Expert and deep learning model identification of iEEG seizures and seizure onset times.

Hundreds of 90-s iEEG records are typically captured from each NeuroPace RNS System patient between clinic visits. While these records provide invaluable information about the patient's electrographic seizure and interictal activity patterns, manually classifying them into electrographic seizure/non-seizure activity, and manually identifying the seizure onset channels and times is an extremely time-consuming process. A convolutional neural network based Electrographic Seizure Classifier (ESC) model was developed in an earlier study. In this study, the classification model is tested against iEEG annotations provided by three expert reviewers board certified in epilepsy. The three experts individually annotated 3,874 iEEG channels from 36, 29, and 35 patients with leads in the mesiotemporal (MTL), neocortical (NEO), and MTL + NEO regions, respectively. The ESC model's seizure/non-seizure classification scores agreed with the three reviewers at 88.7%, 89.6%, and 84.3% which was similar to how reviewers agreed with each other (92.9%-86.4%). On iEEG channels with all 3 experts in agreement (83.2%), the ESC model had an agreement score of 93.2%. Additionally, the ESC model's certainty scores reflected combined reviewer certainty scores. When 0, 1, 2 and 3 (out of 3) reviewers annotated iEEG channels as electrographic seizures, the ESC model's seizure certainty scores were in the range: [0.12-0.19], [0.32-0.42], [0.61-0.70], and [0.92-0.95] respectively. The ESC model was used as a starting-point model for training a second Seizure Onset Detection (SOD) model. For this task, seizure onset times were manually annotated on a relatively small number of iEEG channels (4,859 from 50 patients). Experiments showed that fine-tuning the ESC models with augmented data (30,768 iEEG channels) resulted in a better validation performance (on 20% of the manually annotated data) compared to training with only the original data (3.1s vs 4.4s median absolute error). Similarly, using the ESC model weights as the starting point for fine-tuning instead of other model weight initialization methods provided significant advantage in SOD model validation performance (3.1s vs 4.7s and 3.5s median absolute error). Finally, on iEEG channels where three expert annotations of seizure onset times were within 1.5 s, the SOD model's seizure onset time prediction was within 1.7 s of expert annotation.

Long-term treatment with responsive brain stimulation in adults with refractory partial seizures.

OBJECTIVE: The long-term efficacy and safety of responsive direct neurostimulation was assessed in adults with medically refractory partial onset seizures. METHODS: All participants were treated with a cranially implanted responsive neurostimulator that delivers stimulation to 1 or 2 seizure foci via chronically implanted electrodes when specific electrocorticographic patterns are detected (RNS System). Participants had completed a 2-year primarily open-label safety study (n = 65) or a 2-year randomized blinded controlled safety and efficacy study (n = 191); 230 participants transitioned into an ongoing 7-year study to assess safety and efficacy. RESULTS: The average participant was 34 (±11.4) years old with epilepsy for 19.6 (±11.4) years. The median preimplant frequency of disabling partial or generalized tonic-clonic seizures was 10.2 seizures a month. The median percent seizure reduction in the randomized blinded controlled trial was 44% at 1 year and 53% at 2 years (p < 0.0001, generalized estimating equation) and ranged from 48% to 66% over postimplant years 3 through 6 in the long-term study. Improvements in quality of life were maintained (p < 0.05). The most common serious device-related adverse events over the mean 5.4 years of follow-up were implant site infection (9.0%) involving soft tissue and neurostimulator explantation (4.7%). CONCLUSIONS: The RNS System is the first direct brain responsive neurostimulator. Acute and sustained efficacy and safety were demonstrated in adults with medically refractory partial onset seizures arising from 1 or 2 foci over a mean follow-up of 5.4 years. This experience supports the RNS System as a treatment option for refractory partial seizures. CLASSIFICATION OF EVIDENCE: This study provides Class IV evidence that for adults with medically refractory partial onset seizures, responsive direct cortical stimulation reduces seizures and improves quality of life over a mean follow-up of 5.4 years.

Two-year seizure reduction in adults with medically intractable partial onset epilepsy treated with responsive neurostimulation: final results of the RNS System Pivotal trial.

