n-3 and n-6 Fatty acids are independently associated with lipoprotein-associated phospholipase A₂ in the Multi-Ethnic Study of Atherosclerosis

Abstract

Lipoprotein-associated phospholipase A₂ (Lp-PLA₂) is an independent risk factor for CVD and has been proposed as a marker of vascular inflammation. Polyunsaturated n-3 fatty acids (FA) and several n-6 FA are known to suppress inflammation and may influence Lp-PLA₂ mass and activity. The associations of n-3 and n-6 plasma FA with Lp-PLA₂ mass and activity were analysed using linear regression analysis in 2246 participants of the Multi-Ethnic Study of Atherosclerosis; statistical adjustments were made to control for body mass, inflammation, lipids, diabetes, and additional clinical and demographic factors. Lp-PLA₂ mass and activity were significantly lower in participants with the higher n-3 FA EPA (β = − 4·72, P< 0·001; β = − 1·53; P= 0·023) and DHA levels (β = − 4·47, β = − 1·87; both P< 0·001). Those in the highest quintiles of plasma EPA and DHA showed 12·71 and 19·15 ng/ml lower Lp-PLA₂ mass and 5·7 and 8·90 nmol/min per ml lower Lp-PLA₂ activity than those in the first quintiles, respectively. In addition, lower Lp-PLA₂ mass and activity were associated with higher levels of n-6 arachidonic acid (β = − 1·63, β = − 1·30; both P< 0·001), while γ-linolenic acid was negatively associated with activity (β = − 27·7, P= 0·027). Lp-PLA₂ mass was significantly higher in participants with greater plasma levels of n-6 linoleic (β = 0·828, P= 0·011) and dihomo-γ-linolenic acids (β = 4·17, P= 0·002). Based on their independent associations with Lp-PLA₂ mass and activity, certain n-3 and n-6 FA may have additional influences on CVD risk. Intervention studies are warranted to assess whether these macronutrients may directly influence Lp-PLA₂ expression or activity.

Lipoprotein-associated phospholipase A₂ (Lp-PLA₂) is a macrophage-derived enzyme that contributes to oxidative stress, vascular inflammation and endothelial activation^{(1, 2)} – hallmarks of atherosclerosis and CVD. Primarily associated with LDL⁽³⁾ and co-localising at sites of atherosclerotic plaques, the Lp-PLA₂ enzyme has been reported as a crucial component in plaque destabilisation and disruption⁽⁴⁾. Indeed, elevated levels of Lp-PLA₂ mass are considered as independent risk factor for CVD⁽²⁾, and high levels of Lp-PLA₂ enzymatic activity have been associated with unfavourable health outcomes – including an increased risk for myocardial infarction and CVD death in older adults⁽⁵⁾. Though a number of drug treatments including specific Lp-PLA₂ inhibitors are undergoing clinical trials⁽⁴⁾, a non-pharmaceutical-based approach such as an increased intake of certain macronutrients may provide a simple, cost-effective alternative without appreciable side effects.

Polyunsaturated n-3 fatty acids (FA) have been shown as important contributors to health outcomes and are widely acknowledged for influencing disease risk⁽⁶⁾. First recognised for their health benefits by Dyerberg et al. ⁽⁷⁾, the fish oil n-3 FA EPA and DHA may reduce atherogenic burden and the risk of CVD development by suppressing inflammation and beneficially influencing lipid profile⁽⁶⁾. Thus far, the evidence that n-3 FA influence Lp-PLA₂ is equivocal, with two interventional studies showing no effect^{(8, 9)} and three demonstrating that EPA or EPA/DHA esters reduce the levels of Lp-PLA₂^(10–12). In contrast to the n-3 FA, no study has examined any potential impact or association of plasma n-6 FA with Lp-PLA₂. By examining levels of Lp-PLA₂ mass/activity, n-3 and n-6 FA across a larger study population, the present analysis aims to provide evidence of an additional benefit of PUFA that is not yet established.

