Neuropathological Correlates of Dysarthria in Progressive Supranuclear Palsy

Background  The dysarthria of progressive supranuclear palsy consists of prominent hypokinetic and spastic components with less prominent ataxic components.

Objective  To correlate the types of dysarthria with neuropathological changes in patients with progressive supranuclear palsy.

Design and Methods  In 14 patients with progressive supranuclear palsy, we correlated the perceptual speech findings with the neuropathological findings. A dysarthria assessment was performed a mean ± SD of 31 ± 15 months (range, 10-53 months) before death. The deviant speech dimensions were rated on a scale of 0 (normal) to 3 (severe). The neuropathological examination consisted of semiquantitative analysis of neuronal loss and gliosis by investigators (A.A.F.S., and L.A.B.) blinded to the clinical findings. Correlation and linear regression analysis were used to correlate the severity of the hypokinetic, spastic, and ataxic components with the degree of neuronal loss and gliosis in predetermined anatomical sites.

Results  All patients had hypokinetic and spastic dysarthria, and 9 also had ataxic components. The severity of the hypokinetic components was significantly correlated with the degree of neuronal loss and gliosis in the substantia nigra pars compacta (r = 0.61, P = .02) and pars reticulata (r = 0.64, P = .01) but not in the subthalamic nucleus (r = 0.51, P = .07) or the striatum or globus pallidus (/r/<0.34, P>.20). The severity of the spastic and ataxic components was not significantly correlated with the neuropathological changes in the frontal cortex (r = 0.20, P = .50) and cerebellum (/r/<0.28, P>.33), respectively.

Conclusion  The hypokinetic dysarthria of progressive supranuclear palsy may result from degenerative changes in the substantia nigra pars compacta and pars reticulata and not from changes in the striatum or globus pallidus.

DYSARTHRIA is a cardinal feature of progressive supranuclear palsy (PSP).¹ Dysarthria consists of a combination of hypokinetic, spastic, and ataxic components, usually with prominent hypokinetic and spastic and less prominent ataxic features.² Neuropathological changes in patients with PSP involve neuronal loss and gliosis with neurofibrillary tangles, argyrophilic and τ-positive threadlike structures, and τ-positive astrocytic tanglelike inclusions in many subcortical regions, including the substantia nigra (SN), striatum, globus pallidus, subthalamic nuclei, periaqueductal gray, pontine nuclei, inferior olives, cerebellar dentate nuclei, and certain cranial nerve nuclei.^1,3-7 The anatomical locations of the neuropathological changes responsible for the dysarthria in patients with PSP have not been determined. In other neurological disorders, the major types of dysarthria have been linked to specific anatomical connections in the nervous system. Bilateral involvement of the corticobulbar pathways is associated with spastic dysarthria, disorders of the cerebellum and its connections with ataxic dysarthria,^8,9 and diseases of the extrapyramidal pathways with hypokinetic dysarthria.⁸

We examined the speech disorders of 14 patients with PSP who later underwent autopsy. We correlated the severity of the specific components of the dysarthria with the neuropathological changes in structures that have been associated with dysarthria in other neurological diseases. Preliminary findings have been reported.¹⁰

Patients with PSP have been studied in research protocols at the University of Michigan, Ann Arbor, since 1984, and many have been followed up with longitudinal clinical assessments using the protocols of the Michigan Alzheimer's Disease Research Center since 1989. The evaluations included neurological examinations, speech pathology assessments, structural and functional imaging studies, and subsequent neuropathological examinations at autopsy. We identified all patients with PSP examined postmortem who had been evaluated by a speech pathologist (K.J.K.) between 1984 and 1994.

