FOCAS (Leiden sporadic CAA cohort)

FOCAS is a prospective follow-up study of sporadic cerebral amyloid angiopathy (CAA) at Leiden University Medical Center, started in February 2018. It shares its visit protocol with AURORA, the centre's study of Dutch-type hereditary CAA, and the two are often analysed together to study how CAA markers on MRI progress over time. The data are not public; papers from the cohort offer them on request from the corresponding author.

Composition

The LUMC study page reports 126 participants seen for one or more research visits; visits have ended. The study registration (NL-OMON52554) asks for age 50 or older and probable CAA by the Boston criteria v2.0 without a family history of hereditary CAA, and also lists age- and sex-matched healthy controls. Published subsets: 67 sporadic CAA participants screened at baseline, of whom 55 (mean age 70) had a 3 T MRI after one year; and 5 patients with one-year follow-up scans in a siderosis quantification study.

Acquisition

Participants were seen yearly for 3 T MRI on a Philips Achieva with a 32-channel head coil, 7 T MRI, blood and CSF sampling and clinical questionnaires. The 3 T protocol includes 4-echo SWI (TR 31 ms, first TE 7.2 ms, 0.6 x 0.6 x 1 mm, 130 slices) and T1- and T2-weighted imaging; a visually stimulated BOLD fMRI measures vascular reactivity.

Annotations

Trained raters scored lobar microbleeds and macrobleeds, cortical superficial siderosis (cSS), convexity subarachnoid haemorrhage, white matter hyperintensities and centrum semiovale perivascular spaces at baseline and follow-up, and measured WMH volume. A cSS quantification study built semi-automatic voxel masks of cSS on SWI (vesselness filter, seeded region growing, manual removal of false positives) for 5 FOCAS patients at baseline and one year, together with 15 patients from AURORA. Repeat masks by the same rater overlapped with a mean Dice of 0.75.

Known limitations

• No public access route, license or data use terms.
• Single centre and single 3 T scanner, so models trained on it may not transfer to other protocols.
• Published counts describe subsets; the sex and age make-up of all 126 participants is not published.
• The cSS masks cover only 20 patients in total, and the tool code is available only on request.