Risk Factors for Mild Cognitive Impairment in the Cardiovascular Health Study Cognition Study: Part 2

Objective  To examine the risk factors for mild cognitive impairment (MCI) in a longitudinal population study—the Cardiovascular Health Study Cognition Study.

Design  We examined the factors that in the period 1991 through 1994 predicted the development of MCI in all participants of the Cardiovascular Health Study Cognition Study. Further examination was conducted in the Pittsburgh, Pa, cohort (n = 927), where participants with MCI were classified as having either the MCI amnestic-type or the MCI multiple cognitive deficits–type.

Setting  Multicenter population study.

Patients  This study includes all participants of the Cardiovascular Health Study Cognition Study (n = 3608) who had a magnetic resonance imaging (MRI) scan of the brain between 1991 and 1994, and detailed neuropsychological, neurological, and medical evaluations to identify the presence of MCI or dementia in the period 1998 to 1999. The mean time between the closest clinical examination to the MRI and the diagnostic evaluation for cognitive disorders was 5.8 years for the Cardiovascular Health Study Cognition Study cohort and 6.0 years for the Pittsburgh cohort.

Main Outcome Measures  Risk factors for MCI at the time of the MRI were identified using logistic regression, controlling for age, race, educational level, baseline Modified Mini-Mental State Examination and Digit Symbol Test scores, measurements of depression, MRI findings (atrophy, ventricular volume, white matter lesions, and infarcts), the presence of the apolipoprotein E (APOE) ϵ4 allele, hypertension, diabetes mellitus, and heart disease.

Results  Mild cognitive impairment (n = 577) was associated with race (African American), low educational level, low Modified Mini-Mental State Examination and Digit Symbol Test scores, cortical atrophy, MRI-identified infarcts, and measurements of depression. The MCI amnestic-type was associated with MRI-identified infarcts, the presence of the APOE ϵ4 allele, and low Modified Mini-Mental State Examination scores. The MCI multiple cognitive deficits–type was associated with low Modified Mini-Mental State Examination and Digit Symbol Test scores.

Conclusions  The development of MCI is associated with measurements of cognition and depression, racial and constitutional factors, and cerebrovascular disease. Early cognitive deficits seem to be a common denominator for the 2 forms of MCI; the presence of cerebrovascular disease and the APOE ϵ4 allele is associated with the amnestic type of MCI.

OLDER PARTICIPANTS with mild cognitive impairment (MCI) have an increased risk of developing dementia, especially Alzheimer disease.¹ Therefore, the proper identification of these individuals is important because they constitute a clinical entity that is suitable for therapeutic interventions. However, one of the problems in identifying risk factors for MCI is that there are several neurological, systemic, and psychiatric syndromes that can cause cognitive impairment. Elderly subjects who have cerebrovascular disease,² white matter lesions (WMLs),³ diabetes mellitus,⁴ and hypertension and heart disease,⁵ or depression⁶ can present with mild cognitive deficits. Furthermore, studies that focused only on risk factors for MCI have found that hypertension diagnosed in midlife⁷ and a history of coronary artery bypass grafting⁸ increased the risk of developing MCI. In this study, we examined the risk factors for MCI in the context of a longitudinal population study, the Cardiovascular Health Study (CHS) Cognition Study. These factors were first examined for the whole CHS cohort, and further analysis was conducted in the Pitts burgh, Pa, sample, where we examined the association between risk factors and MCI subgroups.

Details of the CHS annual evaluations as well as the methods of the CHS Cognition Study are described in the companion article in this issue.⁹

The sulcal prominence (or cortical atrophy), ventricular grade, and white matter signal-intensity changes were assessed on a semiquantitative 10-point scale (grades 0-9) by using predefined visual standard atlases. Details of the CHS MRI examination have been published previously.¹⁰ For this study the MRI measurements were dichotomized as follows: ventricular size grade higher than 4, cortical atrophy grade higher than 4, and WMLs grade higher than 2.

Details of the CHS MCI criteria, as well as the criteria for probable and possible MCI are given in the companion article in the "CHS Cognitive Study MCI Criteria" subsection of the "Methods" section.⁹

Logistic regression models for the risk of MCI compared with healthy participants included age, educational level, the Center for Epidemiologic Studies–Depression (CES-D) Scale, Modified Mini-Mental State Examination (3MSE) and Digit Symbol Test (DST)⁹ scores closest to the MRI, MRI findings, and the presence of the apolipoprotein E ϵ4 (APOE ϵ4) allele, diabetes mellitus, hypertension, and/or ischemic heart disease. The analyses were conducted to determine the factors associated with the presence of MCI in all available participants with MCI and in healthy participants.

