Discussion

EPA and DHA have been widely reported for their beneficial effects on cardiovascular and cognitive health⁽ Reference Calder ^{1 –} Reference Caslake, Miles and Kofler ^{4 ,} Reference Calder ^{18 )} but a high level of variation in associations between intake and blood and tissue status has been observed⁽ Reference Browning, Walker and Mander ^{8 )}. The present analysis aimed to identify factors associated with concentrations of EPA, DPA and DHA in major lipid fractions in plasma from individuals consuming their usual diet in order to identify sources of variation in these concentrations. Identification of the contribution that oily fish intake, sex, age, BMI and APOE genotype make to EPA, DPA and DHA status is important for two reasons. First it will highlight the sources of the heterogeneity in the status of these fatty acids, contributing to a better understanding of the use of fatty acid profiles as a measure of dietary intake amongst different population subgroups. Second, it may allow the development of subgroup-specific recommendations for LC n-3 PUFA intake.

The present study reports associations for multiple confounding variables with the relative and absolute concentrations of EPA, DPA and DHA in different plasma lipids. The relative concentration allows investigation of LC n-3 PUFA concentrations in relation to all other fatty acids within the plasma pool (% unit changes), whereas the absolute concentration allows investigation of µg/ml unit changes in LC n-3 PUFA independently of any other fatty acid within the plasma pool. Both ways of expressing the data are useful and informative and both are used in the literature in the field. The absolute concentration of a fatty acid within any plasma lipid fraction will be influenced by the total concentration of that fraction. The absolute concentration of a particular fatty acid may differ between individuals or between subgroups while the relative concentration of that fatty acid may not be different between those individuals or subgroups. Conversely, the relative concentration could be different, but the absolute concentration may not be. Plasma lipids are involved in the transport of fatty acids between tissues where they have different actions depending upon their structure. Hence, the absolute concentration of a fatty acid in a plasma lipid reflects the exposure of tissues to that fatty acid and hence is likely to be a meaningful way of reporting the fatty acid. Conversely, fatty acids often compete with one another for metabolism or for function and hence the relative concentration of each fatty acid (i.e. %) is also likely to be meaningful.

Quantitatively, PC is the main plasma LC n-3 PUFA pool and the present study reports a greater relative concentration of EPA+DHA in plasma PC (online Supplementary Fig. S4) in individuals consuming 2+ portions of oily fish a week compared with those who reported not consuming oily fish, as well as positive associations between EPA, DPA and DHA in other plasma lipid fractions and oily fish intake. Positive associations for oily fish intake and EPA and DHA are reported for plasma phospholipids⁽ Reference Andersen, Solvoll and Drevon ^{19 –} Reference Visioli, Rise and Marangoni ^{21 )} which are confirmed by data from the present analysis which shows 55 % higher EPA+DHA in plasma PC in those consuming two portions of oily fish (each 150 g) per week compared with those reporting no oily fish consumption. Two portions of oily fish supply about 4–5 g of EPA+DHA per week, equivalent to 600–700 mg/d⁽ Reference Calder ^{22 ,} Reference Lee, O’Keefe and Lavie ^{23 )}. Previous studies report comparable increases of 81 % in plasma phospholipid EPA+DHA, and 8·8 and 8·5 µg/ml in total plasma EPA and DHA, respectively, following 16-week consumption of oily fish providing 485 mg EPA+DHA per d⁽ Reference Harris, Pottala and Sands ^{20 )} and 6-week consumption of oily fish providing 927 mg EPA+DHA per d, respectively⁽ Reference Visioli, Rise and Marangoni ^{21 )}. Overall, the findings of the present analysis support existing reports that oily fish intake is associated with, and at a population level is the main determinant of, LC n-3 PUFA in all major blood lipid pools, which may, therefore, be used as biomarkers of oily fish intake⁽ Reference Caslake, Miles and Kofler ^{4 ,} Reference Andersen, Solvoll and Drevon ^{19 ,} Reference Arab ^{24 ,} Reference Harris, Sands and Windsor ^{25 )}. Our analysis does not clearly indicate which plasma lipid fraction would best reflect dietary intake of EPA and DHA, as, in general all four plasma lipid fractions showed dose-dependent increases in EPA and DHA concentration (both absolute and relative) with increasing frequency of oily fish consumption.

