Abstract
Low physical activity (PA) and obesity are dementia risk factors, but it is unclear whether the association between PA and brain health is dependent upon degree of obesity. We aimed to examine the association between PA and brain structure and explore whether it varies by body mass index (BMI). We examined the associations between accelerometer-measured PA and magnetic resonance imaging-measured brain structure using cross-sectional data from a population-based cohort of people in midlife (UK Biobank). Using regression modelling, we explored whether these associations were mediated by BMI or other cardiometabolic risk factors. Complete data were available for 16,725 participants (median age 65 years, 55% women, median BMI 26). Greater PA was positively associated with total brain (β = 3.67, p < 0.0001), grey matter (β = 3.12, p < 0.0001), white matter (β = 0.43, p < 0.0001) and total hippocampal volumes (β = 0.05, p < 0.0001), and inversely associated with white matter hyperintensity volume (β = - 4.6 × 10(-5), p < 0.0001). Although there was no interaction between PA and BMI in explaining brain volumes, BMI attenuated the PA-brain volume associations by between 18 and 58%. The PA-brain volume associations were further attenuated by other cardiometabolic risk factors and in the case of grey matter and hippocampal volume, was no longer statistically significant in fully adjusted models. Greater PA is associated with a healthier profile in brain volumes. BMI and cardiometabolic factors mediated the relationship between PA and grey matter and hippocampal volume and partially mediated its relationship with other brain volumes. These results support a mechanistic basis for PA in optimizing cardiometabolic risk for dementia prevention.