Maternal thyroid hormones during pregnancy ... - Wiley Online Library

Clinical Endocrinology (2014) 81, 117–125

doi: 10.1111/cen.12399

ORIGINAL ARTICLE

Maternal thyroid hormones during pregnancy, childhood adiposity and cardiovascular risk factors: the Generation R Study Guilherme A. F. Godoy*,†,‡, Tim I. M. Korevaar*,§, Robin P. Peeters§, Albert Hofman†, Yolanda B. de Rijke§,¶, Jacoba J. Bongers-Schokking‡,**, Henning Tiemeier†,††, Vincent W. V. Jaddoe*,†,‡ and Romy Gaillard*,†,‡ *The Generation R Study Group, Erasmus Medical Center, †Department of Epidemiology, Erasmus Medical Center, ‡Department of Paediatrics, Erasmus Medical Center-Sophia Children’s Hospital, §Department of Internal Medicine, Erasmus Medical Center, ¶Department of Clinical Chemistry, Erasmus Medical Center-Sophia Children’s Hospital, **Department of Endocrinology, Erasmus Medical Center-Sophia Children’s Hospital, and ††Department of Child and Adolescent Psychiatry, Erasmus Medical CenterSophia Children’s Hospital, Rotterdam, The Netherlands

Summary Objective Variation in maternal thyroid function during early pregnancy may permanently affect childhood growth and cardiovascular development. We examined the associations of early pregnancy maternal TSH and free T4 (FT4) levels with childhood growth, body composition and cardiovascular characteristics. Methods We performed a population-based prospective cohort study among 5646 mothers and their children. Maternal thyroid parameters were assessed in early pregnancy (median: 13
2 weeks; 95% range: 9
7–17
6 weeks). Childhood growth was repeatedly measured from birth to 6 years. At the age of 6 years, childhood body mass index (BMI), total body and abdominal fat distribution, blood pressure and left ventricular mass were measured. Results Maternal thyroid parameters were not consistently associated with childhood length and weight growth characteristics. Lower maternal TSH levels were associated with lower childhood BMI, total fat mass, abdominal subcutaneous fat mass area and diastolic blood pressure (P-values 10
4 pmol/l and FT4 < 22 pmol/l; n = 187) and hyperthyroidism (TSH < 0
03 mIU/l and FT4 > 22 pmol/l, n = 55). We combined hypothyroidism and subclinical hypothyroidism in one category to study the effects of abnormal lower thyroid activity.

calculated as recommended by the American Society of Echocardiography.21,22 Covariates At enrolment, we obtained information about maternal age, parity, educational level, ethnicity and folic acid supplement use. Information about maternal height and prepregnancy weight was obtained at enrolment. BMI was calculated. Maternal gestational weight gain was measured until 30 weeks of gestation (median: 30
2 weeks; 95% range: 28
6–32
6 weeks). Information about maternal smoking during pregnancy was assessed by questionnaire. Sex, gestational age and weight at birth were obtained from medical records. Information about breastfeeding was obtained by questionnaires. Statistical analysis

Childhood growth measurements Well-trained staff in Community Health Centers obtained postnatal growth characteristics at the ages of 6 months (median: 6
2 months; 95% range: 5
2–8
3 months), 12 months (median: 11
1 months; 95% range: 10
1–12
6 months), 24 months (median: 24
8 months; 95% range: 23
4–28
1 months), 36 months (median: 36
7 months; 95% range: 35
4–40
7 months) and 48 months (median: 45
8 months; 95% range: 44
4– 48
6 months). Standard deviation scores (SDS) for post-natal growth characteristics were obtained with Dutch growth reference charts (Growth Analyzer 3
0; Dutch Growth Research Foundation, Rotterdam, the Netherlands). Childhood body fat and cardiovascular outcomes At the age of 6 years (median: 72
5 months; 95% range: 67
6– 94
7 months), children’s height and weight were measured, and body mass index (BMI) was calculated (kg/m²) in a dedicated research facility in the Erasmus Medical Center. Body fat was measured by dual-energy X-ray absorptiometry (DXA) (iDXA, General Electrics – Lunar, 2008, Madison, WI, USA).17 Total body fat mass was calculated as percentage of total body weight (kg) measured by DXA. The android/gynoid fat mass ratio was calculated.17 Abdominal ultrasound examinations were performed with ultrasound LOGIQ E9 (GE Medical System, Wauwatosa, WI, USA) and ATL-Philips Model HDI 5000 (Seattle, WA, USA), as described previously.18,19 Subcutaneous and preperitoneal fat mass areas were measured as areas of 2 cm length along the midline starting from the xiphoid process in direction of the navel. Systolic blood pressure and diastolic blood pressure were measured at the right brachial artery, four times with one-minute intervals, using the validated automatic sphygmanometer Datascope Accutor Plus TM (Paramus, NJ, USA).20 Two-dimensional M-mode echocardiographic measurements were performed using the ATL-Philips Model HDI 5000 or the Logiq E9 (GE Medical Systems) devices, and left ventricular mass was © 2014 John Wiley & Sons Ltd Clinical Endocrinology (2014), 81, 117–125

