LAUSR

6 University of Rochester Medical Center, Center for Translational Neuromedicine, Rochester, United States Introduction: The recently developed magnetic resonance (MR) brain imaging technique MR Encephalography (MREG) is an ultra-fast MR-sequence that records ten full threedimensional images of the brain per second [1] . Thereby, the sequence enables non-invasive measurements of cardiac, respiration, and low frequency pulsations in the human brain, pulsations previously suggested to reflect ǵlymphatic function’ [ 1 , 2 , 3 ]. Because the glymphatic system in rodents is enhanced during NREM sleep [ 4 , 5 ], the current study aimed to investigate whether MREG-detected brain pulsations are enhanced during NREM sleep to establish supportive evidence for a glymphatic-like system in the human brain. Methods: Twenty healthy men, aged 18-29 years, completed a within-subject controlled sleep study in which MREG was acquired with a 3T Siemens Prisma MR scanner. Participants were MR scanned both at baseline (10 hours awake) and after being sleep deprived (34 hours awake), and EEG (high-density Electrical Geodesics, Inc., Oregon, US) was recorded during scans. EEG was cleaned from scanner noise and heartbeat artifacts using optimal basis sets. All MR scans were performed at the same circadian time point and entailed ∼30 min wakefulness followed by a ∼ 1.5hr sleep opportunity. To capture changes in brain pulsations between different vigilance states (NREMsleep/wake), MREG was continuously recorded in 5-min intervals throughout the scans. MREG-detected brain pulsations were assessed by analyzing whole-brain spectral power in the respiration (0.14 – 0.34 Hz) and cardiac (0.8 1.2 Hz) frequency bands. EEG recordings were evaluated by two independent experts (scoring agreement: 73%). Where experts agreed on classifying EEG as either wakefulness or NREM-sleep (stage N1, N2 or N3), the corresponding MREG data were analyzed. Effects of NREM sleep and sleep deprivation on MREG spectral power density were evaluated with a linear mixed model on log-transformed data with the fixed effects “vigilance” (wakefulness versus NREM sleep) and “condition” (baseline versus sleep deprivation) and the random effect "subject nr" (1-20). Results: It was harder for participants to sleep in the scanner during baseline than after sleep deprivation (NREM: 4.0 ± 6.8 min versus 28.2 ± 13.2 min). When sleep-deprived, subjects also exhibited substantial amounts of deep NREM sleep (N2 + N3: 24.8 ± 13.5 min). Sleep stages N1, N2 and N3 were grouped and included in analysis as ‘NREM sleep’. Compared to wakefulness, NREM sleep was associated with enhanced MREG whole-brain spectral power in the respiration frequency band (“vigilance”: F 1,32 = 17; P < 0.001), and a trend for an increase was observed in the cardiac band (“vigilance”: F 1,31 = 3.1; P < 0.09). No significant effect of sleep deprivation was observed. Conclusion: We find that in humans, NREM sleep enhances MREG-detected brain pulsations in the respiration frequency band. This is in line with what has been seen in rodent investigations of the glymphatic system and supports a biological relevance of propagating brain pulsations during sleep. No conflict of interest