Asymmetric sleep in patients with focal epilepsy

Abstract-Text (inklusive Referenzen und Bildunterschriften)

Introduction: Sleep is a multi-focal phenomenon, and sleep biomarkers engage delimited brain regions rather than the whole brain at once. While this has been extensively studied using physiological paradigms, our understanding of the local disruption of sleep-related brain activity by pathological processes remains limited. Epilepsy is known to disrupt sleep architecture and efficiency, but we lack a comprehensive understanding of the extent of the impact that epilepsy has over sleep markers.

Objectives: Here, we retrospectively included 69 patients with focal epilepsy who were recorded during at least two consecutive nights to investigate whether a focal disease can be associated with a perturbed expression of sleep markers. We provide a comprehensive profiling of the main sleep electrophysiological activities (slow oscillations, spindles and sleep slow waves).

Materials and Methods: From our presurgical evaluation EEG database, we retrospectivity included sixty-nine patients (29 females) with a lateralized epileptic focus (34 left-sided). We measured the inter-hemispheric asymmetries of slow oscillations power (delta range, 0.5-4 Hz, Welch"s power spectral density estimate); spindle density (number per min, identified automatically with a published detector), amplitude, duration and locking to slow oscillations; sleep slow wave density, amplitude, duration and slope. Locking to slow oscillations was measured with inter-trial coherence. Slow waves were identified as negative deflection in the filtered signal (0.5-4 Hz) lasting between 250-1000 ms. To test whether population-based differences represented actual differences at the single patient level, we used a cluster analysis to classify patients with left and right focal epilepsy.

Results: We observed an asymmetry of slow oscillations power in favour of the epileptic hemisphere (p=0.0081) in line with an asymmetry of slow wave amplitude, which was higher in the epileptic hemisphere (p=0.0158). We also observed an asymmetry of slow wave slope, which was steeper in the epileptic hemisphere (p=0.0096). Last, we observed an asymmetry of spindles density, which were more frequent in the non-epileptic hemisphere (p<0.0001) and spindles amplitude, which was higher in the epileptic hemisphere (p=0.0402). We further show that patients with left and right epilepsy can be classified above chance level (accuracy: 65%) using these sleep markers within a tree classifier. Remarkably, combining the asymmetry index of IEDs and sleep markers improved slightly but significantly the classification accuracy of IEDs alone.

Conclusion: Our results establish that focal epilepsy is associated with asymmetries in specific sleep markers features. The significant classification accuracy suggests that differences occur not only at the population level, but also at the single patient level. Sleep markers could thus become valuable supportive EEG graphoelements in the investigation of epilepsy.