In Vivo Voxel-Based Morphometry in Multiple System Atrophy of the Cerebellar Type

ORIGINAL CONTRIBUTION In Vivo Voxel-Based Morphometry in Multiple System Atrophy of the Cerebellar Type

Karsten Specht, PhD; Martina Minnerop, MD; Michael Abele, MD; Ju¨rgen Reul, MD; Ullrich Wu¨llner, MD; Thomas Klockgether, MD

Background: Multiple system atrophy (MSA) is a spo- mented into gray and white matter. Data were analyzed radic neurodegenerative disease. According to the clini- using statistical parametric mapping (SPM99). cal presentation a parkinsonian type and a cerebellar type (MSA-C) are distinguished. Results: Gray matter was reduced in the brainstem and the anterior lobe of the cerebellum. Reduction of white Objective: To study the morphological alterations of matter was observed in the middle cerebellar pe- MSA-C–affected brains in vivo using voxel-based mor- duncles, cerebellar white matter, and brainstem. The in- phometric analysis of magnetic resonance images. verted comparison revealed an increase of white matter density along the pyramidal tracts. Setting: University hospital. Conclusions: Voxel-based morphometry revealed a sig- Patients: Fourteen patients (5 men and 9 women) with nificant loss of cerebellar and brainstem tissue in MSA-C. MSA-C (mean age [SD], 59.4 [7.4] years; mean [SD] dis- It allowed a precise anatomical localization and a disease duration, 3.7 [1.4] years) and 13 healthy control sub- tinction between gray and white matter densities. In ad- jects (5 men and 8 women) (mean [SD] age, 55.1 [6.9] dition, our data point to a particular involvement of the years) were studied. pyramidal tract in MSA-C.

Methods: T1-weighted magnetic resonance images were normalized to a common stereotaxic space and seg- Arch Neurol. 2003;60:1431-1435

ULTIPLE SYSTEM atro- and supratentorial brain structures as well phy (MSA) is a spo- as signal abnormalities.5,7-11 To quantify the radic, adult-onset dis- atrophic changes in MSA, we previously ease characterized by used planimetric and volumetric region of neurodegeneration in interest–guided approaches. Compared Mthe basal ganglia, brainstem, cerebellum, with patients with idiopathic Parkinson and intermediolateral cell columns of disease and healthy control subjects, pa- the spinal cord.1,2 The neuropathological tients with MSA consistently had signifi- hallmark of MSA are ␣-synuclein– cant atrophy of the cerebellum, brain- positive oligodendroglial cytoplasmic in- stem, putamen, and caudate nucleus.5,7 clusions.3 Clinically, patients with MSA In morphometric MRI studies using suffered from parkinsonism, cerebellar region of interest–guided measurements, ataxia, and autonomic failure (most no- an inherent bias is introduced by select- tably orthostatic hypotension and uri- ing a limited number of brain regions for nary incontinence).4-6 According to the study. In addition, the segmentation pro- clinical presentation, a parkinsonian type cedure is often arbitrary and poorly re- and a cerebellar type of MSA (MSA-C) are producible. These problems are avoided distinguished.5,6 by using voxel-based morphometric meth- Magnetic resonance imaging (MRI) ods that allow an automated, unbiased, and From the Medical Centre Bonn (Drs Specht and Reul) and the has been extensively used to study the comprehensive assessment of anatomical Department of Neurology morphologic condition of the brain of pa- differences of gray and white matter 12,13 (Drs Minnerop, Abele, Wu¨llner, tients with MSA. Magnetic resonance im- throughout the brain. Voxel-based mor- and Klockgether), University of aging abnormalities occurring in MSA in- phometry has been recently refined and Bonn, Bonn, Germany. clude progressive atrophy of infratentorial successfully used to study structural brain

