Maternal Thyroid Dysfunction During Pregnancy and Thyroid Function of Her Child in Adolescence

Abstract

Context:

Normal maternal thyroid function is important for fetal development. No knowledge exists on how maternal thyroid function and thyroid antibodies during early pregnancy affect thyroid function of the offspring.

Objective:

The aim of this study was to investigate the relationship between maternal and adolescent thyroid function parameters.

Design, Setting, and Participants:

A total of 3673 mother-child pairs from the prospective, population-based Northern Finland Birth Cohort 1986 participated in the study. Maternal serum samples were drawn in early pregnancy (<20th gestational week), and children's samples were drawn at the age of 16 years and analyzed for TSH, free T₄ (fT4), and thyroid peroxidase antibodies (TPO-Abs).

Main Outcome Measures:

TSH, fT4, and TPO-Ab concentrations were measured at the age of 16 years. Children of mothers with thyroid dysfunction (hypothyroidism, hyperthyroidism, or hypothyroxinemia) or TPO-Ab positivity were compared to those of euthyroid or TPO-Ab-negative mothers. The distributions are expressed as medians with 5th to 95th percentiles.

Results:

Boys of hypothyroid mothers had higher TSH concentrations than those of euthyroid mothers: 2.0 (0.9–4.0) vs 1.7 (0.8–3.3) mU/L; P = .001. Children of hyperthyroid mothers had lower TSH concentrations than those of euthyroid mothers: 1.3 (0.6–4.2) vs 1.7 (0.8–3.3) mU/L, P = .013, for boys; and 1.3 (0.5–3.5) vs 1.6 (0.7–3.4) mU/L, P = .034, for girls. There were no differences in TSH or fT4 concentrations between children of hypothyroxinemic and euthyroid mothers. TPO-Ab-positive mothers more often had TPO-Ab-positive children (prevalence, 9.0 vs 3.7% among boys, and 22.7 vs 7.5% among girls).

Conclusions:

Maternal thyroid dysfunction and TPO-Ab positivity during pregnancy seem to modify thyroid function parameters of offspring even in adolescence. Whether this increases the thyroid disease risk of the children is still unknown.

Maternal thyroid dysfunction affects up to 5% of pregnant women, and thyroid antibodies are prevalent in 5–10% of fertile-aged women (1). Maternal thyroid hormones and antibodies cross the placenta and are important to fetal development during the first trimester (2–4). However, research on the impact of maternal thyroid status on later thyroid function of the child is scarce (5).

It is known that circulating TSH levels in the newborn rise spontaneously after birth and decrease within a few days (6). During the first 3 years of life, children show wide variability of TSH concentrations, and they are higher both in prepuberty and in puberty compared with those in adults (6). Currently there are no data concerning possible associations between maternal thyroid hormone levels during pregnancy and thyroid hormone levels of offspring in later life.

In iodine sufficiency, 3% of schoolchildren are positive for thyroid peroxidase antibodies (TPO-Abs) and thyroglobulin antibodies, with a higher prevalence among girls (7). Children of women with autoimmune thyroiditis are at risk of having thyroid dysfunction and thyroid antibodies later in life (5, 8, 9). Children positive for TPO-Abs also more often have TPO-Ab-positive mothers (5), but there are no studies concerning maternal serum sampling conducted during pregnancy.

The aim of this study was to evaluate the impact of maternal thyroid function parameters and thyroid antibodies during pregnancy on thyroid function parameters and thyroid autoantibody levels of the offspring in adolescence.

Subjects and Methods

Study population and data collection

The prospective, population-based Northern Finland Birth Cohort 1986 (NFBC 1986) covers 99% of all births in northern Finland with an expected delivery date between July 1, 1985, and June 30, 1986, from the 2 northernmost provinces of Finland (9362 mothers and 9479 children). Only singleton pregnancies (n = 9247) were included in this study. The cohort mothers were recruited by 24 weeks gestation, but they were followed since their first visit to free-of-charge maternity welfare clinics (from 10th to 12th gestational week onward). Demographic, biological, health behavior, and socioeconomic as well as maternal health data and data related to birth and neonatal outcome have been collected via questionnaires, which were filled in by the mothers, nurses, or midwives (10, 11).

