The association between midlife lipid levels and late-life brain amyloid deposition

Abstract

Elevated low-density lipoprotein cholesterol and total cholesterol in midlife and decline in total cholesterol from mid- to late-life are associated with incident dementia. Whether brain amyloid deposition mediates this relationship is unclear. We explored the association between midlife blood lipid levels and mid- to late-life change in lipid levels with brain amyloid deposition assessed using florbetapir PET scans in a biracial sample of 325 nondemented participants of the Atherosclerosis Risk in Communities–PET Amyloid Imaging study. Midlife total cholesterol, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, and triglycerides were not significantly associated with late-life amyloid burden after adjusting for covariates. Associations between changes in lipids and late-life amyloid deposition were similarly null. Lipids may contribute to dementia risk through alternate mechanisms.

1. Introduction

Quantifying amyloid plaque deposition with florbetapir PET scans provides an objective measure of Alzheimer’s-related pathologic burden. Because amyloid accumulation begins 10–20 years before the onset of clinical symptoms, vascular risk factor status in midlife is likely more relevant for studies of amyloid burden. One vascular risk factor of particular interest is blood lipid levels; indeed, higher total cholesterol (TC) in midlife and declining TC from mid- to late-life have been associated with increased dementia risk. One study to date reported a nonsignificant association between midlife blood lipids and late-life amyloid 20 years later, although the lack of racial diversity among the participants and small sample size used in PET amyloid analyses merit verification of their findings (Nagga et al., 2018). Thus, our goal was to describe the relationship between midlife lipid levels and mid- to late-life change in lipid levels and brain amyloid deposition in the community-based Atherosclerosis Risk in Communities–PET Amyloid Imaging study.

2. Methods

Our eligible sample included 325 nondemented participants of the Atherosclerosis Risk in Communities study who underwent florbetapir PET scans in 2011–2013. TC, high-density lipoprotein cholesterol (HDL-c), low-density lipoprotein cholesterol (LDL-c), and triglycerides were measured from samples collected in midlife (1987–1989) and in late-life (2011–2013). Elevated brain amyloid deposition was defined as a PET scan–derived global standard uptake value ratio greater than the median of 1.2 (Gottesman et al., 2017). For further details on measurement of blood lipids and brain amyloid deposition, please refer to the Supplementary Material. Associations between midlife TC, HDL-c, LDL-c, triglycerides, and mid- to late-life changes in lipid levels and elevated amyloid deposition were quantified using multivariable logistic regression. Covariates included age, education, gender, and APOE4 status. Because black and white participants were from different study areas, we were unable to separate the effects of race and study center in these analyses, and consequently combined the values of race and center into one race-center variable. We also explored these associations in participants with mild cognitive impairment (MCI). In sensitivity analyses, we used penalized splines to explore the possibility of a nonlinear association between lipids and amyloid burden and used linear mixed effects models to estimate person-specific trajectories of lipids from midlife to late-life.

3. Results

Of the participants included in analyses, 186 (57.2%) were female, 134 (41.2%) were black, and 84 (25.8%) had MCI at the time of their PET amyloid scan. Participants with elevated amyloid deposition in late-life were more likely to have at least one APOE e4 allele (p < 0.0001) and more likely to be female (p = 0.044). Those with elevated amyloid were also older (p = 0.013), had higher BMIs (p = 0.0012), and had higher TC (p = 0.007) and LDL-c (p = 0.045) in midlife (see Supplementary Table 1). Positive associations between TC and LDL-c in midlife and elevated amyloid burden in late-life in unadjusted models (TC: OR = 1.08, 95% CI = 1.02–1.15; LDL-c: OR = 1.06, 95% CI 1.00–1.13) were attenuated after adjusting for age, race-center, gender, and education (see Supplementary Table 2). There were no statistically significant associations between triglycerides or HDL-c in midlife and elevated amyloid burden in unadjusted or adjusted models. Results were similar before and after adjustment for APOE e4 allele status and in analyses restricted to participants with MCI. Notably, using penalized splines to describe the association between lipids and amyloid burden did not provide robust evidence of a nonlinear relationship (see Supplementary Fig. 1).

There was no association between change in lipids from mid- to late-life and late-life amyloid burden in unadjusted or adjusted models (see Supplementary Table 3). Results were unchanged when using linear mixed effects to estimate individual lipid trajectories.

4. Discussion

We found no significant associations between midlife lipids or change in lipids from mid- to late-life and elevated amyloid burden in late life. Our results are largely consistent with those of the Swedish BioFINDER Study (Nagga et al., 2018). Importantly, we verify these results with a larger PET amyloid sample size and extend the generalizability of these findings to black participants. We found no evidence of either a linear or nonlinear association between blood lipids and amyloid burden in spline analyses, but cannot exclude the possibility of an effect at very high lipid levels, such as those common in persons with familial hypercholesterolemia. Given our findings, lipids may be related to another pathological process that leads to dementia independent of the characteristic amyloid deposition associated with Alzheimer’s disease dementia. Future research should consider other mechanisms by which lipids may increase dementia risk.