Deep EEG Recordings of the Mammillary Body in Epilepsy Patients


Address correspondence and reprint requests to Dr. K. van Rijckevorsel at Department of Neurology, Saint Luc University Hospital, Université Catholique de Louvain, Av. Hippocrate 10, 1200 Brussels, Belgium. E-mail:


Summary: Purpose: To our knowledge, the epileptic and nonepileptic electroencephalographic (EEG) discharges recorded within the human mammillary body (MB) and mammillothalamic tract (MTT) areas have never been published. Herein, we present the EEG recordings from these structures in patients with refractory epilepsy (RE).

Methods: Three men (ages 41–43 years) were enrolled in a clinical trial for deep brain stimulation (DBS) of MB-MTT in RE. Previous evaluations had demonstrated a low likelihood of successful response to medication or resective surgery. DBS macroelectrodes were bilaterally implanted within the MB-MTT under general anesthesia and their location checked by magnetic resonance imaging (MRI). We obtained a surface-depth EEG for a 2- to 4-day period, including monitoring of the cardiorespiratory and mnemonic functions.

Results: The background pattern of EEG recorded from MB-MTT was low-amplitude (usually <25 μv for MB and <20 μv for MTT) waves with a variable combination of theta–beta rhythms. In two patients, pseudoperiodic slow spikes were unilaterally recorded with or without clinical signs. For one patient, several focal ictal discharges were recorded in the right MB without scalp EEG changes.

Conclusions: The analysis of our depth EEG revealed that the theta–beta pattern represents the predominant physiologic profile of MB. Paroxysmal epileptiform discharges can be observed in human MB. These data supplement those available from animal observations.

For five decades, a few clinical human studies, supported by experimental animal studies, have conceptualized the “pathophysiologic role” of subcortical neuronal centers in seizure propagation and perhaps initiation (1,2). This role also was supported indirectly by the noteworthy anticonvulsant effect of electrical stimulation or surgical lesions of certain subcortical centers in animals and humans (3–6). Electric deep brain stimulation (DBS) now is under evaluation as a possible therapy for refractory epilepsy. DBS has the advantage of being reversible and adjustable (7). Targeted structures in patients with refractory epilepsy (RE) have included anterior thalamic nucleus (ATN), subthalamic nucleus, and amygdalohippocampus (AHC) (6). The search for other promising targets for DBS continues.

Mirski and coworkers (3,4) reported a significant anticonvulsant effect of electrical stimulation of MB. Similar anticonvulsant effects could be obtained by transection of the mammillothalamic tract (MTT) in guinea-pig seizure models. Uptake of [14C]2-deoxyglucose within the MB and their connections showed significant activation of the MB and MTT in animals infused with pentylenetetrazol (8). With this experimental evidence as a base, our epilepsy group conducted a double-blind, crossover pilot study to examine the potential therapeutic efficacy of DBS of MB-MTT in patients with RE (9). During the prerandomization period, a surface-depth EEG recording was obtained for each patient.

Here we present the neurophysiological findings of this prerandomization period. The MB-MTT EEG is reported in epilepsy patients for the first time and discussed in the light of recent human and animal literature. Furthermore, our results reemphasize the hypothesis that deep neuronal centers can sustain epileptiform activity.


Three men (ages 41–43 years) with RE were enrolled in our study of DBS of MB-MTT after obtaining written informed consent from them, as well as approval from the ethical committee of our hospital. Seizures were related to a hypothalamic hamartoma in the first one, and they were cryptogenic in the others. Because of the risk or inapplicability of resective or disconnecting surgery, vagus nerve stimulation (VNS) was implanted in each of them, but with insufficient therapeutic outcome. The study protocol, the surgical procedure, and safety and efficacy data of this study will be reported elsewhere.

