Authors :
Presenting Author: Dhinakaran Chinappen, M.Eng, MBA – MGH, BU
Anirudh Wodeyar, PhD – Neurology – MGH/HMS; Hunki Kwon, PhD – Research Fellow, Neurology, MGH/HMS; Katherine Walsh, BS – Clinical Research Coordinator, Neurology, MGH; Erin Berja, BA – Neurology – MGH; Wen Shi, PhD – Research Fellow, Neurology, MGH/HMS; bryan Baxter, PhD – Instructor, Neurology/Psychiatry, MGH/HMS; Dimitrios Mylonas, PhD – Instructor, Neurology/Psychiatry, MGH/HMS; dara Manoach, PhD – Professor, Neurology/Psychiatry, MGH/HMS; Mark Kramer, PhD – Professor, Mathematics and Statistics, Boston University; Catherine Chu, MD – Associate Professor, Neurology, MGH/HMS
Rationale:
Slow oscillations (SO) and sleep spindles are prominent brain oscillations during non-rapid eye movement (NREM) sleep that support sleep-dependent memory consolidation. Quiet sounds timed to the up-state of SOs (closed-loop auditory stimulation, CLAS) during NREM is a novel, non-invasive tool to increase SOs and spindles with potential applications across many neurological and psychiatric disorders, including epilepsy. The optimal timing of CLAS to induce SOs and spindles across thalamocortical circuits is not known. Further, the impact of CLAS on these thalamocortical rhythms in patients with epilepsy is not known. We evaluated the impact of different detection parameters and timing of CLAS on SOs and spindles in the thalamus and the cortex in patients with epilepsy.
Methods:
A total of 18 subjects (7F, ages 8.1-59.8 years old) with epilepsy from a variety of etiologies undergoing direct thalamic recordings during their epilepsy surgery evaluations were included. Subjects wore headphones that delivered a quiet auditory click timed to the upstate of SOs (0.5-4 Hz) detected in the scalp electroencephalogram (FZ) during NREM sleep. Following a night of CLAS, subjects slept with headphones and SOs were detected, but no stimulation (sham) delivered (n=13). Electrode locations were confirmed on post-operative MRI or co-registration of pre-operative MRI and post-operative CT. Thalamic nuclei were reconstructed using FreeSurfer. For SO detection, thalamic electrodes and FZ electrodes were referenced to a non-cephalic electrode placed on the second spinous process. For spindle analysis, thalamic electrodes were referenced to an adjacent electrode and FZ referenced to the common average reference. Data were divided into single trials centered on auditory stimulation. Trials were rejected as artifact if the amplitude exceeded ± 500 µ V. Evoked responses were compared between CLAS and sham nights in scalp EEG and the thalamus.
Results:
Across patients, CLAS during NREM sleep increased the peak amplitude and decreased the mean duration of the upstate of the endogenous SO (p< 0.001, p< 0.001) and evoked SOs in the thalamus and cortex compared to sham (p< 0.05, Figure 1 A, B). Across patients, CLAS evoked SOs with an 80% yield (range 62-95%) in the thalamus and a 78% yield (57-95%) at FZ; higher than observed after sham stimulation (thalamus: 67% [52-80%]; FZ: 45% [27-75%]; p< 0.001, p< 0.001). Stimuli that evoked a subsequent SO were delivered closer to the peak of the upstate of the endogenous SO compared to those that did not (thalamus: -8.2° versus -65.5°, p< 0.001; FZ: -6.7° versus -30°, p< 0.001, Figure 1 C, D). We find no evidence that CLAS targeting high amplitude SOs impact spindles in the thalamus (Figure 2 A, C) whereas CLAS targeting lower amplitude delta oscillations may decrease spindles coupled to the endogenous SOs and increase spindles coupled to the evoked SOs in the thalamus (Figure 2B, p< 0.05, uncorrected).