Heterogeneity in normal aging: longitudinal trajectories of grey matter

Abstract

AbstractBackgroundIn dementia research, the role of aging is usually de‐trended as a confounder, implying implicit assumptions on the nature of controls. In fact, normal aging brings about structural and functional damage, possibly reflecting brain resistance and resilience, unrelated to overt cognitive impairment (1). The aim of this exploratory study is to predict longitudinal trajectories of brain change in healthy elderly, based on their heterogeneous patterns at baseline, to uncover potentially lost information and link brain, cognitive and biological profiles. To our knowledge, it´s the first study investigating this in healthy aging.MethodCortical volumes and subcortical thickness were assessed through MRI from 307 healthy elderly from ADNI, JADNI and AIBL initiatives. We applied cross‐sectional clustering using unsupervised random forest (2), followed by multivariate mixture of generalized mixed effect model (3) to predict individual trajectories and group them based on similarities. The longitudinal clustering was replicated in the separate cohorts for validity check.Result4 different clusters were identified, which statistically differ for musculoskeletal comorbidities. Cluster 1 shows the healthiest brain and cognitive profile at the age of 60, but it deteriorates the most across time. Conversely, cluster 3 presents the worst grey and white matter profiles at 60, and the greatest cognitive decline over time. However, showing only minimal brain change, cluster 3 resembles brain resistance, whereas cluster 1 seems compatible with cognitive resilience.ConclusionThese different trajectories suggest the possibility of identifying resilient and resistant individuals already from healthy aging. Clinically, this promotes environmental protective factors and early diagnosis. Methodologically, this warns on the heterogeneity of healthy “controls” in experimental designs.