Elsevier

The Lancet Neurology

Volume 20, Issue 4, April 2021, Pages 294-303
The Lancet Neurology

Articles
Development of imaging-based risk scores for prediction of intracranial haemorrhage and ischaemic stroke in patients taking antithrombotic therapy after ischaemic stroke or transient ischaemic attack: a pooled analysis of individual patient data from cohort studies

https://doi.org/10.1016/S1474-4422(21)00024-7Get rights and content

Summary

Background

Balancing the risks of recurrent ischaemic stroke and intracranial haemorrhage is important for patients treated with antithrombotic therapy after ischaemic stroke or transient ischaemic attack. However, existing predictive models offer insufficient performance, particularly for assessing the risk of intracranial haemorrhage. We aimed to develop new risk scores incorporating clinical variables and cerebral microbleeds, an MRI biomarker of intracranial haemorrhage and ischaemic stroke risk.

Methods

We did a pooled analysis of individual-patient data from the Microbleeds International Collaborative Network (MICON), which includes 38 hospital-based prospective cohort studies from 18 countries. All studies recruited participants with previous ischaemic stroke or transient ischaemic attack, acquired baseline MRI allowing quantification of cerebral microbleeds, and followed-up participants for ischaemic stroke and intracranial haemorrhage. Participants not taking antithrombotic drugs were excluded. We developed Cox regression models to predict the 5-year risks of intracranial haemorrhage and ischaemic stroke, selecting candidate predictors on biological relevance and simplifying models using backward elimination. We derived integer risk scores for clinical use. We assessed model performance in internal validation, adjusted for optimism using bootstrapping. The study is registered on PROSPERO, CRD42016036602.

Findings

The included studies recruited participants between Aug 28, 2001, and Feb 4, 2018. 15 766 participants had follow-up for intracranial haemorrhage, and 15 784 for ischaemic stroke. Over a median follow-up of 2 years, 184 intracranial haemorrhages and 1048 ischaemic strokes were reported. The risk models we developed included cerebral microbleed burden and simple clinical variables. Optimism-adjusted c indices were 0·73 (95% CI 0·69–0·77) with a calibration slope of 0·94 (0·81–1·06) for the intracranial haemorrhage model and 0·63 (0·62–0·65) with a calibration slope of 0·97 (0·87–1·07) for the ischaemic stroke model. There was good agreement between predicted and observed risk for both models.

Interpretation

The MICON risk scores, incorporating clinical variables and cerebral microbleeds, offer predictive value for the long-term risks of intracranial haemorrhage and ischaemic stroke in patients prescribed antithrombotic therapy for secondary stroke prevention; external validation is warranted.

Funding

British Heart Foundation and Stroke Association.

Introduction

Antithrombotic therapy is a key component of secondary prevention after ischaemic stroke or transient ischaemic attack. In patients without atrial fibrillation, antiplatelet treatment reduces overall stroke risk by a quarter,1 whereas oral anticoagulation in patients with atrial fibrillation reduces this risk by two-thirds.2, 3 Although antithrombotic treatment increases the risk of intracranial haemorrhage (by about a quarter for antiplatelets, half for direct oral anticoagulants, and two-times for vitamin K antagonists),1, 2, 3 the substantially lower absolute incidence of intracranial haemorrhage overall means that antithrombotic treatment is recommended for most patients. However, deciding on appropriate antithrombotic therapy for a given patient can be challenging, especially in those with additional risk factors for bleeding, such as uncontrolled hypertension, previous intracerebral haemorrhage, or severe cerebral small vessel disease. Ideally, this decision would be based on an individualised assessment of the risks of ischaemic stroke and intracranial haemorrhage. To this end, risk scores for ischaemic stroke and major bleeding have been developed, mainly in patients with atrial fibrillation. Although these scores show reasonable discrimination for ischaemic stroke4, 5 and all-cause major bleeding,5, 6 studies validating existing bleeding risk scores in predicting intracranial haemorrhage have shown poor to moderate performance, with c indices between 0·50 and 0·62 in patients who received anticoagulants7, 8 and 0·58–0·65 in patients who received antiplatelet drugs.8, 9