There is some evidence that age influences the concentration of EPA and DHA in various plasma fatty acid fractions⁽ Reference Crowe, Skeaff and Green ^{10 )}, which has been attributed in part to the higher habitual fish intake with increasing age. Oily fish intake was controlled for in the present statistical analysis, allowing clearer attribution of any observed associations of age with EPA, DPA and DHA concentrations to altered metabolism and not to dietary differences in intakes of oily fish. Any influence of APOE group distribution was also ruled out as, despite a greater number of individuals aged 50–59 years being included in the present analysis, there was no significant difference in the distribution of APOE2, E3 and E4 genotypes between age groups (data not shown). A 28-d stable isotope tracer study in young (mean age 27 years) v. older (mean age 77 years) adults reported a 1- to 2-fold greater enrichment of ¹³C-DHA in plasma phospholipids and CE in the older age group, suggesting a medium-term age-related difference in DHA homoeostasis associated with accumulation of DHA in the circulation in older people⁽ Reference Plourde, Chouinard-Watkins and Vandal ^{26 )}. The findings of the present analysis support reports of increased plasma DHA with increasing age⁽ Reference Walker, Browning and Mander ^{11 ,} Reference Rees, Miles and Banerjee ^{27 ,} Reference Cunnane, Chouinard-Watkins and Castellano ^{28 )} and we further also report positive associations between age and EPA and DPA, suggesting LC n-3 PUFA accumulate in plasma pools during aging. However, this may in part be due to an increase in circulating cholesterol and CE with age (Table 3). Evidence of positive associations of plasma TC with age dates back to the late 1970s^{( 29 )}, and these have been reported in both males and females⁽ Reference Schaefer, Lamon-Fava and Cohn ^{30 )}. Increased circulating LDL (Table 3) may be reflected in higher absolute total PC and CE concentrations with age (P=0·008 and 0·018, age 20–29 v. 60+ years for PC and CE, respectively, data not shown) and we observed that TC and LDL-cholesterol concentrations were significantly positively correlated with LC n-3 PUFA concentrations in PC (TC, P=<0·001, 0·003, 0·027; LDL-cholesterol P=<0·001, <0·001, 0·003, absolute EPA, DPA and DHA, respectively, data not shown), and that TC, LDL-cholesterol and HDL-cholesterol concentrations were positively correlated with LC n-3 PUFA in CE (TC, P=<0·001, 0·002 absolute EPA and DHA, respectively, LDL, P=<0·001, 0·046, 0·055 absolute EPA, relative DPA and DHA, respectively, HDL-cholesterol, P=0·046 relative DPA, data not shown). These data suggest CE levels may play a significant role in the association of age with LC n-3 PUFA reported in this analysis.

Table 3

Blood cholesterol (mmol/l) concentration according to sex, age, BMI, APOE genotype and oily fish intake (Mean values with their standard errors)

TC, total cholesterol.

* P values obtained from one-way ANOVA model.

† P values obtained from Pearson’s correlation model.

‡ BMI: normal weight=18–25 kg/m², overweight=25·1–30 kg/m² and obese=30·1–46 kg/m².

§ APOE genotype: E2 (E2/E2 and E2/E3), E3 (E3/E3) and E4 (E3/E4 and E4/E4).

|| Oily fish defined as: salmon, herring, mackerel, fresh tuna, sardines, kippers and trout.

Insulin has a role in the regulation of genes involved in whole body lipid homoeostasis including in the removal of lipids from the circulation⁽ Reference Stahl, Evans and Pattel ^{31 )}; in cases of insulin resistance, such removal can be compromised. The occurrence of insulin resistance is reported to rise with increasing age and BMI and despite individuals with diabetes or a fasting glucose concentration >6·5 mmol/l being excluded from the present analysis, differences in fasting glucose were still evident between age and BMI groups (glucose positively correlated with age and BMI; P<0·001 both, data not shown). Thus, insulin resistance may contribute to the higher EPA and DPA concentrations in plasma lipid pools observed with increasing age and BMI.

