Sex-specific Effect of Maternal Thyroid Hormone Trajectories on Preschoolers’ Behavioral Development: A Birth Cohort Study

Abstract

Context

Maternal thyroid hormone trajectories are a better predictor of offspring’s neurodevelopment than hormone levels in single trimester of pregnancy. Programming effect of uterine hormonal environment on offspring’s health is usually sex-specific.

Objective

To examine the sex-specific effect of thyroid hormone trajectories on preschoolers’ behavioral development.

Design

Based on Ma’ anshan Birth Cohort in China, pregnant women were recruited at their first antenatal checkup from May 2013 to September 2014.

Setting

Ma’ anshan Maternal and Child Health Hospital in China.

Patients or Other Participants

1860 mother-child pairs were included in the analysis. Children were followed up at age of 4.

Main Outcome Measures

Maternal thyroid hormones [thyroid-stimulating hormone (TSH), free thyroxine (FT₄)] and thyroid peroxidase antibody in the first, second, and third trimesters of pregnancy were retrospectively assayed. Preschoolers’ behavioral development was assessed by Achenbach Child Behavior Checklist/1.5~5.

Results

Maternal TSH and FT₄ levels were respectively fitted into high, moderate, and low trajectories. In boys, maternal high TSH trajectory was related to withdrawn [odds ratio (OR) = 2.01, 95% CI: 1.16, 3.50) and externalizing problems (OR = 2.69, 95% CI: 1.22, 5.92), and moderate TSH trajectory was associated with aggressive behavior (OR = 3.76, 95% CI: 1.16, 12.23). Maternal high FT₄ trajectory was associated with anxious/depressed (OR = 2.22, 95% CI: 1.08, 4.56) and total problems (OR = 1.74, 95% CI: 1.13, 2.66), and low FT₄ trajectory was associated with aggressive behavior (OR = 4.17, 95% CI: 1.22, 14.24).

Conclusions

Maternal thyroid hormone trajectories impact preschool boys’ behavioral development.

pregnant women, thyroid hormones, children, behavioral development

Thyroid hormones have multiple physiological functions, which are essential for maintaining normal pregnancy and fetal growth, especially for fetal brain development (1,2). Thyroid-stimulating hormone (TSH) and free thyroxine (FT₄) are reliable indicators of thyroid status, and the diagnosis of clinical and subclinical thyroid disease is mostly based on the levels of these 2 hormones (3,4). Studies have shown that both maternal subclinical hypothyroidism (elevated TSH with normal FT₄) and isolated hypothyroxinemia (decreased FT₄ with normal TSH) during pregnancy were associated with a variety of adverse maternal and child outcomes, including preterm delivery and low birth weight (5,6). Researchers had also found that maternal hyperthyroidism or hypothyroidism may be related to the increased risk of offspring’s abnormal neurobehavioral development, such as attention-deficit/hyperactivity disorder (ADHD) and epilepsy (7). Most previous studies have used single-trimester thyroid hormone levels to examine the relationship between maternal thyroid dysfunction and offspring’s health outcomes. There are very limited data on repeated measurements of thyroid function during pregnancy. However, maternal thyroid hormones are dynamic during pregnancy. Thyroid hormone trajectories may be better predictor of children’s behavioral problems than assessment of thyroid function during a single trimester (8).

Children’s behavioral problems would affect their long-term quality of life and have always been the focus of attention in the field of children’s neurodevelopment (9). Behavioral problems can be classified into internalizing problems and externalizing problems. Internalizing problems usually manifest as anxiety, withdrawal, and negative emotions, while externalizing problems as obvious destructive behaviors, such as aggression, defiance, and hyperactivity (10). In previous studies, there are few studies focusing on thyroid function and offspring behavior development during pregnancy, and the findings were controversial (11,12). There were sex differences in offspring’s neurodevelopmental programmed by early life adverse environmental exposures (13). Results toward sex differences between maternal thyroid dysfunction and offspring’s behavioral problems were also controversial (8,14,15).

In the current study, based on a population-based cohort, we had repeatedly measured TSH, FT₄, and thyroid peroxidase antibody (TPOAb) in the first, second, and third trimesters of pregnancy and aimed to understand the sex-specific impact of maternal TSH and FT₄ trajectories on preschoolers’ behavioral development.

