Relating Phthalate and BPA Exposure to Metabolism in Peripubescence: The Role of Exposure Timing, Sex, and Puberty

Abstract

Context:

Exposure to endocrine-disrupting chemicals during development may play a role in the increasing prevalence of metabolic syndrome and type 2 diabetes among children and adolescents by interfering with metabolic homeostasis.

Objective:

To explore associations between in utero and peripubertal urinary phthalate metabolite and bisphenol A (BPA) concentrations and markers of peripubertal metabolic homeostasis.

Design:

Early Life Exposure in Mexico to Environmental Toxicants (ELEMENT): a longitudinal cohort study of pregnant women in Mexico City and their offspring.

Setting:

Public maternity hospitals in Mexico City.

Patients or Other Participants:

Women recruited during pregnancy; offspring recruited for follow-up at age 8–14 years (n = 250).

Interventions:

None.

Main Outcome Measures:

Fasting serum c-peptide, IGF-1, leptin, and glucose concentrations among children at follow-up; calculated measures of insulin secretion and insulin resistance.

Results:

Phthalate metabolites and BPA were associated with metabolism biomarkers at age 8–14 years in patterns that varied by sex, pubertal status, and exposure timing. For example, in utero monoethyl phthalate was associated with lower insulin secretion among pubertal boys (P = .02) and higher leptin among girls (P = .04). In utero di-2-ethylhexyl phthlate was associated with higher IGF-1 among pubertal girls; peripubertal di-2-ethylhexyl phthlate was associated with higher IGF-1, insulin secretion, and resistance among prepubertal girls. In contrast, peripubertal dibutyl phthalate, monobenzyl phthalate, and mono-3-carboxypropyl phthalate were associated with lower IGF-1 among pubertal boys. Peripubertal BPA was associated with higher leptin in boys (P = .01).

Conclusions:

Considering the long-term health effects related to metabolic syndrome, additional research on exposure and metabolic outcomes across developmental periods and early adulthood is needed.

The prevalence of obesity among children and adolescents has risen drastically in recent decades, particularly in more recently industrialized countries. In Mexico, an estimated 29% of people under 20 years of age were overweight or obese in 2013, with similar prevalence in the United States (1). Obese children and adolescents are at increased risk for several health problems, including metabolic syndrome, type 2 diabetes, cardiovascular disease, and cancer later in life. Although the clinical definition of metabolic syndrome in children varies, a recent review estimated that 11% of overweight children and 29% of obese children worldwide are affected, and geography and/or ethnicity may play a role in disease risk (2). Indeed, in a study of Mexican adolescents, 62% of obese participants age 8–16 years also had metabolic syndrome (3). Increasingly sedentary lifestyles and high-calorie diets are thought to be main contributors to obesity trends, but exposure to endocrine-disrupting chemicals (EDCs) is also thought to play a role, possibly by interfering with lipid and glucose metabolic homeostasis (4).

Phthalates are known EDCs present in a range of consumer products, resulting in ubiquitous human exposure (5). In a cross-sectional study of 12–19 year olds in the 2003–2008 National Health and Nutrition Examination Survey (NHANES), urinary di-2-ethylhexyl phthalate (DEHP) metabolite concentrations were associated with an increase in homeostatic model assessment of insulin resistance (HOMA-IR), as well as increased odds of being categorized as insulin resistant (6). Similarly, a study of adult NHANES participants (2001–2008) reported that several urinary phthalate metabolites were positively associated with HOMA-IR and fasting blood glucose, with many associations stronger in blacks and Mexican Americans compared with whites (7).

Phthalate exposure has also been associated with hormonal mechanisms involved in metabolic homeostasis. A cross-sectional study of Danish children age 4–9 years reported that urinary DEHP metabolites and monocarboxyisooctyl phthalate (MCiOP) were associated with decreased IGF-1 (8), a mediator of GH that plays an essential role in maintenance of normal metabolism, growth, and development (9). Another study evaluated in utero phthalate exposure in relation to fetal metabolic function, and found an association between first trimester urinary mono-3-carboxypropyl phthalate (MCPP) and higher cord blood leptin in boys (10). Increased cord blood leptin, a hormone produced by adipocytes that acts on the hypothalamus to suppress appetite and regulate energy homeostasis (11), has been associated with increased birth weight and insulin resistance in newborns (12), whereas serum leptin in adults has been associated with increased body mass index (BMI) and insulin resistance (13, 14).

Bisphenol A (BPA) is an estrogenic high-volume chemical utilized in a number of applications, resulting in ubiquitous human exposure (5). Previous studies have not found significant cross-sectional associations between markers of BPA exposure and metabolic function in childhood (15). However, one longitudinal study found that maternal urinary BPA concentrations during pregnancy were associated with increased leptin in boys and increased adiponectin, a hormone secreted by adipose tissue that promotes insulin sensitivity, in girls at age 9 years (16). In contrast, concurrent BPA was not associated with levels of either hormone. In adults, urinary BPA has been associated with insulin resistance and increased levels of concurrently measured serum leptin and adiponectin (17, 18).

