Cerebral amyloid angiopathy and cognitive function

Study population.

The HAAS is a follow-up to the Honolulu Heart Program (HHP), a longitudinal study of heart disease and stroke. The HHP cohort was comprised of 8006 Japanese-American men born between 1900 and 1919 and living on Oahu at the time of enrollment (1965 to 1968).6 In 1991, surviving HHP cohort members were invited to join the HAAS7: 80% (n = 3734) of the surviving participants were examined. In 1994 through 1996, 2,704 participants returned for a follow-up examination. During the HAAS visits, information on medical history and psychosocial factors was obtained and clinical and neuropsychologic evaluations were performed. Blood collected in 1991 through 1993 was used for APOE genotyping.8 The study was approved by the Institutional Review Boards of the Kuakini Medical Center and the Honolulu Department of Veterans Affairs. All participants gave written informed consent.

In 1991 through 1993, the autopsy study was established. Nondemented participants were asked to sign an autopsy consent form and to share copies with their family. An appropriate family member was asked to sign in the case of demented participants. The entire cohort was eligible for the autopsy study, but participants with a diagnosis of dementia were targeted to ensure an adequate sample size for comparisons of cases to controls. To date, 259 individuals have come to autopsy: 71.0% died in the hospital, 13.1% died at home, 11.2% died in a nursing home, and 4.6% died in a hospice or other place. Consent for the autopsy was obtained from a family representative at the time of death.

Assessment of cognitive function and dementia.

At both the HAAS baseline (1991 through 1993) and follow-up visit (1994 through 1996), participants were evaluated with the Cognitive Abilities Screening Instrument (CASI). The CASI is a cross-culturally validated test of global cognitive function designed for use in comparative studies of dementia in the United States and Japan.9 It is a composite of the Hasegawa Dementia Screening Scale,10 the Mini-Mental State Examination,11 and the Modified Mini-Mental State Test.12 The CASI consists of seven sections that assess attention, concentration, orientation, short- and long-term memory, language ability, visual construction, word list generation, abstraction, and judgment. Scores may range from 0 to 100, with higher scores indicating better overall cognitive function. The CASI score obtained closest to time of death was used in the analyses.

The dementia workup has been described in detail elsewhere.7 Briefly, after the CASI an additional workup was performed on a subsample of individuals that included the following: a repeat CASI, neurologic examination, hearing and vision testing, proxy interview including the Informant Questionnaire on Cognitive Decline in the Elderly,13 neuroimaging, blood tests, and the Consortium to Establish a Registry for AD (CERAD) neuropsychologic battery.14 A panel composed of the study neurologist and at least two other physicians with expertise in geriatrics and dementia reviewed subjects’ information and assigned a clinical diagnosis. The panel used the criteria from the Diagnostic and Statistical Manual of Mental Disorders, 3rd ed.—rev. to diagnose dementia.15 AD was diagnosed using the criteria of the National Institute of Neurological and Communicative Disorders and Stroke/AD and Related Disorders Association16 and VaD was diagnosed using the California AD and Treatment Centers criteria.17

Neuropathologic assessment.

The HAAS autopsy study aims to identify pathology related to neurodegenerative and cerebrovascular pathologies. In this study, we focused on the pathologies of CAA and AD and a subset of other cerebrovascular pathology. Brains were removed from the skull, weighed, and immersed in 10% buffered formalin for 2 to 6 weeks. After fixation, brains were reweighed and the dural membranes and external brain surfaces (including blood vessels and leptomeninges) were removed. A neuropathologist examined 1-cm-thick coronal sections of cerebral cortex and 3-mm-thick transverse sections of brainstem and cerebellum for macroscopic lesions. Microscopic assessments were performed by one of three neuropathologists blinded to clinical information. The analyses of CAA, neurofibrillary tangles (NFT), and neuritic plaques (NP) were performed in a microscopic examination of four neocortical areas—middle frontal gyrus, superior–middle temporal gyri, inferior parietal lobule, and occipital association cortex along the calcarine sulcus. The analysis of microscopic cerebrovascular lesions was performed in evaluations of 32 brain regions, which included the four neocortical areas.

Assessment of CAA.

Eight-micron blocks of paraffin embedded tissue were prepared using standard histologic methods and stained with hematoxylin and eosin. To detect CAA, sections were immunostained for βA4 amyloid (clone 10D5, Athena Neurosciences, San Francisco, CA).18 Blood vessels in each neocortical section were examined under 250× magnification and separate grades were given to meningeal and parenchymal vessels. The measure of CAA obtained from parenchymal vessels (mainly arteries and arterioles) was used in these analyses. CAA assessments were standardized among the three raters according to published guidelines.19 The raters discussed specific problems related to field selection and grading during training sessions. A grade of “mild” CAA was given to sections containing one or two βA4-positive vessels. Sections with three to five positive vessels were designated as “moderate,” and sections with greater than five positive vessels were designated as “severe.” If all vessels within an area were nonreactive the section was designated “CAA absent.” For this study, CAA grades from the four neocortical areas were combined to yield an overall CAA grade. Men free of CAA in all four areas received an overall grade of CAA absent. An overall grade of mild was given to men with only mild CAA in one or more areas. Men with moderate CAA in at least one area received an overall grade of moderate, and men with severe CAA in at least one area received an overall grade of severe.

