Alexa I. Canady

Multimodality imaging for improved detection of epileptogenic foci in tuberous sclerosis complex

Objective: Using interictal α-[11C]methyl-l-tryptophan ([11C]AMT) PET scan, the authors have undertaken a quantitative analysis of all tubers visible on MRI or 2-deoxy-2-[18F]fluoro-d-glucose ([18F]FDG) PET, to determine the relationship between [11C]AMT uptake and epileptic activity on EEG. Background: Tuberous sclerosis complex (TSC) is an autosomal dominant disorder, often associated with cortical tubers and intractable epilepsy. The authors have shown previously that [11C]AMT PET scans show high tracer uptake in some epileptogenic tubers and low uptake in the remaining tubers. Methods: Eighteen children, age 7 months to 16 years, were studied. Patients underwent video-EEG monitoring, PET scans of [11C]AMT and [18F]FDG, and T2-weighted or fluid-attenuated inversion recovery (FLAIR) MRI. [11C]AMT uptake values were measured in 258 cortical tubers delineated with coregistered MRI or [18F]FDG scans. Uptake ratios were calculated between the [11C]AMT uptake in tubers and those for normal cortex (tuber/normal cortex). Using the region of epileptiform activity, the authors performed receiver operator characteristics (ROC) analysis and determined the optimal uptake ratio for detecting presumed epileptogenic tubers. Results: Tuber uptake ratios ranged from 0.6 to 2.0. Tuber uptake ratios in the epileptic lobes were higher than those in the nonepileptic lobes (p < 0.0001). All 15 patients with focal seizure activity showed one or more lesions with uptake ratio above 0.98 in the epileptic lobe. ROC analysis showed that a tuber uptake ratio of 0.98 resulted in a specificity of 0.91. Conclusions: Cortical tubers with [11C]AMT uptake greater than or equal to normal cortex are significantly related to epileptiform activity in that lobe. Together, interictal [11C]AMT PET and FLAIR MRI improve the detection of potentially epileptogenic tubers in patients with TSC being evaluated for epilepsy surgery.

Intracranial EEG versus flumazenil and glucose PET in children with extratemporal lobe epilepsy.

Objective: To compare abnormalities determined in 2-deoxy-2-[18F]fluoro-d-glucose (FDG) and [11C]flumazenil (FMZ) PET images with intracranial EEG data in patients with extratemporal lobe epilepsy. Background: Although PET studies with FDG and FMZ are being used clinically to localize epileptogenic regions in patients with refractory epilepsy, the electrophysiologic significance of the identified PET abnormalities remains poorly understood. Methods: We studied 10 patients, mostly children (4 boys, 6 girls, aged 2 to 19 years; mean age, 11 years), who underwent FDG and FMZ PET scans, intracranial EEG monitoring, and cortical resection for intractable epilepsy. EEG electrode positions relative to the brain surface were determined from MRI image volumes. Cortical areas of abnormal glucose metabolism or FMZ binding were determined objectively based on asymmetry measures derived from homotopic cortical areas at three asymmetry thresholds. PET data were then coregistered with the MRI and overlaid on the MRI surface. A receiver operating characteristics (ROC) analysis was performed to determine the specificity and sensitivity of PET-defined abnormalities against the gold standard of intracranial EEG data. Results: FMZ PET detected at least part of the seizure onset zone in all subjects, whereas FDG PET failed to detect the seizure onset region in two of 10 patients. The area under the ROC curves was higher for FMZ than FDG PET for both seizure onset (p = 0.01) and frequent interictal spiking (p = 0.04). Both FMZ and FDG PET showed poor performance for detection of rapid seizure spread (area under the ROC curve not significantly different from 0.5). Conclusions: [11C]flumazenil (FMZ) PET is significantly more sensitive than 2-deoxy-2-[18F]fluoro-d-glucose (FDG) PET for the detection of cortical regions of seizure onset and frequent spiking in patients with extratemporal lobe epilepsy, whereas both FDG and FMZ PET show low sensitivity in the detection of cortical areas of rapid seizure spread. The application of PET, in particular FMZ PET, in guiding subdural electrode placement in refractory extratemporal lobe epilepsy will enhance coverage of the epileptogenic zone.

