Prenatal Perfluoroalkyl Substances Exposure and Maternal Sex Steroid Hormones Across Pregnancy

Abstract

Background:

Poly- and perfluoroalkyl substances (PFAS) are ubiquitous and persistent environmental contaminants that may act as endocrine disruptors in utero, but the specific endocrine pathways are unknown.

Objective:

We examined associations between maternal serum PFAS and sex steroid hormones at three time points during pregnancy.

Methods:

Pregnant women participating in the Understanding Pregnancy Signals and Infant Development (UPSIDE) study contributed biospecimens, questionnaire, and medical record data in each trimester (n=285). PFAS (including perfluorooctanoic acid (PFOA), perfluorooctanesulfonic acid (PFOS), perfluorohexane sulfonic acid (PFHxS), perfluorononanoic acid (PFNA) and perfluorodecanoic acid (PFDA)) were analyzed in second-trimester serum samples by high-performance liquid chromatography and tandem mass spectrometry (LC-MS/MS). Total testosterone [TT], free testosterone [fT], estrone [E1], estradiol [E2], and estriol [E3]) were measured by LC-MS/MS in serum samples from each trimester. Linear mixed models with random intercepts were used to examine associations between log-transformed PFAS concentrations and hormone levels, adjusting for covariates, and stratifying by fetal sex. Results are presented as the mean percentage difference (Δ%) in hormone levels per ln-unit increase in PFAS concentration.

Results:

In adjusted models, PFHxS was associated with higher TT (%Δ= 20.0, 95%CI: 1.7, 41.6), particularly among women carrying male fetuses (%Δ= 15.3, 95%CI: 1.2, 30.7); this association strengthened as the pregnancy progressed. PFNA (%Δ= 7.9, 95%CI: 3.4, 12.5) and PFDA (%Δ= 7.2, 95%CI: 4.9, 9.7) were associated with higher fT, with associations again observed only in women carrying male fetuses. PFHxS was associated with higher levels of E2 and E3 in women carrying female fetuses (%Δ= 13.2, 95%CI: 0.5, 29.1; %Δ= 17.9, 95%CI: 3.2, 34.8, respectively). No associations were observed for PFOS and PFOA.

Conclusion:

PFHxS, PFNA, and PFDA may disrupt androgenic and estrogenic pathways in pregnancy in a sex-dependent manner.

Graphical Abstract

1.0. INTRODUCTION

Per- and polyfluoroalkyl substances (PFASs) are a large class of anthropogenic chemicals (>3000) characterized by having at least one perfluoroalkyl group (Buck et al. 2011). Since the 1940s, these compounds have been widely used in industrial processes and consumer products including disposable food packaging, cookware, stain-resistant fabrics, and carpet (Buck et al. 2011; Glüge et al. 2020). Outside of occupational exposure, the predominant route of exposure to PFAS is ingestion of contaminated drinking water and food (Kannan and Wu 2019). The strong C-F bonds make PFAS highly persistent in the environment and humans (Buck 2011). For example, the biological half-lives of common PFAS such as perfluorooctanesulfonic acid (PFOS), periluorooctanoic acid (PFOA), and perfluorohexane sulfonic acid (PFHxS) range from 3.8 - 8.5 years (Olsen 2007). Based on evidence of adverse health effects associated with exposure to long-chain (≥ 7 fluorinated carbons) PFAS, starting in 2000, numerous countries including the U.S. began to voluntarily phase out the use of PFOS, and PFOA, now considered “legacy” PFAS (EPA 2000; Sunderland et al. 2019). However, because of their environmental persistence, they are still commonly detected in human populations around the world (Baluyot et al. 2021; Karásková et al. 2016; Macheka et al. 2021).

In the U.S., PFAS have been detected in over 95% of adolescents and adults (Lewis, 2015) including pregnant women (Ashley-Martin et al. 2016; Jaacks et al. 2016; Mamsen et al. 2019; Romano et al. 2021) and they can cross the placenta to reach the developing fetus. An extensive body of literature suggests that prenatal PFAS exposure impacts fetal growth, size at birth, and subsequent risk of childhood obesity (Apelberg et al. 2007; Bjerregaard-Olesen et al. 2019; Liu et al. 2018; Maisonet et al. 2012; Wikström et al. 2020). Although the biological pathways underlying these associations are unknown, several have been proposed including endocrine disruption (Rickard et al. 2022).

