Sleeping with Hippocampal Damage

Abstract

The hippocampus plays a critical role in sleep-related memory processes [1-3], but it is unclear which specific sleep features are dependent upon this brain structure. The examination of sleep physiology in patients with focal bilateral hippocampal damage and amnesia could supply important evidence regarding these links. However, there is a dearth of such studies, despite these patients providing compelling insights into awake cognition [4, 5]. Here, we sought to identify the contribution of the hippocampus to the sleep phenotype by characterizing sleep via comprehensive qualitative and quantitative analyses in memory-impaired patients with selective bilateral hippocampal damage and matched control participants using in-home polysomnography on 4 nights. We found that, compared to control participants, patients had significantly reduced slow-wave sleep-likely due to decreased density of slow waves-as well as slow-wave activity. In contrast, slow and fast spindles were indistinguishable from those of control participants. Moreover, patients expressed slow oscillations (SOs), and SO-fast spindle coupling was observed. However, on closer scrutiny, we noted that the timing of spindles within the SO cycle was delayed in the patients. The shift of patients' spindles into the later phase of the up-state within the SO cycle may indicate a mismatch in timing across the SO-spindle-ripple events that are associated with memory consolidation [6, 7]. The substantial effect of selective bilateral hippocampal damage on large-scale oscillatory activity in the cortex suggests that, as with awake cognition, the hippocampus plays a significant role in sleep physiology, which may, in turn, be necessary for efficacious episodic memory.

Keywords: amnesia; episodic memory; fast spindles; hippocampus; memory consolidation; polysomnography; ripples; sleep; slow oscillations; slow-wave sleep.

Figure 1

Sleep Stage Comparisons between the Patient and Control Participants (A) To assess potential differences in sleep architecture between the patients and control participants, we averaged sleep studies conducted on 3 separate nights (1-2-3; means and SEMs are shown). CTL, control participants; HPC, hippocampal-damaged patients; M1, left mastoid; M2, right mastoid; PSG#, polysomnography recording; REM, rapid eye movement sleep; SWS, slow-wave sleep. Patients spent significantly less time in SWS compared to the control group (^∗p = 0.016; see inset for a magnified view of the group difference in SWS). On the top right, we report the electrode layout we used for the EEG (STAR Methods). Scoring of sleep stages was based on the current American Academy of Sleep Medicine scoring rules. (B) Cumulative percentages of time spent in each sleep stage are shown and highlight the low variability across the 3 nights in both groups. Specifically, SWS was not different across the 3 nights for patients (Friedman statistic = 0.29; degrees of freedon [df] = 2; p = 0.87) or for the controls (Friedman statistic = 0.67; df = 2; p = 0.72). White arrows indicate the percentage of SWS on each night for the controls, whereas SWS was significantly reduced in the patients.

Figure 2

Power Spectra in N2 Sleep in the Patient and Control Participants (A) EEG power from 0.6–20 Hz for N2 sleep in patients (green line) and controls (purple line). This graph is based on the central EEG electrode C4. We used a data-driven bootstrapping approach to assess group differences in EEG power density spectra. Bootstrapped tests showed that the patients had reduced power in N2 delta activity compared to controls, from 2.0 to 3.2 Hz; p < 0.05 (indicated by gray shading on the graph). (B) Topographical head plots of EEG-quantified differences between patients and controls in slow-wave activity (SWA) (2–3.2 Hz) for C4. Cooler colors represent lower values. SWA is represented for patients in the left panel and for controls in the middle panel. The relative difference between the two groups (on the right) shows that the patients had decreased SWA (darker blue) compared to the control participants.

Figure 3

Example Sleep-Related Oscillations from a Patient and a Control Participant in N2 Sleep Here, we show averaged sleep-related oscillations for one patient (HPC, left panels, green lines) and one control participant (CTL, right panels, purple lines) for fast (upper panel) and slow (middle panel) sleep spindles as well as for slow oscillations (bottom panel) in N2 sleep with the events time-locked to the most negative trough (time = 0). These graphs demonstrate that characteristics of these sleep markers, such as their shape and amplitude, were similar between the two groups (STAR Methods; Data S1A and S1B).