Relationship of flumazenil and glucose PET abnormalities to neocortical epilepsy surgery outcome

Background: Cortical areas showing abnormal glucose metabolism and [11C]flumazenil (FMZ) binding are commonly seen on PET scans of patients with intractable partial epilepsy, but it is unclear whether these must be totally resected to achieve seizure control. Objective: To analyze whether the extent of cortex showing 2-deoxy-2-[18F]fluoro-d-glucose (FDG) or FMZ PET abnormalities correlates with the outcome of resective epilepsy surgery. Methods: Cortical FDG and FMZ PET abnormalities in 15 young patients (mean age, 12.2 ± 7.0 years) with intractable partial epilepsy of neocortical origin were marked as regions with abnormal asymmetry using an objective semiautomated software package. These marked regions were then projected and measured on the brain surface reconstructed from the coregistered high-resolution MRI. Following cortical resection, the size of nonresected cortex with preoperative PET abnormalities was also measured (calculated separately for marked areas in the lobe of seizure onset as defined by long-term video EEG monitoring, and in remote cortical areas). Extent of preoperative PET abnormalities and postoperative nonresected cortex abnormalities on PET were correlated with outcome scores. Results: Large preoperative FMZ PET abnormalities were associated with poor outcome (r = 0.57; p = 0.025). Larger areas of nonresected cortex with preoperative FMZ PET abnormalities in the lobe of seizure onset were also associated with worse outcome in the whole group (r = 0.66; p = 0.007) as well as in patients with extratemporal resection (r = 0.73; p = 0.007), and in those with no lesion on MRI (r = 0.60; p = 0.049). Patients with seizure-free outcome had significantly smaller nonresected cortex with preoperative FMZ PET abnormalities than those who continued to have seizures (p = 0.022). No significant correlations between nonresected FDG PET abnormalities and surgical outcome were found. Conclusions: Extensive cortical abnormalities on FMZ PET predict poor outcome in neocortical epilepsy surgery. Resection of FMZ abnormalities in the lobe of seizure onset is associated with excellent outcome even in the absence of a structural lesion. In contrast, although FDG PET abnormalities regionalized the epileptogenic area, their size was not related to the extent of epileptogenic tissue to be removed.

Limitations of the Glasgow Coma Scale in predicting outcome in children with traumatic brain injury

OBJECTIVE To study the hypothesis that, in the absence of an ischemic-hypoxic state, children with severe traumatic brain injury and with unfavorable Glasgow Coma Scale scores may have good recovery. DESIGN Retrospective, observational, cross-sectional study with factorial design. SETTING Inpatient population in a university hospital. PATIENTS Seventy-nine children with traumatic brain injury admitted to the intensive care unit. INTERVENTIONS All patients received close monitoring and strict control of intracranial pressure (less than 20 mm Hg) and cerebral perfusion pressure (greater than 60 mm Hg). MEASUREMENTS AND RESULTS Admission Glasgow Coma Scale score, survival, need for cardiopulmonary resuscitation, presence of shock, peak intracranial pressure, duration of coma, Glasgow Outcome Scale score, and the results of neuropsychologic tests were analyzed. Of 79 children, 70 (89%) survived. Although the mortality rate was higher among patients with Glasgow Coma Scale scores of 3 to 5, 14 (64%) of 22 of these children survived. Nonsurvivors had a significantly higher incidence of shock and need for cardiopulmonary resuscitation. Except for two patients who had prolonged hypoxemia, all children, including those with Glasgow Coma Scale scores of 3 to 5, had a satisfactory outcome (Glasgow Outcome Scale scores of 4 or 5). Neuropsychologic outcome was not significantly different in the survivors with Glasgow Coma Scale scores of 3 to 5 and those with Glasgow Coma Scale scores of 6 or more. CONCLUSIONS A low Glasgow Coma Scale score does not always accurately predict the outcome of severe traumatic brain injury; in the absence of hypoxic-ischemic injury, children with traumatic brain injury and Glasgow Coma Scale scores of 3 to 5 can recover independent function.

Volumetric MRI, pathological, and neuropsychological progression in hippocampal sclerosis

