Interictal [¹⁸F]FDG PET usually demonstrates hypometabolism of one temporal lobe (Figs. 2,3,4) or bilateral temporal hypometabolism with more severe hypometabolism of one temporal lobe (Fig. 5) in adults and children with refractory mesial temporal lobe epilepsy (TLE).^{1,3,10,11,39,40,41,42,60,61,67,71,72,74,75,98,106,108,118,137,151,164,165,167,168,186,187,189,197,204,208,210,212,215} Qualitative visual analysis of [¹⁸F]FDG scans obtained with high-performance tomographs detects unilateral (or bilateral, but asymmetric) temporal lobe hypometabolism in >70% of refractory TLE patients. Higher-resolution tomographic systems support a higher detection rate for hypometabolism, and greater concordance among readers, in qualitative interpretation of [¹⁸F]FDG imaging in localization-related epilepsies.⁷² With quantitative analysis, detection of significant temporal lobe hypometabolism may exceed 90% in this group.^92,98,216

Temporal lobe hypometabolism usually extends over mesial and lateral portions of an interictally dysfunctional temporal lobe on [¹⁸F]FDG scans in mesial TLE.^{72,74,167,215} Regional hypometabolism in mesial TLE typically is diffuse, with graded demarcations from adjacent areas of normal metabolism and a relatively large area of hypometabolism. Even in the presence of a temporal lobe foreign-tissue lesion, patients with refractory mesial temporal seizures usually have widespread temporal lobe hypometabolism rather than focal hypometabolism restricted to the site of the lesion. When using older tomographs with lower spatial resolution, the lateral temporal hypometabolism often appeared more severe than the mesial temporal hypometabolism of an affected temporal lobe on qualitative scan interpretation and even with quantification.^72,167 Using an ultra-high-resolution tomograph, however, one study of mesial TLE found that small volumes of anterior mesial

temporal structures were more severely hypometabolic than were any other temporal or extratemporal areas in many mesial TLE patients.²¹⁵ As shown in FIGURE 3, current tomographs more often provide [¹⁸F]FDG scans that show mesial temporal greater than lateral temporal hypometabolism of the affected temporal lobe. This suggests that partial-volume averaging of severe hypometabolism in the epileptogenic amygdala-hippocampus together with less depressed metabolism of adjacent basal temporal areas may cause the mesial temporal areas to appear less severely hypometabolic than they actually are when older clinical PET systems are used. In contrast to lateralized temporal lobe predominance of metabolic dysfunction most often observed in refractory TLE, some TLE patients have symmetric hypometabolism of both temporal lobes (Fig. 6).^10,93 Normal interictal metabolism also occurs in refractory mesial TLE, but normal [¹⁸F]FDG scans are more common in nonrefractory than in refractory mesial TLE.^55,124

Many TLE patients have unilateral frontal, parietal, thalamic, or basal ganglial hypometabolism ipsilateral to temporal hypometabolism, but occipital hypometabolism is rare in mesial TLE.^{3,11,72,74,161,168,186} The temporal hypometabolism is nearly always more severe than is any extratemporal hypometabolism. The thalamus ipsilateral to the affected temporal lobe is the extratemporal site most likely to demonstrate hypometabolism in limbic TLE.^72,74 Thalamic asymmetries in [¹⁸F]FDG activity are mainly dorsal (Fig. 4), and in some cases are associated with dorsal thalamic hypovolemia on structural MRI. The ipsilateral putamen is significantly more likely to be interictally hypometabolic in patients whose complex partial seizures involve dystonic posturing than in patients without ictal hemidystonia.¹⁶¹ The cortical hypometabolic area typically is contiguous across its entire temporal and extratemporal extent (Fig. 4). In particular, the insular and inferior frontal regions are most often the portions of the ipsilateral frontal lobe that are hypometabolic in association with temporal lobe hypometabolism.^3,11,72,137 Bilateral cerebellar hypometabolism is common.²⁰³ Thus, interictal [¹⁸F]FDG PET in refractory mesial TLE usually reveals unilateral diffuse regional hypometabolism of one mesial-lateral temporal area, with or without ipsilateral extratemporal cortical hypometabolism or contralateral temporal hypometabolism; ipsilateral extratemporal hypometabolism almost always appears to be less severe than does the temporal lobe hypometabolism.

