Stroke and Memory Performance in Elderly without Dementia

Abstract

Background

There is conflicting data showing that stroke is associated with a higher risk of dementia and a more severe decline in persons with cognitive impairment. However, if cerebrovascular disease is directly related to cognitive decline in the absence of cognitive impairment or dementia remains unclear.

Objective

To examine the association between stroke and changes in cognitive function over time in elderly persons without dementia at baseline.

Design

The results of neuropsychological tests from several intervals over a five-year-period were clustered into domains of memory, abstract/visuospatial and language in 1271 elderly without dementia or cognitive decline. Stroke was related to the slope of performance in each cognitive domain using generalized estimating equations.

Results

Memory performance declined over time while abstract/visuospatial and language performance remained stable over the study period. Stroke was associated with a more rapid decline in memory performance, while there was no association between stroke and decline in abstract/visuospatial or language performance. The association between stroke and decline in memory performance was strongest for men and for persons without an APOE4 allele. A significant association between stroke and decline in abstract/visuospatial performance was also observed for persons without the APOE-e4 allele.

Conclusion

A history of stroke is related to a progressive decline in memory and abstract/visuospatial performance especially among men and those without an APOE-e4 allele.

INTRODUCTION

Cerebrovascular disease and dementia are among the most common diseases in aging societies. According to the WHO, cerebrovascular disease is the second leading cause of mortality in western societies and the major cause of long-term disability leaving 30% disabled ¹. About 1 percent of people aged 65-69 years have dementia, and this proportion increases with age to approximately 60% percent for people over the age of 95 ².

The role of stroke in the pathogenesis of cognitive decline remains unclear. Longitudinal population-based studies indicate that vascular risk factors, such as diabetes or hypertension are associated with stroke, which in turn may be related to the development of vascular dementia and Alzheimer's disease (AD) ^3,4. We previously reported a relation between stroke and the risk of AD ⁵. Vascular risk factors have also been associated with mild cognitive impairment (MCI) ^6,7, and there is evidence that cerebrovascular disease is associated with more progressive decline in persons with cognitive impairment ^8,9. However, whether or not cerebrovascular disease is directly related to cognitive decline in the absence of cognitive impairment or dementia remains unclear.

The objective of this study was to determine if the effects of stroke result in a decline in memory and other cognitive functions in elderly persons who do not have cognitive impairment or dementia.

METHODS

Subjects and Setting

Participants were part of a longitudinal study of Medicare recipients, aged 65 years or older, residing in northern Manhattan (Washington Heights, Hamilton Heights, Inwood)¹⁰. Each participant underwent an in-person interview of general health and function at the time of study entry followed by a standard assessment, including medical history, physical and neurological examination as well as a neuropsychological battery ¹¹. Baseline data were collected from 1992 through 1994. Follow-up data were collected during evaluations at sequential intervals of approximately 18 months, performed from 1994 to 1996, 1996 to 1997, and 1997 to 1999. In this elderly population, some participants did not complete follow up at all intervals due to refusal to participate further, relocation or death. About one half of participants were evaluated at the third follow-up visit. This study was approved by the institutional review board of the Columbia-Presbyterian Medical Center.

The participants selected for this study were without dementia or cognitive impairment at baseline, complete stroke information, and with at least 3 follow-up intervals.

Of the 2126 individuals who underwent clinical assessment at baseline, 346 (16.3%) individuals were excluded because they were demented at the initial intake examination. Information on stroke was unavailable in 83 (3.9%) cases and 426 (20.0%) subjects had less than three follow-up visits with neuropsychological evaluation (Figure 1). Thus, the study focused on 1271 individuals without dementia or cognitive impairment at baseline, followed over a 5 year interval.

Figure 1.

Description of sample size.

Clinical assessments

Data included medical, neurological, and neuropsychological evaluations ^12,13. All participants underwent a standardized neuropsychological test battery in either English or Spanish ¹⁴. Orientation was evaluated using parts of the modified Mini-Mental State Examination ¹⁵. Language was assessed using the Boston Naming Test ¹⁶, the Controlled Word Association Test ¹⁷, category naming, and the Complex Ideational Material and Phrase Repetition subtests from the Boston Diagnostic Aphasia Evaluation ¹⁸. Abstract Reasoning was evaluated using WAIS-R Similarities subtest ¹⁹, and the non-verbal Identities and Oddities subtest of the Mattis Dementia Rating Scale ²⁰. Visuospatial ability was examined using the Rosen Drawing Test ²¹, and a matching version of the Benton Visual Retention Test ²². Memory was evaluated using the multiple choice version of the Benton Visual Retention Test ²² and the seven subtests of the Selective Reminding Test ²³: total recall, long-term recall, long-term storage, continuous long-term storage, words recalled on last trial, delayed recall, and delayed recognition. This neuropsychological test battery has established norms for the same community ²⁴.

