Introduction

Breast cancer is the most common cancer diagnosed among women in the United States (USA) [1], and there is an interest in better understanding the role of environmental factors on breast cancer risk [2]. Air pollution is an established environmental risk factor for lung cancer [3], and outdoor air pollution has been classified by the International Agency for Research on Cancer (IARC) as a group 1 carcinogen [4].

Air pollution contains a mixture of many compounds, including polycyclic aromatic hydrocarbons (PAHs), metals, and benzene; these may act as carcinogens or as endocrine disruptors and, thus, be relevant for breast carcinogenesis. Inhaled toxicants have been measured in breast fluid, showing that airborne pollutants can reach the breast tissue [5]. The most well-studied compounds are PAHs [6], a combustion by-product, which has the capacity to bind to DNA and form adducts in the breast tissue [7]. Both PAHs and metals have estrogenic properties [8, 9], produce oxidative stress [10], and induce mammary tumors in animal models [7, 9]. Particulate matter, a complex mixture of small airborne particles including metals and hydrocarbons [11], has been shown to exhibit estrogenic properties and DNA-damaging activity in vitro [12], and benzene induces mammary tumors in rodents [13]. Both indoor and outdoor air pollution have been associated with breast tumor methylation of candidate genes selected a priori based on their relationship with breast cancer [14].

Breast cancer risk has been shown to be elevated in urban areas where air pollution is higher [15,16,17], and ecologic studies suggest that increasing traffic emissions in the USA has been associated with an increase in breast cancer risk [18, 19]. However, the epidemiologic evidence to date has been inconsistent and sparse. In this review, we will summarize the existing literature on air pollution and breast cancer risk and highlight recent studies that have advanced our understanding of the relationship between air pollution and breast cancer. We will summarize the research to date on directly measured air pollution (including particulate matter, nitrogen dioxide (NO2) and nitrogen oxides (NOx), traffic-related air pollution), proximity to roadways and traffic density, and other air pollution assessments (indoor air pollution, hazardous air toxics).

Exposure and Outcome Considerations

The exposure of interest for this review was either indoor or outdoor air pollution exposure, and study characteristics and findings are summarized in Tables 1 and 2. Measurement of air pollution varies widely and includes direct measurement at fixed sites, estimation of residential exposure using modeling methods, or proxies of exposure such as density of traffic or proximity to roadways. Generally, studies utilizing modeling techniques such as air dispersion or land-use regression are believed to better capture the air pollution exposures by at least partially overcoming exposure misclassification [37]. Given the considerable heterogeneity of exposure assessment reviewed here (continuous monitored data, categorical exposure categories based off of modeled estimates, questionnaire responses), it is challenging to directly compare many of the estimates. As such, we include the comparisons used for each measure of effect when describing the study findings.

Table 1 The relationship between air pollution and breast cancer risk by recency of publication: case-control studies
Full size table
Table 2 The relationship between air pollution and breast cancer risk by recency of publication: cohort studies
Full size table

The outcome of interest for this review was incident breast cancer. We considered whether studies considered variation in the association by menopausal status at diagnosis (premenopausal vs. postmenopausal) and by tumor hormone receptor subtype defined by the estrogen receptor (ER) and the progesterone receptor (PR).

Article Search Strategy

To identify articles related to air pollution exposure and breast cancer risk, we did a search of PubMed using the following search criteria: (diesel[Title/Abstract] OR pm2.5[Title/Abstract] OR pm10[Title/Abstract] OR air pollution[Title/Abstract] OR particulate matter[Title/Abstract] OR traffic[Title/Abstract] OR Nitrogen dioxide[Title/Abstract] OR NO2[Title/Abstract] OR NOx[Title/Abstract] OR Nitrogen oxide[Title/Abstract]) AND (breast cancer[Title/Abstract]). We restricted our search to articles that were in English and that addressed either indoor or outdoor air pollution exposure in relation to breast cancer incidence. We first narrowed down the articles identified by the PubMed search criteria based on the article title. We next reviewed the abstracts for relevance to the review. We incorporated three additional studies related to the topic that did not arise from the PubMed search but were cited in the articles identified by the search.

