American Journal of
EPIDEMIOLOGY
Volume 144
Copyright O 1996 by The Johns Hopkins University
Number 4
School of Hygiene and Public Hearth
August 15, 1996
Sponsored by the Society for Epldemlotoglc Research
ORIGINAL CONTRIBUTIONS
Relation of Smoking and Alcohol Consumption to Serum Fatty Acids
Joel A. Simon, 12 Josephine Fong,2 John T. Bernert, Jr.,3 and Warren S. Browner1-2
To examine the relation of cigarette smoking and alcohol consumption to serum fatty acid levels, the authors
conducted a cross-sectional study of 190 men who were enrolled in the Murtiple Risk Factor Intervention Trial
between 1973 and 1976. After controlling for dietary fat, cholesterol, energy intake, and other potential
confounders, the authors found that smoking and alcohol intake were associated with the serum cholesterol
ester and phospholipid levels of several fatty acids. As the number of cigarettes smoked per day increased,
the levels of cholesterol ester and phospholipid palmitoleic acid (16:1) and oleic acid (18:1) and the levels of
phospholipid dihomogammalinolenic acid (20:3) and omega-9 eicosatrienoic acid (20:3) increased (all p's <
0.01). Serum levels of phospholipid omega-3 docosahexaenoic acid (22:6) and cholesterol ester and phospholipid arachidonic acid (20:4) were inversely associated with smoking (all p's £ 0.01). As the number of
alcoholic drinks per week increased, levels of cholesterol ester and phospholipid palmitic acid (16:0) and oleic
acid (18:1), cholesterol ester myristic acid (14:0), and phospholipid palmitoleic acid (16:1), adrenic acid (22:4),
and omega-9 eicosatrienoic acid (20:3) increased (all p's < 0.05), whereas levels of cholesterol ester and
phospholipid linoleic acid (18:2) and phospholipid stearic acid (18:0) and the serum polyunsaturated fat:
saturated fat ratio decreased (all p's £ 0.01). These results suggest that smoking and alcohol consumption
may influence the absorption, synthesis, or metabolism of serum fatty acids. Studies that use serum fatty acid
levels as indicators of dietary fat intake should control for the effects of cigarette smoking and alcohol
consumption. Am J Epidemiol 1996;144:325-34.
alcohol drinking; alcohol, ethyl; coronary disease; diet; fatty acids; smoking; tobacco
smoking and alcohol consumption may affect coronary heart disease risk partly through their effects on
the fatty acid composition of serum cholesterol esters
and phospholipids. Such changes in fatty acid composition could influence blood clotting (4) and, in turn,
the risk of coronary heart disease. We have shown in
previous work that the serum level of cholesterol ester
palmitic acid (16:0) is directly associated with coronary heart disease risk, whereas serum levels of phospholipid docosapentaenoic acid (22:5) and docosahexaenoic acid (22:6), omega-3 fatty acids, are
inversely associated with coronary heart disease risk
(5). We also found that the level of cholesterol ester
alpha-linolenic acid (18:3), the parent compound of
the omega-3 class of fatty acids, is inversely associated with the risk of stroke (6). These associations
Part of the increased coronary heart disease risk
associated with cigarette smoking and the decreased
risk associated with moderate alcohol consumption
may result from changes in lipid and lipoprotein levels
and platelet reactivity (1-3). We hypothesized that
Received for publication July 28, 1995, and In final form April 19,
1996.
Abbreviations: MRFIT, Murtiple Risk Factor Intervention Trial.
1
General Internal Medicine Section, Medical Service, Veterans
Affairs Medical Center, San Francisco, CA.
2
Division of Clinical Epidemiology, Department of Epidemiology
and Biostatistics, School of Medicine, University of California, San
Francisco, CA.
3
Clinical Biochemistry Branch, Division of Environmental Health
Laboratory Sciences, National Center for Environmental Health,
Centers for Disease Control and Prevention, Atlanta, GA.
Reprint requests to Dr. Joel A. Simon, General Internal Medicine
Section (111A1), Veterans Affairs Medical Center, 4150 Clement
Street San Francisco, CA 94121.
325
326
Simon et al.
were independent of other cardiovascular disease risk
factors.
Biochemical fatty acid profiles of serum lipids generally reflect established patterns of fatty acid intake
(7). Correlations between dietary intake, as estimated
by food frequency questionnaires, and the plasma fatty
acid composition of cholesterol esters and phospholipids have been reported to range from approximately
0.2 to 0.3 for diet-derived essential polyunsaturated
fatty acids, such as linoleic acid (18:2) and alphalinolenic acid (18:3), and from 0.2 to 0.4 for eicosapentaenoic acid (20:5) and docosahexaenoic acid
(22:6) (8). Nonessential saturated and monounsaturated fatty acids are less reliable indicators of dietary
fatty acid intake, because these levels also reflect fatty
acid synthesis and metabolism.
In order to examine the relation of smoking and
alcohol consumption to serum fatty acid composition,
we conducted a cross-sectional analysis of data collected from a subset of 190 men in the Multiple Risk
Factor Intervention Trial (MRFIT), using stored frozen serum samples that had been collected at the outset
of the study.
MATERIALS AND METHODS
Subjects
MRFIT was a primary coronary heart disease prevention trial that studied the effects of cholesterol and
blood pressure lowering and smoking cessation among
men at high risk for coronary disease. Between December 1973 and February 1976, 12,866 US men aged
35-57 years were enrolled and randomly assigned to
either a Special Intervention group or a Usual Care
group after screening was completed (9). Men in the
Usual Care group (n = 6,438) continued their usual
medical care and were evaluated yearly by MRFIT
staff (10).
We analyzed data from 190 men in the Special
Intervention and Usual Care groups who served as
control subjects in two nested case-control studies that
examined the relation of serum fatty acid levels to
incident coronary heart disease and incident stroke (5,
6). Because of missing data on plasma lipid and glucose levels, 186 of these men were included in the
analysis.
recorded. Tobacco use (cigarettes per day) and alcohol
intake (drinks per week) were determined by selfreport, and nutrient intake at baseline was estimated
using a 24-hour diet recall. These methods have been
described in detail elsewhere (10, 11).
