Breast Cancer Risk in Relation to Ambient Air Pollution Exposure at Residences in the Sister Study Cohort Free

Several studies suggest an association between breast cancer risk and exposure to ambient fine-particulate matter (PM_2.5) and nitrogen dioxide (NO₂), a marker of traffic-related air pollution (1–3). Most notably, a 2010 case–control study reported a 1.3-fold increased risk of breast cancer [95% confidence intervals (CI) of OR: 1.0–1.7] for every 5 ppb increase in NO₂ assessed via a land-use regression spatial model (3). The California Teachers Study recently reported an increased risk for ER⁻/PR⁻ breast cancer associated with endocrine disruptors present in ambient air, namely cadmium compounds and inorganic arsenic (4). However, to date, relatively few studies have investigated the link between air pollution and breast cancer subtypes. This analysis sought to investigate breast cancer risk in relation to primary components of air pollution, namely PM_2.5, PM₁₀, and NO₂, and potential risk differences by breast cancer subtype.

The Sister Study, a cohort of 50,884 U.S. women between ages 35 to 74 whose sister had breast cancer (5), enrolled participants from August 2003 to July 2009, who were followed for a mean of 4.95 years. Two thousand and eighty-nine breast cancer cases arose between enrollment and January 2013, (of which 316 were in situ), with a mean time to breast cancer of 3.96 years. Air pollution exposure was not ascertained on 1,234 women (24 invasive and 6 in situ; 1,204 noncases) predominantly because they lived outside the conterminous U.S., resulting in 1,749 invasive breast cancers and 47,591 noncases for this analysis. Annual averages of air pollution concentration outside the residence were estimated at each participant's home from a validated regionalized universal kriging model derived from regulatory monitors and a large suite of geographic covariates using previously described methods (6). For primary analyses, air pollution estimates were based on annual average concentrations at baseline home addresses, derived using monitoring data from 2006 (PM_2.5 and NO₂) and 2000 (PM₁₀). The cross-validated R² for PM_2.5 NO₂, and PM₁₀ were 0.88, 0.85, and 0.53, respectively (6). HR and 95% CI were estimated using Cox proportional hazards models. Known breast cancer risk factors were considered for inclusion in the model if the factor was associated with both air pollution and breast cancer. Race, educational attainment, smoking status, and menopausal hormone therapy met these criteria. Results were unchanged when we adjusted for geography using splines. In subset analyses, (i) we examined the effect of air pollution separately for breast cancer subtypes, stratifying by estrogen receptor/progesterone receptor (ER/PR) and stage, calculating RR and 95% CI using polytomous logistic regression; and (ii) we examined residential air pollution concentrations derived from 1990′s estimates among those who had lived long-term at their current residence (i.e., excluding those who changed residences) in order to investigate associations of long-term air pollution with breast cancer.

Breast cancer cases were more likely White, highly educated, and users of menopausal hormone therapy (Table 1). There was no association between invasive breast cancer overall and PM_2.5, PM₁₀, or NO₂ (Table 2). However, the risk associated with NO₂ differed when stratified by ER/PR (P = 0.04). NO₂ was associated with a 1.10-fold increased risk of ER⁺/PR⁺ breast cancer [95% CI, 1.02–1.19 per interquartile range (IQR) of 5.8 ppb] but not with ER⁻/PR⁻ breast cancer (RR = 0.92; 95% CI, 0.77–1.09). We observed a borderline increased risk of breast cancer in situ in relation to NO₂ (HR = 1.10; 95% CI, 0.99–1.24 per IQR of 5.8 ppb; data not shown).

Our analysis did not suggest an association between air pollution and overall invasive breast cancer risk. Multiple studies (2, 3), but not all (4, 7), found that exposure to traffic-related air pollutants, particularly NO₂, increased breast cancer risk. A potential explanation for differences among studies could be differing proportions of ER/PR subtypes, if as our data suggest, NO₂ is only associated with ER⁺/PR⁺ breast cancer. NO₂ probably serves as a marker for traffic-related pollution rather than a causal factor per se (3). As such, it may serve as a proxy for components of air pollution that affect estrogens, such as polycyclic aromatic hydrocarbons (PAH). PAHs have estrogenic properties, as shown by PAH binding to ER-β to induce transcriptional targets (8). Thus, there is biologic plausibility for a differential role of air pollution by hormone receptor status. However, Liu and colleagues reported that estrogen disruptors in ambient air were not associated with ER⁺/PR⁺, but rather with ER⁻/PR⁻ breast cancer (their analysis did not report on NO₂; ref. 4).

This analysis using a prospective, large national sample that systematically evaluated air pollution using state-of-the-art spatial modeling is able to rule out a strong relationship between air pollution and breast cancer risk. One limitation is that air pollution exposure earlier in life could affect breast cancer risk; however, our analysis of long-term air pollution exposure showed results were unchanged. Replication of these results is needed before firm conclusions can be drawn regarding ER⁺/PR⁺ breast cancer risk in relation to traffic-related air pollution.

J.D. Kaufman is a consultant/advisory board member for Health Effects Institute, Diesel Exhaust Epidemiology Panel. No potential conflicts of interest were disclosed by the other authors.

The views expressed in this document are solely those of the authors and the EPA does not endorse any products or commercial services mentioned in this publication.

Conception and design: M.T. Young, C.J. Han, L.A. DeRoo, J.D. Kaufman, D.P. Sandler

Development of methodology: M.T. Young, A.A. Szpiro, C.J. Han, L.A. DeRoo, J.D. Kaufman

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): L.A. DeRoo, C. Weinberg, D.P. Sandler

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): K.W. Reding, M.T. Young, A.A. Szpiro, C.J. Han, J.D. Kaufman

Writing, review, and/or revision of the manuscript: K.W. Reding, M.T. Young, A.A. Szpiro, C.J. Han, L.A. DeRoo, C. Weinberg, J.D. Kaufman, D.P. Sandler

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): C.J. Han, L.A. DeRoo

Study supervision: L.A. DeRoo, J.D. Kaufman, D.P. Sandler

This work was funded in part by the Intramural Research Program of the NIH, National Institute of Environmental Health Sciences (Z01 ES 044005; PI: D.P. Sandler). This publication was also in part developed under STAR research assistance agreements (RD831697; PI: J.D. Kaufman; and RD83479601; PI: S. Vedal; Project Director: J.D. Kaufman) awarded by the U.S. EPA. It has not been formally reviewed by the EPA. Dr. Reding was supported by National Institute of Nursing Research career development award (R00 NR 012232; PI: K.W. Reding).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.