Urinary phthalate metabolites and ovarian reserve among women seeking infertility care

Abstract

STUDY QUESTION

Are urinary phthalate metabolites associated with reduced antral follicle growth among women in an infertility setting?

SUMMARY ANSWER

Higher urinary concentrations of di(2-ethylhexyl) phthalate (DEHP) metabolites were associated with significant decreases in antral follicle count (AFC) among women seeking infertility care.

WHAT IS KNOWN ALREADY

Experimental animal studies show that DEHP accelerates primordial follicle recruitment and inhibits antral follicle growth. Whether phthalates also reduce the growing antral follicle pool in humans remains unknown.

STUDY DESIGN, SIZE, DURATION

We examined the association between urinary phthalate metabolites and AFC using prospective data from 215 females recruited between 2004 and 2012 in the Environment and Reproductive Health (EARTH) study.

PARTICIPANTS/MATERIALS, SETTING, METHODS

We quantified the urinary concentrations of 11 phthalate metabolites. We estimated the geometric mean for all urine samples provided prior to unstimulated day 3 AFC assessment for each woman. We evaluated the association of AFC with ∑DEHP (molar sum of four DEHP metabolites) and individual phthalate metabolites using Poisson regression, adjusting for age, BMI and smoking.

MAIN RESULTS AND THE ROLE OF CHANCE

We observed significant decreases in mean AFC for all higher quartiles of ∑DEHP as compared with the lowest quartile. Compared with women in the first quartile of ∑DEHP, women in the second, third and fourth quartiles had a −24% (95% confidence interval (CI): −32%, −16%), −19% (95% CI: −27%, −9%), and −14% (95% CI: −23%, −5%) decrease in mean AFC. The absolute mean AFC in the first quartile was 14.2 follicles (95% CI: 13.2, 15.2) compared with 10.7 follicles (95% CI: 9.9, 11.6) in the second quartile. We observed similar trends among the four individual DEHP metabolites. There was no consistent change in AFC among the remaining phthalate metabolite concentrations evaluated.

LIMITATIONS, REASONS FOR CAUTION

We demonstrated a negative association between DEHP and a well-established marker of ovarian reserve among a subfertile population. However these findings may not be generalizable to women without fertility concerns, and we cannot rule out co-exposure to other chemicals.

WIDER IMPLICATIONS OF THE FINDINGS

Environmental chemicals that inhibit the size of the growing antral follicle pool can impair fertility and reduce fecundity. This study suggests evidence in need of further investigation on the impact of phthalates on the human oocyte and follicular development.

STUDY FUNDING/COMPETING INTERESTS

Work supported by grants ES009718, ES022955, ES000002, and T32ES007069 from the National Institute of Environmental Health Sciences (NIEHS) and grant T32 DK007703-16 from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). C.M. was supported by a post-doctoral training award from the Canadian Institutes of Health Research. There are no competing interests to declare.

Introduction

Environmental chemicals that interfere with hormonal signaling may disrupt the delicate hormonal balance necessary for healthy reproduction. Epidemiologic evidence now links endocrine disrupting chemicals to adverse reproductive and developmental outcomes (Hauser and Calafat, 2005; Woodruff et al., 2008; Diamanti-Kandarakis et al., 2009; Ehrlich et al., 2012; Toft et al., 2012; Birnbaum, 2013; Swan et al., 2015; Trasande et al., 2015). There is widespread concern over a specific group of chemicals known as phthalates to which most of the general population is exposed. Several phthalates—such as di-n-butyl phthalate (DBP) and di(2-ethylhexyl) phthalate (DEHP)—are established endocrine disruptors and reproductive toxicants in female and male animals (Lovekamp-Swan and Davis, 2003; Gray et al., 2006; Howdeshell et al., 2008; Craig et al., 2014). DEHP and its metabolite mono(2-ethylhexyl) phthalate (MEHP) have been shown to disrupt ovarian functioning and inhibit growth of antral follicles experimentally (Lovekamp-Swan and Davis, 2003; Craig et al., 2014; Hannon et al., 2014).

Primarily used to impart flexibility and durability, phthalates have diverse applications in the manufacturing of plastic. They can be found in a wide variety of products ranging from vinyl tiles and flooring to adhesives, detergents, lubricants, medical devices, pharmaceuticals (in the coating of certain oral medications), clothing, food packing, and toys. As a solubilizing agent, phthalates are also used in the preparation of cosmetics and personal care products (Centres for Disease Control and Prevention, 2009). Such widespread use has led to near universal human exposure (Hauser and Calafat, 2005; Centres for Disease Control and Prevention, 2009). Once ingested, inhaled or absorbed, phthalates have a short half-life, undergoing rapid hydrolysis into bio-active monoesters, some of which may then be further metabolized by oxidation or phase II conjugation. The metabolites are excreted mainly in urine and feces (Wittassek and Angerer, 2008). More than 95% of US and Canadian populations have detectable levels of one or more phthalate metabolites in urine (Saravanabhavan et al., 2013; Zota et al., 2014). Studies show that the fetus is most sensitive to potential adverse effects, and biomonitoring studies have found the highest concentration of many urinary phthalate metabolites in women and children (Hauser and Calafat, 2005; Hogberg et al., 2008; Wittassek and Angerer, 2008; Centres for Disease Control and Prevention, 2009; Trasande et al., 2013).

