Brain glucose metabolism network study in MRI-negative Temporal Lobe Epilepsy prognosis

1551 Objectives: Temporal lobe epilepsy (TLE) represents the most prevalent form of localization-related epilepsies,where many of these cases have demonstrated pharmacoresistance and are potential candidates for epilepsy surgery. Roughly 30% TLE have normal MRI scans on visual inspection (MRI-negative TLE). PET scans utilizing 2-[18F] fluoro-2-deoxy-D-glucose positron emission tomography (FDG-PET) can help identify epilepsy-related metabolic disturbances allowing the number of surgeries performed on MRI-negative TLE increased. While the extensive epileptogenic network in MRI-negative TLE prognosis is largely unknown. In the present study, we intended to explore the brain glucose metabolism network in MRI-negative TLE prognosis using 18F-FDG PET. Methods: 34 patients with MRI-negative left temporal lobe epilepsy (LTLE),30 patients with MRI-negative right temporal lobe epilepsy (RTLE) and 15 matched healthy controls (HCs) underwent an 18F-FDG PET scanning. Patients were treated with anterior temporal lobectomy andthe postoperative follow-up was taken at least 2 years. It were divided in two groups according the prognosis: EngelIa and non-Ia group. LTLE :16 Ia and 18 non-Ia, RTLE: 16 Ia and 14 non-Ia. All images were spatially normalized into MNI space, and images of RTLE were left-right flipped. Then 18F-FDG uptakes were calculated for 90 cerebral regions using AAL atlas with bilateral cerebellum as reference. The group-level metabolic brain network was constructed by measuring Pearson correlation coefficients between each pair of brain regions in an inter-subject manner. Then global efficiency and small-worldness was calculated, and permutation test with 1000 times was performed to compare the statistical significance between the two groups. The hub nodes were identified by ranking three nodal network measures: degree, betweenness centrality and clustering coefficient. If a node had highest nodal degree, second-highest betweenness centrality and third lowest clustering coefficient, it was scored by 90, 89, 88, and its summed score was 267. Then ranking the summed scores for all nodes, the top 20% (18/90), were classified as hub nodes. Results: Compared to ctrl and non-Ia group, the network degree of Ia showed decreased tendency. The metabolism brain network of TLE showed small-worldness (σ>1). The small-worldness of Ia and non-Ia patients was similar to that of healthy controls (σ>1). Also Ia group showed reduced global efficiency compared to nonIa group. No significant small-worldness or global efficiency was found between nonIa and ctrl group. Conclusions: These findings revealed that the Ia and non-Ia group had different brain glucose metabolism network, which furthered our understanding of the mechanisms underlying MRI-negative TLE.It is expected to provide functional markers for the prognosis of patients with MRI-negative TLE.abolism network, which furthered our understanding of the mechanisms underlying MRI-negative TLE.It is expected to provide functional markers for the prognosis of patients with MRI-negative TLE.