High-frequency oscillations for the non-invasive localisation of the seizure onset zone

Abstract

Epilepsy affects more than 50 million people worldwide, accounting for 1% of the global burden of disease. The treatment of epilepsy relies on the use of antiepileptic medication to keep seizures under control. However, one-third of patients with epilepsy will suffer from ongoing seizure activity despite adequate antiepileptic treatment, contributing to nearly 80% of the cost of epilepsy. Drug-resistant epilepsy is a serious problem that is often associated with significant morbidity and increased mortality rate, due to accidents, suicide and sudden unexpected death in epilepsy.Epilepsy surgery offers an opportunity to treat patients with focal drug-resistant epilepsy by the resection of epileptogenic tissue. Surgical treatment of epilepsy is a well-established and effective option for patients with poor seizure control, but it remains one of the most underutilised medical interventions. An important factor that limits the use of surgery in the treatment of drug-resistance epilepsy is difficulty in identifying, localising or delineating the surgical target. The lack of a clearly defined surgical target may prevent patients from being considered surgical candidates and inability to clearly delineate epileptogenic tissue may also reduce the likelihood of seizure-free outcome.Current non-invasive epilepsy biomarkers have limitations in their ability to locate and define the true extent of the surgical target. For many years, interictal epileptiform discharges (IEDs) have been considered the hallmark of epileptogenic tissue. However, IEDs may lead to false localisation by showing a different and frequently much wider distribution than the tissue responsible for seizure generation. It has been proposed that some IEDs may define the epileptogenic zone more specifically but reliable differentiation between the clinically relevant ‘red spike’ and the non-specific and propagated ‘green spike’ is currently not possible.In the last decade, interictal high-frequency oscillations (HFOs), which are brief low-amplitude electrographic oscillations (> 60Hz), have shown promise as a reliable biomarker in epilepsy. In particular, HFOs are localised epileptogenic tissue more specific than the traditional ictal and interictal electrographic patterns, and resection of regions generating HFOs is associated with seizure-freedom. Recent findings have also revealed that HFOs often co-occur with scalp-recorded IEDs, suggesting the possibility that IEDs with concurrent HFOs could represent red spikes.This thesis explores the possibility of using red spikes as an imaging biomarker for the localisation of epileptogenic tissue. Electroencephalographic (EEG) data from long-term EEG monitoring of presurgical epilepsy candidates were collected and compared the scalp voltage distribution between green and red spikes. This revealed significant differences in the voltage distribution between both spike populations. Using topographic voltage maps, the spatial correlation between each map and the EEG recorded during simultaneous EEG and functional magnetic resonance imaging (EEG-fMRI) were calculated. The time course of the correlation analysis was used to examine the regional cerebral haemodynamic changes associated with each green and red spike voltage maps. The analysis of the data showed that brain regions associated with green and red spike voltage maps have similar statistical strength and shared a high level of spatial overlap, albeit not spatially identical.The concordance of red and green spikes with the clinical localisation of the presumed epileptic focus were also analysed, for which each metabolic correlate was compared with (1) standard spike-triggered EEG-fMRI activations and assessed their concordance with (2) presurgical structural and functional neuroimaging findings. The investigation revealed that the neural substrate of red spikes is not more specific than green spikes for the localisation of the epileptogenic tissue, revealing that both spikes show a maximum haemodynamic response ipsilateral and concordant with the lobar localisation of the presumed epileptogenic region.Additionally, this PhD thesis also investigated whether the voltage map BOLD responses of red spikes reflect a shared neural substrate between patients and healthy population. The analysis revealed that (1) HFOs associated with spikes are pathological as their patterns of activation are exclusive to patients with focal epilepsy that are not reproduced in healthy individuals, and thus reflect a pathology-specific epileptic network associated with the occurrence of the red spike map.