The role of prenatal maternal sex steroid hormones in weight and adiposity at birth and growth trajectories during infancy

Abstract

OBJECTIVE:

Intrauterine factors can impact fetal and child growth and may underlie the developmental origins of childhood obesity. Sex steroid hormone exposure during pregnancy is a plausible target because of the impact on placental vascularization, nutrient transportation, adipogenesis, and epigenetic modifications. In this study we assessed maternal sex steroid hormones in each trimester in relation to birthweight, neonatal adiposity, and infant growth trajectories, and evaluated sensitive windows of development.

METHODS:

Participants from a prospective pregnancy cohort who delivered at term were included in the analysis (n = 252). Estrone, estradiol, and estriol, as well as total and free testosterone throughout gestation were assessed using high-performance liquid chromatography and tandem mass spectrometry. Path analyses were used to assess the direct associations of sex steroid hormones in each trimester with birth outcomes and infant growth trajectories (birth to 12 months) adjusting for covariates and considering moderation by sex.

RESULTS:

The associations between prenatal sex steroid hormones and fetal/infant growth varied by sex and timing of hormone exposure. First-trimester estrone was associated with higher birthweight z-scores (β = 0.37, 95% CI: 0.02, 0.73) and truncal skinfold thickness (TST) at birth (β = 0.94, 95% CI: 0.34, 1.54) in female infants. Third-trimester total testosterone was associated with higher TST at birth (β = 0.47, 95% CI: 0.03, 0.86) in both sexes. First-trimester estrone and estradiol and first- and third-trimester testosterone were associated with lower probabilities of high stable weight trajectory compared to low stable weight trajectory (Estrone: β = −3.87, 95% CI: −6.59, −1.16; Estradiol: β = −4.36, 95% CI: −7.62, −1.11; First-trimester testosterone: β = −3.53, 95% CI: −6.63, −0.43; Third-trimester testosterone: β = −3.67, 95% CI: −6.66, −0.69) during infancy in male infants.

CONCLUSIONS:

We observed associations between prenatal sex steroid hormone exposure and birthweight, neonatal adiposity and infant growth that were sex and gestational timing dependent. Our findings suggest further investigation on additional mechanisms linking prenatal sex steroid exposure and fetal/postnatal growth is needed.

INTRODUCTION

Childhood obesity has reached epidemic proportions, affecting nearly one in five youths in the United States [1]. The risk of childhood obesity is evident in early infancy, as both high and low birthweight and rapid infant weight gain have been consistently linked to childhood obesity [2–4]. Emerging evidence has suggested that prenatal factors, including gestational weight gain, gestational diabetes, and smoking during pregnancy play a role in the development of childhood obesity [5, 6]. The underlying mechanism may involve fetal programming, whereby physiological and metabolic set points are modulated by the intrauterine milieu, affecting fetal growth and obesity susceptibility [7]. Therefore, identification of intrauterine factors linked to birthweight and infant growth trajectories is critical to understand the early developmental origins of childhood obesity.

Sex steroid hormones (SSH) are potential key intrauterine factors that modify fetal growth and obesity susceptibility. SSH regulate a wide range of maternal and fetal functions [8, 9], including placental vascularization, endometrial and placental nutrient transporter expression, and epigenetic modification of fetal tissues [9–12]. Additionally, estrogens and testosterone can induce cell proliferation, particularly in bone and adipocytes, and thus potentially contribute to offspring growth and adiposity in utero as well as postnatally [13–15]. Prior research shows that even within the “normal” range in community participants, maternal SSH levels varied in pregnancy, which were associated with factors, such as age, race/ethnicity, pre-pregnancy BMI, parity and stress [16, 17]. Therefore, exposure to the variability in SSH levels during pregnancy may contribute to the intrauterine origins of childhood obesity.

To date, the majority of studies assessing SSH and birthweight have focused on estriol (E3) due to its widespread availability as part of prenatal screening test, reporting positive associations [18–20]. However, fewer studies have examined the relationship between other SSH and birthweight and results have been controversial [21–27]. Also, the majority of previous studies assessed prenatal SSH at a single time point, most often mid-late pregnancy [18, 20, 21, 27]. Recognizing that variations in fetal growth originates in early pregnancy, repeated assessment of SSH across pregnancy may be suitable to identify critical periods during which fetal growth is most sensitive to endocrine activity. Whether associations persist beyond birth into infancy remains an open question as well [18, 27, 28].

In this study, we leveraged clinical data and biospecimens from a pregnancy cohort to assess maternal SSH concentrations (estrone, estradiol, E3, total testosterone, and free testosterone) in each trimester in relation to birthweight, neonatal adiposity, and infant weight and adiposity growth trajectories through the age of 12 months. Based on relationships indicated by prior literature [18–21, 23, 25, 27], we hypothesized that higher estrogen and lower testosterone levels in mid-late pregnancy would be associated with higher birthweight and neonatal adiposity, and higher risk of growth trajectories characterized by high infant weight or rapid weight gain. Additionally, we evaluated moderation by infant sex in light of prior evidence indicating sex differences in fetal growth, placental function and adipose tissue biology [29, 30].

METHODS

Study overview

This study analyzed data from a prospective pregnancy cohort, the Understanding Pregnancy Signals and Infant Development (UPSIDE) study, which is part of the Environmental Influences on Child Health Outcomes (ECHO) program [31, 32]. The UPSIDE study recruited pregnant people (n = 326) in their first trimester from 2015 to 2019 through University of Rochester Medical Center (Rochester, NY, USA) affiliated obstetric clinics. Eligibility criteria for the current study included: (1) available serum SSH in all trimesters, (2) infants born at term (gestational age ≥37 weeks) to assess associations in low-risk population, (3) available data on birth and/or infant growth outcomes. There were 252 mother-child dyads included in the final analyses for birthweight (Supplementary Fig. 1). The UPSIDE study was approved by the institutional review boards at the University of Rochester and Rutgers University. Written informed consent was obtained from all participants.

