Abstract
Advanced EEG technology has revealed that epileptiform activity occurs more frequently in Alzheimer's disease (AD) than previously recognized, prompting debate over the utility of EEG in AD diagnostics. Yet, unlike epilepsy, epileptiform activity is not always observed in AD, leading to skepticism. Historically, this absence has been attributed to limited recording depth or insufficient recording duration. We tested an alternative hypothesis that certain types of epileptiform activity, specifically high frequency oscillations (HFOs, defined as 250-500 Hz fast ripples), inhibit interictal spikes (IIS), which are currently used to assess hyperexcitability clinically. We recorded wideband (0.1-500 Hz) hippocampal local field potentials in three AD (Tg2576, Presenilin 2(-/-), Ts65Dn Down syndrome model) and two epilepsy (intrahippocampal kainic acid, pilocarpine) mouse models during wakefulness and sleep. In both AD and epilepsy, HFOs consistently outnumbered IIS across behavioral states, age and recording contact. However, IIS and HFOs showed divergent relationships: a negative correlation between their rates was observed only in AD, in contrast to a positive correlation in epilepsy. HFOs preceded IIS at much shorter intervals in epilepsy than in AD. Co-occurrence of IIS with ripples did not differ between AD and epilepsy. These findings reveal a novel dissociation between clinically-relevant EEG biomarkers in AD and epilepsy. In AD, HFOs may inhibit IIS, which could lead to underestimation of hyperexcitability and hinder patient stratification for anti-seizure therapies. While non-invasive HFO detection remains challenging, we stress the need for wideband EEG/MEG, particularly in AD, to assess the full extent of hyperexcitability and biomarker interactions that would otherwise remain undetected.