Data from 12 subjects were expected to provide ≥90% probability that the lower limit of the 90% confidence interval (CI) for the geometric mean ratio of Cmax/Cmin values (rivaroxaban/apixaban) would be >1. An additional two subjects were enrolled to allow for early withdrawals. This estimate was based on the assumptions that the expected Cmax/Cmin ratio was ≥30% greater for rivaroxaban than apixaban,11 (link),12 (link) Cmax/Cmin would be log-normally distributed, and intersubject standard deviation would not be greater than 0.22.13 (link)
Mean and individual steady-state concentration–time and AXA–time profiles were plotted for both apixaban and rivaroxaban. Scatter plots of AXA versus plasma concentration were plotted for both compounds and analyzed by linear regression. Individual PK and PD parameters were estimated using noncompartmental methods with WinNonlin® Professional (v5.0.1; Pharsight Corporation, Sunnyvale, CA, USA). Terminal elimination rate constants were estimated using the WinNonlin algorithm and AUC parameters were calculated using the log-linear trapezoidal rule (WinNonlin Method 1). Actual sampling times were used for all parameter calculations. Descriptive statistics for PK and PD parameters were tabulated.
Mean and individual steady-state concentration–time and AXA–time profiles were plotted for both apixaban and rivaroxaban. Scatter plots of AXA versus plasma concentration were plotted for both compounds and analyzed by linear regression. Individual PK and PD parameters were estimated using noncompartmental methods with WinNonlin® Professional (v5.0.1; Pharsight Corporation, Sunnyvale, CA, USA). Terminal elimination rate constants were estimated using the WinNonlin algorithm and AUC parameters were calculated using the log-linear trapezoidal rule (WinNonlin Method 1). Actual sampling times were used for all parameter calculations. Descriptive statistics for PK and PD parameters were tabulated.
