Abstract
Background: It is unknown whether the development of cerebral microbleeds (MBs), small areas of signal loss on T2*-weighted gradient-echo imaging (GRE), follows a slow or a rapid process. We hypothesized that MBs may develop rapidly after certain critical events, such as strokes, and investigated the frequency, location, and factors associated with the formation of new MBs after acute ischemic stroke.
Methods: We retrospectively examined 237 consecutive acute ischemic stroke patients who underwent MRI within 24 hours and follow-up MRI during the week after symptom onset. We defined new MBs as MBs that newly appeared on follow-up GRE outside the infarcted area. We examined the association of new MBs with demographics, risk factors, laboratory data, baseline MBs, and small vessel disease (SVD; leukoaraiosis and lacunar infarctions).
Results: Seventy-five patients (31.6%) had baseline MBs, and 30 (12.7%) developed new MBs. Multiple logistic regression analysis indicated that the presence of baseline MBs (odds ratio [OR] 5.72, 95% confidence interval [CI] 2.12–15.42, p = 0.001) and severe SVD (OR 2.94, 95% CI 1.12–7.77, p = 0.03) independently predicted the development of new MBs. Of the 56 new MBs, 29 (51.8%) appeared in the lobar location, 17 (30.4%) appeared in the deep location, and 10 (17.9%) appeared in the infratentorial location.
Conclusions: This study suggests that new microbleeds (MBs) can develop rapidly after acute ischemic stroke. Baseline MBs and severe small vessel disease are predictors for the development of new MBs. Further studies will be needed to investigate the clinical implications and mechanisms of these findings.
Glossary
- CI=
- confidence interval;
- DWI=
- diffusion-weighted imaging;
- ESR=
- erythrocyte sedimentation rate;
- FLAIR=
- fluid-attenuated inversion recovery;
- GRE=
- T2*-weighted gradient-echo imaging;
- ICH=
- intracerebral hemorrhage;
- MB=
- microbleed;
- OR=
- odds ratio;
- SVD=
- small vessel disease;
- T2WI=
- T2-weighted imaging.
With recent advances in MRI technology, there have been numerous studies of cerebral microbleeds (MBs), small areas of signal loss on T2*-weighted gradient-echo imaging (GRE).1–3 MBs are seen in approximately 60% of patients with primary intracerebral hemorrhage (ICH), 30% of patients with ischemic stroke, and 5% to 38% of asymptomatic elderly individuals.4,5 The presence of MBs is associated with future bleeding risk in ischemic stroke, primary ICH, and cerebral amyloid angiopathy.6–8 MBs are also related to cognitive impairment.9
Based on pathology studies, MBs have been considered to be old microhemorrhages that develop over an individual’s lifetime.10,11 This is supported by the finding that MBs are more common in elderly than in young individuals.4,8 There have been only a few serial GRE studies to investigate the natural history of MBs. According to previous studies, newly developed MBs were frequently observed on long-term follow-up GREs. For patients with cerebral amyloid angiopathy and previous lobar hemorrhage, new MBs have been reported to occur in 38% of patients during the 1.5-year follow-up period.12 For hypertensive patients with ICH, new MBs were found in 30% of those after a median period of 23 months.7 It is unknown, however, whether the development of MBs follows a slow or a rapid process. Some studies suggested that MBs could develop shortly after antiplatelet therapy or after thrombolysis in acute ischemic stroke.13,14 In addition, previous studies showed that MBs were common in ischemic stroke patients but rare in TIA patients,15 and that MBs were more common in patients with recurrent strokes than in those experiencing a first stroke.4 These findings suggest that ischemic stroke may trigger the development of MBs.
These observations led us to hypothesize that MBs may develop rapidly after certain critical events, such as strokes. This study aimed to investigate the frequency, location, and other factors associated with the development of new MBs shortly after acute ischemic stroke.
METHODS
Patients.
