Relation of plasma β‐amyloid, clusterin, and tau with cerebral microbleeds: Framingham Heart Study

Abstract Objective Cerebral microbleeds (CMBs) are associated with higher risk of stroke and dementia, predating clinical diagnosis by several years. CMB are considered markers of cerebral small vessel disease (CSVD): hypertensive (deep CMB) and cerebral amyloid angiopathy (lobar CMB). We related plasma β‐Amyloid (40, 42 and their ratio), clusterin, and tau levels to CMB to elucidate their role as biomarkers for the angiopathies represented by CMB. Methods Dementia, stroke, and other neurological disease‐free Framingham Heart Study participants with available CMB and biomarker measurements were included. We related biomarker levels (standardized for analyses) to CMB presence overall and stratified by brain topography (any, lobar, deep), using multivariable logistic regression analyses. Results CMB were observed in 208 (5.7%) participants (mean age 57 years, 54% women). After multivariable adjustment, Aβ1‐40 was associated with any CMB (OR (95%CI) 1.20 (0.99, 1.45) P = 0.062)) and lobar CMB (OR (95%CI) 1.33 (1.05, 1.68) P = 0.019), but not with deep CMB. Log‐Aβ1‐42 levels were not associated with CMB overall. Clusterin was related to mixed CMB (1.70 [1.05, 2.74], P = 0.031). Tau levels were associated with any CMB (OR (95%CI) 1.26 (1.07, 1.49) P = 0.006), lobar CMB (OR (95%CI) 1.26 (1.05, 1.52) P = 0.013), and with deep CMB (OR (95% CI) 1.46 (1.13, 1.89) P = 0.004). Interpretation We found that plasma Aβ1‐40 and Tau are associated with CMB but further studies are needed to confirm their role in hemorrhage prone CSVD represented by CMB and as indicators of ongoing subclinical neuronal injury.


Introduction
Cerebral small vessel disease (CSVD) is almost invariably present in mild degrees in persons with dementia, 1 is estimated to account for about 30% of ischemic strokes, 2 and constitutes the main cause of primary intracerebral hemorrhage. 3 Cerebral microbleeds (CMBs) are subclinical radiological markers of hemorrhage-prone CSVD, attributed to the main types of sporadic CSVD 4 : hypertensive vasculopathy (arteriolosclerosis) for deep CMB, and cerebral amyloid angiopathy (CAA) for lobar CMB. CMB are associated with higher risk of stroke and dementia, 5 and may predate clinical diagnosis by several years, thus serving as markers to understand the pathophysiology of vascular contributions to neurodegeneration, dementia, and stroke risk in preclinical stages.
The use of plasma biomarkers has been advocated to characterize individuals in preclinical stages in the AD continuum. 6 Similarly, there is much need for the identification of plasma biomarkers to characterize the types of CSVD most commonly seen in dementia and stroke. Plasma b amyloid (Ab), clusterin, and tau are blood biomarkers that have been associated with dementia and stroke, 7-10 and may be involved in CSVD. Ab plays a key role in the pathophysiology of AD and CAA, 11 depositing primarily in brain parenchyma in the former and cerebral small vessels in the latter. Ab 40 in particular has been reported to be deposited in cerebral vessels, while Ab 42 is primarily deposited in brain parenchyma in persons with AD. 12 Clusterin has been suggested to participate in clearance of Ab via cerebral small vessels. 13,14 Higher clusterin plasma levels have been associated with severity of AD and rapid cognitive decline. 15 The role of clusterin in AD, however, may depend on other factors such as Ab levels, age, and APOE ɛ4 genotype. 10 Tau plasma concentrations reflect ongoing neurodegeneration, have been associated with poor cognition and AD through mechanisms possibly independent from clusterin. CSF Ab and tau levels have been related to CMB presence 16 and higher clusterin levels have been associated with lobar CMB. 17 Study of the relation of plasma clusterin, Ab, and tau to CMB presence in a large sample of community dwelling individuals may provide insight into the pathophysiology of CAA and hypertensive vasculopathy in persons free of neurological disease. Study of the relation of these circulating biomarkers with CMB presence will allow further characterization of their role as biomarkers of CSVD represented by CMB topography, and to advance understanding of the role of CMB in neurodegeneration.
We hypothesized that higher clusterin, Ab, and tau levels are associated with higher CMB presence; that these relations vary according to CMB topography; and explored whether these associations are modified by age, gender, and APOE ɛ4 genotype. We posit that the associations of clusterin levels with CMB will depend on Ab levels, but be independent of plasma Tau levels.

