TY  - GEN
ID  - heidok29312
AV  - public
CY  - Heidelberg
TI  - Midazolam Microdosing and Population Pharmacokinetic Modelling to Assess CYP3A Drug-Drug Interactions in Early Clinical Development
Y1  - 2021///
N2  - Due to its abundance in the human liver and gut, as well as the important role it plays in drug metabolism, CYP3A is a significant liability for drug-drug interactions. As such, efficient and accurate methods for detecting CYP3A drug-drug interactions are needed to facilitate assessment both in drug development and in the clinic. Several methods are considered in this work, including midazolam microdosing, a limited sampling approach, and population pharmacokinetic modelling. Microdosing consists of administration of a pharmacologically inactive dose that is no more than 1/100th of the therapeutic dose, which allows such doses to be administered with other substances without worry that they may influence the pharmacokinetics or pharmacodynamics of the co-administered substance. Limited sampling approaches use 1 to 4 samples to determine exposure of a drug, rather than obtaining a full pharmacokinetic profile. These approaches are ideal candidates for combining with population pharmacokinetic modelling, which can estimate full profiles from sparse sampling. Thus, the current dissertation had the following aims: 1) establish a method of incorporating midazolam microdosing in multiple rising dose studies for early detection of CYP3A drug-drug interactions; 2) develop a population pharmacokinetic model to describe midazolam exposure during constitutive, inhibited, and induced CYP3A activity; and 3) assess the capability of limited sampling to complement the two preceding aims. As an exploratory objective, model estimated parameters were assessed for potential cut-points that may allow for determination of drug-drug interactions when a baseline profile is not available. 

For the establishment of midazolam microdosing in early clinical development, three early clinical studies were conducted with substances (Compounds A, B, and C) which gave positive CYP3A perpetrator signals in vitro. A 75 µg dose of midazolam was administered alone (baseline CYP3A activity) followed by administration with the highest dose groups tested for each compound on Day 1/3 and Day 14 or Day 17. Midazolam exposure (AUC0-?, Cmax) during administration with the test substances was compared to baseline data via an analysis of variance on log-transformed data. Partial AUC2-4 ratios were also compared to AUC0-? ratios using linear regression on log-transformed data. The data obtained from these studies were further used for external validation, following development of a composite midazolam-1?-OH midazolam population pharmacokinetic model for CYP3A drug-drug interactions. The composite model was also evaluated using limited sampling profiles, both from the model development set, as well as from an external validation set.

The implementation of midazolam microdosing in early clinical studies proved to be feasible: Midazolam concentrations were quantifiable over the full profiles for all subjects in all studies and AUC and Cmax values could, thus, be accurately determined. Results from the test compounds indicated that, based on the Cmax values exceeding relevant thresholds, drug-drug interactions were expected. Point estimates of the midazolam AUC0-? gMean ratios ranged from 108.3 to 127.1% for Compound A, from 93.3 to 114.5% for Compound B, and from 92.0 to 96.7% for the two highest dose groups of Compound C. Cmax gMean ratios were in the same range. Thus, despite the expectation of drug-drug interactions from in vitro results, midazolam microdosing results indicated no relevant interactions were present. AUC2-4 ratios, based on the limited sampling scheme suggested for subsequent studies (2, 2.5, 3, and 4 h), were comparable to the AUC0-? ratios in the conducted studies.

A composite model was developed, which adequately described midazolam and 1?-OH midazolam concentrations for constitutive, inhibited, and induced CYP3A activity. The model showed good internal and external validity, both with full profiles and limited sampling (2, 2.5, 3, and 4 h), and the model estimated parameters were congruent with values found in clinical studies. Assessment of potential cut-points for model estimated parameters to identify drug-drug interaction liability with a single profile suggested that midazolam clearance may reasonably be used to detect inhibition (6.81-16.8 L/h), induction (43.3-86.6 L/h), and no modulation (16.8-43.3 L/h). Sensitivities for potent inhibition and induction were 98.7% and 100%, respectively, and specificity was 99.2% for no modulation. 

Thus, the current dissertation indicated that 1) midazolam microdosing incorporated into early clinical studies is a feasible tool for reducing dedicated drug-drug interaction studies; 2) a population pharmacokinetic approach can provide efficient and accurate CYP3A drug-drug liability detection; and 3) limited sampling can be a useful complementary tool for both midazolam microdosing and population pharmacokinetic modelling. Therefore, the current dissertation provides three separate, complementary tools for the assessment of CYP3A drug-drug interactions, either in drug development or in clinical practice.
A1  - Wiebe, Sabrina
UR  - https://archiv.ub.uni-heidelberg.de/volltextserver/29312/
ER  -