Association between cerebral microbleeds and hypertension in the Swedish general population “Good Aging in Skåne” study

Abstract Cerebral microbleeds (CMB) on MRI are frequent in healthy aging individuals but precede ischemic and hemorrhagic stroke and dementia. Different etiologies have been suggested for nonlobar CMB, which have a stronger connection to hypertension (HT) than do lobar CMB. This study aimed to investigate the prevalence of CMB and the association between nonlobar/lobar CMB and different blood pressure (BP) and HT treatment conditions in a longitudinal, population‐based cohort of the Good Aging in Skåne (GÅS) study. White matter hyperintensities (WMH), CMB, atrophies, and infarctions were identified with brain 3T MRI, and BP parameters were examined in 344 randomly selected subjects between 70 and 87 years old. CMB were observed in 26% of the whole cohort, increasing from 19% of subjects in their 70s to 30% of those over 80 years of age. Of these subjects, 38% had multiple CMB, and 59% had a lobar localization. CMB were associated with severe confluent WMH (odds ratio = 7.02; 2.16‐18.84). Increasing age, being male, and having HT, impaired cognition, or a history of angina pectoris were associated with CMB. Both lobar and nonlobar CMB were associated with HT. Nonlobar CMB were particularly associated with increased BP, pulse pressure, controlled HT, and uncontrolled HT. After controlling for sex and HT, age was no longer a risk factor for CMB In conclusion, sex and HT are the major risk factors for CMB, especially nonlobar CMB, which suggests stricter implementation of recommended guidelines for HT treatment in the elderly.

distinction is that lobar/superficial microbleeds are more related to CAA, 1,4,6,[8][9][10] while deep/infratentorial CMB are more related to hypertensive vasculopathy. 4 Lobar CMB are overrepresented in dementia, suggesting the importance of CAA in this process. 6 Management of the traditional vascular risk factors, especially HT, for the prevention of cerebral small vessel disease (CSVD) is recommended, although studies on the effects of antihypertensive medications on CMB are lacking. 11 The pathophysiological mechanisms behind the development of CMB are not fully understood. The link between large vessel disease, CSVD, and CMB has yet to be fully investigated. In this study, we aimed to investigate the prevalence of CMB and the connection between deep/infratentorial and lobar CMB and different aspects of peripheral blood pressure (BP) parameters as well as HT treatment in a healthy, elderly population-based cohort.

| Selection and baseline examination
The Good Aging in Skåne (GÅS) study is an ongoing national, longitudinal, population-based aging study, part of the Swedish National study on Aging and Care (SNAC). 12 The baseline examination was conducted with 2931 subjects in nine different age cohorts ranging from 60 to 93 years who were recruited from 2001 to 2004, and an additional 1528 subjects in the age cohorts ranging from 60 to 81 years were recruited from 2007 to 2009. A randomized selection was made from the population register of the respective age-group, and a letter of invitation was sent out.
Participation rate was between 60% and 66%. Re-examinations were offered every third year for subjects above 78 years old and every sixth year for those between 60 and 78 years old. All examinations include the same protocol with a medical examination, a questionnaire, and cognitive and physical functioning tests. For participants unable to attend the test center, home visits were offered to reduce selection bias. In conjunction with the 4th re-examination, 821 subjects, aged 70-87 years, were examined with a 24-hour BP assessment, a standardized tilting test, and an intracranial and carotid Doppler. Six hundred and fifty subjects (79%) fulfilled the extended test protocol, of whom 344 participants, with a mean age of 77 years (median: 76 years; range: 70-87 years), were subsequently invited for a supplementary MRI scan during 2015-2017.
The baseline characteristics included age, sex, cohabitation, education, independence in daily life, physical activity (four-grade scale), and alcohol intake (past month and year). The history of medical diagnosis covered stroke, myocardial infarction, angina pectoris, atrial fibrillation, HT, COPD, dementia, depression, diabetes, and insomnia.
All data on medications, medical diagnosis, lifestyle factors, BP assessments, and cognition were derived from the medical, cognitive and questionnaire data obtained from the fourth revisit (except for two participants), the County Patient Medical Records Registry containing outpatient and inpatient records, National Inpatient and Outpatient Registry, and Swedish National Board of Health and Welfare covering all medical inpatient and outpatient visits coded according to the International Classification of Diseases (ICD) 9 and 10 in Sweden during 1987 to December 2016. The ICD classification coverage rate from the Registry for inpatient visits was 99.95% for all inpatient visits and 99.99% for outpatient visits. All drugs were coded according to the Anatomic Therapeutic Chemical (ATC) classification system. In addition, medications were categorized separately for BO1, for CO2-CO10,   and for BP-lowering drugs combined (CO2, CO3, CO7, CO8, CO9).

