Prediction of Cardiovascular Mortality in Middle-aged Men by Dietary and Serum Linoleic and Polyunsaturated Fatty Acids

Background  Substitution of dietary polyunsaturated for saturated fat has long been recommended for the primary prevention of cardiovascular disease (CVD), but only a few prospective cohort studies have provided support for this advice.

Methods  We assessed the association of dietary linoleic and total polyunsaturated fatty acid (PUFA) intake with cardiovascular and overall mortality in a population-based cohort of 1551 middle-aged men. Dietary fat composition was estimated with a 4-day food record and serum fatty acid composition.

Results  During the 15-year follow-up, 78 men died of CVD and 225 of any cause. Total fat intake was not related to CVD or overall mortality. Men with an energy-adjusted dietary intake of linoleic acid (relative risk [RR] 0.39; 95% confidence interval [CI], 0.21-0.71) and PUFA (RR, 0.38; 95% CI, 0.20-0.70) in the upper third were less likely to die of CVD than men with intake in the lower third after adjustment for age. Multivariate adjustment weakened the association somewhat. Mortality from CVD was also lower for men with proportions of serum esterified linoleic acid (RR, 0.42; 95% CI, 0.21-0.80) and PUFA (RR, 0.25; 95% CI, 0.12-0.50) in the upper vs lower third, with some attenuation in multivariate analyses. Serum and to a lesser extent dietary linoleic acid and PUFA were also inversely associated with overall mortality.

Conclusions  Dietary polyunsaturated and more specifically linoleic fatty acid intake may have a substantial cardioprotective benefit that is also reflected in overall mortality. Dietary fat quality seems more important than fat quantity in the reduction of cardiovascular mortality in men.

Substitution of dietary polyunsaturated for saturated fat has been recommended for several decades in the primary prevention of cardiovascular disease (CVD), but only a few prospective cohort studies have provided support for this advice.^1-4 A biological basis has been provided by metabolic studies showing that polyunsaturated fat lowers, but saturated fat increases, serum low-density lipoprotein cholesterol (LDL-C) concentration.⁵ In contrast, evidence for reduction of total fat intake without modification of fat quality in the prevention of CVD is weak.^6,7

Trials in institutionalized patients have suggested that substitution of polyunsaturated for saturated fat may reduce coronary heart disease (CHD) or CVD.^8,9 A few prospective cohort studies have indicated that increased polyunsaturated fat intake,^1-4 a higher polyunsaturated-saturated fat ratio,⁴ or Keys score^1,2 may decrease incident CHD^1,2,4 or overall mortality,³ but more cohort studies have shown no association.⁶ The discrepancies may be due to small study size, imprecise dietary assessment, and insufficient control of confounding.⁷

Experimental and in vitro studies have suggested that n-3 fatty acids such as α-linolenic acid and fish oils have anti-inflammatory, antithrombotic, and antiarrhythmic properties^10,11 and improve insulin sensitivity.¹² In contrast, n-6 fatty acids such as linoleic and arachidonic acid have even been purported to promote inflammation, thrombosis, and insulin resistance.^11-13 In a review of the evidence, however, linoleic acid also appears to decrease thrombosis,¹⁰ and may also decrease arrhythmias¹⁴ and improve insulin sensitivity.¹⁵ Only a few prospective studies have supported a role of dietary linoleic¹⁶ or α-linolenic acid¹⁷ in primary CVD prevention.

Of the few cohort studies showing an inverse association of dietary polyunsaturated fatty acid (PUFA) intake with CVD, none has been population based, and only 1 study has found an association with overall mortality.³ None of these studies has supported their findings with serum biomarkers. The aim of the present study was to assess the association of dietary fat quantity and quality, specifically linoleic and α-linolenic acid, with CVD and overall mortality during a 15-year follow-up in a population-based cohort of 1551 middle-aged men who were free of CVD, cancer, and diabetes at baseline.

The Kuopio Ischaemic Heart Disease Risk Factor (KIHD) Study is a prospective population-based study.^15,18-22 The study population comprised a random age-stratified sample of 2682 men living in eastern Finland who were 42, 48, 54, or 60 years old at baseline between 1984 and 1989. The University of Kuopio Research Ethics Committee approved the study. All participants gave their written informed consent.

For the present study, all men with a history of CVD, diabetes, or cancer at baseline (n = 1123) were excluded. Men with missing data for both dietary and serum fatty acids were also excluded, leaving 1551 men for the analyses.

