Modeling the Risk of Drug-Induced Liver Injury with Adverse Outcome Pathways and Bayesian Networks
- Authors:
- Center:
-
Contributing OfficeNational Center for Toxicological Research
Abstract
The development of high throughput in vitro assays has had profound impact on toxicological assessment. It has the potential to lead to more efficient, accurate, and less animal-intensive testing. However, how to take advantage of the numerous in vitro tests in actual risk assessment processes is still a significant challenge. On the other hand, the adverse outcome pathway (AOP) framework has emerged as a rich source for mechanistic knowledge and as a potential tool to select and structure in vitro assays in predictive models for toxicity. In this study, we utilize knowledge encoded in AOPs to build a predictive model for drug-induced liver injury (DILI) using a Bayesian network approach. DILI is an important cause for the abandonment of promising drug candidates as well as drugs withdraws from the market. Due to its importance, there is substantial work to define AOPs that lead to liver toxicity. We reviewed AOPwiki and related literature to construct a comprehensive AOP network regarding DILI, which represents our current knowledge for molecular events that lead to liver injury by drugs and other chemicals. This constitutes a graphical guide to develop in vitro assays to detect liver injury. As the basic structure of AOP networks are directed acyclic graphs, they provide a natural opportunity to construct models with Bayesian networks. In this poster, we present a Bayesian network model based on DILI AOP networks using L1000 and Tox21 data for gene expression and nuclear receptor binding. Due to the incorporation of significant expert knowledge in the form of AOP, the Bayesian network model has the advantage of being parsimonious and requires only a small number of assays to predict the risk of toxicity.