Materials and Methods

Participants

Based on Ma’ anshan Birth Cohort in China, pregnant women who had their first antenatal checkup in Ma’ anshan Maternal and Child Health Hospital were recruited from May 2013 to September 2014. The inclusion criteria for women were (1) within 14 weeks of pregnancy; (2) planned to have antenatal checkups and childbirth in the hospital; (3) had no mental illness and could understand and complete the questionnaires; and (4) willing to participate in follow-ups during childhood. A total of 3474 pregnant women were recruited. Further exclusion criteria were set in the current study as (1) having family history of or suffered from thyroid diseases; (2) twin pregnancy; (3) with adverse pregnancy outcomes, including embryo arrest, spontaneous abortion, therapeutic abortion, stillbirth, and ectopic pregnancy; (4) missing data on thyroid hormones in the first, second, or third trimester; and (5) missing data on preschoolers’ behavioral development assessment. Finally, 1860 mother-child pairs were included in the analysis (Fig. 1).

Figure 1.

Flow chart of participants.

Maternal Thyroid Hormones Assessment

Fasting venous blood of pregnant women was collected through routine antenatal checkups in the first trimester (average 10 gestational weeks), second trimester (average 25 gestational weeks), and third trimester (average 34 gestational weeks) of pregnancy. After centrifugation, the serum was stored in refrigerators at −80°C. The levels of thyroid hormones TSH (catalog no. 11731459122, RRID: AB_2756377) and FT₄ (catalog no. 07976887190, RRID: AB_2861411) were retrospectively detected during childhood follow-up by electrochemiluminescence immunoassay (Roche Diagnostics GmbH, Mannheim, Germany). Limit of detection of TSH and FT₄ were 0.005 μIU/mL and 0.300 pmol/L, respectively. The inter- and intragroup coefficients of variation in each indicator were all less than 10%.

Evaluation of Children’s Behavioral Development

Children born by women in the Ma’ anshan Birth Cohort were followed up at the average age of 4. Preschoolers’ behavioral problems were assessed by Achenbach Child Behavior Checklist (CBCL/1.5~5) (16) completed by their caregivers. CBCL/1.5~5 is a behavioral scale designed by Achenbach and Rescorla for preschoolers, covering 100 questions (16). The behavioral problem was assessed according to a 3-level scoring: “not true” was scored as 0 points, “sometimes true” was scored as 1 point, “often true” was scored as 2 points. In this study, we selected 3 summary scales (internalizing problems, externalizing problems, and total problems) and 5 syndrome scales (emotionally reactive, anxious/depressed, withdrawn, attention problems, and aggressive behavior) to evaluate preschoolers’ emotional and behavioral development. In the syndrome scales, emotionally reactive, anxious/depressed, and withdrawn belonged to internalizing problems and attention problems and aggressive behavior belonged to externalizing problems. A T-score not less than 60 on summary scales or a score not less than 65 on syndrome scales was regarded as borderline clinical range, and T-score below 60 on summary scales or score below 65 on syndrome scales was considered as the normal range (17). The reliability and stability of the scale is high, and it is proved to be good instrument for evaluating the behavioral problems of Chinese children (18). In this study, the Cronbach α coefficient of CBCL was 0.948.

Covariates

Literature review and directed acyclic graphs (19) were used to identify confounding factors. The directed acyclic graphs of this study is reported in Supplemental Figure 1 (20).

Data on maternal age, maternal education level, household monthly income per capita, maternal prepregnancy body mass index (BMI), previous adverse pregnancy outcomes, and maternal smoking and alcohol use were self-reported by women in questionnaire survey during recruitment. Previous adverse pregnancy outcomes included abortion, preterm birth, stillbirth, and birth defect. Women with 1 or more of the previously mentioned conditions were defined as having previous adverse pregnancy outcomes. Information on pregnancy complications, parity, gestational weeks, birth weight, and offspring’s sex was abstracted from medical notes. Women with gestational hypertension and/or gestational diabetes were defined as having pregnancy complications. Data on maternal TPOAb (catalog no. 11820818, RRID: AB_2631044) level in the first, second, and third trimesters of pregnancy were abstracted from the retrospectively detected thyroid function data set.

Data on exclusive breast feeding, main caregivers, and time spent in screening and outdoor activities were reported by children’s caregivers in the follow-up questionnaire survey. These variables were included in sensitivity analysis.

