Effect of APOE Genotype on Plasma Docosahexaenoic Acid (DHA),Eicosapentaenoic Acid,Arachidonic Acid,and Hippocampal Volume in the Alzheimer’s Disease Cooperative Study-Sponsored DHA Clinical Trial

Abstract

Background:

Docosahexaenoic acid (DHA), eicosapentaenoic acid (EPA), and arachidonic acid (AA) play key roles in several metabolic processes relevant to Alzheimer’s disease (AD) pathogenesis and neuroinflammation. Carrying the APOE ɛ4 allele (APOE4) accelerates omega-3 polyunsaturated fatty acid (PUFA) oxidation. In a pre-planned subgroup analysis of the Alzheimer’s Disease Cooperative Study-sponsored DHA clinical trial, APOE4 carriers with mild probable AD had no improvements in cognitive outcomes compared to placebo, while APOE 4 non-carriers showed a benefit from DHA supplementation.

Objective:

We sought to clarify the effect of APOE ɛ4/ɛ4 on both the ratio of plasma DHA and EPA to AA, and on hippocampal volumes after DHA supplementation.

Methods:

Plasma fatty acids and APOE genotype were obtained in 275 participants randomized to 18 months of DHA supplementation or placebo. A subset of these participants completed brain MRI imaging (n = 86) and lumbar punctures (n = 53).

Results:

After the intervention, DHA-treated APOE ɛ3/ɛ3 and APOE ɛ2/ɛ3 carriers demonstrated significantly greater increase in plasma DHA/AA compared to ɛ4/ɛ4 carriers. APOE ɛ2/ɛ3 had a greater increase in plasma EPA/AA and less decline in left and right hippocampal volumes compared to compared to ɛ4/ɛ4 carriers. The change in plasma and cerebrospinal fluid DHA/AA was strongly correlated. Greater baseline and increase in plasma EPA/AA was associated with a lower decrease in the right hippocampal volume, but only in APOE 4 non-carriers.

Conclusion:

The lower increase in plasma DHA/AA and EPA/AA in APOE ɛ4/ɛ4 carriers after DHA supplementation reduces brain delivery and affects the efficacy of DHA supplementation.

Keywords

Arachidonic acid Alzheimer’s disease apolipoprotein E docosahexaenoic acid eicosapentaenoic acid

INTRODUCTION

There is evidence that among highly unsaturated fatty acids (HUFAs, Fig. 1) docosahexaenoic acid (DHA), eicosapentaenoic acid (EPA), and arachidonic acid (AA) play key roles in several metabolic processes relevant to Alzheimer’s disease (AD) pathogenesis [1, 2]. The omega-3 (n-3) HUFAs DHA and EPA compete with the omega-6 (n-6) HUFA AA at the sn-2 position of phospholipids affecting their levels in tissues and in the circulation [3]. This results in a diet-determined HUFA tissue balance where increasing dietary consumption of DHA or EPA lowers AA tissue levels. This change has a biological significance. A lower ratio of DHA and EPA to AA in the brain detrimentally affects neuroinflammation, oxidative stress, and alters synaptic activity and hippocampal functions [4, 5].

Fig.1

The categorization of highly unsaturated fatty acid (HUFAs), defined as fatty acids containing 20 or more carbons and 3 or more double bonds.

Carrying the APOE ɛ4 allele (APOE4) affects systemic and brain lipid metabolism in an allele-dose dependent manner, and is the strongest genetic risk factor for late-onset AD [6]. In contrast, carrying the APOE ɛ2 allele (APOE2) protects against AD [7]. After lipolysis, enrichment of lipoproteins with APOE accelerates their clearance from plasma [8]. The preferential binding of APOE4 compared to APOE ɛ3 (APOE3) and APOE2 alleles (APOE4 > APOE3 > APOE2) with larger fat-containing particles contributes to altered lipid and cholesterol metabolism [9 –11]. For example, DHA and EPA are transported on chylomicrons to the liver following absorption, and APOE4 carriers have faster DHA and EPA clearance from plasma compared to non-carriers [12 –15]. APOE4 carriers also have a lower DHA/AA in human plasma and in brains of human APOE-targeted replacement (APOE-TR) mice [16]. In APOE-TR mouse models, APOE4 was associated with an increase in mitochondrial carnitine palmitoyltransferase I (CPT-1) expression, which promotes n-3 polyunstaturated fatty acid (PUFA) oxidation [17]. These studies support the notion that APOE4 affects the ratio of DHA and EPA to AA in tissues.

One biomarker of AD severity is the volume of the hippocampus, which predicts the conversion of patients with mild cognitive impairment to AD [18]. The rate of hippocampal volume loss is greatest in patients with AD carrying APOE 4 [19]. Red blood cell and plasma DHA and EPA measurements positively associate with the volume of the hippocampus in both cross-sectional and longitudinal studies [20 –22]. The lipiDiDiet trial used a multi-nutrient intervention that included high doses of DHA in patients with prodromal AD. The nutrient intervention arm was associated with significantly lower decrease in hippocampal volumes compared to placebo over 24 months [23]. Hippocampal volume may therefore offer a surrogate biomarker of the efficacy of n-3 supplementation in the brain.

DHA supplementation has generally been ineffective in trials conducted in patients with AD [24, 25]. In the Alzheimer’s Disease Cooperative Study (ADCS)-sponsored DHA clinical trial, the main outcomes—change in the cognitive subscale of the Alzheimer’s Disease Assessment Scale (ADAS-cog), the Clinical Dementia Rating (CDR) sum of boxes, and the rate of brain atrophy—did not differ by treatment groups. A preplanned subgroup analysis of this trial showed a significant benefit of DHA supplementation compared to placebo on changes in ADAS-Cog and the Mini-Mental Status Examination (MMSE) when analyzed among APOE4 non-carriers; this benefit was not evident in APOE4 carriers [24]. We subsequently examined cerebrospinal fluid (CSF) samples obtained at baseline and at 18-month follow-up in a small subsample from this DHA trial to understand the changes in CSF DHA levels relative to randomized treatment and APOE4 status. DHA-treated APOE4 carriers had less increase in CSF DHA compared to DHA-treated non-carriers [24]. The mechanisms for the different clinical response to DHA supplementation noted in the trial subgroup analysis and the reduced DHA CSF delivery associated with APOE4 are not well understood.

In the current study, we sought to clarify the effect of APOE4 on plasma DHA/AA, EPA/AA, and hippocampal volumes in the setting of this trial. We hypothesized that APOE ɛ4/ɛ4 would be associated with a smaller increase in plasma phospholipid DHA/AA and EPA/AA after supplementation, as well as a greater loss of hippocampal volume compared with APOE 4 non-carriers.

METHODS

ADCS-sponsored clinical trial

This trial randomized 402 participants with mild AD to two treatment arms over 18 months: DHA arm (2 g DHA per day) and placebo arm. All participants were required to take four 950 mg soft-gel capsules which contained either 510 mg DHA (treatment arm) or corn/soy oil (placebo arm) [24]. Fifty-one US centers participated in this trial after obtaining approval from their local institutional review boards. Randomization was 60% assigned to active treatment, and 40% assigned to placebo. The disproportionate randomization to active treatment was intended to enhance recruitment. Randomization was not stratified by APOE groups. Written informed consent was obtained from study participants, legally authorized representatives, or both.

