NTP scientist helps develop predictive model of drug-induced liver injury
A National Toxicology Program (NTP) scientist was part of a team that developed an improved method for predicting drug-induced liver injury (DILI). The new method incorporates potential mechanisms of toxicity into a computer model. Accurate prediction of DILI will lead to a more streamlined drug development process and prevent compounds that cause liver injury from reaching the market.
The researchers used 17 of the Tox21 quantitative high-throughput screening assays to represent a variety of modes of action (MOA). For each MOA, they developed a quantitative structure activity relationship (QSAR) model, then combined the QSAR models for a more robust prediction of DILI. To test this approach, 333 Tox21 chemicals curated in the Food and Drug Administration’s Liver Toxicity Knowledge Base were used in a hold-out testing model.
For each of the thousand model building repetitions, 222 drugs were randomly used as a training set, and 111 drugs were used to validate the new approach compared with a standard QSAR model. This novel DILI-MOA approach drastically improved predictivity over QSAR alone. In addition, the team identified four MOA assays that were the most predictive of DILI, which involved disturbance of the following pathways — peroxisome proliferator−activated receptor gamma, thyroid receptor, estrogen receptor, and the antioxidant response element. (AD)
Citation: Wu L, Liu Z, Auerbach S, Huang R, Chen M, McEuen K, Xu J, Fang H, Tong W. 2017. Integrating drug’s mode of action into quantitative structure-activity relationships for improved prediction of drug-induced liver injury. J Chem Inf Model 57(4):1000−1006.