What this study adds
Air pollution has been classified as a human carcinogen by International Agency for Research on Cancer, based mostly on results of lung cancer. Little research has been conducted on other sites of cancer. Previous findings by other researchers and our team have suggested positive associations between incident breast cancer in women and spatial variations in NO2. This article represents the second study that has been used to determine whether spatial variability in ultrafine particles may be associated with postmenopausal breast cancer. The present study, therefore, contributes to our meager knowledge of the carcinogenicity of ambient air pollution.
Breast cancer ranks first among all cancers in the world affecting women. In 2012, the International Agency for Research on Cancer estimated 1.67 million new cases (25% of all incident cancer cases1). In Canada in 2016, the age-adjusted rate was 130 per 100,000 women2; it accounts for 25% of all cancers diagnosed in women, the lifetime probability of dying from breast cancer is 1 in 30, and the 5-year survival rate after the diagnosis of breast cancer is 87%. The population-attributable risk percent for accepted risk factors (including alcohol consumption, hormonal therapy, age, age at menarche, age at first birth, family history of breast cancer, prior benign breast disease) has been estimated recently at 70%.3
There have been a number of recent reports regarding possible associations between the incidence of female breast cancer and various markers of ambient air pollution.4–14 We have reported previously the results of a case–control study conducted in the mid-1990s in Montreal, Quebec, in which we found a number of occupational risk factors for postmenopausal breast cancer,15 including combustion-related exposures, especially among women with positive estrogen receptor (ER) and positive progesterone (PR) receptor status. We found in the same study7,8 an association with ambient nitrogen dioxide (NO2) measured at street level, which is an accepted marker for traffic-related air pollution.16,17 In another population-based case–control study of postmenopausal breast cancer in Montreal that we conducted between 2008 and 2011, there was little evidence of associations for ultrafine particles (UFP; <0.1 μm in diameter), another marker of traffic-related air pollution, with odds ratios (OR) about 1.01 for an increase of about 3500 /cm3, although slightly higher ORs were found among cases with positive ER and PR status.18
In this article, using data from our first case–control study in which we have reported associations with ambient NO2 and occupational hazards,7,8,15 we present results of associations between postmenopausal breast cancer and number concentrations at street level of UFPs that were predicted from a land use regression model, using observations made in 2011–2012, and thereafter estimated at participants’ residences at time of diagnosis (1996–1997).
We have provided previously a detailed description of the methods of this study.8,9,15,19 Briefly, we conducted a hospital-based case–control study of incident, invasive, postmenopausal breast cancer. The target population comprised postmenopausal women, aged 50–75 years at time of diagnosis who in 1996 and 1997 were residents of the greater Montreal area. Women were menopausal according to the WHO criteria for menopausal status.20,21 Cases were diagnosed with primary, invasive breast cancer (International Classification of Diseases, 9th revision, code 174) confirmed histologically. We identified cases from all 18 hospitals in the region that treated breast cancer. To minimize the potential for recall bias, control subjects were selected at the same time and from the same hospitals as the cases and had one of 32 other selected sites of incident, histologically confirmed cancers. This design was developed by Siemiatycki et al22 in the late 1970s. The study was designed originally to investigate occupational exposures to selected substances, and thus, certain sites of cancer (i.e., liver and intrahepatic bile duct; pancreas; lung, bronchus and trachea; brain and central nervous system; leukemias and lymphomas) were excluded because of their possible association with occupational exposures. Controls were also approximately frequency-matched to cases by age in 5-year age bins.
One to three months after diagnosis, participants completed an interviewer-administered, structured questionnaire on their occupational history and other personal risk factors, including reproductive history, educational attainment, family history of breast cancer, age at menarche, smoking and alcohol consumption, weight and height 2 years before the interview (hence, body mass index), and home address (with duration of residence at that address) at time of diagnosis. Ethics committees at all participating hospitals and affiliated universities approved the protocol, and informed consent was obtained from participating subjects.
As deprived populations often live in areas characterized by higher concentrations of air pollution,7,23 neighborhood socioeconomic status may be a confounding factor. Thus, we obtained data from the 1996 Canadian Census for unemployment rate, low education, and median household income for subjects’ addresses in 350 census tracts in Montreal (hereafter, referred to as neighborhood ecologic covariates).