Fatty acid analysis and omega-3 dietary intake

Fasting plasma and CSF phospholipid fatty acid levels were determined at DSM Nutritional Products using established methods [26]. The fatty-acid profiles were expressed as a percentage of the total fatty acid in micrograms (weight percent). DHA dietary intake was assessed using a validated fatty fish intake questionnaire at baseline [27].

MR imaging

Trial participants without contraindication to MRI (e.g., pacemaker) who were enrolled at trial sites that were also certified Alzheimer’s Disease Neuroimaging Initiative (ADNI) sites were invited (but not required) to participate in an MRI substudy, with MRIs obtained at baseline and at end of the 18-month period. The MRI sequence, as well as methods for across-site standardization and quality control, were those used in the ADNI study. FreeSurfer software v5.3.0 (https://surfer.nmr.mgh.harvard.edu/) was used to automatically segment the bilateral hippocampus and calculate hippocampal volumes at baseline and 18 months. Quality control of the hippocampal segmentations was performed by a researcher who was blind to genetic and treatment data, using the recommendations of the European Alzheimer’s Disease Consortium and the Alzheimer’s Disease Neuroimaging Initiative (EADC – ADNI) harmonized hippocampal protocol [28, 29]. Intracranial volume (ICV) was estimated using FreeSurfer.

Power analysis

In a previous analysis of the ADCS-sponsored DHA clinical trial [26], forty-four participants completed lumbar punctures at baseline and 18 months after allocation to placebo (n = 15, 4 non-APOE4 carriers, 11 APOE4 carriers) or DHA (n = 29, 4 non-APOE4 carriers, 25 APOE4 carriers). The difference (Δ) in CSF DHA levels between placebo and treatment arms was 59% lower in APOE4 carriers compared to APOE4 non-carriers. With the observed SD of 1.027, this yields an effect size of 0.78, and an effect size for the interaction effect of 0.29. To detect a treatment APOE4 interaction with 80% power and α= 0.05 (two sided) yields a required sample size of 96 (24 per treatment/APOE4 cell). Therefore, a sample size of 275 would be sufficient to detect an interaction of plasma DHA by genotype in this trial.

Statistical analysis

Statistical analysis was conducted using R Studio software version 3.5.

Intervention changes in fatty acid ratios were calculated by taking the difference between 18-month and baseline measures. Baseline hippocampal volumes were divided by respective ICV measures. Analysis for post-intervention hippocampal volumes were also calculated by taking the difference between the absolute values of the 18-month and baseline hippocampus volume (without dividing by ICV). At baseline, the percentage of fatty acids by the individual APOE genotype subgroups (APOE ɛ2/ɛ3, ɛ2/ɛ4, ɛ3/ɛ3, ɛ3/ɛ4, and ɛ4/ɛ4) was compared using a univariate linear regression model. 18-months fatty acid ratios were modeled as a function of treatment group and APOE genotype in individual subgroups and their interaction using a multivariable linear regression model. An interaction model between APOE genotype subgroups and sex was modeled for plasma fatty acid ratios. The association between plasma DHA/AA or EPA/AA and hippocampal volumes was studied using a Pearson correlation coefficient and a multivariate regression model to adjust for APOE combined subgroups (APOE4 and non carriers). An interaction model for APOE genotype and fatty acid ratios on the change in hippocampal volumes was also performed. Given the APOE allele-dose effect, with APOE ɛ4/ɛ4 having a stronger effect on AD risk and lipid metabolism than the other alleles, we used APOE ɛ4/ɛ4 carriers as the reference group. A 2-sided 5.0% alpha-level was defined as significant. For this exploratory analysis, we did not control for multiple comparisons, and therefore, results with a p < 0.05 listed throughout the paper should be considered only nominally significant.

RESULTS

Study sample description

Fatty acid and APOE genotype information was available in 275 participants at baseline and 18 months following randomization to DHA supplementation or placebo. 161 were randomized to 2 grams of DHA per day and 114 were randomized to a placebo. A summary of the number of participants by APOE genotype is presented in Table 1. DHA dietary intake did not differ among APOE genotype subgroups (group comparison, p > 0.3).

Table 1

DHA dietary intake by APOE genotype at baseline

APOE genotype	Sample Size N (%)	Baseline DHA Intake (g) mean (sd)
ɛ2/ɛ3	9 (3)	106.55 (60.27)
ɛ2/ɛ4	9 (3)	81.33 (52.73)
ɛ3/ɛ3	88 (32)	96.40 (51.20)
ɛ3/ɛ4	122 (44)	87.00 (57.22)
ɛ4/ɛ4	47 (18)	93.77 (57.95)

Baseline plasma DHA, EPA, and AA within HUFAs and APOE genotype

Among plasma HUFAs, the percentage of AA was significantly greater in APOE ɛ4/ɛ4 compared to ɛ3/ɛ3 (p = 0.027), ɛ2/ɛ3 (p = 0.049), and ɛ2/ɛ4 carriers (p = 0.0019, Fig. 2A). In contrast, there was a trend toward a lower ratio of plasma DHA to HUFAs in APOE ɛ4/ɛ4 compared to ɛ3/ɛ3 (p = 0.068, Fig. 2B). Similarly, the ratio of ALA to HUFA trended toward a lower value in APOE ɛ4/ɛ4 compared to ɛ3/ɛ3 (p = 0.061, Fig. 2C). In addition, baseline plasma ALA/HUFA was significantly lower in APOE ɛ4/ɛ4 compared to ɛ2/ɛ3 (p = 0.031). The ratio of plasma EPA to HUFA did not differ by genotype at baseline (Fig. 2D). For plasma DHA/AA at baseline, APOE ɛ3/ɛ3 carriers exhibited a small but significantly greater ratio (p = 0.045) than ɛ4/ɛ4 carriers (Fig. 3A). Similarly, DHA/AA trended to be greater in APOE ɛ2/ɛ3 carriers compared to ɛ4/ɛ4 carriers (p = 0.078), and was significantly greater in APOE ɛ2/ɛ4 carriers (p = 0.01) compared to ɛ4/ɛ4 carriers. The plasma EPA/AA at baseline in APOE ɛ3/ɛ3 carriers was not significantly different from ɛ4/ɛ4 carriers (Fig. 3B). However, baseline plasma EPA/AA trended toward being greater in APOE ɛ2/ɛ3 (p = 0.061) and ɛ2/ɛ4 carriers (p = 0.084) compared to ɛ4/ɛ4 carriers. For plasma ALA/AA at baseline, APOE ɛ2/ɛ3 carriers exhibited a small but significantly greater ratio (p = 0.01) than ɛ4/ɛ4 carriers (Fig. 3C). In a similar fashion, baseline plasma ALA/AA was significantly greater in APOE ɛ3/ɛ3 carriers (p = 0.02) compared to ɛ4/ɛ4 carriers. Baseline plasma ALA/AA was also significantly greater in APOE ɛ2/ɛ4 carriers (p = 0.003) compared to ɛ4/ɛ4 carriers. However, after removing the outlier in the APOE ɛ2/ɛ4 group, the difference between ɛ2/ɛ4 and ɛ4/ɛ4 became non-significant. The ratio of all individual fatty acids of the total fatty acids in plasma by APOE genotype is presented in Supplementary Table 1.

