Assay calibration to refine prediction of P-gp mediated DDI based on in vitro vesicular transport assay data

Food and Drug Administration (FDA) draft guidance and European Medicines Agency (EMA) guidance on drug-drug interactions (DDI) define criteria for evaluating whether a new molecular entity that inhibits P-glycoprotein (P-gp; also known as ABCB1 or MDR1) in vitro should be tested in a clinical DDI study with a P-gp probe/index substrate such as digoxin. The decision criteria whether to conduct an in vivo DDI study is based on the steady state concentration of a drug of interest compared to the IC₅₀ value determined in vitro. Cut-off values based on different in vivo parameters are specified in both guidelines. However, significant inter-laboratory variability has been reported for P-gp IC₅₀ data determined in vitro (Bentz, J., et al., Drug Metab Dispos, 2013), hampering the practical implementation of a uniform cut-off value for all laboratories. Hence, for a more tailored risk assessment of DDI potential, calibration studies aimed at determining laboratory- or assay-specific cut-off values are highly recommended, which would increase the predictive value of these in vitro assays.

In this study, our aim was to calibrate our P-gp vesicular transport experimental system using inside-out membrane vesicles to allow for better in vivo prediction and refine cut-off values if necessary. The vesicular transport assay is an ideal choice for IC₅₀ determination due to its relatively high throughput and fewer parameters to consider during inhibition evaluation. In addition, this assay type is applicable for IC₅₀ determination of almost any drug molecule regardless of physicochemical characteristics such as passive permeability, which may confound data from cellular assay systems.

35 known P-gp interactors were chosen for determination of IC₅₀ values using N-methyl quinidine (NMQ) as probe/index substrate in a vesicular transport assay with MDR1-HEK293 membrane vesicles. IC₅₀ values could be determined for 32 drugs and data were compared to results of in vivo DDI studies from the literature. Accuracy, sensitivity and specificity of each ratios were determined using FDA and EMA thresholds. The best accuracy (73%) was obtained for [I_1T]/IC₅₀ ratio using FDA cut off value of 0.1. Data were also subjected to Receiver Operating Characteristic (ROC) analysis, to refine cut-off values specific to our vesicular transport assay system. The aim was to minimize false negative determination, and so the sensitivity was set to be over 90% as default parameter to avoid missing potential safety issues during drug development. The analysis resulted in optimal sensitivity of 94% with a specificity of 73% for [I_1u]/IC₅₀ values at a threshold of 0.002. Our data analysis further confirm the importance of laboratory-specific threshold values for DDI risk assessment of digoxin.