Increased risk of CVD is associated with specific changes in the size and number of individual lipoprotein particles. For example, low HDL-cholesterol and high LDL-cholesterol are independent risk factors for CVD(Reference Schaefer1–Reference Griffin3). The fasting and postprandial TAG concentrations have also been identified as risk factors and are involved in a causal manner in the pathogenesis of CVD(Reference Patsch, Miesenbock and Hopferwieser4, Reference Roche and Gibney5). At least, part of the link between the plasma TAG concentration and the CVD risk lies in the relationship between the size of VLDL and LDL particles, and the extent of the exchange of TAG and cholesteryl esters between VLDL and HDL particles. LDL particles are generated from VLDL remnants. Larger, TAG-rich VLDL1 particles give rise to a range of LDL particles including small, dense LDL3, which have been suggested to be atherogenic(Reference Griffin3, Reference Blackburn, Cote and Lamarche6), although not all studies support this view(Reference El Harchaoui, van der Steeg and Stroes7, Reference Hsia, Otvos and Rossouw8). Smaller VLDL2 particles are the precursors of mainly LDL1 and LDL2 (Reference Griffin3, Reference Blackburn, Cote and Lamarche6, Reference Packard and Shepherd9). The total LDL particle concentration may also contribute to the risk of CVD(Reference Mora, Szklo and Otvos10). Higher VLDL–TAG concentrations are associated with increased exchange of TAG, with HDL–cholesteryl esters leading to the removal of HDL from the circulation and enrichment of VLDL remnants with cholesteryl esters, which contributes to the generation of atherogenic LDL(Reference Griffin3, Reference Blackburn, Cote and Lamarche6, Reference Packard and Shepherd9).
An increased intake of fish oil containing n-3 fatty acids is associated with a lower risk of CVD(Reference Roche and Gibney11–Reference Psota, Gebauer and Kris-Etherton13). A review of seventy-two studies that investigated the effect of increasing dietary intake of fish oil rich in very-long-chain n-3 PUFA, DHA (22 : 6n-3) and EPA (20 : 5n-3) on plasma lipoprotein profiles concluded that supplementation with fish oil lowered fasting plasma TAG and increased the LDL-cholesterol concentration, while the HDL-cholesterol concentration was unaffected(Reference Harris14). Overall, dietary supplementation with n-3 PUFA at approximately 4 g/d induces a decrease in the fasting plasma TAG concentration of 25–30 %(Reference Harris14). Although the reduction in the plasma TAG concentration may be beneficial, the increase in LDL-cholesterol is potentially harmful if it reflects an increase in the number of small atherogenic LDL particles(Reference Minihane, Khan and Leigh-Firbank15, Reference Griffin16). EPA and DHA given as individual fatty acids tend to reduce the fasting TAG concentration to a similar extent(Reference Hansen, Grimsgaard and Nilsen17, Reference Mori, Burke and Puddey18), while reports differ in the extent and direction of the effects of these fatty acids on the LDL- and HDL-cholesterol concentrations(Reference Kelley, Siegel and Vemuri19).
