10970215, 2020, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32308 by Cochrane Mexico, Wiley Online Library on [15/11/2022].

See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
IJC
International Journal of Cancer

Association between ambient air pollution and breast

cancer risk: The multiethnic cohort study
Iona Cheng 1,2, Chiuchen Tseng3, Jun Wu4, Juan Yang1, Shannon M. Conroy1, Salma Shariff-Marco1,2, Lianfa Li3, Andrew Hertz5,
Scarlett Lin Gomez1,2,5, Loïc Le Marchand6, Alice S. Whittemore7, Daniel O. Stram3, Beate Ritz8 and Anna H. Wu 3
1
Department of Epidemiology and Biostatistics, University of California, San Francisco, CA, USA
2
Helen Diller Family Comprehensive Cancer Center, University of California, San Francisco, CA, USA
3
Department of Preventive Medicine, Keck School of Medicine, University of Southern California, Los Angeles, CA, USA
4
Program in Public Health, Susan and Henry Samueli College of Health Sciences, University of Irvine, Irvine, CA, USA
5
Cancer Prevention Institute of California, Fremont, CA, USA
6
Epidemiology Program, University of Hawaii Cancer Center, Honolulu, HI, USA
7
Stanford University School of Medicine, Stanford, CA, USA
8
Department of Epidemiology, School of Public Health, University of California, Los Angeles, Los Angeles, CA, USA

Cancer Epidemiology
Previous studies using different exposure methods to assess air pollution and breast cancer risk among primarily whites have
been inconclusive. Air pollutant exposures of particulate matter and oxides of nitrogen were estimated by kriging (NOx, NO2,
PM10, PM2.5), land use regression (LUR, NOx, NO2) and California Line Source Dispersion model (CALINE4, NOx, PM2.5) for
57,589 females from the Multiethnic Cohort, residing largely in Los Angeles County from recruitment (1993–1996) through
2010. Cox proportional hazards models were used to examine the associations between time-varying air pollution and breast
cancer incidence adjusting for confounding factors. Stratified analyses were conducted by race/ethnicity and distance to major
roads. Among all women, breast cancer risk was positively but not significantly associated with NOx (per 50 parts per billion
[ppb]) and NO2 (per 20 ppb) determined by kriging and LUR and with PM2.5 and PM10 (per 10 μg/m3) determined by kriging.
However, among women who lived within 500 m of major roads, significantly increased risks were observed with NOx (hazard
ratio [HR] = 1.35, 95% confidence interval [95% CI]: 1.02–1.79), NO2 (HR = 1.44, 95% CI: 1.04–1.99), PM10 (HR = 1.29, 95%
CI: 1.07–1.55) and PM2.5 (HR = 1.85, 95% CI: 1.15–2.99) determined by kriging and NOx (HR = 1.21, 95% CI:1.01–1.45) and
NO2 (HR = 1.26, 95% CI: 1.00–1.59) determined by LUR. No overall associations were observed with exposures assessed by
CALINE4. Subgroup analyses suggested stronger associations of NOx and NO2 among African Americans and Japanese
Americans. Further studies of multiethnic populations to confirm the effects of air pollution, particularly near-roadway
exposures, on the risk of breast cancer is warranted.

Introduction (LUR) modeling to estimate long-term exposure to NOX and NO2.

To date, the strongest evidence of an association between exposure In contrast, cohort studies in the U.S.5,6 and Denmark,7,8 which
to ambient air pollution and risk of breast cancer is based on results used other air pollution assessment approaches such as dispersion
from the European Study of Cohorts for Air Pollution Effects modeling and kriging interpolation found largely null associations
(ESCAPE) Project of European cohort studies1 and case–control between risk of breast cancer and exposure to NOX, NO2 and par-
studies from Canada2–4; these studies used land use regression ticulate matter (PM). These inconsistent results may be due to

Key words: air pollution, breast cancer, multiethnic cohort, epidemiology

Abbreviations: CA: California; CALINE4: California Line Source Dispersion model, version 4; LUR: land use regression; MEC: multiethnic cohort
Additional Supporting Information may be found in the online version of this article.
Grant sponsor: California Air Resource Board; Grant number: 04-323; Grant sponsor: National Institute of Environmental Health Sciences;
Grant number: P30 ES007048021; Grant sponsor: Health Effect Institute; Grant number: HEI 4787-RFA09-4110-3; Grant sponsor:
National Cancer Institute; Grant numbers: 5-R21-CA094723-03, U01 CA164973; Grant sponsor: National Institute of Environmental Health
Sciences; Grant number: R21 ES016379; Grant sponsor: Susan G. Komen Foundation; Grant number: IIR13262718; Grant sponsor: National
Cancer Institute; Grant number: P30 CA014089
DOI: 10.1002/ijc.32308
History: Received 30 Jul 2018; Accepted 6 Feb 2019; Online 28 Mar 2019.
Correspondence to: Anna H. Wu, NOR 4443, 1441 Eastlake Avenue, 9175, Los Angeles, California 90033, USA, Tel.: +1 323-865-0484,
E-mail: [email protected]

Int. J. Cancer: 146, 699–711 (2020) © 2019 UICC

 10970215, 2020, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32308 by Cochrane Mexico, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
700 Air pollution and breast cancer in a multiethnic population

What’s new?
Prior studies of air pollution and breast cancer generally have employed one method of exposure assessment and have focused
on white women, with inconsistent results. Here, three exposure assessment methods, kriging interpolation, land use regression
(LUR), and California Line Source Dispersion model (CALINE4), were used to investigate associations between long-term air
pollutant exposure and breast cancer risk in the Southern California Multiethnic Cohort. Breast cancer risk was increased in
association with air pollution exposure among women residing near major roads according to kriging and LUR measures.

several factors including the use of different exposure assessment and 118,441 women aged 45–75 years from five self-reported
methods each with known strengths and limitations in spatial and racial/ethnic groups (African Americans, Japanese Americans,
temporal resolution (discussed below). Furthermore, as each study Latinos, Native Hawaiians and whites), residing in Hawaii (HI) or
typically uses only one method of exposure assessment, a compari- CA (primarily Los Angeles County), were enrolled into the MEC.
son of results across different exposure methods (i.e., kriging, LUR, At baseline, participants completed a 26-page mailed question-
dispersion modeling) has not been conducted within a single study. naire with questions pertaining to demographic characteristics,
Cancer Epidemiology

Prior studies also lacked large numbers of nonwhites to examine anthropometrics, reproductive history and other lifestyle factors.
whether associations between air pollution and breast cancer risk Participants were followed prospectively for diagnosis of incident
may vary by race/ethnicity. In the American Cancer Society study invasive breast cancer through routine linkage with the CA and
of air pollution and lung cancer, nonwhites were particularly sus- HI statewide cancer registries, which are a part of the National
ceptible to the effects of air pollution9 but it is not known whether Cancer Institute’s Surveillance, Epidemiology and End Results
Program, and for vital status through linkages to the National
this extends to studies of air pollution and breast cancer.
Death Index and death certificate files. For our study, eligible
To address the research gaps noted above, we conducted a
female MEC participants were those who completed a baseline
large prospective study of ambient air pollution and breast cancer
questionnaire while living in Southern CA, and provided valid
risk within the California (CA) component of the Multiethnic
addresses that were geocoded at the parcel or street segment level
Cohort (MEC) and used three exposure assessment approaches
across the study period (n = 63,511). We excluded women with a
(kriging, LUR and dispersion modeling). Kriging interpolation
breast cancer diagnosis prior to cohort entry if reported on base-
was used to estimate exposures of NOx, NO2, PM2.5 and PM10
line questionnaire or found through linkage with the tumor regis-
from largely regional sources based on continuously collected air
try, and those with implausible dietary (n = 5,858) or address data
monitoring data. A temporally adjusted LUR model was used to
(n = 64), leaving 57,589 women for analyses. This cohort was
estimate NOx and NO2, traffic-related exposures from regional
followed from the date of entry (1993–1996) to the earliest date of
and local sources based on data from spatially dense air monitor-
diagnosis of invasive breast cancer, death or December 31, 2010
ing campaigns, land use factors and traffic characteristics.10,11 The
(study end date), whichever came earlier (mean  SD follow-up
California Line Source Dispersion model, version 4 (CALINE4),
time = 14.7  4.3 years).
air dispersion model, was used to estimate NOx and PM2.5 from
local traffic sources within 1,500 m of participants’ residences,
capturing a small fraction of total emissions and incorporated data Study characteristics and breast cancer risk factors at
on meteorological conditions, traffic counts and roadway net- baseline
works. These three approaches had varying degrees of spatial and Risk factors for breast cancer that we evaluated were first-degree
temporal resolution in exposure coverage. The temporally family history of breast cancer (yes, no), age at menarche (≤12,
adjusted LUR model captured both high spatial and temporal res- 13–14, >14 years), age at first live birth (no children, <20, 21–30,
olution in contrast to kriging interpolation with modest spatial >30 years), number of children (0, 1, 2–3, 4+), menopausal status
and high temporal resolution, and CALINE4 with high spatial (premenopause, natural menopause, oophorectomy, hysterec-
and modest temporal resolution. To the best of our knowledge, tomy), use of hormone therapy (no estrogen use, past estrogen
this is the first prospective study to apply three commonly used use, current estrogen use only, current estrogen use with past or
methods towards assessment of long-term exposure to NOx and current progesterone use), alcohol intake (nondrinker; drinker
risk of breast cancer in a large multiethnic population. In addition, [>0 g/day]) and smoking status (never, former, current). Self-
exposure to NO2 and PM2.5 were estimated by two of the three reported height and weight were used to calculate body mass
assessment methods, providing additional information. index (BMI; under [<18.5 kg/m2], normal [18.5–24.9 kg/m2] and
overweight [25–29.9 kg/m2] and obese [≥30 kg/m2]). Energy
Methods intake (kilocalories per day; quintiles) was based on dietary infor-
Study subjects mation from a self-administered food frequency questionnaire.
The MEC is a large population-based prospective cohort designed Physical activity was based on hours per day spent engaging in
to investigate the etiology of cancer among a multiethnic popula- moderate or vigorous activities (categorized into 0 and quartiles
tion of U.S. adults.12 Briefly, from 1993 through 1996, 96,810 men of nonzero values).12–14 Education (≤high school graduate, some