Sample
The Framingham Heart Study began in 1948 with the enrollment of the Original cohort, followed in 1971 by enrollment of the Offspring cohort (children of the Original cohort and their spouses) and the Third Generation cohort (grandchildren of the Original Cohort participants and children of the Offspring cohort participants) in 2002. Participants from the three generations of Framingham Heart Study (FHS) Cohorts were included based on the availability of brain MRI allowing for CMB detection and biomarker data. Participants with prevalent stroke, dementia, or other neurological disease that could affect the estimation of CMB were excluded. Plasma biomarkers were obtained in FHS Original Cohort participants attending exam 23 (1992-1996 Ab was available only in Framingham Original and Offspring participants; among the participants who attended either of the respective exam cycle above, 1822 participants had brain MRI including CMB measurements. An additional 111 participants were excluded due to other neurological conditions affecting brain MRI. Clusterin was available only in Framingham Offspring (exam cycle 7) and Gen 3 (exam cycle 2) participants; among the participants who attended the respective exam cycle, 3832 participants had brain MRI including CMB measurements. An additional 170 participants were excluded due to other neurological conditions affecting brain MRI.
Tau measurements were available in the three Framingham Cohorts: at Original cohort exam cycle 28, Offspring cohort exam cycle 8, and at Third generation exam cycle 2; among the participants who attended the respective exam cycle, 3841 participants had brain MRI including CMB measurements. An additional 171 participants were excluded due to other neurological conditions affecting brain MRI. Table 1 shows the sample selection chart for each biomarker. The Institutional Review Board of Boston University Medical Center approved the study protocol and informed consent was obtained from all subjects.

Plasma clusterin, Ab, and Tau measurements
Details of blood sample collection, processing and quality control for measurement of Ab, plasma clusterin and Tau have been previously reported. 18 For Ab the samples were analyzed at the Department of Molecular Pharmacology and Experimental Therapeutics of the Mayo Clinic, Jacksonville, FL. Quantification of Ab in plasma was performed using INNO-BIA assays (Innogenetics, Ghent, Belgium), which is a multiplex microsphere-based Luminex xMAP technique. Intra-assay coefficients of variations (CVs) for Ab1-40 and Ab1-42 were 3.2% and 2.6%, and interassay CVs were 10.5% and 7.6% respectively. 18 Plasma clusterin levels were assessed as part of the Systems Approach to Biomarker Research (SABRe) project 2 using a Luminex xMAP assay. 19 The Multiplex assay had five protein targets (Clusterin, Apo-A1, Apo-B100, Lp(a), and CRP) and used a plasma dilution of 1/10,000. The intra-assay %CV for the clusterin measurements was 3.95%, and Interassay % CV was 8.4%. The assay had a lower limit of quantitation of 1.5 and upper limit of quantitation (ULOQ) of 2.67 9 10 4 , with a seven-point calibration curve. 10 Plasma tau was analyzed by Quanterix (Lexington) using a Simoa TM Tau 2.0 Kit and a Simoa HD-1 analyzer. This assay is a molecule enzyme-linked immunosorbent assay (digital ELISA) with a limit of detection of 0.019 pg/mL. The assay includes a set of monoclonal antibodies reacting to both normal and phosphorylated tau and can detect all tau isoforms. The analytical range was between 0.06 and 360 pg/mL. The intra-and interassay coefficients of variation were 4.1% and 7.5% respectively. 20 Outcomes Brain MRI protocol A 1.5-tesla MR machine (Siemens Magnetom) was used to obtain the following sequences: coronal T2-weighted 2470/ 20 to 80 (TR/ TE), echo train length 8, field of view 22 cm, acquisition matrix 192x256 interpolated to 256x256 with 1 excitation, 4-mm slice thickness from nasion to occiput, sagittal T1-weighted 11.4/4.4, 3D FLASH, 192 mm slab, 128 slices of 1.5-mm thickness, 12-degree flip angle and axial T2*gradient echo 656/26 (TR/TE), field of view 22 cm, acquisition matrix 144 9 256, 30-degree flip angle, 19 slices of 5-mm thickness, and 2 mm gap.
MRI were analyzed using a stroke visualization tool developed at the Imaging of Dementia and Aging (IDeA) laboratory that allows simultaneous visualization of multiple image sequences at three times magnification.

