Pharmacokinetics of the Digoxin-Quinidine Interaction via Mixed-Effect Modelling

SummaryIt was the purpose of this study to evaluate the effect of quinidine administration on the population estimates of the volume of distribution (Vdpop) and clearance (CLpop) of digoxin. The data collected on 94 patients included 230 measured serum digoxin concentrations, height, age, sex, weight (wt), serum creatinine, history of digoxin and quinidine administration and the presence or absence of congestive heart failure (CHF). Using the NONMEM software program, estimates were obtained for CLpop and Vdpop. Variables tested for inclusion in the CLpop model were creatinine clearance (CLCR), CHF, wt, ideal bodyweight, quinidine (QUIN) [both as a discrete variable and in a dose-dependent manner], and body surface area. Variables tested for inclusion in the Vdpop model were CLCR, wt, ideal bodyweight, body surface area and quinidine. During model building a p-value of 0.05 was chosen for variable inclusion. The final model was as follows: $$\matrix{{{\rm{C}}{{\rm{L}}_{{\rm{pop}}}}({\rm{L}}/{\rm{h}}) = (3.1 + 0.0516 \times {\rm{C}}{{\rm{L}}_{{\rm{CR}}}}) \times {\rm{QUIN}}} \cr {{\rm{V}}{{\rm{d}}_{{\rm{pop}}}}({\rm{L}}) = (4.03 + 0.0832 \times {\rm{C}}{{\rm{L}}_{{\rm{CR}}}}) \times {\rm{wt}}} \cr {{\rm{F}} = 0.82} \cr}$$ where F is bioavailability. In the above, QUIN is 0.567 if quinidine is being concurrently administered and 1.0 if it is not. The coefficient of variation (CV) of CLpop was 44% while that of Vdpop was 48%. The residual intrasubject CV was 26%. These results compare favourably with previously derived methods of estimating digoxin CLpop and Vdpop but may improve on those methods due to the inclusion of quinidine in the model. These better estimates should result in improved initial dosage of digoxin.