Not-in trial simulation : Prospective use of Not-In-Trial Simulation
Anne Chain (1), Dinesh de Alwis(2), An Vermeulen (3), Meindert Danhof (1), Miriam CJM Sturkenboom (4), Oscar Della Pasqua (1, 5)
(1) Division of Pharmacology, Leiden University, the Netherlands, (2) Eli Lily & Company Ltd, Surrey, United Kingdom, (3) Pharmaceutical Research & Development, Johnson & Johnson, Belgium, (4) Department of Medical Informatics adn Epidemiology, Erasmus University Medical Center, the Netherlands, (5) GlaxoSmithKline, Clinical Pharmacology / Modelling and Simulation, United Kingdom
Objectives: Previously, the concept of Not-In-Trial Simulation model was introduced to describe QTc in a real life cohort using a model-based approach [1]. The model consisted of QTc = age-dep. baseline value + drug effects + effects from comorbidities and co medications. In this model, however, variability descriptors were not included which are required to enable prospective use of such a tool in drug development. The aim of this study is to further evaluate the variability associated with age and to explore the interactions with other covariates.
Methods: The relationship between age and baseline QTc observations was modelled using NONMEM VI. The age-effect model was created with healthy subjects and patients without comorbidities and comedications. An interaction model was created by including patient data with comorbidities and comedications. Model comparisons were made using objective functions with the criteria of p=0.05 while model performance was tested using diagnostic plots, VPCs and NPDEs. After model completion, we used a QT-prolonging drug (d,l-sotalol) to mimic a drug development scenario, which was previously modelled according to a two- comp. model with weight on the clearance. Drug-induced QT-prolongation was added to the underlying effect of the covariates.
Results: The QTc vs. age relationship is described by a linear model. Gender, arrhythmia, myocardial infarction, diabetes and heart failure are found to be covariates to the intercept of the relationship. Diabetes and heart failure also are found to be covariates on the slope of the equation. Simulations from the improved Not-In-Trial tool confirm greater effects from comorbidities and comedications than the drug-induced QTc prolongation.
Conclusions: Age-effects play an important role in QTc observations in clinical trials and real life cohort irrespective of drug treatment. This study shows that baseline QTc values are also dependent on the various health conditions. In contrast to the current implementation of TQT trials, the assessment of the cardiovascular liability must take into account those factors to accurately describe individual patient risk under therapy in real life conditions.
References:
[1] Chain et al., PAGE 18 (2009) Abstr 1512 [www.page-meeting.org/?abstract=1512]
