A 77-year-old woman presents with acute dysarthria and mild right facial droop, with NIH Stroke Scale score 2. Electrocardiogram shows new onset atrial fibrillation. Echocardiogram shows a mildly dilated left atrium with a normal ejection fraction and no wall motion abnormalities. An MRI brain shows small areas of restricted diffusion in both hemispheres, consistent with acute infarction as the cause of her new dysarthria (Fig. 26.2). MRI SWI shows fifteen lobar microbleeds, without evidence for microbleeds in deep locations. Should this patient be anticoagulated to prevent recurrent ischemic stroke due to atrial fibrillation? What is the risk of anticoagulant-induced intracerebral hemorrhage (ICH) in patients with microbleeds?

The microbleeds indicate the presence of a hemorrhage-prone vasculopathy, and the risk of subsequent ICH is probably increased. The clinical question is whether the increased risk of subsequent ICH outweighs the risk reduction that would be conferred by anticoagulation. This patient has a CHA₂DS₂-VASC (for definition see Chap. 12) score of 5 that is associated with an annual risk of recurrent ischemic stroke of ~6.7 % which would be reduced to ~2.2 % with anticoagulation [12]. Therefore, anticoagulation is indicated in the absence of other factors that would substantially increase the risk of bleeding complications.

Microbleeds appear as small, focal areas of signal loss (hypointensity) on T2* sensitive sequences such as GRE or SWI [13]. They are usually not apparent on other MRI sequences and cannot be seen on CT. In the absence of a few other specific conditions that cause microbleeds (e.g., familial cavernous malformation, infective endocarditis, and traumatic diffuse axonal injury), which should be obvious from the history, microbleeds are nearly always attributed to either arteriosclerosis, related to aging and vascular risk factors, or CAA. Radiopathological correlation studies suggest that microbleeds represent small areas of extravascular hemosiderin deposition from previous small, asymptomatic hemorrhages [14]. Microbleeds must be discriminated from several common mimics. Blood vessels seen in cross section will appear as small round hypointense dots, because deoxygenated hemoglobin also has a susceptibility effect. Calcifications, frequently seen in the globus pallidus, may also appear as hypointensities, mimicking microbleeds. There are two published standardized rating systems to enhance specificity and reliability of microbleed identification [15, 16].

One or more chronic microbleeds are seen in up to 30–70 % of persons with acute ischemic stroke, but only 10–30 % of the general population [17]. The increased prevalence in ischemic stroke patients likely reflects shared risk factors for microbleeds and ischemic stroke, such as hypertension. The clinical concern in our patient is that the microbleeds may signify an increased risk for future ICH, and greater risk from anticoagulation. Unfortunately, the risk of warfarin-related ICH in patients with microbleeds is poorly defined. In the absence of better data from prospective studies, specific recommendations for antithrombotic strategy for atrial fibrillation in the setting of incidentally discovered lobar microbleeds cannot be provided, but general guidelines can be discussed. In general, the clinical approach should be to determine the cause of the microbleeds (hypertension, CAA, or other) with application of the Boston criteria (Table 26.2), to mitigate bleeding risk by carefully controlling hypertension, and to judge the risks and benefits of anticoagulation strategies in light of the number of microbleeds and their underlying cause. Screening with MRI to identify microbleeds before anticoagulation is not recommended, given the current lack of certainty on how they should affect management.

In the patient under discussion, the presence of multiple microbleeds in lobar brain locations, without microbleeds in deep hemispheric locations such as the basal ganglia or thalamus, suggests that CAA is the underlying cause [8]. CAA is caused by vascular amyloid deposition. Vascular beta-amyloid is toxic to vascular smooth muscle cells, leading to fibrosis, necrosis and loss of vascular wall integrity with bleeding. CAA causes about 20 % of all symptomatic ICHs. Chronic, asymptomatic microbleeds or areas of superficial siderosis (Fig. 26.3) are often seen in addition to symptomatic hemorrhagic strokes. The Boston criteria for CAA diagnosis use age, presence and number of lobar hemorrhages, microbleeds, and superficial siderosis to assign a probability of underlying CAA as the cause of ICH (Table 26.2) [18, 19]. These criteria rely on the fact that vascular beta-amyloid preferentially involves the leptomeningeal and cortical vessels, with relative sparing of vessels supplying the basal ganglia. Therefore, hemorrhages and microbleeds affecting the cortex and subcortical white matter in the elderly are potentially related to CAA (although arteriosclerosis due to conventional vascular risk factors can also cause bleeding in these locations), while hemorrhages and microbleeds in the deep hemispheric structures such as the basal ganglia are unlikely to be caused by CAA and more likely to be related to arteriosclerosis due to hypertension and other vascular risk factors (Fig. 26.4). The Boston criteria have been pathologically validated in persons with lobar ICH [19]. Although the criteria have not been pathologically validated in persons like our patient under discussion, who only had multiple lobar microbleeds without symptomatic ICH, it is likely that most elderly with multiple lobar microbleeds do in fact have CAA, based on studies showing similar genetic and risk factor profiles as pathologically proven CAA cases [20].

The finding that microbleeds are probably caused by CAA should heighten concern regarding anticoagulation, because patients with CAA have higher bleeding risk than patients with arteriosclerosis caused by conventional risk factors such as hypertension. In patients with lobar ICH and possible or probable CAA the recurrence rate is 5–10 % per year compared to only 2–3 % per year for ICH not caused by CAA. The risk for recurrent CAA-related ICH is higher in patients with larger numbers of additional asymptomatic microbleeds [21], and in post-ICH aspirin users who have more than 5 microbleeds [22]. Consequently, American Heart Association/American Stroke Association (AHA/ASA) guidelines recommend to avoid resuming oral anticoagulation in patients with lobar ICH, but to consider resuming oral anticoagulation in patients with non-lobar ICH [23]. However, our patient with microbleeds has not had a previous symptomatic ICH; therefore her rate of new symptomatic ICH might be lower. A single cohort study of 69 patients with two or more lobar microbleeds and non-hemorrhagic symptoms of CAA (mostly transient neurological events or cognitive impairment) found that 5 % per year had incident symptomatic ICH [24]. Whether truly asymptomatic microbleeds, as in our patient, confer the same yearly risk is unknown.