Associations of ApoE4 status and DHA supplementation on plasma and CSF lipid profiles and entorhinal cortex thickness

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Request accessible transcript

Transcript (if available)

Content

Associations of ApoE4 status and DHA supplementation on plasma and
CSF lipid profiles and entorhinal cortex thickness

by

Haotian Xian

A Thesis Presented to the
FACULTY OF THE USC KECK SCHOOL OF MEDICINE
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
MASTER OF SCIENCE
(BIOSTATISTICS)

August 2022

ii

Acknowledgments

I would like to extend my sincere appreciation to Dr. Yassine for his guidance and input
at every stage of this project. I would like to thank Dr. Siegmund and Dr. Mack for serving on
my thesis committee and providing invaluable feedback and support. I would also like to thank
Mikaila Ann Bantugan, Xulei He and Mitchell Lee for their insights and feedback.
iii

Table of Contents

Acknowledgments................................................................................................................... ii
List of Tables ......................................................................................................................... iv
List of Figures ......................................................................................................................... v
Abstract .................................................................................................................................. vi
Introduction ............................................................................................................................. 1
Chapter 1: Methods ................................................................................................................. 3
Chapter 2: Results ................................................................................................................... 7
Chapter 3: Discussion ........................................................................................................... 11
References ............................................................................................................................. 15
Figures and Tables ................................................................................................................ 18
Supplementary Figures ......................................................................................................... 26

iv

List of Tables
Table 1. Demographic and baseline clinical characteristics by treatment arm ............................. 18
Table 2. A list of DHA and ARA-containing lipids based on data collected using tandem mass
spectrometry .................................................................................................................................. 21
Table 3. DHA/ARA Grouped Lipids Analysis ............................................................................. 22

v

List of Figures

Figure 1. Boxplots for lipid level changes and heatmap for ranking of differences in CSF lipid
changes .......................................................................................................................................... 19
Figure 2. Ranking of differences in Plasma lipid changes............................................................ 20
Figure 3. Boxplots for TG subspecies........................................................................................... 21
Figure 4. Correlation network of CSF and plasma lipids ............................................................. 23
Figure 5. Correlation network of LPC, PC and PCO lipids .......................................................... 24
Figure 6. Boxplot and correlation plots of change in entorhinal cortex ....................................... 25
Supplementary Figure 1. Ranking of beta coefficients of the CSF lipid changes ....................... 26
Supplementary Figure 2. Boxplots for change of TG lipids in E4-carriers vs non-carriers and
ranking of differences in lipid changes in CSF ............................................................................. 26
Supplementary Figure 3. Correlation heatmaps............................................................................ 28

vi

Abstract
APOE4 expression is associated with a lower increase in cerebrospinal fluid (CSF)
docosahexaenoic acid (DHA) levels following DHA supplementation. The entorhinal cortex
(EC) area in the brain is affected early in Alzheimer's disease (AD) and is enriched by DHA
supplementation. We hypothesize that phospholipid DHA levels correlate with the change in EC
thickness. The purpose of this study is to: (1) analyze 6-month trial changes in lipids measured in
CSF and plasma and test for possible correlations; (2) determine how APOE4 affects the
distribution of polyunsaturated fatty acids (PUFAs) in different lipid pools in CSF and plasma
following DHA supplementation and; (3) identify the lipid species that best correlate with
changes in EC thickness. Plasma and CSF samples were obtained from the DHA Brain Delivery
Pilot, a randomized pilot clinical trial of DHA supplementation treatment (n=13) vs. placebo
(n=13) for six months in non-demented older participants (clinical dementia rating (CDR) ≤ 0.5,
age ≥ 55 years old) and stratified by APOE4. Participants were measured for their CSF and
plasma lipid levels before supplementation or placebo treatment and at 6-month follow-ups.
Following supplementation, levels of phosphatidylcholine DHA (PC 38:6) and cholesterol ester
DHA (CE 22:6) had the largest increase in CSF (treatment effect p<0.001). Among DHA
subspecies measured, 6-month changes in plasma PC DHA had the strongest association with
increase in EC thickness in millimeters (beta=0.103, SE=0.038, p=0.011), independent of
APOE4 status (adjusting for APOE4, beta=0.109, SE=0.037, p=0.007). Our findings demonstrate
the existence of polyunsaturated fatty acids (PUFAs) within a triglyceride (TG) pool in CSF that
correlate with the corresponding plasma TG pool, suggesting an exchange of TG PUFA particles
at the CSF-blood barrier. The effect of APOE4 on the metabolism of PUFAs is strongest on TG-
enriched PUFAs in the brain. However, changes in PC-DHA pools had the strongest correlations
with changes in EC thickness.
1

Introduction
Carrying the APOE4 allele is the strongest genetic risk factor for late-onset Alzheimer’s
disease (AD). Apolipoprotein E protein plays a role in neuroplasticity and aging, and in exchange
of lipids within brain cells. Changes to how APOE4 exchanges lipids is one of the possible
mechanisms for the APOE4-associated increase in AD risk. Among its effects on lipids, APOE4
affects the metabolism of omega3 (n-3) and omega-6 (n-6) polyunsaturated fatty acids (PUFAs)
[1-4]. Through n-3 and n-6 fatty acid competition at the sn-2 position within phospholipids for
calcium-dependent phospholipase A2 (cPLA2) activity [5], an increase in the n-3 docosahexaenoic
acid (DHA) is associated with a decrease in the n-6 arachidonic acid (ARA) containing lipids.
APOE4 is associated with a lower DHA:ARA ratio in both plasma[1] and cerebrospinal fluid
CSF[4], possibly resulting from the greater activation of cPLA2 with APOE4[6].
In blood, PUFAs can be esterified to phospholipids such as phosphatidylcholines (PC),
triglycerides (TG), and cholesterol esters (CE), or transported in free unesterified forms. Among
lipoprotein particles, high-density lipoprotein (HDL) exhibits a higher proportion of PUFAs within
its PC species compared with low-density lipoprotein (LDL) particles[7]. TG PUFAs seem to have
a protective role in AD. For example, a lipidomic analysis of plasma samples from ADNI cohorts
identified a significantly negative correlation between plasma TG PUFAs levels and AD
biomarkers such as β-amyloid42 levels in CSF and entorhinal cortical thickness[8].
Abnormal or deficient entorhinal cortex (EC) neural activity is implicated early in late-
onset AD pathogenesis. DHA supplementation appears to restore the electrophysiology of EC
neurons that are manifested in cognitive behaviors such as object recognition[9]. Physiologically,
the EC feeds into the hippocampus (HC). When the EC functions abnormally, the HC is negatively
2

