A Cognitive Reserve Network That Moderates the Relationship Between Alzheimer’s Disease Pathology and Cognition

Abstract

AbstractBackgroundThe new consensus definition of cognitive reserve (CR) provides a framework to study individual differences in cognitive functioning relative to aging and disease. CR denotes a property of the brain that allows for better than expected cognitive performance given the degree of age‐related brain changes or disease. More specifically, individual differences in patterns of brain activity during fMRI tasks might explain the differential susceptibility to pathological burden.MethodAccording to the consensus definitions, we sought to identify and quantify CR from fMRI novelty‐contrast maps in a large multi‐centric sample (DELCODE) consisting of 215 participants with SCD, 79 with MCI, 30 with AD dementia, 56 AD relatives and 156 cognitively normal controls (CN). CSF amyloid‐42/40 ratio, CSF p‐tau and hippocampal volume (ATN) were reduced to a single number, representing a (Alzheimer’s) disease progression (DP) score. To identify a CR network, voxel‐wise linear regression models determined where functional task‐activation moderates the relationship between DP and cognition. Finally, task‐related activity within the CR network was extracted to obtain individual fMRI‐based CR scores.ResultThe DP score showed a strong negative quadratic association with baseline memory scores. CR voxels, in which higher or lower activation were associated with better cognition, weakening the effect of DP, were mainly located within the novelty network. They included lateral‐occipital and superior‐parietal regions, lingual and fusiform gyrus, cuneus and small parts of cingulate. While there was no association between the CR score and cognition in CNs, higher CR scores in MCI and AD patients were related to higher PACC5 scores.ConclusionWe established a DP score that collapses the ATN measures into a single number, while retaining its deleterious effect on an individual’s memory score. Furthermore, a newly identified task‐dependent CR network could be used to establish a CR score, which was related to higher PACC5 scores in MCI and AD patients, suggesting a functional compensation mechanism at later stages of AD that is not yet present in CN. Thus, targeted alteration of brain activity might be a promising route to modify cognitive trajectories in MCI and AD patients. Further, detailed examination of individuals with high CR might reveal additional disease‐modifying factors.