Changes in maternal hormone levels during pregnancy has been associated with adverse birth outcomes, including growth restriction and low birth weight (Mucci et al. 2003; Noyola-Martínez et al. 2019). Several studies have examined the impact of PFAS on hormonal homeostasis in pregnant women. For example, studies have observed associations between PFAS and thyroid hormones during pregnancy (Aimuzi et al. 2020; Ballesteros et al. 2017; Berg et al. 2015; Chan et al. 2011; Inoue et al. 2019). A study of pregnant women in China (n=752) reported that perfluorobutane sulfonate (PFBS) and perfluoroheptanoic acid (PFHpA) were inversely associated with follicle-stimulating hormone (FSH) and luteinizing hormone (LH). A handful of studies have examined serum PFAS concentrations in relation to sex steroid hormones measured in cord blood at birth. In the Shandong province, China, the LWBC study (n=351) reported a positive association between PFOA and PFHxS and estradiol (E2), testosterone, and E2/testosterone ratio (Yao et al. 2019; Yao et al. 2021). In a population (n=424) from Hebei province, China, PFOS was positively associated with E1 and E3 but negatively related to E2 (Wang et al. 2019). A third Chinese study in Hubei, also reported positive associations between PFOA, PFDA, PFHxS and PFNA and estrogens concentrations but testosterone was not examined (Liu et al. 2021). Similarly a Japanese study (n=189) reported positive associations between cord PFOS and E2 concentrations, although only PFOA and PFOS were studied (Itoh et al. 2016). Recently, Yang and colleagues reported inverse associations between first trimester PFDA and E1 measured three times across pregnancy in from Tangshan City, China (Yang et al. 2022). Here we extend this limited literature by examining PFAS concentrations in relation to sex steroid hormones measured serially across pregnancy in a U.S. cohort.

2.0. METHODS

2.1. Study overview and population

From 2015-2019, 326 first trimester pregnant women from outpatient obstetric clinics affiliated with the University of Rochester were enrolled in the Understanding Pregnancy Signals and Infant Development (UPSIDE) study, an ongoing prospective birth cohort originally designed to study maternal prenatal psychological distress and child outcomes (O'Connor et al. 2021). Inclusion criteria were: 1) singleton pregnancies, 2) ≥18 years old, 3) no known substance abuse problems or history of psychotic illness, 4) ability to communicate in English. Additionally, women with certain endocrine conditions were excluded (e.g., polycystic ovary syndrome). Women completed study visits in each trimester including physical exams, biospecimen collection, and questionnaire completion. This data collection was complemented by medical record review. Sociodemographic information, including self-reported race and ethnicity, was collected at baseline recognizing that these social constructs are strong predictors of environmental exposures as well as maternal and child health outcomes (Bulka et al. 2019; Burris et al. 2011; Goldenberg et al. 2008; Shiao et al. 2005; Wang et al. 2020). The current analysis includes participants with PFAS data as well as sex steroid hormone concentrations from at least one trimester (n=285). The research protocol was approved by the Institutional Review Boards of the University of Rochester School of Medicine and Dentistry and Rutgers University. All study activities were explained to participants before obtaining written consent to participate in the study.

2.2. PFAS Analysis

Blood samples for PFAS measures were collected during the second trimester of pregnancy (21 ± 1.8 weeks). After processing, serum was stored at −80°C until samples were shipped on dry ice for PFAS analysis at the Wadsworth Center’s Human Health Exposure Analysis Resource (HHEAR) Laboratory Hub. Using an established high performance liquid chromatography (HPLC) and tandem mass spectrometry (MS/MS) method, thirteen PFAS were measured: PFOS, PFBS, periluorohexane sulfonate (PFHxS), perfluorodecane sulfonate (PFDS), perfluorohexanoic acid (PFHxA), PFHpA, perfluorononanoic acid (PFNA), perfluorodecanoic acid (PFDA), perfluoroundecanoic acid (PFUnDA), perfluorododecanoic acid (PFDoDA), n-methylperfluoro-1-octanesulfonamidoacetic acid (N-MeFOSAA), and perfluorooctane sulfonamide (PFOSA) (Honda et al. 2018). PFAS were extracted using an ion-pairing extraction procedure followed by HPLC-MS/MS detection and quantification, as described elsewhere (Honda et al. 2018). The limit of detection (LOD) for individual PFAS varied from 0.02 to 0.03 μg/L. For the current analysis, we included PFAS with concentrations above the LOD in ≥ 80% of participants. These included PFOA, PFOS, PFNA, and PFHxS (detected in 100% of participants) as well as PFDA (detected in 83%).

