Arachidonate 5-Lipoxygenase Gene Variants Affect Response to Fish Oil Supplementation by Healthy African Americans^,

Abstract

Arachidonate 5-lipoxygenase (ALOX5) gene variants that are common in people of African ancestry are associated with a differential cardiovascular disease (CVD) risk that may be ameliorated by intake of (n-3) PUFA, such as EPA or DHA. We conducted a double-masked, placebo (PL)-controlled trial of fish oil (FO) supplements to determine if changes in erythrocyte (n-3) PUFA composition, heart rate, blood pressure, and plasma lipid and lipoprotein concentrations are modified by genotype. Participants received 5 g/d FO (2 g EPA, 1 g DHA) or 5 g/d corn/soy oil (PL). A total of 116 healthy adults of African ancestry with selected genotypes (genotypes = “dd,” “d5,” and “55” with “d” representing the deletion of 1 or 2 Sp1 binding sites in the ALOX5 promoter and “5” indicating the common allele with 5 sites) were enrolled and 98 completed the study. FO caused significant increases (relative to PL) in erythrocyte EPA, DHA, and total (n-3) PUFA and a decrease in the (n-6) PUFA:(n-3) PUFA ratio in the low-CVD risk “d5” and “55” genotypes but not in the high-risk “dd” genotype. Similarly, HDL particle concentration decreased with FO relative to PL in the “d5” and “55” but not “dd” genotypes. The plasma TG concentration decreased significantly with FO relative to PL in the “d5” but not “dd” and “55” genotypes. No changes were seen in LDL particle or cholesterol concentrations, heart rate, or blood pressure. These findings indicate that the efficacy of FO supplements vary by ALOX5 genotype.

Introduction

Accumulating evidence suggests that consumption of the long-chain (n-3) PUFA EPA [20:5(n-3] and DHA [22:6(n-3)] from fish or fish oil (FO)⁹ supplements decreases the risk of cardiovascular disease (CVD) through several mechanisms, including lowering plasma TG and decreasing heart rate, blood pressure, and inflammation (1–4). The antiinflammatory effects of EPA and DHA are mediated in part by the modification of arachidonic acid [AA; 20:4(n-6)] metabolism via the 5-lipoxygenase (5-LO) pathway. The 5-LO enzyme encoded by the ALOX5 gene mediates the first step in the synthesis of leukotrienes (LT), such as LTB4, from AA. EPA is also a substrate for 5-LO, but the EPA-derived LT, such as LTB5, have less inflammation-promoting activity than do the corresponding AA-derived LT. Thus, increased (n-3) PUFA intake can decrease inflammation. Because inflammation may itself increase plasma TG concentrations (5), the antiinflammatory effects of FO may play a role in lowering plasma TG, although other mechanisms are also clearly important (6).

The promoter region of the ALOX5 gene has from 3 to 8 tandem repeats of a consensus binding site for the transcription factors Specificity Protein 1 (Sp1) and Early Growth Response protein 1 (Egr1), with the most frequent allele having 5 such sites (7). Variant alleles with 3 or 4 sites (or, rarely, >5 sites) are associated with greater intima-media thickness of the carotid artery in healthy adults, according to a study from Los Angeles (8), and with occurrence of a first myocardial infarction, as shown in a case control study in Cost Rica (9). In both studies, the observed diet-gene interactions were more pronounced in individuals with high dietary AA intake who also carried the ALOX5 deletion variant alleles (i.e., 3 or 4 repeats, referred to jointly as “deletion” or “d” alleles) relative to those with 2 common alleles (i.e., the “55” genotype). Additionally, in the Los Angeles study (8), the association of “d” alleles with CVD risk was lower in individuals with higher EPA and DHA intake. More recently, the “d” alleles were also associated with increased risk of CVD in African Americans ascertained through elective cardiac evaluation (10). However, 2 other case-control studies from the UK (11) and the US (12) did not report associations of CVD with the ALOX5 promoter polymorphism, although these 2 studies did not perform gene-diet interaction analyses.

Although the effects of FO supplements on plasma lipids have been evaluated in many intervention trials (13–15), to our knowledge, such studies have not been conducted in African Americans, who, compared with European Americans, have a higher risk of CVD (16) despite having lower plasma TG concentrations (17). Furthermore, the frequency of the ALOX5 3 allele is ~10-fold higher in Africans than in Asians or Europeans (8). Based on these observations and the previously reported diet-gene interactions with the ALOX5 promoter variants (8), we sought to evaluate the effect of FO supplements in people of African ancestry living in the US who may differ in their response to FO, relative to other ancestry groups, due to genetic and/or environmental (e.g., dietary) differences.

The aim of the present study was to conduct a placebo (PL)-controlled intervention trial of FO supplementation in individuals of African ancestry, randomized within 6 ALOX5 promoter genotypes, to determine whether healthy men and women with 1 or 2 “d” alleles have baseline plasma lipid and lipoprotein profiles consistent with greater CVD risk and have a greater improvement in these profiles in response to FO supplementation relative to those with the control “55” genotype. We previously reported from this study that production of both AA- and EPA-derived 5-LO metabolites by monocytes differs by ALOX5 genotype both at baseline and in response to supplementation (18).

Participants and Methods

Study design

Participant recruiting and genotype analysis.

Healthy adults 20–59 y of age who self-identified as African American, black, or of African ancestry were recruited into the study in Oakland, Davis, and Sacramento as previously described (19). Briefly, after an initial questionnaire and genotype assessment, potentially eligible participants were contacted to schedule an appointment for a screening blood draw to be used for a complete blood count, lipid panel, and chemistry panel. Eligible men and women were then invited for the baseline (wk 0) blood draw at which time randomization to a 6-wk regimen of FO or PL was performed.

Ethical review and trial registration.

The institutional review boards of the University of California, Davis and Alta Bates Summit Medical Center reviewed and approved this protocol. Written informed consent was obtained from all study participants.

Inclusion and exclusion criteria.

Eligible individuals had 1 of the 6 ALOX5 genotype groups of interest to the study (“dd” genotypes: “33,” “34,” “44” “d5” genotypes: “35,” “45” control genotype: “55”). Potential participants who reported a physician-diagnosed, chronic, inflammation-related disease; lipid disorder; or regular use of antiinflammatory or lipid-lowering medications were excluded. Other exclusions have been described (19).

Randomization to treatment groups within genotypes.

Participants within each of the 6 ALOX5 genotypes were allocated to treatment or PL using 6 randomization lists and a randomized-block design with a block size of 2.

FO and PL treatments.

FO and PL capsules (1.0 g/capsule) were provided in bulk by Ocean Nutrition Canada, as previously described (19). Volunteers consumed 5 capsules/d for 6 wk. The FO was 40/20 ethyl ester (lot nos. 10524 and 8980) and the PL was corn/soybean PL (lot nos. 10525 and 8981). The fatty acid profile for the 40/20 ethyl ester capsules consisted of a minimum 40% EPA and 20% DHA by analysis. The 5 g of FO thus contained 2.0 g EPA and 1.0 g DHA. The composition of the corn/soybean oil capsules included the following: 0% DHA, 0% EPA, 25% oleic acid, 53% linoleic acid, 5% α-linolenic acid, 10% palmitic acid, and 7% other fatty acids.

