Brain atrophy in cognitively impaired elderly: The importance of long-chain ω-3 fatty acids and B vitamin status in a randomized controlled trial

Abstract

Increased brain atrophy rates are common in older people with cognitive impairment, particularly in those who eventually convert to Alzheimer disease. Plasma concentrations of omega-3 (ω-3) fatty acids and homocysteine are associated with the development of brain atrophy and dementia.
We investigated whether plasma ω-3 fatty acid concentrations (eicosapentaenoic acid and docosahexaenoic acid) modify the treatment effect of homocysteine-lowering B vitamins on brain atrophy rates in a placebo-controlled trial (VITACOG).
This retrospective analysis included 168 elderly people (≥70 y) with mild cognitive impairment, randomly assigned either to placebo (n = 83) or to daily high-dose B vitamin supplementation (folic acid, 0.8 mg; vitamin B-6, 20 mg; vitamin B-12, 0.5 mg) (n = 85). The subjects underwent cranial magnetic resonance imaging scans at baseline and 2 y later. The effect of the intervention was analyzed according to tertiles of baseline ω-3 fatty acid concentrations.
There was a significant interaction (P = 0.024) between B vitamin treatment and plasma combined ω-3 fatty acids (eicosapentaenoic acid and docosahexaenoic acid) on brain atrophy rates. In subjects with high baseline ω-3 fatty acids (>590 μmol/L), B vitamin treatment slowed the mean atrophy rate by 40.0% compared with placebo (P = 0.023). B vitamin treatment had no significant effect on the rate of atrophy among subjects with low baseline ω-3 fatty acids (<390 μmol/L). High baseline ω-3 fatty acids were associated with a slower rate of brain atrophy in the B vitamin group but not in the placebo group.
The beneficial effect of B vitamin treatment on brain atrophy was observed only in subjects with high plasma ω-3 fatty acids. It is also suggested that the beneficial effect of ω-3 fatty acids on brain atrophy may be confined to subjects with good B vitamin status. The results highlight the importance of identifying subgroups likely to benefit in clinical trials. This trial was registered at www.controlled-trials.com as ISRCTN94410159.
© 2015 American Society for Nutrition.

Full text

Brain atrophy in cognitively impaired elderly: the importance of
long-chain v-3 fatty acids and B vitamin status in a randomized
controlled trial
1–4
Fredrik Jernerén, Amany K Elshorbagy, Abderrahim Oulhaj, Stephen M Smith, Helga Refsum, and A David Smith
ABSTRACT
Background: Increased brain atrophy rates are common in older
people with cognitive impairment, particularly in those who even-
tually convert to Alzheimer disease. Plasma concentrations of
omega-3 (v-3) fatty acids and homocysteine are associated with
the development of brain atrophy and dementia.
Objective: We investigated whether plasma v-3 fatty acid concen-
trations (eicosapentaenoic acid and docosahexaenoic acid) modify
the treatment effect of homocysteine-lowering B vitamins on brain
atrophy rates in a placebo-controlled trial (VITACOG).
Design: This retrospective analysis included 168 elderly people
($70 y) with mild cognitive impairment, randomly assigned either
to placebo (n= 83) or to daily high-dose B vitamin supplementation
(folic acid, 0.8 mg; vitamin B-6, 20 mg; vitamin B-12, 0.5 mg) (n=
85). The subjects underwent cranial magnetic resonance imaging
scans at baseline and 2 y later. The effect of the intervention
was analyzed according to tertiles of baseline v-3 fatty acid
concentrations.
Results: There was a significant interaction (P= 0.024) between B
vitamin treatment and plasma combined v-3 fatty acids (eicosapen-
taenoic acid and docosahexaenoic acid) on brain atrophy rates. In
subjects with high baseline v-3 fatty acids (.590 mmol/L), B vi-
tamin treatment slowed the mean atrophy rate by 40.0% compared
with placebo (P= 0.023). B vitamin treatment had no significant
effect on the rate of atrophy among subjects with low baseline v-3
fatty acids (,390 mmol/L). High baseline v-3 fatty acids were
associated with a slower rate of brain atrophy in the B vitamin
group but not in the placebo group.
Conclusions: The beneficial effect of B vitamin treatment on brain
atrophy was observed only in subjects with high plasma v-3 fatty
acids. It is also suggested that the beneficial effect of v-3 fatty acids
on brain atrophy may be confined to subjects with good B vitamin
status. The results highlight the importance of identifying subgroups
likely to benefit in clinical trials. This trial was registered at www.
controlled-trials.com as ISRCTN94410159. Am J Clin Nutr doi:
10.3945/ajcn.114.103283.
Keywords: B vitamin, brain atrophy, homocysteine, mild cogni-
tive impairment, v-3
INTRODUCTION
Mild cognitive impairment (MCI)
5
is a syndrome character-
ized by a subtle decline in cognitive function and is considered
a transitory state between normal aging and clinical dementia
and Alzheimer disease (AD) (1, 2). A modest rate of brain at-
rophy is observed in normal aging. However, in subjects with
MCI, dementia, or AD, the brain atrophy rates are markedly
faster (3–5). Furthermore, in MCI, the rate of atrophy is usually
higher in the subgroup that eventually converts to AD (6). There
are no available cures for AD, but an alternative approach is
strategies to delay disease progression at an early stage. Cranial
MRI is established as a method to monitor disease progression
(3, 4, 7, 8). Effective interventions may be detected by a slowing
of brain atrophy rate.
The role of v-3 fatty acids in cognitive decline and dementia
is debated. Epidemiologic evidence is consistent with a pro-
tective role of dietary intake of fish oils rich in v-3 fatty acids
such as EPA and DHA (9, 10). Case-control studies have re-
vealed associations between DHA or EPA and brain volume
and lower degrees of white matter hyperintensities (11, 12). In
prospective studies, red blood cell DHA and EPA concentrations
were positively correlated with higher total brain and hippocampal
1
From the Oxford Project to Investigate Memory and Ageing (OPTIMA),
Department of Pharmacology, University of Oxford, Oxford, United Kingdom
(FJ, AKE, HR, and ADS); the Department of Physiology, Faculty of Medicine,
University of Alexandria, Alexandria, Egypt (AKE); the Institute of Public
Health, College of Medicine and Health Sciences, United Arab Emirates
University, Al Ain, United Arab Emirates (AO); Functional Magnetic Reso-
nance Imaging of the Brain Centre, Nuffield Department of Clinical Neuro-
sciences, University of Oxford, John Radcliffe Hospital, Oxford, United
Kingdom (SMS); and the Department of Nutrition, Institute of Basic Medical
Sciences, Faculty of Medicine, University of Oslo, Oslo, Norway (HR).
2
Supported by grants from the Charles Wolfson Charitable Trust, The
Medical Research Council, Norwegian Research Council, Norman Collisson
Foundation, Alzheimer’s Research UK, Henry Smith Charity, John Coates
Charitable Trust, Thames Valley Dementias and Neurodegenerative Diseases
Research Network of the National Institute for Health Research, the Sidney
and Elizabeth Corob Charitable Trust, and Meda AB/Recip AB.
3
Supplemental Tables 1–3 and Supplemental Figure 1 are available from
the “Supplemental data” link in the online posting of the article and from the
same link in the online table of contents at http://ajcn.nutrition.org.
4
Address correspondence to F Jernerén, University of Oxford, Depart-
ment of Pharmacology, Mansfield Road, OX1 3QT, Oxford, United
Kingdom. E-mail: fredrik.jerneren@pharm.ox.ac.uk.
5
Abbreviations used: AD, Alzheimer disease; MCI, mild cognitive impair-
ment; PEMT, phosphatidylethanolamine N-methyltransferase; SAH, S-ad-
enosylhomocysteine; tHcy, total homocysteine; VITACOG, Homocysteine
and B Vitamins in Cognitive Impairment.
Received November 18, 2014. Accepted for publication March 19, 2015.
doi: 10.3945/ajcn.114.103283.
Am J Clin Nutr doi: 10.3945/ajcn.114.103283. Printed in USA. Ó2015 American Society for Nutrition 1of7
AJCN. First published ahead of print April 15, 2015 as doi: 10.3945/ajcn.114.103283.
Copyright (C) 2015 by the American Society for Nutrition

