Randomized Clinical Trials on the Effects of

Dietary Fat and Carbohydrate on Plasma

Lipoproteins and Cardiovascular Disease
Frank M. Sacks, MD, Martijn Katan, PhD

I
Several dietary approaches have reduced cardiovas- n this article, we review the randomized clinical trials
cular events in randomized clinical trials. Replacing that determined the effects of diet on plasma lipopro-
saturated fat with polyunsaturated fat prevented cor- teins and cardiovascular disease (CVD). We discuss,
onary events in men, and a Mediterranean diet and
in order, the effect of amount and type of dietary fats and
fatty fish improved survival. None of these trials had
carbohydrate on plasma lipid risk factors, effects on CVD
much impact on total fat intake but rather increased
vegetable oils, n-3 fatty acids, or many other plant that are predicted by these effects on the lipid risk factors,
foods or nutrients that are linked to coronary preven- and the observed effects of dietary fats on CVD; these
tion. The reductions in cardiovascular disease (CVD) include polyunsaturated fats, low-fat diets, Mediterra-
caused by these dietary therapies compare favorably nean diet, vegetarian diet, fish and fish oil, and trans un-
with drug treatments for hyperlipidemia and hyper- saturated fat. We end with a summary of the evidence,
tension. Improvement in blood lipid risk factors is an and conclusions on optimal diet therapy.
important mechanism to explain the results of trials The randomized clinical trial has the potential to pro-
of unsaturated fats. When saturated or trans unsat-
vide a compelling rationale for accepting or rejecting a
urated fats are replaced with monounsaturated or
n-6 polyunsaturated fats from vegetable oils, primar-
treatment. It is no less important for evaluating dietary
ily low-density lipoprotein (LDL) cholesterol de- therapies than drug treatments for which it is the only
creases. The LDL to high-density lipoprotein (HDL) acceptable study design. By randomizing the patients to a
cholesterol ratio decreases. When carbohydrates are dietary treatment or alternative, self-selection does not
used to replace saturated fats, in a low-fat diet, LDL influence the test of efficacy. In a nonrandomized study,
and HDL decrease similarly, and the ratio is not im- in which patients choose a dietary pattern, important
proved; triglycerides increase as well when carbohy- nondietary factors may be associated with the diet that is
drate increases, except when low glycemic index chosen. If the randomized trial is successfully conducted
foods are used. The n-3 polyunsaturated fats in fish
with few dropouts and few patients lost to end point as-
oils suppress cardiac arrhythmias and reduce tri-
glycerides, but they have little effect on LDL or HDL
certainment, the results should fully and solely reflect the
cholesterol levels. The theme should be that diet has true effect of the diet. One drawback of clinical trials of
benefits that come directly from foods, as well as diet in CVD is their high cost and impracticality for test-
from the reduction in saturated fats, cholesterol, ing many nutrients and foods. Also, clinical trials may not
meats, and fatty dairy foods. It is likely that many last long enough to detect such effects as those on cancer
diets could be designed that could prevent CVD. This that take more than a few years to make themselves seen.
potential diversity is crucial for engaging the diverse An alternative to using CVD as an outcome in a dietary
cultures and tastes of people worldwide in cardio- trial is to select surrogate end points that are in the causal
vascular disease prevention. Am J Med. 2002;113(9B):
path between a food or nutrient and CVD. Trials that use
13S–24S. © 2002 by Excerpta Medica, Inc.
surrogate end points are valuable, because cost, duration
of study, and, at times, ethical issues are lessened. Plasma
lipoproteins are important surrogate end points because
of their strong link to the pathogenesis of atherosclerosis
and their strong predictive association with coronary
heart disease (CHD). In this article, we review the estab-
lished effects of dietary fats on plasma lipoprotein con-
From the Nutrition Department, Harvard School of Public Health, Bos- centrations. We also comprehensively review the effects
ton, Massachusetts, USA (FMS); and Wageningen Centre for Food of dietary treatments on coronary death, myocardial in-
Sciences, and Division of Human Nutrition and Epidemiology,
Wageningen Agricultural University, Wageningen, the Netherlands farction, stroke, and all-cause mortality in randomized
(MK). clinical trials. We did not include trials that combined
Requests for reprints should be addressed to Frank M. Sacks, MD, dietary therapy with other treatments or that primarily
Nutrition Department, Harvard School of Public Health, 665 Hunt-
ington Avenue, Boston, Massachusetts 02115. E-mail: fsacks@hsph. used coronary arteriography rather than coronary death
harvard.edu. or myocardial infarction as study end points. Elsewhere

© 2002 by Excerpta Medica, Inc. 0002-9343/02/$22.00 13S

All rights reserved. PII S0002-9343(02)00987-1
 A Symposium: Effects of Dietary Fat and Carbohydrate on Plasma Lipoproteins and CVD/Sacks and Katan