2.3. Sex Steroid Hormone Analysis

Serum samples from each trimester (12 ± 1.3, 21 ± 1.8 and 31 ± 1.9 weeks) were shipped on dry ice to the Endocrine and Metabolic Research Laboratory at the Lundquist Institute at Harbor-UCLA Medical Center where five hormones were measured: estrone (E1), E2, estriol (E3), total testosterone (TT), and free testosterone (fT). Hormone analysis was done using standard methods with minor modifications to the runtime and system parameters (Day et al. 2020; Sathyanarayana et al. 2017; Shiraishi et al. 2008). For TT (ng/d), LC-MS/MS analyses were conducted using a Shimadzu HPLC system (Columbia, MD) interfaced with an Applied Biosystems API5500 LC-MS/MS (Foster City, CA) with a Turbo-Ion-Spray source in the positive ionization mode. For estrogen quantification (pg/ml), the Shimadzu HPLC system (Columbia, MD) interfaced with a triple quadrupole mass spectrometer (API5000 LC-MS/MS, Foster City, CA) was used. Serum from one women had values below detection limits for first trimester TT and fT; twenty-two women had below detection levels and 10 women had missing values for first trimester E3. Additional details of hormone analysis are provided in the Supplemental Material.

2.4. Statistical Analysis

Descriptive statistics were calculated for maternal hormones, PFAS, and all covariates of interest. Concentrations below the limit of detection for PFDA, TT, fT, E3, and missing values for E3 were replaced by the LOD divided by the squared root of 2. Distributions of PFAS and maternal hormone concentrations were right skewed and thus natural log transformed for all analyses. We first examined correlations between PFAS and hormone concentrations using Spearman correlation. From generalized additive models, we found no evidence of non-linearity. We then fit generalized linear models (GLM) to examine associations between PFAS concentrations and individual maternal hormones in each trimester. To examine longitudinal measures of maternal hormones, we used linear mixed models (LMMs) with separate models fit for each hormone outcome. To deal with variation in gestational age at sample collection, models included a fixed effect for PFAS exposure, a smooth function for gestational age, and a random effect for each individual participants. Results are presented as the mean percentage difference (Δ%) in hormone levels per ln-unit increase in PFAS concentration. Potential confounders and covariates were selected for consideration based on existing literature and directed acyclic graphs. These included maternal age, race and ethnicity, parity, gravidity, fetal sex, pre-pregnancy body mass index (BMI), marital status, education, employment status, smoking, season of biospecimen collection and for the longitudinal models, trimester. In final models, covariates changing the beta coefficient of hormone levels by 10% or more were retained, including maternal age (continuous), parity (categorical), maternal race (categorical), education (categorical), and pre-pregnancy BMI (continuous).

Given prior literature on PFAS (Rickard et al. 2022), we additionally examined effect modification by fetal sex and parity. To examine interaction terms between each PFAS and 1) fetal sex; and 2) parity, we fit separate LMMs. Additionally, we also stratified our analysis by fetal sex and parity to better assess these potential modifiers (Buckley et al. 2017). All statistical analyses were conducted using SAS version 9.4 and p-value <0.05 were considered statistically significant.

3.0. RESULTS

A total of 285 participants with PFAS and hormone measurements were included in this analysis (Table 1). On average, participants were 29.0 ± 5 years old, with an average pre-pregnancy BMI of 28.0 ± 7 kg/m² and 66% had at least one previous pregnancy. The majority had at least a high school education (76%) and were nonsmokers (93%). Most women (62%) identified as White, with 70 (25%) identifying as Black, and 37 (13%) as other races or ethnicities.

