Differential associations between apolipoprotein E alleles and cerebral myelin content in normative aging

Abstract

Mounting evidence indicates that myelin breakdown may represent an early phenomenon in neurodegeneration, including Alzheimer’s disease (AD). Understanding the factors influencing myelin synthesis and breakdown will be essential for the development and evaluation of therapeutic interventions. In this work, we assessed associations between genetic variance in apolipoprotein E (APOE) and cerebral myelin content. Quantitative magnetic resonance imaging (qMRI) was performed on a cohort of 92 cognitively unimpaired adults ranging in age from 24 to 94 years. Using advanced qMRI methodology, we measured whole-brain myelin water fraction (MWF), a direct measure of myelin content, as well as longitudinal and transverse relaxation rates (R₁ and R₂), sensitive measures of myelin content, in carriers of the APOE ε4 or APOE ε2 alleles and individuals with the ε33 genotype. Automated brain mapping algorithms and statistical models were used to evaluate the relationships between MWF or relaxation rates and APOE isoforms, accounting for confounding variables including age, sex, and race, in several cerebral structures. Our results indicate that carriers of APOE ε2 exhibited significantly higher myelin content, that is, higher MWF, R₁ or R₂ values, in most cerebral structures investigated as compared to noncarriers, while ε4 carriers exhibited trends toward lower myelin content compared to noncarriers. Finally, all qMRI metrics exhibited quadratic, inverted U-shape, associations with age; attributed to the development of myelination from young to middle age followed by progressive loss of myelin afterwards. Sex and race effects on myelination were, overall, nonsignificant. These findings suggest that individual genetic background may influence cerebral myelin maintenance. Although preliminary, this work lays the foundation for further investigations to clarify the relationship between APOE genotype and myelination, which may suggest potential targets in treatment or prevention of AD.

1. INTRODUCTION

Growing evidence indicates that the breakdown of oligodendrocytes, the myelin producing cells, and myelin sheaths may be an early phenomenon in neurodegeneration, including that seen in mild cognitive impairment and dementias (1–3). In addition to serving as axonal insulation, permitting saltatory conduction of nerve impulses, myelin plays a role in transporting trophic support to the axons, making it paramount to higher-order integrative cerebral functions. Evidence obtained largely through quantitative magnetic resonance imaging (qMRI) suggests that microstructural degradation, including in peri-axonal myelin sheaths, may precede frank cognitive impairment in several neurological conditions, including dementias (4, 5). Therefore, understanding the factors influencing cerebral myelination, and early detection of related microstructural changes, holds substantial promise for the development and evaluation of targeted therapy.

The apolipoprotein E ε4 (APOE ε4) allele has been linked to increased risk of Alzheimer’s disease (AD) development (6), while emerging evidence suggests a protective effect of the APOE ε2 allele (7–9). While ~25% of the population are ε4/− carriers and exhibit underlying higher risk for AD, most of the population carries the ε33 genotype, with a limited subgroup of about ~15% carrying the ε2/− isoform; this latter isoform is associated with reduced risk for AD (10, 11). Several qMRI studies have examined and corroborated the differential effect of APOE isoforms on cerebral microstructural tissue integrity using qMRI metrics, including most prominently relaxation rates (R₁ and R₂) and the diffusion tensor imaging (DTI) indices of fractional anisotropy (FA) and mean diffusivity (MD) (12–17); these metrics have all been shown to represent sensitive probes of tissue microstructural integrity. The results of these studies were, overall, interpreted in terms of the differential effect of APOE genotypes on myelin production and maintenance. Indeed, the APOE protein is tightly involved in the transport and clearance of cholesterol and lipids, which are the main constituents of myelin. However, while sensitive to myelin content, none of these MR metrics are specific, so that the association between APOE genotype and myelination remains to be established.

Advanced MR methods based on multicomponent relaxometry to assess myelin water fraction (MWF), a surrogate of myelin content, have led to much greater specificity in noninvasive MRI myelin mapping (18–20). Using advanced MRI physics and technology, a whole-brain MWF map can now be obtained in less than 15 min using the multicomponent driven equilibrium single pulse observation of T₁ and T₂ (mcDESPOT) analysis (21–24). In pioneering work, using mcDESPOT-based MWF imaging, Dean and colleagues showed significantly lower myelin content in various brain regions of infants who were carriers of the APOE ε4 allele as compared to measurements derived from noncarriers (25). This finding has been recently corroborated in a longitudinal study, conducted on a large cohort of infants and young children (26), while also suggesting the possibility that ε4 carriers have altered rates of cerebral myelination and cognitive development in childhood. These compelling results support a direct role of APOE in early neurodevelopment and myelination. However, to the best of our knowledge, studies investigating the differential associations between the APOE ε2 and APOE ε4 alleles and cerebral myelination, as measured using MWF, in adult subjects have not been undertaken.