In experimental animal studies on female reproductive toxicity, the ovary is a suspected target for DEHP. An early in vivo study by Davis et al. (1994) showed that DEHP decreased estradiol (E₂) production, prolonged estrous cycles, and delayed ovulation in cycling rats (Davis et al., 1994). Gupta and colleagues hypothesized that the antral follicle was the target site of DEHP, and showed that in mice, DEHP and MEHP inhibited the growth of these follicles through reduced E₂ production (Gupta et al., 2010). More recently, Hannon and colleagues (2014) expanded on these finding by demonstrating that exposure of mice to environmentally low levels of DEHP—levels within the range of human exposure—resulted in changes to estrous cycles and accelerated primordial follicle recruitment (Hannon et al., 2014). They furthermore provided evidence of a mechanism by which DEHP may accelerate early folliculogenesis by interfering with cell cycle signaling through altered gene expression. They were the first to report a non-monotonic dose–response for DEHP on the ovary, suggesting that altered ovarian functioning may not be explicitly due to high toxicity levels (Hannon et al., 2014).

Whether phthalates also reduce the growing antral follicle pool in humans remains unknown. While there is substantive experimental evidence linking phthalates to altered ovarian function, very little is known about its impact on the human ovary. We have previously examined the relationship between antral follicle count (AFC) and bisphenol A (Souter et al., 2013). Other researchers have investigated the effect of phthalates on human luteal cell function in vitro (Romani et al., 2014). However, we are not aware of any published epidemiological studies investigating the potential effects of phthalates exposures on the growing antral follicle pool. Our primary objective was therefore to examine the prospective association between eleven urinary phthalate metabolites and the AFC among women seeking infertility care.

Materials and Methods

Study participants

The Environmental and Reproductive Health (EARTH) study is a prospective cohort of couples seeking infertility investigation and treatment at the Massachusetts General Hospital (MGH) Fertility Center and is designed to evaluate the effects of diet and environmental exposures on fertility and pregnancy outcomes. Details of the study cohort have been described previously (Ehrlich et al., 2012). The EARTH study has been ongoing since 2004 and has recruited approximately 700 women and 400 men to date. Women between the ages of 18 and 46 years were eligible to participate and were followed from time of entry, throughout their infertility care and eventual pregnancy. The present study included all women enrolled in EARTH between November 2004 and April 2012 with valid ultrasound data (N = 612). We excluded a priori: women with oophorectomies (n = 6), ‘difficult to visualize the ovary’ scans (n = 6), and women with polycystic ovaries (n = 36) in whom an AFC value was not entered for either the left or right ovary (or both) due to excessively high counts, leaving 564 eligible women. Among these, we included only women with at least one urine sample collected prior to AFC determination (n = 215), to ensure temporality of exposure relative to outcome. The final study cohort for statistical analysis consisted of 215 women with results for urinary concentrations of phthalate metabolites and one AFC measurement per participant. The study was approved by the Human Studies Institutional Review Boards of MGH, Harvard T. H. Chan School of Public Health, and the Centers for Disease Control and Prevention (CDC). Prior to signing informed consent, a trained research nurse explained all procedures and answered questions.