volumes 8 y later (13), and higher relative concentrations of
plasma EPA were associated with a reduced brain atrophy rate in
the medial temporal lobe (14). However, results from random-
ized clinical trials including v-3 supplementation are not
equally convincing (9, 15). One reason for this discrepancy may
be a failure to identify the relevant subgroups that are likely to
benefit from supplementation (16).
Homocysteine is a nonessential, sulfur-containing amino acid
synthesized endogenously from methionine. Raised plasma total
homocysteine (tHcy) is a recognized modifiable risk factor for
cognitive impairment, dementia, and AD (10, 17, 18). The at-
rophy rate of the brain is faster at low plasma vitamin B-12
concentrations (19) and at high plasma tHcy concentrations (20,
21). Results from Homocysteine and B Vitamins in Cognitive
Impairment (VITACOG), a randomized clinical trial with
homocysteine-lowering B vitamins in older people with MCI,
showed that treatment with high doses of B vitamins markedly
reduced the global brain atrophy rate, as well as atrophy rates in
those gray matter regions most commonly associated with AD
(20, 21).
Multiple links between v-3 fatty acids and homocysteine have
been suggested. There is an inverse correlation between tHcy
and plasma concentrations of v-3 fatty acids (22, 23), and B
vitamins are important for the methylation and assembly of
phospholipids (24, 25). The purpose of this study was to de-
termine whether the plasma long-chain v-3 fatty acid status
modifies the effect of high-dose B vitamin supplementation on
brain atrophy rates in elderly subjects with MCI.
METHODS
Participants
This retrospective study was conducted as a part of the
VITACOG trial (registered at www.controlled-trials.com as
ISRCTN94410159). The study received approval of the
Oxfordshire National Health Service research ethics committee
A (COREC 04/Q1604/100) and was carried out according to the
principles of the Declaration of Helsinki. Participants were
recruited in the Oxford area between April 2004 and November
2006, and all participants gave their written informed consent.
The study protocol and participants, along with inclusion and
exclusion criteria, and details on randomization have been de-
scribed in detail previously (20). In short, 646 volunteers aged
.70 y were assessed for eligibility and a diagnosis of MCI.
After random assignment of eligible MCI subjects and some
withdrawals, 133 subjects received 0.8 mg folic acid, 0.5 mg
vitamin B-12, and 20 mg vitamin B-6 (TrioBe Plus; Meda AB/
Recip AB) once daily for 24 mo, and 133 received placebo,
using a parallel design. Adherence was assessed by counting
tablets returned and by measurement of blood vitamin concen-
trations after the completion of the study (20). Of the 223
subjects completing the trial (110 in the B vitamin group and
113 in the placebo group), 187 had given their written consent to
undergo MRI scans at baseline and follow-up. Of these, 180
completed scans at both time points. In the present study, we
only included participants who completed the trial with tech-
nically suitable MRI scans at both occasions (n= 168). Of these,
85 were in the B vitamin group, and 83 were in the placebo
group.
MRI scans
The MRI protocol used in the VITACOG study has been
described elsewhere (20). Briefly, baseline and follow-up volu-
metric cranial MRI scans were carried out at the Oxford Centre
for Clinical Magnetic Resonance Research by using a 1.5T MRI
system (Sonata; Siemens Medical Solutions). Whole-brain at-
rophy per year was estimated from magnetic resonance images
taken at baseline and follow-up by using the fully automated
SIENA protocol (26). Normalized brain volume at baseline was
estimated from a single image by using a cross-sectional method
(SIENAX).
Biochemical assays
Plasma was prepared from nonfasting blood samples collected
at baseline and after 2 y of intervention. Total fatty acid con-
centrations were analyzed by gas chromatography coupled to
mass spectrometry by using a modified in situ transesterification
protocol for fatty acid methyl ester preparation (see supple-
mental material). Plasma samples were stored between 4 and 6 y
at 2808C before analysis. Under these conditions, fatty acid
compositions in serum have been shown to be stable for up to
10 y (27). Fatty acid concentrations were expressed in absolute
concentrations (mmol/L) unless otherwise stated. The between-
day CVs for DHA and EPA were 4.7% and 3.9%, respectively,
and ,10% for the remaining fatty acids (Supplemental Table
1). ApoE genotype, plasma tHcy, folate, and vitamin B-12 were
analyzed as previously described (19).
Statistical analysis
The main objective of this retrospective study was to investigate
the effect of B vitamin treatment on brain atrophy rates as
a function of baseline v-3 fatty acid status. Combined v-3, here
defined as the sum of DHA and EPA, was used along with DHA
and EPA separately. Because fatty acid concentrations demon-
strated a skewed distribution, all fatty acid concentrations were
transformed by using the natural log before analyses. Paired
ttests were used to investigate whether v-3 fatty acids changed
between baseline and follow-up in subjects treated with B vi-
tamins and placebo. Independent-samples ttests were used for
comparisons between groups and partial correlations for analy-
sis of the relation between baseline fatty acid concentrations and
atrophy rates. Comparisons of correlations were performed by
using the Fisher r-to-ztransformation. For the main objective of
the study, the linear regression model was used as the main
statistical model with brain atrophy as the dependent variable
and B vitamin treatment and v-3 fatty acid status, as expressed
in tertiles, as the main predictors or independent variables. The
model also adjusted for age, sex, initial brain volume, ApoE
status, education level, and baseline levels of diastolic blood
pressure, triglyceride concentration (log), creatinine, and tHcy
(log) [see Smith et al. (20) for why these variables were chosen].
Diagnostic checks to assess assumptions of the linear model
were carried out along with outliers and leverage analysis. To
test that the effect of treatment on brain atrophy is dependent on
the concentration of v-3 fatty acids within the linear regression
model, we used a global Fisher test (Ftest) to test the null
hypothesis that all interaction terms are equal to 0. Two in-
teraction terms were included in the linear regression model: the
2of7 JERNERE
´N ET AL.