Directly Measured Air Pollution

Particulate Matter

Particulate matter (PM) is a complex mixture of extremely small particles and liquid droplets that disperse into the air. PM is classified based on its diameter; the most commonly measured groups are PM10, which are inhalable particles with diameters ≤ 10 μm; PM2.5 which is defined as fine inhalable particles with diameters ≤ 2 μm; and PM10–2.5 which is also referred to as PMcoarse. Ultrafine particles (UFPs) are defined as being < 0.1 μm in diameter and are generated by internal combustion engines and through secondary processes [38]. Prior to the use of PM2.5 and PM10, total suspended particle (TSP) was a more crude measure of air pollution that relied on high volume samplers to capture particles ≤ 50–100 μm in diameter.

In an early study of air pollution and breast cancer, Bonner and colleagues published findings from the Western New York Exposures and Breast Cancer (WEB) case-control study, where exposure to TSP was estimated to be associated with a twofold higher odds of postmenopausal breast cancer for TSP exposure at birth [26]. Notably, TSP levels at other time periods were not positively associated with either premenopausal or postmenopausal breast cancer and in some instances, TSP was inversely associated with breast cancer risk. Nonetheless, this study suggested a possible role for particulate matter in relation to breast cancer risk.

Most subsequent cohort studies evaluating the relationship between more precise PM measurements and breast cancer risk have not observed a positive association [30, 31•, 32•]. No association was observed in the USA-based Sister Study, a cohort of women with a family history of breast cancer (1749 invasive cases, 4.9 years of follow-up) [32•]. Neither PM2.5 nor PM10 levels assessed at the baseline residence were associated with overall invasive breast cancer or when considering ER/PR status of the tumor [32•]. In the Nurses’ Health Study II (3416 cases), residential PM exposure (PM2.5, PMcoarse, PM10) was assessed for a 48-month time period. As in the Sister Study, no associations were observed for any PM metric with breast cancer risk with consideration of menopausal status at diagnosis and ER/PR status of the tumor, with the exception for PMcoarse and ER+/PR+ breast cancer (HR = 1.13, 95% CI: 0.99–1.29, per 10 μg/m3) [31•]. No association was observed between 3-year running mean PM2.5 and PM10 levels in the Danish nurse cohort (N = 1145 cases), with possible differences by menopausal status as PM2.5 was observed to be associated with premenopausal (HR = 1.06, 95% CI: 0.94–1.18, per 3.3 μg/m3) but not postmenopausal breast cancer (HR = 0.93, 95% CI: 0.85–1.05, p for interaction = 0.07) [30]. These three cohort studies assessed PM levels using land-use regression models.

In the ESCAPE project, which pooled nine European prospective cohorts, elevated associations with postmenopausal breast cancer (N = 3612 cases) were observed for PM2.5 (HR = 1.08, 95% CI: 0.77–1.51, per 5 μg/m3), PM10 (HR = 1.07, 95% CI: 0.89–1.30, per 10 μg/m3), PMcoarse (HR = 1.20, 95%CI 0.96–1.49, per 5 μg/m3) but the authors noted considerable heterogeneity between individual cohort estimates [28••]. The ESCAPE project also considered elemental components of PM2.5 and PM10, which were assayed using X-ray fluorescence of PM filters. Specific PM components were selected based on their expected relationship to health effects and representativeness as well as data quality. The authors observed an association for PM10 nickel (HR = 1.30, 95% CI: 1.09–1.55, per 2 ng/m3) and vanadium (HR = 1.30, 95% CI: 0.95–1.77, per 3 ng/m3) with respect to postmenopausal breast cancer [28••]. Nickel and vanadium have estrogenic properties that may make those PM components particularly relevant to breast cancer [9] and were specifically included as PM components based on their hypothesized representation of mixed oil-burning and industry exposure sources.

In a Canadian case-control study (681 cases and 596 controls), Goldberg et al. estimated UFPs applying 2011–2012 monitoring data in a land-use regression model but found little to no association with postmenopausal breast cancer [20]. This is the only study to consider UFPs as a measure of air pollution exposure.

