John Libbey Eurotext

The role of the temporal pole in the genesis of temporal lobe seizures Volume 4, supplément 1, Supplement 1, September 2002

The concept of a mesial temporal lobe epilepsy syndrome

Temporal lobe epilepsies are the most frequent form of surgically remediable partial epilepsy, with an underlying anatomical context of hippocampal sclerosis in the great majority of cases. This lesion is found in a little over 70% of patients suffering from temporal lobe seizures who are operated, according to two recent surgical series which report very similar results [1, 2].

Tribute is due to Falconer who, in the early 1950's, on the basis of histological analyses of surgically removed tissue, emphasized the high incidence of hippocampal sclerosis, as well as the fact that the discovery of such a lesion was a positive indicator with respect to surgical outcome [3]. Subsequently, he noticed the strong association with antecedent of febrile convulsions and with a family history of epilepsy [4], thus describing for the first time some of the characteristics of what would later come to be known as the "mesio-temporal lobe epilepsy syndrome".

More recently published results confirmed that there does indeed exist a form of temporal lobe epilepsy which is particularly refractory to drug treatment [5] but easily amenable to surgery [6-8]. This syndromic entity is characterized by: i) a typical progressive development and a clinical picture of fairly homogenous seizures; ii) a topographic distribution of ictal and interictal EEG abnormalities which tend to be focused around the anterior and basal regions of the temporal lobe; and iii) an anatomical context which is easy to detect in an MRI examination, namely hippocampal sclerosis [9]. The large body of basic work showing the epileptogenic potential of the hippocampus prepared the medical community to accept the existence of this syndrome (diagrammatically summarized in Figure 1). The specific cell loss which occurs in the hippocampus together with the reorganization of synapses which accompanies it are key elements in the generation of ictal discharges and in the onset of the epileptogenic process.

The hippocampus, the amygdala, the hippocampal gyrus...

A number of observations point to the hippocampus having a central ­ although not necessarily exclusive ­ role in epileptogenesis in the temporal lobe: i) the hippocampus is particularly susceptible to experimental epileptogenic manipulations; ii) hippocampal sclerosis ­ whether it be a cause or a consequence of the seizures [10, 11] ­ is the most common abnormality detected in histopathological analysis in human temporal lobe epilepsy; iii) and the well-established fact that the outcome of surgery is significantly more likely to be positive when such a lesion is discovered. However, it is clear that any proposition that the hippocampus plays a causal role in epileptogenesis needs to be softened for at least four different reasons:

1) other structures such as the amygdaloid nucleus and the piriform cortex are just as susceptible to experimental epileptogenic procedures;

2) neuronal loss and a range of different types of histopathological changes can be detected outside the hippocampus in temporal lobe epilepsy, notably in the amygdaloid nucleus [12] but also in the uncus, the temporal pole and the temporal neocortex;

3) MRI detects a volume reduction (which is just as significant as that found in the hippocampus) in other limbic structures (the amygdala, the fornix, the mamillo-thalamic tract and the mamillary bodies) [14-22], with the exception of the cingulate gyrus [23], or in various other paralimbic structures, e.g. the entorhinal cortex [24] and the temporal pole [25];

4) surgical ablation is rarely restricted to just the hippocampus ­ the most commonly performed procedure is full anterior temporal resection ­ while so-called selective procedures usually include at least the hippocampal unit, the amygdala and the hippocampal gyrus. In the small number of studies carried out in humans which have directly set out to determine the role played by hippocampal atrophy in the onset of epileptic seizures, this factor has been clearly shown not to be exclusively responsible [28, 29]. According to Spanedda et al. [29], ictal discharges usually involved concomitantly both the amygdala and the hippocampus (67% of the cases), while they originated exclusively in the atrophic hippocampus in only 20% of the cases, a proportion almost similat to seizures originating exclusively from the amygdala (13%). This high frequency of concomitant hippocampal and amygdala involvement (which is also seen in other forms of temporal lobe epilepsy with highly variable histological patterns' [30]) probably accounts for the very limited success seen when surgical ablation fails to include the amygdala [31], the efficacy of temporal ablation seeming to depend ­ at least in part ­ on the extent of resection of the amygdala and the hippocampus [32].

Undoubtedly, in a number of carefully selected patients with TLE, a selective surgical approach ­ either limited to amygdalo-hippocampectomy [6] or using gammaknife surgery [33] ­ can give a remarkably high success rate, approaching 90%. Results tend to be less impressive (closer to 70%) when larger patient populations are considered [34]. This is particularly true when considering that in a number of publications the "seizure-free" (Engel Class I) population includes not only patients who have not had any kind of seizure since surgery (Class Ia) but also those who have had auras (Class Ia), those who have had a few postoperative seizures but none in the last two years (Class Ic), and those who have experienced atypical, generalized convulsions on the withdrawal of antiepileptic drugs (Class Id). The rate of complete success (i.e. just counting Class Ia patients) may drop below 40% in some series of patients, particularly in older studies for which inclusion criteria were probably less rigorous [35-37].

... and the temporal pole

The fact that a significant fraction of selective temporal lobe surgical procedures are unsuccessful raises the obvious question of whether the epileptogenic zone is not more extensive in some cases, extending to other structures than the amygdalo-hippocampal-parahippocampal complex. In the following section, we will try to summarize the evidence which suggests that the temporal pole might be the "Fourth Man" in this scenario.

