Aims
Mid-life obesity is linked with increased risk for late-life dementia, however relationships between later-life body composition and dementia are less clear. We sought to characterise cross-sectional and longitudinal associations between Body Mass Index, neuroimaging biomarkers and cognitive performance.
Methods
Combined cohort from two observational studies (AIBL and ADNI), Body Mass Index (BMI) categorised by WHO criteria. Multivariable linear mixed-effects models used to determine association between BMI category and outcomes: brain beta-amyloid PET (centiloid); MRI (global, hippocampal, white matter hyperintensity volume); cognition (composite scores for attention, executive function and episodic memory). All models adjusted by age, gender, education, APOEe4, and study (AIBL/ADNI).
Results
4,668 observations from 2,828 individuals were included (49.4% male, mean age 73.8 years, 38.2% APOE e4 carriage, 61.8% normal cognition [NC], 25.6% Mild Cognitive Impairment [MCI], 12.6% dementia due to Alzheimer's disease). The average BMI was 26.8 kg/m2 (0.6% underweight, 37.7% normal-range, 41.7% overweight, 20.0% obese [BMI <18, 18-25, 25-30 and >30 kg/m2, respectively]).
Relative to normal-range BMI, overweight-or-obese range BMI had a significant negative association with PET centiloid level, beta=-0.150 (0.037 se) and -0.309 (0.042), respectively (i.e. greater BMI had lower amyloid level). Obesity was associated with greater hippocampal volume, beta=0.192 (0.046), and stronger performance on attention and episodic memory. Neither obesity category nor BMI were associated with longitudinal change in cognitive scores.
Conclusions
Obesity in later-life was associated with lower brain amyloid levels and less hippocampal atrophy in a mixed cohort including older adults with dementia, mild cognitive impairment and normal cognition. While mid-life obesity is associated with greater risk for AD, later-life associations appear influenced by weight loss with emerging dementia.