impacted physiologically and functionally, thus impairing memory performance[10]. We
hypothesize that EC thickness can be used to assess the neural efficacy of DHA supplementation.
The preferred lipid form that facilitates the brain delivery of DHA is not clear. In animal
models, Lysophosphatidylcholine (LPC) DHA can cross the blood-brain barrier (BBB) more
efficiently due to an interaction with BBB transporter mfsd2a[11] suggesting that LPC-DHA may
be the preferred route to enrich the brain during brain development[12]. However, the preferred
lipid pool in the aging human brain is not clear. We conducted the Brain DHA pilot delivery trial
to assess the effect of APOE genotype on brain DHA delivery in persons without but at risk of
dementia [13]. This study aimed: (1) to analyze 6-month trial changes in lipids measured in CSF
and plasma and test for possible correlations; (2) to determine how APOE4 affects the distribution
of PUFAs in different lipid pools in CSF and plasma following DHA supplementation and; (3) to
identify the lipid species that best correlate with changes in EC thickness.

3

Chapter 1: Methods
Participants:
Between 2016 and 2018, participants were recruited from the Los Angeles area. Inclusion
criteria consisted of cognitively unimpaired men and women aged 55 and older who speak English
or Spanish with a first-degree family history of dementia. Exclusion criteria were: current smokers
(or those with a recent history of smoking within the last 5 years), history of cardiovascular disease
defined by a prior heart attack, coronary artery revascularization, renal failure or blindness, a
cancer diagnosis in the past 6 months, uncontrolled hyper- or hypothyroidism, taking anti-
coagulants such as warfarin, intake of n-3 PUFA capsules within the last 3 months, moderate
physical activity (>2.5 hours of aerobic exercise per week), and heavy drinking (>30 units of
alcohol per week). All participants provided signed informed consent and completed the National
Alzheimer Coordinating Center Uniform Data Set (NACC UDS 3.0) neuropsychology test battery.
Those who met the criteria for mild cognitive impairment [14] or had a clinical diagnosis of
dementia were excluded. The study was approved by the USC IRB (HS-14-00864) and the trial
was registered with clinical trials.gov (NCT02541929).

Intervention:
Participants were randomized to 2,152 grams per day of DHA or placebo and treated for 6
months in a single-centered double-blind trial. Participants were required to consume four 1000
mg soft-gel capsules each day containing either 538 mg DHA (treatment arm) or identically
appearing capsules containing corn/soy oil (placebo arm) manufactured and provided by DSM,
Columbia, MD. The capsules had an eicosapentaenoic acid (EPA) level of <0.1% percent of the
overall fatty acid makeup. Participants received written instructions to minimize their overall
PUFA intake and were provided with and instructed to take two vitamin B complex supplements
4

per day, each comprising 500 mcg of vitamin B12, 400 mcg of folic acid, and 50 mg of vitamin
B6 (Homocysteine Modulators, Solgar, NY). CSF and plasma samples were collected in
polypropylene tubes after overnight fasting. Compliance was determined by pill counting.

Lipidomics:
Laboratory technicians that were blinded to participants’ group membership and other
group characteristics performed sample preparation, lipid extraction, and qualitative and
quantitative mass spectrometry analyses on the plasma and CSF samples from this cohort. Our
laboratory previously reported the lipid extraction technique and nano-LC/MS conditions [4].
Plasma (10 μL) was spiked with 5 μl of a mix of SPLASH LipidoMIX stable isotope and Cer/Sph
Mixture I (Avanti, Polar Lipids, Inc.) diluted 1:10 in SPLASH internal standard (IS) mix. All
solvents were HPLC grade purchased from ThermoFisher Scientific. Samples were extracted using
a modified Folch method. Methanol (80 μL) was added to the samples, then vortexed for 1 min
before adding 120 μL of chloroform and vortexing again for 1 min. Samples were then centrifuged
at 4°C at 20,000 relative centrifugal force (RCF) for 10 min. 40 μL of 0.88% potassium chloride
was added to the supernatant was in a low retention Eppendorf microtube and vortexed for 1 min.
Samples were centrifuged as before, and the lower phase was transferred to another low retention
Eppendorf microtube and evaporated by vacuum centrifugation. For the cleanup procedure, non-
sterile micro-centrifugal filters (Thermo Scientific) were prepared by applying 200 μL of 1:1
chloroform:methanol to the filters and centrifuging at 4°C, 10,000 RCF for 5 min. The flow-
through was discarded, and 1:1 chloroform:methanol was added to the samples which were
vortexed and then applied to the filters and centrifuged as before. The filters were then discarded
and the flow-through transferred to auto-sampler vials with inserts and dried under vacuum and
5

