Validation of Drug Toxicity Prediction using DiscoveRx Model

Data from BioMAP Systems analysis of 776 environmental chemicals, including reference pharmaceuticals and failed drugs, on their ability to disrupt physiologically important human biological pathways were published this week in Nature Biotechnology. The results show that this in vitro approach can reproducibly identify potential toxicities and off-target drug effects, as well as pinpoint cellular mechanisms and specifically affected biomarker endpoints underlying specific types of adverse reactions in humans. DiscoveRx Corporation’s BioSeek division conducted the analysis in collaboration with scientists from the U.S. Food and Drug Administration, National Institutes of Health and U.S. Environmental Protection Agency as part of the EPA ToxCast Program.

Assessing the safety aspects of drugs, consumer products and environmental chemicals has been historically undertaken through animal testing. However, the vast number of chemicals needing such toxicological assessment  and the fact that results in animals often do not translate well to results in humans require the development of alternative, faster, more accurate and humane testing approaches.

“This publication examines an unprecedentedly large data set in terms of number of chemicals, chemical diversity and types of assays screened phenotypically in BioMAP primary human cell models of tissue biology and disease,” said Ellen L. Berg, Ph.D., Scientific Director and General Manager of DiscoveRx’s BioSeek division and an author of the publication.  “In contrast to screening approaches aimed at understanding the actions of a single agent at the molecular and mechanistic level, this method harvests the collective knowledge embedded in reference chemicals with respect to their molecular targets, mechanisms of action, and animal and human toxicity and applies it to characterizing the biological activity of new chemicals or medicines.

“Our results show such systems to be a highly useful and reproducible tool for predictive toxicology that can identify potential chemical targets, toxicological liabilities and molecular mechanisms that elucidate specific adverse outcome pathways for drugs and other chemicals,” Dr. Berg continued. “Even using a limited set of primary human cell systems, we were able to recognize consistent patterns of activity that were closely correlated with diverse drug actions and toxicities. New chemicals falling into profile clusters with known activities suggest specific potential toxicities for more careful evaluation, greatly increasing the efficiency of toxicity testing by focusing resources for follow-up testing on the bioactivities of highest concern.”

For example, drugs and chemicals, including selective estrogen receptor modulators, tamoxifen and raloxifene, that are associated with thrombosis-related side effects, like deep vein thrombosis (DVT), were found to preferentially increase the levels of tissue factor in BioMAP models of vascular inflammation.  Thus, the use of these models for screening earlier in the drug discovery process may help identify new medicines with reduced potential for this debilitating side effect.