Cerebral Amyloid Angiopathy: From Vascular Amyloid to Clinical Disease

The construct

Cerebral amyloid angiopathy is defined pathologically by amyloid deposition in walls of cortical and leptomeningeal vessels. In the common sporadic form, the deposited protein is predominantly amyloid beta. The same peptide participates in Alzheimer disease, but vascular and parenchymal amyloid occupy different compartments and do not produce interchangeable phenotypes. Modern reviews and frameworks emphasize CAA as both a hemorrhagic and nonhemorrhagic small-vessel disease R004 R005 R010.

CAA is a pathology. Boston criteria estimate the probability of that pathology in a defined clinical context; amyloid PET detects fibrillar amyloid signal; hemorrhagic MRI markers describe prior blood products. These are connected but distinct constructs.

1. Where CAA occurs

CAA preferentially affects cortical and leptomeningeal vessels, with posterior predominance often reported at the group level. Capillary involvement distinguishes a commonly used pathological subtype, but classification and regional burden depend on sampling and staining. Deep structures can contain other small-vessel pathologies even in a person with severe CAA.

Severity is not captured by presence versus absence alone. A useful pathological description considers anatomical distribution, vessel type, circumferential extent, mural-cell loss, cracking or splitting, fibrinoid change, inflammation, and evidence of hemorrhagic or ischemic injury. Consensus postmortem protocols improve comparability but do not eliminate sampling error R026.

2. Amyloid production, transport, and deposition

Amyloid beta is produced from amyloid precursor protein and can leave brain tissue through several routes. The intramural periarterial drainage hypothesis proposes movement of interstitial solutes along basement membranes within artery walls. Ageing, vascular stiffening, altered pulsatility, ApoE-related biology, impaired transport, and existing amyloid may reduce clearance. CAA might therefore be both a consequence and an amplifier of impaired drainage R011.

The clearance model is compelling because it connects anatomy, ageing, vessel function, and amyloid distribution. It remains a model with unresolved elements. Visible perivascular spaces are not direct flow measurements; cross-sectional associations do not establish temporal order; and experimental pathways may not transport fully to sporadic late-life human disease.

3. A staged progression framework

A useful framework separates four interacting stages R010:

Susceptibility: age, APOE and other genetic factors, amyloid production, clearance capacity, and systemic vascular context.
Vascular deposition: amyloid accumulates in cortical and leptomeningeal vessel walls.
Vascular dysfunction and remodeling: mural-cell loss, impaired reactivity, altered permeability, inflammation, and structural fragility emerge.
Tissue injury and symptoms: microinfarcts, white-matter injury, microbleeds, cortical superficial siderosis, convexity subarachnoid hemorrhage, and lobar ICH occur.

This is not a universal one-way sequence. Some abnormalities may be parallel consequences; injury may further impair clearance; and different patients may enter clinical observation at different stages. Hereditary CAA provides unusually informative temporal models, but mutations, age of onset, and co-pathology differ from sporadic disease R056.

4. How CAA damages tissue

Hemorrhagic pathways

Amyloid replacement and loss of normal vessel-wall structure can reduce mechanical integrity. Additional remodeling, microaneurysmal change, fibrinoid injury, inflammation, or local mechanical stress may contribute to rupture. The amount of amyloid at a rupture site need not be maximal: the vessel may have lost amyloid-bearing wall during rupture, or advanced remodeling rather than amyloid quantity alone may determine failure R010.

Ischemic and nonhemorrhagic pathways

CAA is associated with cortical microinfarcts, white-matter injury, altered reactivity, network disruption, and cognitive impairment. Candidate mechanisms include luminal compromise, dysfunctional vasoreactivity, impaired clearance, barrier change, and secondary degeneration. These pathways matter because a CAA phenotype can be clinically important even without symptomatic lobar ICH R004 R048.

Inflammation

Perivascular and vessel-wall immune responses may be secondary to amyloid, contribute to clearance, damage the barrier, or amplify vessel injury. CAA-related inflammation is a recognizable clinicoradiologic syndrome, but inflammatory activity also exists below the threshold of that syndrome R014 R057 R061.

5. Clinical manifestations

Lobar intracerebral hemorrhage

CAA is a major cause of spontaneous lobar ICH in older adults. Hemorrhage location, multiplicity, cortical superficial siderosis, and other MRI findings alter recurrence estimates. Clinical decisions must balance future hemorrhage against ischemic and cardiovascular risks R004 R009.

Convexity subarachnoid hemorrhage and transient focal neurological episodes

Small convexity subarachnoid hemorrhage and superficial siderosis can produce spreading positive or negative neurological symptoms. These episodes may mimic TIA, migraine aura, or focal seizure. Misclassification matters because reflex antithrombotic treatment based on a presumed ischemic TIA may increase risk in an active hemorrhagic syndrome. The guide teaches the distinction but does not provide patient-specific treatment advice R004 R009.

