Clinical importance of microbleeds in patients receiving IV thrombolysis

The rate of intracerebral hemorrhage (ICH) following tissue plasminogen activator (tPA) therapy for acute ischemic stroke could be reduced if patients who are at higher risk of developing ICH could be rapidly identified and excluded from treatment.^1–4 This may be of particular importance if the time window for the use of tPA is extended beyond 3 hours, as the relative benefit from tPA decreases with time.

Gradient echo (GRE) imaging can detect abnormal accumulations of deoxygenated hemoglobin or other blood breakdown products in the brain that are indicative of prior bleeding. Small GRE positive lesions have been referred to as microbleeds (MBs).⁵ MBs are visualized as signal loss (hypointense lesions) due to the T2*-shortening magnetic susceptibility effect of hemosiderin. These lesions are thought to result from vascular injury caused by disorders such as chronic hypertension and amyloid angiopathy. Few studies have evaluated if presence of MBs on GRE imaging is a risk factor for subsequent hemorrhagic complications.^6–8 The results of these studies are conflicting, with some studies suggesting that the presence of MBs is a risk factor and others indicating that it is not. If MBs are an important risk factor for symptomatic ICH, then a substantial number of patients might be excluded from receiving thrombolytic therapy, because these lesions are present in about 15% of patients with acute ischemic stroke. We sought to evaluate, using a prospective cohort of consecutive tPA-treated patients, if MBs are a risk factor for the development of symptomatic or asymptomatic ICH.

Methods.

Subjects were enrolled at eight university hospitals in the United States, Canada, and Belgium as part of the Diffusion-weighted imaging Evaluation For Understanding Stroke Evolution (DEFUSE) study. This study is a multi-center, open-label, pilot study of IV tPA therapy administered to selected ischemic stroke patients within 3 to 6 hours after symptom onset. The local Institutional Review Board at each site approved the study and consent was obtained from each patient or an appropriate family member. Patients were eligible to participate if they were >18 years old, had a clinical diagnosis of ischemic stroke with an NIHSSS > 5, and could be treated with IV tPA injection within 3 to 6 hours after symptom onset. Patients were excluded if they were comatose or severely obtunded, had rapidly improving symptoms, had a history of stroke within the last 6 weeks, had a premorbid Rankin score of 3 or higher, had a seizure at symptoms onset, had previously known intracranial hemorrhage, or had evidence of acute hemorrhage or clearly identifiable hypodensity >1/3 of the middle cerebral artery territory on baseline noncontrast brain CT. Eligible patients who agreed to participate underwent MRI of the brain following their baseline CT scan. Patients were included irrespective of the presence of MBs on the baseline MRI scan. All patients were treated with 0.9 mg/kg of IV tPA (10% bolus over 1 minute, followed by continuous infusion of the remaining dose over 60 minutes) as quickly as possible following their initial MRI scan, but no later than 6 hours from the onset of their stroke symptoms. Repeat MRI scans were obtained 3 to 6 hours after initiation of thrombolytic therapy and at day 30. In case of neurologic deterioration during the hospital stay an additional CT or MRI scan was obtained to evaluate for hemorrhagic complications. Neurologic deficits were evaluated with the NIH Stroke Scale (NIHSS) before tPA therapy, 3 to 6 hours after tPA, and at day 30. Clinical characteristics including age at onset, sex, time to treatment, history of hypertension, diabetes, dyslipidemia, and smoking were evaluated.

MRI scans were obtained on 1.5 T scanners, all equipped with high-performance gradient systems and capable of echoplanar imaging. The baseline MRI and follow-up MRI scans included the following sequences: GRE imaging, diffusion weighted imaging, perfusion weighted imaging, MR angiography, conventional T1-weighted images. MRI scans at day 30 also included a fluid-attenuated inversion-recovery (FLAIR) sequence. The GRE imaging was obtained in the axial plane with the following parameters: 5 mm slice thickness, 2.5 mm interslice gap, field of view 240 mm, repetition time (TR) 450 to 800 msec, echo time (TE) 14 to 47 msec, 60 degree flip angle, and 256 × 256 matrix.

MBs were defined as focal homogenous hypointense areas with a diameter of up to 5 mm on GRE imaging. Signal loss secondary to globus pallidus calcifications or thrombus in a cerebral artery was excluded.