(REPRINTED) ARCH NEUROL / VOL 60, OCT 2003 WWW.ARCHNEUROL.COM 1431

Patients

With With Probable With Possible Control MSA-C MSA-C MSA-C Subjects B Variable (n = 14) (n=9) (n=5) (n = 13) Sex, M/F 5/9 3/6 2/3 5/8 Age, mean (SD), y 59.4 (7.4) 59.2 (7.8) 60.0 (6.3) 55.1 (6.9) Disease duration 3.7 (1.4) 4.1 (1.1) 3.2 (1.7) NA mean (SD), y Patients with 3 (21) 3 (33) 0 NA z = 2 mm z = 5 mm z = 8 mm parkinsonism Patients with urinary 13 (93) 8 (89) 5 (100) NA incontinence Patients with 13 (93) 8 (89) 5 (100) NA orthostatic hypotension Patients with 2 (14) 1 (11) 1 (20) NA pyramidal signs z = 11 mm z = 14 mm z = 17 mm Abbreviations: MSA-C, multiple system atrophy-cerebellar type; NA, not applicable. *The diagnosis of probable and possible MSA-C was made according to the clinical criteria of Gilman et al.6 Data are given as the number (percentage) of patients unless otherwise indicated.

correlates of aging13,14 and changes of gray and white matter volumes in neurodegenerative diseases.14-19 In this L z = 29 mm z = 32 mm z = 35 mm R

study, we used voxel-based morphometry to study the T-Value morphology of the MSA-affected brain in vivo. To increase the homogeneity of our patient group we re- 6 stricted our analysis to those with MSA-C.5,6 4

2 METHODS 0 z = 20 mm z = 23 mm z = 26 mm PATIENTS

The study was performed in 14 consecutive patients (5 men and 9 women) with MSA-C whose mean (SD) age was 59.4 (7.4) years and mean (SD) disease duration was 3.7 (1.4) years compared with 13 healthy controls (5 men and 8 women) whose mean (SD) age was 55.1 (6.9) years. A diagnosis of MSA-C was made according to the criteria of Gilman et al.6 Alternative causes of cerebellar ataxia were excluded by history, conventional MRI, genetic testing for spinocerebellar ataxia mutations, cerebro- Figure 1. Gray matter reduction in multiple system atrophy (MSA). spinal fluid studies, by laboratory tests including the levels of A, Reduction of white matter density is shown on the 4-color views of the brain. Morphometric analysis was done in 14 patients (5 men and 9 women) antineuronal antibodies, vitamin B12, vitamin E, and thyroid hor- with MSA compared with 13 healthy control subjects (5 men and 8 women), mones, and by the results of lipid electrophoresis and VDRL. using a corrected threshold of PϽ.05. Gray matter is lost within the upper Of 14 patients, 9 met the clinical criteria for probable MSA-C parts of the cerebellum. B, Reduction of gray matter density is shown on the and 5 for possible MSA-C (Table 1). The study was approved 4-color views of the brain and on 12 consecutive axial T1-weighted images by the ethics committee of the medical faculty of the Univer- of a single subject. Morphometric analysis was done in 14 patients with MSA sity of Bonn, Bonn, Germany. Informed and written consent compared with 13 healthy controls, using an uncorrected voxelwise threshold of PϽ.001 and a corrected threshold for the cluster size of PϽ.05. was obtained from all participants. Gray matter is lost within the cerebellum and brainstem. z-Values given above the view refer to the corresponding slices in the Montreal Neurological DATA ACQUISITION Institute reference space.

Magnetic resonance imaging measurements were performed using a 1.5-T scanner (Siemens Symphony; Siemens AG, Er- MORPHOMETRIC ANALYSIS langen, Germany) with the standard head coil. The MRI protocol consisted of a T1-weighted, magnetization-prepared, Data were preprocessed as described by Good et al.13 To pre- rapid acquisition gradient-echo sequence. Repetition time vent mismatch errors, each preprocessing step was controlled was 11.08 milliseconds; echo time, 4.3 milliseconds; flip and verified for each subject and each sequence separately. Af- angle, 15°; excitation, 1 per phase encoding step; field of view, ter defining the anterior commissure in each image as the ori- 230 mm; acquisition matrix, 256ϫ256 pixels; and slice thick- gin of the individual stereotaxic space, we reoriented all im- ness, 180 mm; yielding 200 sagittal slices and a voxel size of ages to the axial view. For optimizing the stereotactic 0.9ϫ0.9ϫ0.9 mm3. normalization procedure, the images were automatically seg-