Sample size.

The sample size within genotype groups was based on the ability to detect differences in ALOX5 mRNA expression at baseline between deletion and control (“55”) genotypes and on expected reductions in ALOX5-derived LT metabolite production by monocytes. Our goal was to recruit 30 participants in order to retain 24 in all genotype groups, except the homozygous “44” genotype because of the lower prevalence of that genotype (8), which had a frequency of only 2.5% (9/354) in the individuals screened for this study. This manuscript reports results on endpoints not considered in the original power calculation. For this reason, we performed a power calculation to determine the magnitude of the treatment effect that we could expect to detect based on the numbers of participants actually recruited and retained in the “dd,” “d5,” and “55” genotype groups. The minimum differences between means of genotype groups “dd” and “55” and “d5” and “55” that we could expect to detect with 80% power at P < 0.05 were 0.82 and 0.72 SD, respectively.

Laboratory methods

Blood draws.

Blood (80 mL) was drawn from the antecubital vein into sodium-heparin tubes at wk 0 (baseline) and wk 6 after an overnight fast of 12 h. Plasma was stored at −80°C.

Dietary analysis.

A Block FFQ (version 2005; Block Dietary Data Systems) was administered at the baseline study visit to estimate usual dietary intake patterns, including the use of supplements of (n-3) fatty acids (i.e., EPA, DHA, α-linolenic acid, and flaxseed), total SFA, MUFA, and PUFA over the preceding year. The FFQ was self-administered with initial instruction provided by a trained staff member following the blood draw. Serving-size pictures and food models were provided to help estimate portion sizes. A registered dietitian reviewed the questionnaires. Interviewers were contacted about missing information, unusual responses, or discrepancies prior to data analysis.

Blood pressure and heart rate.

Blood pressure and heart rate were measured using an automated instrument (Spot Vital Signs model 4200B-E1, Welch Allyn) while the study participant was at rest. Each participant remained at rest for 5 min prior to each measurement.

Plasma lipid analysis.

At baseline, plasma lipid and lipoprotein concentrations were measured for all 116 participants and at wk 6 for all 98 participants who completed the study. One participant was excluded from analysis (at both visits), because the baseline TG concentration (3.76 mmol/L) was much higher than the screening visit for the same participant (1.22 mmol/L), suggesting that the participant was not fasting. NMR analysis (LipoScience) was used to measure mean lipoprotein particle diameter, total number of particles, and particle concentration within different subclasses of VLDL, LDL, and HDL from frozen plasma samples (20).

Erythrocyte fatty acid analysis.

Erythrocyte fatty acids were analyzed to assess the impact of lipid supplementation on nutritional status. Paired samples were available for 95 of the 98 participants who finished the study. Samples from FO- and PL-treated participants were randomized into batches of 20 containing a laboratory reference material, a procedural blank, and both pre- and post-treatment samples from selected participants. RBC lipids were extracted with methanol containing 20% toluene and lipids were transesterified using methanolic sodium hydroxide in the presence of the analytical surrogate triheptadeca-(17-Z)-eneoyl glyceride. The resulting FAME were back extracted into hexane, enriched with tricosanoic acid (23:0) as an internal standard, separated by GC on a 30m × 0.25-mm × 0.25-μm DB-225ms (Agilent Technologies), and detected on an Agilent Technologies 5973N mass spectral detector operated in simultaneous selected ion monitoring/full scan mode. Quantification was performed on acquired selected ion monitoring signals. Surrogate recoveries were acceptable (87 ± 30%) and random across analytical batches, as were the reference material replicate analyses.

Statistical analysis

Statistical analyses were performed using SAS software (version 9.2; SAS Institute). Correlation analysis was performed using Pearson’s method for continuous variables and Spearman’s method for rank variables. Group comparisons of categorical data by genotype were made using the chi-square test. Two-group comparisons of continuous variables were made by Student’s t test or the Rank-Sum test for variables that were not normally distributed. Comparisons of continuous data for more than 2 groups were made by 1-way or 2-way ANOVA using the generalized linear model (GLM) procedure for normally distributed or rank-transformed variables and the Tukey-Kramer adjustment for multiple comparisons. The GLM procedure was used to adjust comparisons of baseline variables among genotypes for covariates known to affect the level of the dependent variable. In particular, baseline heart rate, blood pressure, and plasma lipid and lipoprotein concentrations were adjusted for sex, age, and BMI. Baseline dietary intake was adjusted for sex and body weight, body weight being used rather than BMI, because body weight is directly related to dietary requirements. The effect of the intervention on change in the level of dependent variables between baseline and 6 wk (the end of the intervention) was analyzed using the GLM procedure to identify treatment effects, genotype effects, and treatment by genotype interactions. These analyses were done in 3 ways: 1) without adjustment; 2) model 1 (M1), with adjustment for body weight (because all participants received the same FO dose regardless of weight) and the baseline level of the dependent change-variable of interest; and 3) model 2 (M2), with adjustment for these 2 variables plus rank of percent compliance with supplement use. Both models gave similar results; thus, only the M2 results are presented.

When the baseline level of the dependent change-variable of interest was not normally distributed, the rank of the baseline variable was used for adjustment. The change variables themselves were analyzed as untransformed variables if normally distributed or as ranks if they were not normally distributed, and this information is provided in the various tables. A 2-tailed P value of 0.05 was used as the benchmark for significant differences. P = 0.10 was used as a trigger for further analysis (e.g., trend analysis across genotypes when the ANOVA P value was between 0.05 and 0.10). Adjustment for multiple comparisons among the genotype groups was made during ANOVA using the Tukey-Kramer method and adjusted P values are reported.

When lipid and lipoprotein variables were examined by ANOVA to identify genotype or genotype × treatment interactions, we adjusted for multiple comparisons with Sidak’s method using the MULTTEST procedure if the P value was <0.05. The unadjusted P values are reported in tables, but footnotes are used to indicate whether the adjusted P value remained significant after adjustment. This adjustment was made because of the large number of lipid and lipoprotein endpoints and was done only with the genotype interaction for baseline data and the genotype × treatment interaction for intervention data, because the effect of genotype was the principal focus of this analysis and because FO treatment effects alone are already well characterized for many of these variables. Adjustments were made for variables from the same hierarchical level. For example, when a significant P value was found for small HDL particle concentration using the M1 adjustment model, then P values for large, medium, and small HDL particle concentrations using the M1 model were all adjusted (i.e., n = 3 variables in this level). Similarly, when a significant P value was found for total HDL particle concentration using the M2 model, then P values for total VLDL, LDL, and HDL particle concentrations (using the M2 model) were adjusted. Such adjustments were not made for other variables in this study when no significant effects were seen (e.g., association of genotype with dietary intake or effect of treatment on heart rate and blood pressure), when specific comparisons were part of the principal hypothesis behind the trial design (e.g., effects on EPA and DHA concentrations in erythrocyte membranes), or when variables at the same hierarchical level lacked independence (e.g., membrane fatty acid concentrations, which are reported as mole %, where increases in one fatty acid necessarily will lead to a decrease in another).

Results

Baseline data

Demographics.

