Abstracts

Rationale: The diverse types of interictal epileptiform discharges (IEDs) observed in humans have been well-characterized using intracranial recordings of local field potentials with macroelectrodes. However, few studies have investigated how populations of single neurons may give rise to IEDs, including different types of IEDs in the same patient. Relatedly, since single neuron recordings often yield sparse units, it remains unknown how IEDs may emerge from unit activity in different laminar positions of human cortical columns.


Methods: We used Neuropixels high-density multielectrode probes to simultaneously record single-unit action potentials and local field potentials (LFPs) across the entire depth of the cortical lamina in seven patients undergoing surgery for epilepsy. Principal component analysis on the 383-channel LFP signal was used to derive a single representative timeseries for each recording, and IEDs were marked on this aggregated LFP trace by a trained epileptologist. Each IED waveform was subsequently time-locked to its maximum absolute deflection point, and non-negative matrix factorization (NMF) was used on a line-length transform intermediate to partition the various IED waveforms into clusters representing the different typical spike morphologies within each patient. We used Kilosort 2.5 to identify putative single units from the Nueropixel probe and aligned their firing to the onset of IEDs.


Results: Of the seven patients, three had frequent IEDs in at least one recording site, whereas two had rare IEDs, and two had no IEDs detected. We found between 2-5 distinct IED morphologies profiled using NMF. On average, there were 112 neurons isolated per recording site in patients with IEDs (range: 96-144) with 35 (range: 13-72) significantly modulated by IEDs. We found heterogeneous patterns of firing rate modulations aligned to IEDs, with increased or decreased firing aligned around the maximum IED amplitude change. IED amplitude and duration features were not consistently related to neuronal firing rates. Depth information along the NP probe revealed that in one patient with frequent IEDs there were higher IED-locked neuronal firing rates for neurons in deeper cortical layers compared to more superficial layers.


Conclusions:
Our results demonstrate detection and subtyping of different IED morphologies using Neuropixels probes, and a diversity of single neuron firing modulation patterns in cortical columns during IEDs. Differential neuron activity patterns across cortical layers could suggest possible laminar-based influences in IED morphology, but additional recordings will be crucial for generalizability across brain regions and patients. Future single unit and laminar perspectives at the cellular scale may elucidate fundamental properties governing the distinct contributions of neuronal populations to IEDs, and help design better mechanistically-informed neuromodulation strategies for severe epilepsy.




Funding: U01NS117765, K23NS110920