(REPRINTED) ARCH NEUROL / VOL 60, OCT 2003 WWW.ARCHNEUROL.COM 1432

Voxelwise Statistics Coordinates

Extent, No. of Voxels t Value P Value Corrected x y z Area Gray matter decrease 18 212 14.26 .00 9 −36 −19 Bilateral cerebellum (anterior and posterior lobes) White matter decrease 3661 10.63 .00 −12 −29 −34 Bilateral pons and cerebellum (posterior lobe) 184 5.6 .34 9 23 24 R anterior cingulate 186 5.15 .62 −16 37 −8 L medial frontal lobe and orbital gyrus 360 4.97 .75 −10 −5 −6 L anterior commissure, thalamus, and caudate nucleus 169 4.52 .96 −34 −53 19 L superior, middle temporal gyrus White matter increase 1049 8.47 .002 28 −16 26 R pyramidal tract 227 5.65 .31 −15 −15 −12 L pyramidal tract 149 5.16 .61 46 −69 −4 R middle occipital gyrus 185 4.86 .82 3 −38 −40 R medulla oblongata

*Morphometric analysis of magnetic resonance imaging density changes was made separately for gray and white matter in 14 patients with multiple system atrophy–cerebellar type compared with 13 healthy control subjects. Each reported anatomical location exceeded a voxelwise statistical threshold of PϽ.001 (uncorrected) and an extent threshold of PϽ.05 (corrected). The extent of each location is expressed as the number of voxels. The coordinates refer to the Montreal Neurological Institute space and denote the location of the most significant voxel within each area.

mented into gray matter, white matter, and cerebrospinal fluid and in adjacent regions of the cerebellar hemispheres probability maps. Afterward, all nonbrain voxels were re- in MSA-C–affected patients. The gray matter loss was moved from the segmented images. The gray and white matter more pronounced in the right part of the cerebellum maps obtained by this procedure were separately normalized (Figure 1A). A more liberal analysis using an uncor- to a gray and white matter template representing the stereo- rected threshold of PϽ.001 showed a widespread reduc- tactic standardized space, defined by a template from the Mon- treal Neurological Institute, Montreal, Quebec. After applying tion of gray matter of the cerebellum sparing only parts both transformations into the Montreal Neurological Institute of the hemispheres and the basal portion of the cerebel- template space to the original images, the spatially normalized lum. In addition, there were significant clusters in the images were resampled with a voxel size of 1.5ϫ1.5ϫ1.5 mm3. tegmentum of the mesencephalon and in lateral and ven- Scans were then segmented into gray and white matter. For fur- tral parts of the pons (Figure 1B and Table 2). The in- ther analysis we used the gray matter probability map derived verted comparison did not reveal any increases of gray from optimized gray matter normalization and the white mat- matter in the MSA-C–affected brains. ter probability map derived from optimized white matter normalization. The final probability maps were smoothed with a WHITE MATTER FINDINGS gaussian smoothing kernel size of 8 mm (full-width at half maxi- mum of the gaussian kernel). Ͻ The normalized, segmented, and smoothed data were ana- Initial analysis using a corrected threshold of P .05 re- lyzed using statistical parametric mapping (SPM99).20 The sta- vealed a bilateral reduction of white matter in the middle tistical comparisons between the groups were calculated as cerebellar peduncles of MSA-C–affected brains. In addi- 2-sample t tests separately for the compartments. Each analysis tion, smaller regions of white matter loss were found in was controlled for global differences in voxel intensity by includ- the right posterior and left anterior parts of the cerebel- ing the overall mean of voxel intensities as a confounding covar- lum (Figure 2A). Subsequent analysis using an uncor- iate in the design matrix. The reported results are based on t con- rected threshold of PϽ.001 showed widespread white mat- Ͻ trasts between the 2 groups at a threshold of P .05 corrected and ter loss in MSA-C–affected brains affecting the middle cluster level of at least 5 voxels, which reduces the false-positive cerebellar peduncles extending into the cerebellar white rate by correcting the threshold for multiple testing. To investi- gate the areas of significant differences in more detail, we ex- matter and the dorsal and lateral aspects of the pons. In plored the statistical results on anatomical slices. For these over- addition, we found small clusters of white matter loss in lays, the statistical parametric mapping (SPM99) results were several supratentorial regions including the right ante- thresholded at an uncorrected value of PϽ.001 for the voxel sig- rior cingulum, the left frontal lobe, the left temporal lobe, nificance and a corrected value of PϽ.05 for the cluster level. the left side of the thalamus, and the left caudate nucleus (Figure 2B and Table 2). RESULTS The inverted comparison revealed an increase of white matter density in MSA-C along the right pyramidal tract GRAY MATTER FINDINGS (corrected threshold, PϽ.05) (Figure 3A). Reanalysis with an uncorrected threshold of PϽ.001 showed white mat- At a corrected threshold of PϽ.05 we found a reduction ter increase in both pyramidal tracts in the right occipital of gray matter in the upper part of the cerebellar vermis lobe and the medulla oblongata (Figure 3B and Table 2).

