Content uploaded by Janet Wittes
Author content
All content in this area was uploaded by Janet Wittes
Content may be subject to copyright.
ABSTRACT Effects of butter and 2 types of margarine on
blood lipid and lipoprotein concentrations were compared in a
controlled diet study with 23 men and 23 women. Table spreads,
added to a common basal diet, provided 8.3% of energy as fat.
Diets averaged 34.6% of energy as fat and 15.5% as protein.
Each diet was fed for 5 wk in a 3 3 3 Latin-square design. One
margarine (TFA-M) approximated the average trans monoene
content of trans fatty acid–containing margarines in the United
States (17% trans fatty acids by dry wt). The other margarine
(PUFA-M) was free of trans unsaturated fatty acids; it contained
approximately twice the polyunsaturated fatty acid content of
TFA-M (49% compared with 27% polyunsaturated fatty acids).
The tub-type margarines had similar physical properties at
ambient temperature. Fasting blood lipids and lipoproteins were
determined in 2 samples taken from the subjects during the fifth
week of each dietary treatment. Compared with butter, total
cholesterol was 3.5% lower (P= 0.009) after consumption of
TFA-M and 5.4% lower (P< 0.001) after consumption of PUFA-
M. Similarly, LDL cholesterol was 4.9% lower (P= 0.005) and
6.7% lower (P< 0.001) after consumption of TFA-M and PUFA-
M, respectively. Neither margarine differed from butter in its
effect on HDL cholesterol or triacylglycerols. Thus,
consumption of TFA-M or PUFA-M improved blood lipid
profiles for the major lipoproteins associated with cardiovascular
risk when compared with butter, with a greater improvement
with PUFA-M than with TFA-M. Am J Clin Nutr
1998;68:768–77.
KEY WORDS Margarine, butter, table spreads, cardio-
vascular disease risk factors, polyunsaturated fatty acids, trans
unsaturated fatty acids, blood lipids, lipoproteins, humans
INTRODUCTION
It is well established that saturated fatty acids (SFAs) raise
plasma total and LDL-cholesterol concentrations compared with
n26 polyunsaturated and n29 monounsaturated fatty acids
(MUFAs) (1–3). Therefore, recommendations to reduce the risk
of cardiovascular disease usually stress the importance of reduc-
ing intake of SFAs (4). In practical terms, this advice frequently
implies restricting the intake of products rich in SFAs and cho-
lesterol, such as butter, and replacing them in part with equiva-
lent products lower in cholesterol and SFAs but higher in unsat-
urated fatty acids, such as margarines.
Over the past 7–8 y, evidence has shown that, in addition to
SFAs, trans unsaturated fatty acids (TFAs) also raise plasma
total cholesterol and LDL cholesterol and may lower plasma
HDL cholesterol concentrations (5–8). In our laboratory, the
hypercholesterolemic effects of dietary TFAs were between
those of cis monounsaturated fatty acids and 12–16-carbon SFAs
(8). Some (9, 10), but not all (11), studies indicated that high
intakes of TFAs increase the risk of cardiovascular disease,
which agrees with the observed effects of TFAs on blood lipids.
The major source of TFAs in the diet are products containing
partially hydrogenated vegetable oils. Margarines are important
contributors to the intake of dietary TFAs in the United States.
Therefore, the issue arises as to whether there are benefits to
replacing a product rich in SFAs, such as butter, with a product
lower in SFAs but higher in cis- and TFAs, such as margarine. A
recent meta-analysis of dietary trials that directly compared the
effects of butter and margarine on blood lipids concluded that
replacement of butter by low-TFA soft margarines favorably
affects the blood lipoprotein profile, whereas high-TFA hard mar-
garines probably do not confer a benefit over use of butter (12).
It is now feasible to manufacture margarines that have no TFAs
and have high amounts of unsaturated fatty acids without major
increases in SFAs. In the current trial, we compared such a novel
margarine with butter and with a margarine containing a TFA con-
Effects of margarine compared with those of butter on blood lipid
profiles related to cardiovascular disease risk factors in
normolipemic adults fed controlled diets1–3
Joseph T Judd, David J Baer, Beverly A Clevidence, Richard A Muesing, Shirley C Chen, Jan A Weststrate, Gert W Meijer,
Janet Wittes, Alice H Lichtenstein, Montserrat Vilella-Bach, and Ernst J Schaefer
1From the Diet and Human Performance Laboratory, Beltsville Human
Nutrition Research Center, Agricultural Research Service, US Department of
Agriculture, Beltsville, MD; the Lipid Research Clinic, The George Wash-
ington University Medical Center, Washington, DC; Lipton, Baltimore;
Unilever Research Laboratory, Vlaardingen, Netherlands; Statistics Collabo-
rative, Washington, DC; and the Lipid Metabolism Laboratory, US Depart-
ment of Agriculture Human Nutrition Research Center on Aging at Tufts Uni-
versity, Boston.
2Supported in part by a Cooperative Research and Development Agree-
ment between the Agricultural Research Service, US Department of Agricul-
ture, and Unilever Research Laboratory.
3Address reprint requests to JT Judd, Diet and Human Performance Lab-
oratory, Beltsville Human Nutrition Research Center, Building 308, Room
214, BARC-East, Beltsville, MD 20705. E-mail: judd@bhnrc.arsusda.gov.
Received September 30, 1997.
Accepted for publication April 3, 1998.
Am J Clin Nutr 1998;68:768–77. Printed in USA. © 1998 American Society for Clinical Nutrition
768
by guest on July 13, 2011www.ajcn.orgDownloaded from
centration typical of most margarines consumed in the United
States to determine effects on plasma lipids, lipoprotein profiles,
vitamin E, and lipid hydroperoxide concentrations.
SUBJECTS AND METHODS
Study design
A controlled, crossover feeding trial was conducted at the
Beltsville Human Nutrition Research Center with 24 men and 24
women. The study was conducted from October 1995 to Febru-
ary 1996. All participants consumed 3 different diets for 5 wk.
Diet assignments were determined according to a 3 33 Latin-
square design. This design was chosen to ensure complete bal-
ance of the number of diets administered in each study period as
well as the number of occurrences of each diet sequence (13).
The random sequence of diet assignments was also balanced
with respect to sex and baseline plasma LDL-cholesterol con-
centration. During the fifth week of each period, duplicate blood
samples were collected. Subjects were switched from one diet to
the next without a washout between periods. Because the timing
of the study included the Christmas and New Year’s Day holi-
days, the study periods were timed to complete 2 study periods
before Christmas. Subjects were maintained on the period 2 diet
during the holiday period, and then switched to the period 3 diet
after January 2.
Blinding of study results
Menus, menu food items and portions, and dietary treatments
were color coded during the study. Although study participants
easily recognized the butter diet, differences in appearance,
taste, and other characteristics between the 2 margarines were
not apparent. All samples were coded to blind analysts to treat-
ments. Analytic data were blinded to those who performed the
controlled feeding and sample collection phases of the investiga-
tion. After all data for an analyte were entered and the database
was locked, the treatments were decoded by the statistician per-
forming statistical analysis of the data.
Subject selection
Volunteers were recruited by advertisement in the area of the
Beltsville Agricultural Research Center, Beltsville, MD. Men
and women of all races between the ages of 25 and 65 y were
recruited regardless of smoking habits. From the 601 respon-
dents, 69 met the eligibility criteria described below; 48 (24 men
and 24 women) were selected and entered the study.
Minimum eligibility criteria were based on general health, eat-
ing habits, age, body mass index (BMI), and fasting plasma HDL-
cholesterol, LDL-cholesterol, and triacylglycerol concentrations.
