Abstracts

Rationale: The aims of the Time to stop (TTS) study are: 1) to evaluate current antiepileptic drug (AED) withdrawal policies after epilepsy surgery in children, 2) to explore determinants of timing of AED discontinuation, and 3) to study the relation between timing of withdrawal and seizure recurrence during/ after AED withdrawal as well as eventual seizure outcome. Methods: The TTS is an international retrospective consecutive cohort study of 778 children, operated between 2000 and 2008 in 15 medical centers in 8 countries, in whom it was decided to start AED withdrawal after postoperative seizure freedom. First, using uni- and multivariate linear regression analysis, preoperative variables previously described to independently predict postoperative seizure outcome were related to time intervals between surgery and start of AED withdrawal (I_start), and complete discontinuation of AED (I_stop). Second, time intervals of withdrawal were related to seizure recurrence during or after AED withdrawal and to eventual outcome (Engel 1 > 1yr) using a Cox regression analysis. To this model we added previously identified other predictors of seizure outcome. Results: Mean I_start (n=778) was 16.8 months (SD 13.8, range 0-82) and mean I_stop (n=446) was 30.4 months (SD 18.0, range 0-105). The interval between start and complete discontinuation of medication (I_withdrawal) was 13.6 months (SD 12.8, range 0-68). Time intervals were independently associated with: participating center, number of preoperative AEDs, direct postoperative seizure freedom, etiology, presence of bilateral MRI abnormalities, type of surgery, IQ scores, intracranial recordings performed and presence of epileptic abnormalities on postoperative EEG. During or after AED withdrawal, 96 children had seizure recurrences. Eventually only 26 children were not seizure free. Shorter I_start, bilateral MRI abnormalities, older age at surgery, higher number of AEDs tried preoperatively, and incomplete resection were associated with seizure recurrence during/ after AED withdrawal. If I_stop was added to the model, however, the only independent predictive variables were I_stop, bilateral MRI abnormalities and incomplete resection. Time intervals were not associated with eventual seizure outcome. Predictors for unfavorable eventual seizure outcome were: higher number of preoperative AEDs, left sided surgery and incomplete resection. Conclusions: Early completion, but not start, of AED withdrawal independently increases the risk of seizure recurrence during or after AED withdrawal. Timing intervals, however, are not related to eventual seizure outcome. Early AED discontinuation may unmasks surgical failure but not at the expense of eventual seizure freedom. These results justify a future multicenter randomized trial to study benefits and safety of very early start of AED withdrawal after epilepsy surgery in children.