A land-use regression model for ultrafine particles
As part of our program of developing land-use regression maps for Canadian cities, we developed a model to estimate ambient concentrations of UFPs across the Island of Montreal.24 In summary, this model was derived from a mobile monitoring campaign conducted between 2011 and 2012.24 Data from 414 road segments, including major and minor roads, were used, and the main prediction variables were parking space (200-m buffer), open space (100-m buffer), local roads (100-m buffer), length of rail (100-m buffer), and annual NOx emissions (100-m buffer). The R2 from the land-use regression model was 0.62.
As our land use model was derived for the Island of Montreal, we included in the present analyses only those women living there. We assigned modeled concentrations of UFPs from the measurement campaign in 2011–2012 to geocoded street addresses or to centroids of the associated six-character postal code areas, of women at the time of interview (1996–1997). (In urban areas, a six-character postal code represents a block face or a large complex, such as an apartment building or office tower.) As there does not exist any historical data on UFPs and other pollutants are not well correlated (e.g., Spearman correlation coefficient of 0.14 for subjects’ addresses from our land use regression model of UFPs and of NO27,8), we could not use any back-extrapolation methods (e.g., Chen et al25) to estimate past concentrations. As a separate analysis, we stratified the data according to whether the women had been living at the same address in the previous 10 years or not. As we did not have lifetime residential histories, we selected the 10-year cutoff to restrict to those women who had longer-term exposures at the same address.
We used logistic regression to estimate ORs and 95% confidence intervals (CI). Continuous covariables were modeled as natural cubic spline functions,26 usually on two or three degrees of freedom, with linear terms used if there was no evidence of nonlinearity. As we did not want to lose power by excluding missing values, and more complex methods such as multiple imputation were not warranted because of the few missing values, we often fitted continuous covariates as categorical ones with cut points chosen to reflect the response pattern, and thus, missing values were included as their own category.
We developed three statistical models that incorporated different sets of covariates. Model 1 included only age as a continuous, linear covariate. Model 2 included additional personal risk factors, including family history, age of bilateral oophorectomy, education, ethnicity, age at menarche, age at first full-term pregnancy, cumulative number of weeks of breast feeding, number of years of oral contraceptive use, use of hormonal replacement therapy, body mass index 2 years before diagnosis, tobacco exposure (personal, passive smoking at home and occupational exposures to tobacco smoke), alcohol status, proxy interview (all as factor effects). Model 3 includes the same set of covariates as Model 2 but, to account for possible spatial clustering or neighborhood effects, we added the contextual variables unemployment rate (as a linear effect), low education (as a linear effect) and median household income (as a linear effect).
Hormonal receptor status was categorized as positive or negative for estrogen (ER+ or ER−) or for progesterone (PR+ or PR−), as provided in the pathology reports. We thus conducted separate analyses according to combinations of these receptors.
A total of 1631 subjects were eligible for this study. We conducted 608 interviews among cases and 667 among control subjects, thus, obtaining response rates of 81.1% for cases and 75.7% for controls. Nonresponse was due mostly to refusal of subjects to be interviewed (18.1% of all subjects) and to doctors not granting permission to contact their patients (3.6%). One hundred six women were premenopausal and were excluded from these analyses of postmenopausal breast cancer, thus, leaving 556 cases and 613 controls. Proxy respondents, who were close family members, completed a total of 75 of the questionnaires. After excluding women not living on the Island, we had for the analyses 375 cases and 413 controls.
Table 1 shows the distribution of accepted and potential risk factors for postmenopausal breast cancer. We found increased risks for accepted risk factors (familial history of breast cancer, previous breast disease, education, early age at menarche, later age at first full-term pregnancy, oral contraception therapy, duration of hormone replacement therapy, alcohol consumption). Finding the usual risk factors for breast cancer suggests that the control group was a representative sample of the target. We did not find associations for body mass index, an accepted risk factor for breast cancer, or for breastfeeding and active smoking or environmental tobacco smoke, for which in our opinion there is insufficient evidence to classify these exposures as established risk factors.
Table 2 shows the distribution of modeled concentrations of UFPs assigned to subjects’ addresses, with median concentrations of UFPs for cases being about 23,370 counts per cubic centimeter (/cm3) and for controls 23,100 /cm3. We also show values by duration of residence at that address and by ER/PR status, and the distributions were similar to those above.