A total of 783 individuals were screened for entry into the study, 116 were randomized to either the FO or PL, and 98 completed the 6-wk intervention, as previously reported (19). Seventy percent of participants were female, ages ranged from 20 to 59 y, and BMI from 20 to 37 kg/m². Age, sex, and BMI did not differ by genotype (Table 1) or between the FO and PL groups (FO and PL group means are not shown).

TABLE 1.

Demographic data, heart rate, and blood pressure at baseline in all participants and grouped by ALOX5 genotype (“dd,” “d5,” and “55”)¹

Variable	All	“dd”	“d5”	“55”	P value2
n	116	33	53	30
Sex, female, % (n)	69.8 (81)	78.8 (26)	71.7 (38)	60.0 (18)	0.27
Age,3 y	34; 25, 46	33; 26, 36	35; 24, 48	32; 26, 49	0.59
BMI, kg/m2	27.6 ± 4.3	27.3 ± 4.6	28.3 ± 4.5	26.8 ± 3.6	0.25
Body weight, kg	78.7 ± 13.5	76.9 ± 14.8	80.1 ± 13.2	78.1 ± 12.5	0.54
Compliance,34 % of capsules	94; 81, 100	91; 85, 100	94; 77, 100	94; 88, 100	0.82
Systolic blood pressure,3 mm Hg	113 ± 13	110 ± 14 (112 ± 1.9)*	115 ± 13 (114 ± 1.5)	113 ± 14 (112 ± 2.0)	0.15 (0.44)
Diastolic blood pressure,3 mm Hg	70.4 ± 9.2	69.5 ± 9.8 (70.6 ± 1.4)	71.6 ± 8.6 (70.9 ± 1.1)	69.0 ± 9.5 (69.0 ± 1.5)	0.27 (0.41)
Heart rate, bpm	68.3 ± 9.2	70.5 ± 10.8 (69.9 ± 1.5)	68.1 ± 7.9 (68.6 ± 1.2)	66.4 ± 9.1 (66.8 ± 1.6)	0.19 (0.41)

¹Values are mean ± SD unless otherwise indicated. *Values in parentheses are mean ± SEM adjusted for sex, age, and BMI, and P values for ANOVA adjusting for these covariates.

²P value for comparison among genotypes by chi-square test or ANOVA. Continuous variables that were not normally distributed were transformed to ranks for analysis.

³Indicates continuous variables that were not normally distributed. Median; 25th, 75th percentiles are shown except in the case of blood pressure, where mean ± SD are shown to allow comparisons to adjusted means.

⁴Data were available for 92 participants: 23, 42, and 25, respectively, for the “dd,” “d5,” and “55” genotypes.

Dietary intake.

Intake of energy, protein, carbohydrate, fat, and cholesterol did not vary by genotype (Supplemental Table 1), nor did they differ between the PL and FO groups (FO and PL group means are not shown).

Erythrocyte fatty acid composition.

Our initial hypothesis that LT metabolism would vary by genotype raised the question of whether baseline levels of fatty acids that are substrates for the 5-LO and related pathways, particularly EPA, DHA, and AA, might also vary by genotype. Although these fatty acids did not vary across genotype when analyzed by ANOVA (Supplemental Table 2), the P value for EPA was marginally significant (P = 0.07), with the means for the “d5” and “55” genotypes being 31 and 44% greater, respectively, than the “dd” mean. Of the other fatty acids evaluated, only palmitic had a significant difference by genotype, with the heterozygous “d5” genotype having a higher level than the “55” genotype, while neither differed from the intermediate level of the “dd” genotype.

Given the apparent trend of EPA concentration across genotype (EPA appeared lowest in “dd” and highest in “55”), a correlation analysis was performed for all fatty acids to identify significant trends across genotypes using the number of common “5” alleles (i.e., “dd” = 0, “d5” = 1, and “55” = 2) as the genotype variable. The erythrocyte content of both EPA (r = 0.21; P = 0.023; Spearman rank-order correlation) and the 22-carbon (n-3) PUFA docosapentaenoic acid (r = 0.19; P = 0.039) correlated positively with the number of “5” alleles, whereas the sum of MUFA correlated negatively (r = −0.19; P = 0.048) with the number of “5” alleles.

Heart rate and blood pressure.

Heart rate and blood pressure did not vary by genotype at baseline when evaluated both with and without adjustment for sex, age, and BMI (Table 1).

Plasma lipid and lipoprotein concentrations.

Plasma lipid and lipoprotein concentrations at baseline generally did not vary by genotype when evaluated with or without adjustment for sex, age, and BMI (Supplemental Table 3). However, small VLDL particle and TG concentrations were both significantly greater in the “dd” than the “d5” genotype, though neither differed from the “55” genotype, which was intermediate. The total cholesterol concentration was also highest in the “dd” genotype (intermediate in “d5” and lowest in “55”), though the difference among the genotypes was only marginally significant (P = 0.09). No significant trends across genotypes were seen by regression analysis (using the number of “5” alleles as the genotype variable) for any of these variables.

Effect of FO intervention

Compliance and response to supplementation.

Compliance was assessed using counts of returned pills, as previously reported (19). The median (25th, 75th percentile) compliance was 94% and did not vary by genotype (Table 1) or treatment group (FO: 94, 79, 100%; PL: 94, 86, 100%; P = 0.89). The change in erythrocyte EPA levels (mole %) over the 6-wk study correlated positively with percent compliance in the FO group (r = 0.35; P = 0.019; n = 44; Pearson product-moment correlation), as did the change in DHA (r = 0.39; P = 0.0084). In addition, we evaluated the association of body weight with change in EPA and DHA, because a fixed dose of FO was given and a higher body weight would thus be expected to produce a lower response. Change in both EPA (r = −0.37; P = 0.0090) and DHA (r = −0.41; P = 0.0037) correlated negatively with body weight, as expected.

Change in erythrocyte fatty acid composition.

Changes in fatty acid composition over the 6-wk intervention period were compared by treatment (FO or PL) and genotype (“dd,” “d5,” and “55”) to identify treatment effects, genotype effects, and interactions between the 2 (Table 2). As expected, FO treatment caused increases in EPA, DHA, and total (n-3) PUFA and a decrease in the total (n-6) PUFA:(n-3) PUFA ratio in the FO group relative to the PL group. In addition, there were significant gene × treatment interactions for these 4 variables, indicating a difference in response to supplementation among the genotypes. When pair-wise comparisons were examined within genotypes, significant increases in EPA, DHA, and total (n-3) PUFA and significant decreases in the (n-6) PUFA:(n-3) PUFA ratio were seen for participants receiving FO in the low-risk “55” and the heterozygous “d5” genotype groups relative to PL participants with the same genotype (P < 0.05). However, these variables did not significantly differ between the FO and PL participants in the high-risk “dd” genotype group (Table 2). These differences among the genotype groups are evident when the adjusted means are presented graphically (Fig. 1). Thus, the “dd” genotype did not respond to FO supplementation with the expected increases in EPA and DHA, whereas the other genotypes showed such a response. With regard to changes in other fatty acids, total MUFA and total (n-6) PUFA both decreased with FO treatment (relative to PL) (Table 2), although with no genotype × treatment interaction.

