Clinical Determinants of Poststroke Dementia

Subjects and Methods

Procedures of the Helsinki Stroke Aging Memory Study cohort were detailed in a report on methods and baseline findings.³ Briefly, 486 consecutive patients aged 55 to 85 years were evaluated 3 months after ischemic stroke. Of these, 451 completed a clinical neurological and mental status examination.³ The excluded patients included 1 who refused, 1 with reduced level of consciousness, 1 with severe hearing impairment, and 32 with severe aphasia. Of the 451 patients, 337 (74.7%) completed MRI of the head and a comprehensive neuropsychological examination. The 114 patients not included were 59 in whom MRI was not performed (contraindication in 27, refusal in 18, claustrophobia in 2, severe illness in 11, obesity in 1) and 55 who did not complete in detail the cognitive test battery (lack of sufficient cooperation in 23, refusal in 14, nonfluency in Finnish language in 5, severe aphasia in 5, severe neglect in 3, hearing or visual impairment in 3, illiteracy in 2). The 114 patients excluded were older (mean age, 73.5 versus 70.2 years; P<.001), less often had small-vessel occlusion (1/114 versus 27/337; P=.0064), and more often stroke of undetermined etiology (82/114 versus 202/337; P=.0219), but they did not differ in terms of main demographic and clinical features including sex, education, and number, side, and site of stroke).

The subjects underwent a structured medical and neurological history based on review of all available hospital charts, interview of the subject and a knowledgeable informant, and a structured clinical and neurological examination performed by board-certified neurologists (T.P., R.V.). In addition, all the cases were reviewed by a senior neurologist (T.E.). The neurological examination focused on factors and features related to dementia and stroke similar to the method of the Memory Research Unit, Department of Neurology, University of Helsinki,¹⁶ and the National Stroke Data Bank.¹⁷ The laboratory evaluations included hemoglobin, hematocrit, white blood cell count, thrombocytes, sodium, potassium, liver function tests, calcium, total and HDL cholesterol, triglycerides, vitamin B₁₂, erythrocyte folate, thyroid function tests, fasting blood glucose, and creatinine. History of main vascular risk factors was obtained as described earlier.³ Types of ischemic stroke were classified according to the TOAST criteria into large-artery arteriosclerosis, cardioembolism, small-vessel occlusion (lacune), and stroke of undetermined etiology.¹⁸ Localization of the stroke was divided into right hemispheric, left hemispheric, and bilateral, as well as anterior (carotid), posterior (vertebrobasilar), and anteroposterior (symptoms and signs from both carotid and vertebrobasilar circulation).^{3 19} Furthermore, the stroke syndromes assessed included major dominant and nondominant hemispheric, minor dominant and nondominant hemispheric, deep/lacune, brain stem and cerebellar, and unknown.¹⁷ Major dominant hemispheric stroke syndrome included mixed aphasia with hemiparesis, hemisensory, and hemianopia/neglect; major motor (Broca) aphasia; sensory (Wernicke) aphasia; anterior cerebral artery syndrome with aphasia; or right hemianopia with dysnomia and/or dysmnesia and dyslexia. Stroke severity was assessed with the Scandinavian Stroke Scale²⁰ and the National Institutes of Health Stroke Scale.²¹ Arterial blood pressure was measured with the patient in the supine position, and orthostatic reaction was defined as decrease of systolic blood pressure over 20 mm Hg or diastolic over 10 mm Hg after 5 minutes with the patient in the upright position.²²

Details of the clinical mental status examination, as well as assessment of emotional and social functions, are given in previous reports.^{3 19} Aphasia was assessed clinically and with the Acute Aphasia Screening Protocol.²³

The clinical criteria for dementia were those of the American Psychiatric Association DSM-III.²⁴ Instead of using the more recent definitions for dementia,³ we chose the DSM-III definition because it was used in a previous comparative stroke cohort.^{2 4 6} The patients with any poststroke dementia were further divided into those with stroke-related dementia (mixed AD+VD excluded=no presence of prestroke cognitive decline)⁶ and dementia after first-ever stroke.⁷

