Abstracts

Rationale: Sleep spindles (waxing-and-waning 10–16 Hz oscillations) are considered one of the main EEG hallmarks of non-REM sleep. Several studies showed that non-REM sleep facilitates epileptic activity in the brain; however, it is rather unknown whether or how an epileptogenic focus (epi-focus) alters the generation of spindles.

Here, we examined children with drug-resistant epilepsy (DRE), who underwent surgery and became seizure-free, and we localized the cortical generators of sleep spindles before and after surgery using routine scalp EEG along with the children’s MRIs.

The objectives of the study are to: (i) localize the spindle cortical generators in children with DRE using a non-invasive technique; (ii) quantitatively assess their spatial relationship with the epileptogenic focus (epi-focus) and with the cortical generators of epileptic spikes (spiking-areas); and (iii) provide a better understanding of the evolution of sleep spindles when an electrophysiological compensation takes place in the epileptic brain (i.e. after surgery).

Methods: We included 19 children who had epilepsy surgery at Boston Children’s Hospital. For each patient, we analyzed N2-sleep EEG (19-24 channels) collected before and after surgery. We identified spindles and labeled them as slow or fast based on their power spectrum (Fig.1A). We estimated the cortical source of each spindle through electrical source imaging (ESI; Fig.1B). We delineated each child’s Epi-Focus (Fig.1B) by identifying the resection on the post-op MRI and performed ESI on interictal spikes to localize spiking-areas. For each spindle, we computed distance from Epi-Focus (distance_EpiF) and from spiking-area (distance_Spikes). We tested effect and interaction of two factors (spindle type: slow/fast; phase: pre-surgery/post-surgery) on spindle characteristics using two-way ANOVA.

Results: We observed surgery-related changes in power and frequency of slow-spindles (Fig.2A), but not fast-spindles. Similarly, the spatial extent decreased after surgery for slow-spindles (p < 0.001) but not for fast-spindles. Slow-spindles presented shorter distance_EpiF and distance_Spikes compared to fast-spindles before surgery, but not after surgery (p < 0.001, Fig.2B). Higher percentage of slow-spindles than fast-spindles co-localized with spiking-areas (Fig.2C; p=0.02). We also found that the frontal lobe activation increased after surgery (p=0.006) and was higher for slow than fast spindles (p < 0.001). Moreover, before surgery, the frontal generators of slow spindles were more ipsi-lateral to the epi-focus than contra-lateral (64%, vs. 38%; p< 0.001); contrarily, after surgery, no difference was found.