Alzheimer’s disease and related dementias (ADRDs) are caused by the pathological accumulation of misfolded proteins, yet most variance in cognitive decline is not explained by postmortem neuropathology. Modifiable lifestyle behaviors (e.g., exercise, social activity) impact risk for developing ADRDs and can influence cognitive trajectories among individuals with ADRDs. However, the relative and combined contributions of lifestyle patterns to cognitive aging is poorly understood, as most studies have examined individual behaviors in isolation and without neuropathological characterization. We examined data-driven behavior patterns across multiple lifestyle domains among older adults, and tested their associations with disease-specific neuropathologies and cognitive trajectories.

Participants included 2059 older adults enrolled in the longitudinal Memory and Aging Project (MAP) at the Rush Alzheimer’s Disease Center. At baseline, participants were either cognitively unimpaired (CU; 73%) or had mild cognitive impairment (MCI; 23%). All participants completed an annual 19 test cognitive battery. Raw scores were standardized and averaged to compute a global z-score. Lifestyle factors were measured during at least one visit and included 1) actigraphy-measured physical activity, and self-reported 2) sleep quality, 3) life space, 4) cognitive activities, 5) social activities, and 6) social network. A subset (n=791) had autopsy data that included burden of Alzheimer’s disease (AD), cerebrovascular disease (CVD), Lewy body disease, and hippocampal sclerosis/TDP-43. Latent profile analysis across all 2059 participants identified distinct subgroups (i.e., classes) of lifestyle patterns. Linear mixed-effects models examined relationships between lifestyle classes and global cognitive trajectories, with and without covarying for all neuropathologies. Classes were also compared on rates of incident MCI/dementia.

Five classes were identified: Class 1_Low (lowest lifestyle engagement), Class 2_PA (high physical activity), Class 3_{Low Avg} (low to average lifestyle engagement), Class 4_Balanced (high average lifestyle engagement) and Class 5_Social (large social network). Compared to other classes, classes 4_Balanced and 5_Social had the lowest AD burden (omnibus p<.001) and Class 2_PA had the lowest CVD burden (omnibus p=.029). Classes 2-5 had significantly less steep cognitive decline compared to Class 1_Low (differences in time slope b range: -0.48 to -0.18, ps<.001). Statistical adjustment for neuropathological burden did not alter the magnitude of class differences in cognitive trajectories; lifestyle classes evidenced larger effect sizes compared to individual lifestyle indicators on cognitive decline (b range = -0.020 [social network] to 0.064 [life space]). Among those who were CU at baseline, rates of incident MCI/dementia were lowest in Classes 4_Balanced (26%) and 5_Social (20%) compared to remaining classes (31% to 57%; χ²=109.4, p<.001).

Latent lifestyle profiles among older adults account for differential rates of cognitive decline and clinical progression. This person-specific and multi-modal approach to lifestyle characterization demonstrated stronger associations with cognitive trajectories than any individual lifestyle behavior. Attenuation of cognitive decline was observed in those with at least average levels of engagement across all lifestyle domains, with some indication that social activity may have additional positive effects on cognitive stability. Findings highlight the importance of moderate and balanced engagement across lifestyle behaviors for later life cognitive health, even among individuals with MCI and ADRD pathology.