Abstracts

Despite decades of research, the interplay between sleep physiology and epileptic activity remains poorly defined. In particular, the contribution of the thalamus - a central hub for both sleep regulation and seizure propagation - is not well understood in human epilepsy. Most current knowledge of thalamic sleep physiology derives from animal studies or indirect, non-invasive measures, leaving a major gap. We address this critical gap by testing the hypothesis that epileptic activity governs fluctuations in thalamic spindles, markers of sleep physiology, with periods of heightened epileptic burden decreasing thalamic sleep spindles. We further investigate whether this decrease in sleep spindles is driven by thalamic interictal epileptiform discharges (IEDs) propagated from the seizure onset zone (SOZ).

We analyzed 56 patients who underwent SEEG investigation with thalamic coverage, and at least three days of overnight data. IEDs were detected in the SOZ and thalamus and sleep spindles in the thalamus, using previously validated automatic detectors. Detections were made on 30-60 minutes of continuous non-rapid-eye movement (NREM) sleep data, at least thirty minutes away from seizures. Granger’s causality estimated cortico-thalamic (CT) effective connectivity (broadband: 0.3-512 Hz) over 10 minutes of NREM data per night. SOZ channels with significant connectivity (more than one standard deviation from connectivity distribution) to the thalamus were considered for analysis. We compared the spindle rates across two contrasting nights: minimum and maximum IED rates in the SOZ (Figure 1) and the thalamus (Figure 2). The median spindle rates were taken across the whole thalamus, and when grouping into different nuclei, the median was taken across each nucleus group: ventrolateral (VL), pulvinar (Pul), centromedian (CM) and anterior (ANT) thalamic groups.

We found that there was significant CT suppression of spindles (p=0.037 d=0.32 n=56), reflected by lower spindle rates on the night with the highest SOZ IED rate compared to the night with the lowest. This effect was most pronounced in the CM nucleus (p=0.07;d=0.35) compared to other thalamic nuclei, whereas in the absence of strong CT connectivity, no significant spindle suppression was observed across the whole thalamus (p=0.23;d=0.14), as well as all subnuclei (VL: p=0.19; Pul: p=0.69; ANT: p=0.79; CM: p=0.18). To determine whether spindle suppression is specifically associated to the SOZ, IED rates in the thalamus were compared against thalamic spindle rates. Indeed, spindle rates did not significantly differ between maximum and minimum IED nights when analyzed within the thalamus (p=0.49;d=0.07), with no particular nuclei impacted by thalamic IED rates (VL: p=0.33; Pul: p=0.16; ANT: p=0.52; CM: p=0.29).