Machine learning approaches for imaging-based prognostication of the outcome of surgery for mesial temporal lobe epilepsy.

OBJECTIVES: Around 30% of patients undergoing surgical resection for drug-resistant mesial temporal lobe epilepsy (MTLE) do not obtain seizure freedom. Success of anterior temporal lobe resection (ATLR) critically depends on the careful selection of surgical candidates, aiming at optimizing seizure freedom while minimizing postoperative morbidity. Structural MRI and FDG-PET neuroimaging are routinely used in presurgical assessment and guide the decision to proceed to surgery. In this study, we evaluate the potential of machine learning techniques applied to standard presurgical MRI and PET imaging features to provide enhanced prognostic value relative to current practice. METHODS: Eighty two patients with drug resistant MTLE were scanned with FDG-PET pre-surgery and T1-weighted MRI pre- and postsurgery. From these images the following features of interest were derived: volume of temporal lobe (TL) hypometabolism, % of extratemporal hypometabolism, presence of contralateral TL hypometabolism, presence of hippocampal sclerosis, laterality of seizure onset volume of tissue resected and % of temporal lobe hypometabolism resected. These measures were used as predictor variables in logistic regression, support vector machines, random forests and artificial neural networks. RESULTS: In the study cohort, 24 of 82 (28.3%) who underwent an ATLR for drug-resistant MTLE did not achieve Engel Class I (i.e., free of disabling seizures) outcome at a minimum of 2 years of postoperative follow-up. We found that machine learning approaches were able to predict up to 73% of the 24 ATLR surgical patients who did not achieve a Class I outcome, at the expense of incorrect prediction for up to 31% of patients who did achieve a Class I outcome. Overall accuracies ranged from 70% to 80%, with an area under the receiver operating characteristic curve (AUC) of .75-.81. We additionally found that information regarding overall extent of both total and significantly hypometabolic tissue resected was crucial to predictive performance, with AUC dropping to .59-.62 using presurgical information alone. Incorporating the laterality of seizure onset and the choice of machine learning algorithm did not significantly change predictive performance. SIGNIFICANCE: Collectively, these results indicate that "acceptable" to "good" patient-specific prognostication for drug-resistant MTLE surgery is feasible with machine learning approaches utilizing commonly collected imaging modalities, but that information on the surgical resection region is critical for optimal prognostication.

Metabolic patterns and seizure outcomes following anterior temporal lobectomy.

OBJECTIVE: We investigated the relationship between the interictal metabolic patterns, the extent of resection of 18 F-fluorodeoxyglucose positron emission tomography (18 FDG-PET) hypometabolism, and seizure outcomes in patients with unilateral drug-resistant mesial temporal lobe epilepsy (MTLE) following anterior temporal lobe (TL) resection. METHODS: Eighty-two patients with hippocampal sclerosis or normal magnetic resonance imaging (MRI) findings, concordant 18 FDG-PET hypometabolism, and at least 2 years of postoperative follow-up were included in this 2-center study. The hypometabolic regions in each patient were identified with reference to 20 healthy controls (p < 0.005). The resected TL volume and the volume of resected TL PET hypometabolism (TLH) were calculated from the pre- and postoperative MRI scans coregistered with interictal 18 FDG-PET. RESULTS: Striking differences in metabolic patterns were observed depending on the lateralization of the epileptogenic TL. The extent of the ipsilateral TLH was significantly greater in left MTLE patients (p < 0.001), whereas right MTLE patients had significantly higher rates of contralateral (CTL) TLH (p = 0.016). In right MTLE patients, CTL hypometabolism was the strongest predictor of an unfavorable seizure outcome, associated with a 5-fold increase in the likelihood of seizure recurrence (odds ratio [OR] = 4.90, 95% confidence interval [CI] = 1.07-22.39, p = 0.04). In left MTLE patients, greater extent of resection of ipsilateral TLH was associated with lower rates of seizure recurrence (p = 0.004) in univariate analysis; however, its predictive value did not reach statistical significance (OR = 0.96, 95% CI = 0.90-1.02, p = 0.19). INTERPRETATION: The difference in metabolic patterns depending on the lateralization of MTLE may represent distinct epileptic networks in patients with right versus left MTLE, and can guide preoperative counseling and surgical planning. Ann Neurol 2019; 1-10 ANN NEUROL 2019;85:241-250.

Serious Cardiac Arrhythmias Detected by Subcutaneous Long-term Cardiac Monitors in Patients With Drug-Resistant Epilepsy.

BACKGROUND AND OBJECTIVES: Epilepsy is associated with an increased risk of cardiovascular disease and premature mortality, including sudden unexpected death in epilepsy (SUDEP). Serious cardiac arrythmias might go undetected in routine epilepsy and cardiac investigations. METHODS: This prospective cohort study aimed to detect cardiac arrhythmias in patients with chronic drug-resistant epilepsy (≥5 years duration) using subcutaneous cardiac monitors for a minimum follow-up duration of 12 months. Participants with known cardiovascular disease or those with abnormal 12-lead ECGs were excluded. The device was programmed to automatically record episodes of tachycardia ≥140 beats per minute (bpm), bradycardia ≤40 bpm for ≥3 seconds, or asystole ≥3 seconds. FINDINGS: Thirty-one patients underwent subcutaneous cardiac monitoring for a median recording duration of 2.2 years (range 0.5-4.2). During this time, 28 patients (90.3%) had episodes of sustained (≥30 seconds) sinus tachycardia, 8/31 (25.8%) had sinus bradycardia, and 3 (9.7%) had asystole. Three patients (9.7%) had serious cardiac arrhythmias requiring additional cardiac interventions. Among them, 2 patients had prolonged sinus arrest and ventricular asystole (>6 seconds), leading to pacemaker insertion in one, and another patient with epileptic encephalopathy had multiple episodes of recurrent nonsustained polymorphic ventricular tachycardia and bundle branch conduction abnormalities. The time to first detection of a clinically significant cardiac arrhythmia ranged between 1.2 and 26.9 months following cardiac monitor insertion. DISCUSSION: Implantable cardiac monitors detected a high incidence of clinically significant cardiac arrhythmias in patients with chronic drug-resistant epilepsy, which may contribute to the incidence of premature mortality, including SUDEP.