John Libbey Eurotext

Temporoporal metabolic abnormalities in temporal lobe epilepsies Volume 4, supplément 1, Supplement 1, September 2002

Since the ground-breaking work at UCLA [1], Positron Emission Tomography (PET) has made a great contribution to the pre-surgical assessment of patients with partial, drug-refractory epilepsy, originating in the temporal lobe or elsewhere [2]. Over the last twenty years, a large number of studies have characterized changes in interictal cerebral metabolic patterns using 18F-deoxyglucose-PET (FDG-PET) thereby contributing to our understanding of the pathogenesis in epilepsy.

These studies have shown that, between seizures, metabolic activity is reduced in this region which therefore functions abnormally.

In temporal lobe epilepsy (TLE), PET is particularly sensitive in that it can detect unilateral, interictal temporal hypometabolism and thereby identify the epileptogenic lobe in 70% to 95% of patients [2-7]. A series of studies in patients with temporal lobe epilepsy have shown that metabolic activity in certain specific regions within the temporal lobe is particularly low: for instance in the temporal pole. This metabolic deficiency and its impact on the rest of the temporal lobe is the subject of this article.

Descriptive anatomy of interictal temporal lobe hypometabolism in temporal lobe epilepsy

Interictal temporal lobe hypometabolism was first reported as an extensive area involving the entire temporal lobe and even extending to other regions such as the thalamus and the frontal cortex (although it was found to be less marked in these secondary regions). Such data were acquired using PET machines with a spatial resolution of the order of one centimeter at a time when MRI was not as reliable as it is today. PET data were used to confirm temporal localization in partial epilepsy and to lateralize the epileptogenic temporal lobe [2, 8-10]. Since then, the topography of the hypometabolism has been more accurately defined thereby enhancing the value of the information provided by FDG-PET examinations.

Studies of the distribution of interictal temporal lobe hypometabolism have made a major contribution to the localization of abnormal areas, and the hippocampal plane is particularly useful for this purpose (Figures 1 and 2) [11-14].

Abnormalities within the temporal lobe

The hypometabolism characteristic of TLE involves a large part of the temporal lobe. Recent studies have shown that the mesial part of the temporal lobe is usually involved as well as a large part of the neocortex (Figures 3 and 4). Most of the quantitative studies have suggested that the most profound metabolic impact can be observed at the temporal pole [3, 15].

Metabolic activity is usually particularly low in the mesial temporal lobe. In a study of 22 patients suffering from mesial temporal lobe epilepsy associated with hippocampal sclerosis, we showed that metabolic activity in the temporal pole was reduced by a factor of between 13 and 17% whereas hippocampal involvement resulted in a mean reduction of just 13% with decreases in the rest of the temporal neocortex varying between 0 and 10% [3]. This study also detected an anteroposterior gradient of involvement in the temporal neocortex with the anterior regions more profoundly affected and the level of metabolic activity in posterior regions almost always perfectly normal. No significant differences could be detected between the relative degree of involvement in the superior, middle and inferior temporal gyri.

Extratemporal cortical abnormalities

Although metabolic problems in patients with TLE concern primarily the temporal lobe, many different research groups have detected interictal hypometabolism in extratemporal regions as well. Here metabolic activity is not reduced to the same extent as it is in the temporal lobe itself. Many different regions may be involved, most commonly the frontal lobe, especially the orbitofrontal cortex [16], but sometimes the parietal lobe [17] and a recent investigation also detected hypometabolism in the insula [17]. Taking the results of all these studies together, it can be concluded that extratemporale cortical involvement is often secondary and that such metabolic abnormalities can spread far and wide. The exact significance of these abnormalities is not understood as yet.

Sub-cortical abnormalities

Metabolic activity is also often reduced in certain sub-cortical structures [16], sometimes to a similar extent as that in the temporal lobe, especially in the thalamus [17]. Both the mechanism underlying this phenomenon and its relationship to the seizures suffered by patients with TLE are controversial (as is the case with extratemporal involvement).

Khan et al. [18] studied 17 patients with unilateral TLE associated with either: hippocampal sclerosis (n = 11); a mesial temporal tumor (n = 3); or neocortical temporal epilepsy (n = 3). Thalamic hypometabolism was only detected in those patients with mesial temporal epilepsy associated with hippocampal sclerosis. The focus of the thalamic hypometabolism appeared to be in the dorsomedial nucleus, and metabolic activity was found to be relatively high in the lateral parts of the structure [19]. Sub-cortical hypometabolism was also found to be associated with certain symptoms which manifested during seizures suggesting the possibility that the discharge may encroach on sub-cortical structures. A correlation has also been established between ictal dystonia and contralateral striatal hypometabolism [20].

The presence of interictal sub-cortical hypometabolism points to the possibility of functional abnormalities remote from the epileptogenic focus without any underlying anatomical lesion.