Volunteers were required to be within 85–120% of sex-specific
ideal BMI specified by life insurance reference tables (14). Vol-
unteers who reported taking lipid-lowering drugs, blood pressure
medications, or dietary supplements, or who had eating habits
inconsistent with the study protocol (eg, those consuming vege-
tarian or low-fat diets) were excluded. Volunteers were evaluated
by a physician and determined to be in good health, with no signs
or symptoms of hypertension, hyperlipemia, diabetes, peripheral
vascular disease, gout, liver or kidney disease, or endocrine dis-
orders. Subjects selected for the study were required to have fast-
ing plasma HDL-cholesterol concentrations >0.65 mmol/L (25
mg/dL) for men and >0.91 mmol/L (35 mg/dL) for women, and
fasting plasma triacylglycerol concentrations <3.39 mmol/L (300
mg/dL). From the volunteers who met all other selection criteria,
those selected to participate had plasma LDL-cholesterol concen-
trations between the 25th and 75th percentile (mean: 3.51
mmol/L, 63rd percentile) of those screened.
Women taking hormones for postmenopausal replacement
therapy (n= 10) or birth control (n= 5) were accepted as sub-
jects with the provision that they continue their normal regimen
(type of hormone, schedule, and dose) for the duration of the
study and that they record their medication on their daily ques-
tionnaire. Exercise was not controlled but subjects were encour-
aged to maintain their normal exercise patterns (type of exercise,
duration, and frequency) throughout the study and were required
to record exercise on their daily questionnaire. Smoking was not
disallowed, but only one of the participants smoked. The number
of cigarettes smoked was reported on the daily questionnaire.
This female subject smoked a maximum of 10 cigarettes/d, and
on most days smoked <10.
Volunteers were fully informed of study requirements. They
were required to read and sign a consent form detailing the study
objectives, risks, and benefits before final selection as subjects
for the study. All procedures were approved by the Johns Hop-
kins University Committee on Human Research.
Basal diet and table spreads
All spreads were fed in amounts similar to those of the US diet.
Spreads were added to a basal diet to ensure that total dietary
intake of maconutrients and major fatty acids was similar to a
typical American diet with respect to fat and fatty acids. A basal
diet with a 7-d menu cycle was designed so that, when the table
spread was included, the percentage of energy from fat would be
<37%, that from protein 15%, and that from carbohydrate 48%.
Data from the US Department of Agriculture Handbook no. 8
series (revised series 1–21) (15), together with analyzed values
for the table spreads, were used to formulate the diets.
One of 3 different table spreads (described below) was fed
along with the basal diet and in an amount that would provide 8
en% from fat. This amount of table spread was selected to
approximate the 90th percentile estimate for both butter and
margarine consumption in the American diet for men and women
35–60 y of age who reported that they consumed butter and mar-
garine in the Nationwide Food Consumption Survey, 1977–78
(16). More recent surveys do not list butter and margarine con-
sumption as separate categories. However, on the basis of trends
in per capita consumption of these spreads (17), the 90th per-
centile estimate for current consumption should be approxi-
mately the same.
Monday through Friday, all subjects consumed breakfast and
dinner at the Beltsville Human Nutrition Research Center’s
Human Study Facility under the supervision of a dietitian. At
breakfast, each subject was provided with a carryout lunch to be
consumed that day. Snack items were included in the daily menu,
and subjects were provided the option of consuming the snacks
at dinner or later in the evening. Table spreads were provided
only at breakfast and dinner. Meals for the weekend were pack-
aged for consumption at home and provided to the subjects, with
written instructions, after dinner on Friday. Unlimited amounts
of coffee, tea, and diet sodas were allowed but all additives
(sugar and milk) for coffee and tea were provided with the meals.
Only foods provided by the Human Study Facility were allowed
to be consumed during the study.
TABLE SPREADS AND PLASMA LIPIDS 769
by guest on July 13, 2011www.ajcn.orgDownloaded from
Each morning, Monday through Friday, subjects were weighed
before breakfast when they arrived at the facility. Energy intake
was adjusted in 840- or 1680-kJ (200- or 400-kcal) increments to
maintain initial body weight. Subjects were fed the same items
and the same proportions of each item relative to total dietary
energy. Therefore, the relative amounts of all nutrients, other than
those provided by the table spread, were constant for all subjects.
Each day, subjects completed a questionnaire detailing beverage
intake, factors related to dietary compliance, exercise, medica-
tions, and illnesses. The questionnaires were reviewed routinely
by a study investigator and all problems identified were discussed
with the subject during the next meal.
Each table spread was from a single production lot prepared
specifically for the study before it began. Two tub-type mar-
garines were prepared by Lipton, Baltimore, to meet criteria
established for the spreads to be used in the investigation. All
table spreads were stored under refrigeration for the duration of
the study. Butter was obtained from a commercial creamery
(Sommer Made Creamery, Doylestown, PA). The fatty acid com-
position of the margarine containing TFAs (TFA-M) was based
on the market share weighted-average fatty acid composition of
all forms of the 21 margarines that comprised 69% of the US
market share during the 12-wk period ending in July 1995 (Niel-
son Scantrack Data, AC Nielson Co, Northbrook, IL). The distri-
bution of 18:1 isomers in TFA-M was typical of US margarines
based on partially hydrogenated soybean oil. The margarine high
in polyunsaturated fatty acids (PUFA-M) was produced by
blending liquid sunflower oil with completely hydrogenated
soybean and canola oils. The product was produced with as
much linoleic acid and as little SFAs as possible while maintain-
ing the desired physical characteristics, and it contained virtually
no TFAs. The sensory and physical characteristics of the 2 mar-
garines were similar and were comparable with commercially
available tub-type spreads.
Chemical analysis of diets and table spreads
During the first feeding period, 2 composites of the 7 diets were
collected at 2 energy levels. The food was prepared as though it
were to be consumed and then was mixed in a blender with ice
added to prevent heat buildup. The blended samples were freeze-
dried in preweighed containers and then reweighed. The samples
were then pulverized and weekly composites were prepared by
mixing 15% of each day’s dry weight. Composite samples of the
table spreads were collected from 10 randomly selected containers
and blended together. Thus, 4 weekly composites and duplicate
composites of the table spreads were analyzed for dry matter,
crude protein, crude fat, total dietary fiber, and ash (Corning
Hazleton, Inc, Madison WI). Fatty acid composition of food com-
posites was determined by gas chromatographic separation of fatty
acid methyl esters. a- and g-Tocopherol and lipid hydroperoxide
contents of the table spreads were also determined.
Blood sample collection and analysis
Baseline blood samples were collected during the week before
initiation of the controlled feeding. Subsequently, samples were
collected at 5-wk intervals. The 48 subjects were randomly
divided into 2 groups. One group had samples drawn on Mon-
days and Wednesdays and the other on Tuesdays and Thursdays.
Procedures for blood sampling and processing were those
described in the protocol for the Lipid Research Clinics Program
(18). Blood samples were drawn after an overnight fast (mini-
mum of 12 h), immediately before breakfast. Samples for blood
lipid, tocopherol, and lipid hydroperoxide analyses were col-
lected by venipuncture using a 19-gauge butterfly needle into
evacuated tubes containing disodium EDTA. After being mixed
gently by inversion, the samples were placed immediately on wet
ice. Within 30 min of collection, plasma was separated by cen-
trifugation at 1400 3gfor 20 min at 48C.
Plasma was removed from the tubes, divided into samples,
and stored in cryogenic vials at 2808C. Before storage, propyl
gallate was added to the samples used for tocopherol analyses,
and the sample to be used for HDL and HDL3determination was
precipitated by using the sequential precipitation procedure of
Gidez et al (19). Supernates from the HDL precipitation were
stored at 2808C for later analysis of cholesterol. Analyses for
cholesterol, triacylglycerols, apolipoproteins, and other blood
components were performed after the final blood collection. All
analyses of the samples from each subject were performed in the
same analytic run.