(REPRINTED) ARCH NEUROL / VOL 60, OCT 2003 WWW.ARCHNEUROL.COM 1433

T-Value

B 6

B 4

0 z = 2 mm z = 5 mm z = 8 mm Figure 3. White matter increase in MSA. Increase of white matter density is shown on glass slide views of the brain. A, Morphometric analysis was done in 14 patients (5 men and 9 women) with MSA compared with 13 healthy control subjects (5 men and 8 women), using a corrected threshold of PϽ.05. White matter density of the right pyramidal tract is increased. B, Increase of white matter density is shown on 3 representative images displayed on a diffusion-weighted template. Morphometric analysis was done in 14 patients with MSA compared with 13 healthy controls, using an uncorrected voxelwise threshold of PϽ.001 and a corrected threshold for the z = 11 mm z = 14 mm z = 17 mm cluster size of PϽ.05. White matter density of the both pyramidal tracts is increased.

brainstem; and (3) an increase of white matter density along the pyramidal tracts. In contrast with region of interest–guided morphometric approaches, voxel-based morphometry is not limited to a number of preselected brain regions, but it does L z = 20 mm z = 23 mm z = 26 mm R provide a comprehensive analysis of the entire brain al- T-Value lowing improved anatomical localization and differen- 21 6 tiation between tissue loss in gray and white matter. To minimize the problems of normalization,22 we used an 4 optimized protocol based on the gray and white matter 2 compartments and a verification of each step. This refined method considerably improves the match be- 0 z = 29 mm z = 32 mm z = 35 mm tween the individual brain images and the templates used. In an attempt to increase the sensitivity of the method, we performed a dual statistical analysis, with an initial conservative statistical approach using a corrected threshold of PϽ.05, and a subsequent more liberal approach with an uncorrected threshold of PϽ.001. Our findings of prominent gray matter loss in the cerebellum and brainstem confirm earlier MRI studies5,7,10 and are in good agreement with the distribution of neuronal Figure 2. White matter reduction in multiple system atrophy (MSA). loss in MSA-C–affected brains at autopsy.23 Gray matter A, Reduction of white matter density is shown on the 4-color views of the brain. Morphometric analysis was done in 14 patients (5 men and 9 women) loss was pronounced in the upper cerebellar vermis and with MSA compared with 13 healthy control subjects (5 men and 8 women), adjacent areas of the hemispheres, areas of the cerebel- using a corrected threshold of PϽ.05. White matter is lost in the middle lum that are anatomically referred to as the anterior lobe cerebellar peduncles of both sides. B, Reduction of white matter density is shown on the 4-color views of the brain and on 12 consecutive axial and receive mainly spinal afferents. It is generally as- T1-weighted images of a single subject. Morphometric analysis was done in sumed that cerebellar atrophy in MSA-C is diffuse whereas 14 patients with MSA compared with 13 healthy controls, using an other types of cerebellar degeneration, in particular those uncorrected voxelwise threshold of PϽ.001 and a corrected threshold for the Ͻ caused by chronic alcoholism, are characterized by promi- cluster size of P .05. White matter is lost within the middle cerebellar 24-26 peduncles, cerebellum, pons, and several supratentorial structures. z-Values nent involvement of the anterior lobe. Our findings seem given above the view refer to the corresponding slices in the Montreal to contradict this common view, suggesting that anterior Neurological Institute reference space. lobe degeneration is also typical for MSA-C. Since our patients were in a medium disease stage, one may conclude that the cerebellar anterior lobe is the most vulnerable part COMMENT of the cerebellum in MSA-C where degeneration starts whereas advanced cases that come to autopsy have a dif- We used voxel-based morphometry to evaluate the MRIs fuse cerebellar degeneration. of patients with MSA-C. The major findings were (1) a Multiple system atrophy is unique among the neuro- loss of gray matter in the brainstem and anterior lobe of degenerative disorders as oligodendroglial cells forming cen- the cerebellum; (2) a reduction of white matter in the tral myelin are the primary cell type affected. ␣-Synuclein– middle cerebellar peduncles, cerebellar white matter, and positive inclusions are abundant in oligodendroglial cells

