2. Health Benefits and Bioavailability

## **Omega-3 Index and Cardiovascular Health**

#### **Clemens von Schacky**

**Abstract:** Recent large trials with eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) in the cardiovascular field did not demonstrate a beneficial effect in terms of reductions of clinical endpoints like total mortality, sudden cardiac arrest or other major adverse cardiac events. Pertinent guidelines do not uniformly recommend EPA + DHA for cardiac patients. In contrast, in epidemiologic findings, higher blood levels of EPA + DHA were consistently associated with a lower risk for the endpoints mentioned. Because of low biological and analytical variability, a standardized analytical procedure, a large database and for other reasons, blood levels of EPA + DHA are frequently assessed in erythrocytes, using the HS-Omega-3 Index® methodology. A low Omega-3 Index fulfills the current criteria for a novel cardiovascular risk factor. Neutral results of intervention trials can be explained by issues of bioavailability and trial design that surfaced after the trials were initiated. In the future, incorporating the Omega-3 Index into trial designs by recruiting participants with a low Omega-3 Index and treating them within a pre-specified target range (e.g., 8%–11%), will make more efficient trials possible and provide clearer answers to the questions asked than previously possible.

Reprinted from *Nutrients*. Cite as: Von Schacky, C. Omega-3 Index and Cardiovascular Health. *Nutrients* **2014**, *6*, 799–814.

#### **1. Introduction**

Fish, marine oils, and their concentrates all serve as sources of the two marine omega-3 fatty acids eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA), as do some products from algae. To demonstrate an effect of EPA + DHA on heart health, a number of randomized, controlled intervention studies with clinical endpoints like overall mortality or a combination of adverse cardiac events were conducted in populations with elevated cardiovascular risk. One early intervention study with oily fish, rich in EPA + DHA, and some early studies with fish oil or fish oil concentrate or even purified EPA at doses ranging between 0.9 and 1.8 g/day indeed demonstrated effects in terms of fewer sudden cardiac deaths, fatal or non-fatal myocardial infarctions, or a combination of adverse cardiac events [1–6]. More recent trials did not demonstrate such effects [7–12]. Recent meta-analyses found no significant benefits on total mortality, cardiovascular mortality, and other adverse cardiac or cardiovascular events [13–18]. This is in contrast to findings in epidemiologic studies, where intake of EPA + DHA had been found to correlate generally with an up to 50% lower incidence of adverse cardiac events [18,19], and in even sharper contrast to epidemiologic studies based on levels of EPA + DHA, demonstrating e.g., a 10-fold lower incidence of sudden cardiac death associated with high levels of the fatty acids, as compared to low levels [20,21]. This seemingly contradictory evidence has led the American Heart Association to recommend "omega-3 fatty acids from fish or fish oil capsules (1 g/day) for cardiovascular disease risk reduction" for secondary prevention, whereas the European Society for Cardiology recommends "Fish at least twice a week, one of which to be oily fish", but no supplements for cardiovascular prevention [22,23]. The more recent guidelines on treating patients with stable ischemic heart disease or patients after a myocardial infarction, targeting similar patient populations, do not recommend EPA + DHA [24,25]. At least in Europe, cardiologists do not routinely use EPA + DHA to reduce cardiovascular risk.

A similar picture emerges for atrial fibrillation: In epidemiologic studies, consumption of EPA + DHA or higher levels of EPA + DHA were associated with lower risk for developing atrial fibrillation, while intervention studies found no effect [26–28]. Pertinent guidelines do not mention EPA + DHA [29]. A similar picture also emerges for severe ventricular rhythm disturbances [20,21,30,31].

Why is it that trial results are at odds with results from epidemiology? What needs to be done to better translate the epidemiologic findings into trial results? The current review will try to shed some light on this issue, with a special consideration of the Omega-3 Index.

#### **2. The Omega-3 Index as a Cardiovascular Risk Factor**

At least some nutritional surveys do not provide valid data [32]. This may explain, why the relation of EPA + DHA in the diet to clinical events has been found to be looser than the relation of levels of EPA + DHA measured in blood to clinical events (e.g., [20,33]). A detailed discussion of the pros and cons of the various fatty acid compartments in which levels of omega-3 fatty acids (whole blood, whole plasma, plasma phospholipids, and others) should be measured is outside the scope of this review and can be found elsewhere [34]. The following points argue for the use of erythrocytes: erythrocyte fatty acid composition has a low biological variability, erythrocyte fat consists almost exclusively of phospholipids, erythrocyte fatty acid composition reflects tissue fatty acid composition, pre-analytical stability, and other points [34–38]. In 2004, EPA + DHA in erythrocyte fatty acids were defined as the Omega-3 Index and suggested as a risk factor for sudden cardiac death [39]. Integral to the definition was a specific and standardized analytical procedure, conforming the quality management routinely implemented in the field of clinical chemistry [39] In fatty acid analysis, methods have a large impact on results: when one sample was sent to five different laboratories offering determination of an Omega-3 Index, results differed by a factor of 3.5 [34]. While results may be internally valid in one laboratory, a difference by a factor of 3.5 makes it impossible to compare results among laboratories. Therefore, the Omega-3 Index was renamed HS-Omega-3 Index®. In contrast, the laboratories adhering to the HS-Omega-3 Index methodology perform regular proficiency testing, as mandated in routine Clinical Chemistry labs [34]. So far, the HS-Omega-3 Index is the only analytical procedure used in several laboratories. A standardized analytical procedure is a prerequisite to generate the data base necessary to transport a laboratory parameter from research into clinical routine. Moreover, standardization of the analytical procedure is the first important criterion for establishing a new biomarker for cardiovascular risk set forth by the American Heart Association and the US Preventive Services Task Force [40,41].

As exemplified by Table 1, the HS-Omega-3 Index has been measured in many populations. Of note, a lower HS-Omega-3 Index was always associated with a poorer clinical condition (Table 1).

**Table 1.** Mean HS-Omega-3 Index values in various populations, Mean (±standard deviation (SD)). Please note that in every population studied, a lower value was found to be associated with a worse condition than a higher value. References are given, if not, unpublished, *n =* number of individuals measured.



**Table 1.** *Cont*.

All levels of fatty acids are determined by the balance of substance entering the body and those leaving the body. Neither a recent meal, even if rich in EPA + DHA, nor severe cardiac events altered the HS-Omega-3 Index [38,58–61]. However, while long-term intake of EPA + DHA, e.g., as assessed with food questionnaires, was the main predictor of the HS-Omega-3 Index, long-term intake explained only 12%–25% of its variability [46,62,63]. A hereditary component of 24% exists [64]. A number of other factors correlated positively (+) or negatively (−), like age (+), body mass index (−), socioeconomic status (+), smoking (−), but no other conventional cardiac risk factors [47,64–71]. More factors determining the level of the HS-Omega-3 Index, especially regarding efflux remain to be defined. Therefore, it is impossible to predict the HS-Omega-3 Index in an individual, as it is impossible to predict the increase in the HS-Omega-3 Index in an individual in response to a given dose of EPA + DHA [42,46,62,63]. In Table 2, current evidence is presented on the relation of the HS-Omega-3 Index to cardiovascular events.

Unselected men (*n* = 262), (SD not reported) 9.58%

This evidence is supported by measurements of EPA + DHA in other fatty acid compartments, as discussed in more detail elsewhere [72,73]. Within the framework of "Heart and Soul" and "Triumph", it was investigated whether determination of the HS-Omega-3 Index added to the information obtained by assessing cardiovascular risk with a conventional scoring system, like the Framingham or GRACE scores for predicting fatal events. The HS-Omega-3 Index provided additional information, as demonstrated by larger areas under the curves in various c-statistics for fatal [74] and non-fatal events [53,75]. Taken together, the HS-Omega-3 Index predicts risk, appears largely independent of conventional risk factors, and adds to the information obtained by conventional risk scoring, thus fulfilling the second criterion for establishing a new biomarker for cardiovascular risk set forth by the American Heart Association and the US Preventive Services Task Force [40,41].


**Table 2.** Summary of epidemiologic studies relating the Omega-3 Index to cardiovascular events.

Moreover, the HS-Omega-3 Index has made it possible to reclassify individuals from intermediate cardiovascular risk into the respective high risk and low risk strata [74,75], the third criterion for establishing a new biomarker for cardiovascular risk [40,41].

Increasing the HS-Omega-3 Index by increased intake of EPA + DHA in randomized controlled trials improved a number of surrogate parameters for cardiovascular risk: heart rate was reduced, heart rate variability was increased, blood pressure was reduced, platelet reactivity was reduced, triglycerides were reduced, large buoyant low-density lipoprotein (LDL)-particles were increased and small dense LDL-particles were reduced, large buoyant high-density lipoproteins (HDL)2 were increased, very low-density lipoprotein (VLDL1) + 2 was reduced, pro-inflammatory cytokines (e.g., tumor necrosis factor alpha, interleukin-1β, interleukins-6,8,10 and monocyte chemoattractant protein-1) were reduced, anti-inflammatory oxylipins were increased [43,81–94]. Importantly, in a two-year randomized double-blind angiographic intervention trial, increased erythrocyte EPA + DHA reduced progression and increased regression of coronary lesions, an intermediate parameter [95]. Taken together, increasing the HS-Omega-3 Index improved surrogate and intermediate parameters for cardiovascular events. A large intervention trial with clinical endpoints based on the HS-Omega-3 Index remains to be conducted. Therefore, the fourth criterion, proof of therapeutic consequence of determining the HS-Omega-3 Index, is only partially fulfilled [40,41].

#### **3. Discussion of Neutral Results of Large Intervention Trials**

Why is it that a low HS-Omega-3 Index can be a cardiovascular risk factor, and yet the results of the large trials testing EPA + DHA on clinical endpoints were neutral?

#### *3*.*1*. *Bioavailability Issues*

According to personal information from the respective first authors, participants of recent large intervention trials were advised to take their supplements, frequently an encapsulated EPA + DHA ethyl-ester with breakfast—in many countries a low-fat meal [7–11]. As discussed in more detail in a recent review, bioavailability of EPA + DHA depends on the chemical form in which they are bound (phospholipids > recombined triglycerides > triglycerides > free fatty acids > ethyl-esters) [96,97], on matrix effects (capsule ingestion with concomitant intake of food, fat content in food) or galenic form (*i*.*e*., microencapsulation, emulsification). The chemical binding form impacts on bioavailability roughly with a factor of two, whereas matrix effects can impact bioavailability up to a factor of 13, and the galenic form up to a factor of 21 [96–99]. When the large trials mentioned here were designed, the bioavailability issues just mentioned were unknown. Thus, involuntarily, the combination used in many of the large trials—An unemulsified ethyl-ester or triglyceride with a low fat meal—guaranteed a very low bioavailability of EPA + DHA.

#### *3*.*2*. *Issues in Trial Design*

In all large intervention trials conducted so far, study participants were recruited based on clinical conditions, but irrespective of their baseline omega-3 fatty acid status [1–12]. In all populations studied so far, the HS-Omega-3 Index had a statistically normal distribution (Table 1). Thus, the proportion of the study population with high levels was not prone to the effects of EPA + DHA, if any. In order to recruit a study population, in which an effect of EPA + DHA can be demonstrated, recruiting study participants with a low HS-Omega-3 Index is a logical choice.

In all large intervention trials conducted so far, study participants were exposed to a trial-specific, but fixed dose of EPA + DHA or placebo [1–12]. The inter-individual variability in response to a fixed dose of EPA + DHA has been found to be large, *i*.*e*., vary up to a factor of 13 [42,61]. This fact alone suggests individualizing the dose given in a trial, in order to reach a predefined target range of the HS-Omega-3 Index, e.g., 8%–11%. The statistically normal distribution of the baseline HS-Omega-3 Index further complicates this problem: A large overlap of omega-3 levels in the EPA + DHA group and placebo or control group can be expected, and has been seen in at least one large trial (Mühlhäusler, B., personal communication) [100]. With levels of omega-3 fatty acids not differing between intervention and placebo or control groups, a difference in study outcome cannot be expected, even if the condition studied would be susceptible to treatment with EPA + DHA. It is worth noting that when a neutral intervention trial was analyzed in a cross-sectional way, EPA + DHA levels directly related to study outcome and less to treatment allocation [101].

