In probable CAA, lower CSF Aβ40 and Aβ42 levels are associated with established MRI markers of disease severity, independently of clinical presentation and concomitant Alzheimer-like profile, suggesting that CSF Aβ levels may reflect vascular amyloid burden rather than downstream neurodegeneration.
Key Findings
Results
Lower CSF Aβ40 levels were significantly associated with greater cortical superficial siderosis (cSS) burden in probable CAA.
Standardized β = -0.49 (95% CI -0.8 to -0.17), p = 0.002
Association remained significant after false discovery rate correction for multiple testing
Association held after stratification by CSF Aβ42/Aβ40-defined Alzheimer-like profile
Multivariable models were adjusted for age, sex, MRI sequence type, and delay between onset and lumbar puncture
Results
Lower CSF Aβ42 levels were significantly associated with greater cortical superficial siderosis (cSS) burden in probable CAA.
Standardized β = -0.42 (95% CI -0.73 to -0.11), p = 0.007
Association remained significant after false discovery rate correction
Association was independent of concomitant Alzheimer-like CSF profile
cSS was one of the hemorrhagic MRI markers evaluated alongside lobar cerebral microbleeds
Results
Lower CSF Aβ40 levels were significantly associated with higher Fazekas score (white matter hyperintensity burden) in probable CAA.
Standardized β = -0.18 (95% CI -0.33 to -0.03), p = 0.02
Fazekas score served as a measure of white matter hyperintensity burden
Association survived false discovery rate correction for multiple comparisons
Association was independent of clinical presentation (cognitive impairment, intracerebral hemorrhage, or transient focal neurologic episodes)
Results
Lower CSF Aβ42 levels were significantly associated with higher Fazekas score in probable CAA.
Standardized β = -0.23 (95% CI -0.38 to -0.08), p = 0.002
This was one of the stronger associations observed across all MRI markers tested
Association remained significant after stratification by Alzheimer-like profile
White matter hyperintensity burden represents a nonhemorrhagic MRI marker of disease severity
Results
Neither CSF Aβ40 nor Aβ42 levels were significantly associated with lobar cerebral microbleed (CMB) count.
Lobar CMBs are a primary hemorrhagic marker used in Boston criteria v2.0 for CAA diagnosis
This lack of association contrasts with the significant associations found for cSS and Fazekas score
Multivariable linear regression models adjusted for age, sex, MRI sequence type, and delay between onset and LP were used
The study included 102 patients with mean age 72.0 ± 7.5 years
Results
No significant association was found between CSF tau species (total tau or p-tau181) and any MRI markers of disease severity.
Both total tau and phosphorylated-tau 181 (p-tau181) were measured using standardized immunoassays
MRI markers tested included lobar CMBs, cSS, white matter hyperintensity burden, Fazekas score, and enlarged perivascular spaces in the centrum semiovale
This null finding differentiates tau from amyloid-beta as a biomarker of CAA severity
Authors interpret this as suggesting CSF Aβ levels reflect vascular amyloid burden rather than downstream neurodegeneration
Methods
The study cohort consisted of 102 patients with probable CAA diagnosed according to Boston criteria v2.0 at two tertiary centers between 2014 and 2023.
Mean age at lumbar puncture was 72.0 ± 7.5 years; 65% were male
Initial presentations were cognitive impairment in 53%, intracerebral hemorrhage in 34%, and transient focal neurologic episodes in 13%
All patients underwent lumbar puncture as part of clinical evaluation
This was a retrospective multicenter cohort study design
Results
Associations between CSF Aβ levels and MRI severity markers remained significant after stratification by CSF Aβ42/Aβ40-defined Alzheimer-like profile.
The Aβ42/Aβ40 ratio was used to define an Alzheimer-like CSF profile
This stratification was performed to determine whether co-existing Alzheimer pathology confounded the associations
Findings suggest the CSF Aβ associations with MRI severity are specific to CAA rather than driven by concomitant Alzheimer's disease pathology
Principal component analysis was also used to explore associations between MRI and CSF biomarkers
What This Means
Cerebral amyloid angiopathy (CAA) is a condition where a protein called amyloid builds up in the walls of blood vessels in the brain, leading to bleeding and other types of brain damage. It is currently diagnosed using MRI brain scans, but these scans show the damage that has already occurred rather than directly measuring the disease process itself. This study examined whether proteins measured in cerebrospinal fluid (the fluid surrounding the brain and spinal cord) could provide a more direct window into how severe CAA has become. Specifically, researchers looked at levels of amyloid-beta proteins (Aβ40 and Aβ42) and tau proteins in 102 patients with probable CAA.
This research suggests that lower levels of both Aβ40 and Aβ42 in cerebrospinal fluid are linked to more severe signs of CAA on MRI, specifically more cortical superficial siderosis (a type of iron deposit from prior bleeding on the brain's surface) and greater white matter damage. Notably, these fluid biomarkers were not associated with the number of tiny brain bleeds called microbleeds, which are the primary marker used to diagnose CAA on MRI. Tau proteins, which are associated with Alzheimer's disease and nerve cell death, showed no meaningful association with any MRI severity markers. Importantly, the associations between low amyloid-beta levels and MRI severity held even when accounting for whether patients also had Alzheimer's-like changes in their fluid biomarkers.
These findings matter because they suggest that measuring amyloid-beta proteins in cerebrospinal fluid could offer a complementary tool for assessing how advanced CAA is, beyond what MRI alone can show. The fact that tau proteins were not associated with disease severity, while amyloid-beta proteins were, supports the idea that low fluid amyloid-beta reflects amyloid being trapped in blood vessel walls rather than general brain cell damage. The authors call for future studies that follow patients over time to determine whether these fluid biomarkers can predict how the disease will progress.
Beaufort Q, Charbonnier C, Ruellan P, Ozkul-Wermester O, Amar V, Turpin D, et al.. (2026). Evaluation of Disease Severity Using CSF Biomarkers in Patients With Probable Cerebral Amyloid Angiopathy.. Neurology. https://doi.org/10.1212/WNL.0000000000218458