Predictors of surgical intervention following stereo-electroencephalography in drug-resistant epilepsy

Stereo-electroencephalography (sEEG) is increasingly used in the presurgical evaluation of drug-resistant epilepsy (DRE). While most studies emphasize predictors of long-term seizure freedom, less is known about which patients actually proceed to a therapeutic intervention following invasive monitoring.

We performed a retrospective review of consecutive patients with DRE who underwent sEEG at Albany Medical Center between 2019 and 2024. Clinical, radiographic, and electrophysiologic variables were compared between patients who proceeded to intervention-defined as resection, laser ablation, or neuromodulation-and those who did not. Multivariable logistic regression identified independent predictors of intervention.

Twenty-seven patients underwent sEEG during the study period; 18 (66.7%) proceeded to a therapeutic intervention. Event capture occurred in all patients who underwent intervention versus 55.6% of those who did not (p = 0.007), while habitual seizure capture was more common among treated patients (83.3% vs. 33.3%, p = 0.026). Longer epilepsy duration (>10 years) was associated with intervention (88.9% vs. 44.4%, p = 0.023). In multivariable analysis, habitual seizure capture was the strongest predictor of treatment candidacy (OR 15.3, 95% CI 0.99-233.96, p = 0.050). Subgroup analysis suggested that resection/ablation was associated with more focal implantation strategies (fewer electrodes, less bilateral coverage), whereas neuromodulation was more often chosen after broader or bilateral implantation.

In this contemporary sEEG cohort, habitual seizure capture and longer epilepsy duration were associated with proceeding to intervention, while implantation scope influenced treatment type. These findings highlight that predictors of surgical candidacy differ from those of long-term seizure outcomes, with important implications for patient selection, counseling, and efficient use of invasive monitoring.