Results show BPA to be a pervasive contaminant with 94% or higher detection rates in this study population of inner-city mothers and children. We found that geometric means for maternal prenatal BPA concentrations were significantly lower than paired children’s postnatal concentrations. BPA in pregnant women may be temporarily diverted from the excretory process due to transfer across the placenta. Human placenta does not act as a barrier to BPA (Schönfelder et al. 2002
). Studies of radiolabeled-BPA in pregnant CD1 mice determined that after 24 hours, only 6% of BPA was excreted in urine while a disaccharide conjugate of BPA accounted for 60%, 20% and 10% of radioactivity in placenta, amniotic fluid and fetuses, respectively (Zalko, 2003). In humans, placenta has only a synctiotrophoblast monolayer separating the maternal and fetal blood vessels as opposed to the mouse placenta which has three layers. Early in pregnancy, amniotic fluid is comparable to maternal or fetal plasma and may be composed of secretions from the umbilical cord, placental membrane and developing epithelium (Beall et al, 2007
). Towards the end of pregnancy, an estimated 10 ml/day of amniotic fluid cross into the maternal circulation via the transmembranous pathway (Beall et al. 2007
; Underwood et al. 2005
).While the literature has yet to describe it in humans, there remains the possibility that pregnant women may have less BPA in their urine due to transfer across the placenta. Alternatively, lower total BPA urinary concentration in the mothers than in their children may be related to the fact that diet is the main pathway of exposure to BPA (Morgan et al. 2011
) and children eat more relative to body weight than adults (US EPA, 2011
). Such a trend has been observed for other non-persistent chemicals to which diet is the main source of exposure (CDC 2012
). In contrast, some animal pharmacokinetic evidence suggests that total daily human exposure to BPA cannot be from diet alone and is much higher than previously understood (Taylor et al. 2011
Our unadjusted geometric means are similar to data from two other studies comparing prenatal and childhood BPA concentrations. A European Spanish cohort of mothers and their 4 year old children, reported lower urinary concentrations of BPA in spot urine samples collected prenatally from the mothers as compared to their children [2.2 ng/mL, n=120 vs. 4.2 ng/mL, n=30, p<0.05, respective medians] (Casas et al. 2011
). In a Cincinnati, Ohio study of 240 mother-child dyad spot urine samples, the mean prenatal BPA concentration was 2.0 μg/L and the childhood mean of ages 1, 2, and 3 years was 4.1 μg/L, while the median age 3 year BPA concentration was 2.6 μg/L (Braun et al. 2011b
Our overall prenatalBPA geometric mean of 1.8 ng/ml is in accordance with findings from other investigators. In a US study looking at repeated measures of BPA in 249 women both prenatally and postnatally, Braun et al. (2011b)
found median BPA concentrations of 1.8 g/L (16 weeks), 1.7 g/L (26 weeks) and 1.2 g/L (at birth). Using NHANES data, Woodruff et al. (2011)
, reported a lower geometric mean in pregnant women compared to non-pregnant women (2.53 g/L, n=86 vs. 2.89 g/L, n=489, respectively). After adjusting for creatinine and socio-demographics (age, race/ethnicity, education, smoking, parity, BMI, albumin, duration of fasting prior to sample collection) the BPA concentration disparity increased between pregnant women (1.63 g/L, n=72) and non-pregnant women (2.83 g/L, n=371) (Woodruff et al. 2011
Due to the metabolic characteristics of pregnant women and young children, we controlled for urinary specific gravity as opposed to creatinine in our regression and correlation analyses and this could result in some differences across studies. Renal clearance of creatinine is directly proportional to and dependent on glomerular filtration, such that changes in renal clearance that commonly occur during pregnancy could also cause changes in creatinine concentrations (Jatlow et al. 2003
; Mahalingaiah et al. 2008
). Creatinine, a product of muscle metabolism, relies on the liver and the kidney which are still maturing in children. Childhood levels of creatinine tend to be lower with more variability over time compared to levels in non-elderly adults. Creatinine comparisons in the NHANES III (1988-1994) study describe a mean of 102.1 mg/dL in children ages 6 to 11 years, whereas adult means were consistently higher by decade of age from20 to 49 years (Barr et al. 2005
). Little is known about BPA renal clearance during human pregnancy. Specific gravity is a unit-less ratio of the density of urine to the density of water, dependent on fluid intake, renal perfusion and renal function and thus should be less influenced by muscle mass and muscle retention.
