Genetic risk of late-onset Alzheimer's disease is associated with longitudinal loss of functional brain network segregation in middle-aged cognitively healthy individuals: The PREVENT-Dementia Study

It is well acknowledged that the pathological processes of Alzheimers disease (AD) start decades before clinical manifestations, but early indicators of AD in midlife remain unclear. Functional segregation of brain networks has recently emerged as a key indicator of brain health. In this study, we investigated the vulnerability of intrinsic brain networks to loss of functional segregation during healthy adult lifespan and in cognitively healthy midlife individuals at risk of late-onset AD, and the association between segregation loss and cognition in midlife. Network segregation was measured using the participation coefficient metric within a graph-theoretic framework. In a healthy adult lifespan cohort (18-88 years, N=652), linear relationships of network segregation with age and cortical grey matter volume (GMV) were assessed using multiple regression models. In a cognitively healthy midlife cohort (40-59 years, N=210), associations between network segregation and established risk factors for AD were examined cross-sectionally and longitudinally (over 2 years). Across the healthy adult lifespan, global network segregation was positively associated with GMV and negatively associated with age, replicating previous findings. Three high-order networks [default mode (DMN), frontal-parietal control, and salience] and two sensorimotor networks (visual and motor) showed prominent age-related changes in functional segregation throughout adulthood. At midlife, cross-sectionally, cognitively healthy apolipoprotein (APOE) {varepsilon}4 carriers had higher global segregation than non-carriers. The DMN was the only individual network to show such an effect of APOE genotype. Higher global and DMN segregation was associated with better episodic and relational memory. Critically, APOE {varepsilon}4 carriers, but not non-carriers, showed a significant longitudinal loss of segregation in the DMN over 2 years. Overall, our findings suggest that functional network segregation constitutes a novel and early substrate for the impact of the genetic AD risk on the brain in midlife and thus have implications for the early detection and intervention in AD.