Cerebral Atherosclerosis Is Associated With Cystic Infarcts and Microinfarcts but Not Alzheimer Pathologic Changes
Background and Purpose—Some studies have reported associations between intracranial atherosclerosis and Alzheimer disease pathology. We aimed to correlate severity of cerebral atherosclerosis, arteriolosclerosis, and cerebral amyloid angiopathy with neurofibrillary tangles, neuritic plaques, and cerebral infarcts.
Methods—This autopsy study (n=163) was drawn from a longitudinal study of subcortical ischemic vascular disease, Alzheimer disease, and normal aging. Multivariable logistic regression models were used to test associations among the 3 forms of cerebrovascular disease and the presence of ischemic and neurodegenerative brain lesions. Apolipoprotein E genotype was included as a covariate in these multivariable models.
Results—Cerebral atherosclerosis was positively associated with microinfarcts (odds ratio [OR], 2.3; 95% confidence interval [CI], 1.2–4.4) and cystic infarcts (OR, 2.0; 95% CI, 1.0–4.2) but not Alzheimer disease pathology. Arteriolosclerosis showed a positive correlation with lacunar infarcts (OR, 2.0; 95% CI, 1.0–4.2) but not Alzheimer disease pathology. Cerebral amyloid angiopathy was inversely associated with lacunar infarcts (OR, 0.6; 95% CI, 0.41–1.1), but positively associated with Braak and Braak stage (OR, 1.5; 95% CI, 1.1–2.1) and Consortium to Establish a Registry for Alzheimer Disease plaque score (OR, 1.5; 95% CI, 1.1–2.2).
Conclusions—Microinfarcts, which have been correlated with severity of cognitive impairment, were most strongly associated with atherosclerosis. Possible pathogenetic mechanisms include artery-to-artery emboli, especially microemboli that may include atheroemboli or platelet-fibrin emboli. Arteriolosclerosis was positively, whereas cerebral amyloid angiopathy was negatively correlated with lacunar infarcts, which might prove helpful in clinical differentiation of arteriolosclerotic from cerebral amyloid angiopathy–related vascular brain injury.
The 3 most prevalent forms of late-life cerebrovascular disease (CVD) are atherosclerosis, arteriolosclerosis, and cerebral amyloid angiopathy (CAA). CVD is a well-established risk factor for ischemic vascular brain injury—regions of encephalomalacia which range in size from large cystic infarcts to lacunar infarcts to microinfarcts. Microinfarcts are strongly associated with cognitive impairment, especially in nondemented subjects,1,2 but their associations with type of CVD are unclear. CVD may also be associated with Alzheimer disease (AD) pathology. CAA and parenchymal AD pathology (namely, neurofibrillary tangles and amyloid-neuritic plaques) are frequently seen together; both are associated with apolipoprotein E ε4 (apoE ε4) and reduced clearance of β-amyloid from the brain.3 Associations between cerebral atherosclerosis and AD pathology have also been reported.4,5
We examined the relationship between CVD pathology, including atherosclerosis, arteriolosclerosis, and CAA, with measures of AD pathology and cerebral infarcts. The autopsy sample was derived from a longitudinal study focused on the effects of subcortical ischemic vascular disease (SIVD) and AD on brain structure and cognitive function.6,7 In this study, atherosclerosis refers to intimal thickening affecting basal large arteries; arteriolosclerosis refers to vessel wall thickening and sclerosis affecting arteries <300 µm in external diameter. We hypothesized that (1) atherosclerosis and arteriosclerosis would be associated with ischemic but not AD pathology; and (2) CAA would be associated with apoE ε4, AD pathology, and microinfarcts.
The sample comprises 163 autopsy cases from a multicenter longitudinal study of cognitively normal, SIVD, and AD participants (IVD Program Project, February 2011 neuropathology database). Among the first consecutive 175 autopsy subjects, 1 with fronto-temporal lobar degeneration and 11 with dementia with Lewy bodies (score≥3) were excluded from this analysis. The 175 autopsy cases were drawn from a sample of 736 subjects (291 deceased; autopsy rate 60%). Written informed consent was obtained from all subjects or surrogate decision maker after the protocols approved by the institutional review boards at each participating institution.