Nitrogen Dioxide (NO2) and Nitrogen Oxides (NOx)

Nitrogen oxide (NOx) represents the total concentration of NO and NO2 produced from combustion processes. NO2 levels are considered to be a marker for traffic-related pollution [39], and thus may be a proxy for other components of traffic-related air pollution such as PAHs. NO2 levels have been shown to be correlated with breast cancer incidence in ecological studies [18, 19].

In a mid-1990s case-control study in Quebec (383 invasive cases, 416 controls), a land-use regression model was used to assess NO2 levels in relation to postmenopausal breast cancer [24]. Elevated ORs were observed in association with NO2 across a number of different exposure time periods; for example, for NO2 levels estimated near the time of breast cancer diagnosis in 1996, a 5 ppb increase in NO2 exposure was associated with an OR of 1.31 (95% CI: 1.00–1.71) [24]. Results were similar to those later reported for a larger population-based case-control study conducted in 8 Canadian provinces (1569 cases, 1872 controls) [22••]. In Hystad et al. [22••], NO2 levels for the 20 years prior to diagnosis were estimated using three different methods: satellite-derived observations, satellite-derived observations scaled with historical fixed-site measurements of NO2 and a national land-use regression model [22••]. All three NO2 measurements were associated with both premenopausal and postmenopausal breast cancer cases, with estimates more pronounced for premenopausal breast cancer [22••].

Between 2008 and 2011, another population-based case-control study of postmenopausal breast cancer in Montreal assessed exposure to NO2 for 2005–2006 using a land-use regression model (681 cases and 596 controls) [20]. For an interquartile range (IQR) increase in NO2 (3.75 ppb), the OR for postmenopausal breast cancer was 1.08 (95% CI: 0.92–1.27); the estimated OR was higher in women who lived in their homes for 10 years before the study and for women with ER + PR+ breast cancer (OR = 1.13, 95% CI: 0.94–1.35) [20].

Elevated associations with NO2 have also been observed in cohort studies. In the Sister Study cohort, NO2 assessed using a land-use regression model at the baseline residence was associated with risk of ER + PR+ breast cancer (947 ER + PR+ cases; RR = 1.10, 95% CI: 1.02–1.19 per IQR of 5.8 bbp) but not ER−PR− breast cancer (223 ER−PR− cases, RR = 0.92, 95% CI: 0.77–1.09, p for interaction = 0.04) [32•]. In the ESCAPE project, both NO2 and NOx were both associated with postmenopausal breast cancer (NO2, HR = 1.02, 95% CI: 0.98–1.07 per 10 μg/m3; NOx, HR = 1.04, 95% CI: 1.00–1.08 per 20 μg/m3) [28••]. The ESCAPE project included data from the Danish Diet Cancer and Health Study (n = 987 breast cancer cases), which separately reported that a 100 μg/m3 increase in modeled NOx estimated over a 5-year period was associated with breast cancer risk (IRR = 1.16, 95% CI: 0.89–1.51) [36]. In contrast, no increase in risk was observed with NO2 levels in the Danish Nurse Cohort (N = 1145 cases) using a 3-year running mean average level [30].

Traffic-Related Pollution Models

The Long Island Breast Cancer Study Project (LIBCSP) and the WEB study both employed similar validated traffic-related pollution models that estimated residential exposure to benzo[a]pyrene. In the population-based LIBCSP (1508 cases and 1556 controls) [21], women in the top 5% of exposure when compared to those below the median had an elevated but imprecise OR for overall breast cancer (OR = 1.47, 95% CI: 0.70–3.08) when considering longer-term exposure (1960–1990); the OR was slightly more elevated for ER−PR− breast cancer (OR = 1.67, 95% CI: 0.91–3.05) also for longer-term traffic-related exposure [21].

In the WEB study, they reported a twofold higher odds of premenopausal breast cancer for traffic exposure at age at menarche and a twofold higher odds for postmenopausal breast cancer in relation to traffic exposure at age at first birth [25]. However, associations were not evident for traffic exposure at age at first birth or for exposure at 10 or 20 years prior to diagnosis when considering both premenopausal and postmenopausal breast cancer [25].

