pollution de l’air provenant de véhicules
TRANSCRIPT
Pollution de l’air provenant de véhicules
et cancer de la prostate à Montréal, Canada
Colloque de l’IRSPUM, octobre 2013 Marie-Élise Parent, PhD
Collaborateurs
Mark S Goldberg ̶ Dept. of Medicine, McGill University, Montréal
Dan L Crouse ̶ Statistics Canada, Ottawa
Nancy A Ross ̶ Dept. of Geography, McGill University
Hong Chen ̶ Public Health Ontario,Toronto
Alexandre Liautaud ̶ School of Population and Public Health, University of British Columbia
Prostate cancer (PCa)
• Most frequent cancer among men
• Etiology poorly understood
• Established risk factors: age, family history of PCa,
ancestry / geography
• Evidence from migrant studies suggests that risk is
influenced by the environment
Role for the environment?
Prostate is under hormone regulation. Could
environmental factors modulating androgens & estrogens
play an etiological role? Still not substantiated
A few studies have reported associations between
traffic-related air pollution and cancer risk
• Breast cancer (2010, Montréal)
• Cervical cancer (2012, Denmark)
• Brain cancer (2012, Denmark)
• Childhood cancers (2013, California)
• Lung cancer (2013, Europe)
Prostate cancer
• Positive associations in ecological mortality
studies (1960s, New York and Tennessee)
• No association (2012, Denmark)
Traffic-related air pollution
• Local vehicular traffic is the primary contributor to air
pollution in urban areas
• Vehicular emissions include gases, particles, volatile
organic compounds, some of which are potential
carcinogens
• Polycyclic aromatic hydrocarbons and
benzo(a)pyrene have hormone-modulating properties
Our question
Is traffic-related air pollution associated with prostate
cancer risk?
A largely unexplored territory…
Parent et al., Occup Environ Med 2013;70:511-8.
PROtEuS:
The Prostate cancer and Environment Study
• Study base: Greater Montreal
• Case-control
• Age less 76 yrs
• 2,000 cases and 2,000 controls
This analysis (restricted to residents of Island of Montreal):
• 803 incident prostate cancers, histologically confirmed
across French Montreal hospitals (2005-2007)
• 969 population controls from French electoral list, age –
matched (± 5 yrs)
• Response rates: 86% and 63% for cases and controls
PROtEuS
Data collection
• Face-to-face interviews
• Questionnaires eliciting
- Socio-demographics (including current address)
- Lifestyle factors
- Detailed work history
• Biological specimens
Characteristics of the 803 cases and 969 controls
Cases Controls
Mean age (yrs) 64 65
Ancestry (%)
French 71 52
African 9 5
Asian 1 3
Other 18 38
Completed ≥ high school (%) 73 78
Mean BMI 27 27
Family history of PCa (%) 20 11
Assessment of exposure to traffic-related
air pollution
NO2 used as marker of traffic-related air pollution
Montreal survey of NO2 levels
• Estimated by measuring 2-week integrated
concentrations of nitrogen dioxide (NO2) using Ogawa
samplers at 129 locations at 3 times in 2005 and 2006
• Based on land use, road, and traffic data, a
regression model was developed to predict
concentrations of NO2 at all points in the city
• Using historical data from fixed-site monitors, the land
use regression model was extrapolated backward to
1996 and 1985
Crouse D et al., Atmospheric Environment, 2009
Montreal survey of NO2 levels
• NO2 levels estimated by measuring 2-week integrated
concentrations of nitrogen dioxide (NO2) using Ogawa
samplers at 129 locations at 3 times in 2005 and 2006
• Based on land use, road, and traffic data, a
regression model was developed to predict
concentrations of NO2 at all points in the city
• Using historical data from fixed-site monitors, the land
use regression model was extrapolated backward to
1996 and 1985
Pictures (1)
Anjou, Boul. Metropolitain Park-Extension
ISEE August 2009 15
Montreal survey of NO2 levels
• Estimated by measuring 2-week integrated
concentrations of nitrogen dioxide (NO2) using Ogawa
samplers at 129 locations at 3 times in 2005 and 2006
• Based on land use, road, and traffic data, a
regression model was developed to predict
concentrations of NO2 at all points in the city
• Using historical data from fixed-site monitors, the land
use regression model was extrapolated backward to
1996 and 1985
Selected
predictor
variables
• Distance from nearest highway
• Traffic count on nearest
highway
• Length of major roads within
100m
• Area of industrial space within
500m
• Population density within
1000m
Model R2 = .80
Land use regression
model to predict annual
concentrations of NO2 at
a resolution of
5 m X 5 m
ISEE August 2009 18
Park Mont Royal
Downtown
Land Use Regression Map Averaged over three Seasons
Montreal survey of NO2 levels
• Estimated by measuring 2-week integrated
concentrations of nitrogen dioxide (NO2) using Ogawa
samplers at 129 locations at 3 times in 2005 and 2006
• Based on land use, road, and traffic data, a
regression model was developed to predict
concentrations of NO2 at all points in the city
• Using historical data from fixed-site monitors, the land
use regression model was extrapolated backward to
1996 and 1985
Distribution of estimates of exposure to
NO2 (ppb)
Minimum Mean Maximum
2006 4.3 11.3 37.4
1996 5.9 15.7 44.5
1985 7.9 20.1 66.8
Hong Chen et al., Atmos Environ 2010
Assigning NO2 exposures to our case-
control series of prostate cancer
1st step: geocoded current addresses
- GeoPinpoint V6.4 (DMTI Spatial, Markham, ON)
- ArcGIS 9.3 geographic information system (GIS)
(ESRI, Redlands, CA)
2nd step: linked current addresses to NO2 exposure
estimates
Results
23
OR for prostate cancer per increase of 5ppb of NO2
Age adjusted Fully adjusted*
Years of estimate of
exposure OR 95%CI OR 95%CI
2005-2006 1.47 1.24-1.75 1.44 1.21-1.73
1996 1.45 1.28-1.65 1.41 1.24-1.62
*Age, ancestry, education, family history of prostate cancer
OR for prostate cancer per increase
of 5ppb of NO2
excluding controls not screened for PC
Fully adjusted*
Years of estimate of
exposure OR 95%CI
2005-2006 1.43 1.18-1.72
1996 1.41 1.22-1.62
*Age, ancestry, education, family history of prostate cancer
Conclusion
NO2, used as a marker of traffic related air pollution,
was associated with prostate cancer risk
Discussion
• Surprising…
• Findings due to chance?
• Confounding? All known risk factors already
accounted for in model
• Bias?
Next step:
Expand NO2 levels to lifetime addresses
The future
• Air pollution and cancer research – still embryonic
• Spatially-referenced data may be particularly useful
for studying the etiology of cancers showing
geographic variations
• Need to further develop (and use!) spatially-
referenced databases
• Associations observed with geographic patterns
can help generate new hypotheses