Data Analysis

Data on women’s thyroid hormones available in the first, second, and third trimesters of pregnancy were used to estimate maternal thyroid hormone trajectories and fitted by growth mixture modeling, which is a form of longitudinal hybrid modeling. The modeling not only describes the trend of maternal thyroid hormones over time but also can be used to examine interindividual (between-person) differences in intraindividual (within-person) changes (21). More information on the application of this model is available in the Supplemental Methods (22).

We selected the reference group in which most pregnant women were covered and the level of maternal thyroid hormones were closest to the normal reference ranges established in our laboratory [Supplemental Table 1 (22)]. That is, the low TSH trajectory was set as the reference group of TSH, and the moderate FT₄ trajectory was the reference group of FT₄.

Poisson regression models were applied to examine the relationship between maternal thyroid hormone trajectories and preschoolers’ behavioral development adjusting for the potential confounders as maternal age, maternal education level, household monthly income per capita, maternal BMI, parity, previous adverse pregnancy outcomes, pregnancy complications, maternal smoking, maternal drinking, and maternal TPOAb. To further investigate the sex-specific effect of maternal thyroid hormone trajectories, the regression models were performed stratified by children’s sex.

Two sensitivity analyses have been performed. First, abnormal maternal thyroid function is a risk factor for preterm birth and low birth weight, which in turn are associated with children’s behavioral development (5,6,23,24). Therefore, gestational weeks and birth weight might be potential mediators in the association of maternal thyroid hormones with children’s behavioral problems. Because direct adjustment for birth weight or gestational weeks may induce severe collider-stratification bias (25,26), birth weight z-scores were calculated (27) and were further adjusted in Model 1. In the second sensitivity analysis, as exclusive breast feeding (28), main caregivers (29) before 3 years, and timing spent in screening and outdoor activities (30,31) were strongly associated with the children’s behavioral development but not necessarily with the exposure of interest thereby not being strict confounders. These variables were further adjusted in Model 2 to increase precision.

EpiData 3.0 software was used for data entry, and SPSS 23.0 software was used for statistical analysis. Mplus7.0 software was used to fit maternal thyroid hormone trajectories during pregnancy.

Ethics

This study was approved by the Biomedical Ethics Committee of Anhui Medical University (No. 20131401). All the participants had signed the informed consent forms.

Results

Basic Characteristics of Included and Excluded Participants

Basic characteristics of included and excluded participants were compared and listed in Table 1. Mothers included in the study had higher educational level, longer gestational period, and larger proportion of nulliparous compared to those who were excluded. Higher birth weight and higher percentage of boys was observed among the included children than those excluded participants.

Table 1.

Basic characteristics of included and excluded participants

Characteristics	Included subjects (n = 1860)	Excluded subjects (n = 1614)	Missing data, n (%)	P-value
Maternal age, years	26.6 ± 3.5	26.7 ± 3.9	0 (0.0)	0.467
Maternal education level, years	13.6 ± 3.1	13.1 ± 3.2	21 (0.6)	<0.001
Maternal prepregnancy BMI, kg/m2			0 (0.0)	0.05
<18.5	399 (21.5)	294 (18.2)
18.5~23.9	1242 (66.8)	1112 (68.9)
≥24.0	219 (11.8)	208 (12.9)
Household monthly income per capital, yuan			0 (0.0)	0.570
≤2500	505 (27.2)	413 (25.6)
2500~4000	794 (42.7)	699 (43.3)
>4000	561 (30.2)	502 (31.1)
Parity			0 (0.0)	0.001
Nulliparous	1697 (91.2)	1419 (87.9)
Multiparous	163 (8.8)	195 (12.1)
Maternal smoking			0 (0.0)	0.221
No	1787 (96.1)	1537 (95.2)
Yes	73 (3.9)	77 (4.8)
Maternal alcohol-drinking			0 (0.0)	0.328
No	1720 (92.5)	1478 (91.6)
Yes	140 (7.5)	136 (8.4)
Previous adverse pregnancy outcomes			0 (0.0)	0.917
No	1095 (58.9)	953 (59.0)
Yes	765 (41.1)	661 (41.0)
Pregnancy complications			153 (4.4)	0.159
No	1551 (83.4)	1191 (81.5)
Yes	309 (16.6)	270 (18.5)
Gestational weeks	39.1 ± 1.2	38.9 ± 1.6	200 (5.8)	<0.001
Birth weight, g	3378.0 ± 420.4	3344.9 ± 486.0	206 (5.9)	0.042
Children’s sex			172 (5.0)	<0.001
Boys	967 (52.0)	718 (44.5)
Girls	893 (48.0)	724 (44.9)

Data are given as Mean ± SD or n (%).