Fig.2

Distribution of HUFAs by APOE genotype at baseline. Comparison of baseline plasma ratios of (A) AA/HUFA, (B) DHA/HUFA, (C) ALA/HUFA, and (D) EPA/HUFA among APOE subgroups was performed using a linear regression and p < 0.05 was defined as significant (n = 275, ɛ2/ɛ3 n = 9, ɛ2/ɛ4 n = 9, ɛ3/ɛ3 n = 88, ɛ3/ɛ4 n = 122, and ɛ4/ɛ4 n = 47). The group comparison p values to ɛ4/ɛ4 are shown in the figures.

Fig.3

Distribution of DHA/AA, EPA/AA, and ALA/AA by APOE genotype at baseline. Comparison of baseline plasma ratios of (A) DHA/AA, (B) EPA/AA, and (C) ALA/AA, among APOE subgroups was performed using a linear regression and sample sizes (n = 275, ɛ2/ɛ3 n = 9, ɛ2/ɛ4 n = 9, ɛ3/ɛ3 n = 88, ɛ3/ɛ4 n = 122, and ɛ4/ɛ4 n = 47). The group comparison p values to ɛ4/ɛ4 are shown in the figures. After exclusion of the APOE ɛ2/ɛ4 outlier in C, the difference between the baseline plasma ALA/AA levels became non-significant between APOE ɛ2/ɛ4 and ɛ4/ɛ4.

Effect of sex on the association of APOE genotype with plasma fatty acids

Female sex appeared to influence the association of genotype with HUFAs at baseline, and this effect was stronger for EPA than DHA. We examined the interaction of APOE genotype and female sex on plasma fatty acid ratios. The percentage of DHA of total fatty acids (interaction p value = 0.071) and as a ratio to AA (interaction p value = 0.083) trended to differ by genotype and sex (Fig. 4A, B). For men, APOE4 carriers had a higher ratio than non-carriers. In contrast, in women, APOE4 carriers had a lower ratio than non-carriers. The interaction of sex and genotype on the ratio of DHA to HUFA was not significant. However, after the removal of an outlier in the APOE4 group of women, the sex-genotype interaction on fatty acid ratio became significant (p value = 0.048, Fig. 4C). EPA as a percentage of total fatty acids, as a ratio to HUFAs, and as a ratio to AA significantly differed in APOE4 women compared to men (interaction p value < 0.05 for all) in a similar pattern to DHA (Fig. 4D-F).

Fig.4

Interaction of sex and APOE genotype. The interaction between sex and APOE4 at baseline for (A) percent plasma DHA of total fatty acids, (B) ratio of plasma DHA/AA, (C) ratio of plasma DHA/HUFA, (D) percent plasma EPA of total fatty acids, (E) ratio of plasma EPA/AA, and (F) ratio of plasma EPA/HUFA. Analysis was performed using a linear regression and sample sizes (n = 275, non-APOE4 males n = 29, non-APOE4 females n = 68, APOE4 males n = 64, and APOE4 females n = 114). After removing the APOE4 female outlier, the interaction between sex and APOE4 status became more significant in all cases, making ratio of plasma DHA/HUFA significant, however percent plasma DHA of total fatty acids and ratio of plasma DHA/AA remained non-significant. The red point indicates the mean per group.

Ratio of DHA and EPA of HUFAs at 18 months following supplementation

At 18 months following DHA supplementation, the ratios of plasma n-6 DPA, AA, and n-3 DPA to HUFAs were significantly decreased (p < 0.0001 for all, Fig. 5A-C). It is known that plasma EPA levels increase after DHA supplementation by retro conversion of DHA to EPA [30]. The ratios of both EPA and DHA to HUFAs increased after DHA supplementation (p < 0.0001 for all, Fig. 5D, E). These changes reflect the competition of dietary DHA with other HUFAs at the sn-2 position of phospholipids where an increase in DHA intake is able to incorporate into phospholipids replacing AA, n-6 DPA, and n-3 DPA. To understand the effect of APOE genotype on these fatty acid ratios after supplementation, AA/HUFA in APOE ɛ4/ɛ4 had the lowest decrease in the treatment arm (Fig. 5B), compared to all other genotypes except ɛ2/ɛ4. There were no differences by genotype in the change of n-3 and n-6 DPA/HUFA (Fig. 5A, C). The increase in EPA/HUFA at 18-months was greater in APOE ɛ2/ɛ3 compared with ɛ4/ɛ4 (p = 0.006, Fig. 5D). There was a significant interaction between APOE ɛ4/ɛ4 subgroup and treatment arm on the 18 months EPA/HUFA (interaction p value = 0.041). The increase in DHA/HUFA was least pronounced in the APOE ɛ4/ɛ4 group compared to all other genotypes except APOE ɛ2/ɛ4 (p < 0.05 for all, Fig. 5E). There was no significant difference in change in DHA/HUFA and EPA/HUFA as a function of sex. The 18-month ratio of individual fatty acids of total fatty acids by APOE genotype are presented in Supplementary Table 2.

Fig.5

Post-intervention difference in 18-month and baseline ratios of (A) n-6 DPA/HUFA, B) AA/HUFA, (C) n-3 DPA/HUFA, (D) EPA/HUFA, and (E) DHA/HUFA in plasma of both treatment groups. The change in plasma fatty acids was modeled as a function of treatment arm and APOE subgroups using a multivariate linear regression. The group comparison p values to APOE ɛ4/ɛ4 are shown in the figures. Sample sizes (total n = 275; DHA-treated total n = 161, DHA-treated ɛ2/ɛ3 n = 5, DHA-treated ɛ2/ɛ4 n = 6, DHA-treated ɛ3/ɛ3 n = 50, DHA-treated ɛ3ɛ4 n = 75, and DHA-treated ɛ4/ɛ4 n = 25; placebo-treated total n = 114, placebo-treated ɛ2/ɛ3 n = 4, placebo-treated ɛ2/ɛ4 n = 3, placebo-treated ɛ3/ɛ3 n = 38, placebo-treated ɛ3/ɛ4 n = 47, and placebo-treated ɛ4/ɛ4 n = 22).

Ratio of plasma DHA and EPA to AA at 18 months following supplementation

There was a significant difference between the placebo and DHA-treatment arms in plasma DHA/AA and EPA/AA after the intervention (p < 0.001, Fig. 6A, B). DHA-treated APOE ɛ2/ɛ3 carriers exhibited a 23% greater increase in plasma DHA/AA compared to DHA-treated ɛ4/ɛ4 carriers (p = 0.002, Fig. 6A). DHA-treated APOE ɛ2/ɛ3 carriers exhibited a 38% greater increase in plasma DHA/AA compared to DHA-treated ɛ4/ɛ4 carriers (p = 0.006). DHA-treated APOE ɛ3/ɛ4 carriers exhibited a small, but significantly greater increase in plasma DHA/AA compared to DHA-treated ɛ4ɛ4 carriers (p = 0.031). APOE ɛ2/ɛ4 trended toward a slightly greater increase in plasma DHA/AA compared to ɛ4/ɛ4; however, this trend was non-significant. There was no significant sex and genotype interaction on the change of plasma DHA/AA after supplementation.