re-suspended in 50 μL of 70:30 mobile phase A–B, with mobile phase A consisting of 27%
isopropanol, 42% water, 31% acetonitrile, 10 mM ammonium formate with the addition of 0.1%
formic Acid and 90% isopropanol, 10% acetonitrile, 10 mM ammonium formate, and 0.1% formic
acid made up Mobile phase B.
An Easy-nanoLC 1000 instrument, a nanoflex ESI source, and a Thermo QE/Orbitrap mass
spectrometer were utilized. Samples were injected into an Acclaim PepMap 100, 75 μm × 2 cm
nanoViper C18, 3 μm, 100Å trapping column and Acclaim PepMap RSLC, 75 μm × 15 cm
nanoViper C18, 2 μm, 100Å analytical column for lipid chromatographic separation, running the
following gradient at a constant flow rate of 250 nL/min. The starting conditions were 30% B, then
from 1 to 50 min program from 50 to 98% B, then switch to 30% B from 50 to 65 min. All samples
were run in triplicate in batches of 8 along with a blank and quality control (QC) sample. Full-scan
MS data were acquired in both positive and negative ion modes, with a mass range of m/z 130–
2,000 in the positive ion mode, and m/z 220–2,000 in the negative ion mode, at a resolution of
30,000 for both. The heated capillary was maintained at 200°C, with a spray voltage of 1,500 V.
A maximum inject time of 200 ms was used with 13 microscans/acquired scans. Peak areas were
integrated using the Tracefinder
TM
software using a target compound list of lipids of interest
containing m/z and retention time for each target lipid and internal standard (IS) for that specific
lipid class. For each lipid class, the concentration of lipid species was calculated using the spiked
IS corresponding to that class, by dividing the target compound area by the IS area and multiplying
by the known IS concentration spiked in. Each species in a sample run that had a coefficient of
variation (CV) > 25% were excluded from further analysis as considered not to have been
measured reliably. Each analytical batch was normalized using its quality control (QC) samples
([sample] × [batch QC/normalizing QC]).
6

Statistical analysis:
Baseline participant characteristics were compared between randomized groups using
Fisher’s exact test and Wilcoxon rank-sum test. Changes in fatty acids (6-month follow-up minus
baseline) as a function of treatment and APOE4 genotype were modeled using an ANCOVA model
within a general linear model framework. The model covariates were baseline lipid measurements
and either APOE4 status or treatment status. And to test for an interaction effect, an interaction
term of treatment status times APOE4 status was added in the model. Model residuals were
evaluated for normality and homoscedasticity. The network graphs were generated using a
Pearson’s correlation matrix constructed from the change of different types of lipids, including
cholesterol esters (CE), ceramides (Cer), diacylglycerol (DG), Phosphatidylcholines (PC),
Sphingomyelin (SM) and Triglycerides (TG). The different subspecies of lipids (e.g., PC(30:0),
PC(30:1), et cetera) were then filtered for those with statistically significant changes post-treatment
using the ANCOVA models described above. There were in total 252 such subspecies of lipids in
CSF and plasma before filtering. And there were 64 subspecies of lipids after filtering. Next, a
Pearson’s correlation matrix was calculated for the correlations between those lipids. The matrix
was then used to generate a weighted correlation network with a force-embedded layout, which
uses an iterative algorithm that makes the nodes in an initial circle layout either repulse or attract
each other at each iteration based on their weights. The weights in the network were the Pearson’s
correlation coefficients between the lipids. In the analyses of the correlations between EC
thickness and lipid pools, linear models were run using EC thickness changes post-treatment as
the outcome variable and APOE4, DHA treatment, and lipid changes as the explanatory variables,
with the addition of sex and BMI as covariates in some models. If a variable needed to be
standardized, the variable was divided by its standard deviation. P-values below 0.05 (two-sided)
7

were considered statistically significant. Multiple testing correction was not performed as the
results were mostly exploratory given the relatively small sample size (n=22). The analysis was
done on a modified intention-to-treat basis (limited to trial completers with available follow-up
outcome measures) and all subjects were included regardless of adherence. All statistical analyses
used R (http://www.R-project.org/).
Chapter 2: Results
Study Sample Description:
A total of 26 participants were randomized into either treatment or placebo group, with 13
in each group. However, 4 of those individuals (3 from placebo group, 1 from treatment group)
did not have available follow-up plasma data even though they had follow-up CSF data and
completed the trial, so they were excluded from the analyses. Information regarding APOE4 carrier
status, gender, race, Clinical Dementia Rating (CDR), age, body mass index (BMI), and years of
education from 22 randomized participants who completed the trial and had complete follow-up
data were used in the analyses. A summary of baseline participant characteristics is presented in
Table 1. No participants showed signs of clinical dementia. In the total sample of 22 participants,
the median (IQR) was 69.07 (7.19); participants were primarily female (86%) and highly educated
(median (IQR) years of education was 16.5 (2.75)). By trial design, 54% of participants were
APOE4 carriers. A statistically significant difference in median BMI was noted between the
treatment group and the placebo group (p = 0.007), with higher average BMI in the placebo group.

8

Difference of Lipid Subspecies 6-month Trial Changes in CSF Between DHA Treatment and
Placebo Groups:
Critical insight into lipid exchange mechanisms can be provided by the identification of
enriched lipid subspecies following DHA supplementation. There were 20 statistically significant
(compared with placebo, p<0.05) lipid subspecies changes that occurred in triglycerides (TG),
phosphatidylcholines (PC), and cholesterol esters (CE) among other types of lipids in CSF
following DHA supplementation. These results were based on running the linear model change =
baseline + treatment on each subspecies and using the p-values of the treatment variable beta
coefficients to select statistically significant results. Some of these lipids are plotted in boxplots to
compare their concentration changes (Fig 1a) in the DHA group versus placebo group. Notably,
CE(22:5), which is arachidonic acid (ARA)-containing, had overall negative changes in the DHA
group compared to the placebo group. The lipids were also ranked from highest to lowest based
on the differences between mean percentage changes in the DHA group and placebo group (Fig.
1b). As expected, DHA-containing lipids
1
such as PC(38:6), PC(40:6), PC(36:5), PC(40:7),
PC(36:6), CE(22:6), TG(56:8) and TG(56:7) showed some of the greatest positive differences
between mean percentage change in DHA and placebo, whereas arachidonic acid (ARA)
containing lipids
2
such as CE(20:4), CE(18:3), CE(22:5), PC(40:4) had overall negative
differences between DHA and placebo (Fig. 1b). These results reflect the competition of DHA and
ARA in these lipid pools, consistent with results from the grouped lipids analysis where treatment
effect of CE and PC lipids in CSF were positive for those lipids in the DHA-containing group
while being negative for those in the ARA-containing group (Table 3).