Urinary phthalate metabolite concentrations

EARTH participants provided a urine sample at study entry, and twice during each subsequent treatment cycle, corresponding to Days 3–9 of the monitoring phase of their cycle and again at the time of oocyte retrieval or intrauterine insemination. All urine samples collected prior to the AFC scan date (ranging from 1 to 11 urine samples per woman) were included in the analysis. Urine samples were collected using a sterile phthalate-free polypropylene cup. Each urine sample was analyzed for specific gravity (SG) with a handheld refractometer (National Instrument Company, Inc., Baltimore, MD, USA) divided into aliquots and frozen at −80°C. Samples were shipped on dry ice overnight to the CDC (Atlanta, GA, USA) for quantification of urinary phthalate metabolite concentrations using solid phase extraction coupled with high performance liquid chromatography-isotope dilution tandem mass spectrometry (Silva et al. 2007). The eleven phthalate metabolites were: mono(2-ethylhexyl) phthalate (MEHP), mono(2-ethyl-5-hydroxyhexyl) phthalate (MEHHP), mono(2-ethyl-5-oxohexyl) phthalate (MEOHP), mono(2-ethyl-5-carboxypentyl) phthalate (MECPP), mono(3-carboxypropyl) phthalate (MCPP), monocarboxyisooctyl phthalate (MCOP), monocarboxyisononyl phthalate (MCNP), monobenzyl phthalate (MBzP), monoethyl phthalate (MEP), mono-isobutyl phthalate (MiBP) and mono-n-butyl phthalate (MBP). The limits of detection (LOD) were 0.5–1.2 µg/l (MEHP), 0.2–0.7 µg/l (MEHHP and MEOHP), 0.2–0.6 µg/l (MECPP), 0.1–0.2 µg/l (MCPP), 0.2–0.7 µg/l (MCOP), 0.2–0.6 µg/l (MCNP), 0.2–0.3 µg/l (MBzP), 0.4–0.8 µg/l (MEP), 0.4–0.6 µg/l (MBP), and 0.2–0.3 µg/l (MiBP). We calculated the molar sum of DEHP metabolites (∑DEHP) by dividing each metabolite concentration by its molecular weight and then summing: [(MEHP*(1/278.34)) + (MEHHP*(1/294.34)) + (MEOHP*(1/292.33)) + (MECPP*(1/308.33))]. Values below LOD were assigned the value of LOD divided by the square root of 2 (Hornung and Reed, 1990).

Infertility data and AFC measurement

Routine infertility evaluation at MGH includes the determination of the ovarian AFC through transvaginal ultrasonography on Day 3 of an unstimulated cycle, and measurement of FSH and E₂ concentrations in serum. AFC was defined as the sum of antral follicles in both ovaries as measured by transvaginal ultrasound in the early follicular phase of a cycle (American Society for Reproductive Medicine Practice Committee, 2015). An infertility diagnosis was assigned by the treating infertility physician using the Society for Assisted Reproductive Technology definitions. Other pertinent demographic and clinical information, such as age and race, were obtained from a baseline questionnaire, and parity and gravida were abstracted from the patient's electronic medical records by a trained study staff. Age of participant was collected at time of study enrollment. Height and weight were measured at enrollment by the study nurse. BMI was estimated as weight (kilograms) divided by height (meters) squared.

Statistical analysis

We calculated the geometric mean of each participant's SG-adjusted urinary phthalate metabolite concentrations including all urine samples provided prior to AFC assessment. Urinary phthalate metabolite concentrations were adjusted for urinary dilution by multiplying the metabolite concentration by [(1.015-1)/(SG-1)], where 1.015 is the mean SG level for all included study urine samples, and SG is the specific gravity of the participant's urine sample (Boeniger et al., 1993). We used the geometric mean for each of the phthalate metabolites and for participant's summary estimate of ∑DEHP.

We examined the clinical and demographic characteristics of study participants in the total cohort and by quartiles of ∑DEHP concentration, reported as means (±SD) or number of women (%). We fit multivariable generalized linear models with a Poisson distribution and log-link function to estimate the mean AFC and 95% confidence interval (CI) by quartile of phthalate metabolite concentration. We estimated the absolute difference and percentage change using quartile one as the reference. We used the adjusted mean ratio (aMR) and corresponding CIs from the parameter estimates to calculate percentage change as: [(aMR − 1)*100]. Statistical tests for trend were conducted across quartiles using the median urinary phthalate metabolite concentration in each quartile in the regression models. We also modeled urinary phthalate metabolite concentrations using the geometric mean as a continuous variable on the log-scale, adjusting for covariates, and assessed the overall linear trend with the outcome. We selected a priori a minimum set of covariates, based on substantive knowledge, and included maternal age (continuous), BMI (continuous), and smoking status (never smoked versus ever smoked, defined as a current or former smoker) in adjusted models. In order to explore potential effect modification by age, we included interaction terms in models using the median level of metabolite concentration*age (centered), and further stratified the primary analyses by age into younger women <37 (55%) and older women ≥37 years of age. All tests were two-tailed and the level of statistical significance was set at 0.05. Statistical analyses were performed using SAS v9.4 software (SAS Institute, Inc., Cary, NC, USA).

Sensitivity analysis

As we wanted to examine whether DEHP or its metabolites predicted a potentially relevant clinical marker of diminished ovarian reserve, we dichotomized AFC as low at ≤8 total count compared with an AFC >8. We estimated adjusted risk ratios and 95% CIs for the risk of low AFC across quartiles of ∑DEHP and individual DEHP metabolites with quartile one as the reference using a log-binomial model, in the total sample and stratified by age (<37 versus ≥37 years).