Cognitive domains of the DSM-III definition assessed by the neuropsychological test battery included memory functions (short- and long-term memory), abstract thinking, judgment, aphasia, apraxia, agnosia, and constructional difficulty including visuospatial and constructional functions. The norms were adjusted for age but not education. The results were evaluated in different age groups. Adjustment for education was not done because most of the individuals in the age cohorts had completed grade school. In addition, personality change, work, and social activities were assessed by the neurologist. Abnormality in each domain was judged with the use of norms based on a random healthy Finnish population (2 SD or, if more than one test was used, 1 SD below the level of the norm on several tests indicated abnormality).²⁵ The diagnosis of abnormality in neuropsychological domains was based not only on individual low scores compared with normative data. In addition, we used the process approach, in which qualitative functional features known to be associated with distinct neuropsychological syndromes also guide the diagnostic decision; for example, if impairment of executive functions was the primary feature in the syndrome, the perseverative errors and other executive difficulties in the Token test were not regarded as primary dysphasia, but rather as secondary language difficulty.

Short-term memory was assessed with the Fuld Object-Memory Evaluation (learning curve over five trials, delayed recall, and recognition)²⁶ and the Logical Memory and Visual Reproduction of the Wechsler Memory Scale, Revised (immediate and delayed verbal and visual memory).²⁷ Long-term semantic memory and ability to recall previously learned knowledge was assessed with the short form of Information Task (10 questions on general knowledge and Finnish history) of the WAIS-R.²⁸ Abstract thinking (conceptualization/abstraction) was evaluated by the Similarities Subtest of the WAIS-R. Judgment was evaluated by analysis of four questions in the comprehension portion of the WAIS-R.²⁸ Aphasia was assessed by the short version of the Token test (comprehension; parts 4 and 6, total count of correct responses)²⁹ and the Boston naming test (15 items).³⁰ Verbal fluency was evaluated in both semantic category (animal naming in 1 minute) and letter generation (letter k in 1 minute).³¹ Overall evaluation of speech function was based on the modified Boston Diagnostic Aphasia Examination.^{29 32} Motor control and praxis were evaluated with items from the D test,^{30 32} including reciprocal coordination and serial and spatial organization of hand movements. Agnosia (perceptual functions) was assessed with the recognition tasks from the D test battery and Poppelreuters pictures.^{33 34} Constructional functions were assessed by the block design section of the WAIS-R,²⁸ by the clock test (recognizing and setting time),³³ and by copying a triangle, a flag, a three-dimensional cube, and a Greek cross.^{33 34} According to the neuropsychologist’s best judgment, all the patients included in the study were able to complete the tests to arrive at clinical decisions of normality and abnormality in abstract thinking, judgment, agnosia, and apraxia.

The patients clinically scored as having dysphasia had only mild symptoms: the mean (SD) score of the Acute Aphasia Screening Protocol was 45.1 (4.3) in those with and 49.1 (3.1) in those without dysphasia (P<.001).

Education was divided into two categories: low with 0 to 6 years and high with more than 6 years of formal education.

Results

Of the 377 patients with ischemic stroke, dementia was present in 107 (31.8%). The patients with dementia were older and more often had less than 6 years of education (Table 1⇓).

Among vascular risk factors, only current smoking and cardiac failure were associated with dementia, and high total cholesterol was more frequent in the nondemented group (Table 2⇓). History of any prior CVD (28.0% versus 17.4%; P=.0295) and prior ischemic stroke (26.2% versus 15.7%; P=.0220), but not that of TIA (5.6% versus 16.5%; P=.0056), were more frequent in the demented than in the nondemented group.

Characteristics of ischemic stroke associated with dementia were left hemispheric localization (P=.0366) and major dominant hemispheric stroke syndrome (P=.0002), but not stroke type. The nondemented patients more often had stroke located in the right hemisphere (P=.0482) or had minor nondominant stroke syndrome (P=.0140) compared with the demented patients (Table 3⇓).

Dysphasia in the clinical neurological examination was more frequent in the demented group (20.6% versus 3.5%; P<.001) (Table 4⇓). Of the 30 patients with dysphasia, 22 (73.3%) had dementia. Compared with the nondemented dysphasic patients, the demented patients did not differ significantly in regard to stroke localization on the left side (87.5% versus 81.8%), presence of major stroke syndrome (12.5% versus 36.4%), history of previous stoke (25.0% versus 18.2%), or presence of cognitive decline before index stroke (25.0% versus 31.8%). Of the 30 patients with dysphasia, 9 (30%) had major dominant stroke syndrome, 17 (56.7%) minor dominant stroke syndrome, and 4 (13.3%) another stroke syndrome. Of the 29 patients with major dominant stroke syndrome, 9 (31%) had dysphasia and 20 (69%) had hemianopia and dysmnesia.