Abbreviation: BMI, body mass index.

Grouping for Thyroid Hormone Trajectories

Based on the results of optimal model fitting and clinical significance, we obtained 3 trajectories of maternal TSH levels during pregnancy. Three groups of trajectories of TSH were defined as high level (82 women, 4.4%), moderate level (517 women, 27.8%), and low level (1261 women, 67.8%). FT₄ trajectories during pregnancy were divided into 3 groups as high level (271 women, 14.6%), moderate level (975 women, 52.4%), and low level (614 women, 33.0%). According to the grouping of thyroid hormone trajectories fitted in the first, second, and third trimesters of pregnancy, maternal TSH levels were found to be lowest in the first trimester, then elevated in the second trimester, and followed by a small decline in the third trimester, while FT₄ levels showed a declining trend during the first 2 trimesters and then rose in the third trimester of pregnancy (Fig. 2).

Figure 2.

Maternal thyroid hormone trajectories grouping during pregnancy.

We also fitted maternal TPOAb trajectories during pregnancy, which were divided into 2 groups as group 1 (65 women, 3.5%), high and decreasing TPOAb level, and group 2 (1795 women, 96.5%), low and stable TPOAb level. Maternal TPOAb in the grouping was adjusted regression model as potential confounder.

Basic Characteristics of Participants by Thyroid Hormone Grouping

Basic characteristics in participated mothers and children by thyroid hormone grouping are shown in Table 2. Among 3 TSH trajectory groups, no significant differences in maternal basic characteristics were observed. In FT₄ trajectory groups, mothers with older age and higher education levels were more likely to be with the low trajectory. Offspring of women with the low trajectory of FT₄ had a higher birth weight compared with the moderate and high trajectory groups.

Table 2.

Basic characteristics of participants by thyroid hormone grouping

Characteristics	TSH level			FT4 level
	High	Moderate	Low	High	Moderate	Low
Maternal characteristics
Maternal age, years	26.7 ± 4.4	26.5 ± 3.4	26.7 ± 3.5	26.1 ± 3.5	26.4 ± 3.4	27.3 ± 3.7
Maternal education level,a years	13.0 ± 3.5	13.6 ± 3.1	13.6 ± 3.1	13.2 ± 3.0	13.5 ± 3.1	13.8 ± 3.2
Household monthly income per capita, yuan
≤2500	26/5.1	150/29.7	329/65.1	76/15.0	276/54.7	153/30.3
2500~4000	35/4.4	212/26.7	547/68.9	123/15.5	395/49.7	276/34.8
>4000	21/3.7	155/27.6	385/68.6	72/12.8	304/54.2	185/33.0
Maternal prepregnancy BMI, kg/m2
<18.5	19/4.8	124/31.1	256/64.2	75/18.8	228/57.1	96/24.1
18.5~23.9	53/4.3	344/27.7	845/68.0	170/13.7	639/51.4	433/34.9
≥24.0	10/4.6	49/22.4	160/73.1	26/11.9	108/49.3	85/38.8
Parity
Nulliparous	75/4.4	471/27.8	1151/67.8	244/14.4	892/52.6	561/33.1
Multiparous	7/4.3	46/28.2	110/67.5	27/16.6	83/50.9	53/32.5
Previous adverse pregnancy outcomes
No	51/4.7	306/27.9	738/67.4	166/15.2	588/53.7	341/31.1
Yes	31/4.1	211/27.6	523/68.4	105/13.7	387/50.6	273/35.7
Pregnancy complications
No	70/4.5	441/28.4	1040/67.1	235/15.2	818/52.7	498/32.1
Yes	12/3.9	76/24.6	221/71.5	36/11.7	157/50.8	116/37.5
Maternal smoking
No	76/4.3	495/27.7	1216/68.0	261/14.6	944/52.8	582/32.6
Yes	6/8.2	22/30.1	45/61.6	10/13.7	31/42.5	32/43.8
Maternal alcohol-drinking
No	78/4.5	476/27.7	1166/67.8	252/14.7	903/52.5	565/32.8
Yes	4/2.9	41/29.3	95/67.9	19/13.6	72/51.4	49/35.0
Maternal TPOAb trajectories
Group 1	12/18.5	20/30.8	33/50.8	9/13.8	38/58.5	18/27.7
Group 2	70/3.9	497/27.7	1228/68.4	262/14.6	937/52.2	596/33.2
Children’s characteristics
Gestational weeks	39.0 ± 1.2	39.2 ± 1.2	39.1 ± 1.3	39.1 ± 1.2	39.1 ± 1.2	39.1 ± 1.3
Birth weight,a g	3392.3 ± 432.6	3388.4 ± 401.0	3372.8 ± 427.6	3293.6 ± 388.3	3366.8 ± 426.2	3432.8 ± 418.0
Sex
Boy	41/4.2	286/29.6	640/66.2	138/14.3	520/53.8	309/32.0
Girl	41/4.6	231/25.9	621/69.5	133/14.9	455/51.0	305/34.2
Exclusive breast feedinga
No	68/4.2	454/27.9	1107/68.0	233/14.3	851/52.2	545/33.5
Yes	10/5.1	57/28.9	130/66.0	29/14.7	111/56.3	57/28.9
Main caregivers before 3 yearsa
Parents	52/5.7	263/28.8	598/65.5	138/15.1	468/51.3	307/33.6
Grandparents	28/3.1	248/27.1	639/69.8	124/13.6	494/54.0	297/32.5
Time spent in screeninga
≤1 h/d	32/4.9	188/28.9	430/66.2	94/14.5	344/52.9	212/32.6
>1 h/d	50/4.1	329/27.2	830/68.7	177/14.6	630/52.1	402/33.3
Time spent in outdoor activities a
≤1 h/d	15/3.4	131/30.1	289/66.4	66/15.2	223/51.3	146/33.6
>1 h/d	67/4.7	386/27.1	971/68.2	205/14.4	751/52.7	468/32.9