Here, we investigated the association between APOE isoforms and regional myelin content in a relatively large cohort of well-characterized adults without cognitive impairment (N = 92), across the extended age range of 24 to 94 years. The inclusion of subjects across a wide age range ensures a large dynamic range for myelin content measurements. Myelin content was measured using the Bayesian Monte Carlo (BMC)-mcDESPOT-based MWF method (27, 28). We also measured longitudinal and transverse relaxation rates (R₁ and R₂), sensitive measures of myelin, to provide contact with previous studies. Thus, the main goal of this work is to characterize the regional associations between cerebral myelination and APOE isoforms in cognitively unimpaired adults.

2. MATERIAL & METHODS

2.1. Study cohort

Participants were drawn from the Baltimore Longitudinal Study of Aging (BLSA) and the Genetic and Epigenetic Signatures of Translational Aging Laboratory Testing (GESTALT) study (29, 30). The goal of the BLSA and GESTALT studies is to evaluate multiple biomarkers related to aging, and their inclusion and exclusion criteria are essentially identical. Participants were excluded if they have metallic implants, or major neurologic or medical disorders. All participants were administered the Mini Mental State Examination (MMSE) test. Experimental procedures were performed in compliance with our local Institutional Review Board, and participants provided written informed consent.

2.2. MR imaging

MRI scans were performed on a 3T whole body Philips MRI system (Achieva, Best, The Netherlands) using the internal quadrature body coil for transmission and an eight-channel phased-array head coil for reception. Each participant underwent our BMC-mcDESPOT protocol for MWF, R₁, and R₂ mapping (28, 31). This imaging protocol consisted of 3D spoiled gradient recalled echo (SPGR) images acquired with flip angles (FAs) of [2 4 6 8 10 12 14 16 18 20]°, echo time (TE) of 1.37 ms, repetition time (TR) of 5 ms, with an acquisition time of ~5 min, as well as 3D balanced steady state free precession (bSSFP) images acquired with FAs of [2 4 7 11 16 24 32 40 50 60]°, TE of 2.8 ms, TR of 5.8 ms, and acquisition time of ~6 min. The bSSFP images were acquired with radiofrequency (RF) excitation pulse phase increments of 0 or π to account for off-resonance effects for a total scan time of ~12 min (~6 min for each phase-cycling scan). All SPGR and bSSFP images were acquired with an acquisition matrix of 150 × 130 × 94, voxel size 1.6 mm × 1.6 mm × 1.6 mm. Further, we used the double-angle method (DAM) to correct for excitation RF inhomogeneity (32). For that, two fast spin-echo images were acquired with FAs of 45° and 90°, TE of 102 ms, TR of 3000 ms, acquisition voxel size of 2.6 mm × 2.6 mm × 4 mm, and acquisition time of ~4 min. All images were acquired with a field-of-view of 240 mm × 208 mm × 150 mm, SENSE factor of 2, and reconstructed to a voxel size of 1 mm × 1 mm × 1 mm. The total acquisition time was ~21 min. We emphasize that all MRI studies and ancillary measurements were performed with the same MRI system, running the same pulse sequences, and at the same facility for both BLSA and GESTALT participants.

2.3. MR parameters mapping, registration, and segmentation

For each participant, using the FLIRT analysis as implemented in the FSL software (33), all SPGR, bSSFP, or DAM images were linearly registered to the SPGR image obtained at FA of 8° and the respective derived transformation matrices were then applied to the original SPGR, bSSFP, or DAM images. Then, a whole-brain MWF map was generated using BMC-mcDESPOT from these co-registered SPGR, bSSFP, and DAM datasets (3, 27, 28, 34). Briefly, BMC-mcDESPOT assumes a two-relaxation time component system consisting of short, attributed to myelin water, and long, attribute to intra/extra cellular water, component. We used the signal model explicitly accounting for nonzero TE (28, 31, 34). This emerging method provides rapid and reliable whole brain MWF maps (27, 28, 34, 35), and has been used to provide quantitative evidence of myelin loss in mild cognitive impairment and dementias, and to investigate factors influencing cerebral myelination in normative aging (3, 36–45). A whole-brain R₁ map was generated from the co-registered SPGR and DAM datasets using DESPOT1 and assuming a single component (46), and a whole-brain R₂ map was generated from the co-registered bSSFP and DAM datasets using DESPOT2 and assuming a single component (46). Indeed, R₁ and R₂ values depend on water mobility as well as macromolecular tissue composition including local lipid and iron content, and so are expected to be directly associated with a number of microstructural features of brain tissue, including myelin content. Further, using FSL software (33), the averaged SPGR image over FAs underwent nonlinear registration to the Montreal Neurological Institute (MNI) standard space, and the computed transformation matrix was then applied to the corresponding MWF, R₁, and R₂ maps. Eighteen white matter (WM) or gray matter (GM) regions of interest (ROIs) were defined from the MNI structural atlas corresponding to the whole brain (WB) WM or GM, and the frontal (FL), parietal (PL), temporal (TL), and occipital (OL) WM or GM lobes, cerebellum (CRB) WM or GM, corpus callosum (CC), corona radiata (CR), thalamic radiation (ThR), fronto-occipital fasciculus (FOF), longitudinal fasciculus (LF), and forceps (FOR). ROIs were eroded to reduce partial volume effect. Within each ROI, the mean MWF, R₁, and R₂ values were calculated. We note that all ROI measurements were performed in the MNI template space.