Int. J. Cancer: 146, 699–711 (2020) © 2019 UICC

 10970215, 2020, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32308 by Cochrane Mexico, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Cheng et al. 701

college, college graduate, graduate and professional school) refers Table S1). A correlation matrix of air pollutants (Supporting Infor-
to the highest level attained. Missing categories were included as mation Table S2) showed a weak correlation between kriging
applicable and missing data for variables such as education, BMI, assessed regional pollutants (oxides and PM) and CALINE4 NOx
smoking status and age at menarche were low (1.6%–2.7%). (R2 < 0.22). LUR modeled pollutants (oxides), representing
regional and local pollutants, were modestly correlated with kriging
Address history, geocoding and contextual data and CALINE4 assessed pollutants (R2 = 0.26–0.61).
The MEC actively maintains accurate and up-to-date addresses on
all participants via periodic mailings of newsletters, follow-up ques- Statistical analysis
tionnaires and linkages to administrative data and registries. For As air pollutant exposures varied over time and the duration of
the 57,589 female Southern CA MEC participants included in our exposures differed across participants, we employed time-
study, there were 94,256 addresses recorded during the follow-up dependent approaches to assess air pollutant exposures and
period. Residential addresses were geocoded to latitude and longi- evaluate their effects on breast cancer risk. For every participant’s
tude coordinates of parcels or street segments whenever parcels residential history across the study period, we calculated a set of
could not be identified. Geocoded addresses from 1993 through cumulative average exposures for a series of time intervals, which
2010 were linked to 1990 (1993–1996 addresses), 2000 (1997–2005 were defined as the time between month (kriging and LUR)/year

Cancer Epidemiology
addresses) and 2010 (2006–2010 addresses) U.S. Census block (CALINE4) of cohort entry and each month/year during the
groups. Each MEC participant was assigned a composite measure follow-up until the censor month/year (i.e., time of breast cancer
of neighborhood socioeconomic status (nSES)15,16 based on the diagnosis, death or study end). These series of average exposures
Census block group of her residential history across the study entered into the Cox proportional hazard models as time-
period that was categorized into quintiles based on the nSES distri- dependent variables via the counting process style of input. Dur-
bution of Los Angeles County block groups. Straight line distances ing the regression calculation, the average exposure across the
were calculated from baseline residential addresses to different road time interval starting from entry time until the time of the event
classes as defined by the U.S. Census Topologically Integrated Geo- was used for risk calculations. The regression model used age at
graphic Encoding and Referencing (TIGER) files: A1 (primary cohort entry (5-year age categories) as a strata variable and also
roads, typically interstate highways, with limited access, division adjusted for breast cancer risk factors at baseline and yearly esti-
between the opposing directions of traffic and defined exits), A2 mates of nSES at baseline and at time of event. Hazard ratios
(primary major, noninterstate highways and major roads without (HR) and 95% confidence intervals (CI) for a standard size
access restrictions) and A3 (smaller, secondary roads, usually with increase in an air pollutant were calculated to compare effect esti-
more than two lanes). mates derived from different exposure assessment approaches for
a specific pollutant. We selected a standard size increase for each
Air pollution exposure assessment pollutant as follows. For NOx, which was measured by all three
In brief, kriging interpolation was used to estimate largely regional methods, we chose 50 ppb, which was close to the IQR of krigged
air pollution exposures for nitrogen dioxide (NO2), nitrogen oxides (50.2 ppb) and LUR (45.6 ppb) although it was almost four times
(NOx), particulate matter with aerodynamic diameter less than higher than the IQR for CALINE4 NOx (8.7 ppb). By selecting
10 μm (PM10) and less than 2.5 μm (PM2.5).17 Measured concen- this common fixed unit, it allowed for comparison across the
trations of NO2, NOx, PM10 in 1993–2010 and PM2.5 in three exposure assessment methods. For the same reason, we
2000–2010 were obtained from U.S. EPA routine air monitoring chose 20 ppb for NO2 as the IQR for NO2 was 16.5 ppb and
data. PM2.5 concentrations in 1993–1999 were estimated from a 18.6 ppb determined by kriging and LUR, respectively. For PM10
published spatiotemporal model that used PM10, meteorological and PM2.5, we chose 10 μg/m3 as this was close to the IQR of
variables and spatial effect as predictors.18 The LUR model was krigged PM10 (8.9 μg/m3) but was higher than the respective
used to estimate regional and local NO2 and NOx exposures based IQRs for krigged (3.8 μg/m3) and CALINE4 PM2.5 (2.5 μg/m3).
on air monitoring data from spatially dense air monitoring cam- We also examined copollutant models in which kriging models of
paigns and incorporated data on land use and traffic characteris- NOx or NO2 were adjusted for PM2.5 and PM10, and LUR models
tics; monthly scaling factors for temporal adjustment were applied of NOx or NO2 were also adjusted for gaseous pollutants deter-
based on routinely collected long-term air monitoring data nearest mined by kriging. As similar findings were observed (data not
to the participant’s home.10,11 We have shown strong correlations shown), single pollutant models are presented.
(R2 = 0.88–0.92) between measured and modeled NOx.19 The We checked the proportional hazard assumption in a model
CALINE4 model was used to estimate local traffic exposures of with all covariates by graphing Schoenfeld residuals against time
NOx and PM2.5 within 1,500 m of a residential location based on and found no violation of this assumption. Stratified analyses
traffic emissions, meteorological and roadway data.20–22 Additional were conducted by race/ethnicity and baseline nSES as associa-
detailed methods are described in the Supporting Information tions may be stronger in nonwhites and by distance of the resi-
Material and Supporting Information Figure S1, including the dis- dential address at baseline to major roads (<200, 200 to <500,
tribution of study participants with available air pollutant data for 500 to <1,600, >1,600 m) as proximity to major roads has been
each exposure assessment approach (Supporting Information found to influence risk in some air pollution studies of lung23–25

Int. J. Cancer: 146, 699–711 (2020) © 2019 UICC

 10970215, 2020, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32308 by Cochrane Mexico, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
702 Air pollution and breast cancer in a multiethnic population

and breast cancers.5 In addition, we conducted analyses separately Among all women, a positive association with breast cancer
for risk of hormone receptor-positive (ER+ and/or PR+; ER+/PR+) was suggested for NOx and NO2 exposure assessed by kriging and
and hormone receptor-negative (ER− and PR−; ER−/PR−) breast LUR as well as PM10 and PM2.5 exposure assessed by kriging
tumors as heterogeneity in breast cancer risk factors by hormone (HR range = 1.04–1.12; p-value range = 0.15–0.62; Table 2).
receptor subtypes is well recognized.26 We conducted sensitivity Although there was no formal statistical evidence of heterogeneity
analyses to examine risk patterns by follow-up time (<10 years vs. in effects by race/ethnicity, statistically significant positive associa-
≥10 years) between air pollution exposure and breast cancer inci- tions were observed in African Americans and Japanese Ameri-
dence. We tested for heterogeneity in associations by race/ethnicity cans but not for Latinos and whites. For example, for LUR NOx,
and distance to a major road by including an interaction term for an increase in exposure by 50 ppb was associated with a 26%
each pollutant with race/ethnicity and distance as applicable. All (95% CI: 1.01–1.58) and 42% (95% CI: 1.05–1.91) increased risk
p values presented are two-sided with a significance level of 0.05. of breast cancer in African Americans and Japanese Americans,
Analyses were performed using SAS 9.2 statistical software (SAS respectively. For CALINE4 NOx., an increase in exposure by
Institute, Cary, NC). 50 ppb was associated with an increased risk of breast cancer in
We conducted a meta-analysis of breast cancer risk in rela- Japanese Americans (HR = 1.97; 95% CI: 0.99–3.92) but a reduced
tion to LUR NOx and NO2 by including published studies that risk (HR = 0.62, 95% CI: 0.39–0.99) in Latinos (pheterogeneity = 0.05).
Cancer Epidemiology

