Introduction

Cerebral microbleeds (CMBs) are primarily a radiological construct (small MRI signal voids) indicative of specific underlying microscopic pathological changes, i.e., perivascular collections of hemosiderin deposits that indicate prior micro-hemorrhages [1]. Histopathological analyses of CMBs generally reveal two types of vascular pathological changes: hypertensive vasculopathy and cerebral amyloid angiopathy. The distribution of CMBs is different in these two disorders: lobar CMBs are associated with cerebral amyloid angiopathy, whereas deep CMBs are associated with hypertensive vasculopathy [1, 2].

The prevailing view is that microbleeds are clinically silent [3]. However, recent histopathological data have revealed that CMBs are not only circumscribed hemosiderin deposits, but also affect the surrounding gliosis and cause frank necrosis or microinfarctions, which suggests that they may be of clinical importance [4]. Recent evidence also suggests that CMBs may be an important factor in cognitive impairments in subcortical vascular dementia [5], where they are associated with cognitive dysfunction, particularly executive dysfunction [6]. CMBs are common in ischemic stroke [7], and they may predict the recurrence of stroke [8] and the emergence of emotional lability [9]. These microbleeds are also associated with factors that may affect health-related quality of life (HRQoL) in stroke, including advanced small artery disease of the brain [3], leukoaraiosis [10], and recurrent stroke [8].

Patients with stroke have been shown to have poor HRQoL [11]. The correlates of poor HRQoL 1 month to 3 years after stroke include an older age [10, 11], the female sex [12], being single or widowed [13], a lack of social support [14], a high degree of neurological deficit [15], cognitive impairment [16], and depression [17].

Data on the impact of infarcts and other radiological findings on HRQoL are limited, although large lesion volumes [18] have been found to predict poor HRQoL following stroke [19]. To date, only one study has examined the role that acute infarct location and white matter lesions (WMLs) play in HRQoL, with no correlation found between either lesion location or WMLs and HRQoL [19].

To the best of our knowledge, no data have been published on the impact of CMBs on the HRQoL of stroke patients, and hence their effect on HRQoL in stroke remains unknown. This study set out to examine whether the presence or location of CMBs contributes to HRQoL in acute ischemic stroke survivors.

Materials and methods

Subjects

Of the 2,337 patients admitted with first-ever or recurrent stroke to the Acute Stroke Unit (ASU) of the Prince of Wales Hospital (PWH)—a university-affiliated general hospital serving a population of 800,000 in Hong Kong—from December 2004 to May 2007, 874 received an MRI examination. Due to limited access to an MRI machine, MRI is generally reserved for stroke patients with stable neurological conditions. Compared to patients who did not receive an MRI examination, those who did had a lower mean age (67.8 + 12.1 vs. 74.5 + 12.0; p < 0.001) and less severe stroke, as reflected by a lower National Institutes of Health Stroke Scale (NIHSS) score (5.6 + 4.6 vs. 10.6 + 9.8, p < 0.001), and were more likely to be men (58.8 vs. 46.6%; p < 0.001). Four hundred and fifty-eight (52.2%) patients who had received an MRI were recruited for the study. The inclusion criteria were (1) Chinese ethnicity; (2) aged 18 or above; (3) well-documented (clinical presentation and CT scan of the brain) first or recurrent acute stroke occurring within the 7 days prior to admission; (4) Cantonese dialect as the primary language; and (5) ability and willingness to give consent. The exclusion criteria included: (1) transient ischemic attack, cerebral hemorrhage, subdural hematoma, or subarachnoid hemorrhage; (2) history of a central nervous system (CNS) disease other than stroke, such as tumor, trauma, hydrocephalus, or Parkinson’s disease; (3) severe cognitive impairment, as defined by a Mini-Mental State Examination (MMSE) [20] score of <19; (4) aphasia; and (5) medical conditions that would preclude follow-up 3 months after the index stroke, such as recurrent stroke, prolonged hospitalization (>3 months), and physical frailty.

The study protocol was approved by the Clinical Research Ethics Committee of the Chinese University of Hong Kong. All participants signed a consent form.

Collection of demographic and clinical data

A research nurse collected subjects’ demographic data (age, sex, and educational level in terms of school years) and assessed their stroke severity using the NIHSS [21].

Assessment of HRQoL

Three months after the onset of the index stroke, a trained research assistant, who was blind to the subjects’ radiological data, assessed all participants in a research clinic with the Chinese (Hong Kong) version of the Short Form-36 (SF-36) [22], a valid and reliable instrument for measuring HRQoL in stroke [23]. The SF-36 focuses on subjective perceptions of health. It contains eight subscales that cover general health (GH), mental health (MH), role limitations due to emotional problems (RE), role limitations due to physical problems (RP), social functions (SF), vitality (VT), bodily pain (BP), and physical functioning (PF). The subscale scores range from 0 to 100, with higher values representing better functions. The SF-36 has two composite scores: the physical component (PCS) and mental component (MCS) summary scores.

