Plasma α-Linolenic and Long-Chain ω-3 Fatty Acids Are Associated with a Lower Risk of Acute Myocardial Infarction in Singapore Chinese Adults^,,

Abstract

Background: Long-chain marine omega-3 polyunsaturated fatty acids (n–3 PUFAs) are associated with a lower risk of acute myocardial infarction (AMI), but results for plant-derived α-linolenic acid (ALA; 18:3n–3) are inconsistent.

Objective: We aimed to examine the association between plasma n–3 PUFAs and AMI risk and to explore potential mediation by cardiovascular disease risk factors.

Methods: A nested case-control study with 744 incident AMI cases and 744 matched controls was conducted within the Singapore Chinese Health Study for participants aged 47–83 y. Conditional logistic regression was used to calculate the multivariable ORs for AMI with and without adjustment for cardiovascular disease risk factors, including blood lipids, blood pressure, C-reactive protein, serum creatinine, and glycated hemoglobin.

Results: Plasma long-chain n–3 PUFAs were associated with lower AMI risk (multivariable OR: 0.62; 95% CI: 0.41, 0.94; for the highest compared with the lowest quartile; P-trend = 0.03). This association was not substantially changed after adjustment for cardiovascular disease risk factors. Dietary intakes of fish and long-chain n–3 PUFAs were similarly inversely associated with AMI risk. Plasma ALA was marginally associated with a lower risk of AMI (multivariable OR: 0.73; 95% CI: 0.51, 1.05; P-trend = 0.07) even in persons with high plasma concentrations of long-chain n–3 PUFAs. This association became significantly weaker after adjustment for blood pressure and LDL cholesterol.

Conclusions: Plasma long-chain n–3 PUFAs are associated with a lower risk of AMI in this Asian population. Plasma ALA may be marginally associated with reduced AMI risk, even in persons with high concentrations of long-chain n–3 PUFAs, and this association may be partially mediated by lower blood pressure and LDL cholesterol.

Introduction

Coronary artery disease (CAD)¹² is one of the leading causes of death worldwide, and the disease burden is increasing rapidly in Asia (1–3). Dietary FAs, among other modifiable lifestyle factors, are implicated in the cause of CAD. ω-3 (n–3) PUFAs are associated with lower risk of cardiovascular diseases, particularly fatal CAD (4). Two main subtypes of n–3 PUFAs are identified: the marine-originated long-chain n–3 PUFAs EPA (20:5 n–3) and DHA (22:6 n–3) and the plant-derived intermediate-chain n–3 PUFA α-linolenic acid (ALA; 18:3n–3) found mainly in nuts, seeds, and their oils. Dietary and circulating long-chain n–3 PUFAs are shown to be inversely associated with fatal and nonfatal CAD in large prospective studies (5, 6). In some clinical trials administration of long-chain n–3 PUFAs reduced total mortality and CAD incidence (7, 8), but other trials did not suggest beneficial effects (9). The relation between ALA and CAD risk has been examined in fewer studies than long-chain n–3 PUFAs. ALA can be used for the production of EPA and DHA in the body through desaturase and elongase enzymes and may also affect CAD risk independently (10). Findings of limited epidemiologic studies supported a beneficial effect of ALA intake on nonfatal or fatal CAD particularly sudden death (11–15), whereas others reported no beneficial or even detrimental effects on cardiovascular health (16, 17).

Most studies on n–3 PUFAs and risk of CAD were conducted in Western populations. Food sources, amounts of intake, and physiologic responses may be different for Asian people from their Western counterparts. Few studies have examined dietary (15, 18–22) or circulating (23, 24) long-chain n–3 PUFAs in relation to cardiovascular diseases in Asian populations. Furthermore, most studies focused on CAD mortality (4). Data on ALA in Asia are sparse (15), and there are no prospective studies on plasma ALA and CAD risk.

EPA and DHA are suggested to affect cardiovascular disease risk through a variety of cardiometabolic pathways, including reduction in serum TGs and blood pressure (4, 9). ALA has shown beneficial effects on platelet function, arterial compliance, and arrhythmia (25). To date, data are limited on the association of n–3 PUFAs on biological markers for cardiovascular disease risk in population-based studies.

Therefore, we conducted a case-control study nested within an established population-based prospective cohort, the Singapore Chinese Health Study (SCHS), to examine the association between plasma n–3 PUFAs, including ALA, EPA, and DHA, and risk of acute myocardial infarction (AMI), which has the highest rate of morbidity and mortality among the clinical spectrum of CAD (26). In addition, we examined the possible mediation of the associations of these n–3 PUFAs via established cardiovascular disease risk factors, including blood lipids, blood pressure, and inflammation.