Stimulation macroelectrodes no. 3389 with four contacts, which were used for depth-EEG recording as well (contact 0, 1.5 mm, and intercontacts space, 0.5 mm, DBS; Medtronic Inc., Minneapolis, MN, U.S.A.) were implanted bilaterally and medially within the MB-MTT area. The deepest contact was placed within the MB, whereas the others were positioned within the MTT area. The postoperative magnetic resonance imaging (MRI) confirmed the correct placement of electrodes.

Surface-depth video-EEG was performed for 2 to 4 days. This period corresponds to the prerandomization phase of the study that begins with the day of electrode implantation and ends with the day of electrode and battery internalization. This was done with careful monitoring of the cardiorespiratory and mnemonic functions.

Surface EEG was obtained according to the 10–20 international system with linked ear reference. Depth EEG was recorded with a bipolar montage. A filter of 0–70 Hz and a sampling rate of 400 samples/s were used in surface as well as in depth recordings (Grass Telefactor Twin 3.3.30 Digital EEG-video).


Case 1

This man, born in 1961, has had RE since 1966. The etiology of his epilepsy was considered to be cryptogenic, with frontal hypometabolism on fluorodeoxyglucose–positron emission tomography (FDG-PET) scan. However, the preimplantation MRI demonstrated a small hypothalamic hamartoma. His seizures were characterized by a facial tonic spasm and tonic–dystonic elevation of the right arm, or by prolonged loss of consciousness and complex automatisms. Scalp EEG showed diffuse slow spike-and-wave discharges or low-amplitude fast activity. He was implanted in May 2003, after the failure of VNS therapy.

EEG recorded from MB-MTT was mostly asymmetric, with diffusion of scalp epileptiform activity to the left MTT and pseudoperiodic slow spikes within the right MTT electrodes after a clinical seizure or a seizure with subtle clinical signs (biting movements) (Fig. 1).

Figure 1.

In all EEG traces, 1 s is represented between two vertical lines. R, right; L, left; 1, mammillary body (MB); 2–4, mammillothalamic tract (MTT). Patient 1. Pseudoperiodic spikes on the right MTT without scalp EEG changes. During these discharges, the patient had tonic contractions of jaws translated by muscle artifact on scalp EEG.

Case 2

In this man, born in 1960, seizures developed in 1997, when he had an episode of prolonged cryptogenic convulsive status epilepticus. Afterward, refractory partial seizures developed with automatisms and occasional secondary generalization, with left scalp EEG discharges (Fig. 2A). MRI was considered unremarkable, and FDG-PET scan showed right mesiotemporal hypometabolism. The patient was not considered to be a good candidate for resective or interruptive surgery. After VNS failure, the patient was implanted in July 2003.

Figure 2.

Figure 2.

A: Patient 2. Usual seizure with automatisms and alteration of consciousness. During these seizures, left temporal rhythmic theta discharges were recorded. B: Patient 2. On right mammillary body (MB) (R1), several ictal discharges have been recorded, starting with fast activity of ∼20 Hz, followed by repetitive spikes. These discharges remained localized in the right MB, except if they had a duration >100 s.

Figure 2.

Figure 2.

A: Patient 2. Usual seizure with automatisms and alteration of consciousness. During these seizures, left temporal rhythmic theta discharges were recorded. B: Patient 2. On right mammillary body (MB) (R1), several ictal discharges have been recorded, starting with fast activity of ∼20 Hz, followed by repetitive spikes. These discharges remained localized in the right MB, except if they had a duration >100 s.

Several epileptiform discharges were recorded within the right MB, starting as low-voltage fast discharge (20 Hz) followed by repetitive spikes of progressive higher voltage, and then by low-voltage slow waves (Fig. 2B). We recorded 33 ictal discharges within the right MB only, with a duration ranging between 4 and 108 s. When these discharges exhibited a duration >100 s, the epileptiform activity spread to the right MTT, and then to the left side, but without scalp EEG changes. At that time, the patient became confused. Additionally, several isolated spikes were recorded in the right MB.