Sex steroid hormone assays

Blood was collected during study visits in each trimester (1st trimester: 12.2 ± 1.3 weeks; 2nd trimester: 21.2 ± 1.8 weeks; 3rd trimester: 31.4 ± 2.0 weeks). The detail of SSH assessment was described in our previous study [33]. Briefly, SSH from serum samples were quantified using validated liquid chromatography with tandem mass spectrometry (LC-MS/MS) methods at the Endocrine and Metabolic Research Laboratory at the Lundquist Institute at Harbor-UCLA Medical Center. Total testosterone (TT) was measured using a Shimadzu HPLC system (Columbia, MD) and an Applied Biosystems API5500 LC–MS/MS (Foster City, CA) equipped with a Turbo-Ion-Spray source with positive mode. Equilibrium dialysis using labeled testosterone was used to quantify free testosterone (fT). The Shimadzu HPLC system (Columbia, MD) and a triple quadrupole mass spectrometer (API5000 LC–MS/MS, Foster City, CA) were used to measure estrogens. The limit of quantification (LOQ) was 50 pg/mL for E3. $\text{LOQ}/\sqrt{2}$ was used to replace E3 values less than $\text{LOQ}/\sqrt{2}$ or missing (n = 54) in the 1st trimester. The LOQ for E1 and E2 is 2 pg/mL, and The LOQ for total testosterone is 2 ng/dL. E1, E2, and TT levels at all trimesters were above the LOQ.

Birth and infant growth outcomes

We considered four outcomes: (1) birthweight, (2) newborn truncal skinfold thickness (TST [as a surrogate of adiposity]), (3) growth trajectories for weight-for-age percentile, and (4) TST trajectories from birth to 12 months of age, as both birthweight and rapid infant weight gain have been consistently linked to childhood obesity [2–4] and excessive accumulation of fat or adipose tissue is the defining characteristic of obesity. Birthweights were abstracted from medical records and converted to standardized (Z) scores adjusting for gestational age and infant sex according to Fenton growth standards [34]. At 1, 6, and 12 months, infant weight was measured using a Seca Infant Scale. Age- and sex-specific infant weight-for-age percentile was calculated following the WHO growth standards [35]. Subscapular and suprailiac skinfold thicknesses (mm) were measured twice by trained study coordinators with Holtain calipers following the National Health and Nutrition Examination Survey protocols at each visit. A third measure was obtained if the difference between the first two measures was more than 1.0 mm or either of the two measures was out of reference ranges. For each anatomic site, skinfold thickness was calculated as the average value of the two closest measures. TST, a proxy for central fat mass, was the average of subscapular and suprailiac skinfold thickness [36, 37].

Covariates

Maternal and infant factors previously determined to be associated with SSH concentrations and/or birth outcomes were selected a priori as covariates [16, 38]. Maternal factors included age, race/ethnicity, education, parity, gestational age at the time of blood sample collection, early-pregnancy body mass index (BMI), smoking, and pregnancy complications (i.e., gestational hypertension and preeclampsia, and gestational diabetes). Infant factors included gestational age at birth and sex. Feeding method at 6 months of age was also included in the analysis of infant growth outcomes. Detail categories of covariates are described in the Supplementary Materials. The bivariate associations between covariates and outcomes were presented in Supplementary Table 1.

Statistical analysis

Descriptive statistics were calculated for all variables. SSH were log-transformed. As individual infants have different growth patterns [39], group-based trajectory modeling (GBTM) was used to identify growth trajectories for weight-for-age (WFA) percentile and TST measured from birth to 12 months of age (birth, 1, 6, 12 months). Individuals with 2 or more measures at the study visits were included in the identification of growth trajectories. The GBTM method can identify distinct groups of infants who follow similar patterns of growth over time and accounts for variability in growth rates and the nonlinear nature of infant development. GBTM is a discrete mixture model that uses maximum likelihood estimation to define growth trajectories as a mixture of polynomial regressions. The number of trajectories was determined by the Bayesian information criterion (BIC) and relative clinical significance. Posterior probability of group membership was used to determine the growth trajectory for an infant. GBTM was performed using the traj package (https://www.andrew.cmu.edu/user/bjones/) in STATA 18.0 [40, 41].

To identify potential sensitive windows when prenatal SSH may impact fetal/infant growth, path analysis was conducted to simultaneously assess individual SSH at all trimesters and their independent associations with birth and infant growth outcomes using structural equation modeling adjusting for covariates (Supplementary Fig. 2). Bias-corrected bootstrap methods were used to estimate the confidence interval for the direct effect of individual SSH in each trimester in the path analysis. Additionally, due to prior evidence indicating sex differences in fetal growth, we examined the modification effects of infant sex by including interactions between infant sex and individual SSH concentrations in each trimester using linear regression models for birth outcomes and multinomial logistic regression models for infant growth trajectories. To understand the effect of birth outcomes on the relationship between prenatal SSH and postnatal growth, we fitted the multinomial logistic regression models with and without birth size as a covariate. Additionally, we conducted sensitivity analyses of infant growth outcomes to include feeding method at 6 months as an extra covariate. Power analysis was described in the Supplementary Materials. All analyses were conducted using STATA 18.0 (StataCorp, College Station, TX, USA).

RESULTS

Characteristics of the study cohort

The majority of the mothers were white (62.7%), overweight/obesity in early pregnancy (63.6%), and had education beyond high school (67.5%) (Table 1). Maternal serum estrogen levels (p < 0.001), but not testosterone levels (TT: p = 0.48; fT: p = 0.09), increased during pregnancy.

Table 1.

Characteristics of UPSIDE mother-infant dyads in this study.

