Patients and methods

Twenty three patients with striatonigral degeneration-type (SND) MSA (10 men and 13 women) and 21 patients with PSP (14 men and seven women) consecutively referred to our institute from June 1993 to July 1996 as outpatients or admitted patients were enrolled. Eighteen patients with PD (10 men and eight women) were selected as having the same disease severity as the patients with PSP or SND, as assessed by the Hoehn and Yahr scale and the activity of daily living (ADL) section of the unified Parkinson's disease rating scale (UPDRS, table1).20 Probable PD was diagnosed according to the UK Brain Bank criteria,21 probable SND-type MSA according to Quinn's criteria22 and probable PSP according to the criteria of Litvan et al.23

The clinical diagnoses of SND and PSP were supported by MRI studies, which showed putaminal signal hypointensity in T2 weighted images (suggesting deposition of iron or other paramagnetic substances) in all patients with SND, and slight to severe midbrain atrophy in all patients with PSP.24 Dementia and major depression were diagnosed according to DSM IV criteria.25 Based on the Edinburgh Inventory26 all but two patients (one with PD and one with PSP) were right handed. All patients with PD or SND were receiving levodopa; seven patients with PD were also taking dopamine agonists. All patients with PD and six with SND had a good levodopa response and were experiencing motor fluctuations and dyskinesias. Ten of the 21 patients with PSP were receiving levodopa with small benefit. The cognitive examination was performed when patients on levodopa were at peak motor response.

Neuropsychological examination comprised the mini mental state examination (MMSE) a screening test for mental decay27; the Raven progressive matrices, which measures logical reasoning and visuospatial organisation28; the short tale test for long term verbal memory29; the phonemic verbal fluency test which evaluates word searching strategy30; the visual search test, to assess attention and visual scanning30; the visuospatial orientation line test of Benton31 which assesses the visual ability to distinguish the orientation of lines in space; and the Nelson modification of the Wisconsin card sorting test32 which examines set shifting and categorisation abilities. In the last test we scored the number of sorting categories reached by the patients. Raven test scores were adjusted for age and education according to Basso et al 33; MMSE scores were similarly adjusted using the method of Measso et al.34 Other cognitive test scores were adjusted for age and education according to Spinnler and Tognoni,35 except for the Nelson test, for which this adjustment is not available.

Only 12 patients with PD (66%), 14 with SND (60%), and 11 with PSP (52%) were retested a mean of 21 months (range 18–24 months) after the initial evaluation: five patients with PD, one with PSP, and two patients with SND refused; one patient with PD, five with SND, and six with PSP were too disabled to attend for retesting; and two with SND and three with PSP died before retesting.

STATISTICAL ANALYSIS

Only the patients who presented for both tests were included in the statistical analyses. Between group comparisons employed the non-parametric Kruskal-Wallis analysis of variance (ANOVA). Post hoc comparisons with the non-parametric Mann-WhitneyU test were applied where appropriate. Linear correlations between indices of cognitive and motor decline using Pearson's correlation coefficient were sought when useful to clarify the results.

Results

INITIAL EVALUATION

Although the PD group was selected to have the same disease severity as the SND and PSP groups, evaluation considering only the patients who presented for both tests showed that at the outset the three groups differed significantly in motor performance as assessed by the Hoehn and Yahr and ADL scales, with patients with PD less compromised than patients with SND or PSP. The PD group also had longer disease duration than the SND and PSP groups, but there were no differences in age or education (table 2). Post hoc comparison of the dropout and retested patients in each disease group for illness duration, Hoehn and Yahr, ADL, and MMSE showed a significant difference only for patients with PD, where mean illness duration was shorter (8.0 (SD 3.0) v 12.1 (SD 5.0) years, p=0.03) and mean ADL score was greater 22.5 (SD 5.3) v12.3 (SD 6.6), p=0.01) in those who dropped out. Motor disability, as assessed by the ADL scale, was lower in patients with PSP who dropped out, but not significantly so.

Between group comparison at first evaluation on patients who attended both test sessions showed that the PSP group performed significantly worse than the PD and SND groups in the short tale, verbal fluency, visual search and Benton's test, while the SND group was significantly worse than the PD group only in the verbal fluency test (table 3). Two patients with PSP but no patients with PD or SND had an MMSE score<24 (indicating dementia); the two patients with PSP were also demented according to DSM IV criteria. None of the patients had major depression according to DSM IV criteria.