Dietary intake of PUFA and saturated (SAFA), monounsaturated (MUFA), and linoleic and α-linolenic fatty acids was assessed with 4-day food records.²³ The consumption of foods was assessed at the time of blood sampling at baseline. The dietary records were collected on 3 workdays and 1 weekend day. Data collection was carried out year-round, except for July, when most Finns are on vacation. The participants were instructed on the use of household measures to quantitatively record their food intake during the 4 days of data collection. A nutritionist gave the instructions and checked the completed food intake records. Dietary intake of nutrients and foods was calculated using NUTRICA software (version 2.5; National Public Health Institute, Turku, Finland). The software is based on mainly Finnish values of nutrient composition of foods. The nutrient compositions of foods in NUTRICA software version 2.5 have been analyzed mainly in the 1990s. For the assessment of dietary fatty acids, an earlier version (1.0) was used because of the changes in fatty acid contents of margarines in Finland during the last 15 years. In all, the database contains comprehensive data for 1300 food items and dishes and 30 nutrients. Intake of fatty acids and fiber was calculated in grams per day.

Serum esterified and nonesterified fatty acids were determined in 1 gas chromatographic run without preseparation as described previously.^15,20,21 Fatty acids were chromatographed in an NB-351 capillary column (HNU-Nordion, Helsinki, Finland) by a Hewlett-Packard 5890 Series II gas chromatograph (Hewlett-Packard Company, Avondale, Pa) with a flame ionization detector. The coefficient of variation (CV) for repeated measurements of major esterified fatty acids was about 5%. Because the relative degree of saturation of fatty acids varies among esterified fatty acid types (ie, cholesterol esters, phospholipids, and triacylglycerols), the esterified fatty acid concentrations were adjusted for serum LDL-C, high-density lipoprotein cholesterol (HDL-C), and triacylglycerol concentrations. The CV for major nonesterified fatty acids was about 15%. No adjustment was necessary for nonesterified fatty acids.

Blood glucose was measured using a glucose dehydrogenase method after precipitation of proteins by trichloroacetic acid. Serum insulin was determined with a Novo Biolabs radioimmunoassay kit (Novo Nordisk, Bagsvaerd, Denmark). Serum LDL and HDL fractions were separated by combined ultracentrifugation and precipitation. Lipoprotein fractions and triacylglycerols were measured enzymatically. Plasma α-tocopherol was measured by high-performance liquid chromatography. Plasma ascorbate was determined by a chromatographic method. Serum C-reactive protein was measured with an immunometric assay (Immulite High Sensitivity CR Assay; Diagnostic Products Corporation, Los Angeles, Calif).

Assessment of medical history, medications, family history of diseases, smoking, alcohol consumption, adult socioeconomic status, blood pressure, and leisure-time physical activity has been described previously.^18,19,22 Body mass index (BMI) was calculated as weight in kilograms divided by the square of height in meters.

Deaths were ascertained by computer linkage to the national death registry using the Finnish Social Security number. There were no losses to follow-up. All deaths that occurred between study entry (March 1984–December 1989) and December 2001 were included. Deaths were coded with the International Classification of Diseases, Ninth Revision. Deaths coded as CHD or stroke were validated according to the international criteria adopted by the World Health Organization MONICA (Monitoring of Trends and Determinants of Cardiovascular Disease) Project.^24,25

All serum fatty acids were analyzed as proportions (percentage in moles of total nonesterified or esterified fatty acids). To minimize confounding by energy, dietary variables from the 4-day food records were recorded as grams per day or milligrams per day and adjusted by total energy intake before further analysis (residual method).²⁶ The associations of dietary PUFA and linoleic acid intake (expressed as percentage of energy) with CVD mortality were nonetheless only slightly weaker than when using energy-adjusted variables (data not shown). Because plasma vitamin E is strongly associated with serum lipid and lipoprotein concentrations, α-tocopherol concentrations were standardized for LDL-C, HDL-C, and triglycerides. Associations of serum fatty acid proportions and dietary linoleic acid categorized into thirds were analyzed with forced Cox proportional hazards models. Variables with a skewed distribution (eg, serum insulin, C-reactive protein, and triacylglycerol concentrations, and dietary fat and fatty acid intake) were natural log transformed for analyses involving continuous variables. Covariates with missing values were assigned the group mean. Statistical significance was considered to be P<.05. All statistical analyses were performed with SPSS 11.0 for Windows (SPSS Inc, Chicago, Ill).