This retrospective analysis considered all stroke patients admitted to Asan Medical Center, Seoul, Republic of Korea, between November 1, 2002, and June 30, 2005. We included patients who 1) had a diagnosis of ischemic stroke confirmed by diffusion-weighted imaging (DWI); 2) underwent MRI, including DWI, T2-weighted imaging (T2WI) or fluid-attenuated inversion recovery (FLAIR) imaging, GRE, and magnetic resonance angiography within 24 hours of symptom onset; and 3) underwent follow-up MRI, including GRE, within 7 days of onset. The onset was considered to be the time when a patient was last known to be without ischemic symptoms. According to our MRI protocol, all patients who were admitted within 24 hours of onset were scheduled for scans acutely and 3 to 5 days after onset. Because of clinical care requirements or patient requests, follow-up scans were sometimes performed outside of the target range of times or not at all. We excluded patients whose image quality was poor because of motion artifacts. During the acute phase of stroke, all patients received appropriate management (e.g., thrombolysis, antiplatelet agents, anticoagulants, statins), as standardized by our stroke center’s care pathway.
Standard protocol approvals, registrations, and patient consents.
This study was approved by the institutional review board of our center. Written informed consents were not obtained because of the retrospective design.
MRI.
MRI examinations were performed with a 1.5-T MRI unit, as previously described.16 The GRE parameters were 5-mm slice thickness, 2-mm interslice gap, 20 axial slices, 250-mm field of view, 400-msec repetition time, 30-msec echo time, 20° flip angle, and 256 × 192 matrix.
MBs were defined as unambiguous homogeneous round signal loss lesions with diameters of up to 5 mm, as determined by GRE. Hypointense lesions within the subarachnoid space were regarded as pial blood vessels. Hypointense lesions in the areas of symmetric hypointensity of the globus pallidus were regarded as calcifications. New MBs were defined as MBs that newly appeared on the follow-up GRE. Newly appearing MBs in the acute infarct area were not regarded as new MBs. First, we described the numbers and locations of all MBs on baseline and follow-up GREs independently. Then, we performed slice-to-slice comparison to determine whether the MBs on follow-up GRE were new, had disappeared, or were unchanged. MBs that were questionably hypointense at the initial GRE but became conspicuous at the follow-up were not regarded as new MBs. The locations of baseline and new MBs were classified as 1) lobar if present in the cortex, subcortex, and white matter of the frontal, parietal, temporal, occipital, and insular lobes; 2) deep if present in the head of the caudate, putamen, globus pallidus, internal capsule, and thalamus; and 3) infratentorial if present in the midbrain, pons, medulla, and cerebellum. We also counted the number of baseline and new MBs. By definition, old ICH (as determined by GRE) had a >5-mm slitlike hyperintense or isointense core surrounded by a hypointense rim.17
We used the Scheltens scale to score the degree of small vessel disease (SVD; leukoaraiosis and lacunar infarctions), because this scale reflects the volume of SVD and covers both supratentorial and infratentorial lesions.18
All pairs of GRE were interpreted jointly by 2 investigators who were blinded to the clinical data and other MRI sequences (except DWI to confirm that new MBs occurred outside the infarct area). The order of GRE images was not blinded, because follow-up GRE often reflected infarct areas as hyperintensities, and the time order of images could be estimated. T2WI or FLAIR was reviewed in a different session jointly by 2 investigators who were blinded to the clinical and GRE data. A third investigator was consulted in cases of disagreement.
Clinical assessments.
We obtained clinical and laboratory data on admission, and stroke subtypes from review of the prospectively collected stroke registry. Stroke subtypes were determined according to the classification of the Trial of Org 10172 in Acute Stroke Treatment (TOAST).19
Data analysis.
We compared baseline characteristics of patients with and without baseline MBs and also of those with and without new MBs. For univariate analysis, we used the Pearson χ2 test with the exact method, Student t test, Mann–Whitney U test, or Spearman rank correlation test as appropriate. We tried to evaluate whether new MBs occur in the same distribution as the initial MBs. For the analysis of this regional association between initial and new MBs, we compared the number of baseline MBs and that of new MBs in each location using the Fisher exact test only in the patients having both initial and new MBs. We also investigated whether new MBs are more frequent than MBs that disappeared (baseline MBs that were not observed on follow-up GRE) with the McNemar test using binominal distribution. Multiple logistic regression analysis was performed to estimate the independent contributions of variables to baseline MBs and new MBs. Variables with a p value ≤0.1 by univariate analysis were selected for entry into multiple logistic regression analysis. A 2-tailed p value <0.05 was considered to indicate a significant difference. SPSS for Windows (version 13.0, SPSS Inc., Chicago, IL) was used for all statistical analyses.