Hypotheses on the pathogenesis of polar temporal abnormalities

Despite the extensive literature on metabolic abnormalities in human partial epilepsy, the exact mechanism which causes interictal hypometabolism is controversial. In patients with epilepsy due to an expanding process, the area of hypometabolism often corresponds exactly to the lesion (or sometimes extends slightly beyond it). In contrast, in TLE associated with hippocampal sclerosis, the area of hypometabolism is generally far more extensive. Among the possible mechanisms for this hypometabolism, it seems reasonable to suggest that neuronal loss and consequent deafferentation phenomena might be particularly important. Some studies have found a correlation between the extent of neuronal loss and the degree of hypometabolism although others have found no such relationship. Thus, although it is almost certain that neuronal loss plays a role, it cannot be the only factor causing all the metabolic abnormalities. A deafferentation-based mechanism probably accounts for the polar temporal hypometabolism observed in mesial temporal lobe epilepsy [3]. Other mechanisms may also play a role, e.g. postictal inhibition. A local reduction in synaptic density in specific pathological conditions like certain forms of cortical dysplasia could also cause hypometabolism.

The effect of underlying lesions

A number of reports have shown that there is not necessarily a strict correspondence between tissue damage and the area of reduced metabolic activity. Often, the area of hypometabolism is more extensive than that of the lesion [21] although this pattern depends on the location of the lesion.

This point was illustrated in a study of 22 patients with a cavernous hemangioma associated with epilepsy [22]. In only four of these patients did the area of reduced metabolic activity go far beyond the confines of the hemangioma and, in all four, the lesion was blocking communications between the temporal paralimbic area and the temporal neocortex. In the other 18 patients, the cavernous hemangioma was located elsewhere and the area of reduced metabolic activity coincided fairly closely with the lesion.

The nature of the lesion associated with the epilepsy also affects characteristics of the interictal hypometabolism. In mesial temporal lobe epilepsy, hippocampal sclerosis is associated with a more drastic reduction in metabolic activity than is seen in mesial temporal expanding lesions [5].

Several authors have investigated the relationship between neuronal loss in the hippocampus and mesial temporal hypometabolism. These studies have shown that there is no close relationship between hippocampal metabolic activity (as measured by PET) and neuronal loss in the hippocampus (as estimated either by a neuropathogist or by MRI measurement of volume). Nevertheless, some degree of correlation has been detected between hippocampal neuronal density and metabolic activity in various other regions, including the ipsilateral hemisphere [23], the entire temporal lobe [24], the thalamus [25], the inferior temporal pole [3] and the inferior mesial temporal region [6]. Therefore, although there is no close relationship between neuronal loss in the hippocampus and hippocampal hypometabolism, these two phenomena are nevertheless related in some way.

When it comes to hypometabolism in the temporal pole, the idea of neuronal loss being the only cause could never account for the observed extent of reduction in metabolic activity [23]. In some patients, temporal pole hypometabolism is associated with a focal lesion (Figure 5) but in most cases morphological changes are extremely subtle [26], strongly contrasting with the profundity of the metabolic effect. More recently, MRI studies have revealed lesions in the white matter of the temporal pole which are dedifferentiation between the gray and white matter; such lesions appear to be fairly common in TLE (Ryvlin et al., this volume). Choi et al. [27] have shown that the degree of hypometabolism tends to be greater in patients with this type of abnormality.

Deafferentation

The existence of metabolic abnormalities at a distance from epileptogenic and damaged tissue is suggestive of a deafferentation-based mechanism (Figures 6 and 7). Evidence for this hypothesis comes from clinical studies and experimental anatomical investigations.

Changes in temporal pole hypometabolism following ablation of the hippocampus

Strong evidence in favor of the deafferentation hypothesis comes from data on post-surgical changes in metabolic patterns acquired by a number of different groups.

Hajek et al. [28] showed that after resection of the mesial part of the temporal lobe in patients with epilepsy originating in this region, temporal pole metabolic activity diminished five months following the procedure and was still low after one year; lateral temporal pole metabolism which had been normal prior to the surgery was hardly affected at all. In a comparable study, we showed a similar reduction in metabolic activity after selective ablation of the amygdala and the hippocampus which did not touch the temporal pole [29].

This is similar to what was observed after radiosurgical ablation of the hippocampus in patients suffering from TLE with hippocampal sclerosis [30]: after radiosurgery, temporal pole metabolic activity fell off sharply while seizures stopped and did not recur within the subsequent two years.

These changes in temporal pole metabolic activity strongly suggest that the destruction of mesial temporal structures or of some hippocampo-cortical pathways to the temporal neocortex causes exacerbation of temporal pole hypometabolism. This mechanism is identical to that at play in the case of lesions (often vascular) which involve certain sub-cortical structures and lead to contralateral cortical hypometabolism.

