Based Modeling of Toxicity

To describe a topic to which a whole book could be devoted in just one chapter is a difficult task. One has to address which topics to include and which to omit. For example, there have been several books that address ligand-based approaches for modeling various toxicity endpoints in detail over many chapters, and the reader is recommended to read these to see the broad diversity of computational approaches applied to date. A simple explanation of the 

RSC Drug Discovery Series No. 13 Drug Design Strategies Quantitative Approaches Edited by David J. Livingstone and Andrew M. Davis Royal Society of Chemistry 2012 

Due to space restrictions this chapter will not cover in any detail the structure-based or docking methods which we have compared extensively with machine learning methods in the case of PXR. During the chapter we will highlight how our understanding of molecular causes of toxicity has allowed quantitative modeling, how various pharmaceutical companies actively use such computational models and their impact on drug discovery and toxicity-related attrition. 

3-aminopyrrolidinone farnesyltransferase inhibitors,have also been used to produce individual pharmacophores that were combined, suggesting common areas of positive ionizable features and hydrophobicity from the aromatic 

A recent homology modeling and docking study has suggested that 

---

Less sophisticated but highly useful models have been developed to assess the relative importance of different sources of toxics to the lakes. These screening level models can be equilibrium, steady state or dynamic models. Examples of these are the fugacity-based models of Mackay.17... 

Paquin PR, Zoltay V, Winfield RP, Wu KB, Mathew R, Santore RC, Di Toro DM. 2002b. Extension of the biotic ligand model of acute toxicity to a physiologically-based model of the survival time of rainbow trout (Oncorhynchus mykiss) exposed to silver. Comp Biochem Physiol C 133 305-343. 

GEMS (1986) Graphical Exposure Modeling System. FAP, Fate Atmospheric Pollution. Data Base, Office of Toxic Substances. U.S. Environmental Protection Agency. 

Once developed, these integrated models could be used for several purposes. When combined with Monte Carlo simulation, biomarker models can aid in designing clinical trials that are efficient, powerful, informative, and robust. This integration will continue to improve as mechanism-based models of disease are defined, mechanism-based therapeutic interventions are developed and described, and the relationships between drug exposure and clinical response and toxicity are defined. Functional genomics, proteomics, and lipomics will provide support for defining each of these three factors. 

For small molecules, most prospective (aka predictive ) models are used to identify potential chemical structure-mediated, off-target effects so that chemists can design away from unwanted features and chose the candidate with the overall best profile. The application of prospective models largely encompasses the evaluation of (i) selectivity/promis-cuity, (ii) secondary and safety pharmacology, (iii) intrinsic cytotoxicity, (iv) ADME-based drivers of toxicity, and (v) genotoxidty. This core profiling battery is based on drivers... 

Table 3.2 indicates the criteria by which the US EPA determines that exposure, whether estimated from these models or based on information provided in the PMN, may be significant. If the manufacture or use of a new chemical substance could meet or exceed these criteria, then the agency may require exposure-based testing of toxicity or ecotoxicity in a consent order negotiated under TSCA 5 [50,51,52]. 

The trichloroethylene is oxidized, the gaseous products are removed by the flowing air, and the ehlorine is eaptured by the solid soda and forms salt. The solid salt is removed by diseharging the used OXITOX at the bottom of the reaetor. This is a relatively slow reaetion and the central interest is in removing the last traees of toxic chlorinated compounds (for which TCE is only a model eompound), therefore a very simple model was used. Based on conservation prineiples, it was assumed that chloride removed from the gas phase ends up in the solid phase. This was proven in several material balanee ealeulations. No HCl or other ehlorinated compound was found in the gas phase. The eonsumption rate for TCE was expressed as ... 

Physiologically Based Pharmacodynamic (PBPD) Model—A type of physiologically-based dose-response model which quantitatively describes the relationship between target tissue dose and toxic end points. These models advance the importance of physiologically based models in that they clearly describe the biological effect (response) produced by the system following exposure to an exogenous substance. 

A similarity-related approach is k-nearest neighbor (KNN) analysis, based on the premise that similar compounds have similar properties. Compounds are distributed in multidimensional space according to their values of a number of selected properties the toxicity of a compound of interest is then taken as the mean of the toxicides of a number (k) of nearest neighbors. Cronin et al. [65] used KNN to model the toxicity of 91 heterogeneous organic chemicals to the alga Chlorella vulgaris, but found it no better than MLR. 

Benigni R, Richard AM. Quantitative structure-based modeling applied to characterization and prediction of chemical toxicity. Methods 1998 14 264-76. 

Actively try to include materials that are known to be environmentally benign. Estimate the risk of any proposed new material using computer-based models for eco-toxicity and human toxicity. 

---