Do cerebral microbleeds increase the risk of intracerebral hemorrhage after thrombolysis for acute ischemic stroke?


  • Conflicts of interest: None declared.

Correspondence: David J. Werring, Reader in Neurology, National Hospital for Neurology and Neurosurgery, Box 6, Queen Square, London WC1N 3BG, UK.


Dear Editor,

We read with interest the systematic review and meta-analysis by Shoamanesh and colleagues demonstrating an increased risk of post-thrombolysis symptomatic ICH in ischemic stroke patients with cerebral microbleeds (CMBs) [1]. We also recently performed a systematic review and meta-analysis on this topic [2, 3]. Although we used a slightly different meta-analytic method (fixed-effects as opposed to random-effects model) and calculated the pooled risk ratios (RR) instead of odds ratio (OR), our conclusions were similar: in nearly 800 thrombolysed patients with acute ischemic stroke the pooled RR of symptomatic ICH in patients with versus without CMBs on pre-treatment MRI scan, was 1·90 (95% CI 0·92 to 3·93; P = 0·082) [3], compared to an OR of 1·98 (95% CI 0·90 to 4·35; P = 0·09) in the study by Shoamanesh et al. [1]. Since data were collected longitudinally in all of the studies and the sequence of events was known, we presented RR instead of OR. Moreover, RR have been suggested to be most aligned to the way clinicians think about adverse risks of treatments [4].

We note that Shoamanesh et al. found a borderline significant increase in the odds of symptomatic ICH when they excluded one study because of loss to follow-up [1], but this limitation is unlikely to have affected the reporting of acute ICH in the first 36 hours, so this sensitivity analysis must be interpreted with caution.

Despite the methodological limitations of included studies and small total amount of data available (for a systematic critical appraisal of the studies see [3]) these analyses all highlight a potentially clinically relevant increased hazard for thrombolysis in individuals with CMBs. They reinforce the need for more robust data from a large well-designed and adequately powered prospective multicentre study (Table 1). Until definitive data are available, many important questions remain. Is there a threshold number of CMBs that shifts the risk–benefit ratio towards withholding thrombolysis? Although Shoamanesh et al. provide some data suggesting that CMB number might be important in predicting risk [1], we currently have no data on the importance of CMB location: is the ICH risk highest in those patients with multiple lobar CMBs indicative of cerebral amyloid angiopathy, which has been suggested to be particular hazard for post-thrombolysis ICHs?[5-7] And if CMBs are relevant, how does this help us understand the fundamental pathophysiology of thrombolysis-related ICH? Currently, detecting CMBs should not prevent thrombolytic treatment based on present evidence. Indeed, many acute stroke patients do not undergo MRI. If CMBs do prove to be a powerful predictor of ICH risk, this may strengthen the argument for the more widespread use of MRI in the assessment of acute stroke, with major practical implications for stroke service infrastructure.

Table 1. Recommendations for a study to investigate whether the presence, number and distribution of cerebral microbleeds (CMBs) on pre-thrombolysis MRI scans is associated with an increased risk of symptomatic intracerebral hemorrhage (ICH)
Study design
■ Prospective, multicentre, observational study
Study population
■ Clearly defined clinical characteristics and selection criteria
■ Representative of the population of patients with acute ischemic stroke
(consecutive cases in centers using MRI routinely for acute stroke assessment)
■ Adequate sample size to detect an effect of CMBs based on power calculations
(large collaborative multicentre studies)
■ Report of number and characteristics of patients excluded
Detection and rating of CMBs
■ Standardized MRI parameters for T2*-GRE or susceptibility-weighted imaging:
● field strength
● echo time
● slice thickness
● slice gap
■ Clear definition of CMBs and mimics
■ Use of a standardized CMB rating instrument with clearly defined anatomical regions
■ Rating instrument demonstrated to have good inter- and intra-rater reliability
● Trained observers (ideally a single observer for all analysis in a study)
■ Classification of CMB distribution (deep, lobar, mixed) and number
Definition of outcome
■ Post-thrombolysis ICH definition criteria and methods and timing of assessment clearly defined
■ Clinically relevant standardized definition of post-thrombolysis ICH
(e.g. associated with significant clinical deterioration)
■ Secondary outcomes could also be mortality and functional independence at 3 months
■ Results adjusted for confounding from other baseline risk factors known to be associated with thrombolysis-related ICH and CMBs (including age, leukoaraiosis, hyperglycemia, hypertension, prior use of antiplatelet and anticoagulant medications, recurrent stroke etc.)
■ Results presented according to number and anatomical distribution of CMBs