specifically conducted intensive field campaigns to capture Risk associations were generally similar by nSES or by smoking
LUR NOx and NO2 in urban areas. The specific LUR studies status (data not shown).
included in our meta-analyses were the 11 cohorts within the Risk of breast cancer was not significantly associated with resi-
ESCAPE Project1 that collected data on both NOx and NO2 dential distance to major roads. Compared to women who lived
and three Canadian case–control studies that assessed expo- furthest away from major roads (>1,600 m), those who lived
sure to LUR NO2. The published study-specific adjusted risk 500 to <1,600, 200 to <500 and <200 m showed HRs of 1.07 (95%
estimates1 and MEC results were meta-analyzed according to CI: 0.97–1.18), 1.06 (95% CI: 0.94–1.19), and 1.02 (95% CI:
the DerSimonian and Laird random effects model,27 using 0.80–1.18), respectively (Supporting Information Table S3). How-
comparable standardized units of exposure across studies. ever, the risk patterns associated with krigged and LUR pollutants
Between-study heterogeneity was tested by the Cochran’s Q differed in analyses stratified by distance to major roads (i.e., <200,
test and quantified by I2. Meta-analysis was performed using 200 to <500, 500 to <1,600, >1,600 m; Supporting Information
STATA 11 (Stata-Corp, College Station, TX) and additional Table S3). For krigged PM10, LUR NOx and LUR NO2, the respec-
methods are described (Supporting Information Material). tive HR was highest and statistically significant among those living
closest (<200 m) to major roads; the HR was 1.39 (95% CI:
1.02–1.90) for krigged PM10 and 1.39 (95% CI: 1.04–1.86) for LUR
Results NOx and 1.73 (95% CI: 1.19–2.52) for LUR NO2. Risks associated
The study population consisted of 57,589 women (36% African with exposure to krigged NOx and krigged PM2.5 were also higher
American, 10% Japanese American, 38% Latino and 15% white) among those who lived 200 to <500 m of major roads; the respec-
with racial/ethnic differences in the prevalence of obesity, smoking, tive HRs were 1.49 (95% CI: 1.03–2.14) and 1.86 (95% CI:
parity, age at first birth and other breast cancer risk factors 1.02–3.41). When we combined categories for those living <200 m
(Table 1). African Americans (37%) and Latinos (27%) were more and 200 to <500 m of major roads, women who lived within
likely to live in the lowest SES neighborhoods (quintile 1) at baseline 500 m to major roads (Table 3) showed statistically significant
in comparison with Japanese Americans (5%) and whites (7%). increased risks of breast cancer for all krigged (NOx, NO2, PM10,
The proportion of nonmovers over the study period was highest in PM2.5) and LUR (NOx, NO2) pollutants but not for CALINE4 pol-
Japanese Americans (71%) but similar in other racial/ethnic groups lutants (NOx, PM2.5). Results by distance were largely the same in
(55–57%). Thirty percent of the residential addresses at baseline movers and nonmovers (data not shown). In contrast, there were
were within 500 m of major roads, ranging from 23% among no significant increased risks associated with exposure to any of
African Americans to 35% among Latinos. these air pollutants among women who lived >500 m of roadways.
Secular changes and spatial patterns of air pollutant exposure In subgroup analyses by hormone receptor status (Table 4),
were captured in our study (Supporting Information Fig. S2). increased risk of ER−/PR− breast cancer was suggested with a
Using LUR NOx as an example, we presented levels at three time per 10 μg/m3 increase in krigged PM10 for all women (HR = 1.25;
points, such as baseline (1993), mid-study period (2003) and 95% CI: 0.96–1.63) with a larger effect estimate among Japanese
study end (2010), showing a steady decline in this pollutant so Americans (HR = 3.90; 95% CI: 1.34–11.39). Interestingly, in sub-
that the median levels in 2010 were about half of that in 1993 group analyses by hormone receptor status and distance to major
(Supporting Information Fig. S3). Similar patterns of declines roads (data not shown), risk of ER+/PR+ breast cancer was signif-
were observed for the other pollutants we investigated although icantly increased in association with krigged NOx (HR = 1.54,
the magnitude of decline differed. These changing air pollutant 95% CI:1.07–2.21). Risk of ER−/PR− breast cancer was bor-
levels were captured in our Cox proportional hazard models with derline significantly increased in association with LUR NO2
time-varying exposures (see Statistical Analysis). (HR = 1.80, 95% CI: 0.97–3.34) and statistically significantly

Int. J. Cancer: 146, 699–711 (2020) © 2019 UICC

 Table 1. Distributions of breast cancer risk factors and neighborhood factors by race/ethnicity among CA MEC women at baseline, 1993–1996
All women African Americans Japanese Americans Latinos Whites
Cheng et al.

(n = 57,589)1 (n = 20,695) (n = 6,016) (n = 22,037) (n = 8,766)

n % n % n % n % n %
Age at cohort entry
45–49 7,644 13.30 3,120 15.10 753 12.50 2,747 12.50 1,007 11.50
50–54 8,093 14.10 2,988 14.40 736 12.20 3,285 14.90 1,070 12.20
55–59 10,837 18.80 3,064 14.80 912 15.20 5,162 23.40 1,680 19.20
60–64 10,750 18.70 2,852 13.80 1,064 17.70 5,064 23.00 1,752 20.00
65–69 10,090 17.50 4,114 19.90 1,102 18.30 3,263 14.80 1,607 18.30
70–74 8,684 15.10 3,843 18.60 1,186 19.70 2,225 10.10 1,429 16.30
75+ 1,491 2.60 714 3.50 263 4.40 291 1.30 221 2.50

Int. J. Cancer: 146, 699–711 (2020) © 2019 UICC

BMI
Underweight 1,039 1.80 225 1.10 414 6.90 200 0.90 197 2.20
Normal 19,271 33.50 4,760 23.00 4,078 67.80 6,530 29.60 3,882 44.30
Overweight 19,963 34.70 7,277 35.20 1,232 20.50 8,697 39.50 2,726 31.10
Obese 16,080 27.90 7,580 36.60 275 4.60 6,278 28.50 1,927 22.00
Family history of breast cancer
in mother or sisters
No family history 47,172 81.90 16,762 81.00 5,136 85.40 18,069 82.00 7,141 81.50
At least one of mother of 5,781 10.00 2,215 10.70 583 9.70 1856 8.40 1,121 12.80
sisters had breast cancer
Age at first birth
No children 6,852 11.90 2,732 13.20 1,032 17.20 1,781 8.10 1,303 14.90
<20 years 20,976 36.40 9,304 45.00 424 7.00 8,741 39.70 2,487 28.40
21–30 years 24,661 42.80 6,959 33.60 3,795 63.10 9,679 43.90 4,183 47.70
>30 years 3,261 5.70 867 4.20 636 10.60 1,141 5.20 615 7.00
Age at menarche
<12 years 27,218 47.30 9,972 48.20 2,734 45.40 10,118 45.90 4,353 49.70
13–14 years 21,719 37.70 7,552 36.50 2,376 39.50 8,390 38.10 3,376 38.50
>14 years 7,488 13.00 2,654 12.80 816 13.60 3,080 14.00 931 10.60
Parity
0 children (nulliparous) 6,695 11.60 2,684 13.00 1,023 17.00 1,700 7.70 1,284 14.60
1 child 6,652 11.60 3,284 15.90 718 11.90 1,545 7.00 1,094 12.50
2–3 children 21,828 37.90 7,319 35.40 3,307 55.00 7,020 31.90 4,153 47.40
>4 children 21,404 37.20 7,021 33.90 924 15.40 11,297 51.30 2,132 24.30
(Continues)
703

10970215, 2020, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32308 by Cochrane Mexico, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Cancer Epidemiology
 Cancer Epidemiology

