Clinical manifestations

Presentation and course

It is important to distinguish the following disorders related to deposition of amyloid beta in the central nervous system from the systemic amyloidoses in which amyloid is deposited in the peripheral nervous system and other organs. Although the familial oculoleptomeningeal amyloidoses can affect sites both in and outside the central nervous system, brain and vascular involvement are typically not prominent features of these disorders.

Spontaneous intracerebral hemorrhage. Cerebral amyloid angiopathy most commonly appears in a sporadic form associated with aging. A classic vascular presentation of sporadic cerebral amyloid angiopathy is lobar hemorrhage, often recurrent or multifocal, in an elderly individual.

Cerebral amyloid angiopathy has been identified as responsible for 10% to 20% of primary intracerebral hemorrhages in autopsy series and 34% of primary intracerebral hemorrhages in a clinical series drawn from various populations (92; 96; 75). The associated hemorrhages often rupture into the subarachnoid space; these features contrast with hypertensive hemorrhages, which preferentially rupture into the ventricular space. In several series, 9% to 53% of clinically significant hemorrhages related to cerebral amyloid angiopathy were multiple.

Controlling for lobar volume, cerebral amyloid angiopathy-related hemorrhage tends to occur most commonly per cc of tissue in the occipital cortex, though all lobes can be affected (195; 165). Cerebellar and basal ganglia intracerebral hemorrhages secondary to cerebral amyloid angiopathy occur infrequently, and brainstem hemorrhage is extremely rare. Subclinical superficial sulcal/meningeal bleeding is common, appearing in 23 of 28 (61%) pathologically proven cases of cerebral amyloid angiopathy (113), and it may be associated with increased risk of recurrent intracerebral hemorrhage (50; 2017a 162; 152; 112) and further sulcal bleeding events (151; 150). It has been shown that sulcal bleeding events can be the first presentation for patients with cerebral amyloid angiopathy (38). Cortical superficial siderosis, which results from these sulcal and meningeal bleeding events, is relatively rare in healthy individuals in the population (149; 203). In patients with cerebral amyloid angiopathy, superficial siderosis progression can be easily recognized and semiquantitatively measured using serial MRIs. In patients with symptomatic cerebral amyloid angiopathy presenting without lobar intracerebral hemorrhage, superficial siderosis evolution assessed on MRI follow-up independently predicts increased symptomatic intracerebral hemorrhage risk (HR: 3.76; 95%CI: 1.37-10.35; P=0.010) at roughly 1 year (42). The locations of cerebral amyloid angiopathy-related brain hemorrhages reflect the distribution of the cerebral amyloid angiopathy pathology, a microangiopathy that most frequently affects neocortex and the overlying leptomeninges, rarely the cerebellum (including meninges), and almost never the deep hemispheric or brainstem structures (140). This pattern of localization helps distinguish cerebral amyloid angiopathy-related intracerebral hemorrhage from the more characteristically deep hemispheric bleeds associated with hypertension. It has been recognized that a combination of both hypertensive and vascular amyloid pathologies is not uncommon in patients with intracerebral hemorrhage (144). The distribution of cerebellar hemorrhagic features may also help distinguish the underlying microangiopathy at play. In patients with spontaneous intracerebral hemorrhage, strictly superficial cerebellar microbleeds have been associated with clinically diagnosed and pathologically proven cerebral amyloid angiopathy (145). Additionally, cerebellar superficial siderosis has recently been associated with cerebral amyloid angiopathy (104).

The clinical features of intracerebral hemorrhage secondary to cerebral amyloid angiopathy include an abrupt onset of headache, focal lobar neurologic deficits, altered level of consciousness, nausea, and vomiting. Head injury and surgical procedures have been anecdotally reported to precipitate intracerebral hemorrhages, though a causal relationship remains to be established. Stronger evidence supports anticoagulant, thrombolytic, or antiplatelet agents as triggers for cerebral amyloid angiopathy-related hemorrhages (164; 123; 19). Individuals with exclusively lobar microbleeds may be at increased risk of incident intracerebral hemorrhage, even in the absence of prior intracerebral hemorrhage (183; 01); however, it is important to recognize that clinical context determines degree of risk (39). Hypertension frequently coexists with cerebral amyloid angiopathy in older populations. Growing evidence suggests that hypertension may increase risk of hemorrhage in cerebral amyloid angiopathy, but further studies in this area are required (16).

Other presentations. In addition to lobar hemorrhage, patients with sporadic cerebral amyloid angiopathy can present clinically with mild cognitive impairment or dementia (47; 26) as well as with transient focal neurologic episodes (TFNEs).

Cognitive impairment. Cognitive impairment and dementia are frequently encountered in patients with cerebral amyloid angiopathy. In Vinters’ series, over 40% of patients with cerebral amyloid angiopathy–related hemorrhage exhibited dementia (195). Dementia frequently occurs in patients after lobar intracerebral hemorrhage. Patients are at risk of developing dementia both immediately after intracerebral hemorrhage (early post-ICH dementia) and in the chronic phase, after hemorrhage (delayed post-ICH dementia). The yearly incidence of dementia appears to be 5.8% (95% CI: 5.1% to 7.0%). Larger hematoma size is associated with early risk but not late risk. Educational level, incident mood symptoms, and white matter disease appear to be associated with late-onset dementia (21). Although dementia might be triggered or exacerbated by the occurrence of cerebral amyloid angiopathy–related intracerebral hemorrhage, studies have suggested that as much as 24% of patients with cerebral amyloid angiopathy may have been cognitively impaired prior to hemorrhage (174; 12). Several studies have further suggested that the underlying small vessel disease in cerebral amyloid angiopathy, rather than the intracerebral hemorrhage itself, is a predictor of cognitive impairment in the disease (14; 18; 136). A study by Smith and colleagues also suggested that cerebral amyloid angiopathy–associated leukoaraiosis was an important mechanism for cognitive impairment in these subjects; severe white matter hyperintensities were present in 60% of subjects with pre-hemorrhage cognitive impairment versus 24% of those without. Along with leukoaraiosis, other mechanisms likely to contribute to cognitive impairment in cerebral amyloid angiopathy include lobar hemorrhages, microbleeds, ischemic infarcts, and parenchymal amyloid (ie, Alzheimer disease). Advanced cerebral amyloid angiopathy also appears to cause cognitive impairments in individuals without intracerebral hemorrhage. A clinical-pathological study of 404 community-dwelling individuals found that those with moderate to very severe cerebral amyloid angiopathy (79 of 404, 18.8%) had worse performance on measures of perceptual speed and episodic memory, even after controlling for potential confounders, such as age and Alzheimer disease pathology (05). A study has suggested that cerebral amyloid angiopathy is an important independent predictor of which patients with Alzheimer disease pathology go on to develop dementia. Patients with more severe cerebral amyloid angiopathy had faster rates of cognitive decline (27). Pathologic studies suggest that along with other age-related pathologies, cerebral amyloid angiopathy appears to be a frequent contributor to cognitive impairment in the elderly (28). The cognitive profile of patients with cerebral amyloid angiopathy resembles other types of vascular cognitive impairment, with most prominent deficits in executive function and processing speed (32; 206). Cognitive impairment in cerebral amyloid angiopathy is likely related to the accumulation of multiple pathologic lesions in the disease, including microbleeds, white matter hyperintensities, and cerebral microinfarctions (08; 07).

Transient focal neurologic episodes (TFNEs). Small hemorrhages or microbleeds in cerebral amyloid angiopathy occur in superficial sulcal/meningeal, cortical, and subcortical sites, most often without definite clinical manifestations. Some of these individuals, however, develop recurrent, transient neurologic symptoms (78; 154; 49). The symptoms are typically stereotyped, recurrent, transient neurologic events characterized by paresthesias, numbness, or weakness with a march over seconds or minutes (49; 139; 173). An analysis of 25 individuals with transient focal neurologic episodes from a multicenter cohort of 172 individuals diagnosed with cerebral amyloid angiopathy found approximately equal numbers with predominantly positive symptoms (spreading paresthesias, positive visual phenomena, or limb jerking) and negative symptoms (weakness, language impairment, or visual loss) (49). Among those individuals who underwent MRI, chronic superficial sulcal/meningeal hemorrhage was identified in 50% with transient focal neurologic episodes versus 19% without (p=0.001), suggesting a possible causative role for the superficial blood products in generating these symptoms. Acute convexity subarachnoid hemorrhage and cortical superficial siderosis could potentially trigger cortical spreading depolarizations, manifesting clinically as TFNEs (109). Importantly, these symptoms may be misdiagnosed as transient ischemic attacks, focal seizures, or migraine with aura. Neuroimaging, especially susceptibility-weighted MRI, might be warranted to ensure correct diagnosis (33; 106).

Several studies have shown an association between sulcal bleeding events and future intracerebral hemorrhage, raising the possibility that these spells might also be markers for future bleeding (49; 35; 13; 139; 162). In cohorts of patients with acute convexity subarachnoid hemorrhage and probable cerebral amyloid angiopathy, who predominantly presented with TFNEs, the rate of intracerebral hemorrhage was remarkably high, estimated in 19% per year (95% CI: 13% to 27%) (202; 33). Although antithrombotic treatment could reduce the risk of subsequent ischemic stroke in patients with transient ischemic attacks, it could exacerbate the hemorrhagic risk of patients with TFNEs related to cerebral amyloid angiopathy. Hence, the correct diagnosis of the underlying cause of neurologic symptoms is paramount (106).