Conversely, if a trial reports a positive result, it is likely to have been conducted in a study population with low baseline levels of EPA and DHA, like congestive heart failure: a positive result of a large trial was reported [6], and we found a low mean HS-Omega-3 Index in patients with congestive heart failure (unpublished data, Table 1). A similar case can be made for major depression (Table 1, references [44,50,51,84]).

In the future, recruiting study participants with a low baseline HS-Omega-3 Index and treating them within a predefined target range will allow clearer trial results to be a distinct possibility. Dose adjustments will need to be performed in the placebo group. Since a larger treatment effect can be assumed in the study size estimation, it can be expected that study sizes will be smaller and thus studies less expensive. Clearly, these thoughts are not restricted to trials with patients with cardiovascular risk, atrial fibrillation or ventricular arrhythmia, but can be extended to all areas of omega-3 fatty acid research. This will facilitate scientific progress and lead to a faster recognition of the effects of EPA + DHA.

#### **4. Conclusions**

In an inconsistent manner, EPA and DHA are either recommended or not included in guidelines of cardiac scientific societies. The use of EPA and DHA is not supported by results of recent intervention trials or their meta-analyses. However, epidemiologic data based on assessments of diet and, even more so, data based on levels of EPA + DHA measured in humans, clearly demonstrate that EPA + DHA are associated with a low risk for total mortality, sudden cardiac arrest, and fatal and non-fatal myocardial infarctions. For a number of reasons, like a standardized analytical procedure and a large data base, levels of EPA + DHA are best assessed with the HS-Omega-3 Index. According to current criteria of the American Heart Association and others, the HS-Omega-3 Index is a novel cardiovascular risk factor. Moreover, the HS-Omega-3 Index has led to a fresh look at the field of omega-3 fatty acids and has made it possible to identify issues of bioavailability and study design, explaining at least in part the neutral results of previous intervention trials. In the future, more efficient

intervention studies can be conducted based on the HS-Omega-3 Index, thus providing a clearer picture of the effects of EPA + DHA.

#### **Conflicts of Interest**

CvS operates Omegametrix, a laboratory for fatty acid analyses. Speaker honoraria were received from Reckitt-Benckiser and the Portuguese National Fisheries.

#### **Acknowledgments**

Research grants were provided by several government agencies, AkerBiomarine, Neptune, and Fresenius Kabi. The cost of travel to the AAOCS meeting was provided by Sanofi Consumer Health Care.

#### **References**



## **Relationship between Erythrocyte Omega-3 Content and Obesity Is Gender Dependent**

### **Peter R. C. Howe, Jonathan D. Buckley, Karen J. Murphy, Tahna Pettman, Catherine Milte and Alison M. Coates**

**Abstract:** Epidemiological evidence of an inverse association between consumption of long-chain omega-3 polyunsaturated fatty acids (LC *n*-3 PUFA) and obesity has been conflicting, even though studies in animal models of obesity and limited human trials suggest that LC *n*-3 PUFA consumption may contribute to weight loss. We used baseline data from a convenience sample of 476 adults (291 women, 185 men) participating in clinical trials at our Centre to explore relationships between erythrocyte levels of LC *n*-3 PUFA (a reliable indicator of habitual intake) and measures of adiposity, viz. body mass index (BMI), waist circumference (WC) and body fat (BF) assessed by dual-energy X-ray absorptiometry. Means ± SD of assessments were BMI: 34 ± 7 and 31 ± 5 kg/m<sup>2</sup> ; WC: 105 ± 16 and 110 ± 13 cm; BF: 48 ± 5 and 35% ± 6% in women and men respectively. Erythrocyte levels of eicosapentaenoic acid (EPA) and docosahexaenoic acid (DHA) were similar in men and women while docosapentaenoic acid (DPA) was higher and EPA + DHA (Omega-3 Index) slightly lower in men than in women. Both DHA and EPA + DHA correlated inversely with BMI, WC and BF in women while DPA correlated inversely with BF in men. Quartile distributions and curvilinear regression of the Omega-3 Index *versus* BMI revealed a steep rise of BMI in the lower range of the Omega-3 Index in women, but no association in men. Thus the results highlight important gender differences in relationships of specific LC *n*-3 PUFA in erythrocytes to markers of adiposity. If these reflect causal relationships between LC *n*-3 PUFA consumption and risk of obesity, gender specific targeted interventions should be considered.

Reprinted from *Nutrients*. Cite as: Howe, P.R.C.; Buckley, J.D.; Murphy, K.J.; Pettman, T.; Milte, C.; Coates, A.M. Relationship between Erythrocyte Omega-3 Content and Obesity Is Gender Dependent. *Nutrients* **2014**, *6*, 1850–1860.

#### **1. Introduction**

Polyunsaturated fatty acids (PUFA) are known to beneficially influence fat metabolism and there are numerous studies in animal models of obesity showing that consumption of PUFA, particularly the long-chain omega-3 (LC *n*-3) PUFA from marine sources, can increase fat loss and counteract adiposity [1,2]. This has been supported by a limited number of human trials of LC *n*-3 PUFA supplementation [1–3], although epidemiological evidence of an inverse association between consumption of LC *n*-3 PUFA and obesity has been conflicting [4,5].

In the Health Professionals Follow-Up Study, men with high fish consumption were less likely to be overweight than those with low fish consumption and the proportion of overweight volunteers was inversely related to LC *n*-3 PUFA intake [4]. The Nurses' Health Study, on the other hand, found that higher intakes of fish and LC*n*-3 PUFA were associated with a higher prevalence of obesity [5]. While this unexpected effect of fish intake could be accounted for by higher energy intakes, this was not the

case for LC *n*-3 PUFA intake. However, both these large studies estimated dietary intakes from semi-quantitative food frequency questionnaires which are limited in their ability to accurately assess intakes of different types of fat, particularly LC *n*-3 PUFA.

An alternative approach to assess relationships between LC *n*-3 PUFA consumption and obesity is to evaluate a surrogate biomarker of LC *n*-3 PUFA intake. Several studies have measured LC *n*-3 PUFA as a percentage of total fatty acids in plasma phospholipids with conflicting outcomes. Three early studies conducted in populations of varying ethnicity in Canada found that plasma phospholipid LC *n*-3 PUFA correlated positively with waist circumference [6–8] whereas more recent studies have reported inverse correlations with measures of adiposity [9–11].

Fatty acid levels in plasma phospholipids reflect consumption of dietary fatty acids over a relatively short period (weeks), whereas erythrocyte levels reflect intake over several months [12,13]. Hence the latter is regarded as the most reliable surrogate marker of habitual dietary intake of LC *n*-3 PUFA. Docosahexaenoic acid (DHA), in particular, is incorporated and retained predominantly inside the plasma membrane for the 4 month life of the erythrocyte [12]. Surprisingly, there is little information on relationships between erythrocyte LC *n*-3 PUFA levels and adiposity, although a recent analysis of a cohort of almost 3000 subjects from the Framingham Heart Study indicated a modest inverse relationship between erythrocyte LC *n*-3 PUFA and waist circumference [14].

As we routinely measure erythrocyte fatty acids in nutritional intervention trials, we have chosen to examine relationships between erythrocyte LC *n*-3 PUFA levels and measures of adiposity in baseline data obtained from a convenience sample of trial participants, most of whom had undergone DEXA assessments of body composition. In particular, we have sought to explore potential gender differences in such relationships.

#### **2. Methods**

#### *2.1. Participants and Data*

A secondary analysis was undertaken using de-identified pooled data obtained from volunteers who had participated in nutritional intervention trials conducted by the University of South Australia's Nutritional Physiology Research Centre between 2005 and 2009. Five trials were selected in which measures of weight and adiposity, together with analysis of erythrocyte fatty acid levels, had been undertaken at baseline. Each trial had been approved by the University's Human Research Ethics Committee.

Participants were free-living, non-smoking men and women from both metropolitan and regional locations who were predominantly overweight/obese (inclusion criterion for three of the five trials) but otherwise healthy (*i.e.*, without a diagnosed disease condition) and had limited consumption of fish or fish oil (inclusion criterion for four trials). Baseline anthropometric measurements and blood samples for determination of erythrocyte fatty acid profiles were obtained from 476 participants prior to undergoing dietary interventions. Additionally, dual-energy X-ray absorptiometry (DEXA) assessments of body composition were obtained at the same time from 376 of these participants.

#### *2.2. Assessments*

#### 2.2.1. Anthropometric Measurements

Each participant's height and weight were recorded to calculate body mass index (BMI). Height was measured to the nearest 0·1 cm whilst barefoot using a wall-mounted stadiometer (SECA; Vogel & Halke, Hamburg, Germany). Body weight was measured to the nearest 0.1 kg with participants wearing light clothing using a TANITA Ultimate Scale 2000 (Tanita Corporation, Tokyo, Japan). Waist circumference was measured using a metric tape according ISAK international guidelines [15].

#### 2.2.2. Body Composition

Each participant underwent a whole body DEXA scan (Lunar Prodigy, General Electric, Madison, WI, USA) to determine fat mass and lean mass, from which percentage body fat was estimated.

#### 2.2.3. Assessment of Fatty Acid Profiles

Relative proportions of individual fatty acids in erythrocytes were assessed using a procedure adapted from previously published methods [16]. Erythrocytes were isolated within 2 h of collection by centrifugation, washed in isotonic saline and stored at −80 °C. They were subsequently thawed and the lipids were extracted with chloroform and isopropanol (2:1). The organic phase containing the lipid was evaporated to dryness under a stream of N2 gas. The lipids were then transesterified with acetyl chloride in methanol toluene (4:1, *v*/*v*) at 100 °C for 1 h. The resultant fatty acid methyl esters were extracted with 10% potassium carbonate. Fatty acid methyl esters were separated and quantified using a Shimadzu 2010 gas chromatograph equipped with a 50 m capillary column (0·32 mm, inner diameter) coated with BPX-70 (0·25 mm film thickness; SGE Analytical Science Pty Ltd., Ringwood, VIC, Australia). The injector temperature was set at 250 °C and the detector (flame ionisation) temperature at 260 °C. The initial oven temperature was 130 °C and was programmed to rise to 220 °C at 58 °C/min. H2 was used as the carrier gas at a velocity of 36.4 cm/s. Fatty acid methyl esters were identified based on the retention time to authentic lipid standards (GLC-463; Nu-Chek Prep, Inc., Elysian, MN, USA).

Erythrocyte contents of eicosapentaenoic acid (EPA), docosapentaenoic acid (DPA) and docosahexanoic acid (DHA) were expressed as percentages of total erythrocyte fatty acids. The Omega-3 Index was calculated as the sum of the EPA and DHA contents.

#### *2.3. Statistical Analysis*

Data were analysed using SPSS for Windows (Version 21.0, 2012) and presented as means ± SD (standard deviations). Gender differences in outcome measures were determined by Student's *t*-test; statistical significance was set at *p* < 0.05. Relationships between markers of adiposity and erythrocyte fatty acid contents were assessed by correlation analysis and expressed as Pearson correlation coefficients (*r*). A Bonferroni correction was made for comparisons of each adiposity measure with multiple fatty acids whereby statistical significance was set at *p* < 0.01. Univariate models were used to test for gender interactions. Quartiles of Omega-3 Index were determined for each gender and mean BMI values for each quartile were compared by ANOVA with statistical significance set at *p* < 0.05.

#### **3. Results**

#### *3.1. Participant Characteristics*

Participants were middle-aged and predominantly women. Table 1 presents anthropometric and body compositional assessments for each gender. Due to the selection of overweight/obese adults for the majority of clinical trials, average values of BMI fell within obese classifications for men and women, although there was a wide range (18–59 kg/m<sup>2</sup> ).


**Table 1.** Participant characteristics \*.

\* Data are presented as mean ± standard deviation (number of observations provided in brackets).

#### *3.2. Erythrocyte Fatty Acids*

Table 2 shows mean values of erythrocyte fatty acid levels for each gender. Men had significantly higher erythrocyte DPA while the Omega-3 Index (EPA + DHA) was significantly higher in women.


**Table 2.** Erythrocyte fatty acids (% of total; mean ± SD).

Significant gender difference: \* *p* < 0.05; \*\* *p* < 0.0001.

Table 3 summarises the linear correlation analysis of relationships between erythrocyte fatty acids and measures of adiposity. Pearson correlation coefficients (r) are presented for all participants and for men and women separately. There were strong inverse correlations in the whole dataset between DHA, DPA and the Omega-3 Index and measures of adiposity. DHA and the Omega-3 Index were associated with BMI and waist circumference, whereas DPA predicted body fat. However, the apparent influence of the Omega-3 Index can be attributed to DHA alone as EPA was weakly associated with waist circumference only.


