Fatty acids in blood lipid fractions

Proportions of n-3 PUFAs in blood lipid fractions as biomarkers of n-3 fatty acid intake We found increases in the proportion of ALA after CSO intake and in the proportion of EPA and DHA after fatty fish intake in all lipid fractions, as expected. These findings are in accordance with previous studies that have found an increase in the ALA content of PL, CE, TG and EM after increased ALA intake either from dietary sources (42,48,49,53,56) or supplements (47) and in the EPA and DHA content of same lipid pools after intake of fish (20,56,71,80,81,83) or fish oil supplements (20,44,47,71). Increases in the EPA and DHA in different lipid pools have mostly been found after intake of only fatty fish or combined intake of fatty and lean fish.

However, increases in DHA have also been found in PL after daily cod intake for 15 days (81) and in leucocytes after weekly cod intake of 750 g for 8 weeks (82). In the present study, lean fish intake did not result in an increase in either EPA or DHA.

Biomarkers of marinen-3 PUFAs have been widely investigated, and EPA and DHA content of erythrocytes and plasma PL, CE, TG have been considered effective biomarkers for EPA and DHA intake (10). Furthermore, EPA and DHA especially in plasma PL have been found sensitive to the supplementation dose. However, also other blood lipid fractions seem to respond well to EPA+DHA intakes as was shown in our study with an average intake of approximately 1.8 g of EPA+DHA per day in the fatty fish group and in previous studies with various doses ranging between less than 0.5 g to almost 5 g per day (20,44,47,51,56,71,80,81,83,85). Similarly, we found that increased ALA intake was equally reflected in all lipid pools. However, Hodson et al. (9) previously concluded that there is no additional benefit to use TG for

76

assessing dietary fatty acid intake due to higher variations in the fatty acid levels than in other lipid pools. The TG fraction may, however, be useful for assessing hepatic enzyme activities (9). We also observed larger variation in TG fraction in the proportions of EPA and DHA, which is in line with the previous fish oil intervention study (65). Furthermore, Browning et al. (65) suggested that plasma CE may not be a preferable biomarker for EPA+DHA intake due to low incorporation of DHA into this lipid pool.

Although several lipid fractions seem to be suitable biomarkers forn-3 PUFAs, every lipid pool has its own characteristics (9,64). However, studies often use only one lipid pool as a biomarker of fatty acid intake, which could cause challenges to the interpretation of the results. Moreover, very little is known about the correlations between n-3 fatty acid compositions of different lipid pools. To elucidate these interrelations and to get an insight of the potential shared pathways of metabolic regulation, we investigated the correlations between proportions of ALA, EPA and DHA in blood lipid fractions.

We found the strongest correlations of the proportions ofn-3 PUFAs between PL and CE (r=0.83–0.89). Previous findings from a meta-analysis and a cohort study suggest that there is a strong correlation ofn-3 PUFAs between PL and CE (r=0.55–

0.80 for ALA, r=0.85–0.94 for EPA, r=0.63–0.93 for DHA) (215,216). In a fish oil intervention study, strong correlations of EPA+DHA were found between PC and CE (r=0.84) and among CE, PC and TG (r > 0.70) (65). In a subset of a cohort study EPA+DHA proportions also correlated strongly among erythrocytes and PL, CE and TG (r=0.83–0.94) (217). However, Browning et al. (65) found slightly weaker correlations for EPA+DHA between erythrocytes and PC, CE and TG after 1-year of fish oil intake (r=0.41–0.51). We found strong correlations of ALA, EPA and DHA between EM and CE and PL (r=0.60–0.80) whereas the correlations of these n-3 PUFAs between EM and TG were slightly weaker (r=0.42–0.56). Taken together, these results suggest that then-3 PUFA compositions of different blood lipid fractions are strongly interrelated and that the metabolic regulation of the fatty acid composition of these lipid pools may occur via shared pathways.

Dietary intake ofn-3 PUFAs have been considered to correlate well with the their respective levels in biomarkers due to their largely exogenous synthesis (9).

However, in cross-sectional studies the correlations between dietary intake ofn-3 fatty acids and levels in lipid pools have varied considerably:r=0.34–0.62 for ALA, r=0.15–0.66 for EPA andr=0.18–0.66 for DHA (9,88). In cohort studies, correlations between dietary intake and circulatingn-3 fatty acids have been moderate for EPA (r=0.34–0.40) and DHA (r=0.40–0.50) and very weak (r=0.05) or no correlation for ALA (216,218).