Lipid analyses (cholesterol, triacylglycerols, and HDL choles-
terol) were performed at the Lipid Research Clinic Laboratory,
The George Washington University Medical Center, which main-
tains standardization with the Centers for Disease Control and
Prevention, US Department of Health and Human Services.
Plasma total cholesterol, HDL cholesterol, HDL3cholesterol, and
triacylglycerols were determined enzymatically with commercial
kits (Sigma Chemical Company, St Louis) on an Abbott VP ana-
lyzer (Abbott Laboratories, Chicago). HDL2 cholesterol was
determined as the difference between HDL cholesterol and HDL3
cholesterol. LDL cholesterol was calculated by using the Friede-
wald equation (20). Plasma apolipoprotein (apo) A-I and B con-
centrations were determined by rate nephelometry (Beckman ICS
Immunochemical analyzer; Beckman Instruments, Fullerton,
CA). Apo A-II was determined by radial immunodiffusion.
Plasma lipoprotein(a) [Lp(a)] was analyzed as described previ-
ously (21–23) by using a commercially available enzyme-linked
immunosorbent assay (Strategic Diagnostics, Newark, DE).
a- and g-Tocopherols were determined by using HPLC. Sam-
ples were extracted with isopropanol containing propyl gallate,
and distilled water and hexane containing butylated hydroxy-
toluene (BHT). BHT and an internal standard (tocol) were
added. After vortex mixing, the mixture was allowed to stand and
an aliquot removed from the hexane layer. This aliquot was dried
under nitrogen gas and the residue was redissolved in 120 mL
ethanol:methylene chloride (90:10, by vol) and propyl gallate.
Fifty microliters was injected onto a reversed-phase column
(Microsorb-MV C18, 4.6 mm internal diameter 325 cm, Rainin
Instrument Co, Inc, Woburn, MA) with a mobile phase of
methanol:acetonitrile:methelene chloride (75:20:5) at a flow rate
of 1.5 mL/min. Tocol and tocopherols were detected by spectro-
fluorescence at an excitation wavelength of 294 nm and an emis-
sion wavelength of 324 nm.
Lipid hydroperoxides were determined by using a commer-
cially available kit (Kamiya Biomedical Company, Thousand
Oaks, CA). The samples were dissolved in isopropanol and water
containing triton X-100. The assays were performed on a
microtiter plate. LDL size was determined by nuclear magnetic
resonance spectroscopy (24).
Statistical analysis
All analyses were performed by using SAS for Windows ver-
sion 6.11 or S-Plus (SAS Institute, Cary, NC). The analytic plan
770 JUDD ET AL
by guest on July 13, 2011www.ajcn.orgDownloaded from
was designed a priori and described a mixed-effects model for
analysis of the data (25). For each variable, the average of 2 sam-
ple measurements taken during week 5 of each feeding period
was analyzed by using an analysis of variance model that
included terms for sex, period, and carryover of a diet from one
period to the next. Because the investigators predicted that the
interaction terms would account for only a small amount of the
variation in the data, the analytic plan specified that an interac-
tion term would be included in the final model if it were signifi-
cant at the nominal 0.15 level in the presence of the other terms.
The contrasts between diets were tested by an F test for differ-
ences between groups.
RESULTS
Subjects
Twenty-four men and 24 women completed the screening
process and began the controlled feeding. One man withdrew
because of a personal conflict and one woman withdrew because
of travel that interfered with the feeding protocol. Thus, 23 men
and 23 women completed the feeding phase of the study. Data
were analyzed statistically only for subjects who completed all 3
feeding periods. Characteristics of the participants at baseline
are presented in Table 1. The age range of subjects who com-
pleted the study was 28–65 y with a mean of 46.9 y for men and
46.7 y for women. Baseline BMI (in kg/m2) for men was 25.7
and mean calculated metabolizable energy intake during the
feeding phase was 12.66 MJ/d (3025 kcal/d). For women, base-
line BMI was 24.8 and mean metabolizable energy intake was
9.10 MJ/d (2175 kcal/d). At baseline, mean plasma lipid- and
lipoprotein-cholesterol concentrations for men and women com-
bined were: 1.24 mmol/L for HDL cholesterol, 3.36 mmol/L for
LDL cholesterol, 5.12 mmol/L for total cholesterol, and 1.15
mmol/L for triacylglycerol. Total cholesterol and triacylglycerol
concentrations were higher at baseline than after any of the diets
with table spread added.
Diets and table spreads
Analyzed macronutrient and fatty acid compositions of the
basal diet and table spreads are presented in Table 2. The 2 mar-
garines differed in oil blend but had similar product characteris-
tics (taste, hardness, mouth feel, spreadability, and melting
behavior). The oil blend in TFA-M was liquid soybean and par-
tially hydrogenated soybean oils; the oil blend in PUFA-M was
liquid sunflower oil and completely hydrogenated soybean and
canola oils. On a dry weight basis, the SFA content of TFA-M
and PUFA-M was 31% and 39%, respectively, of that of butter.
The butter contained appreciable amounts of myristic and lauric
acids (13% and 3% of SFAs, respectively), whereas the mar-
garines contained relatively small amounts of myristic acid and
only trace amounts of lauric acid. Although the absolute amount
TABLE SPREADS AND PLASMA LIPIDS 771
TABLE 1
Characteristics of the participants at baseline1
Men Women All
(n= 23) (n= 23) (n= 46)
Age (y) 46.9 ±2.23 46.7 ±1.88 46.8 ±1.44
Weight (kg) 83.4 ±2.34 65.9 ±2.15 74.6 ±2.04
Body mass index (kg/m2) 25.7 ±0.56 24.8 ±0.67 25.3 ±0.44
Total cholesterol (mmol/L) 5.11 ±0.11 5.14 ±0.15 5.12 ±0.09
LDL cholesterol (mmol/L) 3.51 ±0.09 3.21 ±0.13 3.36 ±0.08
HDL cholesterol (mmol/L) 1.08 ±0.06 1.40 ±0.09 1.24 ±0.06
HDL2 cholesterol (mmol/L) 0.30 ±0.03 0.46 ±0.07 0.38 ±0.04
HDL3cholesterol (mmol/L) 0.78 ±0.03 0.94 ±0.04 0.86 ±0.03
Triacylglycerols (mmol/L) 1.15 ±0.10 1.15 ±0.10 1.15 ±0.07
Apo A-I (g/L) 1.49 ±0.04 1.80 ±0.07 1.65 ±0.05
Apo A-II (g/L) 0.32 ±0.02 0.38 ±0.02 0.35 ±0.01
Apo B (g/L) 0.80 ±0.03 0.78 ±0.03 0.79 ±0.02
1x
–±SEM. Apo, apolipoprotein.
TABLE 2
Analyzed composition of basal diet and table spreads fed to subjects for 5 wk1
Table spread
Basal diet Butter TFA-M PUFA-M
Protein (%) 19.3 0.74 0.63 1.59
Total carbohydrate (%) 62.19 1.35 1.96 1.6
Total dietary fiber (%) 4.03 0 0 0
Fat (%) 14.91 96.09 95.66 95.23
a-Tocopherol (mg/g) — 76 76 397
g-Tocopherol (mg/g) — 38 275 129
a-Tocopherol equivalents (mg/g) — 80 104 410
Short-chain fatty acids, 4–10 carbons (%) 0.05 2.43 0 0
Total saturated fatty acids, 12–24 carbons (%) 3.64 53.1 16.44 20.55
Lauric acid 0.03 1.67 ND2TR3
Myristic acid 0.16 7.14 0.07 0.12
Palmitic acid 1.97 27.58 7.93 5.56
Stearic acid 1.22 15 7.14 13.16
Total monounsaturated fatty acids, 14:1–20:1 (%) 6.35 30.63 47.87 20.5
Oleic (cis 18:1) 4.81 24.64 30.18 20.22
trans Monoenes (trans 18:1) 1.33 3.14 17.35 ND2
Total polyunsaturated fatty acids (%) 3.74 5.12 26.57 49.41
Linoleic 3.22 3.29 22.66 48.22
Linolenic 0.33 0.47 3.46 1.1
1Dry weight basis. TFA-M, margarine containing trans fatty acids; PUFA-M, margarine containing polyunsaturated fatty acids.
