Bisphenol A Exposure and the Development of Wheeze and Lung Function in Children through Age Five Years

Abstract

Importance

Bisphenol A (BPA), a prevalent endocrine disrupting chemical, has been associated with wheezing in children, but few studies have examined its impact on lung function or wheeze in older children.

Objective

The objectives of this study were to test whether BPA exposure was associated with lung function, with wheeze, and with pattern of wheeze in children over the first five years.

Design

A birth cohort study, enrolled during early pregnancy.

Setting

Greater Cincinnati, Ohio area.

Participants

398 mother-infant dyads,.

Exposure

We collected maternal urine during pregnancy (16 and 28 weeks) and child urine annually to assess gestational and child BPA exposure.

Main Outcome Measures

We assessed parent-reported wheeze every 6 months for 5 years and measured child forced expiratory volume in one second (FEV1) at age 4 and 5 years. We evaluated associations of BPA with respiratory outcomes: FEV1, child wheeze, and wheeze phenotype.

Results

Urinary BPA concentrations and FEV1 data were available for 208 children, and urinary BPA and parent-reported wheeze data were available for 360 children. Mean maternal urinary BPA ranged from 0.5 to 316 μg/g of creatinine. In multivariable analysis, every 10-fold increase in mean maternal urinary BPA was associated with 14.2% decrease in %FEV1 at 4 years (95% CI −24.5, −3.9) but no association was found at 5 years. In multivariable analysis, every 10-fold increase in mean maternal urinary BPA concentration was marginally associated with a 55% increase in the odds of wheezing (OR 1.55, 95% CI 0.91, 2.63). While mean maternal urinary BPA concentration was not associated with wheeze phenotypes, a 10-fold increase in 16 week maternal BPA was associated with a 4.3 fold increase in odds of persistent wheeze (OR 4.3, 95% CI 1.4, 13.3). Child BPA concentrations were not associated with FEV1 or wheeze.

Conclusions and Relevance

These results provide evidence that suggest that prenatal, but not postnatal, exposure to BPA is associated with diminished lung function and the development of persistent wheeze in children.

Introduction

Asthma rates have risen over the past three decades; one in ten US children have asthma.^{1, 2} Environmental factors, such as tobacco exposure and airborne pollutants, have been identified as risk factors for asthma, but reasons for the increased prevalence of asthma remains poorly understood.^{3, 4} Some investigators have suggested that exposure to endocrine disrupting chemicals, such as phthalates and bisphenol A (BPA), may contribute to the development of asthma in children.^5–8

BPA, a chemical used in some plastics and epoxy resins, is found in many consumer products, and most Americans have detectable BPA in their urine.⁹ Mice pups that were exposed to BPA prenatally developed an asthma phenotype.^{10, 11} We previously reported an association of prenatal BPA exposure with increased odds of developing parent reported wheeze in children through age three years, but we did not examine objective measures of lung function, like spirometry.¹² Others reported that postnatal BPA exposure was associated with child asthma and wheeze, but they did not find an association of prenatal BPA exposure.¹³

Spirometry is a valuable diagnostic tool for identification of respiratory diseases in children.^14–16 Most guidelines recommend using forced expiratory volume in one second (FEV1) for assessing respiratory status in children.¹⁶ The objectives of this study were to test whether BPA exposure was associated with lung function using FEV1, with wheeze, and with pattern of wheeze in children over the first five years.

Methods

This study was comprised of participants in the Health Outcomes and Measures of the Environment (HOME) Study, a prospective birth cohort designed to investigate the effects of exposure to environmental toxicants on child health.^{12, 17} Between March 2003 and January 2006, we enrolled 398 English-speaking women who were 18 years or older, at 16 (± 3) weeks gestation, and lived in a home built before 1979. We tracked the women through pregnancy and followed their children through age 5 years. Women resided within five counties surrounding Cincinnati, received prenatal care from participating obstetrical clinics (9), and delivered at a participating hospital (3). The study included an embedded randomized trial of a lead hazard reduction intervention and injury hazard reduction. The Cincinnati Children’s Hospital Medical Center and the Centers for Disease Control and Prevention (CDC) institutional review boards approved the HOME Study.

