Background Cerebral microbleeds (MBs), defined as haemorrhagic microvascular lesions or microangiopathy in the brain, have traditionally been considered clinically silent. Recent studies, however, suggest that MBs are associated with a decline in cognitive function.
Objective To determine whether an association between MBs and cognitive function exists, we conducted a systematic review of the literature using the Cochrane Library, MEDLINE, EMBASE and the China National Knowledge Infrastructure database. We also searched the reference lists of relevant studies and review articles.
Results A total of seven studies were included. Qualitative meta-analysis of two studies suggested that the presence of MBs was significantly associated with cognitive impairment, while quantitative meta-analysis revealed an association between MBs and cognitive dysfunction in two studies (OR 3.06, 95% CI1.59 to 5.89) and implicated MBs as important in cognitive function decline in three other studies (standardised mean difference −1.06, 95% CI −2.10 to −0.02). MBs in the frontal or temporal region and the basal ganglia might also be related to cognitive dysfunction.
Conclusions These results suggest that rather than being clinically silent, cerebral MBs might be a factor inducing cognitive function decline.
Statistics from Altmetric.com
Cerebral microbleeds (MBs) are defined as haemorrhagic microvascular lesions or microangiopathy in the brain. They are usually identified as foci of signal loss on T2* weighted gradient echo MRI.1 ,2 MBs are associated with lacunar stroke and cerebral white matter lesions (WMLs), suggesting that they are a manifestation of pathology affecting cerebral small vessels. Such cerebral small vessel diseases, such as WMLs and lacunar infarcts, occur more frequently in the elderly and are linked with cognitive decline.3–5 Although it would be easy to conclude that MBs are associated with cognitive decline, MBs have also been detected in healthy individuals and in patients with non-cognitive neurological disorders,1 ,2 leading some to consider them clinically silent.
However, several pieces of evidence suggest that MBs are associated with a decline in cognitive function. MBs often occur in demented individuals, such as patients with Alzheimer's disease, cerebral autosomal dominant arteriopathy with subcortical infarcts and leukoencephalopathy subcortical vascular dementia.6–14 They usually occur as multiple foci throughout cortical and basal ganglia regions, and are identified histologically as tissue damage that may interrupt functionally important white matter tracts or frontal–subcortical circuits.5–7 Several recent studies have identified the presence and number of cerebral MBs as independent predictors of cognitive impairment that is associated with WMLs and lacunar infarcts.1–7
To determine whether an association exists between MBs and cognitive function decline, we conducted a systematic review of the literature and performed non-randomised meta-analyses.
Materials and methods
We restricted this review to clinical studies on cognitive function and cerebral MBs. Cochrane systematic review methods were applied as much as possible.
Two reviewers (CL and SL) independently identified studies through searches of the Cochrane Library (issue 12, 2011), MEDLINE (1966 to December 2011), EMBASE (1980 to December 2011) and the China National Knowledge Infrastructure database of Chinese biomedical research literature (1999 to December 2011). Reference lists of all relevant articles were also searched for additional studies.
Studies of cerebral MBs and cognitive function that included patients of any age and either sex were eligible. Cerebral MBs were defined as homogeneous, round focal areas observed throughout the brain, with a diameter <10 mm and very low signal intensity on MRI. Symmetrical hypointensities in the globi pallidi (calcification or iron deposition) and flow voids from cortical vessels were disregarded.1–5
To be eligible, studies had to evaluate either global cognitive function in the aggregate or at least domains of cognitive function separately, including psychomotor speed, fluency, attention, memory, processing speed, executive function, naming, calculation, language, recall and abstraction. Outcome measures had to include the presence of MB lesions. Studies had to be controlled, with participants divided into an MBs group and a non-MBs group. Cognitive impairment, which had to be scored according to a widely accepted standardised scale, was defined as statistically significant deterioration in global cognitive function in any domain of cognitive function.
Data extraction and study quality
Two reviewers (CL and SL) independently selected studies that met the inclusion criteria and extracted data on the presence, location and basic characteristics of MBs, as well as on mean cognitive function scores in the MBs and non-MBs groups. They also extracted data on the mean difference in cognitive function scores between the two groups. Missing data were obtained from the authors whenever possible.
