In our study using whole-brain voxel-based 18F-FDG-PET, a striking difference in metabolic patterns between MTLE and NTLE was observed. Patients with MTLE displayed a characteristic ipsilateral hypometabolism in the temporal lobe and bilateral cortical-subcortical hypermetabolism pattern. In contrast, NTLE patients showed more extensive hypometabolic regions including the temporal and front-parietal cortex, and hypermetabolic in certain cortical regions. The difference in metabolic patterns between the two TLE subtypes may represent distinct epileptic networks that can guide the clinical classification of epilepsy. Hypometabolism involving ipsilateral thalamus and bilateral frontal lobe could be a predictor of an unfavorable seizure outcome in TLE, which could facilitate preoperative counseling and surgical planning.
18F-FDG-PET has been widely used in patients with medically refractory focal epilepsy for detection of the epileptogenic lesion[11, 14, 15]. Pathologic reduction in glucose metabolism indicates focal neuronal and synaptic loss associated with epileptogenic brain region. Our study has identified two patterns of cerebral hypometabolism in the MTLE and NTLE groups. Glucose hypometabolism of metabolism is present in ipsilateral hippocampus and insular lobe in MTLE. It is well accepted that the medial structure of the temporal lobe including hippocampus is the seizure origination location in MTLE, and as spikes propagate, the cortical insula commonly becomes the earliest involved structure via the Papez circuit. Recurrent spike distribution and seizure propagation from mesial temporal lobe to insular cortex contribute to focal metabolic impairment in insular, or even lead to insular atrophy,. In NTLE, the hypometabolic region is mainly located in the temporal neocortex associated with EZ. Notably, we have also found some restricted extratemporal cortical regions in frontal and parietal lobes showing decreased glucose metabolism in NTLE patients, suggesting more extensive brain regions and connected propagated networks affected in NTLE. Therefore, it is plausible that patients with NTLE usually demonstrate behavioral arrest with awareness impairment at the early stage[27, 28] and followed with motor signs as the seizure activity spread to the frontoparietal convexity[29–31].
A more significant disparity in hypermetabolic patterns is found between MTLE and NTLE patients. Both showed hypermetabolism in cortical and subcortical regions but with different locations. Glucose metabolism is commonly enhanced in the bilateral basal ganglia, brainstem, thalamus, corpus callosum, cingulate gyrus, and frontoparietal-occipital cortex, suggesting the propagation pathway in these structures probably be involved in the MTLE group[32–36]. The thalamus and brainstem are distinctly connected with the cingulate gyrus and cerebral cortex[37, 38]. A focal seizure that originates from the mesial structure of temporal lobe usually spreads to key subcortical regions such as the brainstem and bilateral thalamus, resulting in widespread metabolic disturbance in midline subcortical structures and neocortex[39–42]. Additionally, large areas of hypermetabolism in contralateral cerebral regions of MTLE seem to result from the restoration of chemical homeostasis.
Although cortical and subcortical hypermetabolism extra the EZ is also present among the patient with NTLE, it is much more restricted in the range of glucose metabolism. Notably, the key hypermetabolic region of NTLE is typically bilateral thalamus, suggestive of an underlying propagation pathway from the lateral neocortex of temporal lobe through the basal ganglia to the contralateral hemisphere that contributes to bilateral tonic-clonic seizures. Considering the extent of involvement, we speculate that except for the trans-thalamic circuits, other propagating pathways may exist in NTLE.
In addition, we observed significant hypermetabolism in the cerebellar, consistent with previous functional MRI and SPECT studies[46, 47]. Sufficient data indicate that cerebellum is engaged during seizures, manifesting as reduced gray matter volumes and an increased cerebellar blood flow and neuronal activities during seizures[49, 50]. Experimentally, increased interictal metabolic activities in the cerebellum have also been reported in animal models of epilepsy. The cerebellar metabolic changes in TLE patients in our study suggest a functional significance of temporal-cerebellar connections and the potential ability for cerebellar activation to inhibit seizures. Another possibility is that the abnormal glucose metabolism in the cerebellum may be compensatory as one of the downstream targets through divergent output pathways from temporal lobe. Of interest, bilateral cerebellum is likely to be affected in MTLE, while abnormal metabolism is only found in the contralateral cerebellar hemisphere in the NTLE group, highlighting the potential importance of the cerebellum in differentiation of epilepsy phenotypes in TLE.
We have further assessed metabolic features and surgical outcomes in patients with two subtypes of TLE. We find that patients with MTLE have a worse surgical prognosis if thalamic or frontal hypermetabolism is present. Lesional mesial temporal lobe epilepsy usually include hippocampal sclerosis, focal cortical dysplasia, or local neurodevelopmental tumors[53, 54]. Due to their limited focal damages, standard anterior temporal lobectomy offers comparatively favorable outcomes (50–80% seizure-free rate)[55, 56]. However, if combined with an extended frontal or thalamic 18F-FDG metabolic disturbance, actual lesions associated with epileptogenesis or seizure related networks in MTLE may be larger than they appear on structural imaging. As a result, the removal of only the anterior temporal lobe region may not be sufficient for postsurgical seizure control. For this reason, surgical procedure selection should be comprehensively determined if MTLE patients show thalamic or frontal involvement on the 18F-FDG-PET images. Precise and sufficient location of epileptogenic foci, including invasive EEG, should be carefully carried out for better surgical outcomes.
Several limitations in our study should be mentioned. This is a retrospective study, which inevitably brings selection biases. The sample size is relatively small, especially for the NTLE group that only accounts for 10% of TLE. Even though ASMs were discontinued for at least 24 h before the PET scan, they may still have residual effects on brain metabolism, which should be avoided in future studies. The decision to undergo a resection procedure is complex involving consideration of the patients’ preoperative structural MRI, video-EEG, and invasive SEEG.