2.4. Statistical analyses

To investigate the effect of APOE isoforms on MWF, R₁, or R₂, multi-linear regression analysis was applied using the mean MWF, R₁, or R₂ value within each ROI as the dependent variable and the APOE status, sex, race, age, and age² as the independent variables, after mean age centering. APOE was defined as a 3-level categorical variable with ε33 as the reference group while sex was defined as a two-level categorical variable with women as the reference group. The inclusion of age² is based on our and others’ observations of a quadratic relationship between MWF, R₁, and R₂, and age (40, 41, 47, 48). The initial model incorporated an interaction between age and APOE but was removed as it was found not to be significant for all ROIs and MRI metrics. The resulting parsimonious model was then constructed without this interaction term. Furthermore, to investigate the associations between APOE and MWF, R₁, and R₂ between the APOE ε2/− group and APOE ε4/− group, we performed a between-group ANCOVA analysis for each ROI. All between-group comparisons were controlled for sex, race, age, and age². Given the exploratory nature of this investigation, analyses were conducted with and without correcting for multiple comparisons given the Type II error concerns in exploratory studies.

3. RESULTS

Demographic characteristics of participants

Table 1 shows the demographic characteristics of the participants after exclusion of seven participants with cognitive impairment and four with excessive motion artefacts on their MRI studies. The final cohort consisted of 92 cognitively unimpaired volunteers (mean ± standard deviation MMSE = 28.7 ± 1.6) ranging in age from 24 to 94 years (61.3 ± 19.8 years), of which 53 were men and 39 were women. The cohort consisted of 58 APOE ε33, 13 APOE ε2/− homozygous or heterozygous, and 21 APOE ε4/− homozygous or heterozygous participants. Age and MMSE were not statistically significantly different (p > .05) between genders, APOE groups, or races.

Table 1.

Demographic characteristics of participants.

Characteristic	Value
Number of participants	92
Mean ± SD age (yrs.)	61.3 ± 19.8
Min - Max age (yrs.)	24 – 94
Sex (Women/Men)	39/53
Race (Black/White)	26/66
MMSE score	28.7 ± 1.6
APOE4 (ε4/2, ε4/3, ε4/4)	21 (0, 19, 2)
APOE2 (ε2/2, ε2/3, ε2/4)	13 (1, 12, 0)

SD: standard deviation; Min: minimum; Max: maximum; Yrs.: years; MMSE: Mini Mental State Examination; APOE4: apolipoprotein E4 allele; APOE2: apolipoprotein E2 allele.

Effects of APOE isoforms on regional MWF, R₁, or R₂ qMRI measures

Table 2 summarizes the results of the multiple regression analysis of MWF vs. APOE status. This analysis indicates that the APOE ε2/− group has higher MWF values, that is, higher myelin content, than the reference group, APOE ε33, in all ROIs. This effect was statistically significant (p_ε2/− < .05) or close to significance (p_ε2/− < .1) in most cerebral WM and GM structures investigated even after accounting for the effects of sex, race, and age, before FDR correction. Several ROIs remained close to significance after FDR. Further, although the APOE ε4/− group exhibited, overall, lower MWF values as compared to the reference group in most ROIs, this effect was close to significance only in the whole brain and the frontal and temporal GM lobes before FDR. Furthermore, Fig. 1 shows the mean MWF values, calculated over all participants within the APOE ε2/− or APOE ε4/− groups, relative to the mean MWF value calculated over all participants within the APOE ε33 group, for the selected eighteen WM and GM cerebral regions. These results show that subjects in the APOE ε2 group have, overall, higher MWF values as compared to the reference group or the APOE ε4 group. The ANCOVA analysis indicates that these differences in mean MWF values between APOE ε2/− and APOE ε4/− are statistically significant or close to significance, before FDR correction, even after accounting for sex, race, and age in various cerebral regions including the occipital, parietal, and temporal WM lobes as well as the cerebellum, forceps, thalamic radiation, whole brain GM, and the frontal, occipital, parietal and temporal GM lobes. Several ROIs remained close to statistical significance (p < .1) after FDR as indicated in the caption of Fig. 1.

Table 2.

