AMYPAD: Amyloid PET Imaging Data for Alzheimer's Disease Research
The Amyloid Imaging to Prevent Alzheimer's Disease (AMYPAD) project was an Innovative Medicines Initiative (IMI) funded public-private consortium that aimed to determine the value of ß-amyloid imaging as a diagnostic and therapeutic marker for Alzheimer’s disease. The AMYPAD consortium assembled and continues to expand the Prognostic and Natural History Study (PNHS), which combines amyloid PET imaging data from multiple cohorts to create one of the largest amyloid imaging resources available for the pre-dementia phase of Alzheimer's disease.
AMYPAD PNHS is an open-label, prospective, multi‐center, longitudinal cohort study linked to several ongoing European parent cohorts. Access to this data can help researchers understand the role of amyloid imaging in the earliest stages of Alzheimer’s disease and increase the success of secondary prevention trials.
Developed through AMYPAD's partnership with Amsterdam UMC and GE HealthCare, the dataset was updated in 2026 to add harmonized polygenic risk scores and MRI-derived endophenotypes, extending its use for individualized risk stratification and disease modeling research. Amyloid PET imaging captures both the extent of amyloid accumulation and its distribution across the brain.
What the Datasets Contain
The dataset contains:
- 4 dataset options: 2026 Raw, 2026 Harmonized, 2023 Raw, 2023 Harmonized,
- 3,000+ participants from a variety of European parent cohorts of preclinical and prodromal individuals: ALFA+, AMYPAD DPMS, EMIF-AD (60++), EMIF-AD (90+), EPAD LCS, FACEHBI, FPACK, Microbiota, and UCL-2010-412,
- 1,500+ participants with baseline amyloid PET scans; 1,454 with Centiloid quantification
- 800+ participants with at least one follow-up PET scan; 700+ with Centiloid quantification
- Demographic, clinical outcomes, disease biomarkers (imaging and CSF), risk factor (e.g., genetics and environmental), and other data
- 2023 data and 2026 update with computed polygenic risk scores and imaging-derived endophenotypes from harmonized genotype array and MRI-processing pipelines
What Researchers Can Do with the Datasets
Researchers can use AMYPAD PNHS to study the extent and topographical distribution of amyloid accumulation in preclinical and prodromal Alzheimer's disease, model disease progression across cohorts, or refine risk stratification approaches. The dataset's longitudinal follow-up PET data supports research into the natural history of amyloid burden.
Because AMYPAD PNHS pools standardized amyloid PET data from multiple European cohorts, it gives researchers valuable scale and geographic breadth for preclinical cohorts.
Recent studies using the AMYPAD datasets via AD Workbench:
- "Cortical gray–white matter contrast alterations precede amyloid-β positivity and macrostructural changes in older adults without dementia"
Pieperhoff, L., et al. Alzheimer's & Dementia, 2026;22:e71609. Analyzed MRI and amyloid PET data from 1,323 non-demented AMYPAD participants to show that cortical microstructural changes precede detectable atrophy and amyloid positivity. - "Harmonizing genotype array data to understand genetic risk for brain amyloid burden in the AMYPAD PNHS Consortium"
Luckett, E.S., Abakkouy, Y., et al. Alzheimer's & Dementia, 2025;21(9):e70376. Established a harmonization methodology across AMYPAD PNHS parent cohorts and computed polygenic risk scores linking genetic risk variants to amyloid PET burden. - "Amyloid PET predicts atrophy in older adults without dementia: Results from the AMYPAD Prognostic & Natural History Study"
Pieperhoff, L., et al. NeuroImage: Clinical, 2025;48:103912. Followed 1,329 participants, with longitudinal data for 684, to show that baseline amyloid PET burden predicts region-specific brain atrophy independent of tau pathology.
How to Access the Datasets
The AMYPAD PNHS datasets are available through the AD DISCOVERY PORTAL ↗ and accessible within individually permissioned AD Workbench workspaces at no cost to qualified researchers. AD Workbench provides free compute, virtual machines, and multimodal analysis tools and allows researchers to bring their own code, models, and approved external datasets into their workspace.