1
A list of DHA-containing lipids can be found in Table 2.
2
A list of ARA-containing lipids can be found in Table 2.
9

Difference of Lipid Subspecies 6-month Trial Changes in Plasma Between DHA Treatment and
Placebo Groups:
Consistent with the above results in CSF, DHA-containing lipids such as CE(22:6),
TG(56:8), TG(56:7) and PC(40:6) also had the greatest positive difference between DHA group
and placebo in plasma. Meanwhile, ARA-containing lipids such as PC(40:4), CE(22:5), and
PC(36:4) had overall negative changes between DHA group and placebo group (Fig. 2). Such
inverse relationships between DHA and ARA-containing lipids were also confirmed by the
grouped lipids analysis results, where increases in concentrations in the same DHA-containing
CEs and PCs were complemented by decreases in concentrations in ARA-containing CEs and PCs
in plasma (Table 3). The results from this section and the previous section were also consistent
with the correlations observed between the 6-month trial changes of the same DHA-containing
lipids in CSF and plasma shown by heatmaps (Supplementary Fig. 3). As DHA-containing lipids
increased in CSF, they also increased in plasma, and as ARA-containing lipids decreased in CSF,
they also decreased in plasma.

Interaction Effect of Treatment and APOE4 in DHA-Containing TGs in Plasma:
In plasma, APOE4 carrier status had statistically significant interaction effects with
treatment status on 6-month trial changes of some DHA-containing TGs. Regression models with
interaction terms between APOE4 and treatment showed that TG(54:7), TG(56:8), and TG(56:7)
levels increased after DHA supplementation, but E4 carriers had less increase compared to non-
E4 carriers (Fig. 3). Similar results were found in grouped lipids analysis (Table 3) where DHA-
containing TGs in plasma had lower positive changes in E-4 carriers compared to non-E4 carriers
following DHA treatment (interaction beta = -26.67, p = 0.029). These results suggest that the
10

protective role of DHA in increasing TG PUFAs in plasma against AD is negatively modified by
APOE4.

Network Analysis of Lipid Subspecies and DHA and ARA in CSF and Plasma:
Network analysis of lipid changes (6-month follow-up measurements minus baseline
measurements) in plasma and CSF showed clustering of DHA-containing lipids and ARA-
containing lipids due to their correlations with each other. The changes observed in CE, PC, and
TG DHA-containing lipids such as TG(56:7), PC(38:6) and CE(22:6) were strongly and positively
correlated within CSF and plasma (Fig. 4). Negative correlations were observed in changes in
ARA-containing lipids such as PC(36:4), PC(40:4) and PC(38:4). Negative correlations between
the DHA and ARA lipid pools showed that as DHA levels increased, ARA levels decreased. These
results were consistent with the previously mentioned inverse relationship between DHA and ARA
that competed at the sn-2 position within the phospholipid [15]. In an additional network analysis
of PC and LPC-DHA lipids, our data also showed that LPC-DHA positively correlated with DHA-
containing PC lipids in CSF and plasma (Fig. 5) such as PC(40:6) and PC(38:6). Similar to the
previous network, clustering of DHA and ARA-containing lipids were also observed in this
network. Positive correlations were observed within DHA and ARA clusters while negative
correlations were observed between DHA and ARA clusters.

Entorhinal Cortex Thickness Changes Following DHA Supplementation and Effect of APOE4:
The change of PC and CE DHA in plasma was positively correlated with the change in
entorhinal cortex thickness following DHA supplementation. Notably, these changes were not
influenced by APOE4 carrier status. As PC and CE levels in plasma and CSF increased, the change
11

in entorhinal cortex thickness increased in both non-carriers and carriers of E4 (Fig. 6b-d). After
6 months of DHA treatment, entorhinal cortex thickness was preserved and did not deteriorate
(Fig. 6a). Our data showed that DHA treatment, adjusted for APOE4, was associated with a
statistically significant increase in the entorhinal cortex area thickness compared with placebo (p
= 0.046). The same statistically significant results were observed for the treatment effect when
additional covariates, BMI and sex, were added. In the DHA treatment group, although a trend of
lower increase in entorhinal cortex thickness was observed in APOE4 carriers compared with non-
carriers, the difference was not statistically significant. Greater levels of DHA in plasma were
significantly associated with thicker ERC volume.
Chapter 3: Discussion
This study investigated the lipidomic profile in both the CSF and plasma based on
treatment with DHA and the presence of the APOE4 genotype. We showed that DHA
supplementation increased the levels of DHA-containing lipids in both CSF and plasma. We
additionally demonstrated significant differences in the change of CSF TG PUFAs based on the
presence of the APOE4 genotype. A significant correlation between the changes of PUFA TGs
in CSF and plasma was observed, indicating the importance of TG pools for increasing DHA-
containing lipids in the brain. We hypothesized that an increased oxidation of triglycerides in the
brain within the TG pool makes APOE4 carriers more susceptible to omega-3 loss.
Notably, DHA-containing PCs increased in CSF because of DHA treatment. A previous
study demonstrated an increase in CSF DHA after dietary supplementation over a period of 6
months[2]. The findings from this study demonstrate the same increase of DHA in CSF as
observed in our data, and additionally provide insight into the specific DHA-containing lipids
that increase because of supplementation. In plasma, the largest mean changes in lipids after
12