2None detected.
3Trace detected.
by guest on July 13, 2011www.ajcn.orgDownloaded from
of SFA in PUFA-M was higher than that in TFA-M, the ratio of
stearic acid to the sum of lauric, myristic, and palmitic acids was
higher for PUFA-M than for TFA-M (<2.3:1 and 0.9:1 for PUFA-
M and TFA-M, respectively). Oleic acid (cis-18:1n29) was 33%
lower in PUFA-M than in TFA-M. The oleic acid content of butter
was between that of PUFA-M and TFA-M.
Contributions to total dietary energy from the fat and fatty acids
in the basal diet and each of the table spreads are shown in Table
3. Differences in the fatty acid composition among the 3 diets
were not as great as among the 3 table spreads. All diets contained
similar amounts of MUFA. The 2 margarine diets had comparable
energy from SFAs. The major differences among the 3 diets were
that the butter diet had higher amounts of 12+14+16-carbon fatty
acids than did the other diets, the PUFA-M diet had more PUFAs
than did the butter diet and the TFA-M diet, and the TFA-M diet
had more TFAs than did the butter diet and the PUFA-M diet.
The basal diet provided <0.065 mmol cholesterol/MJ (25.1
mg/MJ). Because the margarines contained no cholesterol, the diets
with margarine provided cholesterol at the level in the basal diet.
Butter added to the basal diet provided an additional 0.015
mmol/MJ (5.8 mg/MJ) for a diet total of 0.079 mmol choles-
terol/MJ (30.5 mg/MJ).
Energy from fat in all diets as fed, ie, basal plus table spread, was
<34.6 en%. This is about that reported now for a typical Ameri-
can diet (26). It is, however, slightly lower than our targeted 37%
of energy from fat. Fat from table spreads contributed 8.2–8.4% of
energy as compared with our targeted value of 8% of energy.
Because these differences from the targeted values occurred in all
diets, they should not affect interpretation of the relative effects of
the spreads on plasma lipids and other variables.
The basal diet contributed slightly more than 3 times the
energy from fat as did the table spreads (ie, 26.3% compared
with 8.3% of energy; Table 3). This diluted the effect of fatty
acids from the spreads on biological responses to the total diet.
However, this represents how table spreads are typically con-
sumed and allows evaluation of their biological effects in exper-
imental diets when they are fed at average intakes.
Macronutrient and fatty acid intakes from the 3 diets for men
and women are presented in Table 4. Mean daily intake of table
spreads for men was 35 g for butter, 41 g for TFA-M, and 41 g
for PUFA-M. These amounts of spreads each provided 28 g fat.
For women, intake was 25 g for butter, 30 g for TFA-M, and 30
g for PUFA-M, which provided 20 g fat each. TFA intakes from
the 3 diets were 2.4 en% (PUFA-M), 2.7 en% (butter), and 3.9
en% (TFA-M), which is similar to estimates for intake of TFAs
in the US diet of 3–6 en% (8). Total dietary fiber was supplied
by the basal diet and was therefore constant across all treat-
ments. Mean daily total dietary fiber intake was 24 g for men and
17 g for women.
The a-tocopherol concentrations of butter and TFA-M were
similar. The a-tocopherol concentration of PUFA-M was 5 times
greater than that of the other table spreads. The g-tocopherol
concentration was <3 times greater in PUFA-M than in the but-
ter diet and 2 times greater in TFA-M than in PUFA-M (Table 2).
Vitamin E activity expressed as RRR-a-tocopherol equivalents
was calculated as a-tocopherol + 0.1 3g-tocopherol (in mg),
and was 80, 104, and 410 as RRR-a-tocopherol equivalents/g for
butter, TFA-M, and PUFA-M, respectively (Table 2).
Biological responses to diets
Sex differences
As expected from documented, inherent sex differences,
women had higher concentrations of HDL cholesterol, HDL2
cholesterol, HDL3 cholesterol, apo A-I, and apo A-II and lower
concentrations of LDL cholesterol and apo B than did men. The
responses to all table spreads were, however, similar for men and
772 JUDD ET AL
TABLE 3
Composition of the test diets1
Basal diet + table spread
Butter TFA-M PUFA-M
% of energy
Protein 15.5 ±0.16 15.5 ±0.16 15.5 ±0.16
Carbohydrate 50.0 ±0.15 50.0 ±0.15 49.9 ±0.15
Fat 34.5 ±0.09 34.6 ±0.09 34.6 ±0.09
Saturated fatty acids 11.2 ±0.03 7.9 ±0.01 8.3 ±0.01
(12–24 carbons)
Stearic acid 3.5 ±0.01 2.8 ±0.00 3.3 ±0.01
cis Monoenes 10.8 ±0.01 11.2 ±0.01 10.4 ±0.01
trans Monoenes 2.7 ±0.00 3.9 ±0.01 2.4 ±0.00
Polyunsaturated fatty acids 7.2 ±0.00 9.0 ±0.01 10.8 ±0.01
1x
–±SEM for analyses of 4 composites of the 7-d menu cycle for each
diet. TFA-M, margarine containing trans fatty acids; PUFA-M, margarine
containing polyunsaturated fatty acids.
TABLE 4
Daily nutrient intake of men and women, determined by using analytical data on the diets with table spreads added1
Basal diet + table spread
Butter TFA-M PUFA-M
Men Women Men Women Men Women
g
Protein 117 84 117 84 118 84
Carbohydrate 378 272 378 272 377 271
Fat 116 84 116 84 116 84
Saturated fatty acids (12–24 carbons) 38 27 27 19 28 20
Stearic acid 12 9 9 7 11 8
cis Monoenes 36 26 38 27 35 25
trans Monoenes 9 7 13 9 8 6
Polyunsaturated fatty acids 24 17 30 22 37 26
1Table values are means. Mean (±SD) energy intake for men (n= 23) was 12.66 ±1.27 MJ/d (3025 ±304 kcal/d) and for women (n= 23) was 9.10 ±0.89
MJ/d (2175 ±212 kcal/d). TFA-M, margarine containing trans fatty acids; PUFA-M, margarine containing polyunsaturated fatty acids.
by guest on July 13, 2011www.ajcn.orgDownloaded from
women; therefore, the overall least-square means are reported
for the sexes combined (Table 5).
Lipids and lipoproteins
Concentrations of plasma triacylglycerol, lipoprotein choles-
terol, and apolipoproteins for all subjects are presented in Table
5. There were no diet-by-period interactions detected for any
variables. Contrasts were used to estimate differences between
diets containing butter and PUFA-M, butter and TFA-M, and
PUFA-M and TFA-M. Because P values were calculated sequen-
tially, the significance of the effects of diet on the measured vari-
ables were corrected for subject, sex, period, and carryover.
After consumption of the butter diet, mean LDL-cholesterol
concentration was 5.2% (0.17 mmol/L) higher than after con-
sumption of the TFA-M diet (P= 0.005) and 7.2% (0.23
mmol/L) higher than after consumption of the PUFA-M diet (P<
0.001). There was an average 1.8% (0.06 mmol/L) decrease in
LDL cholesterol after consumption of the PUFA-M diet com-
pared with the TFA-M diet (P= 0.017).
