Abstracts

Most sleep-onset partial seizures occur during NREM sleep, possibly due to neuronal hypersynchrony facilitated by the same thalamocortical networks that modulate sleep spindles during normal Stage 2 NREM sleep. Previous studies of sleep-onset partial seizures have lacked precise localization of the epileptogenic focus. We studied sleep stage prior to seizure onset in well-localized mesial temporal epilepsy, hypothesizing that sleep-onset seizures would predominantly occur during Stage 2. We identified consecutive seizure-free patients following anterior temporal lobectomy (ATL) from 1993-2001 with video-EEG captured seizures in both wakefulness and sleep. We analyzed each seizure for sleep stage at onset. Sleep stage was determined by standard Rechtshaffen and Kales criteria modified as follows: no chin EMG, frontopolar leads for eye movement determination, bipolar montage for slow wave amplitude, and NREM stages 3 and 4 grouped together as slow wave sleep. We excluded all simple partial seizures without accompanying ictal EEG change and seizures not containing an adequate montage for visual sleep staging. 40 patients were seizure-free following ATL. 23 (10 men and 13 women) patients had seizures in both sleep and wake states. The mean number of seizures recorded per patient was 13 (range: 5-43). There were a total of 335 (176 right and 159 left) temporal onset seizures. 106 (32%) arose from sleep. Sleep seizures were evenly distributed between Stages 1 (54 seizures) and 2 (51 seizures) NREM sleep, with a single seizure from slow wave sleep and none in REM. Sleep-onset seizures in mesial temporal lobe epilepsy occur almost exclusively in light NREM sleep, and only rarely from slow wave sleep. REM onset seizures were not seen. Limitations of our methodology imposed by typical video-EEG recording specifications including lack of electrooculogram leads and chin electromyography could have lead to errant staging of tonic REM sleep as Stage 1 NREM sleep. Also, slow wave sleep could have been underestimated due to utilization of bipolar montages for delta slow wave amplitude instead of central leads referenced to the alternate ear as required by Rechtschaffen and Kales criteria. We conclude that mesial temporal lobe sleep-onset partial seizures occur predominantly in light NREM sleep. Future prospective studies utilizing video-EEG polysomnography techniques would allow for more accurate sleep staging prior to seizure onsets.