Subjects with cognitive impairment and dementia were recruited mainly from university-affiliated memory clinics; cognitively normal subjects were recruited from the community. The sample was enriched for subjects with SIVD, defined by the presence in proton-density MRI of discrete gray matter and white matter hyperintensities >2 mm in diameter, operationally defined as lacunes. Evidence of frank cerebral hemorrhage or cortical infarction excluded a subject from initial study enrollment, but not for continued clinical follow-up and autopsy.
Initial clinical diagnosis was on the basis of a comprehensive evaluation,8 including medical history, activities of daily living, physical and neurologic examination, Mini-Mental State exam (MMSE),9 laboratory testing, serial neuropsychological testing,10 and quantitative MRI measures.11 ApoE4 genotype was obtained on 141 of 163 autopsy cases by polymerase chain reaction, after standard restriction isotyping. Presence or absence of hypertension, hyperlipidemia, diabetes mellitus, heart disease, transient ischemic attack, and stroke was assessed initially and annually. For this analysis, a vascular factor was considered to be present if noted on any annual assessments.
Neuropathology Tissue Protocol
After death, the brain was removed, weighed, and fixed in 10% neutral buffered formalin for at least 2 weeks. After removal of the brainstem, each cerebral hemisphere was sectioned coronally at 5-mm thickness. Macroscopic infarcts were measured, photographed, and blocked for microscopic examination. Tissue was obtained from 12 standardized regions in 1 cerebral hemisphere using a standardized protocol.8 Tissue blocks were dehydrated through graded alcohols, embedded in paraffin, sectioned at 10-μ thickness, and stained with hematoxylin and eosin, Cresyl violet, Congo red, and Bielschowsky silver. At the pathologist’s discretion, cases were immuno-labeled using antibodies against α-synuclein, ubiquitin, glial fibrillary acidic protein, phosphorylated τ, and β-amyloid. The range of neuropathologic lesions has been described.12
Each case was reviewed at Consensus Neuropathology Conferences, including 2 Board-certified neuropathologists (H.V. Vinters, W.G. Ellis) blinded to clinical diagnosis and apoE4 genotype. Severity of atherosclerosis and arteriolosclerosis was rated on a 4-point scale (0=none, 1=mild, 2=moderate, 3=severe). Large arterial vessels were defined as vessels with diameter ≥1.5 mm (anterior, middle, and posterior cerebral arteries of the circle of Willis); small arterial vessels were those with diameter 0.2 to 1.5 mm. CAA was assessed using the modified 0–4 Vonsattel scale, in which 4 reflects the presence of 1 or more CAA-associated microangiopathies (eg, microaneurysm formation).13,14 Large cystic infarcts had infarct size >1 cm in greatest dimension. Lacunar infarcts were visible grossly and had infarct size <1 cm in greatest dimension. Microinfarcts were only visible upon microscropic examination. Although microinfarcts were almost always visualized on routine hematoxylin and eosin–stained sections, they were sometimes highlighted using immunohistochemistry, especially using a macrophage microglial marker (CD68).
For each autopsy case, Braak and Braak stage (B&B),15 Consortium to Establish a Registry for Alzheimer Disease (CERAD)-neuritic plaque,16 and Lewy body score17,18 were recorded. The severity of cerebrovascular ischemic brain injury was rated using a vascular brain injury pathology scoring (CVDPS) developed within this project, and previously described.8 Subscores for cystic infarcts, lacunar infarcts, and microinfarcts summed the individual scores across all brain regions and normalized to a scale of 0 to 100. The 3 subscores were summed to a total CVDPS score (0–300). Acute infarcts or hemorrhages near the time of death were noted, but not included in the CVDPS score.
