Domain‐specific cognitive decline associates with differential regional tau burden in the Alzheimer’s disease spectrum

Abstract

AbstractBackgroundPatients presenting with distinct clinical variants of Alzheimer’s disease (AD) present differential patterns of tau pathology measured with positron emission tomography (PET). Yet, a single set of brain regions like a temporal meta‐ROI is often chosen to study the association between cognition and tau. Instead, we aimed to map the associations between regional tau and multiple cognitive domains, and to test whether using whole brain information yields stronger associations with cognition compared to a singular composite region.MethodWe included 372 amyloid‐positive ADNI participants that had at least one tau‐PET scan (163 cognitively unimpaired [CU], 132 with mild cognitive impairment [MCI], 77 with AD dementia). Composite cognitive scores for memory, executive functioning, visuospatial and language were used for cross‐sectional (closest score to tau‐PET) and longitudinal analyses. Individual rate of cognitive decline was computed using linear mixed effect models, leveraging all visits available for each participant. We did region‐wise analyses, relating tau in the 64 regions of the Desikan‐Killiany atlas plus the amygdalae with cognitive performance and rate of cognitive decline. We also derived a “spatial extent index” where we computed the number of abnormal tau regions and compared the performance of this index to detect cognitive impairment to what would have been found with a classical temporal meta‐ROI SUVR value.ResultNo regional association between tau and cognition survived FDR correction in CU participants. Cross‐sectionally and longitudinally, participants with MCI and with AD dementia, associations between tau and memory were predominant in the temporal lobe, associations with language were restricted mostly to the left hemisphere and associations with executive functioning spanned the entire cortex. (Fig 1) Comparing the sum of tau positive regions (spatial extent index) to tau SUVR in the temporal meta‐ROI, both measures were similarly associated with cognition, except for executive functioning at baseline where the spatial extent index was more strongly associated with cognition than the meta‐ROI. (Fig 2)ConclusionWhile temporal meta‐ROI tau SUVR may be sufficient to track non‐executive cognitive performance in individual with cognitive impairments, a whole brain measure may provide additional information regarding executive impairment occurring in AD.