Methods

SCHS

The SCHS is a population-based, prospective cohort study of 63,257 Chinese men and women (Hokkien or Cantonese dialect group) aged 45–74 y who reside in public housing estates, where 86% Singaporeans resided at the time of recruitment. Recruitment and assessment of baseline diet, lifestyle, and medical history took place from 1993 to 1998. To date, follow-up surveys were conducted twice (1999–2004 and 2006–2010) to update information on use of tobacco and alcohol, medical history, and menopausal status. Blood specimens were also collected from 28,439 subjects during 1994–2005. The cohort was followed for morbidity and mortality through record linkage with Singapore Registry of Births and Deaths, Hospital Discharge Database (HDD), and Singapore Cancer Registry. This study was approved by the institutional review board at the National University of Singapore, and all participants gave informed consent.

Nested case-control study

We have established a nested case-control study that involved 744 incident AMI cases and 744 matched controls. These cases and controls were selected from the SCHS participants who provided blood and did not have a history of CAD or stroke at the time of blood collection on the basis of self-reported diagnosis and data from the HDD [International Classification of Diseases (ICD)-9 codes: 410–414, 427, 428, 430–434, 438].

Cases include incident nonfatal or fatal AMI occurring after blood specimen collection until 31 December 2010. We identified and verified cases of AMI through linkage with the following 3 databases. 1) The national Mediclaim System HDD has been in use in Singapore since 1990 and records up to 3 diagnoses per patient for inpatient discharges from public and private hospitals on the basis of ICD-9 (27). We selected all participants who were identified as AMI cases (ICD-9: 410) in this database until 31 December 2010. The medical records of these participants were then reviewed by a cardiologist with the use of the criteria of the Multi-Ethnic Study of Atherosclerosis (28), and only confirmed cases of AMI were included in our study. 2) The Singapore Myocardial Infarction Registry is a centralized population-based registry that uses similar procedures as we used for HDD to confirm AMI on the basis of a review of medical records (29). We only had access to cases up to 31 December 2005 from the Singapore Myocardial Infarction Registry, but these cases largely overlapped with the cases identified through the other databases. Through steps 1 and 2, 555 incident AMI cases were identified. 3) The Singapore Registry of Births and Deaths codes causes of death according to ICD-9, and participants with codes 410–414 (ischemic heart disease) stated as the primary cause of death were selected, which are assumed to be death from MI in practice, giving 289 fatal AMI cases until 31 December 2010. Among them, 100 belonged to incident AMI cases described above. We thus identified 744 unique fatal and nonfatal AMI cases.

Controls were selected with the use of the risk-set sampling strategy (30); controls were participants who were alive and free of CAD at the time of the diagnosis or death of the respective indexed cases and matched (in ratio of 1 to 1) for sex, father’s dialect group, date of birth (±2 y), date of recruitment (±1 y), and date of blood collection (±6 mo).

Assessment of dietary intakes and relevant covariates

All data on diet and lifestyle were collected with the use of a validated 165-item semiquantitative FFQ. Intakes of dietary nutrients were calculated with the use of the Singapore food composition table (31). The FFQ was validated by two 24-h dietary recalls and re-administration of the FFQ. The correlation coefficients between these 2 methods for fat-related nutrients ranged from 0.24 to 0.79 (31). The intakes of n–3 PUFAs from marine/nonmarine sources and their main food sources were calculated. The main dietary sources of long-chain n–3 PUFAs (mainly EPA and DHA) were listed in 14 seafood items in the FFQ, including fresh fish (fish ball or cake, deep-fried fish, pan- or stir-fried fish, boiled or steamed fish), fresh shellfish (shrimp or prawn, squid or cuttlefish), dried/salted fish (salted fish, ikan bilis, dried fish, other dried seafood such as dried shrimp, dried oyster, dried cuttlefish), and canned fish (canned tuna, canned sardine) (32). The main dietary sources of intermediate-chain n–3 PUFAs (mainly ALA) were grains (21%), cooking oils (11%), and legumes and soy (9%) (15). Information on education, BMI (in kg/m²), physical activity, smoking, alcohol intake, and history of diabetes and hypertension were collected as previously described (29, 33, 34).