Total cholesterol followed a pattern similar to that of LDL
cholesterol. After consumption of the butter diet, mean total cho-
lesterol was 3.6% (0.18 mmol/L) higher than after consumption
of TFA-M (P= 0.009) and 5.7% (0.28 mmol/L) higher than after
consumption of PUFA-M (P< 0.001). Mean total cholesterol
decreased 2.0% (0.10 mmol/L) after consumption of PUFA-M
compared with TFA-M (P= 0.010).
There were no significant differences among the diets for
HDL cholesterol, HDL2 and HDL3 cholesterol fractions, and tri-
acylglycerols (Table 5). The ratio of total to HDL cholesterol
decreased by an average of 3.9% (P= 0.032) after consumption
of the PUFA-M diet compared with the butter diet (Table 5).
There were no significant differences in this ratio between the
butter diet and the TFA-M diet or between the PUFA-M and
TFA-M diets. A similar significant decrease was observed in the
ratio of LDL to HDL cholesterol after consumption of PUFA-M
compared with the butter diet (6.1%, P= 0.023).
There were no significant differences among the diets for apo
A-II concentrations. Apo A-I was significantly lower after con-
sumption of PUFA-M than after TFA-M (Table 5). There was no
difference in apo B concentration after consumption of the but-
ter diet compared with the TFA-M diet; however, apo B was
lower after PUFA-M than after both TFA-M (P= 0.002) and the
butter diet (P< 0.001).
Differences in LDL particle size after consumption of the
spreads were <0.5%, and were considered to be biologically
unimportant. The largest LDL particles were associated with the
butter diet (21.0 ±0.06 nm). LDL particle size for both the TFA-
M and PUFA-M diets was 20.9 ±0.06 nm.
Lipoprotein(a)
Eight of the subjects who completed the study had baseline
Lp(a) concentrations ≤10 mg/L. All of these showed no response
to diet. Because these observations were at the lowest detectable
level for the assay and thus deflate the estimated variance inap-
propriately, we considered them missing for the purpose of sta-
tistical analysis. Data for all subjects having Lp(a) concentrations
>10 mg/L are included in Table 5. For the 38 subjects with base-
line Lp(a) concentrations >10 mg/L, there was a highly signifi-
cant response to diet (Table 5). Compared with butter, consump-
tion of TFA-M resulted in an average 8.6% increase in Lp(a) (P<
0.001) whereas consumption of PUFA-M resulted in an average
5.9% increase (P= 0.008). There was no significant difference in
Lp(a) concentration between the margarine treatments (P= 0.08).
Lp(a) response to diet was examined statistically for subjects
grouped by plasma concentrations at baseline in a way similar to
the procedure reported from a previous study in our laboratory
(27). Subjects with baseline concentrations ≤50 mg/L (n= 11)
showed no response to diet. Subjects with medium baseline con-
centrations, between 50 and 200 mg/L, and high baseline concen-
trations, ≥200 mg/L, showed significant responses to diet. The
response pattern was similar for the medium and high Lp(a)
groups. Analysis of the data based on this grouping still showed no
significant difference in response to the 2 margarine treatments.
Tocopherols and lipid hydroperoxides
Plasma a-tocopherol concentration was higher after consump-
tion of PUFA-M than after consumption of both the TFA-M and
butter diets; however, there was no difference in a-tocopherol
concentration after consumption of TFA-M compared with the
butter diet (Table 5). g-Tocopherol concentration was higher
after consumption of TFA-M than after consumption of the but-
ter diet and higher after consumption of the butter diet than after
PUFA-M. There was a carryover effect of diet for g- but not for
a-tocopherol concentration. There were no significant differ-
ences in plasma lipid hydroperoxide concentrations among the
diets with the spreads added.
DISCUSSION
This investigation was a large dietary trial comparing the
effects of butter and margarines on plasma lipids, lipoproteins,
and antioxidant status. Two margarines were used: a margarine
high in PUFA made with liquid sunflower and completely hydro-
genated soybean and canola oils, and a margarine made with liq-
uid soybean and partially hydrogenated soybean oils and having
TFAs approximating that in an average margarine on the US
TABLE SPREADS AND PLASMA LIPIDS 773
TABLE 5
Plasma lipids of adults after 5 wk of consuming each table spread1
Butter TFA-M PUFA-M
(n= 46) (n= 46) (n= 46)
Total cholesterol (mmol/L) 5.15 ±0.09a4.97 ±0.09b4.87 ±0.09c
LDL cholesterol (mmol/L) 3.44 ±0.09a3.27 ±0.09b3.21 ±0.09c
HDL cholesterol (mmol/L) 1.27 ±0.04 1.24 ±0.04 1.24 ±0.04
HDL2cholesterol (mmol/L) 0.42 ±0.03 0.40 ±0.03 0.41 ±0.03
HDL3cholesterol (mmol/L) 0.84 ±0.02 0.84 ±0.02 0.84 ±0.02
Triacylglycerols (mmol/L) 0.98 ±0.05 0.99 ±0.05 0.92 ±0.05
Apo A-I (g/L) 1.61 ±0.04a,b 1.62 ±0.04a1.59 ±0.04b
Apo A-II (g/L) 0.32 ±0.01 0.32 ±0.01 0.32 ±0.01
Apo B (g/L) 0.76 ±0.02a0.74 ±0.02a0.72 ±0.02b
Lipoprotein(a) (mg/L)2186 ±28a202 ±28b197 ±28b
Total:HDL cholesterol 4.33 ±0.14a4.22 ±0.14a,b 4.16 ±0.14b
LDL:HDL cholesterol 2.96 ±0.13a2.83 ±0.13a,b 2.79 ±0.13b
a-Tocopherol (mmol/L) 24.8 ±0.6a24.7 ±0.6a26.1 ±0.6b
g-Tocopherol (mmol/L) 4.9 ±0.16a5.4 ±0.16b4.5 ±0.16c
1Estimated means ±SEEs based on a mixed model that adjusted for
sex, dietary period, and carryover. Means within a row with different super-
script letters are significantly different, P ≤ 0.05. Apo, apolipoprotein; TFA-
M, margarine containing trans fatty acids; PUFA-M, margarine containing
polyunsaturated fatty acids.
2n= 38 subjects; the 8 subjects who had baseline lipoprotein(a) con-
centrations ≤10 mg/L all showed no response to diet and were excluded
from the analysis.
by guest on July 13, 2011www.ajcn.orgDownloaded from
market. The margarines were prepared so that they would have
very similar physical and sensory characteristics and were con-
stituted to compare favorably with commercial tub-type mar-
garines available on the open market. Compliance with the diets
was judged by direct observation of consumption of weekday
meals, weight being maintained with only infrequent changes in
the energy provided in the meals, and by evaluation of the daily
questionnaires completed by the subjects. Dietary compliance
was excellent for the subjects who consumed all 3 diets and
whose data are included in the results.
Estimates for the daily consumption of margarine and butter
are 11 and 6 g/person, respectively, in the United States (17), 22
and 3 g/person in Netherlands (28), and 12 and 7 g/person in
Britain (29). Because per capita estimates include both those who
consume and do not consume butter or margarine, as well as chil-
dren, who generally consume less than adults, in the investigation
reported here the table spreads were fed in amounts based on
adult consumers of the products. The table spreads were fed in
amounts comparable with the 90th percentile estimate for US
intake of butter or margarine by male and female consumers of
table spreads in the age range of participants in this study (16).
The spreads were included in an experimental diet that approxi-
mated a typical US diet with respect to fat and fatty acid contents.
Protein, carbohydrate, and fat in our diets provided 15.5%,
50.0%, and 34.6% of energy, respectively, whereas intake esti-
mates in the survey data for 1994 (26) were 15.5%, 51.4%, and
33.1% of energy, respectively. The energy from fat was, how-
ever, slightly lower than the targeted 37% that was calculated on
the basis of US Department of Agriculture Handbook no. 8 data
(15). Although no attempt to determine the reason for this dif-
ference has yet been made, it is not unlikely that fat values in the
handbook are higher than they should be for some foods in view
of recent trends toward less fat in many food products. This may
be especially true for meats and meat products.