The precise mechanism by which fish oil induces the lower plasma TAG concentration is a matter of debate(Reference Davidson20), but involves changing the size distribution and concentration of plasma lipoproteins(Reference Kelley, Siegel and Vemuri19, Reference Lu, Windsor and Harris21, Reference Mostad, Bjerve and Lydersen22). Many of the effects of EPA and DHA on the cell function are dependent upon the incorporation of these fatty acids into membrane phospholipids over a prolonged period of time(Reference Calder23). However, the TAG lowering effect of fish oil also involves the direct effects of EPA or DHA, or their peroxidation products, on specific transcription factors in the liver(Reference Davidson20). The overall effect of activation of these transcription factors is to modify the metabolic partitioning of fatty acids away from storage towards β-oxidation, thus reducing the availability of TAG for incorporation into TAG and secretion into the circulation. Administration of a hyperinsulinaemic–euglycaemic clamp reduced VLDL secretion, possibly by reducing the flux of NEFA to the liver(Reference Bulow, Simonsen and Wiggins24). Thus, VLDL–TAG secretion may be regulated by the changes in the supply of fatty acid substrates to the liver. DHA was incorporated preferentially into the NEFA pool within 90 min following meals containing 2 or 30 g fish oil(Reference Lu, Windsor and Harris21, Reference Mostad, Bjerve and Lydersen22Reference Heath, Karpe and Milne25, Reference Burdge, Sala-Vila and West26) and incorporated into VLDL–TAG within 3 h when compared with EPA(Reference Heath, Karpe and Milne25). This suggests that DHA released from chylomicron TAG by the lipoprotein lipase activity is relatively poorly trapped, and is rapidly assimilated by the liver and incorporated into TAG destined for secretion in VLDL(Reference Heath, Karpe and Milne25). This raises the possibility that an acute increase in fish oil intake may induce rapid changes in the hepatic VLDL metabolism. If so, then the differences between individuals in the regulation of fatty acid partitioning in the liver may contribute to any variation in the effectiveness of fish oil supplementation in reducing the plasma TAG concentration and hence the risk of CVD. Fish oil may also alter the postprandial TAG metabolism by changing the competition between chylomicrons and VLDL for lipoprotein lipase, and hence altering the rate of hydrolysis and/or altering the interactions between circulating lipoproteins such as the exchange of apolipoproteins.
The aim of the present study was to test the hypothesis that increasing the fish oil content of a single meal alters the size and number of VLDL, LDL and HDL particles in the plasma during the postprandial period. Healthy men aged 50–65 years consumed either a REF meal or a meal containing 7·3-fold more EPA and DHA in a random order, and lipoprotein size and number was measured over the subsequent 6 h period by NMR spectroscopy.
Materials and methods
Subjects
Details of the subjects and the postprandial study have been reported previously(Reference Burdge, Powell and Calder27). The subjects were eleven healthy men aged 58 (sd 5) years, who had a BMI of 25·8 (sd 3·3) kg/m2, a fasting TAG concentration of 1·3 (sd 0·4) mmol/l, a fasting total cholesterol concentration of 4·7 (sd 0·7) mmol/l, a fasting glucose concentration of 5·0 (sd 0·4) mmol/l and a fasting insulin concentration of 4·6 (sd 2·6; mU/l) measured at a pre-study health screen. The volunteers who consumed restricted diets (e.g. vegetarians and vegans), consumed more than two portions of oily fish per week, habitually took vegetable or fish oil or vitamin dietary supplements, had diagnosed type 2 diabetes mellitus or smoked tobacco were excluded from the study. The study received ethical approval from the South and West Hampshire Local Research Ethics Committee. All volunteers gave their written informed consent.
Study design
Each participant took part in the two postprandial studies in a randomised order, with an interval of at least 14 d between each study(Reference Burdge, Powell and Calder27). The subjects consumed their habitual diet throughout the study. On the day preceding a postprandial study, the subjects were asked not to consume alcohol and to refrain from strenuous exercise. They were asked to consume their evening meal by 20.00 hours, but could consume water freely, and to fast until the postprandial study commenced. The subjects arrived at the Wellcome Trust Clinical Research Facility (Southampton General Hospital, Southampton, UK) at approximately 07.30 hours. A cannula was inserted into a forearm vein and a baseline blood sample (12 ml) drawn. Part of the sample (10 ml) was placed into a tube containing heparin sulphate, and the remainder was placed into a tube and allowed to clot. The participants were given one of two test meals at approximately 08.00 hours. These were a reference (REF) meal based on the pattern of fatty acids in the UK diet or a meal with the increased EPA and DHA content (high fish oil, HFO)(Reference Burdge, Powell and Calder27). The macronutrient and fatty acid compositions of the test meals have been reported in detail previously(Reference Burdge, Powell and Calder27) and are summarised in Table 1. The fat and protein component of the meals was administered in an emulsion(Reference Burdge, Powell and Calder27) that was composed of double cream, a blend of sunflower, olive, linseed and fish oils, Nesquik milkshake powder, casein and carbohydrate. The total volume was made up to 160 ml with water. The carbohydrate component of the meal was given as toast with jam or marmalade. The subjects consumed the test meal during a period of approximately 15 min. The blood samples (12 ml) were collected at 0.5, 1, 1.5, 2, 2.5, 3, 4, 5 and 6 h after the consumption of the test meal began. The participants remained resting throughout the study and were allowed free access to water.