Figure 1 shows the fitted response function for UFPs, modeled using a natural cubic spline on three degrees of freedom and adjusted for age, personal, and ecological covariates. The curve was consistent with a linear response, and linearity was also found in all the subanalyses. We, thus, computed for presentation ORs for increases in the IQR of UFPs.
Table 3 shows the ORs for the three statistical models, and we also show risk estimates by duration of residence in their homes at time of diagnosis. The fully adjusted OR per IQR increase (4719 particles/cm3) (Model 3) was 1.08 (95% CI = 0.96–1.21), and the OR for women living at their residence of 10 or more years was 1.06 (95% CI = 0.90–1.24). A larger risk was found for those living at their homes for less than 10 years (OR = 1.27; 95% CI = 0.94–1.71).
Table 4 shows the results of the analyses in which cases were classified by estrogen (ER+ or ER−) and progesterone (PR+ and PR−) status, and by duration of residence at their address at time of interview. We found no increase in risk for cases classified as ER+/PR+ (OR = 0.98) but did find elevated ORs for ER+PR− and ER−/PR− (OR = 1.23; 95% CI: = 1.04–1.45 and OR = 1.23; 95% CI = 0.99–1.54). Analyses by duration of residence was restricted to cases with ER+/PR+ status, as the other subtypes had too few cases, and the ORs were close to the null.
In this hospital-based, case–control study of postmenopausal breast cancer in Montreal, Canada, we determined the association between incidence and estimates of ambient UFPs near participants’ residences. We found an increased risk of about 8% for an increase equal to the IQR, and among cases with ER+/PR− status, we found an increase of 21%.
The findings of our previous analyses of the present study7,8 showed a 31% increase in risk for an increase of 5 parts per billion (ppb) of ambient nitrogen dioxide (NO2) measured at street level. In our previous occupational analysis of this study, we also found associations with occupationally generated carbon monoxide (occurring as combustion products), monoaromatic hydrocarbons such as benzene, especially among cases with ER+/PR− status, and with polycyclic aromatic hydrocarbons from combustion of petroleum liquids, especially among cases with ER+/PR+ status.15
An important strength of the present study was the high response rates, and so it is unlikely that there was an important selection bias, especially as we found associations in the expected directions with most of the accepted risk factors, except for body mass index that did not show a monotonic increase in the ORs. It is possible that the lack of an association for body mass index may suggest some bias in the sample, possibly due to using cancer controls, although the findings could be due to chance. As in all epidemiological studies, residual confounding effects may still be due to misclassification of exposure of UFPs (measured at the residence and not taking into account workplace address or commuting routes) or to excluding other occupational and nonoccupational risk factors (e.g., X- or gamma-radiation, dieldrin, digoxin, ethylene oxide, polychlorinated biphenyls, or shift work),27 although these errors would likely be nondifferential and would have drawn our risk estimates toward the null.
The finding of higher risks among women living in their homes for less than 10 years is counter to the hypothesis that UFPs may be acting as early-stage carcinogens but may have resulted from them acting at a later stage, or from chance, or from other undetected biases.
We have published recently results from a population-based case–control study conducted in Montreal between 2008 and 2011, where we did not find evidence of associations for UFPs across all subjects (ORs in the order of unity per IQR), but we found slightly larger associations among cases with positive estrogen and progesterone receptor status (ORs of about 1.03).18 Differences in the ORs between studies are not dramatic (1.08 in the present study versus unity in the former), and such differences could be expected by chance. There were, however, notable differences between the design and conduct of the two studies. We used the same land use regression model for UFPs as in the present study, but the information on addresses in the preceding study corresponded closely with the measurement campaign that formed the basis of the land use model. In the population-based study, we had relatively low response rates (about 50%), and the associations with accepted risk factors were not as strong as expected. With only two studies of breast cancer and ultrafine particles, it is of course impossible to make any firm conclusions regarding possible associations.
Ground-level UFPs are generated by internal combustion engines and through secondary processes, such as nucleation. The usage of diesel in Canada is much lower than in other jurisdictions; in 2009, the diesel fleet was estimated to comprise 3.1% of light vehicles (19.8 million total), 72% of medium trucks (450,000), and 100% of heavy trucks (320,000).28 (Buses, boats and train locomotives also use diesel fuels.) The distributions of UFPs found in Montreal24 were similar to those reported elsewhere.29 The mean concentration measured across Montreal was about 39,000 /cm3, and higher concentrations of UFPs were found near major roads, rail yards, the international airport, and highways24 (see eFigure 1; http://links.lww.com/EE/A1), and the spatial distribution of UFPs was different from that of NO2.7,8 In our study population, the mean concentration was around 25,000 /cm3, implying that few subjects lived near the major sources of UFPs. Thus, the relatively low ORs for UFPs found in the present study may have been due to a lack of highly exposed subjects.
