JLE

Epileptic Disorders

MENU

From hypothalamic hamartoma to cortex: what can be learnt from depth recordings and stimulation? Volume 5, issue 4, December 2003

Figures

See all figures

Auteur(s) : Philippe Kahane1, Philippe Ryvlin2, Dominique Hoffmann3, Lorella Minotti1, and Alim Louis Benabid3

1. Neurophysiopathologie de l’Epilepsie, CHU de Grenoble (France)
2. Neurologie fonctionnelle et Epileptologie & CERMEP, CHU de Lyon (France)
3. Neurochirurgie & INSERM U318, CHU de Grenoble (France)

Presented at the International Symposium on Hypothalamic Hamartoma and Epilepsy, Montreal Neurological Institute, Montreal, Canada, November 29th 2001.

In patients suffering from the epileptic syndrome associated with hypothalamic hamartoma (HH) [1], there is now strong evidence that resection or disconnection of HH, as well as radiofrequency thermocoagulation or gamma knife radiosurgery, may lead to complete seizure control, and to improvement of behavioural disturbances [2-14]. Thanks to neurosurgical advances, these therapeutic options are now associated with limited morbidity which, together with recent electrophysiological and functional neuroimaging findings, make surgical treatment a reasonable choice when compared with the relatively severe evolution of medically-treated HH associated epilepsy [1, 15].
A few years ago, however, the pathophysiology of this puzzling syndrome was poorly understood. Though some authors had discussed the possible role of the hamartoma in generating gelastic – or more rarely dacrystic – seizures [16], evidence was lacking. In fact, the electro-clinical presentation of the syndrome corresponded to that of a symptomatic generalised epileptic encephalopathy in many patients, with multiple seizure types, diffuse interictal scalp-EEG epileptiform abnormalities, progressive intellectual decline, frequent behavioural problems, and poor response to antiepileptic drugs [1, 15]. The relation between these diffuse abnormalities and the hypothalamic lesion remained elusive, in as much as isolated laughing (or crying) attacks, the hallmark feature of the syndrome, were not associated with any convincing concurrent scalp-EEG changes. In other patients, surface and intracranial EEG data suggested a temporal or a frontal lobe ictal onset, but resections of these regions consistently failed at reducing seizure frequency, while corpus callosum section only seemed to reduce the number of drop attacks in some patients [17].
At the beginning of the 90’, stereotactic intracerebral EEG (SEEG) recordings showed for the first time that gelastic seizures [18] as well as dacrystic attacks [19] correlated with ictal discharges confined to the hamartoma. SEEG recordings and stimulations in a few additional patients have confirmed this intrinsic HH epileptogenicity [5, 20], which was subsequently corroborated by ictal SPECT findings [5, 11, 21, 22] and magnetic resonance spectroscopy data [23]. However, Munari and coworkers emphasized that the other associated seizure types (i.e. without laughing or crying component) did not arise from the hypothalamic mass but originated in various cortical areas, thus suggesting that these seizures might result from secondary epileptogenesis [18-20].
To date, five patients suffering from drug-resistant seizures associated with a hypothalamic hamartoma have underwent a SEEG study at our institution. Some of these observations have been published elsewhere [18, 20, 24, 25], and we report here the main electro-clinical features of all five cases, focusing our attention on the relationships between the different seizure types, the hamartoma, and the cortex.

Patients and methods

Population studied

All pertinent clinical data are presented in Tables 1 and  2.
The five patients, one of whom had a Pallister-Hall syndrome, were suffering from a childhood onset long-lasting medically intractable epilepsy. All patients experienced gelastic and/or dacrystic attacks, whereas four of them also presented other seizure types.
MRI showed an intra-hypothalamic mass in all cases (see figures 3 and  10), the characteristics of which corresponded to those described as type IIb’ by Valdueza et al. [4]. The hamartoma was predominantly lateralized to the right side in three patients, to the left side in one, and non-lateralized in the remaining patient.
Interictal scalp-EEG recordings showed diffuse abnormalities in most patients (figure 1). Interestingly, these abnormalities predominated on the side ipsilateral to the hamartoma in the four patients where the latter clearly showed a side predominance. Ictal scalp-EEG proved inconclusive during gelastic and/or dacrystic seizures which were recorded in all but one patient (figure 2A). Conversely, asymmetric bilateral EEG changes were observed at ictal onset during the other seizure types also recorded in four of the five patients (figure 2B). As for interictal EEG abnormalities, the lateralization of ictal discharge were ipsilateral to the predominating side of HH.

