Seizure Detection and Automatic Magnet Mode Performance Study

Subjects were admitted to the EMU and underwent standard continuous data collection of vEEG and ECG for 3 to 5 days. If a seizure occurred during the EMU stay, clinical investigators annotated the start and stop times, the type of seizure, the presumed seizure onset location, and the lobe of origin as applicable.

Following the EMU data collection phase of the trial, the de-identified, continuous electronic records (per patient) from the EMU period were provided to an independent and blinded triple review panel. This panel evaluated the EEG data and annotated seizure onset, seizure offset, and a description of seizure type. In the absence of video, seizure types could only be specified as: partial (particular type not denoted), generalized (non-absence), absence, or partial with secondary generalization.

Sensitivity is the total number of seizures detected divided by the total number of seizures during EMU stay.Data used to support sensitivity analyses included digital ECG/EEG files,corresponding M106 device downloads,and CRF data.Seizure and non-seizure EEG segments were provided to independent reviewers to confirm seizure occurrence and define EEG seizure onset times.Seizure onset times were then compared with observed M106 device detections at the detection threshold setting for AutoStim that the patient was randomized to(SDA 2;60%,SDA 4;40%,SDA 6;20%).Sensitivity is only reported if the heart rate surpassed the programmed detection threshold.Number of participants is total number of subjects who had seizures during the EMU stay.

An "Ictal tachycardia Seizure" is a seizure with Ictal Heart rate >= 100 bpm & at least 55% increase, or 35 bpm increase from baseline) Bootstrap confidence intervals using 3000 bootstrap samples. n=total number of seizures; N= number of participants

Sensitivity is defined as the total number of seizures detected divided by the total number of seizures during the EMU stay. Data used to support sensitivity analyses included digital ECG/EEG files, corresponding M106 device downloads, and CRF data. Seizure onset times were compared with modeled M106 device detections at the least sensitive setting capable of detecting the seizure based on the corresponding change in heart rate. The participants' surface ECG data collected during the trial and passed through DMSDAT, a validated bench-top simulant of the Automatic Stimulation feature, was used to produce modeled results for each threshold for AutoStim setting (1;70%, 2;60%, 3;50%, 4;40%, 5;30% and 6;20%). Number of participants is total number of subjects who experienced seizures during the EMU stay.

Bootstrap confidence intervals using 3000 bootstrap samples.

Potential false positive rate is defined as the sum across all patients of the total number of potential false positive detections divided by the sum across all patients of the appropriate monitoring time during the EMU stay. Data used to support the potential false positive rate analyses included digital ECG/EEG files retrieved from the EMU evaluation, corresponding M106 device downloads, and triple review results of EEG recordings.

The evaluated EMU monitoring time includes a daily 3 minutes stepping exercise during which patients stepped up and down on a step stool at a submaximal effort leve.

Cardiac R-wave detection was evaluated against concurrent ECG data (i.e. detailed R-wave test) collected during implant, the first titration visit, at the beginning of the EMU stay, and at the 12 month visit. R-R intervals were calculated using detected R-waves from the Implantable Pulse Generator (IPG) and from a standard ECG monitor during a pre-specified time interval. A time series 10 seconds was recorded using the IPG SyncPulse feature. Simultaneously, a corresponding time series over the same interval was recorded using a standard ECG monitor. The total number of beats accurately detected in the entire study population is reported.

Latency is defined as the time difference between SDA detection time and the annotated seizure onset time. The earliest SDA detection was considered for each seizure. Seizure onset times were compared with M106 device detections at the randomized SDA setting. Negative latencies indicate that the SDA detection preceded the seizure onset time. The median latency is presented for seizures which met the definition of ictal tachycardia as well as all seizure types and indicate the observed latency range.

Usability survey data were collected from all site personnel who used the handheld programmer to evaluate the usability of the AspireSR® VNS Therapy® System.The device usability survey contained 17 questions that measure usability on a five-point Likert scale ranging from "Extremely Difficult" (5) to "Extremely Easy" (1). Site personnel were asked to assess usability of the software features, instructions for use, training materials, and overall usability of the system at four different time points. The time points include implant/recovery and the end of EMU.

