Operational Characteristics of Linear Concentration-QT Models for Assessing QTc Interval in the Thorough QT and Phase I Clinical Studies.

Abstract

Concentration-QTc (C-QTc) analysis can be used as an alternative to the standard statistical methods in clinical QT studies. Pharmacokinetic/pharmacodynamics (PK/PD) simulations were performed to assess the operating characteristics of four C-QTc models. False negatives were 2-6% for crossover and 2-9% for parallel studies, with 12 to 60 subjects per treatment for a dose with 10-ms mean effect. All C-QTc models tested gave less than +1 ms mean bias in the ΔΔQTcmax prediction. The power to exclude 10 ms was >80% across all study designs and sizes, for a dose with 3-ms mean effect. The study demonstrates that linear C-QTc models have adequate sensitivity and specificity when the simulation and data analytical models are the same. C-QTc models that incorporate time- and treatment-specific terms give the least biased ΔΔQTcmax predictions under scenarios of model-misspecifications and offer an advantage when applying to real clinical data where the underlying relationship is not known.