Hypothalamic hamartomas and gelastic epilepsy

Hypothalamic hamartomas are rare malformations that usually take the form of nodules attached to the tuber cinereum or mamillary bodies. They resemble gray matter, with varying proportions of neurons, glia, and myelinated fibrous bundles. The neurons may resemble normal hypothalamic neurons, containing neurosecretory granules, or they may be dysplastic.^1,2

These rare lesions may be asymptomatic or may be associated with precocious puberty, with a remarkable epileptic syndrome, or both. The hallmark of the epileptic syndrome is the gelastic seizure-a brief, repetitive, stereotyped attack of laughter-which begins in early childhood, often in the neonatal period. Later in the first decade, a generalized epileptic encephalopathy develops, characterized by tonic, atonic, and other seizure types in association with a slow spike-and-wave discharge and cognitive deterioration.^3,4

In the past it was hypothesized that gelastic seizures originated in the temporal lobe (TL).⁵ Some patients with hypothalamic hamartomas may have focal epileptiform discharges in the temporal or frontotemporal regions on scalp EEG. Ictal surface and depth recordings in the past have suggested origin in the anterior-mesial temporal region.^5,6 However, focal cortical resection has been uniformly unsuccessful in these patients, whereas resection or radiofrequency lesioning of the hamartomas themselves have yielded good results.^7,8 In addition, ictal SPECT performed during typical gelastic seizures has demonstrated hyperperfusion in the hamartoma and adjacent hypothalamus.^7-9 Ictal depth electrode recording from the hamartoma has shown that gelastic attacks are linked strictly to ictal discharges confined to the hamartoma.¹⁰

Proton magnetic resonance spectroscopic imaging (MRSI) is a highly sensitive method of detecting neuronal dysfunction in the TLs of patients with temporal lobe epilepsy (TLE).^10-14 If patients with gelastic epilepsy have seizures secondary to neuronal damage in their TL structures, one might expect to be able to detect these abnormalities via MRSI. We studied five patients with hypothalamic hamartomas and gelastic seizures using MRSI to quantify the amount of neuronal dysfunction in the TLs and in the hamartomas themselves.

Methods. Selection of patients. Five patients with hypothalamic hamartomas and medically intractable epilepsy were evaluated at the Montreal Neurological Hospital. Studies included neurologic examination, routine and extended video and EEG monitoring, and neuropsychological studies. Normal control subjects were healthy students and laboratory personnel in good health.

Proton magnetic resonance spectroscopic imaging. We performed proton MRSI a using 1.5-T combined imaging and spectroscopy system (Philips Medical Systems, Best, The Netherlands). After making scout images in the axial and sagittal planes, a multislice spin echo MRI (repetition time [TR], 2,000 msec; echo time [TE], 30 msec) was obtained along the axis of the TL. A large region of interest was defined including both TLs and the hamartoma, and excluding bone. Water-suppressed proton MRSI was obtained from that region using a TR of 2,000 msec, a TE of 272 msec, with a 250 × 250-mm field of view (FOV), and 16 × 16 phase encoding profiles. After zero filling the latter to 32 × 32 profiles, water-suppressed MRSI was divided by the nonwater-suppressed MRSI value to correct for artifacts resulting from magnetic field inhomogeneity. The resulting time domain signal was left shifted and subtracted from itself to improve water suppression.¹⁵ This reduces the amplitude of creatine (Cr) and of water, and thus increases the ratio of N-acetylaspartate (NAA) to Cr. The signals were then filtered mildly and Fourier transformed in three dimensions.

The value of NAA/Cr was determined for an average spectrum obtained from voxels located in the TL, and for an average spectrum obtained from the hamartoma. We defined TL arbitrarily as the region between the plane of the head of the hippocampus and the plane of the splenium of the corpus callosum. In normal subjects, NAA/Cr was determined for an average spectrum obtained from voxels located in the hypothalamus, defined arbitrarily as those immediately adjacent to the third ventricle bilaterally in our FOV.