Case 3

This male patient, born in 1964, had tonic seizures since 1972. They started as reflex tooth-brushing seizures, and then they began to occur spontaneously. Seizures were brief, without loss of consciousness, and more frequent during sleep. Seizures presented as a tonic–dystonic posture of the arms during wakefulness and hypermotor seizures during sleep. Preoperative MRI showed an atypical scar on the left parietal cortex. The FDG-PET scan showed left parietal and mesial temporal hypometabolism. The patient was implanted in November of 2003, after VNS failure.

EEG during a typical seizure was recorded from the MB-MTT electrodes, showing 10 to 15 min of rhythmic delta waves on the left electrode, and occasional pseudoperiodic slow spikes from the right MTT (Fig. 3).

Figure 3.

Patient 3. Just after a seizure, you may have rhythmic delta on the left depth electrode for minutes and pseudoperiodic slow spike on the right mammillothalamic tract (MTT) adjacent to the mammillary body (MB) (R2).


MBs are a component of the posterior hypothalamus and are anatomically divided in two main groups of nuclei, medial and lateral (10). Both groups have two main anatomic projections, one with the anterior thalamus via the MTT and the other with the tegmentum via the mammillotegmental tract (10). Animal experiments have recorded theta rhythms within MB neurons. Theta rhythms also can be identified in other components of the Papez circuit, such AHC and ATN (11). This rhythm is believed to play a role in spatial navigation and cognitive functions (10,12). These observations in animals contrast with our findings in patients, showing a predominant electrophysiological pattern recorded in MB-MTT, during the different sleep–wake phases, of low-amplitude mixed beta-theta rhythms. In humans, the theta rhythm was not often recorded from structures interconnected with MB, such as the hippocampus formation (12). Other human studies demonstrated that the appearance of theta activity is a task-dependent phenomenon (12). Furthermore, this background rhythm was modified by spontaneous seizures, as shown in two of our patients. In patient 3, delta rhythms were recorded for several minutes after a seizure, without scalp EEG changes and after the clinical recovery. Pseudoperiodic slow spikes were recorded in the left MTT with or without clinical expression (patient 1), or after a typical seizure in the right MTT (patient 3).

Because of the small number of patients and the great differences of methods applied in animal studies and ours, no clear physiological conclusions or comparisons can be made at this moment (13,14).

It has been well established that seizures originate from the cortex. Later, some clinical and experimental studies shed light on the role played by subcortical neuronal structures in seizure development (1,2). These studies did not provide clear evidence that seizures can develop within these structures. Some reports of depth EEG in patients with “petit mal” (absence) have disclosed the considerable role of the thalamus in maintenance and perhaps initiation of 3-Hz spike-and-wave discharges (1). Conversely, many reports recorded ictal activities within the AHC in temporal lobe epilepsy patients, but the origin of these activities is debatable. This is because of the unreliable anatomic barriers separating the AHC from the adjacent cortex. In our study, we were able to record several epileptiform discharges starting with low-voltage fast activity (15) and limited to the right MB in patient 2, whereas surface EEG remained unchanged (Fig. 2B). Only during prolonged discharges did epileptiform activity spread to the ipsilateral MTT, and later to the contralateral MB-MTT area. At that moment, the patient became confused and agitated. The recording of low-voltage fast patterns in a structure usually indicates that this structure has an important relation with a seizure focus (15). Therefore, the recording of these particular ictal discharges within a structure such the MB, which is clearly separated from the cortex and with limited anatomic connections (10), makes their origin less debatable than the ones recorded within the AHC. Nevertheless, the epileptogenic focus may not be confined to the MB, only because depth EEG was not concomitantly recorded from other structures correlated anatomically and functionally to the MB-MTT, such as the AHC, the cingulum gyrus, and the midbrain (10). Finally, this finding provides robust evidence supporting the possibility that subcortical neuronal structures can sustain ictal epileptiform activity in patients. This observation might provide a rational justification for the surgical failure of mesial temporal lobectomy in some cases. Efficacy and safety of posterior hypothalamic stimulation, and its role in relation to stimulation of other brain structures, remains to be determined.