Cut-off scores were selected for B&B and CVDPS scores to operationally define 5 pathologic diagnosis groups. We used B&B ≥IV19 to indicate the AD group (n=81) and CVDPS score ≥20, as described previously7,8,20,21 to define the CVD group (n=21). Cases with B&B ≥IV and CVDPS score ≥20 were defined as the MIXED group (cases with CVD and AD, n=15). Cases with B&B <IV and CVDPS <20 were classified as having no significant pathologic abnormality. We further subdivided this group into normal controls (NC: cognitively normal and no significant pathology, n=23) and OTHER (cognitively impaired without significant pathology, n=23). These pathologic categories are used to describe the study sample (Table 1), but were not used for the data analyses.
Sample characteristics were compared among the 5 pathologic groups using ANOVA for continuous variables and χ2 tests for categorical variables. Our major hypotheses, tested in the entire autopsy sample, examined the associations of pathologic measures of blood vessel disease and presence of an apoE ε4 allele to lesions within the brain parenchyma. Clinical history of vascular factors and pathologic measures of blood vessel disease (CAA, atherosclerosis, arteriolosclerosis) served as the main independent variables. Measures of pathologic changes in the brain parenchyma served as the dependent variables, including B&B, CERAD neuritic plaque score, CVDPS score, and the subscores for cystic, lacunar, and microinfarcts.
Ordinal logistic regression was used for the primary analysis. Dependent variables modeled as ordinal categorical variables included B&B (0–III, IV–V, VI), CERAD score (none-sparse, moderate, frequent), and CVDPS (CVDPS=0, 0<CVDPS<20, CVDPS ≥20). Cystic infarct (CYSTIC=0, CYSTIC>0), lacunar infarct (LACUNAR=0, LACUNAR >0), and microinfarct (MICRO=0, MICRO>0) scores were modeled separately as binary dependent variables using logistic regression. Atherosclerosis (0–3), arteriolosclerosis (0–3), and CAA (0–4) were jointly modeled in the multivariable ordinal logistic regression models with adjustment for age at death, sex, race/ethnicity, and years of education. In a separate model, apoE4 genotype was added as a covariate. Evaluation of the proportional odds assumption held for all models. Results are presented as proportional odds ratios (OR) with 95% confidence interval (CI). The proportional ORs are interpreted as the odds of being at medium or higher categories of the dependent variable, relative to the low category (or at the high category, relative to medium or lower categories), per unit increase of the independent variable. Multicollinearity statistics showed a variance inflation factor of 2.51 for atherosclerosis and 2.49 for arteriolosclerosis; values exceeding 10 are often regarded as indicating multicollinearity.22 Therefore, we retained atherosclerosis and arteriolosclerosis jointly in the multivariable logistic regression models. All statistical testing was performed at a 5% level of significance using SAS version 9.3 (SAS Institute Inc, Cary, NC).
The sample included 81 AD, 21 CVD, 15 MIXED AD/CVD, 23 OTHER, and 23 NC cases (Table 1). These pathologic groups did not differ in age at death, years of education, or duration of illness. Higher proportions of women were seen in the NC and OTHER groups. Ethnic minorities and a history of stroke were more frequently represented in the CVD and MIXED groups. Cognitive impairment (MMSE) was more severe and presence of apoE ε4 was more frequent in the AD and MIXED groups. These pathologic groups are presented for descriptive purposes only and were not used in the primary analyses.
The distribution of atherosclerosis, arteriolosclerosis, and CAA are shown by pathology group (Figure 1). Atherosclerosis and arteriolosclerosis were more severe in the CVD group (P<0.0001), whereas CAA was more severe in the AD and MIXED groups (P<0.0001). Atherosclerosis and arteriolosclerosis were highly correlated (Spearman r=0.70) but not with CAA (Spearman r=0.01 for atherosclerosis, 0.09 for arteriolosclerosis). The distribution of AD pathology and infarcts for the total sample are shown in Figure 2. B&B and CERAD scores were highly correlated (Spearman r=0.80) but not with CVDPS (Spearman r=−0.13 for B&B, −0.09 for CERAD).