The variation in the correlations for EPA and DHA may be related to use of different assessment methods (i.e. food frequency questionnaire, food record and 24 h recall) and the large variation in the fish consumption among different populations (25). Furthermore, the inclusion or exclusion of the use of fish oil supplements in the assessment methods affects these correlations. Weak correlations for ALA could be

77 due to inaccuracies in assessment of dietary intake or the rapid oxidation of ingested ALA (14). However, a stronger correlation (r=0.64) was found in an intervention study between the intake of ALA and the respective proportion in erythrocytes (54).

We found weak baseline correlations of dietary ALA and EPA with corresponding ALA and EPA proportions in all fractions, whereas dietary DHA was found to correlate with the proportion of DHA only in TG. However, correlations at the end of the study and correlations between changes in n-3 PUFAs and their respective proportions in lipid pools were stronger. These differences could be explained by the number of days in the collected food records: baseline food records were from only 4 days and during the intervention from 12 days. Intakes of EPA and DHA may have been underestimated with the 4-day food record at baseline since it may have not accurately captured the intake of fish or seafood which are not typically consumed every day. Furthermore, it has been estimated that at least 7 days of food record data is required to accurately assess PUFA intake (219).

We found that habitual fish consumption correlated with the proportion of EPA in all blood lipid pools except in CE, with the proportion of DHA in all blood lipid pools and with the serum concentration of DHA at baseline. Previously, intake of fish, especially oily fish, has been found to be associated with EPA and DHA in erythrocytes (220). However, also in a population consuming mainly lean fish, significant associations of fish consumption with EPA and DHA in serum and in LDL CE and LDL PL were observed (221).

In previous studies, only a weak or no correlation of DPA in total plasma, plasma PL and erythrocytes with fish intake has been observed (33,222). Similarly, we found no correlation between fish intake and DPA in PL, TG or EM at baseline or at the end of the study. However, in previous studies DPA and EPA in lipid pools have been found to correlate, which suggests that circulating DPA could primarily be derived from endogenous elongation from EPA (33,222). In line with these findings, we also found positive correlations between EPA and DPA in PL, EM and TG at baseline (r=0.49–0.82), at the end of the study (r=0.47–0.80) and between changes in the proportions of EPA and DPA (r=0.55–0.82).

Despite the similar responses in different lipid fractions ton-3 PUFA intake, these fatty acids have been found to have different kinetic profiles (47,65,69,85). EPA has been found to incorporate faster into plasma lipid fractions and erythrocytes and to clear more rapidly from them than DHA. Mechanisms behind these differences are not well-understood, but it has been suggested that there could be structural or functional preference for DHA over EPA, and that DHA levels in tissues could be more strongly regulated (47,223). Furthermore, DHA preferentially incorporates into inner leaflet of phospholipid bilayer and EPA into the outer leaflet in erythrocyte membranes, which makes the EPA pool more labile (224). These differences should be considered especially in intervention studies in which compliance is not monitored with changes in these fatty acids separately (64). Rapid changes in EPA in the lipid pools could potentially mask non-compliance.

78

Although EPA and DHA seem to differ in their incorporation to different lipid pools, less is known about the kinetic profile of ALA. Duration of the intervention studies related to ALA intake using blood lipid fractions as biomarkers have varied between 4 and 24 weeks (42,44,45,47–49,51,53–57). Furthermore, the daily dose of ALA that has been found to increase the ALA content in lipid pools has ranged from 2.4 g to 14 g in these studies. However, a high ALA intake of ~13 g per day has been found to increase the ALA content in erythrocytes after only a week (225). Similar and relatively rapid changes in ALA content of erythrocytes have been found also with lower daily dosages of ALA (2.4 g and 3.5 g) after 4 weeks (47,51). ALA content of erythrocytes has also been found to drop rapidly after intervention period (47).

Furthermore, Skeaff et al. (226) found that saturated andn-6 fatty acids incorporated into erythrocyte PC and plasma PL and CE in a similar manner within 2 weeks. These results together suggest that also ALA could incorporate rapidly into different lipid pools with varying intakes. In the present study, the fatty acid composition of lipid fractions were measured only at baseline and at the end of the study. Therefore, conclusions about the rate of incorporation of different fatty acids into lipid pools cannot be made.

Along with EPA and DHA, another long-chainn-3 fatty acid found in the fish is DPA (32). However, only a few fish intervention studies have reported the changes of DPA in lipid pools, and these findings have been inconsistent (20,83,84).