Abbreviations: BMI, body mass index; TPOAb, thyroid peroxidase antibody.

^a Missing data: 14 in maternal education level, 1 in birth weight, 34 in exclusive breast feeding, 32 in main caregivers before 3 years, 1 in time spent in screening, and 1 in time spent in outdoor activities.

Association Between Maternal Thyroid Hormone Trajectories and Preschoolers’ Emotional and Behavioral Development

The prevalence of behavioral problems in male and female children in different maternal TSH and FT₄ trajectory groups was shown in Figure 3. In boys, the prevalence of externalizing problems was found to be the highest in high TSH trajectory group (14.6%) and lowest in low TSH trajectory group (5.8%). The anxious/depressed problems were found to be highest in high FT₄ trajectory group (8.7%) while lowest in the moderate FT₄ trajectory group (3.7%). The same trend was observed in total problems.

Figure 3.

Association between maternal thyroid hormone trajectories and preschoolers’ behavioral development. (A) Adjusted for maternal age, maternal education level, household monthly income per capita, maternal body mass index, parity, previous adverse pregnancy outcomes, pregnancy complications, maternal smoking, maternal drinking, and maternal thyroid peroxidase antibody. Low thyroid-stimulating hormone trajectory group and the moderate free thyroxine trajectory group was set as the reference group respectively.

After adjusting for potential confounders, in boys, it showed that maternal high TSH trajectory was related to withdrawn [odds ratio (OR) = 2.01, 95% CI: 1.16, 3.50] and externalizing problems (OR = 2.69, 95% CI: 1.22, 5.92), and moderate TSH trajectory during pregnancy was associated with aggressive behavior (OR = 3.76, 95% CI: 1.16, 12.23). Maternal high FT₄ trajectory was associated with anxious/depressed (OR = 2.22, 95% CI: 1.08, 4.56) and total problems (OR = 1.74, 95% CI: 1.13, 2.66), and low FT₄ trajectory during pregnancy was associated with aggressive behavior (OR = 4.17, 95% CI: 1.22, 14.24).

No significant association between maternal thyroid hormone trajectories and preschoolers’ emotional and behavioral problems was observed in girls (Fig. 3).

The sensitivity analysis did not change the main results essentially. [Supplemental Tables 2 and Table 3 (32)].

Discussion

In this prospective longitudinal study, we found that maternal TSH and FT₄ trajectories were associated with preschool boys’ behavioral development. In detail, maternal high-level TSH trajectory was associated with their withdrawn and externalizing problems, and moderate-level TSH trajectory was found to be associated with aggressive behaviors. As to FT₄, high-level trajectory may increase boys’ anxious/depressed emotion and total problems, and low-level trajectory was related to their aggressive behavior.

