The presence of long-chain n-3 fatty acids such as EPA (20 : 5n-3) and DHA (22 : 6n-3) in the diet is associated with a lower risk of CHD(Reference Wang, Harris and Chung1). The Diet and Reinfarction Trial (DART)(Reference Burr, Fehily and Gilbert2), Gruppo Italiano per lo Studio della Sopravvivenza nell'Infarto Miocardico (GISSI)-Prevenzione (Italian group for the study of the survival of myocardial infarction)(Reference Marchioli, Barzi and Bomba3) and the Japan EPA Lipid Intervention Study (JELIS)(Reference Yokoyama, Origasa and Matsuzaki4) have shown that oily fish, EPA and DHA, and EPA alone significantly reduce cardiac death, but do not significantly affect the risk of non-fatal CHD. The association between EPA and DHA and the risk of non-fatal myocardial infarction (MI) is, however, unclear. Higher fish intake is significantly associated with reduced risk for non-fatal MI(Reference Hu, Bronner and Willett5), but the associations between EPA and DHA levels in erythrocytes and reduced risk for non-fatal MI were not significant in the Nurses' Health cohort study(Reference Sun, Ma and Campos6). Two cohort studies of men also found non-significant relationships between fish intake and the risk of non-fatal MI(Reference Ascherio, Rimm and Stampfer7, Reference Morris, Manson and Rosner8). Therefore, despite intensive investigations into the association between n-3 fatty acids and non-fatal MI, no clear pattern has emerged.
The Framingham risk score is the most widely used system for CHD risk prediction(Reference Chambless, Folsom and Sharrett9, Reference Wilson, D'Agostino and Levy10), but additional markers are required. The amount of long-chain n-3 fatty acids in erythrocyte membranes, as assessed by the Omega-3 Index (sum of EPA and DHA in erythrocytes), was shown to be a significant and independent discriminator of CHD(Reference Harris and Von Schacky11). Other studies have shown that fatty acids, particularly trans-fatty acids(Reference Clifton, Keogh and Noakes12) or n-6 fatty acids(Reference Harris, Poston and Haddock13), are positively associated with MI risk. However, the relationship between the fatty acid profile and the risk of non-fatal MI has not been investigated, even though the fatty acid profile or ‘lipidome’ may be key to identifying MI diet-related risk because fatty acids are associated with systemic inflammation, endothelial dysfunction and lipoproteins, all of which have previously been demonstrated to be independent risk factors for CHD(Reference Harris, Park and Isley14–Reference Mori and Woodman16).
To help elucidate the association between n-3 fatty acids and acute non-fatal MI, we determined if the Omega-3 Index is lower in patients with acute non-fatal MI compared with controls. Furthermore, we examined whether fatty acid profiles can independently predict non-fatal MI cases and add to the predictive power of the Framingham risk score.
Subjects and methods
Subjects
Subjects were recruited consecutively from acute non-fatal MI patients admitted to Hanyang University Kuri Hospital between November 2006 and May 2008. Cases consisted of patients diagnosed with their first case of ST-segment elevated MI. Control subjects were age-and sex-matched to the case subjects; patients were excluded if they had a history of CHD, cancer, hyperlipidaemia or diabetes. The present study was conducted according to the guidelines laid down in the Declaration of Helsinki and all procedures involving human subjects and patients were approved by the institutional review board of Hanyang University Kuri Hospital, and informed, written consent was obtained from all participants. Anthropometric data, medical history and socio-economic status were obtained from both medical chart reviews and interviews.
