RESUMO
OBJECTIVE: This study was undertaken to improve understanding of late relapse following epilepsy surgery in pharmacoresistant epilepsy. METHODS: Retrospective comparison was made of 99 of 1278 patients undergoing surgery during 1999-2015 with seizure relapses after at least 2 years of complete seizure freedom with matched controls experiencing continued long-term seizure freedom. Univariate and multivariate analyses were performed. RESULTS: With a mean follow-up of 9.7 years, mean time to seizure relapse was 56.6 months. In multivariate analysis, incomplete resection based on magnetic resonance imaging (MRI), bilateral lesions on preoperative MRI, and epilepsy onset in the first year of life carried a significantly higher risk of late relapse. In patients with late relapse, additional functional imaging with positron emission tomography had been performed significantly more often. Although the differences were not significant in multivariate analysis, doses of antiepileptic drugs were higher in the relapse group preoperatively and in the first 24 months and complete withdrawal was more frequent in the control group (68% vs. 51%). Regarding seizure frequency, most patients had mild seizure relapse (single relapse seizure or <1/month). SIGNIFICANCE: In our predominantly lesional cohort, complete resection of the MRI lesion is the most important factor to maintain long-term seizure freedom. Two patterns of recurrence were identified: (1) incomplete resected lesions with seizure generation in proximity to the initial resection and (2) epileptogenic networks not detected preoperatively or evolving in the postoperative interval and manifesting with new clinical and diagnostic features.