Fig.6

Post-intervention difference in 18-month and baseline ratios of (A) DHA/AA and (B) EPA/AA in plasma of both treatment groups. The change in DHA/AA and EPA/AA were modeled as function of the treatment arm and APOE subgroups using a multivariate linear regression. The group comparison p values to APOE ɛ4/ɛ4 are shown in the figures. Sample sizes (total n = 275; DHA-treated total n = 161, DHA-treated ɛ2/ɛ3 n = 5, DHA-treated ɛ2/ɛ4 n = 6, DHA-treated ɛ3/ɛ3 n = 50, DHA-treated ɛ3/ɛ4 n = 75, and DHA-treated ɛ4/ɛ4 n = 25; placebo-treated total n = 114, placebo-treated ɛ2/ɛ3 n = 4, placebo-treated ɛ2/ɛ4 n = 3, placebo-treated ɛ3/ɛ3 n = 38, placebo-treated ɛ3/ɛ4 n = 47, and placebo-treated ɛ4/ɛ4 n = 22).

Similarly, DHA-treated APOE ɛ3/ɛ3 carriers exhibited a 23% greater increase in plasma EPA/AA compared to DHA-treated ɛ4/ɛ4 carriers (p = 0.036 Fig. 6B). In addition, DHA-treated APOE ɛ2/ɛ3 carriers exhibited a 57% significantly greater increase in plasma EPA/AA compared to DHA-treated ɛ4/ɛ4 carriers (p = 0.0001). APOE ɛ4/ɛ4 carriers trended toward slightly less change in plasma EPA/AA after supplementation compared to both ɛ2/ɛ4 and ɛ3/ɛ4 carriers. There was a significant interaction between APOE ɛ4/ɛ4 subgroup and treatment arm on the 18 months EPA/AA (interaction p value = 0.021). There was no significant interaction between the change in DHA/AA and EPA/AA as a function of sex and genotype.

Association of plasma and CSF fatty acids by APOE genotype

There was a significant association between the ratio of DHA/AA (r = 0.47, p = 0.0004) and EPA/AA (r = 0.32, p = 0.019) measured in plasma and CSF in a subset of individuals completing both studies (n = 53). APOE genotype affected this relationship. There was a trend toward a stronger association of CSF DHA/AA to plasma DHA/AA in APOE4 non-carriers than in APOE4 carriers (interaction p value = 0.06, Fig. 7A). In the treatment arm (n = 29), there was a strong and significant association between the change in plasma with CSF DHA/AA (r = 0.82, p < 0.0001). Similarly, this association trended to be stronger in APOE4 non-carriers compared to carriers (interaction p value = 0.09, Fig. 7B). The association of the change in CSF EPA/AA with plasma EPA/AA was non-significant (p = 0.18) and was not influenced by APOE genotype (Fig. 7C, D).

Fig.7

Correlation between (A) DHA/AA in plasma and CSF at baseline (total n = 53, non- APOE4 n = 12, APOE4 n = 41; r = 0.47, p = 0.0004), (B) change in DHA/AA in plasma and CSF post-18-month-intervention (total n = 29, non-APOE4 n = 4, APOE4 n = 25; r = 0.82, p < 0.0001), (C) EPA/AA in plasma and CSF at baseline (total n = 53, non-APOE4 n = 12, APOE4 n = 41; r = 0.32, p = 0.019), and (D) change in EPA/AA in plasma and CSF post-18-month-intervention (total n = 29, non-APOE4 n = 4, APOE4 n = 25; r = 0.26, p = 0.18) by APOE4 status. The interaction between APOE genotype subgroups and among baseline and changes in plasma DHA/AA were non-significant. Analysis was done using Pearson’s correlation and interaction modeling.

Association of APOE genotype with hippocampal volume

Hippocampal volumes provide a surrogate brain efficacy biomarker for n-3 fatty acid supplementation and are affected by APOE genotype. In the whole group, baseline left hippocampal volume did not differ significantly among APOE ɛ2/ɛ3 carriers, ɛ2/ɛ4 carriers, and ɛ3/ɛ3 carriers when compared to ɛ4/ɛ4 carriers. Similarly, the right hippocampal volume did not differ among APOE ɛ2/ɛ3 carriers, ɛ2/ɛ4 carriers, and ɛ3/ɛ3 carriers compared to ɛ4/ɛ4 carriers (Table 2).

Table 2

Baseline Hippocampus Volume as percent of ICV (ratio)

APOE genotype	Sample Size N (%)	Baseline Left Hippocampus mean (sd)	Baseline Right Hippocampus mean (sd)
ɛ2/ɛ3	3 (3.49)	0.27 (0.12)	0.27 (0.11)
ɛ2/ɛ4	2 (2.33)	0.26 (0.004)	0.27 (0.05)
ɛ3/ɛ3	28 (32.56)	0.26 (0.06)	0.26 (0.06)
ɛ3/ɛ4	41 (47.67)	0.25 (0.05)	0.25 (0.05)
ɛ4/ɛ4	12 (13.95)	0.23 (0.05)	0.23 (0.04)

All baseline hippocampal volume/ICV ratios were multiplied by100 for presentation purposes. Data presented are means and SD.

In the combined sample of DHA- and placebo-treated participants, the 18-month change in left and right hippocampal volume was investigated by APOE genotype individual groups. APOE ɛ2/ɛ3 carriers exhibited a 10% less decline in left hippocampal volume than APOE ɛ4/ɛ4 carriers (p < 0.001, Fig. 8A). Additionally, APOE ɛ2/ɛ3 carriers (p = 0.0052) and APOE ɛ2/ɛ4 (p = 0.007) carriers experienced significantly less decline in the right hippocampal volume than APOE ɛ4/ɛ4 carriers (p < 0.01, Fig. 8B).

Fig.8

Post-intervention difference in 18-month and baseline A) right and B) left hippocampal volume (cm³) among APOE subgroups in both DHA-treated participants (n = 86, ɛ2/ɛ3 n = 3, ɛ2/ɛ4 n = 2, ɛ3/ɛ3 n = 28, ɛ3/ɛ4 n = 41, and ɛ4/ɛ4 n = 12). Comparison of the change in hippocampal volumes as a function of APOE subgroup was performed using a linear regression. The group comparison p values to ɛ4/ɛ4 are shown in the figures.

The association of plasma DHA/AA and EPA/AA with hippocampal volume

At baseline, plasma DHA/AA and EPA/AA ratios were not correlated with the baseline left or right hippocampal volumes. However, greater baseline plasma DHA/AA was significantly correlated with less decline in right hippocampal volume (r = 0.38, p = 0.0003, Fig. 9A). Using a multivariable linear regression model, baseline DHA/AA and APOE groups were both independently correlated (p < 0.001) with the right hippocampal volume change and there was no significant interaction between plasma DHA/AA and APOE genotype on the change in right hippocampal volume. Furthermore, baseline plasma DHA/AA was not significantly associated with decline in left hippocampal volume.