**Table 3.** Correlations between erythrocyte fatty acids and adiposity measures.

Values are Pearson *r*. Significant correlations: \* *p* < 0.01, \*\* *p* < 0.001, \*\*\* *p* < 0.0001.

Univariate analysis confirmed that there were significant gender interactions in the relationships between Omega-3 Index and BMI (*p* = 0.015) and Omega-3 Index and waist circumference (*p* = 0.028) but not between Omega-3 index and % body fat. Hence it was appropriate to split the data by gender, although gender differences in correlations with % body fat should be interpreted with caution. Erythrocyte DPA was the only significant correlate of adiposity in men; higher DPA predicted a lower percentage body fat. However, DPA was not a significant predictor in women. On the other hand, there were strong inverse correlations between all three markers of adiposity and DHA and consequently the Omega-3 Index in women.

#### *3.3. Associations between the Omega-3 Index and BMI*

BMI is the most widely used measure of obesity and the Omega-3 Index is the most widely accepted marker of habitual intake of LC *n*-3 PUFA. Hence it was of interest to further illustrate the relationship between these measures. Figure 1 shows the mean values for BMI in each quartile of the Omega-3 Index for each gender. The lack of a significant relationship in men was apparent. It was also apparent that the relationship between the Omega-3 Index and BMI in women was not linear. BMI values appeared to rise steeply in the lower quartiles of the Omega-3 Index. There were highly significant differences between mean BMI in the lowest quartile and mean BMI in the two highest quartiles of the Omega-3 Index in women.

**Figure 1.** Average BMI values in quartiles of the Omega-3 Index.

Curvilinear analysis confirmed the skewness of the relationship between Omega-3 Index and BMI in women (Figure 2). A sigmoidal curve gave the most significant fit (*r* 2 = 0.078, *p* < 0.001) and suggested a possible threshold for the Omega-3 Index around 6%, below which BMI tends to rise steeply. There was no such relationship for men (*r* 2 = 0.006, *p* = 0.74).

#### **4. Discussion**

The results of this study confirm previous indications of an inverse relationship between LC *n*-3 PUFA levels in erythrocytes and adiposity in humans [14]. Moreover, they extend previous research by revealing a primary role for DHA in this relationship. Most importantly, however, they highlight a striking gender difference, whereby the association of DHA with lower adiposity was evident in women only; men, on the other hand, tended to show an inverse association between erythrocyte DPA and adiposity. It was also apparent that erythrocyte EPA had little relationship with adiposity.

**Figure 2.** Sigmoidal relationship between BMI and the Omega-3 Index in women.

Recognising that erythrocyte levels reflect habitual intakes of LC *n*-3 PUFA, it is tempting to speculate that increased consumption of DHA-rich fish or fish oil may help to counteract obesity. However, the correlations derived from our cross-sectional analysis cannot imply causation. One could equally argue that being obese inclines individuals to include less fish or fish oil in their diet. Alternatively, lower intakes of fish or fish oil and a predisposition to adiposity may both be secondary to another independent factor, e.g., poor quality diet. The influence of independent factors may also account for previous anomalies in reported relationships of LC *n*-3 PUFA intake with adiposity. For example, the observation that larger waist circumferences were associated with higher plasma phospholipid LC *n*-3 PUFA levels in Canadian Inuits and Cree Indians may have been attributable to other aspects of diet in these populations, notwithstanding their habitually high intakes of LC *n*-3 PUFA [6–8].

However, preclinical research in animal models of obesity and limited data from human intervention trials suggests that LC *n*-3 PUFA consumption is causally related to adiposity. LC *n*-3 PUFA can suppress fat synthesis and increase metabolism in adipose tissue via multiple mechanisms involving altered expression of transcription factors, viz. SREBP-1 and PPARs [17]. *In-vitro* studies with lipid droplets specifically implicate DHA in these mechanisms [18]. Coincidentally, DHA was the predominant LC *n*-3 PUFA consumed in a small number of human intervention trials which reported weight loss or fat loss following supplementation [3,19,20]. Hence the highly significant inverse correlations between erythrocyte DHA and diverse measures of adiposity observed in the present study suggest that increasing DHA intake may help to reduce the incidence of adiposity (*r* = 0.353 indicates that erythrocyte DHA levels account for 12.5% of the variance of waist circumference in women).

The limitation of this association to women is noteworthy, particularly considering that there was no significant difference between men and women in the mean erythrocyte DHA level. Interestingly, Decsi and Kennedy [21] reported that plasma phospholipid DHA levels for almost 3000 participants in the EPIC study were approximately 10% higher in women than men, an effect that they attributed to

enhanced conversion of α-linolenic acid through to DHA in women. However, no differences were reported for DPA. The 6% higher erythrocyte DPA level for men in the present study appears to be a unique observation but would be consistent with the hypothesis of limited conversion of DPA to DHA in men relative to women. It is of interest, therefore, that DPA was a significant predictor of body fat in men, whereas DHA was the predominant predictor of all measures of adiposity in women. Garg and colleagues recently reported that DHA supplementation was effective in reducing platelet aggregation in women, whereas EPA supplementation reduced platelet aggregation in men [22]. Clearly there is a need to further characterise gender differences in LC *n*-3 PUFA and their respective functions. A useful starting point would be large epidemiological studies such as EPIC and the Framingham Heart Study, where blood samples have been routinely analysed for LC *n*-3 PUFA contents.

There is increasing recognition of the limitations of dietary intake assessment tools to estimate LC *n*-3 PUFA intake and increasing acknowledgement of the need for reliable blood biomarkers of an individual's LC *n*-3 PUFA status. Unfortunately the use of different biomarkers can lead to different interpretations. Thus the relatively simple measure of LC *n*-3 PUFA in whole plasma is at best a reflection of recent consumption, whereas assessment in a plasma phospholipid fraction reflects both consumption and incorporation of LC *n*-3 PUFA in a stable pool over a period of weeks. However, the "gold standard" biomarker for habitual LC *n*-3 PUFA consumption is their relative content in erythrocytes, reflecting, as stated earlier, their uptake and retention in the erythrocyte pool over several months [12,13]. It is unfortunate that a number of important epidemiological studies have chosen to use plasma phospholipid determinations when there is increasing recognition of the superiority of erythrocyte fatty acid determinations. Indeed the Omega-3 Index, *i.e.*, the sum of EPA and DHA in erythrocytes, has been widely promoted as both a biomarker of LC *n*-3 PUFA consumption/status and a risk factor for cardiovascular disease [23] and serves as a useful standard for comparison across populations. Hence, we quantified relationships between the Omega-3 Index and measures of adiposity in the present study, even though it was evident that erythrocyte DHA alone was a stronger predictor of adiposity than the combination of EPA + DHA.

Whilst there was no apparent relationship between the Omega-3 Index and measures of adiposity in men, examination of quartiles of the Omega-3 Index in women revealed a non-linear relationship with BMI (Figure 1), wherein BMI was similar in the two highest quartiles but rose sharply in the lower quartiles. This was even more evident when curvilinear relationships were tested. The best fit (shown in Figure 2) was a sigmoidal curve, indicating a plateau effect within an approximate range of 5%–9%, below which BMI appeared to increase exponentially. Bearing in mind that the Omega-3 Index predicts greater risk of cardiovascular disease below 4% and lesser risk above 8%, it appears that extremes of the Omega-3 Index may also be associated with other risk factors, including adiposity and depression [24,25].

BMI is a relatively crude measure of obesity; gender differences may reflect differences between men and women in the relative contribution of fat and lean mass to BMI. However, significant curvilinear relationships were found in women between the Omega-3 Index and both % body fat and waist circumference as well as BMI, strengthening the argument that omega-3 intake is inversely related to adiposity in women.

In conclusion, the outcomes of this cross-sectional analysis of erythrocyte LC *n*-3 PUFA content and measures of adiposity in a convenience sample of Australian adults are consistent with other

evidence suggesting an inverse relationship between LC *n*-3 PUFA intakes and obesity. In particular, DHA intake was a negative predictor of BMI, waist circumference and body fat content in women, whereas DPA was a weaker negative predictor of body fat content in men. Analysis of the Omega-3 Index indicates that women in the lower range of the Index may have increased risk of obesity. These data warrant further confirmation in larger studies where potential gender-specific effects of individual LC *n*-3 PUFA are also taken into account.

#### **Acknowledgments**

Nil.

#### **Author Contributions**

PRCH, AMC and JDB initiated and designed this secondary analysis while KJM, TP and CM were largely responsible for data collection in the contributory intervention trials. AMC undertook the data analysis and PRCH drafted the manuscript. All authors contributed to and approved the final version.

#### **Abbreviations**

BMI, body mass index; WC, waist circumference; BF, body fat; LC *n*-3 PUFA, long-chain omega-3 polyunsaturated fatty acids; EPA, eicosapentaenoic acid; DPA, docosapentaenoic acid; DHA, docosahexaenoic acid.

#### **Conflicts of Interest**

The authors declare no conflict of interest.

#### **References**


## **Does Consumption of LC Omega-3 PUFA Enhance Cognitive Performance in Healthy School-Aged Children and throughout Adulthood? Evidence from Clinical Trials**

#### **Welma Stonehouse**

**Abstract:** Long-chain (LC) omega-3 PUFA derived from marine sources may play an important role in cognitive performance throughout all life stages. Docosahexaenoic acid (DHA), the dominant omega-3 in the brain, is a major component of neuronal cell membranes and affects various neurological pathways and processess. Despite its critical role in brain function, human's capacity to synthesize DHA *de novo* is limited and its consumption through the diet is important. However, many individuals do not or rarely consume seafood. The aim of this review is to critically evaluate the current evidence from randomised controlled trials (RCT) in healthy school-aged children, younger and older adults to determine whether consumption of LC omega-3 PUFA improves cognitive performance and to make recommendations for future research. Current evidence suggests that consumption of LC omega-3 PUFA, particularly DHA, may enhance cognitive performance relating to learning, cognitive development, memory and speed of performing cognitive tasks. Those who habitually consume diets low in DHA, children with low literacy ability and malnourished and older adults with age-related cognitive decline and mild cognitive impairment seem to benefit most. However, study design limitations in many RCTs hamper firm conclusions. The measurement of a uniform biomarker, e.g., % DHA in red blood cells, is essential to establish baseline DHA-status, to determine targets for cognitive performance and to facilitate dosage recommendations. It is recommended that future studies be at least 16 weeks in duration, account for potential interaction effects of gender, age and apolipoprotein E genotype, include vegan/vegetarian populations, include measures of speed of cognitive performance and include brain imaging technologies as supportive information on working mechanisms of LC omega-3 PUFA.

Reprinted from *Nutrients*. Cite as: Stonehouse, W. Does Consumption of LC Omega-3 PUFA Enhance Cognitive Performance in Healthy School-Aged Children and throughout Adulthood? Evidence from Clinical Trials. *Nutrients* **2014**, *6*, 2730–2758.

#### **1. Introduction**

Optimal cognitive performance is vital throughout all stages of life. During childhood it is critical to optimize brain development; throughout adulthood it is important to maintain optimal cognitive functioning; and during old-age it is imperative to defer cognitive decline and prevent dementia. The long-chain (LC) omega-3 polyunsaturated fatty acids (PUFA) derived from marine sources, docosaehexaenoic acid (C22:6, DHA) and eicosapentaenoic acid (C20:5, EPA), may play an important role in achieving these objectives. DHA is the dominant LC omega-3 PUFA in the brain [1] and has been shown to accumulate in areas of the brain associated with learning and memory, such as the cerebral cortex and hippocampus [2,3]. DHA is incorporated into neuronal membrane glycerophospholipids at the *sn*-2 position where it regulates numerous neuronal and glial cell processes including neurogenesis,

**36**

neuroplasticity, neurite outgrowth, synaptogenesis and membrane fluidity which in turn supports membrane protein functions impacting on speed of signal transduction and neurotransmission [4–8]. In addition, DHA improves vascular tone which results in increased cerebral blood flow during cognitive tasks [9] and it regulates the transport and uptake of glucose by the endothelial cells of the blood brain barrier [7,10]. Unesterified DHA released from glycerophospholipids by phospholipase A2 are natural ligands for several nuclear receptors that regulate gene expression, and they are precursors for neuroprotectins and resolvins that counteract neuroinflammation, oxidative stress and increases neuronal survival [4,8,10]. EPA and the plant derived omega-3 fatty acid, alpha-linolenic acid (ALA), also crosses the blood brain barrier, but >99% of these fatty acids are β-oxidised and some EPA is de-esterified from glycerophospholipids [11]. Both ALA and EPA may contribute to brain function by facilitating fuel supply to the brain through ketogenesis, particularly during aging [12]. In addition, unesterified EPA may further contribute to cognitive function through the synthesis of eicosanoids that offset neuroinflammation and improve cerebral blood flow due to its antithrombotic and vasodilatory properties [7]. Thus, DHA seems to be more important for brain function, but ALA and EPA also play minor roles.