There were 31 participants for whom exposure to ovarian stimulation in the cycle preceding AFC measurement could not be ruled out. In order to assess whether this could have inadvertently biased the results, we conducted a sensitivity analysis by repeating analyses excluding these women. We furthermore assessed the correlation between ∑DEHP and FSH and E₂ levels and examined differences in proportion of infertility diagnosis by quartiles of ∑DEHP.

Results

The study cohort comprised 215 women, predominantly Caucasian (81%) and never-smokers (74%), with an average age of 35.7 years at time of enrollment (Table I). Most women were nulliparous (88%), and 42% had a female factor as the primary cause of infertility. The distribution of the SG-adjusted urinary phthalate metabolite concentrations measured from 471 urine samples provided by 215 women is shown in Table II. The percentage of urine samples with detectable concentrations of phthalate metabolites ranged from 72% for MEHP to 100% for MEP and MECPP. The geometric mean of ∑DEHP did not differ by number of urine samples collected (Supplementary Table SI).

Table I.

Characteristics and primary Society for Assisted Reproductive Technology (SART) diagnosis among 215 women with antral follicle count (AFC) measurement enrolled in the Environment and Reproductive Health (EARTH) study between 2004 and 2012.

Characteristic	Total cohort (N = 215)
	N (%)
Age at study entry (years)
Mean ± SD	35.7 ± 4.6
Range	20–45
Age ≥37	97 (45%)
BMI (kg/m2)
Mean ± SD	25.0 ± 5.0
Range	16.1–40.5
Underweight or normal (<25)	128 (60%)
Overweight or obese (>=25)	87 (40%)
Smoking
Never smoked	159 (74%)
Ever smoked
Current smoker	11 (5%)
Former smoker	45 (21%)
Race
Caucasian	175 (81%)
Black/African American	7 (3%)
Asian	16 (7%)
Other	17 (8%)
Nulligravida	143 (67%)
Nullipara	190 (88%)
AFC
Mean ± SD	12.3 ± 6.9
Range	2–40
FSH IU/l1
Mean ± SD	7.5 ± 2.9
Range	1.0–23.8
Serum day 3 E2 pmol/l2
Mean ± SD	47.2 ± 28.9
Range	14.0–297
Trying to conceive3 ≥24 months	55(26%)
Primary SART diagnosis at study entry
Female factor	90 (42%)
Diminished ovarian reserve	25 (12%)
Ovulatory dysfunction	31 (14%)
Endometriosis	18 (8%)
Uterine factor	4 (2%)
Tubal factor	12 (6%)
Male factor	49 (23%)
Unexplained	75 (35%)
Year at study entry
2004–2006	45 (21%)
2007–2009	111 (52%)
2010–2012	59 (27%)

¹FSH measured in 214 women.

²Serum estradiol (E₂) levels measured in 211 women.

³Number of months trying to conceive was self-reported at time of study entry.

Table II.

Distribution of urinary phthalate metabolite concentrations (metabolite or molar sum) measured among 215 women with AFC measurements and 471 urine samples, enrolled in EARTH between 2004 and 2012.

Urine sample specific				AFC sample specific (N = 215) (µg/l)
Metabolite	Sample size	LOD (µg/l)	% Detect2	SG-adjusted GM1 (GSD)	SG-adjusted Median	25th, 75th
MEP	471	0.4–0.8	100	67.6 (5.9)	54.2	27.6, 139
MBP	471	0.4–0.6	97	12.5 (0.76)	12.8	7.4, 22.5
MiBP	471	0.2–0.3	97	6.3 (0.39)	6.8	3.6, 10.3
MBzP	471	0.2–0.3	93	3.3 (0.24)	3.2	1.7, 6.4
MEHP	471	0.5–1.2	72	3.8 (0.28)	3.5	1.6, 6.7
MEHHP	471	0.2–0.7	99	18.8 (1.6)	17.9	8.2, 41.1
MEOHP	471	0.2–0.7	99	12.1 (0.99)	11.2	5.1, 25.0
MECPP	471	0.2–0.6	100	31.6 (2.4)	29.6	13.5, 59.1
∑DEHP3	–	–	–	0.23 (0.02)	0.21	0.10, 0.46
MCPP	471	0.1–0.2	96	3.1 (0.20)	2.7	1.7, 5.6
MCOP	390	0.2–0.7	96	15.6 (1.5)	14.2	5.8, 42.8
MCNP	390	0.2–0.6	93	3.9 (0.27)	3.5	2.1, 6.5

N: number of urinary samples; LOD: limit of detection; SG: specific gravity; GM: Geometric Mean; GSD: Geometric Standard Deviation; MEP: monoethyl phthalate; MBP: mono-n-butyl phthalate; MiBP: mono-isobutyl phthalate; MBzP: monobenzyl phthalate; MEHP: mono(2-ethylhexyl) phthalate; MEHHP: mono(2-ethyl-5-hydroxyhexyl) phthalate; MEOHP: mono(2-ethyl-5-oxohexyl) phthalate; MECPP: mono(2-ethyl-5-carboxypentyl) phthalate; DEHP: di(2-ethylhexyl) phthalate; MCPP: mono(3-carboxypropyl) phthalate; MCOP: monocarboxyisooctyl phthalate; MCNP: monocarboxyisononyl phthalate.