The pathophysiologic basis of regional hypometabolism imaged with [¹⁸F]FDG interictally in TLE is unclear. Ablative structural lesions must contribute to localized decreases in glucose metabolism. Nonetheless, it has been recognized for some time that the volume of hypometabolism is greater than the volume of associated structural lesions in localization-related epilepsies.³⁹ Acute-subacute cerebral infarction is associated with hypometabolism at the site of neuronal loss, and additional extra-infarctional sites of hypometabolism are considered to represent “diaschisis” as passive and usually impersistent effects of a focal insult on remote brain regions that receive projections from the insulted area. Neuronal loss and diaschisis were considered the causes of the anatomically
distributed interictal hypometabolism in TLE patients with hippocampal sclerosis. This hypothesis was refuted, however, by a study of quantified preoperative [¹⁸F]FDG PET in patients whose resected temporal tissue underwent quantitative neuronal volumetric densitometry.⁶⁷ Subsequent studies, using different methodologies, confirmed this observation.^15,33,47,112 Similarly, several studies of MRI volumetry and interictal [¹⁸F]FDG PET found that hippocampal atrophy and temporal lobe hypometabolism are not highly correlated in mesial TLE.^{53,99,139,205} Furthermore, MR spectroscopic determination of the neuron-specific peaks of N-acetylaspartate (NAA) also found limited association of mesial temporal glucose hypometabolism and NAA decreases.^2,97 Although neuronal loss and diaschisis probably cause some of the disseminated glucose metabolic depression in mesial TLE, other factors must also influence regional metabolism interictally. Indeed, surgical ablation of epileptogenic temporal lobe tissue might be expected to increase diaschisis effects on extratemporal glucose metabolism. Instead, comparisons of preoperative and postoperative [¹⁸F]FDG scans have shown postoperative increases in [¹⁸F]FDG activity of the contralateral temporal lobe and of ipsilateral thalamus and frontal sites.^37,61,183 Several pathophysiologic processes other than diaschisis have been proposed, including reduced transmembrane glucose transport rates and reduced mitochondrial glucose oxidation of viable neurons in sites of ictal onset and habitual propagation.^155,217 On comparing the anatomic distributions of ictal hyperperfusion (using SPECT) and interictal glucose hypometabolism (using [¹⁸F]FDG PET) within individual TLE patients, it is evident that many TLE patients have an identical or highly similar distribution of ictal hyperperfusion–interictal hypometabolism that predominates in the temporal lobe and thalamus ipsilateral to the epileptogenic site (including but extending beyond the electrographic ictal onset zone), often with less severe involvement of ipsilateral frontoparietal and basal ganglial sites and of contralateral temporal cortex⁷⁶; one exception is the occipital hyperperfusion often observed in the absence of occipital hypometabolism in TLE. Possibly long-duration postictal dysfunctions may underlie some of the reduction in [¹⁸F]FDG activity, as suggested by differential associations of specific distributions of extrahippocampal hypometabolism with specific patterns of electrographic seizure propagation in groups with unilateral hippocampal sclerosis, for example, patients with bitemporal hypometabolism having earlier spread of ictal discharges from one hippocampus to the other.^16,171 Altered energy use in neurotransmitter metabolism might mediate such spatial relationships between ictal dysfunctions and postictal dysfunctions, as suggested by a study using MR spectroscopic determination of the glutamate-glutamine peaks in which localized interictal decreases and ictal increases of [¹⁸F]FDG activity and glutamate-glutamine peaks were coupled.¹⁴⁴ Dual pathology undetectable with MRI—specifically, microscopic cortical dysplasia accompanying some but not all cases of hippocampal sclerosis—might generate some of the heterogeneity of [¹⁸F]FDG topography in TLE.³² The diagnostically robust patterns of interictal glucose hypometabolism in TLE are not fully explained by macrostructural, microstructural, and biochemical alterations.