Participants with MCI were significantly older than the healthy participants. There were more African American participants in the MCI group than among the healthy participants. The participants with MCI had lower levels of education (ie, high school or less) and more of them carried the APOE ϵ4 allele. Participants with MCI had lower 3MSE and DST scores than did the healthy participants (Table 1).

Significantly greater proportions of participants with MCI had diabetes mellitus, heart disease (including myocardial infarction, congestive heart failure, or angina), hypertension, and a CES-D score exceeding 7 than the healthy participants. There was a greater proportion of participants with MCI who had a white matter lesion grade higher than 2, ventricular volume higher than 4, cortical atrophy higher than 4, and MRI-identified infarcts than the healthy participants (Table 2).

Mild cognitive impairment was associated with race (being African American), educational level, MRI-identified infarcts, a cortical atrophy grade higher than 4, low 3MSE and DST scores, and a CES-D score exceeding 7 (Table 3).

Distribution of risk factors for MCI in the Pittsburgh sample were similar to the results for the 4 centers (Sacramento, Calif; Winston-Salem, NC; Hagerstown, Md; and Pittsburgh) combined (Table 4). Participants with MCI amnestic-type (AT) and MCI multiple cognitive deficits–type (MCDT) were significantly different from healthy participants for lower 3MSE and DST scores. The MCI-MCDT group was older, less educated, and had higher percentages of African American participants. The 2 MCI types differed significantly from one another for educational level and 3MSE and DST scores. Participants with MCI-MCDT were less educated and had lower cognitive test scores than participants with MCI-AT (Table 4).

Significantly higher percentages of the participants with MCI had hypertension, diabetes mellitus, depression, and poor neuroradiological findings. Participants with MCI-AT had poor neuroradiological findings on all 4 measurements, while participants with MCI-MCDT had significantly higher percentages of diabetes mellitus, a WML grade higher than 2, and a ventricular volume higher than 4 (Table 5).

Mild cognitive impairment in the Pittsburgh subsample was associated with the presence of the APOE ϵ4 allele, MRI-identified infarcts, and low baseline 3MSE and DST scores. Mild cognitive impairment amnestic-type was associated with the same risk factors except for low DST scores; MCI-MCDT was significantly associated with 3MSE and DST scores (Table 6).

Further exploratory analysis revealed interesting associations that require further studies in larger cohorts. For example, probable MCI-AT (n = 10) was associated with the presence of the APOE ϵ4 allele (odds ratio [OR], 5.6; 95% confidence interval [CI], 1.35-23.4) and low 3MSE scores (OR, 0.8; 95% CI, 0.72-.94). Possible MCI-AT (n = 23) was associated with MRI-identified infarcts (OR, 4.3; 95% CI, 1.69-10.7) and WMLs (OR, 3.25; 95% CI, 1.37-8.77). Probable MCI-MCDT (n = 26) was associated with low DST scores (OR, 0.9; 95% CI, 0.88-0.99) and the presence of the APOE ϵ4 allele (OR, 3.3; 95% CI, 1.16-9.44). Possible MCI-MCDT (n = 79) was associated with MRI-identified infarcts (OR, 2.0; 95% CI, 1.10-3.80) and low 3MSE scores (OR, 0.8; 95% CI, 0.80-0.90).

Mild cognitive impairment was associated with measurements of cognition and depression, racial and constitutional factors, and the presence of cerebrovascular disease. Low scores on neuropsychological measurements were the common predictor for the 2 types of MCI, and they may have reflected the early effect of several disease processes (eg, heart disease, hypertension, diabetes mellitus, and cerebrovascular disease), an incipient neurodegenerative process, or both. To some extent, neuropsychological test results must be interpreted cautiously as they also formed the basis of our diagnosis.

The increased frequency of cerebrovascular risk factors and the presence of the APOE ϵ4 allele among African American participants may explain why elderly African Americans had a greater risk of developing MCI.¹¹ However, in this study, we found an association between being African American and having MCI independent of the presence of cerebrovascular risk factors and the presence of the APOE ϵ4 allele.

