Physiologically-based pharmacokinetic modelling for the prediction of adverse drug reactions

  • Physiologiebasierte Modellierung für die Vorhersage von unerwünschten Medikamentenwirkungen

Baier, Vanessa; Blank, Lars M. (Thesis advisor); Küpfer, Lars (Thesis advisor)

1. Auflage. - Aachen : Apprimus Verlag (2023)
Book, Dissertation / PhD Thesis

In: Applied microbiology 29
Page(s)/Article-Nr.: 1 Online-Ressource : Illustrationen, Diagramme

Dissertation, RWTH Aachen University, 2022


Adverse drug reactions endanger patients’ health and pose a considerable challenge to drug development and medical care. Despite a variety of approaches ranging from in silico up to clinical studies, predicting drug toxicity still fails in many cases due to limited inter-assay or cross-species translatability and the idiosyncrasy of many drug effects. Thus, findings from diverse sources, such as in vitro assays or animal models, need to be jointly analysed and contextualised with individual patient conditions, e.g., diseases, specific genotypes, or co-medications. Thereby, a systemic understanding and reliable predictions of adverse reaction risks become possible. However, experiments mimicking realistic patient scenarios are frequently expensive, infeasible, and insufficient. Therefore, integrating data from different levels into mechanistic in silico models has emerged as a promising and cost-effective alternative to overcome the imbalance between the lack of viable and sound models and the necessity to predict adverse drug reactions effectively. In this work, computational modelling was applied to identify drugs with a high risk of inducing hepatic adverse drug reactions as well as patients prone to experience such. Predisposing patient factors associated with drug toxicity were considered throughout the studies to account for the idiosyncrasy of adverse drug reactions. A model of bile acid circulation was developed to investigate drug-induced cholestasis by coupling it to a drug-specific whole-body physiologically-based pharmacokinetic model. Through contextualisation of physiological knowledge, pharmacokinetic data, genotype, and in vitro inhibition data, the model allowed the simulation of bile acid levels in healthy individuals and confirmed cholestasis susceptibility for familial cholestasis genotypes during cyclosporine A treatment. The further integration of time-resolved expression data from a drug-treated in vitro assay into the model enabled a systematic categorisation of the cholestasis risk of several hepatotoxic drugs. By providing a framework to benchmark potentially cholestatic drugs against a reference dataset of ten drugs, this approach could support the identification of drug-induced cholestasis in drug development in the future. Finally, to assist patient safety in clinical care, computational modelling was utilised to guide a clinical test strategy striving for a personalised treatment decision by investigating the individual metabolic phenotype of a patient. The simulations of virtual populations permitted to differentiate between biometric and metabolic contributions to drug exposure. Subsequently, recommendations for the test strategy were derived to support optimal study design in terms of sampling time points or selection of compounds. The presented approaches support the early identification of adverse drug reactions during drug development as well as in routine health care. Thus, by elucidating the link between individual patient factors and adverse drug reactions, this work can be employed to increase patients’ safety and optimise drug development in the future.