During enrolment, of the approximately 470 women who visited the surgery clinic for breast-related health issues, 203 were enrolled into the study. This paper presents results of serum and urine samples collected at the time of enrolment, limited to the 170 women (75 cases, 95 controls) for whom we had sufficient quantities for laboratory analyses. The 33 women who were excluded from analyses did not differ with respect to demographic variables from the 170 women who were included (data not shown).

Descriptive characteristics are shown in Table I. Age at enrolment ranged from 30 to 88 years, with a mean age of 51 years. Women of Eskimo (54.7%), Indian (21.8%), and Aleut (15.9%) ethnicities were represented. All 5 regions of the state were represented within the study population. The majority (58.2%) of participants resided in Alaska's south central region at the time of enrolment. The fewest lived in the southeast (5.9%) and the interior (1.2%) regions. Almost one-fifth (17.6%) of women reported a family history of breast cancer. Almost all (92.4%) of the women reported having children. A history of smoking was reported by 77.6% of the women; however, at the time of enrolment, only 59.3% of respondents were current smokers. Similarly, 90% of the women reported ever drinking alcohol, but only 51.3% of respondents reported any alcohol use at the time of enrolment.

Case women reported a significantly lower median cumulative number of months of breast feeding (2 months) than control women (8 months). The variable “expecting more periods” was used as a proxy measure for menopausal status – women who responded “no” were assumed to be menopausal. More case women (63.9%) responded “no, they were not expecting more menstrual periods” than control women (45.6%). That is, more case women were menopausal than control women, a statistically significant difference (p=0.03). No other descriptive variables differed significantly between cases and controls.

Sixty-two of 75 (83%) case women had invasive cancer tumours. We compared the descriptive characteristics and known risk factors for breast cancer shown in Table I between case women with invasive vs. non-invasive cancers and found no significant differences between them for these variables (data not shown).

GM concentrations and confidence intervals for persistent pesticides, indicator PCBs, and BDE-47 measured in serum, and phthalate metabolites measured in urine are shown in Table II. GM serum concentrations of most pesticides were lower in case women than in controls. Only gamma (γ)-hexachlorocyclohexane was higher in cases (8.3 ng/g lipid) than controls (7.7 ng/g lipid); however, this difference was not statistically significant. The GM serum concentrations of all 3 indicator PCBs were also lower among cases than controls, although again, these differences were not statistically significant. The GM serum concentration of BDE-47 was statistically significantly higher in case women (38.84 ng/g lipid) than in controls (25.11 ng/g lipid) (p=0.04). Although GM urinary concentrations for most of the 7 phthalate metabolites were higher among cases than controls, these differences were not statistically significant.

Results of the univariate logistic regression analyses showed no association between the median concentrations of persistent pesticides, PCBs, or most of the phthalate metabolites and breast cancer status (Table III). However, urinary concentrations above the median of mono-(2-ethylhexyl) phthalate (MEHP) were associated with breast cancer, and the association was statistically significant in both univariate and multivariable analyses (OR 2.16, 95% CI 1.16–4.05, p=0.02 and OR_adj 2.43, 95% CI 1.13–5.25, p=0.02, respectively). The result of univariate logistic regression analysis of BDE-47 approached statistical significance (OR 1.79, 95% CI=0.97–3.32; p=0.06). However, this association was not statistically significant in the multivariable analysis. Conversely, the result of the same analysis for p,p′-DDT suggested a slight protective effect (OR 0.55, 95% CI 0.30–1.01, p=0.05). No significant associations between continuous concentrations of environmental chemicals and breast cancer status were found.

Of the 62 women who had invasive tumours, 40 (65%) had tumours that were positive for both oestrogen- and progesterone-binding capacity (ER+/PR+), while 13 (17.3%) were negative for both hormone markers (ER–/PR–). Women with ER–/PR– tumours (n=13) had GM serum concentrations of persistent pesticides, PCBs, and BDE-47 that were consistently (7.7–48%) higher, although not significantly different, than women with ER+/PR+ tumours (n=40). Urine concentrations were higher in 3 of 7 phthalate metabolites among women with ER–/PR– tumours. The differences were not statistically significant (Table IV). We estimated the odds ratios for the ER+ and ER– groups, modelling the probability of ER–. While results for MEHP were not significant, the estimates indicated a positive association.

The scientific literature contains conflicting reports about the association between exposure to persistent organic pollutants and breast cancer (4–16). Results of some studies demonstrated an association between the DDT metabolite DDE and breast cancer (7–9), while others did not (10–14). Studies of exposure to PCBs and breast cancer incidence also reported conflicting results (7–9,12,15,16). Few studies have focused on the association between exposure to environmental contaminants and breast cancer specifically among AN women, and only one epidemiologic study was found that examined the relation between PBDE and breast cancer in any population (21). Rubin et al. (4) confirmed DDT and PCB exposure in AN women, but did not find that exposures to these chemicals were associated with breast cancer. The study did not include PBDEs or phthalates. Day et al. (32) suggested that increases of breast cancer among AN women over the past 30 years may be attributable to increased body mass index (BMI) and a shift away from a traditional diet.

We believe this is the first study to measure and report PBDE serum concentrations among AN women and the first to suggest a possible association between breast cancer and PBDE exposure in this population. The serum concentrations of BDE-47 measured in our participants ranged from 0.71 to 1,370 ng/g lipid, and the GM concentration of BDE-47 was 30.4 ng/g lipid (95% CI 24.9–37.2). AN women have GM serum concentrations of BDE-47 similar to concentrations previously reported in studies from other areas of the Unites States.