The median follow-up for the cohort of 1551 men was 14.6 years (range, 0.8-17.8 years), thus representing 22 645 person-years. During this time, 78 men died of CVD and 225 died of any cause. Because some men with dietary measures of fat intake had missing serum fatty acid measures and vice versa, there were 78 deaths from CVD and 220 deaths from any cause in analyses with dietary measures of fat intake. Corresponding numbers for analyses with serum fatty acid composition were 69 and 202, respectively. Smoking, blood pressure, BMI, and C-reactive protein were positively associated with CVD or overall mortality (Table 1), and socioeconomic status, plasma vitamin E, plasma ascorbic acid, and dietary fiber intake were inversely associated. In analyses with continuous variables, men with a lower dietary intake of linoleic and α-linolenic acid and PUFA had a higher cardiovascular and overall mortality after adjustment for age and year of examination (P<.01 to P<.05; Table 2). Proportions of esterified linoleic and α-linolenic acid and total PUFA and proportions of nonesterified linoleic acid were also inversely associated with death from CVD or any cause (P<.001 to P<.05; Table 2). Intake of total fat, SAFA, MUFA, and cholesterol were not associated with CVD.

Dietary linoleic acid intake was correlated with nonesterified (r = 0.34) and esterified (r = 0.49) linoleic acid proportions. Total PUFA intake was composed of, on average, 77% dietary linoleic acid and was highly correlated with total PUFA intake (r = 0.95). Dietary PUFA and SAFA intake were inversely correlated (r = –0.34). Dietary PUFA intake and serum PUFA esterified proportions were also correlated (r = 0.50). These correlations were highly significant (P<.001).

Nonesterified and esterified linoleic acid and PUFA proportions were also inversely associated with BMI and fasting insulin and glucose concentrations, whereas the corresponding fatty acids estimated from food records were not or were only weakly associated. In multivariate linear regression analyses, the determinants of serum esterified linoleic acid proportions were dietary linoleic acid intake (β = .51), alcohol intake (β = –.17), BMI (β = –.11), blood glucose concentrations (β = –.07), serum insulin concentrations (β = –.05), and age (β = –.02). The corresponding determinants of esterified PUFA proportions were dietary PUFA intake (β = .51), alcohol intake (β = –.10), BMI (β = –.03), glucose (β = –.09), insulin (β = –.05), and age (β = –.06).

Men with dietary linoleic acid intake in the upper third were up to 61% less likely to die of CVD than their counterparts whose intake was in the lower third after adjustment for age and year of examination (Table 3). Men whose α-linolenic acid intake was in the upper third were 30% to 42% less likely to die of CVD than men whose intake was in the lower third, but the associations were not significant. When using dietary α-linolenic acid intake as a natural log–transformed continuous variable, the association was statistically significant in all models: model 1: relative risk (RR), 0.37 (95% confidence interval [CI], 0.18-0.77); model 2: RR, 0.43 (95% CI, 0.20-0.90); model 3: RR, 0.41 (95% CI, 0.18-0.91); and model 4: RR, 0.42 (95% CI, 0.18-0.95).

Dietary PUFA intake in the upper third was associated with up to a 62% lower risk of CVD mortality after adjustment for age and year of examination (Table 3). The associations of the dietary PUFA/SAFA ratio with CVD mortality were somewhat weaker or similar (upper vs lower third, model 1: RR, 0.46 [95% CI, 0.25-0.83]; model 2: RR, 0.60 [95% CI, 0.33-1.11]; model 3: RR, 0.40 [95% CI, 0.18-0.89]; and model 4: RR, 0.44 [95% CI, 0.20-1.00]).

Dietary linoleic acid intake was associated with a lower overall mortality during follow-up after adjustment for age and examination year (upper vs lower third, model 1: RR, 0.66 [95% CI, 0.48-0.92], P for trend, .06), but not significantly after adjustment for lifestyle or dietary factors (data not shown). Men whose α-linolenic acid intake was in the upper third were 15% to 33% less likely to die of any cause than men whose intake was in the lower third, but the trend at best approached significance. Total PUFA intake was not significantly associated with overall mortality. The association of the dietary PUFA/SAFA ratio with overall mortality was significant (upper third vs lower third, model 1: RR, 0.71 [95% CI, 0.51-0.98]), but the association was not significant in models 2 through 4 (data not shown).

The associations of serum esterified fatty acid proportions with CVD mortality generally mirrored those of dietary fatty acids recorded in the food diary, but for linoleic acid proportions more attenuation was seen in models 3 and 4 (Table 4). The serum PUFA/SAFA ratio was also associated with CVD mortality except for model 4 (upper vs lower third, model 1: RR, 0.31 [95% CI, 0.15-0.64]; model 2: RR, 0.36 [95% CI, 0.17-0.76]; model 3: RR, 0.40 [95% CI, 0.18-0.87]; and model 4: RR, 0.49 [95% CI, 0.22-1.12]).