treatment included several TGs. In a study that investigated DHA uptake in the liver, heart, and
brain using radioactive tracers, it was found that the liver has a selectivity for DHA and converts
these lipids into triglycerides[16]. Our results are confirmed by this study as all the TGs that had
the largest mean changes are DHA-containing lipids.
Evidence supporting the effect of the APOE4 allele on lipid metabolism includes an
observed differential clearance in lipid metabolism in the E2, E3, and E4 alleles of the APOE
gene. Following supplementation of a vitamin A-containing fatty diet and a vitamin A-fat
loading test, E4 carriers had a faster clearance of lipids than E3 and E2 carriers[17].
Our results show significant decreases in the positive change of CSF TGs based on
APOE4 carrier status. This effect was observed in plasma as well. Even though DHA-containing
TGs such as TG (54:7), (56:8), and (56:7) levels increased in plasma after DHA
supplementation, E4 carriers had less increase compared to non-E4 carriers. A study
investigating astrocyte uptake of TGs between APOE4 carriers and non-carriers demonstrated
differential metabolism of fatty acids based on genotype[18]. Also, a study investigating serum
triglycerides carriers found significant associations between PUFA TG component scores and
entorhinal cortical thickness and CSF Aβ1-42 levels in APOE4 carriers that were not found in
non-carriers[19]. Another study found an association between decreased circulating TGs and
increased cortical thinning in middle-aged adults, suggesting neuroprotective effects of increased
TGs in the aging brain[20]. Together, these findings suggest that APOE4 causes increased
clearance of CSF TGs (lower amounts of certain lipid subspecies) and hypolipidation
independent of the treatment effect, which may be related to early biomarkers of AD such as
CSF β-amyloid42 levels and entorhinal cortical thickness[8]. This means that the E4 allele may
be responsible for the decrease in circulating TGs and ultimately for cortical thinning in the
13

aging brain. Our findings offer additional insight into these studies’ findings, suggesting the
selective oxidation and metabolism of TGs is a result of the E4 allele.
The study presents evidence indicating that the accumulation and/or metabolism of lipids
is differentiated by the APOE genotype, though future studies may place more of a focus on the
mechanisms that are responsible for these differences. A potential future study may use
radioactive tracers in DHA supplementation to actively track specific lipids between the CSF and
plasma. Additionally, understanding how these differences ultimately affect cognitive
development/decline is important in understanding how lipidomics contribute to the onset of
Alzheimer’s disease. Investigating which enzymes are responsible for these differences and how
they can be potentially modulated in clinical applications would be important for mitigating any
potentially harmful effects of carrying the APOE4 genotype.
Furthermore, investigating the entorhinal cortex provides insight into the functional
consequences of this study. Entorhinal thickness appears to be responsive to DHA
supplementation in these data, though validation is needed. Changes in plasma PC and CE DHA
have similar and stronger associations with entorhinal cortex change than LPC-DHA, this data
argues against the LPC-DHA as a unique brain DHA supply pool. PCs and CEs are found on
HDL particles. This implicates a role for HDL particles carrying DHA within PCs and CE as a
brain source. This prompts the question of whether there is a brain-plasma HDL exchange.
However, APOE4 does not modify the association of DHA supplementation with entorhinal
cortex change in persons before the onset of dementia. Thus, questioning if the effect of APOE4
is preventable. This study targeted people prior to the onset of dementia. If these participants had
waited longer after the onset of dementia, E4 may be unresponsive to DHA treatment.
14

Lastly, previous studies have shown the impact of increased DHA supplementation on
lipids in the plasma and CSF. APOE4 also had the largest effect on TGs, not HDLs. On their
own, HDLs, whether PC or CE, are correlated with entorhinal cortex thickness. Furthermore, this
study proposes a model where HDLs are potential markers of changes in brain DHA levels.
Thus, HDLs may be important to study in neurodegenerative diseases such as Alzheimer’s
Disease because of their observed effects on brain DHA. We suggest that the HDL pool is worth
further investigation.

15

References
1. Tomaszewski N, He X, Solomon V, Lee M, Mack WJ, Quinn JF, et al. 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. J Alzheimers Dis. 2020;74(3):975-90.

2. Arellanes IC, Choe N, Solomon V, He X, Kavin B, Martinez AE, et al. Brain delivery of
supplemental docosahexaenoic acid (DHA): A randomized placebo-controlled clinical trial.
EBioMedicine. 2020;59:102883.

3. Yassine HN, Croteau E, Rawat V, Hibbeln JR, Rapoport SI, Cunnane SC, et al. DHA
brain uptake and APOE4 status: a PET study with [1-11 C]-DHA. Alzheimer's Research &
Therapy. 2017;9(1):23.

4. Yassine HN RV, Mack WJ, Quinn JF, Yurko-Mauro K, Bailey-Hall E, Aisen PS, Chui
HC, Schneider LS. The effect of APOE genotype on the delivery of DHA to cerebrospinal fluid
in Alzheimer's disease. Alzheimer’s Reaserch and Therapy. 2016;8:25.

5. Bibus DLB. Balancing proportions of competing omega-3 and omega-6 highly
unsaturated fatty acids (HUFA) in tissue lipids. Prostaglandins, Leukotrienes & Essential Fatty
Acids. 2015;99:19-23.

6. Wang S, Li B, Solomon V, Fonteh A, Rapoport SI, Bennett DA, et al. Calcium-
dependent cytosolic phospholipase A2 activation is implicated in neuroinflammation and
oxidative stress associated with ApoE4. Mol Neurodegener. 2021;16(1):26.

7. Wiesner P, Leidl K, Boettcher A, Schmitz G, Liebisch G. Lipid profiling of FPLC-
separated lipoprotein fractions by electrospray ionization tandem mass
spectrometry. Journal of Lipid Research. 2009;50(3):574-85.