Definition of dementia and cognitive impairment

Results from the neurological, psychiatric and neuropsychological examinations were reviewed in a consensus conference comprised of physicians, neurologists, neuropsychologists and psychiatrists. Based on this review all participants were assigned to one of three categories: normal cognitive function, cognitive impairment without dementia, or dementia. Cognitive impairment without dementia was defined as the presence of abnormal neuropsychological tests for age, sex, and education group without significant cognitive impairment, and a Clinical Dementia Rating (CDR) of 0.5 ²⁵. Dementia was defined as the presence of abnormalities in several cognitive domains in neuropsychiatric testing accompanied by significant functional impairment (CDR ≥ 1).

Stroke

Stroke was defined according to the WHO criteria ²⁶. At baseline, the presence of stroke was ascertained from an interview with participants and their informants. Positive response(s) to any 1 of the 8 questions shown in Figure 2 was considered as suggestive of a history of stroke. Persons with stroke were confirmed through their medical records, 85% of which included results of brain imaging. The remainder were confirmed by direct examination.

Figure 2.

Survey questions assessing stroke. Stroke was defined as an affirmative answer to one of these questions.

APOE Genotyping

APOE genotypes were determined as described by Hixson and Vernier ²⁷ with slight modification ²⁸. We classified persons as homozygeous or heterozygeous for the APOE ε4 allele or not having any ε4 allele.

Other covariates

Diabetes mellitus and hypertension were defined by self-report at baseline and at each follow-up interval or by the use of disease specific medications. Blood pressure measurements were also considered in the definition of hypertension. Heart disease was defined as a history of myocardial infarction, congestive heart failure or angina pectoris at any time during life. Body mass index (BMI) was calculated by the formula BMI = weight (Kg)/height (m)². Smoking was assessed by self-report and categorized as never, past and current smoking.

Statistical Methods

A factor analysis was performed using data from the entire cohort with the 15 neuropsychological measures using a principal component analysis with varimax rotation and Kaiser normalization ²⁹. This analysis resulted in three factors: 1) a memory factor, in which the seven subtests of the Selective Reminding Test were the main contributors; 2) a abstract/visuospatial factor, where visuospatial and tests of reasoning were the main contributors; and 3) a language factor, in which language measures from the Boston Naming Test (21), Controlled Oral Word Association Test ¹⁷, and the WAIS-R Similarities ¹⁹ were the main contributors. We calculated cognitive scores for each participant at each visit by adding the scores of the measures that contributed most to each factor (tests with correlations of 0.5 or higher). Each factor score was normally distributed.

Generalized estimating equations (GEE) ³⁰ were used to examine changes in each cognitive domain over time. The dependent variables were the factor scores, and the independent variables were stroke, time (included as a continuous variable, and representing the time of follow-up of each participant), and the interaction of stroke and time. After adjusting for age and gender, subsequent models were adjusted for age, gender, education, ethnic group, APOEε4 genotype, BMI, hypertension, heart disease, diabetes and smoking. In these full models age, education and BMI were included as continuous variables, ethnic group, APOEε4 genotype and smoking as multilevel categorical variables, and hypertension, heart disease and diabetes as dichotomized (not present vs. present) variables.

The GEE analysis yielded coefficient values that represent the associations between a factor score and variables included in the model. There were three main coefficients of interest in each model: one comparing the stroke groups (stroke yes/no) at baseline, one relating the change in cognitive scores with time, and an interaction term for stroke and time. A significant p value for the coefficient comparing stroke groups at baseline indicates a difference between two groups at baseline. A significant p value for the coefficient of time indicates a statistically significant change in a cognitive score over the total duration of follow-up. A significant p value for the interaction coefficient indicates a difference in the rate of change in a factor score depending on the stroke group; this is the main variable of interest for the interpretation of the analyses. All analyses were repeated after stratifying for gender and APOEε4genotype.

