How can routinely collected comedication information from phase II/III trials be used for screening of potential effects on model parameters? A case study with the oral direct thrombin inhibitor Dabigatran etexilate
K.H. Liesenfeld, J. Stangier, H.G. Schaefer, A. Staab
Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach, Germany
Objectives: The objective was to develop a process that allows screening for the impact of comedication on model parameters using comedication information routinely collected during phase II/III. The process should be applied using data from a phase II study with dabigatran etexilate.
Methods: The comedication information routinely collected is the start and stop date of concomitant administration and the trade name or ingredient name which is coded in the clinical database using the WHO drug dictionary (WHO-DD). In order to establish comedication classes, Special Search Categories (SSCs) were developed which were either based on the Anatomical Therapeutic Chemical (ATC) classification system, e.g. digitalis glycosides, or on the drug substance names (e.g. all clinical relevant P-gp inhibitors were collected in one SSC). A SAS program was developed that generates for each SSC the WHO-DD code numbers of the respective trade names/ingredient names for comparison with the comedication records of the study. During NONMEM dataset generation an identifier was assigned to all records between start and stop date +1 of a comedication defined in a SSC. The SSCs were investigated as categorical covariates applying a standard forward inclusion/backward elimination model building strategy.
The dabigatran dataset for comedication screening contained 7782 observations from 1483 patients. Thirty-one SSCs were defined. Of the 31 SSCs, 19 were associated with > 5% of the observations. These 19 SSCs were investigated using the PopPK model developed before.
Results: Once the SSCs were defined the incorporation of the comedication information into the dataset was very fast. Three comedication classes were found to impact the dabigatran exposure. The changes caused by opioids (20% increase in exposure at day of surgery), antacids (35% decrease in exposure at day of surgery and 11% decrease of steady state exposure) and proton pump inhibitors (7.4% decrease in steady state exposure) were not considered clinically relevant because the magnitudes of the effect were minor.
Conclusions: An efficient process was developed and successfully applied to include routinely collected comedication information for screening of drug-drug interactions. The SSCs can be applied across studies and projects. The results obtained for dabigatran were included in the proposed drug label.