Of the 19 phenol and phthalate metabolites measured in this study, two showed higher concentrations than those reported in other U.S. populations: 2,5-DCP [median, 54 μg/L in our study vs. 30 μg/L as reported by Hill et al. (1995)
] and MEP [median, 380 μg/L vs. 178 μg/L among female participants of all ages in the 1999–2000 National Health and Nutrition Examination Survey (NHANES) (Centers for Disease Control and Prevention 2005
)]. 2,5-DCP was also relatively high in a population of New York City minority children compared with those at two other sites in the United States (Wolff et al. 2007b
). Total phthalate biomarker concentrations were also relatively high in this study, approaching 1 mg/L total (~3 μM). Our population has a large proportion of minority women, and therefore the levels are consistent with those seen in NHANES data where several phthalate biomarkers were elevated among blacks and Hispanics compared with whites (Centers for Disease Control and Prevention 2005
). Concentrations of BPA were relatively low in this population as in other recent reports of nonoccupational exposures (Kuklenyik et al. 2003
; Liu et al. 2005
; Matsumoto et al. 2003
; Ouchi and Watanabe 2002
; Wolff et al. 2007b
; Ye et al. 2005
Environmental sources of phenols and their precursors include personal care and home cleaning products. 1,4-DCB is used in mothballs and in room deodorizers; it is metabolized to 2,5-DCP. The high correlation of 2,4-DCP with 2,5-DCP suggests that 2,4-DCP is a metabolite of 1,3-dichloro-benzene, a minor contaminant of 1,4-DCB (National Toxicology Program 2005
). TCS is a microbicide, and BP3 exposure comes mainly from sunscreen. Environmental sources of phthalates are numerous. MEP and MBP are found in cosmetics, shampoo, perfume, and products with fragrance. The higher-molecular-weight phthalates, including DEHP and butylbenzylphthalate, are found in soft plastics, vinyl wrap, plastic tubing, and home construction components such as vinyl floor tile.
We observed sex-specific associations of phenols with birth weight and length. Third-trimester 2,5-DCP exposure was associated with lower birth weight among male infants, and BP3 was associated with lower birth weight among female infants. For both biomarkers, the third versus first tertile of prenatal phenols predicted about 200-g-lower difference; this deficit is comparable to the reduction in birth weight seen for active smoking during pregnancy (Bernstein et al. 2005
). This difference is also similar to that between males and females at birth, where females are 135 g (median) lighter than males at 39 weeks of gestation (Oken et al. 2003
Like 2,5-DCP, TCS had sex-specific inverse but nonsignificant associations with birth weight and length among boy infants in this cohort. TCS is 5-chloro-2-(2,4-dichloro-phenoxy) phenol, and thus it is structurally similar to 2,5-DCP. Our finding of increased male birth weight with higher maternal BP3 concentrations is unexpected and has no clear biologic basis. Although BP3 levels were higher in whites, consistent with putative use of sunscreen, the associations of BP3 with birth weight did not differ by race/ethnicity in this study. We saw no effects with BPA in our study, but BPA urinary concentrations were much lower than those of 2,5-DCP, TCS, and BP3 and may not have reached a level of biologic significance.
Pregnant ewes treated with BPA had offspring with reduced birth weights, and their blood levels were greater than 35 μg/L on average, with adipose concentrations of 200 mg/kg (Savabieasfahani et al. 2006
). Experimental findings for other phenols are consistent with our results, supporting a possible mechanism for reduced birth weight in boys prenatally exposed to 1,4-DCB or girls to BP3. In rats, 1,4-DCB reduced body weight at high doses (30–270 mg/kg; Bornatowicz et al. 1994
). In addition, 1,4-DCB is an animal carcinogen and “reasonably anticipated to be a human carcinogen” by the National Toxicology Program and “possibly carcinogenic” by the International Agency for Research on Cancer (National Toxicology Program 2005
). 1,4-DCB is also a respiratory toxin (Elliott et al. 2006
) and was banned in schools in New York State in 2004 because of potential to exacerbate childhood asthma and in California in 2006 for use as room deodorizers.