treatment by second tertile indicator and the treatment by third
tertile indicator. First tertile was then considered as reference.
The Ftest tested the null hypothesis (H
0
): treatment by second
tertile interaction = treatment by third tertile interaction = 0.
This has been done by using the function “linearHypothesis” in
the R package “car.” Because this is an interaction term, the
same procedure was used to test whether the effect of baseline
v-3 fatty acids on brain atrophy is dependent on B vitamin
treatment. Pairwise comparisons among the fatty acid tertiles
following ANCOVA, adjusted for covariates specified above (no
adjustment for multiple comparisons was made), were used to
examine the differences in brain atrophy rates between the
placebo group and the B vitamin–treated group.
In further analyses based on tHcy status, a threshold of 11.3
mmol/L was used to define low and high tHcy groups. This value
corresponds to the median baseline tHcy concentration in the
VITACOG cohort: previous studies showed that the beneficial
effect of B vitamin treatment on global and regional brain at-
rophy (20, 21) and cognitive decline (28) was dependent on
baseline tHcy. The impact of baseline tHcy status on the
TABLE 1
Characteristics of participants in the VITACOG study
1
All (n= 168) Placebo (n= 83) B Vitamins (n= 85)
Age,
2
y 76.6 (75.9, 77.3) 76.2 (75.3, 77.2) 77.0 (75.8, 78.1)
Women, n(%) 102 (60.7) 52 (62.7) 50 (58.8)
Brain volume,
2
mL 1381 (1369, 1393) 1376 (1360, 1391) 1387 (1368, 1405)
BMI,
2
kg/m
2
26.0 (25.4, 26.5) 26.6 (25.7, 27.6) 25.3 (24.6, 26.0)
4
Systolic BP,
2
mm Hg 148 (144, 151) 147 (143, 152) 148 (143, 153)
Diastolic BP,
2
mm Hg 80 (78, 82) 80 (78, 83) 80 (77, 82)
Depression score (GDS)
2
6.5 (5.8, 7.2) 7.5 (6.4, 8.9) 5.6 (4.5, 6.4)
4
Alcohol consumption,
2
U/wk 8.1 (6.6, 9.6)
3
7.2 (5.1, 9.2) 9.0 (6.8, 11.2)
Antidiabetic drugs, n(%) 14 (8.3) 10 (12) 4 (4.7)
Smoker, anytime, n(%) 81 (48.2) 43 (51.8) 38 (44.7)
Use of B vitamins, n(%) 31 (18.5) 17 (20.5) 14 (16.5)
Use of fish oils, v-3, n(%) 67 (39.9) 31 (37.3) 36 (42.4)
ApoE4 carriers, n(%) 51 (30.4) 29 (35) 22 (25.9)
tHcy, baseline,
5
mmol/L 11.3 (10.8, 11.8) 11.3 (10.6, 12.0) 11.3 (10.6, 12.0)
tHcy, follow-up,
5
mmol/L 10.3 (9.8, 10.8) 12.1 (11.4, 12.9)
6
8.7 (8.3, 9.2)
4,6
Vitamin B-12, baseline,
5
pmol/L 331 (313, 350) 333 (310, 357) 330 (303, 360)
Vitamin B-12, follow-up,
5
pmol/L 497 (462, 535) 366 (335, 400)
6
672 (626, 722)
4,6
Folate, baseline,
5
nmol/L 23.3 (21.2, 25.7) 24.2 (21.4, 27.5) 22.4 (19.4, 25.9)
Folate, follow-up,
5
nmol/L 45.4 (39.9, 51.6) 24.9 (21.4, 29.1) 82.1 (74.6, 90.4)
4,6
v-3, baseline,
5
mmol/L 472 (439, 508) 488 (442, 539) 457 (411, 508)
v-3, follow-up,
5
mmol/L 465 (434, 499) 479 (433, 530) 452 (410, 499)
EPA, baseline,
5
mmol/L 177 (161, 195) 181 (158, 208) 173 (151, 197)
EPA, follow-up,
5
mmol/L 177 (162, 195) 181 (158, 207) 174 (153, 198)
DHA, baseline,
5
mmol/L 288 (270, 307) 299 (276, 325) 277 (252, 306)
DHA, follow-up,
5
mmol/L 280 (264, 298) 290 (266, 316) 271 (248, 296)
1
VITACOG subjects with available MRI data at start and finish. ApoE, apolipoprotein E4; BP, blood pressure; GDS,
geriatric depression scale; tHcy, total homocysteine. v-3 denotes the sum of the amounts of EPA+DHA.
2
Values are means; 95% CIs in parentheses.
3
Excluding one high outlier.
4
P,0.05 (2-tailed) compared with placebo group as assessed by independent ttests.
5
Values are geometric means; 95% CIs in parentheses.
6
P,0.05 (2-tailed) compared with baseline value as assessed by paired ttests.
TABLE 2
Plasma fatty acid concentrations (absolute) at baseline as predictors of yearly brain atrophy rate (%)
1
All Placebo B Vitamins
Pfor differenceb(SE) Partial rP b(SE) Partial rP b(SE) Partial rP
v-3 20.29 (0.12) 20.21 0.009 20.08 (0.16) 20.06 0.618 20.47 (0.16) 20.36 0.002 0.047
EPA 20.17 (0.09) 20.17 0.037 20.04 (0.16) 20.04 0.738 20.29 (0.13) 20.27 0.019 0.129
DHA 20.36 (0.13) 20.22 0.005 20.11 (0.20) 20.06 0.592 20.56 (0.16) 20.39 0.001 0.026
1
Unstandardized coefficients (b) with their SE and partial correlation coefficients with their Pvalues. The linear
regression model was adjusted for age, sex, initial brain volume, ApoE status, education level, diastolic blood pressure at
baseline, and baseline concentrations of triglycerides (log), creatinine, and total homocysteine (log). v-3 denotes the sum of
the amounts of EPA+DHA. All fatty acid variables were entered as the natural log of the baseline concentrations in mmol/L
to ensure normal distribution. Pfor the difference in slopes between placebo and B vitamins was calculated by using the
Fisher r-to-ztransformation. n= 168 (all), n= 83 (placebo), and n= 85 (B vitamins).
v-3, B VITAMINS, AND BRAIN ATROPHY 3of7