Fig.9

Post-intervention difference in 18-month and baseline right hippocampal volume (cm³) was correlated with (A) baseline plasma DHA/AA (total n = 86, non-APOE4 n = 31, APOE4 n = 55; r = 0.38, p = 0.0003) and (B) change in plasma DHA/AA (total n = 48, non-APOE4 n = 19, APOE4 n = 29; r = 0.25, p = 0.09) in the DHA-treated arm. Post-intervention difference in 18-month and baseline right hippocampal volume (cm³) was correlated with (C) baseline plasma EPA/AA (total n = 86, non-APOE4 n = 31, APOE4 n = 55; r = 0.24, p = 0.03) and (D) change in plasma EPA/AA (total n = 48, non-APOE4 n = 19, APOE4 n = 29; r = 0.32, p = 0.03) in the DHA-treated arm. The interaction between APOE genotype subgroups and the change in plasma EPA/AA on the change in right hippocampal volume in was significant (p = 0.01). Analysis was done using Pearson’s correlation and an interaction model. Exclusion of the ɛ4 outlier made both the correlation between change in right hippocampal volume and both (C) baseline plasma EPA/AA (r = 0.30, p = 0.005), as well as (D) change in plasma EPA/AA (r = 0.42, p = 0.003) more significant. Exclusion of the outlier, however, made the interaction between APOE subgroups and change in plasma EPA/AA on the change in right hippocampal volume non-significant (p = 0.06).

Higher baseline plasma EPA/AA was significantly correlated with less decline in right hippocampal volume (r = 0.24, p = 0.03, Fig. 9C). Using a multivariable linear regression model, baseline EPA/AA and APOE groups were independently correlated (p < 0.01 for both) with the right hippocampal volume change. In addition, there was no significant interaction between plasma EPA/AA and APOE genotype on the change of the right hippocampal volume. Baseline plasma EPA/AA was not significantly associated with decline in left hippocampal volume.

After the intervention, changes in right and left hippocampal volumes did not significantly differ between the DHA and placebo study groups. Furthermore, in the DHA arm, there was no significant association between increasing plasma DHA/AA ratio and decline in right or left hippocampal volume after supplementation (Fig. 9B). However, there was a significant association between increasing plasma EPA/AA ratio and decline in right hippocampal volume after DHA supplementation (r = 0.31, p = 0.03, Fig. 9D). Using a multivariable linear regression model, the change of EPA/AA and APOE groups were independently correlated (p < 0.01 for both) with right hippocampal volume change. In addition, there was a significant interaction between plasma EPA/AA and APOE genotype on the change of right hippocampal volume (p = 0.01). The increase in plasma EPA/AA was correlated with the less decline in right hippocampal volume in APOE4 non-carriers, but not in APOE4 carriers.

DISCUSSION

In this study, patients with APOE4 and mild probable AD had a lower increase in DHA/AA and EPA/AA after DHA supplementation compared to patients with mild AD carrying APOE2 or APOE3. Plasma and CSF DHA/AA were strongly correlated but higher plasma DHA/AA did not translate as efficiently into higher CSF DHA/AA in APOE4. APOE2 carriers with mild AD had the least decline in hippocampal volumes after the 18-month follow-up. The greater increase in EPA/AA in the DHA treatment arm was associated with lower decline in the right hippocampal volume, but only in APOE4 non-carriers. These data suggest that APOE4 reduces CSF DHA and EPA delivery by either affecting transport across the blood-brain barrier or through increased catabolism, in addition to its effects on plasma DHA and EPA.

These findings confirm earlier reports that n-3 PUFA metabolism is dysregulated in APOE4 (illustrated in Fig. 10C). We hypothesize that the inability of APOE4 carriers to incorporate DHA into plasma phospholipids may be due to the preferential oxidation of this fatty acid in tissues. In addition, we hypothesize that APOE2 limits this oxidation. This hypothesis is supported by several lines of investigation. After consumption of [¹³C]DHA, APOE4 participants had a 77% shorter whole-body half-life of [¹³C]DHA compared to non-ɛ4 carriers [14]. In an earlier study, we found an increase in [¹¹C]DHA brain uptake using PET scanning in younger cognitively normal APOE4 carriers compared to non-carriers; supporting greater brain DHA loss that is compensated for by an increase in DHA transport from plasma to brain in APOE4. The mechanisms for this DHA brain loss may include greater n-3 PUFA oxidation with APOE4. CPT1, the rate limiting enzyme of long-chain fatty acid oxidation, was expressed at greater levels in APOE4-TR mice livers than in APOE3-TR mice [17]. This was associated with lower n-3 PUFA concentrations in the liver, plasma, and adipose tissues of APOE4-TR mice models. Taken together, APOE4 appears to accelerate n-3 PUFA’s mitochondrial oxidation leading to lower levels of esterified DHA and EPA and compromising brain delivery after supplementation [17].

Fig.10

A) Competition between esterified DHA and AA in triglycerides (TG) and phospholipids (PL) from circulatory lipoproteins into the sn-2 position of phospholipids of the membrane bilayer by lipases. Due to greater intake of AA in the western diet, more AA is incorporated into the membrane bilayer. B) Ingestion of dietary DHA after supplementation remodels membrane phospholipid bilayer by the activities of cPLA₂ and the acyl-coA transferanse. In addition DHA is able to retroconvert into EPA. EPA also incorporated into the bilayer. C) Preferential mitochondrial oxidation of DHA by the APOE4 genotype is associated with greater expression of mitochondrial CPT1, providing a mechanism for the lower incorporation of DHA into the membrane bilayer.

The ratio of n-3 HUFAs to total HUFAs or to AA represents a meaningful assessment of tissue levels [3]. The absolute concentrations of these fatty acids also serve as a useful indicator of tissue levels as there is a strong correlation between absolute and relative (% weight) plasma ratios [32]. Although the weight % value (percentage of all plasma fatty acids) has a simple rationale and is widely reported, it has no clear metabolic or biological significance. This is because the HUFA balance in tissues results from the HUFA indiscrimate action of calcium-dependent phospholipase A2 (cPLA2) or the acyl transferase that regulate HUFA recycling into and out of the sn-2 position of phospholipids. Unlike HUFAs, less saturated, unsaturated, and shorter fatty acids have less ability to recycle in and out of the sn-2 position of phospholipids. When DHA and EPA intake are increased, they replace n-6 DPA, n-3 DPA, or AA as seen in Fig. 5 and illustrated in Fig. 10.

The biological effects of these fatty acids (for example, eicosanoid release) is dependent on the activity of cPLA2 [33], which is increased in association with amyloid plaques in the human brain [34]. Increased cPLA2 activity is involved in AD pathogenesis [1] by releasing the relatively abundant AA from membrane phospholipids; this process directly contributes to changes in synaptic activity, neuroinflammation, and cognitive changes in AD mouse models [1]. At baseline, the mean ratio of plasma DHA/AA was 0.26 and the mean ratio of plasma EPA/AA was 0.07, reflecting the relatively greater consumption of n-6 s compared to n-3 s in the American diet. In the treatment arm, the mean 18-month plasma DHA/AA increased to 1.24, and the mean 18-month plasma EPA/AA increased to 0.23. This increase after supplementation was modest and was further attenuated by APOE4. In comparison, an EPA/AA of greater than 0.7 in a Japanese population with a high consumption of n-3 s predicted lower incidence of cardiovascular disease events [35]. It is interesting to note that the greatest increase in EPA/AA was observed in APOE ɛ2/ɛ3 carriers (mean ratio of 0.5 after supplementation); this genotype had the least decline in hippocampal volume after 18 months, and is protected against AD.

