João A. Abrantes (1,2), Anabela Almeida (2), Francisco Sales (3), AmÃlcar Falcão (1,2), Siv Jönsson (4)
(1) Department of Pharmacology, Faculty of Pharmacy, University of Coimbra, Portugal, (2) CNC – Centre for Neuroscience and Cell Biology, University of Coimbra, Portugal, (3) Department of Neurology, Coimbra University Hospital, Portugal, (4) Department of Pharmaceutical Biosciences, Uppsala University, Uppsala, Sweden
Objectives: Video-electroencephalography (VEEG) monitoring is the gold standard for diagnosis of patients with seizure-related disorders. The aims of this study were to develop a repeated time-to-event (RTTE) model for the occurrence of epileptic seizures during VEEG, and to explore clinical information as predictors of the occurrence of seizures.
Methods: A retrospective analysis was carried out with data collected from epileptic patients submitted to VEEG at the Hospitals of the University of Coimbra, Portugal (1998-2005). To stimulate the occurrence of seizures during the VEEG, a drug discontinuation protocol was followed where drugs were withdrawn in a sequential fashion. To describe the time to occurrence of seizures, a parametric survival model defined in terms of hazard was fit to the data using NONMEM 7.3 [1]. Initially, the event distribution was identified. Thereafter, the number of anti-epileptic drugs (AEDs) each patient was taking during the study period was assessed as a time-varying covariate. Finally, an exploration of other covariates (age, sex, body weight, type of epilepsy, location of the epileptogenic focus) was done. The predictive performance of the model was evaluated with Visual Predictive Checks of the Kaplan Meier curves [2].
Results: Data from 111 inpatients were analysed (12-64 years) of which 81 experienced at least one epileptic seizure. The mean evaluative period was ~161 hours and the median number of seizures per patient was 2 (range 0-16 seizures). The RTTE data were best described by a Weibull distribution with parameters estimated independently for the first and for subsequent events. Inclusion of the change in the AEDs on the baseline hazard improved the model statistically significantly (p<0.001) implying an increasing risk for a seizure with a decreasing number of AEDs. The relationship was allowed to differ for the first event and repeated events. Patients with extra-temporal lobe epilepsy were found to have a higher risk for having a seizure (p<0.05) compared with patients with temporal lobe or generalized epilepsy.
Conclusions: A parametric RTTE model including the effect of number of AEDs and allowing for a difference in the distribution function of events for the first and for the subsequent seizures, described the data well. Patients with extra-temporal lobe epilepsy were found to have a higher risk of experiencing a seizure.
Acknowledgments: Leonardo da Vinci programme T4CD by the University of Coimbra.
References:
[1] Beal, S., Sheiner, L.B., Boeckmann, A., & Bauer, R.J. NONMEM User’s Guides 1989-2009; Icon Development Solutions, Ellicott City, MD, USA, 2009.
[2] Holford N, Lavielle M. A tutorial on time to event analysis for mixed effects modellers. PAGE 20 (2011) Abstr 2281 [www.page-meeting.org/?abstract=2281].
Reference: PAGE 23 (2014) Abstr 3180 [www.page-meeting.org/?abstract=3180]
Poster: Drug/Disease modeling - CNS