Increasing body fatness and obesity influence many aspects of fatty acid and lipid metabolism and contribute to disease states such as hypertriglyceridemia, diabetes, and fatty liver disease⁽ Reference Thifault, Cormier and Bouchard-Mercier ^{12 ,} Reference Sorci-Thomas, Babiak and Rudel ^{32 )}; loss of insulin sensitivity with increasing adiposity results in adipose tissue lipolysis and associated higher plasma NEFA concentrations⁽ Reference Sorci-Thomas, Babiak and Rudel ^{32 –} Reference Jensen, Haymond and Rizza ^{34 )}. In the present analysis, there was no correlation between total NEFA concentrations and BMI (data not shown); however, significant, but complex, associations between BMI and LC n-3 PUFA were evident in plasma TAG, with an overall trend towards lower relative concentrations of EPA and DHA with increasing BMI, which is consistent with previous observations⁽ Reference Micallef, Munro and Phang ^{33 ,} Reference Klein-Platat, Drai and Oujaa ^{35 )}. Increased β-oxidation of DHA associated with increased BMI may in part explain lower proportions of LC n-3 PUFA in TAG⁽ Reference Plourde, Chouinard-Watkins and Rioux-Perreault ^{36 )} although altered TAG synthesis and/or selective tissue uptake and partitioning in obesity may also be involved. We observed no association of BMI with absolute plasma concentrations of LC n-3 PUFA and suggest the lower relative concentrations (i.e. %) of EPA and DHA are likely to be offset by increases in total TAG concentrations with increasing BMI.

The proteins encoded by the APOE gene play a major role in the transport and metabolism of lipids via interaction with LDL receptors (LDLR). Two common polymorphisms (rs7412 and rs429358) of the APOE gene in humans result in three protein isoforms, APOE2, E3 and E4. APOE2 and APOE3 are found in the circulation mainly on HDL whereas APOE4 is found preferentially on VLDL with lower concentrations residing on HDL⁽ Reference Gregg, Zech and Schaefer ^{37 )}. The APOE4 allele has been associated with reduced longevity⁽ Reference Soerensen, Dato and Tan ^{38 )} and enhanced the risk of CVD⁽ Reference Bennet, Di Angelantonio and Ye ^{39 )} and Alzheimer’s disease⁽ Reference Bertram, McQueen and Mullin ^{40 )}. Although centrally involved in fatty acid transport and handling in plasma and tissues (and in particular within the brain where APOE is almost the only apo present), the impact of APOE genotype on these processes, and the contribution of dysregulated EPA and DHA metabolism to disease risk is unknown. However, ¹³C-DHA labelling studies provide evidence that DHA metabolism is disturbed in those who are APOE4 carriers⁽ Reference Chouinard-Watkins, Rioux-Perreault and Fortier ^{41 )}.

In the present analysis, APOE4 carriers had significantly higher concentrations of TC and HDLC, and lower concentrations of LDL-cholesterol (Table 3); however, sex×APOE genotype interactions were evident and in male APOE4 carriers we observed significantly higher concentrations of LDL-cholesterol as well as of total CE (data not shown). One advantage of investigating associations in individual plasma lipid classes as opposed to total lipid is that possible effects of APOE and lipoprotein transport and metabolism may be more easily identified. If the associations between APOE and LC n-3 PUFA are seen to occur in lipid pools which are predominantly related to LDL and VLDL particles, they may reflect the dysregulation in lipoprotein handling in people with the E4 allele. However, if the associations between LC n-3 PUFA and APOE genotype are seen to occur across all lipid pools, they may be indicative of alternative mechanisms. Further subgroup analysis indicated higher EPA, DPA and DHA concentrations in CE, EPA and DPA in PC, and EPA in TAG in male APOE4 carriers relative to the non-carrier groups. The higher EPA and DHA may reflect higher overall CE and PC concentrations; however, the lack of association between APOE genotype and fatty acid concentrations in females is suggestive of a sex-specific association independent of CE and PC metabolism.