A normalized measure of asymmetry of NAA/Cr between the two TLs was defined as the difference between the right and left sides divided by the mean of the two sides: (right - left)/[(right + left)/2]. A ratio of more than 2 SDs from the mean of normal control subjects was considered abnormal.

Statistics. Student's t-test was used for statistical analysis of the MRSI data. Statistical significance was considered to be present for p < 0.05.

Ethics. Informed consent was obtained from all patients and normal control subjects. This study is part of a project for MRSI investigation of patients with partial seizures that has been approved by the Research and Ethics Committee of the Montreal Neurological Institute and Hospital.

Results. The clinical findings are summarized in table 1. Laughing seizures were the presenting symptom in all patients. The parents described brief stereotyped episodes of giggling occurring many times daily. Initially these attacks of laughter were described as pleasant, leading to a delay in diagnosis. Later in the first decade they took on a more mechanical, noncontagious, and nonhumorous character.

Multiple additional seizure types developed between the ages of 3 and 10 years. Patient 1 had complex partial seizures with head turning to the left, followed by staring and lip smacking. Patients 2 and 3 had cephalic warnings. During prolonged seizures all patients could experience unilateral clonic jerking of face and neck muscles. Autonomic features such as facial flushing, pupillary dilatation, and urinary incontinence were noted frequently. Patients 2, 3, and 4 had secondary generalized tonic-clonic attacks. Patient 2 had occasional myoclonic seizures culminating in generalized tonic-clonic convulsions. Drop attacks developed in Patients 4 and 5, and a callosotomy was eventually performed in Patient 4 without significant improvement. Brief attacks of laughter continued in all patients, although none reported being amused. All were treated with multiple anticonvulsants and none remain well controlled.

Electroencephalography. Findings are summarized in table 1. Interictal records showed occasional focal epileptic discharges in the TLs of Patients 1, 2, 3, and 5. Ictal recording from surface EEGs showed no changes in Patient 4, and generalized attenuation of background activity in the others. Chronic depth electrode recording from the hamartoma was performed in Patient 1. Interictal epileptic abnormalities were recorded from both the hamartoma and the left TL. Ictal recordings demonstrated seizure onset in the hamartoma with generalization and left-hemispheric predominance.

Magnetic resonance imaging. The studies revealed sessile hypothalamic masses in all patients (figure 1). The lesions slightly distorted the third ventricle and extended into the interpeduncular cistern.

• Download figure
• Open in new tab
• Download powerpoint

Figure 1. Transverse T1-weighted MR image shows the hypothalamic hamartomas in four patients (Patient 5 not shown). The arrows point to the hamartoma in each image.

Magnetic resonance spectroscopic imaging. The patients and four normal control subjects were studied (table 2). as a group, the patients' TL NAA/Cr and normalized asymmetry for NAA/Cr-in other words, (right - left)/[(right + left)/2]-were not significantly statistically different from those of the normal control subjects. Analysis of spectra taken from the hamartomas themselves revealed the presence of an NAA resonance intensity signal, and NAA/Cr was decreased in the hamartomas compared with the control subjects (t = 6.6, p < 0.001). Figure 2 shows an example of spectra obtained from Patient 4.

• Download figure
• Open in new tab
• Download powerpoint

Figure 2. MRI and MR spectroscopic imaging results from Patient 4. Each spectrum below the image represents an average of the spectra recorded from a volume of brain outlined on the image directly above. A and C are temporal lobe spectra. Spectrum B is measured from the hamartoma. This illustrates the symmetry of N-acetylaspartate (NAA) to creatine (Cr) from each temporal lobe as well as the lower NAA/Cr ratio from the hamartoma.