We showed in an exposure survey in three cities in Canada that the strongest positive predictor variables for UFPs were traffic density, diesel counts, and roadways, and the strongest negative predictor was residential land use.30 The land use map in the supplement figure bears this out, with the highest concentrations found on highways and major roads. It is possible that the mobile monitoring may have overestimated concentrations, as they are not taken at stationary sites adjacent to the roads, but it is not known whether the use of fixed-site monitors would have led to different ORs. In the end, use of any land use regression approach to infer exposures to individuals will have error, and Berkson errors may predominate,31 but the total measurement error is likely to be independent of case status.
We note that diesel vehicles mainly emit particles in the accumulation mode (0.12–2 μm) and gasoline vehicles emissions are mainly composed of nucleation-mode particles (<0.01 μm) and that measured emissions in particle numbers (from about 6-1000 nm) were higher for a diesel versus a gasoline vehicle.32 Under high-speed conditions, emission rates were similar between both types of vehicles, although under most operating conditions, spark-ignition vehicles produced lower emission rates.33 Fruin et al34 observed that on freeways, diesel-fueled vehicles are mainly responsible for UFP concentrations. However, on arterial roads, they found that gasoline-powered vehicles subject to hard accelerations mainly drove UFP concentrations. Hu et al35 partially explained elevated UFP concentrations in a residential neighborhood of Los Angeles by the presence of numerous stop signs and stop lights.
Our exposure model for UFPs was developed in 2011–2012 and, thus, reflects spatial patterns in ambient concentrations of UFPs about 15 years after the identification of subjects. Spatial concentrations of UFPs depend on patterns of both gasoline and diesel traffic (which tend to be concentrated on major roadways), and we cannot say for certain that the spatial patterns in 2011 reflected patterns 15 years previously, especially as there are no fixed-site monitoring stations for this pollutant, and thus, back-extrapolation methods cannot be used. Despite concentrations in Montreal of NO2 and inhalable and fine particles having decreased over time, there is some indirect evidence suggesting that the spatial pattern of UFPs may not have changed dramatically over the 15-year period. First, the road network in Montreal has not changed importantly over the last 15 years, although traffic flows have increased, and there has been recently considerable road construction. Second, Kozawa et al36 found between 2009 and 2011 no important reductions in emissions of UFPs from heavy-duty diesel trucks. Third, late model trucks equipped with new technologies, such as diesel particulate filters, have higher emissions37 of smaller particles (10–20 nm) than older models.38
The association between air pollution and breast cancer
It may be worthwhile to place the results of the present study in context with other studies estimating associations between breast cancer and air pollution. The associations with particulate matter were first investigated in two case-control studies in New York State. Associations in one of two counties on Long Island, New York, were found with “high density” vehicular traffic, but based on a handful of exposed cases.4 In a case–control study in upper New York State, higher concentrations of total suspended particulates measured at the address of time of birth of subjects were associated positively with postmenopausal breast cancer but not with premenopausal breast cancer.5 In a subsequent paper from this study,6 a dispersion model for exposures to benzo[a]pyrene was used as a proxy for emissions to traffic-related pollution, and associations were found among premenopausal women using their address at time of menarche. In the Nurses’ Health Study II,12 no associations were found for incident breast cancer and fine particulate matter (PM2.5), but increased rates were found among premenopausal and postmenopausal women living within 50 m of major roads. In the Sister Cohort,10 adjusted hazards ratios (HR) for incident breast cancer for an increase in the IQR of PM2.5 (3.6 μg/m3), from a smoothed national surface from fixed-site monitors, was 1.03 (95% CI = 0.96–1.11). In the Danish Nurse’s Cohort study, Andersen et al,13 using an atmospheric chemistry transport model (THOR, AirGIS), found no associations with PM2.5 (HR = 0.99 per increase of 3.3 μg/m3; 95% CI = 0.94–1.10) or PM10 (HR = 1.02 per 2.9 μg/m3; 95% CI = 0.94–1.10). In a paper describing the results of analyses of incident postmenopausal breast cancer in 15 European cohorts (ESCAPE), Andersen et al14 made use of separate land use regression models for NO2, PM10, and PM2.5, and back-extrapolated values were assigned to the residential addresses of participants. The pooled estimates were as follows: for PM2.5, the HR was 1.05 (95% CI = 0.77–1.51) for an increase of 5 μg/m3, and for PM10, the HR was 1.00 (95% CI = 0.80–1.25) for an increase of 10 μg/m3.