Proximity to Roadways and Traffic Volume

No association was observed with proximity to nearest roads or vehicular traffic volume metrics in Hystad et al. [22••], in the Danish Diet and Cancer Study [36] or in the ESCAPE project [28••]. In the Nurses’ Health Study II, living within 50 m of a major road was suggestively associated with breast cancer, although confidence intervals were wide [31•]. High traffic density measures were not significantly elevated in relation to breast cancer risk in another case-control study conducted on Long Island, New York (793 cases, 966 controls) [27].

Using self-reported childhood residential characteristics of main road and nearest intersecting road as proxies for traffic exposure (i.e., number of lanes, presence of a median or barrier, traffic), Shmuel et al. reported an elevated risk of breast cancer for living on a road with a median or barrier (HR = 1.2, 95% CI: 0.9–1.7) and for living near an intersecting road with high proxy measures of traffic (HR = 1.4, 95% CI: 1.0–1.9) in the Sister Study [29].

Other Measures

Indoor Air Pollution

Burning wood or gas in the home for indoor heating and cooking purposes can release compounds similar to that observed in outdoor air, including particulate matter, PAHs, and benzene [40]. Two studies have evaluated the relationship between air pollution from using indoor stove/fireplaces in the home and breast cancer risk. The first study was conducted in the LIBCSP and reported a OR = 1.42 (95% CI: 1.11–1.84) of breast cancer for burning synthetic or artificial firelogs in the home relative to no indoor stove/fireplace use [23]. In the prospective Sister Study cohort, using an indoor stove/fireplace at the longest adult residence was also associated with breast cancer; risk was higher with increasing frequency of use (at least once per week relative to no stove/fireplace use, HR = 1.17, 95% CI: 1.02–1.34) and was most evident for burning wood or gas [35•].

Hazardous Air Toxics

The California Teacher’s Study used the 2002 Environmental Protection Agency’s (EPA) National Air Toxics Assessment (NATA) database to evaluate the association between quintiles of hazardous air toxic pollutants estimated at the census-track level and breast cancer risk [33, 41]. In a study of 24 air toxics selected based on toxicological data for the ability to induce mammary gland tumors, they reported an association for acrylamide, carbon tetrachloride, chloroprene, 4,4′-methylene bis(2-choloraniline), propylene oxide, and vinyl chloride with overall breast cancer, noting that associations were largely not monotonic [41]. Interestingly, the authors found variability in the associations by hormone receptor subtype; ER+ or PR+ tumors were associated with higher levels of acrylamide, benzidine, carbon tetrachloride, ethylidene dichloride, and vinyl chloride whereas ER−/PR− tumors were associated with higher levels of benzene [41]. When focusing on 11 a priori selected endocrine disrupting hazardous air pollutants, the authors found little evidence for an association with these compounds except for an elevated risk in a select subgroup of never-smoking non-movers for ER−/PR− breast cancer in relation to higher exposure to cadmium compounds and inorganic arsenic [33].

Conclusions

In this review, we have summarized the results for eight case-control studies and nine cohort studies that have used a range of metrics to analyze the relationship between breast cancer and exposure to indoor and outdoor air pollution. The association with breast cancer tended to vary based on the air pollutant assessed, with more consistent findings reported for elevated NO2 or NOx levels and traffic-related air pollution models.

Together, the studies to date suggest little evidence to support a relationship between particulate matter and breast cancer risk. However, it is possible that individual components of PM, such as vanadium and nickel as demonstrated in the ESCAPE project [28••], may be relevant for breast cancer. Additional research is needed to better understand PM components and their potential for a relationship with breast cancer risk.

The results from studies that considered NOx and NO2, indicators of vehicular traffic exposure, are more suggestive of a role for air pollution in breast carcinogenesis than those for particulate matter. Most studies reported positive estimates in relation to higher NO2 and NOx levels, with the possibility that risk may vary by ER/PR subtype as observed in the Sister Study cohort. Consistent with the results for NOx and NO2, the vehicular traffic B[a]P models applied in the LIBCSP and in the WEB study also suggest an elevated risk of breast cancer in relation to PAH exposure from vehicular traffic.

There is little to no association observed across studies that used distance to nearest road or traffic volume as a proxy for air pollution exposure in relation to breast cancer risk. However, some suggestion of an association was observed for childhood exposure to high residential traffic.