in this issue, Ascherio reviews the recent epidemiologic saturated fat, LDL decreases proportionately more than
evidence on dietary fats and CHD.1 The evidence from HDL, and the LDL–HDL ratio decreases.13
clinical trials and epidemiology are largely concordant; Plasma triglyceride concentration has now been shown
together they demonstate a strong relation between diet to be an independent risk factor for coronary disease,
and CHD, and they provide a compelling rationale for whether measured in the fasting12 or nonfasting states.14
dietary treatment. Low-fat, high-carbohydrate diets increase fasting triglyc-
eride levels when carbohydrate replaces saturated or un-
DIETARY FAT AND PLASMA LIPID RISK saturated fat even when the unsaturated fat contains none
FACTORS of the fish fatty acids that are so effective in lowering
triglycerides.13 There is a controversy regarding whether
The causal relationship between total and low-density li- high-fat compared with low-fat diets increase postpran-
poprotein (LDL) cholesterol levels and CHD has been dial triglycerides.15,16 Acute high-fat intakes may indeed
supported strongly by a consistent body of evidence from increase postprandial lipid levels, but chronic substitu-
clinical trials2–5 and epidemiological studies.6 –9 National tion of carbohydrate for fat does not appear to decrease
health organizations advocate dietary changes that de- postprandial and 24-hour triglyceride levels, probably
crease intake of saturated and trans-unsaturated fat and because the high-carbohydrate intake increases endoge-
cholesterol to prevent CHD.10 The rationale is to reduce nous triglyceride synthesis and very low-density lipopro-
LDL concentration. However, diets affect not only LDL tein secretion. However, this issue is still open, because
but also high-density lipoprotein (HDL), triglycerides, measuring postprandial lipoproteins is laborious, and
which are independent lipid risk factors.11,12 Thus, as- there is no clear agreement as to which are the most
sessment of the effects of diet on CHD should include atherogenic postprandial lipoprotein particles.
consideration of the concomitant changes in HDL and Dietary patterns illustrate the contrasting effects of car-
triglycerides. This leads to a controversy about whether bohydrate compared with fat on plasma lipid risk factors.
carbohydrate or unsaturated oils should replace the en- The predicted effects on plasma lipids13 of 3 therapeutic
ergy from saturated and trans unsaturated fat. dietary approaches are shown in Figure 1.13,17 They are
Long-chain saturated fatty acids, commonly found in compared with a standard Western diet, similar to diets
the diet, increase LDL-cholesterol concentrations.13 in North American and Northern European countries:
Palmitic acid (C16:0), which is found in meats, dairy fat, total fat 38%, saturated fat 17%, monounsaturated fat
and palm oil, is the most common saturated fatty acid. 14%, polyunsaturated fat 7%, carbohydrates 42%, and
Palm oil is a major fat source in Latin America, Asia, and dietary cholesterol 400 mg/day. The plasma lipid levels
Europe, although not in the United States. Western diets are illustrative of patients who have mild hyperlipidemia,
with dairy fat and meats also contain myristic acid, for whom diet is recommended as first-line treatment.
whereas other tropical oils, such as coconut or palm ker- The standard Step 1 lower fat approach reduces total fat
nel oils, have mainly lauric acid. Myristic acid has the intake to 30%, saturated fat to 10%, and dietary choles-
strongest effect on LDL followed by lauric and palmitic terol to 300 mg, replacing the saturated fat mainly with
acids. If saturated fat (e.g., 25 g or 10% of total daily carbohydrate. Monounsaturated and polyunsaturated fat
energy intake [such as from about 100 to 150 g beef plus content is similar to the initial diet. The predicted effects
100 to 150 mL ice cream]) is replaced by carbohydrate, on plasma lipids were computed by the Mensink and Ka-
monounsaturated fat, or polyunsaturated fat, LDL will tan equations,13 derived from a meta-analysis of 27 well-
decrease by an average of 13 mg/dL, 15 mg/dL, or 18 controlled dietary trials lasting 3 to 8 weeks, under strict
mg/dL, respectively.13 Thus, monounsaturated fats re- conditions. The quality and number of the individual
duce LDL slightly more than does carbohydrate, and studies makes this analysis definitive. Total cholesterol
polyunsaturated fatty acids have still more of this effect. decreases by 5% on the 30% fat diet, and the LDL choles-
Considering just the changes in LDL, any of these dietary terol decreases by 6%. However, these modest improve-
changes would be considered beneficial. Fish contain ments contrast with a decrease in HDL of 9% and an
small amounts of a special class of polyunsaturated fatty increase in triglycerides of 9%. A more stringent low-fat
acids, the very long-chain n-3 fatty acids. These have little approach is shown next, in which total fat is reduced to
effect on LDL cholesterol, but they lower both fasting and 20% and saturated fat to 7%. Monunsaturated and poly-
postprandial triglycerides. unsaturated fats are reduced by necessity to 10% and 5%,
The effects of dietary fat and carbohydrate on HDL are respectively, to achieve 20% total fat, and carbohydrate is
quite different from the effects on LDL. Low-fat, high- increased to 65%. Dietary cholesterol is reduced to 100
carbohydrate diets are well known to reduce HDL. The mg/day commensurate with the reduction in meats and
ratio of total or LDL cholesterol to HDL cholesterol does eggs. The predicted plasma lipid changes that occurred
not change on a low-fat diet, because HDL is reduced as with the 30% fat diet were intensifed; total cholesterol
much as LDL. However, when unsaturated fat replaces decreases by 9%, LDL decreases by 12%, HDL decreases

14S December 30, 2002 THE AMERICAN JOURNAL OF MEDICINE威 Volume 113 (9B)
 A Symposium: Effects of Dietary Fat and Carbohydrate on Plasma Lipoproteins and CVD/Sacks and Katan

Figure 1. Predicted changes in plasma cholesterol and triglyceride concentrations caused by 3 types of diet treatment: 30% fat (step
1), 20% low-fat, and Mediterranean. The standard Western and Mediterranean diets have 38% fat. The dietary changes are described
more fully in the text. The meta-analysis of Mensink and Katan13 was used for changes in dietary fatty acids and that of Clarke et al17
for changes in dietary cholesterol. Note that the effects of carbohydrate on triglycerides are less if the carbohydrate comes from
low-glycemic-index foods, as described in the text. HDL ⫽ high-density lipoprotein; LDL ⫽ low-density lipoprotein.

by 20%, and triglyceride level increases by 20%. Thus, as TYPE OF CARBOHYDRATE

total fat is reduced, the adverse changes in HDL and trig- The usual mix of carbohydrates in the Western diet con-
lycerides become more prominent. Lastly, a Mediterra- tains much refined polysaccharides, such as in bread and
nean dietary approach is shown that differs from the pre- baked goods, and sugars in juices and soda. These have a
vious 2 diets. Saturated fat is reduced, but it is replaced high glycemic index, causing glucose and insulin to in-
with vegetable oils, such as olive, canola, corn, safflower, crease substantially. Other types of carbohydrates that are
or sunflower oil, or oils from nuts, which contain mainly in certain whole grains, beans, nuts, and vegetables have a
monounsaturated and polyunsaturated fatty acids. Total lower glycemic index, probably because the digestion and
fat content remains at 38%, similar to what is eaten tra- absorption of the glucose is slow. These cause less of an
ditionally in Greece.18 Dietary cholesterol is reduced to increase in serum glucose and insulin. Low glycemic in-
100 mg/day, concordant with this plant-based dietary dex is only partly explained by the fiber content of foods:
pattern. The high unsaturated-fat diet decreases total thus, white bread and whole wheat bread have the same
cholesterol by 11% and LDL by 13%. HDL decreases by glycemic index even though whole wheat bread is much
7%, and triglycerides do not change. In summary, these richer in fiber. An important determinant of glycemic
diets all lower LDL, but the low-fat diets decrease HDL index is the glucose content of the carbohydrates in the
proportionately more than they decrease LDL, and they food, because only glucose increases serum glucose. As a
increase triglycerides. The total cholesterol to HDL cho- result, ordinary sugar, which is a disaccharide of fructose
lesterol ratio increases on the 30% or 20% fat diets but and glucose, has a relatively low glycemic index, as has
decreases only on the Mediterranean diet. These findings milk sugar, which consists of galactose and glucose.20
raise questions about the eventual effects of low-fat diets Lastly, the processing and intrinsic digestability of the
on CHD.19 food has an influence. Some foods with a low glycemic