Nonhemorrhagic imaging markers. Accumulating data suggest an association between cerebral amyloid angiopathy and white matter lesions or leukoaraiosis. Supporting lines of evidence include: (1) prominent leukoaraiosis in several hereditary forms of cerebral amyloid angiopathy (25; 70); (2) neuropathological correlations between advanced cerebral amyloid angiopathy and white matter disease (83); and (3) volumetric MRI analysis showing almost twice as much white matter disease in advanced cerebral amyloid angiopathy as in Alzheimer disease or mild cognitive impairment, conditions that are themselves associated with leukoaraiosis (82). The association between cerebral amyloid angiopathy and leukoaraiosis is striking, in that cerebral amyloid angiopathy generally spares the small vessels within the regions of white matter pallor. Topographically remote convexity or meningeal cerebral amyloid angiopathy have been suggested to generate leukoaraiosis indirectly by producing hypoperfusion of the white matter through long, pial-penetrating arteries (71). Like amyloid pathology in the disease, white matter lesions appear to have a posterior predominance in patients with cerebral amyloid angiopathy (178). The white matter lesion pattern of subcortical spots also appears to be more common in cerebral amyloid angiopathy compared to patients with hypertensive arteriopathy who more commonly harbor peribasal ganglia lesions (40). Hence, this multispot white matter hyperintensity pattern has been included in the updated Boston criteria v2.0 (34).

Cerebral amyloid angiopathy also appears to trigger small, clinically asymptomatic ischemic infarcts (189; 192). In a neuropathological study using microglial immunostaining to detect microinfarcts, these lesions were found to be significantly more common and more numerous in brains with severe rather than mild cerebral amyloid angiopathy (175). Analysis of living patients by diffusion-weighted MRI reached similar conclusions: small lesions consistent with subacute infarctions were identified in 12 of 78 (15%) with probable cerebral amyloid angiopathy versus 0 of 55 microbleed-free control subjects (99). The lesions were located mostly in cortex and subcortical white matter and were associated with higher numbers of cerebral microbleeds. An analysis of 114 patients with recent intracerebral hemorrhage found diffusion-weighted lesions in 9 of 39 (23%) with probable cerebral amyloid angiopathy versus 6 of 75 (8%) in other types of intracerebral hemorrhage and 0 of 47 age-matched control subjects (79), suggesting a particular association with advanced cerebral amyloid angiopathy. These diffusion-weighted lesions also appear to occur in cerebral amyloid angiopathy-related subarachnoid hemorrhages (13). A neuroimaging-neuropathological study has shown that the number of microbleeds on MRI correlates with the number of microinfarcts on pathology (108). An in-vivo-MRI-ex-vivo-MRI-histopathological approach revealed that DWI+ lesions are more common in cerebral amyloid angiopathy cases than in controls and represent acute microinfarcts (177).

Cortical microinfarctions may be an important contributor to cognitive impairment and dementia in cerebral amyloid angiopathy. A longitudinal study revealed that patients with cerebral amyloid angiopathy with cortical microinfarcts have a higher cumulative incidence of dementia than those without them (207). In a neuropathological study involving 284 autopsy cases of demented and nondemented individuals, the presence and severity of cerebral amyloid angiopathy was associated with microinfarcts comprising the CA1 region of the hippocampus and correlated with cognitive decline (87).

In concert, cerebral microinfarcts may disrupt structural connectivity networks in the brain, resulting in impairments in cognitive domains, such as executive functioning and processing speed, as well as reduced gait velocity (156). Furthermore, in patients followed longitudinally (mean follow-up time: 1.3±0.4 years), progression of impairment in structural connectivity networks was associated with worse executive functioning (Beta=0.41, p=0.04) (157).

Enlarged perivascular spaces (EPVS) are another neuroimaging marker of small vessel disease (SVD) (88; 166; 57; 214; 215; 199), and in cerebral amyloid angiopathy in particular (48; 121; 45). Previous studies on enlarged perivascular spaces have generally distinguished between enlarged perivascular spaces in the basal ganglia (BG-EPVS) and enlarged perivascular spaces in the white matter (WM-DPVS). Several studies have suggested that patients with probable or possible cerebral amyloid angiopathy appear to have a higher prevalence of severe enlarged perivascular spaces in the white matter (48; 121). Additionally, EPVS in the white matter appears to correlate with pathologically proven cerebral amyloid angiopathy (45). It has been hypothesized that in patients with cerebral amyloid angiopathy, interstitial fluid blockage due to amyloid beta accumulation within the perivascular space might favor the dilation of these spaces in white matter regions. This may be a result of changes in perivascular drainage due to aging and other factors that contribute to amyloid beta accumulation in patients with cerebral amyloid angiopathy (85). Indeed, neuropathologic evidence suggests that perivascular space dilatation is most severe in areas of high amyloid deposition (188). Furthermore, a neuroimaging-neuropathological study in cerebral amyloid angiopathy revealed that cortical arterioles overlying EPVS in the subcortical white matter have markedly reduced smooth muscle cells and increased vascular amyloid beta accumulation (147). Perivascular space dilatation may, thus, reflect impaired outward flow along arterioles, supporting the notion that perivascular clearance mechanisms are compromised in cerebral amyloid angiopathy.

Hereditary cerebral amyloid angiopathy. Several familial forms of cerebral amyloid angiopathy reported to date demonstrate autosomal dominant inheritance (15). The Icelandic form (hereditary cerebral hemorrhage with amyloidosis-Icelandic type) typically presents in the third decade with multiple intracerebral hemorrhages and progresses to dementia, paralysis, and early mortality (142). Headache (experienced in 63% of patients) and epilepsy (in 25%) are frequent. As opposed to other forms of cerebral amyloid angiopathy, most hemorrhages in the Icelandic form occur in the basal ganglia. Spinal cord vessels may also be affected. Survival is usually 10 to 20 years from symptom onset.

The Dutch form (Dutch-type hereditary cerebral amyloid angiopathy, also referred to as hereditary cerebral hemorrhage with amyloidosis-Dutch type) has been described in several families in two fishing villages on the North Sea coast of the Netherlands (213). This form of cerebral amyloid angiopathy has a later onset, in the fourth to sixth decades. Although chronic headaches may precede other symptoms by years, intracerebral hemorrhage is the usual presenting symptom. Another 13% of patients experience ischemic strokes, and white matter ischemic changes are common. Hemorrhages are frequently located in the parietal lobe and often progress. The mortality for the initial hemorrhage is 50% to 70% and is higher in females and when the disorder is paternally transmitted. Neuropsychiatric symptoms or dementia are a constant feature and likely represent a dementia predominantly of vascular etiology, although Alzheimer disease pathology may play a role in older subjects. The course can be varied, with some patients showing a relatively benign picture and others experiencing recurrent hemorrhages and death over a short period of time. White matter hyperintensities and cerebral microinfarctions appear to be more prevalent in presymptomatic subjects and may, thus, precede symptomatic intracerebral hemorrhage and cognitive symptoms (187). Evidence suggests that mutation carriers may have decreased levels of Aβ40 and Aβ42 in the CSF prior to the onset of clinical symptoms and, thus, may be an early biomarker in the disease (184). Decreased amyloid levels in the CSF are thought to be caused by increased vascular amyloid deposition in the disease, leading to vessel dysfunction. To further support this hypothesis, evidence suggests early changes in vascular reactivity in mutation carriers prior to symptom onset (185).

Several other hereditary forms of cerebral amyloid angiopathy have been identified. Continuing the convention of naming the forms according to the location where they were first observed, these include the Flemish, Italian, Arctic, Iowa, and Piedmont forms of hereditary cerebral amyloid angiopathy (213). Each of the forms has been reported to present clinically as either a dementing illness or lobar hemorrhage (with the exception of the Arctic form), with onset in approximately the sixth decade. The genetic basis for Dutch-type hereditary disease and these other forms of cerebral amyloid angiopathy is mutation in the amyloid precursor protein (APP) gene (see the paragraph on Genetics in the Pathogenesis and pathophysiology section of this article). Another set of families identified in France demonstrates a similar combination of dementia with or without hemorrhagic stroke, beginning in the fifth to sixth decade (167). Intriguingly, this hereditary form of disease appears related to duplication rather than mutation of amyloid precursor protein. The effect of gene dosage on amyloid deposition is similarly reflected by the noted association between Down syndrome and both Alzheimer disease and cerebral amyloid angiopathy. Anecdotal reports suggest older patients with Down syndrome are at increased risk of intracerebral hemorrhage (124), and cerebral amyloid angiopathy has been reported in patients with Down syndrome (129; 168). Because Alzheimer dementia develops universally in Down syndrome, it is difficult to determine the independent contribution of cerebral amyloid angiopathy to patients’ cognitive decline.

