Hypothalamic Hamartomas and Seizures: Distinct Natural History of Isolated and Pallister-Hall Syndrome Cases


Address correspondence and reprint requests to Dr. W.H. Theodore at Clinical Epilepsy Section National Institute of Neurological Disorders and Stroke, NIH, 10 Center Drive, Building 10, Room 5N250, MSC 1408, Bethesda, MD 20892-1408, U.S.A. E-mail: Theodorw@ninds.nih.gov


Summary: Purpose: Hypothalamic hamartomas (HHs) have been associated with uncontrolled seizures, and aggressive therapy including surgery is often recommended. However, some patients, particularly those with other findings associated with Pallister-Hall syndrome (PHS), have a more benign course.

Methods: Thirty-seven of 40 PHS patients and 16 of 16 patients with isolated HH had a lesion confirmed on magnetic resonance imaging (MRI). Records for all patients were reviewed for the following information: presence of seizures, age at seizure onset, seizure type, seizure frequency, number of antiepileptic medications (AEDs) at the time of evaluation, past AEDs, MRI characteristics of the HH, presence of endocrine dysfunction, and presence of developmental and behavioral problems.

Results: All isolated HH patients had a history of seizures, compared with 13 of 40 PHS patients (all PHS patients with seizures had hamartomas). In isolated HH, seizures started earlier in life, occurred more frequently, and were harder to control than those in patients with PHS. Isolated HH patients were more likely to have behavioral and developmental problems than were PHS patients. The T2 signal of the hamartoma was isointense to gray matter in the majority of PHS patients, but showed a significant increase in all but one patient with isolated HH.

Conclusions: Patients with isolated HH have a distinct clinical phenotype, showing more severe seizures and neurologic dysfunction, HH showing increased T2 signal, and are more likely to have precocious puberty. In contrast, PHS patients usually have well-controlled seizures and other endocrine disturbances than precocious puberty. Patients with HH with or without seizures should be evaluated carefully for other clinical manifestations of PHS, particularly before surgery is considered.

Hypothalamic hamartomas (HHs) are nonprogressive lesions typically located between the tuber cinereum and mammillary bodies near the floor of the third ventricle; their association with seizures is well known (1–4). Composed of gray matter–like cells, with variable proportions of fiber bundles, glia, and neurons (3), HHs may occur either as an isolated sporadic lesion (3), or in association with other anomalies, most commonly in the Pallister-Hall syndrome (PHS) (5). Brief gelastic seizures, marked only by a few seconds of facial grimacing, smiling, or laughter, generally not mirthful, are most frequently associated with HHs (3,6,7). Gelastic seizures also have been reported with a number of other brain lesions (8,9).

Patients with HHs usually are reported to have seizures refractory to treatment with antiepileptic medications (AEDs) and behavioral and developmental problems (3,4,10,11). The clinical course of patients with PHS-associated HH is less certain, as most reports focused on isolated HHs, although occasionally patients with a history of HH and polydactyly, suggestive of PHS, were included (11).

PHS is characterized by HHs, central polydactyly, and other abnormalities including imperforate anus, bifid epiglottis, and panhypopituitarism. It is inherited as an autosomal dominant, but many cases are due to spontaneous mutations (12), and is associated with frameshift mutations of the GLI3 gene (chromosomal location 7p13) that functions in the sonic hedgehog pathway (13). This pathway is a key developmental morphogen of the CNS and many other organ systems. Although the original description of PHS included six infants who died in the neonatal period with severe developmental anomalies (5), most patients have much less severe disease (12). Our experience suggested that patients with HH associated with PHS might have a more benign neurologic course than those with isolated HHs, and that clear distinction of the syndromes is important for clinical management. To test this hypothesis and to explore the relation of HHs and seizures, we compared the clinical, radiographic, and endocrine features of these two groups.


We reviewed charts of all patients seen by the Clinical Epilepsy Section with isolated HHs and all patients enrolled in the National Human Genome Research Institute PHS protocol between 1997 and 2002. Protocols for patient evaluation had been approved by the appropriate NIH Institutional Review Boards. Forty patients with PHS were identified. Their mean age was 26.7 years (range, 1–75 years; SD, ±21.6 years). Of these 40, one did not have HH, two had not undergone diagnostic imaging studies, and 37 had HH confirmed on magnetic resonance imaging (MRI). All 37 had a mutation in GLI3. For six PHS and two isolated HH patients who did not have their imaging studies at the NIH, outside films were obtained. Sixteen patients with isolated HH evaluated by the Clinical Epilepsy Section were identified. Their mean age was 6.9 years (range, 1–14 years; SD, ±4.1 years).