General cognitive performance was assessed by the Mini-Mental
State Examination (MMSE), which has a score range from 0 to 30, 13 and participants were categorized into three groups: normal cognition, scores of 28-30; intermediate cognition, scores of 25-27; and impaired cognition, scores ≤24, see Table 1. Immediate and delayed recall was assessed by a 5-item test in which two or more false answers were categorized as a pathological response. Participants were asked whether they had suffered from some degree or a significant loss of memory during recent years.

| Blood pressure measurements
Resting BP was bilaterally assessed three times by a physician as part of the medical examination; after approximately 10 minutes of rest, with the subject in a lying position, BP was assessed using a manual, auscultatory method with an appropriately sized cuff positioned at the heart level. The highest BP value between the left and right arm was used for the analysis. A 12-channel ECG was recorded. HT was defined according to WHO criteria: systolic BP ≥ 140 mmHg, diastolic BP ≥ 90 mmHg, or use of a BP-lowering medication. 14 Systolic ankle pressure was determined on the tibial posterior artery using a Doppler probe and manual BP cuff. An ankle-brachial index (ABI) below 0.9 was categorized as impaired peripheral circulation. The highest value from the left and right brachial pressures was used. Orthostatic BP was obtained immediately and 1, 3, 5, and 10 minutes after tilting from a supine position. An orthostatic reaction was defined as an immediate systolic BP drop ≥30 mm Hg and/or a systolic or diastolic BP drop of ≥20 mm Hg or 10 mm Hg, respectively, after 1-10 minutes. 15

| Blood pressure phenotypes
Blood pressure phenotypes were categorized into the five groups: (a) for an indication other than HT and no HT diagnosis; of these cases, seven participants were hypertensive, and nine were normotensive).

| Magnetic resonance image acquisition and processing
All patients were examined with a three Tesla MRI (General Electric, discovery MR 750w). MR examinations included the following: axial T2-weighted fluid-attenuated inversion recovery (T2 FLAIR), axial diffusion-weighted images (DWI), and axial susceptibility-weighted angiography (SWAN). SWAN sequences were acquired at two settings: 3-mm-thick SWAN images and 5-mm-thick phase images in order to differentiate blood from calcifications. Furthermore, sagittal T1-weighted 0.9-mm isotropic 3D fast spoiled gradient echo (3D-FSPGR) images were performed and reconstructed in the axial and coronal planes. MR images were assessed by an experienced neuroradiologist, and the MRI examinations were evaluated with regard to occurrence of the following: (a) white matter changes, classified according to Fazekas 16 grading, and white matter changes in the brain stem, (b) lacunar infarcts (less than 1 cm in diameter), (c) CMB defined as small (2-5 mm) hypointense lesions on MRI, 17

| Statistical analysis
Demographic variables were selected a priori from the baseline examinations and questionnaires. Age cohorts were determined based on sample sizes. Possible correlations of background characteristics with CMB were initially investigated using Pearson chisquare or Fisher's exact tests when appropriate and subsequently using univariate logistic regression analysis with "CMB yes/no" set as the dependent variable. The significance level was set to 0.05 and for the covariates, that is, sociodemographics, lifestyle, medical diagnosis, and medications, where the P < 0.05, a multivariate analysis adjusted for age and sex was performed. BP variables were investigated where CMB were stratified by localization as nonlobar and lobar. We examined the relationship between the number of CMB found and different aspects of HT using a linear regression model. The number of CMB was restricted to 0-10 for graphical reasons, and therefore, cases with >=10 microbleeds were set as 10 microbleeds. IBM SPSS Statistics v.24 was used for analyses.
Informed consent was obtained from each patient. Ethical approval was obtained from the Ethical Committee at Lund University, 2015/859. TA B L E 1 Baseline characteristics of study participants according to age-groups and the total cohort. The prevalence of CMB and odds ratios (ORs) for coexisting CMB is from the multivariate logistic regression analysis

| MRI findings
All grades of white matter hyperintensities (WMH) were observed in 84% of the study group, while the proportion of severe grades increased from 8.6% among subjects in their 70s to 18% among those over 80 years old ( for global cortical atrophy. Lacunar infarcts and cerebral infarcts were observed in 9.5% and 13.6% of all cases, respectively, and were not associated with age. Severe WMH, defined as a Fazekas score of 3, were associated with CMB (OR = 7.02; CI: 2.62-18.84). Global cortical atrophy, MTA, or cerebral or lacunar infarcts were associated with CMB.

| Association between CMB and blood pressure measurements
In the model adjusted for age and sex, CMB were associated with systolic and diastolic HT and peripheral pulse pressure (PPP) but not with ABI or orthostatic hypotension (Table 3). After stratification by nonlobar and lobar CMB, the associations between HT, systolic HT, diastolic HT, and increased PPP remained significant with the nonlobar CMB, but not with the lobar CMB (Table 4).

