Characterization of a developmental toxicity dose-response model.

The Rai and Van Ryzin dose-response model proposed for teratology experiments has been characterized for its appropriateness and applicability in modeling the dichotomous response data from developmental toxicity studies. Modifications were made in the initial probability statements to reflect more accurately biological events underlying developmental toxicity. Data sets used for the evaluation were obtained from the National Toxicology Program and U.S. EPA laboratories. The studies included developmental evaluations of ethylene glycol, diethylhexyl phthalate, di- and triethylene glycol dimethyl ethers, and nitrofen in rats, mice, or rabbits. Graphic examination and statistical evaluation demonstrate that this model is sensitive to the data when compared to directly measured experimental outcomes. The model was used to interpolate to low-risk dose levels, and comparisons were made between the values obtained and the no-observed-adverse-effect levels (NOAELs) divided by an uncertainty factor. Our investigation suggests that the Rai and Van Ryzin model is sensitive to the developmental toxicity end points, prenatal deaths, and malformations, and appears to model closely their relationship to dose.