Of the main neurological signs, gait impairment (43.9% versus 25.7%; P=.001) and urinary incontinence (14.0% versus 3.5%; P<.001) were more frequent in the demented group (Table 4⇑). The demented patients had lower mean values of systolic (P=.0338) and diastolic (P=.0075) arterial blood pressure when supine and had more frequently an orthostatic reaction (21.5% versus 14.3%, P=.045). None of the laboratory measures showed differences between the two groups (data not shown).

We used a logistic regression model to identify the correlates of dementia in the whole series of the aforementioned variables. The following correlates were found (n=337, demented n=107, model A): dysphasia (OR, 5.64; 95% CI,2.26 to 15.5), major dominant stroke syndrome (OR, 5.0; 95% CI, 1.92 to 14.1), any prior CVD (OR, 2.01; 95% CI, 1.073 to 3.74), and low level of education (OR, 1.13; 95% CI, 1.05 to 1.22) (Table 5⇓).

In the second approach, when those with mixed AD+VD (n=306, demented n=87, model B) were excluded, the correlates of dementia remained the same. Furthermore, in the third approach in which only the patients with first-ever stroke were included (n=273, demented n=79, model C) dysphasia, major dominant stroke syndrome and education were the correlates of dementia. Finally, in the fourth approach in which patients with dysphasia were excluded (n=307, demented n=85, model D), major dominant stroke syndrome and education remained correlates of dementia.

Discussion

We report here the largest well-defined stroke cohort thus far to examine cross-sectional features related to risk of dementia. The patients with poststroke dementia in the present series were more often older and had a lower level of education, history of prior CVD and stroke, history of current smoking, cardiac failure, left hemispheric stroke (reflecting laterality), a major dominant stroke syndrome (reflecting laterality and size), dysphasia, gait impairment, urinary incontinence, lower arterial blood pressure values, and more frequently had an orthostatic reaction compared with the nondemented stroke patients.

The correlates of dementia in logistic regression analysis were dysphasia, major dominant stroke syndrome, history of previous CVD, and low level of education. When those with CVD+AD or those with recurrent stroke were excluded, the order of correlates remained the same. When the patients with dysphasia (n=30) were excluded from the logistic model, the correlates were major dominant syndrome and education.

In our series, presence of dysphasia in the clinical examination was associated with dementia in the whole series, in those with stroke-related dementia (mixed AD+VD excluded), and in those with first-ever stroke. From the series, we had excluded patients with severe aphasia after attempting to test everyone and included only those who could be tested, as suggested in recent guidelines.³⁷ In a similar stroke cohort using the DSM-III definition of dementia (n=251, demented n=61), Tatemichi et al⁶ did not find an association. In both series the frequency of dysphasia was close (8.9% versus 9.7%), but in our study the percentage of demented patients with dysphasia was higher (20.6% versus 12.6%). In the series of Ladurner et al⁸ (n=71, demented n=41), dysphasia was also more frequent in the demented subjects. Furthermore, in a smaller series (n=110, demented n=15) focused on dementia after first-ever stroke, dysphasia was also independently associated with dementia.⁷ The use of different criteria for judging aphasia to be severe enough to necessitate exclusion from a study may explain differences between the cited studies.

Major dominant stroke syndrome was also a predictor of dementia independent of the effect of dysphasia in the present study, which agrees with the findings of Tatemichi et al⁶ and Ladurner et al.⁸ Left hemispheric stroke localization was also more common in demented patients, as shown previously.^{6 9} In the series of Censori et al,⁷ dominant hemisphere stroke was more frequent among the demented than nondemented patients (80.0% versus 50.6%), but the independent correlates were total anterior circulation stroke and frontal lesions. Because the focus of this report was on clinical factors, we did not examine quantitative brain imaging features. Although dysphasia is closely connected to the concept of major dominant stroke syndrome, in the present series 69% of the patients with major dominant stroke syndrome had dysmnesia and 31% dysphasia. In the series of Tatemichi et al,⁶ major dominant stroke syndrome was found in 12% of the patients, and of these it was reported that 10 (40.0%) did not have cognitive impairment, including language impairment.

