Abstract
BACKGROUND: We determined the association between ratios of plasma ceramide species of differing fatty-acyl chain lengths and incident dementia and Alzheimer's disease (AD) dementia in a large, community-based sample. METHODS: We measured plasma ceramide levels in 1892 [54% women, mean age 70.1 (SD 6.9) yr.] dementia-free Framingham Offspring Study cohort participants between 2005 and 2008. We related ratios of very long-chain (C24:0, C22:0) to long-chain (C16:0) ceramides to subsequent risk of incident dementia and AD dementia. Structural MRI brain measures were included as secondary outcomes. RESULTS: During a median 6.5 year follow-up, 81 participants developed dementia, of whom 60 were diagnosed with AD dementia. In multivariable Cox-proportional hazards analyses, each standard deviation (SD) increment in the ratio of ceramides C24:0/C16:0 was associated with a 27% reduction in the risk of dementia (HR 0.73, 95% CI 0.56-0.96) and AD dementia (HR 0.73, 95% CI 0.53-1.00). The ratio of ceramides C22:0/C16:0 was also inversely associated with incident dementia (HR per SD 0.75, 95% CI 0.57-0.98), and approached statistical significance for AD (HR 0.73, 95% CI 0.53-1.01, P = 0.056). Higher ratios of ceramides C24:0/C16:0 and C22:0/C16:0 were also cross-sectionally associated with lower white matter hyperintensity burden on MRI (-0.05 ± 0.02, P = 0.02; -0.06 ± 0.02, P = 0.003; respectively per SD increase), but not with other MRI brain measures. CONCLUSIONS: Higher plasma ratios of very long-chain to long-chain ceramides are associated with a reduced risk of incident dementia and AD dementia in our community-based sample. Circulating ceramide ratios may serve as potential biomarkers for predicting dementia risk in cognitively healthy adults.