This study included the subset of the 398 live-born HOME study infants for whom both urinary BPA concentrations and respiratory outcome data were available. BPA and wheeze data were available for 360 (90%) children. BPA and spirometry data were available for at least one time point (4 or 5 years) for 208 children (155 at age 4 years and 193 at age 5 years). Reasons for missing spirometry data included: not completing the 4 year or 5 year clinic visit, completing the visit before IRB approval for FEV1, child noncooperation, or parental time constraints.

BPA Assessment

We measured BPA concentrations in serial, spot maternal and child urine samples. We collected urine in glass containers at 16 weeks gestation and 26 weeks gestation home visits and annual child visits. The CDC quantified total urinary BPA concentrations using online solid phase extraction coupled with high-performance liquid chromatography isotope dilution tandem mass spectrometry.^{18, 19} We replaced ~10% of BPA concentrations below the limit of detection (LOD = 0.4 μg/L) with a value equal to the LOD divided by √2.²⁰ We standardized BPA concentrations for urinary creatinine (μg BPA/g creatinine) at each time point and log₁₀ transformed the BPA concentrations. We used the mean of creatinine standardized maternal urinary BPA concentrations as our primary measure of maternal BPA exposure.

Covariates

Research assistants conducted surveys at baseline and every 6 months after the children were born to collect demographic characteristics (maternal education, occupation, income, self-reported maternal race, child sex, and health insurance status). We measured covariates with plausible FEV1 or wheeze associations including: prenatal tobacco exposure, season, breastfeeding history, family history of asthma, family history of allergy, child eczema, child allergy, birth weight, maternal parity, pet ownership, and cockroach exposure (by report).^21–25

We measured prenatal tobacco exposure using maternal serum cotinine. Cotinine concentrations at 16 weeks, 26 weeks, and delivery were determined using high-performance liquid chromatography/atmospheric-pressure chemical ionization tandem mass spectrometry.^{26, 27} We imputed values for undetectable concentrations (~35%) by sampling randomly from the left tail of the lognormal distribution of serum cotinine (excluding smokers), and then we calculated the mean of maternal cotinine measures as the measure of prenatal tobacco exposure.

Outcome Measures

During the 4 and 5 year clinic visits, research assistants measured FEV1 using a Piko-1 meter (nSpire Health Inc., Longmont, CO).^{28, 29} We attempted to collect at least three acceptable (meter determined) FEV1 measures from each child. FEV1 was recorded in liters (resolution 0.01 L). We used the maximum acceptable FEV1 obtained as the FEV1 for each participant. We calculated the percent predicted FEV1 (%FEV1) and multiplied the calculated %FEV1 value by 0.9 for children whose mothers reported their race as Black.³⁰

Research assistants surveyed parents every six months for five years to collect parent-reported wheeze data. We used a question from the National Health and Nutrition Examination Survey, asking, “has (child’s name) had wheezing or whistling in his/her chest,” in the last 6 months?³¹ We used this dichotomous value as our child wheeze outcome. We also conducted a trajectory analysis to identify distinct groups of wheeze trajectories/phenotypes.^{32, 33}

Statistical Analysis

We calculated descriptive statistics for all demographic, exposure, and outcome data. We used the geometric mean and 95% confidence interval (CI) to describe central tendency and dispersion for BPA concentrations. We used SAS Version 9.3 (SAS Institute, Inc., Cary, NC) and employed a two sided, 5% level to test statistical significance. We log₁₀ transformed urinary BPA, therefore, the beta estimates represent a 10-fold change in BPA.