Cellular studies demonstrate that DHA, EPA, and AA play key roles in several metabolic processes relevant to AD pathogenesis and neuroinflammation [1, 36]. High levels of AA affect hippocampal function and AD in many detrimental ways. In one process, AA increases the activity of α-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid receptors (AMPARs) [37]. These receptors cause increased stimulation of nerves by neurotransmitters due to increased mediation in the c-Jun n-terminal kinase (JNK) pathway [38], providing excitotoxic injury to the brain. AA in the brain may also cause oxidative stress [1]. In the presence of high levels of AA, neuronal activity undergoes an abnormal increase that reduces neuron survivability [4]. Additionally, acrolein, a byproduct of AA, has been shown to be detrimental to AD brains as it contributes to oxidative stress and lipid peroxidation [39]. In contrast to AA, DHA increases the functionality of the hippocampus with regards to volume and memory capacity/capability. In a rodent study, hippocampal volumes increased along with increased performance in behavior and learning tests after DHA supplementation [5]. EPA has stronger anti-inflammatory effects than DHA [40], and higher doses of EPA supplementation have recently been shown to reduce cardiovascular events in a large clinical trial [41].

Female sex affects how the APOE genotype influences the abundance of HUFAs. Women in the age range of 65–75 may have a higher increased risk of AD than men [42]. APOE-TR mice exhibit a sex-specific effect of APOE4 on plasma fatty acids [43]. Female APOE4-TR mice had lower DHA/AA in the cortex at 18 months of age (equivalent to 60 years in humans) compared to APOE3-TR mice. These differences were less prominent in male mice. Consistent with these observations, we found a trend for lower DHA/AA and significantly lower EPA/AA in APOE4 women compared to men at baseline. One possible explanation for this observation is the greater increase in mitochondrial PUFA oxidation in women carrying APOE4 that is magnified by aging and the post-menopausal state.

Our findings of greater hippocampal volume loss in APOE4 carriers and lower loss in APOE2 carriers is consistent with prior studies. In an Alzheimer’s Disease Neuroimaging Initiative (ADNI) study, 112 cognitive normal elderly individuals, 226 individuals with mild cognitive impairment, and 96 AD patients completed three MRIs over a 1-year span [44]. The presence of the APOE4 allele was associated with higher rates of hippocampal volume loss in AD patients [44]. Expansion of this study into additional ADNI cohorts (n = 725) confirmed lower hippocampal volumes in both APOE4 carriers with and without AD; the difference in volumes was more pronounced with APOE4 homozygosity [45]. In another study, the rate of hippocampal atrophy was examined in APOE2 elderly normal subjects [46]. With a total of 134 subjects, 27 of whom were APOE ɛ2/ɛ3 or APOE ɛ2/ɛ3 and 107 of whom were APOE ɛ3/ɛ3, the rate of hippocampal decrease in volume in APOE 2 carriers was significantly slower compared to the other groups.

Both cross-sectional and longitudinal studies have reported a positive association between plasma DHA and EPA levels with hippocampal volumes [20, 22]. In this study, we found an association between greater baseline DHA/AA and EPA/AA plasma levels and lower loss of the right but not left hippocampal volumes. The reason for this asymmetric association is not clear. In one longitudinal study of 245 participants who had been followed for up to 18 years, the volume of the right hippocampus but not left was associated with time to symptom onset in persons with mild cognitive impairment [47]. In the ADNI cohorts, there was greater atrophy of the left hippocampus than the right, and this asymmetry was more pronounced in APOE4 homozygotes than heterozygotes [45, 48]. However, we did not find significant differences between left and right hippocampal volume loss in this study population and it was not affected by APOE4 homozygosity.

This study has several strengths and weaknesses. The main strength of this study is the relatively large number of samples from participants with APOE ɛ4/ɛ4 (n = 47) allowing evaluation of contributions of E4 homozygosity on plasma DHA, EPA, and AA. Participants were carefully followed with multiple visits over 18 months in the setting of a clinical trial that included assessments of dietary n-3 intake, compliance by pill counting, imaging, and clinical outcomes. Even so, the study also has several limitations. Our findings are a post-hoc analysis and hypothesis generating. Notwithstanding, the aims of this analysis were internally reviewed and approved by the ADCS publication committee prior to the data analysis. In addition, we did not correct for multiple comparisons, and therefore, our results can only be considered nominally significant. These results should be validated in additional large sample studies. The placebo intervention contained soy/corn oil. However, we did not observe significant changes in the placebo arm in plasma fatty acids (% by weight) at 18 months compared to baseline. The sample size for the APOE 2 carriers with hippocampal volumes was low, but this would be expected given the protective effect of APOE 2 on AD. In addition, our finding of a protective effect of APOE 2 on hippocampal volume loss is consistent with past literature from ADNI. Another limitation of this study was that DHA/AA and EPA/AA ratios were associated with change in right but not left hippocampal volumes; this may have been a chance finding related to normal variability in a limited sample size. In addition, the lack of significant differences in hippocampal volumes in some of the genotype groups is likely due to the small sample size of this sub-study. Finally, the AD diagnosis was clinical as there was no amyloid imaging and limited CSF exams. Therefore, the probability of making an accurate diagnosis of AD is less certain, particularly among non-APOE 4 carriers. Therefore, interpretation of the differences in plasma DHA/AA and EPA/AA may be confounded by the clinical diagnosis.

We conclude that APOE4 carriers with mild AD may be less responsive to increases in plasma DHA/AA or EPA/AA after DHA supplementation, providing one explanation for the differential clinical response to DHA supplementation by APOE4 status. The increase in plasma DHA/AA and EPA/AA was greatest in APOE2 carriers with mild AD and was associated with lower decline in right hippocampal volume. Understanding how APOE genotype affects plasma DHA/AA and EPA/AA ratios may offer critical insights helping to reverse or slow down APOE4-driven AD.

Footnotes

ACKNOWLEDGMENTS

HNY was supported by R21AG056518, R01AG055770, and R01AG054434 from the National Institute of Aging. This work was also supported by P50AG05142 from the National Institutes of Health. We thank Dr. Carol Evans, Curtis Taylor, and the ADCS DHA Coordinating Center for facilitating access to the data. We thank Dr. Michael Weiner for critically reviewing the manuscript.

Authors’ disclosures available online ().

The supplementary material is available in the electronic version of this article:

References

Sanchez-Mejia

, Mucke

(2010) Phospholipase A2 and arachidonic acid in Alzheimer’s disease. Biochim Biophys Acta 1801, 784–790.

Yassine

, Braskie

, Mack

, Castor

, Fonteh

, Schneider

, Harrington

, Chui

(2017) Association of docosahexaenoic acid supplementation with Alzheimer disease stage in apolipoprotein E epsilon4 carriers: A review. JAMA Neurol 74, 339–347.

Bibus

, Lands

(2015) Balancing proportions of competing omega-3 and omega-6 highly unsaturated fatty acids (HUFA) in tissue lipids. Prostaglandins Leukot Essent Fatty Acids 99, 19–23.

Okuda

, Saito

, Katsuki

(1994) Arachidonic acid: Toxic and trophic effects on cultured hippocampal neurons. Neuroscience 63, 691–699.