Esterified α-linolenic acid proportions were not associated with CVD mortality (Table 4). The linoleic/α-linolenic acid ratio (upper vs lower third, model 1: RR, 0.80 [95% CI, 0.46-1.40]), n-6/n-3 ratio (upper vs lower third, model 1: RR, 0.80 [95% CI, 0.45-1.43]), n-3 fatty acids (upper vs lower third, model 1: RR, 0.88, 95% CI, 0.43-1.80), or long chain PUFA (upper vs lower third, model 1: RR, 1.19 [95% CI, 0.67-2.09]) were not significantly associated with CVD mortality. Nonesterified linoleic acid proportions were also associated with CVD mortality, but the gradient was nonlinear, with the lowest risk in the middle third (middle vs lower third, model 1: RR, 0.40 [95% CI, 0.22-0.75]). Adjustment for confounding or mediating variables (models 2-4) weakened the association (data not shown).

Esterified linoleic acids proportions were or tended to be associated with a lower overall mortality (upper vs lower third, model 1: RR, 0.44 [95% CI, 0.30-0.67], P for trend, <.001; model 2: RR, 0.59 [95% CI, 0.40-0.86], P for trend, .005; model 3: RR, 0.66 [95% CI, 0.43-0.97], P for trend, .03; model 4: RR, 0.69 [95% CI, 0.41-1.03], P for trend, .08). The inverse associations for proportions of n-6 fatty acids and especially PUFA were even stronger and significant in all models (data not shown; P for trend, <.001 to.02). The serum PUFA/SAFA ratio was also associated with overall mortality (upper vs lower third, model 1: RR, 0.44 [95% CI, 0.30-0.64]; model 2: RR, 0.57 [95% CI, 0.39-0.84]; model 3: RR, 0.60 [95% CI, 0.40-0.90]; and model 4: RR, 0.65 [95% CI, 0.43-1.00]). Esterified α-linolenic acid proportions had a borderline association with overall mortality after adjustment for age and year of examination (upper vs lower third, model 1: RR, 0.72 [95% CI, 0.51-1.03], P for trend, .05), but not after adjustment for potential confounding or mediating variables (data not shown).

Middle-aged men with proportions of serum linoleic acid, n-6 fatty acids, and especially PUFA in the upper third were up to 3 times less likely to die of CVD than men with proportions in the lower third. Dietary intake of linoleic acid and total PUFA as assessed with a 4-day food record was also inversely associated with CVD, but total fat intake was not. Importantly, both serum and dietary fatty acid composition were assessed in this study.

Men with energy-adjusted dietary PUFA intake in the upper third had less than half the risk of premature CVD mortality as men with intake in the lower third, even after multivariate adjustment. The inverse correlations of dietary PUFA and SAFA intake indicate that the apparent benefit of PUFA intake probably comes about in part through substitution of PUFA for SAFA intake (in these men, by substitution of margarine for butter).²⁷ This, coupled with the inverse association of the dietary PUFA/SAFA ratio with CVD mortality, provides support for increasing PUFA intake at the expense of SAFA intake in the primary prevention of CVD and underscores the importance of dietary fat quality over quantity. These findings agree with those from the Western Electric Study, in which coronary death was the outcome,¹ and the Nurses Health Study, in which myocardial infarction was the outcome.⁴ Monounsaturated fatty acid intake was not associated with CVD or overall mortality. Possible beneficial effects in our study may be obscured by the high collinearity of MUFA intake with SAFA intake.

Serum esterified PUFA proportions were even more strongly associated with CVD mortality than dietary PUFA intake. Fasting insulin and glucose levels, BMI, smoking, and alcohol intake were also associated with serum fatty acid composition, probably, at least in part, independently of dietary fatty acid intake.²⁸ Although attenuated somewhat, the inverse association of especially serum PUFA proportions with CVD mortality remained significant even when adjusting for these variables.

After adjustment for lipids, the Pearson correlation of serum esterified linoleic acid proportions measured from whole serum with the energy-adjusted dietary intake of linoleic acid from the 4-day food diary was 0.50, much higher than that of serum cholesterol ester linoleic acid proportions with linoleic acid intake (r = 0.28), as estimated by a semiquantitative food frequency questionnaire in the Atherosclerosis Risk in Communities Study.²⁸ The correlation of serum linoleic acid proportions measured 4 years later with dietary linoleic acid intake at baseline in 895 men participating in the KIHD Study 4-year follow-up was 0.36 (data not shown). Serum esterified fatty acid proportions are thus a good measure of habitual dietary fat composition. Saturated fat intake in Finland has decreased since the mid-1980s²⁷ when the KIHD Study began, but our data indicate that the relative ranking of these men with respect to dietary fat quality may be stabile, at least during the first 4 years of follow-up.