Variablea		All participants (n = 252)
Mothers
Age (years)		29.3 ± 4.4
Race/Ethnicity
White, Non-Hispanic		158 (62.7%)
Black, Non-Hispanic		54 (21.4%)
Hispanic		21 (8.3%)
Others		19 (7.5%)
Nulliparous		84 (33.3%)
High school or less education		82 (32.5%)
Smoking during pregnancy		17 (6.8%)
Gestational hypertension or preeclampsia		30 (11.9%)
Gestational diabetes		14 (5.6%)
Early-pregnancy BMI
Normal/underweight		117 (46.4%)
Overweight		62 (24.6%)
Obesity		73 (29.0%)
Trimester 1	E1 (pg/mL)	1152.4 ±900.5
	E2 (pg/mL)	1853.9 ± 1017.1
	E3 (pg/mL)	291.3 ±260.3
	TT (ng/dL)	69.1 ± 44.6
	fT (ng/dL)	0.4 ± 0.3
Trimester 2	E1 (pg/mL)	4255.6 ±2987.7
	E2 (pg/mL)	6539.7 ±3021.5
	E3 (pg/mL)	3247.4 ± 1337.1
	TT (ng/dL)	77.4 ±53.6
	fT (ng/dL)	0.4 ± 0.3
Trimester 3	E1 (pg/mL)	6664.7 ± 4492.2
	E2 (pg/mL)	11,868.0 ±4756.9
	E3 (pg/mL)	7026.4 ±3147.2
	TT (ng/dL)	75.6 ±55.3
	fT (ng/dL)	0.4 ± 0.3
Infants
Gestational age (weeks)		39.7 ± 1.1
Female		126 (50%)
Birthweight (g)		3422 ± 472
WFA percentile at birth		48.3 ± 26.7
WFA percentile at 1 month (n = 199)		46.0 ± 28.7
WFA percentile at 6 months (n = 193)		46.4 ±29.1
WFA percentile at 12 months (n = 168)		53.6 ±29.4
TST (mm) at birth (n = 234)		4.4 ± 1.3
TST (mm) at 1 month (n = 202)		5.9 ± 1.5
TST (mm) at 6 months (n = 189)		7.0 ± 1.8
TST (mm) at 12 months (n = 161)		6.9 ± 1.6

E1 estrone, E2 estradiol, E3 estriol, TT total testosterone, fT free testosterone, WFA weight for age, TST truncal skinfold thickness.

^aContinuous variables are summarized using mean and standard deviation; Categorical variables are summarized using count and percentage.

Associations of prenatal sex steroid hormones with birth size

Birthweight z-scores.

Direct associations between individual SSH in each trimester and birth outcomes were estimated using path analysis (Table 2). E3 levels in the 3rd trimester were directly associated with higher birthweight z-scores in the full cohort (β = 0.53, 95% CI: 0.23, 0.78) and in both sexes. Sex differences were evident in the associations between other SSH and birthweight (Fig. 1A and Table 2). In female, but not male, infants, higher E1 concentrations in the 1st trimester were associated with greater birthweight z-scores (E1: β = 0.37, 95% CI: 0.02, 0.73). In male infants, fT levels in the 2nd trimester were positively associated with birthweight z-scores (β = 0.32, 95% CI: 0.005, 0.64), whereas the association was reversed in the 3rd trimester (β = −0.31, 95% CI: −0.61, −0.02).

Table 2.

Associations between prenatal sex steroid hormones and birthweight and central adiposity in total samples and by infant sex.

	Total				Male				Female
	Birthweight (z scores)		Truncal skinfold thickness (mm)		Birthweight (z scores)		Truncal skinfold thickness (mm)		Birthweight (z scores)		Truncal skinfold thickness (mm)
	β	95% CI	β	95% CI	β	95% CI	β	95% CI	β	95% CI*	β	95% CI
Estrone
Trimester 1	0.24	−0.03, 0.52	0.32	−0.10, 0.86	−0.05	−0.39, 0.30	−0.39*	−0.99, 0.21	0.37	0.02, 0.73	0.94 *	0.34, 1.54
Trimester 2	−0.10	−0.52, 0.22	−0.15	−0.78, 0.40	−0.003	−0.44, 0.43	0.08	−0.65, 0.81	−0.12	−0.60, 0.36	−0.48	−1.29, 0.34
Trimester 3	−0.10	−0.43, 0.27	−0.09	−0.62, 0.42	−0.21	−0.65, 0.23	0.13	−0.62, 0.88	0.04	−0.40, 0.48	−0.22	−0.97, 0.54
Estradiol
Trimester 1	0.11	−0.28, 0.47	0.11	−0.46, 0.74	−0.27*	−0.73, 0.19	−0.80*	−1.57, −0.04	0.34*	−0.09, 0.77	0.79 *	0.07, 1.51
Trimester 2	−0.11	−0.72, 0.45	0.22	−0.63, 1.03	0.002	−0.65, 0.65	0.33	−0.74, 1.40	−0.22	−0.89, 0.45	0.05	−1.09, 1.19
Trimester 3	0.26	−0.17, 0.70	−0.18	−1.08, 0.63	0.20	−0.39, 0.79	0.30	−0.73, 1.34	0.54	−0.09, 1.18	−0.32	−1.40, 0.75
Estriol
Trimester 1	0.004	−0.11, 0.12	−0.22	−0.40, −0.03	−0.06	−0.20, 0.09	−0.22	−0.47, 0.02	0.09	−0.06, 0.24	−0.19	−0.44, 0.06
Trimester 2	−0.16	−0.49, 0.16	0.42	−0.16, 0.98	−0.31	−0.73, 0.12	0.04	−0.69, 0.77	−0.06	−0.42, 0.29	0.70	0.09, 1.32
Trimester 3	0.53	0.23, 0.78	0.06	−0.63, 0.50	0.48	0.14, 0.82	0.09	−0.55, 0.74	0.63	0.20, 1.05	−0.25	−0.99, 0.49
Total testosterone
Trimester 1	0.13	−0.17, 0.40	0.12	−0.29, 0.56	−0.003	−0.40, 0.39	−0.10	−0.76, 0.55	0.27	−0.13, 0.67	0.33	−0.34, 0.99
Trimester 2	−0.03	−0.37, 0.28	−0.36	−0.80, 0.04	0.19	−0.21, 0.59	−0.36	−1.01, 0.29	−0.30	−0.74, 0.14	−0.33	−1.14, 0.49
Trimester 3	−0.09	−0.34, 0.22	0.47	0.03, 0.86	−0.22	−0.57, 0.14	0.61	0.02, 1.21	0.09	−0.31, 0.48	0.31	−0.39, 1.01
Free testosterone
Trimester 1	0.06	−0.19, 0.28	0.11	−0.27, 0.46	0.06	−0.25, 0.37	0.04	−0.48, 0.56	0.07	−0.26, 0.40	0.25	−0.30, 0.80
Trimester 2	0.09	−0.16, 0.31	−0.34	−0.72, 0.03	0.32 *	0.005, 0.64	−0.08	−0.60, 0.45	−0.20*	−0.55, 0.15	−0.75	−1.39, −0.10
Trimester 3	−0.07	−0.30, 0.18	0.45	0.09, 0.87	−0.31*	−0.61, −0.02	0.29	−0.23, 0.81	0.20*	−0.11, 0.51	0.67	0.13, 1.20