Close links exist between the temporal pole ­ more specifically Brodmann area 38 ­ and both the entorhinal cortex to which it projects directly and the hippocampus from which it receives afferents via the subiculum (for details, refer to the article by Chabardès et al. in this volume). Therefore, it is extensively connected "in parallel" to a circuit which has been convincingly shown in experimental studies to be important in hippocampal epileptogenesis. Moreover, the temporal pole is integrated into the vast insulo-orbito-cingulo-temporal complex [38] by virtue of which it has strong connections with structures which tend to be involved in the propagation of discharges originating in the temporal lobe. It is the "crossroads" location of this hitherto little-investigated region coupled with the fact that it is often ablated in surgery to control epilepsy, which led Munari et al. in the early 1990's [39] to propose stereo-electroencephalography (SEEG) of the temporal pole as a more or less routine examination in TLE.

Intracerebral recordings showed that although most temporal seizures did indeed initially only involve the amygdalo-hippocampal complex (Figure 2), simultaneous onset in the hippocampus, the amygdala and the temporal pole was far to be a rare phenomenon (Figure 2b). This was consistent with the high incidence of profoundly perturbed temporal pole activity between seizures with major abnormalities of baseline activity and many slow waves and spikes which could produce a sub-continuous pattern (Figure 3). In some cases of "pauci-symptomatic" seizures, the discharges were even observed to be entirely confined to the temporal pole (Figure 4). Electrical stimulation of this region using low frequencies (1 Hz) and to greater effect high frequencies (50 Hz) was soon found to be as efficient as stimulation of the amygdala, the hippocampus or the hippocampal gyrus when it comes to inducing seizures (similar in all respects to those which occured spontaneously [40]).

Since then, large amounts of data based on brain imaging have emerged which tend to support the concept that the temporal pole is often a key region in the generation of temporal lobe seizures [41]. We will not repeat these results in detail here (see other articles in this volume) but will briefly review the main findings:

1) positron emission tomography (PET) has revealed that mesial temporal epilepsy is almost always associated with a reduced level of glucose metabolism between seizures not only in mesial temporal lobe structures but also in the pole and the anterio-lateral temporal neocortex [42]; this is probably associated with a deafferentation phenomenon. Moreover, abnormal benzodiazepine receptor density has been detected in the temporal pole [43, 44];

2) conventionally, the blood flow rate pattern during a mesial temporal seizure as recorded by single photon emission computed tomography (SPECT) is characterized by extensive temporal hyperperfusion which almost always includes the pole and is often associated with massive peripheral and/or contralateral hypoperfusion [45]; nevertheless, certain ictal discharges ­ notably polar discharges ­ are not associated with any change in blood flow rate and may even be associated with reduced blood flow, as has been documented in some cases using PET [46];

3) as mentioned above, there may be loss of substance in the temporal pole [25] but it is mainly the recent MRI demonstration of poor demarcation between white and gray matter which has brought the temporal pole to the attention of workers in the surgical management of epilepsy. This type of poor differentiation is observed in 32% to 65% of all patients with temporal lobe epilepsy associated with hippocampal sclerosis [47-49] and corresponds to abnormalities detected in the early 1990's [50] in the form of T2 hypersignals which seemed to be correlated with anterior temporal and neocortical hypometabolism.

Recent SEEG data

We conducted a retrospective review of the records of 25 consecutive patients (age: 17-43 years) with temporal lobe epilepsy and who had been operated on following an SEEG examination [51]. The mean follow-up was 26 months. The respective roles of the temporal pole and the amygdalo-hippocampal complex were investigated in all patients, by analysis of the SEEG recordings made during the most typical spontaneous seizure occurring during the recording.

In 13 of the 25 patients (the TP group, 52%) the ictal discharge involved the temporal pole either from the outset (8/13) or within the first five seconds (5/13). In the other 12 patients (the AH group, 48%), the initial ictal discharge mainly involved the hippocampus and/or the amygdala with later involvement of the temporal pole (after more than 10 seconds in most of the cases). Table I shows that there were no obvious differences between the two groups either in terms of the existence of hippocampal sclerosis or in the proportion of patients in whom surgery completely suppressed all seizures. On the other hand, more members of the AH group had a history of febrile seizures whereas abnormal MRI signals from the temporal pole seemed to be more common in the TP group.

We recently extended these results to a larger population (48 patients) (Chabardès et al., manuscript in preparation) using a similar methodology. Preliminary results confirm that the temporal pole is often (in nearly two-thirds of cases) involved in the onset of temporal lobe seizures and that the symptoms of this type of seizures do not significantly differ from those in which only the amygdalo-hippocampal complex was initially involved (apart from an earlier consciousness impairment). Febrile convulsions were not significantly more common in the AH group unless only complicated febrile convulsions were considered ­ which probably accounts for the greater proportion of patients with hippocampal sclerosis in this group. The two groups could not be clearly distinguished on the basis of MRI data although temporal pole MRI abnormalities were somewhat more common in the TP group.

Therefore, the apparently homogenous syndrome of mesio-temporal lobe epilepsy appears to cover two different entities which are difficult to distinguish by any modality other than intracerebral recordings. The existence of these different entities may explain why supra-selective surgical treatment modalities sometimes fail.

CONCLUSION

Converging anatomical, electrophysiological and functional data strongly suggest that the temporal pole can be involved at the onset of otherwise typical mesiotemporal seizures, regardless of the presence of hippocampal sclerosis. These findings raise the issue of the criteria used when deciding which of the two surgical procedures, anterior temporal resection or selective amygdalo-hippocampectomy, is the most appropriate for the so-called "mesiotemporal epilepsy". The final word should perhaps be left to Walker who, in 1967 observed that "Although the technique of temporal lobectomy varies somewhat, the goal of the different operators is similar, namely to remove the temporal pole" [54].