Proton Magnetic Resonance Spectroscopy in Kennedy Syndrome FREE

KENNEDY DISEASE (KD), or X-linked spinobulbar muscular atrophy, is a motor neuronopathy caused by an expansion of a cytosine-adenine-guanine (CAG) repeat motif in the androgen receptor gene.¹ The clinical picture of KD is distinctive. In their article published 30 years ago, Kennedy and coworkers² described the typical phenotype and the X-linked recessive pattern of inheritance. Clinical signs include weakness of the limbs, mainly with proximal distribution, and bulbar symptoms with prominent weakness and fasciculation of the facial muscles and tongue. Additional symptoms are gynecomastia, postural tremor, and diabetes in some patients. The disease typically manifests in the third to fifth decade of life. Although similarities of KD and amyotrophic lateral sclerosis (ALS) do exist, there are no clinical findings suggesting upper motor neuron involvement in KD.

The pathomechanism leading to neuronal death in KD is unknown, but probably involves the toxicity of the expanded polyglutamine tract in the androgen receptor protein.³ Pathologic protein to protein interactions with subsequent impairment of oxidative energy metabolism may be a common pathway in neuronal degeneration in CAG-repeat diseases.^{4 - 5} In Huntington disease, the hypothesis of impaired energy metabolism is further supported by in vivo proton magnetic resonance spectroscopy (¹H-MRS) studies demonstrating increased lactate (Lac) in different brain regions.^{6 - 7} As shown in ALS, proton spectroscopy provides an efficient tool to assess neuronal loss both in the primary motor cortex and the brainstem.^{8 - 9}

Noninvasive ¹H-MRS offers the opportunity to investigate changes in the metabolite composition of different brain regions. Proton metabolites detectable using this method with long echo times (TEs) of 272 milliseconds are N-acetyl (NA) groups, phosphocreatine (Cr), choline (Cho), and lactate (Lac). Since NA groups are confined to neurons and Cr is distributed relatively evenly through all cells,^{10 - 11} a reduction in the NA/Cr metabolic ratio can be explained by a reduction in levels of neuronal NA due to the degenerative process. Lactate is detectable in the brain in pathologic states associated with impaired oxidative energy metabolism such as in mitochondrial encephalomyopathies.¹²

In this study, we analyzed the relative resonance intensities of NA, Cr, Cho, and Lac in the brainstem and primary motor cortex of patients with Kennedy syndrome. The aim was to investigate whether ¹H-MRS is suitable to detect altered metabolite ratios reflecting neuronal dysfunction and/or pathologic Lac peaks indicating impaired energy metabolism. We have studied the brainstem because bulbar symptoms occur early in the clinical course of spinobulbar muscular atrophy, and overt neuronal loss has been shown by neuropathologic studies in this region. As a control region we investigated the motor cortex. This region is thought not to be involved in KD as reported in most clinical and neuropathologic studies,^{2 ,13 - 15} but recently, subtle corticospinal tract abnormalities have been reported in 2 KD cases.¹⁶

Nine male patients with KD from 6 separate kinships, including 1 sporadic case, were studied. All patients showed the typical clinical phenotype with bulbar symptoms including tongue atrophy and fasciculations. Muscular weakness and atrophy of limb muscles were present in most cases. Some patients showed additional clinical features such as gynecomastia, postural tremor, and diabetes. Clinical signs of upper motor neuron (UMN) involvement were absent. Diagnosis was confirmed in all cases by genetic testing. Clinical data are given in Table 1. There was a family history of neurodegenerative disease in family A and in the isolated case. The mothers of the probands—both obligate carriers for the disease gene—developed senile dementia in their seventh and eighth decades. The control group included 17 healthy male volunteers without any known personal history of neurodegenerative disease. The mean (SD) age of the KD group (48.1 [6.0] years) did not significantly differ from that of healthy controls (48.1 [16.7] years, P = .95). Spectroscopic data of all patients and controls are listed in Table 2. All participants provided informed consent to the study protocol.