Assessment of AD neuropathology.

The assessment for AD pathology has been described elsewhere.20 Briefly, senile plaques and NFT were visualized with modified Bielschowsky21 and Gallyas silver stains and examined under magnifications of 100× and 200×. Plaques containing silver-positive neurites (NP) were differentiated from neurite-free (diffuse) plaques and separate counts were made. To obtain maximum NP and NFT counts, five fields standardized to 1 mm² in diameter were examined for each neocortical area. In each region, the field with the highest count was taken to represent the area.22 NP counts were truncated at 17/mm² but there was no upper limit for NFT counts.

Assessment of cerebrovascular lesions.

Large and small infarcts were detected in the macroscopic examination and defined as circumscribed, cavitary lesions traversed by gliovascular trabeculae and distant from surface structures in the brain. Infarcts less than 1 mm in size were designated as lacunae. Lesions with a substantial collection of blood or hemosiderin and minimal evidence of ischemic infarction were designated large (>1 cm) or small (<1 cm) hemorrhages; these were detected on the macroscopic or microscopic examination.

Neuropathologic diagnosis of AD.

The CERAD neuropathology protocol23 was adapted to assign a postmortem diagnosis of neuropathologically confirmed AD (nAD). To meet the requirements for this diagnosis individuals were required to have a clinical history of dementia and a minimum number of NP—4 for probable AD and 17 for definite AD. Of the 22 men in our sample with a clinical diagnosis of AD, 12 had definite or probable nAD. Of the 14 men with a clinical diagnosis of AD with cerebrovascular disease (AD with CVD), four met the criteria for nAD. The cases that were neuropathologically confirmed did not differ from those who were not confirmed with regard to age, education, time between last CASI examination and autopsy, frequency of the APOE-ε4 allele, or presence of infarct. The confirmed cases had a slightly lower CASI score and more lacunae. Further neuropathologic descriptions of the clinical cases of AD and AD with CVD that did not meet CERAD criteria have been reported elsewhere.20

Analytical sample.

Our final sample consisted of 211 (81.5%) of the 259 completed autopsy cases; 41 individuals died before the CASI could be administered and seven individuals could not be tested. The mean length of time between the HAAS baseline examination and death was 3.9 years (range, <1 to 7.6 years) and the mean length of time between CASI administration and death was 2.4 ± 0.09 years. The mean age at death in our sample (84.7 ± 5.4 years) was similar to that for all nonautopsied decedents (85.3 ± 6.0 years). Years of education completed (9.8 ± 3.3 vs 10.0 ± 3.2 vs 10.7 ± 3.2 years) and late-life total cholesterol values (186.0 ± 34.5 vs 183.1 ± 35.4 vs 192.3 ± 31.8 mg/dL) were similar among our sample, nonautopsied decedents, and surviving HAAS cohort members. The percentage of individuals in our sample with at least one APOE-ε4 allele (12.8%) was slightly lower compared with the other groups (18%). By design, approximately 35% of autopsied decedents were demented (n = 73) compared with 18.4% of nonautopsied decedents. Participants still alive at the time of analysis were significantly younger and less often demented (4.3%).

Data analysis.

Sixty-seven subjects had severe CAA, 20 had moderate CAA and 10 had mild CAA. To increase statistical power we dichotomized the CAA variables as “CAA present” (overall CAA grade of mild, moderate, or severe CAA) and “CAA absent.” The CAA groups were compared with analysis of variance for continuous variables and with χ² tests for categorical variables. Linear regression was used to examine the association of CASI score to CAA and clinical dementia diagnosis (all dementia, AD, AD with CVD, and VaD). We tested the joint effect of CAA and dementia on cognitive function by subdividing the subjects into four groups: nondemented/CAA absent (reference group), nondemented/CAA present, demented (subtype)/CAA absent, and demented (subtype)/CAA present. We also tested whether the interaction between CAA and dementia (subtype) was significant by putting the cross-product term (CAA × dementia diagnosis) into the models predicting CASI score. Possible confounding or mediating factors considered in the models included the following: APOE genotype (ε3,3 [n = 160], ε2,3 [n = 14], ε2,4/3,4/4,4 [n = 27], and missing [n = 10]), age at death, time between CASI administration and death, education level, and neuropathologic findings, including NP and NFT count and presence of infarcts and hemorrhages.

Two models are shown. The basic model (Model 1) is adjusted for age at death, time between CASI administration and death, and education level. The second model is fully adjusted for all confounders and includes the Model 1 variables as well as APOE genotype and neuropathologic measures. We also checked whether the separate addition of APOE genotype or any of the neuropathologic variables to Model 1 significantly altered the relationship between dementia (subtype), CAA, and CASI score. The AD/CAA present and AD/CAA absent groups were analyzed based on both clinical (n = 22) and neuropathologic (n = 16) diagnoses of AD. All analyses were conducted with SAS statistical software, release 6.12 (SAS Institute, Cary, NC).24