A British kindred has hereditary cerebral amyloid angiopathy that leads to progressive dementia, spasticity, and ataxia (familial British dementia, previously known as familial amyloidosis-British type) (194). Inheritance is autosomal dominant. In this form, the entire central nervous system is involved, including white matter, cerebellum, brainstem, and spinal cord. Age of onset is usually in the fifth or sixth decade. Unlike the Dutch and Icelandic forms, hemorrhage is not a prominent feature.

Familial oculoleptomeningeal amyloidosis is described in Japanese, Italian, and North American kindreds (181; 182). Clinical presentation varies from one kindred to another but can include dementia, ataxia, spasticity, ischemic strokes, seizures, peripheral neuropathy, hemiplegic migraine, myelopathy, blindness, and deafness. Intracerebral hemorrhage is rare. These patients have vascular amyloid deposition in the vitreous and retinal vessels, as well as in the leptomeningeal vessels and other organs, but not in brain parenchymal vessels.

Cerebral amyloid angiopathy-related inflammation. Growing numbers of patients have been reported to have episodes of inflammation associated with cerebral amyloid angiopathy (62; 170). These patients typically present with subacute mental status changes, headaches, and seizures, often at a slightly younger age than those presenting with cerebral amyloid angiopathy-related intracerebral hemorrhage. Characteristic imaging findings are prominent, asymmetric, white matter lesions with the MRI appearance of edema (hyperintense on T2-weighted sequences, increased diffusivity) that can extend to the overlying cerebral cortex (100). The clinical and neuroimaging criteria have a high sensitivity (82%) and specificity (95%) and can be used to make the diagnosis of cerebral amyloid angiopathy-related inflammation without brain biopsy (06).

Patients with this presentation often demonstrate clinical and neuroimaging improvement after immunosuppressive therapy, making this potentially the most therapeutically responsive subtype of cerebral amyloid angiopathy. The neuropathological appearance of cerebral amyloid angiopathy-related inflammation suggests that it represents an immune response to the vascular amyloid deposits. An autoimmune etiology has been further supported by the finding of high levels of antiamyloid autoantibodies in the cerebrospinal fluid of 10 individuals with probable cerebral amyloid angiopathy-related vascular inflammation (52) during the acute phase of the illness compared to several different control groups, including patients diagnosed with noninflammatory cerebra amyloid angiopathy or multiple sclerosis. As the patients with cerebral amyloid angiopathy-related inflammation entered remission after immunosuppressive treatment, autoantibodies returned to control levels (148).

Prognosis and complications

The outcome of acute lobar hemorrhage related to cerebral amyloid angiopathy is similar to that resulting from deep hemispheric hemorrhage related to hypertension. Both are associated with 3-month mortalities of approximately 25% and good outcomes (no or moderate disability) in only approximately 30% (163). Although lobar hemorrhage may be less likely than deep hemispheric hemorrhages to involve critical brainstem structures, this favorable feature is balanced by the tendency for lobar hemorrhages to be larger and to affect older patients.

Those who survive the initial hemorrhage remain at risk for recurrent hemorrhages, seizures, and cognitive deterioration. The cumulative rate of recurrent hemorrhage in survivors of lobar hemorrhage has been estimated at 21% over 2 years (141). Predictors of higher risk for recurrence are (1) history of a previous hemorrhage prior to the index presenting intracerebral hemorrhage; (2) possession of an apolipoprotein E2 or E4 allele (see Genetics subheading of Pathogenesis and pathophysiology section); (3) a larger number of hemorrhagic lesions counted on gradient-echo MRI scan; (4) presence of strictly lobar versus deep or mixed patterns of hemorrhage distribution (144); and (5) presence of cortical superficial siderosis (50; 105; 162; 26). In a study of 94 consecutive subjects with gradient-echo MRI at the time of lobar intracerebral hemorrhage, the 3-year cumulative risks for recurrent intracerebral hemorrhage were 14% for those with only one hemorrhage (ie, the presenting lesion) on MRI, 17% for those with two hemorrhagic lesions at baseline, 38% for those with three to five hemorrhages, and 51% for subjects with more than five hemorrhages (75).

Cortical superficial siderosis is considered one of the strongest predictors of first-ever and recurrent intracerebral hemorrhage and appears to also be associated with very early recurrent events in hemorrhage survivors (ie, recurrent event occurs within 6 months of index hemorrhage) (162). Multifocality, extent, and progression of cortical superficial siderosis have been associated with increased risk of recurrent hemorrhage (41; 150), with the annual incidence of intracerebral hemorrhage reaching 26.9% in those with bilateral disseminated cortical superficial siderosis (41). MRI and CT-visible convexity subarachnoid hemorrhage adjacent or remote from lobar acute intracerebral hemorrhage also herald an increased risk of recurrent bleeding (152; 112).

There is some evidence that posteriorly located white matter damage may increase intracerebral hemorrhage recurrence in patients with cerebral amyloid angiopathy (19). The presence of silent ischemic infarcts in the form of DWI+ lesions has also been associated with a higher number of total hemorrhages in patients with cerebral amyloid angiopathy (99). The presence of lacunes in individuals with cerebral amyloid angiopathy also relates to increased intracerebral hemorrhage risk (68).

Clinical vignette

A 65-year-old, right-handed man with history of multiple controlled vascular risk factors (diabetes mellitus, hypertension, hyperlipidemia) presented with acute onset of difficulty reading. He had a history of vague spells of word-finding difficulty in the previous year, as well as slightly increased forgetfulness, but remained fully functional and intellectually active. At the time of his symptoms, he was taking aspirin 81 mg daily, an oral hypoglycemic, an antihypertensive, and a statin. The family history was notable for dementia in the mother in her 70s and otherwise negative for stroke or other neurologic disease. Despite some ongoing complaints of mild difficulty reading, the neurologic exam showed no focal deficits and normal cognition, with zero errors on the Blessed Dementia Information-Memory-Concentration subscale.

Gradient-echo MRI sequences obtained after presentation demonstrated a subacute to chronic left occipital hematoma. These images also showed approximately 20 chronic hemorrhages or microbleeds, primarily in the left occipital and parietal lobes, but also involving the right occipital lobe. Moderately severe white matter hyperintensities were evident on T2-weighted sequences. There was no suggestion of mass lesion or vascular malformation underlying the hemorrhages, and a full laboratory examination disclosed no other cause of hemorrhage such as coagulopathy. The patient was diagnosed with probable cerebral amyloid angiopathy based on the finding of multiple strictly lobar hemorrhages without other definite cause.

The patient returned fully to his neurologic and functional baseline following his hemorrhagic stroke. Aspirin was stopped at the time of the acute hemorrhage but was restarted at 81 mg daily 2 weeks after presentation based on the presence of multiple vascular risk factors. The statin was continued for similar reasons. The antihypertensive medication was increased with the goal of bringing the blood pressure towards the low end of the normal range as tolerated. The patient was instructed not to use anticoagulants or drink heavily and otherwise to pursue his customary physical and mental activities.

Biological basis

Etiology and pathogenesis

The underlying explanation for why particular individuals without family history develop amyloid beta deposition in cerebral blood vessels remains uncertain. In the absence of identified environmental risk factors, genetic factors seem likely, presumably multiple genes interacting in a complex manner. Supporting this possibility is the strong effect of family history on risk of lobar intracerebral hemorrhage (205). In hereditary cerebral amyloid angiopathy, genetic protein defects trigger the accumulation of beta-pleated sheet fibrils in cerebral vessels.

“Amyloid” is a general term for insoluble proteinaceous material composed of beta-pleated sheet fibrils. Different forms of amyloid occur in a variety of diseases and are deposited in different parts of the body; the beta-pleated sheet is a constant feature of all forms. Amyloids are typically resistant to proteolysis and have affinity for certain dyes. The predominant form of amyloid in sporadic cerebral amyloid angiopathy and the hereditary forms caused by mutation or duplication of amyloid precursor protein is the amyloid beta peptide.

Molecular biology. Amyloid beta protein (Aß) is a hydrophobic, nonglycosylated peptide of 39 to 43 amino acids derived from the amyloid precursor protein encoded by a 19-exon gene located on the long arm of chromosome 21.

The amyloid precursor protein is a 659- to 770-amino acid residue, membrane-associated glycoprotein processed by secretase enzymes to yield various fragments, including amyloid beta protein. Subtle biochemical differences distinguish the vascular amyloid of cerebral amyloid angiopathy from parenchymal amyloid deposits (eg, amyloid plaques in Alzheimer disease). Vascular amyloid is predominantly composed of the 39- to 40-amino acid amyloid beta protein species (Aß40), whereas parenchymal plaque amyloid is predominantly composed of a 42- to 43-amino acid amyloid beta protein species (Aß42).