(REPRINTED) ARCH NEUROL / VOL 60, OCT 2003 WWW.ARCHNEUROL.COM 1434

©2003 American Medical Association. All rights reserved. Downloaded From: https://jamanetwork.com/ on 09/27/2021 and distributed throughout the brain beyond areas of overt REFERENCES neuronal loss.3 A recent study using antibodies that spe- cifically recognize an epitope of myelin basic protein that 1. Graham JG, Oppenheimer DR. Orthostatic hypotension and nicotine sensitivity in a is exposed in areas of myelin degeneration detected wide- case of multiple system atrophy. J Neurol Neurosurg Psychiatry. 1969;32:28-34. 2. Quinn N. Multiple system atrophy—the nature of the beast. J Neurol Neurosurg spread immunoreactivity in oligodendrocytic processes sug- Psychiatry. 1989;suppl:78-89. 27 gesting significant myelin degeneration in MSA. Thus, 3. Papp MI, Lantos PL. The distribution of oligodendroglial inclusions in multiple structural abnormalities observed in myelinated fiber tracts system atrophy and its relevance to clinical symptomatology. Brain. 1994;117: 235-243. in MSA may be a direct consequence of myelin pathology 4. Wenning GK, Ben Shlomo Y, Magalhaes M, Daniel SE, Quinn N. Clinical features due to oligodendroglial cell dysfunction. In addition, fi- and natural history of multiple system atrophy: an analysis of 100 cases. Brain. ber tracts may be also affected by axonal loss secondary to 1994;117:835-845. 5. Schulz JB, Klockgether T, Petersen D, et al. Multiple system atrophy: natural his- neuronal degeneration. tory, MRI morphology, and dopamine receptor imaging with 123IBZM-SPECT. Our study shows loss of white matter, which was J Neurol Neurosurg Psychiatry. 1994;57:1047-1056. 6. Gilman S, Low PA, Quinn N, et al. Consensus statement on the diagnosis of mul- most prominent in the middle cerebellar peduncles. This tiple system atrophy. J Neurol Sci. 1999;163:94-98. confirms earlier MRI studies showing that the middle cer- 7. Schulz JB, Skalej M, Wedekind D, et al. Magnetic resonance imaging-based volu- ebellar peduncles undergo the most severe shrinkage of metry differentiates idiopathic Parkinson’s syndrome from multiple system at- 5 rophy and progressive supranuclear palsy. Ann Neurol. 1999;45:65-74. all brain structures analyzed. In addition, our data sug- 8. Savoiardo M, Strada L, Girotti F, et al. Olivopontocerebellar atrophy: MR gest that white matter loss in the cerebellum and brain- diagnosis and relationship to multisystem atrophy. Radiology. 1990;174:693- stem significantly contributes to the infratentorial atro- 696. 9. Schwarz J, Weis S, Kraft E, et al. Signal changes on MRI and increases in reac- phy observed on conventional MRIs in MSA. Further, tive microgliosis, astrogliosis, and iron in the putamen of two patients with mul- there was less pronounced white matter loss in the su- tiple system atrophy. J Neurol Neurosurg Psychiatry. 1996;60:98-101. 10. Schrag A, Good CD, Miszkiel K, et al. Differentiation of atypical parkinsonian syn- pratentorial brain regions. dromes with routine MRI. Neurology. 2000;54:697-702. To our surprise, the pyramidal tracts showed an 11. Schocke MF, Seppi K, Esterhammer R, et al. Diffusion-weighted MRI differenti- increase of white matter density. This finding cannot be ates the Parkinson variant of multiple system atrophy from PD. Neurology. 2002; 58:575-580. explained by degeneration of the pyramidal tracts since 12. Ashburner J, Friston KJ. Voxel-based morphometry—the methods. Neuroim- degeneration would result in decreased T1 signal. Cor- age. 2000;11:805-821. respondingly, most of our patients had no clinical signs 13. Good CD, Johnsrude IS, Ashburner J, Henson RN, Friston KJ, Frackowiak RS. A voxel-based morphometric study of aging in 465 normal adult human brains. of pyramidal tract dysfunction. Another observation ar- Neuroimage. 