Abstracts

Rationale: The VNS Therapy System (Cyberonics, Inc) is programmed to deliver vagus nerve stimulation (VNS) in two modes for the treatment of epilepsy: periodic (automatic) and on-demand (manual). On-demand initiation of VNS, which requires moving a magnet over the generator, may abort or shorten seizure duration, especially if administered at onset of ictal symptoms (Morris GL Epilepsy Behav 2003;4:740-5). Unfortunately, many patients lose consciousness or are physically incapable of using the magnet at seizure onset. In order to reliably activate VNS following electrographic seizure onset, we have developed a prototype system capable of automatically activating the VNS Therapy System following real-time detection of a pre-specified abnormality in the surface EEG. The prototype’s ease of use, and sensitivity and specificity are being prospectively evaluated in a clinical study at the Beth Israel Deaconess Medical Center in Boston, Massachusetts.Methods: Study participants are long-term users of the VNS Therapy System who report no adverse effects from on-demand VNS and who are monitored by video-surface EEG (DigiTrace 1800 Plus Recorder and Digital Video System) for up to five days. After the first 24 hours of the study an electroencephalographer reviews the collected EEG and highlights ictal or interictal EEG patterns; one of these patterns is then selected as the target pattern to be detected by the computerized system (“pre-specified EEG pattern”). On subsequent days, the computerized system is allowed to automatically activate the subject’s VNS Therapy System by energizing an electromagnet that is worn over the VNS generator following real-time detection of the pre-specified EEG pattern using a patient-specific detection algorithm (Shoeb A et al Epilepsy Behav 2004;5:483-98).Results: The study is ongoing, and illustrative results from the first subject are shown here. Figure 1 shows the computerized detection of a new instance of a pre-specified abnormality and automatic activation of the vagus nerve stimulator in response to the detection; activation of the vagus nerve stimulator appears as a spiking pattern starting just after the 245 second mark on the 'VNS' channel. We confirmed that the computerized system activated the subject's vagus nerve stimulator in magnet mode by 1) reviewing the collected video and noting that the subject did not use their hand-held magnet at the time of the abnormality and 2) confirming that the duration of the delivered stimulus matched the subject's on-demand, magnet-mode stimulation duration which differed from the periodic, automatic-mode stimulation duration. Conclusions: Our results support the feasibility of implementing a real-time system that quickly triggers the VNS Therapy System in response to a pre-specified surface EEG abnormality. Further results of the prospective study will be presented at the meeting. To our knowledge, this is the first demonstration of a non-invasive, closed-loop detection-treatment system based on surface EEG.