Laboratory measurements
Fasting blood samples were collected in EDTA and serum separator (SST) blood tubes during the day of admission, centrifuged and then divided into samples for storage at − 80 °C. Erythrocytes were directly methylated by adding boron trifluoride methanol-benzene (B1252; Sigma-Aldrich, St Louis, MO, USA) and heated for 10 min at 100 °C. Fatty acids methyl esters were analysed by GC (Shimadzu 2010AF; Shimadzu Scientific Instrument, Kyoto, Japan) with a 100 m SP2560 capillary column (Supelco, Bellefonte, PA, USA). Fatty acids were identified by comparison with known standards (GLC-727; Nu-Check Prep, Elysian, MN, USA). In the standard, the trans-18 : 1 peak was a mixture of trans-18 : 1n-12, trans-18 : 1n-9 and trans-18 : 1n-7, while the trans-18 : 2n-6 peak contained trans, trans-18 : 2n-6. The Omega-3 Index was calculated as the sum of erythrocyte concentrations of EPA and DHA and expressed as percentage of total fatty acids in erythrocyte membrane(Reference Harris and Von Schacky11). The quality-control sample comprised pooled erythrocytes, and the CV was 5·6 %.
Plasma leptin (human leptin; Invitrogen Inc., Carlsbad, CA, USA), adiponectin (Human Adiponectin/Acrp30 Immunoassay; R&D Systems Inc., Minneapolis, MN, USA) and insulin levels (human insulin; Invitrogen Inc.) were measured by ELISA. Serum lipid profiles (TBA-30FR; Toshiba, Tokyo, Japan), blood chemicals (Coulter LH 750 analyser; Beckman Coulter Inc., Fullerton, CA, USA), liver function (Variant; Bio-RAD, Hercules, CA, USA) and high-sensitivity C-reactive protein concentrations (Wr-CRP Advia 1650; Bayer, Berkeley, CA, USA) were measured with autoanalysers.
Statistical analysis
Continuous variables were expressed as mean values with their standard errors to compare cases and controls using independent t tests. Proportions of nominal variables were compared using the χReference Burr, Fehily and Gilbert2 test and correlation between variables was tested by partial correlation analysis after adjusting for age and sex. Fatty acids were categorised into tertiles based on control values. Binary logistic regression was adjusted for matching variables and used to find an association between baseline concentration and dependent factors. Analyses of fatty acids such as EPA, α-linolenic acid (ALA; 18 : 3n-3), trans-18 : 1n-9 (trans-oleic acid) and 20 : 4n-6 (arachidonic acid) were developed using stepwise selection to reduce the number of fatty acids to a subset of the ones with greatest impact. The calculation of the Framingham risk scores includes age, sex, smoking status, total cholesterol, HDL-cholesterol, and a history of diabetes and hypertension(Reference Wilson, D'Agostino and Levy17). A combination model was developed that included the fatty acids selected in multivariable logistic regression and the Framingham risk score used in the multivariable logistic regression. The area under the receiver operating characteristic curve was determined to compare the explanatory power of fatty acids, Framingham risk score, and a combination of fatty acids and Framingham risk score. A P value of < 0·05 was considered statistically significant. Statistical analysis was performed using SPSS version 12.0 (SPSS Inc., Chicago, IL, USA).
Results
Subject characteristics
The characteristics of subjects are shown in Table 1. There was a greater incidence of a history of diabetes and a family history of CHD in the cases compared with the controls. Although not significant, BMI was higher in the cases than the controls. In addition, sex, age, education level, exercise, smoking, drinking, period of smoking and drinking, and history of hypertension did not differ significantly between the cases and the controls.
Table 1 Characteristics of the study subjects
(Mean values with their standard errors or numbers of subjects)
* P < 0·05 (χ2 test).
† By independent t test.
Metabolic parameters
Leucocyte counts, the aspartic acid transaminase:alanine transaminase ratio, and levels of aspartic acid transaminase, total cholesterol, LDL-cholesterol, high-sensitivity C-reactive protein and leptin were significantly higher in the cases than the controls (P < 0·05), while adiponectin levels were significantly lower in the cases than the controls (P = 0·037; Table 2). Erythrocyte and platelet counts, Hb, packed cell volume, mean platelet volume, Hb-A1c, albumin, Ca, fasting glucose, blood urea N, alanine transaminase, TAG and HDL-cholesterol levels were not significantly different between groups.
