WRAP

The Wisconsin Registry for Alzheimer's Prevention (WRAP) is a longitudinal cohort at the University of Wisconsin-Madison, established in 2001. It enrolled adults at midlife, most of them with a parent who had Alzheimer's dementia, to follow cognition and biomarkers from well before any symptoms. It is used to study preclinical Alzheimer's disease, early cognitive decline and how genetic, health and lifestyle factors affect them.

Composition

By the 2018 review, 1,561 people had enrolled, aged 40 to 65 at entry with a mean baseline age of 54; 73% had a parental history of Alzheimer's dementia and 40% carried APOE e4. The WRAP data request page reports statistics from the 31 May 2025 data freeze: 1,353 women and 570 men, a mean age of 55.7 years at the first cognitive assessment and, at the last visit within the freeze, 1,691 cognitively normal participants, 120 with MCI, 65 with other cognitive impairment and 40 with dementia. 954 participants have MRI, 740 amyloid PET and 702 tau PET. Many have repeated scans: 641 participants have at least two MRIs.

Acquisition

Study visits come about four years after baseline and then about every two years. A standard MRI protocol has been used since 2009 with a 3D inversion-recovery T1-weighted scan, 3D T2 FLAIR, pseudo-continuous ASL, multi-shell diffusion and 4D flow imaging. Amyloid PET uses PiB (dynamic 70-minute acquisition) or NAV4694, and tau PET uses MK-6240. Scanner models are not described in the sources used here. Shared MRI includes T1, T2 and FLAIR in NIfTI format, and PET is shared as NIfTI with a T1 image, both with derived values in tables.

Annotations

No manual image labels are published. Derived measures include brain and hippocampal volumes, white matter hyperintensity volume and regional amyloid and tau PET uptake.

Known limitations

• The cohort is enriched for parental history and is mostly white; the 2025 freeze lists 408 Black or African American participants.
• Imaging was added over time and is only available for a subset.
• The data use agreement is not published, and analyses using machine learning or AI need an extra university risk evaluation before data release.
• The 2025 freeze statistics do not give one total, and their sub-tables add up to slightly different numbers.