| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Neuroradiology |
1 From the Departments of Radiology (M.A., R.R.) and Internal Medicine (A.S., A.T.), Uppsala Hospital, Uppsala, Sweden; and the Department of Radiology, Turku University Hospital, Turku, Finland (P.S.). From the 2002 RSNA Annual Meeting. Received March 26, 2004; revision requested June 4; final revision received February 10, 2005; accepted February 22. Supported by grants from the Swedish Medical Research Council, the Stroke Foundation, and the Selander Foundation. Address correspondence to M.A., Hospital Municipal de Badalona (Resonancia Magnética), Via Augusta 9-13, 08911 Badalona, Barcelona, Spain (e-mail: malemany{at}crccorp.es).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
Materials and Methods: Before examinations, informed consents were signed by either the patient or a relative. The study was carried out with the approval of the local ethics committee. MR imaging was performed in 90 patients with acute stroke: 45 with acute spontaneous IPHs (24 men and 21 women; median age, 65 and 68 years, respectively) and 45 age-matched control subjects without intracranial hemorrhages (26 men and 19 women; median age for both, 67 years), as determined at computed tomography. MR imaging included transverse T1- and T2-weighted spin-echo, transverse fluid-attenuated inversion recovery, transverse and coronal T2*-weighted gradient-echo, and, in 50 patients, diffusion-weighted sequences. Presence of MHs and signs of microangiopathy, such as T2 hyperintensities or lacunae, were recorded in the white and deep gray matter. The relationships between MH and IPH and between MH and T2 hyperintensities were analyzed by means of regression analysis. Different clinical features, such as arterial hypertension or diabetes, were registered and correlated with the image findings by means of regression analysis.
Results: MHs were found in 64% of patients with IPH (29 of 45) and 18% of control subjects (eight of 45). A statistically significant relationship between MH and IPH was determined (P < .001). Among the 29 patients with IPH and MH, 24 (83%) had T2 hyperintensities and 13 (45%) had lacunae; among the 16 patients without MH, seven (44%) had T2 hyperintensities and three (19%) had lacunae. A relationship between MH and occurrence and extent of T2 hyperintensities was also identified (P < .001). There was no clear relationship with the clinical data studied.
Conclusion: The results support a correlation between the presence of imaging signs of cerebral microangiopathy, clinically silent MHs, and acute IPHs.
© RSNA, 2006
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The incidence of MHs in healthy volunteers and in patients has been studied (1–3,5,6,8–13), and results indicate that clinically silent MHs (ie, MHs with no detectable signs or symptoms) occur in 3%–6% of healthy elderly subjects (11,13) and 20% of patients with prior ischemic stroke (1,6,9). MHs have been reported as additional findings in 54%–71% of patients with acute spontaneous intracerebral hematoma, with a statistically significant association with arterial hypertension (2,5,6,8,10). A connection between MHs and cerebral microangiopathy has been suggested (8–11). The majority of these studies were retrospective and were performed in Asian populations (1–3,5,6,8,12,13).
The clinical importance and prognostic value of MH is still debated (10). These lesions have been interpreted as markers for vessel wall disorders with a consequent vascular vulnerability that increases the tendency of intraparenchymal bleeding (4). Patients with such lesions may have a higher risk of symptomatic spontaneous hematoma, and contradictory data have been published on the regional association between MHs and spontaneous hemorrhage (10,14); MHs can represent an increased risk for recurrent bleeding after an acute hemorrhagic event (10) or a risk of cerebral bleeding after brain ischemia (1,9,15). Furthermore, patients with ischemic stroke and MHs may have a higher risk of bleeding complications with oral anticoagulation (12) or after thrombolysis, with controversial results in the literature (16,17); because MHs directly indicate blood extravasation, this MR imaging evidence, rather than other clinical or morphologic variables, could serve to identify patients at high risk of hemorrhagic complication (12).
