Abstracts

Rationale: Some patients with epilepsy have auras which allow them find safety or take a rescue medication. We propose to provide patients with a warning signal using their implanted responsive neurostimulator (RNS). This device is implanted in patients with intractable focal epilepsy not amenable to surgical cure by resection. It is programmed to detect seizure onset patterns and deliver bursts of stimulation in response to these detections. A known side effect is that some programming settings, particularly at high charge densities, can induce a clinical response. The RNS can provide up to five sequential stimuli for each detected “episode”. One of the later stimulations can be customized to provide the patient with a warning signal.  One patient has used this system successfully for several years. Methods: Timestamps of therapies that reach each level (1 through 5) were obtained from interrogation reports in the online patient data management system (PDMS). Missing times were censored. A seizure surrogate was identified from recorded electrocorticograms (usually a “long episode”, sometimes a saturation). The time from each therapy to the next seizure surrogate was calculated to determine if the frequency of each therapy level increased prior to a seizure. During a clinic visit for RNS programming, a search protocol was employed to find stimulation settings that could be sensed by the patient in a comfortable manner; we then narrowed the stimulation to the smallest set of electrode contacts. These parameters were programmed into a later stimulation (most commonly therapy 5 of 5) based on the PDMS data analysis. Results: Review of data from selected patients showed an increase in reaching later (typically 4^th or 5^th) stimulations (i.e., having longer episodes with ongoing detections) prior to a seizure surrogate. The 5^th therapy was typically reached in less than 2% of all episodes. In the first 5 patients tested, we could reliably induce a sensory response in every patient that they felt they could tolerate on days when they have more seizure activity. With occipito-temporal depth electrodes this was typically a photopsia with targeted stimulation of the most posterior electrodes using monopolar stimulation around 5 µC/cm². With lateral subdural electrodes this was typically a scalp paresthesia with stimulation of the basal-most electrodes from around 6 µC/cm². In one case with a depth electrode in the centromedian nucleus of the thalamus this was a contralateral upper limb paresthesia at 2.5 µC/cm². In one patient with bitemporal depth electrodes we induced an eyelid twitch sensation at 5 µC/cm². In this patient, one of the bursts of the 5^th therapy was programmed to provide a warning signal, which has led to administration of rescue medication and prevention of seizure clusters. Conclusions: In patients with responsive neurostimulation, longer episodes that receive trains of repeated stimulations increase in frequency prior to electrographic seizures. Stimulation parameters can be found that induce a tolerable clinical response. These parameters can be programmed into a later stimulation therapy burst to provide a seizure warning signal when the patient is reaching therapies 4 or 5. A trial is underway to assess responses and tolerability in a cohort of patients in the ambulatory setting. Funding: None