Natural product – drug interactions: From quantitative prediction to clinical assessment

Patients often supplement their prescribed drug regimens with herbal and other natural products (NPs), raising concern for negative NP-drug interactions. Pharmacokinetic mechanisms underlying these interactions include inhibition of the pre-systemic and/or systemic metabolism of the 'victim' drug by one or more 'perpetrator' NP constituents, leading to an increase in systemic drug exposure and potential unwanted effects. Unlike most drug products, NPs typically are mixtures of multiple constituents that vary in composition both between manufacturers and between lots from a single manufacturer. In addition, clinical pharmacokinetic data for individual NP constituents rarely are available to explain differences between studies. Lastly, because NPs usually are self-administered, standard dosing regimens may not exist. These inherent limitations render static models used to predict drug-drug interactions less useful to predict NP-drug interactions. Dynamic modeling approaches, including physiologically-based pharmacokinetic (PBPK) modeling and simulation, could help improve prediction accuracy of NP-drug interactions. The exemplar NP, milk thistle (Silybum marianum), and victim drugs, warfarin and midazolam, were selected to test this approach. PBPK model predictions were evaluated with a proof-of-concept healthy volunteer study. Clinical PK endpoints were consistent with model predictions, providing credence to this translational approach, as well a framework that could be applied to other NP-drug combinations. Successful quantitative predictions of NP-drug interactions would aid in the development of systematic guidelines to identify clinically meaningful interactions, with the ultimate goal of providing definitive information to patients and health care providers about the risk or safety of adding a given NP to conventional drug regimens.