PRediction of Vagal Nerve Stimulation EfficaCy In Drug-reSistant Epilepsy (PRECISE)

The patients will be recruited within cooperating epilepsy centres. The recruitment limited to cooperating centres will ensure the correct indication for VNS therapy and the adjustment of VNS parameters according to the best medical practice, which are not part of the study protocol itself, but are crucial for the study results.

The VNS will be indicated in a given patient based on a clinical decision as a standard of patient´s therapy (i.e., the indication of VNS is not a part of the proposed project). The patient will be informed about the study on a pre-implantation visit or by a phone call, during which initial eligibility for the study will be determined, and study protocol and design will be explained by research staff.

G. Assessment 0 - pre-implantation assessment Assessment 0 includes screening for eligibility criteria and obtaining baseline data on subjects participating in the study. After the signature of informed consent, the patient will be given a unique code by which all data will be labelled. The demographic information, health history, including psychiatric disease and type of epilepsy, concomitant medication (AEDs and other drugs), and seizure frequency will be collected. The seizure frequency will be analysed based on the patient diary. Standard EEG will be recorded according to the protocol described below.

Protocol for EEG recording The EEG will be recorded on the EEG recording system with a sampling frequency of 250 Hz or higher, with electrodes placed on the scalp according to 10-20 system (electrodes names as follows: Fp1, Fp2, F7, F3, Fz, F4, F8, T3, C3, Cz, C4, T4, T5, P3, Pz, P4, T6, O1, and O2). Different electrodes names are allowed, but their position must correspond with the position of previously defined electrodes. Additional electrodes are also allowed, but will not be evaluated within the study. EEGs with lower numbers or different positions of electrodes are not allowed.

The EEG is recorded in a supine position with eyes closed except for periods with eyes opening. The recorded EEG will have a duration of 20 minutes or more and must contain defined segments in the following order (Figure 1):

Segment 1 - Rest 1 - duration at least 2 min Segment 2 - Eyes opening/closing - duration at least 10 s Segment 3 - Rest 2 - duration at least 10 s; immediately after eye closure Segment 4 - Photic stimulation - stimulation frequencies: 5 Hz (10 s) - 10 Hz (10 s) - 15 Hz (10s) - 20 Hz (10 s) - 25 Hz (10 s) - 30 Hz (10 s) - 25 Hz (5 s) - 20 Hz (5 s) - 15 Hz (5 s) - 10 Hz (5 s) - 5 Hz (5 s); light intensity at least 0.7 Joule Segment 5 - Hyperventilation - duration at least 4 min (2 minutes hyperventilation by nose + 2 minutes hyperventilation by mouth) Segment 6 - Eyes opening/closing - duration at least 10 s Segment 7 - Rest 3 - duration at least 10 s, immediately after eye closure Segment 8 - Rest 4 - duration at least 2 minutes The recorded EEG will be anonymized and labelled by a unique code previously attributed to a patient. In the next step, the EEG will be sent for mathematical/statistical processing. Based on this processing, the patients´ predicted response to VNS will be determined in terms of predicted responders (seizure reduction ≥ 50%) vs. predicted non-responders (seizure reduction < 50%). Neither the patient nor the physician will be informed about the classification results.

Follow-up assessments - Assessments 1, 2 Two assessments are designed in patients´ follow-up: Assessment 1 - 1 year after stimulation initialization and Assessment 2 - 2 years after stimulation initialization.

Assessment 1, 2 The patients will be evaluated 1 and 2 years after stimulation initialization. The patient will be asked about changes in health history, including psychiatric disease, concomitant medication (AEDs and other drugs), VNS parameters setting, and seizure frequency. The seizure frequency will be collected by reviewing patients 'diary. The reported seizure frequency will be compared with pre-implantation seizure frequency, and real-life response will be calculated in terms of real-life responders (seizure reduction ≥ 50%) or real-life non-responder (seizure reduction < 50%). If the patient does not tolerate at least the minimal VNS setting (0.75 mA) because of serious adverse events, he/she will be excluded from further analysis (the patient´s data will be reported separately).

The information collected in Assessment 1 and 2, including real-life response to VNS, will be labelled by the patient´s unique code and will be sent to the coordinating centre, where it will be compared with the patient´s predicted response.

At the end of Assessment 2, the patient will be asked if he/she is interested in the predicted response to VNS. If yes, the study physician will contact the coordinating centre, where the predicted response for a given patient will be conveyed. The patient will be informed about the predicted response by a phone call or during a regular visit (not a part of the study protocol, described below).

The study is concerned with the verification of the real-life validity of our statistic classifier for the prediction of VNS response based on pre-implantation EEG. The patients based on the pre-implantation EEG will be classified based on their predicted response as predicted responders (≥50% seizure reduction) vs. predicted non-responders (<50% seizure reduction). In Assessment 1 a 2, respectively, the patients´ real-life response will be determined: real-life responders (≥50% seizure reduction) vs. real-life non-responder (< 50% seizure-reduction). The predicted and real-life responses will be statistically compared in terms of accuracy, sensitivity, and specificity.

The individual patients´ groups (predicted responders vs. predicted non-responders, real-life responders vs. real-life non-responders) will be compared based on their demographic data, health history, AEDs, and VNS parameter. A suitable statistical test will be chosen concerning the data´s characteristics (Pearson´s Chi-squared test for categorical variables, Wilcoxon-Mann-Whitney test for continuous variables).