Authors :
Presenting Author: Ajay Deep Kachhvah, PhD – The University of Texas Health Science Center at Houston
Yash Vakilna, MS – The University of Texas Health Science Center at Houston; Omar Alamoudi, PhD – The University of Texas Health Science Center at Houston; Biomedical Engineering Program, King Abdulaziz University, Jeddah, Saudi Arabia; Surya Suresh, MBBS – The University of Texas Health Science Center at Houston; Sandipan Pati, MD – Associate Professor, Neurology, The University of Texas Health Science Center at Houston
Rationale:
Cardiorespiratory comorbidities are increasingly recognized in persons with epilepsy, highlighting the need to understand the intricate connections between the brain and heart. It is reported that cardiorespiratory dysfunction and postictal generalized EEG suppression (PGES) occur following convulsive seizures and can be a cause of sudden unexpected death in epilepsy (SUDEP). Therefore, these physiological interaction dynamics can be leveraged to advance brain-heart neuromodulatory embedded devices, such as vagus nerve stimulation, potentially treating or preventing seizures and mitigating their cardiorespiratory dysfunction. This study investigated the dynamics of brain-heart interactions during selected focal to bilateral tonic-clonic seizures (FBTC) with confirmed PGES, which is more prone to SUDEP. Our key objective was to identify unique fingerprints of thalamo-cardiac interactions during PGES by using physiological (sleep or awake) and pathological (focal seizures) states as reference points.
Methods:
It is a single-center, IRB-approved prospective study where patients with suspected TLE undergoing SEEG investigation had thalamic implantation. We selected eight patients (eight FBTC with PGES), matched against eight focal seizures, awake and sleep baselines. A bidirectional relationship of thalamus (TH), hippocampus (AHP), and cingulate (CING) with the heart (H) was derived for eight patients using convergent-cross-mapping (CCM)
from their SEEG and ECG signals recorded concurrently during post-FBTC (10 minutes), post-focal seizures (10 minutes), sleep (10 hours) and awake (10 hours) periods. The CCM was performed on 30 seconds long time-varying sliding windows extracted from SEEG and ECG signals (bandpass filtered 0.05-50 Hz) at each five seconds interval. Further, to evaluate the uncovered discriminatory power of the obtained bivariate trajectories, a classification problem was formulated based on XGBoost classifier to differentiate the PGES trajectories from post-focal, sleep, and awake ones.
Results:
The bivariate trajectories revealed more significant signatures for the H-TH interaction, effectively distinguishing the PGES state from post-focal and baseline states, than the H-AHP and H-CING ones. It is found that the bivariate H-TH information flow returned to baseline in four minutes during the FBTC recovery, however the recovery from focal seizures showed a gradual decline over nine minutes, with the trajectories failing to return to the physiological state. Moreover, the XGBoost classifier manifested its robustness by parsing the PGES state from a pool of mixed states with high accuracy.
Conclusions:
Our findings highlight the distinct characteristics of the
H-TH interaction in effectively differentiating the PGES state from focal seizures and baseline physiological states. The CCM analyses revealed rapid restoration of bidirectional information flow between the thalamus and heart during recovery from FBTC, while the recovery from focal seizures exhibited a slower and incomplete return to the physiological state.
Funding: Nothing to disclose