Composition of the test meals
REF, reference; HFO, high fish oil.
Measurements of lipoprotein particle size and number, and of 8-isoprostane
The plasma was separated from the cells by centrifugation and frozen. The number and size of particles within VLDL, LDL and HDL lipoprotein classes were measured by NMR spectroscopy (LipoScience, Inc., Raleigh, NC, USA). This method has been validated in the studies of lipoproteins in a number of physiological and pathological states(Reference Otvos, Jeyarajah and Cromwell28), which include the analysis of samples from fasting(Reference Couture, Otvos and Cupples29–Reference Bhalodkar, Blum and Rana32) and postprandial(Reference Schaefer, McNamara and Tayler33) subjects. The particle concentrations were determined from the amplitudes of their spectroscopically distinct NMR signals and weighted average particle diameters were derived from the sum of the diameters of each lipoprotein class multiplied by its relative mass(Reference Kelley, Siegel and Vemuri19, Reference Otvos34, Reference Heijmans, Beekman and Houwing-Duistermaat35). The analyses were performed in accordance with commercial quality control measures. To determine whether fish oil consumption increased the level of lipid peroxides, the plasma 8-isoprostane concentration was measured by DELFIA (Perkin-Elmer, Waltham, MA, USA) time-resolved competitive fluorescence immunoassay using a monoclonal antibody, which has been described previously(Reference Theobald, Goodall and Sattar36).
Statistical analysis
The data were analysed using a standard linear mixed-modelling approach with repeated measures in Statistical Analysis Systems statistical software package version 9.1.3 via Proc Mixed (SAS Institute, Cary, NC, USA), the response variable being the change from baseline. Fixed effects were included in the model to account for the baseline value within the subject/diet combination according to standard practice(37), and the diet × time interaction. Random effects were also included to account for subject and subject × diet interaction. The diet × time interaction term was then decomposed in order to provide an estimate for the difference between diets at each time point. No correction for multiple testing was made. The studentised conditional residuals from each model were inspected in order to assess the model fit and identify the presence of outliers.
Results
All subjects completed both postprandial studies. All data are presented as means with their standard errors for a total of eleven subjects. There were significant differences between the meals at baseline for VLDL particle size (REF 52·1 (sd 2·9) nm, HFO 48·4 (sd 2·8) nm; paired Student's t test, P = 0·021). There were no significant differences between the meals at baseline for VLDL concentration (REF 72·9 (sd 10·8) nmol/l, HFO 66·2 (sd 30·1) nmol/l), LDL particle size (REF 21·1 (sd 0·2) nm, HFO 21·2 (sd 0·3) nm) or concentration (REF 1148·8 (sd 131·3) nmol/l, HFO 1141·5 (sd 134·4) nmol/l), or HDL particle size (REF 8·8 (sd 0·1) nm, HFO 8·8 (sd 0·1) nm) or concentration (REF 31·1 (sd 1·2) μmol/l, HFO 30·4 (sd 1·3) μmol/l).
