IDEAS

IDEAS (Imaging Dementia - Evidence for Amyloid Scanning, ACRIN 7155, NCT02420756) was a US coverage-with-evidence-development study that paid for clinical amyloid PET in Medicare beneficiaries aged 65 and older with mild cognitive impairment (MCI) or dementia of uncertain cause who met appropriate use criteria. The American College of Radiology sponsored and ran the study, and the Alzheimer's Association directed it. The primary question was whether knowing the amyloid status changes how dementia specialists manage their patients. De-identified images and tabular clinical data are distributed by LONI through the Image and Data Archive under a data use agreement.

Composition

Enrollment ran from February 2016 to September 2017, with follow-up to January 2018. The primary analysis covers 11,409 of 16,008 registered beneficiaries: 6,905 with MCI and 4,504 with dementia. About half were women and the median age was 75. Amyloid PET was read as positive in 55.3% of the MCI group and 70.1% of the dementia group. Participants were recruited by 946 dementia specialists at 595 practices.

Acquisition

Scans were acquired as routine clinical studies at 343 PET facilities across the US with one of three FDA-approved amyloid tracers: florbetapir, florbetaben or flutemetamol. Sites later shared 10,774 raw scans with the study for central processing, and 10,361 passed quality control for Centiloid quantification. Florbetapir accounts for about two thirds of them.

Annotations

Each scan has the local visual read (positive or negative) by the site's imaging specialist. The study recorded the intended management plan, the suspected cause and the diagnostic confidence before the scan, and the implemented plan with any change in diagnosis and confidence at a visit about 90 days after it. Medicare claims outcomes were analyzed for the second aim.

Known limitations

The size and content of the LONI release are not documented publicly, so the counts here describe the study and its analyzed scan archive, not necessarily what LONI provides. Scanners, protocols and reconstructions vary across sites, as in routine care. Visual reads were done locally, not centrally. No MRI was part of the protocol. The cohort is mostly White, which limits generalization.