Authors :
Marco Malaga Julca, MD – Baylor College of Medicine
Gina Kayal, MD – Baylor College of Medicine
Presenting Author: Christopher Bi, BS – Baylor College of Medicine
Anthony Allam, BS – Baylor Colloge of Medicne
Alisa Lu, BS – Baylor College of Medicine
Sophie Li, BS – Baylor College of Medicine
Zulfi Haneef, MBBS MD – Baylor College of Medicine
Rationale:
It is challenging to obtain seizure counts in patients to assess efficacy of treatment, or disease course over time due to the recognized unreliability of patient-reported seizures. Long Episodes (LE) in Responsive Neurostimulation System (RNS) recordings identify prolonged epileptiform activity and are typically consistent with seizures, and may be a surrogate for patient-reported seizures.
This study evaluated the relationship between RNS-detected LEs and patient-reported clinical seizures in patients with RNS implants for a 2-year period.
Methods:
We included patients with RNS hippocampal depth electrodes for the management of temporal lobe epilepsy. Patient-reported seizure counts were retrospectively obtained through chart review. Patients were classified as responders if there was a decrease of 50% in seizure frequency from baseline. Long electrographic events (LE frequency captured by the RNS device was obtained from the manufacturer's Patient Data Management System (Fig 1). The frequency of events was transformed logarithmically before analysis to normalize the distribution and minimize the influence of extreme values. We used a negative binomial model to LE rate was predictive of patient-reported seizures in all patients and in responders only.
Results:
The study included 23 patients with hippocampal depth electrodes (13 responders, 10 non-responders). Clinical seizure average baseline rate was 7.4±8.7, and LE average was 209±301. Following treatment, average seizure reduction (Fig 1) was 36.%± 73.5, (Responders: 86.7%±16.2 ; Non-responders -28.4%±67.5) and average LE reduction was 49.5% ±113.1 (Responders: 42.4%±137.4, Non-responders: 58.6%±76.7). The log of the LE rate was not a significant predictor of seizure count when accounting for within-time variation when evaluating all patients (IRR 1.10 95%CI 1.00;1.21, p=0.059) (Fig 2), and when evaluating responders only (IRR 1.11 95%CI 0.96;1.28, p=0.177).
Conclusions:
In patients with temporal lobe epilepsy with RNS depth electrodes followed for 2 years, the rate of device-detected electrocorticographic events was did not show an association with patient-reported seizure frequency, and may be a poor surrogate for clinical response. Ongoing analysis will examine if device-detected seizure counts forecast future worsening of clinical seizure counts.
Funding: ZH was supported by DoD grant Award # HT94252410355.