(948) Increased sleep spindle power predicts decrease in intellectual function in individuals with epilepsy.

Sleep spindles play an important role in neuroplasticity, including memory consolidation and cortical development. Sleep spindles have also been demonstrated to be related to intelligence (IQ). In healthy individuals, the number of sleep spindles and higher sigma power (12-14 Hz) has been associated with higher IQ. However, preliminary data in an initial cohort of 15 participants suggested that among individuals with epilepsy more sleep spindles related to lower IQ. Here we wished to confirm and extend our previous results by assessing the correlation between sleep spindles power and IQ decline in a larger cohort of adult participants with drug-refractory epilepsy.

Methods:

Forty-seven individuals with drug-refractory epilepsy (26 female, 32 focal [25 with temporal lobe epilepsy] and 13 generalized) underwent high-density EEG (HDEEG, 256 electrodes) over the course of a night in the Epilepsy Monitoring Unit of the University of Wisconsin, Madison. Education levels ranged from 8-18 years. IQ was evaluated using the 3-subtest (Block Design, Digit Span, Vocabulary); mean IQ was 83. Data were compared to 256 electrode HDEEG sleep data from 47 aged-matched healthy volunteers acquired in a sleep lab. After sleep scoring using standard AASM criteria, semi-automated artifact rejection was performed on non-rapid eye movement sleep stage 2 and 3 epochs to reject bad epochs and bad channels. In participants, one-second epochs centered on interictal spikes detected by a certified epileptologist were also rejected from the data. Power spectrum of HDEEG data was then computed in the 12-16 Hz range for both for the first and the last hour of sleep. Spindle power data were then converted to 2D images. We used Statistical Parametric Mapping (SPM) for statistical analysis to compare spindle power in participants vs controls then test the hypothesis of a negative correlation between sleep spindle power and IQ in participants, using a random effects approach to take into account inter-subject variability. Results were corrected for multiple comparisons using SPM family-wise-error (FWE)-corrected p-value < 0.05.

Results:

Group comparison revealed widespread increases spindle power in participants with epilepsy compared to controls that were most significant in bilateral temporal areas. Random effects regression analysis in the epilepsy group confirmed a significant negative correlation between HDEEG spindle power and IQ both in the first hour of sleep (peak-level FWE-corrected p=0.037 in central electrodes; cluster-level FWE-corrected p< 0.002) and in the last hour of sleep (set-level FWE-corrected p< 0.013).

Conclusions:

Our results confirm and extend our previous finding that increased sleep spindle activity in individuals with epilepsy is paradoxically associated with reduced IQ. These findings suggest that seizures may create maladaptive plastic changes which may compete with normal synaptic processes associated with learning. Future studies will correlate other neuropsychological measures with HDEEG spindle and slow-wave characteristics.

Funding: Please list any funding that was received in support of this abstract.: Tiny Blue Dot Foundation
Lily's Fund for Epilepsy Research