Location of Cerebral Microbleeds And Their Association with Carotid Intima-media Thickness: A Community-based Study

To assess whether high cerebral microbleeds (CMBs) are associated with carotid intima-media thickness (CIMT), a marker of systemic atherosclerosis, we cross-sectionally evaluated participants from a community-based study, the I-Lan Longitudinal Aging Study. The participants’ demographics and cardiovascular risk factors were determined by questionnaire and/or laboratory measurements. CIMT was measured by ultrasonography. CMBs were assessed by susceptibility-weighted-imaging on 3 T MRI. Of the 962 subjects [62.5(8.6) years, 44.2% men] included, CMBs were found in 134(14.0%) subjects. Among the subjects with identified CMB’s, 85(63.4%) had deep or infratentorial (DI) and 49(36.6%) had strictly lobar(SL) CMBs. After the results were adjusted for age and sex, the analysis revealed that hypertension, hyperlipidemia, obesity, and higher triglyceride levels correlated with DI but not SL CMBs. The subjects with DI CMBs also had a higher mean CIMT and higher prevalence of top quartile CIMT. The multivariate analysis demonstrated that high CIMT (top quartile) significantly predicted the presence of DI CMBs (odds ratio = 2.1; 95% confidence interval = 1.3–3.4; P = 0.004), independent of age, sex, cardiovascular risk factors, and other cerebral small vessel diseases, lacune, and white matter hyperintensity. There was no association between CIMT and SL CMBs. Our results support that there are distinct pathogenesis in DI and SL CMBs.


Risk Factors for the Presence of Overall, DI and SL CMBs.
The risk factor evaluation for the presence of overall, DI, and SL CMBs are shown in Table 1. The subjects with the presence of overall CMBs (without regard to CMBs location) were older. There was no difference in the sex distribution between the subjects with and without CMBs. After the data were adjusted for age and sex, the results revealed that subjects with CMBs had a higher prevalence of CAD, a higher number of lacunes, and more severe WMH. There was a trend toward an association with the presence of CMBs and a higher mean CIMT. When compared to subjects without CMBs, the prevalence of top quartile CIMT was significant higher in subjects with CMBs, after adjusting for age and sex.
Both DI and SL CMBs were associated with increased age, but the correlations with cardiovascular risk factors were different between the two types of CMBs. After adjusting for age and sex, we found that DI CMBs were associated with hypertension, hyperlipidemia, obesity, CAD and higher circulatory triglyceride. In contrast, SL CMBs did not correlate with any cardiovascular risk factors. Notably, almost all subjects with hyperlipidemia had taken Stains (55/58, 94.8%). The association between hyperlipidemia and CMBs had been adjusted for Statin use.
The subjects with DI CMBs had a higher number of lacunes and more severe WMH than subjects without DI CMBs. SL CMBs were associated with WMH severity but not the number of lacunes. The association with CIMT also differed between DI and SL CMBs. When compared with subjects without DI CMBs, subjects with DI CMBs had significantly higher CIMT and a higher prevalence of top quartile CIMT. There was no association between CIMT and SL CMBs.
There were 38 (4%) subjects taking antiplatelets; 33 aspirin (100 mg per day) and 5 ticlopidine (100 mg twice per day). No subject in our population was prescribed anticoagulants. There was no association between CMBs and antiplatelet usage in the present study.
Risk Factors for Top Quartile CIMT. Since a higher CIMT was a stronger predictor for vascular events and more representative of systemic atherosclerosis 15 , we used the top quartile CIMT as a marker of atherosclerosis in our analyses. The risk factors for high CIMT are presented in Table 2. Subjects with older age and male sex were more likely to have a higher CIMT. After adjusting for age and sex, the top quartile of CIMT was associated with hypertension and obesity. Table 3 reveals the results of the multivariate logistic regression analyses. A higher CIMT (top quartile) was associated with the presence of CMBs independent of age, sex, cardiovascular risk factors, CAD, and CKD (model 1). The association was still significant after further adjustment for other CSVDs, the presence of lacunes, and the WMH (model 2). When the CMBs locations were considered, the results showed that the correlation between CMBs and CIMT was significant only in DI but not in SL CMBs. A higher CIMT predicted the presence of DI CMBs with an Odds ratio (OR) of 2.5, independent of age, sex, cardiovascular risk factors, CAD, CKD, and other CSVDs. SL CMBs were not associated with CIMT.

