G. Ferber,1 U. Lorch,2 J. Taubel2; 1Statistik Georg Ferber GmbH, Riehen, Switzerland, 2Richmond Pharmacology. Ltd., London, United Kingdom
BACKGROUND: The ICH E14 guidance suggests that concentration-effect (CE) modeling can complement the primary analysis. Recently, the understanding of this technique has improved and novel methods have been proposed. Based on three Phase I studies; we compared CE to conventional per time point analysis to prove its validity for QT prolongation detection.
METHODS: The studies were designed as double-blind, randomized, placebo-controlled, four-way, crossover TQT studies in healthy male and female subjects. A per time point analysis was performed as primary analysis. CE analysis based on the change from baseline of QTcF or QTcI were performed as secondary analyses. We compared these results with respect to the size of the QT effect.
RESULTS: In the first study, modeling was done regardless the circadian variability. The effects of lower dose (no relevant QTc prolongation) and higher dose (slightly above the threshold of regulatory concern) were well reproduced with the CE analysis. For subsequent negative studies, CE analyses were based on the placebo corrected change from baseline. They showed a slight shortening of QTcF. As expected, confidence intervals for the predicted effect were narrower than those obtained by a per time point analysis.
CONCLUSION: CE analysis has evolved as a valuable tool: it complements the per time point analysis, as expected; and the confidence intervals obtained at relevant plasma drug concentrations are narrower than those obtained from the conventional per-time point analysis. The CE analysis enables the information on QT changes to be collected in early phase studies when relationship between the drug or its metabolite and QT interval is to be evaluated, as it is sufficiently sensitive to detect a QT change of 5 ms and cost-effective, i.e. may be used on a smaller sample size.