Thus far, epidemiologic studies of phthalate and BPA exposure and metabolic homeostasis have been primarily cross-sectional, limiting the ability to determine causality. In addition, investigations of in utero phthalate and BPA exposure have been sparse despite potential long-term metabolic consequences of EDC exposure during this critical period. Our objective was to examine the impact of phthalate and BPA exposure during 2 sensitive periods of development, in utero and peripubescence, on markers of metabolic homeostasis among participants in a prospective birth cohort at age 8–14 years. We investigated urinary phthalate metabolite and BPA concentrations in relation to c-peptide and IGF-1, 2 biomarkers that have not been previously assessed in relation to in utero phthalate and BPA exposure, and leptin and fasting glucose, 2 established biomarkers of energy homeostasis and metabolism.

Materials and Methods

Study population

Participants were originally recruited for the Early Life Exposure in Mexico to Environmental Toxicants project (ELEMENT), a longitudinal cohort study of pregnant women in Mexico City and their offspring. Our analysis includes women who were recruited in 1997–2004 from public maternity hospitals during their first trimester. We followed mothers throughout pregnancy, and both mothers and offspring through postnatal visits as previously described (19). During their third trimester, mothers provided a urine sample and completed interview-based questionnaires. In 2010, we recontacted a subset of their offspring, who were then 8–14 years of age, to participate in follow-up studies (n = 250). Child participants provided urine samples, serum samples, anthropometry, and completed an interview-based questionnaire. In the present analysis we included children who had phthalate metabolite and BPA measurements from their mother's urine sample (n = 219) and/or their urine sample collected at follow-up (n = 240), as well as fasting serum metabolic markers. Most children had urine samples from both in utero and peripubertal time points (n = 212). The ethics and research committees of the Mexico National Institute of Public Health and the University of Michigan approved research protocols and participants provided informed consent before enrollment.

Urinary phthalate metabolites and BPA

Spot urine samples were frozen at −80°C until analysis at NSF International. BPA and 9 phthalate metabolites (monoethyl phthalate [MEP], mono-n-butyl phthalate [MnBP], monoisobutyl phthalate [MiBP], monobenzyl phthalate [MBzP], MCPP, mono-2-ethylhexyl phthalate [MEHP], mono-2-ethyl-5-hydroxyhexyl phthalate [MEHHP], mono-2-ethyl-5-oxohexyl phthalate [MEOHP], and mono-2-ethyl-5-carboxypentyl phthalate [MECPP]) were measured in urine using isotope dilution-liquid chromatography-tandem mass spectrometry as previously described (19). We calculated a DEHP metabolite summary measure (ΣDEHP) for each sample by dividing individual MEHP, MEHHP, MEOHP, and MECPP concentrations by their molar mass and summing them. We calculated a summary dibutyl phthalate (DBP) metabolite measure (ΣDBP) as the molar sum of MnBP and MiBP. Specific gravity was measured using a handheld digital refractometer (Atago Co, Ltd) at the time of sample analysis. Values below the limit of detection (LOD) were replaced with LOD/√2.

Serum biomarkers of metabolic homeostasis

We measured c-peptide, IGF-1, leptin, and glucose in fasting serum samples collected from children at follow-up. These biomarkers provide a broad characterization of glucose and lipid metabolic homeostasis. C-peptide is a marker of insulin secretory function, because it is secreted in quantities equal to insulin from pancreatic β-cells (20), whereas fasting serum glucose is a marker of glucose metabolism and a diabetes screening tool. Serum aliquots were frozen at −80°C and sent on dry ice to the Michigan Diabetes Research and Training Center Chemistry Lab. C-peptide and IGF-1 were measured using automated chemiluminescence immunoassay (Immulite 1000). Leptin was measured using RIA (Millipore), and glucose was measured using automated enzymatic assay (Cobas Mira Plus). We calculated a c-peptide index (CPI) as an indicator of β-cell function using the next equation: [(fasting serum c-peptide/fasting serum glucose) × 100] (21), and a measure of insulin resistance similar to HOMA-IR (c-peptide-based measure of insulin resistance [CP-IR]) with the next equation: [fasting serum c-peptide × fasting serum glucose/405] (22). Values below the LOD were replaced with LOD/√2.

Puberty status

We used Tanner staging as a measure of sexual maturation, where stage 1 indicates no development and stage 5 indicates full development (23). Each child was examined according to a standardized protocol by a trained pediatrician, who assessed Tanner stages in genital (boys), breast (girls), and pubic hair (boys and girls) development. Participants were considered to have entered puberty if they had a Tanner stage more than 1 for genitalia development in boys and breast development in girls.

Statistical analysis

Urinary phthalate metabolite, BPA, and serum biomarker concentrations were log-normally distributed, and ln-transformed before regression analysis. We used linear regression to assess associations of ln-transformed urinary phthalate metabolite and BPA concentrations with ln-transformed serum biomarkers of metabolic homeostasis. Age and BMI z-score, based on World Health Organization growth standards (24), were included in models as potential confounders, whereas urinary-specific gravity was included to adjust for urine dilution. Because BMI may be on the causal pathway between exposure and serum leptin, we ran these models both with BMI z-score to investigate associations independent of BMI, and without BMI z-score to investigate associations via this pathway. Results are presented as the percent change in serum biomarkers (95% confidence interval [CI]) per interquartile range (IQR) increase in continuous urinary phthalate metabolite or BPA. We presented R² values for significant findings to indicate the proportion of outcome variance explained by models.

In sensitivity analyses, we stratified regression models by puberty status and sex to determine if associations between phthalate exposure and metabolic function were different before and during this period of hormonal and metabolic change. We also entered in utero and peripubertal phthalate metabolite and BPA concentrations into models together to investigate associations of combined exposure with metabolic function. All analyses were performed using SAS version 9.4.

