Authors :
Presenting Author: Nicholas Paleologos, BS – NYU Langone Health
David Collins, MD, PhD – UCSF Medical Center
Jiyun Shin, PhD – NYU Langone Health
Anna Maslarova, MD, PhD – NYU Langone Health
Simon Henin, PhD – NYU Langone Health
Alia Seedat, BS – NYU Langone Health
Gyorgy Buzsaki, MD, PhD – NYU Neuroscience Institute
Anli Liu, MD, MA – NYU Langone Health
Rationale:
Hippocampal sharp-wave ripples (SWRs) play an important role in human memory. Interictal epileptiform discharges (IEDs) compete with SWRs during NREM sleep, likely contributing to memory deficits in epilepsy patients (Gelinas 2016). The anterior and posterior hippocampus (HPC) have distinct connections with cortical and subcortical areas, suggesting unique physiological functions are affected in temporal lobe epilepsy (TLE) depending on where the seizure onset zone lies within the long axis of the HPC. Here we study the relationships between HPC SWRs and IEDs along this axis in patients undergoing intracranial EEG monitoring for surgical evaluation.
Methods:
24 hours of continuous iEEG were analyzed from surgical epilepsy patients (n = 6; three females). Patients had an average age of 30 (range: 17 to 48). Seizure onset zones (SOZ) varied from the temporal lobe (1 left, 1 right, 1 bilateral), parieto-occipital lobe (n = 1), and multifocal (n = 2). LFPs were obtained by resampling the data at 512 Hz. Channels were re-referenced by subtracting the white matter signal from neighboring contacts. NREM sleep periods were determined by calculating the spectral power ratio between delta (0.5-4 Hz) and gamma (20-30 Hz) frequency bands. Only stable NREM epochs of >= 5 minutes were used for analysis. IEDs were detected using a 10-60 Hz passband with a peak threshold of 15 SDs from background. SWRs were identified using an 80-200 Hz passband and >= 5 SDs for peak detection. To avoid conflation of IEDs and SWRs, events detected within 500 ms of an IED from the same electrode were discarded. A ripple channel was selected from each HPC electrode by confirming CA1 placement and inspecting spectral features of individual events. The 1/f-corrected PSD was computed for events from the same channel and used as input for UMAP analysis to further distinguish between event types. Candidate events were then manually curated. Localization of channels within anterior or posterior HPC was performed by standard segmentation tools and visual confirmation.Results:
The mean NREM sleep duration was 195 ± 23 minutes (1 SE). Five patients had anterior HPC electrodes (1-3/patient, total = 7), while two had posterior HPC coverage (1-2/patient, total = 3). The mean anterior ripple rate was 1.20 ± 0.36/min while the posterior ripple rate was 0.71 ± 0.42/min (p = 0.67). The mean anterior IED rate was significantly lower than in the posterior HPC (1.75 ± 0.44/min vs 5.35 ± 0.90/min; Welch’s t = -3.59, p = 0.037). Ripple rates were negatively correlated with IED rates on the individual patient level (Figure 1D). Ripple peak frequency for individual events was significantly lower in the anterior HPC compared to posterior HPC (Figure 1E; median: 83 ± 0.24/min vs 85 ± 0.68/min Hz; Mann-Whitney U = 274,553, n1 = 1561, n2 = 397, p = 0.0004 two-tailed).
Conclusions:
Our findings suggest that population activity in the anterior and posterior HPC is independent of each other in surgical epilepsy patients with different SOZs. In both poles of the longitudinal axis, IEDs are shown to suppress intrinsic HPC SWR rates, supporting their contribution to cognitive dysfunction in TLE patients.
Funding: FACES FY2025 Research Grant