We analyzed the association of the mean maternal urinary BPA concentration and child urinary BPA concentration at 4 and 5 years with %FEV1 at 4 and 5 years after adjusting for covariates using mixed effects linear regression.³⁴ We first evaluated associations of mean maternal urinary BPA, then child BPA, and then both in the same analysis.

We analyzed the association of mean maternal urinary BPA concentration with wheeze over the previous six months using generalized estimating equations (GEE) with a logit link to account for within-subject correlation resulting from repeated outcome measurements.¹² We made some assumptions for the analysis of associations of child BPA with wheeze because wheeze was measured twice a year and BPA was measured annually. First, we evaluated the intraclass correlation coefficient (ICC) of child BPA to determine if mean child BPA could be a proxy for child exposures. Second, we evaluated associations of the annual child BPA with concurrent wheeze outcomes collapsed over each year (i.e. 12 month BPA aligned with child wheeze during 0–12 months). Third, we evaluated associations of the annual child BPA projected on future wheeze outcomes collapsed over each year (i.e. 12 month BPA aligned with child wheeze from 12–24 months). Lastly, we added mean maternal urinary BPA to these concurrent and future child BPA and wheeze analyses.

We also analyzed the association of mean maternal urinary BPA concentration with wheeze phenotype using developmental trajectory analysis.^{32, 33} This method identifies, rather than assumes, groups of trajectories, using an objective algorithm for determining groups and estimates the probability that each participant belongs to each distinctive group. Individuals with missing data are estimated with less certainty using this approach. We did not adjust for other covariates, except creatinine (where appropriate), due to inherent power limitations.

We tested interactions of urinary BPA concentration with time (where applicable) in multivariable analyses. We also conducted secondary analyses to explore potential windows of vulnerability by replacing the mean maternal urinary BPA concentration variable with urinary BPA concentration measured at each of the maternal time points (i.e. 16 weeks or 26 weeks) in separate analyses; we used non-standardized BPA and adjusted for concurrent urinary creatinine in these analyses. Lastly, to determine consistency of results, we revisited all analyses with BPA without accounting for creatinine.

Results

Participants for whom FEV1 outcome data were available were more likely to be White, have a mother with more than a high school education, have married parents, have a higher household income, and have a lower prenatal cotinine level than participants without FEV1 data (Table 1). Maternal urinary BPA concentration was similar between the children with FEV1 data and those without. Mean maternal urinary BPA ranged from 0.53 to 293.55 μg/g of creatinine with a geometric mean of 2.4 μg/g of creatinine (95% CI 2.1, 2.7). Maternal urinary BPA concentrations at 16 and 26 weeks gestation were weakly correlated (r=0.18, p=0.03).

Table 1.

Characteristics and Exposures of Study Participants with Available Maternal BPA Concentration and FEV1 or Wheeze Data

	FEV1 (N=208)	Wheeze (N=360)
Child sex
Female	116 (55.8%)	198 (55%)
Male	92 (44.2%)	162 (45%)
Maternal race
Black	61 (29.3%)	104 (28.9%)
White	134 (64.4%)	234 (65%)
Other	13 (6.3%)	22 (6.1%)
Maternal education
≤ High School	41 (19.7%)	78 (21.7%)
> High School	167 (80.3%)	282 (78.3%)
Maternal marital status
Married, living apart	1 (0.5%)	3 (0.8%)
Married, living together	143 (68.8%)	240 (66.7%)
Not married, but living with someone	20 (9.6%)	46 (12.8%)
Not married, living alone	44 (21.2%)	71 (19.7%)
Household income
≤$25,000)	53 (25.7%)	89 (25.1%)
$25,000–$50,000	35 (17%)	75 (21.2%)
$50,000–$80,000	63 (30.6%)	90 (25.4%)
>$80,000	55 (26.7%)	100 (28.2%)
History of maternal allergy
No	118 (56.7%)	206 (57.2%)
Yes	90 (43.3%)	154 (42.8%)
Child insurance status
Private	151 (74%)	247 (73.3%)
Public	53 (26%)	90 (26.7%)
Log 10 mean cotinine from 16 and 26 weeks
Mean (SD), μg/L	−1.3 (1.12)	−1.2 (1.23)
Median, μg/L	−1.6	−1.6
Range, μg/L	(−3.3–2.5)	(−3.3–2.6)
Log 10 mean BPA from 16 and 26 weeks
Mean (SD), μg/g of creatinine	0.4 (0.35)	0.4 (0.32)
Median, μg/g of creatinine	0.3	0.3
Range, μg/g of creatinine	(−0.3–2.5)	(−0.3–2.5)
Gestational age (weeks)
Mean (SD)	39.1 (1.59)	39.0 (1.69)
Median	39.4	39.3
Range	(32.7–42.1)	(30.4–42.1)

