In this case-control study, we investigated the association between prenatal exposure to consumer product chemicals and the odds of being born SGA or LGA, using both single- and multi-pollutant approaches. We observed inverse associations between individual biomarkers of exposure to OPEs, phthalates, and phenols and LGA births. Furthermore, results from our mixtures analyses showed potentially strong joint effects of overall chemical classes on odds of LGA. Specifically, the overall mixture of OPE metabolites as well as that of phthalate metabolites were associated with 2–5 fold reductions in the odds of LGA. Associations between consumer product chemicals and SGA were considerably more mixed and closer to the null. These results suggest that exposure to some consumer product chemicals may reduce fetal growth, but that effects may only be observed on the high end of the birth weight distribution.
This is among the first studies to examine the relationship between OPE exposure and birth weight. We observed that levels of urinary DPhP as well as the mixture of OPE metabolites were associated with reduced odds of LGA births (i.e., lower birth weight) although there was no association with SGA. These findings are consistent with two other epidemiologic studies that indicate a potential link between OPE exposure and altered fetal growth (17, 18). For example, a recent prospective case-control study of low birth weight (n = 339) based in China reported that the odds of low birth weight in women with the highest tertile of urinary DPhP was 4.62 (95% CI: 1.72, 12.40) times that observed in the lowest tertile (17).
Toxicologic evidence also supports the biologic plausibility of a relationship between OPE exposure and fetal growth. One study in mice demonstrated that in utero exposure to triphenyl phosphate (TPhP), the parent compound of DPhP, alters fetal and maternal liver insulin growth factor signaling, which is critical to the control of fetal growth and development (49, 50). Additional research using a human placental cell line has also established that TPhP exposure increases both progesterone and human chorionic gonadotropin secretion via activation of the peroxisome proliferator-activated receptor gamma (51). Changes to placental hormone production has important implications for both placental and fetal growth and development (50). Taken together, these findings clearly indicate the potential for prenatal OPE exposure to affect these processes.
Out of the 12 phthalate metabolites included in this analysis, MEP was most strongly associated with lower odds of LGA births both in single-pollutant models and within quantile g-computation models for the mixture of phthalate metabolites. In support of this observation, a previous prospective study reported inverse associations between maternal urinary MEP concentrations, measured repeatedly across gestation, and birth weight (52). However, the overall literature on prenatal phthalate exposure and fetal growth is inconsistent with respect to both the implicated phthalate species and the direction of effect (1). This is also true for studies that have investigated the potential for sex-specific effects of phthalates in relation to fetal growth outcomes (1). In this study, we observed suggestive sex-specific effects for several DEHP metabolites in relation to SGA births. Specifically, DEHP metabolites were associated with higher odds of SGA among female infants compared to males. However, this contrasts with the prevailing hypothesis that males are more susceptible to in utero environmental exposures than females (53). Given the small sample size and wide confidence intervals for our sex-specific results, we caution against their over-interpretation.
Among the phenols included in this study, MPB and PPB were both highly detected in maternal urine and inversely associated with LGA. While quantile g-computation models did not identify a significant association with the mixture of phenols and LGA births (OR: 0.54 [95% CI: 0.19, 1.56]), MPB was assigned the largest weight for the partial negative effect in the mixture. Previous studies examining prenatal paraben exposure have reported inconsistent associations with fetal growth, with varying estimated directions and magnitude of effect (19–21, 54–56). More broadly, however, parabens are suspected to have effects on body composition and altered weight (57–61). For example, a study in the National Health and Nutrition Examination Survey (NHANES) reported that higher urinary parabens were associated with lower BMI in both children and adults (59). Reports of pro- or anti-obesogenic properties of parabens may have relevance to our findings because LGA infants have disproportionately higher total body fat compared to AGA infants (62).
While we observed several inverse associations between consumer product chemicals and LGA births, our results with SGA were much more variable and closer to the null. Babies born LGA are larger than AGA babies in numerous respects, including birth weight, length, and head circumference. However, LGA infants are also characterized by a disproportionate increase in total body fat and a decrease in lean body mass (62). This phenotypic difference suggests that whether a baby is born LGA is controlled, at least in part, by mechanisms related to maternal-fetal nutrient transfer and adipogenesis (63, 64). Therefore, chemical exposures that impact these processes should be considered in relation to fetal overgrowth. On the other end of the spectrum, SGA captures a diverse set of infants that include both those who are pathologically small (i.e., growth restricted) and those who are constitutionally small (65, 66). We expect that adverse chemical exposures during pregnancy would be related to growth restriction rather than small, but normal fetal growth. However, infants experiencing pathological growth restriction may be a minority of infants classified as SGA, particularly when considering births at term (i.e., > 37 weeks gestation) (65). Without being able to distinguish between these sets of infants, our SGA results may be biased towards the null. A better approach in future studies may be the use of a clinical definition for intrauterine growth restriction or fetal growth trajectories to appropriately isolate pathologically small infants from constitutionally small infants.
This study is not without other limitations. First, this study has a relatively small sample size. This prevented us from examining SGA births more closely to isolate growth restricted infants from constitutionally small infants. It also limited our ability to explore heterogeneity with greater precision or to examine non-linearity of the dose-response relationship. Second, we were lacking information on several potentially important confounders. For example, recent studies have shown that parabens sequester into adipocytes (67–69). This sequestration may explain inverse associations between paraben exposure and body weight measures, including birth weight. In the context of this study, maternal adiposity is an important risk factor for LGA births (70, 71). However, it is possible that we were unable to fully account for this potential confounding by solely adjusting for maternal pre-pregnancy BMI as other measure measures may better account for maternal adiposity. In addition, seasonality is another potential confounder as season has been shown to be a strong predictor of exposure to some consumer product chemicals, such as OPEs, and may be related to changes in fetal growth (48, 72). Lastly, although this study is among the first to consider associations between OPEs and fetal growth, only two out of nine selected flame retardants (i.e. DPhP and BDCPP) were detected in maternal urine frequently enough for analysis.
Despite these limitations, this study also had several strengths relative to prior studies. First, we were able to assess prenatal exposure to consumer product chemicals using measurements from up to three study visits during pregnancy. As this study and others have demonstrated, the stability of consumer product chemical concentrations in urine is often low given their short half-lives (10, 11, 48). Yet, many studies continue to rely on a single spot urine sample to assess prenatal exposure to these chemicals (1), and therefore, may be more susceptible to bias due to exposure misclassification (73). Because this study uses averaged exposure measures from up to three study visits, it may be less susceptible to such misclassification. Second, this study examined consumer product chemical exposure in relation to both SGA and LGA births. While many studies have considered the potential for chemical exposures to reduce fetal growth, few include outcomes related to fetal overgrowth (1). By including both SGA and LGA, this study identified associations at one end of the birth weight distribution (i.e. LGA), but not the other (i.e. SGA). Lastly, we went beyond single-pollutant models to estimate the joint effect of consumer product chemical mixtures on fetal growth (46). Estimating the joint effects of consumer product chemicals is of particular interest because behavioral or regulatory changes to the use of consumer products would likely simultaneously affect exposure to classes of chemicals, rather than individual chemicals. Thus, joint effects correspond better to possible real-world actions, while simultaneously controlling for confounding between elements of the mixture.