Attenuation of the Post-Movement Beta Rebound in Patients with Epilepsy

Abstract

Epilepsy is a condition defined by persistent risk of seizures: episodes of aberrant electrical activity that disrupt normal brain function and place individuals at higher risk of psychiatric disorders, cognitive impairment, and death. Clinicians today do not have high performing, highly accessible measures for diagnosing epilepsy. The current mainstay of data gathering in epilepsy is electroencephalography (EEG) of spontaneous brain activity interpreted by human experts. EEG often does not yield abnormal electrical signatures of epilepsy, but these known signatures have high sensitivity and specificity for epilepsy. Because long-term neural changes in response to seizures can make future seizures more likely, a good prognosis in epilepsy requires early diagnosis and treatment. Because antiseizure medications (ASMs) can have serious adverse cognitive and behavioral effects, avoiding harm also requires accurate diagnosis. Improving the performance and accessibility of epilepsy diagnostic tools is therefore crucial for improving patient outcomes. Seizures occur in the brain when the level of aberrant, synchronous excitatory activity in a focal epileptogenic circuit or generalized epileptogenic network exceeds the brain’s “seizure threshold” and involves wider areas of the brain. The seizure threshold depends on a balance of excitatory and inhibitory activity. A measure of cortical inhibitory tone could therefore inform epilepsy diagnosis, yet no clinical tool currently reports this property of the brain. Post-movement beta rebound (PMBR) is a response that occurs after movement detectable on quantitative EEG (qEEG) and is a putative marker of cortical inhibition. This dissertation introduces a method for quantifying the PMBR in human patients and tests the hypothesis that the PMBR is reduced in participants with epilepsy, congruent with the reduced seizure threshold in epilepsy, to evaluate whether the PMBR might be used to identify individuals with epilepsy. This dissertation advances a method of systematic data processing for artifact rejection from EEG recordings as well as a method of PMBR quantification. Both methods are based on principled searches for parameters in a pilot dataset composed of EEG from 14 adults without epilepsy who completed a cue-response task. These protocols were then applied to a second dataset of 28 children and adults without epilepsy and 53 children and adults with epilepsy. These 81 participants completed an auditory discrimination go/no-go task during recording, which enabled assessment of behavioral performance and the analysis of trials containing stereotypical movements performed in a controlled context. A permutation testing protocol was devised to create an individualized measure of PMBR and enable evaluation of the test characteristics of the PMBR. Participants without epilepsy displayed a positive correlation between age and PMBR power (r2 = 0.23, p = 1.210-2). PMBR power sharply increased shortly before adulthood (age 17.6 years). Participants with epilepsy did not have an association between age and PMBR power, and mature (> 17.6 years) participants without epilepsy had significantly greater PMBR power than participants with epilepsy (t-test p = 4.610-3). Participants without epilepsy still had significantly greater PMBR power than participants with epilepsy when participants with epilepsy were restricted to 1) patients without gross structural anomalies, 2) patients without moderate to severe cognitive disorders, 3) patients with focal epilepsy, 4) patients with generalized epilepsy, 5) patients with only absence seizures, 6) patients with drug resistant epilepsy, 7) patients without psychiatric disorders, 8) patients without any GABAergic ASMs, and 9) patients for whom recordings were captured at both their full dose of ASMs and when having withheld all ASMs. Permutation testing confirmed that more participants without epilepsy had greater PMBR power than participants with epilepsy (Z-test p = 1.2*10-4). Classification of epilepsy or not based on PMBR permutation testing had an ROC AUC of 0.92-0.95, with 67% specificity at 100% sensitivity and 69% sensitivity at 100% specificity, and high intra-individual reliability (r = 0.631). Overall, this dissertation characterizes the maturational trajectory of the PMBR as detected on EEG, finds that the PMBR is attenuated in epilepsy, develops an individualized PMBR test, and assesses the PMBR test’s characteristics as a potential diagnostic biomarker for epilepsy. The findings of this work demonstrate the promise of the PMBR as an exemplar of a qEEG biomarker that could improve epilepsy diagnosis, care, and outcomes.