However, these facts have not yet been established. Computed tomography (CT), which is unable to demonstrate MHs, is the imaging modality most widely used for evaluating patients after stroke. Therefore, the incidence of MHs, their correlation with signs of cerebral microangiopathy, and their prognostic implications are difficult to assess in daily practice. Further MR imaging studies in patients with stroke are required for a concrete understanding of the importance of the presence of MH. The purposes of our study were to evaluate prospectively with MR imaging the coexistence of MHs in white patients with acute spontaneous intraparenchymal hemorrhage (IPH) and acute ischemic stroke and to examine the association with imaging findings of microangiopathy and various clinical data.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
MR Imaging Technique
The MR imaging examinations were performed at 1.5 T. The imaging protocol included transverse spin-echo (SE) T1-weighted (500/14 [repetition time msec/echo time msec], intermediate- and T2-weighted (2300/16–120), fluid-attenuated inversion-recovery (FLAIR) (10000/140/2000 [repetition time msec/echo time msec/inversion time msec]), and transverse and coronal gradient-echo T2*-weighted (500/14; flip angle, 30°) sequences. The pixel size was 0.98 x 0.78 mm, and the section thickness was 5 mm for all pulse sequences. The intersection gap was 0.5 mm for SE and gradient-echo and 1.0 mm for FLAIR sequences. Diffusion-weighted images were obtained for 50 patients, including 44 of 45 control subjects (the examination could not be completed in one who complained of shoulder pain) and six patients with acute IPH. Transverse isotropic single-shot diffusion-weighted imaging was performed with an echo-planar FLAIR sequence (10000/95/2100) that had two b values: 0 and 1000 sec/mm2, with diffusion gradients applied in the x, y, and z directions at b = 1000 sec/mm2. Apparent diffusion coefficient maps were calculated.
Image Analysis
The images were assessed in consensus by a radiologist (M.A.) and a neuroradiologist (R.R.) with 7 and 19 years of experience in brain MR imaging, respectively, and without prior knowledge of clinical or laboratory data. The following abnormalities were recorded: acute hematomas, MHs, hyperintensities on FLAIR images, lacunar infarcts, residuals from old hematomas, and ischemic lesions on diffusion-weighted images. Acute hematomas were grouped as (a) lobar, involving the cortex and/or white matter regions in both cerebral hemispheres; (b) striate or thalamic; or (c) infratentorial, involving the cerebellum or the brain stem. MHs were defined as homogeneous, rounded, and well-demarcated foci of signal intensity loss on T2*-weighted gradient-echo images, smaller than 5 mm in diameter and without surrounding edema (Fig 1). MHs were counted and grouped as lobar, striate or thalamic, or infratentorial (in the same locations as acute hematomas). Care was taken to exclude any calcifications detected with CT.
|
|
The presence of hyperintensities in the white matter and/or the deep gray matter (basal ganglia or thalamus) on FLAIR images was noted. The changes were visually graded as described in Table 1. The presence and number of lacunar infarcts were recorded; lacunar infarcts were defined as areas less than 10 mm in diameter, hypointense on T1-weighted SE MR images and hyperintense on T2-weighted SE and FLAIR MR images, and with a center isointense to cerebrospinal fluid with all sequences. Small areas of ischemic parenchymal destruction, smaller than 10 mm and isointense to cerebrospinal fluid, were registered as lacunae. In some instances, it was not possible to differentiate lacunae from local widening of Wirchow-Robin spaces.
|
Clinical Data
The charts for all patients were reviewed independently by a neurologist (A.S.), and the following data were recorded: (a) arterial hypertension, defined as a history of hypertension or repeated blood pressure measurements over 140/90 mm Hg (patients with high blood pressure at arrival at the hospital [or during the first days] that normalized without treatment were not considered to be hypertensive); (b) diabetes, defined as either a history of diabetes or a fasting blood glucose level above 6.1 mmol/L (if that level rose during the first few days but normalized without treatment, the patient was not denoted as having diabetes); (c) coronary artery disease, defined as a history of myocardial infarction or current treatment for angina pectoris; (d) previous stroke, defined as physician-diagnosed symptomatic infarction without evidence of intracranial blood at CT (transient ischemic attacks were not included); (e) atrial fibrillation, permanent or paroxysmal, confirmed with electrocardiography; and (f) current medical treatment with anticoagulants or antiplatelet agents.