treatment effect according to v-3 fatty acid tertiles was in-
vestigated by using a linear regression model with brain atrophy
as the dependent variable and B vitamin treatment, v-3 fatty
acid tertiles, and tHcy status (high/low) as the main predictors or
independent variables. The model adjusted for the same vari-
ables as specified above. Two 3-way interaction terms were
included in the model: the treatment by tHcy status by second
tertile indicator and the treatment by tHcy status by third tertile
indicator. First tertile was then considered as reference. Statis-
tical analysis was carried out by using the R statistical program
version 3.03 (The R Foundation, www.R-project.org). Pvalues
(2-tailed) ,0.05 (P,0.1 for interaction term) were considered
statistically significant.
RESULTS
Participants
Selected characteristics of the study population are summa-
rized in Table 1. As previously reported (20), the baseline
characteristics and losses to follow-up in the active and placebo
groups were similar. Demographic comparison of those who
completed the MRI scans compared with the whole cohort can
be found in our 2 previous reports (20, 28). The adherence was
good in both groups, as assessed by counting returned tablets
and measuring plasma vitamins and related compounds (20).
There were no significant safety issues and no difference in the
adverse events between the intervention groups. At follow-up,
the B vitamin group showed a marked improvement in plasma
vitamin status, whereas the placebo group showed little or no
change (20). Baseline concentrations of v-3 fatty acids did not
differ between the treatment groups and did not change signif-
icantly from baseline to follow-up in either the B vitamin or
placebo group, as judged by paired ttests (Table 1).
Partial correlation analyses were performed to investigate
associations between absolute v-3 fatty acid concentrations and
brain atrophy rates in the whole study group and stratified by
treatment group. In the total study group, the combined v-3,
DHA, and EPA showed significant negative correlations with
brain atrophy rates (Table 2).
In the B vitamin group, inverse correlations between the
combined v-3, DHA, and EPA and brain atrophy rates were
significant (Table 2). None of these correlations were significant
in the placebo group. The differences in correlation coefficients
in the placebo group and the B vitamin–treated group were
statistically significant for the combined v-3 fatty acids and
DHA (Table 2). Correlations of all fatty acids assayed, using
both absolute and relative fatty acid concentrations, can be
found in Supplemental Tables 2 and 3.
Effect of B vitamins on brain atrophy according to v-3
concentration
Yearly atrophy rates in the placebo and B vitamin groups
according to tertiles of baseline combined v-3, DHA, and EPA
concentrations are shown in Figure 1. Using linear regression,
we found a significant interaction between B vitamin treatment
and combined v-3 (P= 0.024) and for EPA tertiles (P= 0.085).
The interaction was not significant for DHA (P= 0.134). In B
vitamin–treated patients, linear regression analysis revealed
significant trends for the combined v-3 (P= 0.018), EPA (P=
0.038), and DHA (P= 0.012); in other words, the atrophy rate
decreases with increasing v-3 fatty acids.
Adjusted pairwise comparisons following ANCOVA showed
that, in subjects with high baseline combined v-3 (.590 mmol/L),
B vitamin treatment slowed atrophy rates by 40.0% (P= 0.023)
compared with placebo (Figure 1). A similar result was observed
in those with high EPA (.222 mmol/L) (45.8% reduction, P= 0.011)
FIGURE 1. Brain atrophy rates among subjects receiving placebo (black)
and high-dose B vitamins (gray) (mean 6SEM) according to tertiles of
plasma baseline combined v-3 (top), EPA (middle), and DHA (bottom),
adjusted for age, sex, and initial brain volume, ApoE status, educational
level, baseline diastolic blood pressure, and baseline plasma concentrations
of triglycerides (log), creatinine, and total homocysteine (log). Interaction
between treatment groups and v-3 concentrations by tertiles were evaluated
by using a linear regression model, and differences between tertiles within
each treatment group were assessed by ANCOVA. *P,0.05 (2-tailed)
between placebo and B vitamin–treated subjects by adjusted pairwise com-
parisons. Group sizes in the placebo group and B vitamin group were 27–28
among the v-3 tertiles, 26–29 among the EPA tertiles, and 24–32 among the
DHA tertiles. v-3 denotes the sum of the amounts of EPA+DHA.
4of7 JERNERE
´N ET AL.

and for DHA (.345 mmol/L) (43.4% reduction, P= 0.004).
B vitamin treatment had no significant effect on atrophy rates
among subjects in the bottom tertiles of fatty acid concentra-
tions (combined v-3 ,390 mmol/L; EPA ,136 mmol/L; DHA
,245 mmol/L).
Effect of B vitamins on brain atrophy according to baseline
tHcy and v-3 concentration
To follow up our previous results that the effect of B vitamins
depends on baseline plasma tHcy (20, 28), we examined the
association of baseline fatty acid with atrophy in subjects with
low and high baseline tHcy values (threshold, 11.3 mmol/L). B
vitamin treatment had no significant effect compared with pla-
cebo in subjects with low baseline tHcy concentrations, in-
dependent of DHA, EPA, or v-3 concentration (Table 3 and
Supplemental Figure 1). In contrast, in subjects with high
baseline tHcy, the effect of B vitamin treatment changed ac-
cording to plasma v-3 fatty acid concentration: the rate of at-
rophy was significantly slowed by treatment (by ;70%) in
subjects in the upper v-3 fatty acid tertiles (Table 3 and Sup-
plemental Figure 1), whereas no effect was observed in the
lower tertile. Group sizes in these analyses were not ideal and
ranged from 4 to 23 (see Supplemental Figure 1).
DISCUSSION
In this retrospective exploratory analysis of data from a ran-
domized, placebo-controlled trial, we observed a significant
interaction effect between high-dose B vitamin treatment and v-3
fatty acid concentrations on rate of atrophy of the whole brain.
The beneficial effect of high-dose B vitamin supplementation
was augmented by a high baseline status of v-3 fatty acid. In
subjects with high plasma concentrations of v-3 fatty acids
(EPA+DHA .590 mmol/L), B vitamin supplementation slowed
the mean brain atrophy rate by 40% compared with subjects in
the placebo group. In contrast, in subjects with low v-3 fatty
acid concentrations (,390 mmol/L), there was no beneficial
effect of B vitamins on brain atrophy.
One major effect of the high-dose B vitamin treatment is to
lower plasma tHcy. We found that the effect of B vitamins in the
higher tertiles of v-3 fatty acids is limited to patients with
baseline tHcy concentrations above the median ($11.3 mmol/L).
In this subgroup, the brain atrophy rate among patients in the
upper tertile of v-3 fatty acid concentration (.590 mmol/L) was
reduced by ;70% by B vitamin treatment compared with pla-
cebo. Although these results should be interpreted with some
caution due to the small group sizes, our results indicate that the
effect of B vitamins in subjects with moderate to high v-3 fatty
acid concentrations is driven mainly by beneficial effects in
subjects with elevated baseline tHcy concentrations. We there-
fore hypothesize that low tHcy concentrations, which are the
consequence of B vitamin treatment, facilitate the protective
effect of v-3 fatty acids against brain atrophy (Table 2).
Long-chain v-3 fatty acids have been associated with pro-
tective roles in dementia and AD in epidemiologic studies (see
Introduction). Recently, Witte and coworkers (29) showed that
daily fish-oil supplementation (880 mg DHA and 1320 mg EPA)
in healthy elderly for 26 wk prevented the loss of total gray
matter volume. Only 2 studies investigating v-3 fatty acids
along with B vitamins have been reported. One of these in-
vestigated a nutritional supplement that also included v-3 fatty
acids (EPA, 300 mg; DHA, 1200 mg) and B vitamins (folic acid,
0.4 mg; vitamin B-6, 1 mg; vitamin B-12, 0.003 mg) (30). The
supplement produced some beneficial effects in mild AD when
given for 24 wk, but this was not confirmed in a larger follow-up
study (31). The second study used a 2 32 factorial design, with
one arm including B vitamins (folate, 0.56 mg; vitamin B-6,
3 mg; vitamin B-12, 0.02 mg) and the other including v-3 fatty
acids (EPA, 400 mg; DHA, 200 mg), and found that the com-
bination of both nutrient groups decreased the likelihood of
a lower score on a temporal orientation task in a subgroup with
prior stroke (32). Both studies were in populations with different
characteristics and used lower doses of B vitamins compared
with VITACOG, and none of these studies reported brain
TABLE 3
Brain atrophy rates per year (%) after B vitamin treatment compared with placebo, stratified by baseline v-3 and tHcy status
1
v-3 EPA DHA
Placebo B vitamins Difference, % Pvalue Placebo B vitamins Difference, % Pvalue Placebo B vitamins Difference, % Pvalue
tHcy ,11.3 mmol/L
Low tertile 1.00 1.00 0.0 0.991 0.99 0.94 5.1 0.860 1.16 1.04 10.3 0.635
Middle tertile 1.07 0.83 22.4 0.354 0.98 0.84 14.3 0.595 0.88 0.82 6.8 0.807
High tertile 0.89 0.79 11.2 0.600 0.92 0.79 14.1 0.557 0.89 0.70 21.3 0.338
tHcy $11.3 mmol/L
Low tertile 1.03 1.23 219.4 0.409 1.09 1.14 24.6 0.821 1.07 1.10 22.8 0.898
Middle tertile 1.39 0.54 61.2 ,0.001 1.33 0.52 60.9 ,0.001 1.33 0.55 58.6 ,0.001
High tertile 1.48 0.47 68.2 ,0.001 1.36 0.37 72.8 ,0.001 1.59 0.43 73.0 0.002
1
Brain atrophy rates per year for the placebo group and B vitamin–treated group are given in percentages. Percent difference is defined as follows: brain
atrophy in placebo group minus brain atrophy in B vitamin group, divided by the atrophy rate in the placebo group, multiplied by hundred. The linear
regression model included treatment, age, sex, initial brain volume, ApoE status, education level, baseline diastolic blood pressure, and baseline plasma
concentrations of triglycerides (log), creatinine, total homocysteine (median split, 11.3 mmol/L), and the tertiles of the 3 fatty acid variables. Pvalues for the
3-way interactions between treatment 3tHcy status (high/low) 3fatty acid status (tertiles) were significant for the high tertiles: combined v-3 (P= 0.026),
EPA (P= 0.055), and DHA (P= 0.040). Group sizes in these analyses ranged from 4 to 23, with a median of 14 (see Supplemental Figure 1). v-3 denotes the
sum of the amounts of EPA+DHA. For v-3, the low, middle, and high tertiles were defined as ,390, 390–590, and .590 mmol/L, respectively. The
corresponding values were ,136, 136–222, and .222 mmol/L for EPA and ,245, 245–345, and .345 mmol/L for DHA.
v-3, B VITAMINS, AND BRAIN ATROPHY 5of7