RESULTS

The mean age of the sample was 76.2 ± 6.0 years, 69.6% were women, 45.1% were Hispanic, 20.6% were White, and 33.7% were Blacks. The mean of years of education was 8.6 ± 4.6, and 20.8% were homozygous or heterozygous for the APOE-ε4 allele. The mean BMI was 27.1 ± 5.1, and 29.8% of the subjects reported having diabetes, 55.1% hypertension and 29.5% heart disease. 7.6% had a history of stroke. Persons with stroke at baseline had a higher prevalence of diabetes and hypertension than persons without stroke (Table 1). There were no significant differences in stroke prevalence among gender or ethnic groups.

Table 1.

Comparison of demographic characteristics between persons with and without stroke at baseline

Covariates	No Stroke (n=1174)	Stroke (n=97)
Men	359 (30.6)	27 (27.8)
Women	815 (69.4)	70 (72.2)
Education, mean (SD), year	8.6 (4.6)	8.9 (4.3)
Age, mean (SD), year	76.2 (6.0)	76.3 (5.9)
Body mass index, mean (SD)	27.1 (5.1)	27.3 (4.6)
Ethnic group ‡
White/Non-Hispanic	239 (20.4)	23 (23.7)
Black/Non-Hispanic	390 (33.2)	38 (39.2)
Hispanic	538 (45.8)	35 (36.1)
APOE genotype 4/4	21 (2.2)	-
APOE genotype 4/-	255 (26.2)	19 (26.0)
APOE genotype -/-	699 (71.7)	54 (74.0)
Diabetes	352 (29.9)	30 (35.4)*
Heart disease	343 (29.2)	28 (29.1)
Hypertension	630 (53.9)	71 (75.5)*

Values are expressed as number (percentage) unless otherwise indicated. Some percentages are based on an incomplete sample due to small amounts of missing data.

^‡Classified by self-report using the format of the 1990 US census ⁵⁷.

^*significant at a 0.05 level vs. group without stroke

In the GEE analysis memory declined significantly over time (ß=-1.6, p=0.005), while abstract/visuospatial and language performance remained stable over the study period (Table 2). A history of stroke was associated with more rapid decline in memory performance over time (ß =-3.6, p for interaction of stroke and time = 0.04). There was no relation between stroke and decline in abstract/visuospatial (ß =-0.1, p for interaction of stroke and time = 0.9) or language performance (ß =0.1, p for interaction of stroke and time = 0.5). There was also no relation when analyses were repeated for individual tests in the abstract/visuospatial and language domains.

Table 2.

Relationship of Stroke and Time of Follow-up to Memory, Abstract/visuospatial and Language Performance in 1271 Healthy Elderly Over 7 years

	Model 1		Model 2
Variable	Estimated β (SE)	p-value	Estimated β (SE)	p-value
Memory Performance
Time	-1.6 (0.6)	0.005	-1.6 (0.6)	0.006
Stroke	-2.3 (5.9)	0.7	-1.2 (6.2)	0.8
Time*Stroke	-3.6 (1.8)	0.04	-3.5 (1.9)	0.05
Abstract/visuospatial Performance
Time	0.1 (0.3)	0.7	0.1 (0.3)	0.6
Stroke	-2.2 (2.8)	0.4	0.3 (3.1)	0.9
Time*Stroke	-0.1 (0.6)	0.9	-0.2 (0.6)	0.7
Language Performance
Time	-0.1 (0.1)	0.9	0.1 (0.1)	0.9
Stroke	-0.1 (0.5)	0.8	-0.1 (0.5)	0.8
Time*Stroke	0.1 (0.1)	0.5	0.1 (0.1)	0.4

Model 1 is adjusted for age and gender, Model 2 is adjusted for age, gender, education, ethnic group, APOEε4 genotype, BMI, hypertension, heart disease, diabetes and smoking

All analyses were repeated stratifying by gender and APOEε4 genotype. While in both men and women as well as APOE ε4 carriers and non-carriers memory performance significantly declined over time, the association between stroke and decline in memory performance over time (stroke*time interaction) was stronger in men (ß=-10.1, p=0.005, Table 3; p for interaction gender*stroke*time = 0.07) and persons without APOEε4 allele (ß =-4.1, p=0.07, Table 4; p for interaction APOEε4*stroke*time=0.09). Persons without APOEε4 allele also showed a significant stroke*time interaction indicating that abstract/visuospatial function declined faster among APOEε4-non-carriers (ß=-1.1, p=0.04; p for interaction APOEε4*stroke*time=0.07). Thus, memory and abstract/visuospatial function declined at a faster rate in men or persons who lacked the APOEε4 allele with stroke compared to women or APOEε4 carriers. These associations remained unchanged after adjusting for age, education, ethnic group, BMI, hypertension, heart disease, diabetes and smoking. There was no association between stroke and language performance.