Pathophysiology

Aggregation. The central pathophysiologic biochemical event in cerebral amyloid angiopathy is the aggregation of soluble subunit proteins into progressively larger, less soluble, and more biologically active structures. Studies of synthetic amyloid beta protein species demonstrate a tendency to self-aggregate to fibrils in vitro, a predilection more pronounced in Aß42 fragments than Aß40. Mutant peptides mimicking the codon 693 or 694 point mutation linked to the Dutch, Italian, Arctic, or Iowa types of hereditary cerebral hemorrhage with amyloidosis exhibit increased beta-sheet content, increased amyloidogenicity, and enhanced fibril stability. Factors favoring in vivo amyloid beta formation and deposition include (1) overproduction or increased levels of amyloid precursor (as in gene overdose in trisomy 21 Down disease or amyloid precursor protein duplication); (2) a highly amyloidogenic sequence (as in the codon 693 or 694 mutations); (3) altered proteolysis of the amyloid precursor (as in familial Alzheimer disease due to codon 692 mutation); (4) a seeding nucleation event analogous to seeding in crystal formation (132); (5) reduced clearance of brain amyloid (also suggested to occur with mutant forms of amyloid beta protein) (54); and (6) time. Finally, there is mounting evidence that decreases in cerebral blood flow could lead to increased amyloid deposition in cerebral amyloid angiopathy. This may be related to changes in amyloid clearance pathways associated with decreases in cerebral blood flow (66; 111).

Production and clearance of brain amyloid beta protein. A variety of theories regarding the source of cerebrovascular amyloid beta protein have been proposed over the years, including the possibility that it is derived from the systemic blood circulation or produced locally within brain blood vessels. There is some evidence that peripherally formed amyloid beta protein may influence cerebral amyloid deposition (60; 176). However, the prevailing model based on current data posits that amyloid beta protein is generated primarily by neurons, diffuses through the brain parenchyma (where it may deposit as the senile plaques of Alzheimer disease), and ultimately accompanies the brain interstitial fluid to the periarterial space (201). From the periarterial space, amyloid beta protein may either deposit within vessel walls as cerebral amyloid angiopathy or be cleared from the brain via efflux transport across the blood-brain barrier, proteolytic degradation, or centrifugal flow of interstitial fluid out of the brain (169; 24; 30; 135).

The following are among the experimental observations supporting this paradigm: (1) neuronal expression of amyloid precursor protein in transgenic mice is sufficient to produce robust cerebral amyloid angiopathy (demonstrating that amyloid beta protein produced by neurons can reach the vessels); (2) the earliest site of vascular amyloid beta protein deposition is located in proximity to the periarterial space, at the junction of the media and adventitia; (3) there is reduced clearance by efflux transport of the mutant forms of amyloid beta protein that tend to accumulate in vessels; (4) shifting the population of amyloid beta protein from Aß40 to Aß42 by coexpression of mutant presenilin-1 causes more of the peptide to deposit as senile plaques and less to reach the periarterial space and deposit in vessels (90); and (5) genetic reduction or elimination of the protease neprilysin causes cerebral amyloid angiopathy to appear in a transgenic mouse model that otherwise does not develop cerebral amyloid angiopathy at a comparable age (63).

Important implications of this model are that cerebral amyloid angiopathy progression might be slowed by agents that lower its production, decrease its tendency to aggregate in the periarterial space, stimulate its degradation, or enhance its clearance. Many of these predictions remain to be tested in mouse models and humans with cerebral amyloid angiopathy. Data testing an anti-amyloid immunotherapy in mice have suggested that it can both remove vascular amyloid deposits and improve vascular reactivity in animals (09). However, a phase 2 multicenter randomized control trial testing the effect of the same anti-amyloid immunotherapy on blood vessel function in patients with cerebral amyloid angiopathy showed that, though safe and well-tolerated, the drug was associated with a trend towards reduced cerebrovascular reactivity, opposite to the expected direction (http://clinicaltrials.gov/show/NCT01821118) (110).

Perivascular clearance has taken central stage in explaining the accumulation of amyloid beta in the brain and is considered an important shared pathogenic mechanism between Alzheimer disease and cerebral amyloid angiopathy (72). Although incompletely understood, perivascular drainage seems to be driven in part by arterial dilations, including both neurovascular coupling (91) and spontaneous vasomotion (72; 190). Evidence from studies in humans and animal models supports that vascular amyloid beta deposition interferes with cerebrovascular reactivity. Impaired perivascular clearance caused by vascular amyloid beta deposition could, thus, lead to a self-reinforcing cycle in which decreased perivascular clearance leads to accrued amyloid beta vascular accumulation and more damage to vascular structure and function (72). Reduced perivascular clearance of amyloid beta could also increase parenchymal amyloid beta accumulation. This hypothesis is supported by studies of patients participating in passive immunotherapy trials for Alzheimer disease. These studies suggest that by increasing the clearance of amyloid beta from plaques into the perivascular space, anti-amyloid immunotherapy could lead to an accumulation of vascular amyloid beta (24). More recently, the term amyloid-related imaging abnormalities (ARIA) was coined in relation to newer anti-amyloid immunotherapies. These include ARIA-E (edema) and ARIA-H (hemorrhage) and have imaging features that closely resemble cerebral amyloid angiopathy–related inflammation. Though incompletely understood, ARIA is thought to result from a combination of increased cerebral amyloid angiopathy in the setting of perivascular amyloid beta clearance as well as vascular amyloid beta removal leading to perivascular inflammation, blood-brain barrier leakage, and hemorrhage. Rapid mobilization of parenchymal amyloid beta from plaques into the vessel walls could also explain the reduction in cerebrovascular reactivity observed in patients with cerebral amyloid angiopathy treated with amyloid immunotherapy (110). Cerebral amyloid angiopathy has been identified as a significant risk factor for ARIA; therefore, leaders in the cerebral amyloid angiopathy research and clinical community have strongly recommended against the off-label use of currently available anti-amyloid immunotherapies for cerebral amyloid angiopathy (74).

There is mounting evidence that amyloid beta protein may propagate and accumulate in the central nervous system similar to prion proteins (86). Similar to prion proteins, amyloid beta protein has been detected in dura mater and pituitary glands. Furthermore, amyloid beta pathology has been replicated following amyloid beta inoculation in amyloid precursor protein (APP) transgenic mice, suggesting that amyloid beta pathology is transmissible, although unlike prion disease, there is no evidence that Alzheimer disease is transmissible (132; 60; 134; 209; 29). Numerous independent studies have reported significant amyloid beta pathology in cases of iatrogenic Creutzfeldt-Jakob disease either from cadaveric growth hormone injections or dura mater grafts (94; 65; 84; 160; 31; 58). The majority of cases in these studies demonstrate both parenchymal and vascular amyloid beta deposition (160; 31). One possible explanation for this is that both amyloid beta and prion protein seeds are transmitted during these procedures and propagate and accumulate in the brain concurrently. However, Ritchie and colleagues have observed amyloid protein accumulation even in the absence of prion protein in patients who received cadaveric growth hormone (160). Data from several groups have suggested that cerebral amyloid angiopathy itself may be transmissible through cadaveric dura matter exposure, termed “iatrogenic cerebral amyloid angiopathy” (89; 10). This is strong evidence that amyloid beta may have seeding properties that allows propagation and accumulation. In fact, the number of iatrogenic cerebral amyloid angiopathy cases reported in the literature has reached 49 (37), and diagnostic criteria have been proposed (11). Interestingly, cerebral vessels, rather than brain parenchymal, are the preferential site of amyloid beta propagation in iatrogenic cerebral amyloid angiopathy (37). The time interval between exposure and symptoms is estimated to be 3 to 4 decades, similar to the gap observed between initial detection of brain amyloid beta deposition in the form of reduced cerebrospinal fluid amyloid beta and first intracerebral hemorrhage among carriers of the Dutch-type hereditary cerebral amyloid angiopathy mutation (37; 103).

Pathogenesis of hemorrhage. Clinicopathologic studies identify several mechanisms through which patients with cerebral amyloid angiopathy may develop microvascular weakening leading to lobar and subarachnoid hemorrhage. Heavily amyloid-laden microvessels appear structurally brittle, with reduced fibrillary collagen in the vascular extracellular matrix, and potentially vulnerable to rupture (211). Maeda and colleagues found spindle-shaped microaneurysms in small cortical arteries with severe amyloid deposition (118). They suggested that (1) amyloid deposition leads to medial and adventitial damage; (2) vessel wall damage leads to microaneurysm formation; (3) plasma components invade the vessel wall leading to areas of fibrinoid necrosis; and (4) hemorrhage develops from rupture of microaneurysm. The finding by Vonsattel and colleagues that, among 17 brains with cerebral amyloid angiopathy, fibrinoid necrosis was present only in the 12 cases with hemorrhage supports the hypothesis that fibrinoid necrosis is the final common denominator determining whether hemorrhage will develop in cases of cerebral amyloid angiopathy (198). Of note, pathologic evidence demonstrates that vessel rupture sites associated with microbleeds, in fact, have no amyloid beta at the precise site of vessel rupture (191), supporting the concept that more than amyloid deposition alone is required to lead to vascular rupture. These vessels have evidence of local vascular remodeling, including fibrinoid necrosis. A neuropathological study examining vessel segments with evidence of vascular remodeling (including decreased vascular amyloid beta and fibrinoid necrosis) demonstrated that local perivascular inflammation was consistently observed surrounding these vessels (107). These findings suggest inflammation may be an essential component in the final steps preceding vascular rupture.