RESULTS
During the study period, 1,741 patients with ischemic stroke were admitted to our stroke center. Of those, 1,504 patients were excluded for the following reasons: 1,311 underwent initial MRI beyond 24 hours of symptom onset, 183 did not undergo follow-up MRI within 7 days of onset, and 10 patients’ MRI quality was poor for analysis. Thus, 237 remaining patients were included for the final analysis. Demographics and risk factors were not different between included and excluded subjects, except that diabetes was more frequent in excluded patients (468/1,504, 31.1% vs 57/237, 24.1%; p = 0.027). Baseline stroke severity measured by NIH Stroke Scale scores was more severe in included patients (median 7, range 0–28 vs median 5, range 0–36; p < 0.001). Regarding stroke subtypes, cardioembolism was more common in included subjects (74/237, 31.2% vs 308/1504, 20.5%; p < 0.001), and small vessel occlusion was more frequent in excluded subjects (393/1504, 26.1% vs 42/237, 17.7%; p = 0.005). Of the finally included patients, 142 (59.9%) were men and 95 were women, with a median age of 65 years (mean 64.0 ± 12.8 years, range 23–93 years). The median time from symptom onset to initial GRE was 4.7 hours (range 0.3–23.9 hours). The median time from symptom onset to follow-up GRE was 4 days (range 1.0–7.0 days). The most common stroke subtype was large artery atherosclerosis (n = 87, 36.7%), followed by cardioembolism (n = 74, 31.2%) and small vessel occlusion (n = 42, 17.7%). The median SVD score was 9.0 (range 0–41). SVD scores in quartiles 1, 2, 3, and 4 were 0–2, 3–8, 9–16, and 17–41. Fifty-two patients received thrombolytic therapy (25 received IV tissue plasminogen activator, 19 received intra-arterial urokinase, and 8 received both). One hundred sixteen patients received anticoagulants, 120 patients received antiplatelet agents, and 1 patient received none because of infective endocarditis.
Frequency of baseline MBs and new MBs.
On initial GRE, we observed 424 MBs in 75 patients (31.6%). The median number of baseline MBs was 2 (range 1–33). On follow-up GRE, we observed 56 new MBs in 30 patients (12.7%) (figure). The median number of new MBs among these patients was 1 (range 1–5). Baseline MBs were not observed on follow-up GRE in 7 patients (3.0%). One MB disappeared in each of 6 patients, and 2 MBs disappeared in 1 patient. New MBs occurred more frequently than MBs that disappeared (p < 0.001).
Figure Examples of new microbleeds in 6 patients
For each patient (A–F), the baseline T2*-weighted gradient-echo imaging (GRE) is on the left, and the follow-up GRE is on the right. Arrows indicate new microbleeds.
Predictors of baseline MBs.
The presence of MBs on initial GRE was associated with old age (p < 0.001), hypertension (p = 0.02), high systolic blood pressure (p = 0.002), prolonged prothrombin time (p = 0.04), elevated erythrocyte sedimentation rate (p = 0.02), old ICH on GRE (p = 0.001), high SVD score (p < 0.001), and no current smoking (p = 0.01). The time between onset and initial GRE was not associated with MBs. Multiple logistic regression analysis indicated that the presence of old ICH on GRE (odds ratio [OR] 3.37, 95% confidence interval [CI] 1.14–9.94, p = 0.03) and SVD score (OR 1.12, 95% CI 1.07–1.17, p < 0.001) were independently associated with MBs at the initial GRE.
Predictors of new MBs.