Magnetic resonance imaging (MRI) was performed on a 1.5-T, whole-body MRI system (Gyroscan ACS-NT; Philips Medical Systems, Best, the Netherlands) using a head coil suited for MRI and ¹H-MRS. Coronal T2-weighted turbo spin echo sequences with repetition time (TR)/echo time (TE) of 2700/120 milliseconds, transaxial proton density and T2-weighted spin echo sequences with TR/TE₁ and TR/TE₂ of 2500/20 and 90 milliseconds, respectively, and multistack T1-weighted spin echo sequences with TR/TE 300/15 milliseconds were obtained for image-guided localization of the spectroscopic volume of interest (VOI). A cubic VOI of 18 × 20 × 26 mm was placed in the brainstem (pons and upper medulla), and another VOI of 40 × 30 × 25 mm was placed anterior to the central sulcus in the motor cortex and subjacent white matter (Figure 1). Proton magnetic resonance spectra were acquired with point-resolved spectroscopy volume selection and water suppression. With a TR/TE of 2000/272 milliseconds and 128-signal averages, the acquisition of each spectrum took approximately 4 minutes. Analysis of the spectra was performed with the manufacturer-supplied spectroscopy software package of the MRI system. Relative metabolite concentrations for NA, Cho, Cr, and Lac were determined by fitting the resonance line shape of each metabolite to a Lorentz curve before ascertaining the peak integral. In all patients with KD, ¹H-MRS was performed in the motor region of both hemispheres; spectra of the brainstem could not be quantified in 1 patient. In the control group, spectroscopic data were available for the motor area in all volunteers. Data acquisition for the brainstem VOI was performed in 9 controls.

Descriptive analyses were performed on NA/Cho, NA/Cr, and Cho/Cr metabolite ratios. Statistical comparison of the relative metabolite ratios in KD and controls was carried out using the Mann-Whitney U test. We chose a conservative statistical analysis using a nonparametric test because of the small sample size and because there are no data showing the normality of the distribution of metabolite ratios in a larger group of patients and controls. Correlations between age at onset (determined by age at first occurence of limb weakness or tremor, if limb weakness was absent), disease duration, CAG repeat length, and ¹H-MRS ratios were determined using Pearson product moment correlation.

Proton magnetic resonance spectra were obtained from 8 brainstem regions and 18 motor regions of 9 patients with KD. The results were compared with data obtained from 17 healthy volunteers. Spectroscopic data were available for the motor region in all controls, but data acquisition for the brainstem was performed only in 9 controls. Figure 2 shows a set of ¹H-MR spectra acquired with TE 272 milliseconds in the brainstem and the motor region comparing exemplary results from a patient with KD and a healthy volunteer. These spectra show 3 peaks corresponding to NA, Cho, and Cr. Metabolic ratios of NA/Cho, NA/Cr, Cho/Cr in KD compared with healthy controls are listed in Table 2.

In patients with KD and in normal controls, there were no pathologic Lac peaks. Patients with KD showed a significant reduction in the mean NA/Cho ratios in the brainstem (P = .01). The Cho/Cr and NA/Cr ratios did not differ significantly from those of the control group (P = .61 and .14, respectively). In the motor region of patients with KD there was a significant reduction of mean NA/Cho and NA/Cr ratios (P = .04 and .03, respectively), while Cho/Cr ratios did not differ from those of the control group (P = .48). Figure 3 shows graphically the ratios for the 3 metabolites in individual patients and controls for the brainstem and the motor region. For the NA/Cho ratio, the difference in the motor cortex seemed to result from very low ratios in 6 motor regions of 3 patients with KD. The corresponding spectra were acquired in the 2 brothers of family A and the proband of family C (Table 1). In these patients, spectrocopy findings revealed low values for NA/Cho ratios in the brainstem (Figure 3). There were high values for Cho/Cr ratios both in the motor region and brainstem, but mean Cho/Cr ratios did not differ from the controls. Clinical phenotype was not different from other patients with KD except for the occurrence of diabetes mellitus in the 2 brothers. There were markedly expanded CAG repeats (52 repeats) in the brothers, but the proband of family C showed only a moderate expansion (40 repeats).