8. Bernath MM, Bhattacharyya S, Nho K, Barupal DK, Fiehn O, Baillie R, et al. Serum
triglycerides in Alzheimer disease: Relation to neuroimaging and CSF biomarkers. Neurology.
2020;94(20):e2088-e98.

9. Arsenault D, Julien C, Tremblay C, Calon F. DHA improves cognition and prevents
dysfunction of entorhinal cortex neurons in 3xTg-AD mice. PloS one. 2011;6(2):e17397-e.

16

10. Frank LM, Brown EN, Wilson M. Trajectory Encoding in the Hippocampus and
Entorhinal Cortex. Neuron. 2000;27(1):169-78.

11. Nguyen LN, Ma D, Shui G, Wong P, Cazenave-Gassiot A, Zhang X, et al. Mfsd2a is a
transporter for the essential omega-3 fatty acid docosahexaenoic acid. Nature (London).
2014;509(7501):503-6.

12. Sugasini D, Thomas R, Yalagala PCR, Tai LM, Subbaiah PV. Dietary docosahexaenoic
acid (DHA) as lysophosphatidylcholine, but not as free acid, enriches brain DHA and improves
memory in adult mice. Scientific reports. 2017;7(1):11263-11.

13. Arellanes IC, Choe N, Solomon V, He X, Kavin B, Martinez AE, et al. Brain delivery of
supplemental docosahexaenoic acid (DHA): A randomized placebo-controlled clinical trial.
EBioMedicine. 2020;59.

14. Albert MS, DeKosky ST, Dickson D, Dubois B, Feldman HH, Fox NC, et al. The
diagnosis of mild cognitive impairment due to Alzheimer's disease: recommendations from the
National Institute on Aging-Alzheimer's Association workgroups on diagnostic guidelines for
Alzheimer's disease. Alzheimer's & dementia : the journal of the Alzheimer's Association.
2011;7(3):270-9.

15. Bibus D, Lands B. Balancing proportions of competing omega-3 and omega-6 highly
unsaturated fatty acids (HUFA) in tissue lipids. Prostaglandins, leukotrienes and essential fatty
acids. 2015;99:19-23.

16. Polozova A, Salem N, Jr. Role of liver and plasma lipoproteins in selective transport of n-
3 fatty acids to tissues: a comparative study of 14C-DHA and 3H-oleic acid tracers. J Mol
Neurosci. 2007;33(1):56-66.

17. Weintraub MS, Eisenberg S, Breslow JL. Dietary fat clearance in normal subjects is
regulated by genetic variation in apolipoprotein E. The Journal of clinical investigation.
1987;80(6):1571-7.

18. Farmer BC, Kluemper J, Johnson LA. Apolipoprotein E4 Alters Astrocyte Fatty Acid
Metabolism and Lipid Droplet Formation. Cells. 2019;8(2):182.

17

19. Bernath MM, Bhattacharyya S, Nho K, Barupal DK, Fiehn O, Baillie R, et al. Serum
triglycerides in Alzheimer disease. Relation to neuroimaging and CSF biomarkers.
2020;94(20):e2088-e98.

20. Sliz E, Shin J, Syme C, Black S, Seshadri S, Paus T, et al. Thickness of the cerebral
cortex shows positive association with blood levels of triacylglycerols carrying 18-carbon
fatty acids. Communications Biology. 2020;3(1):456.

18

Figures and Tables
Variable N Placebo, N = 10 DHA, N = 12 p-value
1

Age (years), Median (IQR) 22 69.54 (63.36, 70.87) 68.10 (65.97, 71.16) 0.92
Gender, n / N (%) 22 0.57
Female 8 / 10 (80%) 11 / 12 (92%)
Male 2 / 10 (20%) 1 / 12 (8.3%)
Race, n / N (%) 22 0.52
Other 1 / 10 (10%) 0 / 12 (0%)
Asian 1 / 10 (10%) 0 / 12 (0%)
White 4 / 10 (40%) 7 / 12 (58%)
Black 0 / 10 (0%) 1 / 12 (8.3%)
Hispanic 4 / 10 (40%) 4 / 12 (33%)
Education (years), Median (IQR) 22 17.00 (14.25, 18.00) 16.50 (16.00, 17.25) 0.89
BMI (kg/m
2
), Median (IQR) 22 32.77 (31.00, 35.06) 26.52 (24.54, 29.26) 0.007
CDR, n / N (%) 22 >0.99
0 10 / 10 (100%) 11 / 12 (92%)
0.5 0 / 10 (0%) 1 / 12 (8.3%)
APOE4, n / N (%) 22 >0.99
Non-carriers 5 / 10 (50%) 5 / 12 (42%)
Carriers 5 / 10 (50%) 7 / 12 (58%)
*4 participants from the 26 randomized participants were excluded due to missing
follow-up plasma lipid data
1
Wilcoxon rank sum exact test; Fisher's exact test; Wilcoxon rank sum test
Table 1. Demographic and baseline clinical characteristics by treatment arm

19

Figure 1. Boxplots for lipid level changes and heatmap for ranking of differences in CSF lipid changes
(A) Boxplots comparing mean changes in concentration levels of specific PC and CE lipid subspecies in CSF from baseline to
follow-up in DHA treatment group and placebo group. CE(22:5) had an overall decrease in the treatment group since it is ARA-
containing. (B) Ranking of the mean percentage changes of CSF lipid levels in DHA treated group minus mean percentage
changes of CSF lipid levels in placebo group. Mean percentage change is calculated as mean of
𝑙𝑖𝑝𝑖𝑑 (𝑐 ℎ𝑎𝑛𝑔𝑒 )
𝑙𝑖𝑝𝑖𝑑 (𝑏𝑎𝑠𝑒𝑙𝑖𝑛𝑒 )
× 100%. Specific
lipid subspecies used in this ranking were selected based on having statistically significant changes from baseline to follow-up,
adjusting for DHA treatment. The regression model used for selection was lipid(change) = lipid(baseline) + treatment.
20

Figure 2. Ranking of differences in Plasma lipid changes
Ranking of mean percentage change of plasma lipid levels in DHA treated group minus mean percentage change of plasma lipid
levels in placebo group. Mean percentage change is calculated as mean of
𝑙𝑖𝑝𝑖𝑑 (𝑐 ℎ𝑎𝑛𝑔𝑒 )
𝑙𝑖𝑝𝑖𝑑 (𝑏𝑎𝑠𝑒𝑙𝑖𝑛𝑒 )
× 100%. Specific lipid subspecies used
in this ranking were selected based on having statistically significant changes from baseline to follow-up, adjusting for DHA
treatment. The regression model used for selection was lipid(change) = lipid(baseline) + treatment.

