Cerebral amyloid angiopathy (CAA) is a disorder in which amyloid proteins accumulate in the walls of small and medium-sized blood vessels of the brain and its coverings. In its common, age-associated form, the deposited material consists predominantly of amyloid-beta peptides. Affected vessels become structurally abnormal and susceptible to bleeding, impaired blood flow, and inflammatory injury. CAA is an important cause of spontaneous lobar intracerebral hemorrhage in older adults and contributes to cognitive impairment both independently and alongside Alzheimer’s disease.
The disease ranges from microscopic deposits discovered incidentally at autopsy to recurrent disabling hemorrhages. Other manifestations include small cortical infarcts, brief spreading neurological episodes, blood leakage over the brain surface, and a less common inflammatory syndrome. Modern magnetic resonance imaging (MRI) allows characteristic patterns of vascular injury to be recognized during life, although definitive demonstration of vascular amyloid requires examination of tissue. Its clinical importance extends beyond hemorrhagic stroke to the interpretation of dementia biomarkers and the safety of antithrombotic and amyloid-targeting treatments.
Vascular pathology and anatomical distribution
CAA principally affects arteries and arterioles in the cerebral cortex and leptomeninges, the delicate layers of the meninges covering the brain. Capillaries are also involved in some cases. This distribution distinguishes typical CAA from hypertensive arteriopathy, which preferentially damages small penetrating vessels supplying deep structures such as the basal ganglia, thalamus, and brainstem. Amyloid deposition is often patchy, with posterior cortical regions, especially the occipital lobes, commonly showing substantial involvement. The distribution is not restricted to a single vascular territory.
On histopathological examination, amyloid forms extracellular deposits within vessel walls. Congo red staining produces characteristic apple-green birefringence under polarized light, while immunohistochemical staining identifies the deposited protein. Early deposits may be limited and circumferential; advanced disease replaces vascular smooth muscle and disrupts the normal layered architecture. Associated abnormalities include splitting of the vessel wall, fibrinoid degeneration, local dilatation, and microaneurysm formation. These changes weaken vessels, but the amount of amyloid alone does not fully explain which vessels rupture.
A pathological classification separates capillary-involving CAA, often called type 1, from CAA without capillary involvement, called type 2. Both can involve cortical and leptomeningeal arteries. Capillary involvement has associations with particular genetic backgrounds and with Alzheimer-related pathology, but these pathological categories are not equivalent to distinct, reliably separable clinical syndromes. Routine imaging cannot directly determine whether capillary amyloid is present.
The most recognizable consequences are hemorrhagic, but the tissue injury is broader. Cortical microinfarcts, loss of vascular reactivity, disruption of the blood–brain barrier, and injury to underlying white matter may accompany vascular amyloid. Some microscopic infarcts are too small for conventional clinical MRI to detect. Consequently, a person can have substantial CAA-related brain injury without a history of a large hemorrhage.
CAA is usually a disease localized to the central nervous system rather than one manifestation of systemic amyloid deposition. The term “amyloid” describes a shared protein-folding architecture, not a single disease or protein. Common age-associated amyloid-beta CAA therefore differs biologically from systemic light-chain or transthyretin amyloidosis. Rare cerebral angiopathies caused by other amyloid proteins also exist, making identification of the deposited protein especially relevant in unusual hereditary presentations.
Amyloid production, clearance, and vessel injury
Amyloid-beta is generated by enzymatic cleavage of amyloid precursor protein (APP), a membrane-associated protein expressed in the nervous system and other tissues. Sequential processing by beta-secretase and gamma-secretase produces peptides of different lengths. Amyloid-beta containing 40 amino acids, usually abbreviated Aβ40, predominates in many vascular deposits; the 42-amino-acid peptide, Aβ42, is relatively enriched in the parenchymal plaques characteristic of Alzheimer’s disease. Both peptides may occur in either location, so this distinction is one of predominance rather than exclusivity.
CAA is closely connected to the clearance of amyloid-beta from brain tissue. Soluble peptides can be degraded enzymatically, transported across the blood–brain barrier, or removed along pathways associated with interstitial fluid and cerebrospinal fluid. One influential model emphasizes intramural periarterial drainage: solutes move along basement membranes within the walls of capillaries and arteries. Aging and vascular structural changes may reduce the efficiency of this route, favoring deposition along the same pathways used for clearance.
Deposition can then further damage clearance mechanisms. Amyloid accumulation injures smooth muscle cells, alters basement membranes, stiffens vessel walls, and reduces normal vascular responses. Experimental and human physiological studies indicate impaired responses to neuronal activity and changes in blood gases. These abnormalities can reduce the ability to match blood supply to local metabolic requirements. Clearance failure and vascular injury may therefore reinforce one another rather than represent a simple one-way sequence.