Table 1General characteristics.



Pts (sex) Age at SEEG G/D szrs (onset) Other szrs (onset) Seizure frequency PP (onset) Clinical exam Cognitive decline AEDs at SEEG
1 (F) 16 yrs G (30 mths)  + (14 yrs)  > 5/day  –  small height moderate CBZ-GVG-CLB
2 (F) 19 yrs D + / – G (10 yrs)  + (5 yrs)  > 1/day  + (5 yrs) obesity moderate CBZ-PHT-CLB
3 (M) 27 yrs G- > D (2 yrs)  + (5 mths) 1-5/day  –  normal mild CBZ-PHT-GVG-PB-CLB
4 (M) 12 yrs G (< 1 yr)    > 5/day  –  normal mild CBZ
5 (F) 30 yrs G (infancy)  + (13 yrs) 30/mth  –  polydactyly none CBZ-CLB
Pts: patients; F / M = female / male; G / D szrs = gelastic / dacrystic seizures; PP: precocious puberty; AEDs: antiepileptic drugs; yr(s): year(s); mth(s): month(s); SEEG: stereotactic intracerebral EEG; CBZ: carbamazepine; GVG: vigabatrin; CLB: clobazam; PHT: phenytoïne; PB: phenobarbital.

Table 2. Video-EEG monitoring.



Pts Interictal findings Gelastic/dacrystic seizures Other types of seizures
    EEG semiology EEG Semiology
1 bursts of diffuse PSp and long-lasting sequences of bilat Sp & SpW diffuse flattening smiling and/or laughing biF low-voltage fast activity backward fall, contraction of the mouth corners, R eyes deviation, R arm hypertonia ± R or L head turning
2 bursts of diffuse PSp and R F-C PSp delayed R F SW crying ± preceded by an epigastric sensation bilat (R > L) high-voltage fast activity Staring, eyes blinking, flushing, swallowing ± preceded by a thoracic warm, auditory illusions, dizziness and pain in the L leg
        R F-C-T low-voltage fast activity same as above, followed by bradycardia, R eyes deviation, L facial contraction, bilat (L > R) hypertonia, moaning, fall
3 bursts of diffuse PSp R anterior flattening laughing leading to crying ± preceded by an epigastric sensation bilat (R > L) high-voltage fast activity sudden fall ± preceded by R oculocephalic deviation; possible secondary tonic-clonic generalisation
4 L hemispheric multifocal SW & Sp L T-F-C-P theta activity indefinable aura, flushing, laughing, gesticulating, R arm dystonia ± R body rotation; possible R hemibody hypertonia

5 R T-P SW & Sp not recorded cephalic sensation or pressure to laugh ± laughing, rarely déjà-vécu’ at onset biT(-P) flattening (R > L) eyes blinking, swallowing, gestural automatisms, L oculocephalic deviation
Pts: patients; PSp: polyspikes; Sp: spikes; SpW: spike-and-waves; SW: slow waves; R: right; L: left; bi/bilat: bilateral; T: temporal; F: frontal; C: central; P: parietal.