Cerebral microbleeds (CMB)
CMB presence, number, and topography were graded as per previously described methods using published guidelines. 4 CMB topography was divided into lobar only (confined to cortex and subcortical white matter), deep only (internal capsule/external capsule, thalamus, basal ganglia, brainstem) or deep + mixed (deep only or concurrent lobar and deep location, including cerebellum). We also studied CMB burden within individuals graded as absent versus present (≥1 CMBs), ≥2 CMBs, or ≥5 CMBs, and the subgroup of 2 or more lobar only CMB representing probable CAA as per the recently modified Boston criteria. We have previously related the reproducibility of CMB ratings. The intrarater reliability based on blinded reading of 200 scans on two separate occasions was excellent (kappa statistic 0.78). Inter-rater reliability comparing two independent readers in a subset of 200 scans was excellent (Kappa 0.78). Detailed description of our CMB rating methods have been published. 21 Clinical characteristicscovariates Clinical and demographic characteristics were measured at examination cycle closest to the biomarker measurement. Hypertension was defined using JNC-7 criteria as systolic blood pressure ≥140 mm Hg, diastolic blood pressure ≥90 mm Hg or use of antihypertensive medications. Current smoking was defined as self-reported smoking of at least one cigarette per day within the year preceding examination. Medication use was ascertained by self-report. Prevalent cardiovascular disease was defined as coronary heart disease, peripheral arterial disease and/or heart failure. APOE genotype was determined as previously described. 22 We used presence of any e4 allele, previously related to CMB presence and associated with biomarker measurements. 21

Statistical analyses
Descriptive statistics were obtained for the clinical, demographic, and biomarker variables (Tables 2-4). The distribution of Ab was skewed, thus natural logarithmic transformation was used. Log Ab, clusterin, and tau levels were standardized to 0 mean and unit standard deviation for analyses. We used multivariable logistic regression analyses to estimate odds ratios (OR) and 95% confidence intervals for the association of plasma clusterin, Ab, and Tau levels with CMB outcomes (any CMBs, lobar, deep, mixed location). We first assessed if each of the biomarkers was associated with CMB presence in models adjusted for age, gender, cohort-generation, and time between biomarker and CMB measurements. We then evaluated a multivariable adjusted model including covariates that are known to be associated with CMB presence or considered potential confounders: hypertension, statin use, antithrombotic use, and APOE ɛ4 status. A third model was additionally adjusted for C-reactive protein levels because prior studies suggest that clusterin levels may be related to inflammation, 23 (Table 3). A fourth model was additionally adjusted for white matter hyperintensity volume and covert brain infarcts to assess if the associations were independent of ischemic markers of small vessel disease.
For each biomarker we conducted stratified analysis by the following factors: age (55 years old or older vs. younger), gender, and APOE genotype (any ɛ4 risk allele vs. none). For clusterin, we additionally conducted stratified analyses by plasma Ab levels (above vs. below median). We evaluated nonlinear associations by relating tertiles of each biomarker (with the lowest tertile as referent) to CMB presence. Lastly, we performed separate exploratory logistic regression analyses to assess the association of biomarker levels and CMB burden, with three CMB burden groups: ≥1, ≥2, and ≥5 CMB, with no CMB as reference group; and with the subgroup of 2 or more lobar only CMB representing probable CAA. All analyses were performed using SAS version 9.4 (Cary, NC). A Pvalue <0.05 was considered statistically significant.