December 30, 2002 THE AMERICAN JOURNAL OF MEDICINE威 Volume 113 (9B) 15S
 A Symposium: Effects of Dietary Fat and Carbohydrate on Plasma Lipoproteins and CVD/Sacks and Katan

index cause less of an increase in plasma triglycerides than Table 1. Predicted Changes in Coronary Heart Disease (CHD)*
do the more commonly eaten higher glycemic index car- High
bohydrates. HDL decreases when dietary fat is decreased, 30% Fat Unsaturated Fat
whatever the type of carbohydrate. A low-fat diet that is Blood Lipid (Step 1) 20% Fat (Mediterranean)
truly based on high-fiber, low– glycemic index foods †
LDL plasma level ⫺11 mg/dL ⫺23 mg/dL ⫺25 mg/dL
rather than the high– glycemic index, low-fat foods that CHD incidence ⫺11% ⫺23% ⫺25%
are a major part of contemporary low-fat diets may avoid HDL‡ plasma level ⫺4 mg/dL ⫺9 mg/dL ⫺3 mg/dL
the undesirable effects on plasma triglycerides, serum CHD incidence
glucose, and insulin. For example, the Dietary Ap- Men ⫹8% ⫹18% ⫹6%
proaches to Stop Hypertension (DASH) diet, a 27%-fat Women ⫹12% ⫹27% ⫹9%
diet that emphasizes fruit, vegetables, and low-fat dairy TG§ plasma level ⫹18 mg/dL ⫹40 mg/dL ⫹1 mg/dL
products and which includes whole grains, poultry, fish, CHD incidence
and nuts and contains only small amounts of sweets and Men ⫹3% ⫹6% 0%
sugar-containing beverages, did not increase fasting tri- Women ⫹7% ⫹17% 0%
glycerides compared with a typical US diet with 37% fat. HDL ⫽ high-density lipoprotein; LDL ⫽ low-density lipoprotein;
21
The type of carbohydrate and its relevance to plasma TG ⫽ triglycerides.
lipid control, treatment of diabetes, and prevention of * See Figure 1 for initial diet and plasma lipid concentrations.
†
⌬LDL ⫺ 1 mg/dL 3 ⌬CAD ⫺ 1%. (Data from N Engl J Med.30)
CAD is discussed elsewhere in this issue by Jenkins et al.22 ‡
Framingham, Lipid Research Clinics Coronary Primary Prevention
Trial (LRC-CPPT), Multiple Risk Factor Intervention Trial (MRFIT),
PREDICTED EFFECTS OF DIETARY British Regional Heart Study combined; adjusted for other lipids.
⌬HDL ⫺ 1 mg/dL3 ⌬CAD ⫹2% in men, ⫹3% in women. (Data from
THERAPIES ON CORONARY DISEASE Circulation.11)
Unfortunately, there are no satisfactory clinical trials that §
Adjusted for other lipids, including HDL. ⌬TG ⫹88 mg/dL (1
evaluate the effect of a 30% fat (Step 1) or a 20% low-fat mmol/L) 3 ⌬CAD ⫹14% in men, ⫹37% in women. Note that the
effects of carbohydrate on TG are less if the carbohydrate comes from
diet on clinical CHD. The trials were either too small in low glycemic index foods, as described in the text. (Data from J Cardio-
population or too short in duration to properly deter- vasc Risk.12)
mine the effects.23,24 Other trials included a low-fat diet
as part of a multifactorial prevention program,25–28 so
that the dietary effect cannot be independently deter- unsaturated fat diet based on blood lipid effects is consis-
mined, or used coronary angiography27–29 rather than tent with prospective epidemiological studies that di-
myocardial infarction or coronary death to assess the ef- rectly studied the relation between dietary fats and
fect of the diet. Because there are no definitive clinical carbohydrate on coronary rates (reviewed in this supple-
trials of a low-fat, high-carbohydrate diet on coronary ment by Ascherio1). The predicted null effect in men
events, it is of interest to calculate the predicted effects of raises questions about futility of the low-fat intervention,
low-fat diets on coronary events from the known effects whereas the predicted adverse effect in women raises fur-
on plasma lipid risk factors. To do this, the epidemiologic ther questions about safety.
relation between change in LDL, HDL, and triglyceride
and change in coronary disease risk is used.11,12,30 The CONTROLLED TRIALS OF DIETARY
overall predicted effects on CHD differ between men and THERAPY AND CORONARY EVENTS
women, because HDL and triglycerides are stronger risk High-Polyunsaturated Fat Diets
factors in women (Table 1 and Figure 2). In men, the In the period 1956 to 1957, 4 research groups reported
20% or 30% fat diet produces no change in coronary that polyunsaturated vegetable oils substantially reduced
disease incidence, because the beneficial decreases in LDL blood cholesterol concentrations when they replaced sat-
are completely counterbalanced by adverse decreases in urated fats,31–34 and predictive equations were designed
HDL and increases in triglycerides (Figure 2 and Table 1). by Keys et al.32 The polyunsaturated fatty acids studied
In women, the 20% or 30% fat diets are predicted to were mainly the n-6 class and did not include the n-3
increase coronary incidence (Figure 2). In contrast, the polyunsaturated fatty acids from fish. Many subsequent
high-unsaturated-fat diet is predicted to reduce coronary studies confirmed that polyunsaturated fatty acids are the
incidence by 19% in men and 16% in women (Figure 2). most potent fatty acids for reducing LDL cholesterol.13
It is important to recognize that these predictions of Corn, safflower, sunflower, and soybean oils were pre-
the effect of diet on coronary disease incidence have the scribed to hypercholesterolemic patients in the 1960s and
inherent limitations of epidemiology. However, it is just 1970s. The treatment was relatively simple for dietary
as important to recognize that neither the 30% nor the therapy, because patients were often instructed to drink
20% low-fat diets have been tested in a randomized clin- the vegetable oils, as well as to use them in cooking or
ical end point trial. The calculated advantage of the high- salads. Four randomized clinical trials tested the effect

16S December 30, 2002 THE AMERICAN JOURNAL OF MEDICINE威 Volume 113 (9B)
 A Symposium: Effects of Dietary Fat and Carbohydrate on Plasma Lipoproteins and CVD/Sacks and Katan

Figure 2. Predicted effects on coronary artery disease of a 30% fat (Step 1) diet, a 20% low-fat diet, and a Mediterranean diet.
Epidemiologic studies were used to predict the change in coronary heart disease from the changes in low-density lipoprotein
cholesterol, high-density lipoprotein cholesterol, and triglycerides caused by the diets (see also Figure 1).11,12,30 Diets are described
in the text. The changes in plasma lipids are shown in Figure 1. Solid bars ⫽ males (M); open bars ⫽ females (F).

