TRAJECTORY ANALYSES OF MORPHOMETRIC IMAGING MEASURES PROVIDE NOVEL INSIGHTS INTO THE DYNAMICS OF BRAIN AGING AS MODERATED BY SEX AND APOE

Longitudinal models of non-pathological aging are required to achieve a better understanding of specific change characteristics (level and slope) and prediction interactions (precision moderators) associated with exacerbated decline and transitions to cognitive impairment and Alzheimer's disease (AD). We used a sequence of dynamic quantitative trajectory modeling methods to analyze multi-wave longitudinal magnetic resonance imaging (MRI) data. We replicated these analyses across four morphometric measures (left/right hippocampal volume, cortical white matter volume, mean cortical thickness). Specifically, for each of these measures we modelled (1) normative changes in individualized trajectory patterns, (2) data-driven latent class differences (to identify subclasses of change), and (3) precision roles of AD risk moderators (sex, APOE). Cognitively normal adults were drawn from Alzheimer's Disease Neuroimaging Initiative (ADNI) data (n=402; mean 75.5 years (56.3-90.0) at baseline; 48% female). All participants contributed MRI data between one and 12 times (up to eight years). MRI data were processed using Freesurfer 6.0 to extract the four cortical and subcortical measures automatically. We used Mplus 8.2 to perform longitudinal quantitative modeling, including invariance testing (across waves), growth modeling (trajectory analyses, moderation), and latent class growth analyses (to identify subclasses with an algorithm of level and slope). First, we observed significant variability in individualized trajectories for all imaging measures. Second, we identified distinct trajectory subclasses within the overall sample for each measure. Specifically, we identified four subclasses for right and left hippocampal volume and cortical white matter volume, with three subclasses identified for cortical thickness. Third, the overall growth curves for each measure varied when stratified by sex and APOE. As expected, females showed lower hippocampal volume, lower cortical white matter volume, and higher cortical thickness than males. Females showed steeper decline in hippocampal volume and that was exacerbated by the possession of a ε4 allele. Both sexes showed similar decline in cortical thickness and white matter volume but females with a ε4 allele showed the steepest decline. Overall, females possessing a ε4 allele showed the highest risk for brain atrophy. These results will have implications for the establishment of precision-based, clinically-applicable norms and future biomarker prediction analyses.