Research article Related Commentary, page 1089

Consuming fructose-sweetened, not glucose-

sweetened, beverages increases visceral
adiposity and lipids and decreases insulin
sensitivity in overweight/obese humans
Kimber L. Stanhope,1,2 Jean Marc Schwarz,3,4 Nancy L. Keim,5 Steven C. Griffen,6
Andrew A. Bremer,7 James L. Graham,1,2 Bonnie Hatcher,2 Chad L. Cox,2 Artem Dyachenko,3
Wei Zhang,6 John P. McGahan,8 Anthony Seibert,8 Ronald M. Krauss,9 Sally Chiu,9
Ernst J. Schaefer,10 Masumi Ai,10 Seiko Otokozawa,10 Katsuyuki Nakajima,10,11 Takamitsu Nakano,11
Carine Beysen,12 Marc K. Hellerstein,12,13 Lars Berglund,6,14 and Peter J. Havel1,2
1Department of Molecular Biosciences, School of Veterinary Medicine, and 2Department of Nutrition, UCD, Davis, California, USA.
3Collegeof Osteopathic Medicine, Touro University, Vallejo, California, USA. 4UCSF, San Francisco, California, USA.
5United States Department of Agriculture, Western Human Nutrition Research Center, Davis, California, USA. 6Department of Internal Medicine and
7Department of Pediatrics, School of Medicine, UCD, Sacramento, California, USA. 8Department of Radiology, UCD Medical Center, Sacramento,

California, USA. 9Children’s Hospital Oakland Research Institute, Oakland, California, USA. 10Lipid Metabolism Laboratory,
Jean Mayer United States Department of Agriculture Human Nutrition Research Center on Aging at Tufts University, and
Tufts University School of Medicine, Boston, Massachusetts, USA. 11Diagnostic Division, Otsuka Pharmaceutical Co., Tokyo, Japan.
12KineMed, Emeryville, California, USA. 13Nutritional Sciences and Toxicology, University of California, Berkeley, California, USA.
14Veterans Affairs Northern California Health Care System, Sacramento, California, USA.

Studies in animals have documented that, compared with glucose, dietary fructose induces dyslipidemia
and insulin resistance. To assess the relative effects of these dietary sugars during sustained consumption in
humans, overweight and obese subjects consumed glucose- or fructose-sweetened beverages providing 25% of
energy requirements for 10 weeks. Although both groups exhibited similar weight gain during the interven-
tion, visceral adipose volume was significantly increased only in subjects consuming fructose. Fasting plasma
triglyceride concentrations increased by approximately 10% during 10 weeks of glucose consumption but not
after fructose consumption. In contrast, hepatic de novo lipogenesis (DNL) and the 23-hour postprandial
triglyceride AUC were increased specifically during fructose consumption. Similarly, markers of altered lipid
metabolism and lipoprotein remodeling, including fasting apoB, LDL, small dense LDL, oxidized LDL, and
postprandial concentrations of remnant-like particle–triglyceride and –cholesterol significantly increased
during fructose but not glucose consumption. In addition, fasting plasma glucose and insulin levels increased
and insulin sensitivity decreased in subjects consuming fructose but not in those consuming glucose. These
data suggest that dietary fructose specifically increases DNL, promotes dyslipidemia, decreases insulin sensi-
tivity, and increases visceral adiposity in overweight/obese adults.