Data on children have been collected prospectively antenatally, at birth, and at the ages of 7 and 16 years. The latest follow-up, in 2001–2002 when the children were 16 years old, involved questionnaires for parents and children (participation rate, 80%) and clinical examination of the children (participation rate, 74%) among those participants who were alive and traceable (n = 6798).

The Ethics Committees of the Northern Ostrobothnia Hospital District and the National Institute of Health and Welfare approved this study. Informed written consent was obtained from all subjects.

Serum samples and laboratory assays

In 2001–2002, the children had serum sampling included in their clinical examination. The samples were drawn in the morning after overnight fasting, went through primary analyses, and have been stored thereafter at −80°C. The samples have been further thawed 1 to 6 times for various analyses; 14% of the samples were thawed for the second time and 84% for the third time for analyses connected with this study in 2010. To study the effect of repeated freezing and thawing, 7 samples with a 5-year storage time at −80°C from healthy nonpregnant volunteers were analyzed after up to 9 freeze-thaw cycles. Concentrations of TSH, free T₄ (fT4), and TPO-Ab were measured after every other cycle (after thaw numbers 1, 3, 5, 7, and 9), and their concentrations did not change even after repeated freezing and thawing (data not shown).

Altogether, 5765 adolescent samples were analyzed for at least 1 thyroid function test in year 2011. When the sample size was not sufficient for all analyses, TPO-Ab analyses were carried out primarily.

Maternal serum samples were obtained from the Finnish Maternity Cohort (FMC), a serum bank with early pregnancy serum samples from all Finnish pregnant women. Data concerning maternal serum sample collection has been published earlier (12, 13). A total of 5805 maternal samples with sufficient serum samples and sampling conducted before the 20th gestational week (61.2% of the whole cohort with mean gestational age at sampling of 10.7 weeks [SD, 2.8]) were analyzed for at least 1 thyroid function test in year 2006. The excluded population (gestational week at sampling >20; n = 187, insufficient serum sample; n = 3014) did not differ significantly from those included regarding maternal demographic characteristics or pregnancy or neonatal outcomes (data not shown). Ninety-eight mothers were under some thyroid medication during the index pregnancy or had used thyroid medication previously. Ninety-four mothers used levothyroxine, and four mothers used thyrostatic drugs.

Quantitative analyses of TSH, fT4, and TPO-Ab from both maternal and adolescent serum samples were performed by way of chemiluminescent microparticle immunoassays using an Architect i2000 automatic analyzer (Abbott Diagnostics, Abbott Park, Illinois). Lower limits of detection and intra- and interassay coefficients of variation were: 0.0025 mU/L, 1.7 and 5.3% for TSH; 5.1 pmol/L, 3.6 and 7.8% for fT4; and 1.0 IU/mL, 2.5 and 9.8% for TPO-Ab, respectively.

Maternal-adolescent sample pairs and definition of thyroid dysfunction

A total of 3673 adolescent samples had matching maternal sample pairs (1746 boys, 1927 girls) with at least TPO-Ab concentrations analyzed (all tests were performed to 3586 pairs). There were no statistically significant differences in maternal TSH, fT4, or TPO-Ab concentrations when comparing adolescent samples with missing or matching maternal sample pair (data not shown).

The manufacturer's reference intervals for TPO-Ab used in connection with children are <5.61 IU/mL. For mothers, earlier published reference intervals for TSH and fT4 are 0.07–3.1 mU/L and 11.4–22.4 pmol/L in the first trimester and 0.10–3.5 mU/L and 11.09–18.9 pmol/L in the second trimester, respectively. The limit for maternal TPO-Ab positivity was TPO-Ab concentration over the 95th percentile (>167.7 IU/ml) (14).

Mothers were divided into 6 groups according to their fT4, TSH, and TPO-Ab concentrations (Figure 1): 1) euthyroidism, maternal TSH and fT4 both between the reference intervals (n = 3189); 2) hypothyroidism, TSH above its upper limit with low or normal fT4 concentrations (n = 207); 3) hyperthyroidism, TSH below its lower reference limit with high or normal fT4 concentrations (n = 81); 4) hypothyroxinemia, TSH between its reference intervals with low fT4 concentration (n = 45); 5) TPO-Ab-negative, with TPO-Ab concentration <167.7 IU/mL (n = 3481); and 6) TPO-Ab-positive, with TPO-Ab concentration ≥167.7 IU/mL (n = 164).

Figure 1.