Oxidative stress, altered endothelial function, and inflammation contribute to vascular dysfunction. Amyloid-beta can interact with vascular cells and immune cells, promoting inflammatory signaling and extracellular-matrix degradation. In severe disease, barrier leakage and wall degeneration can facilitate hemorrhage. At the same time, luminal narrowing and impaired perfusion provide mechanisms for ischemic injury. This combination explains why CAA cannot be understood solely as a bleeding disorder.
Several details remain unresolved. The relative importance of arterial pulsation, smooth-muscle-driven movements, sleep-related fluid dynamics, and other clearance processes continues to be investigated. Experimental observations do not establish that any one drainage pathway accounts for all human disease. Likewise, vascular amyloid burden, inflammation, vessel-wall remodeling, and accompanying arteriolosclerosis interact in ways that make the relationship between deposition and clinical events variable. Some people with extensive deposits never experience a recognized hemorrhage, whereas others develop recurrent bleeding.
Epidemiology, genetics, and uncommon acquired forms
Age is the strongest demographic association of sporadic CAA. The condition is uncommon in younger adults and becomes increasingly frequent in later life. Estimates of prevalence vary widely because autopsy studies, community imaging studies, and hospital cohorts measure different aspects of disease. Microscopic vascular amyloid is more common than clinically symptomatic CAA, and small deposits do not necessarily imply the same risk as advanced angiopathy with hemorrhagic lesions.
CAA frequently coexists with Alzheimer’s disease, but neither condition is necessary for the other. Autopsy series of Alzheimer’s disease often show vascular amyloid, sometimes extensively. Conversely, CAA can cause lobar hemorrhage or cognitive impairment without the full neuropathological changes of Alzheimer’s disease. Estimates derived from dementia clinics or brain banks therefore cannot be applied directly to the general population. Selection of participants, age at death, tissue sampling, and grading methods all influence reported frequencies.
The apolipoprotein E gene, APOE, is an important genetic modifier. The ε4 allele is associated with increased vascular amyloid deposition, while ε2 has been associated with vessel-wall degenerative changes and hemorrhagic manifestations. These are probabilistic associations, not deterministic tests. Many affected individuals carry neither allele, and most carriers do not develop a clinically recognized CAA syndrome. APOE status alone does not establish the diagnosis.
Rare inherited forms can present substantially earlier than sporadic disease. Some are caused by pathogenic variants in APP and follow autosomal dominant inheritance. The Dutch-type form, associated with the APP E693Q variant, is a well-characterized example featuring recurrent cerebral hemorrhage and cognitive deterioration. Other APP variants produce different mixtures of vascular amyloid, plaques, hemorrhage, and dementia. An inherited Icelandic angiopathy is instead caused by a variant in the cystatin C gene, CST3, and involves a different amyloid protein. These disorders illustrate why cerebral amyloid angiopathy is a pathological category with more than one molecular cause.
An uncommon acquired form has been described after historical exposure to cadaver-derived medical materials, including certain dura mater grafts and pituitary-derived growth hormone preparations. Reported patients may develop amyloid-beta pathology and CAA decades after exposure, sometimes at unusually young ages. The proposed mechanism is transfer of amyloid-beta seeds that promote further aggregation. These observations concern specific historical medical exposures; they do not indicate transmission through ordinary social contact. Their epidemiology and the frequency of clinically significant disease remain less well defined than those of age-associated CAA.
Hemorrhagic and transient neurological presentations
The classic major presentation is spontaneous lobar hemorrhage: bleeding within cortical or immediately subcortical regions of a cerebral hemisphere. Symptoms depend on location and size and may include weakness, visual-field loss, aphasia, headache, seizures, or impaired consciousness. Hemorrhages can extend into nearby subarachnoid spaces or the ventricular system. Compared with hemorrhage caused predominantly by deep perforating-vessel disease, the superficial anatomical distribution provides an important clue to the underlying vascular disorder.
CAA-related hemorrhage may recur, sometimes in a different lobe or hemisphere. Multiple old and new lobar hemorrhages are strongly suggestive in an appropriate clinical setting. However, a single lobar bleed is not specific: vascular malformations, tumors, venous disease, trauma, and other causes remain possible. Hypertension can coexist with CAA and influence hemorrhage risk, so the presence of elevated blood pressure does not identify or exclude the amyloid-related mechanism.