Regression coefficient, β, ± standard error, and significance, p, before false discovery rate (FDR) correction (88, 89), of the regression terms incorporated in the multiple linear regression given by MWF ~ β₀ + β_age × age + β_age² × age² + β_sex × sex + β_APOE × APOE + β_race × race. Sex and race results are not shown as they exhibited overall non-significant associations with MWF in most ROIs. Bold indicates statistical significance (p < 0.05) while Bold/Italic indicates close to significance (p < 0.1) before FDR.

MWF	Age		Age2		ε4 vs. ε3		ε2 vs. ε3
	β (× 10−4)	p	β (× 10−5)	p	β (× 10−3)	p	β (× 10−3)	p
Whole Brain-WM	24.0±11.0	0.0248 **	−2.8±0.88	0.002 **	−1.5±6.8	0.8244	17.0±8.3	0.0436 *
Frontal Lobe-WM	−10.0±1.2	<0.001 **	−2.6±0.75	0.001 **	−0.84±5.8	0.8853	13.0±7.1	0.0783
Occipital Lobe-WM	−6.5±1.2	<0.001 **	−2.5±0.77	0.002 **	−4.8±5.9	0.4217	15.0±7.2	0.0348 *
Parietal Lobe-WM	−9.5±1.1	<0.001 **	−2.1±0.70	0.004 **	−3.6±5.4	0.4991	15.0±6.5	0.0241 *
Temporal Lobe-WM	−7.5±1.3	<0.001 **	−2.4±0.79	0.003 **	−4.6±6.1	0.4575	16.0±7.4	0.0362 *
Cerebellum-WM	−5.1±0.86	<0.001 **	−1.1±0.53	0.045 **	−1.3±4.1	0.7448	11.0±5.0	0.0268 *
Corpus Callosum	−10.0±1.7	<0.001 **	−3.1±1.0	0.004 **	2.9±8.1	0.7229	20.0±9.8	0.0440 *
Corona Radiata	−13.0±1.9	<0.001 **	−3.3±1.2	0.005 **	0.85±8.9	0.9242	14.0±11.0	0.2014
Thalamic Radiation	−10.0±1.4	<0.001 **	−2.5±0.84	0.003 **	−2.4±6.5	0.7148	16.0±7.9	0.0463 *
Fronto-occipital fasciculus	−11.0±1.5	<0.001 **	−3.1±0.9	0.001 **	1.1±7.0	0.8790	13.0±8.5	0.1432
Longitudinal Fasciculus	−10.0±1.6	<0.001 **	−2.8±0.96	0.004 **	1.0±7.4	0.8888	16.0±9.0	0.0867
Forceps	−9.1±1.4	<0.001 **	−2.7±0.86	0.002 **	−3.6±6.6	0.5881	19.0±8.0	0.0216 *
Whole Brain-GM	−5.6±0.82	<0.001 **	−1.5±0.5	0.003 **	−6.7±3.9	0.0864	7.4±4.7	0.1166
Frontal Lobe-GM	−4.6±0.6	<0.001 **	−1.2±0.37	0.002 **	−5.6±2.8	0.0528	3.7±3.4	0.2820
Occipital Lobe-GM	−4.4±0.76	<0.001 **	−1.6±0.47	0.001 **	−4.9±3.6	0.1776	9.3±4.4	0.0374 *
Parietal Lobe-GM	−5.1±0.68	<0.001 **	−0.93±0.42	0.029 **	−5.1±3.2	0.1164	6.4±3.9	0.1052
Temporal Lobe-GM	−3.6±0.68	<0.001 **	−1.1±0.42	0.008 **	−6.3±3.2	0.0523	7.3±3.9	0.0650
Cerebellum-GM	−2.9±0.63	<0.001 **	−1.0±0.39	0.010 **	−2.8±3.0	0.3471	3.1±3.6	0.3867

WM: white matter, GM: gray matter.

^*indicates p-value close to significance (p < 0.1) following FDR correction, while

^**indicates p-value significant (p <0.05) following FDR correction.

Figure 1.

Mean MWF values of the APOE ε4 (gray bars) and APOE ε2 (black bars) groups relative to the mean MWF value of the reference group (APOE ε33). Results are shown for the eighteen investigated WM and GM brain structures/ROIs. ** indicates statistically significantly different (p < .05) and * indicates statistically close to significantly different (p < 0.1) mean MWF values between the APOE ε4 and APOE ε2 groups after controlling for sex, age, age², and race using the ANCOVA analysis, before FDR correction. We note that the OL, PL, TL, CRB, FOR, wb, ol, pl, and tl ROIs remained close to significance (p < 0.1) after FDR. WB: whole brain white matter (WM), wb: whole brain gray matter (GM), FL: frontal lobes WM, fl: frontal lobes GM, OL: occipital lobes WM, ol: occipital lobes GM, PL: Parietal lobes WM, pl: parietal lobes GM, TL: temporal lobes WM, tl: temporal loves GM, CRB: cerebellum WM, crb: cerebellum GM, CC: corpus callosum, CR: corona radiata, TR: thalamic radiation, FOC: fronto-occipital fasciculus, LF: longitudinal fasciculus (LF), and FOR: forceps.

