∑PBDE concentrations ranged from 1 to 2,010 ng/g lipid in this large study of North Carolina breast milk samples collected between 2004 and 2006. These concentrations are higher than milk concentrations reported by most similarly timed studies from the United States (Johnson-Restrepo et al. 2007
; Schecter et al. 2003
; She et al. 2007
; Wu et al. 2007
) and considerably higher than those in reports from abroad (reviewed by Erdogrul et al. 2004
; Frederiksen et al. 2009
; Johnson-Restrepo et al. 2007
; Lind et al. 2003
; Toms et al. 2007
). Concentrations in milk samples from this study were also higher, comparing lipid-adjusted concentrations (nanograms per gram milk lipid), than were corresponding concentrations in blood samples provided by mostly African-American pregnant women in a Maryland study during the same time period (Herbstman et al. 2007
). The higher concentrations reported in these North Carolina samples could have resulted from a slightly younger population, regional differences in regulation and use of PBDEs across the United States, or simply a greater range revealed by the larger study size. Differences are not likely to be due to dramatic temporal increases over such a short period of time.
As reported in other studies, BDE-47 was present in highest concentration, followed by BDEs 99, 100, and 153, which we detected in nearly all samples. We detected BDEs 66, 85, 154, and 183 in < 70% of the samples. In this study, ΣPBDE included only congeners present in > 70% of the samples. This method is somewhat conservative and may underestimate total exposure, but minimizes assumptions inherent when summing data below the LOD. PBDE concentrations were not correlated with PCB or DDE concentrations, which supports reports suggesting that sources of exposure for these persistent pollutants are different (Frederiksen et al. 2009
; Herbstman et al. 2007
; Kiviranta et al. 2004
; Schecter et al. 2005
PBDE concentrations in milk varied by age and BMI but were not affected by parity or race. We had limited power to detect variability by race because the study population was predominantly white. We also lacked information needed to evaluate the effect of breast-feeding history among parous women, although it seems likely that most women breast-feeding their second child also breast-fed their first. When adjusted for other covariates, we observed the lower milk concentration among the oldest women compared with those 25–29 years of age for all congeners except BDE-28. Milk concentrations were significantly higher among obese women compared with those of normal weight for BDEs 47, 99, and 100, but were also slightly higher in milk from women with low BMI. The variability by weight and age, and across congeners, is similar to observations of prenatal serum samples from Maryland women (Herbstman et al. 2007
) and general-population serum samples from the 2003–2004 National Health and Nutrition Examination Survey (Sjodin et al. 2008b
). Higher concentrations among women in younger age groups may suggest that the environment and habits of younger generations expose them to higher levels of PBDEs than do those of women ≥ 35 years of age. Higher concentrations among heavier women may reflect the lipophilic nature of the chemicals. The slightly different pattern for BDE-153 may reflect differences in exposure sources and/or metabolism compared with other congeners.
We assessed variability by individual characteristics only among the congeners detected in > 70% of the participants. We were unable to measure BDE-209, the primary congener of the decaBDE formulation, which is the only brominated flame retardant still produced in the United States. BDE-209 is stable but less likely to bioaccumulate and be detected at remarkable levels in human tissue compared with the lower brominated congeners because of its short half-life (i.e., 2 weeks in humans) (Sjodin et al. 2003
; Thuresson et al. 2006
). According to the Agency for Toxic Substances and Disease Registry (2004)
, “the lower brominated PBDEs are much more likely than decaBDE to be stored in the mother’s body and released during pregnancy, cross the placenta, and enter fetal tissues. Because lower brominated PBDEs dissolve readily in fat, they can accumulate in breast milk fat and be transferred to babies and young children.”
For infants, the primary source of exposure to PBDEs is breast milk. Darnerud et al. (2001)
estimated that infant exposure in Sweden in the late 1990s through breast milk averaged 0.11 μg/day for ΣPBDE, based on Swedish data assuming 3.7% fat in milk and 0.7 L/day consumption. The actual volume of milk consumed by the children in our study has not been determined; however, if we apply the same algorithm and substitute the median concentration observed in the 3-month samples from North Carolina, median daily exposure to infants through breast milk would be considerably higher: 0.8 μg BDE-47 and 1.5 μg ΣPBDEs.
