The Clinical Relevance of Microbleeds in Stroke study (CROMIS-2): rationale,design,and methods

Abstract

Background and rationale

The increasing use of oral anticoagulants, mostly to prevent ischemic stroke due to atrial fibrillation in an ageing population, has been associated with a fivefold increased incidence of oral anticoagulant-associated intracerebral hemorrhage: a rare, serious, and unpredictable complication. We hypothesize that cerebral microbleeds and other markers of cerebral small vessel disease on magnetic resonance imaging, and genetic polymorphisms (e.g. influencing cerebral small vessel integrity or anticoagulation stability), are associated with an increased risk of oral anticoagulant-associated intracerebral hemorrhage, with potential to improve risk prediction.

Aims

(1) To determine the incidence, clinical, radiological, and genetic associations of oral anticoagulant-associated intracerebral hemorrhage in a prospective, multicentre cohort study of patients with atrial fibrillation-related ischemic stroke or transient ischemic attack started on oral anticoagulants; (2) To investigate characteristics of oral anticoagulant-associated intracerebral hemorrhage compared with non-oral anticoagulants related intracerebral hemorrhage in a prospective study.

Design and methods

Study 1: Prospective, multicentre, inception cohort study of 1425 adults started on oral anticoagulants (including vitamin K antagonists and the nonvitamin K oral anticoagulants) after recent ischemic stroke and concurrent atrial fibrillation. Participants will have standardized brain magnetic resonance imaging (including a T2*-weighted gradient-recalled echo sequence) and DNA sample collection at baseline, with two-year follow-up by postal questionnaire and medical records surveillance for symptomatic intracranial hemorrhage, other serious vascular events, and death. We will compare the rates of symptomatic intracranial hemorrhage (primary outcome; subclassified as intracerebral, subdural, extradural, intraventricular), other vascular events, and death (secondary outcomes) in participants with one or more cerebral microbleeds to the rates in those without cerebral microbleeds. Study 2: Prospective case-control study of oral anticoagulant-associated intracerebral hemorrhage compared with non-oral anticoagulant-associated intracerebral hemorrhage to investigate genetic, clinical, and radiological associations with oral anticoagulant-associated intracerebral hemorrhage. In participants with intracerebral hemorrhage (including at least 300 with oral anticoagulant-associated intracerebral hemorrhage), we will collect a DNA sample, standardized clinical data and routine brain imaging (magnetic resonance imaging or computed tomography), and information on functional outcome.

Expected outcomes

We will identify the factors associated with increased intracranial hemorrhage risk after oral anticoagulants for secondary prevention after ischemic stroke due to atrial fibrillation. We will determine clinical, radiological and genetic factors, and clinical outcomes associated with oral anticoagulant-associated intracerebral hemorrhage.

Keywords

atrial fibrillation cerebral microbleeds cerebral small vessel disease intracerebral hemorrhage oral anticoagulants

Introduction and rationale

Oral anticoagulants [OACs, including warfarin and nonvitamin K OACs (NOACS)] are effective in reducing the relative risk of future ischemic stroke in people with atrial fibrillation (AF) by over 60% (1). The most feared of all bleeding complications of OAC is intracranial hemorrhage [bleeding within the skull, of which ∼70% are intracerebral hemorrhage (ICH), i.e. bleeding within the brain substance] (2). OACs are used increasingly in an ageing population; there has been a fivefold increase in the incidence of OAC-associated ICH (OAC-ICH) in population-based studies (3 –5), so that it now accounts for about 15% of all ICH (3).

People with ischemic stroke due to AF are at high risk of future ischemic stroke, but paradoxically, they may also have the highest risk of ICH (1). Concern over bleeding risks may account for low rates of anticoagulation for AF in recent studies (6,7). A key research goal in stroke medicine is to improve judgment of the risks and benefits of anticoagulation to increase the use of OAC and allow personalized, safe, and effective stroke prevention treatment.

Magnetic resonance imaging (MRI) can provide detailed information about cerebral small vessel disease, a strong candidate risk factor for OAC-ICH. Potential markers of small vessel disease include subcortical infarcts (including lacunes), white matter hyperintensities of presumed vascular origin (sometimes termed leukoaraiosis), and cerebral microbleeds (CMBs) (8). CMBs - small rounded hypointense lesions seen on blood-sensitive MRI sequences - are hypothesized to be a marker of bleeding-prone small vessel diseases, with high future risk of ICH (9). Genetic variants related to cerebral small vessel integrity or anticoagulant metabolism could also influence the likelihood of ICH during anticoagulation. The Clinical Relevance of Microbleeds in Stroke (CROMIS-2) study is a multicentre prospective UK study designed to determine whether MRI markers of small vessel diseases, and genetic factors, are related to the risk of intracranial hemorrhage (including ICH), other vascular events, or death during oral anticoagulation (OAC) after ischemic stroke associated with AF.