Table 2. Clinical Trials of Diet Therapy to Reduce Coronary Events: Substitution of Polyunsat-
urated Fat for Saturated Fat
Dietary Duration ⌬ Cholesterol ⌬ CVD
N Fat (%) (yr) (%)* (%)†
Finnish Mental Hospital37 676 34 6 ⫺15‡ ⫺43‡
Oslo35 412 39 5 ⫺14‡ ⫺25‡
MRC soy oil36 393 46 4 ⫺15‡ ⫺12%
Los Angeles34 846 40 8 ⫺13‡ ⫺34‡
CVD ⫽ cardiovascular disease; MRC ⫽ Medical Research Council.
Trials with at least 2 years of average follow-up were included. See text for more information on the indi-
vidual trials.
* Percentage change in serum cholesterol in the treatment group compared with the change in the control
group.
†
Percentage difference in coronary event rates in the treatment compared with the control group. CVD is
defined as myocardial infarction or sudden death for the Finnish, Oslo, and MRC trials, and myocardial
infarction, sudden death, or stroke for the Los Angeles trial.
‡
P ⱕ0.05.

of soy and other unhydrogenated vegetable oils on was monitored and confirmed objectively by fatty acid
CHD.35–38 Both linoleic and ␣-linolenic acids were in- measurements in the patients’ adipose tissue. A descrip-
creased in these trials. CVD events were significantly re- tion of the patients, the experimental and control diets,
duced in 3 of them35,36,38 (Table 2). Overall, these trials and serum cholesterol are shown in Table 3. Corn, soy-
support the relationship found in trials of drug therapy bean, safflower, and cottonseed oils were used in the ex-
that coronary events are reduced by 2% for every 1% perimental diet, rather than the animal fat used in the
reduction in total cholesterol.2 control diet. Linoleic acid, an n-6 polyunsaturated fatty
The trial at the Wadsworth Hospital and Veterans Ad- acid, was the major fatty acid in the experimental diet.
ministration Center in Los Angeles35 was the most rigor- The patients were 65 years old on average; 74% were cig-
ous methodologically, had the most patients (846), and arette smokers, 7% had a previous myocardial infarction,
lasted the longest (8 years). The patients lived and re- 12% had angina pectoris, and 7% had a stroke. The ex-
ceived their meals at the domiciliary center, each diet perimental diet reduced total cardiovascular end points
group in a separate dining room. Moreover, the investi- significantly by 31% from 22.7% to 15.6% (P ⫽ 0.01),
gators established double-blind conditions. Adherence and the combined end point of myocardial infarction,

December 30, 2002 THE AMERICAN JOURNAL OF MEDICINE威 Volume 113 (9B) 17S
 A Symposium: Effects of Dietary Fat and Carbohydrate on Plasma Lipoproteins and CVD/Sacks and Katan

Table 3. Clinical Trial of a High-Polyunsaturated-Fat Diet study demonstrated that a varied and palatable diet could
Experimental Control be developed in line with these specifications and that it
was acceptable to most potential subjects, whereas low-
N 424 422 fat diets were rejected forthwith. The diet chosen for
Baseline characteristics study offered the incidental advantage that if it were ul-
Age 65 years 66 years timately found to be of value, it might be adopted by the
Previous myocardial infarction 7% 7% general public more readily than would a diet requiring
Angina 12% 12% gastronomic sacrifice [pp 3– 4].35
Stroke 7% 6%
Smokers 74% 76% The Oslo Diet-Heart Study36 was a trial that replaced
Serum cholesterol 233 mg/dL 234 mg/dL saturated with polyunsaturated fat, mainly soybean oil, in
Diet analysis during the trial 412 men who had a myocardial infarction. It was not a
Total fat 39% 40% domiciliary trial, and the men were taught to select rather
Polyunsaturated fat 17% 5.5%
than being given their meals. After 5 years, the experi-
Linoleic acid 15% 4%
Linolenic acid 2% 0.5% mental group had fewer recurrent myocardial infarctions
Monounsaturated fat 16% 15% or sudden deaths than the control group (61 [30%] vs. 81
Saturated fat 8% 19% [39%]; P ⫽ 0.05) and fewer cardiovascular deaths (38
Cholesterol 365 mg 653 mg [18%] vs. 52 [25%]; P ⫽ 0.09). The experimental group
Serum cholesterol change ⫺16% ⫺3% continued to experience these trends toward reduced car-
From Circulation.35
diovascular mortality for an additional 6 years after the
trial ended.33 Unlike the Los Angeles dietary trial, non-
cardiovascular deaths were equal in the 2 groups, and
sudden death, or stroke was reduced by 34% (P ⫽ 0.04) total deaths were fewer in the experimental than control
(Table 4).35 Stroke was reduced by 41% (P ⫽ 0.055). group (41 vs. 55; P ⫽ 0.13).
However clear-cut these results appear, acceptance of the The British Medical Research Council conducted a
findings was hampered by the lack of reduction in total trial of soybean oil in 393 men after myocardial infarc-
mortality. The significant reduction in deaths from car- tion, instructing them to drink 43 g/day of soybean oil
diovascular causes, 48 deaths in the experimental diet with fruit juice and to use another 43 g/day in cooking.36
group versus 70 in the control group, was partly mitigated After 4 years, reduction in recurrent myocardial infarc-
by a higher number of deaths from noncardiovascular tion or sudden death was not significant, 45 events in the
causes in the experimental compared with the control
experimental group versus 51 in the control group.
group, notably 7 versus 2 cancers and 4 versus 0 traumas,
Finally, the Finnish Mental Hospital Study carried out
as well as from an uncertain cause in 32 versus 25 pa-
a crossover study in 2 hospitals, in which 1 of the hospitals
tients.35 Such imbalances in small numbers of noncardio-
first fed to the residential patients a control diet and then
vascular causes of deaths have occurred in many much
the experimental diet, and the other hospital fed the diets
larger studies, and it is recognized better now than before
in the reverse order.38 The experimental diet replaced sat-
that these are usually the result of chance.
urated with polyunsaturated fat, mainly soybean oil. Each
A fascinating sidelight of the Los Angeles diet trial was
diet period lasted 6 years, and, for the most part, the same
that a pilot study was conducted to determine whether
patients ate both diets. Coronary end points, CHD death
reduction of saturated fat intake was more acceptable to
or myocardial infarction, were significantly lower during
the patients if accomplished by increasing carbohydrate
or unsaturated oils. The results of the pilot study offer a the soybean oil periods than during the control period
relevant commentary on the current dietary controver- (Table 2 and Figure 3). The coronary event rates at each
sies, ⬎30 years later: hospital showed the effects of the crossover, decreasing in
1 and increasing in the other after the change in diets
The circumstances of this trial—in particular the fact (Figure 3). In summary, replacement of saturated with
that most of the subjects were not highly motivated to- polyunsaturated fat in the diet reduced coronary event
ward self-deprivation—rendered it mandatory that both rates in all 4 trials and significantly in 3 of them.
diets be highly palatable by conventional criteria. It ap- Polyunsaturated fat is prone to oxidation in vitro. If
peared on culinary grounds that a staff skilled in dietetics this also occurs in vivo, it might increase susceptibility of
. . . could achieve a diet closely simulating the conven-
LDL particles to oxidative modification, which would
tional institutional diet if, and only if, the total amount of
dietary fat remained nearly unchanged. This restriction
make them more atherogenic. However, the positive out-
led to a decision to test a replacement of the saturated comes of randomized controlled trials suggest that oxi-
animal fats and hydrogenated shortenings of the conven- dizability of polyunsaturated fatty acids may be less of a
tional diet by equal quantities of unsaturated fat in the factor in the genesis of atherosclerosis and CAD in hu-
form of vegetable oils in the experimental diet. . . . A pilot mans than previously thought.