Table 1.

Characteristics of the UPSIDE study population (n=285).

Characteristic	Mean	STD
Age (years)	29.0	4.6
Pre-pregnancy BMI (kg/m2)	28.0	7.1
	n	%
Parity
0	98	34
>0	187	66
Gravidity
0	57	20
>0	227	80
Race
Whites	178	62
Blacks	70	25
Other	37	13
Ethnicity
Hispanic	28	10
Non-Hispanic	257	90
Marital Status
Married/living as married	170	60
Other	109	38
Education
HS or less	107	37
Some College/Bachelors degree	107	37
Graduate degree	71	26
Smoking during pregnancy
None	265	93
Any	20	7
Fetal Sex
Male	143	50
Female	142	50

BMI: Body Mass Index, STD: Standard Deviation. % may not add up to 100% due to missing.

PFOA, PFOS, PFNA, and PFHXS were above the LOD in all participants with median concentrations of 0.6 μg/L, 2.5 μg/L, 0.3 μg/L, 1.7 μg/L, respectively (Table 2). PFDA was present in 83% of participants and median concentration was 0.1 μg/L. Caucasian women, nulliparas, and those with higher education attainment had higher PFOA and PFHxS concentrations than their counterparts (p<0.05; data not shown). Descriptive statistics for all maternal hormones are shown in Table 3. Overall, mean concentrations for estrogens increased sharply across pregnancy whereas TT and fT remained relatively stable over gestation (Table 3).

Table 2.

Distribution of perfluoroalkyl substances (PFAS) (μg/L) in the UPSIDE study population (n=285).

PFAS	N	%<LOD	Mean	STD	Min	50th	90th	95th	Max
PFOA	285	0	0.69	0.50	0.04	0.59	1.16	1.47	5.45
PFOS	285	0	2.86	1.98	0.61	2.50	4.29	5.52	21.30
PFNA	285	0	0.32	0.32	0.06	0.25	0.50	0.67	2.87
PFHxS	285	0	1.90	0.87	0.56	1.72	3.09	3.46	7.17
PFDA	285	17	0.08	0.10	<LOD	0.06	0.15	0.2	1.04

PFOS: perfluorooctanesulfonic acid, PFOA: perfluorooctanoic acid, PFHxS: perfluorohexane sulfonic acid, PFNA: perfluorononanoic acid, PFDA: perfluorodecanoic acid, STD: Standard Deviation, Min: minimum, Max: Maximum, LOD: limit of detection

Table 3.

Distribution of maternal sex steroid hormones in the UPSIDE study population (n=285).

Hormone	Trimester	Mean	STD	Min	50th	90th	95th	Max
Testosterone (ng/dl)	1	71.3	45.8	8.8	61.5	129.0	156.0	303.0
	2	79.6	57.2	4.6	62.9	144.0	192.0	413.5
	3	78.3	57.8	9.6	63.5	152.0	192.5	349.5
Free Testosterone (ng/dl)	1	0.6	0.2	0.2	0.6	0.8	0.9	1.2
	2	0.5	0.1	0.2	0.5	0.7	0.76	1.0
	3	0.5	0.1	0.2	0.5	0.7	0.80	1.0
Estrone (pg/ml)	1	1143.5	882.5	124.0	937.0	2150.0	2840.0	5640.0
	2	4180.0	2948.9	423.0	3550.0	7760.0	6890.0	20500.0
	3	6809.4	5128.3	751.0	5900.0	12300.0	15400.0	37700.0
Estradiol (pg/ml)	1	1846.3	989.1	356.0	1610.0	3120.0	3720.0	6390.0
	2	6519.8	2945.0	1750.0	5940.0	9870.0	12100.0	27700.0
	3	11961.5	4853.7	2380.0	11100.0	18300.0	19900.0	31900.0
Estriol (pg/ml)	1	323.2	264.3	10.0	267.0	670.0	783.0	1470.0
	2	3239.5	1313.8	351.0	3130.0	4970.0	5400.0	11900.0
	3	7169.5	3364.0	1030.0	6620.0	10800.0	127000	26300.0

STD: Standard Deviation, Min: minimum, Max: Maximum.