Assessment of cardiovascular disease risk factors

Blood samples were collected during home visits during 1994–2005 and stored at −80°C. Systolic and diastolic blood pressures were measured by Omron blood pressure monitors 3 times with the use of a standard protocol, and the average of the 3 measurements was used for statistical analysis. Conventional biochemical risk factors of CAD were measured by National University Hospital referral laboratories, including total, HDL, and LDL (directly measured) cholesterol; TGs; high-sensitivity C-reactive protein (CRP); serum creatinine; and glycated hemoglobin (HbA1c). Detailed analyzing methods and CVs are described in the Supplemental Text. Case-control status was blinded to all laboratory staff. Two biosamples of a given case-control pair were placed and tested in the same plate.

Measurement of plasma FAs

Plasma FAs were measured with the use of gas chromatography-tandem mass spectrometry on an Agilent 7890 GC system equipped with a 7001B QQQ triple quadruple mass detector (Agilent) and an auto sample injector (35). With the use of this method, FAs from both free and esterified (TGs, phospholipids, cholesterol esters) fractions were measured in total. Nineteen FAs, covering major saturated FAs, MUFAs, and PUFAs, were quantified. Plasma total FAs were calculated as the sum of the quantities of the 19 FAs. The within-batch CVs for the 3 FAs of interest (ALA, EPA, and DHA) ranged from 5.3% to 8.1%, and the between-batch CVs ranged from 12.2% to 16.8%.

Statistical analysis

Individual plasma FAs were expressed as a percentage of plasma total FAs. Corresponding FA intakes were expressed as a percentage of total energy intake. Univariate distributions were assessed on all continuous variables. Plasma ALA, EPA, DHA, creatinine, CRP, and HbA1c were log-transformed to improve normality. The continuous variables were categorized into quartiles according to the distribution among the control subjects.

Demographic characteristics and cardiovascular disease risk factors were compared between the cases and controls with the use of univariate conditional logistic regression. Cross-sectional associations between plasma FAs and various cardiovascular disease risk factors were assessed by multivariable linear regression among the control subjects with unique individuals. To improve the comparability across different risk factors, all plasma FAs and risk factors were standardized to z scores (36), and standardized β coefficients were presented.

Conditional logistic regression models were used to compute multivariable-adjusted ORs and 95% CIs for higher quartiles of the exposure variables with the use of the lowest quartile as the reference group. Tests for trend were evaluated by modeling the medians of each quartile of exposure variables as continuous variables. In the multivariable model, besides the 5 matching factors, we adjusted for age (years), total energy intake (kcal/d), plasma total FAs (mg/L), hours of fasting before blood collection (<2 h, 2 h to <8 h, ≥8 h), education level (none, primary school, secondary school or above), cigarette smoking (never, ex-smoker, current smoker with <13 cigarettes/d, current smoker with ≥13 cigarettes/d), alcohol consumption (g/d), physical activity (0 h/wk, <4 h/wk of moderate and <2 h/wk of strenuous activity, ≥4 h/wk of moderate or ≥2 h/wk of strenuous activity), BMI, cholesterol intake (mg/d, residual adjusted), fiber intake (g/d, residual adjusted), total fat intake (percentage of total energy), history of hypertension, and history of diabetes. Analyses were conducted for plasma and dietary n–3 PUFAs and for fish consumption. The sum of the standardized z scores (36) of plasma and dietary n–3 PUFAs were also examined as an integrated measure of exposure. Analyses were also repeated separately for fatal and nonfatal AMI cases.

The possible mediation effect of plasma n–3 PUFAs on AMI risk through various established cardiovascular disease risk factors was assessed by comparing the ORs of plasma n–3 PUFAs with or without each of the potential mediators in the fully adjusted model. We expressed plasma n–3 PUFAs as continuous variables for the mediation analysis to increase the statistical power by using information from all study subjects. The percentage of total association that is mediated by each risk factor was calculated as the percentage change in the β coefficient of the plasma FAs, and the level of significance (P value) of such mediation effect were evaluated with the use of the Karlson-Holm-Breen method (37). We also assessed the combined mediation effect of all the risk factors by including them in the Karlson-Holm-Breen mediation model together. Several risk factors had missing values. Blood pressure was not measured for 87 subjects. Because only nonfasting blood was collected, the plasma TG concentration was beyond the limit of detection for 62 samples. Two participants lacked HbA1c information because no red blood cell sample was available. To compare the mediation effect with the use of the same group of subjects, the above-mentioned missing values were imputed for the mediation analyses by fitting a linear regression equation with age, sex, BMI, and other cardiovascular disease risk factors without missing values (LDL and HDL cholesterol, creatinine, and CRP).

All statistical analyses were conducted with the use of Stata version 11 (StataCorp), and statistical significance was set as P < 0.05.

