Population pharmacokinetics and pharmacodynamics of warfarin in healthy young adults

The population pharmacokinetics and pharmacological response — prothrombin complex activity and factor VII activity — were studied in a group of 48 normal, healthy young volunteers. Population parameter estimates were obtained using a standard two-stage method, a nonlinear mixed effect model (NONMEM) and a two-stage Bayesian method (EM algorithm). A modified sigmoid-Imax model was used to relate the concentration of s-warfarin to the rate of clotting factor synthesis. The three methods produced similar estimates of the population pharmacokinetic parameters, although the standard two-stage method overestimated the contribution of the pharmacokinetic parameters to the interindividual variability. It was not possible to partition the interindividual variability in response between the pharmacodynamic parameters with the NONMEM procedure: the estimates obtained from the EM algorithm were generally in good agreement with those obtained using the standard two-stage approach. The variability in the warfarin concentration contributed at most only 40% of the observed variability in the pharmacological response, and then only for times greater than 96 h after the dose. Most of the variability in the pharmacodynamics was due to interindividual differences in the clotting factor degradation rate constant and C_50,s, the s-warfarin concentration causing a 50% decrease in synthesis rate.