Trimester-specific associations between PFAS and sex steroid hormones are shown in Supplemental Table 1. In adjusted models, a ln-unit increase in PFHxS was associated with higher TT in each trimester (Trimester 1: %Δ= 14.6, 95%CI: −3.9, 36.3; Trimester 2: %Δ= 19.5, 95%CI: −1.0, 44.8; Trimester 3: %Δ= 30.3, 95%CI: 8.3, 56.8). After stratification by fetal sex, only associations in women carrying boys were observed (Trimester 1: %Δ= 22.9, 95%CI: −5.8, 60.0; Trimester 2: %Δ= 34.6, 95%CI: 0.30, 80.6; Trimester 3: %Δ= 49.2, 95%CI: 9.9, 101.4) (Supplemental Tables 2 and 3). In both full cohort and stratified analyses, associations strengthened as pregnancy progressed. PFNA (%Δ= 7.6, 95%CI: 2.0, 13.9; %Δ= 6.5, 95%CI: 1.0, 12.8; %Δ= 12.0, 95%CI: 5.1, 18.5), and PFDA concentrations (%Δ= 5.1, 95%CI: 0.7, 9.7; %Δ= 7.6, 95%CI: 3.5,11.9; %Δ= 11.1, 95%CI: 6.7,15.8) were associated with higher fT levels in all trimesters in the full cohort as well as in analyses limited to mothers carrying male fetuses. In trimester-specific models, few associations were observed between PFAS and estrogens, however, PFHxS was associated with higher second trimester E2 (%Δ= 12.3, 95%CI: −0.2, 25.9) and PFDA was associated with lower second trimester E3 (%Δ= −7.8, 95%CI: −13.6, −1.6). No associations were observed for PFOS and PFOA.

In our main adjusted linear mixed models examining longitudinal measurements of maternal hormones across pregnancy, a ln-unit increase of PFHxS was associated with higher TT (%Δ= 20.0, 95%CI: 1.7, 41.6) and ln-unit increases in PFNA (%Δ= 7.9, 95%CI: 3.4, 12.5) and PFDA (%Δ= 7.2, 95%CI: 4.9, 9.7) were associated with higher fT levels (Figure 1, Supplemental Material Table 4). In models stratified by fetal sex, positive associations between PFHxS and PFNA with TT levels was observed only in women carrying male fetuses (PFHxS: %Δ= 31.7, 95%CI: 1.7, 70.6; PFNA: %Δ= 26.2, 95%CI: 3.2, 54.2), however associations between PFNA and PFDA and fT were similar in both sexes. Finally, PFHxS was associated with higher E2 and E3 in women carrying female, but not male, fetuses (%Δ= 13.2, 95%CI: −0.1, 29.1; %Δ= 17.9, 95%CI: 3.2, 34.8, respectively). In addition to stratified analysis, we evaluated effect modification by including interaction terms in the mixed models. Interaction terms for PFAS*fetal sex were largely non-significant (with the exception of PFNA*E3), whereas interaction terms for PFAS*parity where significant between PFOA and PFOS with E2 and between PFOS and PFNA with E3. (Supplemental Table 5). However, results from analyses stratified by parity were similar to our main linear mixed models with the exception of inverse associations between PFNA (%Δ= −11.1, 95%CI: −18.4,−3.5) and PFDA with E3 (%Δ=−7.1, 95%CI: −11.6, −2.5) among parous women (Supplemental Table 6).

Figure 1.

Percent difference maternal sex steroid hormone levels from longitudinal models in xrelation to ln-unit of PFAS concentrations in UPSIDE mothers by fetal sex, (n=285).

Models adjusted for maternal age, race, parity, education, pre pregnancy BMI, and gestational age at sample collection. PFOS: perfluorooctanesulfonic acid, PFOA: perfluorooctanoic acid, PFHxS: perfluorohexane sulfonic acid, PFNA: perfluorononanoic acid, PFDA: perfluorodecanoic acid, T: testosterone, FT: free testosterone, E1: estrone, E2: estradiol, E3: estriol, 95%CI: 96% Confidence Interval

4.0. DISCUSSION

In this sample of pregnant women, we examined associations between prenatal PFAS concentrations and maternal serum sex steroid hormones. We observed that PFHxS was associated with higher concentrations of testosterone, an association that grew stronger as pregnancy progressed, particularly in women carrying male fetuses. Additionally, PFNA and PFDA were associated with higher levels of fT. PFNA concentrations were associated with lower E3 concentrations only in women carrying male fetuses, while PFHxS concentrations were associated with higher E3 levels only in women carrying female fetuses. Our results extend prior work demonstrating that PFAS are endocrine disruptors during pregnancy with impacts that may differ by fetal sex. Our results also suggest that some PFAS may exert androgenic effects.