SFAs provided from 7.9% (TFA-M) to 11.2% (butter) of
energy in our diets compared with 11.4% of energy in the survey
(26). The diet with butter provided 7.2% of energy as PUFAs and
13.5% of energy as MUFAs; these values were 6.5% and 12.6%,
respectively, in the survey. By design, both margarines were
higher in PUFAs and MUFAs than was butter.
There were substantial improvements in plasma LDL-choles-
terol concentrations after consumption of both margarine diets
compared with the butter diet (on average, concentrations were
4.9–6.7% lower). In addition, the LDL-cholesterol average was
1.8% lower after consumption of PUFA-M than after consump-
tion of TFA-M. Apo B, the major apolipoprotein in LDL, was
lower after consumption of both margarine diets than after the
butter diet, but the difference was significant only after PUFA-
M. Although a predominance of small, dense LDL particles was
indicated to be an important determinant of cardiovascular dis-
ease risk (30–35), we observed no difference in particle size
associated with the type of table spread consumed.
There were no differences in HDL cholesterol, HDL2- and
HDL3 cholesterol fractions, or apo A-II resulting from the type of
table spread consumed. Apo A-I concentration was significantly
lower after consumption of PUFA-M than after TFA-M. High
intakes of PUFAs have been shown to lower HDL-cholesterol
concentrations, but such effects are usually observed when
PUFAs exceed <20% of energy in the diet (36). These effects
would not be expected for PUFA intakes of 11% of energy, as in
the PUFA-M diet fed here. Studies in our laboratory showed that
SFAs raise HDL- as well as LDL-cholesterol concentrations
(37). Because the diet with butter was considerably higher in
SFAs, a similar effect on HDL cholesterol may have occurred in
this study. However, this effect, if present, was not great enough
to produce a significant elevation in HDL cholesterol with the
butter diet as compared with the 2 margarine diets.
In the 1990 report of strategies for blood cholesterol reduction
from the National Cholesterol Education Program (38), it was
estimated that for every 1% reduction in cholesterol concentra-
tion, the risk of cardiovascular disease decreased by an average
of 2%. Application of this prediction to results of the present
study would indicate an average reduction in risk for cardiovas-
cular disease of 7% for TFA-M compared with butter and 11%
for PUFA-M compared with butter. Individuals consuming a
smaller or larger amount of butter than in this study would
expect smaller or larger changes in blood cholesterol and risk for
cardiovascular disease when switching from butter to margarine.
In addition, the prediction equation that relates reductions in
blood cholesterol concentrations to reduction in risk for cardio-
vascular disease is not likely to be valid for every person; it will
underestimate risk for some individuals while overestimating
risk for others.
Lp(a), a particle similar to LDL in lipid composition, but with
2 major proteins, apo B-100 and apo(a) (21), may be a risk fac-
tor for the development of cardiovascular disease (21, 22).
Although the role of Lp(a) in promoting heart disease has not
been definitively elucidated, the particle was reported to be more
atherogenic than LDL (39, 40). Although Lp(a) concentrations
are thought to be largely under genetic control, there have been
reports that dietary TFAs can raise (41, 42) and SFAs can lower
(7, 27) Lp(a). In this investigation, Lp(a) concentrations were
significantly lower after the butter diet than after both margarine
diets. This is probably because of the higher concentration of
SFAs in butter than in either margarine. Although TFAs have
been reported to raise Lp(a) (41, 42), no apparent effect of
dietary TFAs on Lp(a) was evident while comparing a margarine
containing no TFAs and high amounts of PUFAs with a mar-
garine made with partially hydrogenated vegetable oil. Our
observation of lower Lp(a) concentration with butter as com-
pared with TFA-M is consistent with that of Almendingen et al
(7), who found that butter consumption resulted in lower con-
centrations of Lp(a) than did consumption of partially hydro-
genated fish or vegetable oils. Although there may be some pos-
itive effect of dietary SFAs in lowering Lp(a) concentrations,
there is much stronger evidence that saturates raise LDL choles-
terol, the major risk factor for cardiovascular disease associated
with diet in most people. Until stronger evidence for effects of
diet on Lp(a) and the risk of heart disease is available, the effects
of dietary fat on LDL cholesterol must remain the major con-
cern.
The hardness of a spread depends directly on its content of
TFAs, SFAs, or both. Margarines vary greatly in their hardness,
with harder margarines containing more SFAs or TFAs.This
study takes a practical approach to comparing butter and tub-
type spreads on an isoenergetic basis within the range of normal
consumption in the United States. Because the spreads differed
considerably in fatty acid profiles, this study does not directly
compare intake of individual fatty acids on an equienergetic
basis. Neither can these results be generalized for all commercial
margarines because they vary widely in fatty acid composition.
Compared with butter, both types of spreads used in this study
774 JUDD ET AL
by guest on July 13, 2011www.ajcn.orgDownloaded from
had lower amounts of SFAs + TFAs, and both produced
improved profiles for the major lipoproteins. In the margarine
designated PUFA-M, the concentration of SFAs was less than the
concentration of SFAs + TFAs in TFA-M; PUFA-M produced the
most desirable blood lipid profile.
Changes in the ratios of total or LDL cholesterol to HDL cho-
lesterol may be better predictors of risk for cardiovascular dis-
ease than are changes in LDL cholesterol alone (43, 44). In this
study, there was an improvement (lowering) in the ratio of total
to HDL cholesterol after consumption of PUFA-M compared
with after consumption of butter that was not observed after
consumption of TFA-M. Thus, margarines with reasonably high
concentrations of PUFAs may offer additional benefits for risk
reduction. In contrast, lowering of HDL by TFAs was reported
(5), but at TFA intakes greatly in excess of those that would be
expected from tub-type soft margarines like those used in this
study. Although we did not observe an effect of TFAs on HDL
cholesterol or on the ratio of total or LDL cholesterol to HDL
cholesterol in this study, such effects at higher TFA concentra-
tions cannot be ruled out. High PUFA concentrations may, how-
ever, be present in margarines containing liquid vegetable oils
even when they are hardened with partially hydrogenated fats
containing TFAs. There was no difference in the ratios after
consumption of PUFA-M compared with after consumption of
TFA-M. However, effects on the ratios resulting from consump-
tion of margarines with high amounts of PUFAs with and with-
out TFAs over the broad range of fatty acid composition in com-
mercial margarines remain to be determined.
Hegsted et al (2) used data from published studies of the
effects of dietary fatty acids and cholesterol on lipoprotein cho-
lesterol to develop an equation for predicting changes in blood
cholesterol concentrations. According to this prediction equa-
tion, decreases in dietary cholesterol and SFAs and increases in
PUFAs of the magnitudes seen in the current investigation can
be predicted to result in changes in plasma total and LDL cho-
lesterol of a magnitude similar to that observed. Predicted and
observed changes, respectively, in total and LDL cholesterol,
based on changes in dietary fatty acid and cholesterol intakes
for the diets fed in this study are, in mmol/L (mg/dL): total cho-
lesterol for butter compared with PUFA-M, 0.31 and 0.28 (12
and 11), and LDL cholesterol for butter compared with PUFA-
M, 0.23 and 0.23 (9 and 9); total cholesterol for butter compared
with TFA-M, 0.26 and 0.18 (10 and 7), and LDL cholesterol for
butter compared with TFA-M, 0.21 and 0.18 (8 and 7). Changes
predicted because of changes in dietary fatty acid intake with
PUFA-M compared with TFA-M when using the Hegsted equa-
tion (2) were not as close to observed changes as they were for
the change from butter to margarine. Predicted compared with
observed lipid concentrations for PUFA-M compared with TFA-
M were 0.04 and 0.1 mmol/L (1.5 and 4 mg/dL) for total cho-
lesterol, and 0.05 and 0.08 mmol/L (2 and 3 mg/dL) for LDL
cholesterol. Dietary TFAs were shown to be hypercholes-
terolemic compared with oleic acid (5, 8). Therefore, when
TFAs are considered to be hypercholesterolemic and are added
to SFAs, as reported by Lichtenstein et al (45), the prediction is
closer to the observed change for PUFA-M compared with TFA-
M, ie, 0.13 compared with 0.10 mmol/L (5 compared with 4
mg/dL) for total cholesterol, and 0.08 compared with 0.08
mmol/L (3 compared with 3 mg/dL) for LDL cholesterol.