AF and OAC

AF causes up to about 20% of all ischemic strokes, with a lifetime risk of one in four men and women over the age of 40 (10,11). It affects about 12% of people over the age of 75 and is more common in ageing populations (11). Untreated, AF increases the overall risk of stroke fivefold, with the highest risk in elderly patients with previous stroke or transient ischemic attack (TIA) (11). OAC reduces ischemic stroke risk by about 65% in patients with AF, but this benefit has to be balanced against an increased risk of intracranial hemorrhage, including ICH. ICH is the most feared complication of OAC, causing death or severe disability in up to 75% of patients (12). Although RCTs suggest that the rate of symptomatic ICH in anticoagulated patients is low (particularly with the use of NOACs), RCTs may not be generalizable to populations given these drugs in routine clinical practice, and are not designed or powered to investigate predictors of rare adverse events (13,14). A recent observational inception cohort study of patients treated with OAC (of whom a quarter had a previous history of stroke) reported a 2.5% (95% CI 1.1-4.7%) risk of ICH at one-year (15).

Because OAC-related ICH is so often fatal or disabling, even a small increase in the absolute risk of ICH of only 1-2% per year could modify the net benefit of OAC treatment (16). Clinical risk factors for OAC-related bleeding (including ICH) include increasing age, previous ischemic stroke, and intensity of OAC. However, these are not sufficiently reliable for clinical decision making (1), and overlap strongly with the risk factors for ischemic stroke, so it remains extremely difficult to balance the risks and benefits of OAC in practice.

Cerebral small vessel disease, CMBs, and ICH risk

Neuroimaging can identify biomarkers of the presence and severity of age-related cerebral small vessel diseases, including cerebral amyloid angiopathy (CAA) and disease of the deep perforating arteries (often termed hypertensive arteriopathy), which may predispose to ICH in elderly patients treated with OACs (17). CMBs are a key MRI marker of small vessel disease and potential predictor of future stroke including ICH (14,18). CMBs are small, rounded, hypointense lesions on blood-sensitive MRI sequences, including gradient-recalled echo T2*-weighted MRI (T2*-GRE) and susceptibility-weighted imaging (SWI) (9,19). Radiologically defined CMBs correlate pathologically with small collections of blood-breakdown products (in particular, hemosiderin-laded macrophages) adjacent to small vessels mainly affected by hypertensive arteriopathy (including lipohyalinosis and arteriolosclerosis) or CAA (20 –22). Hence, CMBs are a promising new marker for bleeding-prone arterial pathology affecting cerebral small vessels (19). CMB distribution in the brain parallels the underlying small vessel disease: strictly lobar (cortical-subcortical) CMBs are characteristic of CAA, with high specificity for CAA when used according to the Boston criteria (23), while deep CMBs likely reflect hypertensive arteriopathy (9,17,19). A mixed pattern of CMBs may reflect either severe widespread hypertensive arteriopathy or a combination of CAA and hypertensive arteriopathy.

CMBs are common in the populations likely to be exposed to OAC (24): They are found in about 30% of patients with ischemic stroke, up to 40% of the healthy population over the age of 80 (25,26), and 60% of patients with ICH (27). CMBs are associated with an increased risk of ICH in patients with previous ischemic stroke or ICH (28). A recent systematic review of case-case comparisons of 1460 patients with ICH and 3817 patients with ischemic stroke or TIA showed that CMBs are more common in warfarin-related ICH than nonwarfarin-related ICH (OR: 2.7; 95% CI: 1.6-4.4; P < 0.001) (18).