Results

Demographic characteristics of the study population have been previously described (19) and are presented in Supplemental Table 1. Mean serum concentrations of c-peptide, IGF-1, leptin, and glucose were within normal range for children (Table 1) (25). Females had significantly higher concentrations of c-peptide (P = .03), IGF-1 (P = .0001), leptin (P < .0001), and higher CPI (P = .01) compared with males but similar levels of serum glucose (P = .12). BMI z-score was positively associated with leptin, whereas both age and BMI z-score were positively associated with c-peptide, CPI, and CP-IR (Supplemental Table 2). Distributions of phthalate metabolite and BPA concentrations measured during in utero development and at 8–14 years are presented in Supplemental Table 3. Spearman correlations between phthalate metabolite and BPA concentrations measured during in utero vs peripubertal development were weakly correlated (range, 0.06–0.25) (Supplemental Table 4). Concurrently measured phthalate metabolite and BPA concentrations from within time points were moderately to strongly correlated (range, 0.30–0.85) (Supplemental Table 5).

Table 1.

Distribution of Serum Measures of Metabolic Function at Age 8–14 Years by Pubertal Status

	Mean	(SD)	Percentiles
			25th	50th	75th	Max
All participants (n = 248)
C-peptide (ng/mL)	1.7a	(1.2)	1.0	1.5	2.1	10.1
IGF-1 (ng/mL)	257a	(105)	178	226	329	606
Leptin (ng/mL)	11a	(9.0)	4.9	8.1	15	62
Glucose (mg/dL)	87	(9.4)	82	87	92	146
CPI	2.0a	(1.2)	1.1	1.6	2.4	8.4
CP-IR	0.38	(0.33)	0.22	0.31	0.44	3.22
Prepubertal boys (n = 56)
C-peptide (ng/mL)	1.4b	(0.85)	0.89	1.2	1.6	6.3
IGF-1 (ng/mL)	192b	(61)	148	181	222	442
Leptin (ng/mL)	8.0	(5.9)	3.7	6.0	12	32
Glucose (mg/dL)	89	(8.2)	83	89	96	112
CPI	1.5b	(0.93)	1.0	1.3	1.8	7.0
CP-IR	0.30b	(0.20)	0.19	0.26	0.36	1.40
Pubertal boys (n = 58)
C-peptide (ng/mL)	1.8b	(1.2)	1.0	1.5	1.9	7.7
IGF-1 (ng/mL)	274b	(112)	186	232	351	568
Leptin (ng/mL)	8.8	(6.9)	3.8	7.2	11	34
Glucose (mg/dL)	87	(7.6)	83	87	92	100
CPI	2.0b	(1.3)	1.1	1.7	2.3	8.4
CP-IR	0.39b	(0.29)	0.22	0.30	0.42	1.75
Prepubertal girls (n = 86)
C-peptide (ng/mL)	1.7b	(1.4)	1.0	1.4	2.1	10.1
IGF-1 (ng/mL)	232b	(76)	176	219	272	470
Leptin (ng/mL)	12b	(9.4)	5.8	8.9	17	42
Glucose (mg/dL)	87	(11)	80	87	92	146
CPI	2.0b	(1.2)	1.1	1.6	2.3	7.8
CP-IR	0.40b	(0.45)	0.21	0.30	0.44	3.22
Pubertal girls (n = 45)
C-peptide (ng/mL)	2.1b	(0.90)	1.4	2.0	2.4	4.8
IGF-1 (ng/mL)	368b	(95)	306	368	437	606
Leptin (ng/mL)	17b	(11)	10	15	22	62
Glucose (mg/dL)	85	(8.3)	81	84	89	118
CPI	2.5b	(1.1)	1.6	2.4	3.0	5.4
CP-IR	0.4b	(0.19)	0.30	0.40	0.53	1.07

^a

Significant differences in mean serum concentrations in boys vs girls (P < .05).

^b

Differences in mean serum concentrations in prepubertal vs pubertal boys or prepubertal vs pubertal girls were statistically significant (P < .05).

We observed differences in associations between exposure and select outcomes in participants that had begun puberty compared with prepubertal participants. Associations between exposure and IGF-1, fasting glucose, CPI, and CP-IR were modified, so we have presented these results stratified by sex and pubertal status. Associations of exposure with c-peptide and leptin did not differ by pubertal status so we stratified these model results only by sex. Because phthalates have been shown to be antiandrogenic (26), we alternatively stratified by sex and Tanner stage for pubic hair development (1 vs >1), a process driven by a rise in adrenal hormones, and findings were similar (data not shown).

In utero phthalate and BPA in relation to metabolic homeostasis

Among boys, in utero phthalate metabolite and BPA concentrations were not significantly associated with c-peptide, leptin, IGF-1, or CP-IR at age 8–14 years after adjustment for covariates (Tables 2 and 3). Among girls, an IQR increase in in utero MEP was associated with 8.0% higher leptin when adjusting for BMI z-score and other covariates (P = .04, R² = 0.78) (Table 2). However, when BMI was not included this association was not apparent, and an IQR increase in in utero MBzP was associated with 16% lower leptin among girls (P = .03, R² = 0.08) (Table 2).

Table 2.