As defined by the DEFUSE protocol, an urgent CT scan was required to look for hemorrhagic complications if a patient was noted to have a worsening of two or more points on the NIHSS. In addition, all follow-up MRI scans were evaluated for evidence of asymptomatic hemorrhages. According to prespecified criteria,² both symptomatic and asymptomatic hemorrhages were classified into four categories based on their appearance on CT or MRI: hemorrhagic infarction (HI)-1, HI-2, parenchymal hematoma (PH)-1, and PH-2. HI-1 was defined as a small area of hemorrhage along the margin of the infarct and HI-2 as a more confluent hemorrhagic area within the infarct, but without space-occupying effect. PH-1 was defined as a hematoma smaller than 30% of the infarct with mild space-occupying effect and PH-2 as a hematoma exceeding 30% of the infarct with significant space-occupying effect. This assessment was based on a blinded reading of the scans by a senior neuroradiologist (M.M.). Hemorrhages were defined as minor symptomatic hemorrhages if they were associated with a worsening of two or three points on the NIHSS within 36 hours of tPA administration and major symptomatic hemorrhages if associated with a worsening of four or more points within 36 hours of tPA administration.

Two independent stroke neurologists (W.K. and V.T.) assessed the number and location of all MBs on the baseline GRE imaging. These individuals were blinded to neurologic status, subsequent imaging studies, and clinical characteristics. In case of disagreement between observers, the final determination was based on an additional blinded reading of the scan by a senior neuroradiologist (M.M.).

Results.

Between April 2001 and January 2005, 72 patients were enrolled in the DEFUSE study. Two patients were excluded from this analysis because the quality of their baseline GRE scans was poor. MRI scans were obtained 3 to 6 hours after tPA administration in 63 of 70 remaining patients (7 patients had the early follow-up scan obtained beyond the 3 to 6 hour window). Day 30 MRI scans were obtained in 57 patients. Five patients refused the MRI at 30 days and 8 patients died within the 30-day follow-up period.

Of all the subjects, 11 patients (15.7%) had MBs on their baseline GRE imaging. Eight patients had a single MB and three had multiple MBs (2, 3, and 6 MBs). Demographic and baseline characteristics were similar in patients with and without MBs, as well as concomitant medical conditions (table 1).

Overall, 7 patients (10% of all the subjects) developed symptomatic hemorrhages. Of these, 5 were categorized as major (total major hemorrhage rate, 7.1%) and 2 were categorized as minor symptomatic hemorrhages. All 5 major symptomatic hemorrhages were classified as PH-2 and both of the minor symptomatic hemorrhages as PH-1. No patient with a baseline MB had a symptomatic hemorrhage (0%) compared with 7 of 59 patients (11.9%) without baseline MBs. Twenty-nine of 59 patients without MBs (49.2%) had evidence of any type of hemorrhage on follow-up imaging compared to 3 of 11 patients with MBs (27.3%). There was no significant difference in frequency of any type of hemorrhage (p value = 0.18). In the three patients with baseline MBs who later developed asymptomatic hemorrhages, all the new hemorrhages occurred within the location of the acute infarct (a remote location from the baseline MBs). Of all 32 hemorrhagic lesions, 18 were visualized on the first follow-up (3–6 hour) MRI scan and 14 on the second (30 day) follow-up MRI scan. Table 2 summarizes the hemorrhagic complications, categorized by hemorrhage subtype, in patients with and without MBs.

Among patients with MBs at baseline, there were no differences in patient age or number or site of the MBs between patients who developed asymptomatic hemorrhages and those who did not. A new, asymptomatic MB was detected on a follow-up scan in only one patient. This patient did not have any baseline MBs and the new MB was detected on the first follow-up scan in an area remote from the infarct.

As the overall symptomatic hemorrhage rate was 10% with 0 observed among patients with baseline MBs, we calculated the probability that symptomatic hemorrhage rate was greater than 10% for patients with baseline MBs, with a beta-binomial distribution using a non-informative prior. Based on our data, there is an 88% probability that symptomatic hemorrhage rate is less than 10% in patients with baseline MBs.

In eight patients both readers agreed on the presence and location of all MBs. Of the four patients for whom the readers disagreed, three were determined to have MBs by the adjudicator. The inter-rater reliability for detection of MBs had a kappa coefficient of 0.77 (95% CI: 0.55 to 0.99).

Discussion.

This study demonstrates that the presence of MBs on GRE imaging does not appear to be a major risk factor for hemorrhagic complications following IV tPA administered between 3 and 6 hours after stroke onset. No significant difference in the occurrence of symptomatic or asymptomatic hemorrhages was detected between patients with pre-existing MBs compared with patients without MBs. All hemorrhagic complications occurred in the region of the acute infarct and not at the site of pre-existing MBs, further strengthening the notion that pre-existing MBs are not a significant risk factor for hemorrhagic complications after tPA. We believe that this is the first prospective study of consecutively enrolled patients investigating the relationship between pre-treatment existence of MBs and hemorrhagic complications after IV tPA. Our results support the view that the presence of a small number of MBs on baseline GRE imaging should not exclude patients with acute stroke from receiving tPA therapy.