TABLE 2.

Change in erythrocyte membrane fatty acid composition by FO and PL TRT over the 6-wk treatment period for all participants and grouped by ALOX5 genotype (“dd,” “d5,” and “55”)¹

					ANOVA P values2
Variable and TRT	All participants	“dd”	“d5”	“55”	Adjust	GT	TRT	GT x TRT
Participants, n
All	95	22	48	25
FO	49	12	24	13
PL	46	10	24	12
	mol %
16:0 (palmitic)
All	0.32 ± 3.24	−0.53 ± 2.37	0.087 ± 2.98	1.51 ± 4.07	−	0.063	0.76	0.37
FO	0.093 ± 2.88	−0.18 ± 2.44	0.17 ± 2.67	0.21 ± 3.75	+	0.21	0.78	0.13
PL	0.56 ± 3.60	−0.94 ± 2.33	0.007 ± 3.32	2.93 ± 4.08
18:0 (stearic)
All	0.073 ± 4.10	−0.18 ± 2.49	0.61 ± 3.48	−0.73 ± 5.94	−	0.80	0.75	0.78
FO	0.25 ± 3.24	−0.58 ± 2.30	0.73 ± 3.24	0.12 ± 3.99	+	0.25	0.73	0.98
PL	−0.11 ± 4.88	0.30 ± 2.75	0.48 ± 3.78	−1.65 ± 7.60
18:1(n-9) (oleic)
All	−0.099 ± 2.32	0.032 ± 1.39	−0.61 ± 2.00	0.77 ± 3.18	−	0.23	0.0024	0.25
FO	−0.65 ± 2.00*	−0.36 ± 1.12	−0.86 ± 2.56	−0.52 ± 1.49	+	0.84	0.0048	0.42
PL	0.48 ± 2.50	0.50 ± 1.60	−0.37 ± 1.24	2.17 ± 3.94
18:1(n-7) (cis-vallenic)
All	0.024 ± 0.43	0.053 ± 0.40	0.006 ± 0.46	0.034 ± 0.433	−	0.82	0.026	0.13
FO	−0.086 ± 0.33*	−0.045 ± 0.24	−0.056 ± 0.39	−0.18 ± 0.28	+	0.79	0.10	0.28
PL	0.14 ± 0.50	0.17 ± 0.52	0.068 ± 0.51	0.26 ± 0.46
18:2(n-6) (linoleic)
All	−0.56 ± 2.14	−0.18 ± 1.72	−0.68 ± 1.92	−0.65 ± 2.81	−	0.66	0.060	0.044
FO	−1.07 ± 1.96*	0.21 ± 1.17	−1.47 ± 1.25	−1.53 ± 2.98	+	0.78	0.042	0.48
PL	−0.009 ± 2.20	−0.64 ± 2.18	0.11 ± 2.16	0.29 ± 2.39
20:3(n-6) (dihomo-γ-linolenic)
All	−0.15 ± 0.32	−0.19 ± 0.25a,b	−0.073 ± 0.35a	−0.27 ± 0.30b	−	0.033	0.0004	0.34
FO	−0.23 ± 0.36*	−0.25 ± 0.28	−0.16 ± 0.44	−0.33 ± 0.24	+	0.032	<0.0001	0.38
PL	−0.071 ± 0.26	−0.13 ± 0.21	0.018 ± 0.19	−0.20 ± 0.35
20:4(n-6) (AA)
All	−0.73 ± 3.16	−0.27 ± 2.52	−0.69 ± 3.45	−1.22 ± 3.11	−	0.18	0.37	0.42
FO	−0.67 ± 2.46	−0.52 ± 2.78	−0.85 ± 2.04	−0.46 ± 3.01	+	0.063	0.33	0.55
PL	−0.80 ± 3.79	0.029 ± 2.29	0.53 ± 4.49	−2.04 ± 3.13
22:4(n-6) (adrenic)
All	−0.26 ± 0.75	−0.21 ± 0.83	−0.24 ± 0.61	−0.32 ± 0.94	−	0.28	0.0076	0.26
FO	−0.42 ± 0.74*	−0.42 ± 0.88	−0.52 ± 0.58	−0.22 ± 0.88	+	0.062	0.0015	0.44
PL	−0.081 ± 0.74	0.037 ± 0.74	0.046 ± 0.51	−0.43 ± 1.03
22:5(n-6)
All	−0.058 ± 0.37	−0.096 ± 0.30	0.006 ± 0.22	−0.15 ± 0.58	−	0.63	0.055	0.56
FO	−0.14 ± 0.44*	−0.16 ± 0.35	−0.062 ± 0.22	−0.26 ± 0.74	+	0.39	0.035	0.91
PL	0.027 ± 0.24	−0.016 ± 0.23	0.074 ± 0.19	−0.032 ± 0.33
18:3(n-3) (α-linolenic)
All	0.004 ± 0.12	−0.035 ± 0.19	0.019 ± 0.095	0.010 ± 0.096	−	0.77	0.58	0.48
FO	0.000 ± 0.15	−0.062 ± 0.26	0.027 ± 0.097	0.007 ± 0.098	+	0.65	0.32	0.34
PL	0.008 ± 0.086	−0.003 ± 0.040	0.011 ± 0.095	0.013 ± 0.098
20:4(n-3)
All	−0.005 ± 0.062	−0.020 ± 0.069	0.006 ± 0.057	−0.012 ± 0.065	−	0.82	0.68	0.38
FO	−0.003 ± 0.072	−0.038 ± 0.089	0.012 ± 0.076	0.003 ± 0.030	+	0.94	0.55	0.68
PL	−0.007 ± 0.050	0.002 ± 0.021	−0.001 ± 0.026	−0.029 ± 0.088
20:5(n-3) (EPA)
All	0.65 ± 1.00	0.59 ± 0.85	0.66 ± 1.03	0.69 ± 1.11	−	0.90	<0.0001	0.14
FO	1.29 ± 1.02*	0.97 ± 0.96	1.37 ± 1.04*	1.44 ± 1.07*	+	0.81	<0.0001	0.028
PL	−0.021 ± 0.25	0.15 ± 0.37	−0.045 ± 0.18	−0.11 ± 0.23
22:5(n-3)
All	0.28 ± 0.55	0.33 ± 0.50	0.32 ± 0.56	0.15 ± 0.58	−	0.40	<0.0001	0.16
FO	0.56 ± 0.52*	0.53 ± 0.44	0.58 ± 0.62	0.55 ± 0.37	+	0.16	<0.0001	0.084
PL	−0.025 ± 0.40	0.10 ± 0.48	0.05 ± 0.31	−0.28 ± 0.43
22:6(n-3) (DHA)
All	0.51 ± 1.24	0.66 ± 1.04	0.58 ± 1.18	0.25 ± 1.51	−	0.28	<0.0001	0.11
FO	1.16 ± 1.14*	1.05 ± 1.03	1.16 ± 1.27*	1.24 ± 1.04*	+	0.17	<0.0001	0.043
PL	−0.18 ± 0.96	0.19 ± 0.90	−0.01 ± 0.73	−0.83 ± 1.16
∑ SFA
All	0.40 ± 4.52	−0.64 ± 4.05	0.71 ± 3.82	0.74 ± 5.97	−	0.26	0.73	0.85
FO	0.30 ± 3.72	−0.82 ± 3.75	0.86 ± 3.91	0.31 ± 3.36	+	0.072	0.96	0.36
PL	0.51 ± 5.27	−0.43 ± 4.57	0.56 ± 3.81	1.20 ± 8.06
∑ MUFA
All	−0.084 ± 2.65	0.090 ± 1.75	−0.63 ± 2.31	0.82 ± 3.58	−	0.27	0.0032	0.27
FO	−0.77 ± 2.28*	−0.42 ± 1.29	−0.98 ± 2.91	−0.71 ± 1.67	+	0.90	0.0048	0.56
PL	0.65 ± 2.84	0.70 ± 2.09	−0.29 ± 1.48	2.48 ± 4.39
∑(n-6) PUFA
All	−1.76 ± 4.08	−0.97 ± 4.60a	−1.67 ± 3.62a,b	−2.65 ± 4.42b	−	0.17	0.018	0.34
FO	−2.53 ± 3.59*	−1.17 ± 4.25	−3.04 ± 3.02	−2.84 ± 3.87	+	0.035	0.012	0.57
PL	−0.95 ± 4.44	−0.72 ± 5.21	−0.29 ± 3.70	−2.45 ± 5.12
∑(n-3) PUFA
All	1.44 ± 2.58	1.52 ± 2.19	1.59 ± 2.55	1.09 ± 3.01	−	0.49	<0.0001	0.083
FO	2.99 ± 2.44*	2.41 ± 2.28	3.16 ± 2.67*	3.23 ± 2.22*	+	0.30	<0.0001	0.018
PL	−0.21 ± 1.48	0.45 ± 1.58	0.02 ± 1.02	−1.23 ± 1.76
	mole % ratio
∑(n-6) PUFA:∑(n-3) PUFA
All	−1.03 ± 2.80	−1.49 ± 2.12	−1.00 ± 3.25	−0.69 ± 2.39	−	0.38	<0.0001	0.12
FO	−2.36 ± 2.57*	−2.41 ± 2.49	−2.30 ± 3.10*	−2.41 ± 1.55*	+	0.49	<0.0001	0.042
PL	0.38 ± 2.31	−0.38 ± 0.66	0.31 ± 2.90	1.17 ± 1.62