Tables 3&4 and Figures 2&3 summarize comparable results for R₁ and R₂ vs. APOE status. Similar to the MWF results, this analysis indicates that the APOE ε2/− group has higher R₁ and R₂ values than the reference group, APOE ε33, in all ROIs. This effect was statistically significant or close to significance in most cerebral WM and GM structures investigated even after accounting for the effects of sex, race and age, before FDR. Several ROIs retained significance after FDR. However, although the APOE ε4/−group exhibited, overall, lower R₁ or R₂ values as compared to the reference group in most ROIs, this effect was close to significance only in the temporal GM lobes for R₁. Furthermore, Figs. 2 and 3 show, respectively, the mean R₁ and R₂ values calculated over all participants within the APOE ε2/− or APOE ε4/− groups, relative to the corresponding mean R₁ or R₂ values calculated over all participants within the reference group, APOE ε33, for the WM and GM cerebral regions we studied. As seen, subjects in the APOE ε2/− group have, overall, higher R₁ and R₂ values as compared to the reference group or the APOE ε4/− group. The ANCOVA analysis indicates that the differences in the mean R₁ values between the APOE ε2/− and APOE ε4/− groups, accounting for sex, race, and age effects, are statistically significant in most GM and WM cerebral regions studied, while the differences in the mean R₂ values between the APOE ε2/− and APOE ε4/− groups were limited to the occipital and parietal WM lobes, the thalamic radiation and the GM the structures, before FDR. After FDR, several ROIs retained significance or near-significance for R₁ while none of the ROIs retained significance or near-significance for R₂ as indicated in the captions of Figs. 2–3.

Table 3.

Regression coefficient, β, ± standard error, and significance, p, before false discovery rate (FDR) correction, of the regression terms incorporated in the multiple linear regression given by R₁ ~ β₀ + β_age × age + β_age² × age² + β_sex × sex + β_APOE × APOE + β_race × race. Sex and race results are not shown as they exhibited overall non-significant associations with R₁ in most ROIs. Bold indicates statistical significance (p < 0.05) while Bold/Italic indicates close to significance (p < 0.1) before FDR.

R 1	Age		Age2		ε4 vs. ε3		ε2 vs. ε3
	β (× 10−6)	p	β (× 10−8)	p	β (× 10−6)	p	β (× 10−5)	p
Whole Brain-WM	7.3±2.7	0.0089 **	−7.7±2.3	0.001 **	−8.6±17.0	0.6219	4.5±2.1	0.0369 *
Frontal Lobe-WM	−3.0±0.43	<0.001 **	−10.0±2.6	<0.001 **	−4.7±20.0	0.8163	5.2±2.5	0.0367 *
Occipital Lobe-WM	−2.2±0.47	<0.001 **	−9.8±2.9	0.001 **	−21.0±22.0	0.3383	6.9±2.7	0.0123 **
Parietal Lobe-WM	−3.3±0.44	<0.001 **	−6.9±2.7	0.011 **	−15.0±20.0	0.4539	6.3±2.5	0.0126 **
Temporal Lobe-WM	−2.3±0.42	<0.001 **	−9.8±2.6	<0.001 **	−24.0±20.0	0.2223	4.7±2.4	0.0582 *
Cerebellum-WM	−1.9±0.41	<0.001 **	−6.8±2.5	0.009 **	−12.0±19.0	0.5322	4.1±2.4	0.0853
Corpus Callosum	−2.6±0.55	<0.001 **	−9.4±3.4	0.007 **	1.3±26.0	0.9606	4.3±3.2	0.1811
Corona Radiata	−2.7±0.46	<0.001 **	−8.0±2.8	0.006 **	−6.3±22.0	0.7725	4.1±2.7	0.1299
Thalamic Radiation	−2.8±0.32	<0.001 **	−7.7±1.9	<0.001 **	−9.0±15.0	0.5510	4.2±1.8	0.0245 **
Fronto-occipital fasciculus	−2.4±0.38	<0.001 **	−9.0±2.3	<0.001 **	−6.9±18.0	0.6974	4.2±2.1	0.0562 *
Longitudinal Fasciculus	−2.1±0.38	<0.001 **	−7.7±2.3	0.001 **	−12.0±18.0	0.5027	3.9±2.2	0.0744
Forceps	−2.7±0.3	<0.001 **	−9.2±1.8	<0.001 **	−8.5±14.0	0.5446	4.8±1.7	0.0053 **
Whole Brain-GM	−2.2±0.22	<0.001 **	−5.8±1.3	<0.001 **	−9.1±10.0	0.3797	3.4±1.2	0.0087 **
Frontal Lobe-GM	−4.0±0.37	<0.001 **	−8.9±2.3	<0.001 **	−5.1±17.0	0.7706	5.9±2.1	0.0067 **
Occipital Lobe-GM	−2.9±0.4	<0.001 **	−9.5±2.4	<0.001 **	−19.0±19.0	0.3153	6.3±2.3	0.0068 **
Parietal Lobe-GM	−4.2±0.41	<0.001 **	−6.3±2.5	0.013 **	−13.0±19.0	0.5143	5.8±2.3	0.0150 **
Temporal Lobe-GM	−2.9±0.34	<0.001 **	−8.8±2.1	<0.001 **	−27.0±16.0	0.0978	3.0±1.9	0.1218
Cerebellum-GM	−1.6±0.36	<0.001 **	−6.8±2.2	0.002 **	−13.0±17.0	0.4481	2.3±2.0	0.2522