¹Geometric mean of the specific gravity-adjusted phthalate metabolite concentrations by participant (N = 215) was calculated. For MCOP and MCNP, N = 193.

²Percentage of phthalate metabolite concentrations above the detection limit.

³∑DEHP: Molar sum of DEHP metabolites (MEHP, MEHHP, MEOHP and MECPP) expressed in µmol/l.

In the Poisson regression models adjusted for age, BMI and smoking status, we observed a decrease in mean AFC across quartiles of ∑DEHP with an association that reached a plateau after the second quartile (Table III). Compared with women in the first quartile of ∑DEHP, women in the second, third and fourth quartiles had a −24% (95% CI: −32%, −16%), −19% (95% CI: −27%, −9%), and −14% (95% CI: −23%, −5%) decrease in mean AFC. The absolute mean AFC in the first quartile was 14.2 follicles (95% CI: 13.2, 15.2) compared with 10.7 follicles (95% CI: 9.9, 11.6) in the second quartile. Similar trends were observed among the four individual DEHP metabolites with a decrease in mean AFC as quartiles increased with a plateau effect seen after the third quartile of MEHP, MEHHP, and MEOHP, and after the second quartile of MECPP. A maximum decrease of −15% (95% CI: −23, −5), −28% (95% CI: −36, −20) and −13% (95% CI: −22, −3) was observed for MEHP quartiles two, three and four, respectively, compared with quartile one. While trend tests by quartile for dose–response were not significant, the log-linear relationships of the geometric mean for ∑DEHP and individual DEHP metabolites in continuous models were significant (P < 0.05). With the exception of MBP that exhibited more modest decreases—ranging from −5% in the fourth quartile to −16% in the third quartile—there was no consistent change in mean AFC among the remaining six phthalate metabolites evaluated.

Table III.

Mean AFC by quartiles of urinary phthalate metabolite concentrations, adjusted for age, BMI, and smoking among 215 women with 215 AFC measurements in EARTH.