SECOND EVALUATION

At the second evaluation the three groups still differed significantly in disease severity as assessed by the Hoehn and Yahr (p=0.003) and ADL (p=0.001) scales. Post hoc comparison showed that the mean Hoehn and Yahr score for patients with PD (3.5 (SD 0.7)) was significantly lower than those of the SND and PSP groups (4.1 (SD 0.6) and 4.5 (SD 0.5) respectively). The mean ADL score of the PD group (16.7 (SD 9.0)) was significantly lower than those of the SND and PSP groups (25.1 (SD 8.4) and 31.9 (SD 5.6) respectively); in addition, patients with SND scored significantly lower than the patients with PSP.

Between group comparison of percentage changes in cognitive and motor scores between the first and second evaluations showed significantly greater deterioration of performance in the Nelson test by the PSP group compared with the other two groups and in the visual search test by the PSP and SND groups compared with the PD group. The same statistical test failed to show significant differences between the groups in the progression of motor compromise (table4).

Six patients with PSP (54%) but no patients with PD or SND had MMSE scores below 24 at second evaluation; the six patients with PSP also met the DSM IV criteria for dementia. Two of the six demented patients with PSP had dementia at the initial evaluation.

A global cognitive decay index (DI) was calculated for each patient as the algebraic sum of the percentage change in each cognitive test, divided by the number of tests. Percentage changes (deterioration) in DI and ADL scores ⩾30% were arbitrarily considered as “clinically significant” markers of mental decay and motor disability, respectively.

None of the 12 patients with PD had a DI⩾0%, but in seven the ADL score worsened by⩾30%. In the SND group, two of 14 patients had a DI⩾30% and nine had an ADL change⩾0%. Of the 11 patients with PSP, six had a DI of⩾30% (all of whom were demented according to DSM IV) and eight had an ADL change⩾0% (figure).

• Download figure
• Open in new tab
• Download powerpoint

Change in cognitive (dark bars) and motor performance (light bars) of individual patients with time.

No significant correlations were found between the indices of cognitive (DI) and of motor decline (ADL percentage change), Furthermore, we found no within group correlation of percentage changes in Nelson and visual search scores with motor disability at first evaluation, or with changes in ADL scores.

Discussion

At first evaluation the PSP group did worse than the PD and SND groups in the short tale, verbal fluency, visual search, and Benton tests. This is in agreement with the existing literature, which finds greater cognitive impairment in patients withPSP than patients with PD or SND4 5 when the groups are well matched for age and education (as was the case in the present study).

Greater cognitive impairment in PSP is presumably due to the marked deafferentation, particularly of the premotor and prefrontal areas, which occurs as a result of the striatal-thalamocortical pathway alteration well known in this disease.36 However, dysfunction of the ascending reticular formation of the pontomesencephalic tegmentum,37 which projects diffusely to prefrontal cortical areas, as well as direct cortical alterations, recently demonstrated in prefrontal and premotor areas,38 39 could also play a part in mental decay in PSP.

Analysis of the percentage changes in cognitive and motor performance over time showed that the groups differed significantly only on the visual search test, where deterioration was more marked in SND and PSP than in PD, and on the Nelson test, where deterioration was greater in PSP than in PD and SND. From clinical experience and the results of previous cross sectional studies,4 5 we expected to see increasingly marked differences between the groups on cognitive performance with time, with PSP deteriorating faster. Instead, deterioration in most cognitive tests did not differ greatly between the three groups. High dropout rate, large within group variation in test performance, a floor effect in test performance, and relatively short follow up may all have contributed to this.

Post hoc comparison showed that dropout patients with PD had shorter disease duration and greater mean motor disability at first evaluation than the remaining patients with PD, whereas dropout patients with SND or PSP did not differ significantly in any clinical features from those who reattended. However, most of the patients with PD were reluctant rather than unable to come, whereas the main reason why patients with PSP or SND did not attend for retesting was, they said, because of motor disability; in addition two patients with SND and three with PSP died before retesting. It is reasonable to suppose that the patients who did not attend, especially the patients with PSP or SND who complained of disability, deteriorated in all respects more than those who did attend.

The large within group variation in test performance in the patients with PSP or SND, evident by inspection of the SDs in table 4, may also have contributed to the lack of significant differences between the groups. One reason for this intergroup variability could have been differences in disease stage within the groups. However, there were no significant correlations between cognitive decline (DI) and motor decline (as assessed by percentage changes). Furthermore decline in Nelson and visual search tests did not correlate either with age or motor disability at first evaluation, or deterioration in ADL score, within each group. For this reason we do not think that differences in initial motor compromise had any effect on the mean cognitive changes over time.