WM: white matter, GM: gray matter.

^*indicates p-value close to significance (p < 0.1) following FDR correction, while

^**indicates p-value significant (p <0.05) following FDR correction.

Table 4.

Regression coefficient, β, ± standard error, and significance, p, before false discovery rate (FDR) correction, of the regression terms incorporated in the multiple linear regression given by R₂ ~ β₀ + β_age × age + β_age² × age² + β_sex × sex + β_APOE × APOE + β_race × race. Sex and race results are not shown as they exhibited overall non-significant associations with R₂ in most ROIs. Bold indicates statistical significance (p < 0.05) while Bold/Italic indicates close to significance (p < 0.1) before FDR.

R 2	Age		Age2		ε4 vs. ε3		ε2 vs. ε3
	β (× 10−4)	p	β (× 10−6)	p	β (× 10−4)	p	β (× 10−3)	p
Whole Brain-WM	1.7±0.98	0.094 *	−2.3±0.82	0.007 **	−1.2±6.3	0.8523	1.9±0.76	0.0161 *
Frontal Lobe-WM	−1.5±0.16	<0.001 **	−2.8±0.95	0.004 **	0.93±7.3	0.8992	2.3±0.89	0.0103 *
Occipital Lobe-WM	−1.7±0.15	<0.001 **	−3.3±0.95	0.001 **	−6.0±7.3	0.4136	1.8±0.88	0.0403 *
Parietal Lobe-WM	−1.6±0.14	<0.001 **	−1.8±0.83	0.030 **	4.8±6.4	0.4506	2.4±0.77	0.0022 **
Temporal Lobe-WM	−1.3±0.17	<0.001 **	−3.3±1.0	0.002 **	−5.2±7.9	0.5123	2.1±0.96	0.0275 *
Cerebellum-WM	−0.77±0.15	<0.001 **	−0.77±0.9	0.394	−0.59±6.9	0.9323	1.6±0.84	0.0572 *
Corpus Callosum	−1.4±0.19	<0.001 **	−3.1±1.2	0.011 **	−2.9±9.0	0.7507	2.0±1.1	0.0691 *
Corona Radiata	−0.93±0.14	<0.001 **	−1.9±0.89	0.036 **	−3.3±6.8	0.6286	1.4±0.83	0.0858
Thalamic Radiation	−0.99±0.09	<0.001 **	−1.4±0.52	0.009 **	−2.1±4.0	0.5941	1.1±0.48	0.0318 *
Fronto-occipital fasciculus	−1.2±0.11	<0.001 **	−2.3±0.65	0.001 **	−1.6±5.0	0.7458	1.2±0.61	0.0542 *
Longitudinal Fasciculus	−0.79±0.12	<0.001 **	−1.7±0.71	0.017 **	−1.6±5.4	0.7743	1.4±0.66	0.0445 *
Forceps	−0.85±0.07	<0.001 **	−1.5±0.42	0.001 **	−0.98±3.3	0.7635	1.2±0.4	0.0041 **
Whole Brain-GM	−1.1±0.072	<0.001 **	−1.1±0.44	0.011 **	0.67±3.4	0.8425	1.0±0.41	0.0132 *
Frontal Lobe-GM	−1.4±0.12	<0.001 **	−0.93±0.73	0.206	2.1±5.6	0.7130	1.6±0.68	0.0184 *
Occipital Lobe-GM	−1.9±0.14	<0.001 **	−2.4±0.84	0.005 **	−1.6±6.5	0.8100	1.2±0.79	0.1378
Parietal Lobe-GM	−1.4±0.11	<0.001 **	−0.17±0.68	0.807	7.2±5.2	0.1672	0.96±0.63	0.1315
Temporal Lobe-GM	−1.8±0.12	<0.001 **	−2.5±0.71	0.001 **	−3.6±5.5	0.5121	1.2±0.67	0.0754 *
Cerebellum-GM	−1.0±0.11	<0.001 **	−0.79±0.68	0.248	0.46±5.2	0.9299	0.66±0.63	0.3008

WM: white matter, GM: gray matter.

^*indicates p-value close to significance (p < 0.1) following FDR correction, while

^**indicates p-value significant (p <0.05) following FDR correction.

