Cerebral amyloid angiopathy,cerebral microbleeds and implications for anticoagulation decisions: The need for a balanced approach

Cerebral amyloid angiopathy (CAA) is a common small vessel disease characterized by amyloid deposition in small cortical and leptomeningeal arteries. ¹ CAA commonly accompanies stroke due to spontaneous lobar intracerebral hemorrhage (ICH) in elderly people, often associated with high risk of recurrence, in the range of 7% (95% CI: 3–12%) per year in well-phenotyped patients. ² For this reason, when there is suspicion of CAA, clinicians might remain hesitant in prescribing oral anticoagulation in patients in whom it is otherwise indicated, including the case of non-valvular atrial fibrillation. This point is well illustrated in a recent comment piece by DeSimone et al., ³ as well as various review articles and opinion pieces on the topic. Though we agree with the general approach that clinicians should consider the possibility of CAA in weighing the risks and benefits of anticoagulation in elderly patients, ³ the clinical setting and context are key in these “tightrope balancing acts.”

While CAA is estimated to have prevalence of 20–40% in non-demented, and 50–60% in demented elderly subjects based on population-based autopsy studies, ⁴ only a small proportion of these individuals go on to develop CAA-related clinical events including symptomatic ICH. ¹ Hence, the majority of individuals with CAA won’t be clinically affected by the disease and are never diagnosed during life, even through the use of the validated clinical-neuroimaging Boston criteria. ⁵ The most devastating clinical presentation of CAA is spontaneous lobar ICH. ¹ Indeed, given the high recurrence rates after CAA-related ICH, a common approach has been to avoid anticoagulation when possible, to mitigate recurrence risk. ⁶ The role of oral anticoagulation for atrial fibrillation in CAA patients with a history of ICH thus remains hotly debated, despite the absence of real-world data on the effect of CAA on risk of anticoagulation-related ICH. ⁷ CAA is associated with a high risk of ICH recurrence, but in recent years, it has been better appreciated that this risk is not uniform, with the potential for different CAA phenotypes with varying propensities towards bleeding. Particularly, it seems that cortical superficial siderosis ⁸ – a recently discovered hemorrhagic MRI signature of CAA – is a potent marker of individuals at highest risk for recurrent ICH. This has implications in balancing ischemic versus hemorrhagic stroke risk, in that even after a CAA-related ICH, the presence of atrial fibrillation might confer enough risk for ischemic stroke, to offset the presumed risk of ICH recurrence in many patients (especially in the absence of siderosis and a CHA2DS2 score >1–2).

Along these lines, a recent patient-level meta-analysis demonstrated that restarting oral anticoagulation – predominantly with warfarin – is associated with decreased mortality and subsequent ischemic strokes, without an increase in recurrent ICH risk regardless of ICH subtype, including subgroup analysis of patients diagnosed with CAA as per Boston Criteria. ⁹ In particular, non-Vitamin-K antagonists oral anticoagulants (NOACs) might be an attractive option in CAA-related ICH survivors, in view of their 50% lower risk of ICH relative to warfarin, with absolute rates of ICH that are similar to aspirin monotherapy. ¹⁰ Although patients with previous ICH were excluded from all previous randomized trials investigating NOACs for stroke prevention in atrial fibrillation, ¹¹ several Phase II randomized controlled trials assessing whether the favorable safety profile of NOACs persists in ICH survivors with atrial fibrillation, including patients with CAA, are underway (Table 1). ¹²

OPEN IN VIEWER

Table 1.

Summary of the ongoing randomized trials of anticoagulation for atrial fibrillation after intracranial hemorrhage

Trial	Methods	Participants	Interventions	Primary outcome (follow-up)
APACHE-AF (NCT02565693)	Multicenter open label RCT in The Netherlands (2014–2018)	Patients aged ≥18 years with anticoagulant-associated ICH, AF and CHA2DS2-VASc score ≥3 (target n = 100)	Two parallel groups: apixaban vs. aspirin, or carbasalate calcium, or clopidogrel, or dipyridamole, or no antithrombotic drug	Vascular death or non-fatal stroke (≥12 months)
NASPAF-ICH (NCT02998905)	Multicenter open label RCT in Canada (2017–2019)	Patients aged ≥45 years with AF, a previous ICH and a CHADS2 score ≥2 (target n = 100)	Two parallel groups: non-vitamin K antagonist oral anticoagulant vs. aspirin	Composite outcome of ischemic stroke and recurrent ICH (≥6 months)
SoSTART (NCT03153150)	Multicenter, open label RCT recruiting in the UK (2017–2023)	Spontaneous ICH, AF, and a CHA2DS2-VASc score ≥2 (target n = 60 in the pilot phase)	Two parallel groups: start oral anticoagulant vs. avoid oral anticoagulant	All symptomatic serious vascular events (≥12 months)
A3ICH (NCT03243175)	Multicenter, open label RCT recruiting in France (2017–2020)	≥18 years old, with AF and CHA2DS2VASc ≥2 (target n = 300)	Three parallel groups: Apixaban or Left Atrial Appendage Closure or neither	All symptomatic serious (fatal and non-fatal) vascular events (24 months)
STATICH (EudraCT: 2014-002636-13)	Multicenter, open label RCT recruiting in Norway, Denmark, and Sweden (2017–2022)	Spontaneous, primary ICH (target n = 500)	Two parallel groups: antithrombotic drugs vs. no antithrombotic drugs	Recurrent symptomatic ICH

The risk/benefit of anticoagulation therapy seems most reassuring for individuals with other CAA manifestations, including asymptomatic lobar microbleeds, particularly in the context of secondary ischemic stroke prevention. In this setting, clinicians should carefully weight risk-benefit for recurrent ischemic stroke vs. hemorrhagic complications, while keeping in mind that cerebral microbleeds are not only a marker of ICH, but also of ischemic stroke. A meta-analysis of ischemic stroke/TIA cohorts showed that in the presence of strictly lobar microbleeds, the pooled absolute event rate was 9.3% for future ischemic stroke versus 3.6% for ICH. ¹³ Moreover, it is important to keep in mind that there exists strong evidence from large randomized trials for the benefits of oral anticoagulation for people with atrial fibrillation, while data on ICH risk in relation to CAA and cerebral microbleeds mainly come from relatively small (and likely prone to selection bias) observational studies.^2,13 In light of these, a scientific statement from the American Heart Association/American Stroke Association on prevention of stroke in patients with silent cerebrovascular disease, recommends that MRI screening for microbleeds is not needed before the initiation of antithrombotic therapies. ¹⁴

We suggest that a clinical approach based on the absolute risks of ischemic stroke vs. intracerebral hemorrhage, and considering the specific clinical context (symptomatic CAA presenting with ICH vs. other presentations or incidentally found, etc.) might be helpful in guiding anticoagulation decisions in the setting of CAA and atrial fibrillation. Randomized trials are clearly required to settle these important questions in survivors of CAA-ICH. A major barrier to overcome in future randomized trials of different oral anticoagulation strategies after ICH, is the threat of low recruitment rates, due to the reluctance of investigators to enroll patients with lobar ICH or CAA out of fear of bleeding complications based on small observational studies alone. ¹⁵ In patients with non-ICH CAA manifestations, no trials are underway, but the weight of evidence in this setting might favor oral anticoagulation.