Furthermore, supplementation studies have reported mixed results regarding the changes in DPA in lipid pools (20,47,51,85). These discrepancies in study findings are probably due to varying amounts of DPA in diet and supplements. Furthermore, metabolic changes in the n-3 fatty acid pathway may have an influence, since increases in the DPA content may also occur via elongation from EPA or via retroconversion from DHA (21). Furthermore, supplementation with purified EPA was previously found to increase DPA in plasma and platelet PL, whereas purified DHA resulted in a decrease in DPA in platelets (227). Together these results indicate that the metabolic relationship of DPA with EPA and DHA is complex and remains poorly understood. In the present study, we found an increase in DPA only in TG in the fatty fish group. However, we have no information on the amount of DPA in the diet or on the kinetics of DPA incorporation during the study. Nonetheless, the incorporation of DPA into different lipid fractions seems to differ, at least in the short-term. In a 1-week intervention study, purified DPA (2 g/day) was found to increase the proportion of DPA in TG and PL but not in CE or erythrocytes (34).

Furthermore, increased DPA intake also was found to increase EPA and DHA in the TG fraction. This finding suggests that circulating DPA could provide a source for both EPA and DHA.

In summary, there appears to be a range of useful biomarkers for ALA, EPA and DHA intake in humans. Taken together, these results suggest thatn-3 PUFAs in CE, PL, TG and EM can be used interchangeably as biomarkers ofn-3 fatty acid intake with various doses. However, EPA in the blood lipid fractions appears to serve as a more dynamic pool that responds rapidly to its dietary intake, whereas changes in

79 DHA in these lipid pools tend to be more subtle. Furthermore, circulating DPA may have an important metabolic role as a reservoir for EPA and DHA but appears to be a weak biomarker for fish intake. Several factors also need to be considered when choosing the most appropriate biomarker forn-3 PUFA intake to be used in studies, such as the fatty acid status of the study subjects at baseline, clearance of fatty acids in case of a washout period, duration of the study and appropriate sample type.

Furthermore, intra- ja inter-individual variability in the uptake of fatty acids to different lipid pools should be taken into consideration (73). Of note, most of the biomarker studies have been conducted with European or North American populations, and these results need to be substantiated across other populations.

Conversion of ALA to long-chain n-3 PUFAs

ALA is a substrate for the synthesis of long-chainn-3 PUFAs (14). Increased intake of ALA with varying doses has been found to result in increased EPA content in erythrocytes and plasma PL, CE, TG in several studies (42,47–49,51,55,57). In the present study, however, a daily ALA intake of over 10 g from CSO and diet did not significantly increase the proportion of EPA in any blood lipid fraction. Although this finding is in contrast with the majority of previous findings, there are also studies that have observed no effect of increased ALA intake on the EPA content of lipid pools (52–54,56,228). In addition to the lack of effect on the proportion of EPA, we observed no change or a decrease in the proportion of DHA after increased ALA intake. This finding is supported by previous studies (42,45,47–49,51,53–57).

The lack of conversion of ALA to EPA and DHA in our and previous studies could be explained by several factors. First, increased availability of conversion products (EPA+DHA) may have downregulated the conversion. As previously reported by Egert et al. (54), we also observed high baseline n-3 PUFA status of subjects as indicated by the omega-3 index. Another explanation for the findings of Egert et al.

(54) could be that they did not include a run-in period and fish consumption was not allowed during the study. Therefore, it is possible that due to limited fish intake the decreases in the EPA content of erythrocytes during the study may have counteracted the increases of EPA from ALA conversion. Secondly, ALA conversion is affected by an abundance ofn-6 PUFAs, especially linoleic acid, in the diet (16).

Previous studies have shown that the most effective way to enhance the conversion is to increase ALA intake and concomitantly decrease linoleic acid intake (16). In addition to ALA, CSO contains a considerable amount of linoleic acid (16%), which was also reflected in the present study as an increased proportion of linoleic acid within the CSO group in all lipid fractions except TG. This abundance of linoleic acid in our and previous studies (53,56) may partly explain the low conversion rate of ALA to long-chainn-3 PUFAs.

Other possible reasons for the lack of conversion in the previous studies may be related to amount of ALA provided in the study, the length of the study, lack of washout between study periods or gender of study subjects. In the study by James

80

et al. (228), the ALA intake may have been too low (0.5 g for the first 3 weeks and 1.5 g for the next three weeks) to significantly increase long-chain n-3 PUFAs in erythrocyte PL. In addition, the length of some studies may have been too short (4 or 6 weeks) to see changes in the proportions of EPA and DHA in erythrocytes (53,54,56,228). The lack of washout period in the study of Rajaram et al. (53) may have diluted the effects of increased ALA intake since their study also included a period of fatty fish intake. Moreover, female sex hormones upregulate the synthesis of long-chainn-3 PUFAs (35), which could explain the lack of conversion in the study of Dodin et al. (52). In their study, subjects were postmenopausal women which had not received hormone replacement therapy in the preceding 6 months before the study. It is also noteworthy that none of the studies that included both males and females assessed the conversion of ALA to long-chain n-3 PUFAs separately by gender. Due to the small sample size, we also could not investigate the gender-specific ALA conversion in the CSO group.