Few studies have examined Lp-PLA₂ mass and activity in the context of n-3 FA, and to our knowledge no study has examined Lp-PLA₂ and polyunsaturated n-6 FA. The present cross-sectional analysis aimed to determine the relationship of plasma n-3 and n-6 FA levels with Lp-PLA₂ mass and activity in 2246 generally healthy participants enrolled in the Multi-Ethnic Study of Atherosclerosis (MESA). No dietary intervention was used in the present study. To avoid the inherent difficulties in accurately assessing individual FA with dietary questionnaires, polyunsaturated n-3 and n-6 FA were directly measured in the phospholipid fraction of plasma, which has been demonstrated to strongly associate with dietary intake⁽¹³⁾ and erythrocyte cell membrane composition⁽¹⁴⁾. Relative levels of the following plasma PUFA were analysed: linoleic acid (LA, 18 : 2n-6); γ-linolenic acid (18 : 3n-6); dihomo-γ-linolenic acid (20 : 3n-6); arachidonic acid (AA, 20 : 4n-6); α-linolenic acid (18 : 3n-3); EPA (20 : 5n-3); DHA (22 : 6n-3).

Materials and methods

Population

The MESA was initiated to investigate the prevalence, correlates and progression of subclinical CVD⁽¹⁵⁾, and further information about the MESA protocol is available online (http://www.mesa-nhlbi.org). Briefly, the MESA comprises 6814 men and women, 38·6 % White, 27·6 % Black, 11·8 % Chinese and 22·0 % Hispanic, who were 45–84 years of age and free of clinical CVD at baseline, July 2000–August 2002. Baseline examinations included anthropometry, medical and lifestyle histories and blood collection. The present analysis consists of a random sample of 2246 adults represented by approximately equal numbers of Black (n 534), Asian (of Chinese descent n 604), Hispanic (n 572) and White (n 536) participants. The present study was conducted according to the guidelines laid down in the Declaration of Helsinki, and all procedures involving human subjects were approved by the institutional review boards for research involving human subjects of the respective field centres. Written informed consent was obtained from all subjects.

Measurements

Questionnaire information was obtained regarding age, sex, race/ethnicity, education and lifestyle factors including smoking status, drinking alcohol and physical activity. Height (cm) and weight (kg) were measured according to standard procedures⁽¹⁵⁾. Fasting blood was drawn and serum and EDTA-anticoagulant tubes were collected and processed using a standardised protocol⁽¹⁵⁾. Serum and plasma samples were aliquoted and stored at − 70°C until time of use.

Plasma fatty acid profile

Phospholipid FA were extracted from EDTA plasma using the method described previously by Cao et al. ⁽¹⁶⁾. In brief, lipids were extracted from the plasma using a chloroform–methanol extraction method, and cholesterol esters, TAG, phospholipids and NEFA were separated by TLC. FA from the phospholipids were derivatised to methyl esters and detected by GC flame ionisation. The FA detected were expressed as a percentage of total FA. The following representative coefficients of variance were obtained from intra-laboratory quality-control testing (n 20): LA, 2·6 %; AA, 2·4 %; α-linolenic acid, 2·4 %; EPA, 3·3 %; DHA, 2·7 %.

Lipoprotein-associated phospholipase A₂ mass and activity

Blood collection and laboratory procedures have been described previously by Cushman et al. ⁽¹⁷⁾. Lp-PLA₂ activity was measured by radiometric assay using a ³H-labelled platelet-activating factor substrate, analytical CV 7·5 % (GlaxoSmithKline), and expressed as nmol platelet-activating factor hydrolysed/min per ml plasma sample (nmol/min per ml). Lp-PLA₂ mass was measured using the PLAC^® Test, analytical CV 6·3 % (Diadexus). All measurements were made on baseline blood samples not previously thawed.