Results
We observed CMB in 208 (5.7%) participants, with a predominant lobar location: 73% lobar only versus 16.3% deep only and 10.6% mixed. The mean (SD) age in our participants was 55 (13) and 54% were women. Participants with CMB were older, had a higher prevalence of hypertension, were more likely to use statin and antithrombotic treatments (Tables 2-4).
Primary results (model 2, There was a borderline association with mixed only CMB, attenuated after adjustment for vascular risk factors. No significant association was observed between Ab1-40 and deep only or deep and mixed CMB, or Ab1-42 and any of the CMB groups. The Ab1-42/1-40 ratio was significantly associated with any CMB and borderline associated Additional adjustment for CRP levels in model 3 did not change the association between any of the biomarkers and CMB presence (data not shown). Similarly, additional adjustment for MRI markers of ischemic small vessel disease, white matter hyperintensity volume and presence of covert brain infarcts, did not change the results. Overall The prespecified stratified analyses (i.e., by age, gender, and APOE ɛ4 status for clusterin, Ab, and Tau, and by Ab levels for clusterin) did not reveal any meaningful differences in the association of the biomarkers with CMB presence. Analyses of the relation between biomarker levels and CMB burden were limited by the small sample of participants with high CMB burden, but showed that the strength of association between CMB and Ab1-40 and the Ab1-42/1-40 ratio increased as CMB burden increased. For

Discussion
Our study included a large sample of community dwelling middle aged individuals that were free of neurological disease at the time of biomarker measurement and brain MRI. We observed that plasma Ab1-40, the Ab1-40/1-42 ratio, and tau levels were associated with CMB presence overall and with strictly lobar CMB in particular. Clusterin was associated with higher odds of mixed location CMB, and lower odds of strictly deep CMB but multivariate analyses were limited by small number of CMB in the latter subgroup. There was an association between plasma tau and deep CMB, while there were no significant associations with mixed CMB.

Ab and CMB
In agreement with prior studies including smaller samples of elderly persons and individuals with impaired cognition and dementia, 12 we observed that Ab40 but not Ab1-42 was associated with CMBs and the Ab42/40 ratio showed an inverse relation with markers of SVD including CMB. However, the same study also reported an association of Ab40 with lobar and deep CMB, in contrast with our findings of a significant association only with lobar CMB. This difference may be due to differences in the samples studied, such as older age and higher prevalence of hypertension in their study, i.e., a sample more likely to have more advanced forms of both types of angiopathy represented by lobar and deep CMB. In another study in elderly and patients with impaired Log-transformed and standardized to 0 mean and unit standard deviation. 3 Standardized to 0 mean and unit standard deviation.

cognition (Swedish BioFINDER study Biomarkers for
Identifying Neurodegenerative Disorders Early and Reliably) it was found that higher plasma Ab40 and Ab42 levels were associated with CMB presence. 8 The differential findings of an association of Ab1-40 but not Ab42 with CMB in our study, in particular lobar CMB, is supportive of the vascular deposition of the more soluble Ab 40, whereas Ab 42 is a predominant component in amyloid plaques in AD. However, it is possible that Ab 42 also becomes related to CMB in more advanced stages of angiopathy or neurodegenerative process as can be seen in older age and in impaired cognition as in the Swedish BioFINDER study sample. This is also suggested by the observation that the associations of Ab40 with mixed only CMB were stronger in our study, though not statistically significant likely due to the small sample of participants with mixed CMB. Our finding of low Ab42/Ab40 ratio associations with CMB is in line with prior reports showing that decreased ratios are associated with higher Ab brain burden. 24 Given that lobar CMB are considered to represent predominantly CAA (in support FHS participants with lobar CMB have higher prevalence of APOE ɛ4 genotype), our findings would suggest that plasma Ab1-40 may play a role in hemorrhage prone CSVD in preclinical stages, but should be confirmed in additional studies.
Of note, plasma Ab may not be a marker solely for hemorrhagic CSVD as prior reports have shown that higher Ab40 levels may also be increased in participants with lacunar infarctions and WMH, 12 and in elderly with CAA. 25