Correlations of Vascular Factors With Parenchymal Pathology
Diabetes mellitus was inversely associated with B&B (P=0.002) and CERAD score (P=0.04). History of transient ischemic attack, stroke, and severity of arteriolosclerosis and atherosclerosis showed significant positive associations with the CVDPS score (Table 2).
Associations Between CVD Pathology and Parenchymal Pathology
Atherosclerosis was significantly positively associated with CVDPS (OR, 2.1; 95% CI, 1.2–3.7; P=0.01; Table 3). Neither atherosclerosis nor arteriolosclerosis was associated with AD pathology. CAA was not associated with CVDPS, but was associated with B&B stage (OR, 1.5; 95% CI, 1.1–2.1; P=0.03) and CERAD neuritic plaque score (OR, 1.5; 95% CI, 1.1–2.2; P=0.02). ApoE ε4 was independently associated with AD pathology (B&B stage: OR, 2.98; 95% CI, 1.43–6.2; P=0.004; CERAD score: OR, 2.82; 95% CI, 1.34–5.93; P=0.006).
We ran separate multivariable ordinal logistic regression models without apoE ε4 (data not shown) and compared these results with Table 3. Addition of apoE ε4 genotype to the multivariable analyses did not significantly alter the associations between CVD-type and parenchymal pathology.
Associations Between CVD Pathology and Type of Cerebral Infarcts
Atherosclerosis was positively associated with microinfarcts (OR, 2.3; 95% CI, 1.2–4.4) and cystic infarcts (OR, 2.0; 95% CI, 1.0–4.2; Table 4). Arteriolosclerosis showed borderline positive correlation with lacunar infarcts (OR, 2.0; 95% CI, 1.0–4.2). CAA was inversely associated with lacunar infarcts (OR, 0.5; 95% CI, 0.3–0.8). ApoE ε4 was not associated with any type of cerebral infarction.
In this autopsy sample enriched for cases with AD and SIVD, we found differential associations between types of CVD and parenchymal brain pathology. We report a strong and novel association between cerebral atherosclerosis and microinfarcts. Atherosclerosis was also associated with cystic infarcts but not AD pathology. Arteriolosclerosis showed a positive correlation of borderline statistical significance with lacunar infarcts but not AD pathology. CAA, on the other hand, was positively associated with AD pathology, but negatively associated with lacunar infarcts. Addition of apoE genotype to the multivariate analyses did not alter these findings. The independent contributions of CAA to B&B and CERAD score were slightly attenuated as expected, because the apoE ε4 genotype is associated with both CAA and AD pathology.
Associations between microinfarcts and cognitive impairment have been highlighted in several large autopsy studies. In the Honolulu Asia Aging study (HAAS) of Japanese-American men, microinfarcts were found in 64% of 436 autopsy cases.2 Microinfarcts contributed significantly and independently of neurofibrillary tangles to brain atrophy and cognitive impairment, especially in cases without overt dementia. In the Religious Orders study (ROS), microinfarcts were observed in 30% of 425 autopsied cases.1 People with multiple cortical microinfarcts had higher odds of dementia 1.89 (95% CI, 1.03–3.47). Microinfarcts contributed in an additive fashion to neurofibrillary tangles to lower cognition, including perceptual speed and semantic and episodic memory. Correlations between microinfarcts and type of CVD, however, were not reported either in the HAAS or ROS.
In this study, microinfarcts were found in 40% of cases and were most strongly correlated with cerebral atherosclerosis. In several cases with abundant microinfarcts, we noted evidence of thrombi in neighboring meningeal arteries, suggesting the possibility of artery-to-artery thrombo-emboli. These observations suggest paradoxically, but perhaps not surprisingly, that microinfarcts may result from large artery disease. They also raise the possibility that statins or antiplatelet medications might reduce the incidence of microinfarcts.
Microinfarcts have been observed in cases of severe CAA.23 We also observed microinfarcts in several of our cases with Grade 4 CAA, although the sample size (n=5) was small and the association was not statistically significant. We further observed qualitative differences in the morphological appearance of microinfarcts. In CAA, the microinfarcts tended to align along radial penetrating cortical arteries, whereas in cases with atherosclerosis the microinfarcts tended to seem as individual star-shaped lesions.