In the U.S. population, levels of PFOA and PFOS declined from 2000 to 2014 by more than 60% and 80%, respectively as they have been phased out of manufacturing and use (CDC 2020). Nevertheless, of the PFAS studied here, PFOS was detected in the highest concentrations (median: 2.5 μg/L, range: 0.6–21.3 μg/L) in these samples from 2015-2019, reflecting its persistence in the environment including drinking water sources and manufactured consumer products. PFOS concentrations were similar to those reported in a Colorado pregnant women sampled from 2009-2014 (median: 2.4 μg/L), but lower than those in a 2008-2014 Canadian pregnancy cohort (median range for five sites: 4.1- 4.8 μg/L) (Arbuckle et al. 2020; Starling et al. 2017). PFHxS concentrations (median: 1.7 μg/L) were higher than some previous studies including the Canadian (median range for five sites: 0.9-1.3 μg/L) and Colorado (0.8 μg/L) studies as well as the 2018 National Health and Nutrition Examination Survey (NHANES) (median: 1.1 μg/L) (Arbuckle et al. 2020; CDC 2022; Starling et al. 2017). There are documented demographic, geographic, and temporal differences in PFAS concentrations among different populations (Kato et al. 2014; Lauritzen et al. 2016; Lien et al. 2013). Similar to other reports, our data suggest that Caucasian women, nulliparas, and those with higher education attainment had higher PFAS exposures compared to their counterparts (Kato et al. 2014; Sagiv et al. 2015). Detectable levels were much lower for other PFAS including shorter chain PFAS, PFBS (8% detection) and PFHpA (7% detection) suggesting potentially fewer exposed individuals and/or shorter half-lives and more rapid elimination, resulting in reduced participant body burden.

To our knowledge, no other epidemiological study has examined PFAS exposure and maternal sex steroid hormone levels in a U.S. pregnant population. Five studies have measured PFAS and sex steroids in cord blood at birth (Itoh et al. 2016; Liu et al. 2021; Wang et al. 2019; Yao et al. 2019; Yao et al. 2021) while only one other study have measured PFAS and sex steroid during pregnancy (Yang et al. 2022). All of the studies were conducted in Chinese or Japanese populations, and half of them measured only estrogens. In the first study, PFOA and PFOS were measured in cord blood from 189 Japanese pregnant women while E2 and TT were measured in cord blood (Itoh et al. 2016). The LWBC study in China, measured 10 PFAS E2, and TT in cord blood serum from 351 pregnancies (Yao et al. 2019; Yao et al. 2021). The LWBC study observed a positive association between cord blood PFNA and TT, while the Japanese study reported no associations between either PFOA or PFOS and TT. However, in a subsequent analysis from the latter cohort, PFOS was positively associated with cord concentrations of the adrenal androgen dehydroepiandrosterone (DHEA), whereas inverse associations between PFOA and DHEA were observed, with stronger associations in women who gave birth to boys versus girls (Goudarzi et al. 2017). In that study, moreover, prenatal PFOS was associated with a decrease in the cord glucocorticoid/androgenic hormone ratio, suggesting that PFOS may shift steroidogenesis towards androgenic pathways. This hypothesis is further supported by in vitro data showing that a mixture of PFOA, PFOS, PFNA, PFHxS, and PFDA may increase the ratio between estrogens and androgens by enhancing transcription of the aromatase enzyme, but can also affect androgen receptors leading to hormonal imbalance (Kjeldsen and Bonefeld-Jørgensen 2013; Liu et al. 2018). Our study extends this work on steroidogenic pathways as measured by maternal prenatal sex steroid hormones.