Lipid pools may differ in their ability to reflect hepaticn-3 PUFA biosynthesis (229,230). Pignitter et al. (230) found that LDL is a sensitive compartment to investigate the hepatic conversion because it is produced in the liver. Authors of that study also suggested that erythrocytes may not be the most sensitive lipid pool for this purpose since they observed no increase in long-chainn-3 PUFAs after ALA intake. Furthermore, Goyens et al. (229) proposed that ALA conversion could be better reflected in plasma PL than in CE or TG. They did not observe detectable amounts of long-chain n-3 PUFAs in TG and discussed that only CE, which are derived from liver, could reflect hepatic conversion. Plasma CE, however, are also formed in intestine and in the circulation from cholesterol released from peripheral tissues by the action of LCAT.

In the present study, the proportion of EPA or DHA was not increased statistically significantly in any fraction in the CSO group. Therefore, we cannot conclude which blood lipid fraction could most accurately reflect the ALA conversion to long-chain n-3 PUFAs. However, unlike in other fractions, in EM there was no trend towards an increase in the proportion of EPA in the CSO group.

Changes in the proportions of other fatty acids after n-3 PUFA intake

As already discussed, a change in the dietary intake ofn-3 PUFAs is reflected by a change in respective fatty acids in different lipid pools (9). However, an increase in the proportion one fatty acid typically drives down the proportion of another (11). A common finding in the studies with marinen-3 PUFA-enriched diets or intake of fish oil supplement is the increase of these fatty acids in lipid pools at the expense ofn-6 PUFAs (9,20,54). Especially -linolenic acid (18:3n-6), dihomo- -linolenic acid (20:3n-6) and arachidonic acid (20:4n-(20:3n-6) have been found to decrease in CE and PL (20,56,80).

However, increases of EPA and DHA in erythrocytes have also been found to occur at the expense of palmitic acid (16:0) and oleic acid (18:1n-9) (45). Similarly, an increase in ALA has been found to decrease palmitic acid in erythrocytes, -linolenic

81 acid in TG and CE and myristic acid (14:0) and arachidonic acid in CE (45,48). In accordance with these findings, EPA and DHA replaced mainlyn-6 PUFAs in the fatty fish group, especially arachidonic acid, adrenic acid (22:4n-6) and osbond acid (22:5n-6). In the CSO group, n-6 PUFAs (e.g. dihomo- -linolenic acid, osbond acid and arachidonic acid) decreased, and to a lesser extent, MUFAs (e.g. oleic acid) and SFAs (e.g. palmitic acid). These changes were most notable in PL and EM.

6.2.2 Lipoprotein subclasses

Effects of fish intake on lipoprotein subclasses

Fatty fish intake was found to increase the average size of the HDL particle whereas lean fish intake had no effect on the size of HDL. These results are supported by earlier studies reporting increased HDL particle size after fatty fish intake (142,143) and no changes after lean fish intake (143,146). This effect of fatty fish may be related to its DHA content, which has been found to increase the size of HDL (231). However, there are also a few intervention studies that included fatty fish but observed no increase in HDL particle size (84,140,147). In the studies of Li et al. (140) and Rundblad et al. (84), the fish diets contained both fatty fish and lean fish, which could partly explain the lack of effect on HDL size. Instead, in the study of Raatz et al. (147) subjects consumed only fatty fish but the duration of this study (4 weeks) may have been too short to detect changes in the size of HDL. Furthermore, all these studies

Fatty fish intake was found to increase the average size of the HDL particle whereas lean fish intake had no effect on the size of HDL. These results are supported by earlier studies reporting increased HDL particle size after fatty fish intake (142,143) and no changes after lean fish intake (143,146). This effect of fatty fish may be related to its DHA content, which has been found to increase the size of HDL (231). However, there are also a few intervention studies that included fatty fish but observed no increase in HDL particle size (84,140,147). In the studies of Li et al. (140) and Rundblad et al. (84), the fish diets contained both fatty fish and lean fish, which could partly explain the lack of effect on HDL size. Instead, in the study of Raatz et al. (147) subjects consumed only fatty fish but the duration of this study (4 weeks) may have been too short to detect changes in the size of HDL. Furthermore, all these studies