Cognitive impairment

CAA can contribute through microinfarcts, white-matter injury, hemorrhagic lesions, network disruption, and reduced vascular reserve. Attribution remains difficult because ADNC and other age-related pathologies frequently coexist. Cognitive presentation without hemorrhage also presents a different diagnostic spectrum from referral after lobar ICH R052.

CAA-ri may present with subacute cognitive or behavioral change, seizures, headache, focal deficits, asymmetric vasogenic edema-like FLAIR abnormality, and hemorrhagic markers. Clinicoradiologic criteria can support recognition, but infection, neoplasm, primary CNS vasculitis, posterior reversible encephalopathy, and other mimics require specialist evaluation R008 R061.

6. MRI phenotype

Strictly lobar hemorrhagic lesions

Lobar ICH and lobar cerebral microbleeds increase the probability of CAA when alternative causes are excluded. Strictly lobar is a distribution rule, not a guarantee of amyloid. False-positive pathology studies and lesion-level analyses show that non-CAA processes can produce apparently compatible patterns R015.

Cortical superficial siderosis

cSS represents chronic blood products along cortical surfaces and is a powerful CAA-associated marker. Extent and dissemination can have prognostic relevance, while mimics and sequence quality still matter R030 R047.

Centrum semiovale perivascular spaces and multispot WMH

These nonhemorrhagic markers were incorporated into Boston v2.0 in combinations that improve sensitivity. They carry less specificity in isolation and depend on definitions, rating reliability, and clinical spectrum R006 R051 R052.

Microinfarcts and advanced MRI

Conventional MRI detects only a subset of microinfarcts. Higher-field imaging, diffusion, perfusion, CVR, permeability imaging, and susceptibility methods may reveal earlier or different disease dimensions, but most lack a universally validated patient-level CAA threshold R001.

7. Boston criteria v2.0 as a probabilistic instrument

Boston v2.0 was developed and validated against neuropathology in a multicentre dataset. Probable CAA can be reached through multiple strictly lobar hemorrhagic lesions or through a combination of one lobar hemorrhagic lesion and an eligible white-matter feature, within the specified clinical and exclusion framework R006.

The development study does not justify using the criteria as population screening. Subsequent community and nonhemorrhagic validation found materially different performance, illustrating spectrum dependence R051 R052. A criterion can be well validated for symptomatic referral populations while performing less well in memory-clinic or community settings.

When applying the criteria in research, record:

Age and qualifying clinical presentation.
MRI sequences, field strength, and image quality.
Exact hemorrhagic lesion number and distribution.
cSS distribution and extent.
White-matter feature definitions.
Deep hemorrhagic lesions and alternative causes.
Criterion category and version.
Residual uncertainty and reference standard.

8. Molecular, fluid, and genetic biomarkers

Amyloid PET detects fibrillar amyloid but cannot reliably assign signal to vascular versus parenchymal compartments in an individual. Regional patterns may add group-level information without functioning as a stand-alone CAA diagnosis R032 R059. CSF and blood panels may capture amyloid, endothelial, inflammatory, or injury biology, but routine stand-alone specificity remains unestablished R031 R053 R058.

APOE modifies susceptibility and phenotype, but genotype is neither necessary nor sufficient for sporadic CAA diagnosis. Genetic information can add value in selected frameworks, such as the Edinburgh approach after lobar ICH, while availability and population transportability remain limitations R007.

9. Management and safety as evidence domains

CAA affects decisions about antithrombotic treatment, blood-pressure management, secondary prevention, management of manifestations, immunosuppression for inflammatory disease, and eligibility or monitoring for anti-amyloid treatment. The 2025 ICAAA/WSO scientific statement is the best current organizing source for these domains R009. Recommendations must be read at their actual strength and population; absence of randomized CAA-specific trials frequently limits certainty.

ARIA during anti-amyloid therapy overlaps with spontaneous CAA in vascular amyloid, edema-like change, and hemorrhagic findings, but it is treatment-associated and protocol-specific. Drug, dose, genotype, baseline MRI, symptom status, and timing all matter R040 R055.

10. Major live controversies

Is impaired perivascular clearance an initiating driver or a reinforcing consequence?
Which vascular dysfunction appears before structural injury?
Why do only some amyloid-laden vessels rupture?
How should nonhemorrhagic CAA be recognized without sacrificing specificity?
Which features predict severe or symptomatic ARIA?
Can fluid or PET biomarkers distinguish vascular from parenchymal amyloid at patient level?
How do CAA, arteriolosclerosis, ADNC, LATE-NC, and reserve jointly produce cognition?

11. Reading strategy

Read the broad CAA review first R004, then the emerging-concepts review R005 and progression framework R010. Study Boston v2.0 with its development table and validation populations R006, then compare community and nonhemorrhagic performance R051 R052. Finish with the current management statement R009 and the ARIA workgroup update R055.

Verification and further study

Comprehensive CAA review: R004
Mechanistic progression framework: R010
Boston criteria v2.0: R006
Community validation: R051
Nonhemorrhagic validation: R052
Diagnosis and management statement: R009
ARIA workgroup update: R055

Sources cited in this chapter

Open the original or authoritative record to verify wording, population, methods, and limitations.