Abstract
P1353
Introduction: Drug-resistant mesial temporal lobe epilepsy (MTLE) patients commonly resort to anterior temporal lobectomy for seizure control; however, many still suffer from seizures post-surgery, which may relate to their extensive brain network damage. 18F-FDG-PET is very sensitive to metabolic changes in the seizure onset zone (SOZ), which typically captures large areas of hypometabolism extending to regions beyond the presumed SOZ. Diffusion Tensor Imaging (DTI) has been widely used to assess the integrity of white matter (WM) tracts. It is unclear whether the WM diffusivity changes topographically coincide with cortical hypometabolism observed in 18F-FDG-PET in MTLE. In this study, we investigated the changes in glucose uptake, white matter tracts, and their differences in brain networks of MR-HS versus MR-negative MTLE patients using simultaneous PET/MR.
Methods: Data acquisition:
Fifty-five drug-refractory unilateral MTLE patients were scanned (age³18; without other lesions) on a hybrid PET/MR scanner (Siemens Biograph mMR VE11P). One subject was excluded from the analysis due to low image quality. The 18F-FDG PET images were reconstructed using events acquired at 45-60 minutes post injection (mean dose of 3.7 MBq/kg). During the PET acquisition, MR data were acquired: T1-weighted MPRAGE (1×1×1 mm3, TR/TE/TI = 1900/2.44/900 ms) and diffusion MRI (2.0×2.0 mm2, thickness 2.0 mm, TR/TE = 10200/90 ms, 30 directions at b = 1000 s/mm2). Other clinical records were also collected, such as semiology, long-term video-EEG monitoring, radiological reports indicating MR-HS and MR-negative cases, and presurgical diagnosis were also collected.
Data processing:
Brain region parcellation was performed on the MPRAGE images using Freesurfer (v7.0). The 18F-FDG standardized uptake value ratios (SUVR) of the ROIs were obtained and normalized by cerebellar gray matter.
Diffusion-weighted images were first preprocessed for motion and eddy current correction using eddy, and then QSDR reconstruction was used (DSI-studio). Voxel-wise FA (fractional anisotropy) and MD (mean diffusivity) maps were generated.
Data analysis:
Correlational tractography analysis was conducted with DSI-studio. A multiple regression model was used to remove the effects of sex, age, disease laterality, and encoding direction (T-score = 2.5). A false discovery rate (FDR) threshold of 0.05 was used.
The SUVR discordance of bilateral brain regions was first obtained as SUVR_diff = SUVR_ipsi – SUVR_contra.
Metabolic connectivity matrices of the regional 18F-FDG SUVRs were generated by calculating pair-wise Pearson correlation between two ROIs across subjects.
Results: Patient demographics are summarized in Table 1.
Mesial temporal structures, such as hippocampus, amygdala, parahippocampus, and middle temporal cortex, were found to have a significantly larger SUVR_diff in MR-HS group compared to MR-negative group (P<0.001). Notably, the thalamus, a key region often involved in seizure propagation, was also affected (P = 0.00758). Table 2 shows the results of all the ROIs.
Figure 1 demonstrates the tracts with higher MD or lower FA in the MR-HS group. Noteworthily, the fornix of the brain showed reduced FA and increased MD, this could reflect subtle microstructural damages due to recurrent seizures.
Furthermore, the inter-subject metabolic connectivity matrices (Figure 2) demonstrate a loss of structure in the MR-HS group. The root-mean-square errors were: 4.19, 3.42 for the ipsilateral and contralateral connectivity matrices of the MR-HS group when compared to the ipsilateral matrix of the MR-negative group, indicating the ipsilateral brain of the MR-HS group is more affected.
Conclusions: This study identified extensive network damage in MR-HS patients among patients with unilateral mesial temporal lobe epilepsy. Further investigation is warranted to study if other more personalized treatment approaches are needed beyond anterior temporal lobotomy alone.
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.