A population pharmacokinetic-pharmacodynamic analysis and model validation of azimilide

Authors: Phillips L, Agnew JR, Ludwig EA, Thompson GA
Publication: Clin Pharmacol Ther
Keywords: analysis, analysis of variance, Anti-Arrhythmia Agents, arrhythmia, bilirubin, blood, body weight, clinical trials, Dose-Response Relationship, double-blind method, drug, drug effects, drug therapy, Electrocardiography, Phase III as Topic
Division: Cognigen

Abstract

BACKGROUND:

Pharmacokinetic (PK) and pharmacodynamic (PD) models for azimilide were developed and validated with sparse blood sampling and QTc interval data obtained during three clinical trials of azimilide for prevention of supraventricular arrhythmia recurrence.

METHODS:

Patients were orally administered placebo or azimilide dihydrochloride, 35, 50, 75, 100, or 125 mg/d, for 6 to 9 months. NONMEM was used for data fitting and assessment of selected patient covariates and concomitant medication classes for PK/PD relationships.

RESULTS:

Results indicate that azimilide clearance (CL) was dependent on body weight (WTKG), gender, and current tobacco use, where CL (L/h) = 3.92 x (WTKG – 43)(0.208), with a 17% increase for male subjects and a 15.5% increase for current tobacco use. Volume of distribution (V) was also dependent on WTKG and total bilirubin (BIL), where V (L) = 9.88 x (WTKG – 43) + 717 x (BIL)(0.348). The PK/PD analysis indicated that the baseline QTc interval was dependent on gender, New York Heart Association Class, digoxin, and paced artificial pacemaker spike, whereas the 50% effective concentration (EC(50)) was dependent on the serum potassium (K) level, where EC(50) = 107 x K. The change in EC(50) was not clinically significant within the normal range for potassium. The mean E(max) (maximum increase in the QTc interval for the E(max) models) was a 61.7 ms increase from baseline. At 125 mg/d the predicted percent increase in the QTc interval at the maximum plasma drug concentration at steady state was 9% and 10% for male and female patients, respectively. The values of the median prediction error were -3% and -0.4% for the PK and PK/PD models, respectively, and the values of the absolute prediction error were 21% and 4% for the PK and PK/PD models, respectively, indicating that both models are essentially unbiased and acceptably accurate.

CONCLUSIONS:

Azimilide PK parameters are dependent on body weight, gender, smoking status, and bilirubin and are independent of the coadministration of digoxin, warfarin, and cytochrome P4503A4 inhibitors and inducers. The relationship between azimilide concentration and change in QTc is primarily dependent on serum potassium.

By, Phillips L, Thaddeus H Grasela, Agnew JR, Elizabeth A Ludwig, Thompson GA