Several genetic studies have shown that participants who do not have dementia but are carrying the APOE ϵ4 allele had lower global cognitive performance than those without the allele, and participants with focal memory deficits who are carrying the APOE ϵ4 allele have the highest risk of subsequent development of dementia.¹ However, we found that the presence of the APOE ϵ4 allele was associated with MCI-AT, and especially, with the "pure" (probable) forms of MCI. Further studies are necessary to determine the relationship between the APOE ϵ4 allele and the specific types of MCI and dementia.

Magnetic resonance imaging–identified infarcts were important predictors of MCI, especially of the possible MCI-AT and MCI-MCDT subtypes. This finding converges with the notion that the presence of cerebrovascular risk factors (hypertension, diabetes mellitus, and heart disease) is associated with MCI.¹² However, in this study, the effect of cerebrovascular risk factors on the clinical expression of MCI was attenuated when neuroradiological findings were entered into the multivariate analysis. Furthermore, although participants with MCI had more WMLs than healthy participants, this association was not significant in the multiple logistic analysis. It is possible that the presence of other MRI findings and cardiovascular factors attenuated the effect of WMLs on the development of MCI. Importantly, the possible MCI-AT group was associated with both MRI-identified infarcts and WMLs, suggesting a greater cerebrovascular process in this subgroup. Clinical studies have shown that the presence of WMLs was closely related to MRI-identified infarcts and cerebrovascular risk factors.¹⁰ Therefore, additional studies are necessary to investigate the role that the association (and interaction) between cerebrovascular risk factors and the severity of cerebrovascular disease have in the development of specific MCI subtypes.

There are several issues that make depression an important factor in the study of the natural history of MCI and Alzheimer disease. (1) Idiopathic depression can cause cognitive deficits in elderly persons.⁶ (2) Elderly persons who do not have dementia but do have cardiovascular disease, strokes, and WMLs¹³ have more depression than those without ischemic lesions. (3) Cardiovascular disease increases the risk of depression in elderly persons.¹⁴ (4) Elderly persons who do not have dementia but are depressed have smaller hippocampal volumes than those who are not depressed.¹⁵

One limitation of our study is that we did not have detailed annual neuropsychological assessments, which would have allowed us to determine the exact time of onset of MCI. Longitudinal studies are necessary to determine how MCI subgroups evolve to dementia and especially to identify those participants with MCI whose conditions do not progress to dementia.

Corresponding author: Oscar L. Lopez, MD, 3501 Forbes Ave, Suite 830, Oxford Bldg, Pittsburgh, PA 15213 (e-mail: lopezol@msx.upmc.edu).

Accepted for publication May 28, 2003.

Author contributions: Study concept and design (Drs Lopez, Becker, DeKosky, Fitzpatrick, Lyketsos, Carlson, and Kuller and Mr Jagust); acquisition of data (Drs Lopez, Becker, DeKosky, Fitzpatrick, Breitner, Lyketsos, Jones, and Kuller and Mr Jagust); analysis and interpretation of data (Drs Lopez, Dulberg, Becker, DeKosky, Fitzpatrick, Breitner, Lyketsos, Jones, Kawas, and Kuller and Mr Jagust); drafting of the manuscript (Drs Lopez, DeKosky, and Kuller and Mr Jagust); critical revision of the manuscript for important intellectual content (Drs Lopez, Dulberg, Becker, DeKosky, Fitzpatrick, Breitner, Lyketsos, Jones, Kawas, Carlson, and Kuller and Mr Jagust); statistical expertise (Drs Lopez, Dulberg, and Kuller); obtained funding (Mr Jagust and Drs Becker, DeKosky, and Kuller); administrative, technical, and material support (Drs Lopez, Becker, DeKosky, Fitzpatrick, and Kuller); study supervision (Drs Lopez, Becker, and Fitzpatrick and Mr Jagust).

This study was supported by contracts N01-HC-85079 through N01-HC-85086, N01-HC-35129, and N01-HC-15103 from the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Md, and grants AG15928 and AG20098 from the National Institute on Aging, National Institutes of Health. Dr Becker is the recipient of a Research Scientist Development Award, Level II (K02-MH01077) from the National Institute of Mental Health, National Institutes of Health, Bethesda.