Breast cancer is the most common cancer among women worldwide, with an estimated 2.09 million new cancer cases diagnosed globally in 2018 [1]. Over the past 30 years, its incidence has continuously increased in France [1,2]. The rapid increase in the worldwide incidence of breast cancer has been associated with mass screening, menopausal hormonal therapy, and societal changes impacting individuals’ lifestyles. Epidemiological studies have suggested an important role of lifestyle and environmental factors, supported by geographical variations of breast cancer incidence and time trends in incidence rates among migrant populations [3-5]. However, a considerable proportion of the risk remains unexplained, and the impact of environmental factors in the etiology of breast cancer has not been fully explored. Epidemiology and laboratory findings suggest that exposure to environmental pollutants, particularly those with estrogenic potential, may influence the development of breast cancer [6,7].

Ambient air pollution is a major public health concern related to a range of adverse health effects, including cancer, and accounting for an estimated 4.2 million deaths per year [8]. In 2013, the International Agency for Research on Cancer (IARC) classified outdoor air pollution as a whole and particulate matter as carcinogenic in humans, principally based on studies of lung and bladder cancers [9]. However, studies on breast cancer are scarce, and results remain inconsistent [10]. A recent meta-analysis of individual data from 15 European cohorts revealed no association between postmenopausal breast cancer and exposure to various classes of particulate matter (PM_2.5, PM₁₀, and PM coarse); however, they reported a statistically significant positive relationship per 20 µg/m³ increase in nitrogen dioxide (NO₂) exposure [11]. Epidemiological evidence suggests an association between breast cancer and NO₂ from traffic-related air pollution [12-15]. Furthermore, women with extremely dense mammography density, a well-established risk factor for breast cancer, were less likely to have higher levels of ozone (O₃) exposure [16]. Breast cancer has also been linked to exposure to endocrine-disrupting pollutants, such as polychlorinated biphenyls (PCBs) [17,18], dioxins [19-21], benzo[a]pyrene (BaP) [12,22], and cadmium [23,24] with sometimes diverging results. Furthermore, studies have reported positive associations between traffic-related BaP exposure as a surrogate for Polycyclic aromatic hydrocarbon (PAH) exposure and breast cancer [25,26]. Cadmium has been classified by IARC as carcinogenic to humans (Group 1), with sufficient evidence for lung cancer [27]. The only study performed on airborne cadmium exposure and breast cancer risk found no evidence of overall increased risk; however, elevated risks for hormone receptor-negative tumors (estrogen receptor and progesterone receptor-negative (ER-PR–)) were observed with higher exposure to cadmium [28].

Indoor pollution is also an important challenge for global health. Around 3 billion people use traditional biomass fuels for household cooking and heating [29], which is a major source of indoor air pollution and exposure to dioxins, PAH, particulate matter, O₃, and NO₂ [30-35]. However, only two studies have investigated the impact of indoor pollution on breast cancer risk [36,37]. Overall, current epidemiological evidence regarding the impact of exposure to ambient air pollution and indoor pollution on breast cancer risk is not fully consistent. In addition, the effects of simultaneous or sequential exposures to multiple compounds have been insufficiently explored, and the evidence supporting the differential effects of menopausal status on breast cancer risk and receptor status is also limited [23]. Breast cancer is no longer considered a homogeneous disease but is a heterogeneous disease composed of several distinct molecular subtypes according to hormone receptor status (ER, PR, and human epidermal growth factor receptor 2, or HER2) [38]. These subtypes each have different prognoses and can affect women differently. Molecular pathological epidemiology, which integrates molecular pathology into epidemiological studies has emerged in order to test for a difference between the association of a specific environmental exposure with subtypes classified by molecular features in determining disease incidence/mortality [39,40].

One of the major limitations of previous studies is the lack of past residential history and/or historical ambient air pollution exposure estimates, which may have resulted in exposure misclassification, contributing to imprecise risk estimates and bias towards the null [41]. Many epidemiological studies have relied on a variety of exposure assessment techniques such as using data from centrally located ambient air quality monitoring networks, surrogates for exposures. However, such methods are often insufficient for capturing the spatial variability of pollutant concentrations at the local scale, both at the intra-urban scale, and the suburban and rural scale [42]. More complex techniques have been used in recent studies to adequately represent the spatial-temporal variation of pollutants, including land-use regression models, dispersion modeling, chemistry transport models, and hybrid models [43]. These complex methods can be combined to assess exposure to each pollutant at a fine spatial scale and over a large area between 1990 and 2011.