Maternal PFAS have also been studied in relation to indirect biomarkers of prenatal androgen activity, such as infant anogenital distance (Dean and Sharpe 2013; Thankamony et al. 2016). Evidence from animal models and humans suggests higher androgen exposure during the “male programming window” in early pregnancy is reflected in longer, more androgenized AGD in males at birth. Prior work demonstrates that AGD is sensitive to gestational exposures to other endocrine disruptors including phthalates and glyphosate (Lesseur et al. 2021; Swan et al. 2015; Zarean et al. 2019). To date, four studies have examined maternal PFAS in relation to infant AGD (Arbuckle et al. 2020; Christensen et al. 2021; Lind et al. 2017; Tian et al. 2019). Three of these studies observed longer AGD in boys in relation to higher prenatal PFAS exposure, suggesting potentially androgenic effects that are consistent with the current observations (Arbuckle et al. 2020; Christensen et al. 2021; Lind et al. 2017). A smaller study reported positive associations between prenatal exposure to PFOS, PFOA, and PFHxS and TT levels in 15 year old girls (n=72) (Maisonet et al. 2015). Taken together, these results suggest that developmental PFAS exposures may impact androgenic pathways in a sex-specific manner.

We also observed possible estrogenic effects of PFHxS in women carrying female fetuses. In Hubei, China, investigators measured PFAS and estrogens in cord blood (n=942). They reported small, positive associations between PFDA and PFNA with all three estrogens (E1, E2, and E3) (Liu et al. 2021). They also reported similar findings when the PFAS mixture was evaluated (PFOA, PFDA, PFNA, PFOS, PFHxS, and the alternative PFAS, F-53B). The LWBC study also reported a positive association between PFHxS and E2 levels in cord blood of female infants, while the Japanese study reported positive associations between PFOS and E2 only in male infants. In separate analyses, the LWBC study examined the role of steroidogenic enzymes in these relationships, reporting that placental 3β-hydroxysteroid dehydrogenase (3β-HSD1) and 17β-hydroxysteroid dehydrogenase (17β-HSD1) mediated the relationship between PFHxS and E2. In vitro studies offer further support for the hypothesis that PFAS can alter steroidogenesis however, both estrogenic and anti-estrogenic effects have been observed (Henry and Fair 2013; Kraugerud et al. 2011). For example, in H295R cells, PFOA and PFOS increased E1 levels (Behr et al. 2018), while in vitro, some evidence suggests that PFOS can suppress biosynthesis of E2 by reducing acetylation of steroidogenic acute regulatory protein (StAR) (Feng et al. 2015). In the only other epidemiological study evaluating PFAS and estrogens across pregnancy (n=557, Tangshan City, China), Yang and colleagues reported an inverse association between first trimester PFDA and E1 in each trimester (Yang et al. 2022). Our results indicating higher levels of E2 and E3 associated with PFHxS exposure support estrogenic effects.

Associations between PFAS and estrogen production may be further complicated by normative changes during pregnancy; indeed some studies in non-pregnant populations report anti-estrogenic effects. In a NHANES sample from 2015-2016, PFHxS, PFNA PFOS and PFOA were negatively associated with E2 in 12-19 year old girls. Barrett and colleagues also reported negative associations between PFOS and PFOSA and salivary E2 in 178 non-pregnant, reproductive age women (Barrett et al. 2015). The C8 Health Project, moreover, reported significant inverse associations between PFOS and E2 in peri-menopausal women (Knox et al. 2011). (Nian et al. 2020). Changes in estrogen synthesis during pregnancy may contribute to differences between pregnant and non-pregnant populations. During early pregnancy, estrogen production is primarily maternal in origin, whereas later in pregnancy, the placenta becomes the main site for E2 and E3 synthesis (Behr et al. 2018). This change in the sources of estrogen steroidogenic activity may account for differing results in pregnant and non-pregnant individuals.

