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.

Anterior temporal encephaloceles: Elusive, important, and rewarding to treat.

OBJECTIVE: To investigate the etiology and longitudinal clinical, neuropsychological, psychosocial, and surgical outcome profile of patients with medication refractory epilepsy and temporal encephaloceles with a view to highlight diagnostic clues and management strategies. METHODS: The comprehensive epilepsy program databases at two surgical epilepsy centers from January 2000 to October 2018 were reviewed for this observational study, to identify patients with encephaloceles causing temporal lobe epilepsy (TLE) and treated with surgical resection. Their clinical, radiological, neuropsychological, psychiatric, and surgical data were obtained. Body mass index (BMI) data were also reviewed due to possible correlation between idiopathic intracranial hypertension and encephaloceles. RESULTS: Thirteen patients (eight female) were identified; only three were recognized on initial magnetic resonance imaging (MRI) report. Temporal encephaloceles were identified on the left in eight patients, on the right in three patients, and bilaterally in two patients. One patient had a strong family history of encephaloceles. The median BMI for patients with seizure onset ≤20 years of age was 22.4, whereas for patients with onset >20 years median BMI was 32.6 (P = .06). Five patients underwent a focal lesionectomy, three patients had limited temporal lobectomy, and five patients had standard anterior temporal lobectomy. Median postoperative follow-up was 5.5 years. All but one patient were free of disabling seizures. Nine of ten neuropsychologically tested patients had no discernable cognitive decline postoperatively. Postoperative psychosocial adjustment features were present in four patients. SIGNIFICANCE: Genetic factors and a possible association with idiopathic intracranial hypertension (given female predominance and elevated BMI) may contribute to the causation of temporal lobe encephaloceles. It is notable that a targeted surgical approach in the management of patients with TLE associated with encephaloceles has an excellent long-term clinical and neuropsychological outcome. Subtle encephaloceles should be actively searched for in patients with drug-resistant TLE because they significantly change surgical strategy and 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.

The surgically remediable syndrome of epilepsy associated with bottom-of-sulcus dysplasia.

OBJECTIVE: To determine clinical and EEG features that might help identify patients with epilepsy harboring small, intrinsically epileptogenic, surgically treatable, bottom-of-sulcus dysplasias (BOSDs). METHODS: Retrospective review of clinical records, EEG, MRI, and histopathology in 32 patients with drug-resistant epilepsy and MRI-positive (72% 3.0 tesla), pathologically proven (type 2B cortical dysplasia) BOSDs operated at our centers during 2005-2013. RESULTS: Localization of BOSDs was frontal in 19, insula in 5, parietal in 5, and temporal in 3, on the convexity or interhemispheric surfaces. BOSDs were missed on initial MRI at our centers in 22% of patients. Patients presented with focal seizures during infancy in 9, preschool years in 15, and school years in 8 (median age 5 years). Seizures were stereotyped, predominantly nocturnal, and typically nonconvulsive, with semiology referable to the fronto-central or perisylvian regions. Seizures occurred at high frequency during active periods, but often went into prolonged remission with carbamazepine or phenytoin. Intellect was normal or borderline, except in patients with seizure onset during infancy. Scalp EEG frequently revealed localized interictal epileptiform discharges and ictal rhythms. Patients underwent lesionectomy (median age 14 years) guided by electrocorticography and MRI, with prior intracranial EEG monitoring in only one patient. Twenty-eight patients (88%) became seizure-free, and 20 discontinued antiepileptic medication (median follow-up 4.1 years). CONCLUSIONS: In patients with cryptogenic focal epilepsy, this clinical presentation and course should prompt review of or repeat MRI, looking for a BOSD in the frontal, parietal, or insula cortex. If a BOSD is identified, the patient might be considered for single-stage lesionectomy.

Long-term seizure outcome and risk factors for recurrence after extratemporal epilepsy surgery.