The factors associated with new MBs were old age (p = 0.02), presence of baseline MBs (p < 0.001), and high SVD score (p < 0.001). New MBs were more frequently observed in patients with baseline MBs (23/75, 30.7%) than in those without (7/162, 4.3%). The proportion of new MBs increased as SVD quartile increased. In addition, the number of new MBs was associated with the number of baseline MBs (Spearman ρ = 0.57, p < 0.001) and SVD score (Spearman ρ = 0.36, p < 0.001). Increased systolic blood pressure (p = 0.05), diastolic blood pressure (p = 0.05), body temperature (p = 0.08), and the presence of old ICH on initial GRE (p = 0.07) were marginally associated with new MBs. However, the time between initial and follow-up GRE was not associated with the presence of new MBs (table 1). Multiple logistic regression analysis indicated that the number of baseline MBs (OR 1.11, 95% CI 1.03–1.21, p = 0.01), SVD score (OR 1.06, 95% CI 1.01–1.12, p = 0.02), and increased body temperature (OR 2.84, 95% CI 1.04–7.77, p = 0.04) independently predicted the presence of new MBs. When we replaced the numerical variables of the number of baseline MBs and SVD score with dichotomized variables, namely, the presence of baseline MBs and severe SVD (≥75 percentile [SVD score ≥17]), the ORs of these variables were 5.72 (95% CI 2.12–15.42, p = 0.001) and 2.94 (95% CI 1.12–7.77, p = 0.03) (table 2).
Table 1 Clinical, laboratory, and imaging characteristics of new MBs
Table 2 Factors associated with new MBs based on multiple logistic regression analysis
Location of baseline MBs and new MBs.
There were 137 (32.3%) baseline MBs in the lobar location (89 in the cortex or subcortex and 48 in the white matter), 225 (53.1%) in the deep location, and 62 (14.6%) in the infratentorial location. For the 56 new MBs, 29 (51.8%) were lobar (14 in the cortex or subcortex and 15 in the white matter), 17 (30.4%) were deep, and 10 (17.9%) were infratentorial. Among 75 patients with baseline MBs, 23 had lobar MBs, 17 had deep MBs, 2 had infratentorial MBs, 7 had lobar and deep MBs, 2 had lobar and infratentorial MBs, 6 had deep and infratentorial MBs, and 18 had lobar, deep, and infratentorial MBs. Among 30 patients with new MBs, 14 had lobar MBs, 7 had deep MBs, 4 had infratentorial MBs, 3 had lobar and deep MBs, and 2 had lobar and infratentorial MBs. In 23 patients with both baseline MBs and new MBs, the regional association was observed in the lobar location (OR 16.25, 95% CI 1.44–183.09, p = 0.018).
DISCUSSION
Our study showed that approximately 13% of acute ischemic stroke patients developed new MBs within 1 week of stroke onset. Approximately 30% of ischemic stroke patients had MBs on initial GRE, in agreement with previous studies,4,20 and approximately 30% of those with baseline MBs developed new MBs. Our results indicate that MBs may occur rapidly during the acute period of ischemic stroke.
MBs were first described by studies using GRE in the mid-1990s.1–4 Since then, there have been few histopathologic reports of MBs.10,11 These studies have shown the presence of focal hemosiderin-laden macrophages, organized miliary pseudoaneurysms, amyloid angiopathy, and fibrohyalinosis in areas of GRE-detected MBs. Thereafter, MBs on GRE have been regarded as old microhemorrhages. To our knowledge, no systematic studies have investigated short-term serial GREs to explore the rapid appearance of new MBs.
In our study, patients with multiple MBs at baseline had increased risks of development of new MBs. If multiple MBs were present at the initial clinical presentation, those lesions might have occurred at various times before or even after the onset of stroke. This suggests that there may be prolonged risks of MBs in some stroke patients. Moreover, it agrees with previous observations that the total number of baseline MBs predicted the risk of future hemorrhages in patients with cerebral amyloid angiopathy, ICH, and ischemic stroke.7,8,21 In those previous studies, the researchers suggested that the number of baseline hemorrhages was a marker of the severity and aggressiveness of the underlying vascular disease.
Our study also showed that severe SVD was an independent risk factor for development of new MBs. The SVD severity is known to be well correlated with the number of MBs.22 MBs result from microangiopathic changes after chronic hypertension; thus, the number of MBs may reflect bleeding-prone and SVD-prone microangiopathy.11
The cellular mechanism that underlies the formation of new MBs is unknown. Endothelial activation and damage with subsequent breakdown of the blood-brain barrier are key features in cerebral small vessel diseases.23 In susceptible elderly patients with long-standing hypertension, reflected by the presence of baseline MBs and severe SVD, presence of an acute ischemic state may precipitate endothelial dysfunction and result in the formation of new MBs. Sudden elevation of blood pressure associated with stroke may accelerate the leakage of blood. Our results indicate that hyperthermia may also contribute to this process, because patients with new MBs had higher body temperatures, and hyperthermia is known to induce blood-brain barrier disruption after ischemic stroke.24 It is unlikely that the new MBs that we observed resulted from tiny hemorrhagic transformations of index ischemic stroke, because we counted only new MBs outside the infarcted area. Medications such as thrombolytic agents, anticoagulants, and antiplatelet agents may affect the development of new MBs. However, we could not observe the association between these agents and the formation of new MBs.
