.2	45.4±10.1	ns	0.011
Isthmus	21.3±6.7	19.3±4.8	ns	0.040
Splenium	168.8±29.5	160.8±27.8	ns	0.020
Rostrum	12.5±8.4	11.9±4.9	ns	0.003
Total CC	488.1±91.0	492.3±68.9	ns	0.122
Proportions^b
Anterior CC	0.584±0.04	0.610±0.03	0.019	0.130
Posterior CC	0.392±0.04	0.365±0.03	0.018	0.132
Genu	0.222±0.03	0.242±0.03	0.059	0.086
Anterior midbody	0.274±0.02	0.276±0.03	ns	0.002
Posterior midbody	0.088±0.01	0.092±0.02	ns	0.022
Isthmus	0.043±0.01	0.040±0.01	ns	0.040
Splenium	0.348±0.04	0.326±0.03	0.045	0.097
Rostrum	0.025±0.02	0.024±0.01	ns	0.001

Does not include the two cases whose mothers had normal TSH values throughout pregnancy.

Adjusted for total CC area.

CC, corpus callosum.

Table 4.

Comparison of Means Between Groups for Linear, Proportion, and Angle Measurements

	HYPO^a	Control	p
α-angle	36.22±5.61	34.60±6.24	0.384
δ-angle	141.37±6.88	139.67±8.68	0.487
ɛ-angle	116.07±9.10	117.29±5.47	0.596
Height of genu	15.05±3.15	15.13±2.17	0.923
Width of genu	9.23±1.93	9.76±1.52	0.332
Length of splenium	22.92±2.14	22.50±2.77	0.588
Width of splenium	11.15±1.41	11.05±1.67	0.851
Width over height of genu	0.64±0.18	0.66±0.13	0.693
Width over length of splenium	0.49±0.08	0.50±0.11	0.726

Does not include the two cases whose mothers had normal TSH values throughout pregnancy.

No significant correlations were obtained between CC indices and trimester-specific maternal TSH values. However, when children were classified by the number of trimesters (zero to three) in which their mothers' TSH levels were elevated, results revealed an effect for the genu (b=−0.45, t(22)=−2.26, p=0.035), with TSH explaining 20% of variance in its area. This suggests that for mothers with a longer period of TH insufficiency, their children's genua were smaller. Within the entire anterior sub-region, the number of hypothyroid trimesters also significantly predicted its reduced size (b=−0.44, t(18)=6.53, p=0.054) accounting for 44% of variance in its area (Table 5). Although no effects were observed for shape indices, children exposed to more hypothyroid trimesters tended to have more bulbous genua (α-angle; p=0.07).

Table 5.

Results for Regression Analyses Predicting Segmental and Regional Area from Number of Trimesters Exposed to Elevated TSH

		Unstandardized coefficients		Standardized coefficients			CI for B
Variable	n	B	Std. error	Beta	t	Sig.	Lower bound	Upper bound
Genu/CC	22	−0.018	0.008	−0.451	−2.257	0.035	−0.034	−0.001
	20	−0.017	0.011	−0.352	−1.595	0.128	−0.039	0.005
Anterior midbody/CC	22	0.005	0.005	0.213	0.975	0.341	−0.006	0.016
	20	−0.004	0.006	−0.166	−0.715	0.484	−0.018	0.009
Posterior midbody/CC	22	−0.001	0.003	−0.098	−0.441	0.664	−0.008	0.005
	20	−0.004	0.004	−0.222	−0.967	0.346	−0.012	0.005
Isthmus/CC	22	0.000	0.002	0.025	0.111	0.913	−0.005	0.005
	20	−0.002	0.003	−0.152	−0.650	0.524	−0.008	0.004
Splenium/CC	22	0.011	0.010	0.244	1.126	0.273	−0.009	0.032
	20	0.022	0.013	0.376	1.721	0.102	−0.005	0.048
Rostrum/CC	22	0.002	0.004	0.140	0.633	0.534	−0.006	0.010
	20	0.005	0.005	0.243	1.061	0.303	−0.005	0.016
Anterior CC/CC	22	−0.014	0.009	−0.318	−1.500	0.149	−0.034	0.006
	20	−0.025	0.012	−0.437	−2.061	0.054	−0.051	0.000
Posterior CC/CC	22	0.011	0.010	0.250	1.153	0.263	−0.009	0.031
	20	0.020	0.013	0.345	1.561	0.136	−0.007	0.047
α-angle	22	0.320	1.466	0.049	0.218	0.829	−2.739	3.379
	20	3.232	1.678	0.413	1.927	0.070	−0.292	6.757