Table 3.

Relationship of Stroke and Time of Follow-up to Memory, Abstract/visuospatial and Language Performance in 1271 Elderly Persons Over 7 years of Follow-up stratified by gender.

	Model 1		Model 2
Variable	Estimated β (SE)	p-value	Estimated β (SE)	p-value
Memory Performance
Men
Time	-1.8 (1.1)	0.1	-3.2 (1.2)	0.009
Stroke	8.6 (13.3)	0.5	12.0 (15.8)	0.5
Time*Stroke	-10.1 (3.6)	0.005	-9.9 (3.7)	0.008
Women
Time	-1.5 (0.7)	0.02	-1.6 (0.8)	0.04
Stroke	-6.7 (6.1)	0.2	-2.4 (7.6)	0.8
Time*Stroke	-1.3 (1.8)	0.4	-0.7 (2.1)	0.7
Abstract/visuospatial Performance
Men
Time	0.6 (0.5)	0.2	0.4 (0.5)	0.4
Stroke	3.6 (4.9)	0.4	8.6 (4.6)	0.07
Time*Stroke	-0.8 (1.2)	0.5	-0.9 (1.4)	0.5
Women
Time	-0.2 (0.3)	0.6	-0.1 (0.3)	0.8
Stroke	-4.7 (3.3)	0.2	-1.7 (4.39	0.7
Time*Stroke	0.2 (0.6)	0.8	0.2 (0.6)	0.8
Language Performance
Men
Time	-0.1 (0.1)	0.5	-0.1 (0.1)	0.4
Stroke	0.5 (0.8)	0.5	0.8 (0.9)	0.4
Time*Stroke	0.3 (0.2)	0.2	0.2 (0.2)	0.4
Women
Time	0.1 (0.1)	0.7	0.1 (0.1)	0.8
Stroke	-0.4 (0.6)	0.5	0.2 (0.7)	0.8
Time*Stroke	0.1 (0.1)	0.9	0.1 (0.2)	0.7

Model 1 is adjusted for age, Model 2 is adjusted for age, education, ethnic group, APOEε4 genotype, BMI, hypertension, heart disease, diabetes and smoking

Table 4.

Relationship of Stroke and Time of Follow-up to Memory, Abstract/visuospatial and Language Performance in 1271 Elderly Persons Over 7 years of Follow-up stratified by APOE genotype.

	Model 1		Model 2
Variable	Estimated β (SE)	p-value	Estimated β (SE)	p-value
Memory Performance
-/- APOEε4 genotype
Time	-1.3 (0.7)	0.06	-1.7 (0.8)	0.04
Stroke	-1.5 (7.3)	0.8	3.4 (8.4)	0.7
Time*Stroke	-4.1 (2.2)	0.07	-4.2 (2.4)	0.09
-/4 or 4/4 APOEε4 genotype
Time	-2.5 (1.1)	0.02	-3.2 (1.2)	0.008
Stroke	-3.4 (10.8)	0.8	-1.3 (14.0)	0.9
Time*Stroke	-2.6 (2.9)	0.4	-1.4 (3.2)	0.6
Abstract/visuospatial Performance
-/- APOEε4 genotype
Time	1.0 (3.5)	0.8	3.8 (4.1)	0.4
Stroke	-0.1 (3.1)	0.9	-0.1 (0.3)	0.9
Time*Stroke	-1.1 (0.5)	0.04	-1.0 (0.6)	0.06
-/4 or 4/4 APOEε4 genotype
Time	0.4 (0.4)	0.4	0.4 (0.4)	0.4
Stroke	-6.7 (4.7)	0.1	-3.3 (6.0)	0.6
Time*Stroke	1.7 (1.1)	0.1	2.0 (1.0)	0.07
Language Performance
-/- APOEε4 genotype
Time	0.1 (0.1)	0.8	-0.1 (0.1)	0.9
Stroke	-0.1 (0.6)	0.9	0.3 (0.6)	0.6
Time*Stroke	0.1 (0.1)	0.2	0.1 (0.1)	0.2
-/4 or 4/4 APOEε4 genotype
Time	-0.1 (0.1)	0.8	-0.1 (0.1)	0.8
Stroke	-0.4 (0.8)	0.6	0.1 (0.8)	0.9
Time*Stroke	-0.1 (0.2)	0.8	-0.1 (0.3)	0.7

Model 1 is adjusted for age and gender, Model 2 is adjusted for age, gender, education, ethnic group, BMI, hypertension, heart disease, diabetes and smoking

COMMENT

In this study the performance in memory, abstract/visuospatial and language domains declined over time in individuals free of dementia or cognitive impairment at baseline. A history of stroke was associated with faster decline only in memory performance. When stratified by sex or APOEε4 genotype, stroke was associated with a faster decline in memory or abstract/visuospatial performance in men or persons lacking the APOEε4 allele.

