Ineris Direction des risques chroniques Parc technologique Alata BP 2 60550 Verneuil-en-Halatte France
It is essential to discuss the plausibility of “low-dose” effects from environmental exposures. The question, nonetheless, is wrongly labelled, for it is not the magnitude of the dose that matters, but rather the effect. The question thus concerns “doses with low effects”. More precisely, because the low effects on large populations are not that small, even when epidemiological tools fail to detect them, it would be more accurate to talk about “doses with undetectable or barely detectable effects.” Hereafter, we describe this "low-effect dose" concept from the viewpoint of toxicology and epidemiology and discuss the fragile boundary line for these low-effect doses. Next, we review the different types of inference from observed situations (i.e., with high effects) to situations relevant to public health, to characterize the level of confidence to be accorded them. The first type is extrapolation – from higher to lower doses or from higher to lower dose rates. The second type is transposition – from humans to other humans or from animals to humans. The third type can be called “analogy” as in “read across” approaches, where QSAR (Quantitative Structure Activity Relationship) methodology can be used. These three types of inferences can be based on an estimate of the “distance” between observed and predicted areas, but can also rely on knowledge and theories of the relevant mechanisms. The new tools of predictive toxicology are helpful both in deriving quantitative estimates and grounding inferences on sound bases.