SEEG procedure

Intracerebral EEG recordings were performed in the first four patients with the aim to provide a three-dimensional assessment of the epileptogenic network, and to precisely define the relationships between the latter and the hamartoma. Thus, the stereotactic placement of the intracerebral electrodes, as well as their total number (n = 11 to 13), varied from one patient to another one depending on the suspected origin and areas of early seizure spread, but always targeted the HH (figure 3). Electrodes were 0.8 mm in diameter and included 5, 10, or 15 leads 2 mm in length, 1.5 mm apart (Dixi, Besançon, France), depending on the intra-cerebral target. Such multileads electrodes allow to investigate all the structures crossed along their trajectory, including the mesial and lateral aspects of the different lobes, but also fissural cortices, as well as the hamartoma.
In the fifth patient (no 5), intracerebral EEG recordings were conducted as part of a therapeutic procedure aiming at chronically stimulate the HH. The rationale of this therapeutic approach was based on our previous findings regarding the intrinsic epileptogenicity of hypothalamic hamartoma, on the potential benefit of chronic high frequency stimulation of the epileptogenic zone as suggested by the Velasco’s group in temporal lobe epilepsy patients, on the safety of deep brain stimulation in patients with movement disorders, and on the possible reversibility of the procedure. The hamartoma was investigated using three 4-leads deep-brain stimulation electrodes (Medtronic DBS lead 3389, four contacts 1.27 mm in diameter, 1.5 mm long, space by 0.5 mm) in order to sample the entire hypothalamic lesion (see figure 10). In this patient, intra-cortical EEG recordings were not performed, but scalp-EEG was recorded together with HH in order to provide information on the cortical EEG activity.

Intracerebral recordings and stimulation

In all patients, concurrent video and intracerebral EEG were monitored for 7 to 23 days (Biomedical Monitoring System, Campbell, CA, USA; and since 1996, Micromed, Treviso, Italy). We used as a reference, one of the lead located in the white matter in four patients, and the Cz scalp-EEG electrode in patient no 5. Depth EEG activity was displayed using bipolar montage between contiguous contacts. Electrical stimulations were performed under continuous video-EEG control, during sessions that last one to three hours. According to our standard clinical practice [26], and validated safety parameters [27], stimulations were performed between contiguous leads at 1 Hz (pulse width = 3 ms) and 50 Hz (pulse width = 1 ms), using a constant current rectangular pulse generator (World Precision Instruments, New Haven, CT, USA; and since 1996, Micromed, Treviso, Italy). Chronic stimulation in patient n° 5 was applied according to the parameters used in movement disorders.

Results

The main depth EEG findings are summarized in Table 3.

Interictal SEEG findings

As expected from scalp-EEG recordings, spikes and spike-and-waves discharges were recorded in the cortex in the four patients in whom cortical recordings were available. These paroxysmal discharges, though widely extensive, were usually not recorded within the hamartoma (figure 4). Conversely, the latter usually exhibited independent well-localized spikes or spike-and-waves (see also figure 10), the frequency and amplitude of which varied from one patient to another. In one patient (no 4), however, the lesion HH was almost electrically silent, without any detectable paroxysmal discharge. Interestingly, in one patient (no 2), HH spikes were influenced by sleep, where they became continuous, but disappeared during short bursts of spindle-like’ activity of unclear significance, located in the hippocampus and the amygdala (figure 5).

Table 3. SEEG findings, surgery and outcome.

Pts interictal spikes spontaneous seizures electrically-induced seizures surgery outcome (follow-up)
HH cortex G/D other types G/D other types
1  +   +  HH biF HH  –  stereotactic radiosurgery (2) no change (4) (6 years)
2  +   +  HH R F-C-T HH  –  partial removal of the HH (3) mild improvement (5 years)
3  +   +  HH R F ?  –   –   –  no change (9 years)
4 -  +   ? (CG ?) - A (Hc) - L temporal lobectomy no change (9 years)
5  +  not recorded not recorded not recorded HH  – (1) chronic HFS of the HH No change (3 years)
Pts: patients; HH: hypothalamic hamartoma; G/D: gelastic/dacrystic; bi: bilateral; R: right; L: left; F: frontal; C: central; T: temporal; CG: cingulate gyrus; A: amygdala; Hc: hippocampus; HFS: high frequency stimulation; (1): cortical stimulation not performed; (2) Lin Ac, 18 grays – no change of the hypothalamic mass on MRI suggesting that the low dose of X-rays delivered was insufficient; (3) despite two operations, the hypothalamic lesion could not be totally removed and the patient showed only a mild decrease in the frequency of seizures associated with falls; (4) patient deceased during sleep (probable SUDEP).