Figure 2.

Mean R₁ values of the APOE ε4 (gray bars) and APOE ε2 (black bars) groups relative to the mean R₁ value of the reference group (APOE ε33). Results are shown for the eighteen investigated WM and GM brain structures/ROIs. ** indicates statistically significantly different (p < .05) and * indicates statistically close to significantly different (p < 0.1) mean R₁ values between the APOE ε4 and APOE ε2 groups after controlling for sex, age, age², and race using the ANCOVA analysis, before FDR. All significant ROIs remained significant or close to statistical significance after FDR. WB: whole brain white matter (WM), wb: whole brain gray matter (GM), FL: frontal lobes WM, fl: frontal lobes GM, OL: occipital lobes WM, ol: occipital lobes GM, PL: Parietal lobes WM, pl: parietal lobes GM, TL: temporal lobes WM, tl: temporal loves GM, CRB: cerebellum WM, crb: cerebellum GM, CC: corpus callosum, CR: corona radiata, TR: thalamic radiation, FOC: fronto-occipital fasciculus, LF: longitudinal fasciculus (LF), and FOR: forceps.

Figure 3.

Mean R₂ values of the APOE ε4 (gray bars) and APOE ε2 (black bars) groups relative to the mean R₂ value of the reference group (APOE ε33). Results are shown for the eighteen investigated WM and GM brain structures/ROIs. ** indicates statistically significantly different (p < .05) and * indicates statistically close to significantly different (p < 0.1) mean R₂ values between the APOE ε4 and APOE ε2 groups after controlling for sex, age, age², and race using the ANCOVA analysis, before FDR. None of the ROIs retained significance (or near-significance) after FDR. WB: whole brain white matter (WM), wb: whole brain gray matter (GM), FL: frontal lobes WM, fl: frontal lobes GM, OL: occipital lobes WM, ol: occipital lobes GM, PL: Parietal lobes WM, pl: parietal lobes GM, TL: temporal lobes WM, tl: temporal loves GM, CRB: cerebellum WM, crb: cerebellum GM, CC: corpus callosum, CR: corona radiata, TR: thalamic radiation, FOC: fronto-occipital fasciculus, LF: longitudinal fasciculus (LF), and FOR: forceps.

Effects of age, sex, and race on regional MWF, R₁, or R₂ qMRI measures

Significant age effects on MWF, R₁ or R₂ were found for all brain regions evaluated (Tables 2–4). Similarly, the quadratic effect of age, age², on MWF, R₁ or R₂ was significant in all brain regions except in the cerebellum and frontal and parietal GM lobes for R₂. Furthermore, while there were nonsignificant main effects of sex on regional MWF, R₁ or R₂ values, women exhibited trends toward higher parameter values than men (~8–10%). Finally, the effect of race on MWF, R₁ or R₂ was not significant.

4. DISCUSSION

In this cross-sectional study, conducted on a cohort of cognitively unimpaired adults and using MWF mapping and relaxation rate measurements, we find that variation of the APOE gene is correlated with myelin content in several key brain regions. Our analysis indicates that adult carriers of the ε2 allele exhibit significantly higher myelin content as compared to noncarriers in several cerebral regions evaluated, while carriers of the ε4 allele exhibit regional trends toward lower myelin content as compared to noncarriers with significance observed in various brain structures. This exploratory study, while should be considered preliminary, provides further evidence of a relationship between genetic background and life-course brain myelination.

Studies by Bartzokis and colleagues showed that APOE ε2 carriers exhibited significantly higher transverse relaxation rates, R₂, values as compared to noncarriers in several brain regions, whereas APOE ε4 carriers exhibited a more rapid decrease in R₂ values as compared to the APOE ε3 or APOE ε2 carrier groups, especially in late-myelinating cerebral structures (49, 50). Although interpreted in terms of differential effects of the APOE genotypes on myelin production and maintenance, relaxation rates, while sensitive to myelin content, are also sensitive to a number of other factors including axonal degeneration, iron content, hydration, temperature, flow, and macromolecular content (51). Nevertheless, our R₂ results agree with those of Bartzokis and colleagues, and provide further evidence of a relationship between APOE genotypes and myelin content using the more specific myelin content measure, MWF.