The sample size for birthweight was 252 and the sample size for truncal skinfold thickness was 234. Direct associations of individual sex steroid hormones at each trimester were estimated using path analysis in the total samples and linear regression models for sex differences. Sex steroid hormones were log transformed. Model was adjusted for gestational age of blood draw, maternal age, race/ethnicity, parity, early-pregnancy BMI, gestational hypertension and preeclampsia, gestational diabetes, education, smoking during pregnancy, infant sex and gestational age. Bold values are significant results based on 95% CI.

^*indicates interaction with infant sex significant at p < 0.05.

Fig. 1. Sex differences in the associations between sex steroid hormones and birth outcomes.

A The associations between sex steroid hormones and birthweight z scores by trimester. B The associations between sex steroid hormones and truncal skinfold thickness at birth by trimester. Black: total participants; Blue: male infants; Red: female infants.

TST at birth.

TST was considered a proxy for central adiposity. In the full cohort, E3 in the 1st trimester was inversely associated with TST (mm) at birth (β = −0.22, 95% CI: −0.40, −0.03), while TT and fT in the 3rd trimester were positively associated with TST (TT: β = 0.47, 95% CI: 0.03, 0.86; fT: β = 0.45, 95% CI: 0.09, 0.87). Again, sex differences were evident (Fig. 1B and Table 2). In the 1st trimester, higher E1 and E2 levels were associated with greater TST in female infants (E1: β = 0.94, 95% CI: 0.34, 1.54; E2: β = 0.79, 95% CI: 0.07, 1.51), whereas in male infants, 1st trimester E2 was associated with lower TST (β = −0.80, 95% CI: −1.57, −0.04). The positive association between 3rd trimester TT and TST was more prominent in male infants, whereas the association between 3rd trimester fT and TST was more prominent in female infants. Additionally, fT in the 2nd trimester was negatively associated with TST in female infants (β = −0.70, 95% CI: −1.39, −0.10).

Overall, 1st trimester E1 and E2 were associated with higher birthweight and increased neonatal central adiposity in female infants. 1st trimester E3 was associated with lower neonatal central adiposity, whereas 3rd trimester E3 was associated with higher birthweight in both sexes. Additionally, 3rd trimester testosterone was linked to increased neonatal central adiposity in both sexes (Fig. 2).

Fig. 2. Summary of the significant associations between maternal sex steroid hormone levels during pregnancy and birth and infant growth outcomes.

TST indicates truncal skinfold thickness; WFA indicates weight-for-age percentile. T1 = trimester 1; T2 = trimester 2; T3 = trimester 3. Gray arrows indicate associations in the combined cohort; blue arrows indicate associations in male infants; red arrows indicate associations in female infants. ↑ indicates a positive association and ↓ indicates a negative association.

Infant growth trajectories and associations with prenatal sex steroid hormones adjusting for birth size

Four growth trajectories during infancy were identified for WFA percentiles based on BIC (Supplementary Table 2): (1) low stable WFA percentiles (W1: 26.7%); (2) declining WFA percentiles in early infancy (W2: 22.7%); (3) accelerating WFA percentiles (W3: 22.4%); and (4) high stable WFA percentiles (W4: 28.2%) (Fig. 3A). Four growth trajectories were also identified for TST: (1) low stable TST (T1: 38.9%); (2) moderate stable TST (T2: 44.9%); (3) accelerating TST through 6 months followed by decelerating TST (T3: 11.3%); and (4) high TST that steadily increased through 12 months of age (T4: 4.9%) (Fig. 3B). The correlation between WFA and TST trajectories was relatively weak (Spearman’s ρ = 0.29, p < 0.01).

Fig. 3. Growth trajectory patterns during infancy.

A Weight-for-age percentile growth patterns. B Truncal skinfold thickness growth patterns.

Infant WFA trajectories.