Gelastic and dacrystic seizures

Three of the four patients in whom gelastic or dacrystic seizures were recorded demonstrated a concurrent low voltage fast activity, followed by a spike-and-waves discharge which remained confined to the hamartoma (figure 6A). One of these patients (no 2) also presented many such HH ictal discharges during sleep, which remained clinically silent (figure 7A).
In all three patients, the HH discharges were associated with a diffuse flattening of the cortical EEG activity and a disappearance of interictal abnormalities. Subtle cortical changes (fast activity and/or rythmic slow waves or spikes) were latter observed over the two cingulate gyri in patient no 1, the right fronto-centro-temporal region in patient no 2, or the right orbito-cingulate cortex in patient no 3.
In contrast, patient no 4 showed an ictal discharge which predominated at the cortical level, with a diffuse flattening of EEG activity associated with a marked low voltage fast activity over the left cingulate gyrus, and to a lesser degree, the left hippocampus, whereas only subtle changes were recorded in the hamartoma.

Other seizure types

The three patients in whom other types of seizure were recorded showed that the latter were associated with cortical ictal discharges not affecting the HH (see figure 6B). In patient no 2, similar cortical ictal discharges were also recorded during sleep (see figure 7B), but proved clinically silent.
Ictal onset appeared either bifrontal (patient no 1), right fronto-central and lateral temporal (patient no 2), or bifrontal with a right side predominance (patient no 3). The lateralization of these ictal discharges was always ipsilateral to the predominating side of the hamartoma.

Interestingly, ictal discharges involved the cingulate gyrus in all three patients, as well as the regions which showed subtle depth EEG changes during gelastic/dacrystic seizures. In fact, the latter could precede the other seizure types, suggesting a causal relationship between the two (figure 9).

Electrical stimulations

Acute stimulation of the hamartoma and the cortex

Acute stimulation of the hamartoma could reproduce gelastic or dacrystic seizures in three of the five patients, when using high frequency parameters (50 Hz), but in only one patient when stimulating at 1 Hz. These stimulation-induced HH seizures were always associated with an epigastric sensation, which was recognized as being part of the dacrystic attacks in one patient (no 2). Other symptoms could also be induced in patient no 5, the type of which varied depending on the part of HH which was stimulated (figure 10).
Conversely, cortical stimulations, performed in two of the three patients with stimulation-induced HH seizures, failed to elicit any ictal discharge or symptom.
In one of the two patients without electrically-induced HH seizure, the stimulation of the hamartoma induced an unknown sensation of warmth over the face (no 3). In the other patient (no 4), stimulation of the amygdala could reproduce his usual and complex gelastic seizures, whereas hippocampal stimulation elicited comparable episodes but which lacked their laughing component. However, the left temporal lobectomy performed in this patient failed to control these seizures.

Chronic stimulation of the hamartoma

In patient no 5, the implantation of three intra-hamartoma electrodes was designed to chronically stimulate the HH.
The patient first underwent subacute stimulation (130 Hz-100 µs – 0.4 mA) during depth-EEG monitoring in order to evaluate the tolerability as well as the impact of the stimulation on the HH firing pattern. Interictal spikes recorded from the hamartoma and from scalp-EEG, almost totally disappeared during the stimulation, and reappeared when the stimulation was interrupted (figure 11). No side effect was reported.
These findings encouraged us to pursue with chronic stimulation which was started in November 1999. We initially decided to stimulate all the leads of the three electrodes, simultaneously, using the following parameters: 130 Hz/90 µs/0.5V, and then 185 Hz/60 µs/0.1V. The patient progressively complained of increasingly frequent cephalic auras and pressure to laugh (up to 30 per day), as well as from weight gain (from 63 kgs to 68 kgs during a 84 days period), leading us to stop the stimulation. The latter was resumed five months later, by only stimulating the electrode contact which had been stimulated during the subacute stimulation protocol. Seizure frequency, as well as seizure type, were not clearly modified over a 12 months period during which stimulation was alternatively turned off and on (figure 12), while the patient reported frequent headaches and a weight gain from 66 kgs to 72 kgs. Stimulation was then definitively stopped, allowing the weight to progressively return to 61 kgs during the following months.

