Determination Of A Geographic Information System Based Indicator To Assess Environmental Dioxins Exposure In Lyon And Through Comparisons With An Atmospheric Dispersion Model Results

Thomas COUDON, Centre Léon Bérard, France
FAURE E. 1 , DANJOU A. 1,2 , CLAVEL-CHAPELON F. 3,4,5 , SALIZZONI P. 6 , FERVERS B. 1,2

1 Département Cancer & Environnement, Centre Léon Bérard, Lyon, France
2 Université Claude Bernard Lyon 1, Villeurbanne, France
3 Inserm, Centre for research in Epidemiology and Population Health (CESP), U1018 Team "Lifestyle, genes and health ", Villejuif, France
4 Paris Sud University, UMRS 1018, Villejuif, France
5 Gustave Roussy Institute, Villejuif, France
6 Laboratoire de Mécanique des Fluides et Acoustique, Ecole Centrale de Lyon, Ecully, France

Purpose 
To investigate the association between environmental dioxin exposure and breast cancer in the French national E3N prospective cohort using a Geographic Information System (GIS), the purpose of the present study was to evaluate the accuracy of a GIS based exposure indicator (GISBEI) of the classification of participants historical dioxin exposure levels, through a comparison with dioxin concentrations computed by an atmospheric dispersion model.
 
Methods
Industrials sources, the main contributors for dioxin exposure over the study period (CITEPA, 2015), were selected and characterized using the Standardized Toolkit for Identification and Quantification of Dioxin and Furan Releases of the United Nation Environment Program. Atmospheric dioxin dispersion was modeled with SIRANE, an urban Gaussian model, for four years: 1996, 2002, 2007 and 2008. The SIRANE model performances were compared to weekly average dioxin concentrations measured within the Lyon urban area in 2007 and 2008.
Through a sensitivity analysis of the SIRANE results, we identified the meteorological and source parameters mostly affecting dioxin concentrations at the residences of the 300 study subjects. We further compared the correlation (Cohen’s kappa coefficients) of study subjects estimated dioxin exposure levels (classified in quintiles) between the 2 methods, the SIRANE model and the GISBEI, for each year, and before and after integration of identified meteorological and source parameters into the GISBEI.
 
Results
Kappa coefficients for study subjects estimated dioxin exposure levels (classified in quintiles) were all below 0.55 for a GISBEI based solely on proximity and ranged from 0.71 to 0.82 after inclusion into the GISBEI of the meteorological and source parameters identified.
 
Conclusions
This study shows the advantages provided by the use of an atmospheric dispersion model in building and evaluating a GIS based indicator for historical environmental dioxin exposure assessment (1990-2008).
 
Funding sources:
Geo3N : ADEME, CLARA, UCBL
E3N :  MGEN, LNCC, IGR, Inserm