Ambient NO2 measured at street level is an accepted marker for traffic-related air pollution.16,17 As noted in the Introduction, we found for the present study population an association with a land use model of ambient NO2, as measured at street level; namely, for each increase of 5 ppb, the adjusted OR was 1.31 (95% CI = 1.00–1.71).7,8 As well, we found in the same study associations with a number of occupational risk factors for postmenopausal breast cancer, including combustion-related exposures, especially among women with ER+/PR+ receptor status.15 In the Danish Diet Cancer and Health Cohort,9 the adjusted HR for mostly postmenopausal incident, breast cancer, for an increase of 100 μg/m3 of NOx was 1.16 (95% CI = 0.89–1.51), according to computed concentrations from the Danish AirGIS system. In another case–control study conducted in eight Canadian provinces, we found associations between both postmenopausal and premenopausal breast cancer and concentrations of NO211: for an increase of 10 ppb in scaled satellite observations of NO2 for premenopausal women, we found an OR of 1.32 (95% CI = 1.05–1.67), and for postmenopausal women, we observed an OR of 1.10 (95% CI = 0.94–1.28). In the Sister Cohort,10 for an increase in the IQR of concentrations of NO2 (5.8 ppb), the HR was 1.02 (95% CI = 0.97–1.02), and increased risks were found for cases with positive ER and positive PR status (HR = 1.10; 95% CI = 1.02–1.19). In the Danish Nurse’s Cohort study,13 no association was again found for NO2 (HR = 0.99 per 7.4 μg/m3; 95% CI = 0.93–1.05). In the ESCAPE cohort,14 the pooled estimates were as follows: for NOx, the HR was 1.04 (95% CI = 1.00–1.08) for an increase of 20 μg/m3, and for NO2, the HR was 1.07 (95% CI = 0.98–1.07) for an increase of 10 μg/m3.
Given that the data are still limited, although the ESCAPE project did add considerably to the body of literature, it may be premature to make causal statements regarding breast cancer, especially for premenopausal cancer, but the results are suggestive of a positive association for specific markers of traffic-related pollution (NOx and NO2).
Our findings suggest that exposure to ambient UFPs may increase the risk of incident postmenopausal breast cancer. This is the second study to estimate the relationship between ambient UFPs and breast cancer, and our findings require replication, especially in areas in which diesel usage is much higher.
Conflicts of interest statement
The authors declares that they have no conflicts of interest with regard to the content of this report.
The data collection for the breast cancer study was funded by the Quebec Breast Cancer Foundation, and analyses for this paper were funded through a contract with by Health Canada. Dr. Weichenthal also received support from a GRePEC salary award funded by the Cancer Research Society, the Quebec Ministry of Economy, Science and Innovation, and les Fonds de Recherche du Québec- Santé (FRQS). Because the consent form did not allow for sharing of personal data, individual data cannot be sent to third parties. Use of the data may be possible under certain conditions that the McGill University Institutional Review Board would need to approve. The ultrafine particle data may be obtained from Professor Weichenthal.
2. Canadian Cancer Society’s Advisory Committee on Cancer Statistics. Canadian Cancer Statistics 2015 2015; Toronto, ON: Canadian Cancer Society.
3. Tamimi RM, Spiegelman D, Smith-Warner SA, Wang M, Pazaris M, Willett WC, Eliassen AH, Hunter DJ. Population attributable risk of modifiable and nonmodifiable breast cancer risk factors in postmenopausal breast cancer
. Am J Epidemiol 2016; 184:884–893.
4. Lewis-Michl EL, Melius JM, Kallenbach LR, Ju CL, Talbot TO, Orr MF, Lauridsen PE. Breast cancer risk and residence near industry or traffic in Nassau and Suffolk Counties, Long Island, New York. Arch Environ Health 1996; 51:255–265.