21

Figure 3. Boxplots for TG subspecies
Effect of APOE4 carrier status on TG subspecies levels in plasma following DHA treatment. TGs (54:7), (56:8), and (56:7) levels
increased after DHA supplementation, but E4 carriers had less increase compared to non-E4 carriers. The interaction model
used was lipid(change) = treatment + apoe4 + treatment*apoe4 + lipid(baseline). The p values for the interaction term are
indicated below each figure.

DHA-
containing
lipids
CE22:6, CE 20:5
PC 36:6, PC37:6, PC 38:6, PC 39:6, PC 40:6, PC 40:7, PC 40:8, PC 36:5
TG 52:6, TG 54:6, TG 54:7, TG 56:7, TG 56:8, TG 58:7
ARA-
containing
lipids
CE20:4, CE 18:3, CE 22:4, CE:22:5
PC 34:4, PC 35:4, PC 36:4, PC 37:4, PC 38:4, PC 38:5, PC 40:4
TG 54:6, TG 56:5
Table 2. A list of DHA and ARA-containing lipids based on data collected using tandem mass spectrometry

22

Sample Group CE PC TG
Treatment effect: Lipid (change) = treatment + apoe4 + lipid (baseline)
CSF
DHA 1.1246 (0.0001) 0.3489 (2.32e-08) 0.0161 (0.552)
ARA -2.325 (0.0143) -0.2043 (0.118) -0.0048 (0.747)
Plasma
DHA 548.008 (2.82e-05) 284.647 (5.38e-06) 51.7585 (0.018)
ARA -63.956 (0.663) -142.86 (0.0069) -8.831 (0.2312)
APOE4 effect: Lipid (change) = apoe4 + lipid (baseline)

CSF
DHA 0.1183 (0.729) -0.0425 (0.611) -0.0539 (0.0139)
ARA 0.880 (0.380) -0.1328 (0.311) -0.0283 (0.022)

Plasma
DHA 99.1712 (0.550) -25.252 (0.746) -16.994 (0.500)
ARA 252.993 (0.091) -15.229 (0.778) -0.820 (0.9159)
Interaction effect: Lipid (change) = treatment + apoe4 + treatment*apoe4 + lipid (baseline)

CSF
DHA 0.22 (0.671) -5.34e-03 (0.949) -0.02 (0.572)
ARA 0.52 (0.772) 0.13 (0.628) -8.26e-03 (0.733)

Plasma
DHA 50.20 (0.817) 4.60 (0.963) -87.71 (0.029)
ARA 0.60 (0.998) 38.72 (0.680) -26.67 (0.072)
Table 3. DHA/ARA Grouped Lipids Analysis
Grouped lipids analysis based on whether the lipid is DHA containing or ARA-containing (See Table 2).
After grouping into DHA or ARA-containing lipids, total baseline or change in concentrations in each
group are summed within each participant. These sums were then fitted into one of the three
regression models to test for possible treatment effect, APOE4 effect, or interaction effect. The numbers
in each cell represent the beta coefficient of the variable of interest (treatment, APOE4, or interaction
of treatment and APOE4) and its p-value. Significant p-values were highlighted in red.

23

Figure 4. Correlation network of CSF and plasma lipids
Correlation network of CSF and plasma lipid changes (defined as 6-month follow-up minus baseline concentration). Green lines
represent positive correlations while red lines represent negative correlations. Thickness of lines indicates strength of the
correlation. The node shapes represent species of lipids (square: CE; circle: Cer, DG, SM; diamond: PC, PCO, LPC; triangle:
TG). Red nodes indicate lipids in CSF while blue nodes indicate lipids in plasma. The lipids selected for this network were
significantly affect by DHA treatment according to having significant p-values of treatment beta coefficient in the model change
= baseline + treatment. The pairwise correlations selected for this network were statistically significant. The pairs that have
absolute values of Person’s correlation coefficients less than 0.6 are hidden from view. DHA-containing lipids are marked with
white square symbols, ARA-containing lipids are marked with black square symbols.

24

Figure 5. Correlation network of LPC, PC and PCO lipids
Correlation network of LPC, PC and PCO lipid changes (defined as 6-month follow-up minus baseline concentration) in CSF
and plasma. Green lines represent positive correlations while red lines represent negative correlations. Thickness of lines
indicates strength of the correlation. Red nodes indicate lipids in CSF while blue nodes indicate lipids in plasma. The lipids
selected for this network were significantly affect by DHA treatment according to having significant p values of treatment beta
coefficient in the model change = baseline + treatment. The pairwise correlations selected for this network were statistically
significant. DHA-containing lipids are marked with white square symbols. ARA-containing lipids are marked with black square
symbols.