Statistical methods

SAS, version 9.3 (SAS Institute) was used to analyse the data. Baseline characteristics are reported as means and standard deviations for continuous variables and frequency (%) for categorical variables. Levels of analytes with skewed distributions were log-transformed before analysis and results were back-transformed and presented as geometric means. Generalised linear regression analysis evaluated the associations between Lp-PLA₂ and plasma FA, adjusting for potential confounding factors including, age, sex, race/ethnicity, BMI, education, field centre, current smoking, alcohol use, HDL-cholesterol, LDL-cholesterol, high-sensitivity C-reactive protein (hs-CRP) level, TAG level, the presence of diabetes, and use of statins, fibrates or aspirin. Diet was not adjusted for, as phospholipid FA strongly correlate with this variable. Weight categories were defined as normal weight (BMI < 25 kg/m²), overweight (BMI 25– < 30 kg/m²) and obese (BMI 30+ kg/m²). TAG and hs-CRP levels were divided by risk categories according to the American Heart Association guidelines: hs-CRP – low risk ( < 1·0 mg/l), average risk (1–3 mg/l), high risk (>3·0 mg/l); TAG levels – normal ( < 1500 mg/l), borderline high (1500–1990 mg/l), high (2000–4990 mg/l) and very high (>5000 mg/l). Correlations are expressed as β-coefficients, where one unit change in plasma phospholipid FA (%) is equal to the indicated change in Lp-PLA₂ mass or activity, and observations were deemed significant with a P value ≤ 0·05.

Results

Unadjusted mean levels of Lp-PLA₂ mass and activity by demographic characteristics are shown in Table 1. Significant differences in Lp-PLA₂ mass and activity were observed between sexes as well as among races/ethnicities. Females had lower levels of both Lp-PLA₂ mass and activity compared with males (P _trend< 0·001). With respect to race/ethnicity, Lp-PLA₂ mass was highest in White and Hispanic adults followed by Chinese and Black participants (P _trend< 0·001). Similarly, mean levels of Lp-PLA₂ enzymatic activity was significantly lower in Black participants, but no differences were observed among Hispanic, White and Chinese adults (P _trend< 0·001). An association of Lp-PLA₂ mass with a maximum level of education attained did not reach significance (P _trend= 0·052). No significant differences in Lp-PLA₂ were found among the age groups.

Table 1 Mean lipoprotein-associated phospholipase A₂ (Lp-PLA₂) mass and activity levels* of the Multi-Ethnic Study of Atherosclerosis participants by demographic characteristics and lifestyle factors

HS, high school graduate.

^a,b,c,d,eMean values with unlike superscript letters were significantly different for the intra-group (P< 0·05).

* Unadjusted demographic characteristics.

Lp-PLA₂ mass and activity were significantly correlated (r 0·49, P< 0·001) following adjustment for age, sex, race, education and field centre (data not shown). Differences in both Lp-PLA₂ mass and activity were observed based on lifestyle factors including cigarette smoking and alcohol use (Table 1). Non-smokers demonstrated significantly lower levels of Lp-PLA₂ mass than current smokers (P< 0·001), while those who never consumed alcohol had lower levels of both Lp-PLA₂ mass and activity than those who currently consumed or formerly consumed alcohol (P _trend< 0·001). No differences in the mean levels of Lp-PLA₂ mass or activity were observed among the quartiles of intentional physical activity.

The mean levels of Lp-PLA₂ mass and activity by clinical characteristics are shown in Table 2. Statistically significant trends were found based on BMI as well as TAG and hs-CRP levels. Compared with either overweight or obese individuals, normal-weight individuals had the lowest levels of Lp-PLA₂ activity (P _trend= 0·001) and mass (P= 0·007). Higher levels of Lp-PLA₂ mass were found to correlate with higher levels of TAG (P _trend= 0·002) and hs-CRP (P _trend< 0·001, respectively). Finally, differences were found in Lp-PLA₂ mass and activity based on medication use. Lower mean levels of Lp-PLA₂ mass and activity were observed in individuals taking statins (P _trend< 0·001, P _trend< 0·001), fibrates (P _trend= 0·01, P _trend= 0·04) or postmenopausal therapy (P _trend< 0·001, P _trend< 0·001) compared with those not taking these medicinal regimens. Multiple demographic, lifestyle and clinical characteristics were found to influence Lp-PLA₂ mass and activity, thus additional covariate adjustments were made in subsequent analyses of plasma FA.

Table 2 Lipoprotein-associated phospholipase A₂ (Lp-PLA₂) mass and activity levels* of the Multi-Ethnic Study of Atherosclerosis participants categorised by clinical characteristics

hs-CRP, high-sensitivity C-reactive protein.

^a,b,c,dValues with unlike letters were significantly different for the intra-group (P< 0·05).

* Adjusted for age, sex, race, education and field centre.

