The objectives of this study were to investigate the long-term associations between changes in physical activity levels and hippocampal volumes over time, while considering the influence of age, sex, and APOE-ε4 genotype. We investigated the effects of change in physical activity on hippocampal volumes in 411 middle age (mean age = 47.2 years) and 375 older age (mean age = 63.1 years) adults followed up to 12 years. An annual volume decrease was observed in the left (middle age: 0.46%; older age: 0.51%) but not in the right hippocampus. Each additional 10 metabolic equivalents (METs, ~2 h of moderate exercise) increase in weekly physical activity was associated with 0.33% larger hippocampal volume in middle age (equivalent to ~1 year of typical aging). In older age, each additional MET was associated with 0.05% larger hippocampal volume; however, the effects declined with time by 0.005% per year. For older age APOE-ε4 carriers, each additional MET was associated with a 0.10% increase in hippocampal volume. No sex effects of physical activity change were found. Increasing physical activity has long-term positive effects on hippocampal volumes and appears especially beneficial for older APOE-ε4 carriers. To optimize healthy brain aging, physical activity programs should focus on creating long-term exercise habits.
The brain is vulnerable to DNA damage and cardiometabolic risk. Yet, whether genetic variation in DNA repair interacts with cardiometabolic factors to explain cognitive variability remains unclear. Participants ( n = 376,533) of white-British ancestry from the UK biobank with cognitive, neuroimaging, and whole-exome sequencing data were included. Six cognitive outcomes were assessed: fluid intelligence (FIQ), symbol-digit matching task (SDMT), visual matching (MATCH), trail making (TRAIL1 and TRAIL2), and prospective memory (PMEM). Seven brain regions of interest were assessed: total brain (TBV), grey matter (GMV), left and right white matter (LWM/RWM), left and right hippocampi (LHC/RHC), and white matter hyperintensities (WMH) volumes. A total of 3487 genetic variants across 39 DNA repair genes were tested. SNP and gene/gene-set level associations were tested using regression models adjusted for age, sex, APOE ε4, ancestry, and outcome-specific covariates. Genetic interactions with a multidimensional cardiometabolic risk index (CMRI), encompassing established risk factors, were assessed. We detected 107 genetic variants (mostly extremely rare) across 36 DNA repair genes associated at Bonferroni-significance ( p ≤ 1.4 × 10 −5 ) with neurocognitive and brain outcomes. Most associations were observed for WMH (43 variants across 27 genes) and SDMT (26 variants across 17 genes). Most associations (60.8% of variants) were identified only in interaction models with CMRI. Associations across 35 of the 36 previously identified genes were also observed ( p < 0.05) for dementia. Interactions between rare genetic variants involved in DNA repair mechanisms and cardiometabolic risk may explain some of the observed cognitive variability. Graphical Abstract