Among women with samples at 3 and 12 months postpartum, the concentrations in milk were highly correlated and, for most congeners, did not significantly change. The range of change among all congeners was very large and tended to be in a positive (increasing) direction. Our results from this large sample generally differ from those of a small case series studying depuration, which reported subtle decreases in concentration over the course of lactation among eight women (Hooper et al. 2007
). For BDE-153, however, the median concentration in our data significantly increased between 3 and 12 months postpartum. It is unclear why BDE-153 increased whereas other congeners did not. It may reflect variability in the distribution and skew of the change among congeners, which can affect the statistical power to detect change. It could also reflect different levels of continuous exposure over the postpartum period among the congeners. However, it is also important to consider possible variability in the metabolism and storage of the congeners. At 3 months postpartum, retention of excess fat and fluid from pregnancy may dilute the circulating chemical concentration and the corresponding concentration in milk samples. By 12 months, women have less pregnancy-associated fat and fluid retention, so some chemicals mobilized from fat stores could become more concentrated in milk. Differences in the level of change among congeners over the course of lactation may reflect variability in the metabolism, storage, and release of particular congeners from fat. Studies of other pollutants have suggested that weight loss increases chemical concentrations measured in fat tissue (Chevrier et al. 2000
; Pelletier et al. 2003
). In this cohort, neither maternal weight gain during pregnancy nor percentage of body fat was associated with PBDE concentration in milk. Furthermore, changes in the woman’s percentage of body fat postpartum were not associated with changes in milk PBDE concentration.
All women in this study breast-fed at least 3 months, which was required to enter the study. Among these women, PBDE concentration was not associated with breast-feeding duration through 12 months postpartum. However, we were unable to assess whether exposure to PBDEs could impair a woman’s ability to initiate breast-feeding or decrease milk supply enough cause her to discontinue breast-feeding before 3 months postpartum, when women entered our study.
This is the largest study of breast milk PBDE concentration in the United States to date and the only one that evaluated differences by individual characteristics such as age, BMI, parity, and duration of breast-feeding. However, this study was limited to mostly Caucasian, well-educated mothers from central North Carolina. Other U.S. studies of PBDE concentrations in breast milk also included predominantly white and well-educated populations (She et al. 2007
; Wu et al. 2007
), which is reflective of breast-feeding women in the United States (van Rossem et al. 2009
). Although there is no reason to suspect that this North Carolina population would systematically differ from others, the data should be interpreted as regional and not representative of the United States.
The persistence of PBDEs raises our vigilance for potential adverse environmental and human health consequences. Similar persistent organic pollutants such as p
-DDE and PCBs have been extensively investigated over the past 20–30 years for reproductive and developmental toxicity (Branchi et al. 2003
; Tilson and Kodavanti 1997
; Tilson et al. 1990
), resulting in restrictions on many compounds in the United States (Cordle et al. 1982
). Although levels of these other organic pollutants are steadily decreasing, levels of PBDEs are still elevated in the United States compared with other countries with limited recent use.
The consequences of exposure to PBDEs are unknown. Rodent studies have suggested that perinatal exposure to PBDEs may cause learning delays and behavioral problems that persist into adulthood, possibly worsening with age (Eriksson et al. 2001a
; Rice et al. 2007
; Viberg et al. 2002
). Most rodent studies have focused on the pentaBDE congeners, exposing mice during gestation and soon after birth to various doses.
To date, little information is available to assess human health effects. Given the extensive nutritional and immunologic benefits of breast milk, promotion of breast-feeding as the best method for infant feeding should continue (Oddy 2001
; Schack-Nielsen and Michaelsen 2007
). However, timely studies are needed to continue to monitor levels, to evaluate health effects, and to minimize exposure to all pervasive, persistent pollutants (Hooper and She 2003
Ideally, as PBDE containing products exit the market, levels will begin to decline in the United States, as they have in Europe and other areas where use ended in the late 1990s (Lind et al. 2003
). However, because household products containing PBDEs will continue to be in use for some time, it is important to observe generations with greater exposure and monitor whether adverse health outcomes are associated with high-level exposures, especially among the most vulnerable populations. Human milk samples offer a minimally invasive opportunity to monitor population level trends in PBDE exposure in the coming years (Hooper and She 2003
; Meironyte et al. 1999
; Wu et al. 2009