Abstracts

Metabolic divergence of focal epilepsy syndromes

Abstract number : 1.227
Submission category : 5. Neuro Imaging / 5C. Functional Imaging
Year : 2016
Submission ID : 187150
Source : www.aesnet.org
Presentation date : 12/3/2016 12:00:00 AM
Published date : Nov 21, 2016, 18:00 PM

Authors :
Boris Bernhardt, Montreal Neurological Institute, McGill University; Seok-Jun Hong, Montreal Neurological Institute, McGill University, Montreal, Canada; Jean-Paul Soucy, Montreal Neurological Institute and Hospital, McGill University, Canada; PET Unit, M

Rationale: Previous studies have suggested clinical utility of 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) analysis in the study drug- resistant focal epilepsy. While the sensitivity of this technique to unveil metabolic anomalies in the proximity of the pathological core substrate is unquestioned, specificity to dissociate epilepsy syndromes remains unknown. The current study utilized a surface-mapping procedure to evaluate the divergence of FDG-PET anomalies in cohorts with temporal lobe epilepsy (TLE) and frontal lobe epilepsy (FLE). Methods: We studied 14 patients with TLE (3110 years; 10 women; 6/8 left-/right-sided seizure focus) and 10 with dysplasia-related FLE (309 years; 6 women; 3/7 left-/right-sided seizure focus). In all patients, we acquired FDG-PET data on a Siemens tomograph as well as high-resolution T1-weighted MRI images on a 3T Siemens TimTrio. On T1-weighted images, we generated cortical surfaces which were linearly registered to minimally processed PET volumes. We subsequently sampled tracer uptake along cortical surfaces in each individual, followed by surface-based smoothing and across-subjects co-registration with regards to cortical folding patterns. Uptake maps were z-normalized in each individual, prior to sorting hemispheric maps relative to the focus. Using Student's t-tests with correction for multiple comparisons, we compared normalized tracer uptake maps between cohort . For individualized prediction, we utilized a surface-wide support vector machine classifier with leave-one-out cross-validation. Results: While TLE patients showed bilateral temporo-polar, lateral temporal, and ipsilateral medial prefrontal FDG decreases, FLE patients presented with decreased glucose metabolism in ipsilateral medial prefrontal regions (P < 0.05, FWE-corrected; Figure 1). Comparing histopathological subgroups among patients that were operated, we observed a tendency for more marked ipsilateral temporal hypometabolism in TLE patients with hippocampal sclerosis (n=4) compared to those with gliosis only (n=6; t=1.9; p < 0.05 uncorrected). On the other hand, there was no difference in cortical glucose metabolism between FLE patients with FCD Type IIA (n=5) relative to those with Type FCD Type IIB (n=5). Surface-wide support vector machine learning accurately discriminated TLE from FLE patients (23/24 correctly classified i.e., 96% accuracy; permutation-based significance level: p < 0.01). Rerunning our analyses after correcting PET data for corresponding cortical thickness measurements yielded virtually identical findings, suggesting that metabolic divergences occur independently of syndrome-specific morphological anomalies. Conclusions: Our findings demonstrating divergent cortical metabolic signatures across the two most common drug-resistant focal epilepsy syndromes both at the group- and individual-patient level, indicate that FDG-PET may offer sensitive and specific biomarkers, which may complement morphological MRI studies, in the presurgical assessment of epileptic patients. Funding: This research was funded by the Canadian Institutes of Health Research (CIHR MOP-57840 and CIHR MOP-123520). BCB received a CIHR postdoctoral fellowship.
Neuroimaging