Population PK-PD modeling of thorough QT/QTc data allows for mechanistic understanding of observed QTc effects
O. Lillin-de Vries, R. de Greef, T. Kerbusch
Objectives: Thorough QT/QTc trials (TQT) are designed for a yes/no outcome (ICH-E14 statistical analysis). Beyond yes/no more quantitative information can be retrieved by performing a population PK-PD analysis on data available from TQTs, information that is useful for internal decision making when the TQT is positive.
Methods: A PK-PD model (NONMEM VI) was fitted to QTc data from 44 adult HV receiving D mg of drug X (therapeutic dose), 38 HV receiving 5D mg of drug X (supra-therapeutic dose) and 44 HV receiving placebo (the active comparator data were not used). The PK-PD model was built in two steps: a baseline model incorporating demographic effects on QTc (sex, age, circadian rhythm) was built using baseline data and subsequently drug effect and placebo effect were quantified using all QTc data. Linear and non-linear concentration-effect models were tested as well as direct and indirect effect models. The observed time delay between Cmax and dQTc_max was accounted for with a hypothetical effect site model.
Results: 112 HV contributed to 1188 PK samples after doses of 0.7D - 5D mg X. 126 HV contributed to 2320 QTc samples after D mg X, 5D mg X and placebo. The final PK-PD model included covariates sex, age and circadian rhythm on the baseline QTc. An indirect linear drug effect model described the data best, since a time delay of about 2 h was observed between peak levels of X and peak dQTc. The most important model parameters were: baseline QTc for a typical woman was 409 ms and the slope of drug effect was 0.00177 ms/(ng/mL). The predicted dQTc at Tmax - mean (upper 95%CI) - was 3.2 (3.8) ms after D mg of X and 9.4 (11.2) ms after 5D mg X.
Discussion: In spite of the negative hERg tests performed on drug X, and a metabolite, the PK-PD model confirmed a QTc effect. The predicted concentrations of X in the hypothetical compartment of the PK-PD model matched observed concentrations of a metabolite of X. The PK-PD model indicates that the QTc prolongation could potentially be caused by the metabolite.
Conclusion: A population PK-PD model was successfully fitted to the data of a TQT trial. The TQT was positive and the PK-PD model could confirm a QTc prolongation effect slightly above the threshold of regulatory concern.
 C Garnett et al, Concentration-QT Relationships Play a Key Role in the Evaluation of Proarrhythmic Risk During Regulatory Review, J. Clin. Pharmacol. 2008; 48; 13
 V Piotrovsky et al, Pharmacokinetic-Pharmacodynamic Modeling in the Data Analysis and Interpretation of Drug-induced QT/QTc Prolongation, The AAPS Journal 2005; 7 (3) Article 63