The mechanisms by which stroke increases the risk of cognitive decline are not clear. Stroke could increase the risk of cognitive decline by destruction of brain parenchyma and atrophy such as in the case of vascular dementia or AD associated with stroke ^31,32, or by causing damage in strategic locations that lead to amnestic syndromes, such as thalamic strokes ^33,34. Stroke could also increase the risk of cognitive decline by increasing the deposition of amyloid β, the key step in the pathogenesis of Alzheimer's disease ^35,36 or by a combination of these different mechanisms. It is also possible that the occurrence of stroke adds cognitive deficits in persons with subclinical AD that bring them over the diagnostic threshold, without directly affecting the deposition of amyloid beta, and that stroke does not have a direct specific effect on AD.

Studies examining the role of stroke in cognitive function reported inconsistent results. The Framingham Study reported in a nested case-control study a doubled risk of dementia after baseline stroke ³⁷, and a similar observation has been made earlier by a longitudinal study assessing the risk of incident dementia after cerebral infarction in 971 subjects in Minnesota ³⁸. Hospital-based cohorts with a follow-up shorter than 3 months also observed an increased risk of incident dementia after stroke ^39-41, and we previously reported an increased risk of dementia after stroke ⁴². Others have not found an association between cerebrovascular disease and cognitive impairment or dementia ^43,44.

Our results are consistent with studies showing an increased risk of AD in persons with stroke ^45,46. The main cognitive domain affected in AD is memory ^47,48 and it seems reasonable to postulate that if stroke is related to a higher risk of AD, it must be related to decline in memory. Furthermore, it seems that this effect is independent of APOE genotype, which is in agreement with studies indicating an increased risk of AD with stroke in persons without the APOEε4 allele ⁴⁹.

Stroke has been found to be related to impairment in frontal executive functions ^50-52. The domain in our study that better represents this construct is abstract/visuospatial performance, and we found no association of stroke to differences in this domain at baseline or with follow-up. The reasons for this negative finding may be that our cognitive battery lacked better measures of frontal/executive functions, such as the Color trails ⁵³.

There are several potential alternative explanations for our findings. One is chance, particularly in the context of multiple comparisons. However, this study was based on our previous findings relating stroke to a higher risk of AD ⁵⁴. Also, this study is consistent with other studies as described in the previous paragraph; these facts make chance due to multiple comparisons an unlikely explanation for our findings ⁵⁵. One of our findings was that stroke was related to faster cognitive decline in men. The strata for men was much smaller than for women, and only 27 men had stroke, and this could also result in chance findings. These findings should be reproduced in a larger cohort. Another potential explanation is bias. For example, that only subjects with preclinical AD reported stroke while subjects that would not develop AD did not. This type of reporting bias seems unlikely and we excluded cases of prevalent dementia or cognitive impairment that could have influenced our results. Further, if lower education is related to stroke, and persons with lower education are more likely to be diagnosed with AD, then it is possible that a relation between stroke and cognitive decline could be confounded by socioeconomic status. We adjusted for years of education and ethnicity as markers of socioeconomic status to account for this possibility. However, it is possible that stroke is related to other behaviors related to poor health, that in turn may increase the risk of AD, that we could not adjust for, and we cannot eliminate the possibility of lack of control for unknown confounders as a potential explanation for our findings. Finally, a potentially major source of bias, and the main limitation of our study, is the lack of ascertainment of sub-clinical cerebrovascular disease in persons without stroke. We also lacked information on the location and severity of cerebrovascular disease. If sub-clinical stroke is associated with cognitive decline as we hypothesized for clinical stroke, then our results are biased toward the null. Thus, our findings seem to underestimate the true relation of stroke to memory decline, and our negative findings for language and visuospatial abilities may be explained by this source of bias.

This study has several strengths. We had a comprehensive and sensitive neuropsychological battery validated for use in the communities of northern Manhattan ⁵⁶. We also excluded from our analyses persons with dementia and cognitive impairment without dementia at baseline that may have biased the analyses, and had several evaluation time points that allowed prospective analyses.