704

Table 1. Distributions of breast cancer risk factors and neighborhood factors by race/ethnicity among CA MEC women at baseline, 1993–1996 (Continued)
All women African Americans Japanese Americans Latinos Whites
(n = 57,589)1 (n = 20,695) (n = 6,016) (n = 22,037) (n = 8,766)
n % n % n % n % n %
Menopause status
Premenopause 6,140 10.70 2,284 11.00 805 13.40 2,241 10.20 798 9.10
Natural menopause 27,606 47.90 8,120 39.20 3,399 56.50 11,470 52.00 4,575 52.20
Oophorectomy 7,937 13.80 3,374 16.30 721 12.00 2,449 11.10 1,388 15.80
Hysterectomy 9,848 17.10 4,445 21.50 557 9.30 3,479 15.80 1,358 15.50
Hormone therapy usage
Never estrogen use, with or without 30,958 53.80 11,873 57.40 3,253 54.10 11,966 54.30 3,829 43.70
past or current progesterone use
Past estrogen use, with or without 10,042 17.40 3,987 19.30 757 12.60 3,699 16.80 1,582 18.00
past progesterone use
Current estrogen use alone 6,617 11.50 2,201 10.60 782 13.00 2,262 10.30 1,361 15.50
Current estrogen use, with past 6,307 11.00 1,311 6.30 1,034 17.20 2,267 10.30 1,688 19.30
or current progesterone use
Physical activity, hr/day
0 4,641 8.10 1,210 5.80 149 2.50 2,927 13.30 352 4.00
Quartile 1, <0.4 9,669 16.80 3,631 17.50 868 14.40 4,176 18.90 983 11.20
Quartile 2, 0.4 to <0.7 14,047 24.40 5,936 28.70 1,565 26.00 4,643 21.10 1,886 21.50
Quartile 3, 0.7 to <1.2 12,657 22.00 4,643 22.40 1,483 24.70 4,346 19.70 2,167 24.70
Quartile 4, 1.2 to <13.3 14,308 24.80 4,354 21.00 1,862 31.00 4,847 22.00 3,219 36.70
Energy intake, kcal/day
Quintile 1, 417.4 to <1,158.5 10,994 19.10 4,789 23.10 1,084 18.00 3,504 15.90 1,608 18.30
Quintile 2, 1,158.5 to <1,539.8 10,995 19.10 3,972 19.20 1,451 24.10 3,565 16.20 1,993 22.70
Quintile 3, 1,539.8 to <1,961.1 10,995 19.10 3,716 18.00 1,465 24.40 3,879 17.60 1,920 21.90
Quintile 4, 1,961.1 to <2,633.8 10,995 19.10 3,646 17.60 1,185 19.70 4,402 20.00 1,746 19.90
Quintile 5, 2,633.8 to <7,211.3 10,995 19.10 3,601 17.40 596 9.90 5,733 26.00 1,046 11.90
Alcohol intake
Nondrinker 34,396 59.70 12,455 60.20 4,224 70.20 13,758 62.40 3,910 44.60
Drinker 20,578 35.70 7,269 35.10 1,557 25.90 7,325 33.20 4,403 50.20
(Continues)
Air pollution and breast cancer in a multiethnic population

Int. J. Cancer: 146, 699–711 (2020) © 2019 UICC

10970215, 2020, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32308 by Cochrane Mexico, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
 Table 1. Distributions of breast cancer risk factors and neighborhood factors by race/ethnicity among CA MEC women at baseline, 1993–1996 (Continued)
All women African Americans Japanese Americans Latinos Whites
(n = 57,589)1 (n = 20,695) (n = 6,016) (n = 22,037) (n = 8,766)
Cheng et al.

n % n % n % n % n %
Smoking status
Never smoker 30,985 53.80 9,203 44.50 4,000 66.50 13,719 62.30 4,020 45.90
Former smoker 16,696 29.00 7,013 33.90 1,417 23.60 5,127 23.30 3,122 35.60
Current smoker 8,346 14.50 4,139 20.00 541 9.00 2,150 9.80 1,502 17.10
Education
≤High school graduate 30,329 52.70 8,581 41.50 2,140 35.60 16,046 72.80 3,532 40.30
Some college 16,353 28.40 7,407 35.80 2,167 36.00 3,904 17.70 2,844 32.40
College graduate 5,184 9.00 2,274 11.00 1,028 17.10 776 3.50 1,098 12.50
Graduate and professional school 4,787 8.30 2,108 10.20 624 10.40 856 3.90 1,195 13.60

Int. J. Cancer: 146, 699–711 (2020) © 2019 UICC

Baseline neighborhood SES
Quintile 1—Low 14,597 25.30 7,671 37.10 293 4.90 5,970 27.10 652 7.40
Quintile 2 15,080 26.20 6,006 29.00 754 12.50 6,886 31.20 1,423 16.20
Quintile 3 11,286 19.60 3,156 15.30 1,475 24.50 4,667 21.20 1,966 22.40
Quintile 4 9,740 16.90 2,711 13.10 1,826 30.40 2,832 12.90 2,352 26.80
Quintile 5—High 6,255 10.90 806 3.90 1,646 27.40 1,475 6.70 2,318 26.40
Number of residential moves
over follow-up
0 33,367 57.90 11,840 57.20 4,266 70.90 12,208 55.40 5,010 57.20
1–2 19,592 34.00 7,030 34.00 1,596 26.50 7,748 35.20 3,191 36.40
3–5 4,402 7.60 1,729 8.40 148 2.50 1976 9.00 544 6.20
6+ 228 0.40 96 0.50 6 0.10 105 0.50 21 0.20
Distance of baseline address
to the nearest major road2, meters
<500 17,075 29.60 4,738 22.90 1,939 32.20 7,628 34.60 2,770 31.60
≥500 40,514 70.40 15,957 77.10 4,077 67.80 14,409 65.40 5,996 68.40
Numbers may not total to 100% due to missing.
1
Includes Native Hawaiians (9 cases/75 cohorts); numbers not shown separately.
2
Major roads classified according to U.S. Census: A1 (primary roads, typically interstate highways, with limited access, division between the opposing directions of traffic and defined exits),
A2 (primary major, noninterstate highways and major roads without access restrictions) and A3 (smaller, secondary roads, usually with more than two lanes).
705

10970215, 2020, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32308 by Cochrane Mexico, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Cancer Epidemiology
 10970215, 2020, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32308 by Cochrane Mexico, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
706 Air pollution and breast cancer in a multiethnic population

increased with krigging PM10 (HR = 1.94, 95% CI: 1.15–3.24)

adjusted for age at entry (as a strata variable, 5-year categories), race/ethnicity for all women, BMI, family history of breast cancer, age at first live birth, age at menarche, number of children, menopausal
status, hormone replacement therapy, physical activity, energy intake, alcohol use, smoking, education and neighborhood SES. Phet by race/ethnicity ≥0.05 for all pollutants. Among all women HR and
HR represents the increase in breast cancer per 50 ppb NOX, 20 ppb NO2, 10 μg/m3 PM10, PM2.5 (Kriging); per 50 ppb NOX, 20 ppb NO2 (LUR), per 50 ppb NOX and 1 μg/m3 PM2.5 (CALINE4). Models

95% CI representing increase in interquartile range (IQR) of kriging NOX (50.2 ppb) HR = 1.12 (0.96–1.13), NO2 (16.5 ppb) HR = 1.08 (0.93–1.25), PM10 (8.9 μg/m3) HR = 1.04 (0.94–1.14), PM2.5
(3.8 μg/m3) HR = 1.04 (0.94–1.14); LUR NOX (45.6 ppb) HR = 1.08 (0.97–1.20), NO2 (18.6 ppb) HR = 1.03 (0.91–1.18); CALINE4 NOX (8.7 ppb) HR = 0.99 (0.95–1.04), PM2.5 (0.5 μg/m 3) HR = 1.00
(0.68–1.40)
(0.69–1.47)
(0.85–1.39)
(0.60–1.97)
(0.71–1.20)
(0.75–1.32)
(0.59–2.05)
(0.84–1.26)
and PM2.5 (HR = 5.30, 95% CI: 1.24–22.64).
95% CI
We did not find any differences in results by follow-up
time (<10 years vs. ≥10 years) between air pollutants and risk
of breast cancer (data not shown). As an example, for LUR
0.98
1.00
1.09
1.09
0.92
0.99
1.10
1.03
HR

NOx, the HR was 1.01 (95% CI: 0.87–1.16) for <10 years vs.
1.07 (95% CI: 0.85–1.35) for ≥10 years of follow-up.

Interquartlie range: Kriging NOX (50.2 ppb), NO2 (16.5 ppb), PM10 (8.9 μg/m3), PM 2.5 (3.8 μg/m3), LUR NOX (45.6 ppb), NO2 (18.6 ppb), CALINE4 NOX (8.7 ppb), PM 2.5 (2.5 μg/m3).
Case (n)
Whites

The meta-analysis of the associations between LUR NOx and

512
512
513
513
493
498
447
447
breast cancer among 11 cohorts within the ESCAPE Project1 and
the MEC provide further evidence that NOx exposure impacts
(0.92–1.74)
(0.75–1.62)
(0.90–1.44)
(0.83–2.71)
(0.80–1.24)
(0.67–1.15)
(0.39–0.99)
(0.75–1.01)
breast cancer risk (per 20 μg/m3 meta-analysis HR = 1.023; 95%
95% CI

CI: 1.002–1.046; I2 = 7.2, 11 df; Supporting Information Fig. S4).