Percent Difference in C-Peptide and Leptin per IQR Increase in In Utero ln-Transformed Phthalate Metabolite or BPA Concentrations, Stratified by Sex

	Boys (n = 105)			Girls (n = 114)
	% Diff	95% CI	P Value	% Diff	95% CI	P Value
C-peptidea
BPA	0.2	(−15, 18)	.98	14	(−3.4, 36)	.12
ΣDEHP	−1.9	(−15, 14)	.79	−0.5	(−13, 13)	.94
ΣDBP	−4.9	(−16, 7.7)	.42	2.3	(−9.0, 15)	.70
MBzP	−9.2	(−18, 1.1)	.08	−5.8	(−15, 5.0)	.28
MCPP	−4.1	(−16, 9.8)	.54	1.9	(−11, 17)	.79
MEP	−8.4	(−18, 2.6)	.13	4.6	(−5.2, 15)	.37
Leptina
BPA	−2.4	(−17, 14)	.76	7.9	(−5.2, 23)	.25
ΣDEHP	3.8	(−9.9, 20)	.60	−7.0	(−16, 2.6)	.15
ΣDBP	3.5	(−8.1, 16)	.57	−1.9	(−10, 7.2)	.67
MBzP	3.8	(−6.4, 15)	.47	−2.6	(−10, 5.7)	.52
MCPP	3.7	(−8.9, 18)	.58	−6.3	(−16, 4.0)	.22
MEP	2.0	(−8.5, 14)	.72	8.0	(0.3, 16)	.04
Leptinb (not adjusted for BMI z-score)
BPA	−11	(−31, 16)	.39	−6.7	(−28, 21)	.60
ΣDEHP	−3.5	(−24, 22)	.77	−9.0	(−25, 11)	.35
ΣDBP	5.5	(−13, 29)	.59	−11	(−25, 6.6)	.21
MBzP	0.8	(−15, 20)	.93	−16	(−29, −1.7)	.03
MCPP	3.0	(−17, 28)	.78	−15	(−31, 5.2)	.14
MEP	−2.3	(−19, 17)	.80	1.0	(−13, 18)	.90

ΣDEHP: molar sum of MEHP, MEHHP, MEOHP, and MECPP; ΣDBP, molar sum of MnBP and MiBP.

^a

Adjusted for age, BMI z-score, and urinary-specific gravity.

^b

Adjusted for age and urinary-specific gravity.

Table 3.

Percent Difference in Serum Outcome Measures per IQR Increase in In Utero Phthalate Metabolite or BPA Concentrations, Adjusted for Age, BMI z-Score, and Urinary-Specific Gravity and Stratified by Sex and Pubertal Status

	Prepubertal Boys (n = 55)			Pubertal Boys (n = 47)			Prepubertal Girls (n = 82)			Pubertal Girls (n = 32)
	% Diff	(95% CI)	P Value	% Diff	(95% CI)	P Value	% Diff	(95% CI)	P Value	% Diff	(95% CI)	P Value
IGF-1
BPA	3.5	(−10, 20)	.63	2.1	(−17, 25)	.84	−7.5	(−20, 7.0)	.29	8.9	(−11, 33)	.39
ΣDEHP	1.0	(−12, 16)	.89	−8.4	(−23, 9.1)	.32	−3.3	(−13, 7.4)	.53	24	(6.1, 46)	.01
ΣDBP	6.0	(−6.7, 20)	.36	−3.9	(−15, 9.1)	.53	1.5	(−7.4, 11)	.74	6.6	(−9.5, 26)	.43
MBzP	0.3	(−9.5, 11)	.95	−8.7	(−18, 1.9)	.10	2.0	(−6.3, 11)	.65	−8.1	(−22, 7.8)	.29
MCPP	1.6	(−11, 16)	.81	−9.4	(−21, 4.6)	.17	1.3	(−8.9, 13)	.81	7.5	(−14, 35)	.52
MEP	3.5	(−7.2, 15)	.53	−4.2	(−15, 7.5)	.45	3.8	(−4.3, 13)	.36	8.6	(−4.0, 23)	.18
Glucose
BPA	2.9	(−1.4, 7.3)	.19	−3.2	(−8.3, 2.1)	.22	1.9	(−4.0, 8.1)	.53	0.3	(−5.0, 5.8)	.92
ΣDEHP	2.5	(−1.6, 6.9)	.23	−4.1	(−8.3, 0.4)	.07	3.5	(−0.8, 7.9)	.11	−1.2	(−5.8, 3.6)	.61
ΣDBP	2.3	(−1.4, 6.2)	.22	−3.4	(−6.4, −0.3)	.03	2.2	(−1.5, 6.0)	.24	−2.1	(−6.2, 2.2)	.33
MBzP	−1.0	(−4.0, 2.0)	.50	−1.5	(−4.4, 1.5)	.31	0.3	(−3.1, 3.7)	.88	0.3	(−3.9, 4.7)	.87
MCPP	1.1	(−2.8, 5.2)	.57	−4.0	(−7.4, −0.4)	.03	1.6	(−2.6, 6.0)	.46	−0.6	(−6.4, 5.7)	.85
MEP	0.3	(−2.9, 3.6)	.86	1.5	(−1.5, 4.7)	.32	2.1	(−1.2, 5.4)	.22	0.00	(−3.3, 3.4)	1.00
CPI
BPA	2.5	(−15, 24)	.79	−10	(−33, 20)	.46	12	(−9.8, 38)	.31	8.8	(−14, 37)	.46
ΣDEHP	2.7	(−14, 23)	.77	−9.1	(−29, 17)	.44	−1.7	(−16, 15)	.82	−2.9	(−21, 19)	.77
ΣDBP	−6.9	(−21, 9.8)	.39	−3.9	(−20, 15)	.66	3.4	(−9.5, 18)	.62	−1.0	(−18, 19)	.91
MBzP	−9.2	(−20, 3.5)	.15	−6.8	(−21, 9.4)	.38	−3.8	(−15, 8.7)	.53	−14	(−28, 2.0)	.08
MCPP	−9.4	(−24, 7.6)	.25	−4.6	(−22, 17)	.65	2.5	(−12, 19)	.74	10	(−15, 43)	.44
MEP	−1.8	(−15, 13)	.80	−17	(−29, −3.3)	.02	3.8	(−7.8, 17)	.53	3.5	(−10, 19)	.63
CP-IR
BPA	8.4	(−13, 36)	.47	−16	(−38, 14)	.26	16	(−10, 50)	.26	9.4	(−14, 39)	.45
ΣDEHP	8.0	(−13, 34)	.48	−16	(−36, 8.5)	.17	5.3	(−13, 27)	.58	−5.1	(−24, 18)	.62
ΣDBP	−2.5	(−20, 19)	.80	−10	(−26, 8.3)	.25	7.9	(−8.1, 27)	.35	−5.1	(−22, 16)	.59
MBzP	−11	(−24, 4.0)	.14	−9.6	(−24, 7.0)	.23	−3.3	(−17, 12)	.65	−14	(−28, 3.9)	.11
MCPP	−7.4	(−25, 14)	.46	−12	(−29, 9.4)	.24	5.8	(−12, 27)	.54	9.1	(−17, 43)	.52
MEP	−1.3	(−17, 17)	.88	−15	(−28, 1.0)	.07	8.1	(−6.3, 25)	.28	3.5	(−11, 20)	.65