Statistical Analysis
The statistical analysis was done together by our statistical consultant and one of us (M.A.). A statistical software package (STATISTICA 2003, version 6; StatSoft, Uppsala, Sweden) was used for analyses. The relationships between MH and IPH, ischemic lesions, and grades of FLAIR hyperintensities were analyzed by means of regression analysis. The number of MHs was plotted against age to determine whether a relationship between those variables could be established. The relationship between MHs and clinical data were analyzed for patients with IPH and for the control subjects (patients without IPH), also by means of regression analysis. For all tests, P values of less than .05 were considered to indicate a significant difference.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
|
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
The coexistence of MHs and acute IPH in our study was similar to that identified in other studies (5,10). The distribution of MHs in different parts of the brain was similar to that found by Roob et al (10), who found 39% of MHs in cortical-subcortical regions and 38% in the basal ganglia or thalami (compared with 40% in cortical-subcortical regions and 42% in the basal ganglia or thalami in our study). With respect to regional associations between MH and the site of acute IPH, no specific patterns of MH distribution were identified; these results concurred with those of Roob et al (10).
Our results differed from those of Lee et al (14), who claimed that cerebral microbleeds are regionally associated with intracerebral hemorrhage. In their study, the parenchyma was divided into large regions for analysis, making it easier to find MHs within the same region as the IPH. MHs in patients with IPH are often multiple, which makes it difficult to analyze regional association. In our opinion, only a follow-up study can reveal a regional relationship; when the hematoma spreads over an area after a bleeding event, it is not possible to determine whether there has been an earlier MH in the same location. MHs may be a marker of generalized small-vessel disease that increases the risk of bleeding, but their presence may not predict the specific region of the brain where IPH will occur.
In our study, MHs were detected in 23% of the control subjects in whom acute ischemic changes were demonstrated with diffusion-weighted MR imaging. This result was similar to the results of Nighoghossian et al (9), who found MHs in 20% of patients with ischemic stroke. Further prospective studies are needed to assess whether such lesions are indicative of a higher risk for hemorrhagic transformation.
An association between MH and radiologic signs of cerebral microangiopathy, measured as the grade of leukoaraiosis and number of lacunae, has been reported (8–11). The results presented in our study provide further evidence that MH reflects severe microangiopathy, with vessel wall injury thus increasing the tendency for bleeding. The changes recorded in both white and deep gray matter and the strong correlation (P < .001) between hyperintensities, nonlacunar changes, and the presence of MH, raise the question of how related they are to the risk of hemorrhagic complications.
A strong association between MH and arterial hypertension has previously been reported (5,10,12). The association was not verified in our study, although a large proportion of patients with hypertension was observed among patients also with MHs.
No association was found between MH and the other clinical conditions analyzed (diabetes, previous stroke, coronariopathy, atrial fibrillation, or anticoagulant or antiplatelet therapy). These results differed from previously published results that indicated an association with diabetes (9) but were in concordance with the results of Roob et al (10). However, the data presented here are not extensive and cannot exclude these associations.
MH has been correlated with age (9,12). In our study, regression analyses for the number of MHs versus patient age indicated no correlation between the two factors for the whole age range. MHs have also been associated with microangiopathy due to cerebral amyloid angiopathy, in which fibrinoid necrosis due to progressive deposition of amyloid within the vessel wall is seen (4). The IPHs in cerebral amyloid angiopathy are typically cortical and lobar. Concomitant MR imaging evidence of cortical-subcortical microbleeding (18) supports this diagnosis. The MHs identified in our study had no predilection sites, and cerebral amyloid angiopathy was not suspected in any of the patients.