18S December 30, 2002 THE AMERICAN JOURNAL OF MEDICINE威 Volume 113 (9B)
 A Symposium: Effects of Dietary Fat and Carbohydrate on Plasma Lipoproteins and CVD/Sacks and Katan

Table 4. Clinical Trial of a High-Polyunsaturated-Fat Diet: Clinical Cardiovascular End Points

Experimental Control Risk Reduction
(N ⫽ 424) (N ⫽ 422) (%) P Value
MI, definite 27 (6.4%)* 40 (9.5%) 32 NS
Sudden death 18 (4.2%) 27 (6.4%) 33 NS
Stroke 13 (3.1%) 22 (5.2%) 41 0.055
Fatal cardiovascular 48 (11.3%) 70 (16.6%) 31 ⬍0.05
Nonfatal cardiovascular 37 (8.7%) 49 (11.6%) 24 NS
Total cardiovascular† 66 (15.6%) 96 (22.7%) 31 0.01
MI, sudden death, stroke 60 (14.2%) 87 (20.6%) 34 0.04
MI ⫽ myocardial infarction; NS ⫽ not significant.
* Number (percent) patients with event.
†
Myocardial infarction, sudden death, stroke, cerebral hemorrhage, limb amputation.
From Circulation.35

Figure 3. Effects of a high-polyunsaturated-fat diet on coronary heart disease (CHD): the Finnish Mental Hospital Study.38 A
crossover study was conducted in 2 hospitals, the patients in 1 (hospital K) receiving the control diet first and then the experimental
diet, and the patients in the other (hospital N) receiving the experimental diet first and then the control diet. In each hospital,
coronary death and myocardial infarction (MI) rates were lower during the polyunsaturated fat diet.

rum cholesterol, it is likely that dietary adherence was low

Clinical Coronary End Point Trials with a
in both studies.
Low-Fat Diet
Only 2 clinical coronary end point trials tested the effect Mediterranean Diet (Lyon Heart Study)
of a low-fat diet without involving other treatments, such The effect of a Mediterranean-style diet on coronary
as blood pressure– or plasma lipid–lowering medica- death after myocardial infarction was studied in the Lyon
tion.23,24 Both trials enrolled patients with myocardial Diet Heart Study.39 – 41 The dietary instruction started
infarction soon after they left the hospital. Reduction in relatively early, 2 weeks, after the acute myocardial infarc-
saturated fat was intended to reduce total fat intake, and tion. The control group received Step 1 dietary advice.
carbohydrate-rich foods were recommended. Neither The Mediterranean diet replaced animal fat with polyun-
trial showed a significant benefit (Table 5). However, saturated vegetable oil rich in ␣-linolenic acid, an n-3
both are unsatisfactory tests of the low-fat hypothesis, fatty acid that is the mother compound of the n-3 poly-
because the diets lowered serum cholesterol only mini- unsaturated fatty acids from fish but that does not share
mally, the durations of 2 years in the 1989 study24 or 3 their effect on plasma triglycerides. Total fat was similar
years in the 1965 study23 years were insufficient for lipid in both groups at 31% of energy. Meat, butter, and cream
treatments to produce a reduction in coronary events, or were reduced, and legumes, bread, fruits, and vegetables
they had a very small sample size23 (Table 5). Because it is were increased (Table 6). Dietary cholesterol intake de-
well established that reduction of saturated fat lowers se- creased. Dietary adherence was proved by higher serum

December 30, 2002 THE AMERICAN JOURNAL OF MEDICINE威 Volume 113 (9B) 19S
 A Symposium: Effects of Dietary Fat and Carbohydrate on Plasma Lipoproteins and CVD/Sacks and Katan

Table 5. Clinical Trials of Diet to Reduce Coronary Events: Low-Fat Diets

Effects of the Low-Fat
Diet
Dietary Fat Duration
N (% energy) (yr) Cholesterol CHD
London 23
264 22% vs. 41% 3 ⫺5% ⫺5%
DART24 2,033 32% vs. 35% 2 ⫺3.5% ⫺9%
CHD ⫽ heart disease; DART ⫽ Diet and Reinfarction Trial.
Trials with ⱖ2 years of average follow-up were included.

Table 6. Mediterranean Diet to Prevent Death After Myocardial Infarction: Changes in Foods,
Nutrients, and Coronary Risk Factors
Increased Decreased Unchanged
Dietary changes
Fruits Meats Total fat (31%–33%)
Vegetables Butter, cream Cheese, poultry, fish
Beans Cholesterol
Bread
Vegetable oil
Fiber
Body weight Unchanged
Blood pressure Unchanged
Dietary fatty acid changes
Oleic acid Linoleic acid
␣-Linolenic acid Saturated fat
EPA, DHA
Blood levels
Ascorbic acid Linoleic acid Cholesterol, LDL, HDL
␣-Tocopherol Trans fatty acids
Oleic acid Platelet aggregation
␣-Linolenic acid
EPA, DHA
DHA ⫽ docosahexaenoic acid; EPA ⫽ eicosapentaenoic acid; HDL ⫽ high-density lipoprotein; LDL ⫽ low-
density lipoprotein.
Adapted from Lancet.39

levels of oleic acid, ␣-linolenic acid, eicosapentaenoic the diet improved the lipid profile as expected. The results
acid, and antioxidant vitamins. Particularly in view of support several other hypotheses on diet and heart dis-
these changes in dietary fat intake, it was surprising that ease, including benefits of ␻-3 fatty acids in general (␣-
there were no changes in plasma lipid levels. The trial was linolenic acid in particular), monounsaturated fats, anti-
stopped after a mean duration of 27 months, with highly oxidant vitamins, and fiber.
significant reductions in coronary death in the Mediter-
ranean diet group.39 Adherence to the assigned diets con- Indian Vegetarian Diet
tinued for another 19 months after the trial ended, for a The Indian Heart Study42 tested the effect of fish and
total of 46 months of treatment. All-cause mortality was plant foods, including fruit, vegetables, legumes, and nuts
reduced from 12% to 6%, coronary death from 9% to 3%, in patients with acute myocardial infarction. The diets of
and nonfatal myocardial infarction from 12% to 4%, all the experimental and control groups both were reduced
statistically significant39 (Figure 4). Cancer incidence was in meats, eggs, butter, hydrogenated oils and increased in
also reduced from 6% to 2%41 (Figure 4). Thus, a Medi- vegetable oils. Total fat intake was 24% to 26% of energy
terranean dietary approach that did not reduce total fat in both groups. This trial is unique in that it began the
but improved many other aspects of the diet (Table 6) dietary intervention while the patients were in the hospi-
prevented death, recurrent coronary events, and cancer tal only 2 to 3 days after the acute myocardial infarction
in patients after myocardial infarction (Figure 4). Per- and that a significant reduction in coronary events of 36%
haps the clinical benefits would have been even greater if occurred in the experimental group after only 12 weeks of

20S December 30, 2002 THE AMERICAN JOURNAL OF MEDICINE威 Volume 113 (9B)
 A Symposium: Effects of Dietary Fat and Carbohydrate on Plasma Lipoproteins and CVD/Sacks and Katan