Most margarines are rich sources of tocopherols, both natu-
rally occurring tocopherols in vegetable oils and tocopherols
added to margarines as antioxidants during production. In this
study, plasma a-tocopherol concentration but not g-tocopherol
concentration appeared to reflect the content of the table
spreads. PUFA-M contained the greatest amount of a-toco-
pherol and TFA-M contained the greatest amount of g-toco-
pherol. Butter and TFA-M contained similar amounts of a-toco-
pherol and plasma concentrations were similar as well.
PUFA-M contained more a-tocopherol than did either butter or
TFA-M, and plasma a-tocopherol concentrations were highest
after consumption of PUFA-M. For g-tocopherol, however,
plasma concentrations were highest after consumption of TFA-
M, which contained more g-tocopherol than did the other 2
spreads. Although PUFA-M contained <3 times more g-toco-
pherol than did butter, plasma g-tocopherol concentration was
higher after butter was consumed. There was a significant car-
ryover effect for g-tocopherol but not for a-tocopherol, sug-
gesting that the length of the dietary periods for this study might
be shorter than the plasma half-life of tocopherols.
There was no difference in the effect of the 3 spreads on
plasma lipid hydroperoxide concentrations even though the
difference in lipid hydroperoxide concentrations of the
spreads themselves was about 10-fold. Thus, there appear to
be sufficient circulating amounts of tocopherols from all
spreads to prevent changes in plasma lipid hydroperoxide
concentrations.
Several studies reported plasma lipid responses to butter,
margarine, or vegetable oils added to controlled diets (7, 12,
23, 46–48). However, in most of these studies, the amounts of
butter, margarine, or vegetable oil were unrealistically higher
than what might be reasonably consumed as part of a self-
selected diet, which makes it difficult to extrapolate the results
to what might be expected from a typical diet. Although infer-
ences may be drawn from such studies of expected effects of
margarine, the amounts of oils and fatty acid profiles of the
oils may not be attainable in actual margarine-type products
having acceptable physical and sensory properties.
In conclusion, we showed that within the range of con-
sumption of table spreads in a typical American diet, the fatty
acid profile of a margarine can be controlled in a product with
excellent physical and sensory properties. When such a mar-
garine is consumed in place of butter, and presumably in place
of other fats high in SFAs, appreciable improvement in blood
lipid profiles of the major lipoproteins can be expected for
most people. Because table spreads represent a major portion
of the visible fat in the diets of many countries, including the
United States, United Kingdom, and Netherlands, selection of
products with desirable fatty acid profiles can be an easy
means to begin consuming a more healthful diet and reducing
the risk of cardiovascular disease.
We thank Evelyn Lashley and the staff of the Beltsville Human Study Facil-
ity, BHNRC, for preparing and feeding the controlled diets; the staff of the
Lipid Research Clinic, The George Washington University, for lipid and
lipoprotein analyses; Joseph Sampugna, University of Maryland, for fatty acid
analyses; Edward Emken, National Center for Agricultural Utilization
Research, Agricultural Research Service, US Department of Agriculture, for
analysis of cis and trans fatty acid positional isomers; the staff of the Diet and
Human Performance Laboratory, BHNRC, for assistance in performing the
study; and Benjamin Caballero, Johns Hopkins School of Public Health, for
medical supervision of the study. We are indebted to William Stone, East Ten-
nessee State University, for tocopherol and lipid hydroperoxide measurements;
and James Otvos, North Carolina State University, for determinations of LDL
TABLE SPREADS AND PLASMA LIPIDS 775
by guest on July 13, 2011www.ajcn.orgDownloaded from
particle size. We thank Lipton, Baltimore, for preparing the study margarines.
REFERENCES
1. Grundy SM, Denke MA. Dietary influences on serum lipids and
lipoproteins. J Lipid Res 1990;31:1149–72.
2. Hegsted DM, Ausman LM, Johnson JA, Dallal GE. Dietary fat and
serum lipids: an evaluation of the experimental data. Am J Clin Nutr
1993;57:875–83.
3. Clarke R, Frost C, Collins R, Appleby P, Peto R. Dietary lipids and
blood cholesterol: quantitative meta-analysis of metabolic ward
studies. Br Med J 1997;314:112–7.
4. National Cholesterol Education Program. Second Report of the
Expert Panel on Detection, Evaluation, and Treatment of High
Blood Cholesterol in Adults (Adult Treatment Panel II). II. Dietary
therapy and physical activity. Circulation 1994;89:1364–404.
5. Mensink RP, Katan MB. Effect of dietary trans fatty acids on high-
density and low-density lipoprotein cholesterol levels in healthy
subjects. N Engl J Med 1990;323:439–45.
6. Zock PL, Katan MB. Hydrogenation alternatives: effects of trans
fatty acids and stearic acid versus linoleic acid on serum lipids and
lipoproteins in humans. J Lipid Res 1992;33:399–410.
7. Almendingen K, Jordal O, Kierulf P, Sandstad B, Pedersen JI.
Effects of partially hydrogenated fish oil, partially hydrogenated
soybean oil, and butter on serum lipoproteins and Lp(a) in men.
J Lipid Res 1995;36:1370–84.
8. Judd JT, Clevidence BA, Muesing RA, Wittes J, Sunkin ME, Pod-
czasy JJ. Dietary trans fatty acids: effects on plasma lipids and
lipoproteins of healthy men and women. Am J Clin Nutr
1994;59:861–8.
9. Willett WC, Stampfer MJ, Manson JE, et al. Intake of trans fatty
acids and risk of coronary heart disease among women. Lancet
1993;341:581–5.
10. Ascherio A, Hennekens CH, Buring JE, Master C, Stampfer MJ,
Willett WC. Trans fatty acid intake and risk of myocardial infarc-
tion. Circulation 1994;89:94–101.
11. Aro A, Kardinaal AFM, Salminen I, et al. Adipose tissue isomeric
trans fatty acids and risk of myocardial infarction in nine countries:
the EURAMIC study. Lancet 1995;345:273–8.
12. Zock RL, Katan MB. Butter, margarine and serum lipoproteins.
Atherosclerosis 1997;131:7–16.
13. Hinkelman K, Kempthorne O. Design and analysis of experiments.
Vol 1. New York: John Wiley and Sons, Inc, 1994.
14. Society of Actuaries and Association of Life Insurance Medical
Directors of America. 1979 Build study. Weight of insured persons
in the United States associated with lowest mortality. Philadelphia:
Association of Life Insurance Medical Directors of America, 1980.
15. US Department of Agriculture, Human Nutrition Information Ser-
vice. Composition of foods. Agriculture handbook no. 8, sections
1–22. Washington, DC: US Government Printing Office,
1976–1990.
16. Pao EM, Fleming KH, Guenther PM, Mickle SJ. Foods commonly
eaten by individuals: amount per day and per eating occasion.