At age 3 years, the children in our study had significantly higher BPA concentrations than they did at follow-up ages 5 and 7 years, a difference that may be related to changes in behavior as children age. While young children tend to have greater oral contact with their world than older children and adults, diet should still be considered the primary route of BPA exposure in young children (Reed et al. 1999
). It is known that children between the ages of 3 and 6 years eat more relative to body weight than adults (US EPA, 2011
). Additionally, children under the age of 6 years tend to have less varied diets (Goldman, 1995
). Diet in pre-school children may be more dependent on the feeding behaviors of their mothers whereas school-aged children are able to make food choices outside the presence of their mothers. Adult pregnant female eating behaviors are also likely to be different than those of children.
In our study population, African Americans had consistently significantly higher BPA concentrations than Dominican Americans, except among 3 year olds where BPA was elevated for African Americans but not significantly. Additionally we saw higher concentrations among older children of never married mothers compared to older children of ever married mothers. This suggests that BPA concentrations may vary by socioeconomic status. However, we did not see any differences based on household income nor maternal education. In contrast, other studies have found BPA concentration to be negatively associated with income and maternal education (Calafat et al. 2008
; Braun et al. 2011b
). Little is known about how dietary differences across race/ethnicity and maternal characteristics may affect BPA exposures in pregnancy or childhood. Our results have implications for guiding socio-demographic considerations in future health effects studies of BPA exposures.
We found postnatal BPA concentrations to be higher during the summer months than during the rest of the year. This could be due to different dietary patterns among children during the summer relative to other seasons. It could also be related to diurnal variation in renal clearance of BPA by season. This finding is consistent with previously reported BPA urinary geometric means in a population of prepubertal girls from three sites in the US, including New York City (Wolff et al. 2010
Urinary concentrations of BPAand phthalate metabolites were correlated. These findings are similar to Braun et al. (2011a)
in the US but are in contrast to those of Casas et al. (2011)
in Spain. Given the ubiquitous presence of both BPA and phthalates in the environment, it is not surprising that they would both be widely detectable in our study population. However, the observation of correlations between BPA concentrations and these phthalate metabolites highlights the complex issue of analyzing the health effects of chemical mixtures in people.
We saw a consistent pattern of BPA exposure for race/ethnicity, marital status and season of collection despite no correlation between BPA concentrations in repeat samples. High variability and poor interclass correlations for BPA concentrations in spot urine samples have been reported by others, however spot urine samples may adequately reflect BPA exposure at the group level (Braun et al. 2011a
; Teitelbaum et al. 2008
; Ye et al. 2011). If, as expected, the errors in measures of BPA concentrations are non-differential, when BPA concentrations are analyzed as a dependent variable in the regression models, the standard errors for the regression coefficients are expected to be increased, resulting in lower statistical power (Cook and Campbell, 1979
). Our significant findings indicate it is likely the differences are greater than we can detect due to the poor reliability of the biomarker. Regardless, our use of single spot urines for BPA analysis provides clear evidence that the children in our study population are exposed to BPA between ages 3 and 7 years. Cumulative exposure effects on health and development beginning prenatally and over a lifetime are a concern which should be investigated further.
Limitations of this study include the use of single spot urines and the lack of recording time of day for urine collection. The data on time of urine collection are not available for the prenatal samples or the age 3 samples, and are only available for a minority of the samples collected at ages 5 and 7. Diurnal variations in urinary BPA concentrations have been reported and it is possible collection time may confound our results. We also did not previously collect any dietary data but we have begun administering food frequency assessments among cohort children at older ages and these may give insight on dietary behaviors in future analyses.