The research assistant also administered the 15-item version of the Geriatric Depression Scale (GDS) [24], the MMSE, and the Lubben Social Network Scale (LSNS) [25] to assess participants’ depressive symptoms, cognitive functioning, and degree of social support, respectively. The GDS and MMSE scores range from 0 to 15 and 0 to 30, respectively. A higher GDS score indicates more severe depressive symptoms, whereas a lower MMSE score represents poorer global cognitive functioning. The LSNS is a composite social network scale that was specifically designed for use in the elderly [26]. It measures the level of social support that patients receive and their degree of social interaction with relatives and friends. The maximum score is 50, with a higher score indicating better social support.

Radiological examination

MRI was performed with a 1.5 T system (Sonata, Siemens Medical, Erlangen, Germany) within 7 days of admission. Diffusion weighted imaging (DWI) (TR/TE/excitation = 180/122/4, matrix = 128 × 128, field of view (FOV) = 230 mm, slice thickness/gap = 5 mm/1 mm, echo planar imaging (EPI) factor = 90, acquisition time = 55 s) with three orthogonally applied gradients (b values of 1,000, 500, and 0) was used. Other sequences included axial gradient echo (GRE) (TR/TE/excitation = 350/30/2, flip angle of 30°, slice thickness/gap = 5 mm/0.5 mm, FOV = 230 mm, matrix = 256 × 256, time of acquisition = 5 min, 4 s), axial spin echo (SE) T1-weighted (TR/TE/excitation = 425/14/2, FOV = 230 mm, slice thickness/gap = 5 mm/0.5 mm, matrix = 256 × 256, time of acquisition = 4 min, 28 s), turbo spin echo (TSE), proton density (PD), T2-weighted (TR/TE/excitation = 2500/13-120/1, turbo factor = 5–15, FOV = 230 mm, slice thickness/gap = 5 mm/0.5 mm, matrix = 256 × 256, time of acquisition = 3 min, 30 s), and axial flair (fluid attenuated inversion recovery) (TR/TE/TI/excitation = 9000/117/2500/2, turbo factor = 31, FOV = 230 mm, slice thickness/gap = 5 mm/1 mm, matrix = 256 × 256, time of acquisition = 3 min, 20 s) sequences.

A neurologist (YKC), who was blind to the results of the other assessments, evaluated the MRIs for the following.

  1. 1.

    Brain infarcts Lesions approaching the signal characteristics of cerebrospinal fluid on the T1-weighted images and measuring over 3 mm in diameter were regarded as brain infarcts [27]. Both the number and size of infarcts were measured. The total area of acute infarcts was measured with manual outlines of all areas with restricted water diffusion that were identified on the DWI with b values of 1,000. Their total volume was calculated by multiplying the total area by the sum of the slice thickness and gap. Inter- and intra-rater reliability tests were performed on 20 participants, with good agreement found (volume of acute infarcts: inter-rater kappa = 0.93, intra-rater kappa = 0.96; number of infarcts: inter-rater kappa = 0.89, intra-rater kappa = 0.94).

  2. 2.

    CMBs CMBs were defined as small (2–10 mm) hypointense lesions on the T2*-weighted GRE sequence, but symmetric basal ganglia calcification and flow void artifacts of the pial blood vessels were excluded [28]. The CMB locations were divided into lobar (cortex and sub-cortical white matter), deep (caudate, lentiform, thalamus, internal, and external capsule), and posterior fossa (brainstem and cerebellum) groups [29]. The number of CMBs in each region was also recorded. Intra-rater reliability tests of the CMB measurements were performed on 30 participants, with good agreement found [presence of CMBs: intra-rater kappa = 0.85; number of CMBs: intra-rater intra-class correlation (ICC) = 0.95].

  3. 3.

    WMLs The extent of WMLs was graded using a modified version of Fazekas’ scale [30], which scores deep and subcortical WMLs as mild, moderate, or severe [31]. The WMLs were scored on the basis of the PD images. The inter- and intra-rater weighted kappas for the WMLs were 0.85 and 0.90, respectively.

Statistical analysis

Descriptive data are presented as proportions or means, as appropriate. The correlation between the participants’ clinical (age, sex, education, NIHSS, MMSE, and GDS scores) and radiological (number and volume of infarcts, presence of CMBs, and severity of WMLs) characteristics and the SF-36 domain and summary scores were examined using Spearman’s correlation, except for the correlations between age and the SF-36 scores, which were examined with Pearson’s correlation because not all of the other variables were normally distributed. Both the unadjusted and Bonferroni-adjusted p values were computed. Variables with an unadjusted p < 0.05 in these analyses were entered into different multivariate stepwise linear regression models to determine the partial correlations of the predictors of the SF-36 domain and summary scores. All statistical tests were performed using SPSS for Windows (Release 14.0; SPSS Inc., Chicago, IL, USA).