Discussion. We studied five patients with hypothalamic hamartomas and the typical evolution of the epileptic syndrome.^3,4,6-8 MRSI revealed normal and symmetric TL NAA/Cr in all patients compared with control subjects. However, the NAA/Cr measured from the hamartoma was reduced compared with the hypothalami of normal control subjects.

It is becoming increasingly clear that seizures in these patients do not arise from TL structures, although surface ictal and interictal recordings, and even depth electrode recordings, may suggest a TL origin.⁶ However, recent studies have presented evidence that gelastic seizures can originate in the hamartoma itself. Depth electrodes placed within hamartomas have revealed ictal activity arising from these structures, which correlates with clinical laughing attacks in our Patient 1 as well as in other reported patients.^7,10,16 Ictal SPECT has also shown focal hyperperfusion in hamartomas associated with clinical seizure onset.^7,9

Surgical experience supports these findings. Temporal resections based on surface EEG recordings have produced uniformly poor outcomes.⁶ In contrast, removal of the hamartomas,^8,17,18 or stereotactic radiofrequency lesioning,⁷ has resulted in good clinical outcome in these patients after limited follow-up. One surgical series⁸ suggested that the location of the hamartoma with respect to its neighboring structures and its shape determined the clinical phenotype; epilepsy was associated only with hamartomas that were sessile and compressed the thalamus, rather than with pedunculated lesions. It is tempting to speculate that lesions might influence normal corticothalamic loops to cause slow spike-and-wave discharge associated with multiple seizure types.^19,20 An Alternative mechanism, however, is secondary epileptogenesis related to the discharges in the hamartoma.³ This could explain the generalized Lennox-Gastautlike syndrome that develops in patients late in the first decade of life.^19,20

This study provides further evidence that the seizures of these patients do not arise in their TLs. Proton MRSI is a highly sensitive method of detecting neuronal dysfunction in the TLs of patients with TLE. In a recent study of 100 consecutive patients with TLE, all had abnormal NAA/Cr (defined as 2 SDs below a group of control subjects) in at least one TL.^11,21 In some patients this occurred despite the absence of identifiable lesions, such as mesial temporal atrophy, on MRI. Although MRSI is not a direct indicator of epileptogenicity, given previous experience with TLE, the absence of detectable abnormality in the TL in patients with medically intractable epilepsy provides evidence that the epileptogenic area is not in that region. The normal and symmetric NAA/Cr in our patients with hypothalamic hamartomas therefore provides further proof that their seizures do not originate in the TLs, and thus this syndrome does not represent TLE.

This is the first report of spectroscopic imaging from hypothalamic hamartomas. Attempting to recover spectra from hypothalamic hamartomas is fraught with several difficulties. First, the size of the hamartoma is small and is at the limit of resolution of the technique. Second, the location of the hamartoma is near the bony structures at the base of the skull. Spectra from this region are therefore particularly prone to contamination from bone and lipid artifacts. Finally, there is no perfectly analogous region in the normal brain that can serve as a control. Nevertheless, the reduced NAA/Cr compared with the hypothalami of normal control subjects implies that the hamartomas contain neurons that are dysfunctional compared with normal hypothalamus.

Pathologic findings in hypothalamic hamartomas from patients without epilepsy are usually reported to reveal normal neurons, myelinated fibers, and glia.^1,22 In patients with hypothalamic hamartomas and epilepsy, both normal neurons and dysplastic neurons are described.^4,23 The epileptogenicity of these hamartomas may be because of the aberrant connectivity of neuronal structures or because of the intrinsic epileptogenicity of dysplastic neurons.²⁴ Intrinsic epileptogenicity is well documented in cortical dysplasia both in vitro and by electrocorticography.^24,25 Hypothalamic hamartomas may behave quite similarly. Both have also been demonstrated to show focal hyperperfusion during ictal SPECT.²⁶ As the resolution of spectroscopic imaging improves, future work correlating various components of hamartomas with epilepsy are likely to lead to improved understanding of the epileptogenicity of these lesions.