B-64: BOLD Signal Variability, Protein Pathology, and Memory Performance

Abstract

Objective: Blood oxygen level dependent (BOLD) signal variability is a promising measure of brain functioning. However, its relationship with Alzheimer’s disease (AD) protein pathology remains unclear, despite evidence that pathology may disrupt cerebral blood flow. We used a multiverse approach to evaluate associations between BOLD variability, positron emission tomography (PET)-derived amyloid and tau, and memory performance in adults with mild cognitive impairment (MCI) and Dementia of the Alzheimer’s type (DAT). Method/Data Selection: Participants included adults with MCI (N = 141) and DAT (N = 66), ages 51–88 (M = 71.94, SD = 6.95). Resting-state fMRI data were preprocessed using fMRIPrep and FSL in subject-native space. BOLD variability and PET data (i.e., amyloid and tau) were extracted from subject-specific medial temporal lobe regions. Variability was calculated using mean squared successive difference (MSSD) and standard deviation (SD). Memory measures included visual design, list, and story tests (RBANS, Benson, Craft, HVLT). Multiple linear regression models were estimated using multiverse, varying predictors, covariates, and outcome variables (1,472 total models). Results/Data Synthesis: All associations between BOLD variability and protein pathology were nonsignificant (p > .05). Similarly, only 1.20% of models were significant between BOLD variability and memory performance. In contrast, PET pathology was associated with worse memory (73.6% significant models, p < .05). Conclusions: BOLD variability was not associated with the severity of protein pathology deposition or memory; however, greater AD protein pathology was associated with worse memory. These findings underscore the importance of multiverse approaches and suggest that BOLD variability may reflect functional processes beyond protein pathology or memory performance.