Interestingly, we have previously reported APOE genotype mediated differences in the response of plasma EPA and DHA to a fish oil supplement given over 8 weeks in males, with lower enrichment in total lipid and phospholipid EPA and DHA in APOE4 carriers relative to the wild-type APOE3/E3 genotype, but only in overweight participants⁽ Reference Chouinard-Watkins, Conway and Minihane ^{14 )}. The aetiology of these associations with LC n-3 PUFA metabolism is currently unknown. As with the association with age, higher plasma LC n-3 PUFA in APOE4 carriers may reflect reduced tissue uptake and DHA accumulating in the circulation. Although lower overall concentrations of APOE were observed in APOE4 carriers (data not shown) no difference in plasma APOE concentrations were evident between sexes, which potentially could have contributed to the differential associations of APOE genotype with EPA, DPA and DHA concentrations. The preferential binding of VLDL by APOE4 and possible associations of APOE genotype with PC and CE synthesis and cellular uptake of EPA and DHA via the LDLR family, LDLR concentrations and specific LC PUFA transporters such as the MFSD2A transporter in the brain⁽ Reference Nguyen, Ma and Shui ^{42 )} may be involved, and are worthy of future investigations. Associations between sex and the activity of these transporters and receptors would also be of interest, along with sex and APOE associations with FADS and ELOVL genes which encode desaturation and elongation enzymes required for the synthesis of LC n-3 PUFA. Differential synthesis of EPA and DHA has been reported between sexes; Pawlosky et al. report greater ability of females to convert ALA to DHA through increased conversion of DPA to DHA compared with males when consuming a beef-based diet. These results were not observed when consuming a fish based diet in which the capacity to convert DPA to DHA was equal between males and females. These findings suggest LC n-3 PUFA metabolism in females may be more sensitive to dietary alterations or may be affected by hormonal regulation⁽ Reference Pawlosky, Hibbeln and Lin ^{43 )}. Indeed there is evidence for up-regulation of the desaturase–elongase pathway via oestrogenic actions resulting in the increased conversion of ALA to EPA⁽ Reference Andersen, Solvoll and Drevon ^{19 ,} Reference Katan, Deslypere and van Birgelen ^{44 ,} Reference Childs, Kew and Finnegan ^{45 )} and DHA⁽ Reference Walker, Browning and Mander ^{11 ,} Reference Giltay, Gooren and Toorians ^{13 ,} Reference Childs, Romeu-Nadal and Burdge ^{46 )} indicating significant effects of female sex hormones on the metabolism of LC n-3 PUFA. Consistent with these observations, there is evidence for an increase in DHA in relation to EPA and DPA at baseline and response to EPA+DHA intake in females compared with males⁽ Reference Metherel, Armstrong and Patterson ^{47 ,} Reference Patterson, Chalil and Aristizabal Henao ^{48 )}. The present analysis further reports lower concentrations of both DPA (−36 % lower absolute concentration in TAG) and EPA (20 % lower absolute concentration in TAG) in females but does not report higher concentrations of DHA in females or find a significant effect of sex on the ratio of DPA: DHA (P>0·50, data not shown). However, these results are also in contrast to other reports describing increased concentrations of EPA and DHA in females⁽ Reference Andersen, Solvoll and Drevon ^{19 ,} Reference Katan, Deslypere and van Birgelen ^{44 ,} Reference Childs, Kew and Finnegan ^{45 )}. These data from the present analysis suggest investigation into associations between sex, APOE, and fatty acid synthesis enzymes and transporters would be of worthwhile to understand further the mechanisms by which these associations occur.

In conclusion, we report concentrations of EPA, DPA and DHA to vary across APOE genotype and that sex is an important factor to consider when evaluating LC n-3 PUFA concentrations in these genotypic subgroups. Our results also confirm that concentrations of EPA, DPA and DHA in plasma pools are suitable population markers of oily fish consumption and show that age and sex are important contributors to the variation in EPA, DPA and DHA concentrations in plasma lipids independent of APOE genotype. These variables should be considered when interpreting LC n-3 PUFA concentrations as a marker of dietary intake and when suggesting dietary LC n-3 PUFA recommendations to ensure benefits are achieved across population subgroups. The investigation into the handling of supplemental EPA and DHA in these subgroups is to be addressed in a further publication and could provide the basis for more detailed advice. However, the aetiology and physiological significance of the interaction between sex and APOE genotype and its association with EPA, DPA and DHA status still requires further investigation.