Purpose: We illustrate a promising approach to integrating mechanistic knowledge with multiple data sources for in vitro assays, using drug-induced liver injury (DILI) as an example. Methods: The adverse outcome pathway (AOP) framework was used as an integration tool for mechanistic knowledge and for the selection of a subset of predictors from high-dimensional in vitro data sets. A subset of assays from the Tox21 and L1000 projects was selected as predictors in the predictive modeling of DILI risk with the guidance of DILI related AOPs. A sparse logistic model with elastic net penalty was then constructed with these predictors. Results: Most drugs classified as most-DILI-concern were mapped to AOPs related to liver toxicity found in AOPwiki. This model predictive model using selected assays obtained a high accuracy. The prediction results have also been corroborated by data for adverse event reporting from the FDA FAERS database. Conclusion: The results confirm the value of AOPs as a knowledge source for understanding adverse events. They demonstrate the power of integrating mechanistic knowledge with high throughput assays for toxicological evaluations. Corroboration with independent FAERS data further lend credence to the proposed approach.
References: Khadka, K. K., Chen, M., Liu, Z., Tong, W., & Wang, D. (2020). Integrating adverse outcome pathways (AOPs) and high throughput in vitro assays for better risk evaluations, a study with drug-induced liver injury (DILI).