Table 2. Associations of gaseous and particulate matter air pollutants and risk of breast cancer by race/ethnicity among CA MEC women, 1993–2010

The meta-analysis of the association between LUR NO2 (per

10 μg/m3) and breast cancer risk also showed an elevated risk that
1.26
1.10
1.14
1.50
0.99
0.88
0.62
0.87
HR

was not statistically significant (HR = 1.020; 95% CI: 0.991–1.049;

Cancer Epidemiology

I2 = 11.1, 14 df; data not shown).

Case (n)
Latinos

834
834
834
833
810
811
742
742

Discussion
To the best of our knowledge, this is the first large multiethnic
(1.05–1.91)
(0.72–2.27)
(0.79–2.55)
(0.86–1.83)
(0.61–3.89)

(0.93–1.92)
(0.99–3.92)
(1.01–1.54)

cohort to employ three exposure assessment approaches (kriging

95% CI

interpolation, LUR and CALINE4 models, Supporting Informa-

tion Fig. S1) to estimate NOx and risk of breast cancer. In addi-
Japanese Americans

tion, exposure history to NO2 and PM2.5 was assessed by two of

1.42
1.28
1.42
1.26
1.54

1.34
1.97
1.25
HR

the three assessment approaches so that we comprehensively and

prospectively evaluated the impact of both gaseous and particulate
Case (n)

matter air pollutants on breast cancer risk in the MEC. Although

315
315
315
315
312
312
273
273

our prior expectations were that temporally adjusted LUR would

provide increased precision and variability of individual exposure,
(1.01–1.58)
(0.93–1.51)
(0.86–1.53)
(0.88–1.17)
(0.72–1.50)

(0.87–1.49)
(0.67–1.86)
(0.90–1.26)

both LUR and kriging were positively but not significantly associ-
95% CI

ated with breast cancer risk. Nevertheless, increased risks of breast

cancer (p < 0.05) were most prominent with exposure to LUR
African Americans

NOx among Japanese Americans and African Americans, and

1.26
1.19
1.15
1.02
1.04

1.14
1.11
1.06
HR

while the effect estimates for NOx based on LUR models had
narrower confidence intervals than for NOx from kriging or CAL-
Case (n)

934
961
882
882
1,024
1,058
1,059
1,057

INE4 models all of these exposures were consistently positively

associated with the outcome. Among women living close to major
roads (<200 m and 200 to <500 m), increased risk of breast cancer
(0.96–1.31)
(0.91–1.31)
(0.95–1.16)
(0.85–1.42)
(0.96–1.22)
(0.90–1.20)
(0.73–1.26)
(0.92–1.09)

was significantly associated with krigged (NOX, NO2, PM10,

95% CI

PM2.5) and LUR (NOX and NO2) derived pollutant measures and
Abbreviations: HR, hazard ratio; CI, confidence intervals.

displayed comparable effect sizes of NOX and NO2 measured by

LUR and kriging. Results for CALINE4 models were less consis-
1.12
1.09
1.05
1.10
1.08
1.04
0.97
1.00
HR

tent as the stratified analysis by proximity to major roadways

(<500 vs. ≥500 m) yielded nearly identical risk estimates, which
All women
Case (n)

may reflect the difficulty to refine exposures when assessing the

2,693
2,727
2,729
2,726
2,557
2,590
2,352
2,352
Note: Values in bold represent p < 0.05.

impact of local traffic within 1,500 m (see below). Finally, the

meta-analysis of MEC results with those from ESCAPE, compris-
Air pollutant1

ing 11 European cohorts1 also indicated a positive association

between LUR NOx and breast cancer.
PM2.5

PM2.5
PM10
NO2

NO2
NOX

NOX

NOX

The significant associations and larger effect sizes were seen

for both particulate matter (PM10, PM2.5) and nitrogen oxides
(NOX, NO2) determined by kriging and LUR models among
assessment

(0.96–1.04).
Exposure

women residing within 500 m of major roadways are intriguing

CALINE4
Kriging

as the proximity to major roadways measure (<200, 200 to <500,

LUR

500 to <1,600, >1,600 m) was not significantly associated with

1

Int. J. Cancer: 146, 699–711 (2020) © 2019 UICC

 10970215, 2020, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32308 by Cochrane Mexico, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Cheng et al. 707

Table 3. Associations of gaseous and particulate matter air pollutants and risk of breast cancer by distance to major roads among CA MEC
women, 1993–2010
Distance to major roads1 <500 m Distance to major roads1 ≥500 m
Exposure assessment Air pollutant Case (n) HR 95% CI Case (n) HR 95% CI
Kriging NOX 781 1.35 (1.02–1.79) 1,912 1.05 (0.87–1.27)
p-value 0.04 0.61
NO2 791 1.44 (1.04–1.99) 1,936 0.98 (0.78–1.21)
p-value 0.03 0.82
PM10 792 1.29 (1.07–1.55) 1,937 0.95 (0.85–1.08)
p-value 0.01 0.45
PM2.5 791 1.85 (1.15–2.99) 1,935 0.89 (0.65–1.21)
p-value 0.01 0.45
LUR NOX 738 1.21 (1.01–1.45) 1,819 1.02 (0.86–1.22)
p-value 0.04 0.80

Cancer Epidemiology
NO2 748 1.26 (1.00–1.59) 1,842 0.93 (0.77–1.12)
p-value 0.05 0.43
CALINE4 NOX 671 0.99 (0.68–1.44) 1,681 1.07 (0.59–1.94)
p-value 0.95 0.83
PM2.5 671 1.01 (0.90–1.14) 1,681 1.04 (0.85–1.26)
p-value 0.85 0.73
Note: Values in bold represent p ≤ 0.05.
HR represents the increase in breast cancer per 50 ppb NOX, 20 ppb NO2, 10 μg/m3 PM10, PM2.5 (Kriging); per 50 ppb NOX, 20 ppb NO2 (LUR), per
50 ppb NOX, 1 μg/m3 PM2.5 (CALINE4). Models adjusted for age at entry (as a strata variable, 5-year categories), race/ethnicity for all women, BMI, fam-
ily history of breast cancer, age at first live birth, age at menarche, number of children, menopausal status, hormone replacement therapy, physical
activity, energy intake, alcohol use, smoking, education and neighborhood SES. Phet by distance to major road >0.05 for all pollutants.
1
Major roads classified according to U.S. Census: A1 (primary roads, typically interstate highways, with limited access, division between the opposing
directions of traffic and defined exits), A2 (primary major, noninterstate highways and major roads without access restrictions) and A3 (smaller, second-
ary roads, usually with more than two lanes).
Abbreviations: HR, hazard ratio; CI, confidence intervals.

risk. Previous studies have examined proximity effects using understand whether air pollution composition, sources of expo-
buffers of 100,23,24 2005 and 300 m.11 We selected multiple cut sure (e.g., vehicle vs. nonvehicle) or some other factors are
points including <200, 200 to <500 and <500 m as previous stud- involved.
ies showed an impact zone of primarily local traffic emission of These findings in the MEC also support recent results from
approximately 300 m in the daytime with good mixing28 and a the ESCAPE Project, which exclusively used LUR models to
wider impact zone before sunrise with stable atmosphere.29 In a assess air pollution exposure, and reported an increased risk of
study of predictors of intracommunity variation in air pollution breast cancer associated with LUR NOx.1 For LUR NOx and
in Los Angeles County, Franklin and colleagues30 found that breast cancer, meta-analysis of MEC results with those from
compared to living at least 1,500 m from a freeway, living within ESCAPE1 demonstrated a significant association with no evi-
250 m of a freeway was associated with 41–75% increase in dence of heterogeneity (Supporting Information Fig. S4) and a
traffic-related air pollutants depending on the size of the urban statistically nonsignificant increased risk in association with LUR
area. These results support our finding of consistent positive HRs NO2. These collective findings suggest that the LUR approach
associated with krigged and LUR air pollutant exposure for those with high temporal and high spatial resolution may be a valuable
who lived <200 and 200 to <500 m (Supporting Information approach for capturing both regional and local long-term air pol-
Table S3). Our results are compatible with studies of lung cancer, lutant exposures.
coronary heart disease and other health endpoints, which have It is of note that CALINE4 (dispersion modeling) estimate of
identified the importance of near-roadway air pollution.31–33 Few NOx, local traffic exposures was not significantly associated with
studies on air pollution and breast cancer have examined proxim- breast cancer risk in our study. Several limitations may have con-
ity to major roads, but results from the Nurses’ Health Study also tributed to uncertainties in CALINE4 model estimates in our
point to the importance of air pollution effects by proximity to study. In CALINE4 simulations, traffic counts in 2002 were scaled
major roads5 with the suggestion of increased risk for women liv- to other years using total vehicle miles traveled for the region in
ing <50 m of the largest road type compared to those living each year, assuming traffic counts decreased or increased at the
>200 m away. These findings support the need for further investi- same scale on all roads. This likely introduced uncertainties in traf-
gations of near-roadway exposures and breast cancer risk to fic counts on individual roads and subsequently affected CALINE4

