Subarachnoid CSF hyperintensities at 7 tesla FLAIR MRI: A novel marker in cerebral amyloid angiopathy.

BACKGROUND: We observed subarachnoid cerebrospinal fluid (CSF) hyperintensities at non-contrast 7-tesla (T) fluid-attenuated inversion recovery (FLAIR) MRI, frequently topographically associated with cortical superficial siderosis (cSS), in participants with cerebral amyloid angiopathy (CAA). To systemically evaluate these CSF hyperintensities we investigated their frequency and anatomical and temporal relationship with cSS on 7T and 3T MRI in hereditary Dutch-type CAA (D-CAA), sporadic CAA (sCAA), and non-CAA controls.

METHODS: CAA participants were included from two prospective natural history studies and non-CAA controls from a 7T study in healthy females and females with ischemic stroke. CSF hyperintensities were scored by two independent observers.

RESULTS: We included 38 sCAA participants (mean age 72y), 50 D-CAA participants (mean age 50y) and 44 non-CAA controls (mean age 53y, 15 with stroke). In total 27/38 (71 %, 95 %CI 56-84) sCAA and 23/50 (46 %, 95 %CI 33-60) D-CAA participants had subarachnoid CSF hyperintensities at baseline 7T. Most (96 %) of those had cSS, in 54 % there was complete topographical overlap with cSS. The remaining 46 % had ≥1 sulcus with CSF hyperintensities without co-localizing cSS. None of the healthy controls and 2/15 (13 %, 95 %CI 2-41, 100 % cSS overlap) of the stroke controls had CSF hyperintensities. In 85 % of the CAA participants CSF hyperintensities could retrospectively be identified at 3T. Of the 35 CAA participants with follow-up 7T after two years, 17/35 (49 %) showed increase and 6/35 (17 %) decrease of regional CSF hyperintensities. In 2/11 (18 %) of participants with follow-up who had baseline CSF hyperintensities without overlapping cSS, new cSS developed at those locations.

CONCLUSIONS: Subarachnoid CSF hyperintensities at 7T FLAIR MRI occur frequently in CAA and are associated with cSS, although without complete overlap. We hypothesize that the phenomenon could be a sign of subtle plasma protein or blood product leakage into the CSF, resulting in CSF T1-shortening.