1 s2.0 S0002916523055235 Main
1 s2.0 S0002916523055235 Main
1 s2.0 S0002916523055235 Main
1134 Am J Clin Nutr 2013;97:1134–43. Printed in USA. Ó 2013 American Society for Nutrition
OMEGA-3 DHA AND COGNITIVE PERFORMANCE 1135
Apolipoprotein E genotype (APOE) is a major genetic risk sules/d. DHA capsules provided 1.16 g DHA/d and 0.17 g EPA/d,
factor for Alzheimer’s disease with carriers of the APOE4 allelic and placebo capsules contained high–oleic acid sunflower oil.
variant (w25% of whites) at several-fold increased risk (21, 22). Fatty acid profiles of treatments are summarized in Table 1. The
Although the evidence is controversial, it is likely that the dosage of DHA was chosen to be physiologically relevant and
APOE4 allele also affects cognitive performance in cognitively achievable through diet (equivalent to w2–3 portions oily fish/wk),
healthy adults (23). Some studies showed poorer performance and the duration of 6 mo was chosen to ensure saturation of the
on cognitive tasks in healthy adult APOE4 carriers (23), whereas tissues with DHA. Erythrocyte DHA levels which have been
other studies showed no difference (24) or even better perfor- shown to correlate with brain tissue levels (36), reach a plateau
mance (25, 26) compared with APOE4 noncarriers. Structural after 6 mo (37). Placebo and treatment capsules were identical in
and functional neurologic changes are seen in APOE4 carriers size and shape. Capsules were provided in identical opaque drug
decades before the appearance of any cognitive or clinical containers that were coded and distributed by staff from Health
symptoms (27–30). Prospective and some intervention studies & Herbs International Ltd according to the randomization
have shown APOE-modulating effects on the relation between scheme. Both research staff and participants were blind as to
LC n23 PUFAs and cognitive function, although the results which participants received DHA or placebo treatments until after
have been controversial (10, 31–34). data analysis. Participants were requested to consume capsules
The primary aim of the study was to investigate whether a high- with a meal and to store capsules in the fridge.
DHA supplement for 6 mo would improve memory (episodic and General demographic information, including ethnicity, level of
working memory), attention, reaction times (RTs) of memory and education, and first language, was obtained by using a structured
attention, and processing speed in young, healthy adults (age questionnaire. Cognitive assessments, fasting blood samples, and
range: 18–45 y) whose habitual diets were low in DHA. A sec- anthropometric measurements were obtained at baseline and after
ondary aim was to investigate whether sex and APOE would 6 mo. Participants were requested not to consume any fatty fish or
modulate the response to the intervention. fish-oil supplements (other than those provided) and to maintain
their normal daily routine (eating pattern, physical activity, and
SUBJECTS AND METHODS
alcohol consumption) for the duration of the study. Participants
kept weekly diaries to record the consumption of DHA or placebo
This dietary intervention (http://www.anzctr.org.au; capsules, seafood (type and amount), and any deviations from the
ACTRN12610000212055) was conducted at Massey Uni- study protocol (eg, illness and use of medication or other nutri-
versity’s Albany campus (New Zealand) between March and tional supplements). The average weekly consumption of seafood
December 2010 according to the guidelines of the Declaration of portions was calculated from diaries and categorized into fatty,
Helsinki. Ethical approval for the trial was obtained from the medium-fat, and low-fat sources (as previously described). Treat-
Massey University Human Ethics Committee (Southern A; ref- ment compliance was determined by using a combination of weekly
erence 10/07), and written informed consent was obtained from diary records, pill-counting of leftover capsules, and analysis of
all participants. erythrocyte LC n23 PUFA levels which is a valid biomarker for
the intake of LC n23 PUFA (37).
Participants At the end of the study, each participant completed a computer-
A total of 228 adults (83 men and 145 women), aged 18–45 y, based tolerance questionnaire adapted from Freeman and Sinha (38).
were recruited in Auckland, New Zealand. Inclusion criteria for
participants were no known major medical condition or disease Blood sample collection and assays
and not taking medication for any condition or disease, non-
EDTA blood was collected. The EDTA buffy coat (white
smoking, low habitual intake of LC n23 PUFAs (less than w200
blood cell–rich layer) was used for the extraction of DNA for
mg EPA + DHA/wk), no consumption of fish-oil supplements
APOE analysis. Erythrocytes were washed 3 times with saline
over the past 6 mo, no allergies to seafood, and not pregnant or
(0.9% NaCl) for fatty acid analysis. Analysis of APOE were
lactating. LC n23 PUFA intake was estimated by asking potential
carried out by Canterbury Health Laboratories, which is fully
participants to record the frequency of habitual consumption of
seafood. Seafood was categorized into fatty (greater than w1 g
n23/100 g), medium-fat (w0.5–1 g n23/100g), and low-fat (less TABLE 1
Fatty acid composition (g/2.25-g daily dosage) of DHA and placebo
than w0.5 g n23/100 g) sources (35), and frequency options
capsules1
included never, 1, 2, or 3 times/mo, and 1, 2, or .23/wk.