Results

We selected 744 AMI cases and 744 of their matched controls. The mean age at blood collection was 66 y (range: 47–83 y), and 65% of participants were men. Participants who developed AMI were more likely to be a current smoker, to have less regular physical activity, and to have a history of physician-diagnosed hypertension and diabetes. They also had higher blood pressure, LDL cholesterol, TGs, creatinine, CRP, and HbA1c but lower HDL cholesterol. The AMI cases had significantly lower plasma concentrations of EPA and DHA than the controls, but the difference in plasma ALA was not significant (Table 1).

TABLE 1.

Distributions of selected characteristics in patients who developed incidental acute myocardial infarction and their matched control subjects in the Singapore Chinese Health Study¹

Characteristics	Cases (n = 744)	Controls (n = 744)	P-difference
Age, y	66.1 ± 7.8	66.0 ± 7.7	NA (matched)
Sex, % of men	64.7	64.7	NA (matched)
Education, % ≥secondary level	25.9	28.2	0.28
Current smoker, %	31.1	21.9	<0.001
No regular physical activity,2 %	74.6	67.2	0.002
BMI, kg/m2	23.2 ± 3.1	22.9 ± 2.9	0.06
Energy intake, kcal/d	1610 ± 591	1610 ± 588	0.87
Ethanol intake, g/d	2.3 ± 9.4	2.3 ± 8.0	0.88
History of hypertension, %	46.9	36.8	<0.001
History of diabetes, %	25.0	12.5	<0.001
Fasting status, %	21.2	21.4	0.95
Plasma ALA, % total FAs	0.33 ± 0.20	0.34 ± 0.28	0.40
Plasma EPA, % total FAs	0.44 ± 0.25	0.52 ± 0.43	<0.001
Plasma DHA, % total FAs	2.24 ± 1.24	2.42 ± 1.45	0.001
Systolic blood pressure, mm Hg	149 ± 23.7	141 ± 21.7	<0.001
Diastolic blood pressure, mm Hg	83 ± 11.7	81 ± 10.9	0.001
LDL cholesterol, mmol/L	3.3 ± 0.9	3.2 ± 0.8	0.001
HDL cholesterol, mmol/L	1.3 ± 0.3	1.4 ± 0.3	<0.001
TGs, mmol/L	1.7 ± 0.7	1.6 ± 0.7	0.001
Creatinine, μmol/L	77.2 ± 57.0	71.2 ± 37.1	0.03
C-reactive protein, mg/L	3.8 ± 9.1	2.8 ± 9.1	0.05
HbA1c, %	6.7 ± 1.6	6.1 ± 1.1	<0.001

¹Values are means ± SDs for continuous variables and % for categorical variables. P-difference was derived from univariate conditional logistic regression. Missing values for systolic and diastolic blood pressures (n = 87), TGs (n = 62), and HbA1c (n = 2). ALA, α-linolenic acid; HbA1c, glycated hemoglobin; NA, not applicable.

²Only physical activities of moderate or vigorous intensity were compared.

The distributions of selected characteristics in relation to quartiles of plasma n–3 PUFAs among controls is shown in Supplemental Table 1. Participants with higher plasma ALA were more likely to have higher education, to be nonsmokers, to have no regular physical activity, and to have higher intake of dietary ALA than participants with lower plasma ALA concentrations. Participants with higher plasma long-chain n–3 PUFAs (EPA and DHA) were more likely to be women and to have a higher intake of dietary long-chain n–3 PUFAs than participants with lower plasma EPA and DHA concentrations. The Pearson correlation coefficient between plasma and dietary n–3 PUFAs was 0.21 (P < 0.0001) for EPA/DHA and 0.11 (P = 0.003) for ALA. These moderate correlations were as expected because the dietary assessment was conducted 6.6 y before blood collection on average.

Cross-sectional associations between plasma n–3 PUFAs and cardiovascular disease risk factors in the control group were assessed with the use of multivariable linear regression and are presented in Table 2. Plasma ALA was significantly associated with lower systolic and diastolic blood pressures, lower LDL- and HDL-cholesterol concentrations, and higher TG concentrations. Plasma EPA was significantly associated with higher HDL-cholesterol concentration, and plasma DHA was significantly associated with higher LDL-cholesterol and TG concentrations.

TABLE 2.