OBJECTIVE: To demonstrate the safety and effectiveness of responsive stimulation at the seizure focus as an adjunctive therapy to reduce the frequency of seizures in adults with medically intractable partial onset seizures arising from one or two seizure foci. METHODS: Randomized multicenter double-blinded controlled trial of responsive focal cortical stimulation (RNS System). Subjects with medically intractable partial onset seizures from one or two foci were implanted, and 1 month postimplant were randomized 1:1 to active or sham stimulation. After the fifth postimplant month, all subjects received responsive stimulation in an open label period (OLP) to complete 2 years of postimplant follow-up. RESULTS: All 191 subjects were randomized. The percent change in seizures at the end of the blinded period was -37.9% in the active and -17.3% in the sham stimulation group (p = 0.012, Generalized Estimating Equations). The median percent reduction in seizures in the OLP was 44% at 1 year and 53% at 2 years, which represents a progressive and significant improvement with time (p < 0.0001). The serious adverse event rate was not different between subjects receiving active and sham stimulation. Adverse events were consistent with the known risks of an implanted medical device, seizures, and of other epilepsy treatments. There were no adverse effects on neuropsychological function or mood. SIGNIFICANCE: Responsive stimulation to the seizure focus reduced the frequency of partial-onset seizures acutely, showed improving seizure reduction over time, was well tolerated, and was acceptably safe. The RNS System provides an additional treatment option for patients with medically intractable partial-onset seizures.

Bone mass and turnover in women with epilepsy on antiepileptic drug monotherapy.

Antiepileptic drugs, particularly cytochrome P450 enzyme inducers, are associated with disorders of bone metabolism. We studied premenopausal women with epilepsy receiving antiepileptic drug monotherapy (phenytoin, carbamazepine, valproate, and lamotrigine). Subjects completed exercise and nutrition questionnaires and bone mineral density studies. Serum was analyzed for indices of bone metabolism including calcium, 25-hydroxyvitamin D, parathyroid hormone, insulin growth factor I, insulin binding protein III, and bone formation markers, bone-specific alkaline phosphatase, and osteocalcin. Urine was analyzed for cross-linked N-telopeptide of type I collagen, a bone resorption marker. Calcium concentrations were significantly less in subjects receiving carbamazepine, phenytoin, and valproate than in those receiving lamotrigine (p = 0.008). Insulin growth factor-I was significantly reduced in subjects receiving phenytoin compared with those receiving lamotrigine (p = 0.017). Subjects receiving phenytoin had significantly greater levels of bone-specific alkaline phosphatase (p = 0.007). Our results demonstrate that phenytoin is associated with changes in bone metabolism and increased bone turnover. The lower calcium concentrations in subjects taking carbamazepine or valproate compared with those taking other antiepileptic drugs suggest that these antiepileptic drugs may have long-term effects. Subjects receiving lamotrigine had no significant reductions in calcium or increases in markers of bone turnover, suggesting this agent is less likely to have long-term adverse effects on bone.

Predictors of ovulatory failure in women with epilepsy.

Women with epilepsy (WWE) are at increased risk for reproductive disorders. This study was designed to evaluate whether WWE are more likely to have anovulatory cycles and to assess the relative association of the epilepsy syndrome category and antiepileptic drugs (AEDs) to ovulatory dysfunction. Subjects included women aged 18 to 40 years not receiving hormones. Women without epilepsy (23 controls) and women with localization-related epilepsy (LRE, n = 59) or idiopathic (primary) generalized epilepsy (IGE, n = 35) receiving either a cytochrome P450 enzyme (cP450) inducing AED (carbamazepine, phenytoin, and phenobarbital), a cP450 inhibiting AED (valproate), or an AED that does not alter cP450 enzymes (lamotrigine and gabapentin) in monotherapy for 6 months or more were followed for three menstrual cycles. A transvaginal ovarian ultrasound was obtained. Endocrine and metabolic variables were measured and luteinizing hormone sampled over 8 hours on days 2 to 5 of one cycle. Anovulatory cycles occurred in 10.9% of cycles in controls, 14.3% of cycles with LRE, and 27.1% of cycles with IGE. Of women using valproate currently or within the preceding 3 years, 38.1% had at least one anovulatory cycle in contrast with 10.7% of women not using valproate within the preceding 3 years. Predictors of ovulatory failure included IGE syndrome, use of valproate currently or within 3 years, high free testosterone, and fewer numbers of luteinizing hormone pulses, but not polycystic-appearing ovaries. WWE are more likely to experience anovulatory menstrual cycles and the effects of epilepsy syndrome, and AED therapy may be additive. Women with IGE receiving valproate were at highest risk for anovulatory cycles, polycystic-appearing ovaries, elevated body mass index, and hyperandrogynism. WWE with anovulatory cycles may have no other signs of reproductive dysfunction. Therefore, clinicians must be alert to this potential complication of epilepsy.