Abstracts

Rationale: Hippocampal ripples recorded via IEEG can synchronize with the cortex, garnering attention in both clinical epilepsy and cognitive neuroscience. However, this coupling could be either physiological or pathological. Cognitive studies sometimes overlook epileptic backgrounds, while clinical studies require new features to discern the nature of ripples not superimposed on IEDs. We aimed to validate whether physiological hippocampus-temporal neocortex ripple coupling exists during human NREM sleep and to investigate the impact of epileptogenicity on the coupling features of hippocampal ripples.


Methods: We retrospective reviewed 88 patients with drug-resistant epilepsy who underwent simultaneous intracranial and scalp EEG monitoring and 71 patients from the MNI Normal IEEG Atlas. Grey matter IEEG channels were categorized as the SOZ and non-SOZ (NSOZ). NSOZ was further classified into the exclusive irritative zone (EIZ, with IEDs) and the normal zone (NOZ, without IEDs). NREM sleep stage 2-3 recordings from 15 hippocampi categorized as the NOZ were selected. Data from 12 and 14 hippocampi classified as the SOZ and EIZ, respectively, were also included. Ripples were classified as type I (superimposed on IEDs) or type II (not superimposed on IEDs). The coupling between ripples in the hippocampus and ipsilateral temporal neocortex was evaluated by ripple co-occurrence and ripple-band activity modulation (RAM). Ripple co-occurrence intervals within the hippocampus were also analyzed. The classification potential of ripple coupling features for discriminating hippocampus belonging to the presumed EZ/non-EZ was evaluated using AUROC and AIC.

Results: Only 12.6% of sampled hippocampi were classified as the NOZ. Among the NOZ hippocampi, 16.7% exhibited ripple co-occurrence and 25% demonstrated RAM between hippocampus and temporal cortex. In contrast, when the hippocampus was the SOZ, the proportions were 91.7% and 91.7% for all ripples, and 75% and 91.7% for type II ripples. Multivariate analysis showed that ripple co-occurrence and RAM phenomena were more likely to be observed when the hippocampus was the SOZ compared to the EIZ or NOZ (p< 0.001, p< 0.001), even for type II ripples (p=0.009, p=0.003). In addition, ripple co-occurrence intervals within the hippocampus were significantly longer when the hippocampus was the SOZ than when it belonged to the EIZ or NOZ (p=0.02, p< 0.001, all ripples; p=0.003, p< 0.001, type II ripples). A stepwise regression incorporating ripple coupling features and ripple rate showed that the optimal model distinguished hippocampus belonging to the EZ/non-EZ with an AUROC of 0.914.