One additional mechanism by which amyloid beta protein may promote intracerebral hemorrhage is by increased local anticoagulant activity; fibrillar amyloid beta protein has been found to bind amyloid precursor protein and potentiate its inhibition of clotting factor XIa (200).

Genetics

Hereditary cerebral amyloid angiopathy. The hereditary cerebral amyloid angiopathies are all associated with alterations of the amyloid precursor protein gene. The gene for amyloid precursor protein resides on the long arm of chromosome 21 and contains 19 exons. The amyloid beta protein fragment of amyloid precursor protein is encoded by parts of exon 16 and 17. The Dutch, Flemish, Italian, Arctic, Iowa, and Piedmont forms of hereditary cerebral amyloid angiopathy are each associated with single nucleotide mutations within the amyloid beta protein-coding region of amyloid precursor protein, resulting in single amino acid changes in amyloid beta protein. The French hereditary form of cerebral amyloid angiopathy is associated with duplication of the amyloid precursor protein gene.

The location of the amyloid precursor protein mutations associated with prominent cerebral amyloid angiopathy contrasts with the amyloid precursor protein mutations associated primarily with familial Alzheimer disease, located immediately flanking rather than within the amyloid beta protein-coding regions. The majority of cases of familial Alzheimer disease are caused by defects in the presenilin genes located on chromosomes 14 (presenilin-1) and 1 (presenilin-2). Advanced cerebral amyloid angiopathy can occur in familial Alzheimer disease due to these gene defects but, in general, is not prominent.

Among nonamyloid beta protein forms of hereditary cerebral amyloid angiopathy, the Icelandic type of hereditary cerebral hemorrhage with amyloidosis is transmitted in autosomal dominant fashion. The cystatin C protein in affected patients lacks the first 10 amino acids and has a Glu-for-Leu substitution at residue 68. The gene for cystatin C lies on chromosome 20p11.2 and contains three exons. All patients with genetic analysis reported to date have demonstrated a single T-2-A transversion in codon 68. Familial oculoleptomeningeal amyloidosis is transmitted in autosomal dominant fashion. Like the more general class of familial amyloid polyneuropathies, these disorders are associated with mutations of the transthyretin gene located on chromosome 18. As multiple distinct transthyretin point mutations have been reported to associate with oculoleptomeningeal involvement in at least some families, the genotype-phenotype correlation of this spectrum of disorders remains to be worked out. However, patients with certain mutations that cause transthyretin amyloidosis develop cerebral amyloid angiopathy (171). Cerebral amyloid angiopathy may continue to develop in the brain in these patients even after therapeutic liver transplants because of choroid plexus production of variant amyloid. Maia and colleagues report a retrospective analysis of 87 consecutive posttransplant ATTR Val30Met FAP patients (119). They report that 31% of the patients developed cerebral amyloid angiopathy type symptoms including transient focal neurologic episodes, seizures, and hemorrhagic and ischemic stroke. Data from a cohort of subjects with most common Val30Met mutation of hereditary transthyretin amyloidosis (ATTR) suggest that cerebral amyloid angiopathy can be visualized in the brain in patients using Pittsburg compound B-based PET scans (172).

The genetic cause of familial British dementia has been identified as mutational loss of a stop codon within the BRI gene on chromosome 13 (194). The absence of the normal stop codon leads to generation of an insoluble 34 amino acid cleavage product that comprises the vascular amyloid deposits in this disorder. The early stages in the disease have been characterized. The most common MRI findings are extensive white matter hyperintensities. Infarction or atrophy of the corpus callosum was also frequently observed. Patients most commonly manifested isolated memory loss on neuropsychological testing. Spastic paraplegia and cerebellar ataxia are reported to be very common findings as the disease progresses (128).

Sporadic cerebral amyloid angiopathy. The major genetic risk factor for sporadic cerebral amyloid angiopathy (as well as sporadic Alzheimer disease) is the apolipoprotein E gene. Apolipoprotein E is located on chromosome 19q13.2 and exists in three major allelic variants differing by single nucleotide and amino acid differences: E2, E3, and E4. Both the apolipoprotein E2 and E4 alleles appear to increase the risk of cerebral amyloid angiopathy-related intracerebral hemorrhage. This situation contrasts with Alzheimer disease, where E4 increases risk but E2 decreases risk. The reason for this discrepancy may lie in differences between the cerebral amyloid angiopathy and Alzheimer disease processes: Apolipoprotein E4 enhances deposition of amyloid beta protein in both vessels and senile plaques, whereas apolipoprotein E2 may promote the breakdown of amyloid-laden vessels, a step that is unique to the cerebral amyloid angiopathy pathway (126). Pathologic studies have suggested that having the apolipoprotein E4 allele appears to increase severity of cerebral amyloid angiopathy (210). Although the apolipoprotein E2 genotype appears to be more common in cerebral amyloid angiopathy patients with intracerebral hemorrhage, the apolipoprotein E4 genotype is more common in patients with cognitive impairment (47).

A population-based study of hemorrhagic stroke in the greater Cincinnati/northern Kentucky region found possession of an E2 or E4 allele to confer 2.3-fold increased odds for lobar hemorrhage (205). The 29% attributable risk associated with apolipoprotein E genotype made it the single greatest attributable risk factor for lobar hemorrhage in this study. An analysis of 2189 individuals with intracerebral hemorrhage and 4041 control subjects drawn from seven international centers confirmed associations of lobar hemorrhage with apolipoprotein E2 (odds ratio 1.82) and E4 (odds ratio 2.2) at the extremely high levels of statistical significance typically used in genome-wide association studies (p< 1 x 10-9 for both alleles) (23). The same alleles are also associated with earlier hemorrhage recurrence. In a prospective study of 71 consecutive elderly patients who survived lobar hemorrhage, those who carried apolipoprotein E2 or E4 had a 2-year cumulative rate of recurrence of 28% compared to 10% for those with the common apolipoprotein E3/E3 genotype (141). The E2 allele further confers risk for larger hemorrhage volumes and worse outcomes following lobar intracerebral hemorrhage (17) suggesting specific effects of this genetic variant on the vascular structure in cerebral amyloid angiopathy.

Even adjusting for apolipoprotein E genotype, family history of hemorrhagic stroke remains another risk factor for lobar hemorrhage (205). A novel candidate genetic factor is the CR1 gene rs6656401 polymorphism, which was found to be associated with cerebral amyloid angiopathy-related lobar hemorrhage (odds ratio 1.61), risk of recurrent hemorrhage (hazard ratio 1.35), and severe cerebral amyloid angiopathy pathology at autopsy (odds ratio 1.34) (22). Heritability estimates suggest that, beyond the apolipoprotein E genotype, there is a substantial genetic contribution to risk of intracerebral hemorrhage in cerebral amyloid angiopathy (55). Future studies are, thus, likely to identify further genetic risks for sporadic cerebral amyloid angiopathy.

Neuropathology. Identification of cerebral amyloid angiopathy at autopsy is common in asymptomatic elderly individuals. Tomonaga found an incidence of 36% in an autopsy series of normal brains (179). The incidence of cerebral amyloid angiopathy in asymptomatic individuals increases with age. The prevalence of cerebral amyloid angiopathy pathology is even higher in the presence of accompanying Alzheimer disease (208). Cerebral amyloid angiopathy is noted in up to 92% of brains from patients with Alzheimer disease and is widely accepted as one of four microscopic features that, when present in abundance, allow for the neuropathologic diagnosis of Alzheimer disease. It is nonetheless important to note that cerebral amyloid angiopathy can occur without other Alzheimer pathologies, and Alzheimer disease can occur without evidence of cerebral amyloid angiopathy (27; 28). Patients with combined Alzheimer disease and cerebral amyloid angiopathy are at risk for lobar hemorrhage in addition to dementia. Evidence suggests that patients with Alzheimer disease and cerebral amyloid angiopathy appear to have an increased frequency of TFNEs (159). These symptoms might help identify patients with the disease without lobar hemorrhage in memory clinic populations. In a neuropathologic study of 117 brains with definite Alzheimer disease, the one quarter with moderate to severe cerebral amyloid angiopathy had a higher incidence of hemorrhages and ischemic lesions than those with less severe or absent cerebral amyloid angiopathy (61).

Cerebral amyloid angiopathy can be strongly suspected in brain tissue by examination of routine (hematoxylin- and eosin-stained) sections. Under routine staining, cerebral amyloid angiopathy appears in cortical microvessels as an effacement of the normal arteriolar smooth muscle cell component, which is replaced by a hyaline eosinophilic material. Its existence and extent can be confirmed by using cytochemical techniques, such as Congo red with polarization microscopy and thioflavin S or T with fluorescence microscopy. Although immunohistochemical techniques are more specific, conventional cytochemical methods (eg, Congo red) are still valuable tools for screening brain tissue removed at autopsy or biopsy for the presence of cerebral amyloid angiopathy.