Using the low stable WFA trajectory (W1) as the referent, 3rd trimester E3 was associated with higher probability (β = 2.10, 95% CI: 0.18, 3.27) of membership in the declining WFA trajectory (W2, Table 3). Numerous sex differences were evident in the associations between other SSH and WFA trajectories (Supplementary Table 3). Furthermore, we refitted models with additional adjustment for birthweight to evaluate the associations between prenatal SSH and postnatal growth beyond birth outcomes. Birthweight was associated with the WFA trajectories (p < 0.001). Overall, when including birthweight (Table 4), models had the same directions of associations between SSH and WFA trajectories, however the magnitude of associations tended to be enhanced in male infants and attenuated in female infants compared to models without adjustment of birthweight (Supplementary Table 3). Particularly, in male infants, the associations of 1st trimester E1, E2, TT, and fT with lower probability of the declining WFA trajectory (W2; E1: β = −4.00, 95% CI: −6.82, −1.17; E2: β = −3.23, 95% CI: −6.35, −0.10; TT: β = −3.64, 95% CI: −6.70, −0.58; fT: β = −2.56, 95% CI: −4.96, −0.16) and the high stable WFA trajectory (W4: E1: β = −3.87, 95% CI: −6.59, −1.16; E2: β = −4.36, 95% CI: −7.62, −1.11; TT: β = −3.53, 95% CI: −6.63, −0.43; fT: β = −2.47, 95% CI: −4.85, −0.10) were strengthened. Also, 2nd trimester TT was associated with increased probability of the declining WFA trajectory (W2: β = 4.48, 95% CI: 0.55, 8.41) and the accelerating WFA trajectory (W3; β = 2.89, 95% CI: 0.03, 5.76). 3rd trimester E3 was associated with higher probability of the declining WFA trajectory (W2: β = 3.93, 95% CI: 0.99, 6.88), while TT and fT were associated with lower probability of the accelerating WFA trajectory (W3; TT: β = −2.69, 95% CI: −4.69, −0.69; fT: β = −1.72, 95% CI: −3.32, −0.12) and the high stable WFA trajectory (W4: TT: β = −3.67, 95% CI: −6.66, −0.69; fT: β = −2.79, 95% CI: −5.02, −0.56).

Table 3.

The direct association of sex steroid hormone at each trimester and weight-for-age and central adiposity growth patterns during infancy.

	Weight-for-age percentile trajectories (n = 218) β, 95% CIa				Truncal skinfold thickness trajectories (n = 215) β, 95% CIa
	W1 Low stable	W2 Declining	W3 Accelerating	W4 High stable	T1 Low stable	T2 Moderate stable	T3 Accelerating-decelerating
Estrone
Trimester 1	Ref	−0.43 (−1.87, 0.97)	0.67 (−11.04, 1.82)	−0.44 (−11.75, 0.87)	Ref	0.05 (−11.09, 0.97)	1.52 (−11.20, 6.14)
Trimester 2		0.70 (−11.14, 2.35)	−0.73 (−12.47, 1.09)	1.10 (−0.86, 2.79)		0.96 (−0.30, 2.22)	−2.58 (−16.71, 1.20)
Trimester 3		−0.13 (−11.77, 1.30)	0.55 (−0.94, 2.04)	−0.74 (−12.25, 1.07)		−1.16 (−12.40, 0.06)	2.19 (−14.42, 5.01)
Estradiol
Trimester 1	Ref	−0.48 (−12.33, 1.17)	−0.32 (−12.12, 1.42)	−0.83 (−12.51, 1.08)	Ref	0.13 (−11.28, 1.32)	1.15 (−12.68, 5.19)
Trimester 2		0.59 (−12.30, 3.31)	−0.61 (−13.21, 2.26)	1.38 (−11.36, 4.51)		1.35 (−0.47, 3.41)	−1.83 (−16.23, 5.68)
Trimester 3		1.31 (−11.23, 3.58)	1.94 (−0.97, 3.62)	0.04 (−12.71, 2.24)		−0.90 (−12.80, 0.80)	0.86 (−13.01, 4.53)
Estriol
Trimester 1	Ref	0.12 (−0.58, 0.60)	−0.12 (−0.70, 0.55)	0.03 (−0.59, 0.64)	Ref	−0.13 (−0.55, 0.28)	−0.66 (−11.75, 0.74)
Trimester 2		0.35 (−11.29, 2.05)	−0.35 (−11.93, 1.57)	0.39 (−11.18, 2.23)		1.67 (0.33, 2.82)	0.40 (−11.87, 2.98)
Trimester 3		2.10 (0.18, 3.27)	0.82 (−0.84, 2.12)	1.09 (−0.74, 2.40)		−0.41 (−11.50, 0.66)	−0.35 (−12.48, 3.27)
Total testosterone
Trimester 1	Ref	0.03 (−11.59, 1.53)	−0.09 (−11.99, 1.44)	0.44 (−11.20, 1.81)	Ref	−0.79 (−11.85, 0.26)	−0.90 (−13.94, 4.20)
Trimester 2		0.56 (−0.95, 2.73)	0.40 (−11.27, 2.46)	0.04 (−11.49, 1.83)		0.13 (−11.27, 1.34)	0.15 (−13.42, 4.54)
Trimester 3		−0.93 (−12.08, 0.54)	−0.26 (−11.62, 1.20)	−0.78 (−11.97, 0.66)		0.57 (−0.43, 1.68)	0.28 (−12.80, 2.31)
Free testosterone
Trimester 1	Ref	−0.25 (−11.47, 0.82)	−0.08 (−11.45, 1.30)	−0.01 (−11.29, 1.15)	Ref	−0.46 (−11.35, 0.37)	0.07 (−13.42, 2.38)
Trimester 2		0.34 (−0.78, 1.64)	0.35 (−11.02, 1.58)	0.11 (−0.94, 1.32)		−0.20 (−11.07, 0.67)	−0.02 (−12.31, 3.52)
Trimester 3		−0.42 (−11.50, 0.85)	−0.21 (−11.21, 1.02)	−0.24 (−11.30, 0.87)		0.31 (−0.47, 1.20)	−0.47 (−12.59, 2.08)

Path analysis was used to estimate direct association. Sex steroid hormones were log transformed. Model was adjusted for gestational age of blood draw, maternal age, race/ethnicity, parity, early-pregnancy BMI, gestational hypertension and preeclampsia, gestational diabetes, education, smoking during pregnancy, infant sex and gestational age. Bold values are significant results based on 95% CI.

^aBias-corrected bootstrap confidence interval. The associations between sex steroid hormones and the T4 pattern for truncal skinfold thickness were not estimated due to limited sample size.

Table 4.

The direct association of sex steroid hormone at each trimester and weight-for-age and central adiposity growth patterns by infant sex adjusting for birth outcomes.