Fasting plasma total cholesterol, low density lipoprotein cholesterol, high density lipoprotein cholesterol, triglyceride, and glucose levels were determined
at baseline. At the Centers for Disease Control and
Prevention, we measured lipoprotein fatty acid levels
from serum specimens obtained at baseline and frozen
at -55°C for the entire interim period. We extracted
serum aliquots by the procedure of Folch et al. (12),
isolated the cholesterol esters and phospholipids by
thin-layer chromatography, and transesterified them to
the methyl esters (13). We purified the fatty acid
methyl esters on a small silicic acid column and analyzed them by capillary gas-liquid chromatography on
a 0.2-mm X 50-m FFAP column mounted in a
Hewlett-Packard 5880 gas chromatograph (HewlettPackard Company, Palo Alto, California) (14). The
instrument was calibrated before each series of analyses using a quantitative standards mixture (GLC68A) from Nu Chek Prep (Elysian, Minnesota). Specific fatty acid levels were expressed as the percentage
of total fatty acids in the cholesterol ester and phospholipid fractions. Only identified fatty acid peaks
were used to estimate the fatty acid composition of the
cholesterol esters and phospholipids. We chose to
measure and examine the fatty acid composition of
serum cholesterol esters and phospholipids rather than
that of serum triglycerides, because of their longer
half-life. The fatty acid composition of serum cholesterol esters and phospholipids may better reflect usual
dietary fatty acid intake patterns (13).
We evaluated the stability of the frozen serum specimens prior to this analysis. We analyzed 12 randomly
selected samples for their malondialdehyde (15) and
conjugated diene (16) content. We also assayed Folch
extracts of the samples for fluorescent degradation
products, their vitamin A and E content (17), and fatty
acid profiles. Comparison of the results with reference
pools and fresh serum indicated that very little oxidative damage had occurred. There were minor but statistically significant increases in malondialdehyde and
conjugated diene content in comparison with fresh
serum (p < 0.02).
Measurements
At baseline, subjects were weighed after removing
their shoes and outdoor clothing. Seated blood pressure was measured in mmHg using a random-zero
manometer and a cuff size appropriate for arm circumference. Three blood pressure readings were obtained,
and the average of the second and third readings was
Statistical methods
We analyzed the association of smoking (0 = nonsmoker, 1 = 1-20 cigarettes/day, 2 = 21-40 cigarettes/day, 3 = >40 cigarettes/day) and alcohol consumption (per 10 drinks/week) with serum fatty acid
levels (measured as a percentage of fatty acid compoAm J Epidemiol
Vol. 144, No. 4, 1996
Smoking, Alcohol, and Fatty Acids
sition) using general linear models. We verified that a
linear model was appropriate by examining the association between smoking and serum fatty acid levels
across all four categories of smoking, checking for
evidence of nonlinear relations. We performed univariate regression, examining the relation of smoking and
alcohol consumption to each fatty acid, and then performed multivariate regression, adjusting for age,
body mass index, plasma glucose, total energy intake,
cholesterol intake, polyunsaturated fatty acid intake,
monounsaturated fatty acid intake, saturated fatty acid
intake, and the MRFTT selection criteria (plasma cholesterol level and diastolic blood pressure). Smoking
and alcohol consumption were included in all multivariate models. Because the quadratic terms for alcohol intake were statistically significant for two fatty
acids, cholesterol ester stearic acid (18:0) and palmitoleic acid (16:1), multivariate models that examined
the relation between smoking and alcohol consumption and the cholesterol ester fatty acids stearic acid
and palmitoleic acid included alcohol intake as both a
linear and a quadratic term.
We determined the correlations between the dietary
intake variables to evaluate the possibility of collinearity. The strongest correlation was between monounsaturated fat intake and polyunsaturated fat intake (r =
0.6). Because multivariate models that included and
excluded monounsaturated fat intake produced virtually identical results, we chose to present the findings
from models that included monounsaturated fat intake.
We calculated univariate and multivariate regression coefficients and their 95 percent confidence intervals. With the exception of phospholipid eicosapentaenoic acid (20:5) values, which required logarithmic
transformation, data on the fatty acid variables generally had approximately normal distributions. Using the
slopes and 95 percent confidence intervals derived
from the multivariate models, we estimated the relative percentage differences in serum fatty acid levels
accounted for by smoking and alcohol consumption
(i.e., the predicted difference divided by the absolute
percentage of each fatty acid in the cholesterol ester
and phospholipid fractions). We considered two-tailed
p values less than 0.05 to be statistically significant,
unadjusted for multiple comparisons (18). SAS software was used in all statistical analyses (19).
RESULTS
Although the men recruited for MRFIT were at
increased risk for coronary heart disease because they
smoked, had high diastolic blood pressures, or had
high plasma cholesterol levels, over 50 percent of the
MRFIT subjects analyzed in this study were nonsmokers (table 1). Six percent of our study subjects were
nondrinkers, and 4 percent were heavy drinkers (^5
alcoholic drinks per day). On average, however, most
of the men consumed one or two drinks per day.
Consistent with previous observations (20), the principal fatty acids in the serum cholesterol ester fraction
TABLE 1. Baseline characteristics of 190 study participants from the Multiple Risk Factor Intervention
Trial, 1973-1976
Variable
Mean
SD*
Medan
Range
Age (years)
Systolic blood pressure (mmHg)
Diastolic blood pressure (mmHg)
Body mass indext
Tobacco use (cigarettes/day)^
Diet
Total energy intake (kcal/day)
Cholesterol (mg/day)
Monounsaturated fat (% of kcal)
Polyunsaturated fat (% of kcal)
Saturated fat (% of kcal)
Total fat (% of kcal)
Alcohol intake (drinks/week)
Plasma glucose (mg/dl)$
Plasma cholesterol (mg/dl)
Plasma low density lipoprotein cholesterol
(mg/dl)§
Plasma high density lipoprotein cholesterol
(mg/dl)§
Plasma trigjycendes (mg/dl)
* SD, standard deviation,
t Weight (kg)/height (m)».