Exposures occurring early in life and/or during biological windows of greater sensitivity (ie, in utero and during childhood) have been suggested to be more strongly associated with breast cancer risk [44,45]. However, very few studies have investigated these effects, and the majority of research has been based on adulthood exposures within short observation periods. These studies may have missed critical windows or the cumulative effects of lifetime exposure that could impact breast cancer risk [45,46]. A lack of historical measures makes retrospective exposure reconstruction difficult, especially for earlier periods. Higher breast cancer incidence is commonly observed in large cities compared with rural areas [47]. Also, women born in an urban area are at a greater risk of breast cancer versus women born in rural areas [47]. To our knowledge, no study has investigated the effects of life-course residential trajectories on breast cancer. Residence in an urban area has been suggested to be a surrogate of air pollution exposure released from road traffic, industrial facilities, and waste incineration [48], and to be useful to investigate earlier periods where historical air pollution records are unavailable and back-extrapolation unfeasible [47].

Early evidence suggests that cancer risk may not be a linear function of cumulative carcinogen dose [49]. Individuals are exposed simultaneously to a complex and changing mixture of environmental exposures [50]. These may independently, cumulatively, or interactively influence the risk of developing breast cancer. Furthermore, in single pollutant models, it is unclear whether an observed association is due to the effect of the evaluated pollutant or whether it acts as a surrogate for another pollutant from the same source. Multipollutant approaches need to address the complex structure of mixtures that frequently present multicollinearity. However, such epidemiological studies are limited [51,52].

Current evidence on the role of genetic susceptibility (polymorphisms) related to exposure to ambient air pollution remains limited. Previous epidemiological studies have reported interactions between genetic polymorphisms and some pollutants [53,54]. Saintot et al reported that women carrying the Val CYP1B1 allele and who had lived near a waste incinerator for more than 10 years had a greater risk of breast cancer than those with the Leu/Leu genotype and had never been exposed [55]. A positive association was found in postmenopausal women with a CYP1A1 variant genotype [56].

Methylation is one of several epigenetic events involved in the regulation of gene expression, and it can undergo alterations as a consequence of environmental stimuli. Emerging evidence suggests that exposure to ambient air pollutions could influence DNA methylation, producing hypomethylation of repetitive elements in leukocytes and buccal cells, as well as altered methylation at the CpG level in specific genes [57-60]. The biological effects of DNA methylation induced by exposure to air pollution have been investigated in the context of lung cancer risk [61], but little is known about their impact on breast cancer risk.

The overall objective of the XENAIR project is to investigate the chronic long-term effects of exposure to multiple ambient air pollutants and risk of breast cancer in a nested case-control study within the ongoing French prospective E3N (Etude Epidémiologique auprès des femmes de la Mutuelle Générale de l’Education Nationale) cohort. More specifically, the project aims to assess the associations between chronic exposure to selected ambient air pollutants (particulate matter, NO₂, O₃, BaP, dioxins, PCB 153, and cadmium) estimated from individual residential addresses of study subjects from recruitment (1990) and breast cancer risk. The study will (1) analyze exposure trajectory profiles of individual compounds over time and estimate breast cancer risk associated with each of these exposure profiles; (2) estimate breast cancer risk associated with the weighted cumulative duration of urban residence since birth (used as a surrogate for exposure to ambient air pollutions), as well as indoor exposure from domestic heating and combustion activities; (3) explore approaches to estimate multi-pollutant exposure.

The XENAIR project will further assess potential interactions between long-term exposure to low doses of air pollutants and genetic polymorphisms involved in air pollutant metabolism, to address the hypothesis that breast cancer risk associated with pollutants may depend on individual genetic susceptibility. We will also explore the potential role of DNA methylation as a marker of exposure to ambient air pollutions and as a potential mediator of the effect of ambient air pollutions on breast cancer risk. Additionally, we will estimate the fraction of breast cancers attributable to air pollution in France based on the risk estimates and additional costs of breast cancer management attributable to air pollutants.

E3N is an ongoing prospective cohort study launched in 1990 to investigate the main risk factors for cancer and severe chronic conditions in women [62]. Participants were recruited between June 1990 and November 1991 among women aged 40-65 years, living in France and insured with the MGEN, a national health insurance plan covering people working with the French education system and their families, and have been biennially followed-up with self-administered mailed questionnaires. E3N is the French part of the European Prospective Investigation on Cancer (EPIC), a vast European study coordinated by IARC and involving nearly 500,000 Europeans in 10 countries [63]. At recruitment, 98,995 E3N participants filled in a self-administered questionnaire, which included data about lifestyle and reproductive factors, anthropometry, past medical history, and familial history of cancer. To date, twelve questionnaires have been sent to the participants (participation rate at each questionnaire ~80%). Between 1994 and 1998, participants were invited to give a blood specimen. Blood samples were collected from 25,000 women, and saliva samples were later collected from an additional 47,000 women. The occurrence of cancer was self-reported in each questionnaire, and a small number of cancers were further identified from the insurance files or information on causes of death obtained from the National Service on Causes of Deaths. Pathology reports confirming diagnoses of invasive breast cancer (the primary outcome of the present project) were obtained for 93% of self-declared cases, and the proportion of false-positive self-reports was low (<5%). Addresses were recorded at baseline (1990) and the 5- and 9-year follow-up questionnaires (years 1997, 2000, 2002, 2005, 2008, and 2011). Postal codes were recorded at 3- and 4-years follow-up (1993 and 1994). Participants’ place of birth (postal code and municipality) was obtained from the first questionnaire and assigned an urban/rural status based on data from the closest national census [64]. Informed consent was obtained from each participant, and the study was approved by the French National Commission for Data Protection and Privacy (CNIL).