Small-volume bleeding into cortical sulci produces convexity subarachnoid hemorrhage. This differs anatomically from the basal cisternal bleeding commonly associated with a ruptured intracranial aneurysm. In older adults, CAA is an important cause of nontraumatic convexity bleeding. Its symptoms may be relatively focal or transient, and severe thunderclap headache is not obligatory. Other causes of convexity subarachnoid hemorrhage include reversible cerebral vasoconstriction, cerebral venous thrombosis, and inflammatory vascular disorders.
CAA-related transient focal neurological episodes, sometimes called “amyloid spells,” are recurrent brief disturbances of sensation, movement, vision, or language. Symptoms may spread gradually over a limb or from one body region to an adjacent one, often over minutes. Positive symptoms, such as tingling or involuntary movements, and negative symptoms, such as weakness or numbness, can occur. Their stereotyped recurrence and association with superficial blood products help distinguish them from some other transient disorders.
These episodes can resemble a transient ischemic attack, migraine aura, or focal epileptic seizure. A leading mechanistic explanation is cortical spreading depolarization triggered by acute or chronic blood products near the cortical surface; seizures may account for some events. The distinction has clinical consequences because an ischemic interpretation and a hemorrhagic interpretation imply different treatment risks. Observational cohorts show that CAA-related transient episodes, particularly those associated with superficial bleeding, can precede subsequent lobar hemorrhage.
Cognitive impairment and nonhemorrhagic brain injury
CAA can impair cognition even in the absence of a large symptomatic hemorrhage. Its effects reflect multiple forms of tissue injury: cortical microinfarcts, microscopic hemorrhages, white-matter damage, and disruption of distributed neural networks. Large hemorrhages can add focal deficits and abrupt declines, but a gradual course is also possible. Cognitive symptoms may therefore develop through cumulative vascular injury rather than a single recognizable stroke.
Commonly affected domains include processing speed, executive function, and aspects of attention. Memory can also be impaired, especially when Alzheimer-related pathology coexists. No cognitive profile is sufficiently specific to diagnose CAA. The extent and location of lesions, premorbid abilities, accompanying neurodegeneration, and other vascular diseases all influence the clinical picture. A diagnosis of dementia describes functional cognitive decline; it does not by itself identify which of these processes is responsible.
White-matter hyperintensities on MRI are frequent but nonspecific. In CAA, they may reflect chronic vascular dysfunction, secondary degeneration, and impaired tissue perfusion. Their distribution and pattern can add diagnostic information, but similar signal abnormalities occur with aging, hypertensive small-vessel disease, and other conditions. Diffusion-based studies have also demonstrated abnormalities of structural connectivity, supporting a network-level contribution to cognitive dysfunction.
Ischemic lesions range from acute diffusion-positive infarcts to microscopic cortical infarcts visible only on tissue examination or specialized imaging. Their small individual size can obscure their cumulative importance. They may arise from luminal compromise, impaired vascular reactivity, microvascular occlusion, or reduced perfusion. Not every ischemic lesion in a person with CAA is caused by amyloid angiopathy, however; atrial fibrillation, atherosclerosis, and other vascular disorders can coexist.
CAA and Alzheimer’s disease share amyloid-beta biology but differ in the location and consequences of deposition. Alzheimer’s disease is characterized by parenchymal amyloid plaques together with tau-related neurofibrillary pathology, whereas CAA concerns vascular deposits and their effects. This overlap complicates the interpretation of amyloid biomarkers. A positive amyloid scan in a cognitively impaired older adult may reflect contributions from both vascular and parenchymal deposits, and it cannot alone determine the cause of impairment. Tissue and longitudinal imaging studies indicate that vascular amyloid can contribute independently to cognitive decline, while mixed pathology remains common in later life.
Imaging and the Boston diagnostic criteria
Structural imaging provides the principal route to recognizing CAA during life. Computed tomography rapidly demonstrates acute intracerebral hemorrhage and can identify associated subarachnoid blood. MRI adds sensitivity for old, small, and superficial hemorrhagic lesions. T2*-weighted gradient-echo and susceptibility-weighted sequences reveal magnetic effects from blood-breakdown products that may be inconspicuous on conventional sequences. The absence of these sequences can substantially limit an examination’s ability to characterize a hemorrhagic small-vessel disease.
Cerebral microbleeds appear as small rounded susceptibility lesions, generally representing previous microscopic hemorrhage. A strictly lobar distribution supports CAA, whereas deep microbleeds suggest a different or additional arteriopathy. Cortical superficial siderosis appears as curvilinear susceptibility along the brain surface and reflects chronic deposition of iron-containing blood products in superficial tissues. It is particularly important because its presence and extent are associated with future hemorrhage risk. Microbleed counts vary with scanner field strength, sequence sensitivity, and technical parameters.