There are only limited prospective studies of CMBs, including patients with ischemic stroke (29 –31), ICH (32,33), or healthy elderly individuals (34). The largest prospective cohort of CMBs and antithrombotic use included 908 patients with ischemic stroke, treated with a single antithrombotic agent (93% with aspirin) (30). During a mean follow-up of 26 months, the risk of ICH was 0.6% in patients with no CMB, 1.9% in patients with one CMB, 4.6% in patients with two to four CMBs, and 7.6% in patients with five or more CMBs (P < 0.001) (30). Overall, 15/908 (1.7%) patients suffered ICH at an average follow-up time of 27 months; of patients with CMBs, 4.4% developed ICH, while 0.6% of patients without CMBs developed ICH (30), giving a relative risk of CMBs on ICH of 7.3 (4.4/0.6). However, the findings from this Asian cohort may not be generalizable to other populations: CMBs were detected in 27.8% of the patients, most commonly in deep brain regions (e.g. basal ganglia and thalamus), suggesting that hypertensive arteriopathy was the main small vessel pathology. A recent systematic review and meta-analysis pooled data from 10 cohorts including 3067 patients with ischemic stroke or TIA, and found that (35) CMBs were associated with an increased risk of recurrent stroke (OR: 2.25; 95% CI: 1.70-2.98; P < 0.0001), but especially symptomatic ICH (OR: 8.52; 95% CI: 4.23-17.18; P = 0.007). However, in Caucasians, the OR for CMB presence on ICH risk was 3.9 (95% CI 0.91-16.4), while in Asian cohorts, it was 10.4 (95% CI 4.6-23.7) (35). The systematic review found no large prospective studies of CMBs in ischemic stroke Caucasian cohorts treated with anticoagulants for AF.

CMBs are being found more often with the increasing use of MRI in stroke patients (22). The clinical relevance of CMBs for ICH risk needs to be clarified to avoid patients being denied effective treatment due to concerns regarding CMBs; conversely, if patients at very high bleeding risk could be identified, they could have individualized antithrombotic treatment or alternative ischemic stroke prevention methods (e.g. left atrial appendage occlusion devices) and be spared the devastating yet potentially avoidable hazard of ICH related to long-term anticoagulation.

Rationale for the research questions

Because OAC-ICH is associated with increased age, and often occurs with anticoagulation intensity within the therapeutic range (36), the mechanism may be related to individual patient factors, for example, age-related cerebral small vessel disease, such as CAA or disease of the deep perforating arteries (sometimes termed hypertensive arteriopathy) (37 –41). Moreover, some studies suggest that white matter abnormalities of presumed vascular origin - confluent deep white matter abnormality seen as low attenuation on computed tomography (CT) or hyperintensities on T2-weighted MRI – increase the risk of OAC-ICH (42,43). Because CMBs provide evidence of leakage of blood from pathologically fragile small vessels (44), we hypothesize that CMBs may be an independent predictor of OAC-ICH. There is also increasing interest in genetic factors that might predispose to OAC-ICH, with two plausible mechanisms: First, some genes may influence the sensitivity to OAC and stability of anticoagulation (45), and second, other genes may make cerebral small vessels more likely to bleed by affecting the integrity of the arterial wall (46).

Research questions

Study 1: CROMIS-2 AF

(1)

What is the incidence of OAC-associated intracranial bleeding (subclassified as intracerebral, subdural, extradural, intraventricular, and subarachnoid) in a multicentre, prospective, inception cohort study of patients with recent ischemic stroke anticoagulated for nonvalvular AF?

(2)

Are CMBs (or other markers of small vessel disease) associated with the future risk of symptomatic OAC-related intracranial hemorrhage (subclassified as intracerebral, subdural, extradural, intraventricular, and subarachnoid)?

(3)

Are CMBs (or other markers of small vessel disease) associated with an increased risk of other vascular events (including recurrent TIA and ischemic stroke) or death?

(4)

Can our findings be used to develop a risk model for OAC-related intracranial hemorrhage?

Study 2: CROMIS-2 ICH

(5)

Are genetic, clinical, or radiological risk factors associated with OAC-ICH or functional outcome after ICH?

(6)

What are the clinical and radiological characteristics of OAC-ICH (including those associated with warfarin or NOACS)?

Methods

Patient population, sample size, and study design

Study I (AF): observational inception cohort study (n ≥ 1425) of patients started on OAC (without prior anticoagulant use and timing dependant on treating physician) for presumed cardioembolic stroke due to nonvalvular AF with follow-up for the occurrence of ICH, other vascular event including ischemic stroke, for an average of two-years. Our main baseline exposures (risk factors of interest) are the presence of CMBs and other markers of small vessel disease on MRI, and genetic polymorphisms in candidate genes with potential functional relevance to ICH risk.

Study 2 (ICH): prospective study of patients admitted to participating centers with spontaneous (primary) ICH (with a minimum target of at least 300 OAC-ICH). We will collect detailed clinical and imaging data and DNA.