The clinical implication of new MBs is uncertain at present. We found that new MBs were not associated with acute treatment modalities (e.g., thrombolysis, anticoagulation) or hemorrhagic transformation. MBs by themselves may be associated with cognitive dysfunction, independent of the ischemic severity.9 It was hypothesized that cognitive dysfunction resulted from associated tissue damage by MBs. Therefore, new MBs per se may be associated with cognitive decline. However, we did not evaluate cognitive function in our patients.
Our study has limitations. First, although we prospectively collected data using a stroke registry and MRI protocol, our study had a retrospective design and might have a potential risk of selection bias. There were some differences between included and excluded patients in terms of frequency of diabetes, stroke severity, and stroke subtypes. Because we included patients who underwent initial MRI within 24 hours of onset, milder stroke patients with lacunar infarcts might have been excluded because of late arrival to the emergency department. Second, image coregistration was not performed. We tried to do image coregistration, but the image quality became poor after coregistration because of the interslice gap (2 mm) in our GRE protocol. It may be argued that MBs smaller than 2 mm that were missed on the initial GRE might have been misidentified as new MBs at follow-up. In that case, however, a comparable number should have disappeared on the follow-up imaging, but this was not the case. Nonetheless, because the order of GREs could not be blinded, the raters might be biased to detect more new MBs than MBs that disappeared on the follow-up GREs. Third, we used 1.5-T MRI, which is inferior to 3-T MRI in the detection of MBs.25 The use of 3-T MRI may have allowed detection of more MBs. Fourth, we did not perform follow-up CT scans, which might detect even small foci of new hemorrhages.
Our study suggests that new MBs can occur rapidly, within 1 week of ischemic stroke. The presence of baseline MBs and severe SVD may be associated with the development of new MBs. Future prospective studies are necessary to validate this finding and should explore the clinical implications and potential mechanisms of new MBs after acute ischemic stroke.
AUTHOR CONTRIBUTIONS
Statistical analysis was performed by Sang-Beom Jeon, MD, and Sung-Cheol Yun, PhD, Departments of Neurology (S.-B.J.) and Division of Epidemiology and Biostatistics Clinical Research Center (S.-C.Y.), Asan Medical Center, University of Ulsan College of Medicine.
Footnotes
-
Editorial, page 1614
e-Pub ahead of print on September 16, 2009, at www.neurology.org.
Supported by a grant (M103KV010010 06K2201 01010) from the Brain Research Center of the 21st Century Frontier Research Program funded by the Ministry of Science and Technology of Korea and a grant (A060171) of the Korea Health 21 R&D Project, Ministry of Health & Welfare.
Disclosure: The authors report no disclosures.
Received March 17, 2009. Accepted in final form July 27, 2009.
REFERENCES
Letters: Rapid online correspondence
REQUIREMENTS
You must ensure that your Disclosures have been updated within the previous six months. Please go to our Submission Site to add or update your Disclosure information.
Your co-authors must send a completed Publishing Agreement Form to Neurology Staff (not necessary for the lead/corresponding author as the form below will suffice) before you upload your comment.
If you are responding to a comment that was written about an article you originally authored:
You (and co-authors) do not need to fill out forms or check disclosures as author forms are still valid
and apply to letter.
Submission specifications:
- Submissions must be < 200 words with < 5 references. Reference 1 must be the article on which you are commenting.
- Submissions should not have more than 5 authors. (Exception: original author replies can include all original authors of the article)
- Submit only on articles published within 6 months of issue date.
- Do not be redundant. Read any comments already posted on the article prior to submission.
- Submitted comments are subject to editing and editor review prior to posting.