CI, confidence interval.

For the different neuropsychological tests, separate MANOVAs indicated significant between-group differences on the BRIEF Emotional Control subscale (p=0.03) and on the EMQ (p=0.007) and at a trend level (p=0.07) on the BRIEF Working Memory scale. These results reflected the higher scores (i.e., more problems) by HYPO than controls. HYPO also tended to have poorer reading ability than controls (p=0.07; Table 6).

Table 6.

Comparison Between Groups for Significant Differences in Neuropsychological Test Scores Presented as Means Difference±Standard Deviation

	HYPO^a	Control	p-Value
BRIEF Emotional Control T-Score^b	53.20±12.50	46.06±5.82	0.031
BRIEF BRI T-score	52.35±12.26	46.50±5.53	0.067
BRIEF Working Memory T-Score	54.90±13.16	48.06±8.36	0.067
EMQ Score^c	7.93±1.79	6.60±0.90	0.007
WRAT 3 Reading Standard Score	108.75±11.91	1115.81±10.25	0.069

Does not include the two cases whose mothers had normal TSH values throughout pregnancy.

T-score: mean=50, standard deviation=10; standard score: mean=100, standard deviation=15.

Score represents the square root of the sum across all ratings.

Table 7 shows the significant correlations between CC sub-region and segment values and neuropsychological test results for the HYPO group. Those with larger genua had fewer problems on the BRIEF Shift subscale, while those children with plumper genua (larger width:height ratios) scored higher on Rey-O Delay and TOMAL Object Recall, Paired Recall, and Verbal Memory Index. Children with smaller splenium ratios scored higher on the WASI Vocabulary subtest, while children with wider and shorter splenia scored higher on TOMAL Digits Forward and Backward, Paired Recall, and Abstract Visual Memory. Further, those with smaller ɛ-angles (i.e., more upright splenia) performed less well on the TOMAL Visual Sequential Memory subtest. Finally, those with a larger posterior midbody showed more initiative (lower BRIEF Initiate scores) and better narrative recall (higher CMS Delayed Story Recall scores).

Table 7.

Correlation Between CC Measurements and Neuropsychological Test Scores in HYPO

	Neuropsychological test	r	p
Genu/CC	BRIEF Shift T-Score^*	−0.464	0.039
Posterior midbody/CC	BRIEF Initiate T-Score	−0.460	0.041
	CMS Stories Delayed	0.456	0.043
Splenium/CC	WASI Vocab Scaled Score^**	−0.483	0.031
Width over height of genu	Rey-O Delay	0.498	0.025
	TOMAL Object Recall	0.560	0.010
	TOMAL Paired Recall	0.522	0.018
	TOMAL Verbal Memory Index	0.594	0.006
Width over length of splenium	TOMAL Digit Forward	0.570	0.009
	TOMAL Digit Backward	0.678	0.001
	TOMAL Paired Recall	−0.487	0.029
	TOMAL Abstract Visual Memory	0.494	0.027
ɛ-angle	TOMAL Visual Sequential Memory	−0.545	0.013

Higher scores on BRIEF test indicate more problems in the corresponding subtest category.

Higher scores Rey, WASI, TOMAL, and CMS indicate better performance in the corresponding subtest category.