¹Values are mean ± SD. Means for a given GT within the same group (All, FO, or PL) without a common letter differ. *Different between TRT within the same group. Mean ± SD values are not adjusted. AA, arachidonic acid; FO, fish oil; GLM, generalized linear model; GT, genotype; PL, placebo; TRT, treatment group.

²Statistical analysis (2-way ANOVA using a GLM) was performed on change in fatty acid (using ranked data, excepting linoleic acid and DHA, which were normally distributed) without adjustment (indicated by a minus sign in the “Adjust” column) and with adjustment (indicated by a plus sign in the “Adjust” column) for body weight, baseline concentration of the fatty acid of interest (using ranked data, excepting DHA, which were normally distributed at baseline), and rank of percent compliance. The values for these analyses indicate differences between TRT, among GT, or an interaction between the 2 (GT × TRT).

FIGURE 1.

Change in erythrocyte fatty acid composition over a 6-wk supplementation period in participants receiving FO supplements or PL grouped by ALOX5 genotype (“dd,” “d5,” “55”). Values are mean ± SEM. Mean changes (adjusted for baseline fatty acid composition, body weight, and percent compliance with supplementation) are shown for EPA(A), DHA(B), total (n-3) PUFA (C); and the total (n-3) PUFA:(n-6) PUFA ratio (D). *Different between FO and PL within genotypes. P values for 2-way ANOVA are shown in Table 2. FO, fish oil; PL, placebo. Sample sizes for FO/PL were 12/10, 24/24 and 13/12 for the “dd,” “d5,” and “55” genotypes, respectively.

Effect of intervention on heart rate and blood pressure.

Heart rate and blood pressure did not change as a result of the FO treatment (relative to PL) in the total group or within genotypes (Supplemental Table 4).

Effect of FO intervention on plasma lipid and lipoprotein concentrations.

The total TG concentration decreased significantly with FO treatment, relative to PL, over the 6-wk intervention period (Table 3). In addition, there was a gene × treatment interaction showing that the FO-specific decrease was limited to the “d5” genotype group and was not seen in the “dd” or “55” genotypes (Table 3). TG are primarily carried by VLDL particles in fasting plasma. The change in the total VLDL TG concentration had a similar pattern, but the gene × treatment interaction was of marginal significance (P = 0.067). However, there was a significant gene × treatment interaction for the medium VLDL TG concentration in that the FO-induced decrease (relative to PL) was significant in the “d5” but not the “dd” or “55” genotypes. A significant gene × treatment interaction was also seen for the FO-specific change in the medium VLDL particle concentration, though the post hoc comparison of means did not identify significant differences among the 3 genotype groups.

TABLE 3.

Change in plasma total lipids by FO and PL TRT over the 6-wk intervention for all participants and grouped by ALOX5 genotype (“dd,” “d5,” and “55”)¹