All records were reviewed for the following information: absence or presence of seizures; age at seizure onset (before age 1 year, between ages 1 and 10 years, and older than 10 years); seizure type; seizure frequency (at least daily, at least weekly, less frequent than weekly); number of current AEDs; number of previous AEDs tried; and the presence of endocrine abnormalities.

MRI studies including axial and sagittal T1-weighted, axial T2-weighted, axial or coronal fluid-attenuated inversion recovery (FLAIR) or both, axial T1-weighted postcontrast images of the brain parenchyma, coronal T1-weighted with and without contrast, and sagittal T1-weighted postcontrast images primary through the pituitary fossa and suprasellar area were independently reviewed by two neuroradiologists (N.P., A.P.). In five PHS and two isolated HH patients, FLAIR images were not obtained.

The hamartoma volume was calculated in 33 PHS and 12 isolated HH patients by using the ellipsoid volume formula (4/3*π*d1/2*d2/2*d3/2). The greatest diameters in the lateral, craniocaudal, and anteroposterior level were measured manually. Extension of the hamartomas to the suprasellar or hypothalamic areas or both and the mass effect on the optic chiasm, the medial temporal lobes, and the midbrain were graded (grade 1 to 3) by the neuroradiologists, who were blinded to the patient's diagnosis.

All the isolated HH and 10 of the 13 PHS patients with a history of seizures had routine interictal electroencephalography (EEG). The three who did not were not having seizures and were not on any AEDs at the time of evaluation. Five PHS and 14 isolated HH patients had 24- to 48-h video-EEG monitoring.

Neuropsychological testing (n = 16 in the PHS group; mean age, 39.81 years; SD, ±22.09; range, 10–75 years; n = 8 in the isolated HH group, mean age, 9.2 years; SD, ±3; range, 4 years 11 months to 13 years 8 months) was performed by using a separate battery of neuropsychological tests for each group according to age requirements, testability by test-battery procedures, and neuropsychological features presentation. For the PHS group, the test battery was determined by the age of the patient. Adults had the Wechsler Adult Intelligence Scale–Third Edition (WAIS-III) (14), and children were given the Wechsler Intelligence Scale for Children–Third Edition (WISC-III) (15) as part of a larger neuropsychological battery. For the isolated HH group, the Woodcock-Johnson Psycho-Educational Battery–Revised: Tests of Cognitive Ability were administered as part of a larger neuropsychological test battery (16). The Woodcock-Johnson Battery allowed standardized measurement across a wide range of cognitive skills considered relevant for determining intellectual development.

Statistical analyses were performed by using Systat (SPSS Inc., Chicago, IL, U.S.A.).



Thirteen PHS patients had a history of a seizure disorder. All of these patients had HH. Five patients had their first reported seizure before age 1 year, six between ages 1 and 10 years, and two after age 10 years. Ten PHS patients were having seizures at the time of evaluation. Six were having at least daily seizures; one, at least weekly; and three, less frequently. One of the three patients who were not having seizures at the time of evaluation had had a single presumed hypoglycemic seizure at age 8 months; one had not had a seizure since her HH was resected at age 11 months (but had panhypopituitarism); and a third patient (in her forties at the time of evaluation) had been free of seizures since childhood.

Four of the PHS patients with hamartomas had another seizure type in addition to gelastic seizures. One patient had myoclonic seizures; one, generalized tonic–clonic; one complex-partial seizures; and two siblings had a severe epilepsy syndrome consisting of frequent complex-partial and generalized seizures.

Fourteen of the 16 patients in the isolated HH group had their first seizure before age 1 year, and two between the ages of 1 and 10 years. None of the isolated HH patients was seizure free at the time of evaluation. Fourteen patients were having at least daily seizures, one was having at least weekly seizures, and one, less frequent episodes. All had other episodes in addition to gelastic seizures: 10 were having other complex-partial seizure types; four, atonic seizures; four, generalized tonic–clonic seizures; one, myoclonic seizures; and one, tonic seizures. One isolated HH patient underwent surgery elsewhere after being seen at NIH. He reportedly died 24 h after surgery.

The interictal EEG was abnormal in four of the 10 PHS patients studied: one had frequent bilateral spike–wave discharges, one had rhythmic delta activity, and two,background slowing. Two PHS patients with seizures had video-EEG telemetry; each experienced several events without EEG change. One had episodes during which he would look up from his reading and take several deep breaths, reporting “amplified hearing, leg jolts, and head chills; ” and the other, episodes of laughter and smiling. Seven isolated HH patients had diffuse background slowing and polyspikes–wave discharges on interictal EEG; six had discharges alone; one had unilateral epileptiform discharges with central predominance; one, background slowing in the delta range; and one, a normal routine 21-channel and video-EEG monitoring. Only three of 14 who had ictal telemetry showed a change from their background (that showed diffuse epileptiform discharges) during clinical events.