Discussion

The present study found that insufficient maternal TH during pregnancy affects the development of specific CC sub-regions in the offspring. Relative to children born to women with normal thyroid function throughout pregnancy, it was found that children whose mothers underwent a period of TH deficiency in pregnancy due to de novo diagnosis or to inadequately treated pre-existing hypothyroidism had smaller anterior and larger posterior CCs. These changes consist of a smaller genu and a larger splenium size. The CC shape was generally unaffected. Although CC morphometry and morphology indices were not correlated with average maternal TSH levels within a single trimester, an association was observed between the duration of the maternal hypothyroidism and CC morphometry. Specifically, children from women who were hypothyroid for more than one trimester had smaller genua and anterior CCs and larger splenia than did children whose mothers were hypothyroid for a shorter duration. Also within the HYPO group, genu and splenium size and certain shape features were correlated with a number of the neuropsychological test indices.

Since the advent of MRI for research and use of advanced analytical procedures, much interest has been garnered for studying the CC, possibly because this is the most detectable structure within the brain and it can be severely malformed or even missing in certain pediatric populations (24). Furthermore, there is a substantial literature showing that adults with callosal transections have distinct cognitive, social, and emotional deficits (25). Importantly also, findings from animal, histological, and recent MRI studies show that the CC undergoes major development in a non-uniform (17,24) and bidirectional (26) fashion during pregnancy. In animals, callosal development involves a period of exuberant axonal growth followed by axonal pruning in late gestation and myelination of axons postnatally (22). Indeed, this period of exuberant connections is thought to play a significant role in cortical evolution (27). Furthermore, in primates, the axonal properties underlying the complex system of intercommunication with specific regions of the cortex has been found to reflect the specific geometries of the callosal and cortical regions with which they connect (28). Also, populations suffering from psychiatric disorders show CC shape abnormalities (29,30), while preterm infants and children with traumatic brain injury or autism show increased or decreased CC size overall or within selective regions (31,32). The present findings indicating that offspring of hypothyroid women had smaller than normal genua and larger than normal splenia are novel and suggest that the human CC is indeed vulnerable to TH insufficiency during pregnancy.

Rodents with early TH deficiencies exhibit CC abnormalities that reflect atypical topography of callosal projections (7), reduced number and abnormal maturation and myelination of axons (6 –8), and malformations of the CC itself (33). The present findings of CC abnormalities in humans, particularly children who experienced longer periods of gestational TH deficiency, suggest TH insufficiency is the likely culprit. However, it was not possible to pinpoint when within individual trimesters the specific effects occurred, since most women were receiving some (albeit inadequate) treatment or were treated soon after experiencing symptoms (in de novo cases). Nevertheless, the observation of strongest effects in children exposed to TH insufficiency during the third trimester suggests (based on animal evidence) that this may be due to reduced callosal sprouting in the genu and reduced pruning in the splenium (34). Furthermore, the findings also suggest that to exert an effect, the impact of TH loss may be cumulative, since correlations were not shown with averaged TSH values for each trimester.

The present findings also indicate a plumper genu in HYPO than controls, while observed genu abnormalities were correlated with increased difficulties in shifting attention. Since the genu contains fibers connecting homologous frontal lobe regions that play a role in aspects of executive functioning, particularly attention, the current findings offer a mechanism as to why HYPO have specific executive function difficulties. Interestingly, research from the Generation R study shows children exposed to maternal hypothyroxinemia have an increased risk of autism, which may reflect specific executive-function difficulties (14).

The larger than normal splenium in the HYPO group may reflect an abnormality in the pruning process that typically follows exuberant axonal growth (35) and is disrupted by exposure to hypothyroidism in rodents (34). Splenium size was also observed to be inversely correlated with WASI Vocabulary performance, signifying maternal TH insufficiency may have a negative impact on language comprehension and expression. Since the splenium has crossing fibers from occipital, parietal, and temporal lobes, some of which are involved in the formation and storage of language information (e.g., Wernicke's area), the present findings suggest this may be a possible neural substrate for the observed verbal weaknesses in HYPO (36). Furthermore, given that splenium enlargement is observed in obsessive-compulsive disorder and familial major depression disorder (37,38), HYPO may be at risk for these problems as well.