Flow chart of the study population in the Northern Finland Birth Cohort 1986.

Reference intervals

There were 5765 (60.8% of the NFBC 1986) children's samples available for calculations of reference intervals. Data from boys and girls were analyzed together and separately.

Among the TPO-Ab-negative children, outliers were identified by using the detection method described by Horn et al (15) after log-transformation of the data to achieve normality. All outliers were evaluated, and all cases with high outlying TSH concentrations (>5.28 mU/L) were excluded from all analyses because they were assumed to have subclinical hypothyroidism. No identification of outliers was performed among the TPO-Ab-positive children.

The calculated new reference intervals were used to identify the prevalence of thyroid dysfunction among children and to evaluate whether this prevalence was different among those whose mothers had had thyroid dysfunction during pregnancy. Hypothyroidism was identified as having a TSH concentration over the upper reference limit with normal or low fT4 concentrations; hyperthyroidism as having a TSH concentration under the lower reference limit with normal or high fT4 concentrations; and hypothyroxinemia as having fT4 concentration under the lower reference limit with normal TSH concentrations.

Statistical analysis

Student's t test and Fisher's exact test were used when evaluating differences in background factors among those included and excluded from the study due to missing values.

Comparison of children's TSH, fT4, and TPO-Ab concentrations as continuous variables among all children and between genders was conducted by using the Mann-Whitney U test. Because statistically significant differences were seen in TSH and TPO-Ab concentrations, all further analyses were performed for boys and girls separately.

Mann-Whitney U test was used to compare serum TSH, fT4, and TPO-Ab concentrations in children born to mothers with or without thyroid dysfunction or TPO-Ab positivity or negativity. Similarly, this test was used to compare TSH and fT4 concentrations among TPO-Ab-positive and -negative children. Pearson's χ² test and Fisher's exact test were used to evaluate the prevalence of TPO-Ab positivity in children between genders and to evaluate the differences of prevalence of TPO-Ab positivity and thyroid dysfunction among children born to mothers with and without thyroid dysfunction and TPO-Ab positivity. All distributions are presented as medians with the 5th-95th percentile range.

Odds ratios (ORs) with 95% confidence intervals (CIs) for children having the same thyroid dysfunction as their mothers were calculated with logistic regression analysis. To avoid some misclassification error due to categorization of the data, we also performed linear regression analysis with logarithmically transformed continuous TSH levels to estimate the effects of increases in maternal TSH levels on children's thyroid function. The results were adjusted for maternal continuous TPO-Ab levels. Further risk estimates were performed after excluding TPO-Ab-positive mothers.

Sensitivity analyses were performed after excluding mothers under thyroid medication before or during pregnancy (n = 98) and by categorizing mothers using only data on TSH concentration.

Reference intervals were calculated as 2.5th and 97.5th percentiles separately for TPO-Ab-negative and -positive children. Bias-corrected and accelerated 95% CIs of the percentiles were calculated with 1000 bootstrap resamples. Statistical analyses were performed using SPSS version 18.0 software (SPSS Inc, Chicago, Illinois).

Results

At the age of 16 years, children of hypothyroid mothers had higher median TSH concentrations (1.8 [5th–95th percentiles, 0.8–3.9] mU/L; P = .001) and children of hyperthyroid mothers had lower median TSH concentrations (1.3 [0.6–3.5] mU/L; P = .001) than those of euthyroid mothers (1.6 [0.8–3.4] mU/L). Children of hypothyroid mothers also had lower median fT4 concentrations (13.2 [11.3–15.6] pmol/L; P = .05) and higher TPO-Ab concentrations (0.24 [0.00–94.4] IU/mL; P = .004) compared with those of euthyroid mothers (13.4 [11.4–16.0] pmol/L for fT4, and 0.1 [0.00–9.4] IU/mL for TPO-Ab). No statistically significant differences were seen in TSH or fT4 concentrations among children of hypothyroxinemic mothers compared with euthyroid mothers, but children of hypothyroxinemic mothers had higher median TPO-Ab concentrations: 0.4 (0.0–157.1) vs 0.1 (0.0–9.4) IU/mL; P = .004. The results remained the same after excluding mothers with thyroid medication and categorizing mothers using only data on TSH concentration (data not shown).