In contrast to our hypothesis of an inverse effect of phthalate exposure on birth size and gestation, we found a positive association of low-MWP biomarkers with duration of pregnancy and infant head circumference. Effect sizes were small: < 1 day longer gestation per ln-biomarker and 2.8 days between the third and first tertiles of low-MWP biomarkers. Similar but nonsignificant effects on gestational age were found for the DEHP-MWP and high-MWP biomarkers. Our maternal exposures may be too low to elicit the inverse effects we hypothesized based on the birth weight reductions reported in rodents. The lower cut point of the third tertiles were 0.01 μmol/L for BPA, 0.5 μmol/L for 2,5-DCP, 0.4 μmol/L for DEHP-MWP biomarkers, and 3.9 μmol/L for low-MWP biomarkers. Moreover, the effect sizes we observed were modest enough that residual confounding resulting from poorly measured maternal anthropometric features may account for these findings.
In humans, associations have been reported between prenatal and early postnatal phthalate exposures and shorter anogenital distance as well as lower serum testosterone in newborns (Main et al. 2006
; Swan et al. 2005
). In addition, shorter gestational age was associated with cord serum concentrations of DEHP and its metabolite mono-2-ethylhexyl phthalate (MEHP) (Latini et al. 2003
). It is possible that in these studies exposures were higher than in our population, because MEHP levels in serum were slightly greater than 1,000 μg/L on average, which would be comparable to higher urinary concentrations than we observed. Other pre-natal exposure biomarkers have been associated with reduced gestational age (Fenster and Eskenazi 2006
), and both positive and negative associations with head circumference have also been reported (Apelberg et al. 2007
; Eskenazi et al. 2004
; Wolff et al. 2007a
). However, increased weight and lengthened gestation as a result of androgen antagonist exposures have not been reported in children.
Limited support exists for a hormonal mechanism for both shorter and longer gestation following phthalate exposures in animals, depending on dose. DEHP in rats has been reported to cause both longer (Dalgaard et al. 2003
) and shorter (Marsman 1995
) gestational age. Low perinatal exposure can be androgenic in male rats (earlier puberty), but can have the opposite effect at high doses compared with controls (Ge et al. 2007
). High doses, in these and other studies, exceeded 100–3,000 mg/kg/day. Dibutyl phthalate, the precursor of MBP, has estrogenic effects in vitro
at levels typically found for environmental estrogens, including BPA (van Meeuwen et al. 2007
Overall, for both phenols and phthalates, we found few significant associations in this study; for example, the findings in could be attributable to multiple comparisons (five associations at p
< 0.05 among 72 comparisons). An additional limitation is that we had biomarkers measured once in the third trimester for exposures that ordinarily have relatively short half-lives (days). Consistency in levels during pregnancy has been observed for some environmental exposure biomarkers (Longnecker et al. 1999
; Muckle et al. 2001
), whereas pesticide levels, with ambient exposures that are likely to be sporadic, show more variability (Bradman et al. 2005
). However, research in other populations has suggested that phthalate biomarkers are relatively stable for a period of weeks to months (Hauser et al. 2004
; Hoppin et al. 2002
; Teitelbaum et al. 2007
); less is known for phenol biomarkers, but they also appear to have adequate stability to predict exposure over 6–12 months in children (Teitelbaum et al. 2007
). It is reasonable that the biomarkers we describe here have modest intraindividual variability, because use of common products that result in these exposures may be fairly constant over days or months. Nevertheless, to more fully understand relationships between exposures with short half-lives and health outcomes, it may be necessary to investigate additional methods of exposure assessment, especially ones that might offer a more comprehensive and integrated picture of the individual environment, perhaps by evaluating specific products use over long period of time in conjunction with indoor air levels as well as biomarkers of exposure.
Creatinine correction is commonly used for urinary biomarkers of phthalates, pesticides, phenols, and phytoestrogens. There are limitations to the use of creatinine to normalize for urine dilution; other investigators have used specific gravity instead of creatinine to adjust phthalate urinary biomarkers for urine dilution, but we did not have specific gravity measurements. However, specific gravity is highly correlated with creatinine (Barr et al. 2005
), and therefore it is not likely that we overcorrected for urine dilution, especially because we discarded results from very dilute urines. In addition, parameters in our models were little changed by creatinine-corrected values (micrograms per gram creatinine) versus uncorrected values (micrograms per liter) for the biomarkers or by adjustment for creatinine as a covariate in the multivariate models.
The exposures we studied are relatively prevalent, and some biomarker levels approach those with significant effects in experimental models. In a healthy cohort such as ours, effects of hormonally active environmental exposures on birth size may be small, yet more sensitive end points such as infant neurologic development may be affected. A further dimension to consider in future research is multiple exposures of hormonally active agents such as these. In terms of prevention, exposure to these chemicals can be avoided if the product contents are known; unfortunately, they often are not listed on the label because they are not “active” ingredients.