Introduction
Authorship note: Kimber L. Stanhope and Jean Marc Schwarz are co–first authors. Studies investigating the effects of fructose consumption in humans
Conflict of interest: M. Ai was supported by a research fellowship from Denka and animals have been comprehensively reviewed (1–4), and while
Seiken Co. L. Berglund is a shareholder of Pfizer. C. Beysen owns stock in and is an strong evidence exists that consumption of diets high in fructose
employee of KineMed. S.C. Griffen receives income from Bristol-Myers Squibb. P.J. results in increased de novo lipogenesis (DNL), dyslipidemia, insu-
Havel has a research contract with Novo Nordisk to test a proprietary compound for
efficacy in delaying or preventing onset of diabetes in what we believe is a novel rat lin resistance, and obesity in animals, direct experimental evidence
model of type 2 diabetes. M.K. Hellerstein owns stock in and has received consulting that consumption of fructose promotes DNL, dyslipidemia, insulin
income from KineMed. R.M. Krauss has received consulting income from Merck, resistance, glucose intolerance, and obesity in humans is lacking.
Merck-Schering Plough, and Isis Pharmaceuticals and has received grant support
from Merck, Merck-Schering Plough, Sanofi-Aventis, and Metabolex. K. Nakajima has
Thus, we have investigated and compared the biological effects of
received consulting income from Denka Seiken Co. and Otsuka Pharmaceutical Co. S. the 2 major simple sugars in the diet, glucose and fructose, on BW
Otokozawa was supported by a research fellowship from Kyowa Medex Co. J.A. Seibert and regional fat deposition and on indices of lipid and carbohydrate
is a consulting physicist for the American College of Radiology Imaging Network,
National Lung Screening Trial.
metabolism in older, overweight and obese men and women.
Nonstandard abbreviations used: CCRC, Clinical and Translational Science Cen-
We sought to answer the following questions: (a) Does consump-
ter’s Clinical Research Center; DNL, de novo lipogenesis; GLM, general linear model; tion of fructose with an ad libitum diet promote greater BW gain
HFCS, high-fructose corn syrup; LPL, lipoprotein lipase; MSRF, metabolic syndrome and have differential effects on regional adipose deposition and
risk factors; OGTT, oral glucose tolerance test; PROC MIXED, mixed procedures; adipose gene expression compared with consumption of glucose
RLP, remnant-like particle lipoprotein; RLP-C, RLP-cholesterol; RM, repeated mea-
sures; SAT, subcutaneous adipose tissue; sdLDL, small dense LDL; TG, triglyceride; with an ad libitum diet? (b) Does consumption of fructose induce
VAT, visceral adipose tissue. dyslipidemia compared with consumption of glucose? (c) Is fruc-
Citation for this article: J. Clin. Invest. 119:1322–1334 (2009). doi:10.1172/JCI37385. tose-induced hypertriglyceridemia the result of increased rates

1322 The Journal of Clinical Investigation    http://www.jci.org    Volume 119    Number 5    May 2009
 research article

of hepatic DNL and/or decreased triglyceride (TG) clearance? (d) anthropomorphic characteristics or in any of the measured met-
Does consumption of fructose decrease glucose tolerance and abolic parameters (Table 2).
insulin sensitivity? (e) Are there differences between the responses Outpatient food intake, BW and composition, adipose tissue gene expres-
of older men and postmenopausal women to dietary fructose? sion, and blood pressure. During 24-hour food-intake recall interviews
Consumption of fructose-sweetened but not glucose-sweetened conducted on 6 outpatient days, both groups of subjects reported
beverages for 10 weeks increased DNL, promoted dyslipidemia, consuming significantly more energy than their calculated energy
decreased insulin sensitivity, and increased visceral adiposity in requirements. There were no significant differences between men
overweight/obese adults. and women or between subjects consuming glucose and subjects
consuming fructose in fat, sugar, or alcohol intake as a percentage
Results of energy intake or in the amount of energy consumed as a percent-
During the baseline phase of the study, subjects resided in the age of calculated energy requirements (Supplemental Table 2).
UCD Clinical and Translational Science Center’s Clinical Research The changes in anthropomorphic outcomes are summarized in
Center (CCRC) for 2 weeks and consumed an energy-balanced, Table 3, and detailed analyses are presented in Supplemental Table
high–complex carbohydrate (55%) diet (Supplemental Table 1; sup- 3. Despite comparable weight gain, there were differential effects
plemental material available online with this article; doi:10.1172/ of glucose and fructose on regional adipose deposition and gene
JCI37385DS1). Procedures conducted during the baseline CCRC expression. BW was stable during the 2-week inpatient periods
visit included a 24-hour serial blood collection, a 26-hour stable at both the beginning and end of the study. However, during the
isotope infusion for determination of fractional DNL, fasting and 8-week outpatient intervention period, when the subjects con-
postprandial postheparin blood sampling, an oral glucose toler- sumed 25% of daily energy requirement as glucose- or fructose-
ance test (OGTT) and disposal test, a gluteal adipose biopsy, and sweetened beverages along with ad libitum self-selected diets, both
a CT scan of the abdomen. Subjects then began an 8-week outpa- groups of subjects exhibited significant increases of BW (Figure
tient intervention and consumed either fructose- (n = 17) or glu- 1A), fat mass, and waist circumference. Total and visceral adipose
cose-sweetened (n = 15) beverages at 25% of energy requirements tissue (VAT) volumes were not significantly changed in subjects
with self-selected ad libitum diets. The subjects returned to the consuming glucose; however, subcutaneous adipose tissue (SAT)
CCRC after 2 outpatient weeks for 2 days and then again for the volume was significantly increased. In contrast, both total abdom-
final 2 weeks of the intervention for inpatient metabolic studies, inal fat and VAT volume were significantly increased in subjects
during which the glucose- or fructose-sweetened beverages were consuming fructose (Figure 1B).
consumed as part of an energy-balanced diet. Blood was collected SAT from the gluteal region was biopsied at 0 weeks and 10
over four 24-hour periods, during baseline and after 2, 8, and 10 weeks and analyzed for the expression of lipogenic and other
weeks of intervention. The study design is outlined in Table 1. genes (Supplemental Table 4). The percentage changes of gene
Baseline characteristics and parameters. There were no signifi- expression at 10 weeks compared with baseline (0 weeks) were
cant differences between the 2 experimental groups in baseline greater in subjects consuming glucose than in those consuming

