Study Design Parameters Affecting Exposure Response Analysis of QT Data: Results From Simulation Studies

The operating characteristics of dose-escalating studies in terms of false-negative predictions of the QT effect and the power to exclude clinically relevant QT effects are not quantitatively established. One thousand single-ascending-dose (SAD) studies with 7 dose groups with 6/2 subjects on active drug/placebo were generated through simulation for each of 32 scenarios with (1) 8 different QT effects of the study drug and (2) achieved plasma concentration 2- or 4-fold above therapeutic levels. For each subject, drug-free QT data from a thorough QT study were subsampled to capture the circadian profile, on which a drug effect was added. The percentage of false-negative studies was between 4% and 9% when the drug's QT effect was set to 10 milliseconds. If a somewhat lower effect of 6.7 milliseconds was set at therapeutic concentrations, the fraction of negative studies was higher, 40% to 60% when the variability of the QT data was well controlled. When the QT effect was set to 5 milliseconds at therapeutic plasma concentrations, the power of SAD studies to exclude 10 milliseconds QT effect was generally above 70% (74% to 94%) with well-controlled QT variability, whereas the power was reduced to 36% to 69% if supratherapeutic plasma concentrations were not achieved. The rate of false-negative studies was acceptably low in placebo-controlled SAD studies. With a drug with no or a small QT effect, supratherapeutic plasma concentrations, and well-controlled variability of QT data, the power of SAD studies to exclude a relevant effect was above 70%.