Table 1 summarizes the eligibility criteria for both studies.

Table 1

Patient population: inclusion and exclusion criteria

Inclusion criteria	Exclusion criteria
Study I: CROMIS-2 (AF)
1. Adult (≥18 years; no upper limit) patients with a clinical diagnosis of nonvalvular AF (verified by ECG, including paroxysmal AF) and intention to treat with oral anticoagulants (e.g. warfarin) 2. Previous ischemic stroke or TIA (in the last one-month) diagnosed by treating clinician 3. All patients must be able to have the CROMIS-2 MRI protocol (including T2*-GRE) before (or within one week) of starting oral anticoagulant.	1. Any MRI contraindications 2. Previous therapeutic use of oral anticoagulation 3. Definite contra-indication to oral anticoagulation 4. Serious head injury (resulting in loss of consciousness)
Study II: CROMIS-2 (ICH)
Adult (>18 years) patients treated at participating centers with confirmed spontaneous ICH (confirmed on CT or MRI scans) in the past one-month, with or without a history of anticoagulant use at the time of the ICH.	1. Known underlying structural cause for ICH (e.g. arteriovenous malformation, tumor, cavernoma, intracranial aneurysm, hemorrhagic transformation of an infarct) 2. Major head trauma (causing loss of consciousness and thought to be sufficient to have caused the ICH) in the 24 h prior to presentation to hospital

Sample size and recruitment of subjects

Study I (AF): We will recruit a total of 1425 patients from UK centers over 47 months. Sample size estimates are based on the incidence of ICH as this is the only intracranial hemorrhage subtype for which data on the influence of CMBs on risk is available. We expect that 20% of our cohort will have CMBs and that overall 2% will have an ICH over two-years. If we assume a conservative relative risk of 4 for CMBS on ICH risk (30), then we estimate the rate of ICH at two-year follow-up to be 5.0% in patients with CMBs, compared with 1.25% in patients without CMBs. This difference would be clinically important and could tip the risk-benefit judgment in favor of avoiding or reducing the intensity of OAC, or considering alternative treatments (e.g. left atrial appendage occlusion device, an antiplatelet agent, or no antithrombotic treatment) in patients with CMBs. To detect such a difference as statistically significant at the 5% level with 90% power would require 1425 patients. Although the best current evidence for the relative risk associated with CMBs in Caucasian populations is 3.9 (35), we have calculated the power for a range of risk ratios, with all other assumptions kept the same (Fig. 1). We would still have 80% power to detect a statistically significant effect of CMBs on ICH risk if the true relative risk was as low as 3.3. Attrition could further reduce power, but we will still have 80% power even with attrition of 28% (based on a relative of risk of 4).

Fig. 1

Power of the study (without attrition) across a range of risk ratios, based on an overall event rate of 2% over two-year follow-up and 20% of patients having CMBs.

The anticipated ICH event rate of 2% over two-years taking into account attrition suggests that we will observe up to 30 intracranial hemorrhage events in total. The ‘rule of 10’ for developing risk models will allow us to include just three predictor variables (47), though more will be possible through use of modern regression techniques (48). A risk model based solely on CMBs would have a sensitivity of 50% and a specificity of 81% for predicting an ICH within two-years. A risk model based on more predictors should improve on these values. We expect to use existing summary AF prediction risk scores (incorporating multiple variables, e.g. CHADS2VASC, HAS-BLED) as single predictor variables to allow us to assess the additional value of including CMBs as a predictor.

Study 2 (ICH): We will recruit at least 300 patients with OAC-ICH, with an overall target of at least 1000 patients with ICH.

Brain imaging protocol

In Study I (AF), all patients will have a standardized MRI at baseline, as soon as possible after stroke or TIA attributed to AF, and within one- to two-weeks after starting OAC. The protocol includes axial T2, T2*-GRE (essential for detecting CMBS), diffusion-weighted imaging, coronal T1-weighted, and FLAIR MR sequences (Fig. 2). The T2* GRE sequence parameters are specified in Table 2. We will acquire additional 3D GRE SWI at as many sites as possible. Where no CMB-sensitive sequences have been obtained, patients will be withdrawn. Where a 3D GRE SWI but not a T2* GRE has been obtained, patients will be included in the primary analysis if no CMBs are detected. A planned sub-analysis will be undertaken including those patients with CMBs on 3D GRE SWI but no corresponding T2*-GRE given the higher sensitivity for CMBs of 3D GRE SWI compared with T2*-GRE (49).