Anatomical links between the temporal pole and the mesial part of the temporal lobe

Anatomical investigations in animals have revealed strong links between the hippocampus and the cortex of the temporal pole, the perirhinal cortex, the orbitofrontal cortex, and the cingulated gyrus. The amygdala also plays an important role in communications between the hippocampus and the cortex, by virtue of connections between the hippocampus and the amygdala and numerous efferent fibers to the cortex from the amygdaloid nucleus [31]. As shown in studies in monkeys, most of these projections from the amygdala terminate on the orbital or medial side of the frontal lobe, with a few terminating on the lateral side, the insula, the cortex of the temporal pole and the perirhinal cortex. Afferent fibers from association areas are relayed via the amygdala which also massively projects to the entorhinal cortex. The amygdala receives afferent fibers from the orbitofrontal cortex, the anterior cingulate gyrus, the temporal pole cortex, and the anterior part of the insular cortex. As shown by Van Hoesen and Pandya [33, 34], the entorhinal cortex reciprocally receives afferent fibers from various temporal regions, in particular the internal regions (the TF and TH areas in the 1947 Von Bonin and Bailey Classification System [35]), the perirhinal cortex (area 35) and the temporal pole cortex (the TG area). Although the entorhinal cortex appears to receive most of the cortical information, it seems that the subicular complex also receives direct cortical afferents, notably coming from the temporal pole, the perirhinal cortex and the hippocampal gyrus as well as from the cingulate cortex [36].

These strong links which join the temporal pole with internal temporal structures may account for the metabolic consequences of deafferentation described above.

Relationship with seizures

Metabolic abnormalities are often used to confirm the side of the epileptogenic lobe. One of the most important and discussed questions is whether these abnormalities can give us any additional information about the origin of partial seizures and their epileptogenic network. It has already been pointed out that metabolic abnormalities vary according to the presence or absence of frank lesions. Wieser and his group have shown that the localization and degree of interictal hypometabolism were different in patients with mesial temporal lobe epilepsy from those with neocortical epilepsy [5].

In lateral TLE, hypometabolism is usually confined to the external regions and often coincides exactly with the areas of damage that are often seen in this form of epilepsy (e.g. a tumor, abnormal cerebral development, vascular malformation, scar tissue, etc.). In contrast, in medial TLE, metabolic activity tends to be reduced not only on the mesial part of the temporal lobe but also in its lateral cortex and often on the lateral side as well. Thus, hypometabolism tends to be both more profound and more extensive in medial than in lateral temporal lobe epilepsy. When a pattern of hypometabolism typical of medial TLE is detected by PET in a patient known to have a lesion on the lateral part of the temporal lobe, it would be wise to investigate the possibility of the existence of a second lesion on the internal side.

Little work has focused on possible links between hypometabolism and inter- and periictal electroencephalographic abnormalities, although one study of 28 patients with mesial temporal epilepsy by Koutroumanidis et al. [37] revealed a correlation between the presence of interictal slow waves in scalp EEG and reduced metabolic activity in the lateral temporal lobe, especially the posterior regions. In another study [38], Single Photon Emission Computed Tomography (SPECT) was performed between seizures in patients who had previously had a stereo-EEG recordings; a comparison showed that regions in which cerebral blood flow was reduced also gave rise to depth-recorded spikes. Both these sets of results suggest that interictal EEG abnormalities might play a role in the existence and localization of the hypometabolism.

Other mechanisms

Other mechanisms may also be involved in hypometabolism, such as reduced synaptic density (without any neuronal loss) leading to reorganization of the neuronal network at the level of epileptogenic focus [23]; another possible factor could be developmental malformation of the cortex.

The effect of anti-epileptic drugs is also recognized. These drugs are known to inhibit cerebral metabolism in a global fashion but they may also induce regional changes dependent on the underlying cortical epileptogenicity [39-41].

A recent study indicated that inhibition of glucose transport at the blood-brain barrier might also be a factor in interictal hypometabolism [42]. These results should be taken in the context of the uncoupling of cerebral blood flow and cerebral metabolic activity in areas of temporal hypometabolism [43].

Predictive value of temporal pole metabolism measurements

The clinical relevance of PET examinations has been consolidated by demonstration of the correlation which exists between postoperative hypometabolism (both its presence and its degree) and surgical outcome.

Certain authors have suggested that the extent to which metabolic activity in the uncus region is reduced following surgery is a reliable predictor of a satisfactory outcome whereas lateral hypometabolism has no significance [44]. Other groups have shown that it is hypometabolism in the temporal lobe which is the most informative parameter in mesial temporal lobe epilepsy [45]. Similarly, a complete absence of interictal temporal hypometabolism appears to be a poor prognostic indicator in TLE.

It should not be surprising that links can be found between metabolic abnormalities and postoperative outcome in patients with TLE when all the mechanisms underlying hypometabolism are considered together. However, in order to define these relationships more precisely and to make it possible to obtain more information from FDG-PET examinations, specific studies of the relationships between seizures and metabolic abnormalities will have to be carried out.

CONCLUSION

The extent to which interictal temporal pole metabolic activity is reduced in mesial temporal epilepsy points up the importance of this region in this form of disease. A number of pathogenic mechanisms partly explain this phenomenon, most convincingly those based on deafferentation. Nevertheless, much remains to be done to pinpoint down the underlying causes of temporal pole hypometabolism in mesial temporal lobe epilepsy and, more importantly the clinical relevance of this phenomenon.

Acknowledgments

I would like to express my gratitude to S. Dupont, F. Chassoux, V. Bouilleret and B. Devaux for all their help with this work.