Indoor burning of biomass is of concern worldwide [42] and prevalence estimates suggest that frequency of use is high in the USA as well [35•]. Two studies, the LIBCSP and Sister Study cohort, observed associations with indoor stove/fireplace use in relation to breast cancer risk although associations were not consistent in terms of type of material burned [23, 35•]. More research is needed to better understand the importance of indoor air pollution components in relation to breast cancer risk.

Exposure to hazardous air toxics, as demonstrated in the California Teacher’s Study, may be relevant for breast cancer [33, 34]. The EPA NATA relies on modeled data from reported industry emissions and other sources in conjunction with air pollution modeling techniques to assess air toxics on the census-track level [43]. Because hazardous air pollutants are rarely monitored, better assessment of exposure to these air toxics is needed.

Most of the literature on air pollution and breast cancer risk has focused on adult-level air pollution exposure, predominately measuring and estimating air pollution in the years immediately preceding diagnosis. However, early life may represent a potentially susceptible time period for breast tissue [44]. In the Sister Study cohort, self-reported childhood residential characteristics as proxy measures for traffic exposure were associated with an elevated association with breast cancer risk [29]. Similarly, in the WEB study, TSP exposure levels at birth were associated with later postmenopausal breast cancer risk and when using a geographic traffic BaP exposure model, exposure at age of menarche and at age at first childbirth was associated with premenopausal and postmenopausal breast cancer, respectively [25]. Early life exposure to air pollution is challenging to measure accurately as studies usually have women at a range of ages (thus having a range of calendar years relevant for early life exposure), and air pollution monitoring data often does not go back far enough in time to capture the relevant period. Advanced modeling that can reliability extrapolate back would be useful in quantifying early life air pollution exposure.

Few studies have been able to incorporate information regarding ER/PR tumor subtype. The impact of breast cancer risk factors have been shown to vary by ER/PR status suggesting etiologic heterogeneity [45]. The Sister Study found the association with NO2 to be limited to ER+/PR+ breast cancer [32•]. Similarly, in Goldberg et al., an IQR increase in NO2 was most strongly related to ER+/PR+ breast cancer, as was UFPs [20]. The Nurses’ Health Study II found PMcoarse to be related to ER+/PR+ breast cancer [31•]. In contrast, the LIBCSP study found results using the B[a]P traffic model to be strongest for ER−/PR− breast cancer [21]. Hazardous air toxics and breast cancer associations also appeared to vary by ER/PR status of the tumor, with associations reported for both ER+/PR+ breast cancer and ER−/PR− breast cancer depending on the air toxic [33, 34]. Different air pollutant constituents may have differing carcinogenic properties and thus may be relevant for hormone-receptor positive or negative disease. More research is needed, especially in large studies with good power to describe differences in hormone receptor status.

A few studies in this review were limited to only postmenopausal women [20, 24, 28••] although most studies included both pre- and postmenopausal women. Other risk factors for breast cancer have been shown to vary by menopausal status at diagnosis, such as obesity [46]. It is unclear whether associations vary by menopausal status at diagnosis, with some studies reporting stronger associations of air pollution exposures with premenopausal breast cancer [22••, 30, 31•] and some with postmenopausal breast cancer [25,26,27]. As with tumor receptor subtype, more research is needed in well-powered studies to quantify differences in exposure by menopausal status at diagnosis.

The purpose of this review was to evaluate the relationship between metrics of air pollution and breast cancer incidence. However, there is a recent and growing body of research that is worth noting on the relationship between air pollution and mortality after a breast cancer diagnosis. Specifically, a higher risk of mortality after a breast cancer diagnosis has been observed in relation to PM2.5 [47,48,49] and PM10 levels [50]. More research is needed to understand the contribution of air pollution to survival after breast cancer.

In conclusion, epidemiologic research on the association between indoor and outdoor air pollution and breast cancer risk suggests a relationship between air pollution when using NO2 as a marker for traffic-related air pollution. Improved exposure assessment in order to better capture and characterize exposure is needed. Future research needs to consider early life time periods of exposure, stratification by tumor subtype and menopausal status, indoor air pollution metrics, and hazardous air toxics as potential contributors to breast cancer risk.