Boys had significantly higher median serum TSH concentrations than girls: 1.67 (5th–95th percentiles, 0.82–3.37) mU/L compared with 1.57 (0.71–3.54) mU/L; P < .001. However, girls had higher median TPO-Ab concentrations than boys: 0.14 (0.00–44.27) IU/mL compared with 0.08 (0.00–3.94) IU/mL; P = .002.

Boys of hypothyroid mothers had higher median TSH concentrations (2.0 [5th–95th percentiles, 0.9–4.0] vs 1.7 [0.8–3.3] mU/L; P = .001), and girls had higher median TPO-Ab concentrations (0.3 [0.0–255.5] vs 0.1 [0.0–33.9] IU/mL; P = .015) than those born to euthyroid mothers. Children of hyperthyroid mothers had significantly lower median TSH concentrations (1.3 [0.6–4.2] vs 1.7[0.8–3.3] mU/L; P = .013, for boys; and 1.3 [0.5–3.5] vs 1.6 [0.7–3.4] mU/L; P = .034, for girls). Boys of hypothyroxinemic mothers had higher median TPO-Ab concentrations: 0.5 (0.0–160.6) vs 0.1 (0.0–3.9) IU/mL; P = .016 (Table 1).

Table 1.

Comparison of Thyroid Function Parameters in Boys and Girls According to Maternal Thyroid Status in the Northern Finland Birth Cohort 1986

	Maternal Thyroid Status
	Euthyroid	Hypothyroid	Hyperthyroid	Hypothyroxinemic	TPO-Ab-Negative	TPO-Ab-Positive
Boys
n (mother-child pairs)	1589–1591	107–108	32	27	1735	89
TSH, mU/L	1.7 (0.8–3.3)	2.0 (0.9–4.0)c	1.3 (0.6–4.2)b	1.6 (0.6–5.1)	1.7 (0.8–3.4)	1.8 (0.9–3.4)
fT4, pmol/L	13.4 (11.4–15.9)	13.1 (11.4–15.7)	13.0 (10.7–15.1)	13.1 (10.8–16.9)	13.4 (11.4–15.8)	13.2 (11.7–16.2)
TPO-Ab, IU/mL	0.1 (0.0–3.9)	0.2 (0.0–6.5)	0.0 (0.0–4.3)	0.5 (0.0–160.6)a	0.1 (0.0–3.7)	0.2 (0.0–51.1)b
Girls
n (mother-child pairs)	1589–1598	96–99	49	18	1746	75
TSH, mU/L	1.6 (0.7–3.4)	1.6 (0.5–3.9)	1.3 (0.5–3.5)a	1.5 (0.6–3.9)	1.6 (0.7–3.5)	1.6 (0.7–3.8)
fT4, pmol/L	13.5 (11.5–16.0)	13.3 (11.2–15.5)	13.6 (11.6–15.8)	13.1 (10.1–15.1)	13.5 (11.5–16.0)	13.7 (11.5–16.2)
TPO-Ab, IU/mL	0.1 (0.0–33.9)	0.3 (0.0–255.5)a	0.3 (0.0–506.5)	0.4 (0.0–72.4)	0.1 (0.0–32.9)	0.6 (0.0–387.1)c

Numbers vary due to missing samples. Distributions are expressed as median (5th–95th percentiles). Euthyroid indicates both TSH and fT4 between the trimester-specific reference intervals. Hypothyroid indicates TSH above its upper reference limit and low or normal fT4. Hyperthyroid indicates TSH below its lower reference limit and high or normal fT4. Hypothyroxinemic indicates TSH between its reference intervals and fT4 below its lower limit. P values were obtained with the Mann-Whitney U test when comparing individual groups with the maternal euthyroid group or an antibody-positive group with an antibody-negative group.

^aP < .05;

^bP < .01;

^cP < .001.

When comparing children of TPO-Ab-positive and -negative mothers, both boys and girls had significantly higher median TPO-Ab concentrations: 0.2 (5th–95th percentiles, 0.0–51.1) vs 0.1 (0.0–3.9) IU/mL; P = .05, for boys; and 0.6 (0.0–387.1) vs 0.1 (0.0–33.9) IU/mL; P < .001, for girls (Table 1).