$n*> 186.
§n-189.
Am J Epidemiol
Vol. 144, No. 4, 1996
327
5O.0
138.0
91.8
27.1
13.9
5.6
14.0
7.8
3.0
18.6
50.9
137.0
92.0
26.8
0.0
35.3-58.1
100.0-181.0
64.0-116.0
19.6-36.7
0.0-80.0
2,458
450
14.6
6.1
14.3
37.6
11.7
100
244
967
306
4.5
3.0
4.8
9.7
11.4
12
40
2,348
349
14.6
5.6
13.6
37.4
9.0
99
240
775-6,228
44-1,296
0.9-33.0
1.0-14.6
1.6-35.3
5.0-69.1
0.0-70.0
67-171
124-370
163
40
165
40-299
44
188
12
149
42
149
21-88
44-1,495
328
Simon et al.
were oleic acid (18:1) and the omega-6 fatty acid
linoleic acid (18:2) (table 2). There was a wider distribution of fatty acids in the phospholipid fraction,
including the omega-3 fatty acids eicosapentaenoic
acid (20:5), docosapentaenoic acid (22:5), and docosahexaenoic acid (22:6).
Cigarette smoking
In univariate analyses, smoking was associated with
higher serum levels of myristic acid (14:0), palmitoleic acid (16:1), oleic acid (18:1), alpha-linolenic acid
(18:3), dihomogammalinolenic acid (20:3), and
omega-9 eicosatrienoic acid (20:3) (all p's < 0.05).
Smoking was associated with lower serum levels of
docosahexaenoic acid (22:6), linoleic acid (18:2), and
arachidonic acid (20:4) (ally's ^ 0.01). Smoking was
also associated with a lower serum polyunsaturated:
saturated fatty acid ratio (all p's < 0.05). For serum
fatty acids that were significant in univariate analyses,
we compared the mean serum fatty acid level in each
of the four smoking categories (0, 1-20, 21-40, and
>40 cigarettes per day). With the exception of arachidonic acid, there were monotonic increases or decreases in the mean serum fatty acid level as smoking
increased. Mean arachidonic acid levels remained un-
changed at 20 cigarettes or less per day, but they
decreased consistently among subjects who smoked
more than 20 cigarettes per day.
In multivariate analyses, we found that smoking was
independently associated with higher levels of palmitoleic acid (16:1), oleic acid (18:1), phospholipid dihomogammalinolenic acid (20:3), and omega-9 eicosatrienoic acid (20:3) and with lower levels of
docosahexaenoic acid (22:6) and arachidonic acid
(20:4) (all p's < 0.01) (table 3). The serum polyunsaturated : saturated fatty acid ratio was not associated
with smoking after multivariate adjustment.
Alcohol consumption
In univariate analyses, alcohol consumption was
associated with higher serum levels of myristic acid
(14:0), palmitic acid (16:0), palmitoleic acid (16:1),
oleic acid (18:1), alpha-linolenic acid (18:3), adrenic
acid (22:4), log eicosapentaenoic acid (20:5), and
omega-9 eicosatrienoic acid (20:3) (all p's < 0.05).
Alcohol consumption was associated with lower serum levels of stearic acid (18:0) and linoleic acid
(18:2) and a lower serum polyunsaturated fatty acid:
saturated fatty acid ratio (all p's ^ 0.01).
TABLE 2. Percentages of fatty acid composition of cholesterol esters and phosphollpids in 190
participants from the Multiple Risk Factor Intervention Trial, 1973-1976*
Cholesterol ester
PhospttoipM
Mean
Mean
SO
sot
Saturated fatty acids
Myriatic (14:0)
Palmitic (16:0)
Stearic (18:0)
Monounsatu rated fatty acids
Palmitoleic (16:1)
Oleic (18:1)
Polyunsaturated fatty acids
Omaga-3 fatty acids
Alpha-linolenic (18:3)
Eicosapentaenoic (20:5)
Docosapentaenoic (22:5)
Docosahexaenoic (22:6)
Omega-6 tatty acids
Unoteic (18:2)
Eicosadienoic (20:2)
Dihomogammalinotenic (20:3)
Arachidonic (20:4)
Adrenic (22:4)
Docosapentaenoic (22:5)
Omega-9 fatty acids
Eicosatrienoic (20:3)
Potyunsaturated:saturated fatty acid ratio
1.10
0.20
1.08
0.23
0.27
28.18
4.87
0.12
1.83
1.29
3.71
20.09
1.82
2.98
0.72
11.41
0.44
1.99
0.40
0.14
0.13
0.74
1.02
3.36
0.07
0.60
0.27
1.24
52.70
5.24
0.79
8.44
0.17
1.92
21.67
0.33
3.55
12.60
0.55
0.40
3.04
0.06
0.79
2.25
0.14
0.15
0.17
0.11
1.02
0.09
0.60
12.15
4.55
0.68
* Missing fatty acid values were nondetectaWe.
f SD, standard deviation.