Prepubertal: Tanner stage = 1 (boys = genital development; girls = breast development); pubertal: Tanner stage > 1 (boys = genital development; girls = breast development). ΣDEHP: molar sum of MEHP, MEHHP, MEOHP, and MECPP; ΣDPB, molar sum of MnBP and MiBP.

Among pubertal boys, in utero ΣDBP and MCPP were associated with lower fasting glucose, whereas MEP was associated with lower CPI (Table 3). Among pubertal girls, an IQR increase in in utero ΣDEHP was associated with 24% higher IGF-1 (P = .01, R² = 0.32), although the number of girls in this stratum was small (n = 32) (Table 3).

Peripubertal phthalate and BPA in relation to metabolic homeostasis

In cross-sectional analyses, an IQR increase in peripubertal BPA was associated with 16% higher leptin in boys (P = .01, R² = 0.70) after adjustment for BMI z-score and other covariates (Table 4). However, when BMI z-score was not included this association disappeared, and an IQR increase in ΣDBP was associated with 17% lower leptin in girls (P = .02, R² = 0.09) (Table 4).

Table 4.

Percent Difference in C-Peptide and Leptin per IQR Increase in Peripubertal ln-Transformed Phthalate Metabolite or BPA Concentrations, Stratified by Sex

	Boys (n = 112)			Girls (n = 128)
	% Diff	95% CI	P Value	% Diff	95% CI	P Value
C-peptide
BPA	3.4	(−8.5, 17)	.59	9.9	(−2.1, 23)	.11
ΣDEHP	5.8	(−4.1, 17)	.25	12	(−1.0, 26)	.07
ΣDBP	15	(2.2, 31)	.02	1.9	(−8.3, 13)	.72
MBzP	−5.7	(−17, 7.8)	.39	−1.2	(−12, 11)	.84
MCPP	11	(−2.6, 27)	.11	3.8	(−5.8, 14)	.45
MEP	2.5	(−8.7, 15)	.68	6.8	(−3.1, 18)	.18
Leptina
BPA	16	(3.8, 30)	.01	5.3	(−4.0, 15)	.27
ΣDEHP	5.0	(−4.4, 15)	.31	−1.2	(−10, 8.9)	.81
ΣDBP	1.9	(−9.7, 15)	.76	−2.6	(−10, 5.9)	.53
MBzP	−0.1	(−12, 14)	.99	−5.6	(−14, 3.7)	.23
MCPP	2.3	(−10, 16)	.73	4.1	(−3.7, 12)	.31
MEP	−1.7	(−12, 9.9)	.77	−5.5	(−12, 2.1)	.15
Leptinb (not adjusted for BMI z-score)
BPA	9.0	(−11, 33)	.40	8.2	(−10, 30)	.40
ΣDEHP	1.2	(−14, 19)	.88	1.7	(−16, 24)	.87
ΣDBP	1.0	(−18, 25)	.93	−17	(−29, −2.5)	.02
MBzP	−7.2	(−26, 16)	.51	−3.0	(−20, 17)	.75
MCPP	−11	(−29, 11)	.31	−8.0	(−21, 7.2)	.28
MEP	−4.7	(−21, 15)	.62	1.7	(−13, 18)	.83

ΣDEHP: molar sum of MEHP, MEHHP, MEOHP, and MECPP; ΣDBP, molar sum of MnBP and MiBP.