Figure 4. The effect of a Mediterranean diet on total mortality, coronary disease events, and cancer in men after acute myocardial
infarction (MI): the Lyon Heart Study.39 – 41 See text for description and Table 6 for changes in diet and coronary risk factors.
Numbers in bars are the number of events. Percentages over the open bars are percentage risk reduction with P value.

treatment. Perhaps the early timing of the intervention (⫺33%). Nonfatal myocardial infarction was not affected
after myocardial infarction caused the rapid reduction in significantly, 3.2% in the control group versus 4.8% in
coronary events. After 1 year of treatment, significant de- the fish group. This trial tested 2 additional dietary inter-
creases occurred in all-cause mortality by 45%, in coro- ventions, reduced fat and increased fiber, and neither
nary deaths by 42%, and in nonfatal myocardial infarc- showed any tendency to produce benefit (Table 5). The
tion by 38%. The vegetarian diet also reduced ventricular DART study24 and the Lyon Heart Study41 both suggest
ectopy. Fruit and vegetable intake was 3 times higher in that n-3 fatty acids protect against death after myocardial
the experimental than in the control group. The experi- infarction. These results have recently been much rein-
mental group had higher intake of polyunsaturated fats, forced by the Gruppo Italiano per lo Studio della Soprav-
fiber, vitamins C and E, carotene, and potassium and had vivenza nell’Infarto Miocardico (GISSI) trial, a random-
lower intake of saturated fats and cholesterol; carbohy- ized controlled trial in 11,324 Italian patients surviving a
drate intake shifted from refined to complex sources. In recent myocardial infarction.43 Treatment with 1 g/day of
the experimental compared with the control group, body n-3 polyunsaturated fatty acids from fish lowered the
weight, LDL cholesterol, triglycerides, blood pressure, overall risk of death and of coronary death from 6.8% to
and serum glucose significantly decreased, and HDL cho- 4.8% (⫺2%) (Figure 5). In both trials24,43 the death rate
lesterol increased. The patients in the experimental group began to lessen in the fish or fish oil group as early as 3
who lost ⬎0.5 kg after 1 year had greater reductions in months after treatment was started.
serum cholesterol, triglycerides, glucose, and blood pres-
sure and had fewer recurrent cardiovascular events than
those who lost ⬍0.5 kg. Thus, a vegetarian diet initiated
soon after acute myocardial infarction reduced deaths
and recurrent cardiovascular events.
Fish and Fish Oil
The Diet and Reinfarction Trial (DART) tested the effect
of increased intake of fatty fish for 2 years in 2,033 Welsh
men after acute myocardial infarction.24 The patients
were asked to eat mackerel, herring, kipper, sardine,
salmon, or trout twice weekly, beginning 6 weeks after
myocardial infarction. Those who could not eat fatty fish
took fish oil capsules (20% n-3 fatty acids), 1.5 g daily.
This small amount of fatty fish or fish oil significantly
reduced total mortality from 12.8% to 9.3% (⫺29%; Fig- Figure 5. Summaries of trials of fatty fish or fish oil to prevent
ure 5) and coronary mortality from 11.4% to 7.7% coronary artery disease. See text for additional results.

December 30, 2002 THE AMERICAN JOURNAL OF MEDICINE威 Volume 113 (9B) 21S
 A Symposium: Effects of Dietary Fat and Carbohydrate on Plasma Lipoproteins and CVD/Sacks and Katan

Figure 6. The effect of trans unsaturated fat compared with other unsaturated fats on the ratio of total cholesterol to high-density
lipoprotein (HDL) cholesterol. In each instance, saturated fat (10% of total daily calories) is replaced by the same amount of
unsaturated fat. Carb ⫽ carbohydrate; Mono ⫽ monounsaturated fat; Poly ⫽ polyunsaturated fat; Trans ⫽ trans unsaturated fat.51

Several actions of n-3 fatty acids that could explain butter. When you replace 10% of energy intake from sat-
their beneficial effect after myocardial infarction are re- urated fat with trans fatty acids, there is no change in
duction of plasma triglycerides,44 accelerated clearance of LDL, and there is about a 7 mg/dL (15%) decrease in
chylomicrons and very low-density lipoprotein,45 in- HDL. Thus, LDL stays high, and HDL decreases precipi-
creased HDL,46 and reduced blood pressure.47 The n-3 tously. This is worse than the effects of carbohydrate,
fatty acids also have an antiarrhythmic action,48 increas- which decreases both LDL and HDL. By combining the
ing the threshold for ventricular fibrillation.49 Dietary in- changes in total cholesterol and HDL cholesterol into a
take of ␻-3 fatty acids was very low in persons who expe- ratio, the trans fatty acids can be compared conveniently
rienced sudden death.50 This antiarrhythmic action of with the other dietary fatty acids. This analysis is based on
n-3 fatty acids may have been the dominant mechanism 8 carefully controlled trials. Trans fatty acids have the
for prevention of death after acute myocardial infarction worst effects on blood lipids among all dietary fatty acids,
in the Lyon Heart Study,40 the DART study,24 and the even saturates (Figure 6). Monounsaturated and polyun-
GISSI study.43 saturated fatty acids reduce the ratio, whereas carbohy-
drates have a negligible effect. These adverse effects of
Trans Unsaturated Fatty Acids, Serum Lipids, trans unsaturated fatty acids on cholesterol and HDL
and CHD agree with epidemiologic findings of increased risk of
Information on the effects of dietary trans fatty acids on CHD in people who consume these fatty acids, as de-
serum lipids and CHD has increased tremendously dur- scribed by Ascherio.1 This evidence of a harmful effect of
ing the past 10 years. Trans unsaturated fatty acids are trans fatty acids on coronary disease has been sufficient to
formed from the naturally occurring cis unsaturated fatty convince a large part of the fats and oils industry in Eu-
acids during the hydrogenation process that takes liquid rope to get rid of trans fatty acids in packaged foods and
vegetable oil, such as soybean, and makes solid or semi- most margarines. Most Dutch retail margarines are now
solid margarine or shortenings for frying or baking. Most free from trans fatty acids, and the same trend is taking
trans unsaturated fatty acids in the diet are the analogues place in a large part of the world. The United States is
of oleic acid, the main cis monounsaturated fatty acid in lagging behind in this respect, although McDonald’s and
the diet. Among all dietary fats, trans fatty acids have Frito-Lay are progressively reducing the trans fatty acid
unique effects on serum lipids. LDL increases and HDL content of their foods. The term “partially hydrogenated”
decreases when oleic acid is replaced by trans fatty acids, on the label indicates the presence of trans fatty acids.
both when these are isomers of the same chain length and Nonetheless, both in Europe and the United States, there
when they are longer-chain isomers formed by hydroge- are still large amounts of trans fatty acids in the frying fats
nation of fish oils.51–53 As soon as this information was used in restaurants and in baked goods, including cook-
known, people asked whether foods that contain trans ies, snacks, and pastries. This is now the major source of
unsaturated fatty acids, such as margarines, truly do have trans fatty acids, in Eastern Europe, Western Europe,
benefits over food that contain saturated fat, such as North America, and South America, so there is still quite