PURPOSE: We aimed to assess long-term seizure outcome and risk factors for seizure recurrence in a cohort of patients who have undergone extratemporal resection for management of refractory seizures. METHODS: Eighty-one patients underwent extratemporal resection at Austin Health, Melbourne, Australia (1991-2004). Seizure recurrence was any postoperative disabling seizure (complex partial seizure [CPS] ± secondary generalization). Multivariate Cox proportional hazards regression models examined potential preoperative and perioperative risk factors and the risk associated with early postoperative seizures (≤ 28 days postsurgery). The change between preoperative and postoperative seizure frequency was also measured. KEY FINDINGS: Median follow-up was 10.3 years (range 1-17.7). The probabilities of freedom from disabling seizures (on or off antiepileptic medication) were 40.7% (95% confidence interval [CI] 30-51) at 1 month, 23.5% (95% CI 15-33) at 1 year, and 14.7% (95% CI 8-23) at 5 years postoperative. Reduction of disabling seizures to at least 20% of preoperative frequency was attained by 57% of patients at 5 postoperative years. Of the preoperative/perioperative factors, focal cortical dysplasia (FCD) type 1 (hazard ratio [HR] 1.90, 95% CI 1.08-3.34, p = 0.025) and incomplete resection (HR 1.71, 95% CI 1.06-2.76, p = 0.028) were independent recurrence risks. After surgery, an early postoperative seizure was the only factor associated with higher risk (HR 4.28 [2.42-7.57], p = 0.00). SIGNIFICANCE: Distinction between subtypes of focal cortical dysplasia, which can be made using magnetic resonance imaging (MRI) criteria, may be useful for preoperative prognostication. Early seizures after surgery are not benign and may be markers of factors that contribute to seizure recurrence. Most patients achieve substantial reduction in seizure frequency. Further study of the significance of this reduction in terms of surgical "success" or otherwise is required.

A case study of trastuzumab treatment for metastatic breast cancer in pregnancy: fetal risks and management of cerebral metastases.

Trastuzumab increases survival amongst women with human epidermal growth factor receptor (HER)-2 receptor positive metastatic breast cancer, but maternal and fetal risks are associated with advanced disease and its treatment in pregnancy. We present a case of a primigravid with HER-2 positive metastatic breast cancer who received trastuzumab throughout pregnancy. She presented with cerebral metastases, requiring surgical decompression and resection. Reversible oligohydramnios developed during pregnancy. Fetal safety data on trastuzumab in pregnancy is limited, but case reports suggest a recurring pattern of (mostly reversible) oligohydramnios.

Parahippocampal epilepsy with subtle dysplasia: A cause of "imaging negative" partial epilepsy.

PURPOSE: Lesion-negative refractory partial epilepsy is a major challenge in the assessment of patients for potential surgery. Finding a potential epileptogenic lesion simplifies assessment and is associated with good outcome. Here we describe imaging features of subtle parahippocampal dysplasia in five cases that were initially assessed as having imaging-negative frontal or temporal lobe epilepsy. METHODS: We analyzed the clinical and imaging features of five patients with seizures from the parahippocampal region. RESULTS: Five patients had subtle but distinctive magnetic resonance imaging (MRI) abnormalities in the parahippocampal gyrus. This was a unilateral signal abnormality in the parahippocampal white matter extending into gray matter on heavily T(1)- and T(2)-weighted images with relative preservation of the gray-white matter boundary on T(1)-weighted volume sequences. Only one of these patients had typical electroclinical unilateral temporal lobe epilepsy (TLE); one mimicked frontal lobe epilepsy, two showed bitemporal seizures, and one had unlocalized partial seizures. All have had surgery; four are seizure-free (one has occasional auras only, follow-up 6 months to 10 years), and one has a >50% seizure reduction. Histopathologic evaluation suggested dysplastic features in the surgical specimens in all. DISCUSSION: In patients with lesion-negative partial epilepsy with frontal or temporal semiology, or in cases with apparent bitemporal seizures, subtle parahippocampal abnormalities should be carefully excluded. Recognizing the MRI findings of an abnormal parahippocampal gyrus can lead to successful surgery without invasive monitoring, despite apparently incongruent electroclinical features.