Variable and TRT	All participants	“dd”	“d5”	“55”	ANOVA P values2
Adjust	GT	TRT	GT x TRT
Participants, n
All	97	24	48	25
FO	49	13	23	13
PL	48	11	25	12
Plasma total lipids, mmol/L
Cholesterol
All	−0.055 ± 0.431	−0.041 ± 0.389	−0.091 ± 0.454	−0.001 ± 0.436	−	0.64	0.091	0.85
FO	−0.137 ± 0.447	−0.125 ± 0.480	−0.198 ± 0.408	−0.041 ± 0.497	+	0.40	0.22	0.65
PL	0.028 ± 0.402	0.058 ± 0.227	0.008 ± 0.480	0.043 ± 0.375
TG
All	−0.096 ± 0.247	−0.166 ± 0.252	−0.061 ± 0.248	−0.097 ± 0.233	−	0.28	0.0033	0.46
FO	−0.181 ± 0.223*	−0.227 ± 0.267	−0.179 ± 0.185*	−0.140 ± 0.246	+	0.64	0.0067	0.048**
PL	−0.010 ± 0.242	−0.093 ± 0.224	0.046 ± 0.254	−0.050 ± 0.219
VLDL TG, mmol/L
Total
All	−0.094 ± 0.248	−0.168 ± 0.243	−0.051 ± 0.249	−0.106 ± 0.243	−	0.19	0.011	0.49
FO	−0.170 ± 0.222*	−0.222 ± 0.257	−0.157 ± 0.190	−0.141 ± 0.246	+	0.48	0.020	0.067
PL	−0.017 ± 0.252	−0.105 ± 0.221	0.047 ± 0.259	−0.068 ± 0.245
Large
All	−0.032 ± 0.119	−0.058 ± 0.145	−0.040 ± 0.090	0.008 ± 0.134	−	0.36	0.0050	0.56
FO	−0.051 ± 0.137*	−0.091 ± 0.189	−0.064 ± 0.070	0.013 ± 0.156	+	0.30	0.0052	0.19
PL	−0.013 ± 0.093	−0.019 ± 0.049	−0.018 ± 0.102	0.003 ± 0.111
Medium
All	−0.050 ± 0.185	−0.076 ± 0.198	−0.016 ± 0.191	−0.091 ± 0.150	−	0.43	0.21	0.37
FO	−0.083 ± 0.165	−0.076 ± 0.230	−0.074 ± 0.145*	−0.106 ± 0.129	+	0.61	0.18	0.022**
PL	−0.017 ± 0.199	−0.076 ± 0.164	0.036 ± 0.214	−0.075 ± 0.175
Small
All	−0.012 ± 0.062	−0.034 ± 0.062a	0.005 ± 0.055b	−0.023 ± 0.067a,b	−	0.0203	0.0001	0.97
FO	−0.036 ± 0.060*	−0.054 ± 0.058	−0.020 ± 0.049	−0.048 ± 0.075	+	0.20	0.0014	0.88
PL	0.014 ± 0.054	−0.010 ± 0.061	0.029 ± 0.050	0.004 ± 0.046
VLDL particles, nmol/L
Total
All	−5.80 ± 19.21	−12.34 ± 20.28a	−0.75 ± 17.70b	−10.52 ± 18.33a,b	−	0.012#x2020	0.0006	0.77
FO	−12.94 ± 15.38*	−16.85 ± 20.93	−8.42 ± 10.54	−17.03 ± 15.17	+	0.096	0.0030	0.28
PL	1.49 ± 20.11	−7.02 ± 19.04	7.60 ± 19.59	−3.46 ± 19.43
Large
All	−0.30 ± 1.16	−0.60 ± 1.69	−0.37 ± 0.86	0.13 ± 0.93	−	0.53	0.11	0.83
FO	−0.41 ± 1.35*	−0.92 ± 2.26	−0.40 ± 0.72	0.076 ± 0.84	+	0.28	0.049	0.33
PL	−0.18 ± 0.92	−0.22 ± 0.39	−0.35 ± 0.99	0.20 ± 1.05
Medium
All	−3.11 ± 11.40	−5.07 ± 12.49	−0.66 ± 11.18	−5.95 ± 10.05	−	0.31	0.23	0.47
FO	−5.16 ± 9.85	−5.18 ± 14.24	−4.03 ± 7.97	−7.14 ± 7.90	+	0.48	0.21	0.034**
PL	−1.02 ± 12.55	−4.94 ± 10.76	2.45 ± 12.87	−4.66 ± 12.21
Small
All	−2.39 ± 12.29	−6.67 ± 12.58a	0.95 ± 10.88b	−4.70 ± 13.23a,b	−	0.024**	< 0.0001	0.94
FO	−7.37 ± 11.87*	−10.74 ± 11.93	−3.99 ± 9.62	−9.97 ± 14.48	+	0.27	0.0012	0.81
PL	2.69 ± 10.62	−1.85 ± 12.09	5.50 ± 10.09	1.00 ± 9.22
LDL cholesterol, mmol/L
Total
All	0.00 ± 0.346	0.056 ± 0.337	−0.066 ± 0.350	0.073 ± 0.335	−	0.16	0.74	0.87
FO	−0.012 ± 0.360	0.010 ± 0.417	−0.084 ± 0.331	0.094 ± 0.348	+	0.12	0.98	0.91
PL	0.012 ± 0.335	0.111 ± 0.218	−0.049 ± 0.373	0.050 ± 0.336
Intermediate
All	−0.003 ± 0.096	−0.005 ± 0.120	−0.008 ± 0.090	0.011 ± 0.084	−	0.64	0.25	0.50
FO	−0.020 ± 0.066	−0.015 ± 0.096	−0.027 ± 0.061	−0.015 ± 0.034	+	0.29	0.57	0.25
PL	0.016 ± 0.117	0.007 ± 0.147	0.009 ± 0.109	0.038 ± 0.112
Large
All	−0.057 ± 0.419	0.023 ± 0.435	−0.157 ± 0.383	0.059 ± 0.438	−	0.056	0.28	0.78
FO	−0.096 ± 0.425	−0.063 ± 0.455	−0.193 ± 0.372	0.044 ± 0.470	+	0.14	0.52	0.76
PL	−0.018 ± 0.413	0.125 ± 0.406	−0.125 ± 0.398	0.075 ± 0.421
Small
All	0.060 ± 0.325	0.038 ± 0.313	0.100 ± 0.337	0.004 ± 0.312	−	0.42	0.15	0.93
FO	0.104 ± 0.331	0.088 ± 0.285	0.136 ± 0.357	0.065 ± 0.345	+	0.59	0.28	0.94
PL	0.014 ± 0.314	−0.022 ± 0.347	0.067 ± 0.321	−0.063 ± 0.271
LDL particles, nmol/L
Total
All	12.85 ± 172	18.85 ± 178	6.42 ± 182	19.46 ± 153	−	0.94	0.83	0.80
FO	16.22 ± 176	10.22 ± 196	5.57 ± 180	41.08 ± 161	+	0.60	0.92	0.91
PL	9.41 ± 169	29.04 ± 162	7.20 ± 187	−3.98 ± 147
Intermediate
All	−0.81 ± 29.91	−1.46 ± 37.31	−2.62 ± 28.03	3.29 ± 26.09	−	0.64	0.25	0.50
FO	−6.40 ± 20.51	−4.64 ± 29.93	−8.42 ± 18.93	−4.56 ± 10.72	+	0.29	0.57	0.25
PL	4.89 ± 36.50	2.29 ± 45.81	2.73 ± 33.89	11.79 ± 34.76
Large
All	−16.38 ± 122	6.54 ± 125	−44.65 ± 112	15.92 ± 128	−	0.068	0.29	0.75
FO	−26.91 ± 123	−19.23 ± 133	−52.46 ± 109	10.62 ± 136	+	0.18	0.53	0.78
PL	−5.62 ± 120	36.99 ± 114	−37.47 ± 117	21.67 ± 124
Small
All	30.04 ± 200	13.77 ± 197	53.69 ± 208	0.25 ± 191	−	0.48	0.28	0.89
FO	49.53 ± 199	34.10 ± 172	66.45 ± 215	35.03 ± 205	+	0.64	0.49	0.93
PL	10.14 ± 202	−10.24 ± 228	41.95 ± 205	−37.43 ± 176
HDL cholesterol, mmol/L
Total
All	0.012 ± 0.158	−0.005 ± 0.139	−0.019 ± 0.175	−0.005 ± 0.146	−	0.89	0.35	0.81
FO	−0.031 ± 0.176	−0.007 ± 0.163	−0.048 ± 0.198	−0.023 ± 0.155	+	0.99	0.52	0.93
PL	0.006 ± 0.136	−0.004 ± 0.111	0.007 ± 0.150	0.014 ± 0.139
Large
All	0.019 ± 0.157	0.021 ± 0.126	0.005 ± 0.171	0.044 ± 0.158	−	0.65	0.44	0.79
FO	0.033 ± 0.179	0.019 ± 0.158	0.020 ± 0.192	0.071 ± 0.183	+	0.81	0.22	0.39
PL	0.005 ± 0.132	0.025 ± 0.083	−0.008 ± 0.153	0.014 ± 0.127
Medium
All	−0.023 ± 0.107	−0.021 ± 0.140	−0.031 ± 0.084	−0.010 ± 0.113	−	0.62	0.53	0.37
FO	−0.022 ± 0.116	−0.023 ± 0.170	−0.020 ± 0.056	−0.025 ± 0.139	+	0.71	0.67	0.017#x2020
PL	−0.024 ± 0.098	−0.018 ± 0.104	−0.041 ± 0.104	0.006 ± 0.080
Small
All	−0.008 ± 0.103	−0.006 ± 0.116	0.006 ± 0.098	−0.039 ± 0.096	−	0.15	0.0037	0.088
FO	−0.041 ± 0.101*	−0.002 ± 0.141	−0.048 ± 0.075	−0.069 ± 0.090	+	0.093	0.0020	0.12
PL	0.025 ± 0.094	−0.010 ± 0.084	0.056 ± 0.091	−0.006 ± 0.094
HDL cholesterol, % of total cholesterol
All	0.18 ± 3.59	0.23 ± 3.85	0.40 ± 3.54	−0.30 ± 3.51	−	0.73	0.43	0.84
FO	0.45 ± 4.08	0.82 ± 4.92	0.63 ± 3.79	−0.24 ± 3.92	+	0.62	0.41	0.91
PL	−0.10 ± 3.02	−0.48 ± 2.00	0.19 ± 3.37	−0.36 ± 3.19
HDL particles, nmol/L
Total
All	−0.86 ± 3.39	−0.68 ± 3.20	−0.63 ± 3.65	−1.47 ± 3.11	−	0.62	0.0049	0.084
FO	−1.99 ± 3.54*	−0.56 ± 3.68a	−2.28 ± 3.61*a,b	−2.90 ± 3.07*b	+	0.32	0.0084	0.014#x2020
PL	0.29 ± 2.83	−0.83 ± 2.68	0.89 ± 3.02	0.080 ± 2.39
Large
All	0.19 ± 2.00	0.24 ± 2.12	0.066 ± 2.08	0.38 ± 1.81	−	0.80	0.29	0.93
FO	0.35 ± 2.23	0.32 ± 2.77	0.21 ± 2.04	0.61 ± 2.13	+	0.94	0.22	0.60
PL	0.034 ± 1.75	0.16 ± 1.03	−0.068 ± 2.14	0.13 ± 1.43
Medium
All	−0.68 ± 3.19	−0.63 ± 4.19	−0.92 ± 2.51	−0.29 ± 3.38	−	0.62	0.53	0.37
FO	−0.66 ± 3.46	−0.70 ± 5.06	−0.60 ± 1.67	−0.73 ± 4.15	+	0.71	0.67	0.017#x2020
PL	−0.71 ± 2.93	−0.54 ± 3.09	−1.21 ± 3.10	0.18 ± 2.39
Small
All	−0.37 ± 4.02	−0.30 ± 4.50	0.22 ± 3.84	−1.56 ± 3.76	−	0.21	0.0094	0.073
FO	−1.67 ± 3.94*	−0.18 ± 5.46	−1.89 ± 2.95	−2.78 ± 3.54	+	0.14	0.0038	0.068
PL	0.97 ± 3.68	−0.44 ± 3.27	2.17 ± 3.57	−0.24 ± 3.67
Mean particle diameter, nm
VLDL
All	0.40 ± 14.30	−0.092 ± 11.21	−1.32 ± 11.31	4.17 ± 20.65	−	0.38	0.72	0.38
FO	1.72 ± 16.81	2.47 ± 13.30	−2.97 ± 7.40	9.28 ± 27.35	+	0.25	1.00	0.28
PL	−0.95 ± 11.20	−3.12 ± 7.61	0.20 ± 13.97	−1.37 ± 7.07
LDL
All	−0.047 ± 0.47	−0.010 ± 0.47	−0.12 ± 0.45	0.067 ± 0.49	−	0.21	0.36	0.95
FO	−0.093 ± 0.45	−0.042 ± 0.33	−0.19 ± 0.48	0.032 ± 0.51	+	0.37	0.52	0.95
PL	0.000 ± 0.48	0.026 ± 0.61	−0.062 ± 0.42	0.11 ± 0.48
HDL
All	0.034 ± 0.17	0.037 ± 0.14	0.014 ± 0.16	0.072 ± 0.19	−	0.41	0.043	0.17
FO	0.072 ± 0.17*	0.026 ± 0.14	0.057 ± 0.18	0.14 ± 0.19	+	0.60	0.024	0.14
PL	−0.004 ± 0.15	0.049 ± 0.14	−0.026 ± 0.14	−0.007 ± 0.17