Most risk scores for ischaemic stroke and intracranial haemorrhage only include clinical variables, although scores using serum biomarkers have been developed, which might offer improved performance.10, 11, 12 However, the role of MRI biomarkers for cerebrovascular disease (increasingly obtained as part of standard stroke care) in improving risk prediction remains uncertain. Cerebral microbleeds are a MRI biomarker of vascular fragility, associated with hypertensive microangiopathy (also known as arteriolosclerosis or deep perforator arteriopathy) and cerebral amyloid angiopathy, the two cerebral small vessel diseases that cause most spontaneous intracerebral haemorrhage.13 Accordingly, the potential of cerebral microbleeds to predict intracranial haemorrhage has attracted particular interest. In a prospective observational study, the addition of cerebral microbleeds to the HASBLED bleeding risk score improved the c index for intracranial haemorrhage from 0·41 to 0·66,14 and a large individual patient data meta-analysis confirmed a strong association between cerebral microbleeds and intracranial haemorrhage in patients with previous ischaemic stroke or transient ischaemic attack.15 This study15 also found that cerebral microbleeds are associated with ischaemic stroke risk, with a higher absolute risk of ischaemic stroke than intracranial haemorrhage across all levels of cerebral microbleed burden investigated.

Given these findings, we aimed to establish the added predictive value of cerebral microbleeds for intracranial haemorrhage and ischaemic stroke, by using the same large international dataset to develop risk models based on cerebral microbleed burden and simple clinical variables, and to compare these with models using clinical variables alone. From our models, we aimed to derive simple risk scores that could be easily used for risk stratification in clinical practice. We investigated whether the resulting scores identified a group of patients at similar or higher predicted risk of intracranial haemorrhage than ischaemic stroke and whether our new intracranial haemorrhage risk score performed better than existing methods.

Section snippets

Study design and participants

We used pooled individual patient data from the Microbleeds International Collaborative Network (MICON) of prospective observational studies, for which the full methodology and composition has been published.15 MICON includes 38 cohorts from 18 countries in North America, Europe, the Middle East, Asia, and Australasia, collectively including 20 322 participants with previous ischaemic stroke or transient ischaemic attack, baseline MRI—including blood-sensitive paramagnetic sequences to detect

Results

Of the 38 studies and 20 322 participants in the collaboration, one study comprising 3355 (16·5%) participants that collected follow-up for a composite of any stroke outcome only was excluded (figure 1). From the remaining 37 cohorts, 979 (4·8%) participants were excluded because they did not receive antithrombotic medication, and 204 (1·0%) participants were excluded because they were not followed-up for both intracranial haemorrhage and ischaemic stroke. 15 784 (77·7%) participants were

Discussion

Our most important result is the description of the novel MICON-intracranial haemorrhage (MICON-ICH) risk score—which includes clinical variables and MRI-detected cerebral microbleeds—to predict intracranial haemorrhage in patients taking antithrombotic therapy after ischaemic stroke or transient ischaemic attack. The addition of cerebral microbleeds to a score based on clinical variables alone substantially improved performance, and a direct comparison with three existing bleeding risk scores

Data sharing

Requests for access to anonymised study data for legitimate academic purposes should be directed to the corresponding author. Approval by the study steering committee and the principal investigator of each cohort in the study will be required before data can be shared.

References (29)

  • Collaborative meta-analysis of randomised trials of antiplatelet therapy for prevention of death, myocardial infarction, and stroke in high risk patients

    BMJ

    (2002)
  • RG Hart et al.

    Meta-analysis: antithrombotic therapy to prevent stroke in patients who have nonvalvular atrial fibrillation

    Ann Intern Med

    (2007)
  • W Zhu et al.

    Meta-analysis of ATRIA versus CHA2DS2-VASc for predicting stroke and thromboembolism in patients with atrial fibrillation for predicting stroke and thromboembolism in patients with atrial fibrillation

    Int J Cardiol

    (2017)
  • AD Borre et al.

    Predicting thromboembolic and bleeding event risk in patients with non-valvular atrial fibrillation: a systematic review

    Thromb Haemost

    (2018)
  • Cited by (39)

    View all citing articles on Scopus
    View full text