Verena Gotta (1), Karel van Ammel (2), Frank Cools (2), David Gallacher (2), Meindert Danhof (1), Piet H. van der Graaf (1)
(1) Systems Pharmacology, Leiden Academic Center of Drug Research (LACDR), Leiden University, The Netherlands, (2) Safety Pharmacology, Janssen Research & Development, Janssen Pharmaceutica, Belgium
Objectives: This simulation study aimed at investigating the power of PKPD modeling to detect different magnitudes of QT-prolongation in preclinical cardiovascular safety studies in the conscious telemetered dog.
Methods: A PKPD model predicting individual corrected QT intervals (QTc, linear correction to a heart rate of 60 bpm) was developed for the positive control drug sotalol from a standard cardiovascular safety study: 6 animals received on 4 occasions in a crossover setup a vehicle, low, mid and high dose [1]. Maximal plasma concentration (Cmax, population prediction) at the high dose level was 15 777 ng/ml. The model described QTc over 24h as a function of circadian variation (cosine function) and drug concentration (sigmoidal Emax model, Emax=50.8 ms, EC50=1780 ng/ml), while drug concentrations were described by a simple 1-compartment model. The residual variability in QTc was 8.5 ms. This “true” model was simulated 100 times for each study investigating 3%, 1% and 0.5% of the original dose levels, with expected drug-induced QTc prolongation (∆QTc) at high-dose population Cmax of 11.5, 4.8 and 2.6 ms. Because simulated drug concentrations were very low compared to EC50, the simulated studies were re-estimated using a linear model (model 1) and a model not including a drug effect as reference (model 2). A drop of the objective function by 3.84 (likelihood ratio test, alpha = 5%) was considered as significant.
Results: The power of detecting a drug effect was 100%, 98% and 74% at 3%, 1% and 0.5% of the original dose level, respectively. Mean estimated ∆QTcs [95% prediction interval] were 11.7 [8.8-14.7] ms, 4.8 [2.7-6.8] ms and 2.6 [0.8-4.2] ms, respectively.
Conclusions: Compared to the reported power of conventional statistical methods (80% for detecting a 4 ms [2] and 10 ms [3] QTc effect, respectively), these preliminary results suggest superior sensitivity of model-based approaches to quantify QT prolongation in preclinical setting. This underscores the value of PKPD modeling in preclinical safety testing.
References:
[1] Chain AS, Dubois VF, Danhof M, et al.. Identifying the translational gap in the evaluation of drug-induced QTc interval prolongation. Br J Clin Pharmacol. 2013;76(5):708-24 [2] Sivarajah A, Collins S, Sutton MR, et al. Cardiovascular safety assessments in the conscious telemetered dog: utilisation of super-intervals to enhance statistical power. J. Pharmacol. Toxicol. Methods. 2010;62(1):12–9[3] Chiang AY, Bass AS, Cooper MM, et al. ILSI – HESI cardiovascular safety subcommittee dataset: An analysis of the statistical properties of QT interval and rate-corrected QT interval (QTc). J. Pharmacol. Toxicol. Methods. 2007;56(2):95–102
Reference: PAGE 23 (2014) Abstr 3044 [www.page-meeting.org/?abstract=3044]
Poster: Drug/Disease modeling - Safety