18F-FDG PET/MRI allows improved detection of epileptogenic focus in patients with normal MRI

Objectives: In approximately 25% of patients with drug resistant focal epilepsy undergoing pre-surgical evaluation, magnetic resonance imaging (MRI) is normal or discordant with clinical and EEG data. A major challenge in pre-surgical evaluation is the correct delineation of the seizure onset zone (SOZ), which often requires invasive EEG recordings. We aimed at assessing the clinical value of 18F-FDG PET coregistered with MRI (18F-FDG PET/MRI) in localizing epilepsy prior to intracranial EEG recordings. Furthermore, we tested the predictive value of 18F-FDG PET/MRI on the postoperative outcome.

Methods: Sixty patients with neocortical epilepsy (mean age at epilepsy onset 13.3±8.12 years) with normal 1.5T 3D brain MRI were retrospectively evaluated. For each patient, we collected clinical characteristics, localizing features on preoperative diagnostic tests (ictal video-electroencephalogram (VEEG), intracranial EEG (iEEG), FDG PET/MRI images), histopathological findings and surgical outcome. PET images were reprocessed, fused with MRI using FSL software (Oxford Centre for Functional MRI brain) and compared to invasive EEG recordings. Surgery was performed 1 to 43 months after PET (median 9 months). Outcomes were assessed according to Engel’s classification at follow-up until 66 months (median 32 months). A favorable outcome was defined as reduced rate of seizures (Engel IA-IB-IC-2A) and/or reduced antiepileptic therapy.

Results: In 47 patients (73.3%), PET/MRI could detect foci of decreased uptake. Compared to invasive EEG recordings, PET/MRI correctly identified SOZ in 42 patients (70%), detected foci discordant to EEG in 5 (8.3%) and resulted falsely negative in 13 (21.7%). Pooled sensitivity was 76.4%, positive predictive value 89.4%, positive likelihood ratio 76.4%, post-test probability 89.4%. A positive PET/MRI examination did not show a significant correlation with a poorer outcome (p=0.16) or with the need of increasing antiepileptic therapy (p=0.47).

Conclusion: PET/MRI allowed for a correct, noninvasive identification of SOZ in 70% of patients with an unremarkable MRI examination. The presence of a hypometabolic focus on PET/MRI is not predictive of a worse outcome after surgery. Brain PET/MR should be routinely performed before surgery and may limit the need of a preoperative invasive assessment. Research Support: N/A