Deep Phenotyping for Physiologic Biomarkers for Posttraumatic Epilepsy in Children
Abstract
Background: Post-traumatic epilepsy (PTE) is a leading cause of acquired epilepsy, occurring in up to 20% of children following severe traumatic brain injury (TBI) and representing the leading cause of epilepsy with onset in young adulthood. The mechanisms underlying acquired epileptogenesis in humans after TBI remains poorly understood, and a better understanding of the processes between initial insult and PTE development may provide treatment targets towards preventing it. Recent work has identified that after moderate-severe TBI, hemorrhagic temporal lobe injury is related to both a high incidence of early seizures and longitudinal development of PTE. Despite this finding, extensive work has suggested that while early post-traumatic seizures (PTS) are associated with an increased risk for late seizures, the increased risk for PTE is not a direct consequence of early seizures. This raises the possibility that the underlying physiologic environment immediately after TBI carries physiologic biomarkers for post-traumatic epileptogenesis. Hypothesis: We hypothesize that neuroanatomical injury variations in conjunction with high-frequency longitudinal multivariate physiologic data will enable development of models that can predict PTS and PTE with high accuracy. Specific Aims: Our goal is to use advanced multivariate modeling to further our understanding of pediatric post-traumatic epileptogenesis. We propose both a statistical and data mining approach after pediatric severe TBI to identify physiologic biomarkers predictive of PTS and PTE. Aim 1: Retrospective model development for identify physiologic biomarkers of PTS and PTE. Aim 2: Prospectively test multivariate models of PTS and PTE in pediatric TBI patients. Research Strategy: We will achieve Aim 1 by retrospectively exploring our existing clinical database of high-frequency multimodal physiologic data of children with severe TBI. To date, we have collected time-synchronized high-frequency multimodal physiologic data from over 80 children with severe TBI and have identified physiologic derangements associated with early pediatric PTS. Through the application of statistical and data mining techniques, we will extract features from physiologic data, neuroimaging and CDEs (common data elements) identified within the EMR (electronic medical record). We will develop statistical and machine learning models to predict epilepsy after TBI in children. We will also apply model-based indices of cerebral dynamics, multivariate techniques, and structural equations modeling to explore directional relationships within such models. We will achieve Aim 2 by prospectively apply models from Aim 1 to pediatric patients with severe TBI to test its validity for assessing patient risk for PTS or PTE.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Nov 19, 2019
- Source ID
- W81XWH1910514
Entities
People
- Brian Appavu
Organizations
- Phoenix Children's Hospital
- United States Army