Cutuli

, De Bartolo

, Caporali

, Laricchiuta

, Foti

, Ronci

, Rossi

, Neri

, Spalletta

, Caltagirone

, Farioli-Vecchioli

, Petrosini

(2014) n-3 polyunsaturated fatty acids supplementation enhances hippocampal functionality in aged mice. Front Aging Neurosci 6, 220.

Liu

C-C

, Kanekiyo

, Xu

, Bu

(2013) Apolipoprotein E and Alzheimer disease: Risk, mechanisms, and therapy. Nat Rev Neurol 9, 106–118.

Wolters

, Yang

, Biggs

, Jakobsdottir

, Li

, Evans

, Bis

, Harris

, Vasan

, Zilhao

, Ghanbari

, Ikram

, Launer

, Psaty

, Tranah

, Kulminski

, Gudnason

, Seshadri

, E2-CHARGE investigators (2019) The impact of APOE genotype on survival: Results of 38,537 participants from six population-based cohorts (E2-CHARGE). PloS One 14, e0219668e0219668.

Tomiyasu

, Walsh

, Ikewaki

, Judge

, Sacks

(2001) Differential metabolism of human VLDL according to content of ApoE and ApoC-III. Arterioscler Thromb Vasc Biol 21, 1494–1500.

, Dhanasekaran

, Alexander

, Rader

, Phillips

, Lund-Katz

(2013) Molecular mechanisms responsible for the differential effects of apoE3 and apoE4 on plasma lipoprotein-cholesterol levels. Arterioscler Thromb Vasc Biol 33, 687–693.

10.

Michikawa

, Fan

, Isobe

, Yanagisawa

(2000) Apolipoprotein E exhibits isoform-specific promotion of lipid efflux from astrocytes and neurons in culture. J Neurochem 74, 1008–1016.

11.

Weintraub

, Eisenberg

, Breslow

(1987) Dietary fat clearance in normal subjects is regulated by genetic variation in apolipoprotein E. J Clin Invest 80, 1571–1577.

12.

Nock

, Chouinard-Watkins

, Plourde

(2017) Carriers of an apolipoprotein E epsilon 4 allele are more vulnerable to a dietary deficiency in omega-3 fatty acids and cognitive decline. Biochim Biophys Acta 1862, 1068–1078.

13.

Cunnane

, Plourde

, Pifferi

, Bégin

, Féart

, Barberger-Gateau

(2009) Fish, docosahexaenoic acid and Alzheimer’s disease. Prog Lipid Res 48, 239–256.

14.

Chouinard-Watkins

, Rioux-Perreault

, Fortier

, Tremblay-Mercier

, Zhang

, Lawrence

, Vohl

, Perron

, Lorrain

, Brenna

, Cunnane

, Plourde

(2013) Disturbance in uniformly 13C-labelled DHA metabolism in elderly human subjects carrying the apoE epsilon4 allele. Br J Nutr 110, 1751–1759.

15.

Chouinard-Watkins

, Conway

, Minihane

, Jackson

, Lovegrove

, Plourde

(2015) Interaction between BMI and APOE genotype is associated with changes in the plasma long-chain-PUFA response to a fish-oil supplement in healthy participants. Am J Clin Nutr 102, 505–513.

16.

Abdullah

, Evans

, Emmerich

, Crynen

, Shackleton

, Keegan

, Luis

, Tai

, LaDu

, Mullan

, Crawford

, Bachmeier

(2017) APOE epsilon4 specific imbalance of arachidonic acid and docosahexaenoic acid in serum phospholipids identifies individuals with preclinical mild cognitive impairment/Alzheimer’s disease. Aging (Albany NY) 9, 964–985.

17.

Conway

, Larouche

, Alata

, Vandal

, Calon

, Plourde

(2014) Apolipoprotein E isoforms disrupt long-chain fatty acid distribution in the plasma, the liver and the adipose tissue of mice. Prostaglandins Leukot Essent Fatty Acids 91, 261–267.

18.

Jack

Jr , Petersen

, Xu

, O’Brien

, Smith

, Ivnik

, Boeve

, Waring

, Tangalos

, Kokmen

(1999) Prediction of AD with MRI-based hippocampal volume in mild cognitive impairment. Neurology 52, 1397–1403.

19.

Schuff

, Woerner

, Boreta

, Kornfield

, Shaw

, Trojanowski

, Thompson

, Jack

Jr , Weiner

, Alzheimer’s Disease Neuroimaging Initiative (2009) MRI of hippocampal volume loss in early Alzheimer’s disease in relation to ApoE genotype and biomarkers. Brain 132, 1067–1077.

20.

Yassine

, Feng

, Azizkhanian

, Rawat

, Castor

, Fonteh

, Harrington

, Zheng

, Reed

, DeCarli

, Jagust

, Chui

(2016) Association of serum docosahexaenoic acid with cerebral amyloidosis. JAMA Neurol 73, 1208–1216.

21.

Tan

, Harris

, Beiser

, Au

, Himali

, Debette

, Pikula

, DeCarli

, Wolf

, Vasan

(2012) Red blood cell omega-3 fatty acid levels and markers of accelerated brain aging. Neurology 78, 658–664.

22.

Pottala

, Yaffe

, Robinson

, Espeland

, Wallace

, Harris

(2014) Higher RBC EPA + DHA corresponds with larger total brain and hippocampal volumes: WHIMS-MRI study. Neurology 82, 435–442.

23.

Soininen

, Solomon

, Visser

, Hendrix

, Blennow

, Kivipelto

, Hartmann

; LipiDiDiet clinical study group (2017) A randomised, double-blind, controlled trial. Lancet Neurol 16, 965–975.

24.

Quinn

, Raman

, Thomas

, Yurko-Mauro

, Nelson

, Van Dyck

, Galvin

, Emond

, Jack

Jr , Weiner

, Shinto

, Aisen

(2010) Docosahexaenoic acid supplementation and cognitive decline in Alzheimer disease: A randomized trial. JAMA 304, 1903–1911.

25.

Freund-Levi

, Eriksdotter-Jönhagen

, Cederholm

, Basun

, Faxén-Irving

, Garlind

, Vedin

, Vessby

, Wahlund

L-O

, Palmblad

(2006) ω-3 fatty acid treatment in 174 patients with mild to moderate Alzheimer disease: OmegAD Study: A randomized double-blind trial. Arch Neurol 63, 1402–1408.

26.

Yassine

, Rawat

, Mack

, Quinn

, Yurko-Mauro

, Bailey-Hall

, Aisen

, Chui

, Schneider

(2016) The effect of APOE genotype on the delivery of DHA to cerebrospinal fluid in Alzheimer’s disease. Alzheimers Res Ther 8, 25.

27.

Kuratko

(2013) Food-frequency questionnaire for assessing long-chain ω-3 fatty-acid intake. Nutrition 5, 807–808.

28.