Other causes of hemorrhages in the brain include cavernomas and petechial lesions secondary to brain contusions or diffuse axonal injury after head trauma. In an MR imaging classification of cavernous malformations (19), type IV lesions are identical to MHs on T2*-weighted gradient-echo MR images, which means that some of the foci of signal intensity loss in our study might have represented cavernomas.
The diagnosis of brain contusion or diffuse axonal injury was highly unprovable in our study, because patients with previous head trauma were excluded. However, in clinical practice, a bleeding episode is difficult to date because some of these events are clinically silent. In some circumstances, such as after head trauma, the distinction between new MHs, indicating diffuse axonal injury, and old petechial lesions is of vital importance. Small hemorrhages may be undetected with SE MR sequences but are easily detected with T2*-weighted gradient-echo MR sequences (20–23). A study in rabbits (22) demonstrated that it is not possible to differentiate between old (6–7 months) and fresh hemorrhages if they are very small. On susceptibility-weighted MR images, the signal intensity patterns for hematoma residues and small acute hematomas are the same (signal intensity loss). The contrast in both cases is based on the same physical mechanism: the shortening of the T2* relaxation time owing to susceptibility effects, caused by hemosiderin and ferritin in chronic hematomas and by deoxyhemoglobin in intact red blood cells in acute hematomas. Similar results were obtained in the present study; some small lesions were detected only on T2*-weighted MR images, and their age could not be determined by means of image characteristics.
T2*-weighted gradient-echo MR sequences are accurate for detecting small amounts of blood products (20,24–26). In our study, MHs were detected only with T2*-weighted gradient-echo MR imaging and were invisible with CT and with SE and FLAIR MR imaging. Epiplanar sequences are intrinsically susceptibility weighted, and diffusion-weighted MR imaging has been recommended for the diagnosis of hemorrhages (26,27); in our study, however, diffusion-weighted MR imaging missed 77% (17 of 22) of MHs depicted with gradient-echo MR sequences. In addition, diffusion-weighted MR imaging resulted in 24 false-positive findings because of the low signal-to-noise ratio of the sequence, which resulted in the presence of dark pixels that did not correspond to blood products. In an earlier study (28), 27 of 30 MHs (90%) were missed with use of an SE-based diffusion-weighted MR imaging sequence, but there were no false-positive findings. Results of animal experiments have shown (22) that the areas of susceptibility changes are larger than are residuals of hemorrhages at histopathologic evaluation, and iron deposits in a limited number of cells are disclosed as hypointense foci on MR images. Susceptibility-weighted gradient-echo sequences should be included in MR imaging studies whenever the presence of intracranial blood is suspected, especially if MHs are analyzed.
MHs are better detected at MR imaging with higher field strengths, because susceptibility effects increase to the square of the field strength. High-spatial-resolution MR imaging has improved the detection of qualitative changes of the brain parenchyma, including changes related to microangiopathy (29). Diffuse axonal injury has been more effectively detected with T2*-weighted gradient-echo MR imaging at 3 T than at 1.5 T (30). With the extended clinical use of stronger magnets (eg, 3 T), the detection of MHs and morphologic signs of microangiopathy will improve, thus increasing the understanding of their physiopathologic correlation.
Other causes of signal intensity loss, such as calcium deposits, metallic particles, or air bubbles, can lead to differential diagnostic problems when the brain is imaged with T2*-weighted gradient-echo MR sequences. Signal intensity loss based on vascular flow void at cross sections of blood vessels can give the same MR imaging appearance as small deposits of blood products (hemosiderin). These diagnostic problems can be solved by acquiring MR images in different planes and with the help of other sequences (31) or, occasionally, CT.
The main limitation to this study was the lack of histopathologic confirmation that the so-called MHs actually represented deposits of blood-breakdown products. However, the foci of signal intensity loss in the present study were identical to the MR imaging characteristics that represent deposits of hemosiderin in lesions analyzed histopathologically in both patients (5,7) and animals (20,22); thus, we believe that these foci were signs of earlier hemorrhages.