| BP phenotypes
The uncontrolled HT group (n = 92) had an increased risk of CMB compared to the normotensives (n = 72) (OR = 2.30; CI: 1.05-5.05; P = 0.037; in the multivariate model). When adjusted for age and sex, a significantly higher OR of nonlobar CMB was observed in subjects with controlled HT, uncontrolled HT, and untreated HT than in the healthy controls (Table 4). Intake of the drugs anticoagulants (P = 0.047) and Ca-channel blockers (P = 0.023) was associated to CMB, OR = 1.67, respectively OR = 1.98, in univariate analyses, see Table   S1. Medial temporal lobe atrophy score was associated to CMB, OR = 2.08, (P = 0.046). Age-corrected MTA score showed a similar association, OR = 2.18, (P = 0.062), see Table S2. The pattern between blood pressure measurements and CMB was similar in univariate and multivariate analyses, Table S3.

| D ISCUSS I ON
In our study, we investigated the relationships between different HT treatment conditions and the risk of CMB. The prevalence of CMB strongly increased with age, and in most cases, there was a single instance with a lobar localization. The main effect of HT on the prevalence of CMB was observed in nonlobar CMB, especially among subjects with uncontrolled HT. 5 It is not fully understood what aspect of HT and its mechanisms induce CMB. One theory is that increased arterial stiffness, both an effect of and a contributor to HT, is the main pathological mechanism behind CSVD and CMB. It is believed that the decrease in arterial autoregulation following arterial stiffness makes them less compliant to the pulse wave, which propagates the pulse wave further down the arterial tree. This might explain the main effect of HT, controlled as well as uncontrolled, on nonlobar but not lobar CMB. It seems that organs with penetrating arterioles, such as brain, kidney, and retina, are more sensitive to these changes, likely because they are more pericytes. CSVD has also been associated with chronic kidney disease, 24 and a decreased estimated glomerular filtration rate (eGFR) and arterial stiffness have been associated with cognitive impairment.
Many studies have investigated the effect of HT (defined as a heterogeneous group including increased resting BP, self-reported diagnosis, and/or antihypertensive treatment), but few studies have tried to investigate the different aspects of HT in detail. In our study, we found that an increase in PPP leads to an increased risk of nonlobar CMB. ABI can be viewed as a proxy for both atherosclerosis (<0.9) and extensive peripheral arterial stiffness (>1.3). However, in this study, no associations were noted in any direction between pathological ABI and CMB.
The association between age and CMB disappeared after adjusting for sex and HT. This is inconsistent with most previous studies.
However, previous data were based on younger cohorts. It is possible that the increased effect of age on CMB is saturated by introducing a survival bias, as an increasing number of individuals with less favorable genetic or lifestyle phenotypes have died or failed to come to examinations due to illness. Another contributing factor to that inconsistency is that no other studies included HT in the multivariate analyses, as the impact of this risk factor increases with age.
It is noteworthy that 22% of the participants above 80 years old had a history of atrial fibrillation and 44% of them used anticoagulants. However, CMB have been associated with both IS and ICH, especially in Asian populations and in subjects with lobar CMB associated with CAA. 25 In patients with nonvalvular atrial fibrillation, the presence of five or more CMB was an independent predictor of all-cause mortality and IS mortality, while patients with lobar CMB had increased risk for ICH (hazard ratio: 5.91). 26 A study comparing patients with ICH vs. IS while taking nonvitamin K antagonists showed that the ICH group had a higher proportion of CMB (79% vs 37%) and a higher number of CMB (median: 5 vs 0). 27 Further studies are needed to interpret the risk association between CMB and decisions of anticoagulant treatment.
A strength of the present study is its design-a large, general population-based study, with randomized inclusion of participants and a high proportion of subjects over 80 years of age. Home visits were offered at re-examinations to reduce selection bias. Although the selection of participants for the MRI was randomized from the GÅS study, the inclusion criteria were restricted to subjects able to perform the MRI and clinical investigation, thereby excluding the frail and demented. However, since the study comprises relatively healthy subjects, the results are to a large extent transferrable to the normal elderly urban population. A notable study limitation was that it was conducted in a northern European country on a population with little ethnic diversity, possibly reducing applicability to other ethnic groups.

| SUMMARY
CMB are an independent marker of CSVD, and the presence of nonlobar CMB is explained mainly by uncontrolled HT. CMB are frequent in the general older population, about one-third of those over 75 years of age, and are associated with WMH and cognitive impairment. These results indicate the need to implement recommended guidelines for HT treatment in the elderly population, although further prospective studies are needed to address causality, especially the effects of central BP, arterial stiffness, and pulse wave propagation on occurrence of CMB.

ACK N OWLED G M ENTS
The administrative staff Zinka Tucek and Lena Bergren for administrative support with re-examination and MRI.