Rationale:
Interictal epileptiform discharges (IEDs) can hijack the physiological coupling between hippocampal sharp-wave ripples (SWRs) and thalamocortical spindles during NREM sleep, likely contributing to memory dysfunction in patients with epilepsy. Anterior and posterior hippocampal regions possess distinct connectivity patterns to cortex in rodents and primates. Here we evaluate the relationship between spindles, slow oscillations (SOs), IEDs and SWRs along the hippocampal long axis during NREM sleep, using intracranial EEG (iEEG) in patients with temporal lobe epilepsy (TLE).
Methods:
We analyzed 24 hours of continuous iEEG from five patients undergoing surgical evaluation. NREM sleep was identified by continuously calculating the mean frequency that showed the highest relative power and using a manually determined threshold, confirmed by visual inspection. SOs were identified on neocortical electrodes as zero crossings in the 0.5-2 Hz frequency bands. Spindles were detected on these same channels as sustained 10-20 Hz events above the background ( >= 4SD, 400-3000 ms). SWRs were identified within the CA1 subfield as sustained 80-120 Hz events above background ( >= 4 SD, 20-100 ms). IEDs were detected as in Janca et al., 2014. To avoid conflation of IEDs and SWRs, SWRs occurring within 500 ms of an IED were discarded. IED and SWR events in anterior vs. posterior hippocampus were analyzed separately.
Results:
Analysis of whole-day data showed clear epochs of high power in the delta frequency band (0.4 – 5 Hz,
Fig 1A). Within these epochs, abundant SOs and spindles were detected (
Fig 1B), with a clear relationship between the SO phase and spindle occurrence (
Fig 1C), with spindles most likely to occur during the SO up-state (
Fig 1D).
A total of 27 anterior and 16 posterior hippocampal electrodes were analyzed, showing clear IEDs and SWRs during sleep (
Fig 2A-B). SWR rates were higher in anterior hippocampus than in posterior across the day, and this difference was preserved during NREM sleep (
Fig 1C; anterior vs posterior U=24, p< 0.05; NREM vs whole-day F(3,91)=28.72, p< 0.05). This was not the case for IEDs, which were observed at equivalent rates in both anterior and posterior hippocampus (
Fig 1D; anterior vs posterior U=7, p=0.43; NREM vs whole day F(3,87)=6.00, p< 0.05). The correlation between cortical signals and SWRs was similar between anterior and posterior hippocampus, with SWRs occurring less frequently around the beginning of SO “up-states” and more frequently around spindle peaks (
Fig 1E). Strikingly, anterior IEDs had no clear relationship with frontal spindles despite posterior IEDs being strongly correlated with them (
Fig 1F).
Conclusions:
Our findings suggest that SWRs and IEDs occur more frequently in anterior vs posterior CA1 in patients with TLE, despite comparable sampling and differing SOZ. We also find that the temporal relationship between SWRs and IEDs and cortical markers of NREM sleep differ, and that these relationships change across the anterior-posterior hippocampal axis. These findings suggest that the anterior and posterior hippocampus are differentially engaged in seizure networks.
Funding: K23NS104252 (AL), F30MH117859 (DC), ONR N00014-17-1-2961 (SH)