There was a significant interaction effect of meal composition and time (P = 0·036) on the VLDL concentration (Fig. 1). There was no significant change in the concentration of VLDL particles when compared with baseline following the REF meal. However, following the HFO meal, the concentration of VLDL particles was significantly increased when compared with baseline between 3 and 6 h. Between 3 and 5 h, the change in the VLDL concentration from baseline was significantly greater following the HFO meal when compared with the REF meal. There was a significant interaction effect of meal composition and time (P = 0·005) on VLDL particle size. Consuming the REF meal increased the size of VLDL particles when compared with baseline between 1 and 6 h. Consuming the HFO meal also significantly increased the size of VLDL particles between 1 and 4 h, but returned to a level not significantly different from baseline by 5 h. Between 2.5 and 4 h, VLDL particles were significantly smaller (P < 0·05 to P < 0·001) after the HFO meal than the REF meal.
Postprandial lipoprotein concentration ((a) VLDL, (b) LDL and (c) HDL) and size ((d) VLDL, (e) LDL and (f) HDL) corrected for baseline values (means with their standard errors) following the consumption of the reference (
; REF) or high fish oil (
; HFO) meals in men (n 11). Data were analysed using a standard linear mixed-modelling approach with time and meal composition as the fixed factors. The different symbols indicate different levels of statistical significance: *P < 0·05, †P < 0·01, ‡P < 0·001 and §P < 0·0001. The symbols on the graph indicate the differences between the REF and HFO meals at individual time points. The symbols below the x-axis indicate statistically significant differences from baseline following each meal.
There was no significant effect of time or meal composition on the LDL concentration (Fig. 1). There was a significant effect of time (P = 0·013), but not meal composition, on LDL particle size. Following both the meals, LDL particle size was significantly reduced when compared with baseline between 2 and 5 h.
There was a significant effect of time (P = 0·0002), but not meal composition, on the HDL concentration. The HDL concentration was significantly lower than baseline between 2·5 and 4 h following the REF meal, and between 3 and 4 h following the HFO meal (Fig. 1). There was also a significant effect of time (P < 0·0001), but not meal composition, on HDL particle size. There was an increase in the HDL size following both the meals, but this was only significant following the HFO meal.
There was no significant effect of diet or time on the 8-isoprostane concentration (data not shown).
Discussion
The results of the present study show, for the first time, that increasing the amount of fish oil in a single meal induces class-specific changes in the concentration and particle size of lipoproteins. Consuming the HFO meal increased the VLDL concentration and decreased the VLDL size during the postprandial period. An acute increase in the consumption of fish oil did not significantly alter the LDL or HDL concentration or the particle size. There were modest, but highly variable, changes in the LDL size and in the HDL concentration and size in the late postprandial period.
A number of studies have reported that increased consumption of fish oil, or purified EPA or DHA, lowers the fasting or postprandial TAG concentration(Reference Harris14), and that these fatty acids change the particle size distribution and concentration of specific lipoproteins(Reference Kelley, Siegel and Vemuri19). Since these studies involve increased intake of EPA and/or DHA over a prolonged period, the findings imply the induction of metabolic adaptation rather than an acute effect. The mechanism through which n-3 PUFA lower the plasma TAG concentration is unclear, although it has been argued that altered metabolic partitioning of the fatty acids is critical to this process(Reference Davidson20). In the present study, we present data which show that consuming a single meal enriched with fish oil significantly alters both the VLDL concentration and particle size, and that this effect is specific to VLDL particles. We have shown in the same samples that consuming the HFO meal did not alter the magnitude of the change in total plasma TAG(Reference Burdge, Powell and Calder27). It is possible that any effect of the HFO meal on VLDL–TAG may have been masked by chylomicron TAG from the meal. However, the present data suggest that although the total plasma TAG concentration was not altered by increasing fish oil intake, there was a shift in the VLDL metabolism towards the secretion of a greater number of smaller VLDL particles. These changes are in agreement with the effects of dietary supplementation with fish oil on the VLDL size and concentration(Reference Kelley, Siegel and Vemuri19). Thus, the metabolic basis of the beneficial effects of increased fish oil intake on the plasma TAG concentration may reflect acute changes in the regulation of VLDL particle size and secretion. Smaller VLDL particles are associated with the generation of larger LDL particles and thus lower atherogenic risk(Reference Schaefer1). Thus, these findings, with respect to VLDL, are in agreement with the beneficial effects of fish oil on CVD risk(Reference Hansen, Grimsgaard and Nilsen17).