Am J Epidemiol
Vol. 144, No. 4, 1996
Smoking, Alcohol, and Fatty Acids
329
TABLE 3. Multivariats association* between smoking and percentage difference In serum fatty acid
composition in 190 participants from the Multiple Risk Factor Intervention Trial, 1973-1976
Cholesterol ester
Pnotphot^ld
95% CI*
Stopet
95% Cl
Stopot
Saturated fatty acids
Myristjc(14:o)
Palmitic (16:0)
Stearic (18:0)
0.02
-0.13
0.02
Monounsatu rated fatty adds
Palmitoleic(16:1)
CHeic(18:1)
0.47
0.80
-0.01 to 0.05
-0.31 to 0.05
-0.02 to 0.05
0.22 to 0.72*
0.37 to 1.24*
0.02
-0.18
0.02
0.13
0.61
0.00 to 0.04
-0.47 to 0.10
-0.19 to 0.24
0.07 to 0.20*
0.31 to 0.91*
Polyunaaturated tatty acids
Omega-3 tatty acids
Alpha-linolenic (18:3)
Eicosapentaenoic (2O:5)§
Docosapentaenoic (22:5)
Docoaahexaonotc (22:6)
Omega-6 tatty acids
Unoleic(18:2)
Eicosadienoic (20:2)
Dihomogammalinolenic (20:3)
Arachidonic (20:4)
Adrenic (22:4)
Docosapentaenoic (22:5)
Omega-9 fatty acids
Etcosatrianoic (20:3)
0.01
-0.01 to 0.03
0.01
-0.01
-0.04
-0.33
0.00
-0.09
-0.09
-0.53
to 0.02
to 0.07
to 0.01
to -0.14*
-0.63
-1.41 to 0.14
0.02
-0.57
-0.01 to 0.05
-0.88 to -0.25*
0.17
0.01
0.19
-0.57
0.01
0.00
-0.32
0.00
0.06
-0.94
-0.02
-0.03
to 0.65
to 0.02
to 0.32*
to -0.20*
to 0.03
to 0.02
Polyunsaturated:saturated tatty a d d ratio
-0.05
0.02
-0.16 to 0.06
-0.01
0.01 to 0.04*
-0.03 to 0.00
* p £0.01.
t Slope denotes absolute percentage difference in fatty acid level (e.g., 0.02 denotes 0.02%) for each increment of smoking (0 - nonsmoker, 1 » 1-20 cigarettes/day, 2 - 21-40 dgarettes/day, 3 - >40 cigarettes/day),
adjusted for age, body mass index, alcohol intake, plasma glucose level, total calories, cholesterol intake, intake
of polyunsaturated, monounsaturated, and saturated fatty adds, dastolic blood pressure, and plasma cholesterol
level. Multivariate models for cholesterol ester stearic acid (18:0) and palmitoleic add (16:1) also induded alcohol1.
$ Cl, confidence interval.
§ Slope for log eicosapentaenoic acid.
In multivariate analyses, we found that alcohol consumption was independently associated with higher
levels of cholesterol ester myristic acid (14:0),
palmitic acid (16:0), and oleic acid (18:1) and phospholipid palmitic acid, palmitoleic acid (16:1), oleic
acid, adrenic acid (22:4), and omega-9 eicosatrienoic
acid (20:3). Alcohol consumption was independently
associated with lower levels of phospholipid stearic
acid (18:0) and cholesterol ester and phospholipid
linoleic acid (18:2) (all p's < 0.05) (table 4). The
serum polyunsaturated fatty acid: saturated fatty acid
ratio remained inversely associated with alcohol intake
after multivariate adjustment (p < 0.01).
The relations between alcohol consumption and levels of cholesterol ester stearic acid (18:0) and palmitoleic acid (16:1) were nonlinear. As alcohol consumption increased from 0 to 30 drinks per week, the
level of cholesterol ester stearic acid decreased. At
higher levels of alcohol consumption, however, the
Am J Epidemiol
Vol. 144, No. 4, 1996
level of cholesterol ester stearic acid increased. As the
level of alcohol consumption increased, the level of
cholesterol ester palmitoleic acid generally increased,
although not monotonically, as reflected in a statistically significant alcohol2 term (p = 0.01).
Predicted effects on serum fatty acid levels
Smoking and alcohol consumption had substantial
predicted effects on several cholesterol ester and phospholipid fatty acids (table 5). For example, heavy
smokers (>40 cigarettes per day) had 54 percent
higher levels of phospholipid palmitoleic acid (16:1)
and 30 percent lower levels of phospholipid docosahexaenoic acid (22:6) than nonsmokers. The predicted
effects of alcohol consumption were most notable for
palmitoleic acid (16:1), where every 10 drinks consumed per week were associated with a 21 percent
increase in phospholipid palmitoleic acid levels.
330
Simon et al.
TABLE 4. Multtvarlate association* between aJcohol consumption (per 10 drinks/week) and percentage
difference in serum fatty acid composition in 190 participants from the Multiple Risk Factor Intervention
Trial, 1973-1976
siopet
Saturated tatty acids
Myristic(14:0)
Palmitic (16:0)
Stearic (18:0)
Monounsaturated fatty acids
Palmitoleic(16:1)
Oleic(18:1)
Polyunsaturated fatty acids
Omega-3 fatty acids
Alpha-linolenic (18:3)
Eicosapentaenoic (20:5)ll
Docosapentaenoic (22:5)
Docosahexaenoic (22:6)
Omega-6 fatty acids
Linoleic(18:2)
Eicosadienoic (20:2)
Dihomogammalinolenic (20:3)
Arachidonic (20:4)
Adranic (22:4)
Docosapentasnoic (22:5)
Omega-9 fatty acids
Eicosatrienoic (20:3)
Poryunsaturated:saturated fatty acid ratio
Cholesterol ester
95%CIJ
PhosphoBpkJ
Siopet
9 5 % Cl
0.03
0.22
-0.14
0.002§
0.00 to 0.06*
0.08 to 0.37**
-0.21 to - 0 . 0 8 "
0.001 to 0.003**
0.01
0.57
-0.28
-0.01 to 0.02
0.33 to 0.80**
-0.46 t o - 0 . 1 0 * *
1.24
-0.011§
0.71
0.81 to 1.66**
-0.019 to-0.002*
0.35 to 1.07**
0.15
0.10 to 0.20**
0.28
0.03 to 0.53*
0.02
0.00 to 0.04
-1.81
-2.45
-0.02
0.17
-0.05 to 0.00
-0.09 to 0.43
-0.19
to-1.18**
-0.28 t o - 0 . 1 0 * *
0.00
0.05
-0.01
0.01
-0.01
-0.01
-0.05
-0.16
to 0.01
to 0.12
to 0.03
to 0.17
-0.84
-0.01
-0.10
0.16
0.02
0.01
-1.24
-0.01
-0.21
-0.14
0.00
-0.02
to -0.44**
to 0.00
to 0.00
to 0.47
to 0.04*
to 0.03
0.04
0.02 to 0.05**
-0.02
-0.04 to-0.01 **
* p<0.05; •• p^O.01.
t Slope denotes absolute percentage difference in fatty acid level (e.g., 0.03 denotes 0.03%) for every 10
drinks per week, adjusted for age, body mas3 index, smoking, plasma glucose level, total calories, cholesterol
intake, intake of polyunsaturated, monounsaturated, and saturated fatty acids, diastolic blood pressure, and plasma cholesterol level. Multivariate models for cholesterol ester stearic acid (18:0) and palmitoleic acid (16:1) also
included alcohol 2 .