There were 223 (6.0%) TPO-Ab-positive children, of whom 74 (33.2%) were boys and 149 (66.8%) were girls. TPO-Ab-positive mothers more often had TPO-Ab-positive children (Figure 2), 9.0 vs 3.7% among boys (P = .021), and 22.7 vs 7.5% among girls (P < .001). TPO-Ab-positive boys had higher median serum TSH and lower median fT4 concentrations than TPO-Ab-negative boys (TSH, 1.84 vs 1.67 mU/L, P = .04; and fT4, 13.00 vs 13.41 pmol/L, P = .03), and TPO-Ab-positive girls had higher median TSH concentrations than TPO-Ab-negative girls (TSH, 2.10 vs 1.53 mU/L; P < .001) (Table 2).

Figure 2.

Prevalence of TPO-Ab-positive children (%) grouped by maternal antibody status.

Table 2.

Comparison of TSH and fT4 Concentrations Between TPO-Ab-Negative and TPO-Ab-Positive Children in the Northern Finland Birth Cohort 1986

	TPO-Ab-Negative	TPO-Ab-Positive
Boys
n	1770	74
TSH, mU/L	1.67 (0.95–2.80)	1.84 (0.91–4.70)a
fT4, pmol/L	13.41 (11.80–15.81)	13.00 (11.44–14.72)a
Girls
n	1691	149
TSH, mU/L	1.53 (0.83–2.89)	2.10 (1.01–3.97)b
fT4, pmol/L	13.48 (11.85–15.42)	13.28 (11.50–15.26)

Distributions are expressed as median (5th–95th percentiles). P values were obtained with the Mann-Whitney U test when comparing TPO-Ab-positive and -negative children.

^aP < .05;

^bP < .001.

Children's reference intervals are presented in Table 3 and are 0.64–3.74 mU/L for TSH and 11.01–16.63 pmol/L for fT4. When using these reference values of TSH and fT4 concentrations to evaluate the prevalence of thyroid dysfunction in children, there were 125 (2.2%) cases of overt or subclinical hypothyroidism (of whom 26.2% or 32 of 122 were TPO-Ab-positive), 86 (1.5%) cases of overt or subclinical hyperthyroidism, and 82 (1.4%) cases of hypothyroxinemia.

Table 3.

Reference Intervals (2.5th–97.5th Percentiles With 95% CIs) for TSH and fT4 in TPO-Ab-Positive and -Negative Children in the Northern Finland Birth Cohort 1986

	n	TSH, mU/L		fT4, pmol/L
		2.5th Percentile (95% CI)	97.5th Percentile (95% CI)	2.5th Percentile (95% CI)	97.5th Percentile (95% CI)
TPO-Ab-negative
Boys	2764	0.69 (0.65–0.7)	3.76 (3.61–3.95)	11.02 (10.86–11.14)	16.44 (16.24–16.76)
Girls	2633	0.59 (0.57–0.61)	3.66 (3.56–3.83)	11.08 (10.90–11.22)	16.75 (16.51–16.98)
All TPO-Ab-negative children	5764	0.64 (0.62–0.65)	3.74 (3.62–3.84)	11.01 (10.91–11.12)	16.63 (16.45–16.79)
TPO-Ab-positive
Boys	122	0.35 (0.00–0.88)	11.79 (7.58–64.22)	10.04 (9.26–10.95)	16.23 (15.45–17.29)
Girls	246	0.14 (0.03–0.68)	9.06 (6.08–14.13)	10.57 (9.44–11.21)	17.92 (16.60–21.14)

Reference intervals were calculated as 2.5th and 97.5th percentiles separately for TPO-Ab-negative and -positive children and presented with 95% CIs. Bias-corrected and accelerated 95% CIs of the percentiles were calculated with 1000 bootstrap resamples.

The ORs (95% CIs) for children to have the same thyroid dysfunction as their mothers are presented in Table 4. Hypothyroid mothers more often had hypothyroid children than euthyroid mothers: 8.1 vs 3.2%, P = .001 (3.4 [1.8–6.5]) for all children; 6.6 vs 3.1%, P = .081, for boys (2.9 [1.1–7.7]); and 9.9 vs 3.4%, P = .005, for girls (3.9 [1.7–9.0]). Between children of hypothyroid and euthyroid mothers, the percentage difference (95% CI) of children's TSH concentrations was 13.6% (5.7–22.1); among boys, it was 19.1% (8.7–30.5); and among girls, it was 8.11% (−3.3 to 20.9) with all mothers included, and it did not change after adjusting for TPO-Ab or excluding TPO-Ab-positive mothers from the analysis.