As non-randomised meta-analyses were carried out, study quality was assessed using a modified version of the Newcastle–Ottawa Scale. A score of up to 8 points was assigned to each study based on the quality of MBs and non-MBs group selection, comparability of groups and assessment of cognitive function.
Results were reported as ORs with corresponding 95% CIs for dichotomous data. The weighted mean difference or standardised mean difference was calculated for continuous data. Heterogeneity between study results was assessed using a standard I2 test. All calculations were carried out using statistical software provided by the Cochrane Collaboration (RevMan 5.1).
Description of studies
We identified 242 potential studies from our initial electronic databases and reference lists, of which 201 studies were excluded after review of the titles and abstracts. The full text of the remaining 41 studies was read. Of those studies, 34 were excluded because neither the presence nor the location of the MB lesions was reported, or because the articles were reviews or commentaries. Seven studies, comprising 527 participants in the MBs group and 4476 in the non-MBs group, met all of the inclusion criteria and were included in the systematic review (figure 1).1–7 The basic characteristics of the included studies are summarised in figure 2. Participants in one study were patients with ischaemic cerebral infarction7 while participants in the remaining six studies were healthy individuals.
Two studies4 ,5 received quality scores of 6 out of 8, and the remaining studies received quality scores of 5 (figure 3). Only one study4 included blinded assessment of MB lesions. None of the studies featured blinded assessment of cognitive function. All studies were cross sectional rather than longitudinal.
MB lesions in all seven included studies were defined similarly and were identified by MRI. Cognitive function was evaluated in different ways. Two studies2 ,3 used the Mini-Mental State Examination (MMSE) and another two6 ,7 used the Montreal Cognitive Assessment Scale (MoCA). These generalised screening instruments assess global cognitive function, but not function in specific cognitive domains. Three studies1 ,3 ,5 used a more comprehensive series of neuropsychological tests to examine several cognitive function domains separately.
Two studies4 ,5 did not report the numbers of participants or mean cognitive function scores in the MBs and non-MBs groups, so we performed a qualitative meta-analysis to examine the association between MB lesions and cognitive function. Another two studies1 ,7 reported the number of participants in the MBs and non-MBs groups but not mean function scores, so we performed dichotomous meta-analyses on data from these studies.
Three studies2 ,3 ,6 reported mean cognitive function scores in the MBs and non-MBs groups, allowing us to perform continuous meta-analyses. These three studies showed significant heterogeneity (I2=97%), probably reflecting the quite different numbers of participants and high SD for cognitive function scores.
Qualitative analysis of cognitive function
In one study,4 global cognitive function was assessed using the MMSE, and an MMSE score >1.5 SD below the mean for the participant's age group was considered subnormal. This study showed that there was a significant association between the presence of MBs and cognitive impairment.
Another study5 measured cognitive function using a battery of neuropsychological tests covering seven domains: global cognitive function, verbal memory, visuospatial memory, psychomotor speed, fluency, concept shifting and attention. Global cognitive function was evaluated using the MMSE and the Cognitive Index, defined as a compound score that was calculated as the mean of the Paper–Pencil Memory Scanning Task, Stroop Test, Symbol–Digit Substitution Task, combined score on the three learning trials of the Rey Auditory Verbal and Learning Test and the delayed recall of this last test. Psychomotor speed was the mean of the z scores of the 1 letter subtask of the Paper–Pencil Memory Scanning Task, the reading subtask of the Stroop Test and the Symbol–Digit Substitution Task. Attention was a compound score of the z score of the total time of the Verbal Series Attention Test. Cognitive Index, psychomotor speed and attention were significantly lower in the MBs group than in the non-MBs group.
Neither of these studies4 ,5 reported the number of participants in the MBs and non-MBs groups who experienced a decline in cognitive function. As we were unable to obtain the original data from the study authors, we did not include them in the meta-analysis. Nevertheless, the two studies reported that the presence of MBs was significantly associated with cognitive impairment.