Plasma levels of n-3 and n-6 phospholipid FA were found to correlate with both Lp-PLA₂ mass and activity (Table 3). Plasma LA levels positively correlated with both mass (β = 0·83, P= 0·01) and activity (β = 1·07, P< 0·001), while dihomo-γ-linolenic acid was found to be positively correlated with Lp-PLA₂ mass alone (β = 4·17, P= 0·002). In contrast, plasma γ-linolenic acid negatively correlated with Lp-PLA₂ activity (β = − 27·7, P= 0·03), and plasma AA levels negatively correlated with both Lp-PLA₂ mass (β = − 1·63, P< 0·001) and activity (β = − 1·30, P< 0·001). Similarly, plasma n-3 FA levels of EPA and DHA negatively correlated with both mass (β = − 4·90, P< 0·001; β = − 4·99, P< 0·001) and activity (β = − 1·53, P= 0·02; β = − 1·87, P< 0·001), respectively. Plasma α-linolenic acid was not observed to correlate with either Lp-PLA₂ mass (P= 0·25) or activity (P= 0·21).

Table 3 Correlations of plasma phospholipid fatty acids with lipoprotein-associated phospholipase A₂ (Lp-PLA₂) mass and activity (β-Coefficients)

* Adjusted for age, sex, race, education and field centre.

† Model 1 plus adjustments for smoking, alcohol, BMI, HDL-cholesterol, LDL-cholesterol, TAG, high-sensitivity C-reactive protein, diabetes and use of statins, fibrates and aspirin.

n-6 FA levels have not been examined in the context of Lp-PLA₂ mass and activity, and further statistical analysis is therefore warranted. Mean levels of Lp-PLA₂ mass and activity were examined by quintiles of LA and AA – the most abundant n-6 FA in a typical Western diet. Differences were found in both mass and activity among the quintiles of LA and AA, and remained significant following adjustment for multiple covariates including inflammation (hs-CRP) and standard lipid measures (TAG, HDL-C and LDL-C) as well as demographic and lifestyle factors (Fig. 1(a) and (b)). Unexpectedly, LA and AA were differentially associated with Lp-PLA₂ mass and activity. Individuals with the highest plasma LA levels (fifth quintile) showed an 8·09 ng/ml higher mean level of Lp-PLA₂ mass (P= 0·017) and a 10·5 nmol/min per ml higher mean level of Lp-PLA₂ activity (P< 0·001) compared with those with the lowest levels of plasma LA (first quintile). In contrast, individuals with the highest levels of plasma AA were found to have a 9·75 ng/ml lower mean level of Lp-PLA₂ mass (P= 0·005) and an 8·91 nmol/min per ml lower mean level of Lp-PLA₂ activity (P< 0·001) compared to those in the first quintile of AA.

Fig. 1 Mean levels of lipoprotein-associated phospholipase A₂ (Lp-PLA₂) (a) mass and (b) activity are shown by quintiles of plasma arachidonic acid () and linoleic acid (). Values are means, with their standard errors represented by vertical bars. Values were adjusted for age, sex, race, education, field centre, smoking, alcohol, BMI, HDL-cholesterol, LDL-cholesterol, TAG, high-sensitivity C-reactive protein and use of statins, fibrates and aspirin. * Mean values were significantly different from those of quintile 1 (P< 0·05).

With respect to the n-3 FA, plasma EPA and DHA have been well characterised for cardiovascular benefits and may be increased with fish oil supplements and/or greater fish consumption. Expectedly, differences were found in both mass and activity among the quintiles of EPA and DHA, and remained significant following adjustment for multiple covariates including systemic inflammation (hs-CRP) and standard lipid measures (TAG, HDL-C and LDL-C) as well as demographic and lifestyle factors (Fig. 2(a) and (b)). Compared with those in the first quintile for plasma EPA, individuals in the fifth quintile showed a 12·71 ng/ml lower mean level of Lp-PLA₂ mass (P< 0·001) and a 5·7 nmol/min per ml lower mean level of Lp-PLA₂ activity (P= 0·03). Similarly, individuals in the fifth plasma DHA quintile were found to have a 19·15 ng/ml lower mean level of Lp-PLA₂ mass (P< 0·001) and an 8·90 nmol/min per ml lower mean level of Lp-PLA₂ activity (P< 0·001) compared with those in the first quintile.

