Tiered Approaches to Incorporate the Adverse Outcome Pathway Framework into Chemical-Specific Risk-Based Decision Making

Abstract

Listen

Translate article

The concept of Adverse Outcome Pathways (AOPs) arose as a means of addressing the challenges associated with establishing relationships between high-throughout (HT) in vitro dose response data and in vivo biological outcomes. However, AOP development has also been met with challenges of its own, such as the time, effort, and expertise necessary to achieve a scientifically sound construct able to support ecotoxicology and human health risk assessment. Thus, a staged development process has been developed to match the information content of an AOP with the decision context in which it will be used. This approach allows effort to be spent on detailed evidence evaluation and quantitative assessment of the dose-response characteristics for those AOPs where this level of confidence and precision is needed. In addition, through advances in computational analytical methodologies that integrate HT data (e.g., transcriptomic data) with traditional toxicology information spanning a broad chemical and biological space, computationally predicted AOPs can be rapidly generated to help accelerate the curation of AOPs. AOPs are chemical agnostic thereby allowing a single AOP to be coupled with in vitro dose-response information from a variety of chemicals. To predict an in vivo outcome, however, exposure and pharmacokinetic characteristics (i.e., absorption, metabolism, distribution, and elimination) must be considered. As with the staged development process for AOPs, it is possible to develop ADME predictions in a tiered manner such that lower tiers provide qualitative or semi-quantitative predictions when data is lacking, and higher tiers provide quantitative predictions with increasing confidence when data is abundant. Tiered approaches to AOP development and ADME predictions provide a mechanism for using AOPs, with chemical-specific exposure and pharmacokinetic considerations, for risk assessment both in data poor and data rich scenarios. They also provide a natural mechanism for identifying areas of research that would have the highest impact on risk-based decision making by highlighting AOPs and/or ADME predictions that are insufficient to address the decision context in which they could be used.