The Boston criteria combine age, clinical presentation, lesion distribution, and exclusion of other causes to assign diagnostic confidence. The original criteria were published in 2001, followed by a modified version in 2010 and version 2.0 in 2022. These criteria estimate the likelihood of underlying CAA rather than directly visualizing vascular amyloid. Version 2.0 expanded the role of nonhemorrhagic white-matter markers and included a broader range of clinical presentations than earlier versions.
For probable CAA under version 2.0, an individual aged at least 50 years with a qualifying presentation can meet the MRI-based category through either at least two strictly lobar hemorrhagic lesions or one such lesion plus a qualifying white-matter feature. Hemorrhagic lesions include lobar hemorrhage, lobar microbleeds, and foci of cortical superficial siderosis or convexity subarachnoid hemorrhage. The white-matter features are severe enlarged perivascular spaces in the centrum semiovale or a defined multispot pattern of white-matter hyperintensities. Deep hemorrhagic lesions must be absent, and alternative causes must be excluded. Cerebellar hemorrhagic lesions are not counted as either lobar or deep for these criteria.
“Definite CAA” requires a full postmortem pathological assessment meeting specified conditions. Tissue obtained through biopsy or hematoma evacuation can support a pathological diagnostic category but is subject to sampling limitations because deposition is patchy. Diagnostic performance also depends on the population examined: criteria validated in symptomatic specialist cohorts are not interchangeable with population screening tools. Amyloid positron emission tomography and cerebrospinal-fluid measurements are valuable research approaches, but neither reliably separates vascular from parenchymal amyloid on its own.
CAA-related inflammation and amyloid-beta-related angiitis
A minority of patients develop an inflammatory response associated with vascular amyloid. Two related pathological patterns are recognized. CAA-related inflammation commonly describes predominantly perivascular inflammation, while amyloid-beta-related angiitis involves inflammation extending through and damaging the vessel wall, often with granulomatous features. Terminology varies somewhat across studies, and both patterns belong to the broader spectrum of inflammatory amyloid-associated vascular disease.
The clinical course differs from typical noninflammatory CAA. Symptoms commonly evolve over days to weeks and can include cognitive or behavioral change, reduced alertness, headache, seizures, and focal neurological deficits. Large lobar hemorrhage is not required. Because the syndrome may produce subacute encephalopathy rather than an obvious stroke, its differential diagnosis includes infection, neoplasm, other inflammatory brain disorders, and several forms of vasculitis.
MRI usually shows asymmetric areas of vasogenic edema involving subcortical white matter, often extending toward the cortex. Susceptibility sequences commonly reveal accompanying lobar microbleeds or superficial siderosis that point to underlying CAA. Leptomeningeal or parenchymal enhancement may occur after administration of a contrast agent. The combination of asymmetric edema and characteristic hemorrhagic markers is more informative than either feature alone.
Clinicoradiological criteria can support a probable diagnosis when the presentation and imaging pattern are characteristic and competing causes have been excluded. Cerebrospinal fluid may show increased protein or inflammatory cells, but these findings are nonspecific. Tissue examination can establish the inflammatory pattern and exclude mimics when uncertainty remains. A negative small biopsy does not invariably exclude the condition because affected vessels may be unevenly distributed.
Published cohorts describe frequent clinical and radiological improvement following corticosteroid treatment, sometimes with additional immunosuppressive therapy. Relapses occur, and the evidence base consists largely of observational studies rather than large randomized trials. Improvement of edema does not establish elimination of the underlying vascular amyloid. The inflammatory syndrome is nevertheless important because it represents a potentially reversible component of neurological deterioration. Proposed mechanisms include immune recognition of deposited amyloid-beta, with anti-amyloid-beta antibodies implicated in research studies; these remain mechanistic findings rather than a universally established clinical diagnostic test.
Treatment context, hemorrhage prevention, and competing risks
No established treatment removes vascular amyloid and has been shown to prevent the full range of clinical outcomes in sporadic CAA. Clinical management consequently addresses acute complications, modifiable contributors to vascular injury, and the competing risks created by other diseases. Acute CAA-related intracerebral hemorrhage is managed within the broader framework of hemorrhagic stroke care, including assessment of hematoma expansion, anticoagulant effects, neurological deterioration, and complications. The role of neurosurgery depends on the hemorrhage’s characteristics and the overall clinical circumstances rather than the CAA label alone.
