- Design
- Retrospective multicentre cohort study with multivariable linear regression and false discovery rate correction
- Population
- 102 patients with probable cerebral amyloid angiopathy (Boston criteria v2.0) at two tertiary centres 2014-2023 who underwent lumbar puncture; mean age 72.0, 65% male
- Primary outcome
- Association of CSF amyloid-beta 40 and 42, total tau and p-tau181 with haemorrhagic and non-haemorrhagic MRI severity markers
- Effect
- Amyloid-beta 40 vs siderosis standardised beta -0.49 (95% CI -0.8 to -0.17, p=0.002); amyloid-beta 42 -0.42 (-0.73 to -0.11, p=0.007); both associated with Fazekas score; no association with microbleed count or with tau
Cerebral amyloid angiopathy is diagnosed on MRI using the Boston criteria, and those criteria count downstream consequences - microbleeds, siderosis, white matter change - rather than the vascular amyloid itself. This retrospective two-centre cohort asked whether CSF amyloid measurements track disease severity better.
102 patients with probable amyloid angiopathy by Boston v2.0 who had had a lumbar puncture as part of clinical evaluation were included, mean age 72, 65% male. Just over half presented with cognitive impairment, a third with intracerebral haemorrhage, and 13% with transient focal neurological episodes.
Lower CSF amyloid-beta 40 and 42 were associated with greater cortical superficial siderosis (standardised beta -0.49, 95% CI -0.8 to -0.17, p=0.002; and -0.42, -0.73 to -0.11, p=0.007) and with higher Fazekas score. Neither was associated with lobar microbleed count. Total tau and phosphorylated tau 181 showed no association with any MRI marker, and the amyloid associations survived stratification by whether the patient had an Alzheimer-like amyloid ratio.
The pattern is informative rather than decisive. Amyloid tracks siderosis and white matter disease, which are the markers most tied to clinical severity, and not microbleed count, which is the marker clinicians instinctively use as a severity gauge. Tau not moving suggests these CSF findings reflect vascular amyloid rather than accompanying neurodegeneration - the paper's central claim, and a plausible one.
This is retrospective, in patients selected because someone decided to do a lumbar puncture, and the authors say prospective longitudinal work is needed before any prognostic use. So do not add CSF sampling to the amyloid angiopathy workup on this. Do stop treating a microbleed count as a severity score - the siderosis burden is the marker that mattered here, and it is already on the scan you have.
- Report and follow cortical superficial siderosis burden, not just microbleed counts
- Do not add lumbar puncture to a routine amyloid angiopathy workup on this evidence
- A normal CSF tau does not argue against amyloid angiopathy - tau tracked nothing here
- An Alzheimer-like amyloid ratio does not explain the findings; they held after stratification
- Remember the selection: these patients all had a lumbar puncture for a clinical reason
The statistics, in plain English
A standardised beta of -0.49 means about half a standard deviation less siderosis burden per standard deviation higher CSF amyloid - a moderate association, and the interval (-0.8 to -0.17) is wide because there are 102 patients. Multiple MRI markers were tested against multiple CSF measures, and the authors applied false discovery rate correction, which is the right defence against finding associations by chance. The absence of an association with microbleed count is a null result in a modest sample, so read it as 'not demonstrated' rather than 'definitely absent'.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for neurology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free