Fig. 2

CROMIS-2 (AF) MRI protocol. (A) axial T2-weighted image; (B) axial T2*-weighted gradient echo image; (C) axial diffusion-weighted image; (D) coronal T1-weighted image; (E) coronal FLAIR image; (F) axial susceptibility-weighted image. The arrows in panel B are cerebral microbleeds in the right thalamus.

Table 2

CROMIS-2 (AF) imaging protocol: optimal and accepted limits of important MRI parameters for T2*-weighted gradient-recalled echo (T2*-GRE) sequences

Parameter	Optimal	Accepted limits
Slice thickness (ST)	3 mm	3–5 mm
Slice gap (SG)	Ideally no gap	Not more than 1 mm
Echo time (TE)	20–30 ms	10–40 ms

All imaging analysis for CMBs and other imaging markers of cerebrovascular disease will take place at the coordinating center by raters trained in using appropriate validated rating scales, blind to clinical details. CMB number and anatomical distribution will be rated using The Microbleed Anatomic Rating Scale (50). White matter hyperintensity of presumed vascular origin will be graded using standard validated rating scales or semi-quantitative methods. Other potential markers of small vessel disease (e.g. MRI-visible perivascular spaces, lacunes of presumed vascular origin, cortical superficial siderosis) will be rated. Regional atrophy (including medial temporal lobe atrophy) will be rated using validated rating scales or semi-quantitative methods using either coronal/volumetric T1-weighted or coronal FLAIR imaging.

In Study 2 (ICH), all brain imaging undertaken as part of standard clinical care will be collected and analyzed. Hematoma volume will be estimated on plain CT imaging using ABCD/2 or planimetric methods. Perihematomal edema will be measured using a validated planimetric method. We will analyze all available MRI on patients with ICH for markers of cerebral small vessel disease.

Genetic data

Two vials of blood in EDTA will be collected. DNA will be extracted and stored at the UCL Institute of Neurology, Queen Square. Genotyping will be undertaken in collaboration with the International Stroke Genetics Consortium (http://www.strokegenetics.org), subject to regulatory approvals and funding. We will request the use of publically available control genetic data, including the Wellcome 1956 cohort.

Prospective follow-up schedule

The follow-up schedule is shown in Table 3. Multiple ascertainment methods will be used to ensure high rates of complete follow up data. In CROMIS-2 (AF), patients will be followed up at 6, 12, and 24 months by postal questionnaires to patients and their general practitioners. Data collected will include INR monitoring, antithrombotic drug use, and stroke events (ischemic stroke and intracranial hemorrhages). In CROMIS-2 (ICH), patients will be followed up using similar postal questionnaires at six-months. All outcome events will be notified to the central study team by each participating hospital research team. The central study team will also be notified of all deaths and hospital attendances in study participants from the Health and Social Care Information Centre; this includes data on all admitted patient care episodes, outpatient appointments, and emergency attendances in England. We will request data access reports every six-months for the duration of the study.

Table 3

Summary of data to be collected at baseline and during scheduled follow-up assessments

Data to be collected	Baseline	6 months	12 months	24 months
Study I: CROMIS-2 (AF)
Alive and contacted to check at same address		X	X	X
Clinical and demographic information	X
Electrocardiography (confirmed AF or PAF)	X
Echocardiography	X
Blood tests (including LFTs)	X
CHADS2 score	X
HAS-BLED
NIHSS	X
Modified Rankin Score	X
Quality of Life measure (EQ-5D-5L)	X	X	X	X
MoCA	X		X*
IQCODE (short form)	X
MRI	X
INR values		X	X	X
Stroke event questionnaire		X	X	X
Study II: CROMIS-2 (ICH)
Clinical and demographic information	X
History of transient neurological episodes	X	X
ECG	X
Blood test	X
NIHSS	X
Modified Rankin Score	X
Quality of Life measure (EQ-5D-5L)	X	X
MoCA	X
IQCODE (short form)	X
Imaging (when available CT and/or MRI)	X	X
INR value (on admission if anticoagulation-related ICH)	X
Stroke Event Questionnaire		X

Data to be collected in a planned substudy at participating sites.

Using all available clinical and imaging information, all clinical stroke events will be adjudicated using standardized definitions (51) by a member of the central study team blinded to all baseline clinical and imaging data.

