Abstracts

Rationale: High frequency oscillations (HFOs) are often seen in epileptic brain regions, both before and during seizures. For this reason, they have been suggested as important biomarkers, with the potential to provide both spatial and temporal information about epileptic seizures. However, little is known about the neuronal mechanisms underlying epileptic HFOs. In particular, there is debate about whether HFOs reflect the rhythmic, synchronous activation of neurons shaped by inhibition or if HFOs are just correlates of increased non-rhythmic neuronal activity. Resolving this debate will help us better understand the relationship between HFOs and epilepsy and shed light on the neuronal mechanisms of human seizures. Here, we present the first description of how human neocortical inhibitory and excitatory neurons interact to generate HFOs during seizures. Methods: Four patients were implanted with intracranial grid electrodes as part of the clinical process of identifying the precise site of origin of their drug-resistant epilepsy. A 4x4 mm Neuroport microarray (Blackrock Microsystems) was also placed in a region of the neocortex that was expected to be in the resection site. We used these arrays to simultaneously record the activity of dozens of individual neurons during ictal activity from layer 3 of the neocortex. We then identified putative excitatory and fast-spiking inhibitory neurons using well-established criteria. The local field potential (LFP) was also recorded and used to identify HFO events (> 150 Hz) exceeding a threshold of 3 standard deviations above the mean of the filtered signal. Results: HFOs occurred preferentially at the falling (negative) phase of slower oscillations throughout a seizure. HFOs, along with their underlying waves, rapidly propagated across the array. As expected, HFOs were strongly correlated to single neuron firing, with the firing rate of both inhibitory and excitatory cells increasing with HFO amplitude. Surprisingly, however, we found no evidence of rhythmic neuronal firing during HFOs. Many excitatory cells fired just once during an HFO event and inhibitory neurons fired non-rhythmically during HFOs. As we have previously shown, inhibitory neurons ceased firing altogether less than half-way through spike-and-wave seizures. However, HFOs continued to occur after inhibitory cessation, suggesting that inhibition from FS neurons plays no role in generating HFOs. Conclusions: Our results suggest that HFOs seen during human seizures are not full-fledged, rhythmic oscillations. In fact, they are simply the spectral correlates of increased non-rhythmic neuronal activity. Importantly, inhibitory neurons were not found to fire at a fixed frequency during HFOs, and HFOs continued to occur after the cessation of FS inhibitory firing midway through a seizure. Thus, HFOs are markers of increased neuronal firing and perhaps increased neuronal synchrony, but they do not reflect truly rhythmic neuronal activity and are not sculpted by rhythmic inhibition.