Five of the 13 PHS patients with a seizure history were not taking any AED at the time of evaluation; five were taking one AED, two were taking two, and one was taking three. Four PHS patients had no history of taking other AEDs; six had tried at least one other medication; one had tried four others; and two had tried five others in addition to their current therapy.

In the isolated HH group, one patient was taking caffeine for seizures; four were taking one AED; seven, two; and the other four, three at the time of evaluation. The patient taking caffeine had not tried any standard AED. One patient had not tried any other AED, and 12 had previously tried two or more AEDs (one patient had tried nine).


Results in 11 PHS patients evaluated at NIH were published previously (17). Twenty-one of the 37 PHS patients with hamartomas (including 12 of the 13 with a history of seizures and HH) and eight of 16 patients with isolated HH had endocrine abnormalities. Seven PHS patients had panhypopituitarism, and five, isolated growth hormone deficiency. Three PHS and eight isolated HH patients had precocious puberty.


Hamartoma volume was significantly greater for PHS patients than for the isolated HH patients (Table 1). A significant positive relation was found between seizure frequency and HH volume (F= 2.8, p < 0.05) for PHS patients (F= 4.5, p < 0.02). This relation did not hold for the smaller isolated HH group, although the only patient who did not have daily seizures had the smallest HH. In a logistic regression model using both groups, PHS diagnosis predicted lower seizure frequency (p < 0.01), but HH volume had no significant effect.

Table 1. Magnetic resonance imaging data in each patient group
(n = 37)
(n = 16)

p Value
  1. NS, not significant; HH, hypothalamic hamartoma; PHS, Pallister-Hall syndrome.

HH volume37 cc ± 445 cc ± 8<0.01
Optic distortion172<0.03
Midbrain distortion132NS
Temporal distortion102NS

For the entire group, a significant relation was seen between temporal distortion and HH volume: patients with left or bilateral distortion had a greater volume than those with right-sided or no distortion (F= 12.8, p < 0.001). All PHS patients and all but one isolated patient had involvement of the hypothalamus, assessed by mass effect on the floor of the third ventricle and the tuber cinereum. Although the mammillary bodies were not seen clearly because of anatomic distortion, even the smallest HH was bigger than the tuber cinereum, which was the surface of insertion. Thus almost certainly mamillary body attachment occurred. Suprasellar extension was found in all but one PHS, and three isolated HH patients evaluated. PHS patients were more likely to have distortion of the optic chiasm, nerves, or tract; however, none had clinical visual-field defects (see Table 1).

In an analysis of all patients, those with endocrine problems tended to have a greater degree of both suprasellar and hypothalamic extension and were more likely to have temporal (Pearson χ2= 9.449; p < 0.03) and optic structure distortion (Pearson χ2= 11.635, p < 0.01). These patients also tended to have a greater hamartoma volume (10 cc ± 22 vs. 44 cc ± 46; p < 0.01). In logistic regression, a diagnosis of PHS was associated with a decreased risk (p < 0.01) and greater HH volume (irrespective of diagnosis) with a higher risk (p < 0.03) of having precocious puberty. Suprasellar extension and optic, temporal, and midbrain distortion were not significantly associated with precocious puberty.

The T1 signal of the HH was isointense to gray matter in both patient groups (Figs. 1 and 2). No patient in either group had contrast enhancement. The T2 signal of the hamartoma was isointense to gray matter in the PHS patients, except four subjects imaged as infants or toddlers, who had a subtle increase in signal intensity on the FLAIR images (Figs. 1 and 3). In these four patients, the tissue of the hamartomas demonstrated heterogeneous signal intensities on the T2 and FLAIR technique. The signal changes within the hamartoma were identical to the intensities of the normal grey and the nonmyelinated white matter. The internal arrangement of these tissues, however, was irregular, reflecting the well-recognized abnormal internal architecture of the hamartomas. All but one patient with isolated HH demonstrated a significant increase in the T2 signal compared with the normal gray matter (Fig. 2).

Figure 1.

Axial fluid-attenuated inversion recovery (FLAIR) magnetic resonance (MR) image (A), coronal T1 MR image (B), and sagittal T1 postcontrast MR image (C) of hypothalamic hamartoma in a patient with Pallister-Hall syndrome.

Figure 2.

Axial fluid-attenuated inversion recovery (FLAIR) magnetic resonance (MR) image (A), coronal T1 MR image (B), and sagittal T1 postcontrast MR image (C) of hypothalamic hamartoma in a patient with isolated hypothalamic hamartoma.

Figure 3.

Axial T2 magnetic resonance image of an infant with Pallister-Hall syndrome.