^a

Adjusted for age, BMI z-score, and urinary-specific gravity.

^b

Adjusted for age and urinary-specific gravity.

Among boys, an IQR increase in peripubertal ΣDBP was associated with 15% higher c-peptide (P = .02, R² = 0.37) (Table 4). Among pubertal boys, we observed lower IGF-1 associated with increases in several concurrently measured phthalate metabolites, including ΣDBP, MBzP, and MCPP, but did not observe the same associations among girls or prepubertal boys (Table 5). In contrast, among prepubertal girls we observed higher IGF-1 with increases in ΣDEHP (P = .03, R² = 0.20) (Table 5). Within this stratum, an IQR increase in ΣDEHP was also associated with 20% higher CPI (P = .02, R² = 0.48) and CP-IR (P = .07, R² = 0.43) (Table 5). Among pubertal girls we observed small but significant increases in fasting glucose with increases in concurrent urinary ΣDBP and MBzP (P = .04 and P = .03, respectively; R² = 0.35 for each association).

Table 5.

Percent Difference in Serum Outcome Measures per IQR Increase in Peripubertal Phthalate Metabolite or BPA Concentrations, Adjusted for Age, BMI z-Score, and Urinary-Specific Gravity and Stratified by Sex and Pubertal Status

	Prepubertal Boys (n = 54)			Pubertal Boys (n = 55)			Prepubertal Girls (n = 83)			Pubertal Girls (n = 45)
	% Diff	(95% CI)	P Value	% Diff	(95% CI)	P Value	% Diff	(95% CI)	P Value	% Diff	(95% CI)	P Value
IGF-1
BPA	6.6	(−7.2, 22)	.36	1.3	(−10, 15)	.84	8.4	(−2.4, 20)	.13	−4.4	(−15, 8.1)	.46
ΣDEHP	3.1	(−10, 18)	.65	−5.8	(−14, 3.1)	.19	14	(1.0, 28)	.03	2.9	(−8.6, 16)	.63
ΣDBP	4.1	(−8.3, 18)	.52	−15	(−26, −2.6)	.02	1.6	(−8.1, 12)	.76	0.9	(−9.8, 13)	.87
MBzP	−4.3	(−16, 8.4)	.48	−21	(−32, −8.8)	.002	11	(−0.8, 23)	.07	3.9	(−8.4, 18)	.54
MCPP	0.6	(−12, 15)	.93	−15	(−27, −1.9)	.03	−0.7	(−8.9, 8.3)	.88	−2.1	(−13, 9.6)	.70
MEP	9.2	(−3.0, 23)	.14	−2.4	(−14, 11)	.70	2.1	(−6.9, 12)	.66	7.9	(−2.7, 20)	.14
Glucose
BPA	1.0	(−3.5, 5.6)	.66	−0.2	(−3.5, 3.1)	.89	−1.7	(−5.8, 2.6)	.44	0.7	(−3.1, 4.6)	.73
ΣDEHP	3.0	(−1.4, 7.5)	.18	1.0	(−1.4, 3.5)	.39	−0.04	(−4.8, 5.0)	.99	1.9	(−1.8, 5.6)	.32
ΣDBP	1.6	(−2.5, 5.8)	.45	−1.9	(−5.5, 1.8)	.31	−1.8	(−5.7, 2.2)	.36	3.6	(0.2, 7.1)	.04
MBzP	0.1	(−3.9, 4.3)	.95	0.7	(−3.6, 5.1)	.77	−0.6	(−4.9, 3.9)	.80	4.1	(0.3, 8.1)	.03
MCPP	0.7	(−3.5, 5.1)	.76	−2.6	(−6.4, 1.3)	.19	1.0	(−2.4, 4.6)	.55	3.0	(−0.4, 6.6)	.08
MEP	−0.1	(−3.9, 3.9)	.97	−2.5	(−5.6, 0.7)	.13	0.4	(−3.3, 4.2)	.83	1.8	(−1.5, 5.2)	.27
CPI
BPA	0.5	(−16, 21)	.96	6.6	(−9.1, 25)	.43	14	(−1.3, 31)	.07	9.7	(−6.7, 29)	.26
ΣDEHP	4.0	(−13, 24)	.66	3.4	(−8.1, 16)	.57	20	(2.5, 41)	.02	6.7	(−8.7, 25)	.40
ΣDBP	11	(−5.9, 31)	.21	6.8	(−11, 28)	.47	9.0	(−4.8, 25)	.21	−6.5	(−19, 8.3)	.36
MBzP	−1.5	(−17, 16)	.86	−16	(−32, 2.8)	.09	4.6	(−10, 22)	.55	−5.5	(−20, 12)	.50
MCPP	5.4	(−11, 25)	.55	6.0	(−13, 29)	.56	8.3	(−3.7, 22)	.18	−5.5	(−19, 10)	.44
MEP	−1.2	(−16, 16)	.89	6.8	(−9.2, 26)	.42	12	(−1.1, 27)	.07	3.8	(−9.8, 19)	.59
CP-IR
BPA	2.5	(−18, 27)	.82	6.1	(−10, 25)	.48	10	(−7.8, 32)	.28	11	(−7.3, 33)	.25
ΣDEHP	10	(−11, 36)	.35	5.6	(−6.5, 19)	.37	20	(−1.4, 47)	.07	11	(−6.9, 32)	.24
ΣDBP	15	(−5.7, 39)	.17	2.8	(−15, 24)	.77	5.1	(−11, 24)	.56	0.4	(−15, 19)	.96
MBzP	−1.2	(−19, 20)	.90	−15	(−32, 4.9)	.13	3.4	(−14, 24)	.72	2.5	(−15, 24)	.79
MCPP	6.8	(−13, 31)	.52	0.5	(−18, 23)	.96	11	(−4.3, 28)	.17	0.2	(−15, 19)	.98
MEP	−1.3	(−18, 19)	.89	1.5	(−14, 20)	.86	13	(−3.1, 32)	.12	7.6	(−7.9, 26)	.35