Importantly, placental tissue derives primarily from the fetus and as a result, placental function can differ by sex of the fetus (Clifton 2010; Rosenfeld 2015). For example, sex differences in placental androgen signaling are hypothesized to contribute to increased male growth relative to female growth in utero (Meakin et al. 2019). Evidence also shows that developing males and females have different susceptibilities to endocrine disrupting chemicals (Barrett et al. 2017; Lesseur et al. 2021; Swan et al. 2015). It is known that PFAS can reach the placenta and have been detected in placental tissue (Bangma et al. 2020; Dankers et al. 2013). However, the extent to which PFAS can alter placental hormone production in a sex-dependent manner remains unclear. Additionally, we observed effect differences by individual PFAS suggesting that not all PFAS may act in a sex-dependent manner. Future studies should include other placental hormones and placental signaling pathways in relation to individual PFAS as well as PFAS mixtures.

Our study supports the hypothesis that PFAS may act as endocrine disruptors during pregnancy. Alterations in sex steroid hormones during pregnancy have been associated with inadequate fetal growth, suggesting one pathway by which gestational PFAS exposure may alter growth trajectories (Braun 2017; Costa 2016; Fowden et al. 2015). Epidemiological studies have shown that variation in maternal and placental hormone levels during pregnancy is related to pregnancy complications as well as adverse birth outcomes including growth restriction, preterm birth and low birth weight (Gilles et al. 2018; Jelliffe-Pawlowski et al. 2010; Kumar et al. 2018; Mucci et al. 2003; Noyola-Martínez et al. 2019; Wadhwa et al. 2004). From the developmental origins of health and disease (DOHaD) perspective, through disruption of the prenatal hormone milieu, PFAS may impact health trajectories through childhood and into adulthood (Day et al. 2020; Filippou and Homburg 2017; Gore et al. 2015; Parazzini et al. 2017). Future follow-up of our study population will include assessment of child growth pre- and postnatally to expand our current results.

A notable strength of our study is that to our knowledge, it is the first to examine sex steroid hormones measured serially across pregnancy in relation to PFAS exposure, which is important given the rapidly shifting endocrine environment in pregnancy. We were able to measure hormone levels in early pregnancy, a critical time of fetal development. Additionally, hormone concentrations were measured using LC/MS-MS and free testosterone was measure by equilibrium dialysis, which are gold standard for these hormone measurements. This is particularly important for androgens, which are present in very low concentrations in women; prior work employing other analytic techniques such as immunoassay may have inadequate sensitivity and may not yield accurate results (Rosner et al. 2007). We also report some limitations. PFAS exposure was assessed at a single time-point in mid-pregnancy however, PFAS are persistent chemicals with half-lives between 3 and 8.5 years and we expect relative stability in exposure across pregnancy (Buck et al. 2011). Additionally, others have reported relatively high inter class correlation coefficients for PFAS measures between two pregnancy related measurements (Wang et al. 2019; Yang et al. 2022). Pregnancy is also characterized by metabolic changes including variation in glomerular filtration rate and plasma volume expansion, which may impact PFAS concentrations in a particular time point (Cheung and Lafayette 2013) as well as maternal hormone level. Additionally, hormone levels in maternal serum do not provide information on which of the sources of prenatal hormones (i.e., maternal, fetal, placental) is being impacted by PFAS exposure, however our sex-dependent results suggest fetal and/or placental hormone production may be affected. Finally, our cohort is relatively small and results need to be confirmed in larger and more diverse populations.

5.0. CONCLUSION

Our results suggest that PFHxS, PFNA, PFDA may be sex-specific endocrine disruptors with androgenic effects in pregnancy. Endocrine pathways including alterations in sex-steroid hormones may underlie well-documented associations between PFAS and fetal growth. Although PFOA and PFOS were detected in the highest concentrations, no significant changes in sex steroid hormone levels were observed in relation to these legacy PFAS. Compared to other studies, this may be due to the lower exposure of these legacy PFAS in our study population, difference in biomarker (cord blood vs. maternal serum) and timing. This is the first study on maternal sex steroid hormones across pregnancy in relation to PFAS in a U.S. population and additional studies examining the potential for PFAS to impact the prenatal sex steroid hormonal milieu is warranted.

• Detectable levels were very low for shorter chain PFAS including PFBS and PFHpA

• PFOA and PFOS had minimal impact in maternal sex steroid hormones

• PFHxS, PFNA, and PFDA may exert androgenic effects

• Associations between PFAS and maternal sex steroid hormones differed by fetal sex