22S December 30, 2002 THE AMERICAN JOURNAL OF MEDICINE威 Volume 113 (9B)
 A Symposium: Effects of Dietary Fat and Carbohydrate on Plasma Lipoproteins and CVD/Sacks and Katan

a bit of work left to do to improve the food supply to ductions in rates of CVD caused by these dietary thera-
prevent CHD. pies compare favorably with drug treatments for coro-
nary patients. The theme should be that diet has benefits
SUMMARY AND CONCLUSIONS that come directly from foods, as well as from the reduc-
tion in saturated fats, cholesterol, meats, and fatty dairy
Dietary effects on plasma lipids are the following: foods. It is likely that many diets could be designed with
these principles that can prevent CVD. This potential di-
● Decreasing saturated fat lowers LDL cholesterol.
versity is crucial for engaging the diverse cultures and
● Replacing saturated fat with carbohydrate lowers HDL
as much as it lowers LDL. Triglycerides are increased tastes of people worldwide in coronary disease preven-
with common high glycemic index sugars and tion.
starches.
● Fish oil fatty acids do not lower LDL cholesterol but do REFERENCES
lower triglycerides. 1. Ascherio A. Epidemiologic studies on dietary fats and cor-
● Replacing saturated fat with monounsaturated or onary heart disease. Am J Med. 2002;113(Suppl 9B):13S–
polyunsaturated fat lowers HDL only slightly and does 24S.
not increase triglycerides. 2. Holme I. Cholesterol reduction and its impact on coronary
artery disease and total mortality. Am J Cardiol. 1995;76:
● Trans unsaturated fats produced by hydrogenating 10C–17C.
vegetable oil increase LDL and decrease HDL. 3. Shepherd J, Cobbe SM, Ford I, et al. Prevention of coronary
● The most favorable overall changes in plasma lipid heart disease with pravastatin in men with hypercholester-
concentrations are produced by replacing saturated olemia. N Engl J Med. 1995;333:1301–1307.
and trans unsaturated fats with unhydrogenated 4. Scandinavian Simvastatin Survival Study Group (4S). Ran-
domized trial of cholesterol lowering in 4444 patients with
monounsaturated or polyunsaturated fats. coronary heart disease: the Scandinavian Simvastatin Sur-
vival Study (4S). Lancet. 1994;344:1383–1389.
Dietary therapy that reduced CVD in clinical trials are 5. Sacks FM, Pfeffer MA, Moye LA, et al. The effect of prav-
as follows: astatin on coronary events after myocardial infarction in
patients with average cholesterol levels. N Engl J Med.
● Replacing saturated fat with polyunsaturated vegetable 1996;335:1001–1009.
oil 6. Martin MJ, Hulley SB, Browner WS, et al. Serum choles-
● Mediterranean diet after myocardial infarction terol, blood pressure, and mortality: implications from a
cohort of 361,662 men. Lancet. 1986;2:933–936.
● Indian vegetarian diet after myocardial infarction; and
7. Pekkanen J, Linn S, Heiss G, et al. Ten-year mortality from
● Fish oil after myocardial infarction. cardiovascular disease in relation to cholesterol level
among men with and without preexisting cardiovascular
(No randomized clinical trials of low-fat diets of suffi- disease. N Engl J Med. 1990;322:1700 –1707.
cient duration and size were available.) 8. Rose G, Reid DD, Hamilton PJ, et al. Myocardial ischemia,
The standard dietary approach recommended for cor- risk factors and death from coronary heart disease. Lancet.
onary disease prevention that increases carbohydrates 1977;1:105–109.
9. Kannel WB. Range of serum cholesterol values in the pop-
and reduces saturated fat has never been tested satisfac- ulation developing coronary artery disease. Am J Cardiol.
torily in a randomized clinical trial. The reductions in 1995;76:69C–77C.
HDL and the increases in triglycerides caused by carbo- 10. American Heart Association dietary guidelines. Revision
hydrate would be predicted to diminish or eliminate any 2000: a statement for healthcare professionals from the
benefits on coronary disease gained by lowering LDL. Nutrition Committee of the American Heart Association.
Circulation. 2000;102:2296 –2311.
Prospective epidemiologic studies also do not show re-
11. Gordon DJ, Probstfield JL, Garrison RJ, et al. High-density
duction in risk with carbohydrates. It is possible, but not lipoprotein cholesterol and cardiovascular disease. Four
proven, that low-fat diets could be effective in preventing prospective American studies. Circulation. 1989;79:8 –15.
coronary disease if they are based on whole grain cereals, 12. Hokanson JE, Austin MA. Plasma triglyceride level is a risk
legumes, fruits, and vegetables and are low in rapidly ab- factor for cardiovascular disease independent of high-den-
sity lipoprotein cholesterol level: a meta-analysis of popu-
sorbed refined carbohydrates. In contrast, several dietary
lation-based prospective studies. J Cardiovasc Risk. 1996;
approaches reduced coronary events in randomized clin- 3:213–219.
ical trials. Replacing saturated fat with polyunsaturated 13. Mensink RP, Katan MB. Effect of dietary fatty acids on
fat prevented coronary events in men. In men who had an serum lipids and liporoteins. A meta-analysis of 27 trials.
acute myocardial infarction, a Mediterranean diet and Arterioscler Thromb. 1992;12:911–919.
14. Stampfer MJ, Krauss RM, Ma J, et al. A prospective study
fatty fish improved survival. None of these trials had
of triglycerides, LDL particle diameter, and risk of myocar-
much impact on total fat intake but rather increased veg- dial infarction. JAMA. 1996;276:882–888.
etable oils, n-3 fatty acids, or many other plant foods or 15. Chen YD, Skowronski R, Coulston AM, et al. Effect of acute
nutrients that are linked to coronary prevention. The re- variations in dietary fat and carbohydrate intake on retinyl

