Leading applicant:

Fondazione IRCCS "Istituto Nazionale Tumori", Environmental Epidemiology Unit (INT)


Paolo Contiero (paolo.contiero@istutitumori.mi.it)


It is worth noting that all types of pollution phenomena can widely vary across the European countries, depending on fundamental factors of ecological, climatic and geographical nature.Criteria for the selection of areas with an increased breast cancer risk, suitable for exploring the relationship between pollutants concentration and breast cancer, are to be established by an ad-hoc group of experts. Possible examples include:

  • information on typology (type of chemicals) and general level of chemical pollution of each target area must be accessible via databases or digital archives. This allows the acquisition of data on single pollutants and sources of pollution
  • each target area must have an history of contamination and available data for calendar years taking into account the lag time between exposure and diagnosis (e.g. 10 years)
  • when possible, the selected polluted site is defined also in terms of its predominant surrounding conditions. In other words, the target area should be described for the vocational properties of its peripheral landscape (environments, settlements and human activities) by using general land-use indicators (urban, rural, sub-urban, natural, social, residential, industrial, etc.)
  • information on major natural (physical, geochemical and climatic) variables of the target area is recommended
  • information on concentrations and concentration trends of pollutants over time must be available for ground waters and/or soils, and possibly expressed as TEQ (Toxic Equivalents)
  • the integrative use of data resulting from previous studies on ecological indicators, or from other investigations aimed to clarify the biological and ecological effects of chemical pollution, is not mandatory but recommended
  • density population should be taken into consideration

The necessary data to perform the environmental study are those needed for spatial analysis on breast cancer incidence (sub-area or geocode for each cancer cases, total female population by sub-area, deprivation index of the sub-area). In addition to those, the concentration (e.g. μg/l) by sub-area of the modelled pollutant/pollutants (chosen depending on the selected areas) is to be sought. We use generalized additive models to estimate sub-area cancer risk, a form of non-parametric or semi parametric regression with the ability to analyse area-based data adjusting for covariates [Vieira, Env Healt, 2012]. The model is semi-parametric because it includes both nonparametric and parametric components. LOESS smooth is used which adapts to changes in population density where the amount of smoothing depends on the percentage of the data points in the neighbourhood. As already mentioned, this methodology is applied to estimate risk maps, adjusting for: deprivation index alone, concentration of selected pollutant/pollutants alone, deprivation index and pollutant concentration together, and with their interaction. In order to assess the contribution of these factors to the underlying spatial patterns, we compare these maps with those previously estimated by using the crude model (the unadjusted geographic variation in cancer risk)