Approche multi-niveaux des modifications précritiques dans l'épilepsie
Louis Cousyn · theses.fr (ABES) · 2023
Epilepsy is defined as the repetition of seizures, with varied symptomatology, whose spontaneous and unpredictable nature accounts for the chronic disability in patients. About one-third of patients are resistant to antiepileptic drugs, i.e. uncontrolled epileptic seizures despite at least two adequate molecules. The study of the mechanisms underlying the transition between an interictal state (literally “between two seizures”) and the seizure – also known as ictogenesis – has long suggested the existence of preictal modifications. The preictal state, which refers to the period preceding the seizure, is the cornerstone of seizure prediction methods. Identifying sensitive and specific preictal biomarkers would be an opportunity for therapeutic strategies to stop the “road to the seizure”. My thesis work aimed to characterize preictal states using: i) subjective clinical symptoms; ii) standardized measures of cardiac activity and intracerebral functional connectivity; iii) recordings of in vivo neuronal activities and in vitro epileptiform activities in patients with epilepsy. I divided my approach into several levels: 1) The evaluation of prodromal symptoms using daily four-point self-assessment questionnaires in patients: several machine-learning algorithms, based on support vector machine classifiers, have shown good performance prediction of the preictal state. A clinical study using the mobile application Epiday, which was developed based on my thesis work, will prospectively analyze this approach in real-life conditions. 2) The analysis of heart rate variability – as a reflection of sympathetic/parasympathetic balance – during a daily standardized resting-state protocol, which exhibited a good distinction between inter and preictal states by using a classifier. The pseudoprospective analysis showed correct overall probabilistic predictions of the seizure risk but with a moderate sensitivity that requires some adjustments before considering any clinical application. 3) The study of functional connectivity using intracerebral EEG recordings, also during daily resting-state protocols: the classification algorithm based on phase synchrony showed the best performances in distinguishing between inter and preictal states in the theta band. These results were significantly superior to those of the classifier based on interictal epileptiform discharges. The daily probabilistic forecasts of the seizure risk were also promising. Transposing the connectivity data into a hyperbolic space showed similar results. 4) In vivo recordings of neuronal activity using intracerebral microelectrodes: I was able to identify a large number of neurons and monitor their activity several hours before the seizure. I intend to continue my thesis work by looking for preictal modifications of the individual and interindividual behavior of these neurons. 5) In vitro recordings from postoperative human epileptic brain tissue allowed me to record interictal-like and preictal discharges, which I was able to correlate with data acquired in vivo. My thesis work provided new data suggesting the existence of preictal dynamic modifications, from machine-learning algorithms and the implementation of a daily standardized resting-state protocol. My work paves the way for the analysis of changes in neuronal activities from intracerebral EEG recordings.