December 30, 2002 THE AMERICAN JOURNAL OF MEDICINE威 Volume 113 (9B) 23S
 A Symposium: Effects of Dietary Fat and Carbohydrate on Plasma Lipoproteins and CVD/Sacks and Katan

ester content of intestinally derived lipoproteins. J Clin En- complications of atherosclerosis. Circulation. 1969;40(suppl
docrinol Metab. 1992;74:28 –32. II):II-1–II-63.
16. Garg A. High-monounsaturated-fat diets for patients with 36. Leren P. The Oslo Diet-Heart Study. Eleven-year report.
diabetes mellitus: a meta-analysis. Am J Clin Nutr. 1998; Circulation. 1970;42:935–942.
67(3 suppl):577S–582S. 37. Research Committee. Controlled trial of soya-bean oil in
17. Clarke R, Frost C, Collins R, Appleby P, Peto R. Dietary myocardial infarction. Lancet. 1968;ii:693–700.
lipids and blood cholesterol: quantitative meta-analysis of 38. Turpeinen O, Karvonen MJ, Pekkarinen M, Miettinen M,
metabolic ward studies. BMJ. 1997;314:112–117. Elosuo R, Paavilainen E. Dietary prevention of coronary
18. Tzonou A, Kalandidi A, Trichopoulou A, et al. Diet and heart disease: the Finnish Mental Hospital Study. Int J
coronary heart disease: a case-control study in Athens, Epidemiol. 1979;8:99 –118.
Greece. Epidemiology. 1993;4:511–516. 39. De Lorgeril M, Renaud S, Mamelle N, et al. Mediterranean
19. Sacks FM, Willett WC. More on chewing the fat: the good alpha-linolenic acid-rich diet in secondary prevention of
fat and the good cholesterol [editorial]. N Engl J Med. coronary heart disease. Lancet. 1994;343:1454 –1459.
1991;325:1740 –1742. 40. De Lorgeril M, Salen P, Martin JL, Monjaud I, Delaye J,
20. Katan MB. Are there good and bad carbohydrates for HDL Mamelle N. Mediterranean diet, traditional risk factors, and
cholesterol? Lancet. 1999;353:1029 –1030. the rate of cardiovascular complications after myocardial
21. Obarzanek E, Sacks FM, Vollmer WM, et al. Effects on infarction. Final report of the Lyon Heart Study. Circulation.
blood lipids of a blood pressure-lowering diet: the Dietary 1999;99:779 –785.
Approaches to Stop Hypertension (DASH) Trial. Am J Clin 41. De Lorgeril M, Salen P, Martin JL, Monjaud I, Boucher P,
Nutr. 2001;74:80 –89. Mamelle N. Mediterranean diet in a randomized trial. Pro-
22. Jenkins DJA, Kendall CWC, Augustin LSA, Vuksan V. High- longed survival and possible reduced cancer rate. Arch
complex carbohydrate or lente carbohydrate foods? Am J Intern Med. 1998;158:1181–1187.
Med. 2002;113(Suppl 9B):30S–37S. 42. Singh RB, Rostogi SS, Verma R, et al. Randomised con-
23. Research Committee. Low-fat diet in myocardial infarction. trolled trial of cardioprotective diet in patients with recent
Lancet. 1965;ii:501–504. acute myocardial infarction: results of one year follow up.
24. Burr ML, Fehily AM, Gilbert JF, et al. Effects of changes in BMJ. 1992;304:1015–1019.
fat, fish, and fibre intakes on death and myocardial 43. GISSI-Prevenzione Investigators. Dietary supplementation
infarction: diet and reinfarction trial (DART). Lancet. 1989; with n-3 polyunsaturated fatty acids and vitamin E after
ii:757–761. myocardial infarction: results of the GISSI-Prevenzione trial.
25. The Multiple Risk Factor Intervention Trial Research Group. Lancet. 1999;354:447–455.
Mortality after 16 years for participants randomized to the 44. Harris WS. Fish oils and plasma lipid and lipoprotein me-
Multiple Risk Factor Intervention Trial. Circulation. 1996;94: tabolism in humans: a critical review. J Lipid Res. 1989;30:
946 –951. 785–807.
26. Kallio V, Hamalainen H, Hakkila J, Luurila OJ. Reduction in 45. Weintraub MS, Zechner R, Brown A, et al. Dietary polyun-
sudden deaths by a multifactorial intervention programme saturated fats of the omega-6 and omega-3 series reduce
after acute myocardial infarction. Lancet. 1979;ii:1091– postprandial lipoprotein levels. J Clin Invest. 1988;82:
1094. 1884 –1894.
27. Ornish D, Brown SE, Scherwitz LW, et al. Can lifestyle 46. Sacks FM, Hebert P, Appel LJ, et al. The effect of fish oil on
changes reverse coronary heart disease? The Lifestyle blood pressure and high density lipoprotein cholesterol
Heart Trial. Lancet. 1990;336:129 –133. levels in phase 1 of the Trials of Hypertension Prevention.
28. Schuler G, Hambrecht R, Schlierf G, et al. Regular physical J Hypertension. 1994;12(suppl 7):S23–S31.
exercise and low-fat diet. Effects on progression of coro- 47. Morris MC, Sacks F, Rosner G. Does fish oil lower blood
nary artery disease. Circulation. 1992;86:1–11. pressure? A meta-analysis of controlled trials. Circulation.
29. Watts GF, Lewis B, Brunt JN, et al. Effects on coronary 1993;88:523–533.
artery disease of lipid-lowering diet, or diet plus cholestyra- 48. Billman GE, Hallaq H, Leaf A. Prevention of ischemia-in-
mine, in the St Thomas’ Atherosclerosis Regression Study duced ventricular fibrillation by omega-3 fatty acids. Proc
(STARS). Lancet. 1992;339:563–569. Natl Acad Sci USA. 1994;91:4427–4430.
30. Gordon DJ, Rifkind BM. High-density lipoprotein—the clin- 49. Sellmayer A, Witzgall H, Lorenz RL, Weber PC. Effects of
ical implications of recent studies. N Engl J Med. 1989;321: dietary fish oil on ventricular premature complexes. Am J
1311–1316. Cardiol. 1995;76:974 –976.
31. Bronte-Stewart B, Antonis A, Eales L, Brock JF. Effects of 50. Siscovic DS, Raghunathan TE, King I, et al. Dietary intake
feeding different fats on serum-cholesterol level. Lancet. and cell membrane levels of long-chain n-3 polyunsatu-
1956;i:521–527. rated fatty acids and the risk of primary cardiac arrest.
32. Keys A, Anderson JT, Grande F. Prediction of serum-cho- JAMA. 1995;274:1363–1367.
lesterol responses of man to changes in fats in the diet. 51. Katan MB, Mensink RP, Zock PL. Trans fatty acids and their
Lancet. 1957;ii:959. effect on lipoproteins in humans. Annu Rev Nutr. 1995;15:
33. Ahrens EH, Hirsch J, Insull W, Tsaltas TT, Blomstrand R, 473–493.
Peterson ML. The influence of dietary fats on serum lipid 52. Zock PL, Katan MB. Trans fatty acids, lipoproteins, and
levels in man. Lancet. 1957;i:943–953. coronary risk. Can J Physiol Pharmacol. 1997;75:211–216.
34. Malmros H, Wigand G. The effect on serum cholesterol of 53. Almendingen K, Jordal O, Kierulf P, Sanstad B, Pedersen
diets containing different fats. Lancet. 1957;ii:1–7. JI. Effects of partially hydrogenated fish oil, partially hydro-
35. Dayton S, Pearce ML, Hashimoto S, et al. A controlled genated soybean oil, and butter on serum lipoproteins and
clinical trial of a diet high in unsaturated fat in preventing Lp(a) in men. J Lipid Res. 1995;36:1370 –1384.

24S December 30, 2002 THE AMERICAN JOURNAL OF MEDICINE威 Volume 113 (9B)