Peri-ictal [¹⁸F]FDG PET studies have shown a complex time course of changes in regional cerebral metabolic rate of glucose utilization (CMRGlc) following seizures. In a group of patients with complex partial seizures who had PET at different intervals following the most recent seizure, quantified regional metabolic changes evolved for >48 hours after a single complex partial seizure.¹¹⁶ The most severe regional hypometabolism occurred >48 hours after the seizure, the least severe hypometabolism occurred at 24 to 48 hours postictally, and metabolism was intermediate in the first 24 hours postictally. In a study in which [¹⁸F]FDG studies were performed on average <60 hours after the most recent seizure, the type of seizures that preceded the scan had a strong influence on the regional distribution of hypometabolism.¹⁷¹ In general, the most localized ictal discharges preceded scans with the smallest volumes of hypometabolism, and secondarily generalized seizures preceded scans with the most widespread patterns of unilateral hypometabolism. In both lesional and nonlesional partial epilepsies, the volume of hypometabolism expands with increasing numbers of seizures over the course of the epilepsy, even with nonprogressive lesions,^88,124 and in limbic TLE, the severity of mesial temporal hypometabolism increases with increasing duration of epilepsy.²⁰⁹

Focal mesial temporal hypermetabolism sometimes occurs interictally in children with mesial TLE, but it rarely occurs in adults with localization-related epilepsies.^23,41,210 Continuous or repetitive focal mesial temporal seizures, which are subclinical and not detectable with scalp electrodes, may cause “interictal” deep temporal hypermetabolism.¹⁸⁵ Alternatively, there may be interictal epileptogenic processes that are peculiar to childhood and that generate greater glucose metabolism interictally. The latter speculation is encouraged by the presence of interictal regional hypermetabolism in some young children with Sturge-Weber syndrome or with infantile spasms, which does not occur in older children with Sturge-Weber syndrome or with Lennox-Gastaut syndrome (a common “endpoint” for patients with infantile spasms earlier in life), the older children having exclusively hypometabolism or normal metabolism interictally.

Ictal or peri-ictal (representing mixed ictal–postictal–interictal states during the [¹⁸F]FDG uptake period) [¹⁸F]FDG scans are difficult to obtain and interpret. True ictal imaging with [¹⁸F]FDG is restricted to status epilepticus due to the relatively poor temporal resolution of the [¹⁸F]FDG method. Occurrence of a single complex partial seizure during the [¹⁸F]FDG uptake period may be associated with the usual interictal findings of unilateral temporal hypometabolism. In one reported case, a partial seizure occurred about 2 minutes after [¹⁸F]FDG injection and the scan appeared normal; the same patient later had marked hypometabolism of the epileptogenic temporal lobe on an interictal [¹⁸F]FDG scan.⁷¹ Presumably, ictal hypermetabolism was averaged with interictal–postictal hypometabolism over the temporal lobe to cause “normalization” of [¹⁸F]FDG activity on the peri-ictal scan. In another case, a TLE patient had repeated complex partial seizures following [¹⁸F]FDG injection, and the scan showed hypermetabolism over the epileptogenic temporal lobe, with ipsilateral frontal and thalamic metabolic increases.⁴⁴ Alterations on ictal and peri-ictal [¹⁸F]FDG images likely reflect ictal dysfunction at the site of ictal onset, but they also reflect dysfunctions in areas of ictal propagation and interictal and postictal dysfunction in these areas. It is difficult or impossible to sort out the relative contributions of these various dysfunctions to a single set of [¹⁸F]FDG images.

Cerebral glucose metabolism and CBF normally are “coupled” (increasing and decreasing in parallel with changes in synaptic activity within each region of cortex), but interictal PET measurements show uncoupling of CBF and [¹⁸F]FDG activity in TLE.^53,114 Interictal CBF imaging with PET often shows “diffuse” regional hypoperfusion, consisting of a relatively large area of hypoperfusion with indistinct boundaries from adjacent areas of normal CBF and with inhomogeneous severity of hypoperfusion.^9,48,108,118 Interictal regional CBF decreases often occur predominantly contralateral to the ictal onset zone in mesial TLE.^118,206 Ictal CBF imaging of complex partial seizures with [¹⁵O]H₂O is nearly impossible to obtain, given the 2-minute half-life of [¹⁵O], except with seizure induction by proconvulsant drugs or during complex partial status epilepticus.¹⁹⁶ For these reasons, ictal imaging and resting
interictal imaging with [¹⁵O]H₂O have no clinical role in presurgical evaluation. It has been suggested, however, that detection of regional glucose metabolic–CBF uncoupling might be more sensitive in detecting epileptogenic tissue than is [¹⁸F]FDG mapping alone.²²⁶