gms | German Medical Science

53. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie e. V. (GMDS)

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie

15. bis 18.09.2008, Stuttgart

Impact of delayed effects in the exposure-response analysis of clinical QT studies

Meeting Abstract

Search Medline for

  • Arne Ring - Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach, Deutschland
  • Patricia Glomb - Boehringer Ingelheim Pharma GmbH & Co. KG, Biberach, Deutschland

Deutsche Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie. 53. Jahrestagung der Deutschen Gesellschaft für Medizinische Informatik, Biometrie und Epidemiologie (gmds). Stuttgart, 15.-19.09.2008. Düsseldorf: German Medical Science GMS Publishing House; 2008. DocP-26

The electronic version of this article is the complete one and can be found online at:

Published: September 10, 2008

© 2008 Ring et al.
This is an Open Access article distributed under the terms of the Creative Commons Attribution License ( You are free: to Share – to copy, distribute and transmit the work, provided the original author and source are credited.




The ICH E14 guideline [1] recommends investigating the risk of QT(c) prolongation of new drugs in a dedicated clinical study, the so called “Thorough QT study” (TQT). The primary analysis of such a study is the quantitative comparison of the QTc change from baseline between drug and placebo at each sampling time point. A recommended statistical methodology is the repeated measurements analysis as proposed by Patterson et al [2].

In addition to this analysis, some sponsors had also explored the relationship between the actual drug concentration in the blood (exposure) and the change from baseline of the (heart rate corrected) QT interval effect (response) [3], [4], [5]. This type of analysis had been gained increased visibility since also regulators provided stronger support for it [6]. One advantage of this PK-QT analysis is the use of the full data set for the effect estimation which is statistically more efficient than slicing the data by time points as recommended by the E14 analysis [7].

On the other hand, the PK-QT analysis depends on the underlying modelling assumptions. While first analyses were based on a linear regression, the methodology rapidly developed towards linear and nonlinear mixed effects models.

The current investigation examines the detection and the impact of a potential delay between concentration and QT(c)-effect in the PK-QT analysis.


The pharmacokinetic-pharmacodynamic model for the exposure response analysis of the QT interval is shown in Figure 1 [Fig. 1].

The pharmacokinetic model consists of an absorption system and an elimination system which are based on monoexponential transit times. The parameters of this model are F the fraction of the dose which reaches systemic circulation, 1/ka the mean absorption time and 1/ke the mean elimination time.

The concentration of the drug in the body is determining the effect size of the QT interval prolongation, based on the effect function E. The simplest effect model is a linear model

dQT = c * Conc

where c is a proportionality factor, and dQT is the change from baseline of the QT(c) interval. Other pharmacodynamic models are possible [8] including the non-linear Emax-model

dQT = E0 + (Emax * Conc)/(EC50 + Conc).

The current investigations introduced another change in the pharmacodynamic model: a lag time was added, which leads to delayed onset and offset of the QT(c)-prolongation. The aim was to simulate data with the delayed effects and to study the following questions:

a) Which impact do delayed effects have on the estimation of the model parameters, if the estimation model does not account for the delay?

b) How can delayed effects been identified?

c) If the estimation model accounts for the delay – can the model parameters and the QT(c) effects be estimated efficiently?

Question a) and c) can be addressed by examination of the relationship between the delay parameter and the estimated parameters. Question b) is often investigated by plotting the concentration time profiles of the concentration and the QT(c) change from baseline. In addition, scatter plots of both parameters are examined which shall reveal typical hysteresis loops with counter clockwise direction.

However, there is currently no statistical metric available which indicates whether a delay is of a magnitude that it could potentially bias the effect estimates of the pharmacodynamic model.

The investigations are based on designs and sample sizes of typical TQT studies. Most model parameters are simulated with interindividual variability. The estimations will be based on mixed models. Monte-Carlo-simulations were carried out with 1000 replications of TQT studies for each parameter combination.

Results and Discussion

While hysteresis plays an important role in the pre-clinical studies of QT(c)- prolongation, delays of QT(c)-effects have not been addressed statistically in clinical trials yet. Preliminary studies indicated that the magnitude of a delayed QT(c) effect could potentially be underestimated if the estimation model does not account for the delay. Details of the results are still to be derived; they will be presented in the poster.


ICH Harmonized Tripartite Guideline. The Clinical Evaluation of QT/QTc Interval Prolongation and Proarrhythmic Potential for Non-Antiarrythmic Drugs. ICH E14, 2005.
Patterson SD, Jones B, Zariffa N. Modeling and Interpreting QTc Prolongation in Clinical Pharmacology Studies. Drug Inf J 2005;39:437-45.
Piotrovski V. Pharmacokinetic-Pharmacodynamic Modeling in the Data Analysis and Interpretation of Drug-induced QT/QTc Prolongation. AAPS J 2005; 7(3): Article 63.
Jonker DM, Leishman D, Wallis R, Milligan PE, Jonsson NE. Development and Validation of a Mechanism-based PK/PD Model for the in Vitro-in Vivo Prediction of QT Prolongation by Dofetilide. Population approach group 13th meeting 2004: Abstr 502.
Li J. Use of nonlinear Mixed Modeling Approach to Analyze ECG Data. Informa Life Sciences 2008.
Garnett CE, Beasley N, Bhattaram VA, Jadhav PR, Madabushi R, Stockbridge N, Tornøe CW, Wang Y, Zhu H, Gobburu JV. Concentration-QT Relationships Play a Key Role in the Evaluation of Proarrhythmic Risk During Regulatory Review. J Clin Pharmacol. 2008;48(1):13-8.
Ring A, Grasmaeder K, Ring C. Comparison of the E14 Analysis and a PK-QT Modelling Approach – Case Studies and Simulations. DIA cardiac safety conference. Washington 2007.
Bonate P. Pharmacokinetic-Pharmacodynamic Modeling and Simulation. Berlin: Springer 2006.