The frequency of parent reported wheeze during the previous six month period (n=360) varied between 15.9% and 24.1% (Table 2). The correlation of %FEV1 at 4 and 5 years was 0.34 (95% CI 0.18, 0.48).

Table 2.

Respiratory Outcomes at Each Age (Months)

	6	12	18	24	30	36	42	48	54	60
Reported Wheeze (%)	19.9	24.1	22.0	19.3	15.9	17.1	16.3	18.7	16.2	16.3
FEV1 (L) Mean ± Standard Deviation	--	--	--	--	--	--	--	0.7 ± 0.25	--	0.89 ± 0.26
%FEV1 (%) Mean ± Standard Deviation	--	--	--	--	--	--	--	76 ± 24	--	84.6 ± 22

%FEV1

In adjusted analysis of the association of maternal urinary BPA with %FEV1, there was a significant interaction of BPA with time (p=0.03). The interaction demonstrated an association of increasing concentrations of mean maternal BPA with decreasing %FEV1 at 4 years (β=−14.2, 95% CI −24.5, −3.9, p=0.007). In contrast, there was no statistically significant association at 5 years (β=0.04, 95% CI −9.04, 9.12, p=0.99).

In an adjusted analysis evaluating the association of child urinary BPA with %FEV1, there was no statistically significant association and no interaction with time. Additionally, when maternal urinary BPA was added to the analysis, the results were nearly identical to the analysis of maternal urinary BPA alone (no association of child BPA with %FEV1).

We examined the effect of the timing of maternal urinary BPA exposure by evaluating the two maternal BPA measures separately. Adjusting for the same factors and creatinine, we found a significant interaction of 16 week maternal BPA with time (p=0.04). The interaction demonstrated a borderline association of increasing concentrations of 16 week maternal urinary BPA with decreasing %FEV1 at 4 years (β=−7.6, 95% CI −16.4, 1.3, p=0.09) but no association at 5 years (β=2.4, 95% CI −5.4, 10.2, p=0.55). The relationships were similar for 26 week maternal urinary BPA, as the interaction demonstrated a significant association (p=0.03) of increasing concentrations of 26 week maternal BPA with decreasing %FEV1 at 4 years (β=−9.7, 95% CI −18.2, −1.3, p=0.03) but no association at 5 years (β=−0.2, 95% CI −8.1, 7.7, p=0.95). These findings were statistically unchanged when evaluating BPA without creatinine adjustment.

Wheeze in the previous 6 months

In adjusted analysis, mean maternal urinary BPA concentration was marginally associated with wheeze (AOR 1.55, 95% CI 0.91, 2.63, p=0.11). There was no statistically significant interaction of mean maternal BPA concentration with time.

The intraclass correlation coefficient (ICC) for child urinary BPA was 0.15, so we did not evaluate mean child BPA. In the analysis of child BPA with concurrent wheeze there was no statistically significant association (AOR=1.06, 95% CI 0.75, 1.51, p=0.74), and there was no statistically significant association in the child BPA and future wheeze analysis (AOR=1.08, 95% CI 0.65, 1.78, p=0.76). These associations (concurrent or future) did not change when maternal urinary BPA was added to the analysis.