Discussion
The main finding of this study was that the presence of CMBs correlated with a higher CIMT. We also found that the location of the CMBs determined the association between the CMBs and CIMT; DI, but not SL, CMBs were significantly associated with a higher CIMT, independent of age, sex, cardiovascular risk factors (hypertension, DM, hyperlipidemia, cigarette smoking, obesity, CAD and CKD), and other CSVDs.
Previous studies with smaller population have investigated the relationship between CMBs and atherosclerosis in patients with a history of stroke 12,13 . The present community-based study evaluated the relationship between CMBs and atherosclerosis in the general population. Increased IMT has been deemed to be representative of early stages of atherosclerosis, and is widely used as a marker of atherosclerosis 11,14 . Our results indicate that CMBs in the general population were associated with atherosclerosis. This could explain why in several studies, the presence of CMBs predicts a higher risk of future not only hemorrhagic but also ischemic stroke independent of age, sex, and cardiovascular risk factors 16 . Since our study population included community-dwelling, subjects without a history of stroke, we postulated that atherosclerosis might be involved in the pathophysiology of CMBs at an early stage of the disease.
A recent-published paper of the Framingham Offspring Study revealed similar CMBs location-determined association with atherosclerosis markers 17 . Their results showed that carotid stenosis ≥25% was associated with DI but not SL CMBs. However, this study did not find a significant correlation between baseline CIMT and CMBs. The controversial results may due to the different study population. The other important reason might be that this study directly used CIMT value but not stratified ones into analysis. We classified CIMT into top quartile and the lower three quartiles since CIMT may be the sequela of other etiology instead of atherosclerosis and a higher CIMT was a stronger predictor for vascular events and more representative of systemic atherosclerosis 15 .
Our results also showed that DI but not SL CMBs were associated with atherosclerosis, independent of age, sex and cardiovascular risk factors. DI but not SL CMB, correlated with a history of CAD and lacunes in our study, a notion that also supports this postulation. In our population, DI CMBs were correlated with several atherogenic risk factors such as hypertension, hyperlipidemia, obesity, and higher triglyceride levels (Table 1), which might explain why only DI but not SL CMBs were associated with atherosclerosis.
Endothelial dysfunction, involved in the early stage of atherosclerosis 18 , might lead to impaired cerebral microvascular autoregulation in response to elevated perfusion pressure (such as in elevated BP) and consequent microvascular rupture 19,20 . It is possible that endothelial dysfunction mediates the association between atherosclerosis and DI CMBs.
Several studies have reported an association between arterial stiffness and age-related CSVDs, and showed that the relation of arterial stiffness to CMBs was different between the patients with stroke and the general population [21][22][23] . Previous studies in stroke patients have demonstrated an association between CMBs and arterial stiffness 12,13,24 . We did not find a correlation between CMBs, both DI and SL, and pulse pressure, which result was consistent with a previous study of subjects without a history of cerebrovascular and cardiovascular diseases 23,25 . Arterial stiffness, the reduced capability of an artery to expand and contract in response to pressure changes, usually develops in the elderly and people with long-standing hypertension 26 . The relationship between atherosclerosis and arterial stiffness has been analyzed and discussed in several studies 26 . Their correlation has been shown more consistently in more advanced atherosclerosis [26][27][28] . During the early stage of atherosclerosis, a thickened IMT may maintain vascular circumferential stress and eliminate vascular injury and prevent subsequent vascular  structural changes. More advanced atherosclerosis or arterial stiffness would develop when the adaptive response is surpassed by persisted or profound vascular injury 29,30 . Our results revealed a relationship between DI CMBs with higher CIMT, but not pulse pressure in a community-dwelling, general population without a history of stroke, and it might reflect the pathogenesis of DI CMBs at a prodromal or early stage of disease. Arterial stiffness might follow atherosclerosis and be involved in the pathophysiology of DI CMBs at a later stage of disease. We would need a future longitudinal follow-up study to validate this hypothesis.