		Weight-for-age percentile trajectories (n = 218) β, 95% CIa				Truncal skinfold thickness trajectories (n = 215) β, 95% CIa
		W1 Low stable	W2 Declining	W3 Accelerating	W4 High stable	T1 Low stable	T2 Moderate stable	T3 Accelerating-decelerating
Male
Estrone	Trimester 1	Ref	−4.00* (−6.82, −1.17)	0.26 (−1.39, 1.91)	−3.87 (−6.59, −1.16)	Ref	−0.01 (−1.16, 1.14)	0.70 (−1.70, 3.09)
	Trimester 2		1.79 (−1.82, 5.40)	−0.92 (−3.05, 1.20)	1.87 (−1.70, 5.43)		0.70 (−0.81, 2.22)	−2.53 (−5.94, 0.89)
	Trimester 3		1.20 (−2.42, 4.82)	1.05 (−1.02, 3.11)	−0.09 (−3.71, 3.54)		−1.32 (−2.80, 0.15)	2.32 (−0.51, 5.15)
Estradiol	Trimester 1	Ref	−3.23* (−6.35, −0.10)	−0.28 (−2.24, 1.69)	−4.36* (−7.62, −1.11)	Ref	0.20 (−1.33, 1.74)	1.27 (−1.72, 4.25)
	Trimester 2		0.43 (−4.39, 5.25)	−1.97 (−5.32, 1.39)	1.42 (−3.52, 6.36)		1.13 (−1.12, 3.38)	−3.64 (−8.40, 1.13)
	Trimester 3		4.03 (−0.40, 8.45)	2.72 (−0.36, 5.80)	0.98 (−3.57, 5.53)		−1.25 (−3.31, 0.82)	2.44 (−1.97, 6.85)
Estriol	Trimester 1	Ref	−0.45 (−1.44, 0.54)	−0.17 (−0.85, 0.51)	−0.88* (−1.89, 0.14)	Ref	−0.17 (−0.65, 0.30)	−0.29 (−1.31, 0.72)
	Trimester 2		0.58 (−2.63, 3.78)	−1.18 (−3.36, 0.99)	2.23 (−1.03, 5.48)		1.42 (−0.15, 2.99)	−1.21 (−4.10, 1.68)
	Trimester 3		3.93* (0.99, 6.88)	1.03 (−0.72, 2.78)	1.22 (−1.47, 3.91)		−0.53 (−1.87, 0.82)	−1.32 (−4.11, 1.47)
Total testosterone	Trimester 1	Ref	−3.64* (−6.70, −0.58)	−0.51 (−2.68, 1.65)	−3.53* (−6.63, −0.43)	Ref	−0.65 (−2.01, 0.71)	−0.38 (−2.89, 2.14)
	Trimester 2		4.48* (0.55, 8.41)	2.89 (0.03, 5.76)	3.72* (−0.22, 7.66)		0.55 (−1.05, 2.14)	−1.76* (−5.49, 1.97)
	Trimester 3		−2.80 (−5.80, 0.19)	−2.69* (−4.69, −0.69)	−3.67* (−6.66, −0.69)		−0.06 (−1.35, 1.24)	0.50 (−2.08, 3.07)
Free testosterone	Trimester 1	Ref	−2.56 (−4.96, −0.16)	0.16 (−1.41, 1.73)	−2.47 (−4.85, −0.10)	Ref	−0.51 (−1.57, 0.56)	−0.06 (−1.83, 1.72)
	Trimester 2		1.68 (−0.96, 4.31)	1.62 (−0.45, 3.68)	1.86* (−0.82, 4.54)		0.06 (−1.11, 1.24)	−0.71 (−3.13, 1.70)
	Trimester 3		−1.21 (−3.35, 0.93)	−1.72* (−3.32, −0.12)	−2.79* (−5.02, −0.56)		0.19 (−0.87, 1.25)	−0.34 (−2.81, 2.13)
Female
Estrone	Trimester 1	Ref	0.93* (−1.34, 3.20)	1.24 (−0.74, 3.22)	−0.75 (−3.26, 1.75)	Ref	−0.04 (−1.31, 1.23)	2.55 (−0.64, 5.74)
	Trimester 2		0.77 (−1.58, 3.12)	−0.003 (−2.12, 2.11)	2.53 (−0.11, 5.16)		1.03 (−0.59, 2.65)	−3.12 (−7.61, 1.36)
	Trimester 3		−1.84 (−4.46, 0.78)	−0.73 (−2.95, 1.49)	−2.15 (−5.12, 0.81)		−0.80 (−2.39, 0.79)	3.67 (−0.58, 7.92)
Estradiol	Trimester 1	Ref	1.75* (−0.86, 4.37)	−1.01 (−3.38, 1.36)	−0.23* (−3.00, 2.55)	Ref	0.85 (−0.67, 2.36)	1.54 (−1.72, 4.79)
	Trimester 2		−1.22 (−4.79, 2.35)	1.33 (−1.75, 4.41)	1.79 (−2.19, 5.77)		1.17 (−1.14, 3.49)	−2.18 (−7.05, 2.69)
	Trimester 3		−0.39 (−3.95, 3.18)	1.92 (−1.25, 5.08)	−2.15 (−6.05, 1.75)		−1.67 (−3.91, 0.56)	1.46 (−2.64, 5.56)
Estriol	Trimester 1	Ref	0.70 (−0.17, 1.58)	−0.03 (−0.76, 0.69)	0.77* (−0.18, 1.72)	Ref	0.12 (−0.39, 0.62)	−0.50 (−1.45, 0.46)
	Trimester 2		1.64 (−0.33, 3.61)	0.47 (−1.10, 2.03)	1.55 (−0.61, 3.72)		1.22 (−0.20, 2.64)	1.29 (−1.61, 4.19)
	Trimester 3		−0.11* (−2.53, 2.31)	0.49 (−1.57, 2.54)	−0.89 (−3.53, 1.74)		−0.40 (−1.94, 1.14)	0.58 (−2.32, 3.48)
Total testosterone	Trimester 1	Ref	0.50* (−1.60, 2.61)	−0.74 (−2.63, 1.15)	1.53* (−1.12, 4.19)	Ref	−1.10 (−2.48, 0.29)	−3.76 (−6.91, −0.61)
	Trimester 2		−1.80* (−4.45, 0.84)	−0.96 (−3.59, 1.66)	−3.11* (−6.53, 0.32)		0.62 (−1.15, 2.39)	4.88* (0.54, 9.23)
	Trimester 3		0.12 (−1.87, 2.11)	1.09* (−0.97, 3.16)	1.54* (−1.03, 4.10)		−0.05 (−1.43, 1.34)	−0.47 (−3.46, 2.53)
Free testosterone	Trimester 1	Ref	−0.64 (−2.34, 1.05)	−1.15 (−2.76, 0.46)	−0.05 (−1.99, 1.88)	Ref	−0.52 (−1.69, 0.65)	−0.91 (−3.51, 1.70)
	Trimester 2		−1.29 (−3.27, 0.68)	−0.34 (−2.08, 1.41)	−2.08* (−4.48, 0.31)		−0.11 (−1.45, 1.22)	2.26 (−0.79, 5.32)
	Trimester 3		0.22 (−1.51, 1.96)	0.70* (−0.98, 2.39)	1.39* (−0.63, 3.41)		−0.42 (−1.51, 0.67)	−1.27 (−3.40, 0.85)

