Simultaneous high-resolution whole-brain MR spectroscopy and [ 18 F]FDG PET for temporal lobe epilepsy

Purpose Precise lateralizing the epileptogenic zone in patients with drug-resistant mesial temporal lobe epilepsy (mTLE) remains challenging, particularly when routine MRI scans are inconclusive (MRI-negative). This study aimed to investigate the synergy of fast, high-resolution, whole-brain MRSI in conjunction with simultaneous [ 18 F]FDG PET for the lateralization of mTLE. Methods Forty-eight drug-resistant mTLE patients (M/F 31/17, age 12–58) underwent MRSI and [ 18 F]FDG PET on a hybrid PET/MR scanner. Lateralization of mTLE was evaluated by visual inspection and statistical classifiers of metabolic mappings against routine MRI. Additionally, this study explored how disease status influences the associations between altered N-acetyl aspartate (NAA) and FDG uptake using hierarchical moderated multiple regression. Results The high-resolution whole-brain MRSI data offers metabolite maps at comparable resolution to [ 18 F]FDG PET. Visual examinations of combined MRSI and [ 18 F]FDG PET showed an mTLE lateralization accuracy rate of 91.7% in a 48-patient cohort, surpassing routine MRI (52.1%). Notably, out of 23 MRI-negative mTLE, combined MRSI and [ 18 F]FDG PET helped detect 19 cases. Logistical regression models combining hippocampal NAA level and FDG uptake improved lateralization performance (AUC=0.856), while further incorporating extrahippocampal regions such as amygdala, thalamus, and superior temporal gyrus increased the AUC to 0.939. Concurrent MRSI/PET revealed a moderating influence of disease duration and hippocampal atrophy on the association between hippocampal NAA and glucose uptake, providing significant new insights into the disease’s trajectory. Conclusion This paper reports the first metabolic imaging study using simultaneous high-resolution MRSI and [ 18 F]FDG PET, which help visualize MRI-unidentifiable lesions and may thus advance diagnostic tools and management strategies for drug-resistant mTLE.