For definitive identification of cerebral amyloid angiopathy, histochemical methods have been largely superseded by immunohistochemical staining with primary antibodies to the major amyloid proteins: amyloid beta protein (for sporadic cerebral amyloid angiopathy or familial disease involving amyloid precursor protein mutation) or the specific protein deposits characteristic of other hereditary cerebral amyloid angiopathy syndromes such as transthyretin and cystatin C. Most antibodies work well in paraffin-embedded (therefore archival) material, and many primary antibodies are commercially available. Of interest is the observation that anti-cystatin C antibodies also show some labeling of vascular amyloid in sporadic cerebral amyloid angiopathy (196).

Some patients with severe cerebral amyloid angiopathy develop superimposed cerebral amyloid angiopathy–related vascular remodeling, including microaneurysm formation, vessels with, a “lumen within lumen” appearance, and fibrinoid necrosis (120; 137).

Fibrinoid necrosis especially has been implicated in the occurrence of (fatal) cerebral amyloid angiopathy-related intracerebral hemorrhages (198). Ultrastructural studies of cerebral amyloid angiopathy, though few (because optimally preserved biopsy material is usually required for definitive interpretation of findings), show destruction of smooth muscle cells in affected arterioles, with deposition of characteristic 7 to 10 nm amyloid filaments in place of the smooth muscle and surrounding or encasing remaining smooth muscle cells. Surprisingly, the endothelium appears relatively normal in some instances of arteriolar cerebral amyloid angiopathy (197), though it may be severely damaged in the case of capillary amyloid deposition.

In cases of cerebral amyloid angiopathy-related inflammation, the vascular amyloid is associated with a perivascular inflammatory response or true transmural vasculitis, often of a giant-cell type. In these individuals, a variably severe granulomatous inflammatory response develops in response to the cerebral amyloid angiopathy. The inflammatory response includes macrophage and T-cell components (CD4+ and CD8+ cells), and both leptomeningeal and cortical parenchymal microvessels may be involved. Adjacent brain parenchyma usually shows infarcts or microhemorrhages.

In summary, based on the aforementioned spectrum of pathological changes related to cerebral amyloid angiopathy, neuropathology examination serves as the reference standard for the definite diagnosis of cerebral amyloid angiopathy. To this end, two main classification systems have been proposed. The most widely used in the clinical setting is a modified version of the 5-point cerebral amyloid angiopathy severity scale system proposed by Vonsattel and colleagues in 1991 (198; 77). In this scale, vessels in H&E and amyloid beta immunohistochemical staining sections are graded in the following way:

Grade 0 = absence of amyloid beta vascular staining
Grade 1 = presence of patchy amyloid beta vascular staining in an otherwise normal-appearing vessel
Grade 2 = thickened vessel wall, with media completely replaced by amyloid beta
Grade 3 = splitting of the vessel wall affecting at least 50% of vessel diameter in addition to the complete replacement of the media by amyloid beta, also known as “vessel-in-vessel” or “lumen-within-lumen” appearance
Grade 4 = amyloid beta–laden vessels with evidence of fibrinoid necrosis

The most advanced degree observed in leptomeningeal or parenchymal vessels defines the final grade. Grade 2 or greater in full brain autopsies and grade 1 or greater in brain biopsies or hematoma evacuations are considered diagnostic of cerebral amyloid angiopathy (36).

A new consensus protocol for neuropathological cerebral amyloid angiopathy assessment has been validated, and variations of this protocol are frequently used in research contexts (114). In this protocol, parenchymal and meningeal vessels are separately graded on a 0–3 scale in each lobe (0 = no amyloid beta, 1 = scant amyloid beta, 2 = some circumferential amyloid beta, and 3 = widespread circumferential amyloid beta); capillary cerebral amyloid angiopathy is defined as present or absent; and vasculopathy is graded on a 0–2 scale (0 = absent, 1 = occasional vessel, 2 = many vessels) (114).

Epidemiology

Primary intracerebral hemorrhages account for 10% of all strokes in Western populations and a higher proportion among Asians. Clinical and autopsy series from different populations suggest cerebral amyloid angiopathy is responsible for 10% to 34% of these intracerebral hemorrhages (92; 75). The mean age of patients with cerebral amyloid angiopathy-related hemorrhage is in the 70s.

The most important epidemiological risk factor for sporadic cerebral amyloid angiopathy is age. In an autopsy study of 128 subjects, Tomonaga found cerebral amyloid angiopathy in 8% of subjects in the seventh decade of life, 20% in the eighth, 37% in the ninth, and 58% in the tenth, with an overall incidence of 36% in normal subjects aged 60 to 97 (179). Other studies confirm the high prevalence in older individuals and that moderate to severe cerebral amyloid angiopathy is present in 66% of elders who were nondemented during life (05). Sporadic cerebral amyloid angiopathy is rarely seen in persons under 50 years old. There is no definite age-adjusted sex predilection of cerebral amyloid angiopathy. Some early studies reported a female predominance, but this likely reflected an older mean age of women in these series. With the exception of the hereditary forms of cerebral amyloid angiopathy, no evidence is seen of ethnic or geographic predilection.

Although cerebral amyloid angiopathy–related hemorrhage is not always associated with hypertension (205), antihypertensive treatment appears to reduce hemorrhage occurrence (204; 03). An observational study showed that elevated blood pressure is associated with increased recurrent lobar intracerebral hemorrhage risk in patients with probable or possible cerebral amyloid angiopathy (16). These results are in line with animal data suggesting that elevated blood pressure causes hemorrhage with decreased latency in amyloid transgenic mice compared to normal mice (146). A study demonstrated that total small vessel disease burden in cerebral amyloid angiopathy (approximated by combining the most common neuroimaging features in the disease) correlated with presence of vasculopathic changes on pathology (a hallmark sign of severity) and presentation with intracerebral hemorrhage (OR 2.40 [95% CI 1.06-5.45], p=0.035 and 2.23 [95%CI 1.07-4.64], p=0.033, respectively) (46). Neither diabetes mellitus nor coronary atherosclerosis, also prominent vascular risk factors, appears linked to cerebral amyloid angiopathy–related intracerebral hemorrhage. Other potential risks for lobar hemorrhage identified in epidemiological studies are previous ischemic stroke, frequent alcohol use, and low serum cholesterol. For the moment, it remains unclear whether these factors are associated cerebral amyloid angiopathy-related hemorrhages.

The role of genetic factors and family history as risks for cerebral amyloid angiopathy and lobar intracerebral hemorrhage is discussed in the Pathogenesis and pathophysiology section of this article.

Prevention

In contrast to other forms of stroke, modifiable risk factors to prevent cerebral amyloid angiopathy–related outcomes have not been identified. Preliminary evidence suggests that in cerebral amyloid angiopathy patients with lobar intracerebral hemorrhage, elevated mean blood pressure predicts increased risk of intracerebral hemorrhage recurrence (16). Emerging evidence has linked increased blood pressure variability to elevated risk of cerebral small vessel disease, stroke, and dementia above and beyond the role of mean blood pressure levels. However, it remains unclear whether blood pressure variability may also contribute to subclinical disease course in cerebral amyloid angiopathy patients.

Preliminary evidence suggests that variability in blood pressure can influence tau burden in the brain (117). The total effect of blood pressure variability on cognitive progression was significantly mediated by both vascular pathology (defined as the summary score of microinfarcts, white matter rarefaction, atherosclerosis of the Circle of Willis, and arteriolosclerosis) and neurofibrillary tangle pathology (assessed by Braak stage), showing significant indirect effects. Additionally, the mediating role of putative ischemic vascular pathology appeared more pronounced among cognitively normal participants, whereas the mediating role of neurofibrillary tangles was strongly involved in both progression from normal cognition to mild cognitive impairment and dementia and from mild cognitive impairment to dementia. The data suggest that blood pressure variability may play a strong role in amyloid angiopathy–related cognitive impairment and disease pathology. Further studies are ongoing.

Differential diagnosis

The differential diagnosis for the intracerebral hemorrhage lesions associated with cerebral amyloid angiopathy includes the following:

The differential diagnosis for the advanced white matter disease that can result from cerebral amyloid angiopathy includes the following:

Confusing conditions

The differential diagnosis of cerebral amyloid angiopathy–related hemorrhage includes the following:

These diagnoses should be considered particularly when seeing patients of younger age (< 55 years old, consider arteriovenous or cavernous malformation–related hemorrhages, RCVS); patients with prodromal symptoms, such as headache (seen in cerebral venous thrombosis, RCVS); patients with stroke risk factors (atrial fibrillation, which may suggest ischemic stroke with hemorrhagic transformation); patients with poorly controlled hypertension or other vascular risk factors, such as diabetes (hypertensive-related hemorrhage); or patients with existing cancer (a metastatic hemorrhage–prone tumor to the brain). Neuroimaging features, including MRI and CT-based imaging, may help to further distinguish these possibilities. Angiography studies (CT and MRI angiograms or conventional angiography) may be able to help identify arteriovenous malformations.

Diagnostic workup

MRI and MRA scans add supportive evidence for the presence of cerebral amyloid angiopathy. MR images exclude an underlying tumor or vascular malformation more definitively than CT. Moreover, MRI allows the use of gradient refocused echo sequences (variously referred to as gradient-echo, magnetic susceptibility, or T2-weighted) that are particularly helpful because of their sensitivity to hemosiderin deposits from old hemorrhages, large or small.