Results

The study sample (n = 458) had the following characteristics: 74.5% were married, and their mean age and educational level were 66.2 + 11.9 and 5.2 + 4.5 years, respectively; 99.8% of the MRIs were obtained within 1 week of the stroke. The mean number and volume of acute infarcts were 1.1 + 1.7 and 3.0 + 9.3 ml, respectively. CMBs were found in 111 (24.2%) participants, and the mean number of CMBs was 5.6 + 8.7. A significant correlation was found between the volume of acute infarcts and the presence of CMBs (rho = 0.269, p < 0.001).

The psychiatric, social, and HRQoL assessments took place 14.6 + 2.0 weeks after the index stroke. The mean MMSE, GDS, LSNS, and NIHSS scores were 25.9 + 3.3, 5.0 + 3.7, 30.4 + 8.0, and 4.7 + 3.4, respectively. The PF, RP, BP, GH, VT, SF, RE, and MH subscale scores were 71.6 ± 27.9, 58.1 ± 41.4, 77.5 ± 24.6, 49.8 ± 20.6, 64.6 ± 21.3, 89.9 ± 15.6, 80.0 ± 35.0, and 73.5 ± 19.4, respectively.

Univariate analyses revealed that the female sex, education, neurological deficits (NIHSS), depressive symptoms (GDS), and cognitive function (MMSE) were correlated with all eight subscale and two summary scores (PCS and MCS; Table 1). The presence of lobar CMBs was negatively correlated with the PF, SF, RE, and PCS scores. After Bonferroni adjustment, the correlation between lobar CMBs and PF remained significant, whereas those between lobar CMBs and SF and RE became borderline and nonsignificant, respectively. Deep CMBs were also negatively correlated with the SF and RE scores, whereas posterior fossa CMBs were negatively correlated with the RP score. The severity of WMLs was correlated with four of the SF-36 subscale scores and the PCS score, whereas the number of old infarcts was correlated with the PF and RP scores of the SF-36 (Table 2).

Table 1 Statistically significant correlations between demographic and clinical characteristics and quality of life
Full size table
Table 2 Statistically significant correlations between radiological characteristics and quality of life
Full size table

Linear regression analysis was performed for the PF, SF, RE, RP, PCS, and MCS scores (Table 3), and the presence of lobar CMBs was found to be independently associated with the PF and SF scores (p < 0.05).

Table 3 Partial correlations from linear regression analysis of quality of life (dependent variable) and demographic, clinical and radiological variables
Full size table

Discussion

To the best of our knowledge, this study was the first to examine the impact of CMBs on the HRQoL of stroke patients. The results show that lobar CMBs are correlated with HRQoL in stroke survivors. The study sample comprised a large cohort of stroke patients, and the radiological findings were examined against the background of several clinical and psychosocial correlates of HRQoL.

In addition to being a marker of underlying vascular disease, CMBs may also have a direct effect on neurological function, cognition, psychiatric morbidity, and disability, thereby resulting in poorer HRQoL. A higher number of baseline CMBs was associated with the greater risk of cognitive impairment, functional dependence, and death in a 3-year prospective study of patients with lobar hemorrhages [32]. CMBs have also been associated with executive function impairment in patients with ischemic stroke [6]. Finally, CMBs are known to be an independent correlate of emotional liability in stroke [9].

If CMBs do have a direct effect on HRQoL, then their location would be expected to play a part. Indeed, our findings suggest that only lobar, not deep or posterior fossa, CMBs are independent correlates of HRQoL. This finding also suggests that the CMBs associated with cerebral amyloid angiopathy, which has a lobar distribution, may have a greater impact on HRQoL in stroke. The importance of CMB location has been reported previously in studies of the effects of frontal lobe and basal ganglia CMBs on executive function [6] and thalamic CMBs on emotional liability [9].

The current study has several limitations. The assessment of PSD was made only once, at the 3-month follow-up. Hence, the causality of the relationship between the MRI findings and the long-term HRQoL remains unknown. Moreover, patients who did not receive an MRI examination and those with more severe stroke resulting in prolonged hospitalization and physical frailty were excluded. Another limitation is that the SF-36 is a generic scale, and some investigators have questioned its validity in stroke [33]. Stroke-specific measures, such as the Stroke-Specific Quality of Life Scale, may have produced more valid information and should be used in future studies. Finally, the CMBs were measured in GRE with a slice thickness of 5 mm, which is less sensitive than thin-section susceptibility-weighted imaging [34].

In summary, the results of this study suggest that lobar CMBs have an impact on the HRQoL of stroke survivors. The importance of CMBs to other short- and long-term outcome measures in stroke warrants further investigation.