Table 1
Twelve-week inpatient/outpatient, procedure, and diet schedule

Study week Monday Tuesday Wednesday Thursday Friday Saturday Sunday

Week 2 Baseline InpatientA InpatientA InpatientA InpatientA InpatientA InpatientA InpatientA
DXA Postprandial OGTT Glucose
postheparin disposal test
blood draw
Week 1 Baseline InpatientA InpatientB InpatientA InpatientA InpatientA InpatientA CheckoutC
Gluteal adipose 26-hour stable CT Scan 24-hour Fasting postheparin
biopsy isotope infusion blood collection blood draw
Week 1–2 Intervention OutpatientC OutpatientC OutpatientC OutpatientC OutpatientC OutpatientC OutpatientC
Week 3 Intervention Outpatient C Inpatient
D Inpatient
E OutpatientB OutpatientB OutpatientB OutpatientB
Ad lib buffet 24-hour
blood collection
Week 4–8 Intervention OutpatientC OutpatientC OutpatientC OutpatientC OutpatientC OutpatientC OutpatientC
Week 9 Intervention InpatientD InpatientE InpatientE InpatientE InpatientE InpatientE InpatientE
Ad lib buffet 24-hour DXA Postprandial Oral glucose
blood collection postheparin tolerance and
blood draw disposal test
Week 10 Intervention InpatientE InpatientF InpatientE InpatientE InpatientE InpatientE Checkout
Gluteal adipose 26-hour stable CT scan 24-hour Fasting postheparin
biopsy isotope infusion blood collection blood draw
AEnergy-balanced diet: 55% of energy requirement complex carbohydrate; 30% fat; 15% protein. BSteady-state energy-balanced diet: 55% of energy
requirement complex carbohydrate; 30% fat; 15% protein. CAd libitum usual diet plus 25% of energy requirement as sugar-sweetened beverage. DAd
libitum food-intake trial plus 25% of energy requirement as sugar-sweetened beverage. EEnergy-balanced diet: 25% sugar-sweetened beverage; 30%
complex carbohydrate; 30% fat; 15% protein. FSteady-state energy-balanced diet: 25% sugar beverage; 30% complex carbohydrate; 30% fat; 15% protein.
DXA, dual energy x-ray absorptiometry.