Boccardi

, Bocchetta

, Apostolova

, Barnes

, Bartzokis

, Corbetta

, DeCarli

, deToledo-Morrell

, Firbank

, Ganzola

, Gerritsen

, Henneman

, Killiany

, Malykhin

, Pasqualetti

, Pruessner

, Redolfi

, Robitaille

, Soininen

, Tolomeo

, Wang

, Watson

, Wolf

, Duvernoy

, Duchesne

, Jack

Jr , Frisoni

(2015) Delphi definition of the EADC-ADNI Harmonized Protocol for hippocampal segmentation on magnetic resonance. Alzheimers Dement 11, 126–138.

29.

Frisoni

, Jack

Jr , Bocchetta

, Bauer

, Frederiksen

, Liu

, Preboske

, Swihart

, Blair

, Cavedo

, Grothe

, Lanfredi

, Martinez

, Nishikawa

, Portegies

, Stoub

, Ward

, Apostolova

, Ganzola

, Wolf

, Barkhof

, Bartzokis

, DeCarli

, Csernansky

, deToledo-Morrell

, Geerlings

, Kaye

, Killiany

, Lehericy

, Matsuda

, O’Brien

, Silbert

, Scheltens

, Soininen

, Teipel

, Waldemar

, Fellgiebel

, Barnes

, Firbank

, Gerritsen

, Henneman

, Malykhin

, Pruessner

, Wang

, Watson

, Wolf

, deLeon

, Pantel

, Ferrari

, Bosco

, Pasqualetti

, Duchesne

, Duvernoy

, Boccardi

(2015) The EADC-ADNI Harmonized Protocol for manual hippocampal segmentation on magnetic resonance: Evidence of validity. Alzheimers Dement 11, 111–125.

30.

Metherel

, Chouinard-Watkins

, Trepanier

, Lacombe

RJS

, Bazinet

(2017) Retroconversion is a minor contributor to increases in eicosapentaenoic acid following docosahexaenoic acid feeding as determined by compound specific isotope analysis in rat liver. Nutr Metab (Lond) 14, 75.

31.

Yassine

, Croteau

, Rawat

, Hibbeln

, Rapoport

, Cunnane

, Umhau

(2017) DHA brain uptake and APOE4 status: A PET study with [1-11 C]-DHA. Alzheimers Res Ther 9, 23.

32.

Song

, Schenk

, Diep

, Murphy

, Harris

, Eiriksdottir

, Gudnason

, Casper

, Lampe

, Neuhouser

(2016) Measurement of circulating phospholipid fatty acids: Association between relative weight percentage and absolute concentrations. J Am Coll Nutr 35, 647–656.

33.

Leslie

(2004) Regulation of the specific release of arachidonic acid by cytosolic phospholipase A2. Prostaglandins Leukot Essent Fatty Acids 70, 373–376.

34.

Stephenson

, Lemere

, Selkoe

, Clemens

(1996) Cytosolic phospholipase A2(cPLA2) immunoreactivity is elevated in Alzheimer’s disease brain. Neurobiol Dis 3, 51–63.

35.

Nelson

, Raskin

(2019) The eicosapentaenoic acid: Arachidonic acid ratio and its clinical utility in cardiovascular disease. Postgrad Med 131, 268–277.

36.

Wada

, DeLong

, Hong

, Rieke

, Song

, Sidhu

, Yuan

, Warnock

, Schmaier

, Yokoyama

, Smyth

, Wilson

, FitzGerald

, Garavito

, Sui

, Regan

, Smith

(2007) Enzymes and receptors of prostaglandin pathways with arachidonic acid-derived versus eicosapentaenoic acid-derived substrates and products. J Biol Chem 282, 22254–22266.

37.

Sanchez-Mejia

, Newman

, Toh

, Yu

, Zhou

, Halabisky

, Cisse

, Scearce-Levie

, Cheng

, Gan

, Palop

, Bonventre

, Mucke

(2008) Phospholipase A2 reduction ameliorates cognitive deficits in a mouse model of Alzheimer’s disease. Nat Neurosci 11, 1311–1318.

38.

Vieira

, Fernandes

, Burgeiro

, Thomas

, Huganir

, Duarte

, Carvalho

, Santos

(2010) Excitotoxicity through Ca2+-permeable AMPA receptors requires Ca2+-dependent JNK activation. Neurobiol Dis 40, 645–655.

39.

Lovell

, Xie

, Markesbery

(2001) Acrolein is increased in Alzheimer’s disease brain and is toxic to primary hippocampal cultures. Neurobiol Aging 22, 187–194.

40.

Sierra

, Lara-Villoslada

, Comalada

, Olivares

, Xaus

(2008) Dietary eicosapentaenoic acid and docosahexaenoic acid equally incorporate as decosahexaenoic acid but differ in inflammatory effects. Nutrition 24, 245–254.

41.

Bhatt

, Steg

, Miller

, Brinton

, Jacobson

, Ketchum

, Doyle

, Juliano

, Jiao

, Granowitz

, Tardif

J-C

, Ballantyne

(2018) Cardiovascular risk reduction with icosapent ethyl for hypertriglyceridemia. N Engl J Med 380, 11–22.

42.

Neu

, Pa

, Kukull

, Beekly

, Kuzma

, Gangadharan

, Wang

, Romero

, Arneric

, Redolfi

, Orlandi

, Frisoni

, Au

, Devine

, Auerbach

, Espinosa

, Boada

, Ruiz

, Johnson

, Koscik

, Wang

, Hsu

, Chen

, Toga

(2017) Apolipoprotein E genotype and sex risk factors for alzheimer disease: A meta-analysis. JAMA Neurol 74, 1178–1189.

43.

Martinsen

, Tejera

, Vauzour

, Harden

, Dick

, Shinde

, Barden

, Mori

, Minihane

(2019) Altered SPMs and age-associated decrease in brain DHA in APOE4 female mice. FASEB J 33, 10315–10326.

44.

Schuff

, Woerner

, Boreta

, Kornfield

, Shaw

, Trojanowski

, Thompson

, Jack

Jr , Weiner

(2009) MRI of hippocampal volume loss in early Alzheimer’s disease in relation to ApoE genotype and biomarkers. Brain 132, 1067–1077.

45.

Shi

, Lepore

, Gutman

, Thompson

, Baxter

, Caselli

, Wang

, Alzheimer’s Disease Neuroimaging Initiative (2014) Genetic influence of apolipoprotein E4 genotype on hippocampal morphometry: An N=725 surface-based Alzheimer’s disease neuroimaging initiative study. Hum Brain Mapp 35, 3903–3918.

46.

Chiang

, Insel

, Tosun

, Schuff

, Truran-Sacrey

, Raptentsetsang

, Jack

Jr , Aisen

, Petersen

, Weiner

(2010) Hippocampal atrophy rates and CSF biomarkers in elderly APOE2 normal subjects. Neurology 75, 1976–1981.

47.

Soldan

, Pettigrew

, Lu

, Wang

M-C

, Selnes

, Albert

, Brown

, Ratnanather

, Younes

, Miller

, BIOCARD Research Team (2015) Relationship of medial temporal lobe atrophy, APOE genotype, and cognitive reserve in preclinical Alzheimer’s disease. Hum Brain Mapp 36, 2826–2841.

48.

, Shi

, Gutman

, Baxter

, Thompson

, Caselli

, Wang

, Alzheimer’s Disease Neuroimaging Initiative (2016) Influence of APOE genotype on hippocampal atrophy over time - an N=1925 surface-based ADNI study. PLoS One 11, e0152901.