£ Cl, confidence interval.
§ Slope for alcohol 2 .
II Slope for log eicosapentaenoic acid.
DISCUSSION
Cigarette smoking and alcohol consumption were
independently associated with the serum lipoprotein
levels of several cholesterol ester and phospholipid
fatty acids. These associations were present after adjustment for dietary fat, energy, cholesterol, and body
mass index.
Saturated fatty acids
Although higher blood levels of saturated fatty acids
are associated with an increased risk of coronary heart
disease (5, 21-23), we did not find any association
between smoking and saturated fatty acid levels. However, alcohol consumption was associated with higher
levels of palmitic acid (16:0) and myristic acid (14:0)
and lower levels of phospholipid stearic acid (18:0). In
agreement with our findings, some previous reports
found that alcohol consumption was independently
associated with higher levels of palmitic acid (24) and
lower levels of stearic acid (25). In contrast, Cambien
et al. (25) reported lower levels of myristic acid as
alcohol consumption increased. Because higher serum
levels of palmitic acid have been associated with increased risk of coronary heart disease (5, 22, 23, 26)
and because moderate alcohol consumption is associated with a lower risk of coronary heart disease (7),
these findings suggest that other effects of alcohol,
such as an increase in high density lipoprotein cholesterol levels, outweigh any potentially adverse effects
of alcohol on palmitic acid or myristic acid.
Monounsaturated fatty acids
Several studies have reported that higher serum and
adipose tissue levels of palmitoleic acid (16:1) are
Am J Epidemiol Vol. 144, No. 4, 1996
Smoking, Alcohol, and Fatty Acids
TABLE 5. Multivariate-adjusted predicted levels of senim
fatty acids in 190 participants from the Multiple Risk
Intervention Trial, 1973-1976
Predicted
e«ecl*(%)
95%Clt
38
54
12
16
18 to 58
29 to 83
6 to 19
8 to 24
Smoking^
Monounsaturated fatty actds
Cholesterol ester palmitoteic (16:1)
Phospholipid palmitoleic (16:1)
Cholesterol e3teroleic (18:1)
Phospholipid oleic (18:1)
Polyunsaturated tatty acids
Omega-3 tatty adds
Phospholipid docosahaxaenoic (22:6)
Omega-6 fatty acids
Phospholipid dibomogammaJinolenic
(20:3)
Cholesterol ester arachidonic (20:4)
Phospholipid arachidonic (20:4)
Omega-9 fatty acids
Phospholipid eicosatrienoic (20:3)
-30
-47 to -13
16
-20
-14
5 to 27
-31 to - 9
-22to-5
35
18 to 71
5
Alcohol consumptions
Saturated fatty acids
Cholesterol ester myristic (14:0)
Cholesterol ester palmitic (16:0)
Phospholipid paJmitic (16:0)
Phospholipid stearic (18:0)
2
2
-2
Oto 10
1 to3
1 to3
-3to-1
Monounsaturated fatty acids
Phospholipid paJmitoleic (16:1)
Cholesterol ester oleic (18:1)
Phospholipid oleic (18:1)
21
4
3
14 to 28
2to5
0to5
Polyunsaturated fatty acids
Omega-6 fatty acids
Cholesterol ester linoleic (18:2)
Phospholipid linoleic (18:2)
Phospholipid adrank:(22:4)
Omega-9 fatty acids
Phospholipid eicosatrienoic (20:3)
-3
-4
4
-5to-2
-6to-2
0to7
24
12 to 29
* The predicted effect is the percentage difference in serum fatty
acid levels that is accounted for by smoking and alcohol consumption, divided by the absolute percentage of each fatty acid in the
cholesterol ester and phospholipid fractions. Results were adjusted
br age, body mass index, smoking, alcohol intake, plasma glucose
level, total calories, cholesterol intake, intake of polyunsaturated,
monounsaturated, and saturated fatty acids, diastolic blood pressure, and plasma cholesterol level. Multtvanate models for cholesterol ester paJmitoleic acid (16:1) also included alcohol'.
t Cl, confidence interval.
i Compares the nonsmoking category with the heaviest smoking category (>40 cigarettes/day).
§ Per 10 drinks per week.
associated with coronary heart disease (26-29) and
stroke (30). Other studies have found higher levels of
plasma or platelet oleic acid (18:1) in subjects with
ischemic heart disease (29) and in subjects who died
from fatal cerebrovascular accidents and myocardial
infarctions (31). Although it is possible that these
Am J Epidemiol
Vol. 144, No. 4, 1996
331
associations between monounsaturated fatty acids and
coronary heart disease reflect the metabolism of dietary saturated fatty acids to monounsaturated fatty
acids, none of these studies controlled for the effects
of smoking or alcohol consumption.
On the basis of a report that found alcohol to be
associated with higher levels of cholesterol ester
palmitoleic acid and oleic acid, it was proposed that
palmitoleic acid might serve as a marker of alcohol
consumption (32). Cambien et al. found that smoking
was associated with higher levels of cholesterol ester
palmitoleic acid and oleic acid and that alcohol consumption was associated with higher levels of palmitoleic acid (25). Leng et al. (33) also reported that
smokers had higher levels of cholesterol ester oleic
acid. In agreement with these studies, we found that
smoking and alcohol consumption were independently
associated with higher serum levels of palmitoleic acid
and oleic acid.