The brain retains its DHA content as indicated by a long DHA half-life of ~2.5 years in human brain [13], but studies in animal models have shown that long-term DHA depletion results in significant losses in brain DHA [2,14]. Studies in rats which involved dietary LC omega-3 PUFA depletion over successive generations or even in one-generation showed decreased brain DHA levels, particularly in the frontal cortex and hippocampus areas, with reciprocal increased levels of the omega-6 PUFA, docosapentaenoic acid (DPA, C22:5, *n*-6). With the increase in DPA the level of unsaturation was maintained, but brain function was impaired, including changes in learning, memory, auditory and olfactory responses [2,6,14]. These effects were, however, restored by repletion with dietary DHA [14]. Thus, individuals who follow omega-3 PUFA deficient diets particularly over several generations, for example families who never consume seafood, the major source of DHA, may have depleted levels of brain DHA and their cognitive function may not be optimal. Based on studies in rodents and non-human primates the brain has the capacity to nearly meet its daily needs for DHA through the conversion of ALA to DHA, mainly by the liver, when sufficient dietary ALA (>1.2 g ALA/day) is consumed [15]. The capacity of humans to synthesise DHA *de novo* is limited. While DHA is retro-converted to EPA; the extent of conversion of EPA and ALA to DHA is small [1]. The conversion of ALA to DHA is influenced by several factors; a background diet high in linoleic acid (LA, C18:2, *n*-6) reduces the conversion due to substrate competition; the conversion is more efficient in women [1]; and low dietary intakes of DHA increases conversion [15]. Vegans and vegetarians seem to have similar capacity to convert ALA to DHA than omnivores with ALA supplementation increasing EPA to a small extent, with little effect on DHA in blood lipids [16,17]. Consumption of preformed DHA from fish and seafood, supplements (marine or algae derived) or DHA enriched foods may therefore be a more efficient way of ensuring adequate supply of DHA for optimal brain development and function. Large proportions of populations consume inadequate amounts of LC omega-3 PUFA and fish and seafood. The 2008/2009 New Zealand Adult Nutrition Survey reported that ~30% of adults did not or rarely consumed seafood [18]. Only 21% of Australian children consumed fish or seafood during the two-day Australian National Children's Nutrition and Physical Activity survey [19] and Australian adults reported average consumption of 0.25 g/day of total LC

omega-3 PUFA (including EPA, DHA and docosapentaenoic acid (DPA)) [20]. Median usual intakes of total fish and fish high in LC omega-3 PUFA reported by U.S. adults was 12.2 and 1.98 g/day, respectively and median intakes of DHA from foods plus dietary supplements was only 0.04 g/day [21]. Belgian adults reported median DHA intakes ranging from 0.07 to 0.09 g/day [22]. Several studies have shown that vegan diets are devoid of DHA and vegetarian diets that include dairy food and eggs only provide about 0.02 g DHA/day (reviewed by Sanders [16]). These low intakes were accompanied by substantially lower levels of DHA in plasma, serum, red blood cells (RBC) and plasma phospholipids (PL) in vegans and vegetarians compared to omnivores [16]. Although populations following DHA deficient diets do not seem to exhibit apparent cognitive dysfunction, it is imperative to acertain whether increased consumption of DHA by individuals with low dietary intakes, but otherwise healthy will enhance cognitive performance. In summary, basic research provides strong support for the notion that LC omega-3 PUFA, particularly DHA, play an important role in brain function; but will consumption of LC omega-3 PUFA enhance cognitive performance in healthy school-aged children and throughout adulthood, particularly in populations with low dietary intakes? This review will aim to answer this question by critically examining the evidence from all the clinical trials that have been conducted on healthy school-aged children, younger adults and older adults investigating the effects of LC omega-3 PUFA on cognitive performance. Recommendations for future research will also be made.

#### **2. Evidence from Clinical Trials in Healthy Mainstream School-Aged Children**

DHA may be particularly important during periods of brain growth spurts which take place from the last trimester of pregnancy up to 2 years of age. Thereafter, the frontal lobes continue to develop throughout childhood, adolescence and into the late twenties with spurts of frontal lobe development at age 7–9 years and mid-adolescence [23,24]. Table 1 provides a summary of all nutrition intervention trials that have investigated the effects of LC omega-3 PUFA on cognitive function, learning and school achievement in healthy school-aged children. Kuratko *et al*. [25] have also reviewed studies published until November 2012 on DHA and learning and behavior in healthy children. The evidence from clinical trials focusing on healthy mainstream school-aged children is relatively new as evident from the small number of studies (*n* = 10) published since 2007. Most of the studies were conducted in children aged 6–12 years old. The studies varied widely with regard to duration (from 8 weeks to 12 months), dosage (from 0.1 to 1.2 g LC omega-3/day), type of LC *n*-3 PUFA (fish oil, algal oil, enriched foods) and DHA:EPA ratio of the interventions, type of outcomes measured and type of participants. LC omega-3 PUFA was generally provided in the form of fish oil or algal oil (high in DHA) and in a few studies as LC omega-3 PUFA enriched foods. Most studies were conducted in children with low baseline intakes of LC omega-3 PUFA. A landmark study by McNamara *et al*. [26] showed for the first time in humans the direct link between DHA and brain activation. Supplementation of 0.4 g/day and 1.2 g/day of DHA increased activation of the dorsolateral prefrontal cortex during a sustained attention task in boys aged 8–10 years. However, these effects were not translated into improvements in visual sustained attention performance. Eight weeks may be sufficient for DHA to increase brain activation, but longer periods may be needed to result in improved cognitive performance.

Richardson *et al*. [27] showed that DHA supplementation improved reading in children who underperformed in reading. Children with reading scores ≤20th centile gained an additional 0.8 months in reading age while children in the ≤10th centile gained 1.9 months in reading age with DHA supplementation. Parletta *et al*. [28] showed in Australian indigenous children with low literacy ability improvements with EPA + DHA on cognitive development (draw-a-person) with a larger effect in the 7–12 year old children. They were unable to show improvements in academic achievement (reading and spelling). However, these results should be interpreted in context of the myriad of factors in this indigenous population that may have affected the attainment of English literacy, e.g., language experiences, home support, socio-economic status (SES) and school attendance. Omega-3 supplementation alone may not have been sufficient to overcome these factors. The fact that improvements were seen in the draw-a-person variable, a non-verbal, culture-free test of cognitive development that does not require schooling or specific language skills, supports this argument [28]. Studies in malnourished 7–9 year old South African [29] and 8–12 year old Mexican children [30] showed improvements in learning and cognitive performance with LC omega-3 supplementation. Whereas no effects were seen over 12 months in 6–10 year old malnourished children from India [31] and Indonesia [32]. However, dosages were small and the supplement used in the Indian study was mostly ALA [31,32]. Baumgartner *et al*. [33] conducted the first LC omega-3 trial in children who were purposely recruited with iron deficiency; they showed that EPA + DHA supplementation in children with iron deficiency anemia had negative effects on working memory. They also showed treatment gender interactions where boys with iron deficiency performed better in long-term memory and retrieval with DHA while girls performed worse. These studies demonstrate the complexities of conducting omega-3 supplementation trials on cognitive performance and learning in malnourished populations where multiple other factors and nutrient deficiencies may affect the outcomes. Yet, these are the populations most likely to benefit from supplementation.

Two studies conducted in healthy mainstream school children in the UK did not show any benefit of consuming DHA on cognitive performance and learning [34,35]. The study by Kennedy *et al*. [34] was underpowered and short in duration (8 weeks). In Kirby *et al*. [35], cheek cell EPA and DHA increased in both DHA and placebo groups, although the increase was greater in the DHA group. Thus, parents of children in the placebo group may have become more aware of the benefits of omega-3 PUFA and increased their intakes.

Inconsistencies between studies could be ascribed to potential modulating effects of age and gender. Children at different phases of brain and cognitive development and boys and girls may respond differently to LC omega-3 PUFA supplementation as was shown by Parletta *et al*. [28] and Baumgartner *et al*. [33]. In a large sample of 6–16 year old American children, the relationship between dietary omega-3 PUFA and cognitive test scores was twice as strong in girls as in boys [36]. None of the other studies reviewed investigated potential interaction effects of age and gender and some studies used wide age ranges which may have resulted in greater variability or response modulating effects on outcomes.

Biomarkers of LC omega-3 PUFA intake are often not measured in studies on children because of children's fear of having a blood sample taken and consequently not wanting to volunteer for studies. Cheek cell samples are sometimes collected, which is much less invasive and has been shown to correlate well with dietary intakes, plasma and RBC levels [37]. The studies in Table 1 measured DHA and EPA levels/concentrations in RBC, plasma, RBC or plasma phospholipids (PL) and cheek cells. The levels increased in all studies with supplementation of LC omega-3 PUFA and the magnitude of the increase often reflected the supplementation dose [26,29,31–33,35].

To date, no LC omega-3 PUFA interventions have been conducted in adolescents and the only evidence is from observational studies. A prospective study in >9000 15 year old Swedish school children showed significantly higher school grades at age 16 in adolescents consuming fish more than once a week compared to less than once a week [38]. At age 18, male adolescents who consumed fish more than once per week compared to less than once per week at age 15 had higher IQ scores which was obtained from completed intelligence tests as part of the mandatory Swedish military service conscription examination [39]. De Groot *et al*. [40] recently showed in 700 Dutch adolescents, 12–18 years, that consumption of fish providing the recommended amount of EPA + DHA of ~0.45 g/day compared to no fish intake was associated with more advanced vocabulary and higher end term grades. However, eating more fish than the recommended amount was no more beneficial [40].

In summary, it seems as if children with low literacy ability and who are malnourished with low LC omega-3 PUFA intakes may benefit most from the consumption of LC omega-3 PUFA with regard to cognitive outcomes (e.g., memory, non-verbal cognitive development, processing speed, visual-perceptive capacity, attention and executive function) and school achievement (e.g., reading and spelling). Inconsistencies between studies may have been due to different dosages, duration, other nutrient deficiencies and lack of investigating interaction effects of gender and age. Dosages may have been too low in several of the studies that did not show benefits. Nutrient deficiencies such as iron deficiency in malnourished populations may need to be corrected before supplementation with LC omega-3 fatty acids can commence to avoid potential adverse interaction between nutrient deficiencies and omega-3 supplementation.


**40** 


**Table 1.** *Cont.*


phosphatidylethanolamine; RBC, red blood cell; RCT, randomized controlled trial; a Multiple supplement arms, only omega-3 study arm reported.