Path analysis was used to estimate direct association. Sex steroid hormones were log transformed. Model was adjusted for birth outcomes, gestational age of blood draw, maternal age, race/ethnicity, parity, early-pregnancy BMI, gestational hypertension and preeclampsia, gestational diabetes, education, smoking during pregnancy, infant sex and gestational age. Bold values are significant results based on 95% CI.

^aBias-corrected bootstrap confidence interval. The associations between sex steroid hormones and T4 for truncal skinfold thickness were not estimated due to limited sample size.

^*Indicates interaction with infant sex significant at p < 0.05. The comparisons of sex steroid hormones between the TST T4 and T1 patterns were not estimated due to limited number of infants with the T4 pattern.

Sensitivity analysis adjusting for feeding method at 6 months identified similar results except that in male infants 3rd trimester E2 was associated with higher probability of the W2 and W3 trajectories, while 2nd trimester fT was associated with higher probability of the W3 trajectory. In female infants, 2nd trimester TT and fT were associated with lower probability of the W4 trajectory (Supplementary Table 4).

Infant TST trajectories.

Few associations between prenatal SSH and infant TST trajectories were observed. Using the low stable TST trajectory (T1) as the referent, 2nd trimester E3 was associated with higher probability of the moderate stable TST trajectory (T2) (β = 1.67, 95% CI: 0.33, 2.82) (Table 3). TST at birth was also significantly linked to TST trajectories (p ≤ 0.001). Including TST at birth attenuated the associations of 2nd trimester E3 with the moderate stable TST trajectory (T2) in both male and female infants (Table 4). But, in female infants, 1st trimester TT was associated with lower probability of the accelerating and decelerating TST trajectory (T3; β = −3.76, 95% CI: −6.91, −0.61), while the relationship was reversed in the 2nd trimester (β = 4.88, 95% CI: 0.54, 9.23). Sensitivity analysis adjusting for feeding method at 6 months also identified the positive association between 2nd trimester TT and higher probability of the T3 trajectory (Supplementary Table 4). Of note, the comparisons of SSH between the high accelerating TST trajectory (T4) and the low stable TST trajectory (T1) were not estimated due to limited number of infants with the high accelerating TST trajectory (T4).

To summarize, the associations between SSH and infant WFA trajectories were more prominent in male infants. Specifically, for the two trajectories with higher infant WFA, the accelerating WFA and high stable WFA trajectories (W3 and W4), 1st trimester E1, E2, and testosterone, as well as 3rd trimester testosterone, were associated with lower probability of the W4 trajectory. 3rd trimester testosterone was also associated with a lower probability of the W3 trajectory (Fig. 2). Conversely, there were no strong links between SSH and infant TST trajectories, except that 1st and 2nd trimester testosterone had opposite associations with the accelerating-decelerating TST trajectory (T3) in female infants (Fig. 2).

DISCUSSION

To date, this study is the most comprehensive evaluation of the associations between prenatal maternal SSH and fetal/infant growth with 4 major SSH measured in all trimesters and 4 timepoints of infant anthropometric assessments. Our findings reveal the complex interplay between prenatal SSH, birth size, and postnatal growth, with implications for understanding the developmental origins of childhood obesity. Notably, the associations between prenatal SSH and both birth size and postnatal growth were found to vary by the specific individual SSH and the timing of pregnancy. Furthermore, the study uncovered significant sex differences in these associations, suggesting potential underlying mechanisms that may differ between male and female infants. Importantly, the observed associations between prenatal SSH and postnatal weight and adiposity growth were beyond birth size, underscoring the unique role of prenatal SSH in shaping later childhood growth.

Specifically for birth size, our finding of a positive association between 3rd trimester E3 and birthweight aligns with previous studies [18, 19, 21–23]. However, limited research has explored the associations of other SSH with birthweight, or their interactions with infant sex. To our knowledge, no studies have investigated prenatal SSH in relation to neonatal adiposity, despite its link to later childhood obesity [3]. This study found that 1st trimester E1 and E2 were associated with greater birthweight and/or neonatal TST in female infants. Two prior studies on 1st trimester E2 and birthweight reported inconsistent results: one assessed only female infants and showed a similar but non-significant trend [23], while the other found no associations but did not examine sex differences [26]. Thus, early-pregnancy E1 and E2 may be linked to fetal growth in female infants. Additionally, the negative association between 3rd trimester fT and birthweight in male infants is consistent with previous findings [25, 27, 42]. In contrast to the birthweight findings, 3rd trimester testosterone was positively associated with neonatal adiposity in both sexes. Overall, the relationship between prenatal SSH, birthweight, and adiposity varied by hormone type, infant sex, and the timing of exposure.