Because studies that found monounsaturated fatty
acid levels to be associated with increased risk of
coronary heart disease did not control for the effects of
smoking, the possibility of confounding by smoking
cannot be excluded. This is relevant, because dietary
monounsaturated fat intake and moderate alcohol consumption, both of which may increase blood monounsaturated fatty acid levels, may be protective against
coronary heart disease (7, 34).
Polyunsaturated fatty acids
Omega-3 fatty acids. A high dietary intake of
omega-3 fatty acids has been linked to lower cardiovascular disease risk in certain populations (35). We
found that smoking was associated with lower levels
of phospholipid omega-3 docosahexaenoic acid (22:
6). However, there was no association between alcohol
consumption and omega-3 polyunsaturated fatty acid
levels. These findings agree with those of Leng et al.
(33). Smoking is known to increase platelet aggregation (1), an effect consistent with smoking-associated
differences in docosahexaenoic acid levels.
Omega-6 fatty acids. The omega-6 polyunsaturated fatty acids have not been consistently associated
with coronary heart disease. Increased platelet, adipose tissue, plasma, and serum levels of linoleic acid
(18:2) and increased intake of linoleic acid have been
associated with a lower risk of coronary disease in
some studies (22, 23, 36, 37) and a higher risk of
coronary disease in others (38, 39). None of these
studies controlled for the effects of alcohol. Lower
plasma and adipose tissue levels of dihomogammalinolenic acid have been reported variably to increase
(23, 40) or decrease (29) coronary heart disease risk.
Levels of arachidonic acid have been reported to be
332
Simon et al.
decreased in subjects with coronary heart disease and
vascular disease (21, 23, 31, 37, 41-43), but some of
these studies did not control for the effects of smoking
(21,31,41).
We found that several omega-6 polyunsaturated
fatty acids were associated with smoking and alcohol
consumption. Smoking was associated with higher
serum levels of phospholipid dihomogammalinolenic
acid (20:3), a precursor of arachidonic acid and prostaglandin Ej that has been reported to reduce endothelial prostacyclin production (44, 45). Smoking was
also associated with lower serum levels of cholesterol
ester and phospholipid arachidonic acid (20:4), a precursor of prostacyclin and thromboxane A 2 (44). The
association that we found between smoking and lower
serum levels of arachidonic acid agrees with the findings of other studies (33, 46). In several studies that
did not control for the effects of smoking, higher
levels of arachidonic acid have been found in subjects
with coronary heart disease (28, 29, 47).
Alcohol consumption was associated with lower
serum levels of cholesterol ester linoleic acid (18:2).
These results agree with those of other studies (24, 25,
32, 48). Because linoleic acid inhibits platelet aggregation (44) and lowers low density lipoprotein cholesterol levels when substituted for saturated fats in the
diet (49), it may be related to decreased coronary heart
disease risk. However, it has also been hypothesized
that higher concentrations of linoleic acid may increase lipid peroxidation, which in turn might increase
the risk of coronary heart disease (39). Alcohol consumption was also associated with higher serum levels
of phospholipid adrenic acid (22:4), a metabolite of
arachidonic acid. The effect of adrenic acid on coronary heart disease risk is not known.
Omega-9 fatty acids. Eicosatrienoic acid (20:3)
may be associated with an increased risk of coronary
heart disease (36, 41). Unlike the omega-3 and omega6 fatty acids, eicosatrienoic acid is a nonessential fatty
acid that promotes platelet aggregation (50). Both
smoking and alcohol consumption were associated
with higher levels of phospholipid eicosatrienoic acid.
Smoking, alcohol, and platelet function
We found that smoking was associated with increased levels of dihomogammalinolenic acid (20:3)
and omega-9 eicosatrienoic acid (20:3) and with decreased levels of omega-3 docosahexaenoic acid (22:
6). Each of these smoking-associated differences may
increase platelet aggregation. Alcohol consumption
was associated with increased levels of myristic acid
(14:0), palmitic acid (16:0), and omega-9 eicosatrienoic acid (20:3) and with decreased levels of stearic
acid (18:0) and linoleic acid (18:2). These alcohol-
associated differences in fatty acid composition would
be expected to have contradictory effects on platelet
aggregation; lower levels of stearic acid should reduce
platelet aggregation, whereas lower levels of linoleic
acid and higher levels of myristic acid, palmitic acid,
and eicosatrienoic acid should increase platelet aggregation (35, 44, 51).
Smoking increases platelet reactivity, and alcohol
consumption decreases it (1, 52). Some of these effects
may be mediated by differences in serum fatty acid
composition. On the basis of our findings, this hypothesis seems more tenable for smoking than for alcohol
consumption.
Limitations
Our study had a number of limitations. MRFTT
enrolled middle-aged men who were at high risk for
coronary heart disease; thus, caution in generalizing
our results is warranted. Although we assessed the
stability of the stored frozen serum specimens and
found little oxidative damage, we cannot rule out the
possibility that oxidative changes might have contributed to the observed differences and that an analysis of
fresh serum samples would have demonstrated other
associations. Our study had limited statistical power,
and associations between smoking and other fatty acids might be evident in a larger study. A single 24hour diet recall may lack the precision necessary to
assess usual dietary intake accurately. Additionally,
because of the large number of comparisons performed, it is possible that our findings might be the
result of chance. Finally, because percentages of fatty
acids were used, predicted increases in the levels of
some cholesterol ester and phospholipid fatty acids
would necessarily have been accompanied by corresponding decreases in the levels of other cholesterol
ester and phospholipid fatty acids.
Conclusions
In previous work, using stepwise regression models
that included smoking and alcohol consumption, we
demonstrated that palmitic acid (16:0) is associated
with an increased risk of coronary heart disease and
that the omega-3 fatty acids are associated with a
decreased risk of coronary heart disease and stroke (5,
6). Our current findings, however, raise two methodological issues. First, many studies that have examined
the relation of fatty acids to coronary heart disease did
not adjust for the effects of smoking or alcohol consumption. Our findings—i.e., that smoking and alcohol consumption may have important effects on the
fatty acid composition of serum lipoproteins—suggest
that future studies of fatty acids and coronary heart
Am J Epidemiol
Vol. 144, No. 4, 1996
Smoking, Alcohol, and Fatty Acids
disease should control for these factors. Second, a
number of previous studies of the relation of smoking
and alcohol consumption to serum fatty acids did not
control for differences in dietary intake. Because the
diets of smokers and nonsmokers and drinkers and
nondrinkers differ (53-55), detected differences in
fatty acid composition in these studies may have been
related to differences in dietary intake.