Fatty acids DHA capsules Placebo capsules
Study design Palmitic acid (16:0) 0.02 0.09
Stearic acid (18:0) 0.07 0.06
A randomized, placebo-controlled, double-blind study design
Oleic acid (18:1n29) 0.13 1.61
was used. Volunteers who met eligibility criteria were randomly Linoleic acid (18:2n26) 0.02 0.12
assigned to one of 2 groups (ie, the DHA or placebo groups) for Arachidonic acid (20:4n26) 0.05 ,0.01
a period of 6 mo. The random allocation was done by stratified EPA (20:5n23) 0.17 0.02
random assignment on the basis of sex and age. The randomi- DPA2 (22:5n23) 0.06 ,0.01
zation scheme was generated by using the website Randomi- DHA (22:6n23) 1.16 0.02
zation.com (http://www.randomization.com). DHA and placebo 1
Only fatty acids that were detected at $1% of total fatty acids for
capsules were supplied by Efamol Ltd and Health & Herbs In- either active or placebo capsules are shown.
2
ternational Ltd. Treatment was provided as three 750-mg cap- DPA, docosapentaenoic acid.
1136 STONEHOUSE ET AL
accredited with International Accreditation New Zealand to cognitive domains and tasks were assessed: episodic memory
International Organization for Standardization 15189, by using (immediate and delayed word recall, delayed word recognition,
polymerase chain reaction as described by Hixson and Vernier and delayed picture recognition), working memory (n-back,
(39). Erythrocyte fatty acids were analyzed by using a Shimadzu Corsi blocks, and a letter-number sequencing task), attention
gas chromatograph 2010 ported to a gas chromatograph mass [Stroop test, choice reaction time (CRT), and digit vigilance]; and
spectrometer-QT2010 (Shimadzu Corp) as previously described processing speed (finding As task).
(40). Briefly, a 200-mL solution of 83 mmol/L heptadecanoic The accuracy (percentage of correct responses made) for all
acid 17:0 (S . 98%) as an internal standard that contained tests and RTs (in ms) (only for n-back, word and picture rec-
butylated hydroxytoluene (150 mmol/L) dissolved in methanol ognition, Stroop, CRT, and digit vigilance) were assessed. See
was added to 50 mL erythrocytes followed by 1 mL methanolic online supplementary material under “Supplemental data” in the
HCL (3N). Samples were vortexed and incubated for 4 h at online issue for detailed descriptions of tasks.
908C. After cooling to room temperature, fatty acid methyl es- The cognitive testing was conducted under rigorously con-
ters were extracted by adding 2 mL hexane followed by thor- trolled conditions (see online supplementary material under
ough mixing. The hexane phase that contained fatty acid methyl “Supplemental data” in the online issue). In brief, assessments at
esters was recovered after centrifugation, dried under nitrogen, baseline and end were performed at a similar time of day (be-
resuspended with 100 mL hexane, and transferred to a gas tween 0700 and 1000). On the day before tests, participants were
chromatography vial of which 1 mL was injected onto the gas instructed not to consume alcohol, take recreational drugs, or
chromatography–mass spectrometer via split injection (split undertake any unusual sporting activities, to have a good night’s
ratio of 1:10). The capillary column used was an Rtx-2330 sleep, and not be overly stressed. These aspects were confirmed
(30 m 3 0.25 mm; Restek). Mass spectrometer conditions were before participants were allowed to commence assessments.
as follows: ion source temperature, 2358C; interface tempera- Participants arrived at the research unit fasted from food or
ture, 2508C; and ionization voltage, 70 eV. The injector tem- stimulants (caffeine and alcohol), except for water, for $10 h. A
perature was 2508C. Helium gas was used as the carrier gas at standard breakfast was provided before the commencement of
a flow rate of 0.77 mL/min. An initial oven temperature of 708C cognitive tests. Environmental factors such as noise and tem-
was maintained for 3.0 min, allowed to increase to 1558C at perature were controlled to avoid any distraction during tests.
a rate of 258C/min, held for 6.0 min, increased to 1758C at a rate All assessments were carried out on 5 standardized computers,
of 38C/min, held for 3.0 min, increased to 2058C at a rate of 38C/ with the same computer used by individual participants at
min, and finally increased to 2208C at a rate of 88C/min and held baseline and end assessments. Participants were instructed on
for 2.0 min. The total time for each run was 36 min. The total the procedure for the administration of the cognitive battery and
spectrum of erythrocyte fatty acids in the samples were identi- undertook a training session before administration of the full
fied and quantified and are expressed as the weight percentage of battery of tests at baseline and end assessments.
total fatty acids. CVs for EPA and DHA assays were 2.7% and
3.6%, respectively. The CV for the DHA assay was 4%.
