A forum article published on December 19, 2016 in the peer-reviewed journal Toxicological Sciences discusses Adverse Outcome Pathways (AOPs) and how this framework can support the development and use of computational prediction models for regulatory toxicology. Clemens Wittwehr from the European Commission (EC)’s Joint Research Centre (JRC), together with collaborators from 11 other international organizations, recapitulates the ongoing efforts on transforming regulatory toxicology and chemical safety assessment “from a largely empirical science based on direct observation of apical toxicity outcomes in whole organism toxicity test to a predictive one in which outcomes and risk are inferred from accumulated mechanistic understanding.”
The AOPs framework offers a systematic approach to collecting and organizing knowledge on toxicity mechanisms (see FPF background article), while computational models of biological systems are being developed with the aim to facilitate the inference and extrapolation across scales and species. The authors argue that knowledge organized in AOP networks can be used to guide the development of computational prediction models aimed to support the integration of mechanistic, in silico, and in vitro data into chemical safety assessment workflows. This notion was explored as part of a workshop “AOP-informed predictive modeling approaches for regulatory toxicology,” held in September 2015. The paper presents four examples of AOP-informed model development and use, one on skin sensitization in mammals, and three more on endocrine disruption, namely “aromatase inhibition leading to reproductive impairment in fish,” “inhibition of thyroid hormone synthesis/degradation leading to varied developmental effects,” and “activation of estrogen receptor-α leading to diverse adverse outcomes.”
In addition, the authors discuss the importance of problem formulation in ensuring that the developed predictive models are “fit for regulatory purposes.” All stakeholders, including model developers, intended users, and decision makers, should collaborate during the problem formulation phase in order to collectively define “the scope and regulatory purposes for which a prediction model is needed, the type of model best suited to meet the regulatory objectives, the data criteria, and model’s domain of applicability.” This activity ensures that all parties agree on the “degree of uncertainty that can be tolerated in AOP and/or associated quantitative models.” Given the importance of crowd-sourcing for the success of predictive toxicology initiatives, the authors conclude by calling on the modeling community to actively engage in the development of AOP-informed computational models.
In December 2016, the industry organization European Centre for Ecotoxicology and Toxicology of Chemicals (ECETOC) published a guidance on assessment and application of the endocrine system-relevant AOPs, while a non-government organization Pesticide Action Network (PAN) Europe released a report called “AOP, the Trojan horse for industry lobbying tools?,” advising against the use of AOPs as a final decision-making tool in risk assessment (FPF reported).
JRC (February 3, 2017). “How Adverse Outcome Pathways can support computational approaches to predictive toxicology.”
ECETOC (December, 2016). “Guidance on assessment and application of Adverse Outcome Pathways (AOPs) relevant to the endocrine system.” Technical Report No. 128 (pdf)
PAN Europe (December, 2016). “AOP, the Trojan horse for industry lobbying tools?” (pdf)
Wittwehr, C., et al. (2016). “How adverse outcome pathways can aid the development and use of computational prediction models for regulatory toxicology.” Toxicological Sciences 155: 326-336.