**Table 1.**

#### **3. Evidence from Clinical Trials in Healthy Younger Adults**

The aim during younger adulthood is to maintain optimal brain function. Although brain development is established, neuroplasticity is ongoing [6]. Only seven studies to date investigated the effects of LC omega-3 PUFA on cognitive performance in younger healthy adults (Table 2). The study by Stonehouse *et al*. [41] in healthy young adults who habitually consumed diets low in DHA, has been one of the largest and longest trials to date and showed that DHA supplementation improved memory and reaction time of memory [41]. This was also the only study so far in healthy young adults that investigated whether gender and apolipoprotein E genotype (*APOE*) modulated the response to LC omega-3 PUFA supplementation. They showed that memory domains were affected differently by DHA in men and women; in women episodic memory improved whereas in men, reaction time of working memory improved compared to placebo [41]. This may be explained by men and women using different problem-solving strategies as indicated by differences in the functional organization of the brain when performing memory tasks [42,43]. *APOE* did not affect responses in the group as a whole but when stratified for gender, improvements in reaction time for working memory and attention with DHA compared to placebo were more pronounced in male *APOE4* allele carriers than in non-carriers. However, this effects needs to be further explored since the study was not statistically powered to investigate the three-way interaction of treatment\*gender\**APOE* [41]. Apolipoprotein E is the primary lipid transporter in brain tissue with carriers of the *APOE4* allelic variant at several fold increased risk of Alzheimer's disease (~three- and ~15-fold increase in risk in *APOE3/E4* and *APOE4/E4*, respectively, relative to the wild-type genotype) [44,45]. Structural and functional neurological changes are seen in *APOE4* carrier's decades before the appearance of any cognitive or clinical symptoms [46–48]. Surprisingly, young (20–35 years) *APOE4* carriers have been shown to perform better on cognitive tasks than non-carriers have [49,50]. This may be due to compensatory mechanisms being employed by carriers of the *APOE4* allele as suggested by increased brain activation in the frontal and temporal regions of *APOE4* carriers during memory tasks compared to non-carriers [47]. The *APOE4* carriers may compensate by taking longer to complete the cognitive tasks more accurately. Any effect of DHA supplementation in *APOE4* carriers is therefore more likely to be seen in tasks assessing reaction time as was seen in our study [41]. Considering the relatively high prevalence of *APOE4* carriers, (~24% in Caucasian populations [44,45] and 31% in the New Zealand sample [41]), it may be an important factor to take into account when investigating the cognitive benefits of LC omega-3 PUFA.

None of the other RCTs summarized in Table 2 showed any cognitive benefits with LC omega-3 PUFA [51–55]. Fontani *et al*. [56] showed improvements in sustained attention and reaction time of sustained attention. However, although the trial is described as a RCT, the authors do not report the placebo results and these results should thus be interpreted with caution. None of these trials examined gender or *APOE* interactions. If gender or *APOE* dimorphisms exist, combining groups may cancel out or dilute any potential effects. Some studies used smaller DHA dosages [51,52], had small sample sizes [51,53,54,56], included a wide age range (18–70 years) [55] and all studies were short in duration ranging from 4 to 12 weeks [51–56].






Abbreviations: DHA, docosahexaenoic acid; EPA, eicosapentaenoic acid; *APOE*, apolipoprotein E genotype; LC, long-chain; NR, not reported; RBC, red blood cell; RCT, randomized controlled trial;

Narendran *et al*.[57] not included in table because the focus was mechanistic and did not use a RCT study design.

**Table 2.**

*Cont*.

Jackson *et al*. [9,53] investigated the effects of short-term (12 weeks) supplementation of LC omega-3 PUFA on neural tissue activation and cerebral blood flow using near-infrared spectroscopy to assess oxy-hemoglobin and deoxy-hemoglobin in the frontal cortex of adults during performance of cognitive tasks. DHA at low and high dosages [9,53], but not EPA [9], significantly increased oxyhemoglobin and total hemoglobin during several cognitive tasks indicating increased cerebral blood flow.

In summary, the trial by Stonehouse *et al*. [41] overcame the study design limitations mentioned above, namely the intervention period was adequate (6 months); a relatively large DHA dosage was used (1.2 g DHA/day) which resulted in achieving RBC DHA levels of ~8%; sufficient statistical power and gender and *APOE* interactions were investigated. They showed that DHA improved memory and reaction time of memory, demonstrating that younger adults may benefit from consumption of DHA [41].

#### **4. Evidence from Clinical Trials in Healthy Older Adults**

The main aim for cognitive function during older age is to defer cognitive decline and to prevent dementia. Age-related cognitive decline (ARCD) is decline in cognitive functioning as a consequence of the aging process that is within normal age limits [58]. Mild cognitive impairment (MCI) represents a transitional state between ARCD and dementia, but individuals with MCI are able to function normally in everyday life [59,60]. Clinical trials investigating the effects of LC omega-3 PUFA on cognitive performance in healthy older adults (without dementia) (Table 3) have been inconsistent, with some showing no effects [61–64] and others showing improvements in different measures of cognitive function, mostly memory [58–60,65–68] as well as executive function [68] and visuospatial learning [58]. The outcomes have been affected by various study design limitations such as high baseline LC omega-3 status, wide variations in cognitive impairment scores with MMSE ranging from 21 to 30, small dosages, short trial duration, insensitive outcome measures, insufficient statistical power, wide age ranges, and lack of investigating potential response modulating effects of age, gender and *APOE*. LC omega-3 PUFA were provided in the form of ethyl esters, algal oil, fish oil, enriched margarine (one study [62]) and krill oil (one study investigating effects on brain activation, not cognitive function [69]).

One of the most rigorously designed trials was unable to show any benefit of LC omega-3 PUFA on a range of cognitive outcomes [61]. The authors argued that the population may have already consumed sufficient LC omega-3 PUFA as evident from relatively high serum DHA:DPA (omega-6) ratios in both treatment groups at 24 months. Unfortunately, the authors did not assess the LC omega-3 status at baseline to confirm this. High intakes of dietary LC omega-3 PUFA of ~0.3 g/day may have also precluded any cognitive benefits with fish oil in the study by Van de Rest *et al*. [64]. In addition, wide ranges of mini-mental scale examination (MMSE) scores from 23 to 30 may have resulted in greater variability in cognitive responses that could have resulted in a null effect. The Alpha Omega trial, which had 2911 patients with stable myocardial infarction, has been the largest and longest (40 months) trial so far conducted. Neither ALA, DHA + EPA, nor a combination of ALA + DHA + EPA affected MMSE scores [62]. The study was designed for CVD as primary outcome and MMSE was used as a secondary measure of global cognitive function. MMSE may not be sensitive for detecting small changes in cognitive function with nutrition interventions in a normal aging population. Furthermore, the effects were investigated against a background where >85% of participants were using lipid lowering and anti-hypertensive drugs which in turn may have affected cognitive function through their effects on cardio-metabolic markers, masking the effects of omega-3 PUFA. The most likely explanation for the lack of cognitive benefits in the study by Stough *et al*. [63] was the low dosage (0.25 g/day DHA from tuna oil) consumed over a short duration of 90 days and a wide age-range of 45–80 years which may have increased the variability in the outcome measures.

Yurko-Mauro *et al*. [58] showed significant improvements in several measures of memory as well as visuospatial learning in older adults with subjective memory complaints and ARCD. The improvement in the paired associate learning (PAL) test was related to a gain of 7 years in age compared to reference data. Their study was sufficient in duration (6 months), provided a large dosage (0.9 g/day DHA), had sufficient statistical power and was conducted in individuals with low habitual intakes of DHA.

Two studies were conducted in older adults with MCI [59,60]. Lee *et al*. [59] showed highly significant improvements in memory in older women with MCI. Their sample size was small (*n* = 35), but the effect size was large with a mean Z-score difference between fish oil and placebo of 0.8 (0.34, 1.26). The differences between this and other studies may be that participants with MCI were recruited, leaving more room for improvement in cognitive test scores, participants were from low socioeconomic background likely to consume low amounts of omega-3 rich fish, and a large dosage (1.3 g/day DHA + 0.45 g/day EPA) was consumed over a long duration (12 months). Furthermore, the study was conducted in women only which may have resulted in a more homogeneous sample. The results are consistent with that of Stonehouse *et al*. [41] who showed improvements in memory in younger women [41]. Sinn *et al*. [60], in their study on older adults with MCI, showed improvements in verbal fluency with a high DHA supplement but not with a high EPA supplement. However, this was the only significant effect out of 11 cognitive assessments and could be due to type 1 error. The lack of effects may have been due to insufficient statistical power. However, the recruitment of large numbers of participants with MCI is not an easy task.

Inconsistencies between studies could be ascribed to response modulating effects of gender, age and *APOE*, but very few studies have investigated these effects. Van de Rest *et al*. [64] identified treatment\**APOE* interactions and treatment\*gender interactions with *APOE4* carriers and men showing improvements in attention compared to placebo.

Improvements have been shown in cognitive performance in older adults over short duration with high dosages [66] which is most likely due to the vascular and antithrombotic effects of EPA and DHA rather than their effects on neurological changes. Nilsson *et al*. [66] showed improvements in working memory with a high dosage of 1.5 g/day EPA + 1.05 g/day DHA over a very short period of 5 weeks. They also showed significant improvements in cardio-metabolic risk markers that were inversely related to performance in working memory. Witte *et al*. [68] also showed an inverse relationship between improvements in executive function and fasting insulin.


**49** 




randomized controlled trial; SBP, systolic blood pressure; TG, triglycerides; TNF-α, tumor necrosis factor alpha; a Multiple supplement arms, only omega-3 study arm reported; Richter *et al*.

table because it was not a RCT.

[70] not included in

**Table 3.** *Cont*.

Supplementation with krill oil (0.19 g EPA + 0.09 g DHA/day) and sardine oil (0.49 g EPA + 0.25 g DHA) in 61–72 year old men for 12 weeks resulted in increased activation of the dorsolateral prefrontal cortex during a working memory task using near-infrared spectroscopy and electroencephalography compared to placebo. Krill oil also increased activity during a calculation task in the left frontal area, the dominant area for calculations [69]. Fish oil (1.32 g EPA + 0.88 g DHA/day) supplementation in 50–75 year olds over 26 weeks improved brain white matter microstructural integrity and grey matter volume in frontal, temporal, parietal and limbic brain areas [68].

In summary, many of the RCT had intrinsic design limitations which hamper drawing firm conclusions regarding the efficacy of LC omega-3 PUFA on cognitive performance in healthy older adults. However, the current evidence suggests that DHA may be of benefit for older adults with ARCD and MCI, particularly for improving memory.

#### **5. Discussion**

Trends are emerging from the current evidence suggesting that consumption of LC omega-3 PUFA, particularly DHA, by healthy school-aged children, and younger and older adults may enhance cognitive performance particularly in those who habitually consume diets low in LC omega-3 PUFA. However, the evidence is inconsistent due to various intrinsic design limitations in many of the RCTs which hamper drawing firm conclusions.

Baseline DHA status may have been an important confounding factor in the available research. Since the human brain tenaciously retains DHA [71], individuals who have been following DHA depleted diets over the long-term are most likely to show cognitive benefits with supplementation whereas individuals with already adequate DHA status may not respond. Although several studies excluded participants based on high intakes of LC omega-3 sources (supplements or seafood), the way that this was assessed and the time periods of intake considered (ranging from 2 weeks to 6 months) differed widely between studies. Some studies included biomarkers of DHA intake to verify low baseline status, but it has mostly been used to confirm compliance to LC omega-3 PUFA interventions. The studies used a wide range of biomarkers and units including RBCs, plasma, plasma PL, serum, plasma cholesterolesters (CE), plasma phosphatidylcholine (PC), RBC PC, RBC phosphatidyletanolamine (PE), cheek cells, expressed as either % of total fatty acids or in concentration units. It is therefore difficult to establish the long-term DHA status of participants and to interpret results across studies.

