Graduate Assistant Researcher Texas A&M University College of Medicine
Rationale: Acquired epilepsy is a highly variable condition caused by brain injury, stroke, infections and chemical neurotoxicity. However, there is no specific biomarker for predicting the risk of epilepsy in people with these risk factors. High frequency oscillations (HFOs) have been shown to increase in individuals with epilepsy and localize to epileptic tissue, but their role in epileptogenesis is still unclear. Given their characteristics, HFOs make for an attractive electrographic feature and potential biomarker of epilepsy. Post-traumatic epilepsy occurs in up to 50% of people after brain injury, but an early diagnostic risk stratification to determine who is at high risk for developing epilepsy is invaluable. Here, we sought to utilize HFOs as a potential biomarker for predicting a subject’s individual risk for PTE after brain injury. Methods: A novel algorithm for the detection of HFOs was created and used to retrospectively analyze long-term (four months) intracranial EEG recordings from mice with induced traumatic brain injuries from controlled cortical impacts. The iEEG recordings from TBI mice (n=20) were analyzed for seizure occurrence, followed by detection and characterization of HFOs by the custom algorithm. Results: Six of the twenty mice developed epilepsy with spontaneous recurrent seizures (responders) and their HFOs were compared to the remaining fourteen in the non-epileptic (non-responders) cohort. The epileptic mice had a striking increase in total number of HFOs over the course of the four-month period and a massive increase in the HFOs frequency during the latent period (first 30 days) compared to the non-epileptic mice. Additionally, the approximated primary frequency of the detected HFOs was significantly (p< 0.01) lower in epileptic mice than non-epileptic mice. Conclusions: These results show that HFOs hold great potential as an electrographic biomarker to predict the risk of developing epilepsy, especially for early risk stratification of subjects with known risk factors. Additional studies with dynamic machine learning methods are warranted for clinical translation of HFOs in epilepsy. Funding: Please list any funding that was received in support of this abstract.: Funded by DOD Award #W81XWH-16-1-0660 and TAMU X-Grant