Hyattsville, MD: Consumer Nutrition Center, Human Nutrition
Information Service, US Department of Agriculture, 1982. (Home
Economics Research Report no. 44.)
17. United States Department of Agriculture, National Agricultural Sta-
tistics Service. Agricultural statistics 1995–96. Washington, DC: US
Government Printing Office, 1996.
18. Hainline A, Karon J, Lippel K, eds. Manual of laboratory opera-
tions, Lipid Research Clinics Program. Lipid and lipoprotein analy-
sis. 2nd ed. Bethesda, MD: National Heart Lung and Blood Insti-
tute, NIH, DHHS, 1982.
19. Gidez LI, Miller GJ, Burstein M, Slagle S, Eder HA. Separation and
quantitation of subclasses of human plasma high-density lipopro-
teins by a simple precipitation procedure. J Lipid Res
1982;23:1206–23.
20. Friedewald WT, Levy RI, Fredrickson DS. Estimation of the con-
centration of low-density lipoprotein cholesterol in plasma without
use of the preparative ultracentrifuge. Clin Chem 1972;18:499–502.
21. Bostom AG, Gagnon DR, Cupples LA, et al. A prospective investi-
gation of elevated lipoprotein(a) detected by electrophoresis and
cardiovascular disease in women: The Framingham Heart Study.
Circulation 1994;90:1688–95.
22. Schaefer EJ, Lamon-Fava S, Jenner JL, et al. Lipoprotein(a) levels
and risk of coronary heart disease in men. The Lipid Research Clin-
ics Coronary Primary Prevention Trial. JAMA 1994;271:999–1003.
23. Lichtenstein AH,Ausman LM, Carrasco W, Jenner JL, Ordovas JM,
Schaefer EJ. Hydrogenation impairs the hypolipidemic effect of
corn oil in humans. Hydrogenation, trans fatty acids, and plasma
lipids. Arterioscler Thromb 1993;13:154–61.
24. Otvos JD, Jeyarajah EJ, Bennett DW, Krauss RM. Development of
a proton nuclear magnetic resonance spectroscopic method for
determining plasma lipoprotein concentrations and subspecies dis-
tributions from a single, rapid measurement. Clin Chem
1992;38:1632–8.
25. Littell RC, Milliken GA, Stroup WW, Wolfinger RD. SAS system
for mixed models. Cary, NC: SAS Institute, Inc, 1996.
26. Cleveland LE, Goldman JD, Borrud LG. Data tables: results from
USDA’s 1994 Continuing Survey of Food Intakes by Individuals
and 1994 Diet and Health Knowledge Survey. Riverdale, MD:Agri-
cultural Research Service, US Department of Agriculture, 1996.
27. Clevidence BA, Judd JT, Schaefer EJ, et al. Plasma lipoprotein (a)
levels in men and women consuming diets enriched in saturated,
cis- or trans-monounsaturated fatty acids. Arterioscler Thromb Vasc
Biol 1997;17:1657–61.
28. Voorlichtingsbureau voor de Voeding. (National Education Board
for Nutrition.) Zo eet Nederland 1992. (Food consumption of the
Netherlands in 1992.) Results of the Food Consumption Survey. The
Hague: National Education Board for Nutrition, 1993.
29. Gregory J, Foster K, Tyler H, Wiseman M. Office of Population
Census and Surveys. Social Survey Division. The Dietary and Nutri-
tion Survey of British Adults. London: Her Majesty’s Stationery
Office, 1990.
30. Austin MA, Breslow JL, Hennekens CH, Buring JE, Willett WC,
Krauss RM. Low-density lipoprotein subclass patterns and risk of
myocardial infarction. JAMA 1988;260:1917–21.
31. Griffin BA, Freeman DJ, Tait GW, et al. Role of plasma triacylglyc-
erol in the regulation of plasma low density lipoprotein (LDL) sub-
fractions: relative contribution of small, dense LDL to coronary
heart disease risk. Atherosclerosis 1994;106:241–53.
32. Campos H, Genest JJ, Blijlevens E, et al. Low density particle size
and coronary artery disease. Arterioscler Thromb 1992;12:187–95.
33. Gardner CD, Fortman SP, Krauss RM. Association of small low-
density lipoprotein particles with the incidence of coronary artery
disease in men and women. JAMA 1996;276:875–81.
34. Stampfer MJ, Krauss RM, Ma J, et al. A prospective study of tria-
cylglycerol level, low-density lipoprotein particle diameter, and risk
of myocardial infarction. JAMA 1996;276:882–8.
35. Lamarche B, Tchernof A, Moorjani S, et al. Small, dense low-den-
sity lipoprotein particles as a predictor of the risk of ischemic heart
disease in men. Prospective results from the Québec Cardiovascular
Study. Circulation 1997;95:69–75.
36. Wardlaw GM, Snook JT. Effect of diets high in butter, corn oil, or
high-oleic acid sunflower oil on serum lipids and apolipoproteins in
men. Am J Clin Nutr 1990;51:815–21.
37. Clevidence BA, Judd JT, Schatzkin A, et al. Plasma lipid and
lipoprotein concentrations of men consuming a low-fat, high-fiber
diet. Am J Clin Nutr 1992;55:689–94.
38. National Cholesterol Education Program, Second Report of the
Expert Panel on Detection, Evaluation, and Treatment of High
Blood Cholesterol in Adults (Adult Treatment Panel II). Circulation
1994;89:1329–445.
39. Hoefler G, Harnoncourt F, Paschke E, Mirtl W, Pfeiffer KH, Kost-
ner GM. Lipoprotein Lp(a): a risk factor for myocardial infarction.
776 JUDD ET AL
by guest on July 13, 2011www.ajcn.orgDownloaded from
TABLE SPREADS AND PLASMA LIPIDS 777
Arteriosclerosis 1988;8:398–401.
40. Rosengren A, Wilhelmsen L, Eriksson E, Risberg B, Wedel H.
Lipoprotein (a) and coronary heart disease: a prospective case-con-
trol study in a general population sample of middle aged men. BMJ
1990;301:1248–51.
41. Mensink RP, Zock PL, Katan MB, Hornstra G. Effect of dietary cis
and trans fatty acids on serum lipoprotein(a) levels in humans.
J Lipid Res 1992;33:1493–501.
42. Nestel P, Noakes M, Belling B, et al. Plasma lipoprotein lipid and
Lp(a) changes with substitution of elaidic acid for oleic acid in the
diet. J Lipid Res 1992;33:1029–36.
43. Kinosian B, Glick H, Garland G. Cholesterol and coronary heart
disease: predicting risks by levels and ratios. Ann Intern Med
1994;121:641–7.
44. Kinosian B, Glick H, Preiss L, Puder K. Cholesterol and coronary
heart disease: predicting risks in men by changes in levels and
ratios. J Invest Med 1995;43:443–50.
45. Lichtenstein AH, Ausman, LM, McNamara JR, Schaefer EJ. Trans
and saturated fatty acid content of dietary fat effects plasma lipid
and lipoprotein concentrations. Circulation 1996;94(suppl):I-97
(abstr no. 0559).
46. Chisholm A, Mann J, Sutherland W, Duncan A, Skeaff M, Frampton
C. Effect on lipoprotein profile of replacing butter with margarine in
a low fat diet: randomized crossover study with hypercholestero-
laemic subjects. BMJ 1996;312:931–4.
47. Wood R, Kubena K,Tseng S, Martin G, Crook R. Effect of palm oil,
margarine, butter, and sunflower oil on the serum lipids and lipopro-
teins of normocholesterolemic middle-aged men. J Nutr Biochem
1993;4:286–97.
48. Wood R, Kubena K, O’Brien B,Tseng S, Martin G. Effect of butter,
mono- and polyunsaturated fatty acid-enriched butter, trans fatty
by guest on July 13, 2011www.ajcn.orgDownloaded from