To our knowledge, this is one of the few studies to
have comprehensively examined the relation of smoking and alcohol consumption to serum cholesterol
ester and phospholipid fatty acid concentrations while
adjusting for dietary fat, cholesterol, and energy intake. Because these associations were present after
adjustment, our results are consistent with the hypothesis that smoking and alcohol drinking may affect the
absorption, synthesis, or metabolism of fatty acids.
Understanding the relation between smoking, alcohol
consumption, and fatty acids is important, even for
nonsmokers and nondrinkers, if cardiovascular disease
risk is mediated by modifiable differences in serum
fatty acid composition.
ACKNOWLEDGMENTS
This study was supported by National Heart, Lung, and
Blood Institute grant HL 32338.
The authors gratefully acknowledge the assistance and
advice of the MRFTT Editorial Committee and the staff of
the MRFIT Coordinating Center at the University of Minnesota.
REFERENCES
1. Office of Smoking and Health, Center for Chronic Disease
Prevention and Health Promotion, Centers for Disease Control, US Public Health Service. The health benefits of smoking
cessation: a report of the Surgeon General. Rockville, MD: US
Public Health Service, 1990. (DHHS publication no. (CDC)
90-8416).
2. Langer RD, Criqui MH, Reed DM. Lipoproteins and blood
pressure as biological pathways for effect of moderate alcohol
consumption on coronary heart disease. Circulation 1992;85:
910-15.
3. Renaud S, de Lorgeril M. Wine, alcohol, platelets, and the
French paradox for coronary heart disease. Lancet 1992;339:
1523-6.
4. Hoak JC, Glinsmann WH, Judd JT, eds. Workshop on dietary
fatty acids and thrombosis: Airlie, Virginia, March 18-20,
1991. Am J Clin Nutr 1992;56(suppl):783S-826S.
5. Simon JA, Hodgkins ML, Browner WS, et al. Serum fatty
acids and the risk of coronary heart disease. Am J Epidemiol
1995;142:469-76.
6. Simon JA, Fong J, Bernert JT Jr, et al. Serum fatty acids and
the risk of stroke. Stroke 1995;26:778-82.
7. Willett W. Nutritional epidemiology. New York, NY: Oxford
University Press, 1990.
8. Ma J, Folsom AR, Shahar E, et al. Plasma fatty acid composition as an indicator of habitual dietary fat intake in middleAm J Epidemiol
Vol. 144, No. 4, 1996
333
aged adults. Am J Clin Nutr 1995;62:564-71.
9. Multiple Risk Factor Intervention Trial Research Group. Multiple Risk Factor Intervention Trial: risk factor changes and
mortality results. JAMA 1982;248:1465-77.
10. Sherwin R, Kaelber CT, Kezdi P, et al. The Multiple Risk
Factor Intervention Trial (MRFIT) • H. The development of the
protocol. Prev Med 1981; 10:402-25.
11. Tillotson JL, Gorder DD, Kassim N. Nutrition data collection
in the Multiple Risk Factor Intervention Trial (MRFTT): baseline nutrient intake of a randomized population. J Am Diet
Assoc 1981;78:235-4O.
12. Folch J, Lees M, Stanley GH. A simple method for the
isolation and purification of total lipids from animal tissues.
J Biol Chem 1957;226:497-509.
13. Bernert JT Jr, Akins JR, Miller DT. Direct determination of
linoleate/oleate ratio in serum cholesterol esters by liquid
chromatography. Clin Chem 1982;28:676-80.
14. Bernert JT Jr, Kilboume EM, Akins JR, et al. Compositional
analysis of oil samples implicated in the Spanish toxic oil
syndrome. J Food Sci 1987;52:1562-9.
15. Lee DM. Malondialdehyde formation in stored plasma. Biochem Biophys Res Commun 1980;95:1663-72.
16. Klein RA. The detection of oxidation in liposome preparations. Biochem Biophys Acta 1970,210:486-9.
17. Bieri JG, Tolliver TJ, Catignani GL. Simultaneous determination of a-tocopherol and retinol in plasma or red cells by
high pressure liquid chromatography. Am J Clin Nutr 1979;
32:2143-9.
18. Rothman KJ. No adjustments are needed for multiple comparisons. Epidemiology 1990;l:43-6.
19. SAS Institute, Inc. SAS/STAT user's guide, version 6.07. 4th
ed. Cary, NC: SAS Institute, Inc, 1989.
20. Dougherty RM, Galli C, Ferro-Luzzi A, et al. Lipid and
phospholipid fatty acid composition of plasma, red blood
cells, and platelets and how they are affected by dietary lipids:
a study of normal subjects from Italy, Finland, and the USA.
Am J Clin Nutr 1987;45:443-55.
21. Schrade W, Biegler R, Bohle E. Fatty-acid distribution in the
lipid fractions of healthy persons of different ages, patients
with atherosclerosis and patients with idiopathic hyperlipidaemia. J Atherosclerosis Res 1961; 1:47-61.
22. Simpson HC, Barker K, Carter RD, et al. Low dietary intake
of linoleic acid predisposes to myocardial infarction. Br Med J
1982;285:683-4.
23. Miettinen TA, Naukkarinen V, Huttunen JK, et al. Fatty-acid
composition of serum lipids predicts myocardial infarction. Br
Med J 1982;285:993-6.
24. De Backer G, De Craene I, Rosseneu M, et al. Relationship
between serum cholesteryl ester composition, dietary habits
and coronary risk factors in middle-aged men. Atherosclerosis
1989;78:237-43.