Findings on the relationship between maternal thyroid function and offspring’s internalizing and externalizing behavioral problems are controversial. Studies in the Generation R study (Rotterdam) had shown that higher maternal TSH concentration during pregnancy was associated with increased externalizing problems in children aged 1.5 and 3 years, while no association was observed between maternal FT₄ concentration and children’s behavioral problems (11). Northern Finland birth cohort showed that girls were more likely to have ADHD born of mothers with high TSH levels in the first trimester of pregnancy (15). Endendijk et al examined the relationship between maternal thyroid parameters and behavioral problems in children between 23 and 60 months of age in 442 mother-child dyads (8) They found that children of the mothers with lowest but increasing TSH and highest but decreasing FT₄ had fewer internalizing problems than those of mothers with highest, nonincreasing TSH and lowest FT₄. However, findings in the ALSPAC cohort study revealed that maternal TSH, FT₄, and TPOAb levels during pregnancy were not related with emotional and behavioral problems in children aged 3.5 to 11 years (12). The Rhea Mother-Child Cohort study showed no sex-specific effect of maternal subclinical hypothyroidism on children’s emotional and behavioral development (14). The inconsistency may be due to different assessments for maternal thyroid function, different instruments for children’s emotional and behavioral assessment, different marginal or clinical criteria to define children’s behavioral problems, and the use of different statistical analysis methods.

The biological mechanism underlying the association between maternal thyroid hormone trajectories and preschooler’s behavioral development is unclear. In this study, compared with maternal low level TSH trajectory, both high-level and moderate-level TSH trajectories seemed to increase children’s behavioral problems. The release of TSH is affected by the circulation of triiodothyronine andT₄ in the negative feedback system (33). Elevated TSH may represent maternal low-level thyroid hormones and is regarded as an important sign of abnormal thyroid function (34). The elevated TSH concentration during pregnancy has a long-term effect on offspring’s thickening of the frontal and parietal cortex, which may lead to executive function problems and affect offspring’s behavioral development (35-37).

Compared with maternal moderate level of FT₄ trajectory, both high- and low-level FT₄ would increase boys’ behavioral problems. Thyroid hormones, which are necessary for fetal brain development, depend on the placental transport of maternal thyroid hormones, especially in the first trimester of pregnancy. Thyroid hormones continue to be transferred from the mothers to the fetus in the subsequent stages of pregnancy (38). Deficiency or excess of thyroid hormones can both cause abnormalities in the structure and function of the developing brain, thereby increasing the sensitivity of the offspring to neurodevelopmental disorders (39). The role of thyroid hormones in the fetal central nervous system is achieved through the interaction of triiodothyronine and nuclear receptors. Thyroid hormones regulate the genetic expression of multiple genes involved in neuronal differentiation, migration, signal transmission, myelination, and cortex development (40,41). A population-based prospective cohort study showed that both low and high maternal FT₄ concentrations were related to low children’s cognitive development (42), as it is closely related to their emotional and behavioral problems (43).

It is worth noting that our findings regarding the effect of maternal thyroid hormone trajectories on preschoolers’ behavioral development are only observed in boys. The potential mechanism is unclear. This sex-specific effect is presumed to be due to sex differences in the nervous system development, which can be manifested in neuronal structure, neural connectivity, and neurochemistry (44). Male fetus produces testosterone from 8 to 24 weeks of pregnancy, which will affect neural pathways and cause behavioral virilization (45). It is generally believed that behavioral virilization tends to externalize problems, such as oppositional defiant disorders and aggressive behaviors (46), although studies have shown that internalizing and externalizing problems often co-occur in children (47,48). More and more studies have shown that there is interaction and cross-talk between thyroid hormones and sex steroid hormones (49), and this effect is more direct in males (50). Thyroid hormones are reported to be associated with testicular development, growth, and maturation. They will enhance gonadotropin-induced androgen synthesis and release in the testis (51). Experimental studies indicated that thyroid hormone is essential for male’s typical brain development (44,52). We suspect that boys may be more susceptible to alterations of maternal thyroid hormones and thus further to the subsequent behavioral problems.