2001;14:21-36. guing against pyramidal tract degeneration in our pa- 14. Ohnishi T, Matsuda H, Tabira T, Asada T, Uno M. Changes in brain morphology in Alzheimer disease and normal aging: is Alzheimer disease an exaggerated ag- tients is a decreased T2 signal along the pyramidal ing process? AJNR Am J Neuroradiol. 2001;22:1680-1685. tracts (our unpublished data, 2000). This finding, in 15. Rombouts SA, Barkhof F, Witter MP, Scheltens P. Unbiased whole-brain analy- conjunction with the increased T1 signal, points to de- sis of gray matter loss in Alzheimer’s disease. Neurosci Lett. 2000;285:231- 233. creased water content in the pyramidal tracts, possibly 16. Baron JC, Chetelat G, Desgranges B, et al. In vivo mapping of gray matter loss due to deposition of an unknown biochemical sub- with voxel-based morphometry in mild Alzheimer’s disease. Neuroimage. 2001; strate. Whether these findings are related to altered my- 14:298-309. 17. Burton E, Karas G, Paling S, et al. Patterns of cerebral atrophy in dementia with elin properties due to oligodendroglial pathology re- lewy bodies using voxel-based morphometry. Neuroimage. 2002;17:618. mains a matter of speculation. 18. Thieben MJ, Duggins AJ, Good CD, et al. The distribution of structural neuro- pathology in pre-clinical Huntington’s disease. Brain. 2002;125:1815-1828. 19. Kassubek J, Juengling FD, Hellwig B, Spreer J, Lu¨cking CH. Thalamic gray mat- Accepted for publication March 26, 2003. ter changes in unilateral Parkinsonian resting tremor: a voxel-based morpho- Author contributions: Study concept and design (Drs metric analysis of 3-dimensional magnetic resonance imaging. Neurosci Lett. 2002;323:29-32. Reul and Klockgether); acquisition of data (Drs Specht, 20. Members and Collaborators of the Wellcome Department of Cognitive Neurol- Minnerop, Abele, and Wu¨ llner); analysis and interpreta- ogy. Statistical parametric mapping (SPM99). Available at: http://www.fil.ion.ucl .ac.uk/spm. Accessed July 16, 2003. tion of data (Drs Specht and Klockgether); drafting of the 21. Good CD, Scahill RI, Fox NC, et al. Automatic differentiation of anatomical pat- manuscript (Drs Specht, Minnerop, Abele, Reul, and terns in the human brain: validation with studies of degenerative dementias. Neu- Klockgether); critical revision of the manuscript for im- roimage. 2002;17:29-46. 22. Ashburner J, Friston KJ. Why voxel-based morphometry should be used. Neu- portant intellectual content (Drs Specht, Abele, Wu¨ llner, roimage. 2001;14:1238-1243. and Klockgether); statistical expertise (Drs Specht and 23. Wenning GK, Tison F, Ben Shlomo Y, et al. Multiple system atrophy: a review of Klockgether); obtained funding (Drs Wu¨ llner and Klock- 203 pathologically proven cases. Mov Disord. 1997;12:133-147. 24. Holmes G. An attempt to classify cerebellar disease, with a note on Marie’s he- gether); administrative, technical, and material support (Drs reditary cerebellar ataxia. Brain. 1907;30:545-567. Specht, Minnerop, Abele, Wu¨ llner, and Klockgether); 25. Greenfield JG. The Spino-Cerebellar Degenerations. Springfield, Ill: Charles C Thomas Publisher; 1954. study supervision (Drs Reul and Klockgether ). 26. Victor M, Adams RD, Collins GH. The Wernicke-Korsakoff syndrome: a clinical Corresponding author: Thomas Klockgether, MD, and pathological study of 245 patients, 82 with post-mortem examinations. Con- Department of Neurology, University of Bonn, Sigmund- temp Neurol Ser. 1971;7:1-206. 27. Matsuo A, Akiguchi I, Lee GC, McGeer EG, McGeer PL, Kimura J. Myelin degen- Freud-Strausse 25, D-53105 Bonn, Germany (e-mail: eration in multiple system atrophy detected by unique antibodies. Am J Pathol. [email protected]). 1998;153:735-744.

(REPRINTED) ARCH NEUROL / VOL 60, OCT 2003 WWW.ARCHNEUROL.COM 1435