It is not readily clear why the genu and the splenium were the most affected CC sub-regions. However, this may reflect the exact timing of TH deficiency in relation to the stage of each segment's development as well as to the severity of the hypothyroidism. Other factors may include TH availability from iodine deficiency or exposure to environmental TH disruptors (39). Although the genu and splenium differ in number of myelinated axons (fewer in the splenium), this is an unlikely explanation given that CC myelination only starts after birth (22), which was not when the present sample was TH deficient. Furthermore, TH deficiency may indirectly alter callosal projections by disrupting cortical microarchitecture (40,41), given that manifestations in different cortical areas can influence the integrity of callosal projections (35).

Since the CC is the brain's primary interconnecting fiber bundle, abnormalities in its structural and anatomical organization can affect multiple mental functions in different ways that reflect the specific structure of the CC and the nature of the projections to the cortex such as axonal diameter and temporal delays (42). Previous studies have shown an association between CC surface area and social deficits in children with traumatic brain injury (32) and between CC length and executive function abilities in normal preschoolers (21). Similarly, the present findings suggest a functional impact of reduction in area and plumpness of genu on emotion regulation, as observed in newborns whose mothers were hypothyroxinemic (43). Conversely, callosal enlargement seen in children with neurofibromatosis-1 has been associated with nonverbal learning disability and impaired visuospatial and visuomotor skills (17). Although overgrowth in the present HYPO sample may contribute to similar deficits, the current study did not examine these particular abilities. However, Pop et al. have shown that children exposed gestationally to hypothyroxinemia had weak perceptuomotor skills, and this became more prominent as the number of hypothyroxinemic trimesters increased (11).

A strength of the current research includes our having one of the largest and longest-followed samples of well-characterized offspring born to women treated for hypothyroidism during pregnancy and similarly followed controls, all recruited originally as infants and now studied in peri-adolescence. Likewise, use of MRI and manually tracing the CC in children whose CC maturity is near complete is novel. Additionally, the CC segmentation approach was superior to traditional arbitrary sectioning methods in most CC research (44), since that used in our study was derived from known functional projections to the cortex.

However, this study does have some limitations. Regarding hormonal data, TSH values were averaged within trimesters, and in a few cases, group averages were used if data were missing in a particular trimester. Also, there was insufficient T4 data to perform these correlations. Further, the study did not record when symptoms first appeared in women diagnosed de novo in pregnancy, and it was assumed that this occurred at the onset of pregnancy when in fact it may not have appeared until the second trimester. Two women had normal TSH levels throughout pregnancy, and in only a few cases, mothers were hypothyroid in the third trimester, which is when CC sprouting occurs. Second, there was a lack of hormonal data on the full set of controls, and thus some of these women may have had subclinical if not clinical hypothyroidism during their pregnancy. However, it should be noted that in the subset that was tested, none were hypothyroid, and all mothers of control children verbally reported not being hypothyroid in pregnancy and at every assessment period when their children were seen. A third limitation is that the small sample size represents only about a third of the original cohort given the five- to seven-year gap between the current assessment and one previously at age five. Fourth, some children may have been exposed to factors other than maternal hypothyroidism such as poor nutrition or enrichment that could negatively or positively impact subsequent CC morphology. Fifth, the test battery was limited mostly to memory measures given that this was the main focus of the project for which scans were originally derived. Thus, abilities such as visuospatial skills could not be examined that may more accurately reflect the observed CC abnormalities. Six, the study did not initially set out to analyze sex differences in CC metrics. However, in light of research showing a larger overall CC area in males and a larger splenium in females (45), a supplementary analysis was performed examining for sex differences and none were found. Finally, effects on CC microstructure were not investigated that would determine how the degree of maternal TH insufficiency directly affects myelination and number, size, and integrity of callosal fibers. Therefore, future studies need to use DTI and DTI tractography techniques to evaluate the CC and other white-matter structures better.