We used separate multivariable models to estimate associations of wheeze with creatinine adjusted log-transformed maternal urinary BPA concentration at both time points during pregnancy (16 weeks gestation and 26 weeks gestation). An increase in maternal urinary BPA concentration at 16 weeks gestation was associated with wheeze (AOR=1.79, 95% CI 1.16, 2.78, p=0.01), but maternal urinary BPA concentration at 26 weeks gestation was not associated with wheeze (AOR=1.06, 95% CI 0.65, 1.74, p=0.81). There was no statistically significant interaction of maternal BPA concentration with time. These findings were statistically unchanged when evaluating BPA without creatinine adjustment.

Wheeze Phenotype

We identified four distinct developmental trajectories of wheeze, which we categorized: 1) never, 31.2%, 2) early onset/transient, 41.4%, 3) late onset, 15.7%, and 4) persistent, 11.7% (Figure 1).³⁵ Mean maternal urinary BPA concentration was not associated with wheeze phenotypes, but a 10-fold increase in 16 week BPA (adjusted for log-transformed urinary creatinine) was associated with a 4.3 fold increase in odds of persistent wheeze compared to no wheeze (Table 3). These findings were statistically unchanged when we evaluated BPA without creatinine adjustment. We did not evaluate child BPA in the trajectory analyses because of the low ICC of child BPA.

Figure 1.

Developmental Trajectories of Wheeze

Table 3.

Associations of Log₁₀ Maternal BPA with Wheeze Phenotype

BPA Measure	Early		Late		Persistent
	OR*	95% CI	OR*	95% CI	OR*	95% CI
Mean Maternal BPA†	1.48	0.28,7.88	0.78	0.14, 4.39	2.73	0.66, 11.3
16 week maternal BPA‡	1.31	0.39, 4.34	0.86	0.21, 3.53	4.27	1.37, 13.3
26 week maternal BPA‡	1.17	0.37, 3.69	0.71	0.17, 3.0	0.90	0.26, 3.12

^*Never wheeze is the reference group

^†Standardized for urinary creatinine

^‡Adjusted for same time urinary creatinine

Discussion

We found that prenatal BPA exposure, measured using maternal urinary BPA concentrations at two time points during gestation, was associated with a decrease in children’s lung function at age four years but not at age five years; child BPA concentrations were not associated with lung function at either age. Maternal urinary BPA concentration was marginally associated with increased odds of parent-reported wheeze, and there was a trend for increased odds of persistent wheeze. Child BPA concentrations were not associated with wheeze. BPA exposure during both early and later pregnancy was associated with %FEV1, but BPA exposure during early pregnancy was more strongly associated with wheeze and wheeze phenotype. These results confirm and extend the observed increase in odds of child wheeze associated with prenatal BPA exposure previously described in this same cohort at a younger age.¹²

Investigators have noted an association of exposure to plastic materials with the development of respiratory disorders in animal and epidemiologic studies.^5–8 Animal studies suggest that BPA may affect lung development. Specifically, perinatal exposure to BPA enhanced airway inflammation and responsiveness in a mouse model of asthma.¹⁰ In another study, prenatal but not postnatal BPA exposure promoted the development of allergic asthma in mice.¹¹ The mechanism for this association is unclear, but a recent study noted that rhesus macaques exposed to BPA had accelerated development of secretory cells in the proximal airways.³⁶ However, another animal study reported that maternal exposure to BPA has only subtle association with allergic inflammation that did not lead to airway responsiveness.³⁷ Human studies note an association of BPA and asthma, but the risks of exposure and timing of exposure are inconsistent. We noted an association of prenatal BPA exposure and child wheeze in young children; however, another study reported a postnatal association of BPA exposure with child asthma and wheeze but did not find an association of prenatal BPA exposure.^{12, 13} In an analysis of 2005–2006 National Health and Nutrition Examination Survey data, Vaidya reported a cross-sectional association of urinary BPA and allergic asthma, but only in females.³⁸ The design of these studies varied such that it is not possible to make direct comparison of timing of exposures and associated outcomes.

