Development of New Algorithm May Enable Improved, Animal-Free Testing for Chemical Toxicity
Researchers from Rutgers University and other leading academic institutions recently developed a unique algorithm that compares tested chemical compound fragments with those of untested compounds. In doing so, it enables the assessment of their differences and similarities to predict an untested chemical's toxicity. The results of this research demonstrate the potential of this method and the prospect of it reducing reliance on toxicity testing in animals.
Chemical Toxicity Testing
Determining unsafe levels of exposure to chemicals is critical to the safety of individuals throughout the world, including millions of workers. More than 100,000 chemicals currently on the market, many of which are used in consumer products, have not undergone thorough safety testing and are missing important information on their toxicity. While traditional toxicology assessments involve the use of animal models for chemical testing, this can be costly, time consuming, and fraught with ethical implications. A definite need exists for a better toxicity testing method – one that is faster, more accurate, and more cost effective.
A recent study, published in the journal Environmental Health Perspectives, discusses research conducted to test for an improved means of assessing chemical toxicity. Realizing the limits and feasibility issues associated with animal models, Rutgers-led investigators evaluated the use of a newly developed algorithm as part of a non-animal testing model.
Researchers used a group of 7,385 chemical compounds with known toxicity data and compared that with PubChem data for the same chemicals to create in vitro bioprofiles. Using a clustering algorithm, corresponding bioassay groups were identified. An external test set of 600 new chemical substances was then used to validate predictivity. For a number of chemicals, bioassay clusters were found to exhibit a high level of success in oral toxicity prediction, ranging from 62 to 100 percent.
Currently, there are no in vitro testing methods accepted by regulatory agencies as appropriate substitutions for animal tests. Because of this, the development of feasible computational non-animal testing methods has long been sought by the scientific community. The in vitro profiling strategies discussed show significant promise for helping address the growing needs of chemical toxicity prediction and for reducing the dependence on animal testing models.