Preoperative interhemispheric coherence as a potential predictive marker for seizure outcome after total corpus callosotomy in nonlesional generalized epilepsy: a scalp EEG study.

Objective: Corpus callosotomy (CC) is one of the palliative epilepsy surgical procedures available for nonlesional generalized epilepsy, but it is more invasive than other palliative surgical procedures. The main challenge is proper selection of suitable patients for CC. Coherence analysis is a method for evaluating brain connectivity, but the correlation between preoperative coherence and surgical outcomes has not previously been clarified. The authors aimed to evaluate correlations between preoperative interhemispheric coherence and surgical outcome in patients with nonlesional generalized epilepsy.

Methods: This retrospective study investigated patients with nonlesional generalized epilepsy who underwent total CC. The authors collected data for patients with good seizure outcome (Oguni classification A or B) and bad seizure outcome (Oguni classification D). For coherence analysis, the authors selected a period without interictal discharges. Preoperative interhemispheric coherence values from 8 pairs of symmetrically opposite scalp electrodes were computed across 5 frequency bands. Then, the authors evaluated correlations between coherence and surgical outcomes.

Results: Forty patients were included (19 males and 21 females). The mean (range) age at the time of surgery was 5.1 (1-18) years. Seizure outcomes were good in 15 patients and bad in the other 25 patients. Age at onset of epilepsy, duration of epilepsy before surgery, age at time of surgery, and presence of epileptic spasm did not differ significantly between patients with good and bad seizure outcomes (p = 0.36, p = 0.14, p = 0.10, and p = 0.20, respectively). Significant correlations were identified between higher Fp1-Fp2 interhemispheric coherence values in the delta, theta, and alpha frequency bands and bad surgical outcomes (p = 0.0397, p = 0.0322, and p = 0.0476, respectively). The receiver operating characteristic curves of the Fp1-Fp2 coherence values in these frequency bands showed areas under the curve of 67%, 69%, and 67%, respectively. The optimal cutoff values for Fp1-Fp2 interhemispheric coherence to predict surgical outcomes were 55.6 for delta (66.7% sensitivity and 72.0% specificity), 55.9 for theta (60.0% sensitivity and 76.0% specificity), and 50.3 for alpha (53.3% sensitivity and 84.0% specificity).

Conclusions: This is the first study to identify potential predictive factors for surgical outcomes based on preoperative interhemispheric coherence in nonlesional generalized epilepsy. Higher coherence between Fp1-Fp2 in the delta, theta, and alpha frequencies correlated with bad seizure outcome after CC.