Statistical analysis
The sample-size calculation was based on a difference in
Analysis of supplements z score of 0.5 for memory domains and proved a statistical
Oxidation levels of supplements during the 6-mo duration of the power equal to 0.8 and an a level of 0.05 (2 tailed) (G*Power
study were analyzed by measuring the peroxide and anisidine 3.1.2) (47). The minimum sample size required was 32 partic-
values by using the American Oil Chemists’ Society Official ipants per treatment and sex group. To test for sex 3 treatment
Method Cd8-53 with modifications and the American Oil and APOE 3 treatment interactions, a total sample size of 179
Chemists’ Society Official Method Cd18-90, respectively (41). provided 80% power to detect a medium effect size f of 0.25
Oxidation levels were below maximum permitted levels. The fatty (equivalent to a treatment effect of 0.5 SD) at an a level of 0.05
acid content of the capsules were analyzed by using a Shimadzu (G*Power 3.1.2) (47).
GC-17A gas chromatograph equipped with a flame-ionization Descriptive and comparison statistics (independent t test,
detector as previously described (42). 2 tailed) of baseline characteristics were based on all participants
randomly assigned to treatment groups. Baseline characteristics
of dropouts and participants who completed the study were
Cognitive assessment compared, and dropouts did not differ from study completers.
Cognitive function was assessed with the Computerized The primary analysis was carried out on all participants for
Mental Performance Assessment System (Northumbria Univer- whom baseline and end data were available irrespective of the
sity), which has previously been shown to be sensitive to nutri- level of compliance or protocol violations. The analysis was
tional interventions (43, 44). The following 2 additional tasks were carried out on 7 cognitive domains, including sex (men compared
included; the finding As task from the Kit of Factor-Referenced with. women) and APOE (APOE4 carriers compared with
Cognitive Tests (45) and the letter-number sequencing task, which noncarriers) as independent variables (28 tests). Changes to
is a subtest of the Wechsler Adult Intelligence Scale III In- cognitive domains during the treatment period between DHA
telligence test (46). Cognitive tasks used were all standard tasks of and placebo groups were assessed by using ANCOVA models to
cognitive function that have previously been shown to increase adjust for baseline cognitive-function test scores and other co-
activation of the frontal cortex (15), which is the area of the brain variates as follows: education, first language (English compared
associated with the accumulation of DHA, memory, and attention with other), age, and baseline DHA concentrations. Sex and
(4, 5). The battery of tests took w1 h to administer. The following APOE were added to the model as independent variables to test
OMEGA-3 DHA AND COGNITIVE PERFORMANCE 1137
for sex 3 treatment, APOE 3 treatment, and APOE 3 sex 3 RT of attention ¼ðzRT Stroop test
treatment interactions. þ zRT choice RT þ zRT digit vigilance ÞO3 ð6Þ
For all cognitive outcomes, z scores were calculated by
pooling baseline and 6-mo data as previously described (9).
z scores were clustered into cognitive domains as follows:
Processing speed ¼ zfinding As ð7Þ
Memory ¼ðzimmediate word recall þ zdelayed word recall
þ zdelayed word recognition þ zdelayed picture recognition ÞO4 ð1Þ
Statistical analyses were performed with IBM SPSS statistics
software (version 20; IBM Corp).
Working memory ¼ðznback þ zCorsi blocks
þ zletternumber sequencing task ÞO3 ð2Þ RESULTS
The flow of participants through the study is summarized in
Figure 1. Of 228 participants who were randomly assigned to
treatments, 52 subjects were lost to follow-up or discontinued the
Attention ¼ ðzStroop test þ zCRT þ zdigit vigilance ÞO3 ð3Þ intervention for various reasons (n = 30 in the DHA group; n = 22
in the placebo group) (Figure 1). The final analysis was conducted
in 176 participants [n = 85 in the DHA group (33 men and 52
women); n = 91 in the placebo group (33 men and 58 women)]
for whom baseline and end data were available irrespective of the
RT of memory ¼ðzRT delayedword recognition
level of compliance or protocol violations.
þ zRT delayedpicture recognition ÞO2 ð4Þ Baseline characteristics of subjects are summarized in Table
2. Participants were mostly European, had English as their first
language, and were highly educated (most having tertiary
qualifications). DHA and placebo groups did not differ with
RT of working memory ¼ zRT nback ð5Þ regard to baseline characteristics (Table 2) or cognitive tests
TABLE 3
Changes within and between treatments in erythrocyte arachidonic acid, long-chain omega-3 fatty acids, and omega-3 index from baseline to 6 mo1
DHA Placebo DHA compared
(n = 83) (n = 90) with placebo
1
Includes only participants with cognitive-function scores at baseline and 6 mo. The DHA group included 33 men and 52 women. The placebo group included 33 men and 58 women. A decrease in the
reaction time of cognitive tasks indicates an improvement. P values were derived by using ANCOVA (adjusted for baseline cognitive function score, baseline DHA concentrations, first language, age, and
education). Sex and APOE were added as independent variables to test for sex 3 treatment and APOE 3 treatment interactions.
2
Mean 6 SD (all such values).
3
Mean; 95% CI in parentheses (all such values). All values were adjusted for baseline cognitive function z score, baseline DHA concentrations, first language, age, and education.
4
Baseline and 6-mo z scores differed significantly within treatment (P , 0.05; dependent t test, 2 tailed).
5
RT, reaction time.
1139
1140 STONEHOUSE ET AL