volume or brain atrophy data. The results of these studies are
therefore difficult to compare with ours.
Fatty acids are delivered to various target tissues as compo-
nents of phospholipids, of which phosphatidylcholine is the most
abundant in plasma. Experiments in rodents have shown that
phosphatidylcholine molecules enriched in DHA are distributed
selectively to certain tissues, including the brain (33). It is
therefore conceivable that a reduced phosphatidylcholine syn-
thesis will affect the transport of v-3 fatty acids to the brain, with
possible implications for brain health. Indeed, low plasma con-
centrations of phosphatidylcholine enriched in DHA and EPA have
been linked to the risk of dementia (34, 35). Phosphatidylcholine is
synthesized in the liver via the cytidine 5#-diphosphate–choline
dependent pathway or from phosphatidylethanolamine through
3 consecutive S-adenosylmethionine–dependent methylation re-
actions catalyzed by phosphatidylethanolamine N-methyltransferase
(PEMT). DHA content in phosphatidylcholine has been pro-
posed as a marker of PEMT activity (36), and plasma DHA is
disproportionally reduced by disruption of PEMT in a mouse
model (37). As a consequence, PEMT activity is considered
vital for the delivery and incorporation of v-3 fatty acids into the
brain (37, 38).
PEMT is inhibited by S-adenosylhomocysteine (SAH), the
precursor of homocysteine. At high tHcy concentrations, SAH
accumulates, which in turn may reduce PEMT activity. In pa-
tients with AD, there is an inverse correlation between plasma
SAH and DHA concentrations in erythrocyte phosphatidylcho-
line, possibly because of inhibition of PEMT by SAH (39). In
chick embryos, exposure to homocysteine altered brain lipid
composition, with reduced concentrations of phosphatidylcho-
line and an increase of phosphatidylethanolamine while also
reducing the proportion of DHA in brain cell membranes (40).
In rats, a B vitamin–enriched diet increased plasma total DHA
concentration compared with a B vitamin–deficient diet (41).
These reports are consistent with the hypothesis that a good B
vitamin status and low tHcy concentrations are required for an
optimal utilization and distribution of v-3 fatty acids.
Although a biochemical interaction at the level of phospho-
lipid metabolism seems likely, there are other potential expla-
nations for the observed interaction. For example, it is possible
that both v-3 fatty acids and B vitamins protect against hyper-
phosphorylation of tau, with potential consequences for tangle
formation (42, 43). Also, both B vitamins and v-3 fatty acids
might attenuate inflammation associated with AD. A combina-
tion of B vitamins and v-3 fatty acids was recently shown to
reduce oxidative stress and inflammation in a rodent model of
hypertension (44). Whether any of these mechanism explain the
interaction reported herein is a focus for future studies.
There are some limitations of this study. In this study, we did
not measure phosphatidylcholine, which is probably the best
source of DHA for the brain (34, 38). In future studies, it would be
valuable to also investigate the distribution of v-3 fatty acids
in the various plasma compartments and also examine the effect
of B vitamins on phosphatidylcholine. Finally, our study was
a randomized controlled trial with B vitamins, not v-3 fatty
acids. In future trials, it would be useful to also include treat-
ment with v-3 fatty acids.
In conclusion, we have shown that the effect of B vitamin
supplementation on brain atrophy rates depends on pre-existing
plasma v-3 fatty acid concentrations; this finding could possibly
explain why some B vitamin trials on brain function have failed.
Conversely, our results suggest that tHcy status may also de-
termine the effects of v-3 fatty acids in cognitive decline and
dementia and so could explain why some trials of v-3 fatty acids
have failed. Altogether, our results emphasize the importance of
identifying subgroups in clinical trials. A randomized clinical
trial of B vitamin and v-3 fatty acid supplementation using
a232 factorial design is clearly warranted to shed light on the
roles of homocysteine and v-3 fatty acids in brain atrophy, MCI,
dementia, and AD.
The authors’ responsibilities were as follows—FJ, AKE, HR, and ADS:
designed the research; FJ: conducted the lipid analyses and wrote the first
draft of the manuscript; FJ, AO, and SMS: analyzed data; and all authors:
critically reviewed the analyses and the manuscript. ADS is named as in-
ventor on 3 patents held by the University of Oxford on the use of B vitamins
to treat AD or MCI (US6008221, US6127370, and PCT/GB2010/051557);
HR is named as inventor on patent PCT/GB2010/051557. Under the Uni-
versity of Oxford’s rules, they could benefit financially if the patents are
exploited. FJ, AKE, AO, and SMS reported no personal or financial conflicts
of interest. None of the funders or the sponsor (University of Oxford) played
any role in the design and conduct of the study; collection, management,
analysis, and interpretation of the data; or preparation, review, or approval of
the manuscript.
REFERENCES
1. Gauthier S, Reisberg B, Zaudig M, Petersen RC, Ritchie K, Broich K,
Belleville S, Brodaty H, Bennett D, Chertkow H, et al. Mild cognitive
impairment. Lancet 2006;367:1262–70.
2. Petersen RC. Mild cognitive impairment as a diagnostic entity. J Intern
Med 2004;256:183–94.
3. Fox NC, Scahill RI, Crum WR, Rossor MN. Correlation between rates
of brain atrophy and cognitive decline in AD. Neurology 1999;52:
1687–9.
4. Smith AD. Imaging the progression of Alzheimer pathology through
the brain. Proc Natl Acad Sci USA 2002;99:4135–7.
5. Jack CR Jr., Shiung MM, Gunter JL, O’Brien PC, Weigand SD,
Knopman DS, Boeve BF, Ivnik RJ, Smith GE, Cha RH, et al. Com-
parison of different MRI brain atrophy rate measures with clinical
disease progression in AD. Neurology 2004;62:591–600.
6. Risacher SL, Saykin AJ, West JD, Shen L, Firpi HA, McDonald BC.
Baseline MRI predictors of conversion from MCI to probable AD in
the ADNI cohort. Curr Alzheimer Res 2009;6:347–61.
7. Bradley KM, Bydder GM, Budge MM, Hajnal JV, White SJ, Ripley
BD, Smith AD. Serial brain MRI at 3-6 month intervals as a surrogate
marker for Alzheimer’s disease. Br J Radiol 2002;75:506–13.
8. Jack CR Jr., Shiung MM, Weigand SD, O’Brien PC, Gunter JL, Boeve
BF, Knopman DS, Smith GE, Ivnik RJ, Tangalos EG, et al. Brain at-
rophy rates predict subsequent clinical conversion in normal elderly
and amnestic MCI. Neurology 2005;65:1227–31.
9. Huang TL. Omega-3 fatty acids, cognitive decline, and Alzheimer’s
disease: a critical review and evaluation of the literature. J Alzheimers
Dis 2010;21:673–90.
10. Beydoun MA, Beydoun HA, Gamaldo AA, Teel A, Zonderman AB,
Wang Y. Epidemiologic studies of modifiable factors associated with
cognition and dementia: systematic review and meta-analysis. BMC
Public Health 2014;14:643.
11. Tan ZS, Harris WS, Beiser AS, Au R, Himali JJ, Debette S, Pikula A,
Decarli C, Wolf PA, Vasan RS, et al. Red blood cell omega-3 fatty acid
levels and markers of accelerated brain aging. Neurology 2012;78:
658–64.
12. Virtanen JK, Siscovick DS, Lemaitre RN, Longstreth WT, Spiegelman
D, Rimm EB, King IB, Mozaffarian D. Circulating omega-3 poly-
unsaturated fatty acids and subclinical brain abnormalities on MRI in
older adults: the cardiovascular health study. J Am Heart Assoc 2013;2:
e000305.
13. Pottala JV, Yaffe K, Robinson JG, Espeland MA, Wallace R, Harris
WS. Higher RBC EPA + DHA corresponds with larger total brain
and hippocampal volumes: WHIMS-MRI Study. Neurology 2014;82:
435–42.
6of7 JERNERE
´N ET AL.