Objective: To examine the relationships between age at onset and duration of seizure disorder with severity of hippocampal sclerosis (HS) and cognitive functioning in patients with HS and unilateral temporal lobe epilepsy. Methods: Twenty-six subjects had left temporal lobe seizure onset; 20 had right temporal onset. Measures were age at seizure onset, duration of seizure disorder divided by age (seizure duration), history of febrile convulsion (FC), ratio of the smaller hippocampal volume to the larger (HF) as determined by volumetric MRI, and pathologic HS grade. Results: Results showed that pathologic HS grade and HF were positively related to seizure duration, and negatively related to seizure onset. When subjects were divided into onset prior to age 10 versus later, subjects with earlier onset had higher mean pathologic HS grade and smaller (more asymmetric) mean HF. When subjects were divided into seizure duration <0.5 (i.e., less than half current lifetime) vs greater, subjects with seizure duration ≥0.5 had higher mean pathologic HS grade and lower mean HF. There was also evidence for earlier age at seizure onset and longer seizure duration being associated with worse performance on neuropsychological measures. FC was not related to either seizure duration or age at seizure onset, but patients with a history of FC showed higher pathologic HS grade and lower HF. A history of FC was not related to cognitive functioning. Conclusions: Unilateral HS patients with earlier seizure onset and longer duration of epilepsy have more severe HS and greater hippocampal volume asymmetry. This suggests that HS may be a progressive disorder with risk for cognitive dysfunction.

Outcome after posthemorrhagic ventriculomegaly in comparison with mild hemorrhage without ventriculomegaly

The neurodevelopmental sequelae in 33 low birth weight neonates with moderate or severe hemorrhage and ventriculomegaly (VM group) and in 39 neonates with mild hemorrhage only (non-VM group) were evaluated prospectively. Both groups were comparable in birth weight, gestational age, and socioeconomic status. Ventriculoperitoneal shunts were inserted in 23 of the 33 VM group infants at a mean age of 26 days. Eighty-two shunt revisions were performed, for obstruction (71 revisions) or infection (11 revisions), in 18 of the 23 children. At a mean age of 50 months, 19 of 33 children in the VM group had sequelae; 14 children had moderate or severe neurologic deficits, and 5 children had mild sequelae. In the non-VM group, only 3 of 39 children had deficits, all of which were mild (p less than 0.05). In the VM group, 19 of 33 children had mental developmental delay in comparison with 8 of 39 in the non-VM group (p less than 0.05), and 17 of 33 children in the VM group had motor developmental delay in comparison with 5 of 39 in the non-VM group (p less than 0.01). Within the VM group, the number of children with neurodevelopmental sequelae did not differ significantly among the 23 children with shunts, in comparison with the 10 who did not require shunting. Among the children with shunts, a higher incidence of sequelae occurred when lack of ventricular decompression was noted immediately after shunt insertion (p less than 0.005) and when shunt infections occurred (p less than 0.01). The most important predictor of mental and motor outcome in the group with shunts was lack of ventricular decompression immediately after shunt insertion. We speculate that, in some infants, loss of brain tissue, cerebral atrophy, or both may occur before insertion of the ventriculoperitoneal shunt, even when the shunt is inserted early.

In an Infant with Previously Undiagnosed Hemophilia A Treatment and Peri-operative Considerations

We report a 3-month-old infant who became paraplegic from an epidural hematoma caused by a diagnostic lumbar puncture for work-up of sepsis. The differential diagnosis of the cause of paraplegia was epidural hematoma formation versus spinal abscess. Hemophilia A was diagnosed when coagulation studies were discovered to be abnormal, and non-contrast CT scan revealed an epidural mass with spinal cord displacement. The coagulopathy was rapidly corrected preoperatively with an infusion of cryoprecipitate. A medially limited bilateral T8-L4 laminectomy allowed complete evacuation of the hematoma with maximum preservation of normal bone tissue, but no clinical improvement resulted. Coagulopathy should be highly suspect in an infant who becomes paraplegic after lumbar puncture. The coagulopathy may be rapidly corrected with deficient factor replacement, allowing major spinal surgery to be performed safely.We report a 3-month-old infant who became paraplegic from an epidural hematoma caused by a diagnostic lumbar puncture for work-up of sepsis. The differential diagnosis of the cause of paraplegia was epidural hematoma formation versus spinal abscess. Hemophilia A was diagnosed when coagulation studies were discovered to be abnormal, and non-contrast CT scan revealed an epidural mass with spinal cord displacement. The coagulopathy was rapidly corrected preoperatively with an infusion of cryoprecipitate. A medially limited bilateral T8-L4 laminectomy allowed complete evacuation of the hematoma with maximum preservation of normal bone tissue, but no clinical improvement resulted. Coagulopathy should be highly suspect in an infant who becomes paraplegic after lumbar puncture. The coagulopathy may be rapidly corrected with deficient factor replacement, allowing major spinal surgery to be performed safely.