Int. J. Cancer: 146, 699–711 (2020) © 2019 UICC

 Cancer Epidemiology

708

Table 4. Associations of gaseous and particulate matter air pollutants and risk of breast cancer by hormone receptor status and race/ethnicity among CA MEC women, 1993–2010
All women African Americans Japanese Americans Latinos Whites
Exposure Air Case Case Case Case Case
assessment pollutant1 (n) HR 95% CI (n) HR 95% CI (n) HR 95% CI (n) HR 95% CI (n) HR 95% CI
ER+/PR+ Breast Cancer
Kriging NOX 1,762 1.13 (0.93–1.37) 631 1.19 (0.89–1.61) 233 1.16 (0.59–2.25) 517 1.19 (0.79–1.80) 374 1.11 (0.72–1.71)
NO2 1,789 1.06 (0.85–1.32) 658 1.02 (0.72–1.44) 233 1.09 (0.55–2.16) 517 1.24 (0.76–2.04) 374 1.16 (0.74–1.82)
PM10 1,791 0.97 (0.86–1.09) 659 0.90 (0.76–1.07) 233 0.99 (0.64–1.55) 517 1.12 (0.84–1.50) 375 1.19 (0.89–1.60)
PM2.5 1,791 0.96 (0.71–1.31) 659 0.91 (0.59–1.40) 233 1.12 (0.39–3.25) 517 1.39 (0.67–2.88) 375 1.18 (0.59–2.35)
LUR NOX 1,672 1.07 (0.92–1.25) 570 1.26 (0.94–1.69) 231 1.29 (0.91–1.83) 503 1.00 (0.75–1.32) 361 0.95 (0.69–1.30)
NO2 1,692 0.98 (0.63–1.51) 587 1.07 (0.45–2.52) 231 1.11 (0.38–3.27) 504 0.85 (0.36–1.99) 363 1.03 (0.45–2.38)
CALINE4 NOX 1,560 0.91 (0.64–1.29) 553 1.12 (0.58–2.17) 201 1.16 (0.92–4.65) 466 0.51 (0.27–0.95) 333 1.01 (0.48–1.10)
PM2.5 1,560 0.98 (0.88–1.10) 553 1.07 (0.86–1.33) 201 1.25 (0.97–1.60) 466 0.82 (0.67–1.00) 333 1.01 (0.80–1.28)
ER−/PR− breast cancer
Kriging NOX 430 0.87 (0.59–1.27) 189 0.83 (0.48–1.46) 39 1.00 (0.20–5.00) 140 1.19 (0.54–2.62) 61 0.63 (0.23–1.72)
NO2 435 0.95 (0.60–1.51) 194 1.04 (0.51–2.10) 39 4.28 (0.78–23.40) 140 0.74 (0.29–1.91) 61 0.57 (0.20–1.61)
PM10 435 1.25 (0.96–1.63) 194 1.26 (0.88–1.82) 39 3.90 (1.34–11.39) 140 1.40 (0.78–2.50) 61 0.70 (0.35–1.39)
PM2.5 433 1.25 (0.65–2.44) 193 1.07 (0.44–2.63) 39 5.19 (0.35–76.84) 139 1.63 (0.36–7.34) 61 0.80 (0.14–4.39)
LUR NOX 401 1.22 (0.90–1.64) 169 1.33 (0.79–2.23) 39 2.02 (0.89–4.58) 135 1.02 (0.60–1.73) 57 1.16 (0.54–2.50)
NO2 411 1.20 (0.83–1.72) 176 1.26 (0.67–2.34) 39 3.81 (1.47–9.88) 135 0.77 (0.39–1.53) 60 0.96 (0.42–2.20)
CALINE4 NOX 376 0.96 (0.48–1.94) 159 0.70 (0.19–2.61) 35 2.67 (0.45–15.83) 124 0.63 (0.19–2.08) 57 2.15 (0.46–10.14)
PM2.5 376 1.00 (0.80–1.25) 159 0.88 (0.57–1.37) 35 1.40 (0.83–2.36) 124 0.88 (0.61–1.28) 57 1.25 (0.75–2.08)
Note: Values in bold represent p < 0.05.
HR represents the increase in breast cancer per 50 ppb NOX, 20 ppb NO2, 10 μg/m3 PM10, PM2.5 (Kriging); per 50 ppb NOx, 20 ppb NO2 (LUR), per 50 ppb NOx, 1 μg/m3 PM2.5 (CALINE4). Models
adjusted for age at entry (as a strata variable, 5-year categories), race/ethnicity for all women, BMI, family history of breast cancer, age at first live birth, age at menarche, number of children, meno-
pausal status, hormone replacement therapy, physical activity, energy intake, alcohol use, smoking, education and neighborhood SES. Phet by race/ethnicity ≥0.05 for all pollutants except CALINE4
NOX phet = 0.04.
1
Kriging per 50 ppb NOx, 20 ppb NO2, 10 μg/m3 PM10, PM2.5; LUR per 50 ppb NOx, 20 ppb NO2; CALINE4 per 50 ppb NOx, 1 μg/m3 PM2.5.
Abbreviations: HR, hazard ratio; CI, confidence intervals.
Air pollution and breast cancer in a multiethnic population

Int. J. Cancer: 146, 699–711 (2020) © 2019 UICC

10970215, 2020, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32308 by Cochrane Mexico, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
 10970215, 2020, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32308 by Cochrane Mexico, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Cheng et al. 709

simulations. We also used meteorological data from the closest except for whites. Previous studies by hormone receptor status are
meteorological monitoring stations with more than 75% of com- not consistent. NO2 exposure was associated with ER+/PR+ breast
plete data in a given year. Depending on the missingness in data, cancer in the Sister Study6 and a recent case–control study in
these stations were up to 80 km of each residential location. The Montreal.4 However, high levels of benzene and exposure to traffic-
heterogeneously distributed meteorological stations and frequent related benzo(a)pyrene exposure was implicated for ER−/PR−
missing data in some years likely further affected the quality of the breast cancer respectively in the California Teachers Cohort42 and
CALINE4 estimates. Future studies using higher quality meteoro- Long Island Breast Cancer study.43 Given the incomplete under-
logical data with higher resolution and uniform spatial coverage standing of the etiology of ER−/PR− breast cancers, the potential
may help to refine our CALINE4 assessment and reduce exposure role of air pollution warrants further investigation.
misclassification. The specific mechanism by which gaseous and particulate
Weaker associations for PM2.5, PM10, NO2 and NOx have matter air pollutants may influence breast cancer development is
been reported in prospective cohort studies that used single air not known. Exposure to NOx and NO2 is believed to be a proxy
pollution approaches in the U.S. (spatial–temporal modeling5 of exposure to traffic-related air pollutants, which is a complex
and kriging6), Denmark (dispersion modeling)7,8 and Canada mixture containing numerous polycyclic aromatic hydrocarbons
(satellite-based modeling).34 A recent review of air pollution and (PAHs), benzo(a)pyrene (BaP), benzene, metals and other