¹Values are mean ± SD. Means for a given genotype within the same group (All, FO, or PL) without a common superscript differ. *Different between treatment within the same group.

**P value was no longer significant after adjustment for multiple comparisons. †P value remained significant after adjustment for multiple comparisons. Mean ± SD values are not adjusted. FO, fish oil; GLM, generalized linear model; GT, genotype; PL, placebo; TRT, treatment group.

²Statistical analysis (2-way ANOVA using a GLM) was performed without adjustment (indicated by a minus sign in the “Adjust” column) and with adjustment (indicated by a plus sign in the “Adjust” column) for body weight, baseline concentration of the fatty acid of interest (using ranked data, excepting DHA, which were normally distributed at baseline), and rank of percent compliance. The values for the 3 analyses indicate differences between TRT, among GT, or an interaction between the 2 (GT × TRT).

The concentration of total HDL particles decreased significantly with the FO treatment (relative to PL) in the “55” and “d5” genotype groups, whereas no significant FO-specific decrease was seen in the “dd” genotype group (Table 3). Although there was an apparent decrease in total HDL particle concentration in the “dd” participants receiving FO (though, as indicated above, it did not differ from PL), the magnitude of this decrease was significantly smaller than the change in the “55” participants receiving FO. (The corresponding change in the “d5” FO participants did not differ from either the “dd” or “d5” participants.) This gene × treatment interaction pattern was also seen in the FO-specific changes in medium HDL particle (P = 0.017) and cholesterol (P = 0.017) concentrations, though post hoc comparisons among genotypes were not significant in either case.

Small HDL particle and cholesterol concentrations both decreased with FO treatment (relative to PL). As a result of these changes, the overall mean diameter of HDL particles increased with FO treatment (relative to PL) (Table 3).

Discussion

Previous studies have shown that men and women with the “dd” ALOX5 genotype have a higher CVD risk than do people with at least one common “5” allele (hence the characterization of the “d” allele as “high risk”) and that this risk of CVD may be diminished for the “dd” genotype with high intake of long-chain (n-3) PUFA such as EPA and DHA (8, 9). The reasons for the underlying risk and its association with (n-3) PUFA intake are not known. Our hypothesis in planning this study was that ALOX5 gene expression would vary by genotype and that the resulting differential production of proinflammatory AA-derived LT, such as LTB4 production by monocytes and macrophages, could affect the development of CVD. Further, we postulated that increasing the intake of EPA would increase cellular levels of EPA, which would then compete with AA as a substrate for LT production by 5-LO and that differential effects would be seen by genotype in that individuals with higher basal LT production (due to genotype) might have a greater antiinflammatory benefit from increased EPA intake. However, as we previously reported, ALOX5 mRNA expression, protein levels, and LTB4 production did not vary by genotype in purified monocytes from these participants (19, 21) although mRNA expression was higher in lymphocytes from “dd” and”d5” participants compared with the “55” homozygotes (21).