14. Samieri C, Maillard P, Crivello F, Proust-Lima C, Peuchant E, Helmer
C, Amieva H, Allard M, Dartigues JF, Cunnane SC, et al. Plasma long-
chain omega-3 fatty acids and atrophy of the medial temporal lobe.
Neurology 2012;79:642–50.
15. Sydenham E, Dangour AD, Lim WS. Omega 3 fatty acid for the
prevention of cognitive decline and dementia. Cochrane Database Syst
Rev 2012;6:CD005379.
16. Frautschy SA, Cole GM. What was lost in translation in the DHA trial
is whom you should intend to treat. Alzheimers Res Ther 2011;3:2.
17. Wald DS, Kasturiratne A, Simmonds M. Serum homocysteine and
dementia: meta-analysis of eight cohort studies including 8669 par-
ticipants. Alzheimers Dement 2011;7:412–7.
18. Smith AD. The worldwide challenge of the dementias: a role for
B-vitamins and homocysteine? Food Nutr Bull 2008;29(2 Suppl):
S143–72.
19. Vogiatzoglou A, Refsum H, Johnston C, Smith SM, Bradley KM, de
Jager C, Budge MM, Smith AD. Vitamin B12 status and rate of
brain volume loss in community-dwelling elderly. Neurology 2008;71:
826–32.
20. Smith AD, Smith SM, de Jager CA, Whitbread P, Johnston C,
Agacinski G, Oulhaj A, Bradley KM, Jacoby R, Refsum H. Homo-
cysteine-lowering by B-vitamins slows the rate of accelerated brain
atrophy in mild cognitive impairment: a randomized controlled trial.
PLoS ONE 2010;5:e12244.
21. Douaud G, Refsum H, de Jager CA, Jacoby R, Nichols TE, Smith SM,
Smith AD. Preventing Alzheimer’s disease-related gray matter atrophy
by B-vitamin treatment. Proc Natl Acad Sci U S A 2013;110:9523–8.
22. Mehmetoglu I, Yerlikaya FH, Kurban S, Polat H. Plasma omega-3 fatty
acid levels negatively and omega-6 fatty acid levels pos itively as-
sociated with other cardiovascular risk factors including homo-
cysteine in severe obese subjects. Asia Pac J Clin Nutr 2012;21:
519–25.
23. Huang T, Zheng J, Chen Y, Yang B, Wahlqvist ML, Li D. High con-
sumption of omega-3 polyunsaturated fatty acids decrease plasma
homocysteine: a meta-analysis of randomized, placebo-controlled tri-
als. Nutrition 2011;27:863–7.
24. Akesson B, Fehling C, Ja
¨gerstad M, Stenram U. Effect of experimental
folate deficiency on lipid metabolism in liver and brain. Br J Nutr 1982;
47:505–20.
25. da Silva RP, Kelly KB, Al Rajabi A, Jacobs RL. Novel insights on
interactions between folate and lipid metabolism. Biofactors 2014;40:
277–83.
26. Smith SM, Zhang Y, Jenkinson M, Chen J, Matthews PM, Federico A,
De Stefano N. Accurate, robust, and automated longitudinal and cross-
sectional brain change analysis. Neuroimage 2002;17:479–89.
27. Matthan NR, Ip B, Resteghini N, Ausman LM, Lichtenstein AH. Long-
term fatty acid stability in human serum cholesteryl ester, triglyceride,
and phospholipid fractions. J Lipid Res 2010;51:2826–32.
28. de Jager CA, Oulhaj A, Jacoby R, Refsum H, Smith AD. Cognitive and
clinical outcomes of homocysteine-lowering B-vitamin treatment in
mild cognitive impairment: a randomized controlled trial. Int J Geriatr
Psychiatry 2012;27:592–600.
29. Witte AV, Kerti L, Hermannstadter HM, Fiebach JB, Schreiber SJ,
Schuchardt JP, Hahn A, Floel A. Long-chain omega-3 fatty acids im-
prove brain function and structure in older adults. Cereb Cortex 2014;
24:3059–68.
30. Scheltens P, Twisk JW, Blesa R, Scarpini E, von Arnim CA, Bongers
A, Harrison J, Swinkels SH, Stam CJ, de Waal H, et al. Efficacy of
Souvenaid in mild Alzheimer’s disease: results from a randomized,
controlled trial. J Alzheimers Dis 2012;31:225–36.
31. Shah RC, Kamphuis PJ, Leurgans S, Swinkels SH, Sadowsky CH,
Bongers A, Rappaport SA, Quinn JF, Wieggers RL, Scheltens P, et al.
The S-Connect study: results from a randomized, controlled trial of
Souvenaid in mild-to-moderate Alzheimer’s disease. Alzheimers Res
Ther 2013;5:59.
32. Andreeva VA, Kesse-Guyot E, Barberger-Gateau P, Fezeu L, Hercberg
S, Galan P. Cognitive function after supplementation with B-vitamins
and long-chain omega-3 fatty acids: ancillary findings from the SU.
FOL.OM3 randomized trial. Am J Clin Nutr 2011;94:278–86.
33. Polozova A, Gionfriddo E, Salem N Jr. Effect of docosahexaenoic acid
on tissue targeting and metabolism of plasma lipoproteins. Prosta-
glandins Leukot Essent Fatty Acids 2006;75:183–90.
34. Schaefer EJ, Bongard V, Beiser AS, Lamon-Fava S, Robins SJ, Au R,
Tucker KL, Kyle DJ, Wilson PW, Wolf PA. Plasma phosphatidyl-
choline docosahexaenoic acid content and risk of dementia and
Alzheimer disease: the Framingham Heart Study. Arch Neurol 2006;
63:1545–50.
35. Whiley L, Sen A, Heaton J, Proitsi P, Garcia-Gomez D, Leung R,
Smith N, Thambisetty M, Kloszewska I, Mecocci P, et al. Evidence of
altered phosphatidylcholine metabolism in Alzheimer’s disease. Neu-
robiol Aging 2014;35:271–8.
36. da Costa KA, Sanders LM, Fischer LM, Zeisel SH. Docosahexaenoic
acid in plasma phosphatidylcholine may be a potential marker for in
vivo phosphatidylethanolamine N-methyltransferase activity in hu-
mans. Am J Clin Nutr 2011;93:968–74.
37. Watkins SM, Zhu X, Z eis el SH. Phosphatid yle tha nolamine-N -
methyltransferase activity and dietary choline regulate liver-plasma
lipid flux and essential fatty acid metabolism in mice. J Nutr 2003;133:
3386–91.
38. Magret V, Elkhalil L, Nazih-Sanderson F, Martin F, Bourre JM,
Fruchart JC, Delbart C. Entry of polyunsaturated fatty acids into the
brain: evidence that high-density lipoprotein-induced methylation of
phosphatidylethanolamine and phospholipase A2 are involved. Bio-
chem J 1996;316:805–11.
39. Selley ML. A metabolic link between S-adenosylhomocysteine and
polyunsaturated fatty acid metabolism in Alzheimer’s disease. Neurobiol
Aging 2007;28:1834–9.
40. Miller RR, Leanza CM, Phillips EE, Blacquire KD. Homocysteine-
induced changes in brain membrane composition correlate with
increased brain caspase-3 activities and reduced chick embryo vi-
ability. Comp Biochem Physiol B Biochem Mol Biol 2003;136:
521–32.
41. van Wijk N, Watkins CJ, Hageman RJ, Sijben JC, Kamphuis PG,
Wurtman RJ, Broersen LM. Combined dietary folate, vitamin B-12,
and vitamin B-6 intake influences plasma docosahexaenoic acid con-
centration in rats. Nutr Metab (Lond) 2012;9:49.
42. Cole GM, Frautschy SA. DHA may prevent age-related dementia.
J Nutr 2010;140:869–74.
43. Sontag E, Nunbhakdi-Craig V, Sontag JM, Diaz-Arrastia R, Ogris E,
Dayal S, Lentz SR, Arning E, Bottiglieri T. Protein phosphatase 2A
methyltransferase links homocysteine metabolism with tau and amy-
loid precursor protein regulation. J Neurosci 2007;27:2751–9.
44. Kemse NG, Kale AA, Joshi SR. A combined supplementation of
omega-3 fatty acids and micronutrients (folic acid, vitamin B12) re-
duces oxidative stress markers in a rat model of pregnancy induced
hypertension. PLoS ONE 2014;9:e111902.
v-3, B VITAMINS, AND BRAIN ATROPHY 7of7