Cancer Epidemiology
breast cancer was reported by White et al.35 summarizing these chemicals. PAHs have well-documented mutagenic and carcino-
published studies. Multiple reasons may contribute to these genic effects and have been shown to cause mammary cancers in
inconsistent findings. Accurate assessment of long-term ambient rodent models.44 Traffic-related PAH exposures may increase risk
air pollution exposure is notoriously difficult36,37 with incomplete by increasing the formation of PAH-DNA adducts in breast tis-
residential history as a common limitation. Most prior studies sues.45 The Long Island Breast Cancer Study found a higher risk
have relied on a single residential location at either baseline or at of breast cancer in relation to PAH-adducts in blood lymphocytes
the most recent residence to estimate air pollution exposure with although traffic pollutants were only one of the sources of
much less dense monitoring of exposures prior to 1990. The latter PAHs.46 Methylation may be another potential biologic mecha-
concern also affects case–control studies2,3 and cohort studies6,34 nism as PAH sources have been associated with hypo- and
that typically estimated air pollution exposure based on residen- hypermethylation at multiple promoter regions in breast tumors
tial address at diagnosis of cancer or study enrollment. Some and in blood of control women.35 Two of the ESCAPE cohorts
case–control studies with lifetime residential history38,39 deter- conducted a genome-wide DNA methylation study in relation to
mined historical air pollution monitoring data by imputation of exposure to gaseous and particulate matter pollutants and found
past exposures. Information on residential history prior to cohort an association between global hypomethylation with exposure to
entry was not available in the MEC, but it was complete for up to NOx (p = 0.089) and NO2 (p = 0.014) but not to exposures of par-
18 years after cohort entry—a time of intense air pollution moni- ticulate matter.47 High epigenome-wide DNA methylation in
toring in Los Angeles during which we captured secular changes prediagnostic blood samples has been associated with lower risk
and spatial patterns (Supporting Information Fig. S2). of breast cancer.48 Thus, it is a reasonable hypothesis that hyp-
Previous studies have been largely limited to whites while the omethylation, often a hallmark of genetic instability, in relation to
MEC includes a large, diverse and multiethnic population. Rea- exposure to traffic-related air pollutants may lead to high risk of
sons for the stronger associations observed among African Amer- breast cancer.
icans and Japanese Americans in comparison to Latinos and Despite notable strengths in our study, there are also some
whites in the MEC are not apparent and may be related to geo- limitations. Although exposure to NOx was assessed by all three
graphic differences in residence and associated exposures. African methods (kriging, LUR and CALINE4), NO2 (kriging and LUR)
American and Japanese American communities in the MEC are and PM2.5 (kriging and CALINE4) were assessed by two methods
more clustered in Los Angeles County than whites or Latinos while PM10 was assessed by kriging only. Information on LUR
(Supporting Information Fig. S5). These communities are situ- PM10 and PM2.5 was not available because we used previously
ated along the two sides of the 405 freeway in Los Angeles County published LUR models for Los Angeles,19 Orange49 and San
with exposures not only from vehicle emissions but also refineries, Diego50 counties, which used Ogawa passive air samplers to con-
ports,40 and from the Los Angeles International Airport.41 The duct dense sampling of ambient of NOx and NO2 but not the par-
proximity of African American and Japanese American communi- ticulate matter pollutants. Our CALINE4 approach modeled local
ties to major sources of pollutant exposures may have contributed traffic emissions and characterized Gaussian dispersion but did
to the distinct risk patterns we observed. In addition, we controlled not consider other physical and chemical mechanisms such as
for the established risk factors for breast cancer but found that photochemistry, and hence, we did not have reliable estimates of
racial/ethnic differences in the prevalence of established risk factors NO2 exposure.51,52 While we have estimates of CALINE4 PM2.5,
(e.g., BMI, parity, age at menopause) are unlikely to account for they were highly correlated (R2 = 0.99) with NOx. We do not have
the observed associations. specific emission source information but the krigged estimates
It is intriguing that findings were suggestive for ER−/PR− were based on regional measured pollutant concentrations. Our
breast cancer risk and krigged PM10 exposure in all MEC women LUR models of NOx and NO2 reflected mainly local and regional

Int. J. Cancer: 146, 699–711 (2020) © 2019 UICC

 10970215, 2020, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32308 by Cochrane Mexico, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
710 Air pollution and breast cancer in a multiethnic population

traffic emission as five out of the nine predictor variables were positive associations given the number of comparisons made.
traffic-related but other sources (e.g., industrial, commercial, Finally, we are not aware of and thus unable to control for residual
others) were also predictors of the LUR model.19 We recognize confounding due to any breast cancer risk factors that are also
that the use of distance to major road at baseline address may associated with proximity to major roads or other air pollution
introduce misclassification, yet we observed similar positive asso- exposures.
ciations as seen in Table 3 when restricting our analysis to non- In conclusion, this prospective study of air pollution and
movers. Future studies that also collect emission sources will breast cancer captured long-term spatial and temporal variation
likely advance our understanding of the relationship between in air pollution exposures, using kriging, LUR and CALINE4
ambient air pollution and breast cancer risk. modeling approaches, among Southern California female partici-
Information on residential history prior to cohort entry was pants in the MEC. The collective findings suggest NOx assessed
not available; thus, the influence of exposures in earlier time win- by LUR and air pollutants near major roadways are associated
dows were not investigated. Our assessment is only based on the with breast cancer risk. Additional well-designed studies in multi-
location of residence. Most adults may spend part of the day away ethnic populations comparing air pollution measures based on
from their residences while commuting and working. It is esti- different assessment approaches are warranted to understand the
mated that California adults spent about 87% of their time role of ambient air pollution on breast cancer risk.
Cancer Epidemiology

indoors53; however, we have no assessment of time spent out-

doors for each individual. Thus, while we have captured a key
component of the exposure of interest, random misclassification Acknowledgements
may exist in our assessment of air pollution exposure but this mis- This work was supported by the Susan G. Komen Foundation (IIR13262718)
classification should be nondifferential and would diminish our and the National Cancer Institute (U01 CA164973), National Cancer Insti-
power to identify any true associations.54,55 In addition, it is diffi- tute (P30 CA014089 for AW), National Enviornmental Health Sciences (P30
cult to separate effects from a mixture of air pollutants into its ES007048021 for AW), California Air Resource Board (contract 04-323),
National Institute of Environmental Health Sciences (R21 ES016379), Health
components. Although consistent patterns of associations were Effect Institute (HEI 4787-RFA09-4110-3), National Cancer Institute
observed for certain subgroups (e.g., smoking status, nSES; data (5-R21-CA094723-03). We thank Diana Chingos who served as patient
not shown), we cannot rule out the possibility of chance and false advocate on the study.

References
1. Andersen ZJ, Stafoggia M, Weinmayr G, et al. cohort study. Cancer Epidemiol Biomarkers Prev 16. Yang J, Schupp CW, Harrati A, et al. Developing
Long-term exposure to ambient air pollution 2017;26:428–30. an area-based socioeconomic measure from Ameri-
and incidence of postmenopausal breast cancer 8. Raaschou-Nielsen O, Andersen ZJ, Hvidberg M, can Community Survey data. Fremont, CA: Can-
in 15 European cohorts within the ESCAPE et al. Air pollution from traffic and cancer inci- cer Prevention Institute of California, 2014.
project. Environ Health Perspect 2017;125: dence: a Danish cohort study. Environ Health 17. Laurent O, Hu J, Li L, et al. A statewide nested
107005. 2011;10:67. case-control study of preterm birth and air pollu-
2. Crouse DL, Goldberg MS, Ross NA, et al. Post- 9. Turner MC, Krewski D, Pope CA 3rd, et al. Long- tion by source and composition: California,
menopausal breast cancer is associated with expo- term ambient fine particulate matter air pollution 2001-2008. Environ Health Perspect 2016;124:
sure to traffic-related air pollution in Montreal, and lung cancer in a large cohort of never- 1479–86.
Canada: a case-control study. Environ Health Per- smokers. Am J Respir Crit Care Med 2011;184: 18. Li L, Wu A, Cheng I, et al. Spatiotemporal estima-
spect 2010;118:1578–83. 1374–81. tion of historical PM2.5 concentrations using
3. Hystad P, Villeneuve PJ, Goldberg MS, et al. 10. Jerrett M, Arain A, Kanaroglou P, et al. A review PM10, meteorological variables, and spatial
Canadian cancer registries epidemiology research and evaluation of intraurban air pollution expo- effects. Atmos Environ 2017;166:182–91.
G. Exposure to traffic-related air pollution and sure models. J Expo Anal Environ Epidemiol 2005; 19. Su JG, Jerrett M, Beckerman B, et al. Predicting
the risk of developing breast cancer among 15:185–204. traffic-related air pollution in Los Angeles using a
women in eight Canadian provinces: a case- 11. Wu J, Wilhelm M, Chung J, et al. Comparing distance decay regression selection strategy. Envi-
control study. Environ Int 2015;74:240–8. exposure assessment methods for traffic-related ron Res 2009;109:657–70.
4. Goldberg MS, Labreche F, Weichenthal S, et al. air pollution in an adverse pregnancy outcome 20. Benson P. CALINE4: a dispersion model for
The association between the incidence of post- study. Environ Res 2011;111:685–92. predicting air pollutant concentrations near road-
menopausal breast cancer and concentrations at 12. Kolonel LN, Henderson BE, Hankin JH, et al. A ways. Sacramento, CA: California Department of
street-level of nitrogen dioxide and ultrafine parti- multiethnic cohort in Hawaii and Los Angeles: Transportation, 1989.
cles. Environ Res 2017;158:7–15. baseline characteristics. Am J Epidemiol 2000;151: 21. Wu J, Houston D, Lurmann F, et al. Exposure of
5. Hart JE, Bertrand KA, DuPre N, et al. Long-term 346–57. PM2.5 and EC from diesel and gasoline vehicles
particulate matter exposures during adulthood 13. Cheng I, Kocarnik JM, Dumitrescu L, et al. in communities near the Ports of Los Angeles and
and risk of breast cancer incidence in the Nurses’ Pleiotropic effects of genetic risk variants for Long Beach, California. Atmos Environ 2009;43:
health study II prospective cohort. Cancer other cancers on colorectal cancer risk: PAGE, 1962–71.
Epidemiol Biomarkers Prev 2016;25:1274–6. GECCO and CCFR consortia. Gut 2014;63: 22. Wu J, Laurent O, Li L, et al. Adverse reproductive
6. Reding KW, Young MT, Szpiro AA, et al. Breast 800–7. health outcomes and exposure to gaseous and
cancer risk in relation to ambient air pollution 14. White KK, Park SY, Kolonel LN, et al. Body size particulate-matter air pollution in pregnant
exposure at residences in the sister study cohort. and breast cancer risk: the multiethnic cohort. Int women. Research report 188. Boston, MA: Health
Cancer Epidemiol Biomarkers Prev 2015;24: J Cancer 2012;131:E705–16. Effects Institute, 2016.
1907–9. 15. Yost K, Perkins C, Cohen R, et al. Socioeconomic 23. Beelen R, Hoek G, van den Brandt PA, et al.
7. Andersen ZJ, Ravnskjaer L, Andersen KK, et al. status and breast cancer incidence in California Long-term exposure to traffic-related air pollution
Long-term exposure to fine particulate matter and for different race/ethnic groups. Cancer Causes and lung cancer risk. Epidemiology 2008;19:
breast cancer incidence in the Danish nurse Control 2001;12:703–11. 702–10.