In conclusion, the present study shows that maternal TH is necessary for early CC development and, when deficient, leads to abnormalities in size and shape of the genu and splenium. These abnormalities reflect primarily the duration of maternal TH insufficiency and are associated with suboptimal neuropsychological test performance and behavioral problems, particularly poor emotion regulation. Nevertheless, in order to understand the role of TH better on the many facets of CC development, future research needs to examine other conditions involving insufficient TH at different times during pregnancy or the postpartum, such as iodine deficiency and congenital hypothyroidism.

Footnotes

Acknowledgments

The authors would like to thank Anishka Greene, Dragana Ostojic, Erin Sheard, Michelle Vitti, and Karen Willoughby for their assistance in data collection; Garry Detzler, Tammy Rayner, and Ruth Weiss for obtaining MR scans; Susan Blaser for reading neuroradiological scans; Hayyah Clairman for her assistance in editing the article; and Charles Raybaud for enlightening us on corpus callosum development. Our extreme gratitude goes to the children and families whose dedication to this research and willingness to participate have made these findings possible. Funding for this research was through a Canadian Institute of Health Research operating grant (MOP 49488) and an SSuRe Summer Student Scholarship, while funding to ascertain the original sample was through the American Thyroid Association and March of Dimes.

Author Disclosure Statement

The authors have no conflicts of interest to disclose.

References

Bernal

. 2007. Thyroid hormone receptors in brain development and function. Nat Clin Pract Endocrinol Metab, 3:249–259.

Morreale de Escobar

, Obregon

, Escobar del Rey

. 2004. Role of thyroid hormone in early brain development. Eur J Endocrinol, 151:U25–37.

Hsu

, Stein

, Xu

. 2008. Abnormal growth of the corticospinal axons into the lumbar spinal cord of the hyt/hyt mouse with congenital hypothyroidism. J Neurosci Res, 86:3126–3139.

Auso

, Lavado-Autric

, Cuevas

, Escobar del Rey

, Morreale de Escobar

, Berbel

. 2004. A moderate and transient deficiency of maternal thyroid function at the beginning of fetal neocorticogenesis alters neuronal migration. Endocrinology, 145:4037–4047.

Schoonover

, Seibel

, Jolson

, Stack

, Rahman

, Jones

, Mariash

, Anderson

. 2004. Thyroid hormone regulates oligodendrocyte accumulation in developing rat brain white matter tracts. Endocrinology, 145:5013–5020.

Ferreira

, Pereira

, Manhaes

, Barradas

. 2007. Ultrastructural identification of oligodendrocyte/myelin proteins in corpus callosum of hypothyroid animals. Int J Devel Neurosci, 25:87–94.

Gravel

, Hawkes

. 1990. Maturation of the corpus callosum of the rat: I. Influence of thyroid hormones on the topography of callosal projections. J Comp Neurol, 291:128–146.

Berbel

, Guadafio-Ferraz

, Angulo

, Cerezo

. 1994. Role of thyroid hormones in the maturation of interhemispheric connections in rats. Behav Brain Res, 64:9–14.

Calvo

, Jauniaux

, Gulbis

, Asuncion

, Gervy

, Contempre

, Morreale de Escobar

. 2002. Fetal tissues are exposed to biologically relevant free thyroxine concentrations during early phases of development. J Clin Endocrinol Metab, 87:1768–1777.

10.

Kilby

, Gittoes

, McCabe

, Verhaeg

, Franklyn

. 2000. Expression of thyroid receptor isoforms in the human fetal central nervous system and the effects of intrauterine growth restriction. Clin Endocrinol, 53:469–477.

11.

Pop

, Brouwers

, Vader

, Vulsma

, van Baar

, de Vijlder

. 2003. Maternal hypothyroxinaemia during early pregnancy and subsequent child development: a 3-year follow-up study. Clin Endocrinol, 59:282–288.

12.

