Quantitative risk assessment and the limitations of the linearized multistage model

Publisher Website
Google Scholar
Abstract
1 Quantifying carcinogenic risk is an important objec tive for assisting in the assessment and management of risks from chemical exposure. The most widely used ofthe many mathematical models proposed for extrapolation of carcinogenicity data from animal studies to low dose human exposures is the linearized multistage (LMS) model. This has, in effect, become the default approach for much of Quantitative Risk Assessment (QRA). The practical properties of this model have been investigated. 2 Analysis of simulated data using the LMS model showed (i) that the Maximum Likelihood Estimate (MLE) of the low dose slope, q1, was unstable and extremely sensitive to small changes in the data; (ii) the 95% Upper Confidence Limit (UCL) estimate, q1*, preferred by the US Environmental Protection Agency (EPA) was insensitive with only small changes in values being obtained for large changes in the data; (iii) data sets where there was no statistical significance could give risk estimates similar to those obtained from data sets with clear dose-related effects; (iv) the size of the values of the Virtually Safe Dose (VSD) obtained did not necessarily relate to the biological interpretation of the data sets; (v) the value of q1* obtained was closely related to the top dose used in the study. 3 Limitations of the LMS model were illustrated by examples of its use in assessing the carcinogenicity of 2, 3, 7, 8-TCDD leading to the conclusion that the existing models are not suitable for routine use in the estimation of the risk from chemical carcinogens. The use of the LMS model has been justified in part by its original derivation from a mathematical model based upon a multistage model of carcinogenesis. However, estimates of the parameters of the model used to provide estimates of low dose risk to humans have no direct relationship to specific biological event in carcinogenesis. Further developments in mathematical models and increased understanding of the biological events underlying carcinogenesis will lead to more biologically plausible QRA methods which would then justify serious consideration of QRA by regulatory authorities throughout the world.