Metabolite	Mean AFC (95% CI)1	Difference in mean AFC	% Change (95% CI)	P-value2
∑DEHP (µmol/l)
Q1 (0.02–0.09)	14.2 (13.2, 15.2)	0 (ref)	–	–
Q2 (0.10–0.21)	10.7 (9.9, 11.6)	−3.5	−24% (−32, −16)	<0.001
Q3 (0.21–0.44)	11.5 (10.7, 12.5)	−2.7	−19% (−27, −9)	0.002
Q4 (0.46–18.88)	12.1 (11.2, 13.1)	−2.1	−14% (−23, −5)	0.004
P-trend3				0.34
MEHP (µg/l)
Q1 (0.21–1.62)	14.1 (13.1, 15.2)	0 (ref)	–	–
Q2 (1.63–3.50)	12.0 (11.1, 13.0)	−2.1	−15% (−23, −5)	0.003
Q3 (3.53–6.53)	10.2 (9.3, 11.0)	−3.9	−28% (−36, −20)	<0.001
Q4 (6.71–265)	12.3 (11.4, 13.2)	−1.8	−13% (−22, −3)	0.009
P-trend				0.33
MEHHP (µg/l)
Q1 (1.00–7.50)	13.4 (12.5, 14.5)	0 (ref)	–	–
Q2 (8.21–17.70)	11.7 (10.8, 12.6)	−1.7	−13% (−22, −3)	0.01
Q3 (17.90–40.57)	10.9 (10.1, 11.9)	−2.5	−19% (−27, −9)	0.0002
Q4 (41.10–1977)	12.5 (11.6, 13.5)	−0.9	−7% (−16, 4)	0.20
P-trend				0.92
MEOHP (µg/l)
Q1 (0.94–5.10)	13.9 (12.9, 15.0)	0 (ref)	–	–
Q2 (5.14–11.12)	11.5 (10.7, 12.5)	−2.4	−17% (−26, −7)	0.0009
Q3 (11.22–24.90)	11.4 (10.5, 12.3)	−2.5	−18% (−27, −9)	0.0003
Q4 (25.03–1278)	11.7 (10.9, 12.7)	−2.2	−16% (−24, −6)	0.002
P-trend				0.06
MECPP (µg/l)
Q1 (3.17–13.40)	14.0 (13.0, 15.1)	0 (ref)	–	–
Q2 (13.50–29.34)	10.9 (10.1, 11.8)	−3.1	−22% (−30, −13)	<0.001
Q3 (29.63–58.45)	12.0 (11.1, 12.9)	−2.0	−14% (−23, −5)	0.004
Q4 (59.10–2109)	11.7 (10.8, 12.6)	−2.3	−17% (−25, −7)	0.001
P-trend				0.08
MEP(µg/l)
Q1 (3.43–25.20)	12.6 (11.7, 13.6)	0 (ref)	–	–
Q2 (27.60–54.02)	12.5 (11.6, 13.5)	−0.1	−1% (−11, 10)	0.81
Q3 (54.17–138.85)	11.3 (10.4, 12.2)	−1.3	−11% (−20, 0)	0.04
Q4 (139.11–5995)	12.1 (11.2, 13.1)	−0.5	−4% (−14, 7)	0.47
P-trend				0.68
MBP (µg/l)
Q1 (0.60–7.35)	13.1 (12.2, 14.2)	0 (ref)	–	–
Q2 (7.43–12.75)	11.9 (11.0, 12.8)	−1.2	−10% (−19, 1)	0.07
Q3 (12.79–22.14)	11.0 (10.1, 12.0)	−2.1	−16% (−25, −6)	0.002
Q4 (22.50–140.20)	12.5 (11.6, 13.5)	−0.6	−5% (−15, 6)	0.34
P-trend				0.57
MCPP (µg/l)
Q1 (0.30–1.70)	12.0 (11.1, 13.0)	0 (ref)	–	–
Q2 (1.71–2.65)	11.7 (10.8, 12.7)	−0.3	−2% (−13, 9)	0.67
Q3 (2.66–5.50)	12.0 (11.1, 13.0)	0	0% (−10, 12)	0.94
Q4 (5.56–61.70)	12.8 (11.9, 13.8)	0.8	7% (−4, 19)	0.23
P-trend				0.11
MiBP (µg/l)
Q1 (0.50–3.50)	12.7 (11.8, 13.7)	0 (ref)	--	-
Q2 (3.61–6.75)	12.8 (11.9, 13.8)	0.1	1% (−10, 12)	0.92
Q3 (6.75–10.21)	11.1 (10.3, 12.1)	−1.6	−13% (−22, −2)	0.02
Q4 (10.28–90.0)	11.9 (11.0, 12.9)	−0.8	−6% (−16, 4)	0.23
P-trend				0.12
MBzP (µg/l)
Q1 (0.22–1.63)	12.1 (11.1, 13.0)	0 (ref)	–	–
Q2 (1.71–3.24)	12.0 (11.1, 13.0)	−0.1	0% (−11, 11)	0.93
Q3 (3.30–6.28)	12.6 (11.7, 13.6)	0.5	4% (−6, 17)	0.44
Q4 (6.39–75.19)	11.9 (11.0, 12.9)	−0.2	−1% (−12, 10)	0.81
P-trend				0.80
MCOP (µg/l)
Q1 (0.92–5.56)	12.3 (11.4, 13.4)	0 (ref)	–	–
Q2 (5.80–14.19)	11.7 (10.8, 12.7)	−0.6	−5% (−16, 6)	0.36
Q3 (14.23–42.80)	12.3 (11.4, 13.3)	0	0% (−11, 12)	0.96
Q4 (43.22–520.0)	12.8 (11.8, 13.9)	0.5	4% (−7, 16)	0.52
P-trend				0.23
MCNP (µg/l)
Q1 (0.51–2.06)	12.0 (11.0, 13.0)	0 (ref)	–	–
Q2 (2.07–3.53)	11.6 (10.7, 12.7)	−0.4	−3% (−14, 9)	0.62
Q3 (3.53–6.50)	13.0 (12.0, 14.0)	1.0	8% (−3, 22)	0.16
Q4 (6.75–160.5)	12.6 (11.6, 13.6)	0.6	5% (−6, 18)	0.41
P-trend				0.35

¹Poisson regression models adjusted for maternal age (continuous), BMI (continuous) and smoking status (never, ever).

²P-value is for test of significant difference between the quartile of interest and quartile 1.

³Tests for linear trend were performed using the median level of urinary phthalate metabolite in each quartile in the regression model, adjusting for covariates.