Data on established and potential breast cancer risk factors were available from the self-administered questionnaires at baseline. Regular updates have been collected on smoking, anthropometry (height, weight), physical activity, diabetes, hypertension, benign breast disease, gynecological screening, family history of breast cancer (FHbreast cancer), education, and reproductive factors. Women completed two validated self-administered diet history questionnaires (DHQ) in 1993 and 2005. The E3N DHQ covered the daily consumption of 208 food items by collecting food frequencies and portion sizes for eight meals and snacks during the day [65,66]. Dietary exposure to BaP, dioxins, PCB 153, and cadmium will be assessed for each woman by combining consumption data from the E3N DHQ and food contamination data available for France. Contamination data for BaP, dioxins, PCB 153, cadmium are available from the French agency for food safety (ANSES) [67,68].

The present study is based on a nested case-control subset of the E3N cohort. It involves 5222 histologically confirmed incident breast cancer cases identified during the 1990-2011 follow-up period. Women were included if they had completed their home address at baseline, lived in the French metropolitan territory during the 1990-2011 follow-up time, and not had any cancer at baseline. For each breast cancer case, one control was randomly selected by incidence density sampling, among cohort participants at risk of breast cancer at the time when the case was diagnosed, using the follow-up time since inclusion into the cohort as the time axis. In order to best select appropriate controls according to the planned studies, two complementary groups of cases were set, according to the presence of a blood sample, saliva sample, or no biological sample available. For the first group of cases (with a blood sample), controls were matched to cases on the department of residence, age (±1 year), date (±3 months), and menopausal status at blood collection. Controls for the second group (without a blood sample) were matched on the same criteria but collected at baseline, and additionally matched on the existence or not of a saliva sample.

A structured questionnaire was sent to all selected cases and controls to collect lifetime residential addresses from date of birth to the present (street address, municipality/city, postal code), school and workplace addresses, commute duration ( “less than 30 min,” “between 30 and 60 min,” “more than 60 min”) and type (“walking,” “cycling,” “motorcycle riding,” “driving a diesel car,” “driving a gasoline car,” or “using public transportation”), and information on domestic heating and combustion activities. We also collected information on the age when starting and stopping living in each reported home. Women were asked to report the period during which their home was built (before 1948, between 1948 and 1974, or after 1974), and whether it overlooked a courtyard or a street (courtyard, street, or both). In terms of indoor heating, they reported their main type of heating (collective central, or individual) and their main source of heating (wood, charcoal, electricity, gas, or fuel). Regarding indoor wood-burning stove or fireplace cooking, women were asked whether they used an indoor wood-burning stove or a fireplace in their home (yes, no), and if yes, the type (stove/wood stove, open fireplace, or closed fireplace). Information on the use and frequency of cooking foods on the barbecue was also collected (never, rarely (once to twice per year)), occasionally, or frequently (at least once a month)). Study participants additionally answered questions on whether they burned green waste (yes, no), and if yes the frequency (never, rarely (once or twice a year, occasionally, or frequently (at least once a month)), and the quantity of green waste burned every year (less than 1 m³, 1-5 m³, or more than 5 m³). The overall response rate was 65.4%.

The methods of geocoding residential history and industrial sources have been described in detail elsewhere [69]. Briefly, residential histories from the E3N follow-up questionnaires and the residential questionnaire will be geocoded (X and Y coordinates, addresses) using ArcGIS Software (ArcGIS Locator version 10.0, Environmental System Research Institute) and the national addresses database from the National Geographic Institute (BD Adresse, IGN). Geocoding will be performed by a trained technician blinded to the case-control status of the participants.

Assessment at the national level of exposure to selected pollutants (PM₁₀, PM_2.5, NO₂, O₃, BaP, dioxins, PCB 153, and cadmium) will be based on complementary models according to data availability and pollutant emissions characteristics (Table 1). Specifically, we will use a regional chemistry-transport model (CHIMERE) [70], an urban gaussian dispersion model (SIRANE) [71], a land-use regression model [72] and a GIS-based metric [73]. A detailed description of these models is provided in Multimedia Appendix 1.

Our research is based on the E3N French national cohort. Informed consent was obtained from all participants, and the study was approved by the French National Commission for Data Protection and Privacy (CNIL).