Altered in vitro and in vivo flumazenil binding in human epileptogenic neocortex

In vitro and in vivo parameters of flumazenil (FMZ) binding were measured in spiking and nonspiking neocortex identified by intraoperative elcctrocorticography in epileptic patients who underwent cortical resection for seizure control. In vitro measures of receptor affinity (KD), number (Bmax) and laminar distribution for [3H]-FMZ binding in the epileptic focus (n = 38) were compared to nonspiking cortex from a subgroup of the patients (n = 12) and to tissue obtained from trauma patients (n = 5). The in vitro binding parameters were compared to in vivo [11C]-FMZ binding measured with positron emission tomography (PET) (n = 19). The Bmax was higher in the 38 spiking tissues as compared to the 12 nonspiking tissues (P = .012). Paired comparison of spiking versus nonspiking binding in the 12 patients from whom nonspiking tissue was available showed increases in both KD (P = .037) and Bmax (P = .0047) in spiking cortex. A positive correlation was found between KD and Bmax values for 38 patients (r = 0.55, P < .0001), the magnitude of the KD increase being twice that of the Bmax increase. In addition, there was a significant correlation between the asymmetry indices of the in vivo FMZ binding on PET and in vitro KD of spiking cortex (n = 19, r = 0.52, P = .02). The laminar distribution of [3H]-FMZ showed increased FMZ binding in cortical layers V-VI in spiking cortex compared to nonspiking and control cortex. The increased receptor number in spiking cortical layers V-VI may be a compensatory mechanism to decreased GABAergic input. The increased Bmax in spiking cortex was accompanied by a larger decrease in the affinity of FMZ for the receptor suggesting that decreased FMZ binding in the epileptic focus measured with PET is due to a decrease in the affinity of the tracer for the receptor.

Permanent I-125 brain stem implants in children

Abstract Between 1988 and 1997, 28 children have had iodine-125 implants for CNS tumors performed in our institution. Ten had stereotactic implantation in the brain stem region, and nine had the diagnosis of brain stem glioma (8 diffuse pontine, 1 midbrain tumor). Their ages ranged from 1.8 to 12 years. All patients had histological confirmation of malignancy (7 high-grade glioma, 2 low-grade glioma, 1 PNET). Diffuse pontine glioma patients received external beam radiation (50 Gy) followed by a fractionated stereotactic boost of 3 Gy×4 fractions. After 4–6 weeks, patients were reevaluated for stereotactic interstitial I-125 therapy. The planned implant dose was 82.9 Gy to the enhancing tumor (4 cGy per h). Preliminary results indicated that no surgical complications were associated with the catheter placement. Four patients have died (7–9 months from diagnosis) and four patients remain alive (5–38 months from diagnosis, median 10 months). Two autopsies confirmed the presence of progressive glioblastoma multiforme and intralesional necrosis. In one patient who received an implant alone for midbrain LGA, necrosis without tumor was found on biopsy after 36 months. He was successfully treated with hyperbaric oxygen therapy. The implementation of permanent I-125 implants appears to have a role in the management of pediatric CNS malignancy. This study confirms the results of previous reports regarding the safety of stereotactic interstitial brachytherapy in the brain stem. Tumor control for patients with high-grade brain stem glioma remains poor even with high focal radiation doses.

Evaluation of Shunt Malfunction Using Shunt Site Reservoir

Objective: To determine the usefulness of a separate reservoir placed at the site of the shunt in evaluation of shunt malfunction. Methods and Materials: A ventricular catheter was placed alongside the proximal catheter of the shunt and connected to a subgaleal reservoir in 17 patients, in 9 a double-lumen catheter with integrated reservoir and in 13 patients a dual catheter with a double-port reservoir was used. At presentation of suspected shunt malfunction, a standard shunt function evaluation using shunt tap, CT scan or shunt injection was performed, and subsequently, the pressure from the tap of the reservoir was obtained. Results: Thirty-three patients presented with symptoms of malfunction at an interval of 2.3 ± 3 months (range 2–429 days). The pretest probability of shunt malfunction in this population was 73%. Posttest probability of shunt malfunction was 82.5% with standard evaluation and improved to 100% by the separate reservoir tap pressure measurement. In 4 patients in whom the shunt tap was dry, shunt infection was diagnosed prior to revision using CSF obtained at the reservoir tap. In 5 patients with proximal malfunction and bradycardia, the reservoir tap allowed early ventricular decompression. Conclusion: This study shows that a reservoir placed at the site of the shunt remains patient even when the shunt malfunctions, suggesting that flow rather than catheter position is important in proximal malfunction. It is superior to shunt tap for detection of shunt malfunction and infection, and it allows early ventricular decompression in a sick patient awaiting surgery for shunt revision.