Long-term seizure outcome following surgery for dysembryoplastic neuroepithelial tumor.

OBJECT: Resection of dysembryoplastic neuroepithelial tumor (DNET) is thought to result in favorable seizure outcome, but long-term follow-up data are scarce. The authors present a review of 18 patients who underwent surgical removal of a DNET: 12 via temporal lobectomy and six via lesionectomy. METHODS: The mean long-term follow up was 10.8 years (median 10.4 years, range 7.8 to 14.8 years), and results obtained during this time period were compared with previously reported short-term (mean 2.7 years) seizure outcome data. In the current study, 66.7% patients had an Engel Class I outcome and 55.6% had an Engel Class IA outcome compared with 77.8% and 55.6%, respectively. Temporal lobectomy (Engel Class I, 83.3%; Engel Class IA, 66.7%) led to a better seizure outcome than lesionectomy (Engel Classes I and IA, 33.3%). Two patients (11.1%) required repeated operation and both had an incomplete lesionectomy initially. CONCLUSIONS: Results indicated that complete resection of a DNET leads to a favorable seizure outcome, with epilepsy cure in those who had experienced early postoperative seizure relief. Long-term seizure outcome after surgery is predictable based on the result of short-term follow up.

An unusual case of tumor-to-cavernoma metastasis. A case report and literature review.

BACKGROUND: Metastases of systemic neoplasia to preexisting intracranial mass lesions are uncommon phenomena. Tumor-to-intracranial cavernoma metastases are even more unusual and rarely reported. We describe here a case of melanoma to intracranial cavernoma metastasis. CASE DESCRIPTION: A 39-year-old woman presented after an episode of generalized tonic-clonic seizure on a background of infrequent epilepsy. She was found to have a left parieto-occipital hemorrhagic lesion on imaging studies. The lesion was surgically removed and histopathology showed a metastatic melanoma within a cavernoma. CONCLUSION: This case report represents the third recorded case of tumor-to-intracranial cavernoma metastasis and the first melanoma to intracranial cavernoma metastasis. An extensive literature review of tumor-to-intracranial tumor metastases was conducted and disclosed an increase in reporting of the uncommon phenomenon of metastasis into preexisting intracranial lesions. It should therefore be considered as a differential diagnosis in patients with history of systemic cancer who present with progression of preexistent intracranial lesions.

Temporal lobe dysembryoplastic neuroepithelial tumour: significance of discordant interictal spikes.

PURPOSE: Dysembryoplastic neuroepithelial tumours (DNET) are an important cause of refractory partial epilepsies. They usually occur within dysplastic cortex and tend to affect the temporal lobes. The EEG of these patients is characterised by slowing and/or epileptiform abnormalities with a multifocal distribution. We studied the EEG features of epilepsy patients with a temporal lobe DNET to assess the relationship of EEG abnormalities with the localisation of the tumour and the clinical features. METHODS: We retrospectively reviewed 16 patients with unilateral, temporal lobe DNET on MRI. The EEG abnormalities were classified as concordant to the lesion when the EEG discharges were confined to the ipsilateral temporal lobe or discordant when EEG discharges were found in other areas. Clinical and epilepsy characteristics were compared between patients with concordant and discordant EEG. RESULTS: Focal EEG abnormalities were found in 81% of the patients; 6/16 patients had concordant EEG abnormalities, and 7/16 patients had discordant EEG abnormalities. Epilepsy severity prior to the operation, antecedents and post-operative outcome were not different between patients with concordant or discordant EEG abnormalities. CONCLUSION: Patients with temporal lobe DNET often show EEG discharges discordant to the tumour. However, they do not appear to predict the clinical and epilepsy characteristics of these patients.

Temporal lobectomy: long-term seizure outcome, late recurrence and risks for seizure recurrence.