Outcomes

The primary outcome for Study 1 is symptomatic intracranial hemorrhage (including ICH), confirmed on brain imaging. Intracranial hemorrhage includes any bleeding within the skull, regardless of the site. We will record the incidence of different hemorrhage subtypes (subclassified as intracerebral, subdural, extradural, intraventricular, and subarachnoid). Secondary outcomes in study 1 include ischemic stroke, TIA, cardiac events (including myocardial infarction), death of any cause, major hemorrhagic events other than ICH, quality of life, and long-term physical disability. Similar outcomes will be collected for Study 2.

Statistical analyses

We will compare the rate of ICH between the CMB and CMB-free groups using the log-rank test and will investigate whether the number of CMBs is associated with the risk of ICH using Cox regression. Cox regression will be used to develop a risk prediction model for ICH using TRIPOD guidelines (52). Potential risk factors for the model will be prespecified in the statistical analysis plan based on published literature, and variable selection methods may be used to reduce the number of predictors in the risk model. We expect to include existing validated risk scores (including CHADS2VASC for ischemic stroke and HASBLED for serious bleeding) as single variables in the risk model. Penalized estimation, such as ridge or lasso (48), may be used to guard against overfitting. Cross-validation, used in conjunction with calibration slopes and the c-index, will be used to internally validate the model and assess calibration, discrimination, and predictive accuracy. Missing data, and the reasons for it, will be investigated. Imputation may be used if deemed necessary.

Study organization and funding

CROMIS-2 has been adopted by the National Institute of Health Clinical Research Network (NIHR). Recruitment commenced August 2011 and will continue until summer 2015. The sample size will be kept under review by the Steering Committee and may be revised in line with the best available data on CMB-associated relative risks. There are currently 80 sites open to recruitment across the UK and one other European centre (Maastricht). CROMIS-2 is funded by the British Heart Foundation and the Stroke Association. A study website accessible to the public has been set up providing up-to-date information for participants and researchers on study progress (www.ucl.ac.uk/cromis-2).

Summary

Cerebral small vessel disease is a plausible risk factor for intracranial hemorrhage (including ICH) related to OAC use. CMBs have emerged as a potential marker of bleeding-prone small vessel disease and future ICH risk, but high-quality prospective studies of CMBs and ICH risk on anticoagulation are not available. CROMIS-2 will provide prospective data on how CMBs (and other neuroimaging small vessel disease markers) may contribute to assessing the future risk of intracranial hemorrhage on OAC. This risk assessment may become even more important as new drugs will increase the use of OAC (14,53). CROMIS-2 will assess the feasibility of developing a risk model to identify those patients at higher risk of symptomatic intracranial hemorrhage.

Footnotes

Conflict of interest: None declared.

References

Camm

, Kirchhof

, Lip

. Guidelines for the management of atrial fibrillation: the Task Force for the Management of Atrial Fibrillation of the European Society of Cardiology (ESC). Europace 2010; 12: 1360–420.

Fang

, Go

, Chang

. Death and disability from warfarin-associated intracranial and extracranial hemorrhages. Am J Med 2007; 120: 700–5.

Flaherty

, Kissela

, Woo

. The increasing incidence of anticoagulant-associated intracerebral hemorrhage. Neurology 2007; 68: 116–21.

Lovelock

, Molyneux

, Rothwell

, Oxford Vascular

. Change in incidence and aetiology of intracerebral haemorrhage in Oxfordshire, UK, between 1981 and 2006: a population-based study. Lancet Neurol 2007; 6: 487–93.

Bejot

, Cordonnier

, Durier

, Aboa-Eboule

, Rouaud

, Giroud

. Intracerebral haemorrhage profiles are changing: results from the Dijon population-based study. Brain 2013; 136(Pt 2): 658–64.

Fisher

. MRI screening for chronic anticoagulation in atrial fibrillation. Front Neurol 2013; 4: 137.

Kirchhof

, Nabauer

, Gerth

. Impact of the type of centre on management of AF patients: surprising evidence for differences in antithrombotic therapy decisions. Thromb Haemost 2011; 105: 1010–23.

Wardlaw

, Smith

, Biessels

. Neuroimaging standards for research into small vessel disease and its contribution to ageing and neurodegeneration. Lancet Neurol 2013; 12: 822–38.

Charidimou

, Werring

. Cerebral microbleeds: detection, mechanisms and clinical challenges. Future Neurol 2011; 6: 587–611.

10.

Lloyd-Jones

, Wang

, Leip

. Lifetime risk for development of atrial fibrillation: the Framingham Heart Study. Circulation 2004; 110: 1042–6.