Online Supplemental Material
Brain atrophy in cognitively impaired elderly: the importance of long-chain omega-3
fatty acids and B-vitamin status in a randomized controlled trial
Fredrik Jernerén1*, Amany K Elshorbagy1,2, Abderrahim Oulhaj3, Stephen M Smith4, Helga
Refsum1,5, and A David Smith1
1Oxford Project to Investigate Memory and Ageing (OPTIMA), Department of
Pharmacology, University of Oxford, Oxford, United Kingdom
2Department of Physiology, Faculty of Medicine, University of Alexandria, Alexandria,
Egypt
3Institute of Public Health, College of Medicine and Health Sciences, United Arab Emirates
University, Al Ain, United Arab Emirates
4Functional Magnetic Resonance Imaging of the Brain Centre (FMRIB), Nuffield
Department of Clinical Neurosciences, University of Oxford, John Radcliffe Hospital,
Oxford, United Kingdom
5Department of Nutrition, Institute of Basic Medical Sciences, Faculty of Medicine,
University of Oslo, Oslo, Norway
*Corresponding author: Department of Pharmacology, Mansfield Road, OX1 3QT, Oxford,
United Kingdom, Tel. +44(0)1865 271877, Fax +44(0)1865271853; E-mail:
fredrik.jerneren@pharm.ox.ac.uk
Quantitation of total fatty acid profile
Plasma total fatty acids at baseline and after 24 mo of intervention were analyzed by gas
chromatography coupled to mass spectrometry (GC-MS), using non-fasting plasma samples.
Sample preparation was based on the in situ transesterification protocol as described (1), with
some modifications. A total of 30 μL plasma was mixed with 30 μL hexane, 30 μL internal
standard mix containing heneicosanoic acid (21:0) and glyceryl triheptadecanoat (TG-17:0),

Online Supplemental Material
using silanized glass vials. A volume of 750 μL methanolic-HCl (3N) was added, and the
sample vortexed and incubated in a water bath at 95°C for 2 h. A calibration curve was
constructed using a standard series containing 13 fatty acids in their free forms, processed as
above. After cooling to room temperature, the samples were extracted with 250 μL hexane
containing 2g/L butylated hydroxy toluene. The supernatants containing the fatty acid methyl
esters were transferred to glass vials. An AS3000 autosampler (Thermo Scientific, Hemel
Hempstead, United Kingdom) was used to introduce 1 μL sample, using split injection (1:20)
and an inlet temperature of 220°C, into a Focus GC-DSQ II GC-MS system (Thermo
Scientific, Hemel Hempstead, United Kingdom). The GC was equipped with a BPX70
column (25 m x 0.22 mm, 25 μm film, SGE, Weiterstadt, Germany). Helium was used as
carrier gas at 1.3 mL/min. The column temperature was initially held at 150°C for 2 min,
increased at 4°C/min to 180°C, followed by an increase of 2°C/min to 188°C. The
temperature was finally increased to 220°C at 10°C/min. The total run time was 16.7 min.
The following settings were used: ion source temperature, 200°C; transfer line temperature,
250°C; and electron energy, -70 eV. Detection was achieved using selective ion monitoring
(Supplemental Table 1). 17:0, liberated from TG-17:0, was used as an internal standard. The
ratio between 21:0 and 17:0 was monitored to assess efficiency of the hydrolysis and
derivatization reaction. Xcalibur software pack version 1.4 (Thermo Scientific, Hemel
Hempstead, United Kingdom) was used for data processing.
Correlations of fatty acids with B-vitamin markers and brain atrophy rates
Correlation of omega-3 fatty acids with brain atrophy rates are found in the main results
section of the paper. In addition to the omega-3 fatty acids, arachidonic acid (20:4n-6) and
total fatty acid concentration showed significant negative correlations with brain atrophy
rates (Supplemental Table 2). When analyzed in B-vitamin treated subjects only, palmitic-

