Abstracts

Rationale: Hippocampal sharp-wave-ripples (SWRs) are high-frequency oscillations crucial for memory consolidation in mammals. While extensively characterized in rodents, recent characterization of SWRs in human memory behavior is debated due to variances in recording methods and brain states. Human detection approaches are limited by coarse electrode resolution, uncertain placement, basic bandpass filtering and signal contamination by interictal epileptiform discharges (IED). We address these issues by comparing high-resolution hippocampal recordings in an Alzheimer's disease (AD) mouse model and microwire recordings in epilepsy patients. We propose a multi-step method combining neurophysiological, morphological, and spatial features to enhance SWR detection specificity in humans.


Methods: Using 1024-channel silicon probes, we recorded spontaneous SWRs and IEDs from the entire dorsal hippocampus in APP/PS1 transgenic mice (n=5) during quiet wakefulness. We identified key morphological, spectral and spatial features of these two phenomena and applied them to human hippocampal macro-micro recordings (5 frontal lobe, 8 temporal lobe epilepsy patients) during NREM sleep. Hippocampal subfields were segmented using ASHS software, confirmed by visual inspection. Microwires and macroelectrodes in the hippocampal CA1/2/3, dentate gyrus, and subiculum were analyzed. After IED cleaning, candidate SWRs were detected using conventional filtering-based methods (80-250 Hz) and then verified based on the features derived from rodent recordings and UMAP waveform topology analysis.


Results: In AD mice, SWRs were (1) primarily localized to CA1, (2) exhibited clear ripple cycles (³ 3) in raw LFP, (3) coupled with the positive/negative sharp wave in the pyramidal/dendritic layer, and (4) showed a narrow ripple band power increase. Conversely, IEDs had higher voltage amplitude (~15 fold) and were recorded across all HPC subfields. With these four features, we identified human hippocampal SWRs in microwires and macro-contacts in epilepsy patients. Consistent with rodent results, SWRs were only found in the CA1/3 and subiculum, with 52.3% CA1 electrodes showing SWRs. IEDs were large depolarizations visible across multiple contacts, identified in all hippocampal subfields. In TLE patients, SWR-negative channels showed significantly higher IED rates than SWR-positive channels. IEDs reduced SWR rate by 50% for ~3 s. Nocturnal seizures in 3 patients led to almost complete suppression of SWR rate in the first hour and >50% reduction in the second hour compared to one hour pre-seizure baseline. Seizures did not affect IED rate.


Conclusions: SWR detection in human epilepsy patients should be applied with considerable caution. We provide clear criteria defining SWRs from IEDs and false-positives based on high-density recordings in rodents. By reliably separating SWRs from IEDs, we discovered that IEDs and seizures acutely disrupt hippocampal circuits important for memory consolidation. Our methodology provides a firm foundation for future investigation on the roles of SWRs and IEDs in human memory function.


Funding: DFG MA 10301/1- 1, NYU FACES

R01 NS127954 (Liu), K23-NS 104252 (Liu)

Department of Neurology