11.

Lip

, Tse

, Lane

. Atrial fibrillation. Lancet 2012; 379: 648–61.

12.

Fang

, Go

, Chang

. Death and disability from warfarin-associated intracranial and extracranial hemorrhages. Am J Med 2007; 120: 700–5.

13.

Vandenbroucke

. Why do the results of randomised and observational studies differ? BMJ 2011; 343: d7020.

14.

Charidimou

, Shakeshaft

, Werring

. Cerebral microbleeds on magnetic resonance imaging and anticoagulant-associated intracerebral hemorrhage risk. Front Neurol 2012; 3: 133.

15.

Hylek

, Evans-Molina

, Shea

, Henault

, Regan

. Major hemorrhage and tolerability of warfarin in the first year of therapy among elderly patients with atrial fibrillation. Circulation 2007; 115: 2689–96.

16.

Gustafsson

, Asplund

, Britton

, Norrving

, Olsson

, Marke

. Cost effectiveness of primary stroke prevention in atrial fibrillation: Swedish national perspective. BMJ 1992; 305: 1457–60.

17.

Pantoni

. Cerebral small vessel disease: from pathogenesis and clinical characteristics to therapeutic challenges. Lancet Neurol 2010; 9: 689–701.

18.

Lovelock

, Cordonnier

, Naka

. Antithrombotic drug use, cerebral microbleeds, and intracerebral hemorrhage: a systematic review of published and unpublished studies. Stroke 2010; 41: 1222–8.

19.

Greenberg

, Vernooij

, Cordonnier

. Cerebral microbleeds: a guide to detection and interpretation. Lancet Neurol 2009; 8: 165–74.

20.

Fazekas

, Kleinert

, Roob

. Histopathologic analysis of foci of signal loss on gradient-echo T2*-weighted MR images in patients with spontaneous intracerebral hemorrhage: evidence of microangiopathy-related microbleeds. AJNR Am J Neuroradiol 1999; 20: 637–42.

21.

Shoamanesh

, Kwok

, Benavente

. Cerebral microbleeds: histopathological correlation of neuroimaging. Cerebrovasc Dis 2011; 32: 528–34.

22.

Werring

. Cerebral Microbleeds: Pathophysiology to Clinical Practice. Cambridge, UK, Cambridge University Press, 2011.

23.

Knudsen

, Rosand

, Karluk

, Greenberg

. Clinical diagnosis of cerebral amyloid angiopathy: validation of the Boston criteria. Neurology 2001; 56: 537–9.

24.

Ueno

, Naka

, Ohshita

. Association between cerebral microbleeds on T2*-weighted MR images and recurrent hemorrhagic stroke in patients treated with warfarin following ischemic stroke. AJNR Am J Neuroradiol 2008; 29: 1483–6.

25.

Poels

, Vernooij

, Ikram

. Prevalence and risk factors of cerebral microbleeds: an update of the Rotterdam scan study. Stroke 2010; 41(10 Suppl.): S103–6.

26.

Sveinbjornsdottir

, Sigurdsson

, Aspelund

. Cerebral microbleeds in the population based AGES-Reykjavik study: prevalence and location. J Neurol Neurosurg Psychiatry 2008; 79: 1002–6.

27.

Cordonnier

, Al-Shahi Salman

, Wardlaw

. Spontaneous brain microbleeds: systematic review, subgroup analyses and standards for study design and reporting. Brain 2007; 130(Pt 8):1988–2003.

28.

Naka

, Nomura

, Wakabayashi

. Frequency of asymptomatic microbleeds on T2*-weighted MR images of patients with recurrent stroke: association with combination of stroke subtypes and leukoaraiosis. AJNR Am J Neuroradiol 2004; 25: 714–9.

29.

Fan

, Zhang

, Lam

, Mok

, Wong

. Cerebral microbleeds as a risk factor for subsequent intracerebral hemorrhages among patients with acute ischemic stroke. Stroke 2003; 34: 2459–62.

30.

Soo

, Yang

, Lam

. Risk vs benefit of anti-thrombotic therapy in ischaemic stroke patients with cerebral microbleeds. J Neurol 2008; 255: 1679–86.

31.

Thijs

, Lemmens

, Schoofs

. Microbleeds and the risk of recurrent stroke. Stroke 2010; 41: 2005–9.

32.

Jeon

, Kang

, Cho

. Initial microbleeds at MR imaging can predict recurrent intracerebral hemorrhage. J Neurol 2007; 254: 508–12.

