How pre-marketing data can be used for predicting the weight of drug interactions in clinical practice

Abstract Unexpected drug interactions have led to the withdrawal of many drugs, raising concern about the gap between what is known at the time of approval and the risk of serious effects in the longer term, particularly in high-risk populations generally excluded from drug development. This is because the majority of drug interaction studies are done using in vitro methods, or in healthy young volunteers who may not reflect the complexity of patients, and the settings in which the drug will be used in clinical practice. Pre-marketing interaction studies should therefore be designed to make information easily accessible and clinically transferable. They should be adequate in terms of sample size, population, comorbidity, phenotyping and/or genotyping, end-points and outcome measures, and conducted in conditions of dose, route and timing of co-administration that reproduce the proposed therapeutic indications of the new drug. Although young volunteers have the advantage of minimizing some confounding effects introduced by diseases or polypharmacy, patients drawn from populations for whom the drug is intended would be more relevant and accurate, providing the studies are feasible and safe.