Determinants of cognitive and brain resilience to tau pathology: A longitudinal analysis

Abstract

AbstractBackgroundMechanisms of resilience against tau pathology in individuals across the Alzheimer’s disease (AD) spectrum are insufficiently understood. Longitudinal data is necessary for revealing which factors contribute to maintaining cognition (cognitive resilience, CR) or preserving brain structure (brain resilience, BR) despite elevated tau, and for clarifying whether these factors provide a baseline advantage and/or moderate rates of brain structural and cognitive change. We therefore investigated whether age, sex, education, APOE‐e4 status, intracranial volume (ICV) and cortical thickness (in CR analysis) moderate the association between baseline tau load and subsequent changes in cognition and cortical thickness.MethodsThe study included 341 amyloid‐β‐positive individuals from a multi‐cohort dataset (BioFINDER‐1/AVID/ADNI/UCSF), diagnosed with MCI or AD dementia at the time of the baseline [18F]flortaucipir tau‐PET scan, who had longitudinal cognitive assessments (CR sample). A subset (n = 133) additionally underwent longitudinal structural MRI (BR sample). We used (separate) linear mixed‐effect models with MMSE as outcome to investigate the association of key baseline variables‐of‐interest (age, sex, APOE‐e4 status, education, ICV and global cortical thickness) with cognitive decline in the presence of tau (covaried for cohort). A three‐way interaction between time, tau and the predictor‐of‐interest was initially assessed. In the absence of a moderation effect, we subsequently assessed models including only the two‐way interaction between each predictor and time. These analyses were repeated with global cortical thinning as an outcome variable.ResultsCharacteristics of the CR sample and BR sub‐sample are described in Table‐1. Models revealed that younger age (βInteraction = ‐0.18, p<0.01), higher education (βInteraction = ‐0.62, p<0.001), larger ICV (βInteraction = ‐8.12, p<0.01) and greater cortical thickness (βInteraction = ‐9.9, p<0.05) were related to higher CR, as these factors moderated the relationship between tau pathology and decline in MMSE (Figure‐1,2). No predictor moderated tau effects on cortical thinning nor explained additional variance (Table‐2).ConclusionThis study indicates that tau pathology affects cognitive decline differently across age, education, ICV and cortical thickness levels. Understanding moderators of prospective cognitive decline and cortical thinning related to tau pathology, as well as their complex interactions, may be relevant for improving prognosis and clinical trial design in AD.