Potential systemic availability classification of chemicals for safety assessment

The assessment of chemical safety is essential for protecting human health, yet current approaches have severe limitations. They are unable to address the rapidly growing number of substances requiring evaluation and further rely on animal testing, which the public expects to be phased out for ethical reasons. To address this, the European Partnership for Alternative Approaches to Animal Testing (EPAA) proposed the use of a novel framework for the classification of systemic toxicity based on new approach methodologies. One dimension of this framework is the grouping of substances into different classes of Potential Systemic Availability (PSA) concern (low, medium, high). But so far it remained conceptually and practically unclear how this classification can be achieved. Here, we outline the theoretical considerations for a health-protective definition of PSA concern classes and present a method for the quantitative evaluation of this property. Using high-throughput physiologically based kinetic modelling, we are able to classify the PSA concerns of 139 out of 150 EPAA NAM Designathon compounds. Further, we manually annotate these compounds to evaluate the plausibility of predicted classifications against expert judgement. Our results outline under which circumstances it is appropriate to prioritise or deprioritise chemicals due to their toxicokinetic properties. However, we find that most compounds cannot be assessed on their PSA alone and need to be considered medium PSA concern so that classification into any overall systemic toxicity concern class remains possible. Future integration of bioactivity data will be necessary to fully judge the utility of our method and of the entire framework.