Discussion

In this study, we have clearly demonstrated that epileptic seizures associated with hypothalamic hamartoma could exhibit different types of electroclinical patterns. During the most typical seizures, i.e., laughing and crying episodes, the epileptic discharges usually arose and remained confined within the hamartoma. In addition, interictal spikes were recorded from the hamartoma in the majority of patients, whereas the stimulation of the HH could reproduce gelastic or dacrystic seizures, as observed by others [5]. These findings, corroborated by the results of ictal SPECT which showed increased blood flow changes from the hypothalamic region during gelastic seizures [5, 11, 21, 22], strongly support the intrinsic epileptogenicity of hypothalamic hamartomas, in as much as its removal, disconnection, irradiation, or coagulation can control the seizures [2-14]. These findings are reminiscent of the observation that patients suffering from hemifacial spasms and cerebellar gangliogliomas demonstrate ictal discharges originating in the vincinity of the tumor, which removal results in seizure remission [28-30].
However, the HH of one of our patient failed to demonstrate spikes or either spontaneous or electrically-induced seizures, raising the issue of other ictal onset sites responsible for gelastic seizures. However, the limited number of intra-hamartoma recording leads, as compared to the size of the HH, could also explained our negative findings in that patient. This possibility was well-illustrated in another patient where three 4-leads electrodes were placed within the hamartoma, showing differences in electrically-induced clinical symptoms from one part of the HH to another (see figure 10). Alternatively, both interictal and ictal EEG findings could be consistent with the view that this patients’ gelastic seizures arose from the cortex. Accordingly, the associated ictal semiology was more rich and complex than that observed during laughing attacks in the three previous patients.
If it is now well-established that gelastic and dacrystic seizures usually originate from the hamartoma itself [31], the pathogenesis of the epileptic syndrome associated with HH is far from being fully understood. Particularly, the almost constant occurrence of seizures other than gelastic/dacrystic episodes remains puzzling, in as much as they seem to arise from various cortical areas the removal of which has consistently failed to control the epilepsy. We have no clear explanation for the mechanisms underlying these seizures, but several arguments support the hypothesis of secondary epileptogenesis. First, the epilepsy associated with HH usually starts with gelastic seizures, while the other seizure types will tend to occur later during the evolution of the ilness. Second, scalp-EEG interictal epileptiform abnormalities, which proved to be independent from those recorded in the hamartoma, usually worsen with time, particularly when non-gelastic seizures develop [21]. Third, we observed that the laughing or crying attacks were sometimes immediately followed by the other seizure types, as if ictal discharges within the hamartoma triggered those which seemed to originate in the cortex. Last but not least, data from various centres show that resection of HH, when fully achieved, can control all seizure types.
Secondary epileptogenesis in HH have also been suggested by others [32]. This phenomenon could occur as a consequence of repetitive intra-hamartoma ictal discharges, affecting neighbouring structures involved in the neuromodulation of cortical activity. In that respect, we propose a speculative pathophysiology in which the mamillo-thalamo-cingulate tract would serve as a relay of HH discharges towards the cortex, the excitability of which would then progressively increase, first leading to interictal epileptiform abnormalities and then to seizures (figure 13). This view is supported by the following observations: i) cortical discharges do not seem to occur randomly in the cortex, as showed by the lateralization of scalp EEG abnormalities and intracranially recorded seizures, which proved always ipsilateral to the lesion in those of our patients in whom the hamartoma demonstrated a side predominance; a similar concordance in lateralization was also found between interictal hypometabolism on 18FDG-PET and HH (Ryvlin et al., this issue); ii) hypothalamic hamartomas which give rise to epileptic seizures are located within the hypothalamus with tight connections with the mamillary bodies [4, 33, 34]; iii) ictal SPECT findings during gelastic seizures have showed that thalamic hyperperfusion could coexist with that observed in the HH [5], whereas thalamic hypometabolism, ipsilateral to HH, was demonstrated by interictal 18FDG-PET [Ryvlin et al., this issue]; iv) the cingulate gyrus was consistently involved during non-gelastic seizures in this series.
Whether this proposal is valid or not remains a very important issue, since it could provide arguments on the moment to discuss surgery (e.g., worsening of interictal EEG abnormalities, appearance of seizures other than gelastic/dacrystic, behavioral problems) in order to prevent the development of a definite secondary epileptogenesis. It seems likely that this latter, once established, makes more aleatory the effectiveness of the lesion elimination on both kind of seizures. n