Neuroimaging Biomarkers of Post-Traumatic Epileptogenesis

Principal Investigator: GUGGER, JAMES

Proposal Number: EP210015

Award Number: W81XWH-22-1-0593

Period of Performance: 7/1/2022 - 6/30/2025

PUBLIC ABSTRACT

Background: Post-traumatic epilepsy (PTE) is the near-ideal condition to evaluate therapies that can prevent epilepsy, as it is one of the most common forms of epilepsy and there is a relatively short latency period prior to onset of recurrent seizures. During this latency period, the brain is thought to undergo changes that increase the propensity for seizures, called epileptogenesis. Successful clinical trials of epilepsy prevention therapies require a better understanding of the process of epileptogenesis. Both TBI and epilepsy are disorders of brain connectivity (i.e., the way the brain network is wired), but the relationship between TBI, brain connectivity abnormalities, and epileptogenesis represents a critical knowledge gap in the field. Brain connectivity or network abnormalities after TBI are associated with chronic physical and neurological symptoms such as memory problems. In epilepsy, connectivity abnormalities can help map the location of seizure onset and spread. The widespread abnormalities in brain networks represent a shared trait between TBI and epilepsy and warrant investigation as a tool for epilepsy prediction after TBI.

Objectives/Hypotheses:

The overall objectives of this proposal are to assess alterations in brain connectivity from 2 weeks to 6 months after TBI and use these measurements as features to develop predictive models to classify patients who will develop PTE. The central hypotheses are: (1) individuals who develop PTE have abnormal network profiles distinct from those seen with TBI of comparable injury severity without epilepsy, and (2) network profiles can accurately classify whether a patient will develop PTE.

Specific Aims:

Aim 1: To characterize the longitudinal profile of structural network measures inferred from diffusion magnetic resonance imaging (MRI) in a sample of uninjured controls. Using orthopedic and uninjured friend control data with an interscan interval of 6 months, we will establish the control distribution for structural network measures.

Aim 2: To characterize structural network measures inferred from diffusion MRI obtained 2 weeks and 6 months after TBI in patients with and without PTE. Specifically, we will examine differences in network measures both cross-sectionally at each time point as well as the difference in the longitudinal profile between patients with and without PTE.

Aim 3: To develop a robust machine learning framework to accurately classify patients who will develop PTE. Specifically, we will utilize structural network measures obtained at 2 weeks and 6 months after injury alongside traditional clinical risk factors for PTE to train models to accurately classify whether a patient will develop PTE.

Research Strategy:

This study will utilize data from one of the largest, well-characterized samples of patients with TBI enrolled in the Transforming Research and Clinical Knowledge in Traumatic Brain Injury (TRACK-TBI) Epileptogenesis Project (TRACK-TBI EPI) where PTE diagnosis will be determined by a neurologist who specializes in the care of patients with epilepsy and confirmed by a centralized panel of experts. Patients in TRACK-TBI have diffusion MRI scans, a powerful noninvasive tool for mapping brain connectivity, at both 2 weeks and 6 months after injury. We will utilize linear mixed effects models to compare structural network measures from these diffusion MRI scans in those with and without PTE. We will use these network abnormalities as features alongside traditional clinical risk factors for PTE to train robust machine learning models to accurately classify whether a patient will develop PTE.

Innovation and Impact:

The proposed research is directly aligned with the ERP’s mission and vision and will use an innovative method of analysis of diffusion MRI to determine if individuals with TBI are at risk of developing PTE. The proposed work has the potential to generate innovative deliverables that will have high impact on the millions of active-duty Service Members, Veterans, and civilians with TBI as well as those with or at risk for all forms of epilepsy. By completion of the proposed aims, we expect to improve our understanding of epileptogenesis, which will inform the design of animal models that replicate human epileptogenesis. Furthermore, we expect to identify a brain connectivity signature of epileptogenesis after TBI that can be used as a noninvasive tool to accurately identify patients at risk of developing PTE that would pave the way for feasible clinical trials of epilepsy prevention therapies.

TECHNICAL ABSTRACT

Background: Post-traumatic epilepsy (PTE) is the near-ideal condition to evaluate antiepileptogenic therapies, as it one of the most common forms of acquired epilepsy and there is a relatively short latency period during which epileptogenesis occurs prior to onset of recurrent seizures. Feasible epilepsy prevention clinical trials require a better understanding of the process of epileptogenesis. Both TBI and epilepsy are disorders of abnormal, distributed structural brain networks, but the relationship between TBI, structural network abnormalities, and epileptogenesis represents a critical knowledge gap in the field. Alterations in structural network topology affect recovery and cognitive functioning after TBI. In temporal lobe epilepsy, structural network measures delineate the localization and extent of epileptogenic foci. The widespread abnormalities in the white matter scaffolding of brain networks represent a shared pathophysiologic mechanism between TBI and epilepsy and warrant investigation as a biomarker for epilepsy prediction after TBI.

Objectives/Hypotheses: The overall objectives of this proposal are to assess alterations in structural network measurements during the subacute and chronic periods after TBI and use these measurements as features to develop predictive models to classify patients who will develop PTE. The central hypotheses are: (1) individuals who develop PTE have abnormal structural network profiles distinct from those seen with TBI of comparable injury severity without epilepsy, and (2) structural network measures can accurately classify whether a patient will develop PTE.

Specific Aims:

Aim 2: To characterize structural network measures inferred from diffusion MRI obtained in the subacute and chronic periods after TBI in patients with and without PTE. Specifically, we will examine differences in network measures both cross-sectionally at each time point as well as the difference in the longitudinal profile between patients with and without PTE.

Aim 3: To develop a robust machine learning framework to accurately classify patients who will develop PTE. Specifically, we will utilize structural network measures from the subacute and chronic periods after injury alongside traditional clinical risk factors for PTE to train models to accurately classify whether a patient will develop PTE.

Research Strategy: This study will utilize data from one of the largest, well-phenotyped samples of patients with TBI enrolled in the Transforming Research and Clinical Knowledge in Traumatic Brain Injury (TRACK-TBI) Epileptogenesis Project (TRACK-TBI EPI) where PTE diagnosis will be determined by an epileptologist and confirmed by centralized adjudication. We will utilize linear mixed effects models to compare structural network measures from diffusion MRI acquired at 2 weeks and 6 months after injury between those with and without PTE. We will use these network abnormalities as features alongside traditional clinical risk factors for PTE to train robust machine learning models to accurately classify whether a patient will develop PTE.

Innovation and Impact: The proposed research is directly aligned with the ERP’s mission and vision and will use an innovative method of analysis of diffusion MRI to determine if individuals with TBI are at risk of developing PTE. The proposed work has the potential to generate innovative deliverables that will have high impact on the millions of active-duty Service Members, Veterans, and civilians with TBI as well as those with or at risk for all forms of epilepsy. By completion of the proposed aims, we expect to improve our understanding of epileptogenesis, which will inform the design of preclinical models that replicate human epileptogenesis. Furthermore, we expect to identify a structural connectomic signature of post-traumatic epileptogenesis that can be used as a noninvasive, predictive biomarker of PTE that would pave the way for feasible clinical trials of antiepileptogenic therapies.