This study has several strengths. First, instead of a single measurement of thyroid function, we used large samples of thyroid hormones from the first, second, and third trimesters of pregnancy to fit maternal thyroid hormone trajectories, which can better reflect the fluctuation of thyroid hormone levels during pregnancy and have higher predictive values on preschoolers’ behavioral development. Meanwhile, benefiting from the large sample size in maternal thyroid hormones in our study, we finely stratified children’s sex in regression models to examine the sex-specific effect of thyroid hormone on offspring’s behavioral development. Second, thyroid function was retrospectively assayed in childhood follow-ups by using stored blood samples; thus, women did not know their detailed thyroid function status right away. Except for pregnant women with clinical thyroid diseases diagnosed by hospitals who had potential treatments [the proportion was 0.97% (18/1860)], other women did not receive any interventions associated with other kind of thyroid functions. It will therefore decrease the exposure misclassification and thus allow us to observe the true and natural effect of thyroid hormone trajectories on offspring’s behavioral development. Third, in addition to TSH and FT₄, we also performed trajectory fitting for maternal TPOAb during pregnancy and controlled it in the regression model. Previous studies have shown that TPOAb positivity during pregnancy may increase the risk of neuropsychiatric developmental abnormalities in offspring, including suboptimal intellectual, motor development (53,54), and the risk of autism (55) and ADHD (56). It will reduce the confounding effect of maternal TPOAb exposure on offspring’s behavioral development. Furthermore, this study was based on a prospective birth cohort; besides exposure and outcome, the potential confounders were also prospectively and accurately collected during the study. Precision variables that would affect children’s behavioral development were fully considered in the models and thus further increase the precision and reliability of the findings.

Some limitations must be acknowledged as well. First, due to limited children with aggressive behaviors, the CI for the association of moderate TSH trajectory with aggressive behavior is too wide. Nevertheless, according to the normal reference ranges of maternal thyroid hormones in our laboratory [as shown in Supplemental Table 1 (22)], we can find that the in the moderate TSH trajectory group, the average TSH levels exceeded the normal reference groups in all 3 trimesters (2.66 μIU/mL vs 1.50 μIU/mL in first trimester, 3.64 μIU/mL vs 2.42 μIU/mL in second trimester, and 3.33 μIU/mL vs 2.20 μIU/mL in third trimester, respectively). It may indicate a potential adverse effect of moderate TSH trajectory on children’s aggressive behavior development, as needs to be confirmed by studies with larger samples. Second, preschoolers’ behavioral development was assessed by children’s caregivers. Different caregivers may have different subjective evaluations of children’s behavioral problems (57). In addition, the emotional status of the caregiver may also affect the evaluation of children’s behavioral problems (58), which may lead to certain misclassifications for the outcome. Multi-informant reporting may be a more reliable method to assess children’s emotional and behavioral problems (59). Third, studies had reported that iodine deficiency during pregnancy may impair children’s brain development and be related to the risk of their behavioral problems (60,61). Although the Ma’ anshan was not reported to be iodine deficient area, we still lacked the local basal iodine level in the current study. The interpretation of the findings, therefore, should be made cautiously. Finally, there were many factors that would affect children’s behavioral development. Residual confounders could not be fully ruled out, such as children’s family upbringing environment, parents’ marital status (62), and maternal environmental endocrine disruptors exposures (63,64).

Conclusions

In conclusion, maternal thyroid hormone trajectories impact preschool boys’ behavioral development.

Abbreviations

Abbreviations
 
• ADHD

  attention deficit hyperactivity disorder;

 
• CBCL

  Child Behavior Checklist

 
• FT₄

  free thyroxine

 
• MABC

  Ma’ anshan Birth Cohort

 
• TPOAb

  thyroid peroxidase antibody

 
• TSH

  thyroid-stimulating hormone.

Acknowledgments

The authors are grateful to the mothers and children in the study for their active cooperation and also would like to thank all the investigators in Ma’ anshan Maternal and Child Health Hospital for their assistance and support in the study.

Financial Support

National Natural Science Foundation of China (81872630), The University Synergy Innovation Program of Anhui Province (GXXT-2020-067), Sci-tech Basic Resources Research Program of China (2017FY101107), the Special Project “Reproductive Health, Prevention and Control of Major Birth Defects” of National Key Research and Development Program (2016YFC1000204-2), the Non-profit Central Research Institute Fund of Chinese Academy of Medical Sciences (2019PT310002), and Research Fund of Anhui Institute of translational medicine (ZHYX2020A001).

Disclosures

The authors report no conflict of interest.

Data Availability

Restrictions apply to the availability of some or all data generated or analyzed during this study to preserve patient confidentiality or because they were used under license. The corresponding author will on request detail the restrictions and any conditions under which access to some data may be provided.