Although differential ALOX5 gene expression and LTB4 production were not found in this study, we did report (19) lower baseline production of ALOX5 metabolites by monocytes, including AA-derived 5-hydroxyeicosatetraenoic acid, in the “dd” and “d5” compared with the “55” genotypes. In addition, a gene × treatment interaction was observed following FO supplementation. The “dd” and “d5” genotypes had significantly smaller increases in the production of the EPA-derived ALOX5 metabolite 5-hydroxyeicosapentaenoic acid (HEPE) by monocytes than did participants with the “55” genotype. In addition, no increase in production of the EPA-derived 15-lipoxygenase metabolite 15-HEPE by monocytes was seen in the “dd” genotype, whereas there were significant increases in the “d5” and “55” genotypes.

In the work reported here, we evaluated changes in the EPA, DHA, and total (n-3) PUFA content of erythrocyte membranes as an indicator of response to FO treatment. A significant increase was seen overall (relative to PL) but a gene × treatment interaction was also observed such that significant increases in EPA, DHA, and total (n-3) PUFA were limited to the “d5” and “55” genotypes. Thus, EPA, DHA, and total (n-3) PUFA did not increase (relative to PL) in the high-risk “dd” genotype. This finding, at least for the “dd” and “55” genotypes, is consistent with the smaller increase in EPA-derived 5-HEPE and 15-HEPE seen previously with FO supplements (relative to PL) in the “dd” compared with the “55” genotype (19), suggesting that the differential production of EPA-derived oxylipid metabolites may have resulted at least in part from differential incorporation of EPA into cellular membranes from participants with the different genotypes. How such differential incorporation (or retention) might result from a variant in the ALOX5 promoter is not clear. It is possible, however, that an unrecognized (genetic) factor could have resulted in selection for both the “d” variant and for incorporation of less EPA. Interestingly, baseline erythrocyte EPA levels were also associated with genotype such that “dd” had the lowest and “55” the highest levels, which argues for the existence of differential EPA incorporation preceding the intervention. Further work is needed to confirm that people with these genotypes have differential incorporation or retention of EPA and DHA. If confirmed, this could have implications for recommendations regarding the benefits of FO supplementation in people carrying ALOX5 deletion alleles.

A principal goal of our study, not specifically related to these ALOX5 genotypes, was to determine if FO supplementation would reduce plasma TG concentrations in African Americans, who have a higher risk for CVD than European Americans (16) despite having lower fasting plasma TG concentrations (17). As expected, FO supplementation decreased plasma TG concentrations by ~20%. Our results differ from those of a South African study among Africans that used a dose-escalation design and administered up to a 4-fold higher dose of EPA and DHA than the present study but did not find a decrease in plasma TG concentrations (23). Our results are consistent with those of a number of other studies with FO treatment (13, 22) or with purified EPA and DHA (23–30) in participants primarily of European ancestry. Reduction in plasma TG in response to (n-3) PUFA has been reported in both fasting and postprandial TG in participants with normal and elevated TG (13, 31). The decrease in total plasma TG concentration in our study was accompanied by a concomitant decrease in VLDL particle concentration, also consistent with other studies (32).

Small VLDL particle and TG concentrations were higher in the “dd” genotype than in the “d5” genotype at baseline, with the “55” genotype being intermediate. This difference could be due to an underlying genotype effect, though we would have predicted an allelic dose-response effect that is not evident here. In addition, small VLDL particle and TG concentrations tended to decrease by 6 wk to a greater degree in the “dd” than in the “d5” genotype, with the “55” genotype being intermediate (regardless of treatment group). This pattern was also seen with the change in total VLDL particle concentrations. Thus, the baseline difference was minimized by 6 wk, suggesting that the difference at baseline may have been a chance observation.

Total and LDL cholesterol concentrations did not change with FO treatment in the present study, which is consistent with previous reports (24, 25, 27, 29, 33–35). However, the concentration of total HDL particles, though not total HDL cholesterol, was significantly reduced by FO treatment, which differs from previous studies that have shown small increases or no effect with FO (26, 29, 34, 35). HDL particles are thought to protect against CVD by several mechanisms, including reverse cholesterol transport, antioxidant, and antiinflammatory effects (36). In the present study, HDL particle concentrations decreased as a result of FO treatment only in the “d5” and “55” genotypes and not in the high-risk “dd” genotype. Lowering HDL particle concentrations could increase the risk of CVD, and the lack of decrease in HDL particle concentrations could be considered a relative “benefit” for the “dd” genotype in that HDL levels were maintained rather than reduced. The principal reason for a decrease in the total HDL particle concentration in this study was a decrease in the small HDL particle concentration. HDL particle size may affect CVD risk, with some studies indicating that small HDL particles are relatively more protective than large particles (18, 20), whereas others suggest the opposite (37, 38). Functional characteristics of HDL particles may also vary independent of size (36). Thus, although the present study shows a significant gene effect on change in the HDL particle concentration in response to FO treatment, it is difficult to predict how CVD risk would be affected by this interaction. Confirming this effect in a subsequent study would be a first step in answering this question.

Heart rate and blood pressure were examined as well in the present study, but neither was affected by the FO intervention or genotype. A meta-analysis of controlled intervention trials has concluded that FO intake reduces heart rate by ~1.6 bpm (39), with the greatest impact in participants with higher heart rates at baseline and in studies of longer duration. The lack of effect of the intervention on heart rate in the present study was likely due to the relatively short duration of the intervention, the relatively small sample size given the expected magnitude of the effect, and/or the fact that our participants were healthy adults without a history of CVD. With regard to blood pressure, 2 meta-analyses (40, 41) have concluded that relatively high intakes of (n-3) fatty acids, >3 g/d, can decrease blood pressure in men and women with high blood pressure at baseline, but perhaps not in normotensive adults, where negative results have been reported (42). Thus, the lack of effect of the intervention on blood pressure in the present study is consistent with these results in that individuals with hypertension were excluded and the dose of FO was lower than doses that have proven effective in prior studies.

In summary, the TG-lowering effect of FO in this study helps to confirm that current recommendations (2), which are based largely on data from other ancestry groups, are also apparently applicable to those of African ancestry. However, people of African ancestry have a 10-fold higher frequency of one of the ALOX5 “d” alleles (the “3” allele) than do Asians or Europeans (8). This may have implications for the efficacy of FO supplements, because participants in the high-risk “dd” genotype did not have a significant increase in erythrocyte EPA, DHA, and total (n-3) PUFA, nor did they have a significant decrease in the (n-6) PUFA:(n-3) PUFA ratio in response to FO supplementation. Differential effects on plasma lipid and lipoprotein profiles were also seen by genotype. Further work is needed to reproduce these findings and define potential mechanisms to determine their significance with regard to CVD risk and any possible impact on recommendations for FO supplementation.