Evaluation of dysarthria included assessment of oral motor and oral agility skills and perceptual speech analysis. Oral motor examination consisted of assessment of muscular strength; coordination; accuracy; range of excursion; and symmetry of head and neck, face, mandibular, tongue, palatopharyngeal, and respiratory muscles at rest and during reflex and voluntary movements. Oral agility was assessed by oral diadochokinetic rates and the oral agility skills subtest of the Boston Diagnostic Aphasia Examination.¹¹ Perceptual speech analysis included identification and rating of the severity of the deviant speech dimensions during the examination and from videotaped or audiotaped samples of spontaneous speech, description of the Cookie Theft picture from the Boston Diagnostic Aphasia Examination,¹¹ and oral reading of the "Grandfather Passage."⁸ We used the definitions of deviant speech dimensions of Darley et al⁸ and the University of Michigan classification of hypokinetic, ataxic, and spastic dysarthrias^2,10,12 (Table 1). Each deviant speech dimension identified was assigned a severity score extending from 0 (normal) to 3 (severe). Weighting factors were applied to emphasize the deviant speech dimensions most characteristic of each type of dysarthria.^2,10,12 A total score was obtained reflecting the degree of hypokinesia, spasticity, and ataxia in speech. The possible scores ranged from 0 to 48 for each dysarthria type, with higher scores indicating more severe impairment. To ensure consistency, all speech pathology examinations were performed by a single speech pathologist (K.J.K.). Methods of measuring the severity of speech disorders have been described in previous publications.^2,9,10,12

The neuropathological diagnosis of PSP was based on the findings specified in the National Institute of Neurological Disorders and Stroke criteria,³ including neuronal loss, gliosis, and neurofibrillary tangles; the latter were detected with a modified Bielschowsky silver stain in subcortical nuclei.^3,4,13 Coexistent Alzheimer disease was determined using the Reagan criteria.^3,14 Neuronal loss and gliosis were graded semiquantitatively in 46 brain sections stained with cresyl violet–Luxol fast blue–eosin and phosphotungstic acid–hematoxylin using the following scale: absent is 0; mild, 1; moderate, 2; and severe, 3. Intermediate changes were given half values. The pathological changes were scored by consensus of 2 neuropathologists (A.A.F.S., and L.A.B.), unaware of the clinical findings, who viewed the sections concurrently in a multiheaded microscope. We formulated hypotheses concerning the anatomical sites in the nervous system where neuropathological changes might be correlated with the components of dysarthria. Since the distribution of the data deviated from normality, we used Spearman rank correlation coefficients to analyze the relation between hypokinetic, spastic, and ataxic dysarthria types and neuropathological abnormalities in these predetermined anatomical sites. The scores for neuronal loss and gliosis in each anatomical area were added to produce a composite neuropathological score. We created a composite neuropathological score to limit problems with multiple comparisons, given the modest sample size and the number of analyses conducted. Based on the association of hypokinetic dysarthria with several nuclei of the basal ganglia, we correlated the hypokinetic dysarthria rating with the composite neuropathological scores in the SN pars compacta (SNc), SN pars reticulata (SNr), subthalamic nucleus, caudate nucleus, putamen, lateral globus pallidus, medial globus pallidus, and periaqueductal gray. Based on the association of spastic dysarthria with corticobulbar projections, we correlated the spastic dysarthria rating with the composite neuropathological scores in the frontal cortex. Based on the association of ataxic dysarthria with the cerebellum and related structures, we correlated the ataxic dysarthria rating and the composite neuropathological scores in the cerebellar cortex, dentate nuclei, inferior olives, and red nuclei.

The 14 patients consisted of 7 men and 7 women (mean ± SD age at death, 69.0 ± 5.8 years; range, 59-80 years). The clinical diagnosis at the last clinic visit was PSP in all patients. One patient initially was diagnosed as having Alzheimer disease because of progressive dementia, but later developed rigidity and supranuclear gaze palsy. Several others were diagnosed as having Parkinson disease (PD) elsewhere, but had clear signs of PSP by the time of our initial examination. Twelve patients had received levodopa during their illness. No benefit was noted in 11, and 1 had a "poor" response. Motor speech was assessed at the time of the initial examination. At this time, all 14 patients had cognitive impairments, limb rigidity, supranuclear gaze palsy, and frequent falls. Although we regularly see our patients at 6-month intervals, some subjects were unable to return for examination as they became more impaired or were institutionalized. Consequently, the last clinical neurological examination ratings were performed on average a mean ± SD of 8 ± 8 months before death. The mean ± SD duration of neurological symptoms before death was 7 ± 2 years (range, 3-10 years). Most of the patients had moderate to severe parkinsonism and gaze limitations at their last examination. Bradykinesia and axial rigidity were the most severe motor signs, and 7 patients were unable to stand. Limitation of vertical gaze was greater than limitation of horizontal gaze in all patients. At their last clinical examination, nearly all patients had completely lost voluntary vertical eye movements. Gait ataxia was a common complaint early in the disease, but frequently became less apparent as the patients became increasingly immobile. One patient had resting distal tremor, and 2 others had extensor plantar responses.