Fig. 2 Mean levels of lipoprotein-associated phospholipase A₂ (Lp-PLA₂) (a) mass and (b) activity are shown by quintiles of plasma EPA and DHA. Values are means, with their standard errors represented by vertical bars. Values were adjusted for age, sex, race, education, field centre, smoking, alcohol, BMI, HDL-cholesterol, LDL-cholesterol, TAG, high-sensitivity C-reactive protein and use of statins, fibrates and aspirin. * Mean values were significantly different from those of quintile 1 (P< 0·05).

Discussion

The present cross-sectional analysis showed independent associations of plasma n-3 and n-6 FA levels with both Lp-PLA₂ mass and activity in 2246 adult participants of the MESA. These associations remained significant following multivariate adjustments for traditional CVD risk factors including systemic inflammation, cholesterol and TAG levels as well as additional clinical, demographic and lifestyle factors. Expectedly, Lp-PLA₂ mass and activity varied by race/ethnicity, BMI, sex, hs-CRP and TAG levels, as shown previously^(18–20). In agreement with Hatoum et al. ⁽¹⁹⁾, but in contrast to Brilakis et al. ⁽¹⁸⁾ and Hirschler et al. ⁽²⁰⁾, we found a modest but significant association of Lp-PLA₂ activity with BMI.

Affirmed by a host of prospective studies and a subsequent meta-analysis by the Lp-PLA₂ Studies Collaboration⁽²¹⁾ group, Lp-PLA₂ mass and activity are risk factors for CHD, congestive heart failure and stroke – though only enzyme mass has been recommended as an independent CVD risk factor at the present time. The meta-analysis showed that Lp-PLA₂ mass imparts a slightly higher risk of myocardial infarction and stroke, but not CVD death, than Lp-PLA₂ activity. Revealingly, the relative risk of Lp-PLA₂ mass for CHD following adjustment for Lp-PLA₂ activity remained 1·08 (1·04–1·12) per 1 sd in multivariable analysis⁽²¹⁾. The collective evidence has shown that mass and activity are partially related, and may both be important for assessing risks of future CHD and stroke.

To date, no studies have examined n-6 FA in the context of Lp-PLA₂ mass or activity, and only a few have examined n-3 FA. In agreement with the present findings in the plasma phospholipid fraction, Schmidt et al. ⁽²²⁾ observed that adipose tissue levels of EPA and DHA were inversely associated with Lp-PLA₂ concentrations. Follow-up intervention trials to determine whether EPA or DHA supplementation reduces Lp-PLA₂ mass and/or activity have yielded inconsistent results. For example, two studies have reported no effect of supplementing EPA+DHA on Lp-PLA₂ mass or activity over 6 and 8 weeks^{(8, 9)}. In contrast, Schaefer et al. ⁽¹⁰⁾ found that supplementation with EPA alone reduced Lp-PLA₂ levels by 6 % over 6 weeks in a larger study population. Finally, two clinical trials have demonstrated that administering EPA/DHA ethyl ester in conjunction with simvastatin⁽¹¹⁾ or atorvastatin⁽¹²⁾ lowered Lp-PLA₂ mass to a greater extent than a placebo+statin treatment. Overall, the inverse associations in the present analysis support these latter studies, showing a benefit of the fish oil FA with respect to Lp-PLA₂ – notably, these reductions in Lp-PLA₂ may be explained by a number of biochemical mechanisms.

n-3 FA have been shown to have multiple effects on oxidative stress and inflammatory pathways – several of which overlap with those involved in Lp-PLA₂ expression. Specifically, two classes of EPA and DHA metabolites, resolvins and protectins, have been shown to suppress inflammatory cytokine production, block eicosanoid signalling and reduce oxidative stress⁽²³⁾ – all of which may suppress the activation of signalling elements involved in Lp-PLA₂ induction including p38 mitogen-activated protein kinase and phosphatidylinositol 3-kinase^(24–26). Apart from these metabolites, EPA and DHA have also been shown to disrupt oxidised LDL signalling^{(27, 28)} – also a known inducer of Lp-PLA₂ expression⁽²⁵⁾.