Int. J. Cancer: 146, 699–711 (2020) © 2019 UICC

 10970215, 2020, 3, Downloaded from https://onlinelibrary.wiley.com/doi/10.1002/ijc.32308 by Cochrane Mexico, Wiley Online Library on [15/11/2022]. See the Terms and Conditions (https://onlinelibrary.wiley.com/terms-and-conditions) on Wiley Online Library for rules of use; OA articles are governed by the applicable Creative Commons License
Cheng et al. 711

24. Raaschou-Nielsen O, Andersen ZJ, Beelen R, et al. 35. White AJ, Bradshaw PT, Hamra GB. Air pollution related exposures. IARC Monogr Eval Carcinog
Air pollution and lung cancer incidence in and breast cancer: a review. Curr Epidemiol Rep Risk Hum 2010;92:1–853.
17 European cohorts: prospective analyses from 2018;5:92–100. 46. Gammon MD, Sagiv SK, Eng SM, et al. Polycyclic
the European study of cohorts for air pollution 36. White LF, Yu J, Jerrett M, et al. Temporal aspects aromatic hydrocarbon-DNA adducts and breast
effects (ESCAPE). Lancet Oncol 2013;14:813–22. of air pollutant measures in epidemiologic analy- cancer: a pooled analysis. Arch Environ Health
25. Raaschou-Nielsen O, Andersen ZJ, Hvidberg M, sis: a simulation study. Sci Rep 2016;6:19691. 2004;59:640–9.
et al. Lung cancer incidence and long-term expo- 37. Sheppard L, Burnett RT, Szpiro AA, et al. Con- 47. Plusquin M, Guida F, Polidoro S, et al. DNA
sure to air pollution from traffic. Environ Health founding and exposure measurement error in air methylation and exposure to ambient air pollution
Perspect 2011;119:860–5. pollution epidemiology. Air Qual Atmos Health in two prospective cohorts. Environ Int 2017;108:
26. Yang XR, Chang-Claude J, Goode EL, et al. Asso- 2012;5:203–16. 127–36.
ciations of breast cancer risk factors with tumor 38. Bonner MR, Han D, Nie J, et al. Breast cancer risk 48. van Veldhoven K, Polidoro S, Baglietto L, et al.
subtypes: a pooled analysis from the breast cancer and exposure in early life to polycyclic aromatic Epigenome-wide association study reveals
association consortium studies. J Natl Cancer Inst hydrocarbons using total suspended particulates decreased average methylation levels years before
2011;103:250–63. as a proxy measure. Cancer Epidemiol Biomarkers breast cancer diagnosis. Clin Epigenetics 2015;7:67.
27. DerSimonian R, Laird N. Meta-analysis in clinical Prev 2005;14:53–60. 49. Li LF, Wu J, Wilhelm M, et al. Use of generalized
trials. Control Clin Trials 1986;7:177–88. 39. Beyea J, Stellman SD, Teitelbaum S, et al. Imputa- additive models and cokriging of spatial residuals
28. Zhu Y, Hinds WC, Kim S, et al. Concentration tion method for lifetime exposure assessment in to improve land-use regression estimates of nitro-
and size distribution of ultrafine particles near a air pollution epidemiologic studies. Environ gen oxides in Southern California. Atmos Environ
major highway. J Air Waste Manag Assoc 2002;52: Health 2013;12:62. 2012;55:220–8.
1032–42. 40. Ault AP, Moore MJ, Furutani H, et al. Impact of 50. Ross Z, English PB, Scalf R, et al. Nitrogen diox-

Cancer Epidemiology
29. Hu S, Fruin S, Kozawa K, et al. Wide area of air emissions from the Los Angeles port region on ide prediction in Southern California using land
pollutant impact downwind of a freeway during San Diego air quality during regional transport use regression modeling: potential for environ-
pre-sunrise hours. Atmos Environ (1994) 2009;43: events. Environ Sci Technol 2009;43:3500–6. mental health analyses. J Expo Sci Environ
2541–9. 41. Hudda N, Gould T, Hartin K, et al. Emissions Epidemiol 2006;16:106–14.
30. Franklin M, Vora H, Avol E, et al. Predictors of from an international airport increase particle 51. Finlayson-Pitts BJ, Pitts JN Jr. Chemistry of the upper
intra-community variation in air quality. J Expo number concentrations 4-fold at 10 km down- and lower atmosphere: theory, experiments, and
Sci Environ Epidemiol 2012;22:135–47. wind. Environ Sci Technol 2014;48:6628–35. applications. San Diego, CA: Academic Press, 2000.
31. Heinrich J, Thiering E, Rzehak P, et al. Long-term 42. Garcia E, Hurley S, Nelson DO, et al. Hazardous 52. Sillman S. The relation between ozone, NOx and
exposure to NO2 and PM10 and all-cause and air pollutants and breast cancer risk in California hydrocarbons in urban and polluted rural envi-
cause-specific mortality in a prospective cohort of teachers: a cohort study. Environ Health 2015; ronments. Atmos Environ 1999;33:1821–45.
women. Occup Environ Med 2013;70:179–86. 14:14. 53. Klepeis NE, Nelson WC, Ott WR, et al. The
32. Chen H, Goldberg MS, Villeneuve PJ. A system- 43. Mordukhovich I, Beyea J, Herring AH, et al. National Human Activity Pattern Survey
atic review of the relation between long-term Vehicular traffic-related polycyclic aromatic (NHAPS): a resource for assessing exposure to
exposure to ambient air pollution and chronic dis- hydrocarbon exposure and breast cancer inci- environmental pollutants. J Expo Anal Environ
eases. Rev Environ Health 2008;23:243–97. dence: the Long Island breast cancer study project Epidemiol 2001;11:231–52.
33. Ghosh R, Lurmann F, Perez L, et al. Near- (LIBCSP). Environ Health Perspect 2016;124:30–8. 54. Jurek AM, Greenland S, Maldonado G, et al.
roadway air pollution and coronary heart disease: 44. Maltoni C, Ciliberti A, Pinto C, et al. Results of Proper interpretation of non-differential mis-
burden of disease and potential impact of a green- long-term experimental carcinogenicity studies of classification effects: expectations vs observations.
house gas reduction strategy in Southern Califor- the effects of gasoline, correlated fuels, and major Int J Epidemiol 2005;34:680–7.
nia. Environ Health Perspect 2016;124:193–200. gasoline aromatics on rats. Ann N Y Acad Sci 55. Jurek AM, Greenland S, Maldonado G. How far
34. To T, Zhu J, Villeneuve PJ, et al. Chronic disease 1997;837:15–52. from non-differential does exposure or disease
prevalence in women and air pollution—a 30-year 45. IARC Working Group on the Evaluation of Carci- misclassification have to be to bias measures of
longitudinal cohort study. Environ Int 2015;80: nogenic Risks to Humans. Some non-heterocyclic association away from the null? Int J Epidemiol
26–32. polycyclic aromatic hydrocarbons and some 2008;37:382–5.

Int. J. Cancer: 146, 699–711 (2020) © 2019 UICC