Prepubertal: Tanner stage = 1 (boys = genital development; girls = breast development); pubertal: Tanner stage > 1 (boys = genital development; girls = breast development). ΣDEHP: molar sum of MEHP, MEHHP, MEOHP, and MECPP; ΣDPB, molar sum of MnBP and MiBP.

When we included phthalate metabolite and BPA concentrations from both in utero and peripubertal development in regression models together, our results did not materially change (data not shown).

Discussion

Despite concern that EDCs may play a significant role in the development of obesity (4) and the understanding that environmental exposures during critical developmental periods can have long-term health consequences (27), few studies have investigated relationships between in utero EDC exposure and metabolic function in childhood or adolescence. In this analysis we found that urinary phthalate metabolite and BPA concentrations during in utero development and at age 8–14 years were associated with markers of metabolic homeostasis, and many of these associations were dependent on sex and puberty status.

Comparisons with previous studies

In a previous longitudinal study, maternal urinary BPA concentrations during the third trimester were associated with higher serum leptin among boys at age 9 years, whereas concurrent urinary BPA concentrations during childhood were not (16). We observed a similar, positive association between BPA and serum leptin in boys, but with markers of concurrent childhood exposure rather than in utero exposure. Differences between study populations may be one explanation for this inconsistency. Although the 2 studies had similar sample sizes, utilized similar markers of exposure and metabolic function, and controlled for BMI, participants in the previous study were from a Mexican-American agricultural community with relatively higher BPA exposure and low socioeconomic status, whereas participants in the present study were from an urban environment in Mexico with lower BPA exposure and higher socioeconomic status (19, 28). However, the similarities of our findings are interesting, because both studies observed a sex-specific relationship between markers of BPA exposure and increased serum leptin despite measuring exposure at 2 distinct developmental periods. In addition, the cross-sectional associations between urinary BPA and serum leptin found here are similar to those previously reported in adults while controlling for body size (17). Interestingly, when we did not include BMI z-score in models predicting serum leptin, almost all effect estimates became smaller. Specifically, positive associations of leptin with in utero MEP in girls and peripubertal BPA in boys were no longer significant, whereas other significant negative associations emerged (Tables 2 and 4). These findings suggest that phthalate and BPA exposure may be related to leptin concentrations both independent of BMI as well as via this pathway.

In a cross-sectional analysis of urinary phthalate metabolites and metabolic outcomes among 12- to 19-year-old NHANES participants, DEHP metabolites were associated with increased HOMA-IR, indicating that concurrent exposure was associated with increased insulin resistance (6). A similar study of adult NHANES participants also reported associations between several phthalate metabolites and HOMA-IR and fasting serum glucose, particularly among Mexican-Americans (7). We observed similar, positive, cross-sectional relationships between ΣDEHP metabolites and c-peptide based measures of both insulin production (CPI) and resistance (CP-IR), as well as with increased IGF-1, specifically within prepubertal girls. We also observed similar positive associations between ΣDBP and concurrently measured fasting glucose, but only among pubertal girls.

A study of Dutch 4–9 year olds reported associations between concurrent measures of DEHP metabolites and mono(carboxyoctyl) phthalate and decreased IGF-1 in boys but not girls (8). We did not observe this cross-sectional relationship, although IGF-1 was negatively associated with other phthalate metabolites specifically among pubertal boys, including ΣDBP, MBzP, and MCPP. In contrast, ΣDEHP was associated with higher IGF-1 among prepubertal girls. One explanation for this difference may be that the Dutch study involved prepubertal children, whereas participants in the present study were older when IGF-1 levels may be beginning to rise. In addition, children in the present study had higher phthalate metabolite concentrations compared with children in the Dutch study (8, 19). Similar to previous cross-sectional studies of BPA exposure in childhood (15), we did not observe associations between peripubertal BPA and indicators of insulin resistance.

Potential mechanisms

The mechanisms by which phthalate and BPA exposure may disrupt metabolic function are still uncertain, and likely differ depending on timing of exposure (ie, in utero vs peripubertal). One possibility is via peroxisome proliferator-activated receptors (PPARs), a group of nuclear receptors that play a role in adipogenesis, lipid metabolism, and metabolic homeostasis (29). Animal and in vitro studies have demonstrated that BPA and many phthalate parent compounds and metabolites can alter PPAR-α and PPAR-γ expression (30), resulting in changes in adipocyte differentiation and release of leptin and adiponectin from adipocytes (31). Because PPAR-γ is highly expressed in adipose tissue and controls expression of leptin and other factors that play a role in insulin sensitivity (29), activation of this receptor may be an important pathway by which phthalate or BPA exposure can affect metabolic function. Findings from the present study are consistent with this hypothesis, because we observed associations between markers of in utero and peripubertal phthalate metabolite and BPA exposure and serum leptin concentrations. In addition, we observed positive associations between ΣDBP and c-peptide among boys and ΣDEHP and CPI among prepubertal girls, suggesting increased insulin secretion and pancreatic β-cell function, possibly as a result of increased insulin resistance. PPAR activity also differs by sex (32), which could potentially explain the observed sex-specific associations between exposure and metabolic function.

