You are here:
Up to the Task? Quantitative Evaluation Criteria for Physiologically Based Pharmacokinetic Models (SOT 22)
Citation:
Wambaugh, J. AND D. Kapraun. Up to the Task? Quantitative Evaluation Criteria for Physiologically Based Pharmacokinetic Models (SOT 22). SOT, San Diego, CA, March 27 - 31, 2022.
Impact/Purpose:
Session proposal for the Society of Toxicology annual meeting March 2022
Description:
Scientists are frequently faced with the decision of whether to apply a given mathematical model to make important predictions, but for a given application when is a model “up to the task”? In this workshop, we connect this question to physiologically based pharmacokinetic (PBPK) models, which provide mathematical descriptions of organismal absorption, distribution, metabolism, and excretion (ADME) of a substance and which are used in chemical risk assessments to relate externally applied doses (such as oral or inhaled doses associated with adverse health effects) or environmental exposures (such as estimates of oral ingestion via drinking water) to internal dose metrics (such as blood concentrations) that may be more directly related to toxicity. Specifically, we ask, “When is a PBPK model ‘good enough’ – in a quantitative sense – for a given chemical risk assessment application?” Some sources tentatively recommend a PBPK model evaluation criterion based on a “factor of two”. For example, the World Health Organization (WHO) International Programme on Chemical Safety (IPCS, 2010) has stated, “In PBPK modelling, predictions that are, on average, within a factor of 2 of the experimental data have frequently been considered adequate.” However, in a recent peer review of a U.S. Environmental Protection Agency (EPA) dose-response analysis that involved the application of a PBPK model, one toxicology expert stated “this reviewer has never understood why PBPK modelers consider factors of 2 to 3 as reasonable fits of a model to the experimental data”! We believe this reviewer is not alone. This workshop will explore the basis for and adequacy of various evaluation criteria, including the “factor of two” criterion. A diverse group of speakers will share their experiences and perspectives with respect to evaluation of PBPK model fitness through a series of presentations. First, a U.S. EPA quantitative risk assessment scientist will provide context and focus for the discussion of quantitative evaluation criteria for PBPK models. Then, a veteran of risk assessment science that has worked in government and the pharmaceutical industry will elucidate differences in the types of data that one might use to evaluate PBPK models and why these differences matter. Next, a U.S. FDA senior staff fellow will provide insights into “closeness” criteria based on a case study PBPK models for pregnancy. The next speaker, a university professor, will describe the issues of reproducibility, transparency, and evaluation for PBPK models. The final two presentations from distinguished scientists from academia and industry will focus on methodological tools related to model evaluation: one will discuss the importance and relevance of sensitivity analysis and uncertainty quantification for mathematical models in general (with relevant examples) and another will build on that framework by describing alternatives to “data-fit-based” criteria for PBPK models based on methods from decision analysis. Following the prepared presentations, the workshop co-chair will moderate a panel discussion. This abstract does not reflect EPA policy.