Neuropsychological testing

Neuropsychological data on the isolated HH patients have been published previously (18). The isolated HH group performed poorly on the Woodcock-Johnson Battery; a broad cognitive ability standard score (75.9 ± 14.3; range, 59–98). No correlation was found between age at seizure onset and cognitive abilities, but the frequency of gelastic and complex-partial seizures did correlate negatively with the broad cognitive ability scores. Sixteen patients tested in the PHS group had a mean Weschler Intelligence Scale verbal score of 90.7 (SD, ±21; range, 40–118) and performance score of 88 ± 23.8 (range, 40–116). No correlation was found between either seizure-onset age or seizure frequency, and verbal or nonverbal abilities.


These results show that the clinical course and imaging features differ markedly in patients with isolated HH and PHS, suggesting different approaches to their management. The majority of patients with isolated HH followed the course described previously (2–4,10,19–21). Presenting early in life with gelastic seizures, they have other seizure types (including atonic seizures), behavioral and cognitive dysfunction, associated with generalized bilateral or bitemporal discharges. Seizures are frequent, difficult to control with AEDs, and do not respond to focal corticectomy or anterior mesial temporal resection (10).

In contrast, most patients with PHS and hamartomas do not have seizures; the gelastic seizures that were reported usually responded well to AEDs. Only a minority had other seizure types, and only two siblings appeared to show a “progressive” course similar to that of the isolated HH group. Only a few PHS patients (those with the most frequent and difficult-to-control seizures) had marked behavioral or developmental problems. These data support previous reports of a strong effect of seizures on cognitive and behavioral problems in patients with HH (18,21,22). Neuropsychological deficits tended to be mild in the PHS group compared with those in the isolated patients and may have been overestimated, as not all were tested. No single AED emerged as being superior in the treatment of seizures for either group.

A clear correlation was noted between lesion size and the likelihood of endocrine dysfunction. All but one of the PHS patients with seizures had endocrine dysfunction. This reinforces the previous recommendation that PHS patients with seizures should have endocrine evaluation (12).

A potential confounding variable in our series is that all isolated HH patients were severely affected, having been referred for uncontrolled seizures, possibly reflecting ascertainment bias (23,24). The PHS patients, in contrast, were referred for comprehensive evaluation to the National Human Genome Research Institute. Thus our sample was not population based. Although the PHS patients were older, the difference in age is unlikely to affect our clinical observations. PHS usually was diagnosed early in life, and isolated HH patients do not show spontaneous seizure remission.

Previous studies attempted to relate HH imaging characteristics to patients' clinical course. Individuals with sessile lesions distorting the third ventricle and involved in or enveloped by the hypothalamus had the most severe clinical course (11,25), possibly because of irritation of the mammillary bodies (11). Patients with HH attached to the third ventricle floor or suspended by a peduncle had a much more benign course (23).

In our study, all hamartomas were sessile and involved the hypothalamic parenchyma. PHS patients with seizures had larger hamartomas, but a poor correlation existed between seizure frequency and hamartoma extension. In both PHS and HH groups, the hamartomas were isointense on T1-weighted sequences, but in the isolated HH group, a strong trend was seen for the hamartomas to be hyperintense on T2-weighted sequences, whereas they were generally isointense to gray matter in the PHS group. Similar increases in T2-signal intensity have been reported in isolated HH patients (11,27). A patient with PHS had HH hyperintense to gray matter on T2-weighted sequences, but this was attributed to the patient's youth and myelination status (28). It is interesting that a patient with cognitive delay, HH, growth hormone deficiency, and hypogonadotrophic hypogonadism, but not seizures or PHS, had T2 intensity similar to that of gray matter (29).

Increasing HH size predicted higher seizure frequency in the PHS patients, but not in the HH patients, perhaps because the latter almost all had at least daily seizures. This finding suggests that the mechanisms through which HH produce seizures differ between patients with PHS and isolated HH. They may be more closely related to mass effect in the former and to intrinsic lesion characteristics in the latter. However, T2 changes are nonspecific. It is possible that the increased T2 signal in the isolated HH patients is related to early onset of a severe seizure disorder, rather than intrinsic tumor characteristics leading to epileptogenicity.

Although patients with HH usually are considered as a group, PHS has a clinically distinct neurologic phenotype from the syndrome of isolated HH. Patients with isolated HH have a distinct clinical phenotype, showing more severe seizures and neurologic dysfunction, HH showing increased T2 signal, and are more likely to have precocious puberty. In contrast, PHS patients usually have well-controlled seizures and endocrine disturbances other than precocious puberty. It is very important to evaluate carefully all children with HH for signs of PHS, both for clinical decisions and for genetic counseling. Our data suggest that intrinsic differences exist in the hamartomas between the groups. Patients with PHS-associated hamartomas may respond more readily to AEDs and rarely need surgical intervention.


Acknowledgment:  We thank Jacqueline Greenfield, Kathleen Kelley, and Pat Tyer-Reeves for their excellent technical support.