In most patients, the speech pathology evaluation was performed only at the time of the patient's initial examination, a mean ± SD of 31 ± 15 months (range, 10-53 months) before death. All patients had mixed dysarthria with hypokinetic and spastic components, and 9 also had ataxic components (Table 2). Hypokinetic dysarthria scores ranged from 4 to 39 (mean, 15), spastic dysarthria scores from 6 to 34 (mean, 18), and ataxic dysarthria scores from 0 to 20 (mean, 5). All patients had masked faces. Thirteen patients had impaired lingual rapid alternating movements ranging in severity from mild to severe, with a mean rating of moderate. Ten patients had a brisk jaw jerk, and 9 had a hyperactive gag reflex. The nonverbal agility skills scores ranged from 0 to 11 of a total of 12 (mean, 4.5). The verbal agility skills scores ranged from 0 to 14 of a total of 15 (mean, 9).

Whole brain weight ranged from 951 to 1390 g (mean, 1197 g). Gross inspection demonstrated no abnormal cerebral atrophy. Pallor of the SN was seen in all patients. Extensive subcortical neuropathological changes were found in a distribution characteristic of PSP, with neurofibrillary tangles and extensive neuronal loss and gliosis.^3,13 Neurofibrillary tangles were found in all patients in the subthalamic nucleus, SNc, third nuclear complex, periaqueductal gray, and pontine nuclei. In 13 patients, neurofibrillary tangles were found in the lateral globus pallidus, locus ceruleus, and inferior olivary nucleus. Tangles were also found in the putamen of 6 patients, in the claustrum of 5, and in the caudate nucleus of 3. Neuronal loss and gliosis accompanied these changes, and were most severe in the SNc, periaqueductal gray, subthalamic nucleus, and medial globus pallidus. Neuronal loss, gliosis, and neurofibrillary tangles were apparent in the regions chosen for correlational analysis (Table 3). Two patients had neuropathological changes of Alzheimer disease and PSP, and the other 12 had only pure PSP.

The severity of the hypokinetic component of dysarthria was significantly correlated with the neuropathological score in the SNc and SNr. A sizeable correlation was also found with the degree of neuronal loss and gliosis in the subthalamic nucleus, but this did not reach statistical significance (Table 4 and Figure 1). Correlations with other predetermined anatomical sites in the striatum and globus pallidus were smaller and statistically insignificant (/r/<0.34, P>.20). We adjusted for the duration from the speech pathology evaluation to death by multiple regression with dysarthria as outcome, and the adjusted correlations were similar.

To assess whether the relation between neuropathological changes in the SN and hypokinetic dysarthria might be attributable to symptoms of dysarthria in general, we correlated the sum of 2 nonhypokinetic types of dysarthria, spastic and ataxic, with the severity of the neuropathological changes. We found no significant correlations in any of the regions (P>.12 for all); estimated correlations with the SN were in fact slightly negative (SNc: r = −0.43, P>.12; SNr: r = −0.15, P>.60). Furthermore, the correlations between hypokinetic dysarthria and neuropathological scores in the SN persisted when the sum of spastic and ataxic dysarthria was controlled by linear regression. Specifically, the P values for these partial correlations were .05 for the SNc and .03 for the SNr.

Spastic dysarthria components were not closely correlated with neuropathological changes in the frontal cortex (r = 0.20, P = .50), and ataxic dysarthria components did not correlate significantly with the neuropathological changes in the dentate nucleus, inferior olives, or red nuclei (/r/<0.28, P>.33). All patients except 1 had neuropathological scores of 0 in the cerebellar cortex, preventing correlations of the severity of ataxic dysarthria with pathological features in this site. The dysarthria scores of the 2 patients with Alzheimer disease and PSP were within the same range as those of the patients with pure PSP.