Table 2 Metabolic parameters in cases and controls
(Mean values with their standard errors)
AST, aspartic acid transaminase; ALT, alanine transaminase.
Mean value was significantly different from that of the control group: * P < 0·05, *** P < 0·001.
Erythrocyte fatty acid composition
The Omega-3 Index was significantly lower in the cases than the controls (9·57 v. 11·81 %; P < 0·001), while total trans-fatty acids were significantly higher (1·01 (sem 0·04) v. 0·56 (sem 0·03) %; P < 0·001). EPA and ALA levels were also significantly lower in the cases than the controls, but total trans-fatty acids, trans-18 : 1, arachidonic acid, 16 : 0 and 22 : 4n-6 were significantly higher in the cases than the controls. Multivariable-adjusted regression analysis showed that the Omega-3 Index, EPA and ALA were negatively (P < 0·05) associated with the risk of acute non-fatal MI after adjusting for age, sex, history of diabetes, family history of CHD, smoking status, hypertension status, and levels of glucose, total cholesterol and HDL-cholesterol (Table 3). However, total trans-fatty acids, trans-18 : 1 and arachidonic acid were positively associated with the risk of acute non-fatal MI after adjusting for all confounding variables. The Omega-3 Index, and EPA and ALA levels were negatively related to the concentration of high-sensitivity C-reactive protein, but total trans-fatty acids and trans-oleic acids were positively related to the concentration of high-sensitivity C-reactive protein concentration (Table 4). Trans-oleic acid was significantly related to the levels of leptin and adiponectin, respectively, and trans-linoleic acid was positively related to the concentration of TAG. The Omega-3 Index, and ALA and EPA levels were negatively related to the concentration of glucose, while only EPA was negatively related to the total cholesterol level. Linoleic acid was negatively related to TAG, but was positively associated with LDL- and HDL-cholesterol concentrations.
Table 3 Association of fatty acid composition of erythrocytes with the risk of acute non-fatal myocardial infarction by multivariable regression analysis†
(Odds ratios and 95 % confidence intervals)
Omega-3 Index, sum of EPA and DHA.
* P < 0·05, ** P < 0·01, *** P < 0·001 (logistic regression analysis).
† OR and 95 % CI were analysed by logistic regression. OR in cases and controls were adjusted for age, sex, history of diabetes, family history of CHD, smoking status, hypertension status, and levels of glucose, total cholesterol and HDL-cholesterol.
‡ Estimates of P values for a linear trend were based on linear scores derived from the medians of tertiles of erythrocyte fatty acid composition among controls.
Table 4 Correlation between fatty acid composition of erythrocytes and metabolic parameters determined by partial correlation analysis
(Partial correlation coefficients adjusted for age and sex)
hs-CRP, high-sensitivity C-reactive protein; Omega-3 Index, sum of EPA and DHA in erythrocytes.
* P < 0·05, ** P < 0·01, *** P < 0·001.
The ability of risk scores, such as the traditional risk score and the fatty acids risk score including EPA, ALA, trans-oleic acid and arachidonic acid, to discriminate cases from controls was compared, both alone and in combination (Fig. 1). The area under the receiver operating characteristic curve of the fatty acid risk score alone (0·97) was better than that of the traditional risk score (0·68) as well as the combination of the two risk scores (0·95).
Fig. 1 The ability of the fatty acid profile of erythrocytes to discrimination between myocardial infarction cases and controls was assessed in the validation set using receiver operating characteristic curves. The fatty acids risk score (– - –) included levels of EPA, α-linolenic acid, trans-oleic acid and arachidonic acid; the traditional risk score (– - –) included Framingham risk factors such as age, sex, smoking status, total cholesterol, HDL-cholesterol, and a history of diabetes and hypertension; a combined risk score included traditional plus fatty acids risk score factors (—). (- - -), Reference line.
