Back

Impact of a White Matter Reference Region on the Relationship between Florbetapir PET Measurements of Amyloid Plaque Deposition and Measurements of Cognitive Decline

Bhargava, V.; Wang, M.; Chen, Y.; Luo, J.; Weiner, M.; Landau, S.; Jagust, W.; Sabbagh, M.; Su, Y.; Reiman, E. M.; Chen, K.

2025-09-25 neurology
10.1101/2025.09.19.25336123 medRxiv
Show abstract

The objective of this study was to systematically investigate both cross-sectional and longitudinal associations between amyloid PET tracer, Florbetapir (FBP), and cognition when different reference regions of interest - whole cerebellum versus white matter - are used for Standardized Uptake Value Ratio (SUVR) semi-quantification of amyloid beta deposition. Baseline and 2.2{+/-}0.4 year follow-up Florbetapir PET scans from 1,238 mild AD dementia, mild cognitive impairment (MCI), and cognitively unimpaired (CU) participants from AD Neuroimaging Initiative (ADNI) were used to characterize and compare the impact of using a cerebral white matter versus whole cerebellar reference region on cross-sectional and longitudinal relationships between florbetapir SUVR indicators of amyloid plaque deposition and measurements of cognitive or clinical decline (ADAS-Cog-13, CDR-Sum Boxes, and AVLT-total) after covarying for age and education. In both cross-sectional and longitudinal comparisons, florbetapir PET measurements of amyloid plaque deposition using the cerebral white matter reference region were more closely related to each measure of cognitive or clinical decline in the aggregate mild dementia, MCI and CU group (P<1.3E-06). This study supports the potential use of a cerebral white matter reference region in the detection and tracking of amyloid plaque deposition using florbetapir PET. Additional studies are needed to clarify the generalizability of findings to other amyloid PET ligands.

Matching journals

The top 5 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.