Quantitative analysis of cognitive function
One meta-analysis was carried out on data pooled from two studies.1 ,7 In one study,1 standard neuropsychological tests were used to assess areas of cognition, including naming, visual perception, speed and attention, and executive function. Only executive function declined in the non-MBs group. Executive functions were assessed using two or more of the following tests: Stroop Test, word fluency, Trail Making Test, Weigl Color–Form Sorting Task and the Modified Card Sorting Test. Another study used the MoCA to assess cognition.7 In that study, an MoCA score ≥26 was classified as normal, and a score <26 was classified as cognitive dysfunction. Data pooled from these two studies showed that the OR for cognitive function decline in the MBs group relative to the non-MBs group was 3.06 (95% CI 1.59 to 5.89) (figure 4).
Another meta-analysis was carried out on data pooled from three studies.2 ,3 ,6 One study2 assessed global cognitive function using the MMSE. In that study, total scores <27 or >1.5 SD below the mean for the given age were defined as subnormal. Another study3 used a battery of neurocognitive tests to measure three domains of cognitive function: memory score, processing speed and executive function. Memory function was evaluated using a modified version of the California Verbal Learning Test; processing speed using the Digit Symbol Substitution Test, the Salthouse Figure Comparison Test, and parts 1 and 2 of the Stroop Test; and executive function using the Digits Backward, a shortened version of the CANTAB spatial working memory test and part 3 of the Stroop Test. There was a significant decline in executive function and processing speed in the non-MBs group. The third study6 measured cognition using the MoCA. An MoCA score <26 was regarded as cognitive dysfunction, while a score ≥26 was regarded as normal.
These three studies2 ,3 ,6 reported the means and SD of cognitive function scores for MBs and non-MBs groups. As they showed significant heterogeneity (I2=97%, p<0.001), a random effect model was used. Meta-analysis showed that MBs were an important factor in decline in cognitive function (standardised mean difference −1.06, 95% CI −2.10 to −0.02) (figure 5).
Location of MB lesions and cognitive function
One study reported that participants with executive dysfunction had more MBs than participants without dysfunction in both the frontal region (mean count 1.54 vs 0.03; p=0.002) and the basal ganglia (mean 1.17 vs 0.32; p=0.048).1 Another study indicated that the presence of MBs in the basal ganglia might be related to lower ‘attention and calculation’ scores (p=0.014), while the presence of thalamic MBs might be independently related to lower scores for ‘total function’ (p=0.036) and ‘orientation’ (p=0.012).2 A third study reported that the presence of MBs in the frontal and temporal lobes might be associated with decline in verbal memory, visuospatial memory and psychomotor speed, while the presence of MBs in the basal ganglia might be associated with lower scores for global cognitive function, psychomotor speed and attention.5 The remaining studies did not report a relationship between MB location and cognitive function.3 ,4 ,6 ,7
This systematic review sought to generate definitive evidence about whether MBs might lead to cognitive function decline, but only seven studies satisfying the inclusion criteria were identified. The evidence from these studies suggests that MBs might be associated with cognitive dysfunction, but these findings are tentative in light of the fact that all of the included studies were cross sectional and generally of poor methodological quality. High quality prospective studies are needed to address this question more rigorously.
Four of the included studies2 ,4 ,6 ,7 used the MMSE and MoCA scales to assess global cognitive function; their results suggested that MB lesions in the brain may result in a decline in cognitive function in the general population, especially in the elderly. Unfortunately, the MMSE and MoCA are low resolution screening instruments that do not permit detailed analysis of specific domains of cognitive function. In contrast, the other three included studies1 ,3 ,5 used a series of neuropsychological tests to measure specific cognitive function domains, and their findings further supported the idea that MBs damaged cognitive function. These results explicitly related MB lesions to global cognitive dysfunction and executive dysfunction. However, the fact that most of the studies did not include a detailed analysis of cognitive function domains, and that the remaining studies found different cognitive function domains to be associated with MBs, means that which dimensions of cognitive function are affected by MBs remains controversial. MB lesions may affect different cognitive domains depending on differences in how centres perform neuroimaging, conduct neuropsychological and neurocognitive assessments, and define cognitive decline. Future studies should examine the relationship between MBs and specific cognitive domain function in greater detail.
