APOE ɛ4 exacerbates age-dependent deficits in cortical microstructure

Abstract

The apolipoprotein E ɛ4 allele is the primary genetic risk factor for the sporadic type of Alzheimer's disease. However, the mechanisms by which apolipoprotein E ɛ4 are associated with neurodegeneration are still poorly understood. We applied the Neurite Orientation Dispersion Model to characterize the effects of apolipoprotein ɛ4 and its interactions with age and education on cortical microstructure in cognitively normal individuals. Data from 1954 participants were included from the PREVENT-Dementia and ALFA (ALzheimer and FAmilies) studies (mean age = 57, 1197 non-carriers and 757 apolipoprotein E ɛ4 carriers). Structural MRI datasets were processed with FreeSurfer v7.2. The Microstructure Diffusion Toolbox was used to derive Orientation Dispersion Index maps from diffusion MRI datasets. Primary analyses were focused on (i) the main effects of apolipoprotein E ɛ4, and (ii) the interactions of apolipoprotein E ɛ4 with age and education on lobar and vertex-wise Orientation Dispersion Index and implemented using Permutation Analysis of Linear Models. There were apolipoprotein E ɛ4 × age interactions in the temporo-parietal and frontal lobes, indicating steeper age-dependent Orientation Dispersion Index changes in apolipoprotein E ɛ4 carriers. Steeper age-related Orientation Dispersion Index declines were observed among apolipoprotein E ɛ4 carriers with lower years of education. We demonstrated that apolipoprotein E ɛ4 worsened age-related Orientation Dispersion Index decreases in brain regions typically associated with atrophy patterns of Alzheimer's disease. This finding also suggests that apolipoprotein E ɛ4 may hasten the onset age of dementia by accelerating age-dependent reductions in cortical Orientation Dispersion Index.

Keywords: NODDI; cognitive impairment; neurodegeneration; preclinical dementia.

Graphical Abstract

Figure 1

Violin plots of mean lobar ODI in APOE ɛ4 carriers and non-carriers. Non-parametric permutation tests of group differences using PALM revealed no significant differences between both groups after adjusting for multiple comparisons with FWE (frontal: T = 1.5, P_FWE = 0.82, parietal: T = 2.2, P_FWE = 0.75, temporal: T = 0.9, P_FWE = 0.88, occipital: T = 1.1, P_FWE = 0.86, cingulate: T = 0.2, P_FWE = 0.98, insula: T = 0.4, P_FWE = 0.94). Abbreviations: APOE, Apolipoprotein E gene; FWE, family wise error; ODI, Orientation Dispersion Index; PALM, Permutation Analysis of Linear Models.

Figure 2

Marginal effect graphs depicting the two-way interactions between APOE ɛ4 and age (x-axis) on lobar ODI values (y-axis). Individuals with the APOE ɛ4 allele showed significantly steeper age-related reductions in ODI compared with non-carriers in the cingulate, frontal, insula, parietal and temporal lobes. Statistical significance was determined in PALM using non-parametric permutation models (5000 permutations, P_FWE < 0.05, adjusted for gender, years of formal education and sites). APOE, Apolipoprotein E gene; FWE, family wise error; ODI, Orientation Dispersion Index; PALM, Permutation Analysis of Linear Models.

Figure 3

Differences in standardized lobar ODI between APOE ɛ4 carriers and non-carriers (y-axis) as a function of age (x-axis). Pairwise comparisons were calculated based on marginal effects to predict the earliest hypothetical age at which APOE ɛ4 carriers would show decreased lobar ODI relative to non-carriers, highlighted by the vertical dotted lines on the x-axis. The shaded cyan region represents the 95% confidence interval. For each lobe, the earliest age of significant ODI reductions in APOE ɛ4 carriers is illustrated in the choropleth brain map. Abbreviations: APOE, Apolipoprotein E gene; ODI, Orientation Dispersion Index.

Figure 4

Vertex-wise interactions between APOE ɛ4 and age on cortical ODI. Results were obtained using non-parametric permutation models, implemented using PALM on ComBat harmonized ODI surface maps, and significance was established with TFCE P_FWE < 0.05, adjusted for sex, years of formal education and site. Abbreviations: APOE, Apolipoprotein E gene; TFCE, Threshold Free Cluster Enhancement; FWE, Family Wise Error; ODI, Orientation Dispersion Index; PALM, Permutation Analysis of Linear Models.

Figure 5

Summary statistics and MNI152 coordinates significant cluster associated with the APOE ɛ4×age interaction on vertex-wise ODI (TFCE P_FWE < 0.05, adjusted for sex, years of formal education, and site). Abbreviations: APOE, Apolipoprotein E gene; TFCE, Threshold Free Cluster Enhancement; FWE, family wise error; ODI, Orientation Dispersion Index; MNI152, Montreal Neurological Institute 152.

Figure 6

Three-way interactions between APOE ɛ4 × age × education. Marginal effects are depicted for the frontal and parietal lobes, stratified by APOE ɛ4 status. Lower years of education (mean–1 SD) are associated with steeper age-dependent reductions of ODI among the ɛ4 carriers only (frontal: T = 2.68, parietal: T = 2.37, P_FWE < 0.05, adjusted for site and sex). Abbreviations: APOE, Apolipoprotein E gene; FWE, family wise error; SD, standard deviation; ODI, Orientation Dispersion Index.

Figure 7

Vertex-wise three-way interactions between APOE ɛ4 × age × education. Marginal effects are depicted for the frontal and parietal lobes, stratified by APOE ɛ4 status. Lower years of education (mean–1 SD) are associated with steeper age-dependent reductions of ODI among the ɛ4 carriers only (P_FWE < 0.05, adjusted for site and sex). Abbreviations: APOE, Apolipoprotein E gene; FWE, family wise error; SD, standard deviation; ODI, Orientation Dispersion Index.

Figure 8

Summary statistics and MNI152 coordinates for each significant cluster associated with the three-way interactions of APOE ɛ4×age × education interaction on vertex-wise ODI. Abbreviations: APOE, Apolipoprotein E gene; ODI, Orientation Dispersion Index; MNI152, Montreal Neurological Institute 152.