The use of a uniform biomarker is essential in order to establish baseline DHA status, to determine target levels for optimal cognitive performance as well as threshold levels above which no further benefits are seen. The concept of establishing an omega-3 index for mental health has been suggested by Milte *et al*. [72], based on the omega-3 index for mortality from coronary heart disease developed by Harris and Von Schacky [73]. This index expresses the levels of EPA + DHA in RBC membranes as percentage of total RBC fatty acids and an omega-3 index of ≥8% is associated with the greatest cardio-protection whereas an index of ≤4% is associated with the least protection [73]. Since DHA plays a major role in cognitive performance whereas EPA's role is probably minor, a DHA-index for cognitive performance could be established. A biomarker reflecting long-term intake of DHA may be more appropriate. Plasma DHA reflects recent intakes whereas plasma PL and RBC DHA reflects long-term DHA intakes [1] but RBC DHA has been shown to be more sensitive to long-term intakes than plasma PL [74]. The biomarker also has to correlate well with brain tissue levels. RBC DHA was shown to be the most efficient biomarker for accumulation of DHA in the baboon neonate brain (RBC DHA, *r* = 0.86; plasma DHA, *r* = 0.58) (reviewed by [74]). However, in studies involving children a less invasive biomarker may be more appropriate such as cheek cell DHA. Cheek cell DHA levels have been shown to correlate well with dietary intakes (*r* = 0.65), plasma (*r* = 0.61) and RBC DHA (*r* = 0.58) levels [37]. In a study on piglets cheek cell DHA correlated well with brain DHA levels (*r* = 0.60), but the correlation was not as good as for plasma (*r* = 0.70) and RBC levels (*r* = 0.72) [75]. The analysis of whole blood collected by finger prick and stored on absorbent paper may also provide a non-invasive, rapid, less costly and reliable method for DHA quantification (correlation between RBC DHA and whole blood spot collected by finger prick, *r* = 0.58) [76]. There may therefore be several potential candidate biomarkers, but RBC DHA may be the preferred biomarker because of the established history of the omega-3 index for coronary heart disease [77]. Equations could to be developed to predict a uniform DHA-index level from these different biomarkers. The uniform measurement of a DHA biomarker/index in RCT could facilitate the establishment of target DHA levels at which cognitive performance is optimal which could then guide dietary intake recommendations. We know from kinetic studies that over a period of 6 months, for every 1 g/day DHA consumed, RBC DHA levels increased by 1% [78]. Arterburn *et al*. [1] showed that plasma PL DHA was highly sensitive to dietary intake of DHA up to doses of ~2 g/day after which the DHA levels approached saturation and increased only incrementally. Identification of factors that predict biomarker responses to DHA consumption would be important to estimate dietary requirements for achieving DHA targets. Flock *et al*. [79] identified increased EPA + DHA dose as the strongest predictor of the omega-3 index (% RBC EPA + DHA); lower baseline omega-3 index levels, older age, lower body weight, increased physical activity with increased dose and female sex predicted greater increases in the omega-3 index. The background diet, particularly the omega-6 PUFA content, may also be an important predictor of RBC DHA response [80] that needs to be investigated.

The duration of studies in this review have also been variable ranging from 4 weeks to 2 years. Studies in animal models showed that recovery of brain DHA levels from a state of depletion is a much slower process compared to other tissues. Rats fed an omega-3 repletion diet containing ALA and DHA after being subjected to a low omega-3 PUFA diet through two generations required 8 weeks to reach DHA levels comparable to rats fed omega-3 PUFA adequate diets whereas DHA was almost completely replete in serum and liver after 2 weeks [81]. In rhesus monkeys that were omega-3 PUFA deficient and fed a DHA rich fish oil diet, DHA in phosphatidylethanolamine of the frontal cortex increased after 2 weeks and stabilized after 12 week [82]. The half-life of DHA in the human brain is ~2.5 years [13]. Umhau *et al*. [13] commented that any potential benefit of increasing brain DHA through dietary change may therefore not be fully manifested in clinical trials of only a few weeks and if such rapid improvements occurred it may rather be due to peripheral actions which indirectly affect brain function [13]. The 5-week study by Nilsson *et al*. [66] is an example of this where improvements in working memory correlated with improvements in cardio-metabolic markers. This may also explain why several short term studies failed to show any effects of cognitive function. Several studies of 16 weeks and longer showed improvements in cognitive performance [27–30,41,58–60,68] which is the minimum time needed for RBC DHA to reach a steady state [1,78]. The brain may not be saturated with DHA after 16 weeks of supplementation, but measurable outcomes may become apparent after 16 weeks.

The outcomes that were improved with LC omega-3 PUFA supplementation in children included verbal learning and memory [29], reading [27,29], spelling [29], non-verbal cognitive development [28] and processing speed, visual-perceptive capacity, attention and executive function [30]; in younger adults memory and reaction time of memory were improved [41]; and in older adults several studies showed improvements in memory [27–29,41,58–60,65,68], while executive function [68] and visuospatial learning [58] were also improved. Very few studies assessed the speed of performing cognitive tasks. This represents a fundamental measure of brain function and is equally informative or complementary to information on the accuracy of task performance [83]. Speed of information processing is one of the cognitive abilities in children to develop first and is fundamental to the development and expression of other cognitive abilities such as learning, memory and executive functions [23]. Bearing in mind that DHA improves neural communication through several mechanisms as discussed in the introduction, it is highly likely that DHA may affect speed of cognitive performance. Stonehouse *et al*. [41] showed improvements in reaction time of episodic memory and working memory, but not processing speed; Portillo-Reyes *et al*. [30] showed improvements in processing speed; and McNamara *et al*. [26] showed an inverse relationship between RBC DHA levels and reaction time in a sustained attention task while Muthayya *et al*. [31] could not show any effect on cognitive speediness with a ALA supplement containing a small amount of DHA (0.1 g/day). The significance of any speed change should be interpreted in the context of the function that was assessed [84]. Since memory has been the outcome most often shown to be improved by DHA supplementation, it is likely that DHA may also improve the speed at which memory tasks are performed as was shown by Stonehouse *et al*. [41]. It is suggested that future studies include this outcome in their battery of tests. The use of computerized test batteries allows for the assessment of speed of performing cognitive tasks, but also has the advantage of standardized presentation of cognitive tests, it removes the person-to-person interactions with a researcher that can bias and obfuscate data, and it allows for closely controlled collection of a large amount of data within a short period of time [83]. On the other hand, it may be tempting for researchers to assess multiple cognitive outcomes in the hope to find positive results. However, statistical significant findings from this approach are likely to result from chance alone (type 1 error) [85]. Instead, an approach where a small set of cognitive outcomes are identified and pre-specified (primary outcome) [85] based on current evidence, e.g., memory, and investigated in greater detail will be more valuable in substantiating the effects of LC omega-3 PUFA on cognitive performance than a shot-gun approach.

The increased incorporation of brain imaging technologies in future LC omega-3 PUFA interventions could provide valuable supportive *in vivo* information on the working mechanisms of LC omega-3 PUFA. Brain imaging makers can reliably reflect neurostructural, neurophysiological, neurochemical and functional cerebral changes occurring in response to the intervention. However, these imaging markers cannot be considered as a substitute of clinical endpoints in terms of cognitive or behavioral response to a task or challenge [86].

As discussed above, outcomes may have been confounded by potential response modulating effects of gender, age and *APOE*, but very few studies have investigated these interaction effects. If dimorphisms exist for any of these factors, potential effects may be diluted or cancelled out resulting

in biased conclusions. Future trials of DHA on cognitive function should take these factors into account by either recruiting homogenous samples or by planning gender-, age- or *APOE*-stratification into the study design to ensure stratified randomization and sufficient statistical power.

To date, no studies have been conducted in vegan and vegetarian populations, who have much lower dietary and blood DHA levels compared to omnivores [16], to determine the association between DHA intake and cognitive function. Sarter *et al*. [17] suggest that lifetime DHA insufficiency may put vegans at increased risk for cognitive dysfunction. It is therefore important that future research studies focus on this target population. The availability of vegetarian omega-3 supplements, e.g., algae-sourced DHA, and evidence that supplementation with these preparation result in increased plasma and RBC DHA levels in vegans and vegetarians [16,17,87] makes this possible.

#### **6. Conclusions**

Individuals with low habitual intake of LC omega-3 PUFA, children with low literacy ability and who are malnourished, and older adults with ARCD and MCI may benefit most from consuming LC omega-3 PUFA, particularly DHA. However, the evidence-base is still emerging and RCTs have been inconsistent with many study design limitations. A major challenge ahead is the design and conduct of rigorous RCT to provide the evidence-base for dietary recommendations regarding DHA. It is recommended that future studies include a uniform biomarker, e.g., % DHA in RBC, in order to establish baseline DHA-status, determine targets for improved cognitive performance and to facilitate dosage recommendations. It is also recommended that future studies be at least 16 weeks in duration, account for potential interaction effects of gender, age and apolipoprotein E genotype, include vegan/vegetarian populations, include measures of speed of cognitive performance which could be facilitated by using computerised cognitive test batteries and include brain imaging technologies as supportive information on working mechanisms of LC omega-3 PUFA.

Supplementation with DHA is unlikely to be a "magic bullet" that will create geniuses. However, because of humans' limited capacity to synthesise DHA *de novo* and its critical role in brain function it seems prudent that healthy individuals should include DHA in their diets for optimal cognitive performance through all stages of life. While the evidence is not available yet to make specific recommendations for dietary intake of LC omega-3 PUFA and cognitive performance, we should aim to achieve country specific recommendations of LC omega-3 PUFA. Several international organisations recommend consumption of ≥500 mg/day EPA+DHA or ≥2 fatty fish meals/week [88]. The Australian–New Zealand recommended suggested dietary targets (SDT) for LC omega-3 PUFA is 610 mg/day for men and 430 mg/day for women aged 14 and older [89]. The energy adjusted SDT for 9–13 year old boys and girls are 510 and 410 mg/day and for 4–8 year old boys and girls 400 and 350 mg/day, respectively [90].

#### **Conflicts of Interest**

The author declares no conflict of interest.

#### **References**



## **Randomized Controlled Trial Examining the Effects of Fish Oil and Multivitamin Supplementation on the Incorporation of** *n***-3 and** *n***-6 Fatty Acids into Red Blood Cells**

### **Andrew Pipingas, Robyn Cockerell, Natalie Grima, Andrew Sinclair, Con Stough, Andrew Scholey, Stephen Myers, Kevin Croft, Avni Sali and Matthew P. Pase**

**Abstract:** The present randomized, placebo-controlled, double-blind, parallel-groups clinical trial examined the effects of fish oil and multivitamin supplementation on the incorporation of *n*-3 and *n*-6 fatty acids into red blood cells. Healthy adult humans (*n* = 160) were randomized to receive 6 g of fish oil, 6 g of fish oil plus a multivitamin, 3 g of fish oil plus a multivitamin or a placebo daily for 16 weeks. Treatment with 6 g of fish oil, with or without a daily multivitamin, led to higher eicosapentaenoic acid (EPA) composition at endpoint. Docosahexaenoic acid (DHA) composition was unchanged following treatment. The long chain LC *n*-3 PUFA index was only higher, compared to placebo, in the group receiving the combination of 6 g of fish oil and the multivitamin. Analysis by gender revealed that all treatments increased EPA incorporation in females while, in males, EPA was only significantly increased by the 6 g fish oil multivitamin combination. There was considerable individual variability in the red blood cell incorporation of EPA and DHA at endpoint. Gender contributed to a large proportion of this variability with females generally showing higher LC *n*-3 PUFA composition at endpoint. In conclusion, the incorporation of LC *n*-3 PUFA into red blood cells was influenced by dosage, the concurrent intake of vitamin/minerals and gender.

Reprinted from *Nutrients*. Cite as: Pipingas, A.; Cockerell, R.; Grima, N.; Sinclair, A.; Stough, C.; Scholey, A.; Myers, S.; Croft, K.; Sali, A.; Pase, M.P. Randomized Controlled Trial Examining the Effects of Fish Oil and Multivitamin Supplementation on the Incorporation of *n*-3 and *n*-6 Fatty Acids into Red Blood Cells. *Nutrients* **2014**, *6*, 1956–1970.

#### **1. Introduction**

Two of the most commonly consumed dietary supplements in the Western world are fish oils containing long chain *n*-3 polyunsaturated fatty acids (LC *n*-3 PUFA) and multivitamins [1,2]. The high prevalence of multivitamin use can be attributed to the fact that vitamin deficiencies are common, even in affluent countries [3]. The high prevalence of fish oil use may be in response to recent health messages, made by respected medical authorities such as the American Heart Association, advocating the benefits of increasing dietary LC *n*-3 PUFA intake.

Extensive research has explored the effects of multivitamin and fish oil supplementation in isolation, however, examination into their combined effect on human health remains scarce. Data from the National Health and Nutrition Examination Survey suggests that users of complementary medicine are most likely to use more than one supplement [4] meaning that many people are using both vitamin and fish oil supplements at the same time. There is also some preliminary evidence to suggest that vitamins and fish oils may have synergistic effects. Vitamin and mineral co-factors can influence the biosynthesis of LC *n*-3 PUFA, altering levels of LC *n*-3 PUFA measured *in vivo* [5–7]. In particular, a preclinical study demonstrated that an experimentally induced folic acid deficiency was associated with a fall in LC *n*-3 PUFA levels, suggesting that low levels of antioxidant vitamins may increase lipid peroxidation [7]. Based on their frequency of use and potentially synergistic actions, there is a clear need to understand how multivitamins and fish oils combine to affect potential health outcomes.

