[Application of BMD models in quantitative evaluation of carcinogenic risk]

The characteristics of the relationship between the exposure level and the incidence of harmful health effects is a general objective of assessing health risk related to chemical exposure. A real dose-response relationship, particularly for low doses, is usually unknown thus it can only be deduced on the basis of biological and statistical considerations. Low dose risk estimation requires extrapolation beyond the range of data observed. Nowadays, quantitative risk assessment encounters certain duality as linear models are mostly used in evaluation of environmental exposure, and nonlinear models are most frequently applied in occupational exposure. Bearing in mind that the assumption on the dose-response linear relationship is an apparent simplification built on our lack of knowledge of carcinogenic mechanisms, it seems that the proposal of the US Environmental Protection Agency (US EPA) to use benchmark dose methods (BMD) for exposure effects manifested by neoplasms, is the compromise between a conservative approach of the EPA towards risk modelling resulting from the limited knowledge of these mechanisms and lack of criteria for selection of nonlinear models. In the case of carcinogenic risk assessment, this solution involves the identification of so called 'cut off point' by applying relevant statistical models. The cut off point indicates the exposure level that contributes to the increase in risk by 10%, 5% or 1%. Another possibility is to use interspecies extrapolation methods for the cut of point modification, and then draw a line between the cut off point and the beginning of the system of co-ordinates. Since it is not feasible to apply a complete BMD methods for certain chemicals (e.g. bezo(a)pirene), not because of chemical properties, but due to insufficient experimental experience, it is proposed to use the method of sticking together a straight line and a curve in relation to a two-grade curve and not to its confidence limitation.