You are here:
Predictive Models and Computational Toxicology
Citation:
KNUDSEN, T. B., M. T. MARTIN, K. J. CHANDLER, N. KLEINSTREUER, R. JUDSON, AND N. SIPES. Predictive Models and Computational Toxicology. Chapter 26, Paul C.Barrow (ed.), Teratogenicity Testing; Methods and Protocols. Humana Press Incorporated, Totowa, NJ, 947:343-374, (2013).
Impact/Purpose:
This chapter has addressed some of the critical issues and questions for computational approaches to understanding, modeling, and predicting developmental toxicity. Recent advances in computing power allow for the integration and correlation of vast amounts of data. This greatly extends our ability to identify and understand those biological pathways leading to adverse impacts, to make better predictions about human health risk and to model developing systems with an unprecedented degree of complexity. Computer simulation of ‘virtual tissues’ aims to recapitulate complex cell behaviors, and to provide a cell-agent-based multicellular modeling platform to reproduce developmental events and model the effects of chemical perturbations.
Description:
Understanding the potential health risks posed by environmental chemicals is a significant challenge elevated by the large number of diverse chemicals with generally uncharacterized exposures, mechanisms, and toxicities. The ToxCast computational toxicology research program was launched by EPA in 2007 and is part of the federal Tox21 consortium to develop a cost-effective approach for efficiently prioritizing the toxicity testing of thousands of chemicals and the application of this information to assessing human toxicology. ToxCast addresses this problem through an integrated workflow using high-throughput screening (HTS) of chemical libraries across more than 650 in vitro assays including biochemical assays, human cells and cell lines, and alternative models such as mouse embryonic stem cells and zebrafish embryo development. The initial phase of ToxCast profiled a library of 309 environmental chemicals, mostly pesticidal actives having rich in vivo data from guideline studies that include chronic/cancer bioassays in mice and rats, multigenerational reproductive studies in rats, and prenatal developmental toxicity endpoints in rats and rabbits. The first phase of ToxCast was used to build models that aim to determine how well in vivo animal effects can be predicted solely from the in vitro data. Phase I is now complete and both the in vitro data (ToxCast) and anchoring in vivo database (ToxRefDB) have been made available to the public (http://actor.epa.gov/). As Phase II of ToxCast is now underway, the purpose of this chapter is to review progress to date with ToxCast predictive modeling, using specific examples on developmental and reproductive effects in rats and rabbits with lessons learned during Phase I.
