BALTIMORE – A novel implantable device has demonstrated potential for predicting seizure onset in a preliminary analysis of 15 adult patients with medically refractory complex partial seizures.
The ambulatory intracranial EEG (iEEG) device, NeuroVista’s Seizure Advisory System (SAS), consists of electrodes that are implanted between the skull and the brain surface that continuously record electrical activity. The electrodes are connected by wires to a data storage device implanted in the chest. Signals are transmitted wirelessly to an external handheld device that processes the data and transmits visual and audible signals to the patient. A blue light signifies a low likelihood of seizures, white indicates medium susceptibility, and red alerts to a high likelihood of impending seizure.
"This is something we’ve never been able to do before, to predict when a seizure might happen, which potentially gives the opportunity for patients to make themselves safe, or possibly even take an acute-acting medication long-term. The uncertainty of when a seizure might occur is the most disabling part of seizures for most people. So to be able to have this sort of warning, to be able to structure day-to-day activities around [seizures] potentially, will mean a lot to people being able to control their lives," said Dr. Mark Cook, chair of medicine and director of neurosciences at St. Vincent’s Hospital, Melbourne. Dr. Cook reported the results of the study at the annual meeting of the American Epilepsy Society.
The 15 study subjects were all 18 years and older, had disabling partial and/or secondarily generalized partial seizures, and had failed therapeutic treatment with a minimum of two antiepileptic drugs. They were implanted at one of three clinical centers in Australia: Austin Health, The Royal Melbourne Hospital, and St. Vincent’s Hospital. At baseline, the patients reported experiencing 2-12 disabling partial onset seizures per month. Following intracranial implantation of the device, iEEG was collected to configure a patient-specific algorithm that identified periods of low, moderate, and high seizure likelihood.
In the data collection phase of the study, the estimated performance for the high and low likelihood advisories had to be statistically superior to a time-matched chance predictor with P value of .05 or less, and the high likelihood advisory sensitivity could not be statistically inferior to 65%. Patients who met those criteria entered an Advisory Phase where their system was configured to provide visual and audible advisories that indicate seizure likelihood.
In some instances, seizure incidence as recorded by the device was dramatically different from what the patient had reported. In one patient who initially reported having 7 seizures per month, the SAS recorded 104 per month, based on data covering 70 days. Another patient who reported 6 seizures per month actually had 80 per month, based on 126 days of data. Other patients overestimated their seizures, with one patient who reported having 3 seizures per month experiencing just 0.6 seizures per month, based on 209 days of data.
"Some patients overestimate the number of their seizures they’re having, but the amount by which [others] underestimate their seizures is very dramatic, sometimes by a factor of 10," Dr. Cook commented.
In the data collection (training) phase of the study, the sensitivity of the red advisory among 11 patients who completed the phase ranged from 0.65 to 1.00, with 10 of those patients meeting the performance criteria for the seizure advisories. Among 6 patients who have completed the subsequent 4-month advisory (prospective) study phase, red advisory sensitivity ranged from 0.56 to 1.00. In both study phases, there were no seizures during the low advisory (100% negative predictive value). The other four patients are still in the data collection phase.
The safety profile is consistent with published literature for strip electrodes and subclavicular implants, he said.
NeuroVista is continuing to study the clinical utility of the SAS, a company spokesman said.
Dr. Cook stated that he had no financial disclosures. The study was funded by NeuroVista, and four of the coinvestigators are company employees.