DIAN Observational Study

The DIAN Observational Study follows families that carry a mutation causing dominantly inherited Alzheimer disease (DIAD) in the APP, PSEN1 or PSEN2 gene. Adult children of an affected parent are enrolled whether or not they carry the mutation, so the non-carrier siblings form a natural control group. Because the expected age at symptom onset is known for each mutation, the study can place every participant on a timeline relative to onset and track changes years before symptoms. The network is led by Washington University in St. Louis and runs sites in the Americas, Europe, Asia and Australia.

Composition

As of mid-2023 the study website lists 673 enrolled participants, about three quarters of them without symptoms. Of the 657 with confirmed mutation data, 397 carry a mutation and 260 do not. Participants return every one or two years, so most have several visits. Clinical, cognitive, genetic, CSF and blood biomarker data and neuropathology from autopsies come with the images.

Acquisition

MRI follows ADNI-based protocols on 3T scanners: MPRAGE, FLAIR, T2-weighted, diffusion, resting-state fMRI, ASL and a gradient echo scan for microbleeds. In the 2021 microbleed analysis this was a 2D T2*-GRE with 4 mm slices for 240 participants and SWI with 2 mm slices for 234 at baseline, and 153 participants switched from SWI to T2*-GRE at follow-up. PET covers PiB amyloid, FDG and, at some sites, tau. Imaging is held in the DIAN Central Archive.

Annotations

There are no voxel masks. Radiologists at Mayo Clinic read the gradient echo scans for cerebral microhemorrhages, superficial siderosis and macrohemorrhages, giving counts and lobar, deep or cerebellar location per visit, with longitudinal scans reviewed side by side. In the 511 participants analysed, 32 had microhemorrhages at baseline, 4 had superficial siderosis and 2 had macrohemorrhages. FreeSurfer volumes and white matter hyperintensity volumes are also available.

Known limitations

• The microbleed sequence changed from SWI to T2*-GRE at some sites, which affects detection sensitivity.
• Microbleed findings are counts and locations from visual reads, not segmentations.
• The exact mutation is not stored in the central archives, and mutation status is released only when a request keeps participants de-identified.
• Raw images need a justification in the request; processed data are the default.