Phthalates and BPA may also disrupt metabolic homeostasis via oxidative stress. Previous studies have demonstrated that urinary phthalate metabolite and BPA concentrations are associated with increased 8-isoprostane, a marker of lipid peroxidation, and increased 8-hydroxydeoxyguanosine, a marker of DNA oxidation (33, 34). In vitro and animal studies have demonstrated that increased production of reactive oxygen species can induce insulin resistance and dysregulate adipokines (35). Again, our findings are consistent with this hypothesis as they suggest possible increased insulin production and insulin resistance associated with increased markers of phthalate exposure. However, we did not observe associations between markers of BPA exposure and insulin production or resistance, despite our previous findings of strong associations between BPA and markers of oxidative stress in a cohort of pregnant women (33).

Limitations

One limitation of the current study is that we did not measure insulin in participant serum samples, so were unable to calculate HOMA-IR, a common measure of insulin resistance. Instead, we measured serum concentrations of c-peptide, a peptide secreted from pancreatic β-cells at the same rate as insulin. C-peptide is an excellent measure of β-cell function as it directly reflects insulin secretion, is more stable and has a longer biological half-life than insulin, and is not a component of synthetic insulin (23). However, c-peptide does not directly reflect insulin action, and thus may not be the most precise measure of insulin resistance or sensitivity (22). As a result, our ability to directly assess insulin resistance and to make comparisons with previous studies that utilized HOMA-IR was somewhat limited.

We also did not control for potential dietary sources of BPA and phthalate exposure. Because diet is also related to fasting glucose, insulin resistance, and other markers of metabolic function, intake of certain foods may confound relationships between our exposure and outcome measures. However, it is not clear which specific foods are primary sources of phthalate and BPA exposure (36), or whether these overlap with dietary influences, eg, macronutrient composition, fiber, and glycemic index, relevant to metabolic outcomes.

We measured maternal urinary phthalate metabolites and BPA at one point in time as a measure of in utero exposure to the fetus, even though concentrations fluctuate across pregnancy (37, 38). If phthalate and BPA exposure early in pregnancy has an impact on metabolic development, we may not have observed an association in the present study, because we measured maternal concentrations during the third trimester. However, in ongoing research within this study population we are measuring markers of environmental exposure across pregnancy using archived samples. Concurrently measured markers of metabolic homeostasis and peripubertal phthalate and BPA exposure were also measured at one time point, limiting our ability to evaluate causal relationships.

In addition, because this analysis was exploratory in nature, we made a large number of comparisons, increasing the likelihood of chance significant findings. However, an important strength of this analysis is phthalate metabolite and BPA measurements collected during both in utero and peripubertal development, which allowed us to examine potential sensitive windows of exposure. We also measured multiple biomarkers reflecting lipid and glucose metabolism, which provide insights into the development of different components of metabolic risk.

Conclusion

Urinary phthalate metabolites and BPA measured during both in utero and peripubertal development were associated with markers of glucose and lipid metabolism at age 8–14 years in patterns that varied by both sex and puberty status. Considering the long-term health effects related to childhood obesity and metabolic syndrome, additional research with repeated exposure and outcome measures during childhood and peripubescence are needed to deepen our understanding of these relationships.

Acknowledgments

This work was supported by the following grants: P01ES022844 and T32ES007062 from the National Institute for Environmental Health Sciences (NIHES), and RD 83543601 from the US Environmental Protection Agency (US EPA). It contents are solely the responsibility of the grantee and do not necessarily represent the official views of the US EPA. Further, the US EPA does not endrose the purchase any commerical products or services mentioned in the publication.

Disclosure Summary: The authors have nothing to disclose.

Abbreviations

 
• BMI

  body mass index

 
• BPA

  bisphenol A

 
• CI

  confidence interval

 
• CPI

  c-peptide index

 
• CP-IR

  c-peptide-based measure of insulin resistance

 
• DBP

  dibutyl phthalate

 
• ΣDBP

  DBP metabolite summary measure

 
• DEHP

  di-2-ethylhexyl phthalate

 
• ΣDEHP

  DEHP metabolite summary measure

 
• EDC

  endocrine-disrupting chemical

 
• HOMA-IR

  homeostatic model assessment of insulin resistance

 
• IQR

  interquartile range

 
• LOD

  limit of detection

 
• MBzP

  monobenzyl phthalate

 
• MCPP

  mono-3-carboxypropyl phthalate

 
• MECPP

  mono-2-ethyl-5-carboxypentyl phthalate

 
• MEHHP

  mono-2-ethyl-5-hydroxyhexyl phthalate

 
• MEHP

  mono-2-ethylhexyl phthalate

 
• MEOHP

  mono-2-ethyl-5-oxohexyl phthalate

 
• MEP

  monoethyl phthalate

 
• MiBP

  monoisobutyl phthalate

 
• MnBP

  mono-n-butyl phthalate

 
• NHANES

  National Health and Nutrition Examination Survey

 
• PPAR

  peroxisome proliferator-activated receptor.