Similar to the observations for n-3 FA, associations among plasma n-6 FA and Lp-PLA₂ may be explained by their influence on inflammation. The positive association of LA with Lp-PLA₂ was expected to be due to its promotion of inflammation and activation of phosphatidylinositol 3-kinase in cell-culture models⁽²⁹⁾. Similarly, dihomo-γ-linolenic acid has been found to associate with inflammatory markers in MESA study participants⁽³⁰⁾, though no direct effects on inflammatory signalling have been reported. In contrast, AA is well characterised for being inflammatory, largely for its metabolism into the inflammatory 4-series leukotrienes; however, AA also generates lipoxins – powerful non-classic eicosanoids that suppress inflammatory signalling pathways⁽³¹⁾. Apart from their active role in resolving inflammation, lipoxins have also been shown to suppress an inducer of Lp-PLA₂ expression, p38 mitogen-activated protein kinase⁽³²⁾. Taken together, there is a considerable body of evidence that n-3 and n-6 FA affect inflammatory signalling cascades, which may in turn influence Lp-PLA₂ expression via phosphatidylinositol 3-kinase and p38 mitogen-activated protein kinase – but further research is needed to determine whether such effects may be involved in the induction or suppression of Lp-PLA₂ expression in vivo.

In terms of implications, the present findings are suggestive that a higher intake of certain n-3 FA may have an additional cardiovascular benefit on Lp-PLA₂, in addition to their well-known influences on blood lipids, inflammation, oxidative stress and arrhythmia. Importantly, the observed associations predict the value of Lp-PLA₂ mass or activity based on a one unit change (%) in the given FA. For FA such as AA, DHA or EPA, a 1 % change in the plasma phospholipid fraction is quite feasible, and thus may have implications for maintaining normal or moderately reducing elevated levels of Lp-PLA₂. Those in the top quintile for EPA and DHA demonstrated modestly lower levels of Lp-PLA₂, but whether 10–15 % lower Lp-PLA₂ levels have long-term health implications is unclear and warrants further exploration. The present findings regarding n-6 FA are perhaps more controversial, as their influence on CVD remains inconclusive and require further study. Notably, any potential effect(s) of n-3 or n-6 FA on Lp-PLA₂ would probably be weaker than pharmaceutical interventions^{(33, 34)}.

The present study has a number of strengths including the direct measurement of the phospholipid fraction of plasma FA in a relatively large population compared with previous studies. Direct measurement avoids the inherent problems of dietary recall, and phospholipid PUFA levels are well documented to be highly correlated with dietary intake⁽¹²⁾. Additionally, numerous demographic, lifestyle and clinical factor adjustments were made to better determine whether PUFA associate with LpPLA2 mass and activity. In terms of limitations, the cross-sectional study design prevents the determination of temporality. Though multiple statistical adjustments were made within the present analysis, the potential for other confounding variables remains. In addition, the MESA is a relatively healthy prospective study population, and it must be acknowledged that present observations were limited by a relatively few individuals with levels of Lp-PLA₂ mass considered to be of moderate (200–235 ng/ml) or high risk for CVD (>235 ng/ml) – indeed, stronger associations may be observed in a study population with higher Lp-PLA₂ levels. Finally, it must be recognised that a number of the observed associations were relatively modest in nature. Though associations of LA and AA with Lp-PLA₂ mass and activity reached significance, their potential contribution in driving expression or activity is probably limited or negligible.

In conclusion, the observed associations of Lp-PLA₂ mass and activity with EPA, DHA and certain n-6 FA may indicate additional health benefits of these FA that have not been established. Future dietary intervention studies as well as cell-culture experiments may better evaluate whether FA directly influence Lp-PLA₂ expression or enzymatic activity. Given that higher Lp-PLA₂ levels increase the risk of CVD incidence and destabilising atherosclerotic plaques, strategies should be explored that promote and/or maintain enzyme levels considered low risk, < 200 ng/ml. Though specific enzyme inhibitors, statins and fibrates, are being tested in clinical trials to this end, a non-pharmaceutical-based approach such as long-term nutritional support with EPA/DHA warrants further study.