There were a few limitations to our study. Carotid plaque area and volume are other atherosclerosis markers 31 which we did not measure in the present study. We were concerned that the frequency and severity of plaque in the younger and healthy population in the present study would have been too low to achieve statistical power with those carotid measurements. Therefore, we chose CIMT as the marker of atherosclerosis. Notably, increased CIMT might not always represent atherosclerosis pathology, therefore, we used a higher level of CIMT (top quartile), which is more specific for atherosclerosis 15 . Since both increased IMT and DI CMBs are correlated with hypertension, their relationship might simply be contributed by different hypertensive effects on vessels 31,32 , though we have adjusted with hypertension into our multivariate analyses. A recent study explored the association between carotid plaque volume (total and the subcomponents) and CMBs in 72 patients (CMB prevalence 35.3%); the results showed that an increased volume of the fatty component was associated with the presence and number of CMBs. Their method would be considered in the future study with a larger population to validate the relationship between DI CMBs and systemic atherosclerosis 33 . We also lacked information of inflammatory biomarkers 34 and endothelial function 16 in our population which could have provided clues to the mechanisms and the associations between atherosclerosis and DI CMBs. Finally, a longitudinal follow-up study is needed to see if higher level of CIMT leads to new DI formation. This information would strengthen our postulation that atherosclerosis is a causal factor in the pathophysiology of DI CMBs.
In conclusion, the present study has provided further evidence that there are pathogenic differences between DI and SL CMBs.

Methods
Study population. The I-Lan Longitudinal Aging Study is a community-based aging cohort study in the I-Lan County of Taiwan that aimed to evaluate the interrelationship between the geriatric syndromes and brain structural abnormalities and explore predictors or associated factors for future disable, dementia or mortality in the geriatric population when they are at a prodromal or early stage of disease 35 . In brief, community-dwelling adults aged 50 years or older from Yuanshan Township in I-Lan County were invited to participate in the study. Any subject that having any contraindication to an MRI such as metal implants, having been institutionalized for any reason or having known neuropsychiatric diseases such as dementia, stroke, brain tumor, or major depression was excluded from the study. The whole study was approved by the Institutional Review Board of the National Yang Ming University, Taipei, Taiwan. All the participants provided informed consent and we have conducted the study in accordance with the relevant ethical guidelines and regulations.

Demographics and Cardiovascular Risk Factors.
A questionnaire was used to collect data regarding the demographics and medical history of the study subjects. The heights, weights, and resting blood pressures (BP) of the subjects were measured. Fasting serum lipid levels (total cholesterol, low density lipoprotein, high density lipoprotein and triglyceride), and blood urea nitrogen and creatinine levels were determined by a chemical analyzer (ADVIA 1900, Siemens, Malvern, PA, USA). The presence of cardiovascular risk factors including cigarette smoking was determined by patient history or laboratory investigation. Hypertension was defined as a self-report of a current antihypertensive medication prescription or as a measurement of SBP ≥140 mmHg or DBP ≥90 mmHg 36 . Diabetes mellitus (DM) was defined as a self-report of current DM treatment or a measurement of HgbA1c ≥6.5% 37 . Hyperlipidemia was recorded if there was a self-report of the use of a statin agent or a total blood cholesterol level ≥240 mg/dL 38 . Body mass index (BMI) was calculated as weight in kg/height in m 2 .
Obesity was defined as BMI ≥30 (World Health Organization, 2014). Chronic kidney disease (CKD) was defined as an estimated glomerular filtration rate ≤60 mL/min/1.73m 39 . A history of coronary artery disease (CAD) was determined by the subjects' self-reports.