The cross-sectional association of plasma concentrations of n–3 PUFAs with cardiovascular disease risk factors in control subjects only in the Singapore Chinese Health Study¹

Cardiovascular disease risk factors	ALA	EPA	DHA	EPA + DHA
Systolic blood pressure, mm Hg	−0.100 ± 0.039*	−0.032 ± 0.039	−0.020 ± 0.038	−0.027 ± 0.038
Diastolic blood pressure, mm Hg	−0.117 ± 0.039**	−0.045 ± 0.040	−0.064 ± 0.038	−0.066 ± 0.039
LDL cholesterol, mmol/L	−0.219 ± 0.038***	0.055 ± 0.039	0.081 ± 0.038*	0.078 ± 0.038*
HDL cholesterol, mmol/L	−0.187 ± 0.035***	0.107 ± 0.036**	0.067 ± 0.035	0.076 ± 0.035*
TGs. mmol/L	0.260 ± 0.027***	0.021 ± 0.029	0.072 ± 0.029*	0.063 ± 0.029*
Creatinine, μmol/L	−0.049 ± 0.033	−0.061 ± 0.033	−0.038 ± 0.033	−0.043 ± 0.033
C-reactive protein, mg/L	−0.066 ± 0.040	−0.047 ± 0.040	−0.054 ± 0.039	−0.059 ± 0.039
HbA1c, %	−0.047 ± 0.033	−0.021 ± 0.034	−0.013 ± 0.033	−0.017 ± 0.033

¹Values are standardized β coefficient ± SE from linear regression, expressed as the multiples of SD change in outcome variables (cardiovascular disease risk factors) per SD increase of logged values of respective exposure variables (plasma n–3 PUFAs). Because of skewed distribution, creatinine, C-reactive protein, and HbA1c were log-transformed before being standardized. All models were adjusted for age at blood collection, sex, father’s dialect, date of blood collection, total energy intake, plasma total FAs, hours of fasting before blood collection, education, cigarette smoking, alcohol consumption, physical activity, BMI, cholesterol intake, fiber intake, total fat intake, and history of hypertension and diabetes. Missing values in the regression for systolic and diastolic blood pressure (n = 42), TGs (n = 26), and HbA1c (n = 2); 723 unique individuals were used for the other analyses. *,**,***Significant: *P < 0.05, **P < 0.01, ***P < 0.001. ALA, α-linolenic acid; HbA1c, glycated hemoglobin.

The association between plasma n–3 PUFAs and risk of AMI is shown in Table 3. In the multivariable model, higher plasma concentrations of EPA and DHA were associated with a lower AMI risk. Compared with the lowest quartile, the OR for the highest quartile was 0.68 (95% CI: 0.46, 1.00; P-trend = 0.02) for plasma EPA, 0.59 (95% CI: 0.38, 0.90; P-trend = 0.01) for plasma DHA, and 0.62 (95% CI: 0.41, 0.94; P-trend = 0.03) for the combined plasma concentration of EPA and DHA. For plasma ALA, no clear trend was observed; the higher 3 quartiles of plasma ALA were associated with lower risk of AMI as compared with the lowest quartile, a difference that was significant for the third quartile. These results may have been affected by multiple testing because we examined 3 exposure variables. With the use of Bonferroni correction (α = 0.05/3 = 0.017) the association for plasma DHA in the multivariable model remains statistically significant, whereas associations for plasma ALA and EPA were only nominally significant.

TABLE 3.

ORs (95% CIs) of acute myocardial infarction by quartile concentrations of plasma n–3 PUFAs in adults in the Singapore Chinese Health Study