The correlations between age at onset, disease duration, CAG repeat length, and metabolite ratios both of the brainstem and the motor region were not significant in the group of patients with KD (data not shown). There was an inverse correlation between age at onset and CAG repeat length (P = .03; r = −0.72).

To assess metabolic changes in brain metabolism in patients with KD, ¹H-MRS of the brainstem and the motor cortex was performed. In this study, we detected no pathologic Lac peaks in cases of KD, and our data failed to provide further evidence for an impaired energy metabolism in KD, a mechanism that may be involved in CAG repeat diseases.

The results demonstrate a significant reduction in NA/Cho ratios in the brainstem of patients with KD compared with healthy controls (P = .01). Since the Cho/Cr ratio is unchanged, these data suggest a decrease in the NA signal, which in water-suppressed proton spectra consists of N-acetylaspartate (NAA) and N-acetyl-aspartylglutamate (NAAG). Using long TEs as in our study, the signals of NAA and NAAG cannot be distinguished; both contribute to the NA signal. With data acquisition in the cortex, NA signal mainly results from NAA because the concentration of NAAG is only about 10% that of NAA. In the brainstem, their relative proportions are inverse.^{17 - 18} Since immunocytochemical studies have shown that these NA metabolites are localized in neurons and their processes,¹⁰ reduced NA levels in spectroscopic studies suggest neuronal loss or dysfunction.

In ALS, previous studies using ¹H-MRS provided evidence of an altered pattern of metabolites in the cerebral cortex and brainstem.^{8 - 9} The hallmark of ALS is the progressive degeneration of motor neurons in the cortex, brainstem, and spinal cord. The reports on decreased NA/Cr and NA/Cho ratios suggest that ¹H-MRS is a promising tool to quantify region-specific neuronal loss or dysfunction. Our finding of altered metabolite ratios in the brainstem of patients with KD can be explained by reduced NA concentrations reflecting neuronal impairment. Degeneration of nuclei of cranial nerves V, VII, and XII, along with degeneration of anterior horn cells and accompanying astrocytosis, has been described in autopsy studies.^{2 ,13 - 15} The spectroscopic findings of reduced brainstem NA correlate with the known pathology of KD.

We cannot exclude the possibility that the decrease in the NA/Cho ratio is in part related to an accompanying increase in choline-containing compounds in some patients. Although mean Cho/Cr ratios are not different in KD, there are some patients with high Cho/Cr ratios in the brainstem, as shown in Figure 3. Elevation of Cho levels have been described in ¹H-MRS studies in ALS and Alzheimer disease.^{8 ,19} This finding is commonly explained by membrane damage or by glial proliferation. Reactive gliosis in the brainstem motor nuclei is a well-known pathologic feature in KD.¹³ Additionally, there are reports on changes in the myelinated pathways in the brainstem²⁰ and on macrophage permeation of the pyramidal tracts in the medulla.¹⁶

In Huntington disease, neuropathologic studies have shown a correlation between the CAG repeat number and extent of postmortem pathologic features.^{21 - 22} Imaging studies in Huntington disease using positron emission tomography document this correlation between the number of repeat motifs and the extent of pathologic findings in vivo.²³ In KD there is some debate whether repeat length influences clinical phenotype, as indicated by controversial findings regarding the correlation between CAG repeat number and onset of muscle weakness.^{24 - 26} We found an inverse correlation between CAG number and age at onset. Correlation analysis revealed no linear relation between CAG repeat length, age at onset, disease duration, and metabolite ratios. One possible explanation for this result is the small sample of patients and the fact that they were not selected specifically for anticipated brainstem pathologic features.