Discussion
In the present study, the fatty acid composition of erythrocytes was associated with the risk of occurrence of acute non-fatal MI. The risk of acute non-fatal MI was significantly reduced by a higher Omega-3 Index, n-3 fatty acids, EPA and ALA after adjusting for age, sex, history of diabetes, family history of CHD, smoking status, hypertension status, and levels of glucose, total cholesterol and HDL-cholesterol. In contrast, higher levels of total trans-fatty acids, trans-18 : 1 and 20 : 4n-6 were associated with an increased risk of acute non-fatal MI. To the best of our knowledge, the present study is the first to demonstrate that the fatty acid profile (lipidome) of erythrocytes can allow discrimination between acute non-fatal MI cases and controls. Furthermore, the fatty acid profile significantly and substantially improved acute non-fatal MI case discrimination compared with traditional CHD risk factors, suggesting that the erythrocyte fatty acid profile is more powerful than the Framingham risk score for identifying patients with acute non-fatal MI. The traditional Framingham risk score predicts CHD with an accuracy of 70–80 %(Reference Chambless, Folsom and Sharrett9, Reference Wilson, D'Agostino and Levy17), but measurement of the tissue levels of fatty acids could serve as screening tool to assess MI risk and guide dietary interventions using an evidence-based approach.
Previous CHD studies have found inverse associations between n-3 fatty acids(Reference Sun, Ma and Campos6, Reference Dewailly, Blanchet and Gingras18–Reference Albert, Campos and Stampfer21) and trans-fatty acids(Reference Clifton, Keogh and Noakes12, Reference Mozaffarian, Katan and Ascherio22–Reference Block, Harris and Reid24). Of the four fatty acids included in the model, increasing levels of two (EPA and ALA) were inversely associated with the OR for MI case status, while a direct association between the levels of trans-oleic acid and arachidonic acid levels and the OR for MI was found. Most studies have suggested that n-3 fatty acids, particularly DHA, are more cardioprotective than EPA(Reference Mori and Woodman16), and because tissue phospholipid DHA proportions are considerably higher than EPA (2·1 v. 0·8 %), the former may give a stronger signal and have lower variability. However, EPA reduced major coronary events by 19 % in patients with a history of coronary artery disease in the Japan EPA Lipid Intervention Study (JELIS) when the plasma level of EPA was about 2·6 %(Reference Yokoyama, Origasa and Matsuzaki4). EPA has many beneficial effects(Reference Harris, Park and Isley14), including hypolipidaemic and anti-platelet activities(Reference Park and Harris25), and plaque-stabilising properties(Reference Thies, Garry and Yaqoob26). In the present study, only EPA was associated with MI in Korean patients; the EPA levels of erythrocytes were 3·47 % in controls and 1·68 % in cases, similar to the values reported in a Japanese population and higher than those reported in an American study (0·46 % in cases v. 0·72 % in controls)(Reference Block, Harris and Reid24). Block et al. (Reference Block, Harris and Reid24) also reported DHA levels of 2·93 % in cases and 3·53 % in controls, which were lower than the Korean controls of 8·35 % and cases of 7·89 % in the present study.