5. Bonner MR, Han D, Nie J, Rogerson P, Vena JE, Muti P, Trevisan M, Edge SB, Freudenheim JL. Breast cancer risk and exposure in early life to polycyclic aromatic hydrocarbons using total suspended particulates as a proxy measure. Cancer Epidemiol Biomarkers Prev 2005; 14:53–60.
6. Nie J, Beyea J, Bonner MR, Han D, Vena JE, Rogerson P, Vito D, Muti P, Trevisan M, Edge SB, Freudenheim JL. Exposure to traffic emissions throughout life and risk of breast cancer: the Western New York Exposures and Breast Cancer (WEB) study. Cancer Causes and Control 2007; 18:947–955.
7. Crouse DL, Goldberg MS, Ross NA. A prediction-based approach to modelling temporal and spatial variability of traffic-related air pollution
in Montreal, Canada. Atmos Environ 2009; 43:5075–5084.
8. Crouse DL, Goldberg MS, Ross NA, Chen H, Labreche F. Postmenopausal breast cancer
is associated with exposure to traffic-related air pollution
in Montreal, Canada: a case-control study. Environ Health Perspect 2010; 118:1578–1583.
9. Raaschou-Nielsen O, Andersen ZJ, Hvidberg M, Jensen SS, Ketzel M, Sorensen M, Hansen J, Loft S, Overvad K, Tjonneland A. Air pollution from traffic and cancer incidence
: a Danish cohort study. Environ Health 2011; 10:67.
10. Reding KW, Young MT, Szpiro AA, Han CJ, DeRoo LA, Weinberg C, Kaufman JD, Sandler DP. Breast cancer risk in relation to ambient air pollution exposure at residences in the Sister Study Cohort. Cancer Epidemiol Biomarkers Prev 2015; 24:1907–1909.
11. Hystad P, Villeneuve PJ, Goldberg MS, Crouse DL, Johnson K. Canadian Cancer Registries Epidemiology Research G. Exposure to traffic-related air pollution
and the risk of developing breast cancer among women in eight Canadian provinces: a case-control study. Environ Int 2015; 74:240–248.
12. Hart JE, Bertrand KA, DuPre N, James P, Vieira VM, Tamimi RM, Laden F. Long-term particulate matter exposures during adulthood and risk of breast cancer incidence
in the Nurses’ Health Study II Prospective Cohort. Cancer Epidemiol Biomarkers Prev 2016; 25:1274–1276.
13. Andersen ZJ, Ravnskjaer L, Andersen KK, Loft S, Brandt J, Becker T, Ketzel M, Hertel O, Lynge E, Brauner EV. Long-term exposure to fine particulate matter and breast cancer incidence
in the Danish Nurse Cohort Study. Cancer Epidemiol Biomarkers Prev 2017; 26:428–430.
14. Andersen ZJ, Stafoggia M, Weinmayr G, Pedersen M, Galassi C, Jorgensen JT, Oudin A, Forsberg B, Olsson D, Oftedal B, Marit Aasvang G, Aamodt G, Pyko A, Pershagen G, Korek M, De Faire U, Pedersen NL, Ostenson CG, Fratiglioni L, Eriksen KT, Tjonneland A, Peeters PH, Bueno-de-Mesquita B, Plusquin M, Key TJ, Jaensch A, Nagel G, Lang A, Wang M, Tsai MY, Fournier A, Boutron-Ruault MC, Baglietto L, Grioni S, Marcon A, Krogh V, Ricceri F, Sacerdote C, Migliore E. Long-term exposure to ambient air pollution and incidence
of postmenopausal breast cancer
in 15 European cohorts within the ESCAPE project. Environ Health Perspect 2017; 125:107005.
15. Labreche F, Goldberg MS, Valois MF, Nadon L. Postmenopausal breast cancer
and occupational exposures. Occup Environ Med 2010; 67:263–269.
16. Beckerman B, Jerrett M, Brook JR, Verma DK, Arain MA, Finkelstein MM. Correlation of nitrogen dioxide with other traffic pollutants near a major expressway. Atmos Environ 2008; 42:275–290.
17. HEI Panel on the Health Effects of Traffic-Related Air PollutionTraffic-Related Air Pollution: A Critical Review of the Literature on Emissions, Exposure, and Health Effects. Special Reports 2010; Boston, MA: Health Effects Institute.