There is little information available relevant to long-term seizure outcome after anterior temporal lobectomy, particularly at extended postoperative periods. The aim of this study was an in-depth examination of patterns of longitudinal outcome and potential risk factors for seizure recurrence after lobectomy, utilizing a large patient sample with long follow-up. Included were 325 patients who underwent anterior temporal lobectomy between 1978 and 1998 (mean follow-up 9.6 +/- 4.2 years). Retrospective data were analysed using survival analysis and multivariate regression with Cox proportional hazard models. The probability of complete seizure freedom at 2 years post-surgery was 55.3% [95% confidence interval (CI) 50-61]; at 5 years, 47.7% (95% CI 42-53); and at 10 postoperative years it was 41% (95% CI 36-48). Patients with discrete abnormalities preoperatively (i.e. lesions and hippocampal sclerosis) had a significantly higher probability of seizure freedom than patients without obvious abnormality. The latter group had a pattern of recurrence similar to that in patients with lesions outside the area of excision. After adjustment for preoperative pathology, only the presence of preoperative secondarily generalized seizures had a significant association with recurrence [occasional preoperative generalized seizures, hazard ratio (HR) 1.6, 95% CI 1.1-2.3; frequent seizures, HR 2.0, 95% CI 1.4-2.9 compared with absence of preoperative generalized seizures]. Duration of preoperative epilepsy, age of seizure onset and age at surgery did not have an effect on outcome. Patients with two seizure-free postoperative years had a 74% (95% CI 66-81) probability of seizure freedom by 10 postoperative years. This late seizure recurrence was not associated with any identified risk factors. Specifically, patients with hippocampal sclerosis were not at higher risk. Surprisingly, complete discontinuation of anti-epileptic drugs (AEDs) after two postoperative years was not associated with an increased risk of recurrence (HR 1.03, 95% CI 0.5-2.1). This may be because selection of patients for AED discontinuation is biased towards those individuals perceived as 'low risk'. The results of this study indicate that the lack of an obvious abnormality or the presence of diffuse pathology, and preoperative secondarily generalized seizures are risk factors for recurrence after surgery. Late recurrence after initial seizure freedom is not a rare event; risk factors specific to this phenomenon are as yet unidentified.

A sheep model for the study of focal epilepsy with concurrent intracranial EEG and functional MRI.

PURPOSE: We describe a sheep model of penicillin-induced seizure activity using electroencephalography (EEG) and functional MRI (fMRI). METHODS: Ten adult sheep were used. Spikes and seizures were generated by instillation of 8,000-10,000 IU of penicillin into the right prefrontal cortex via a specially designed port. Bilateral intracranial EEG was acquired by using carbon fiber electrodes. Animals had behavioral characterization of their seizures and were then anesthetized for fMRI studies. Functional MRI was performed at 1.5 and 3 Tesla by measuring blood oxygen level-dependent (BOLD) weighted signal intensity at different times during the evolution of seizures. RESULTS: Behavioral seizures were associated with electrographic seizures. Intracranial EEG obtained in the MR scanner was of high quality. Focal spiking and seizures were seen in all animals and developed 11.3 +/- 11.2 s and 17.3 +/- 12.1 min after penicillin administration, respectively. An average of 13 +/- 4.8 seizures were seen per animal, each lasting 27.3 +/- 12.3 s. Functional MR images with little parenchymal artefact were obtained. Regional BOLD signal-intensity changes were observed during seizures at the seizure focus and ipsilateral amygdala. CONCLUSIONS: We have developed an animal model of partial epilepsy in which seizures can be reliably elicited with concurrent fMRI and intracranial EEG. During unilateral electrographic seizures, focal BOLD signal changes occurred at the seizure focus and ipsilateral amygdala, suggesting the presence of a cortico-subcortical loop. This observation illustrates the potential of the model for understanding seizure generation, spread, and possibly the consequences of repeated seizures on the brain.