EPA competes with arachidonic acid, the precursor of PGE2 and leucotriene B4, thereby producing pro-inflammatory and thrombotic effects(Reference Wada, DeLong and Hong27). Thus, a high intake of n-6 fatty acids is characterised by increases in blood viscosity, vasospasm and vasoconstriction, and n-3 fatty acids are associated with a decrease in blood viscosity and have anti-inflammatory, anti-thrombotic, anti-arrhythmic, hypolipidaemic and vasodilatory properties(Reference Mori and Woodman16). Arachidonic acid content in adipose tissue is associated with increased risk of MI(Reference Block, Harris and Reid24, Reference Pedersen, Ringstad and Almendingen28), but a meta-analysis of case–control and prospective cohort studies found that increased arachidonic acid content in phospholipids or TAG was not significantly associated with CHD(Reference Harris, Poston and Haddock13). However, Block et al. (Reference Block, Harris and Reid24) reported that there was a significant difference in arachidonic acid content between patients with acute coronary syndrome status and controls, and that arachidonic acid had a U-shaped relationship with the risk of acute coronary syndrome status. In the present study, the relationship of arachidonic acid to MI was also U-shaped; compared with the first quartile of arachidonic acid, the OR for case status in the second and third quartiles were 0·55 (95 % CI 0·14, 2·11) and 2·98 (95 % CI 0·95, 9·34), respectively. Fatty acids are measured as proportions; thus higher arachidonic and lower EPA levels may be associated with an increased risk of acute non-fatal MI.
Trans-fatty acids are associated with the risk of CHD(Reference Mozaffarian, Katan and Ascherio22). Clifton et al. (Reference Clifton, Keogh and Noakes12) and Pedersen et al. (Reference Pedersen, Ringstad and Almendingen28) found a positive association between MI and adipose tissue levels of trans-oleic acid. Block et al. (Reference Block, Harris and Reid24) measured fatty acids in erythrocytes and reported that high levels of trans-oleic acid were associated with acute coronary syndrome, whereas levels of trans-linoleic acid did not show a significant association with acute coronary syndrome. In the present study, we consistently found that the trans-oleic acid content of erythrocytes was associated with an increased risk of acute non-fatal MI. Although margarine, beef and dairy consumption have been linked to CHD, it has been difficult to link trans-fatty acids in erythrocytes from these sources to MI. Although the usual case–control design is prone to bias, the use of fatty acid biomarkers may be more reliable than dietary measures, and MI does not alter the tissue fatty acid composition(Reference Kark, Manor and Goldman29). Furthermore, the acute interventions commonly used to treat MI patients upon admission are not expected to alter the fatty acid composition of erythrocytes; however, we could not evaluate this assumption in the present study because blood samples were drawn upon admission.
The hypotriacylglycerolaemic effect of n-3 fatty acids is well established, and has been ascribed to the reduced hepatic synthesis of VLDL and increased catabolism of VLDL(Reference De Caterina, Madonna and Bertolotto30). A meta-analysis indicated that among all classes of fatty acids, trans-fatty acids have the strongest effect on raising serum LDL-cholesterol levels(Reference Mensink, Zock and Kester31). However, we did not find a significant association between erythrocyte fatty acids and lipoprotein levels with the exception of the level of EPA, which was negatively associated with the levels of total cholesterol and trans-linoleic acid and positively associated with the concentration of TAG. Small sample sizes, inadequate biomarkers and uncontrolled confounding factors may explain the discrepancies between these studies.
In conclusion, the fatty acid profile of erythrocytes can discriminate better between acute non-fatal MI cases and controls than the Framingham risk score. Increased levels of n-3 fatty acids and decreased levels of trans-fatty acids and n-6 fatty acids are associated with decreased odds of MI. These findings suggest that the fatty acid profile of erythrocytes may contain metabolomic information, and thus be of prognostic value in the diagnosis of non-fatal MI. Although prospective validation and a larger study are required, fatty acid profiles may have clinical utility for non-fatal MI risk assessment.
Acknowledgements
The present study was supported by a Korean Research Foundation grant funded by the Korean Government (MOEHD, Basic Research Promotion Fund; KRF-2007-331-C00279).
Y. P. is the principal investigator for this project and wrote the manuscript; J. Lim conducted the laboratory and statistical analysis; J. Lee recruited MI patients and collected medical information, and advised on the clinical part of this paper; S. K. recruited control patients and collected medical information, and advised on the clinical part of this paper.
The authors state that there are no conflicts of interest.
