Graphical Evaluation for Cardiovascular Safety Data in Clinical Trials

Speaker: Ihab G. Girgis; Johnson & Johnson Pharmaceutical R&D

Detecting drug-induced effect on cardiac repolarization or QT interval is a closely monitored safety element in drug development and more recently, it is thoroughly scrutinized in regulatory submissions. Length of the QT intervals can be influenced by a number of covariates, such as, heart rate (HR), RR interval (RR= 60/HR), gender, and natural circadian rhythm.

There are other unknown factors that influence this interval, making it highly variable across population and the analysis of such data is complex. In order to evaluate the effect of drug on QT interval, accurate modeling of drug-free baseline QT becomes an important first step; the changes to this baseline model after the administration of investigational drug will reflect the effect of the investigational drug on the QT/QTc interval.

This work focuses on the graphical evaluation of baseline QT clinical data and its modeling. A hierarchical Bayesian approach has been used. The QT-RR relationship is explored using various models and performance of the models is evaluated in comparison to well-known correlation methods (Bazett, Fridericia, Framingham, Hodge and individual correction). Finally, diverse nonlinear functions, ranging from a simple cosine function to multi-harmonics Fourier series are tested to describe the circadian rhythm effect.

Statistical analysis software