There were small postprandial changes in LDL and HDL particles, but these were not affected by the amount of fish oil in the meal. The postprandial changes in the LDL metabolism have been reported previously(Reference Blackburn, Cote and Lamarche6). It is possible that significant changes in the LDL or HDL metabolism associated with fish oil supplementation(Reference Mori, Burke and Puddey18) may have occurred after the period of blood sampling and so were not detected, or that repeated exposure to increased fish oil intake is required to modify the LDL or HDL size and concentration substantially.
Given that significant enrichment of EPA and DHA into cell membranes in human subjects due to increased dietary intake of these fatty acids occurs over a period of several weeks or more(Reference Calder23), the present findings do not support altered membrane fatty acid composition as the primary mechanism underlying the effect of fish oil on the VLDL metabolism. One explanation for the effect of fish oil on the VLDL metabolism is an acute change in the activity and/or expression of transcription factors that are regulated by the fatty acids, including HNF4a, PPARα, PPARγ and SREBP1c, and that modify the fatty acid metabolism(Reference Davidson20). We have shown previously that DHA released from chylomicron TAG is preferentially incorporated into the plasma NEFA pool(Reference Burdge, Sala-Vila and West26) and incorporated into VLDL–TAG within the postprandial period(Reference Heath, Karpe and Milne25). This supports the suggestion that the fatty acids released from fish oil reach the liver within the postprandial period, and hence may be able to interact with transcription factors that regulate the hepatic fatty acid metabolism. It has been suggested that fatty acid oxidation products are more potent agonists of the transcription factors that modify metabolic partitioning of the fatty acids in the liver(Reference Davidson20). However, there was no indication of increased lipid peroxidation following the HFO meal when compared with the REF meal, which suggests that any direct effect of the fatty acids on the regulation of the VLDL metabolism was not mediated primarily through lipid peroxides. The reduction in the NEFA flux to the liver has been suggested to explain the reduction in the VLDL–TAG concentration during a hyperinsulinaemic–euglycaemic clamp(Reference Bulow, Simonsen and Wiggins24). However, we have shown in the same samples as those used for lipoprotein measurements that there was no effect of the HFO meal when compared with the REF meal on the plasma NEFA concentration during the postprandial period(Reference Burdge, Sala-Vila and West26). This supports the suggestion that the effects of the HFO meal on the VLDL metabolism were not due to the lower flux of fatty acids to the liver. However, in the absence of measurements of chylomicron size and concentration, it is also possible that the acute effects of fish oil on the VLDL metabolism may also be due to a change in the competition between VLDL and chylomicrons for lipoprotein lipase, and hence their relative rates of hydrolysis.
Irrespective of the underlying mechanism, these findings suggest that the beneficial effects of fish oil on circulating VLDL levels reflect the acute changes in the VLDL metabolism. Thus, one possible implication is that the variation between individuals in the extent to which fish oil lowers the fasting or postprandial TAG concentration may reflect the differences in the function of pathways that regulate the VLDL metabolism.
Acknowledgements
We thank the study participants and the staff of the Wellcome Trust Clinical Research Facility, Southampton General Hospital, UK for their time and commitment. We also thank LipoScience for carrying out the NMR analysis. G. C. B., J. P., D. T. and P. C. C. designed the study. G. C. B. carried out the study, and G. C. B. and D. T. carried out the analysis. T. D. and J. C. carried out the statistical analysis. G. C. B. wrote the manuscript with contributions from P. C. C., J. P., T. D. and D. T. None of the authors had any financial or personal conflicts of interest. The study was funded by a grant from Unilever R&D Colworth to G. C. B. and P. C. C.