25

Figure 6. Boxplot and correlation plots of change in entorhinal cortex
(A): DHA treatment, adjusted for APOE4, was associated with a significant increase in the entorhinal cortex area thickness
compared with placebo. Regression model: ERC change = treatment + APOE4 (mean group difference in ERC change (95% CI)
0.161 (0.003, 0.319), p=0.046). The same statistically significant results were observed for the treatment effect when additional
covariates were added. Regression model: ERC change = treatment + APOE4 + BMI + sex (mean group difference in ERC
change (95% CI) 0.164 (-0.05, 0.38), p=0.03). (B): Regression of SD-standardized plasma PC DHA concentration and
entorhinal cortex thickness change after adjusting for APOE4 (beta=0.109, SE=0.037 p=0.007) shows statistically significant
association between increase in PC DHA and increase in entorhinal cortex thickness. SD-standardization was done by dividing
the variable by its standard deviation. (C), (D): Change of entorhinal cortex thickness were associated with change of plasma CE
(22:6), which is a DHA-containing lipid, in both plasma and CSF after adjusting for APOE4.

26

Supplementary Figures

Supplementary Figure 1. Ranking of beta coefficients of the CSF lipid changes
Ranking of beta coefficients of the treatment variable from the linear model lipid(change)=lipid(baseline)+treatment, where
change is calculated as 6-month follow-up measurement minus baseline measurement and treatment is a binary indicator for
whether the subject was randomized into DHA treatment group or placebo group. The asterisks represent the value of the p-value
(* p <0.05, ** p <0.001, *** p <0.0001)

Supplementary Figure 2. Boxplots for change of TG lipids in E4-carriers vs non-carriers and ranking of differences in lipid
changes in CSF
(A) TG(50:5), (52:6), and (50:3) in CSF had lower 6-month trial changes in E4 carriers compared to non-E4 carriers (B)
Ranking of mean percentage changes of CSF lipid levels in APOE4 carriers minus mean percentage changes of CSF lipid levels
in non-APOE4 carriers. Mean percentage change is calculated as mean of
𝑙𝑖𝑝𝑖𝑑 (𝑐 ℎ𝑎𝑛𝑔𝑒 )
𝑙𝑖𝑝𝑖𝑑 (𝑏𝑎𝑠𝑒𝑙𝑖𝑛𝑒 )
× 100%. Specific lipid subspecies used
in this ranking were selected based on having statistically significant changes from baseline, adjusting for APOE4 status. The
regression model used for selection was change ~ baseline + APOE4 status.
27

C
28

D
Supplementary Figure 3. Correlation heatmaps
Correlation heat maps between CSF and plasma lipid pools showing Pearson’s correlation coefficient in each cell. Blue
represents positive correlation and red represents negative correlation (A) CE lipid pool (B) DG lipid pool (C) PC lipid pool (D)
TG lipid pool. The correlations were done using 6-month concentration changes in each subspecies.

Asset Metadata

Creator Xian, Haotian (author)

Core Title Associations of ApoE4 status and DHA supplementation on plasma and CSF lipid profiles and entorhinal cortex thickness

Contributor Electronically uploaded by the author (provenance)

School Keck School of Medicine

Degree Master of Science

Degree Program Biostatistics

Degree Conferral Date 2022-08

Publication Date 07/27/2024

Defense Date 06/23/2022

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag docosahexaenoic acid,entorhinal cortex,oai:digitallibrary.usc.edu:usctheses,OAI-PMH Harvest,polyunsaturated fatty acids

Format application/pdf (imt)

Language English

Advisor Mack, Wendy (committee chair), Siegmund, Kimberly (committee member), Yassine, Hussein (committee member)

Creator Email hxian@usc.edu,penn123abc@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-oUC111375436

Unique identifier UC111375436

Legacy Identifier etd-XianHaotia-11007

Document Type Thesis

Format application/pdf (imt)

Rights Xian, Haotian

Type texts

Source 20220728-usctheses-batch-962 (batch), University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright. The original signature page accompanying the original submission of the work to the USC Libraries is retained by the USC Libraries and a copy of it may be obtained by authorized requesters contacting the repository e-mail address given.

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA

Repository Email uscdl@usc.edu

Abstract (if available)

Abstract APOE4 expression is associated with a lower increase in cerebrospinal fluid (CSF) docosahexaenoic acid (DHA) levels following DHA supplementation. The entorhinal cortex (EC) area in the brain is affected early in Alzheimer's disease (AD) and is enriched by DHA supplementation. We hypothesize that phospholipid DHA levels correlate with the change in EC thickness. The purpose of this study is to: (1) analyze 6-month trial changes in lipids measured in CSF and plasma and test for possible correlations; (2) determine how APOE4 affects the distribution of polyunsaturated fatty acids (PUFAs) in different lipid pools in CSF and plasma following DHA supplementation and; (3) identify the lipid species that best correlate with changes in EC thickness. Plasma and CSF samples were obtained from the DHA Brain Delivery Pilot, a randomized pilot clinical trial of DHA supplementation treatment (n=13) vs. placebo (n=13) for six months in non-demented older participants (clinical dementia rating (CDR) ≤ 0.5, age ≥ 55 years old) and stratified by APOE4. Participants were measured for their CSF and plasma lipid levels before supplementation or placebo treatment and at 6-month follow-ups. Following supplementation, levels of phosphatidylcholine DHA (PC 38:6) and cholesterol ester DHA (CE 22:6) had the largest increase in CSF (treatment effect p<0.001). Among DHA subspecies measured, 6-month changes in plasma PC DHA had the strongest association with increase in EC thickness in millimeters (beta=0.103, SE=0.038, p=0.011), independent of APOE4 status (adjusting for APOE4, beta=0.109, SE=0.037, p=0.007). Our findings demonstrate the existence of polyunsaturated fatty acids (PUFAs) within a triglyceride (TG) pool in CSF that correlate with the corresponding plasma TG pool, suggesting an exchange of TG PUFA particles at the CSF-blood barrier. The effect of APOE4 on the metabolism of PUFAs is strongest on TG-enriched PUFAs in the brain. However, changes in PC-DHA pools had the strongest correlations with changes in EC thickness.