This study revealed significant correlations between the intensity of the neuropathological changes in the SNc and SNr and the severity of the hypokinetic components of dysarthria in patients with PSP. We found no significant correlations with neuropathological changes in the striatum, globus pallidus, or periaqueductal gray. We selected the SN for correlation in relation to the hypokinetic components of dysarthria because of the similarity of the hypokinetic dysarthria in patients with PSP to that in patients with PD and the known intense neuropathological changes in this site in those with both PD and PSP. Darley et al⁸ proposed that hypokinetic dysarthria reflected extrapyramidal dysfunction based on examination of patients with PD. In patients with both PSP and PD, there is severe neurodegeneration within the SNc and decreased inhibitory input to the SNr from the globus pallidus, but only in those with PSP is there substantial loss of SNr neurons.¹⁵ In patients with PSP, the striatum and the globus pallidus are also involved, but in those with PD these regions are spared. The severity of neuropathological changes was considerably less in the striatum than in the SN, which may be a reason for the nonsignificant correlations for this structure. The medial globus pallidus, subthalamic nucleus, and periaqueductal gray had severe neuronal loss and gliosis, but the neuropathological changes in these structures were not significantly correlated with the severity of hypokinetic dysarthria. Hence the findings in the present study suggest that the most important structures in the pathophysiology of hypokinetic dysarthria in patients with PSP may be the SNc and the SNr. The SNr is the output station of the basal ganglia and projects to anterior/ventral lateral, mediodorsal, and midline thalamic nuclei, superior colliculus, and pedunculopontine nucleus in the brainstem.¹⁶ The role of the SNr in patients with dysarthria may be related to the projections into the brainstem.

We found no significant correlations between the spastic and ataxic components of dysarthria and the severity of neuropathological abnormalities in sites known to be affected in those with PSP and associated with these types of dysarthria. We also correlated with neuropathological scores the sum of the 2 nonhypokinetic types of dysarthria, spastic and ataxic. We found no correlations in any of the regions studied, indicating that symptoms of dysarthria in general do not account for the correlation between hypokinetic dysarthria and neuropathological changes in the SN. While these results should be interpreted with caution given the limited sample size, they are consistent with a link with the SN that is specific to the hypokinetic dysarthria components.

Dysarthria has been reported as the second most common clinical manifestation of PSP.¹⁷ The SN, globus pallidus, and subthalamic nucleus are the areas with the most severe neuropathological abnormalities in patients with PSP, raising the possibility that the pathological process starts there.¹⁸

The patients described in this study appear to represent an appropriate sample of patients with PSP. All met clinical and neuropathological criteria for PSP.^1,3-5,13 All had mixed dysarthria with hypokinetic and spastic components, and more than half had ataxic components. The spectrum of clinical findings in these patients was similar to those in previous reports.^4,17 In keeping with an earlier report,¹⁹ our series included an equivalent proportion of men and women; however, a recent report²⁰ stated that PSP affects men more frequently than women. As in our sample, in PSP, resting tremor is unusual, dysarthria and gaze limitation are consistently present late in the illness, and vertical eye movements are more severely affected than horizontal eye movements.^4,17,20,21

In this study, the time from diagnosis and speech pathology evaluation to death was long, and many of our patients became anarthric during that time. The neuropathological changes doubtless advanced in the SN and the other structures affected from the time of diagnosis and dysarthria assessment to the time of death. Despite these concerns, the correlations of hypokinetic dysarthria with neuropathological changes in the SNc and SNr were significant, perhaps reflecting the importance of these sites in the pathogenesis of hypokinetic dysarthria.

Accepted for publication August 1, 2000.

This study was supported in part by grant P50AG08671 from the National Institute on Aging, National Institutes of Health, Bethesda, Md (Michigan Alzheimer's Disease Research Center).

Corresponding author and reprints: Karen J. Kluin, MS, CCC, BC-NCD, Department of Speech-Language Pathology, University of Michigan Health System, 1D203 University Hospital, 1500 E Medical Center Dr, Ann Arbor, MI 48109-0043.