Impact of Phase 1 study design on estimation of QT interval prolongation risk using exposure-response analysis

The International Council for Harmonisation (ICH) guidelines have been revised allowing for modeling of concentration-QT (C-QT) data from Phase I dose-escalation studies to be used as primary analysis for QT prolongation risk assessment of new drugs. This work compares three commonly used Phase I dose-escalation study designs regarding their efficiency to accurately identify drug effects on QT interval through C-QT modeling. Parallel group design and 4-period crossover designs with sequential or interleaving cohorts were evaluated. Clinical trial simulations were performed for each design and across different scenarios (e.g. different magnitudes of drug effect, QT variability), assuming a pre-specified linear mixed effect (LME) model for the relationship between drug concentration and change from baseline QT (ΔQT). Analyses suggest no systematic bias in either the predictions of placebo-adjusted ΔQT (ΔΔQT) or the LME model parameter estimates across all evaluated designs. Additionally, false negative rates remained similar and adequately controlled across all evaluated designs. However, compared to the crossover designs, the parallel design had significantly less power to correctly exclude a clinically significant QT effect, especially in the presence of substantial intercept inter-individual variability. In such cases, parallel design is associated with increased uncertainty around ΔΔQT prediction, mainly attributed to the uncertainty around the estimation of the treatment-specific intercept in the model. Throughout all the evaluated scenarios, the crossover design with interleaving cohorts had consistently the best performance characteristics. The results from this investigation will further facilitate informed decision-making during Phase I study design and the interpretation of the associated C-QT modeling output.