Gradient-echo MRI is particularly well suited to identify the multiple lobar hemorrhagic lesions required to meet the criteria for probable cerebral amyloid angiopathy and increase the level of diagnostic certainty (76). It should be noted that gradient-echo MRI is sensitive to deoxyhemoglobin as well as hemosiderin, making it an effective technique for the diagnosis of acute as well as chronic hemorrhage (98). Cerebral angiography does not aid the diagnosis of cerebral amyloid angiopathy except by more firmly ruling out other underlying vascular disorders. The presence of a “mixed” hemorrhagic pattern on gradient-echo MRI with regard to topographical distribution (ie, the coexistence of a lobar hemorrhage with deeply located cerebral microbleeds) remains a difficult clinical scenario with respect to diagnosing the predominant underlying small vessel disease (144; 180).

The diagnosis of cerebral amyloid angiopathy should be suspected in elderly individuals with symptomatic lobar intracerebral hemorrhage. The presence of strictly lobar cerebral microbleeds further increases the likelihood that advanced cerebral amyloid angiopathy is present. A definite diagnosis of cerebral amyloid angiopathy still requires a pathologic specimen. The Boston Criteria for diagnosis of cerebral amyloid angiopathy were designed to deal with this situation by recognizing various levels of diagnostic certainty (102). Diagnoses of definite cerebral amyloid angiopathy or probable cerebral amyloid angiopathy with supporting pathology, the highest levels of diagnostic certainty, require full postmortem exam showing pathologically severe cerebral amyloid angiopathy (for definite cerebral amyloid angiopathy) or a sample of brain tissue obtained by biopsy or hematoma evacuation showing some degree of cerebral amyloid angiopathy (for probable cerebral amyloid angiopathy) without other diagnostic lesion. In the more commonly encountered clinical situation in which brain tissue is unavailable, the in vivo diagnosis of probable cerebral amyloid angiopathy can be reached through a set of clinical and neuroimaging criteria, which have evolved over time. According to the first version of the Boston criteria (version 1.0), developed in the 1990s, multiple hemorrhages restricted to lobar regions in older individuals, not attributable to any other causes, were indicative of probable underlying cerebral amyloid angiopathy pathology in the absence of other definite causes of hemorrhage, such as head trauma or ischemic stroke, brain tumor, vascular malformation, vasculitis, blood dyscrasia, or coagulopathy (113). These criteria emanate from the most distinctive features of cerebral amyloid angiopathy-related hemorrhages: their tendency to be multiple, recurrent, and localized to cortical or corticosubcortical (ie, lobar) rather than deep hemispheric brain regions. In validating studies, 13 of 13 diagnoses of probable cerebral amyloid angiopathy were corroborated by neuropathologic exam, suggesting reasonably high specificity (102) and sensitivity was very high among clinically symptomatic patients known to carry the mutation for Dutch hereditary cerebral amyloid angiopathy (186). The diagnostic sensitivity of the Boston Criteria probable cerebral amyloid angiopathy category was further increased by including the presence of radiographically detected superficial sulcal/meningeal blood products as an additional site of bleeding when determining whether multiple hemorrhages are present (113), named the modified Boston criteria (version 1.5). The diagnosis of possible cerebral amyloid angiopathy is used for situations that are less suggestive, such as a single lobar hemorrhage without other definite cause.

Identification of probable or possible cerebral amyloid angiopathy depends on imaging of hemorrhages, which may be acute or chronic, and may represent sizable symptomatic hemorrhage or small (< 0.5 cm diameter) asymptomatic microbleeds. Acute lobar intracerebral hemorrhage is most commonly diagnosed with CT scan.

Validation of these criteria has been further extended to include patients with lobar microbleeds in the absence of intracerebral hemorrhage (122). It should be stated that although these criteria appear to have a high positive predictive value for cerebral amyloid angiopathy in patients with clinical symptoms, a study showed that microbleeds alone may have less predictive value for cerebral amyloid angiopathy in population-based cohorts (51).

To incorporate the more recently described and non-hemorrhagic white matter neuroimaging markers of cerebral amyloid angiopathy, named the enlarged perivascular spaces in the centrum semiovale and the multi-spot subcortical pattern of white matter hyperintensities, an updated version of the Boston criteria (version 2.0) has been proposed and validated among several multicenter cohorts of symptomatic individuals.

Table 1. Boston Criteria Version 2.0 For Diagnosis of Cerebral Amyloid Angiopathy

These new criteria showed superior accuracy compared to previous versions, improving diagnostic sensitivity without excessively compromising specificity. According to these new criteria, only one lobar hemorrhagic feature (either an intracerebral hematoma, a microbleed, or region of cortical superficial siderosis), in addition to either the presence of severe perivascular spaces in the centrum semiovale or the presence the multi-spot subcortical pattern of white matter hyperintensities, fulfills criteria for probable cerebral amyloid angiopathy in symptomatic individuals. According to these new criteria, the presence of two foci of cortical superficial siderosis also fulfills neuroimaging criteria for probable cerebral amyloid angiopathy. The presence of any deep hemorrhagic lesion remains a definite exclusion criterion in this updated version. It should be emphasized that none of the imaging findings included in the Boston criteria version 2.0 are perfectly sensitive or specific for cerebral amyloid angiopathy (37). Therefore, they should be interpreted in the intended clinical context. Patients must be over 50 years of age and present with either spontaneous intracerebral hemorrhage, TFNEs, cognitive impairment, or dementia to be eligible according to the Boston criteria v2.0. It remains unknown how these criteria perform in asymptomatic populations or in individuals with incidental findings (37).

CT-based criteria for making a diagnosis of cerebral amyloid angiopathy–associated lobar intracerebral hemorrhage in the acute setting have been suggested (the Edinburgh criteria) (161). Three predictors of cerebral amyloid angiopathy–associated lobar intracerebral hemorrhage diagnosis were included in two sets of rule-in or rule-out criteria: subarachnoid hemorrhage and finger-like projections from intracerebral hemorrhage on CT (among the most frequently reported CT features of cerebral amyloid angiopathy–associated lobar intracerebral hemorrhage in clinical routine) and APOE ε4 possession. In combination, these features appear to have strong predictive value for cerebral amyloid angiopathy (161). These criteria may be particularly useful in settings where MRI is scarcely available, such as low-to-middle income countries, and precisely where the incidence of intracerebral hemorrhages is higher (161). Because regular genetic testing is also unavailable worldwide, a simplified CT-based model of the Edinburgh criteria including only evidence of subarachnoid hemorrhage and finger-like projections and excluding APOE genotyping was tested and demonstrated high diagnostic accuracy (161). Patients who are not eligible for MRI, such as those with noncompatible hardware (eg, non-MRI conditional cardiac implantable electronic devices, metallic intraocular foreign bodies, drug infusion pumps, among others) may also benefit from the CT-based Edinburgh criteria.

The emerging approach of imaging amyloid beta deposits in vivo with high-affinity ligands developed for Alzheimer disease, such as the Pittsburgh compound B (PiB), has also been investigated in cerebral amyloid angiopathy (101). Two studies of PET imaging using Pittsburgh Compound B (PiB) performed on subjects with probable cerebral amyloid angiopathy found significantly increased PiB uptake relative to healthy elderly control subjects and a significantly elevated ratio of occipital-to-global PiB uptake relative to subjects with Alzheimer disease (97; 115). Concerning the value and accuracy of amyloid-PET in diagnosing patients with sporadic cerebral amyloid angiopathy, a meta-analysis reported a pooled sensitivity of 79% and specificity of 78% (35). Further studies have also indicated increased PiB retention at both current sites of cerebral amyloid angiopathy-related hemorrhagic lesions (56) and at sites of future bleeding in subjects who underwent follow-up MRI a median of 19 months after PiB-PET (81). The FDA-approved amyloid ligand Florbetapir has been shown to distinguish cerebral amyloid angiopathy-related hemorrhage from hypertensive hemorrhage, opening the question as to whether this tool could be used clinically to diagnosis cerebral amyloid angiopathy in select cases (80; 153). However, analyses could have been confounded by the concomitance of vascular and parenchymal amyloid due to the frequent coexistence of cerebral amyloid angiopathy and Alzheimer disease pathology. One study did not find evidence of regional cerebral amyloid angiopathy burden on pathology contributing to antemortem regional PiB-PET signal (127). In fact, among patients with minimal amyloid beta plaque pathology, no association was found between regional cerebral amyloid angiopathy and PiB-PET binding, suggesting that amyloid beta PET imaging is predominantly driven by parenchymal amyloid and is not likely a direct biomarker of cerebral amyloid angiopathy pathology. More studies are needed to better define the utility of amyloid beta PET imaging in cerebral amyloid angiopathy.