33.

Greenberg

, Eng

, Ning

, Smith

, Rosand

. Hemorrhage burden predicts recurrent intracerebral hemorrhage after lobar hemorrhage. Stroke 2004; 35: 1415–20.

34.

Bokura

, Saika

, Yamaguchi

. Microbleeds are associated with subsequent hemorrhagic and ischemic stroke in healthy elderly individuals. Stroke 2011; 42: 1867–71.

35.

Charidimou

, Kakar

, Fox

, Werring

. Cerebral microbleeds and recurrent stroke risk: systematic review and meta-analysis of prospective ischemic stroke and transient ischemic attack cohorts. Stroke 2013; 44: 995–1001.

36.

Rosand

, Eckman

, Knudsen

, Singer

, Greenberg

. The effect of warfarin and intensity of anticoagulation on outcome of intracerebral hemorrhage. Arch Intern Med 2004; 164: 880–4.

37.

Charidimou

, Gang

, Werring

. Sporadic cerebral amyloid angiopathy revisited: recent insights into pathophysiology and clinical spectrum. J Neurol Neurosurg Psychiatry 2012; 83: 124–37.

38.

Nicoll

, Burnett

, Love

. High frequency of apolipoprotein E epsilon 2 allele in hemorrhage due to cerebral amyloid angiopathy. Ann Neurol 1997; 41: 716–21.

39.

Biffi

, Sonni

, Anderson

. Variants at APOE influence risk of deep and lobar intracerebral hemorrhage. Ann Neurol 2010; 68: 934–43.

40.

Melo

, Bogousslavsky

, Regli

, Janzer

. Fatal hemorrhage during anticoagulation of cardioembolic infarction: role of cerebral amyloid angiopathy. Eur Neurol 1993; 33: 9–12.

41.

McCarron

, Nicoll

. Cerebral amyloid angiopathy and thrombolysis-related intracerebral haemorrhage. Lancet Neurol 2004; 3: 484–92.

42.

Smith

, Rosand

, Knudsen

, Hylek

, Greenberg

. Leukoaraiosis is associated with warfarin-related hemorrhage following ischemic stroke. Neurology 2002; 59: 193–7.

43.

Gorter

. Major bleeding during anticoagulation after cerebral ischemia: patterns and risk factors. Stroke Prevention In Reversible Ischemia Trial (SPIRIT). European Atrial Fibrillation Trial (EAFT) study groups. Neurology 1999; 53: 1319–27.

44.

Shoamanesh

, Kwok

, Benavente

. Cerebral microbleeds: histopathological correlation of neuroimaging. Cerebrovasc Dis 2011; 32: 528–34.

45.

Klein

, Altman

, Eriksson

. Estimation of the warfarin dose with clinical and pharmacogenetic data. N Engl J Med 2009; 360: 753–64.

46.

Rosand

, Hylek

, O'Donnell

, Greenberg

. Warfarin-associated hemorrhage and cerebral amyloid angiopathy: a genetic and pathologic study. Neurology 2000; 55: 947–51.

47.

Harrell

. Regression Modeling Strategies. New York, USA, Springer-Verlag, 2001.

48.

Ambler

, Seaman

, Omar

. An evaluation of penalised survival methods for developing prognostic models with rare events. Stat Med 2012; 31: 1150–61.

49.

Cheng

, Batool

, McCreary

. Susceptibility-weighted imaging is more reliable than T2*-weighted gradient-recalled echo MRI for detecting microbleeds. Stroke 2013; 44: 2782–6.

50.

Gregoire

, Chaudhary

, Brown

. The Microbleed Anatomical Rating Scale (MARS): reliability of a tool to map brain microbleeds. Neurology 2009; 73: 1759–66.

51.

Sacco

, Kasner

, Broderick

. An updated definition of stroke for the 21st century: a statement for healthcare professionals from the American Heart Association/American Stroke Association. Stroke 2013; 44: 2064–89.

52.

Collins

, Reitsma

, Altman

, Moons

. Transparent reporting of a multivariable prediction model for individual prognosis or diagnosis (TRIPOD): the TRIPOD statement. BMJ 2015; 350: g7594.

53.

Diener

, Connolly

, Ezekowitz

. Dabigatran compared with warfarin in patients with atrial fibrillation and previous transient ischaemic attack or stroke: a subgroup analysis of the RE-LY trial. Lancet Neurol 2010; 9: 1157–63.