Data analysis and statistical analysis
The sample size of 20 was determined based on the reported PK parameters of atorvastatin. The coefficient of variances calculated based on a single-dose administration of atorvastatin 10 mg were 45.7% for AUC and 44.2% for Cmax.29 The correlation coefficient was assumed to be 0.7. When there is no drug interaction, the number of subjects whose 90% CI of the geometric mean ratio of PK parameters was 0.80 − 1.25 was calculated to be 20 for AUC and 19 for Cmax with 80% power.
The PK parameters were assessed in all subjects who received more than one dose of the study drug and whose PK data were adequate for the calculation of more than one primary PK parameter (PK analysis set). Safety was assessed in all subjects who received more than one dose of the study drug. Levels of analyte below the level of quantification were entered as 0 for calculations. Descriptive statistics were used to summarize demographics and safety parameters. For plasma drug concentration and PK parameters, summary statistics and the two-sided 95% CIs were calculated. A natural logarithmic transformation of PK parameters except for Tmax was applied for all statistical inference. The 90% CI for ratios of geometric means of logarithmic PK parameters was calculated by the following mixed effects model;
Loge (PK Parameter) = μ + time point + subject + ε
μ: population mean, time point: duration of administration, subject: interindividual variation, ε: error
For each interacting drug, the 90% CI of the geometric mean ratios of the AUC0-t and Cmax of midazolam, prednisolone, and atorvastatin with coadministration of carotegrast methyl (days 7 and 14) or after the end of administration of carotegrast methyl (days 28 and 42) to those prior to the administration of carotegrast methyl (day -1) were calculated. When the 90% CI of the geometric mean ratio fell within the range of 0.80 − 1.25, it was determined that there was no PK interaction. PK parameters were calculated using noncompartmental analysis with WinNonlin Professional Version 6.3 (Phoenix Corporation, Mountain View, California, USA). AEs were coded using the Medical Dictionary for Regulatory Activities (MedDRA) version 19.0. All data processing, summarization, and analyses were conducted using SAS software ver. 9.3 for Windows (SAS Institute Inc., Cary, NC, USA).