Abstracts

Rationale: Any "healthy" brain can be forced to seize, e.g. after a convulsive electroshock. Are these seizures different from those that occur in epileptic patients, or in patients for whom epilepsy is a co-morbidity (i.e. Alzheimer, autism etc.)? Answering this question requires understanding the dynamics of seizures. How seizures propagate also remains an unaddressed issue. In a given patient, seizures show variable spatial and temporal propagation patterns. The goal of this study is to extract general rules of seizure dynamics and propagation. Methods: We built a mathematical model, The Epileptor, which can fully reproduce the dynamics of seizures with partial onset. The model only requires 5 state variables and simple coupled differential equations. The predictions of the model were tested in various species, including in Humans (using randomly picked seizures in the ieeg database). Detailed experiments were performed in vitro using hippocampal preparations from mice. Results: We establish a taxonomy of 16 different types of seizures. One type, The Epileptor, accounts for 83% of seizures recorded in drug-resistant patients. These seizures are characterized by the presence of fast oscillations and a DC shift at onset, and by a logarithmic scaling of activity at offset. The same pattern is present in flies, zebrafish, and rodents. One state variable acting on a very slow timescale accounts for seizure dynamics. Coupling two Epileptors, we could demonstrate that the slow state variable in central to the spatio-temporal pattern of seizure propagation. Conclusions: Seizures with partial onset are very simple mathematical objects, which onset and offset obey simple rules that are invariant across species. A slow state variable, which biophysical correlates remain to be identified, accounts for seizure genesis and propagation.