We found that carriers of the APOE ε4/− allele exhibited, overall, lower MWF values than noncarriers in several GM and WM structures (Figure 1, Table 2). These results support and expand on Dean and colleagues’ and Remer and colleagues’ observations of lower cerebral MWF in several gray matter and white matter regions in infants and young children who carry the ε4 allele (25, 26). Furthermore, it has recently been shown that MWF loss in cognitively impaired individuals is more prominent in APOE ε4/− carriers than in noncarriers, especially in normal appearing white matter regions (52). Additionally, we found that APOE ε2/− carriers have significantly higher MWF values in several ROIs studied (Figure 1, Table 2). To the best of our knowledge, this is the first study reporting results on the association between APOE ε2 and MWF. Despite the fact that our analysis was conducted in cognitively unimpaired participants, our results are complementary to and consistent with previous MRI studies based on DTI, relaxation rate measurements, and volumetric measurements, providing evidence that APOE ε2 carriers exhibit higher cerebral tissue integrity, potentially indicating decreased vulnerability to AD (12, 49, 50, 53–56). However, the effect of APOE on neurodegeneration remains unclear (57, 58). Indeed, although some studies have shown no effects of the ε23 and ε22 genotypes on white matter integrity as compared to ε33 or ε34 (15, 59), others have shown that remyelination may be defective in patients with multiple sclerosis (MS) who are carriers of the APOE ε2 allele (60), and higher prevalence of sporadic Parkinson disease (PD) (61). In contrast, studies suggest that APOE ε2 carriers exhibit a mild MS progression compared to carriers of the other genotypes (62), while the APOE ε4 allele is associated with a higher prevalence of dementia in PD (63). Given these conflicting observations and the limited number of carriers of APOE ε2 in our study, it is premature to ascribe definitive biological significance to our results. Studies with larger cohort sizes are still needed.

In the central nervous system, APOE mediates the synthesis and distribution of lipids and cholesterol to neurons and glial cells, the main constituents of myelin, and plays a paramount role in lipid clearance and recycling (57). Human APOE isoforms, of which there are three, differ in the extent to which they are expressed and folded, as well as in their binding affinities to APOE receptors, while also differentially promote brain energy metabolism and lipid and cholesterol efflux. APOE ε4 has been shown to be less efficient in these functions than APOE ε3 and APOE ε2 (64–68). All these differences likely play a role in their differential effects on amyloid-beta and tau protein accumulation and clearance as well as the amount of cholesterol efflux at sites, and may also indicate a causal role in myelination (57, 69–72). Indeed, APOE is mainly expressed by astrocytes and microglia, two major contributors to myelin synthesis, with studies indicating that APOE ε4 alters astrocytes and microglia function (73, 74). Moreover, it has been shown that APOE protein abundance in human brain follows an isoform-dependent pattern (ε2 > ε3 > ε4) (10). Higher levels of the APOE ε2 isoform are associated with greater cholesterol delivery to APOE receptors, perhaps accounting for higher efficiency in myelin synthesis and lower amyloid-beta deposition, while the APOE ε4 allele is associated with higher amyloid-beta and tau loads, pathologic hallmarks of Alzheimer’s disease. Such accumulations could lead to impairments of the lipid-rich myelin sheath and glia (10, 75–78). Furthermore, the APOE ε4 allele has been associated with decreased metabolism in several brain structures, which has been shown to predict cognitive decline after 2 years of longitudinal follow-up (79). Myelin maintenance through oligodendrocyte metabolism is an energy-demanding process, so that myelin homeostasis could be particularly compromised in APOE ε4 carriers. Moreover, growing evidence suggests that APOE ε4 is associated with widespread decline in cerebral blood flow (80–82). Interestingly, lower cerebral blood flow (CBF) has been recently associated with lower myelin content (38). However, direct studies of the modulation of CBF by APOE alleles are lacking. Finally, we note that the role of APOE in cognition in unimpaired older adults remains unclear. Some investigations suggested that the APOE ε4 allele is linked to cognitive decline in older adults (83, 84), but others indicate that the APOE ε4 allele may not be associated with cognitive performance in adults (85, 86). A recent study in the BLSA reported that APOE effects on longitudinal decline were age and sex specific and varied by cognitive domain (87). Further research is needed across the lifespan to elucidate the underlying mechanisms of APOE genotype in relation to age-related changes in cognition and myelin integrity.

This work has certain limitations. Although our study cohort is relatively large, the sample size of the ε2 and ε4 carrier groups were somewhat limited. Moreover, our study is cross-sectional and with a population sample size lacking sufficient power for Type I error adjustments. Indeed, several investigated regions lost significance after FDR correction. Therefore, further studies are needed to replicate and confirm our reported findings. Nevertheless, our results, along with those of Dean and Remer (25, 26), provide strong motivation for longitudinal investigation of larger cohorts to establish age-associated myelination trajectories as a function of APOE genotype. Finally, certain physiological and experimental parameters could bias qMRI parameter determination. These include, but are not limited to, the effects of magnetization transfer between macromolecules and free water protons, iron content, exchange between water pools, magnetization spoiling, RF pulse duration and shape, gradient duration and shape, and water diffusion within different compartments. These represent major challenges for all qMRI outcome measures, including of myelin content, with additional technical developments required to further improve accuracy.

5. CONCLUSIONS

This study provides additional evidence in support of the hypothesis suggesting a direct role of APOE in cerebral myelination. Our work also establishes a foundation for further investigation of the etiology of demyelination in neurodegeneration, including in Alzheimer’s disease.