The results from the seven included studies were consistent with findings from several studies that we excluded because they failed to satisfy the inclusion criteria. Three excluded studies found that the presence of MBs was consistently associated with cognitive dysfunction.15–17 In fact, one of these studies involved a 6 year follow-up and found that MBs had prognostic significance for progression of vascular cognitive impairment.16 Several studies reported that MB lesions played a role in MB associated vasculopathy—for example, the presence of multiple MBs, especially when in a strictly lobar location, was related to worse performance on cognitive function tests, even after the data were adjusted for vascular risk factors and other imaging markers of small vessel disease.17–21 A recent prospective cohort study showed that the presence of MB lesions located in the frontal and temporal lobes was associated with poorer cognitive performance in the elderly without dementia, independent of the presence of other cerebral small vessel disease related lesions.22 In another study, cerebral MB lesions in stroke clinic patients were consistently related to frontal executive impairment at a follow-up of 5.7 years and had prognostic relevance for long term cognitive outcome.23 Consistent with this outcome, another study reported that the absence of cerebral MB lesions may be associated with a higher likelihood of reversible cognitive dysfunction in patients with acute ischaemic stroke.24
Our meta-analysis suggests that not only the presence of MBs but specifically their location in the frontal or temporal region and the basal ganglia might be associated with a decline in cognitive function.1–5 This result is consistent with some of the studies excluded from our systematic review.9–14 It is also consistent with the fact that the frontal lobes, basal ganglia and thalamus participate in frontal–subcortical circuits involved in cognitive function. Cerebral MBs in these regions are associated with tissue necrosis that damages frontal–subcortical circuits or white matter tracts, inducing cognitive impairment.1 ,2 ,5 In contrast, one excluded study reported that the presence and number of MBs were not associated with cognitive function in patients with lacunar stroke and radiological (ischaemic) leukoaraiosis.18 This discrepancy might reflect a confounding effect, as lacunar stroke and radiological (ischaemic) leukoaraiosis could induce cognitive function decline,19 thereby masking the effects of MB lesions.
This systematic review is limited by the limitations of the included studies. Only seven studies involving a relatively small number of participants were included. Different studies evaluated cognitive function using scales with different sensitivity and specificity, which is likely to cause heterogeneity in scoring. Most of the included studies assessed cognitive function using the MMSE and MoCA; these instruments are not specific or sensitive enough to draw conclusions about specific cognitive domains. In addition, MMSE and MoCA scores are easily affected by age and educational level, such that higher education may lead to false negatives and lower education to false positives among the elderly.2 ,4 ,6 ,7 The remaining studies in our meta-analysis relied on standard neuropsychological tests and neurocognitive tests to examine cognitive domains individually, but it is difficult to compare these results with the aggregate scores on the MMSE and MoCA.
A further limitation in this systematic review is that different studies identified MB lesions using different MRI protocols, which might lead to heterogeneity in the numbers of MB lesions detected. Most of the studies were subject to measurement bias because they did not involve blinded identification of MB lesions or blinded assessment of cognitive function. Most participants in the included studies were volunteers and consecutive patients, which may have introduced selection bias. Indeed, the non-randomised meta-analyses applied here are easily affected by confounding and bias.
Prospective cohort studies are needed to confirm the association between MBs and cognitive function. To examine this association as rigorously as possible, future studies should research all domains of cognitive function using appropriate tools instead of relying on low resolution screening tools such as the MMSE and MoCA. MRI analyses to identify MBs should be performed in a blinded manner, as should assessments of cognitive function. Study participants should come from the general population in order to avoid selection bias.
Contributors BW raised and designed this subject. CL collected and extracted data, and drafted the article. SL collected and extracted data. WT and ZH revised the important intelligent content. ML and BW approved the final version.
Funding This research was supported by the National Natural Science Foundation of China (30900472) and the Science and Technology Support Program of the Department of Science and Technology of Sichuan Province (2012FZ0006).
Competing interest None.
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.