Spirometry is a valuable tool for the early diagnosis of respiratory diseases in pre-school children.^14–16 FEV1 is the key spirometric measure in asthma management; it has been shown to predict future asthma symptoms and exacerbations.^{14, 39} Moreover, prospective studies show that children with wheeze have persistently lower FEV1 than non-wheezing children, and poor airway function in early life is an important risk factor for poor airway function later.^{40, 41} However, the intra-subject variability of FEV1 over time has implications for the use of FEV1 as a long vs. short term measure of lung function.⁴² FEV1 can be useful in assessment of current status, but a spot measure of FEV1 may not represent long term trends.⁴³ The association of maternal BPA exposure with a decrease in child lung function in this study was inconsistent, but the association at age four years represents an objective and clinically relevant effect that extends our earlier work showing that maternal BPA exposure was associated with the development of wheeze in young children.¹² When we evaluated these patterns of wheeze, we found that gestational BPA exposure has a marginal association with increased odds of wheeze through age five, especially the “persistent” wheeze phenotype. Although these results need to be confirmed in other cohorts, this study raises questions about the role of BPA exposures in the development of asthma and the asthma epidemic.

We also tested whether gestational urinary BPA concentration at different time points was more strongly associated with FEV1, wheeze, and wheeze phenotype. Higher BPA concentration at 16 weeks of gestation was associated with an increased odds of wheeze, increased odds of being in a persistent wheeze phenotype, and a borderline association with FEV1 deficits; whereas, higher BPA concentration at 26 weeks was associated with decreased FEV1 but not wheeze or phenotype. These results, taken from a small birth cohort, suggest that maternal BPA exposure is a risk factor for diminished lung function and wheeze in children, and that earlier gestation exposures may be more important than later.

There are limitations to this study. First, BPA concentration was weakly correlated at 16 and 26 weeks, and it is known to vary widely over time;^44–47 this may result in exposure misclassification. We had serial measures of urinary BPA concentration which diminishes, but does not dismiss this limitation. Exposure misclassification would likely reduce rather than enhance the association of gestational BPA exposure with wheeze or lung function. Second, FEV1, which was only available for a subset of the children, may not predict future lung health and may not fully discriminate the effects of BPA. Third, the cohort’s mean %FEV1 was below 100%, which suggests that these children had poorer lung function than the healthy children that were used to generate the reference sample. Fourth, wheezing was based on parent-report and may have been under or over-reported. Fifth, there was differential attrition; minority, low-income families were more likely to drop out of the cohort over time than white, affluent families which could influence the generalizability of our results. Additionally, the cohort was limited to English speaking families. Lastly, these findings may reflect concurrent exposures to other chemicals or other unknown factors that were not accounted for in this analysis.

Conclusion

We found that prenatal BPA exposure that occurred during early pregnancy was inconsistently associated with diminished lung function, increased odds of wheeze, and a persistent wheeze phenotype in young children. Additional research is needed to clarify the contrasting findings in recent human studies. If future studies confirm that prenatal BPA exposure may be a risk factor of impaired respiratory health, it may offer another avenue to prevent the development of asthma.

Abbreviations

AOR

Adjusted Odds Ratio

CDC

Centers for Disease Control and Prevention

CI

Confidence Interval

FEV1

forced expiratory volume in one second

GEE

Generalized Estimating Equations

HOME

Health Outcomes and Measures of the Environment

ICC

Intraclass Correlation Coefficient

L

Liter

LOD

Limit of Detection

OR

Odds Ratio

%FEV1

percent predicted FEV1