High Frequency Oscillations and Functional Networks to Localize the Epileptogenic Tissue for Surgery

This thesis combined two recent EEG signal analysis techniques that seemed promising to improve estimation of the epileptogenic tissue and thereby surgical outcome: high frequency oscillations(HFOs) and functional network analysis.

The first part introduced new concepts in focal epilepsy. Reviewing recent technical and methodological developments in the presurgical evaluation for epilepsy surgery showed a change from epileptic focus to network when determining the most appropriate surgical strategy. Technical progress revealed microscopic structural and electrical brain signal abnormalities contained within macroscopically normal signals. New analyses arose that focus on the interplay between diseased tissue, the epileptogenic network, and the healthy brain. Alternative novel ancillary investigations such as metabolic and molecular imaging expose previously imperceptible epileptogenic tissue. These developments explain some surgical failures, increase our understanding of the disease and provide opportunities for future surgical planning and patient counselling. It is important to unify recording and storage of the data generated by specialized clinical epilepsy care, thereby facilitating collaborative initiatives and increase sample sizes for research. If treated with care, the resulting large standardized databases will quickly progress knowledge.

The second part focused on the translation of HFOs into clinical practice. We performed the world’s first HFO RCT. We could not demonstrate non-inferiority of HFO to spike-based ioECoG-tailoring for epilepsy surgery in the whole population and TLE subgroup, suggesting superiority of spikes. Yet, when correcting for confounding by poor-outcome pathologies, the analyses were inconclusive for the whole population and TLE subgroup, and showed non-inferiority of HFO to spike-based ioECoG-tailoring in the eTLE subgroup. The trial results challenge the clinical value of HFOs as an epilepsy biomarker, especially in TLE, but indicate that HFO-based ioECoG may hold promise in eTLE. While conducting the HFO-trial, new studies reported conflicting results on the performance of HFOs at individual patient level. This could result from difficulties in discriminating pathological from physiological HFOs and artefacts. We further investigated this and showed that correcting for physiological ripple rates improved the ability of ripples to identify the epileptogenic tissue and predict outcome. Especially in patients with an epileptic focus in eloquent cortex and mesiotemporal brain areas. We also showed that high-density recordings facilitated interpretation for fast ripples.

The third part investigated whether interictal functional network analysis could improve delineation of the epileptogenic tissue. We showed that interictal whole brain networks in focal epilepsies are characterized by a more regular network organization as compared to healthy controls. This network disorganization might be responsible for co-morbid cognitive and behavioural impairments often reported in these patients. Since the presumed pathophysiology of epileptic HFOs suggests focal functional misconnections between neurons, we studied the spatial relationship between electrodes showing interictal spikes and HFOs, and local network measures computed from interictal functional connectivity in the high frequencies. We showed that in event-free data, electrodes displaying interictal events are functionally isolated in the gamma-band and seem functionally integrated in the FR-band. In contrast, we found that in data containing events, electrodes displaying interictal events are functionally integrated in the gamma-band and functionally isolated in the FR-band.