Carotid Ultrasonography. All of the subjects underwent ultrasound imaging of the bilateral common carotid arteries (CCAs) in longitudinal projections using an instrument (GE LOGIQ 400 PRO; GE, Cleveland, OH, USA) equipped with a high-resolution broadband width linear array transducer by one technician. Each subject's neck was extended in a supine position, with his/her head turned 30-degree toward the opposite direction of measurement. Then, the CIMT was measured on the far wall of subject's right and left distal CCA by long axis view, and an image was automatically taken during R wave in electrocardiography. After taking an image, an operator selected 2 cm width region between carotid bifurcation and CCA, and then mean values of each CIMT were calculated by automatic methods in ultrasonography machine. The average of the left and right carotid IMTs were measured for analyses. All measurements were performed by one technician who was unaware of the clinical characteristics of the subjects. Measurements were additionally taken on a separate visit in 20 random-sampled subjects, and the intra-rater k = 0.80 (95% confidence interval 0.77-0.89). CMBs were defined as small, rounded or circular, well-defined hypointense lesions within the brain parenchyma with clear margins that were ≦10 mm in size on the SWI image 3,40 . Microbleed mimics such as vessels, calcification, partial volume, air-bone interfaces, and hemorrhages within or adjacent to an infarct were carefully excluded. We distinguised calcification by viewing T1-weighted MRI. CMB mimics showing low density on T1-weight MRI will be regarded as calcification. We used the Microbleed Anatomical Rating Scale to measure the presence, amount, and topographic distribution of the CMBs in each subject, which has been reported a good intra-rater and inter-rater reliability 40 . The microbleeds were classified according to whether they were located in the deep, infratentorial, or lobar categories. Lobar topography was determined according to Stark and Bradley 41 , and included the cortical and subcortical regions (including subcortical U fibers). The lobar CMBs were assessed in the fontal, parietal, temporal and occipital regions. The deep regions included the basal ganglia (BG), thalamus, internal capsule, external capsule, corpus callosum, and deep/periventricular white matter; the infratentorial regions included the brainstem and cerebellum. DPWM was defined as white matter adjacent to or within approximately 10 mm of the lateral ventricular margin. Images were displayed and viewed using the MRIcro software (version 1.40, Chris Rorden's MRIcro) by one neurologist (Dr. Chung) who was blinded to the clinical data during the CMB assessment and analyses. CMBs in 20 random-sampled subjects' images were evaluated again at a separate time, and the intra-rater k = 0.83 (95% confidence interval 0.79-0.90). We also re-assessed CMBs in 25 random-sampled subjects' images by Dr. Chung and her well-trained assistant (Mr. Ching-Sern Yong). The inter-rater k was 0.82 (95% confidence interval 0.79-0.88).

Brain MRI Acquisition and CMBs
Scientific RepoRts | 7: 12058 | DOI:10.1038/s41598-017-12176-y The other manifestations of CSVDs, the numbers of lacunes, and the severity of WMH were also recorded in every subject by FLAIR-T2-weighted MRI. Lacunes are small CSF-containing cavities, ≤15 mm in diameter, located in the deep grey or white matter with adjacent white matter hyperintensity 1 . The severity of WMH was rated by the modified Fazekas scale 42 . The Fazekas scale scores are: 0 as no WMH, 1 as mild WMH, 2 as moderate WMH, and 3 as severe WMH.
Statistical Analysis. The statistical analyses were performed with SAS software, version 9.1 (SAS Institute, Cary, NC, USA). For continuous numeric variables, the nonparametric Mann-Whitney tests were performed as appropriate for group comparisons. The χ2 test or Fisher's exact test was performed for categorical variables. Univariate and multivariate logistic regression analyses were performed to investigate the risk factors of (1) the presence of overall, DI, and SL CMBs respectively and (2) the top quartile CIMT. The associations between CMBs and the top quartile CIMT were also evaluated by multivariate logistic regression analyses. The results were presented as odds ratios (ORs) with 95% confidence intervals (95% CIs).
Data availability statement. The datasets generated during and/or analysed during the current study are available from the corresponding author on reasonable request.