The exclusion of patients whose clinical condition was too poor for them to undergo MR imaging could have caused bias. Their poor condition might be due to the presence of a bigger IPH, accompanied perhaps by more MHs, which might indicate more severe microangiopathy. This fact could have been detected only with MR imaging.
The clinical implications of diagnosing MH could include the prevention of hemorrhagic complications after thrombolysis. Intravenous and intraarterial thrombolysis improve outcomes when administered to selected patients with acute ischemic stroke. A limitation of these therapies is the occurrence of hemorrhagic transformation, which is symptomatic in 6%–10% of patients (17). Although the risk factors for bleeding during anticoagulation after cerebral ischemia have been studied (32–34), none of the patients evaluated were examined with MR imaging, and MHs were not mentioned as a possible risk factor.
Prior intracerebral hemorrhage is a contraindication to thrombolysis, but there are no guidelines for the treatment of patients with previous asymptomatic hemorrhages, most often MHs, detected only with MR imaging (17). Kidwell et al (17) postulated the need to exclude the presence of MH with brain MR imaging before thrombolysis. Although the majority of cases of hemorrhagic transformation after thrombolytic therapy may be caused by disruption of the blood-brain barrier in acutely injured tissue, some bleeding may be associated with small-vessel injury or MH, particularly those hemorrhages located in regions remote from the acute ischemic field (17).
The effect of MHs on the risk of cerebral bleeding after thrombolysis has been evaluated in a retrospective study (16) analyzing the presence of MHs on pretreatment T2-weighted MR images in 44 patients with acute ischemic stroke. The results suggest that patients with stroke with a small number of MHs can be treated safely with thrombolysis. Larger prospective studies including T2*-weighted MR imaging are needed to address the predictive value of the detection of MHs with regard to the risk of tissue plasminogen activator–induced IPH. The results of further MR imaging investigations could lead to changes in the policy for thrombolytic therapy and to the establishment of new guidelines.
In conclusion, the results from a white population confirm previous results and support the hypothesis of a correlation between the presence of clinically silent MHs, radiologic signs of microangiopathy (T2 hyperintensities and lacunae), and acute IPHs. The detection of MHs is of diagnostic importance and may indicate microangiopathy with an increased risk of bleeding. This raises the question of whether brain MR imaging including susceptibility-weighted gradient-echo sequences should be routinely performed for correct management of patients with stroke, especially before thrombolysis for pretreatment evaluation. Whether patients with MH should be excluded from thrombolytic therapy requires further investigation.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Abbreviations: FLAIR = fluid-attenuated inversion recovery • IPH = intraparenchymal hemorrhage • MH = microhemorrhage • SD = standard deviation • SE = spin echo
Authors stated no financial relationship to disclose.
Author contributions: Guarantor of integrity of entire study, R.R.; study concepts, R.R., M.A., A.T.; study design, R.R., M.A., P.S., A.T.; literature research, M.A., R.R.; clinical studies, A.S., A.T.; data acquisition, M.A., R.R., P.S., A.S.; data analysis/interpretation, M.A., R.R., P.S.; statistical analysis, M.A.; manuscript preparation and editing, M.A., R.R.; manuscript definition of intellectual content, M.A., R.R., A.T.; manuscript revision/review, R.R., A.T., M.A., A.S.; manuscript final version approval, all authors
|
References |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  C. Cordonnier, R. Al-Shahi Salman, and J. Wardlaw Spontaneous brain microbleeds: systematic review, subgroup analyses and standards for study design and reporting Brain, August 1, 2007; 130(8): 1988 - 2003. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 V. Palumbo, J. M. Boulanger, M. D. Hill, D. Inzitari, A. M. Buchan, and on behalf of the CASES Investigators Leukoaraiosis and intracerebral hemorrhage after thrombolysis in acute stroke Neurology, March 27, 2007; 68(13): 1020 - 1024. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