Blood pressure is an important modifiable determinant of recurrent intracerebral hemorrhage, including lobar hemorrhage. Observational data and secondary analyses of blood-pressure-lowering trials support its relevance even when amyloid angiopathy, rather than hypertensive arteriopathy, is the presumed underlying pathology. This does not mean hypertension is necessary for CAA to develop. Rather, pressure-related mechanical stress can act on vessels already weakened by amyloid deposition. CAA-specific evidence is less extensive than evidence for intracerebral hemorrhage populations as a whole.
Antithrombotic treatment creates a particularly difficult balance. Anticoagulants reduce embolic stroke in appropriate populations with atrial fibrillation but can increase the consequences of fragile cerebral vessels. Prior lobar hemorrhage, extensive superficial siderosis, and other MRI findings inform estimates of bleeding risk. Trials comparing direct oral anticoagulants with warfarin generally show less intracranial bleeding with the former, but patients with severe CAA or recent lobar hemorrhage have often been underrepresented. These trial averages therefore do not resolve every CAA-specific decision.
Antiplatelet evidence requires similar care. The RESTART randomized trial, published in 2019, investigated restarting antiplatelet treatment after intracerebral hemorrhage occurring during antithrombotic use. It did not demonstrate an increased recurrent hemorrhage risk overall, but its imaging subgroups were not large enough to settle all questions about advanced CAA. Left atrial appendage closure has also been studied as a way to reduce long-term anticoagulant exposure in selected patients, although it entails procedural risks and usually some peri-procedural antithrombotic treatment.
Rehabilitation addresses the functional consequences of hemorrhage and cognitive injury through physical, occupational, language, and cognitive approaches. Seizures, mood symptoms, and swallowing or mobility impairments are treated according to their manifestations. Across these areas, management concerns are not limited to recurrence: disability, cognition, and health-related quality of life are relevant outcomes. The available evidence does not support a single treatment pathway applicable to all patients with incidental imaging abnormalities, symptomatic hemorrhage, inflammatory disease, and mixed neurodegenerative pathology.
Natural history, anti-amyloid therapies, and research directions
CAA has a heterogeneous natural history. Mild pathological deposits may remain clinically unrecognized, whereas symptomatic disease can produce accumulating microvascular injury and recurrent hemorrhage. In cohorts of survivors of CAA-related lobar hemorrhage, recurrent intracerebral hemorrhage rates are often approximately 5–10% per year, although estimates differ with diagnostic definitions, follow-up, and patient selection. This range does not describe the risk of an asymptomatic person with incidental lobar microbleeds.
Cortical superficial siderosis, especially when disseminated across several sulci, is an important marker of future hemorrhage. Previous lobar hemorrhage and the overall burden of hemorrhagic lesions also contribute prognostic information. These associations support risk stratification at the group level but do not predict an individual event with certainty. Mortality and disability depend not only on the underlying angiopathy but also on hematoma volume, location, extension, baseline function, and accompanying disorders.
The use of amyloid-targeting antibodies for Alzheimer’s disease has made vascular amyloid increasingly relevant to drug safety. Agents such as lecanemab and donanemab can cause amyloid-related imaging abnormalities (ARIA). ARIA-E denotes edema or sulcal effusion, while ARIA-H denotes microhemorrhages or superficial siderosis. These abnormalities are often asymptomatic but can occasionally be serious. APOE ε4 status, pre-existing hemorrhagic MRI lesions, and treatment-related factors influence risk. Pretreatment imaging criteria and MRI monitoring are therefore integral to the studied treatment protocols.
ARIA is not identical to spontaneous CAA-related inflammation, although the two share vascular amyloid, permeability changes, and inflammatory features. Mobilization of amyloid and immune responses around amyloid-bearing vessels are proposed mechanisms. Alzheimer treatment trials commonly exclude patients with substantial hemorrhagic MRI abnormalities, limiting the applicability of their safety results to advanced CAA. Clinical benefit demonstrated for an Alzheimer therapy does not establish that the same drug treats symptomatic vascular amyloid disease.
Research aims include earlier detection, more precise prognosis, and treatments directed at vascular injury or amyloid clearance. Candidate biomarkers include quantitative susceptibility measures, high-resolution imaging of cortical microinfarcts, cerebrospinal-fluid amyloid profiles, and blood-based assays. Their value depends on distinguishing CAA from Alzheimer pathology and other small-vessel diseases, not merely detecting amyloid-related change. Clinical trials also face an endpoint problem: large hemorrhage, cognitive decline, MRI lesion progression, and functional disability capture different components of disease. A marker that changes with treatment is not necessarily a validated surrogate endpoint for meaningful clinical benefit.