Background: Non-demented cognitive aging trajectories are characterized by vast level and slope differences and a spectrum of outcomes, including dementia. Objective: The goal of AD risk management (and its corollary, promoting healthy brain aging) is aided by two converging objectives: 1) classifying dynamic distributions of non-demented cognitive trajectories, and 2) identifying modifiable risk-elevating and risk-reducing factors that discriminate stable or normal trajectory patterns from declining or pre-impairment patterns. Method: Using latent class growth analysis we classified three episodic memory aging trajectories for n = 882 older adults (baseline M age=71.6, SD =8.9, range = 53-95, female=66%): Stable (SMA; above average level, sustained slope), Normal (NMA; average level, moderately declining slope), and Declining (DMA; below average level, substantially declining slope). Using random forest analyses, we simultaneously assessed 17 risk/protective factors from non-modifiable demographic, functional, psychological, and lifestyle domains. Within two age strata (Young-Old, Old-Old), three pairwise prediction analyses identified important discriminating factors. Results: Prediction analyses revealed that different modifiable risk predictors, both shared and unique across age strata, discriminated SMA (i.e., education, depressive symptoms, living status, body mass index, heart rate, social activity) and DMA (i.e., lifestyle activities [cognitive, self-maintenance, social], grip strength, heart rate, gait) groups. Conclusion: Memory trajectory analyses produced empirical classes varying in level and slope. Prediction analyses revealed different predictors of SMA and DMA that also varied by age strata. Precision approaches for promoting healthier memory aging—and delaying memory impairment—may identify modifiable factors that constitute specific targets for intervention in the differential context of age and non-demented trajectory patterns.

Although several factors predicted resilience in both sexes, a greater number applied only to women. Sex-specific mechanisms and intervention targets are implied.

Background: Age-related frailty reflects cumulative multisystem physiological and health decline. Frailty increases the risk of adverse brain and cognitive outcomes, including differential decline and dementia. In a longitudinal sample of non-demented older adults, we examine whether (a) the level and/or change in frailty predicts trajectories across three cognitive domains (memory, speed, and executive function (EF)) and (b) prediction patterns are modified by sex or Alzheimer's genetic risk (Apolipoprotein E (APOE)). Methods: Participants (n = 632; M age = 70.7, range 53-95; 3 waves) were from the Victoria Longitudinal Study. After computing a frailty index, we used latent growth modeling and path analysis to test the frailty effects on level and change in three latent variables of cognition. We tested two potential moderators by stratifying by sex and APOE risk (ε4+, ε4-). Results: First, frailty levels predicted speed and EF performance (level) and differential memory change slopes. Second, change in frailty predicted the rate of decline for both speed and EF. Third, sex moderation analyses showed that females were selectively sensitive to (a) frailty effects on memory change and (b) frailty change effects on speed change. In contrast, the frailty effects on EF change were stronger in males. Fourth, genetic moderation analyses showed that APOE risk (e4+) carriers were selectively sensitive to frailty effects on memory change. Conclusion: In non-demented older adults, increasing frailty is strongly associated with the differential decline in cognitive trajectories. For example, higher (worse) frailty was associated with more rapid memory decline than was lower (better) frailty. These effects, however, are moderated by both genetic risk and sex.