Moxifloxacin versus placebo modeling of the QT interval

The objectives were to develop a population model for placebo-corrected moxifloxacin QT interval in healthy subjects using non-linear mixed effects modeling and to examine effect of covariates on the observed QT. Based on the parameters of interest, optimizations of observation times and number of subjects were proposed. A pool of four thorough QT studies was used, representing 99 subjects receiving placebo and moxifloxacin. The data was modeled using Monolix. The placebo effect on QT was satisfactorily described using a 2-oscillator model. It reflected the circadian rhythm variability which is taken into account when assessing the time-matched mean difference on QT between treatment and baseline. Based on this model, the moxifloxacin effect on QT was satisfactorily described by the same equation with the adjunct of a direct and proportional drug concentration-effect. The Emax model provided the best description of the effect. The unique covariate was gender for both baseline QTc and individual heart rate correction factor. The present design included up to 16 observations for pharmacodynamics. Using this model, 9 observation times for pharmacodynamics provided satisfactory estimates for the parameters of interest (Emax). With 15% precision limit on Emax, 60 subjects was optimal. The simultaneous placebo-moxifloxacin QT model proposed is an interesting alternative to the ICH E14 guideline in assessing QT prolongation effect. This approach provides accurate information over a range of concentrations using different relationships (slope or Emax models) to quantify the drug-response relationship versus placebo. This allowed optimizing the observation times and number of subjects.