Online Supplemental Material
(16:1n-7) and oleic acid (18:1n-9) were added to the list of fatty acids significantly inversely
correlated with atrophy rates (Supplemental Table 2). No fatty acid was significantly
correlated with brain atrophy rate in the placebo group.
Since the concentration of total fatty acids might explain some of the associations for
the individual FA, we repeated the analysis using relative amounts of fatty acids
(Supplemental Table 3). The relative proportions of combined omega-3 (r = −0.29,
P = 0.012) and DHA (r = −0.33, P = 0.005), but not EPA (r = −0.21, P = 0.075), were
significantly inversely associated with brain atrophy rates in the B-vitamin group. None of
the additional fatty acids which showed significant correlations with the brain atrophy rate
when analyzed by absolute concentrations, were significant when analyzed by relative
concentrations, regardless of treatment group (Supplemental Table 3).
References
1. Glaser C, Demmelmair H, Koletzko B. High-throughput analysis of total plasma fatty
acid composition with direct in situ transesterification. PLoS One 2010;5(8):e12045.
doi: 10.1371/journal.pone.0012045.

Online Supplemental Material
Supplemental Table 1. Fatty acids analyzed by GC-MS1
Formula
RT (min)
SIM masses
CV (%)
Lauric acid
12:0
2.0
[74,87]
9.2
Myristic acid
14:0
3.1
[74,87]
4.9
Palmitic acid
16:0
4.8
[74,87]
6.0
Palmitoleic acid
16:1n-7
5.2
[55,74]
6.2
Heptadecanoic acid (IS)
17:0
5.9
[74,87]
NA
Stearic acid
18:0
7.2
[74,87]
6.9
Oleic acid
18:1n-9
7.6
[55,74]
4.7
Linoleic acid
18:2n-6
8.3
[55,67]
4.5
γ-Linolenic acid
18:3n-6
8.8
[67,79]
6.7
α-Linolenic acid
18:3n-3
9.3
[67,79]
4.7
Heneicosanoic acid (IS)
21:0
11.4
[74,87]
NA
Dihomo-γ-linolenic acid
20:3n-6
11.7
[67,79]
4.1
Arachidonic acid
20:4n-6
12.1
[67,79]
4.8
Eicosapentaenoic acid
20:5n-3
13.4
[79,91]
3.9
Docosahexaenoic acid
22:6n-3
16.2
[67,79]
4.7
1 CV, Coefficient of variance; GC-MS, Gas chromatography-mass spectrometry; SIM,
selective ion monitoring; RT retention time.

Online Supplemental Material
Supplemental Table 2. Correlation between yearly brain atrophy rate and absolute
concentrations of fatty acids at baseline1
ALL
PLACEBO
B-VITAMINS
Partial r
P
Partial r
P
Partial r
P
12:0
-0.05
0.517
-0.04
0.761
-0.01
0.916
14:0
-0.07
0.403
0.03
0.789
-0.11
0.363
16:0
-0.12
0.153
0.03
0.825
-0.20
0.085
16:1n-7
-0.13
0.119
-0.01
0.959
-0.26*
0.027
18:0
-0.11
0.189
0.00
0.997
-0.17
0.148
18:1n-9
-0.14
0.092
-0.02
0.988
-0.25*
0.076
18:2n-6
-0.14
0.084
-0.06
0.613
-0.16
0.179
18:3n-3
-0.06
0.472
0.04
0.713
-0.12
0.325
18:3n-6
-0.02
0.776
0.09
0.434
-0.07
0.567
20:3n-6
-0.08
0.317
0.07
0.537
-0.20
0.083
20:4n-6
-0.17*
0.037
0.06
0.641
-0.32*
0.006
20:5n-3
-0.17*
0.037
-0.04
0.738
-0.27*
0.019
22:6n-3
-0.22*
0.005
-0.06
0.592
-0.39*
0.001
Omega-3
-0.21*
0.009
-0.06
0.618
-0.36*
0.002
Total FA
-0.16*
0.042
-0.02
0.884
-0.25*
0.030
1 Partial correlation coefficients adjusted for age, sex, initial brain volume, ApoE status,
education level, baseline diastolic blood pressure, and baseline plasma concentrations of
triglycerides (log), creatinine, and total homocysteine (log). All fatty acid concentrations
were transformed using the natural log prior to analyses. *P ˂ 0.05 (two-tailed). N=168 (All),
N=83 (Placebo), and N=85 (B-vitamins).

Online Supplemental Material
Supplemental Table 3. Correlation between yearly brain atrophy rate and relative
concentrations of fatty acids at baseline1
ALL
PLACEBO
B-VITAMINS
Partial r
P
Partial r
P
Partial r
P
12:0
-0.06
0.945
-0.03
0.787
0.10
0.404
14:0
0.04
0.632
0.05
0.695
0.08
0.527
16:0
0.12
0.156
0.14
0.230
0.08
0.481
16:1n-7
-0.05
0.519
0.00
0.975
-0.17
0.144
18:0
0.07
0.370
0.04
0.762
0.10
0.413
18:1n-9
0.04
0.625
0.03
0.791
-0.03
0.793
18:2n-6
0.00
0.965
-0.08
0.490
0.13
0.258
18:3n-3
0.03
0.736
0.06
0.599
0.02
0.850
18:3n-6
0.07
0.364
0.08
0.333
0.09
0.461
20:3n-6
0.03
0.688
0.12
0.397
-0.03
0.789
20:4n-6
-0.06
0.495
0.08
0.524
-0.18
0.139
20:5n-3
-0.13
0.117
-0.04
0.761
-0.21
0.075
22:6n-3
-0.17*
0.038
-0.06
0.611
-0.33*
0.005
Omega-3
-0.16*
0.051
-0.06
0.638
-0.29*
0.012
1 Partial correlation coefficients adjusted for age, sex, initial brain volume, ApoE status,
education level, baseline diastolic blood pressure, and baseline plasma concentrations of
triglycerides (log), creatinine, and total homocysteine (log). All fatty acid concentrations
were transformed using the natural log prior to analyses. *P ˂ 0.05 (two-tailed). N=168 (All),
N=83 (Placebo), and N=85 (B-vitamins).

Online Supplemental Material
Supplemental Figure 1. Comparison of yearly brain atrophy rates (unadjusted) between
placebo (black) and B-vitamin treatment (gray) (mean + SEM) based on tertiles of baseline
combined omega-3 (left panels), EPA (middle panels), and DHA (right panels) in subjects
with baseline tHcy levels ≥11.3 μmol/L (top) and ˂11.3 μmol/L (bottom). Group sizes are
indicated above each column. *P ˂ 0.05 (two-tailed) between groups by independent t-tests.