The finding of a significant decrease of mean NA/Cho and NA/Cr ratios in the motor region of the KD group deserves further consideration. No patient showed clinical signs of upper motor neuron involvement such as spasticity or exaggerated deep tendon reflexes. Clinical signs of upper motor neuron involvement have not been reported in KD except in 1 report of a family with hyperactive tendon jerks and extensor plantar responses otherwise resembling KD.²⁰ Postmortem examination of 1 affected member of this family showed mild changes in the myelinated pathways in the brainstem. Apart from a recent report of Shaw and coworkers,¹⁶ the published autopsy studies report no involvement of the cortex and corticospinal tracts in KD.^{13 - 15} One of the 2 autopsied patients described by Shaw and coworkers had developed presenile dementia. Pathologic examination revealed neuronal loss and gliosis in the hippocampus and in the prefrontal subcortical white matter. Subtle corticospinal tract involvement was identified by immunocytochemical studies in both of these patients. Clinical signs of upper motor neuron involvement were absent on clinical examination in both cases.

In our spectroscopic study, the reduction of mean ratios in the motor cortex of the patient group is due to very low NA/Cho ratios in 2 brothers of family A and the proband of family C, as shown in Figure 3. N-acetyl–Cho ratios in these patients of 1.6 to 2.0 have not been found in the control group and fall below the lower 99% confidence intervals of controls. These patients had the highest Cho/Cr ratios of the KD group, exceeding the upper margin of 99% confidence intervals of controls. The finding of a decrease in NA/Cho ratios may be due to reduced NA and/or elevated Cho levels. Because the mean NA/Cr ratio is statistically significantly reduced in the motor region of patients with KD, a decrease in NA is probable. The finding of high Cho/Cr ratios in the 3 patients may indicate an additional elevation of the Choline signal. A release of Cho phosphoglycerides, which are constituents of membrane lipids and myelin lipids, may contribute to a stronger Cho signal. Another possible reason for elevated Cho levels is an increase in cell membrane material due to reactive gliosis. Our spectroscopic findings may reflect neuronal dysfunction and gliosis in the motor cortex and adjacent white matter. We suggest subclinical involvement of the motor region in a subset of patients with KD. Our ¹H-MRS data complement the recent autopsy results of Shaw and coworkers.¹⁶

The pathomechanism leading to the postulated motor area involvement in some KD cases is unclear. Both brothers of family A showed a marked expansion of CAG repeats (52 triplets), but in the single patient of family C, CAG repeat length was only slightly expanded (40 triplets), and there is no correlation between repeat length and ¹H-MRS data in the whole group. The clinical phenotype of these patients is similar to the other patients except for the occurrence of diabetes mellitus in the brothers. On neurologic examination, signs of dementia were absent in all patients, but the mother of the 2 brothers, an obligate carrier of the disease gene, suffered from senile dementia. Other environmental or genetic factors contributing to the risk of motor region involvement in KD may exist. One possible factor is the influence of modifying genes. The occurrence of diabetes and the family history of dementia in the brothers could support this hypothesis. In a recent phenotype/genotype study in KD, no correlation between CAG repeat length and the probability of concomitant diabetes mellitus was reported.²⁷ This leads the authors to postulate that factors other than CAG repeat length may influence the phenotype in KD.

The androgen receptor gene is widely expressed in neurons of the frontal and parietal cortex, as shown by immunocytochemical studies.¹⁵ One can only speculate that modifying genes may render neurons in regions that are normally not involved in disease pathology, like the motor region, more susceptible to the gene effect of the trinucleotide repeat mutation in KD. This may lead to neuronal degeneration and subsequent development of reactive gliosis.

Proton magnetic resonance spectroscopy is a noninvasive tool to assess metabolic changes in Kennedy syndrome. The lack of a pathologic Lac peak does not provide evidence for impaired energy metabolism in this disease. Alterations in metabolic ratios suggest a reduction in NA, which reflects neuronal loss or dysfunction in the brainstem of patients with Kennedy syndrome. The findings of abnormal ratios in the motor cortex, a region that is clinically not involved, may indicate subclinical involvement of upper motor neurons and reactive gliosis in a subset of patients.

Accepted for publication April 23, 1999.

We wish to thank all the patients and the volunteers for their cooperation in this study. We thank Edith Disput for preparing the photographs.

Reprints: Jochen Karitzky, MD, Department of Neurology, University of Ulm, Oberer Eselsberg 45, D-89081 Ulm, Germany (email: jkaritzky@aol.com).