Nonesterified linoleic acid proportions in the fasting state also reflect adipose and dietary fatty acid composition.²⁹ Men with higher nonesterified linoleic acid proportions seemed to have a lower CVD mortality, but the lowest mortality was in the middle third. The imprecision and low sensitivity of the measurement of nonesterified fatty acids in our study (coefficient of variation, 15%) may explain the discrepancy in results between nonesterified and esterified linoleic acid proportions.

Men with dietary intake of linoleic acid in the upper third were less than half as likely to die of CVD than men with intake in the lower third. Findings for serum proportions of linoleic acid and especially total n-6 fatty acids were similar, although some attenuation was seen in multivariate models. Because linoleic acid (18:2n-6) is elongated to arachidonic acid (20:4n-6) in the metabolism of serum fatty acids, n-6 fatty acid proportions are also an index of linoleic acid intake. The Nurses Health Study also found an inverse association of dietary linoleic acid intake with incident CHD.⁴ Serum linoleic acid proportions have also been inversely associated with incident CHD^30,31 and stroke³² in nested case-control^30,32 and cohort³¹ studies. For serum linoleic and n-6 fatty acids and total PUFA proportions, this apparent protective benefit carried over to lower overall mortality. Because CVD deaths made up only about one third of total deaths, this implies that n-6 fatty acids may also tend to decrease non-CVD mortality.

We found a nonsignificant trend for an inverse association of dietary and serum α-linolenic acid, but not serum fish oils, with CVD and overall mortality. The Lyon Heart Study,³³ with increased α-linolenic acid intake as one of the dietary interventions, and the Gruppo Italiano per lo Studio della Sopravvivenza nell’Infarto miocardico (GISSI)-Prevenzione trial,³⁴ which focused on fish oil supplements, show that n-3 fatty acids reduce risk in secondary CHD prevention. In the present study, α-linolenic acid intake at baseline was low. Furthermore, in this cohort fish oils, which were inversely associated with acute coronary events,²⁰ come mainly from nonfatty lake fish. In Finland, lake fish are also a major dietary source of mercury, which was positively associated with coronary events.²⁰

Total PUFA intake was low in these middle-aged Finnish men, but the n-6/n-3 ratio was high (6.5 ± 1.8). We found no association of the serum n-6/n-3 ratio with CVD or overall mortality. Some in vitro, animal, and cross-sectional studies have suggested that a high n-6/n-3 ratio may increase the risk of CVD, but the evidence suggesting that it may have a clinical or public health relevance is weak⁶ and inconsistent with the present findings.

Strengths of this study include its longitudinal population-based design, detailed assessment of potential confounding and mediating factors, and assessment of dietary fat composition with both food records and serum biomarkers. The consistency of the associations of dietary intake of linoleic acid and PUFA and corresponding serum proportions, especially with CVD mortality, lends further validity to the findings. The main findings suggesting that increased linoleic acid or PUFA intake or substitution of polyunsaturated for saturated fat decreases CVD mortality furthermore agree with those of the Nurses Health Study.⁴ Nonetheless, these associations may not apply to high polyunsaturated fat and low saturated fat diets or to other ethnic groups.

We found strong inverse associations of dietary and serum linoleic acid and PUFA, but no association of dietary total fat intake, with CVD mortality. Dietary fat quality thus seems more important than fat quantity in the reduction of CVD mortality in middle-aged men. Carrying out recommendations to replace saturated fat with polyunsaturated fat in the primary prevention of cardiovascular disease may substantially decrease CVD and to a lesser degree overall mortality.

Correspondence: Kristiina Nyyssönen, PhD, Research Institute of Public Health, University of Kuopio, PO Box 1627, FIN-70211 Kuopio, Finland (Kristiina.Nyyssonen@uku.fi).

Accepted for Publication: August 4, 2004.

Financial Disclosure: None.

Funding/Support: The Kuopio Ischaemic Heart Disease Risk Factor (KIHD) Study was supported by grants 41471, 45155, 1041086, and 2041022 from the Academy of Finland, Helsinki; grants 167/722/96, 157/722/97, and 156/722/98 from the Ministry of Education of Finland, Helsinki; and grant HL44199 from the National Heart, Lung and Blood Institute, Bethesda, Md.

Acknowledgment: We thank the staff of the Research Institute of Public Health, University of Kuopio, and Kuopio Research Institute of Exercise Medicine for data collection in the KIHD study.