In the age-stratified analysis, we observed somewhat larger decreases in mean AFC among women younger than 37 years of age, than among women 37 years or older (Supplementary Table SII). While there was less evidence of effect modification by age within quartiles of ∑DEHP, we noted more pronounced decreases, both in absolute mean AFC and percentage change for MEHP, MEHHP and MEOHP among women <37 years of age, although interaction terms were not significant. Among younger women, on average those in the second, third and fourth quartile of MEHP had a decrease of −20% (95% CI: −31, −8), −32% (95% CI: −41, −22) and −17% (95% CI: −27, −4), respectively compared with the first quartile. These percentage changes were larger than what was observed among women ≥37 years, where the corresponding differences were: −9% (95% CI: −22, 7), −21% (95% CI: −34, −5) and −8% (95% CI: −22, 10), respectively.

Sensitivity analysis

The risk of lower (or decreased) ovarian reserve, defined as an AFC of ≤8, was modestly increased across higher quartiles of ∑DEHP compared with quartile one in the total sample, although CIs included the null in quartiles three and four. Similar trends of increasing risk across quartiles two and three were observed for the four DEHP metabolites (Supplementary Table SIII). Within the age-stratified analysis, we observed potential evidence of effect modification by age, with stronger associations between risk of AFC ≤8 and quartiles of DEHP metabolites among women <37 years of age, with risk ratios ranging from 1.5 to 4.3 compared with quartile one. Women ≥37 years with higher urinary ∑DEHP or individual DEHP metabolite concentrations appeared not to be at increased risk of low ovarian reserve, with risk ratios generally close to 1 (Supplementary Table SIII).

Excluding the 31 women that may have been exposed to ovarian stimulation in the cycle preceding AFC assessment did not materially change the results of the primary analysis, nor were there any appreciable differences in the study characteristics (data not shown). We also computed the correlation between the geometric mean of ∑DEHP by both FSH and E₂ levels. These were −0.02 and −0.03, respectively. There was furthermore no difference in the distributions of either hormone across quartiles of ∑DEHP (chi-square P-value 0.73 and 0.84, respectively). We did not observe a significant difference in the type of infertility diagnosis across quartile of ∑DEHP (chi-square P-value 0.40).

Discussion

As far as we are aware, this is the first human study to show a lower number of growing antral follicles across increasing quartiles of urinary DEHP metabolite concentrations among women seeking infertility treatment. The association was non-linear with a similar decrease in AFC for quartiles 2 through 4 for ∑DEHP and MECPP, and a plateau after quartile 3 for MEHP, MEHHP, and MEOHP. While reduced AFC with higher quartiles of urinary DEHP metabolites was observed in the total study cohort, these decreases were more pronounced among women <37 years of age compared with ≥37. We further observed that the risk of decreased ovarian reserve, defined as an AFC ≤8, was modestly increased across quartiles of ∑DEHP and DEHP metabolites compared with quartile one in the total cohort. These associations were also stronger among women <37 years of age, further suggesting that age may modify the effect of DEHP on AFC, despite interaction terms (P > 0.05).

Our results suggest that DEHP metabolites may adversely impact the size of the growing antral follicle pool among women in this study cohort. Younger women appear to be at higher risk of the potential effects of DEHP exposure. While our study was not designed to elucidate the mechanism through which exposure to phthalates adversely impact follicular development, our results are consistent with evidence from laboratory animal studies (Gupta et al., 2010; Craig et al., 2014; Hannon et al., 2014). However, whether this effect does indeed manifest from a diminished primordial follicle population, a more direct marker of ovarian reserve and ovarian aging, cannot be determined within the parameters of the present study.

While our stratified analysis showed that younger women had larger decreases in mean AFC and this group may be at particular risk of the deleterious effects of DEHP, most of the results from our analysis using AFC as a binary outcome did not reach statistical significance. Although the study was sufficiently powered for the main and stratified analysis using counts, our subgroup analyses looking at low AFC defined as a binary outcome within levels of age had more limited power. This lack of precision, evident in the widened CIs, precluded us from drawing firm conclusions. Nevertheless, the overall trends observed in this sensitivity analysis were consistent with those in the primary analysis. Moreover, it seems intuitive, that among women ≥37 years of age, we would not readily observe the influence of an environmental factor exerting an effect on ovarian parameters given that age would be the most important predictor and determinant of reduced AFC among older women. However, among younger women, environmental chemicals may indeed play a role and act as a putative risk factor, having an effect comparable to what may be expected through the normal aging process. We also cannot preclude the possibility that in a larger sample, the observed effect among older women may become significant.

