Using adverse outcome pathways to contextualise (Q)SAR predictions for reproductive toxicity - A case study with aromatase inhibition

The development and application of (quantitative) structure-activity relationship ((Q)SAR) models for reproductive toxicology remains challenging, given the complexity of the endpoint and the risks associated with subsequent decision making. Adverse outcome pathways (AOPs) organise knowledge and provide context of model outputs, aiding risk assessors' decision making. Using aromatase as an example, we demonstrate how AOPs can be used to contextualise a variety of (Q)SAR approaches. AOPs stemming from aromatase inhibition - leading to adverse outcomes of regulatory significance - were synthesised and annotated with relevant assays, assay data and (Q)SAR models. The resulting framework enabled the deployment of different types of (Q)SAR models that predict for key events along the pathway. The use of models for molecular initiating events enables relevant knowledge to span a wider area of chemical space - compared to using models trained solely on in vivo toxicity data. Utilising such methods, alongside additional assay data and exposure information, could lead to improved risk assessment strategies during compound prioritisation and labelling.

Automated summary

The article introduces the use of adverse outcome pathways (AOPs) to provide context for (Q)SAR methods, using aromatase as an example. By utilizing different types of (Q)SAR models to predict key events, along with additional assay data and exposure information, improved risk assessment strategies for compound prioritization and labeling could be achieved.