Plasma n–3 PUFAs	Quartile 1	Quartile 2	Quartile 3	Quartile 4	P-trend
ALA1
n Cases/n total	220/406	173/359	167/353	184/370
Median, % total FAs	0.19	0.25	0.33	0.50
Model 12	1.00	0.74 (0.55, 1.00)	0.65 (0.47, 0.89)	0.68 (0.49, 0.94)	0.01
Model 23	1.00	0.77 (0.56, 1.04)	0.67 (0.48, 0.93)	0.76 (0.54, 1.08)	0.09
Model 34	1.00	0.75 (0.55, 1.04)	0.68 (0.48, 0.95)	0.73 (0.51, 1.05)	0.07
EPA
n Cases/n total	211/398	233/418	153/339	147/333
Median, % total FAs	0.28	0.37	0.46	0.71
Model 1	1.00	1.12 (0.82, 1.53)	0.68 (0.48, 0.97)	0.65 (0.45, 0.93)	0.004
Model 2	1.00	1.13 (0.82, 1.55)	0.74 (0.51, 1.06)	0.71 (0.49, 1.03)	0.03
Model 3	1.00	1.06 (0.76, 1.48)	0.70 (0.48, 1.02)	0.68 (0.46, 1.00)	0.02
DHA
n Cases/n total	209/395	184/370	198/384	153/339
Median, % total FAs	1.18	1.69	2.37	3.96
Model 1	1.00	0.86 (0.64, 1.17)	0.85 (0.62, 1.17)	0.56 (0.37, 0.84)	0.005
Model 2	1.00	0.92 (0.67, 1.26)	0.91 (0.66, 1.26)	0.60 (0.40, 0.91)	0.02
Model 3	1.00	0.90 (0.65, 1.23)	0.86 (0.61, 1.20)	0.59 (0.38, 0.90)	0.01
EPA + DHA
n Cases/n total	208/394	181/367	201/387	154/340
Median, % total FAs	1.49	2.07	2.83	4.58
Model 1	1.00	0.86 (0.63, 1.17)	0.88 (0.64, 1.22)	0.60 (0.41, 0.89)	0.01
Model 2	1.00	0.91 (0.66, 1.24)	0.94 (0.67, 1.31)	0.66 (0.44, 0.99)	0.04
Model 3	1.00	0.88 (0.63, 1.21)	0.87 (0.62, 1.23)	0.62 (0.41, 0.94)	0.03

¹ALA, α-linolenic acid.

²Model 1 was adjusted for total energy intake, age at blood collection, plasma total FAs, hours of fasting before blood collection, gender, father’s dialect, year of birth, year of recruitment, and year of blood collection.

³Model 2 was additionally adjusted for education, cigarette smoking, alcohol consumption, physical activity, BMI, cholesterol intake, fiber intake, and total fat intake.

⁴Model 3 was additionally adjusted for history of hypertension and diabetes.

Similar trends were observed for dietary intake of long-chain n–3 PUFAs (extreme-quartile OR: 0.64; 95% CI: 0.46, 0.89; P-trend = 0.03), but no substantial association was observed for dietary intake of ALA (Supplemental Table 2). We also conducted separate analyses for fatal and nonfatal AMI as an outcome. Although the point estimates suggested a stronger association for fatal AMI (extreme-quartile OR: 0.62; 95% CI: 0.34, 1.11; P-trend = 0.14) than for nonfatal AMI (extreme-quartile OR: 0.80; 95% CI: 0.52, 1.24; P-trend = 0.41), this may well have been because of chance as indicated by the largely overlapping CIs (Supplemental Table 3).

The joint association of plasma long-chain n–3 PUFAs (EPA and DHA) and ALA with AMI is shown in Supplemental Figure 1. Compared with participants who had the lowest plasma concentrations of both long-chain and intermediate-chain n–3 PUFAs, participants with the highest plasma concentrations of both types of n–3 PUFAs had a 49% lower risk of AMI (OR: 0.51; 95% CI: 0.31, 0.84). The test for multiplicative interaction was not statistically significant (P = 0.24).

The association between plasma n–3 PUFAs and AMI risk after further adjustment for various cardiovascular disease risk factors that could act as potential biological mediators is shown in Table 4. The association between plasma ALA and AMI risk (continuous OR: 0.89; 95% CI: 0.78, 1.02) became significantly weaker after adjustment for systolic blood pressure (OR: 0.92; 95% CI: 0.80, 1.05) and LDL cholesterol (OR: 0.93; 95% CI: 0.81, 1.06). We thus estimated that systolic blood pressure and LDL cholesterol mediated 20.1% and 30.9% of the association between ALA and AMI risk, respectively. Adjustment for cardiovascular disease risk factors did not significantly attenuate the association between plasma long-chain n–3 PUFAs and AMI risk.

TABLE 4.

Change in ORs (95% CIs) of acute myocardial infarction by plasma n–3 PUFAs after addition of cardiovascular disease risk factors in adults in the Singapore Chinese Health Study¹

	ALA		EPA + DHA
Models	OR (95% CI)	P-mediation	OR (95% CI)	P-mediation
Multivariable model2	0.89 (0.78, 1.02)		0.79 (0.68, 0.92)
Add SBP	0.92 (0.80, 1.05)	0.03	0.79 (0.68, 0.92)	0.77
Add LDL cholesterol	0.93 (0.81, 1.06)	0.01	0.77 (0.66, 0.89)	0.01
Add HDL cholesterol	0.87 (0.76, 1.00)	0.02	0.80 (0.69, 0.94)	0.09
Add TGs	0.89 (0.78, 1.02)	0.28	0.79 (0.68, 0.92)	0.49
Add creatinine	0.89 (0.78, 1.02)	0.34	0.80 (0.68, 0.92)	0.62
Add C-reactive protein	0.90 (0.78, 1.03)	0.65	0.79 (0.68, 0.92)	0.25
Add HbA1c	0.90 (0.79, 1.03)	0.49	0.79 (0.68, 0.91)	0.62
Add all risk factors	0.93 (0.81, 1.07)	0.36	0.79 (0.67, 0.92)	0.79