Major stroke syndrome relates to left-sided lesions and reflects both language and verbal and nonverbal memory disturbance that may occur from unilateral left-sided damage.^{6 38 39} Memory and language deficits are essential domains in definitions used for dementia³ and are also known to be potential consequences of stroke. Language impairment seems to also have an independent correlation with dementia that is not explained by the consequences of major dominant hemispheric stroke, partly because nondominant stroke syndromes can also affect language and other cognitive domains.

In the present series, none of the stroke types were associated with dementia. In the series of Tatemichi et al,⁶ small-vessel occlusion (lacunae) was an independent predictor of poststroke dementia. This may well reflect selection since lacunar lesions predominate in patients with small-vessel disease and subcortical ischemic lesions.⁴⁰ However, possible selection and small sample size may have underpowered this study to detect a possible association. In our series, including an older random population, the frequency of stroke of undetermined etiology was 60%. Originally TOAST was designed for clinical drug trials, in which the frequency of undetermined stroke has been more than 40%.⁴¹ Our higher figure is mainly due to the strict classification system, including clinical definition of lacunar syndromes, lack of repeated brain images early in the course, and the absence of systematic cardiac assessments. Accordingly, both stroke due to small-vessel occlusion (lacunar) and cardioembolism are likely underestimated in our series.¹⁸ In addition to our series, the only other stroke cohort examining both the frequency of dementia and types of stroke thus far is that of Tatemichi et al.⁶ In that series the frequency of lacunar stroke was 29.7% among the nondemented and 36.4% in the demented subgroups, and the frequency of DSM-III dementia was 26.3%. Whether difference in distribution of lacunar stroke (8% in the study presented here) could have had an effect on the difference in frequency of dementia compared with our series (31.8%) cannot be confirmed.

Only completed previous strokes seemed to be a risk factor for VD (26.2% versus 15.7%; P=.0220), not the history of TIAs, which actually seemed to occur less often (5.6% versus 16.5%; P=.0056) in patients with dementia in our series. In the series of Tatemichi et al,⁶ demented patients did not differ from nondemented patients in regard the history of prior TIA (17.2% versus 19.1%; P=.732), but demented subjects more often had a history of stroke (31.8% versus 20.0%; P=.064). A history of TIA increases the risk of subsequent stroke,⁴² but a history of TIA at the time of ischemic stroke did not influence outcome in 3 years of follow-up.⁴³

Low level of education was a correlate of dementia, as shown in two previous poststroke series,^{6 9} as well as in population-based series on dementia.⁴⁴ The norms on cognitive tests were not adjusted for education because so few subjects were highly educated in this cohort. However, this may have slightly influenced the diagnosis of dementia. Higher education likely relates to larger functional cognitive reserve⁴⁵ and also to differences in lifestyle and risk factor profile.⁴⁶

Of the vascular risk factors, current smoking was associated with dementia in the present series, as in the series of Gorelic et al,⁹ where it was an independent correlate of dementia poststroke. Previously diabetes,^{6 7} myocardial infarction,⁹ and arterial hypertension⁸ have been related to poststroke dementia, which could not be confirmed in the present series.

In epidemiological studies, risk factors of VD have included arterial hypertension,^{10 12 13} cardiac disorders,^{10 11 12} diabetes,^{10 11} prior stroke,^{11 13} high hematocrit level,¹² and alcohol abuse¹³ in different combinations. Recently, Skoog et al⁴⁷ related high midlife blood pressure to a higher risk of dementia later in life. In the present study mean arterial blood pressure was lower in the demented group. In addition, in the series of Guo et al⁴⁸ the demented patients showed lower arterial blood pressure values at the time of diagnosis, which was not the case in the series of Skoog et al.⁴⁷ Low blood pressure has been related to the severity of dementia,⁴⁸ which may be a factor in explaining these differences. The demented patients in our series more frequently had abnormal orthostatic reaction. Thus far the association between postural hypotension and dementia is still controversial. In a general aged population, postural hypotension was not associated with dementia.⁴⁹ On the other hand, the presence of more frequent orthostatic reaction among patients with poststroke dementia may reflect changes in autoregulation due to small-vessel changes or other ischemic brain lesions.⁴⁷

Our data suggest that a single explanation for poststroke dementia is not adequate and that multiple factors, including stroke features, host characteristics, and prior CVD, each independently contribute to the risk. In addition to ischemic brain changes, coexisting AD-type pathology may also be a factor, as indicated by the number of patients with prestroke cognitive decline and presumed coexisting AD.

Selected Abbreviations and Acronyms