Table 4.

Estimated Risks of Children to Have the Same Thyroid Dysfunction as Their Mothers in the Northern Finland Birth Cohort 1986

Maternal Thyroid Status	OR (95% CIs)				OR (95% CIs)
	No. of Mothers	All Children	Boys	Girls	No. of Mothersa	All Children	Boys	Girls
Hypothyroid	207 (108/99)	2.7 (1.5–4.6)	2.2 (1.0–5.1)	3.1 (1.5–6.6)	118 (60/58)	3.4 (1.8–6.5)	2.9 (1.0–5.1)	3.9 (1.7–9.0)
Hyperthyroid	81 (32/49)	3.7 (1.5–8.8)	4.7 (1.0–20.8)	3.0 (1.0–8.8)	67 (23/44)	4.1 (1.7–9.8)	5.9 (1.3–26.7)	3.1 (1.1–9.0)

Number of mothers includes mothers of all children (boys/girls). Hypothyroid indicates TSH above its upper reference limit and low or normal fT4. Hyperthyroid indicates TSH below its lower reference limit and high or normal fT4.

^aTPO-Ab-positive mothers (TPO-Ab concentration ≥167.7 IU/mL) excluded from analysis.

Hyperthyroid mothers had hyperthyroid children more often than euthyroid mothers: 8.0 vs 2.3%, P = .009, for all children (OR [95% CI], 4.1 [1.7–9.8]); 6.9 vs 1.6%, P = .082 (5.9 [1.3–26.7]) for boys; and 8.7 vs 3.0%, P = .057 (3.1 [1.1–9.1]) for girls. Between children of hyperthyroid and euthyroid mothers, the percentage difference (95% CI) of children's TSH concentrations was −17.6% (−26.1 to −8.0); among boys, −18.0 (−30 to −4.0); and among girls, −16.6% (−28.4 to 2.8), with all mothers included; it did not change after adjusting for TPO-Ab or excluding TPO-Ab-positive mothers from the analysis. The results of OR analyses did not substantially change after the following sensitivity analyses: excluding mothers with thyroid medication, and categorizing mothers by only using data on TSH concentration (data not shown).

Hypothyroxinemic mothers also had hypothyroxinemic children more often, but the differences were not statistically significant: 5.1 vs 2.4%, P = .251 for all children; 4.3 vs 2.8%, P = .488 for boys; and 6.3 vs 2.0%, P = .289 for girls.

Discussion

Previous information concerning the impact of maternal thyroid function on thyroid function parameters of the child is scarce, although thyroid dysfunction during pregnancy is widely studied. Our study describes the association between maternal thyroid status during pregnancy and thyroid function parameters in adolescent offspring. Serum TSH and fT4 concentrations in the children were significantly in accordance with maternal TSH and fT4 concentrations during pregnancy. When comparing the children of hypothyroid and hyperthyroid mothers with the children of euthyroid mothers, significant differences were seen between the groups even after adjusting for maternal TPO-Abs, showing that maternal thyroid status during pregnancy might associate with later thyroid function of the child. To our knowledge, our study is the first to show this association.

Individual levels of fT4 and TSH are largely genetically controlled (16, 17), although variations in fT4 concentrations may be more environmentally driven than previously thought (18). Aggregation of thyroid antibodies in first-degree relatives has been suggested to be controlled by genes (19–21), and female gender seems to be a predictive factor as regards consistently elevated TSH levels (22). In our study, the effect of gender on thyroid function of the child was clear because boys had higher TSH concentrations and girls had a higher prevalence of TPO-Abs. We cannot explain why adolescent boys had higher TSH concentrations than girls, but similar association has been seen in other studies (23). Women seem to be particularly at risk of elevated TSH concentrations later in life, when the prevalence of thyroid disease exceeds that in men (23). The fact that maternal thyroid dysfunction was independently associated with children's TSH concentrations after adjusting for maternal TPO-Ab or excluding TPO-Ab-positive mothers might be a manifestation of the genetic control of TSH.