Understanding the effect of phthalates on ovarian function has significant implications for human fertility. The ovary's primary functions are folliculogenesis and estrogen synthesis, both critical processes to successful reproduction (Hannon et al., 2014). As the growing antral follicular pool is dependent upon primordial follicle development, the potential to accelerate the depletion of the primary follicle can result in reduced ovarian reserve, premature ovarian aging or infertility. Recently, Hannon and colleagues demonstrated in an in vivo mouse model that low—environmentally relevant—levels of DEHP altered estrous cycling in mice and accelerated early folliculogenesis by interfering with P13K signaling pathways, through altered gene expression or protein levels, in a non-monotonic dose–response (Hannon et al., 2014). P13 signaling is considered critical to primordial follicle survival and activation (John et al., 2008; Reddy et al., 2009; Zheng et al., 2012). Earlier animal studies used high doses of DEHP (up to 2 g/kg), resulting in decreased serum E₂ production, inhibition of antral follicle growth, and subsequent anovulation and estrous cycling disruption (Lovekamp-Swan and Davis, 2003; Howdeshell et al., 2008; Ehrlich et al., 2012). However, the work by Hannon and colleagues has since enhanced our understanding of the effects of DEHP by illustrating that short duration and low levels of exposure result in abnormalities in ovarian functioning. Consequently overt toxicity of very high doses may not be an essential component to the deleterious effects on the ovary.

Consistent with the substantial body of evidence from animal studies, more human studies are emerging linking phthalates and other short-lived chemicals with altered male and female reproductive end-points. Hauser et al. recently reported that urinary concentrations of DEHP metabolites were inversely associated with oocyte yield, clinical pregnancy and live births rates following assisted reproduction (Hauser et al., in press). Other studies have shown that phthalates are associated with increased risk of miscarriage (Toft et al., 2012), preterm birth (Ferguson et al., 2014) and reduced anogenital distance (Swan et al., 2015). A study demonstrating reduced couple fecundity through male-mediated exposure underlines the importance of studying the joint effect of a couple's exposure on relevant fertility outcomes (Buck Louis et al., 2014).

Our study provides evidence that the human female ovary may be adversely affected by DEHP exposure. The prospective nature of this design, relying upon an infertile study population from a large academic fertility setting, permitted a careful examination of the direction of the relationship between phthalate metabolite concentrations and the size of the antral follicle pool assessed in the early phase of the menstrual cycle. The urinary concentrations of the phthalate metabolites measured are within the ranges reported for the US general population (Centres for Disease Control and Prevention, 2009). Despite the lack of significant test for a linear dose–response when using quartiles, we observed significant tests of linear trend for DEHP metabolites when assessing the geometric mean concentration as a continuous variable in linear models with the outcome. However, these findings may not be generalizable to women from the general population without fertility concerns, co-exposures to other select chemicals were also not accounted for, exposure to phthalates may be reflective of other unknown lifestyle factors that might influence ovulatory potential, and findings are based on a relatively small cohort. This study design also does not permit an examination of the potential impact of phthalates on polycystic ovaries. Moreover, we lacked information on anti-Mullerian hormone; however, AFC is considered to be the most well-established marker of ovarian reserve available (Hansen et al., 2011; American Society for Reproductive Medicine Practice Committee, 2015). Furthermore, in the body, phthalates are short-lived chemicals, making the assessment of long-term exposure difficult. We attempted to partially account for this by collecting multiple urine samples that preceded the ovarian ultrasound evaluation, with ∼30% of the women having three or more urine samples, and up to 13% having more than five samples (Supplementary Table SI). Multiple phthalate metabolites were also evaluated at the same time to account for various sources of exposure, and all samples were collected in one location and processed under one protocol.

Any environmental chemical that inhibits the size of the growing antral follicle pool, and consequently ovarian reserve, can impair fertility and reduce fecundity. Considering that phthalate exposure is nearly universal, these results may have important clinical and public health relevance. Our findings represent the first human study to suggest that higher urinary concentrations of DEHP metabolites are associated with a negative change in the AFC of women seeking infertility treatment. Because our study suggests probable evidence for the impact of exposure to DEHP on the human oocyte and follicular development, additional research with a larger sample would provide useful information to further elucidate possible mechanisms, and corroborate our findings.

Supplementary data

Supplementary data are available at http://humrep.oxfordjournals.org/.

Authors' roles

I.S., R.H. and C.M. conceived and designed the study. A.M.C. conducted the chemical analysis of urine samples and produced the chemical database. Statistical analysis of the data was done by C.M. in consultation with P.W., A.G. and Y.-H.C. C.M. drafted the manuscript and all authors critically revised the manuscript for important intellectual content.

Funding

Work supported by grants ES009718, ES022955, ES000002, and T32ES007069 from the National Institute of Environmental Health Sciences (NIEHS) and grant T32 DK007703-16 from the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK). C.M. was supported by a post-doctoral training award from the Canadian Institutes of Health Research.

Conflict of interest

The authors have no conflict of interests to declare. The findings and conclusions in this report are those of the authors and do not necessarily represent the views of the Centers for Disease Control and Prevention (CDC). The use of trade names and commercial sources is for identification only and does not constitute endorsement by the US Department of Health and Human Services or CDC.