This lack of correlation between motor and cognitive indices is illustrated graphically in the figure where patients with PSP or SND who deteriorated most cognitively were not those with the greatest deterioration in motor performance. Neurofibrillary tangles are normally present in the basal ganglia and brainstem of necropsied patients with PSP; however, recent pathological studies have shown that such lesions may also occur, to a variable extent, in the premotor cortex, prefrontal cortex, and hippocampus.38 39 It may be, therefore, that the variable presence of these lesions40 is responsible for the variability of cognitive deterioration found in our patients with PSP, with greater DI perhaps related to greater compromise of the cortical premotor and prefrontal areas.

There may also have been a floor effect: performance was poor at initial evaluation in the verbal fluency test and Benton's test in patients with PSP, and the verbal fluency test in patients with SND.

Short follow up was probably the main reason for the lack of deterioration seen in the PD group. Although PD seems to progress at a slower pace than PSP and SND, significant cognitive deterioration becomes evident with sufficient follow up. In a previous longitudinal study we found that 18% of patients with PD became demented 7 years after initial evaluation.15 However, in a longitudinal study by Bayles et al,14 the duration (24 months) of which was comparable with ours, 20% of patients with PD deteriorated by four or more points in MMSE score (defined as significant). None of our patients with PD had similar changes in MMSE score and none was demented at second evaluation. Perhaps the smaller sample size of our study (12v 77 in Bayles et al) and high dropout rate of our patients with PD (34%) also contributed to these differences.

For motor disability, we found that although this deteriorated significantly over time in all groups, between group differences in percentage changes in motor performance were not significant. This is again at variance with clinical experience and the literature. Faster disease progression and greater mortality compared with PD have been reported in both PSP41 42 and SND.43 44 The result is probably due to patient selection by dropout: as noted dropout patients with PD had more severe illness at first evaluation than those who were retested; and although dropout patients with PSP or SND did not differ at the outset from those who reattended, at the time of second evaluation five had died and the others gave severe motor disability as a reason for not reattending.

Nevertheless, we did find significant deterioration differences between the groups in the Nelson and visual search tests, indicating that these tests are the most sensitive in showing differences between degenerative parkinsonisms with time. The Nelson test taps executive functions, and worse performance in PSP is probably related to the greater striatofrontal pathway dysfunction documented in this condition.36 This is also in agreement with the finding of Pillon et al 4 of greater impairment in “reactive flexibility” in their patients with PSP which they contrasted with impairment in “spontaneous flexibility”45 in their patients with SND.

Both patients with PSP and those with SND deteriorated significantly more than patients with PD on the visual search test. Although this test mainly examines focused attention, it also taps other cognitive functions such as ability to shift between contrasting responses, speed of cognitive processing, and visual scanning ability. Previous publications have noted attentional impairment, deficit in cognitive set shifting ability,5 and bradyphrenia in PSP.46 47

The continuation of focused attention requires integrated activity of brainstem, thalamus, and neocortex. The critical brainstem structures involved in this activating system are the reticular formation and adjacent nuclei (raphe nucleus, locus coeruleus, and segmental nuclei) whereas the intralaminar, midline, and reticular thalamic nuclei48 are the most important relay stations; these project diffusely to multimodal associative cortical areas, among which the prefrontal areas are the most important for attention.49 It is not surprising, therefore, that attention was particularly impaired in PSP, as distinct alterations in the midbrain tegmentum and less prominent damage to premotor and prefrontal areas have been demonstrated pathologically.38 39 Marked alterations in the superior colliculi and the periaqueductal region could well be responsible for the poor performance of patients with PSP in this test, as also found by Kimura et al.50

It is more difficult to account for the visual search deterioration in patients with SND. This could have been due to degeneration of the striatal-cortical circuits, although this is less prominent than in PSP. Because cortical oligodendroglial bodies in MSA are restricted to the motor and premotor cortex51 and their relation to cell damage is not fully understood; their pathogenetic role in the cognitive deficits of patients with MSA is unclear.

The similarity between patients with PD and those with SND in cognitive performance emphasises the role of subcortical deafferentation in cognitive impairment in these diseases. However, patients with SND deteriorated slightly more than those with PD and this may be due to neuronal loss in the caudate nucleus in SND.52

In conclusion, our longitudinal study has shown that patients with PSP undergo greater cognitive impairment than patients with SND or PD, and this would correlate with greater anatomical compromise in PSP involving both subcortical (pallidum, mesencephalic tegmentum, striatum) and cortical structures (prefrontal and premotor cortex). Progression to dementia was prominent in PSP but not uniform in all patients, and the differences may be related to differential development of neurofibrillary tangles in the cortex. The cognitive performance of patients with SND was similar to that of patients with PD both at first and second evaluations; however, the SND group was characterised by slightly greater worsening with time. Longitudinal studies with serial examinations over a short period of time on greater numbers of patients are required to further elucidate the pattern of cognitive regression in these diseases.