¹Values are ORs expressed per SD increase of logged values of respective exposure variables (plasma n–3 PUFAs). The statistical significance of the effect of addition of cardiovascular disease risk factors to the model (mediation effect) was tested with the use of the Karlson-Holm-Breen method. Missing values were imputed for SBP (n = 87), TGs (n = 62), and HbA1c (n = 2). ALA, α-linolenic acid; SBP, systolic blood pressure.

²Multivariable model was adjusted for total energy intake, age at blood collection, plasma total FAs, hours of fasting before blood collection, sex, father’s dialect, year of birth, year of recruitment, year of blood collection, education, cigarette smoking, alcohol consumption, physical activity, BMI, cholesterol intake, fiber intake, total fat intake, and history of hypertension and diabetes.

The association for long-chain n–3 PUFAs and AMI risk with the use of different assessment methods (measured plasma FAs compared with dietary intakes) and different subtypes of AMI (fatal compared with nonfatal) is shown in Supplemental Table 4. Associations for dietary intake of long-chain n–3 PUFAs and fish intake were similar to associations for the plasma long-chain n–3 PUFAs. We also integrated information from plasma and dietary long-chain n–3 PUFAs by taking the sum of their respective standardized z scores. The integrated variable of long-chain n–3 PUFAs resulted in stronger associations with both fatal AMI (extreme-quartile OR: 0.37; 95% CI: 0.19, 0.72; P-trend = 0.002) and nonfatal AMI (extreme-quartile OR: 0.61; 95% CI: 0.40, 0.92; P-trend = 0.02). Integration of plasma and dietary measurements of ALA did not show significant associations with risk of AMI, possibly because of limited statistical power (Supplemental Table 5).

Discussion

In this prospective nested case-control study of Singapore Chinese, higher circulating long-chain n–3 PUFAs (EPA and DHA) were associated with a lower incidence of AMI. Dietary intake of long-chain n–3 PUFAs and fish intake showed similar association with AMI risk as did plasma long-chain n–3 PUFAs, and integrating information from plasma biomarker and dietary intake resulted in stronger associations with AMI risk. For plasma ALA, no clear trend was observed, but the highest 3 quartiles were associated with a lower AMI risk compared with the lowest quartile. Our results suggested that lower LDL-cholesterol concentrations and lower blood pressure may contribute to the association between plasma ALA and a lower AMI risk.

Our results are largely consistent with previous prospective studies which observed that consumption of EPA and DHA lowers the risk of CAD (38). A recent meta-analysis of prospective and retrospective studies observed that dietary and circulating ALA was associated with a moderately lower risk of cardiovascular diseases with a pooled RR of 0.80 (95% CI: 0.63, 1.03) (39). Another meta-analysis of prospective studies on blood ALA concentrations also found a nonsignificantly lower risk of CAD with a pooled RR of 0.93 (95% CI: 0.83, 1.03) (40). The modest association that we observed for plasma ALA on AMI risk is similar to these 2 meta-analyses. We also found a suggestion of a more pronounced association between plasma ALA and fatal AMI than nonfatal AMI, consistent with those reported by several other cohort studies (13, 14) and a meta-analysis (41).

The observation that higher plasma ALA concentrations could further reduce AMI risk in the highest quartile of plasma long-chain n–3 PUFAs corroborates the hypothesis that the 2 types of n–3 PUFAs affect the cardiovascular system independently. Indeed, although ALA could be converted to EPA and DHA, such endogenous biochemical conversion is limited in humans (42–44). This independent contribution was also observed in the same cohort on dietary intakes of n–3 PUFAs and cardiovascular death (15). These findings support recommendations to regularly consume n–3 PUFAs from both marine- and plant-based foods.

