Predictions of how toxic chemical compounds are to humans can be made using cell-based methods, without the need for animal testing, finds a study published in Nature Communications this week. This research presents cell-based toxicity models which may aid the development of alternatives to traditional animal testing in measuring compound toxicity.

Ruili Huang and colleagues test more than 10,000 chemicals as part of the Tox21 project that seeks to develop better methods of testing the toxicity of compounds such as pesticides, industrial chemicals, food additives and drugs. They test the activity of the chemicals in 15 different concentrations as they interact with 30 targets, such as nuclear receptors or cellular pathways in human cells. The authors thus generate more than 50 million data points, which they combine with data on the structure of the chemicals to create toxicity models that may be used to predict adverse health outcomes of the chemical compounds in animals or humans.

When these results are compared to toxicity data derived from animal testing or as a result of known exposure effects in humans, the authors find that their models are able to predict both human and animal toxicity. Although these results need to be further tested using additional pathways and targets, the authors propose that cell-based methods can be used in toxicological testing, and could help prioritize which compounds to select for more in-depth toxicological testing.