Objectives: The aims of this study were to evaluate the demographic and clinical characteristics of children with epilepsy who discontinued their antiseizure medication (ASM), to determine potential predictors of seizure relapse, to calculate the rate of seizure relapse, and to detect long-term seizure outcomes.
Methods: A total of 269 seizure-free children with epilepsy who were decided to discontinue their ASM and were followed up for at least 18 months after ASM withdrawal were retrospectively evaluated.
Results: The enrolled children had been followed up for a median of 46 months (range 18-126 months; IQR: 29-61) after ASM withdrawal and 90 (33.5%) of their seizures relapsed. The median time to seizure relapse was 8 months (range 0.23-117 months; IQR: 2-25). Seizure relapse occurred in 16.7% of the 90 children at 1 month, 45.6% at 6 months, 62.2% at 1 year, 74.4% at 2 years, and 94.4% at 5 years. Univariate logistic regression analyses revealed six predictors significantly related to relapse: age at first seizure, age at diagnosis of epilepsy, intellectual disability, EEG findings after ASM withdrawal, ASM tapering time, and number of seizures on ASM. In multivariate logistic regression analyses, age at first seizure, intellectual disability, and ASM tapering time were not significantly associated anymore. The other three remained independently predictive. Pharmacological control of seizures with monotherapy was restored in 93.3% of the children with seizure relapse.
Conclusion: This study evaluated potential predictors of seizure relapse, some of which have rarely been evaluated in previous studies. Adolescent age at diagnosis, abnormal EEG findings after ASM withdrawal, and high number of seizures on ASM were associated with a higher risk of seizure relapse. Abnormal MRI findings such as malformations of cortical development and hydrocephalus may be potential biomarkers for the risk of seizure relapse.
Keywords: Antiseizure medication; Children; Epilepsy; Relapse; Risk factor; Withdrawal.
Copyright © 2022 Elsevier Inc. All rights reserved.