We were unable to confirm previous controversial associations between atherosclerosis and AD-associated neuritic plaques and neurofibrillary tangles.24 In the National Alzheimer Coordinating Center (NACC) database, Honig et al5 found an association between severe atherosclerosis in the Circle of Willis and frequent versus none-to-moderate neuritic plaque scores (OR, 3.9; 95% CI, 2.0–7.5). However, the Baltimore Longitudinal Study on Aging (BLSA) found no relationship between the degree of atherosclerosis in the intracranial vessels, aorta, or heart and the degree of AD-type brain pathology.25 We also found no associations between severity of atherosclerosis and B&B stage of neurofibrillary tangles or CERAD ratings of neuritic plaques (Tables 2–4). These inconsistencies in findings may reflect sample selection. If comparisons are made between cases with dementia (with both AD and atherosclerosis) and normal controls with neither AD nor atherosclerosis, a spurious association may be found between AD and atherosclerosis. A broad representation of atherosclerosis and cognition reflective of that represented in the general population is more likely to be achieved in community-based autopsy series or in a study focusing on vascular dementia than in Alzheimer center brain banks. In our sample, the pathologic groups did not differ in the prevalence of hypertension, hyperlipidemia, and CAD. The distribution of atherosclerosis and arteriolosclerosis in Figure 1 also suggests a good range of atherosclerosis in normal controls.
We examined the relationship between vascular factors and parenchymal pathology (Table 2). Epidemiologic studies report an association between vascular factor and clinically-diagnosed AD26,27; however, autopsy studies derived from epidemiologic cohorts have not shown a relationship between vascular factors and AD pathology.24,25 The epidemiologic literature reports associations between diabetes mellitus and increased incidence of clinically-diagnosed AD.28 In this study, we noted a positive trend between diabetes mellitus and infarct scores, but a negative association with AD pathology. Our ascertainment of diabetes mellitus was relatively crude (on the basis of self-reported or informant-reported history of diabetes mellitus). No associations between diabetes mellitus and AD pathology were reported in the HAAS,29 ROS,30 or in the Vantaa study.31 Taken together, these findings suggest that associations between diabetes mellitus and risk of dementia may be mediated through the additive burden of infarcts, rather than acceleration of AD pathology.
The major strengths of this study are the inclusion of a broad spectrum of subjects with both SIVD and AD, unimpaired elderly, together with a standardized neuropathologic assessment of the brain parenchyma. A limitation of our study was the use of semiquantitative measures to rate severity of vascular pathology. The weighted κ of 0.54 from 15 autopsy subjects indicated moderate test–retest reliability for the arteriolosclerosis score.
In conclusion, microinfarcts were most strongly associated with atherosclerosis. Possible pathogenetic mechanisms include artery-to-artery emboli, especially microemboli that may include atheroemboli or platelet-fibrin emboli. Arteriolosclerosis was positively, whereas CAA was negatively correlated with lacunar infarcts, which might prove helpful in clinical differentiation of arteriolosclerotic from CAA-related vascular brain injury.
Sources of Funding
This research was supported, in part by, National Institute on Aging Grants P01 AG12435, P50 AG05142, and P50 AG16570.
- Received April 29, 2013.
- Revision received June 4, 2013.
- Accepted June 13, 2013.
- © 2013 American Heart Association, Inc.
- Arvanitakis Z,
- Leurgans SE,
- Barnes LL,
- Bennett DA,
- Schneider JA
- Greenberg SM,
- Vonsattel JP
- Reed BR,
- Mungas DM,
- Kramer JH,
- Ellis W,
- Vinters HV,
- Zarow C,
- et al
- Allison PD
- Chui HC,
- Zheng L,
- Reed BR,
- Vinters HV,
- Mack WJ
- Peila R,
- Rodriguez BL,
- Launer LJ