Pharmacokinetic-pharmacodynamic modeling in the data analysis and interpretation of drug-induced QT/QTc prolongation

Abstract

A BSTRACTIn this review, factors affecting the QT interval and the methods that are currently in use in the analysis of drug effects on the QT interval duration are overviewed with the emphasis on (population) pharmacokinetic-pharmacodynamic (PK-PD) modeling. Among which the heart rate (HR) and the circadian rhythm are most important since they may interfere with the drug effect and need to be taken into account in the data analysis. The HR effect or the RR interval (the distance between 2 consecutive R peaks) effect is commonly eliminated before any further analysis, and many formulae have been suggested to correct QT intervals for changes in RR intervals. The most often used are Bazett and Fridericia formulae introduced in 1920. They are both based on the power function and differ in the exponent parameter. However, both assume the same exponent for different individuals. More recent fi ndings do not confi rm this assumption, and individualized correction is necessary to avoid under-or overcorrection that may lead to artifi cial observations of drug-induced QT interval prolongation. Despite the fact that circadian rhythm in QT and QTc intervals is a welldocumented phenomenon, it is usually overlooked when drug effects are evaluated. This may result in a false-positive outcome of the analysis as the QTc peak due to the circadian rhythm may coincide with the peak of the drug plasma concentration. In view of these effects interfering with a potential drug effect on the QTc interval and having in mind low precision of QT interval measurements, a preferable way to evaluate the drug effect is to apply a population PK-PD modeling. In the literature, however, there are only a few publications in which population PK-PD modeling is applied to QT interval prolongation data, and they all refer to antiarrhythmic agents. In this review, after the most important sources of variability are outlined, a comprehensive population PK-PD model is presented that incorporates an individualized QT interval correction, a circadian rhythm in the individually corrected QT intervals, and a drug effect. The model application is illustrated using real data obtained with 2 compounds differing in their QT interval prolongation potential. The usefulness of combining data of several studies is stressed. Finally, the standard approach based on the raw observations and formal statistics, as described in the Preliminary Concept paper of the International Conference on Harmonization, is briefl y compared with the method based on population PK-PD modeling, and the advantages of the latter are outlined.

Population Pharmacokinetic/Pharmacodynamic Modeling of Guanfacine Effects on QTc and Heart Rate in Pediatric Patients

“…An effect of sex on BQTP was the only covariate effect in the final ER model for QTcP and its inclusion resulted in a typical baseline QTcP estimate that is 9 (5-13)ms higher for females compared with males. The magnitude of this difference is consistent with literature reports for sex differences in HR-corrected QT interval (13). The typical (95% CI) SLOP was estimated to be 0.941 (0.627-1.25)ms/ng/mL, resulting in an approximate 1-ms increase from baseline for every 1 ng/mL of guanfacine in plasma.…”

“…More complex linear or nonlinear models were then evaluated if suggested by the graphical evaluation or the results of the linear slope-intercept model. A model incorporating circadian rhythm was also investigated (13). Following development of the base ER model for HR, a covariate analysis was performed (see below for covariate analysis).…”

Phase 1 pharmacokinetic and drug‐interaction study of dasatinib in patients with advanced solid tumors

“…Because of the strong correlation between QT interval and heart rate, we considered several formulas for QT correction. 15 We used baseline ECGs to determine which formula yielded data that were least affected by heart rate. The Pearson correlation coefficients between QT interval, QTc value, and heart rate were À0.869 (QT, uncorrected), 0.282 (Bazett formula), À0.388 (Fridericia formula), À0.213 (linear population), and À0.201 (log-linear population).…”

The Use of Clinical Utility Assessments in Early Clinical Development

“…Additionally, due to a lack of adequate data availability, large variability and complex modeling required, there are only a few examples in literature for safety data modeling (25)(26)(27). Perhaps the most prominent safety data modeling is the PK/ PD modeling of the QTc interval prolongation (28,29) due to some very high profile drug recalls from the market. Any efforts to model safety data in the early stages of development are further complicated by usually smaller sample size in a majority of phase I/II clinical trials.…”