There are currently limited data on reference intervals of serum TSH and fT4 concentrations in adolescent populations, and therefore adult reference intervals are widely used. Our results suggest a decrease in the upper limit of TSH levels from 4.0 mU/L (as reported by the manufacturer) to 3.74 mU/L, and also an increase in the lower limit from 0.4 to 0.64 mU/L among the adolescents. In a recent Finnish study concerning the adult population (24), it was also proposed that a decrease of 10% in the current upper limit of TSH levels would be suitable. In our study, the fT4 reference intervals in adolescents (11.01–16.63 pmol/L) are considerably narrower than the clinically used reference intervals for adults. This may be a result of the adolescent population being healthier than adults, and in addition, exclusion of TPO-Ab-positive subjects reduces the number of cases of subclinical thyroid disease among the study population.

In our study, the girls of TPO-Ab-positive mothers showed a particularly higher prevalence of TPO-Ab positivity. Among girls, the prevalence of TPO-Ab positivity was 3.0-fold higher and among boys 2.4-fold higher if the mother was TPO-Ab positive. Similar results have been reported earlier, although in a smaller population and with serum sampling of the mothers not carried out during pregnancy (5). In addition, in another study children diagnosed with autoimmune thyroiditis had measurable concentrations of TPO-Ab in cord blood sera more often than controls (9).

The early effects of maternal thyroid antibodies on the child are better known than the effects in later life. Maternal TSH receptor antibodies may cause transient congenital thyroid dysfunction (25). Maternal TPO-Ab positivity has been associated with high TSH concentrations in the child 1 week after birth (26). The mothers of congenitally hypothyroid children have been reported to have a high prevalence of decreased fT4 and increased TSH levels and TPO-Ab positivity (27), and 50% of children with subclinical hypothyroidism have been reported to have a positive family history of thyroid disease (28).

One third of TPO-Ab-positive children have been reported to have subclinical hypothyroidism (7), and 15–37% of children with abnormal serum TSH concentrations are TPO-Ab positive (22). Our results confirm that TPO-Ab-positive children have higher TSH concentrations than TPO-Ab-negative children, suggesting that subclinical thyroid diseases affect even adolescents.

The strengths of our current study lie in the well-documented population-based NFBC 1986 cohort with its nearly 3700 mother-child sample pairs and in the high participation rate of the subjects. These mother-child pairs represent the whole cohort for all important background factors. Our study setting is reliable when comparing the various thyroid function groups and observing the analogous nature of thyroid hormone levels in mother-child pairs. All actions have been strictly supervised, and the methods are convergent. In addition, Finland has been iodine-sufficient since the 1940s (29–31). It is also highly important that we were able to use our own population-based trimester-specific reference intervals for maternal serum samples and most of the maternal samples were drawn during the first trimester.

Unfortunately, we were not able to study genetics and paternal impact. The number of mother-child pairs was lower than the number of maternal and adolescent samples available independently, but it was also verified that the adolescent samples without maternal pair did not show any difference in overall TSH, fT4, or TPO-Ab concentrations or maternal background factors compared to the adolescent samples with matching maternal sample. We acknowledge that serum fT4 measurements may not be totally reliable during pregnancy (32–34), although they are clinically used in addition to TSH measurements. To avoid this problem, our classification of study mothers is mostly based on TSH concentrations, and we also performed sensitivity analysis categorizing the data without using data on fT4 concentrations.

There is a possibility that some of the study mothers may have been misclassified by using trimester-specific reference intervals because individual variation in thyroid function during pregnancy is small (35, 36). However, using trimester-specific reference intervals is a more accurate estimate of thyroid function during pregnancy than using reference intervals obtained in nonpregnant populations. We also estimated the linear effects of increases in maternal TSH concentrations on adolescents' thyroid function parameters, and the results were in the same direction as in our categorized analyses.

We also do not know whether the mothers or children with altered thyroid function parameters suffer any symptoms from possible thyroid disease or whether the adolescents with high values are at a higher risk of overt thyroid disease.

In conclusion, altered maternal thyroid function parameters and TPO-Ab positivity during early pregnancy appear to associate with thyroid function of the child in adolescence. Children of hypothyroid mothers had a higher risk of being hypothyroid, and children of hyperthyroid mothers had a higher risk of being hyperthyroid than those born to euthyroid mothers. The children, especially girls, of TPO-Ab-positive mothers had a higher prevalence of TPO-Ab positivity than those of TPO-Ab-negative mothers, and it seems that antibodies may affect the child as early as in utero. This knowledge might help in recognizing risk groups for thyroid dysfunction and in making earlier diagnosis. However, the long-term effects of these changes in thyroid function in adolescence are still unknown.