Haddow

, Palomaki

, Allan

, Williams

, Knight

, Gagnon

, O'Heir

, Mitchell

, Hermos

, Waisbren

, Faix

, Klein

. 1999. Maternal thyroid deficiency during pregnancy and subsequent neuropsychological development of the child. New Engl J Med, 341:549–555.

13.

Willoughby

, McAndrews

, Rovet

. 2013. Effects of early thyroid hormone deficiency on children's autobiographical memory performance. J Intern Neuropsychol Soc, 19:419–429.

14.

Roman

, Ghassabian

, Bongers-Schokking

, Jaddoe

, Hofman

, de Rijke

, Verhulst

, Tiemeier

. 2013. Association of gestational maternal hypothyroxinemia and increased autism risk. Ann Neurol, 74:733–742.

15.

Willoughby

, McAndrews

, Rovet

. 2014. Effects of maternal hypothyroidism on offspring hippocampus and memory. Thyroid, 24:576–584.

16.

Rovet

, Lischinsky

, Clairman

, Skocic

. 2012. Maternal hypothyroidism affects offspring cortical cytoarchitecture. American Thyroid Association Annual Meeting, Quebec City.

17.

Paul

. 2011. Developmental malformation of the corpus callosum: a review of typical callosal development and examples of developmental disorders with callosal involvement. J Neurodevel Disord, 3:3–27.

18.

Hofer

, Frahm

. 2006. Topography of the human corpus callosum revisted—comprehensive fiber tractography using diffusion tensor magnetic resonance image. Neuroimage, 32:989–994.

19.

Piven

, Bailey

, Ranson

, Arndt

. 1997. An MRI study of the corpus callosum in autism. Am J Psychiatry, 154:1051–1056.

20.

Hynd

, Semrud-Clikeman

, Lorys

, Novey

, Eliopulos

, Lyytinen

. 1991. Corpus callosum morphology in attention deficit-hyperactivity disorder: morphometric analysis of MRI. J Learn Disabil, 24:141–146.

21.

Ghassabian

, Herba

, Roza

, Govaert

, Schenk

, Jaddoe

, Hofman

, White

, Verhulst

, Tiemeier

. 2013. Infant brain structures, executive function, and attention deficit/hyperactivity problems at preschool age. A prospective study. J Child Psychol Psychiatry, 54:96–104.

22.

Barkovich

, Kjos

. 1988. Normal postnatal development of the corpus callosum as demonstrated by MR imaging. Am J Neuroradiol, 9:487–491.

23.

Gabrielli

, Salvolini

, Bonifazi

, Ciferri

, Lanza

, Rossi

, Coppa

, Giorgi

. 1993. Morphological studies of the corpus callosum by MRI in children with malformative syndromes. Neuroradiology, 35:109–112.

24.

Raybaud

. 2010. The corpus callosum, the other great forebrain commissures, and the septum pellicudum: anatomy, development, and malformation. Neuroradiology, 52:447–477.

25.

Bogen

, Bogen

. 1988. Creativity and the corpus callosum. Psychiatr Clin North Am, 11:293–301.

26.

Kier

, Truwit

. 1996. The normal and abnormal genu of the corpus callosum: an evolutionary, embryologic, anatomic, and MR analysis. AJNR Am J Neuroradiol, 17:1631–1641.

27.

Innocenti

. 1995. Exuberant development of connections, and its possible permissive role in cortical evolution. Trends Neurosci, 18:397–402.

28.

Innocenti

, Vercelli

, Caminiti

. 2014. The diameter of cortical axons depends both on the area of origin and target. Cereb Cortex, 24:2178–2188.

29.

Frumin

, Golland

, Kikinis

, Hirayasu , Salisbury

, Hennen

, Dickey

, Anderson

, Jolesz

, Grimson

, McCarley

, Shenton

. 2002. Shape differences in the corpus callosum in first-episode schizophrenia and first-episode psychotic affective disorder. Am J Psychiatry, 159:866–868.

30.

Innocenti

, Ansermet

, Parnas

. 2003. Schizophrenia, neurodevelopment and corpus callosum. Mol Psychiatry, 8:261–274.