Clusterin and CMB
Clusterin levels have been related to dementia and stroke, but the relation is complex and influenced by several factors. Although we observed an inverse association of higher clusterin levels with lower odds of CMB in deep brain regions in our initial model, multivariate adjustment was not possible. Our results do not support a role for clusterin in the vascular injury represented by CMB in lobar regions; with regards to a potential role in deep CMB, presumably increased clusterin levels may reflect an attempt to prevent ongoing neurodegeneration that may be sufficient to prevent deep CMB representing hypertensive vasculopathy; however, as higher CMB burden is accrued, clusterin may be ineffective to prevent CMB as suggested by the observation that higher clusterin levels are associated with higher odds of mixed only CMB. However, these results should be interpreted with caution and taken only as hypothesis generating, in view of the small sample of persons with deep only and mixed CMB.

Tau and CMB
The association we observed of higher tau levels with higher CMB presence overall, and lobar and deep in particular, signal that CMB may also be markers of ongoing neurodegeneration. The association of tau levels with mixed only CMB, a subgroup representing higher burden of CMB, though not statistically significant was surprisingly in the opposite direction than with deep CMB. To further understand this finding, we assessed for possible imbalance in covariates that would suggest residual confounding. We observed substantial imbalance in the proportion of participants taking treatments for cardiovascular disease prevention in the mixed CMB subgroup (more likely to use) compared with those with deep only CMB (less likely to use), suggesting that treatment effects are possibly implicated in these findings. Our multivariable analyses considered these treatments as covariates and adjusted analyses were large unchanged; in addition, we assessed for interaction with treatment use, but interactions were not significant. However, treatment related effects such as duration of exposure to medications, control achieved over the long-term could not be accounted for in our analyses, and may represent residual confounding. It is also important to highlight that these analyses are limited by the small sample of mixed CMB outcomes.
Our findings support a role for tau in the group of probable CAA (i.e., 2 or more CMB). Given that the neurovascular unit (neuron-blood vessel) is a fundamental component of the brain where neuronal structure and function are closely inter-related to blood vessels, our findings suggest that vascular disease represented by CMB also indicates neuronal injury, but the relation may be modulated by cardiovascular treatments.

Strengths and Limitations
Among the strengths in our study are the inclusion of a large sample with available biomarker and MRI data, high reliability of exposure and outcome ascertainments, both blinded to each other and to all clinical and demographic characteristics, and use of highly sensitive assays for biomarker measurements. 26 Our study expands previous studies by including a large sample of middle aged community dwelling individuals.
Several limitations are noteworthy. Participants who underwent brain MRI are generally healthier than those who did not have MRI; however, the effect estimates are expected to be unbiased with respect to participant selection as inclusion into the present study is considered unrelated to the exposure or outcome. Our analyses involving prespecified subgroups of CMB burden are ª 2020 The Authors. Annals of Clinical and Translational Neurology published by Wiley Periodicals LLC on behalf of American Neurological Association limited by the smaller sample among high burden subgroups. Analyses of the relation of biomarkers with mixed CMB are also limited by the small sample of outcomes in this group, precluding further exploratory analyses. The relation of the plasma biomarkers and CMB ratings is cross sectional, thus we cannot establish causality in their relationship. Our results, particularly in the relation of Ab and CMB, should be considered hypothesis generating and interpreted with caution given the multiple comparisons performed. The generalizability of our results is limited to participants of White racial group, as Framingham Heart Study participants are primarily of European ancestry; this observation is important as the risk factors for CMB and underlying vasculopathy may differ in other racial groups. Similarly, given the complexity of our samples further studies are needed to replicate our results, particularly in samples of other racial and ethnic groups.

Conclusion
We found plasma Ab and Tau are related to CMB overall and lobar CMB in particular. Our results advance our understanding of the pathophysiology of CMB and their role in neurodegeneration. Our findings provide important insight into the mechanism of vascular disease predating the onset of dementia and stroke, and generate hypotheses to be tested in additional studies, for the possible role of these biomarkers in hemorrhage prone CSVD.