The Journal of Clinical Investigation    http://www.jci.org    Volume 119    Number 5    May 2009 1323
research article

Table 2 ing glucose but were markedly increased in

Baseline anthropomorphic and metabolic parameters subjects consuming fructose (Figure 2, A and
B). Fasting (Figure 3, A and B) and postpran-
Glucose Fructose dial apoB, the apoB/apoA1 ratio, and total and
Parameter Male Female Male Female LDL cholesterol were also unchanged during
(n = 7) (n = 8) (n = 9) (n = 8) consumption of glucose and increased dur-
Age (yr) 54 ± 3 56 ± 2 52 ± 4 53 ± 2 ing consumption of fructose. In both groups
Weight (kg) 88.4 ± 2.9 84.0 ± 4.5 89.3 ± 2.9 81.9 ± 4.2 of subjects, plasma HDL concentrations were
BMI (kg/m2) 29.3 ± 1.1 29.4 ± 1.3 28.4 ± 0.7 30.3 ± 1.0 unchanged at 10 weeks but increased at the 2-
Waist circumference (cm) 98.9 ± 2.6 91.0 ± 4.0 97.3 ± 3.3 91.8 ± 4.4 and 8-week time points.
Body fat (%) 29.4 ± 1.1 43.2 ± 1.5 28.5 ± 1.3 39.6 ± 2.2 In subjects consuming glucose, fasting
TG (mg/dl) 148 ± 31 145 ± 23 131 ± 21 159 ± 30 small dense LDL (sdLDL) concentrations
Total cholesterol (mg/dl) 179 ± 14 193 ± 10 176 ± 6 198 ± 15
(Figure 3, C and D) initially decreased at 2
HDL (mg/dl) 36 ± 3 41 ± 3 39 ± 4 41 ± 3
LDL (mg/dl) 124 ± 5 123 ± 11 107 ± 7 124 ± 15
weeks and were not different from baseline at
Glucose (mg/dl) 89 ± 2 89 ± 3 88 ± 1 90 ± 1 10 weeks. In contrast, fasting sdLDL concen-
Insulin (μU/ml) 14.3 ± 3.2 15.6 ± 2.9 16.3 ± 2.5 12.0 ± 1.6 trations increased progressively in subjects
consuming fructose. sdLDL was the lipid
GLM 2-factor ANOVA (type of sugar and sexual phenotype). There were no significant differ- parameter most affected by preexisting met-
ences among groups. Data represent mean ± SEM.
abolic syndrome risk factors (MSRF), with
increases during fructose consumption more
than 2-fold greater in subjects with 3 MSRF
fructose for stearoyl-CoA desaturase, fatty acid desaturase 1, and than in subjects with 0 to 2 MSRF (Supplemental Table 7). Fasting
fatty acid desaturase 2. oxidized LDL concentrations did not change in subjects consum-
All subjects had normal blood pressure measurements at base- ing glucose but increased in subjects consuming fructose.
line, and blood pressure values did not change during the consump- Fasting plasma remnant-like particle lipoprotein–TG (RLP-TG)
tion of either fructose or glucose over the course of the 10-week and RLP-cholesterol (RLP-C) concentrations were unaffected by
intervention period (Table 3; Supplemental Table 5). consumption of glucose or fructose (data not shown). In subjects
Lipid and lipoprotein concentrations, fractional hepatic DNL, and lipo- consuming glucose, postprandial concentrations of RLP-TG (Fig-
protein lipase activity. Plasma concentrations of lipid and lipoprotein ure 3, E and F) were unchanged; however RLP-C concentrations
parameters measured at 0 weeks, 2 weeks, 8 weeks, and 10 weeks were increased at 8 weeks. During consumption of fructose, post-
with detailed analyses are presented in Supplemental Table 6. prandial concentrations of both RLP-TG and RLP-C were increased.
In general, plasma lipid and lipoprotein concentrations increased FFA exposure over 24 hours was increased in subjects consuming
markedly during fructose consumption and were unchanged during glucose but unchanged in subjects consuming fructose.
glucose consumption (Table 4). In exception, fasting TG concentra- Increased DNL contributed to the increases of postprandial
tions increased in subjects consuming glucose but were unchanged TG during fructose consumption. Fractional hepatic DNL was
in subjects consuming fructose (2, 8, and 10 weeks vs. 0 weeks: unchanged during glucose consumption, both in the fasting
+1.0% ± 5.5%, +1.0% ± 5.0% and +3.9% ± 5.5%; P = 0.92). There was (8.8% ± 1.8% vs. 9.5% ± 1.8%; P = 0.47) and postprandial states
marked variability in the fasting TG responses to fructose consump- (13.4% ± 2.8% vs. 14.2% ± 1.7%; P = 0.31). Fasting DNL was unaf-
tion both within groups and within the individual subject. The fected during fructose consumption (9.9% ± 1.3% vs. 8.3% ± 0.9%;
mean SD of the percentage changes at 2 weeks, 8 weeks, and 10 weeks P = 0.25), but postprandial DNL was significantly increased (11.4% ± 1.3%
compared with 0 weeks in each subject was 13.4% ± 1.5%. In contrast vs. 16.9% ± 1.4%; P = 0.021) (Table 4). The 16-hour AUC for frac-
to fasting TG, indices of postprandial TG — 23-hour AUC, TG expo- tional DNL was not increased compared with baseline in subjects
sure, postprandial TG peak — did not increase in subjects consum- consuming glucose (54% ± 17% vs. 60% ± 8% × 16 h; P = 0.69)

