Computational modeling of high-frequency oscillations at the onset of neocortical partial seizures: from 'altered...

In this paper, a neural mass model is proposed to analyze some mechanisms underlying the generation of fast oscillations (80 Hz and beyond) at the onset of seizures. This model includes one sub-population of pyramidal cells and one sub-population of interneurons targeting the perisomatic region of pyramidal cells where fast GABAergic currents are mediated. We identified some conditions for which the model can reproduce the features of high-frequency, chirp-like (from approximately 100 to approximately 70 Hz) signatures observed in real depth-EEG signals recorded in epileptic patients at seizure onset ("fast onset activity"). These conditions included appropriate alterations in (i) the strengths of GABAergic and glutamatergic connections, and (ii) the amplitude of average EPSPs/IPSPs. Results revealed that a subtle balance between excitatory and inhibitory feedbacks is required in the model for reproducing a 'realistic' fast activity, i.e., showing a reduction of frequency with a simultaneous increase in amplitude, as actually observed in epileptogenic cerebral cortex. Results also demonstrated that the number of scenarios (variation, in time, of model parameters) leading to chirp-like signatures was rather limited. First, to produce high-frequency output signals, the model should operate in a "resonance" region, at the frontier between a stable and an unstable region. Second both EPSP and IPSP amplitudes should decrease with time in order to obey the frequency/amplitude constraint. These scenarios obtained through a mathematical analysis of the model show how some alteration in the structure of neural networks can lead to dysfunction. They also provide insights into potentially important mechanisms for high-frequency epileptic activity generation.