Causal, Predictive or Observational? Different Understandings of Key Event Relationships for Adverse Outcome Pathways and Their Implications on Practice

Abstract

The difference between predictive and causal relationships has been a question in mechanistic toxicology. The adverse outcome pathways (AOPs) methodology is a mechanistic framework that supports the transition to non-animal methods in risk assessment. AOPs include molecular initiating events, key events, and adverse outcomes, linked by key event relationships (KERs) to infer higher-level outcomes from lower biological levels. To better understand the implications of causal and predictive KERs in AOP development, this study aimed to: 1) analyze the use of the terms describing KERs in key AOP guiding documents, and 2) examine their application in AOP development. Our text analysis on the two documents revealed that KER definitions could involve predictive, causal, or both. Examples were also found from studies of AOP development where KER definition is specified to be observational. The implications of these findings are discussed in the context of KER evaluation. We conducted a literature analysis to assess how these terms are adapted across different stages of AOP development, noting that while the use of various KER definitions aligns with specific research objectives, the methodologies for establishing causality in some studies may be questionable.