18. Goldberg MS, Labrèche F, Weichenthal S, Lavigne E, Valois M-E, Hatzopoulou M, Van Ryswyk K, Shekarrizfard M, Villeneuve PJ, Crouse D, Parent M-É. The association between the incidence
of postmenopausal breast cancer
and concentrations at street-level of nitrogen dioxide and ultrafine particles
. Environ Res 2017; 158:7–15.
19. Labreche F, Goldberg MS, Valois MF, Nadon L, Richardson L, Lakhani R, Latreille B. Occupational exposures to extremely low frequency magnetic fields and postmenopausal breast cancer
. Am J Ind Med 2003; 44:643–652.
20. World Health Organisation. Research on the Menopause 1981; Geneva, Switzerland: World Health Organization.
21. WHO Scientific Group on Research on the Menopause in the 1990s. Research on the Menopause in the 1990s 1996; Vol. 866Geneva: World Health Organization1–107.
22. Siemiatycki J, Day NE, Fabry J, Cooper JA. Discovering carcinogens in the occupational environment: a novel epidemiologic approach. J Natl Cancer Inst 1981; 66:217–225.
23. Jerrett M, Burnett RT, Kanaroglou PS, Eyels J, Brook JR, Giovis C, Finkelstein N. A GIS - environmental justice analysis of particulate air pollution in Hamilton, Canada. Environ Plan A 2001; 33:955–973.
24. Weichenthal S, Ryswyk KV, Goldstein A, Bagg S, Shekkarizfard M, Hatzopoulou M. A land use regression model for ambient ultrafine particles
in Montreal, Canada: A comparison of linear regression and a machine learning approach. Environ Res 2016; 146:65–72.
25. Chen H, Goldberg MS, Crouse DL, Burnett RT, Jerrett M, Villeneuve PJ, Wheeler AJ, Labreche F, Ross NA. Back-extrapolation of estimates of exposure from current land-use regression models. Atmos Environ 2010; 44:4346–4354.
26. Cao J, Valois MF, Goldberg MS. An S-Plus function to calculate relative risks and adjusted means for regression models using natural splines. Comput Methods Prog Biomed 2006; 84:58–62.
28. Natural Resources Canada. 2009 Canadian Vehicle Survey Summary Report Factsheet 2009; Ottawa, Canada: Natural Resources Canada.
29. Kumar P, Morawska L, Birmili W, Paasonen P, Hu M, Kulmala M, Harrison RM, Norford L, Britter R. Ultrafine particles
in cities. Environ Int 2014; 66:1–10.
30. Weichenthal S, Van Ryswyk K, Kulka R, Sun L, Wallace L, Joseph L. In-vehicle exposures to particulate air pollution in Canadian metropolitan areas: the urban transportation exposure study. Environ Sci Technol 2015; 49:597–605.
31. Armstrong BG. Effect of measurement error on epidemiological studies of environmental and occupational exposures. Occup Environ Med 1998; 55:651–656.
32. Huang C, Lou D, Hu Z, Feng Q, Chen Y, Chen C, Tan P, Yao D. A PEMS study of the emissions of gaseous pollutants and ultra Fi Ne particles from gasoline- and diesel-fueled vehicles. Atmos Environ 2013; 77:703–710.
33. Rickeard DJ, Bateman JR, Kwon YK, McAughey JJ, Dickens CJ. Exhaust Particulate Size Distribution: Vehicle and Fuel Influences in Light Duty Vehicles 1996; Warrendale, PA:: SAE International.
34. Fruin S, Westerdahl D, Sax T, Sioutas C, Fine PM. Measurements and predictors of on-road ultrafine particle concentrations and associated pollutants in Los Angeles. Atmos Environ 2008; 42:207–219.
35. Hu S, Paulso SE, Fruin S, Kozawa S, Mara S, Winer AM. Observation of elevated air pollutant concentrations in a residential neighborhood of Los Angeles California using a mobile platform. Atmos Environ 2012; 51:311–319.
36. Kozawa KH, Park SS, Mara SL, Herner JD. Verifying emission reductions from heavy-duty diesel trucks operating on Southern California freeways. Environ Sci Technol 2014; 48:1475–1483.
37. Bari S, Marian R. Evolution of risk of diesel engine emissions on health during last 4 decades and comparison with other engine cycles: an innovative survey.201557557V012T15A004.
38. Shi JP, Mark D, Harrison RM. Characterization of particles from a current technology heavy-duty diesel engine. Environ Sci Technol 2000; 34:748–755.