The present study investigated the effects of fish oil supplementation, with and without the addition of a multivitamin, on LC *n*-3 PUFA and LC *n*-6 PUFA incorporation measured in red blood cells. Healthy elderly participants (*n* = 160) were randomized into four groups to receive daily: (1) 6 g of fish oil; (2) 6 g of fish oil plus a multivitamin; (3) 3 g of fish oil and a multivitamin; or (4) a placebo in a double-blind, parallel groups design. The primary outcome of this trial was the effect of treatment on cognitive and cardiovascular function, which has been previously published [8]. This paper is concerned with the secondary aim of this trial which was to examine how high and low dosages of fish oil, in combination with a multivitamin, affected the incorporation of LC *n*-3 PUFA into erythrocytes. Specifically, in participants taking fish oil, we predicted increases in both EPA and DHA given that these were provided in balanced proportions in the fish oil supplements; and a dose response effect between consumption of 3 g and 6 g of fish oil. Additionally, we also examined whether combining 6 g of fish oil with a daily multivitamin increased LC *n*-3 PUFA red blood cell incorporation, over and above the effects of fish oil alone.

#### **2. Methods**

#### *2.1. Participants*

The sample consisted of 160 healthy male and female volunteers aged 50 to 70 years. Participants were recruited from the general community and were non-smoking volunteers, not currently taking any medication or vitamin/herbal supplements. Exclusion criteria were; diagnosis of dementia, diabetes, neurologic (*i.e.*, Epilepsy, Parkinson's disease, head trauma) or psychiatric disorders (*i.e.*, depression, schizophrenia), cardiovascular disease (including stroke) or past or present drug or alcohol abuse. Individuals taking anti-coagulant, anti-cholinergic, anti-depressants or acetyl-cholinesterase inhibitors were also excluded. Further exclusion criteria included those currently taking cognitive enhancing supplements regularly and current or long-term multivitamin or fish oil supplementation. The participant flow diagram is shown in Figure 1. The study randomized 160 participants and 144 completed the trial.

**Figure 1.** Participant flow diagram. MV: multivitamin.

#### *2.2. Setting*

The study was conducted at Swinburne University of Technology, Hawthorn, Australia.

#### *2.3. Interventions, Randomization and Blinding*

The trial was randomised, placebo-controlled and double-blind, using a parallel group design. The participants were randomly assigned to one of the following four daily treatments:


Participants consumed their assigned treatment daily for 16 weeks. The clinical trials supplements and matching placebos were provided by Swisse Wellness Pty Ltd. (Melbourne, Australia). The active fish oil supplement was Swisse Ultiboost Wild Salmon Oil and the active multivitamin supplement was Swisse Ultivite 50+ (Mens and Womens formulations). The constituents of the multivitamins are given in the online supplement (Table S1). All participants took one multivitamin (or its corresponding placebo) daily. Participants allocated to receive 6 g of fish oil daily were required to take six active fish oil capsules daily. Participants randomized to receive 3 g of fish oil took three active fish oil capsules and three matching placebo capsules daily. Participants in the placebo group received six placebo fish oil capsules daily. The placebo fish oil contained 1000 mg of Sunola Oil and 50 IU of vitamin E administered in a soft gelatin capsule. Sunola oil is a mono-unsaturated, high oleic (*n*-9) sunflower oil and was chosen as a control given that it is virtually *trans*-fat free and has a similar profile to olive oil. Small sachets with a few drops of fish oil were included in containers to assist with blinding by providing a fish odour when opened. The placebo multivitamin contained carrot powder with a small amount of riboflavin to produce colouration of the urine similar to the active multivitamin. The placebos were identical to the active tablets in shape, size and colour.

Participants were randomly assigned to one of the four experimental groups using a random permuted block procedure with a block size of four. The randomisation was conducted independently by the supplement supplier and the bottles labeled according to the randomization schedule. The research staff were blinded to this allocation. To ensure adequate blinding, placebo and active treatments were packaged in identical blister packs for multivitamins and sealed plastic containers for fish oil capsules. Participants were allocated the next sequential number upon enrolment in the study. Data was unblinded following the analysis of the main study aims.

#### *2.4. Outcomes Measures*

The outcome measures for this study were the incorporation of LC *n*-3 PUFA and LC *n*-6 PUFA fatty acids into red blood cells, following supplementation. Blood sampling was conducted in the morning following a 12-h fasting period. Blood was collected via venepuncture from the antecubital vein, using the BD-vacutainer system. Samples were analysed by Healthscope Functional Pathology according to standard procedures. Samples were centrifuged at 3000 rpm for 10 min before plasma was removed. Red blood cells were then washed twice by suspending in 0.9% saline, centrifuging at 3000 rpm aspirating off the supernatant. Red cells were then stored at −20 °C until assayed. Methyl ethers of fatty acids were prepared as follows: 350 μL of plasma and 1.5 mL of red cell extract were added to a 10 mL extraction tube. 3.8 mL of a methanol/chloroform mixture was added before vortexing the tube for 6 min. 0.8 mL of 0.1 M KCI solution was added and the tube was then vortexed for a further 3 min and then centrifuged at 3000 rpm for 10 min. The upper aqueous layer was discarded by aspiration. A silane treated glass wool was placed in the bottom of a glass Pasteur pipette and then filled with sodium sulphate. The organic layer was passed through sodium sulphate and the eluate was collected in 2 mL vials. The solvent was evaporated to dryness in heating block (<45 °C) with nitrogen. The dry residue was reconstituted with 130 μL of Meth-Prep II methylation agent. Vials were then closed and left to rest at room temperature overnight. 0.8 μL of the esterification mixture was then injected into a Gas Liquid Chromatography using flame ionisation detection (GLC-FID), on a Schimadzu G-2010 (Shumadzu, Kyoto, Japan), for analysis. Chromatic conditions included a detector temperature of 300 °C and injector temperature of 250 °C. Injector sampling time was 0.5 min. All fatty acid values were expressed as a percentage of red blood cell total fatty acids. The LC *n*-3 PUFA index was calculated as total DHA + total EPA + total DPA. Total *n*-3 fatty acid and total *n*-6 fatty acids were calculated as the combination of both total long and short chain *n*-3 (alpha linolenic acid + EPA + DHA + DPA) and *n*-6 (linoleic acid + gamma linolenic acid + eicosadienoic acid + eicosatrienoic acid + arachidonic acid) fatty acids respectively.

#### *2.5. Sample Size*

The sample size of 160 was determined based on the variance of the cognitive and cardiovascular study outcomes, which are reported separately [8]. Percept changes in cognitive performance were expected to be considerably smaller than changes in LC *n*-3 PUFA over the study period. Thus, the study was believed to be appropriately powered to investigate changes in LC *n*-3 PUFA due to treatment.

#### *2.6. Procedure*

Participants were required to attend testing sessions at our laboratories on three separate occasions; at baseline, following six weeks of supplementation and following 16 weeks of supplementation. Blood samples were taken both at baseline and at week 16 only.

The research was conducted in accordance with the guidelines of the Australian National Health and Medical Research Council and the Declaration of Helsinki (as revised in 2004). The study was approved by the Swinburne University Human Research Ethics Committee. Written informed consent was obtained from all subjects. This trial was registered with the Australian and New Zealand Clinical Trial Registry (ACTRN12611000094976).

#### *2.7. Statistical Analyses*

Results were analyzed using SPSS statistics (IBM, version 20, New York, NY, USA). Univariate analyses of variance (ANOVA) were used to examine whether any significant group differences existed at baseline for the basic demographic and health variables displayed in Table 1. Univariate ANCOVAs were also used to examine the effects of treatment on all outcomes variables at week 16. Significant main effects of treatment were further examined using simple planned contrasts, applying Bonferroni corrections to each contrast in order to adjust for comparisons across the treatment groups. Given that males and females may respond differently to LC *n*-3 PUFA supplementation across different clinical outcomes, we examined whether gender predicted the incorporation of LC *n*-3 PUFA and LC *n*-6 PUFA into red blood cells. For these analyses, gender was entered as a fixed factor and the respective fatty acid variable at endpoint as the dependent variable in ANOVA. All analyses were adjusted for the respective scores at baseline and all results were considered statistically significant at *p* < 0.05.

#### **3. Results**

The trial started in 2010 and was ceased in 2012 due to attainment of the desired sample size. No serious adverse events were reported.

#### *3.1. Cohort Demographics*

The descriptive demographics of the study population at baseline are given in Table 1. The mean age of the sample was 59 years. The sample was roughly gender balanced with slightly more females. On average, the sample was well educated and high functioning. Blood pressure levels were normal across the sample although Low Density Lipoprotein (LDL) cholesterol tended to be elevated across all treatment groups. ANOVA revealed that, at baseline, the treatment groups were well matched

across all continuous variables displayed in Table 1, with no significant group differences noted. ANOVA also suggested that males and females were well matched across all fatty acid variables at baseline. Mean baseline fatty acid values can be seen in Table 2, stratified according to treatment allocation. Across the whole sample, median red blood cell composition of *n*-3 fatty acid and *n*-6 fatty acid tended to be lower than those reported in a normative group of almost 160,000 people [9]. In contrast, median baseline values of saturated and monounsaturated fats were higher in the present cohort.


**Table 1.** Sample demographics (means and standard deviations) stratified by treatment allocation.

Note: MMSE: Mini Mental State Examination, BMI: Body Mass Index, LDL: low density lipoprotein cholesterol, HDL: high density lipoprotein cholesterol, SBP: systolic blood pressure, diastolic blood pressure.

#### *3.2. Main Effects of Treatment on n-3 Fatty Acid and n-6 Fatty Acid Blood Measures*

Table 2 displays red blood cell fatty acid composition before and after treatment. Univariate ANCOVA revealed that week 16 EPA (*F*(3, 136) = 12.20, *p* < 0.001), DPA (*F*(3, 136) = 3.09, *p* < 0.05), LC *n*-3 PUFA index (*F*(3, 136) = 3.98, *p* < 0.01), AA/EPA ratio (*F*(3, 136) = 53.74, *p* < 0.001), total *n*-3 fatty acid (*F*(3, 136) = 3.96, *p* < 0.05), total *n*-6 (*F*(3, 136) = 4.01, *p* < 0.01) and the *n*-3/*n*-6 ratio (*F*(3, 136) = 10.13, *p* < 0.001) differed between treatment groups, when controlling for baseline. Week 16 DHA (*F*(3, 136) = 2.01, *p* = 0.10) did not differ according to treatment allocation, when controlling for baseline.


**Table 2.** Means, standard deviations and percentage change for red blood cell fatty acid status over the course of supplementation.

Note: EPA: Eicosapentaenoic Acid, DHA: Docosahexaenoic Acid, AA: Arachidonic Acid, LC *n*-3 PUFA: Long chain *n*-3 polyunsaturated fatty acid, FA: Fatty acid, ANOVA results are for univariate analysis of variance comparing blood values at end-point, by treatment allocation, whilst controlling for baseline values. Stars display results of simple planned contrasts between the placebo and treatment groups. \* *p* < 0.05, \*\* *p* < 0.01, \*\*\* *p* < 0.001.

#### *3.3. Effects of Treatment, Relative to Placebo, on n-3 Fatty Acid and n-6 Fatty Acid Blood Measures*

Those treatment groups differing significantly from placebo can be seen in Table 2. The week 16 AA/EPA ratio was lower across all treatment groups relative to placebo. EPA was significantly higher at week 16 in the two 6 g fish oil groups. Despite a significant main effect, none of the treatment groups had significantly higher DPA at study endpoint, as compared to placebo. Both the LC *n*-3 PUFA index and total *n*-3 fatty acid were only higher in the group receiving the combination of the 6 g fish oil and multivitamin combination. *n*-6 Fatty acid was lower in the 3 g fish oil multivitamin group as well as in the group receiving 6 g of fish oil in isolation. The *n*-3/*n*-6 fatty acid ratio was significantly higher following 6 g of fish oil supplementation, irrespective of the multivitamin.

#### *3.4. Effects of Fish Oil Dosage on n-3 Fatty Acid and n-6 Fatty Acid Blood Measures*

Red blood cell incorporation of *n*-3 fatty acid at week 16 was expected to be higher following supplementation with 6 g as opposed to 3 g of fish oil, demonstrating a dose response. When directly comparing the 6 g fish oil multivitamin group to the 3 g fish oil multivitamin group, the higher dose fish oil group displayed significantly higher week 16 incorporation of EPA (*p* < 0.01), DPA (*p* < 0.05), LC *n*-3 PUFA index (*p* < 0.05), total *n*-3 fatty acid (*p* < 0.05) and the *n*-3/*n*-6 fatty acid ratio (*p* < 0.01). The AA/EPA ratio was also lower in the higher dose fish oil group (*p* < 0.01).