A representative sample of US women (22) had a lipid-adjusted GM serum BDE-47 concentration of 19.6 ng/g lipid (95% CI 16.4–23.5). AN women in our study had GM concentrations of BDE-47 that more closely approximated concentrations at the 75th percentile (38.4 ng/g lipid; 95% CI 31.8–46.4) of the US sample (22). BDE-47 concentrations measured in 50 serum samples collected in the 1990s from Laotian women living near San Francisco ranged from less than the LOD (10 ng/g lipid) to 511 ng/g lipid (mean 50.6 ng/g lipid; median 16.5 ng/g lipid) (33). The concentration of BDE-47 measured in a 2005 study of pooled serum from 100 discarded anonymous US samples collected in 2003 was 32.5 ng/g lipid (34).

A recent case–control study that measured concentrations of the 5 major PBDE congeners, including BDE-47, in adipose tissue collected from 134 women in the San Francisco Bay Area found no association between exposure to PBDEs and breast cancer (21). In that study population, the median concentration of BDE-47 measured in adipose was 22.0 ng/g lipid. For case women, the concentration was 18.83 ng/g lipid, and for control women, it was 26.85 ng/g lipid. The authors state that these adipose concentrations of PBDE were among the highest ever reported. The GM concentrations of BDE-47 measured in serum in our study population (30.4 ng/g lipid) and among our case women (38.8 ng/g lipid, respectively) were higher than concentrations reported in adipose.

The results of our study suggest a possible association between breast cancer and BDE-47 (OR 1.79, p=0.06). However, the association is not supported in a multivariable analysis of BDE-47 with other known risk factors including age, menopause status, and having a relative with breast cancer. Our data seem to show a protective effect from exposure to p,p′–DDT. This association has not been reported previously. It is possible that the result is spurious, and should be interpreted with care.

No reports describing phthalate exposure among AN people were found in the literature. Our results suggested an association between urinary concentrations dichotomized at the median of the phthalate metabolite MEHP and breast cancer, an association that held in both univariate and multivariable analyses. However, MEHP is a minor metabolite of DEHP, and is the metabolite from which all others are formed. Therefore, we would expect to see associations with the other metabolites as well. The fact that associations were not found among the other DEHP metabolites may suggest the association between MEHP and breast cancer is a chance finding. Furthermore, the concentrations of phthalate metabolites were measured in a single spot urine sample. Urinary concentrations of some phthalate metabolites, including MEHP, can be highly variable over time (35). Although this variability could result in exposure misclassification, it would likely be non-differential with respect to the outcome examined, significantly reducing our ability to observe an effect of DEHP on cancer risk. For these reasons, the findings regarding MEHP should be interpreted with caution.

We found only one study that reported an association between exposure to phthalates and breast cancer (23). A case–control study of more than 450 Mexican women found that case women had higher concentrations (169.58 µg/g creatinine) of the phthalate metabolite MEP than did control women (106.78 µg/g creatinine) – concentrations that were almost double those of MEP among our study population – and that MEP concentrations were associated with breast cancer.

A recent in vitro study implicated phthalates as possible potentiators in breast carcinogenesis (36). One study examined the effect of 2 phthalates – neither of them MEHP – on ER-negative breast cancer cells. The authors reported that the exposure induced cell proliferation, migration, invasion, and tumour formation (37). Animal studies have demonstrated metabolic differences between species, necessitating human studies to determine health effects (22,29). Temporal differences in human exposure, inconsistent study design, and differing methods, however, appear to account for wide variability in findings reported from studies investigating human health effects of phthalate exposure (22,35,38).

It is widely accepted that oestrogen is an important contributor to breast cancer development and progression (39), an effect demonstrated best in studies of breast cancer among males; men who develop breast cancer have higher than normal levels of oestrogen (40). Reproduction-related characteristics, such as early age at menarche, later age at bearing first child, and postmenopausal obesity are associated with ER+ tumours, which respond to treatment with hormone therapy (41,42). ER– tumours are not influenced by oestrogen and do not respond to treatment by hormone therapy, are associated with higher mortality (43), and are believed by some to have different promoters and aetiologies than ER+ tumours (41,44,45). An ecologic study based on county level breast cancer and exposure data concluded that ER+ tumours could be influenced by environmental factors with endocrine disrupting capabilities (44). Høyer et al., in a nested case–control study of Danish women, concluded the opposite: that exposure to potential estrogenic organochlorines does not lead to developing ER+ breast cancer type. This study further reported an association between dieldrin and incidence of ER– tumours (46). Although the differences were not statistically significant, in our study, women with ER–/PR– tumours type had consistently higher concentrations of the persistent pesticides we measured, than ER +/PR+ women.

Limitations of this study included a small sample size, which may have decreased our power to detect statistical differences, and the inability to include all potential confounders (e.g. BMI) in the analyses. Another limitation is the potential exposure misclassification due to the short half-life of phthalates, which can result in high variability of phthalate metabolites, particularly MEHP (35). However, any residual misclassification would be non-differential and bias results towards the null. Furthermore, our controls were women with benign breast disease. It is possible that one or more of the environmental chemicals we measured could be associated with benign breast disease, along with breast cancer. This could have resulted in some of our null findings. Future research should attempt to measure environmental chemical exposures in the general AN population.

AN women may be exposed to types and concentrations of environmental chemicals that are different from other populations of women in the US. Studies of DDT or PCBs have not demonstrated with any certainty that exposure to these chemicals results in increased risk of developing breast cancer. The health effects of exposure to relatively new environmental chemicals, such as PBDE and phthalates, are not well understood. The biologic plausibility exists for these chemicals to affect reproductive and other hormonally-regulated systems, as they are known endocrine disruptors. Larger population-based studies of AN people would provide more statistical power to better determine possible association between exposure to contaminants, in particular those with endocrine disrupting capabilities, and the risk of breast cancer.