The following HTML text is provided to enhance online
readability. Many aspects of typography translate only awkwardly to HTML.
Please use the page image
as the authoritative form to ensure accuracy.
Toxicity Testing in the 21st Century: A Vision and a Strategy
ity, and metabolism of an agent of concern. All are conceptually based on the similar-property principle, that is, that chemicals with similar structure are likely to exhibit similar activity (Tong et al. 2003). Accordingly, biologic properties of new chemicals are often inferred from properties of similar existing chemicals whose hazards are already known. Specifically, SAR analysis involves building mathematical models and databases that use physical properties (such as solubility, molecular weight, dissociation constant, ionization potential energies, and melting point) and chemical properties (such as steric properties, presence or absence of chemical moieties or functional groups, and electrophilicity) to predict biologic or toxicologic activity of chemicals. SAR analyses can be qualitative (for example, recognition of structural alerts, that is chemical functional groups and substructures) or quantitative (for example, use of mathematical modeling to link physical, chemical, and structural properties with biologic or toxic end points) (Benigni 2004). Key factors in the successful application of SAR methods include proper representation and selection of structural, physical, and chemical molecular features; appropriate selection of the initial set of compounds (that is, the “training set”) and methods of analysis; the quality of the biologic data; and knowledge of the mode or mechanism of toxic action (McKinney et al. 2000).
Current applications of SAR analyses include soft drug design, which involves improving the therapeutic index of a drug by manipulating its steric and structural properties (Bodor 1999); design and testing of chemotherapeutic agents (van den Broek et al. 1989); nonviral gene and targeted-gene delivery (Congiu et al. 2004); creating predictive models of carcinogenicity to replace animal models (Benigni 2004); predicting the toxicity of chemicals, particularly pesticides and metals (Walker et al. 2003a); and predicting the environmental fate and ecologic effects of industrial chemicals (Walker et al. 2003b). Among the available predictive-toxicity systems, the most widely used are statistically based cor-