#### *3.5. Effects of Combining Fish Oil with a Multivitamin on n-3 and n-6 Blood Measures*

Adding a multivitamin to the fish oil was expected to increase week 16 *n*-3 fatty acid incorporation into red blood cells, over and above the effects of fish oil alone. When directly comparing the two 6 g fish oils groups, with and without the addition of a multivitamin, there were no significant differences between the two groups across any of the week 16 *n*-3 fatty acid or *n*-6 fatty acid variables. However, when comparing to placebo, the LC *n*-3 PUFA index and total *n*-3 fatty were only increased following the 6 g fish oil multivitamin combination (Table 2) and not the 6g fish oil alone.

#### *3.6. Sources of Variability in Red Blood Cell n-3 Fatty Acid Incorporation*

Figure 2 shows changes in total *n*-3 fatty acid, EPA, DHA and the AA/EPA ratio, over the course of the study, stratified by treatment allocation. Considerable individual variability in *n*-3 fatty acid change is evident. Interestingly, many participants allocated to the fish oil conditions decreased their amount of total *n*-3 fatty acid and DHA as measured from red blood cells over the 16 week study period. However, EPA tended to increase and the AA/EPA ratio decreased in the fish oil treatment arms suggesting that the variability in DHA and total *n*-3 fatty acid may reflect individual differences in incorporation rather than compliance to treatment. The AA/EPA ratio appears to be the best indicator of compliance to treatment as almost all participants receiving active fish oil decreased their ratio, whilst those in the control group tended to remain stable. Across all *n*-3 fatty acid measures, the coefficients of variations using week 16 treatment means and standard deviations (Table 3) tended to be lowest for the 6 g fish oil + multivitamin group and highest for the 3 g fish oil + multivitamin group. The coefficients of variation also tended to be lowest for the AA/EPA ratio.

**Figure 2.** Variability in red blood cell incorporation stratified by treatment allocation for measures of DHA (**A**), EPA (**B**), AA/EPA ratio (**C**) and total *n*-3 fatty acid (**D**). EP: eicosapentaenoic acid, DHA: docosahexaenoic acid, AA: Arachidonic acid, FO: fish oil, MV: multivitamin. Circles and stars represent outliers less than 2 and greater than 2 standard deviations from the mean respectively.

*3.7. Gender and Variability in n-3 Fatty Acid* 

Recent studies suggest that males and females respond differently to LC *n*-3 PUFA supplementation across different clinical outcomes [10,11]. Males and females may therefore differ in their ability to incorporate *n*-3 fatty acids into erythrocytes. The authors investigated whether gender accounted for some of the observed variability in the incorporation of *n*-3 fatty acids into red blood cells. Gender was a significant predictor of week 16 EPA (*F*(3, 135) = 4.54, *p* < 0.05), DHA (*F*(3, 135) = 4.42, *p* < 0.05), LC *n*-3 PUFA index (*F*(3, 135) = 4.53, *p* < 0.05) and total *n*-3 fatty acids (*F*(3, 135) = 5.10, *p* < 0.05). Gender was not predictive of the week 16 AA/EPA ratio (*F*(3, 135) = 0.01, *p* = 0.94) nor DPA (*F*(3, 135) = 2.58, *p* = 0.11). Significant interactions were also found between treatment allocation and gender for EPA (*F*(3, 135) = 3.40, *p* < 0.05), DHA (*F*(3, 135) = 4.99, *p* < 0.01), DPA (*F*(3, 135) = 5.10, *p* < 0.01), LC *n*-3 PUFA index (*F*(3, 135) = 5.37, *p* < 0.01) and total *n*-3 fatty acids (*F*(3, 135) = 4.86, *p* < 0.01). Selected interactions are displayed in Figure 3 (Separate analysis of males and females across all blood measures can be seen in Tables S2 and S3 of the online supplement). It can be seen that females tended to have higher red blood cell incorporation of most *n*-3 fatty acid blood measures at endpoint. The most interesting finding was that, in females, all treatment groups led to an increase in EPA relative to placebo (Figure 3A). In males, EPA only increased following treatment with the combination of 6 g fish oil and a daily multivitamin. Unlike gender, other demographic and clinical factors such as age, height, weight, physical activity, total cholesterol, high sensitivity CRP and general health status did not predict *n*-3 fatty acid incorporation into red blood cells at study endpoint.


**Table 3.** Coefficients of variation for each *n*-3 fatty acid blood measure at week 16, stratified by treatment allocation.

Note: EPA = Eicosapentaenoic Acid, DHA = Docosahexaenoic Acid, AA = Arachidonic Acid, LC *n*-3 PUFA = Long chain *n*-3 polyunsaturated fatty acid. Coefficients of variation calculated as week 16 standard deviation/mean for each measure respectively.

**Figure 3.** Incorporation of EPA (**A**) and DHA (**B**) into red blood cells at study endpoint stratified by treatment allocation and gender. Males are represented by white bars and females by shaded bars. \* Group significantly different to placebo at *p* < 0.05, \*\* Group significantly different to placebo at *p* < 0.01.

#### **4. Discussion**

This study investigated the effects of fish oil supplementation, with and without the addition of a multivitamin, on red blood cell fatty acid composition. Daily supplementation with 6 g of fish oil for 16 weeks led to higher composition of EPA as well as a lower AA/EPA ratio. The LC *n*-3 PUFA index and total *n*-3 fatty acid only increased after 6 g of fish oil was administered in combination with a daily multivitamin. As expected, 6 g of fish oil combined with a multivitamin was more effective at increasing *n*-3 fatty acid erythrocyte composition than 3 g of fish oil combined with a multivitamin. Over the 16 week study period there was considerable individual variability in *n*-3 fatty acid change, much of which was accounted for by gender. A predicted dose response effect in *n*-3 fatty acids was seen between the 3 g and 6 g fish oil supplements.

Fish oils combined with the concomitant use of a multivitamin increased the LC *n*-3 PUFA index. The LC *n*-3 PUFA index was not increased following 6 g of fish oil alone. It thus follows that the intake of vitamins and minerals, from dietary sources, may also affect the efficacy of fish oil supplementation. These results are important because low amounts of LC *n*-3 PUFA are associated with an increased risk of death from coronary heart disease [12]. Combining fish oil supplementation with adequate vitamin/mineral intake, either through diet or supplementation, may help bolster the LC *n*-3 PUFA index thus reducing cardiovascular disease risk. Future fish oil intervention trials are advised to account for habitual intake of vitamins (*i.e.*, through food frequency questionnaires), as this may partly explain individual differences in response to fish oil treatment.

The mechanism by which vitamin/mineral intake interacts with fish oil supplementation to increase the LC *n*-3 PUFA index is not completely understood. Preliminary evidence obtained from animal studies suggests that certain vitamins and minerals, such as B vitamins and iron, influence *in vivo* composition of *n*-3 [5–7]. Although speculative, multivitamin use may boost antioxidant defence, protecting LC *n*-3 PUFA from oxidation.

The present results suggest that considerable variability exists in the individual to uptake and transfer LC *n*-3 PUFA to red blood cells. Gender was identified as one factor contributing to this variability. Females supplemented with fish oil were generally found to have higher incorporation of total *n*-3 fatty acids at the end of the study. No gender differences were found for the AA/EPA ratio suggesting that gender differences are not merely due to compliance to treatment. Instead, these results suggest that males and females differ in their ability to incorporate some specific types of LC *n*-3 PUFA, such as EPA, into red blood cells. In females, all treatments led to a significant increase in EPA over and above the placebo. In contrast, only the 6 g fish oil multivitamin treatment led to an increase in EPA composition in males. These findings are interesting in light of recent studies showing that males and females respond differently to LC *n*-3 PUFA supplementation across clinical outcomes such as platelet aggregation [10] and cognitive performance [11]. If the present findings can be replicated, they may have significant implications for health policy and guidelines because males and females may be required to consume different amounts of fatty fish or fish oil supplements in order to achieve optimal LC *n*-3 PUFA blood composition.

In certain areas of investigation, inconsistencies have been reported regarding the health benefits of fish oil supplementation. For example, randomized controlled trials have produced conflicting results as to whether fish oil supplementation can enhance cognitive performance or mitigate cognitive decline in adults [11,13,14]. These conflicting results are surprising given that observational studies have been far more consistent in suggesting that *n*-3 fatty acid blood composition is associated with cognitive outcomes [15–18]. In light of the present findings, inconsistencies reported in fish oil intervention studies may be partly due to individual differences in the ability to incorporate LC *n*-3 PUFAs into cell membranes. These individual differences may be due to gender or vitamin/mineral intake, either through background diet or concomitant supplement use. Others have also suggested that genetic markers, such as the presence of the APOE e4 allele, may also affect response to fish oil supplementation [19]. To counteract the variability in response to fish oil supplementation, these results highlight the importance of including blood measures of LC *n*-3 PUFA status in future fish oil intervention studies.

The fish oil supplements in the current study contained a balanced ratio of EPA and DHA. Although EPA increased by as much as 96%, DHA red blood cell levels did not significantly increase following fish oil supplementation. These results are consistent with a previous report showing that EPA, as compared to DHA, was better incorporated into erythrocyte membranes following supplementation [20]. In this previous study, EPA increased by 300% while DHA only increased by 42% following 8 weeks of daily supplementation with 1296 mg EPA and 864 mg DHA. Previous studies have also shown that the uptake of DHA into erythrocyte membranes is more variable than that of EPA [20,21].

Limitations of the current study include the relatively small sample size, the relatively short follow-up period and the fact that LC *n*-3 PUFA composition was only measured at baseline and then again following 16 weeks of supplementation. Assessing red blood cell fatty acid incorporation at multiple time points, spread out over the intervention period, would provide a better indicator of *n*-3 fatty acid change across time. The multivitamin formulations used in the present study differed slightly for males and females and this may have inflated some of the observed gender differences. Furthermore, we did not monitor or examine how changes in other dietary factors may have influenced the reported results over the 16 week study period. Lastly, recent studies have shown health benefits of fish oil associated with higher dosages than that used in the present study [11,22] and it is possible that higher dosages would differentially affect *n*-3 fatty blood biomarkers.

#### **5. Conclusions**

Daily supplementation for 16 weeks with 6g of fish oil, with or without a multivitamin, led to higher EPA incorporation into erythrocytes. A dose response effect was demonstrated between 3 g and 6 g of fish oil on *n*-3 fatty acids. Treatment had no effect on DHA composition. At study endpoint, the LC *n*-3 PUFA index was only higher for those receiving a multivitamin in addition to 6 g of daily fish oil, suggesting that some vitamins/minerals aid the incorporation of LC *n*-3 PUFA into red blood cells. There was considerable individual variability in the response to supplementation with females, generally found to incorporate LC *n*-3 PUFA into red blood cells more effectively than males. Relative to placebo, all treatments increased EPA in females whereas only the 6 g fish oil multivitamin combination treatment increased EPA in males. These results suggest that some males may incorporate relatively low amounts of LC *n*-3 PUFA into red blood cells despite adhering to LC *n*-3 PUFA intake guidelines. This is an important area for future research because dietary recommendations around LC *n*-3 PUFA intake may need to be gender specific.

#### **Author Contributions**

Formulating research questions and study design: AP with input from all authors; carrying out research: NG, RC, MP, AP; data analysis: AP, MP, RC; writing manuscript: AP, MP, ASi; Revision of manuscript: all authors.

#### **Conflicts of Interest**

The study was sponsored by Swisse Wellness Pty Ltd. (formerly Swisse Vitamins Pty Ltd., Melbourne, Australia) under contract to Swinburne University of Technology and performed independently by the Centre for Human Psychopharmacology. The National Institute of Integrative Medicine, of which Professor Avni Sali is currently director, receives financial support from Swisse Wellness Pty Ltd. Andrew Pipingas and Avni Sali are currently members of the Scientific Advisory Panel for Swisse Wellness Pty Ltd. Aside from input into the supplements utilized and the broad aims of the study as well as the provision of supplements, Swisse Wellness Pty Ltd. were not involved in any other aspect of the conduct of the trial including analysis, or interpretation of the trial findings.

#### **References**