Neuropathological verification of cerebral amyloid angiopathy is most commonly accomplished by examination of tissue obtained from the wall of an evacuated lobar hematoma or postmortem exam. Brain biopsy is usually not pursued when clinical suspicion of cerebral amyloid angiopathy is high, as confirming a clinically probable diagnosis generally does not lead to major alterations in therapy. The exception to this situation is often those subjects with cerebral amyloid angiopathy-related inflammation, where biopsy can both establish a firm basis for immunosuppressive treatment and exclude other potential causes of brain inflammation. A brain specimen obtained for suspected cerebral amyloid angiopathy should include leptomeningeal tissue whenever possible. Brain tissue is examined by routine histologic stains (hematoxylin and eosin, Congo red) as well as immunostaining for amyloid beta protein, as indicated. In a study in which biopsy was simulated on postmortem brains with known cerebral amyloid angiopathy severity, the identification by Congo red staining of any degree of vascular amyloid in the biopsy specimen was highly sensitive for the diagnosis of definite cerebral amyloid angiopathy. The specificity for this diagnosis increased with greater severities of cerebral amyloid angiopathy in the biopsy specimen (77).

Cerebrospinal fluid markers may have an ancillary role in the diagnostic evaluation of cerebral amyloid angiopathy. In Alzheimer disease, analysis of cerebrospinal fluid has found Aß42 concentration to be lowered and concentration of the microtubule-associated protein tau to be elevated (53). Analysis of cerebrospinal fluid in 17 nondemented individuals with probable cerebral amyloid angiopathy found reduced concentrations of both Aß42 and Aß40 with normal or mildly increased tau (193). A meta-analysis suggests that decreased Aß40 levels can distinguish cerebral amyloid angiopathy from Alzheimer disease (43). Cerebrospinal fluid assays for Aß42 and tau are commercially available and may be considered when seeking confirmation of cerebral amyloid angiopathy.

Genetic testing in cerebral amyloid angiopathy is useful to identify mutations associated with one of the hereditary syndromes. Furthermore, APOE genotype analysis demonstrating at least one e4 allele (APOE e4) is strongly associated with moderate or severe cerebral amyloid angiopathy on pathology and has, thus, been incorporated in the Edinburgh criteria for in vivo diagnosis of cerebral amyloid angiopathy (161).

Cerebral amyloid angiopathy-related inflammation is associated with a distinct set of diagnostic findings. In addition to its defining neuropathologic appearance, this syndrome is also characterized by the striking MRI finding of asymmetric white matter hyperintensities involving periventricular and subcortical white matter and sometimes reaching the overlying cerebral cortex. Multiple lobar microbleeds are often present on gradient-echo sequences, not necessarily restricted to the cortical lobes affected by the T2 hyperintensities. Typically, minimal gadolinium enhancement is seen (although both leptomeningeal and parenchymal enhancement have been reported), with no diagnostic finding on MR-, CT-, or catheter-based angiography, presumably reflecting the small size of the inflamed vessels. An intriguing, though still unexplained, association has been noted between cerebral amyloid angiopathy-related inflammation and the apolipoprotein E4/E4 genotype (100).

Management

Management of acute lobar hemorrhage due to cerebral amyloid angiopathy is similar to hemorrhages of other etiologies. Previous studies suggested that evacuation could benefit patients with the lobar or superficial hemorrhages most commonly encountered in cerebral amyloid angiopathy (130; 131). Despite concerns that brain surgery might be especially hazardous in cerebral amyloid angiopathy because of an increased risk of perioperative bleeding, published series indicated that resections could be performed without elevated postoperative risk (133; 93; 212). The Early MiNimally-invasive Removal of IntraCerebral Hemorrhage (ENRICH) trial (NCT02880878), a multicenter adaptive trial, compared standard of care clinical management against early (< 24h) surgical hematoma evacuation through minimally invasive surgical procedures in patients with acute spontaneous, primary supratentorial intracerebral hemorrhage. This trial was the first to demonstrate the functional benefit of surgical hematoma evacuation. Surgical intervention was associated with improved functional outcome, measured using the modified Rankin scale, and lower 30-day mortality rates (results presented at the 2023 European Stroke Organization Conference). Current medical management for acute intracerebral hemorrhage includes blood pressure control, supportive medical care, discontinuation of antiplatelet agents, and reversal of anticoagulation. Osmotic agents and hyperventilation may be employed for temporary amelioration of raised intracranial pressure. The third multicenter Intensive Care Bundle with Blood Pressure Reduction in Acute Cerebral Haemorrhage Trial (INTERACT3) demonstrated improved functional outcome for patients with acute intracerebral hemorrhage treated with a care bundle protocol that included early control of blood pressure, blood glucose, and body temperature and reversal of abnormal anticoagulation (116).

Currently, no active treatments for prevention of hemorrhage recurrence in cerebral amyloid angiopathy have been definitively established. However, some measures can reasonably be taken. Careful control of hypertension appears to be a prudent course in patients with cerebral amyloid angiopathy. This approach is supported by a secondary analysis of the Perindopril Protection against Recurrent Stroke Study (PROGRESS), in which stroke patients randomized to treatment with perindopril plus optional indapamide demonstrated lower risk of any intracerebral hemorrhage (50% risk reduction, 95% CI 26% to 67%) as well as specifically lower risk of intracerebral hemorrhages meeting criteria for probable cerebral amyloid angiopathy (77% risk reduction, 95% CI 19% to 93%) (03).

Anticoagulation increases both the likelihood and severity of hemorrhagic stroke and in general should be avoided in patients with cerebral amyloid angiopathy-related hemorrhage. Although data from randomized controlled trials are lacking, a decision-analysis model suggested that for patients with cerebral amyloid angiopathy-related hemorrhage, the increased risk of catastrophic recurrent hemorrhage associated with long-term anticoagulation outweighs the beneficial effects of preventing atrial fibrillation-related thromboembolic stroke (59). However, considering observational data suggesting that anticoagulation after intracerebral hemorrhage may be associated with improved outcomes (20) and equivocal data from SoSTART and APACHE-AF trials, future randomized controlled trials, including the ongoing ASPIRE trial (NCT03907046) and ENRICH-AF trial (NCT03950076), are required to definitively address this important issue (44). Antiplatelet agents may also increase the risk of hemorrhage, but the effect is likely substantially smaller. The RESTART trial randomized patients with recent intracerebral hemorrhage on aspirin for secondary prevention at the time of hemorrhage to continue aspirin or stop aspirin (158). No increase in the rates of major hemorrhage (including recurrent intracerebral hemorrhage) was observed. Therefore, it is a reasonable practice to consider continuing antiplatelet medications for secondary prevention and otherwise withhold them from patients with cerebral amyloid angiopathy–related hemorrhage. The risk of antiplatelet and anticoagulant therapy may be increased in the presence of an apolipoprotein E2 allele (125; 164). Similar questions have been raised about use of statin medications, which appear to increase the risk of hemorrhagic stroke (while decreasing the overall risk of stroke) when used for secondary stroke prevention (02; 69). The ongoing SATURN trial (NCT03936361) will help to clarify the safety of statins in this patient population. Prudent practice in treating patients with cerebral amyloid angiopathy is to use statins when clearly indicated for prevention of myocardial infarction or ischemic stroke but to withhold them in the absence of a definite indication.

Although immunosuppressive therapy has no established role for noninflammatory cerebral amyloid angiopathy, it leads to clinical and radiologic improvement in the majority of patients with the syndrome of cerebral amyloid angiopathy-related inflammation. In a follow-up study of 12 patients with cerebral amyloid angiopathy-related inflammation, 10 patients had initial improvement to immunosuppression but three of those 10 had subsequent relapse of symptoms, whereas the remaining two patients showed no response. The volume of T2 hyperintensities generally paralleled clinical improvements and subsequent relapses, making MRI a reasonable marker of treatment effect (138). Data also suggest that treatment of cerebral amyloid angiopathy–related inflammation with immunosuppression decreases relapse rates (155).

Reasonable choices for immunosuppressive regimen are a 5-day course of high-dose corticosteroids (500 mg-1 g/day x 5 days) followed by a slow PO steroid taper (over months). If there is no radiographic or clinical change, cyclophosphamide (1-2 mg/kg/day PO for 2 weeks) could be considered (consideration of brain biopsy to confirm diagnosis prior to initiating treatment is recommended). Other immunosuppressive agents may be further alternatives in high refractory cases, although clinical experience with these agents remains limited.

There are many exciting prospects for novel disease-modifying treatments for cerebral amyloid angiopathy. Cerebral amyloid angiopathy has a defined set of steps in its pathogenesis, including amyloid beta production, perivascular clearance of amyloid beta, vascular amyloid beta accumulation, vascular remodeling, and perivascular inflammation, each of which is a potential therapeutic target. Given improvements in identifying the presence and progression of cerebral amyloid angiopathy and the high volume of investigations into treatment approaches for amyloid beta protein-related disease, future years are likely to see multiple trials of biologically plausible candidate agents aimed at slowing this currently untreatable form of cerebrovascular disorder.

Special considerations

Pregnancy

No evidence indicates increased risk in pregnancy in patients with cerebral amyloid angiopathy. Except in some familial forms, most patients will be beyond childbearing years.

Anesthesia

No evidence exists to show increased risk from anesthesia in patients with cerebral amyloid angiopathy.