Aβ accumulation associated with executive function decline and tau accumulation associated with memory decline in cognitively normal older people

Abstract Background The early accumulation of Aβ and tau in cognitively‐normal people is predictive of future cognitive decline. It has been difficult, however, to disentangle their effects. Aβ and tau target distinct regions in early stages. Frontal‐parietal regions are among the first to show Aβ accumulation and are important for executive control. Temporal regions, especially the medial temporal lobe, are early sites where tau accumulates and are critical for memory. Their differential vulnerability to Aβ and tau allows us to examine whether there are specific pathology‐cognitive associations, where faster decline in memory, beyond its expected level based on executive function, may be related to tau accumulation, and that faster executive function decline may be related to Aβ accumulation. Method 78 cognitively‐normal older adults from the Berkeley Aging Cohort Study underwent longitudinal 11C‐Pittsburgh compound‐B (PiB) PET, 18F‐flortaucipir (FTP) PET, and cognitive assessments of executive function and memory (Fig‐1A). We used linear mixed‐effects models to estimate longitudinal change in Aβ (Fig‐1B), temporal tau (Fig‐1C), executive function (Fig‐1D), and memory (Fig‐1E). We focused on within‐individual variability in cognitive decline in different domains and computed two cognitive decline residual scores (executive function controlling for memory, Fig‐2A; memory controlling for executive function, Fig‐2B) to represent domain‐specific performance decline. Result Aβ accumulation across wide‐spread regions, especially frontal/parietal cortex, was associated with faster executive function decline, but not memory, even after controlling for meta‐temporal tau change (Fig‐3A). On the other hand, temporal tau accumulation, especially in the medial temporal regions, was related to faster memory decline, but not executive function, even after controlling for global Aβ change (Fig‐3B). Using raw cognitive slopes, instead of residual scores, yielded stronger associations with both pathologies but reduced the ability to detect specific pathology‐cognition relationships. Conclusion We found evidence supporting specific associations between accumulating pathology and domain‐specific cognitive decline. The “ Aβ‐executive function, tau‐memory” association may reflect specific vulnerabilities in the executive control network and the memory network to different pathology, leading to cognitive decline in different domains. The variability of cognitive decline within the same individual across different domains provides unique information and may be used as behavioral markers for specific pathology.