In mothers with TSH and FT4 levels within the reference range, we explored the associations of maternal thyroid hormone levels with repeatedly measured infant and childhood growth characteristics (height and weight) using linear regression models. We used age-adjusted SDS of growth characteristics for these analyses. Next, we performed linear regression analyses to assess the associations of maternal thyroid parameters with childhood total body and abdominal fat distribution, and cardiovascular outcomes. For all analyses, maternal TSH and FT4 levels were divided in quintiles and analysed as categorical variables to explore potential nonlinear associations. We used the third quintile as a reference category as the median was in this category. We also analysed maternal TSH and FT4 as continuous variables, showing the associations per standard deviation change of maternal thyroid parameters. We constructed SDS values [(observed value population mean)/ standard deviation (SD) of the population] for all outcome measurements to enable comparison of the effect estimates. Finally, we used linear regression models to study the associations of abnormal maternal thyroid function with childhood growth characteristics, body composition and cardiovascular outcomes at the age of 6 years. We used the euthyroid women (i.e. women with TSH and FT4 levels within the reference range) as the reference group. Regression assumptions for all analyses were met. All models were first only adjusted for gestational age at blood sampling, child’s sex and age at outcome measurement, and additionally adjusted for maternal age, ethnicity, educational level, parity, prepregnancy BMI, gestational weight gain, smoking during pregnancy, folic acid supplementation, gestational age and weight at birth, breastfeeding, and child’s height (fat mass outcomes) or child’s BMI (cardiovascular outcomes). To explore the role of maternal first-trimester thyroid hormone levels only, we performed a sensitivity analysis among 3561 (89%) women with thyroid parameters measurements obtained before a gestational age of 14 weeks (median: 12
5 weeks; 95% range: 9
0– 13
9 weeks). Additional sensitivity analyses were also performed among the complete range of maternal TSH and FT4 levels. Missing data of the covariates were imputed using multiple imputation. All analyses were performed using the Statistical

120 G. A. F. Godoy et al. Package of Social Sciences version 17
0 for Windows (SPSS Inc, Chicago, IL, USA).

follow-up data available were more likely to be lower educated and of non-European descent.

Results

Maternal thyroid function and childhood growth

Subject characteristics Characteristics of the study population are shown in Table 1. Supplemental Table S1 shows that mothers without childhood

Table 1. Characteristics of study population Characteristics (n = 5646) Maternal characteristics Age, mean (SD), years Height, mean (SD), cm Prepregnancy weight, mean (SD), kg Prepregnancy body mass index, mean (SD), kg/m2 Gestational weight gain, mean (SD), kg Gestational age at intake, median (95% range), weeks Parity, No. (%), primipara Education, No. (%) Primary school Secondary school Higher education Ethnicity, No. (%) European Non-European Smoking habits, No. (%) No Yes Folic acid supplements, No. (%) None 1st 10 weeks Periconception use TSH, median (95% range), mIU/L FT4, median (95% range), pmol/L Birth characteristics Male sex, No. (%) Gestational age, median (95% range), weeks Birth weight, mean (SD), g Childhood characteristics Age at follow-up, median (95% range), years Height, mean (SD), cm Weight, mean (SD), kg Body mass index, mean (SD), kg/m2 Total fat mass, mean (SD), % Android/gynoid fat mass ratio, mean (SD) Abdominal subcutaneous fat mass area, mean (SD), cm² Abdominal preperitoneal fat mass area, mean (SD), cm² Systolic blood pressure, mean (SD), mmHg Diastolic blood pressure, mean (SD), mmHg Left ventricular mass, mean (SD), g

Value

29
7 167
5 66
2 23
5

(5
0) (7
4) (12
6) (4
2)

10
6 (5
0) 13
2 (9
7, 17
6) 3214 (57
4) 517 (9
8) 2372 (45
2) 2364 (45) 3299 (60
8) 2127 (39
2) 3613 (72) 1402 (28) 1087 1374 1831 1
35 14
76

(25
3) (32) (42
7) (0
03, 4
49) (10
28, 22
26)

2856 (50
6) 40
1 (35
7, 42
3) 3412
9 (563
8) 6 119
3 23
2 16
2 24
8 0
3 0
6

(5
6, 7
9) (5
9) (4
1) (1
8) (5
6) (0
1) (0
4)

0
5 (0
3) 102
7 (8
1) 60
7 (6
9) 53
3 (11
5)

Values are means (standard deviation) or numbers (percentages). Median (95% range).

In the model adjusted only for gestational age at blood sampling, child’s sex and age at measurement, maternal TSH levels were not associated with childhood growth characteristics, whereas higher maternal FT4 levels were significantly associated with lower childhood weight growth throughout the first six years of life (Supplemental Figure S1). In the fully adjusted models, no significant trends for maternal TSH with childhood growth outcomes were present, whereas a higher maternal FT4 level was associated with a lower childhood weight at 6 months (P-value