The subjects in our study generally had a high consumption of fish and other seafood (mean daily intake: 54.7 ± 31.4 g), which are the main contributors to EPA and DHA intakes (32). However, because many of the commonly consumed types of fish in Singapore are white lean fish, the plasma concentrations of EPA and DHA in our study participants were not among the highest in the world (5, 23, 24, 45, 46). The observation that risk of AMI continued to decrease from the third quartile to the fourth quartile of the plasma long-chain n–3 PUFAs suggested that there was no ceiling effect of these FAs in preventing AMI at the intake amounts in our study population. In contrast, we did not observe a clear trend for the association between plasma ALA and AMI risk with a suggestion that low intakes may increase risk, but that risk does not continue to decrease at higher amounts of ALA intakes. A similar nonlinear association was observed in a case-control study conducted in Costa Rica (14).

n–3 PUFAs affect a wide range of physiologic functions in multiple tissues. The associations between higher plasma EPA and DHA with both higher LDL- and HDL-cholesterol concentrations in our study are supported by meta-analyses of clinical trials (47, 48). Dietary long-chain n–3 PUFAs were also shown to lower plasma TG concentrations and blood pressure in randomized trials (49, 50). In contrast, we only observed a nonsignificant inverse association between plasma EPA/DHA and blood pressure and a direct association between plasma DHA and TG concentrations. These novel observations might partly be explained by the lower intake of the n–3 PUFAs in our participants compared with the higher dose tested in clinical trials of fish oils (50). Despite examining a wide range of cardiovascular disease risk factors, we did not observe substantial mediators of the inverse association of plasma long-chain n–3 PUFAs with AMI risk. It is possible that putative beneficial effects of long-chain n–3 PUFAs are mediated through other pathways such as reduced platelet aggregation response (51), reduced heart rate (52), and antiarrhythmia (53). It is also possible that components of fish other than long-chain n–3 PUFAs, such as healthy protein, vitamin D, and selenium, contribute to the lower AMI risk associated with fish consumption through mechanisms unrelated to effects observed in trials that provide fish oil in isolation (54).

We observed an inverse association between plasma ALA and blood pressure and LDL- and HDL-cholesterol concentrations, which is supported by the findings of several clinical trials (55–58). However, evidence for the effects of ALA on conventional cardiovascular disease risk factors was mixed and controversial (59). The only meta-analysis of clinical trials on this topic found that dietary ALA significantly decreases HDL cholesterol, with no significant effects on blood pressure, LDL cholesterol, and TGs (60). On the basis of our observations, ALA may have both beneficial and detrimental associations with cardiovascular disease risk factors, which warrants further examination in other cohorts and clinical trials. Possibly, the variation in the strength of the association between ALA and AMI risk in populations in previous studies depended on the relative importance of different cardiovascular disease risk factors for the development of AMI in that population. In our study, lower LDL cholesterol and lower blood pressure may partially explain the inverse association between plasma ALA and AMI risk, whereas the strengthened association after further adjustment for HDL cholesterol could be an artifact because the association between HDL cholesterol and AMI may not be causal (61).

Strengths of our nested case-control study included the prospective design that reduces the likelihood of selection and recall bias. In addition, our use of validated FFQs and quantitative laboratory assays provided objective, specific, and more accurate measures of exposures to n–3 PUFAs. Furthermore, the analysis incorporating with a wide range of potential confounders and established cardiovascular disease risk factors minimized the potential confounding bias on the observed associations of n–3 PUFAs to AMI risk. However, this study is also subject to several limitations. First, there might be misclassification because of the use of nonfasting plasma. Yet we observed that plasma concentrations of n–3 PUFAs were comparable among control subjects with fasting and nonfasting plasma, suggesting limited misclassification because of lack of fasting. We addressed this issue by adjustment for fasting status in multivariable models. Second, as in any observational study, we cannot fully exclude the possibility of residual confounding because of diet and lifestyle that were not well measured in our study. Third, we might be underpowered to detect part of the mediation effects. Fourth, the assessment of the association between plasma n–3 PUFAs and cardiovascular disease risk factors is cross-sectional, and we cannot completely exclude the possibility of reverse causation. Finally, we suggest caution in generalizing the findings to other populations with a different ethnicity or substantially different intake amounts.

In conclusion, higher plasma concentrations of EPA and DHA were associated with lower risk of AMI in an ethnic Chinese population in Singapore. This association was largely independent of the established cardiovascular disease risk factors. Higher intakes of fish and long-chain n–3 PUFAs were similarly associated with a lower risk of AMI. These results suggest that higher fish consumption can contribute to a substantial reduction in the incidence of AMI. We did not observe a linear association between plasma ALA and risk of AMI, and the dose-response relation and possible underlying mechanisms require further research. Our results suggest that higher plasma ALA concentrations may be beneficial even in persons with high plasma concentrations of long-chain n–3 PUFAs and that this association may be partially mediated by lower blood pressure and LDL-cholesterol concentrations.