31.

Thompson

, Inder

, Faggian

, Johnston

, Warfield

, Anderson

, Doyle

, Egan

. 2011. Characterization of the corpus callosum in very preterm and full-term infants utilizing MRI. Neuroimage, 55:479–490.

32.

Ewing-Cobbs

, Prasad

, Swank

, Kramer

, Mendez

, Treble

, Payne

, Bachevalier

. 2012. Social communication in young children with traumatic brain injury: relations with corpus callosum morphometry. Int J Dev Neurosci, 30:247–254.

33.

Goodman

, Gilbert

. 2007. Modest thyroid hormone insufficiency during development induces a cellular malformation in the corpus callosum: a model of cortical dysplasia. Endocrinology, 149:2593–2597.

34.

Mohan

, Sinha

, Pathak

, Rastogi

, Kumar

, Pal

, Godbole

. 2012. Maternal thyroid hormone deficiency affects the fetal neocorticogenesis by reducing the proliferating pool, rate of neurogenesis and indirect neurogenesis. Exper Neurol, 237:477–488.

35.

LaMantia

, Rakic

. 1990. Axon overproduction and elimination in the corpus callosum of the developing rhesus monkey. J Neurosci, 10:2156–2175.

36.

Henrichs

, Bongers-Schokking

, Schenk

, Ghassabian

, Schmidt

, Visser

, Hooijkaas

, de Muinck Keizer-Schrama

, Hofman

, Jaddoe

, Visser

, Steegers

, Verhulst

, de Rijke

, Tiemeier

. 2010. Maternal thyroid function during early pregnancy and cognitive functioning in early childhood: the generation R study. J Clin Endocrinol Metab, 95:4227–4234.

37.

Lacerda

, Brambilla

, Sassi

, Nicoletti

, Mallinger

, Frank

, Kupfer

, Keshavan

, Soares

. 2005. Anatomical MRI study of corpus callosum in unipolar depression. J Psychiatr Res, 39:347–354.

38.

Park

, Park

, Kim

, Jang

, Jung

, Choi

, Kang

, Lee

, Kwon

. 2011. Midsagittal structural differences and sexual dimorphism of the corpus callosum in obsessive-compulsive disorder. Psychiatry Res, 192:147–153.

39.

Powell

, Van Nguyen

, Gilbert

, Parekh

, Colon-Perez

, Mareci

, Montie

. 2012. Magnetic resonance imaging and volumetric analysis: novel tools to study the effects of thyroid hormone disruption on white matter development. Neurotoxicol Teratol, 33:1322–1329.

40.

Berbel

, Auso

, Garcia-Velasco

, Molina

, Camacho

. 2001. Role of thyroid hormones in the maturation and organisation of rat barrel cortex. Neuroscience, 107:383–394.

41.

Berbel

, Escobar del Rey

, Morreale de Escobar

, Ruiz-Marcos

. 1985. Effect of hypothyroidism on the size of spines of pyramidal neurons of the cerebral cortex. Brain Res, 337:217–223.

42.

Caminiti

, Carducci

, Piervincenzi

, Battaglia-Mayer

, Confalone

, Visco-Comandini

, Pantano

, Innocenti

. 2013. Diameter, length, speed, and conduction delay of callosal axons in macaque monkeys and humans: comparing data from histology and magnetic resonance imaging diffusion tractography. J Neurosci, 33:14501–14511.

43.

Kooistra

, Crawford

, van Baar

, Brouwers

, Pop

. 2006. Neonatal effects of maternal hypothyroxinemia during early pregnancy. Pediatrics, 117:161–167.

44.

Witelson

. 1989. Hand and sex differences in the isthmus and genu of the human corpus callosum. A postmortem morphological study. Brain, 112:799–835.

45.

Clarke

, Kraftsik

, Van der Loos

, Innocenti

. 1989. Forms and measures of adult and developing human corpus callosum: is there sexual dimorphism?. J Comp Neurol, 280:213–230.