25. Cambien F, Warnet J-M, Vernier V, et al. An epidemiologic
appraisal of the associations between the fatty acids esterifying serum cholesterol and some cardiovascular risk factors in
middle-aged men. Am J Epidemiol 1988;127;75-86.
26. Schrade W, Boehle E, Biegler R. Humoral changes in
arteriosclerosis: investigations on lipids, fatty acids, ketone
bodies, pyruvic acid, lactic acid, and glucose in the blood.
Lancet 1960;2:1409-16.
27. Bucalossi A, Mori S. Fatty acid composition of adipose tissue
in ischemic heart disease and stroke. Gerontol Clin (Basel)
1972;14:339-45.
28. Kirkeby K, Ingvaldsen P, Bjerkedal I. Fatty acid composition
of serum lipids in men with myocardial infarction. Acta Med
Scand 1972;192:513-19.
29. Boberg M, Vessby B, Croon LB. Fatty acid composition of
platelets and of plasma lipid esters in relation to platelet
function in patients with ischaemic heart disease. Atherosclerosis 1985;58:49-63.
30. Ciavatti M, Michel G, Dechavanne M. Platelet phospholipid
in stroke. Clin Chim Acta 1978;84:347-51.
31. Kingsbury KJ, Brett C, Stovold R, et al. Abnormal fatty acid
334
32.
33.
34.
35.
36.
37.
38.
39.
40.
41.
42.
43.
Simon et al.
composition and human atherosclerosis. Postgrad Med J 1974;
50:425-40.
Warnet J-M, Cambien F, Vernier V, et al. Relation between
consumption of alcohol and fatty acids esterifying serum
cholesterol in healthy men. Br Med J 1985;29O:1859-61.
Leng GC, Smith FB, Fowkes FG, et al. Relationship between
plasma essential fatty acids and smoking, serum lipids, blood
pressure and haemostatic and rheological factors. Prostaglandins Leukot Essent Fatty Acids 1994;51:101-8.
Keys A, Menotti A, Karvonen MJ, et al. The diet and 15-year
death rate in the Seven Countries Study. Am J Epidemiol
1986;124:9O3-15.
Nord0y A, Goodnight SH. Dietary lipids and thrombosis:
relationships to atherosclerosis. Arteriosclerosis 1990;10:
149-63.
Wood DA, Riemersma RA, Butler S, et al. Linoleic and
eicosapentaenoic acids in adipose tissue and platelets and risk
of coronary heart disease. Lancet 1987;1:177—83.
Mamiemi J, Lehtonen A, Inberg M, et al. Fatty acid composition of serum lipids in patients with a coronary bypass
operation. J Intern Med 1989;225:343-7.
Blankenhom DH, Johnson RL, Mack WJ, et al. The influence
of diet on the appearance of new lesions in human coronary
arteries. JAMA 199O;263:1646-52.
Hodgson JM, Wahlqvist ML, Boxall JA, et al. Can linoleic
acid contribute to coronary artery disease? Am J Clin Nutr
1993:58:228-34.
Wood DA, Butler S, Riemersma RA, et al. Adipose tissue and
platelet fatty acids and coronary heart disease in Scottish men.
Lancet 1984;2:117—21.
Hagenfeldt L, Paasikivi J, SjOgren A. Plasma levels of free
polyunsaturated fatty acids in patients with ischemic heart
disease. Metabolism 1973;23:1349-56.
Salonen JT, Salonen R, Penttila L et al. Serum fatty acids,
apolipoproteins, selenium and vitamin antioxidants and the
risk of death from coronary artery disease. Am J Cardiol
1985;56:226-31.
Leng GC, Horrobin DF, Fowkes FG, et al. Plasma essential
44.
45.
46.
47.
48.
49.
50.
51.
52.
53.
54.
55.
fatty acids, cigarette smoking, and dietary antioxidants in
peripheral arterial disease: a population-based case-control
study. Arterioscler Thromb 1994;14:471-8.
Goodnight SH Jr, Harris WS, Connor WE, et al. Polyunsaturated fatty acids, hyperlipidemia, and thrombosis. Arteriosclerosis 1982;2:87-113.
Siess W, Siegel FL, Lapetina EG. Dihomogammalinolenic
acid, but not eicosapentaenoic acid, activates washed human
platelets. Biochim Biophys Acta 1984;801:265-76.
Santos MT, Valles J, Aznar J, et al. Effect of smoking on
plasma and platelet fatty acid composition in middle-aged
men. Atherosclerosis 1984;50:53-62.
Skuladottir G, Hardarson T, Sigfusson N, et al. Arachidonic
acid levels in serum phospholipids of patients with angina
pectoris or fatal myocardial infarction. Acta Med Scand 1985;
218:55-8.
Burke V, Croft KD, Puddey IB, et al. Effects of alcohol intake
on plasma fatty acids assessed independently of diet and
smoking habits. Clin Sci 1991;81:785—91.
Grundy SM, Denke MA. Dietary influences on serum lipids
and lipoproteins. J Lipid Res 1990;31:1149-72.
Nord0y A. Dietary fatty acids, platelets, endothelial cells and
coronary heart disease. Acta Med Scand Suppl 1985;701:
15-22.
Homstra G. Effect of dietary lipids on platelet function and
thrombosis. Ann Med 1989;21:53-7.
Renaud SC, Beswick AD, Fehily AM, et al. Alcohol and
platelet aggregation: The Caerphilly Prospective Heart Disease Study. Am J Clin Nutr 1992;55:1012-17.
Thomson M, Fulton M, Elton RA, et al. Alcohol consumption
and nutrient intake in middle-aged Scottish men. Am J Clin
Nutr 1988;47:139-45.
Nuttens MC, Romon M, Ruidavets JB, et al. Relationship
between smoking and diet: the MONICA-France project. J Intern Med 1992;231:349-56.
Thompson RL, Pyke S, Scott EA, et al. Cigarette smoking,
polyunsaturated fats, and coronary heart disease. Ann N Y
AcadSci 1993;686:130-8.
Am J Epidemiol
Vol. 144, No. 4, 1996