One group reported that hemorrhagic complications are more frequent in patients with acute stroke with MBs than in patients without MBs.⁶ However, the presence of MBs was determined on a post-treatment MRI in some patients in this study, which leaves open the possibility that the MBs were not pre-existent, but rather the result of the treatment. Moreover, the majority of the patients in their study were treated with antithrombotic agents, rather than IV tPA, and therefore the results do not directly address the risk of hemorrhagic complications following tPA. Subsequently, the same group investigated the relationship between pre-existing MBs on GRE and hemorrhagic complications in patients who were treated with IV tPA within 7 hours of stroke onset.⁷ The main limitation of this study is the retrospective design and small number of subjects (44 patients) included. Similar to the results of our study, they found that pre-existing MBs are not associated with an increased risk of hemorrhagic complications after tPA. A different group studied the significance of MBs on baseline GRE imaging in 41 patients receiving intra-arterial thrombolytic therapy.⁸ They concluded that pretreatment existence of MBs may be a risk factor for hemorrhagic complications after intra-arterial thrombolysis, however, this was based on only one patient with MBs who had a hemorrhagic complication.

The frequency of MBs on baseline GRE imaging in our study was 15.7%. This rate is similar to two previous studies of the incidence of MBs in acute ischemic stroke, which reported frequencies of 18.2% and 12%.^7,8 In patients with ICH, the frequency of MBs has been reported to exceed 50%, while the frequency of MBs is less than 10% in asymptomatic patients.^9–12 The kappa statistic for detection of MBs on GRE imaging between our two observers was 0.77, indicating a high level of agreement. To our knowledge, no previous studies have evaluated inter-rater reliability for identification of MBs.

Of all the subjects, 5 (7.1%) had major symptomatic intracranial hemorrhages. This is comparable to the 5.2% symptomatic hemorrhagic complication rate reported in a meta-analysis of series of patients treated with IV tPA administered within 3 hours of stroke onset.¹³ Hemorrhages classified as PH-2 were seen in 7.1% (5 patients) of all the subjects. This rate is similar to the rates reported from the pooled analysis of the data from the National Institute of Neurologic Disorders and Stroke, ECASS, and ATLANTIS trials.¹⁴ The pooled analysis reported a frequency of parenchymal hematoma (PH-2) of 5.9% in patients treated with IV tPA between 3 and 4.5 hours of symptom onset and of 6.9% in those between 4.5 to 6 hours of symptom onset.

The frequency of asymptomatic hemorrhages was considerably higher in our study than in previous reports. There are two explanations for this discrepancy. First, GRE imaging used in this study to assess for the presence of asymptomatic hemorrhages is considerably more sensitive than CT, which was used in previous studies.^15,16 Secondly, GRE imaging at day 30 was obtained in addition to a 3- to 6-hour follow-up scan whereas previous studies have obtained only one follow-up CT scan in the first 48 hours after treatment. Of the 24 asymptomatic hemorrhages identified in our study, 12 were detected on the first follow-up MRI and an additional 12 were identified on the day 30 MRI.

Our study has limitations. The sample size is relatively small, yet we were able to exclude with a high degree of certainly that the symptomatic ICH rate in patients with MBs is greater than 10%. Only a few of our patients with MBs developed hemorrhages (all asymptomatic), therefore, we could not clarify the influence of number and site of MBs on the risk of ICH. The relationship between the symptomatic and asymptomatic hemorrhages detected in this study and tPA therapy is not clear as there was no control group that did not receive tPA. GRE imaging at day 30 was not performed in all patients because of patient refusals or deaths. It is unclear whether our results, obtained in the 3- to 6-hour treatment window, can be extrapolated to the 0- to 3-hour time window.

Appendix

DEFUSE investigators are as follows: Stanford University—Gregory Albers, MD, Roland Bammer, PhD, Stephanie Kemp, BS, Maarten Lansberg, MD, PhD, Wataru Kakuda, MD, Michael Marks, MD, Michael Moseley, PhD, Christine Wijman, MD. Beth Israel Deaconess—Gottfried Schlaug, MD, Cheryl Hogan. University of Alberta—Ashfaq Shuaib, MD, Frederika Herbert, RN. University Hospitals of Leuven—Vincent Thijs, MD, PhD, Ingeborg Wauters, RN. University of Pittsburgh—Lawrence Wechsler, MD, Sharon DeCesare. University of Utah—Elaine Skalabrin, MD, Janet Woodruff, RN. Wayne State University—William Coplin, MD, Flicia Mada, RN.