For infant growth, unlike the limited previous research, which only focused on 3rd trimester individual SSH [18, 27, 28], this study assessed multiple SSH across pregnancy and their associations with infant adiposity growth trajectories, which are linked to childhood adiposity and cardiovascular risks [43]. Specifically, we found a positive association between 3rd trimester E3 and the declining WFA trajectory in male infants, contrasting with a previous study that correlated E3 with children’s weight at 2 years of age [18]. This difference may be due to the previous study’s lack of adjustment for birthweight and not assessing sex differences. Additionally, this study found that 1st trimester E1/E2 was associated with a lower probability of the high stable WFA trajectory in male infants, a relationship that has not been previously examined. Regarding testosterone, 2nd trimester TT was linked to a higher probability of the accelerating WFA trajectory, while 1st- and 3rd-trimester TT/fT were associated with lower probabilities of the accelerating and high stable WFA trajectories in male infants. Two prior studies focused on 3rd trimester TT and androgen activity: one study (n = 49) reported a positive association with rapid weight gain (birth to 6 months) in male infants [27], which differs from our findings, while the other identified an association with a higher probability of an accelerated catch-up growth pattern compared to a consistent high weight pattern in male children [28], similar to our finding in male infants (data not shown). Interestingly, prenatal testosterone was more strongly linked to postnatal TST trajectories in female infants. Overall, these findings emphasized the importance of considering longitudinal growth patterns, sex differences, and birth size when assessing the impact of prenatal hormone exposure on postnatal growth, with potential implications for understanding the developmental origins of childhood obesity.

These findings underscored the multifaceted mechanisms potentially influencing fetal and postnatal growth, including individual SSH, developmental timing, and fetal sex. During early pregnancy, E1 and E2, the primary estrogens [44], promote angiogenesis, vascularization, blood flow in myome-trial and placental arteries, and facilitate nutrient transport to the fetus by regulating uteroplacental glucose transporter expression [9, 10]. In mid-late pregnancy, E3, mainly synthesized by the placenta from dehydroepiandrosterone sulfate produced by fetal adrenal glands, becomes the dominant estrogen [18, 44], and serves as an indicator of placental and fetal adrenal function, reflecting fetal well-being [18, 45]. E3 has also been suggested to regulate uteroplacental vascularization and blood flow in late pregnancy, though its full physiological function remains unclear [46]. Postnatally, estrogens are known to interact with epigenetic enzymes and regulate epigenetic modifications [12, 47]. Notably, prenatal E3 exposure affects epigenetic modifications in the brain, muscle, and adipose tissues that may influence postnatal adiposity and growth [12]. However, the sex differences in the relationship between individual estrogens, birth size, and postnatal growth need further investigation.

The associations between prenatal testosterone exposure and fetal and postnatal growth also varied by the timing of exposure and infant sex. 3rd trimester testosterone was linked to lower birthweight but higher adiposity at birth. Animal studies suggest that maternal testosterone can downregulate placental amino acid transport in rats [11] and reduce uterine artery blood flow and placental vasculature, especially in male rat fetuses [48], which may explain the association between high maternal testosterone and lower birthweight in males. Additionally, testosterone may promote visceral preadipocyte proliferation and adiposity in humans [13, 49], with sex-specific responses likely driven by adipose tissue sex dimorphism, such as differences in sex steroid receptor expression, adipocyte precursor cell numbers, and sex chromosome programming [29]. Prenatal testosterone may also affect postnatal growth through interactions with epigenetic enzymes that regulate epigenetic modifications [50]. Again, the sex differences in the relationship between prenatal testosterone and fetal and postnatal growth warrant further investigation.

The strengths of this study include: (1) the prospective repeated measures of the major estrogens and testosterone across all three trimesters; (2) sensitive LC-MS/MS and equilibrium dialysis method were used to quantify estrogens and testosterone, particularly the low concentrations of E3 in early pregnancy and testosterone across pregnancy; (3) by using path analysis models and including interactions with infant sex, we were able to estimate direct associations of SSH in three trimesters simultaneously, giving insight into sensitive periods of hormone exposure and sex-specific responses; (4) birth size was adjusted in the models of postnatal growth, underscoring the role of prenatal SSH in postnatal growth beyond birth size. Several caveats are important to consider when interpreting the findings of this study. The prospective UPSIDE study was not originally powered to examine the particular research questions assessed in this study. We focused on estrogens and testosterone and did not assess additional prenatal hormones, such as dihydrotestosterone, and progesterone, that may mediate the role of estrogens and testosterone in fetal and postnatal growth [28]. Also, the concentrations of fT were low, making them susceptible to methodological issues. Therefore, the results of fT should be interpreted alongside those of TT. Additionally, we used a relatively crude measure of adiposity, and more advanced tools, such as dual-energy x-ray absorptiometry, magnetic resonance imaging, or air displacement plethysmography may provide more accurate measures.

CONCLUSION

Using data on sex steroids across pregnancy, we demonstrate some novel findings that: (1) early pregnancy estrogens are associated with greater weight and adiposity in female newborns; (2) late pregnancy testosterone is associated with neonatal adiposity in both sexes; (3) male infants exposed to higher prenatal maternal sex steroids, particularly in early pregnancy, were less likely to have postnatal high weight trajectories. These findings contribute to a deeper understanding of the complex interplay between prenatal SSH and fetal and postnatal growth, emphasizing the need for further research to elucidate the underlying mechanisms and potential long-term implications for childhood obesity.

Supplementary Material

Supplementary information The online version contains supplementary material available at https://doi.org/10.1038/s41366-025-01743-3.

DATA AVAILABILITY

The UPSIDE-ECHO study data have been uploaded to the NICHD Data and Specimen Hub (https://echochildren.org/dash/). De-identified data on ECHO participants are available to the scientific community.