Usability was calculated as percentage of the users who found the usability of system to be "easy-2" or extremely easy-1".

Seizure frequency was calculated at 3, 6, 12, 18 and 24 month follow-up visits based on seizure diary information and compared to baseline estimates. Response rate was computed and summarized for partial seizures (SPS, CPS and CPS with 2nd GTCs) and overall seizure types as the percentage of patients that achieved ≥50% seizure reduction per month from baseline by visit.

Investigators completed the National Hospital Seizure Severity Scale (NHS3) questionnaire at screening, at the end of the EMU stay (provided a seizure occurred during the EMU stay), and at follow-up visits. Severity was evaluated by seizure type. The range of NHS3 scale is 1-27 with 1 being the least severe and 27 being the most severe.

Negative median value means improvement.

Clinical outcomes such as seizure severity, intensity and post-ictal duration were also assessed during the long-term follow-up visits (3, 6, 12, 18 and 24 months) with patient reported questionnaires (SSQ; Seizure Severity Questionnaire). The range for SSQ (all sub-scores) is 1-7 with 1 being the least severe and 7 being the most severe.

Mean SSQ scores at 3, 6, 12, 18 and 24 months were compared to baseline. A change from baseline is calculated as baseline minus follow-up visit score to correspond to the Minimally Important Change (MIC) criteria as defined in the Scoring Scheme for SSQ v2. Questionnaire. Subscale scores were averaged to compute the SSQ Total Score

Clinical outcomes including seizure duration and cessation were assessed with vEEG during EMU stay. Number of seizures treated with Automatic Stimulation during EMU were evaluated. Of these seizures, those ending during the 60 second course of Automatic Stimulation were assessed and tabulated by seizure type.

Seizure duration was calculated using historical EEG data from patients enrolled in the trial and compared to the duration of seizures that occurred during the study EMU stay. The seizure start and end times were determined via clinical observation and/or through an adjudication process with qualified EEG reviewers.

Post-ictal duration was quantified by identifying the time at which the number of EEG channels within the 95% confidence interval of relative power reaches a number that is consistent with that during the pre-seizure period. This measure represents the amount of time required following a seizure until the EEG recovers to the pre-seizure state. It is used to objectively estimate patient recovery time. Historical seizures were baseline EEG recordings measured during monitoring prior to implantation. Post-ictal duration is only reported for seizures with Post Ictal Duration (seconds) annotated and >= 20% Heart Rate Rise

Quality of life data was collected using patient-completed QOLIE-31-P surveys and compared between baseline and follow-up visits. The MIC score for each subscale defines the threshold for Minimally Important Change. If a score exceeds the MIC Score, the improvement from baseline is considered clinically significant. The range for QOLIE-31-P (all sub-scores) is 0-100 with higher scores reflecting greater well-being.Subscale scores were averaged to compute the QOLIE Total Score.

Quantitative evaluation of EEG was used to characterize the seizures that were treated with Automatic Stimulation. Intensity was evaluated by surveying the average power level from the 10-20 system EEG channel of maximum output during the course of the seizure. Intensity was only reported for seizures with intensity annotated and >= 20% Heart Rate Rise. The relative intensity was calculated by normalizing the power calculations to the pre-seizure state, and thus the reported changes are dimensionless. n= number of seizures

During a 1 hour period during the EMU stay, the VNS Therapy device was programmed to normal mode stimulation ON time 30 seconds, OFF time 5 minutes. AutoStim and Magnet Mode were programmed OFF. In this hour, 10 to 11 normal mode stimulations can be expected. Around each of these stimulations, ECG data were collected to assess potential stimulation related heart rate changes (during stimulation, after stimulation and after black-out time). A black-out time is a period after stimulation during which no seizure detections can occur, to ensure that potential stimulation related heart rate changes were not seen as ictal tachycardia that would trigger false positive detection. During the trial, the black-out time was programmed to 30 seconds. The heart rate changes for all stimulations and all patients were averaged.