Table 3
Baseline values and percentage changes in body composition and blood pressure after consumption of glucose- or fructose-sweetened bev-
erages for 10 weeks

Outcome variable Glucose Glucose Fructose Fructose

(0 weeks) (% change 10 weeks) (0 weeks) (% change 10 weeks)
Systolic BP (mmHg) 122 ± 2 0 ± 2 120 ± 2 +1 ± 1
Diastolic BP (mmHg) 77 ± 1 –2 ± 1 76 ± 1 –1 ± 1
BW (kg) 85.9 ± 2.7 +1.8 ± 0.5A 85.7 ± 2.6 +1.4 ± 0.3B
Total body fat (kg) 30.7 ± 2.2 +3.2 ± 0.6B 28.9 ± 2.2 +2.8 ± 1.0A
Waist circumference (cm) 94.6 ± 2.6 +1.7 ± 0.6C 94.7 ± 2.7 +1.9 ± 0.4B
Total abdominal fat (cc) 765 ± 57 +4.8 ± 2.1 683 ± 55 +8.6 ± 3.0C
Extraabdominal fat (cc) 522 ± 59 +4.6 ± 1.4C 476 ± 43 +7.3 ± 4.0
Intraabdominal fat (cc) 243 ± 21 +3.2 ± 4.4 207 ± 21 +14.0 ± 5.5A
AP < 0.01; BP < 0.001; CP < 0.05, paired Student’s t test, 10 weeks vs. 0 weeks. Data represent mean ± SEM.

1324 The Journal of Clinical Investigation    http://www.jci.org    Volume 119    Number 5    May 2009
 research article

Figure 1
Changes of BW and abdominal fat. (A) Changes of BW during the 2-week inpatient baseline, 8-week outpatient intervention, and 2-week inpa-
tient intervention periods. **P < 0.01; ****P < 0.0001, day 56 outpatient:intervention vs. day 1 outpatient:intervention; paired Student’s t test.
Glucose, n = 15; fructose, n = 17. (B) Changes of total abdominal adipose tissue, SAT, and VAT volume in subjects after consuming glucose- or
fructose-sweetened beverages for 10 weeks. *P < 0.05; **P < 0.01, 10 weeks vs. 0 weeks; paired Student’s t test. Glucose, n = 14; fructose,
n = 17. Data represent mean ± SEM.

but was significantly increased in subjects consuming fructose TG exposure, postprandial TG peak, and postprandial RLP-C in
(21% ± 9% vs. 104% ± 19% × 16 h; P = 0.0043). The increase of the men compared with women (Supplemental Table 10). There were
16-hour AUC for fractional DNL during fructose consumption no significant differences in the effects of fructose on indices of
was significantly larger than that during glucose consumption glucose tolerance/insulin sensitivity between men and women
(83% ± 22% vs. 7% ± 14% × 16 h; P = 0.016) (Figure 4). (Supplemental Table 11). However, overall the changes in insulin
Reduced TG clearance may also contribute to increases of postpran- sensitivity were different between men and women (P = 0.033; Sup-
dial TG in subjects consuming fructose. Postprandial postheparin plemental Table 9), with women exhibiting significantly greater
lipoprotein lipase (LPL) activity tended to increase after 10 weeks of decreases of insulin sensitivity in response to sugar consumption
glucose consumption and to decrease after 10 weeks of fructose con- than men. The insulin sensitivity index decreased by 10.2% ± 12.1%
sumption, and the overall difference between the sugars was signifi- in women consuming glucose but increased by 12.5% ± 12.6% in
cant (P = 0.041). Fasting postheparin LPL activity was not significantly men consuming glucose. The insulin sensitivity index decreased by
affected by consumption of either glucose or fructose (Table 4). 23.6% ± 4.4% in women consuming fructose and by 11.7% ± 5.6% in
Plasma glucose, plasma insulin, and insulin sensitivity. Indices of insu- men consuming fructose.
lin sensitivity/glucose tolerance at the measured time points with Effects of energy intake during the previous day. Subjects consumed sig-
effects of sugar analyses are presented in Supplemental Table 8. In nificantly more energy ad libitum on the days prior to the 2-week and
general, insulin sensitivity and glucose tolerance were not affect- 8-week 24-hour serial blood collections than during the energy-bal-
ed by the consumption of glucose but were decreased during the anced feeding that preceded the 0-week and 10-week 24-hour serial
consumption of fructose (Table 5). Fasting glucose concentra- blood collections (Supplemental Table 12). The previous day’s ener-
tions decreased in subjects consuming glucose but increased in gy intakes were included in the mixed procedures (PROC MIXED)
subjects consuming fructose. Fasting insulin concentrations were repeated measures (RM) ANOVA model as a time-level covariable;
unchanged during glucose consumption but were increased during therefore, the contribution and significance of the effect of the previ-
consumption of fructose beverages. Glucose excursions, as assessed ous day’s energy intake on the variation of the outcome response can
by the 3-hour AUC, increased in both groups of subjects during the be ascertained by the F statistic and P value of the covariable (Supple-
OGTT (Figure 5, A and B). Insulin excursions were unchanged in mental Table 13). Within subjects consuming fructose, postprandial
subjects consuming glucose but increased in subjects consuming apoB was the outcome most significantly affected by energy intake
fructose (Figure 5, C and D). The insulin sensitivity index, assessed during the previous day; TG exposure, fasting apoB, and postpran-
by the deuterated glucose disposal (5), was unchanged in subjects dial RLP-C were significantly affected as well.
consuming glucose but decreased by 17% in subjects consuming
fructose (Figure 5E). The magnitude of the changes of indices of Discussion
insulin sensitivity during fructose consumption were not signifi- BW and body fat. Given the comparable weight gain in the 2 groups
cantly affected by the number of MSRF (Supplemental Table 9). of subjects, the differences in intraabdominal fat gain and in the
Effect of sexual phenotype. The total and percentage increases of fat gene expression of lipogenic enzymes from subcutaneous adipose
mass (men: +4.4% ± 0.8%; women: +1.5% ± 0.7%; P = 0.020) and intraab- biopsies suggest that fructose consumption may specifically pro-
dominal fat volume (men: +18.1% ± 5.1%; women: –0.6% ± 4.4%; mote lipid deposition in VAT, particularly in men, whereas glucose
P = 0.049) were greater in men than in women. Men consuming consumption appears to favor SAT deposition.
fructose also had larger increases of intraabdominal fat compared Dyslipidemia. In agreement with the results from this study, we
with women consuming fructose (P = 0.033; Supplemental Table and other investigators have reported that long-term consumption
3). Fructose consumption resulted in larger increases of 24-hour (≥2 weeks) of fructose at 20%–25% of energy requirement did not

The Journal of Clinical Investigation    http://www.jci.org    Volume 119    Number 5    May 2009 1325
research article

Table 4
Baseline levels and percentage changes in lipid, lipoprotein, DNL, and LPL activity after consumption of glucose- or fructose-sweetened bev-
erages for 10 weeks

Outcome variable Glucose Glucose Fructose Fructose

(0 weeks) (% change 10 weeks) (0 weeks) (% change 10 weeks)
Fasting TG (mg/dl) 146 ± 17 +9.7 ± 3.2A 144 ± 18 +3.9 ± 5.5
23-hour TG AUC (mg/dl × 23 h) 783 ± 118 –32.0 ± 14.8 808 ± 167 +99.2 ± 31.5B
Mean 24-hour TG (mg/dl) 171 ± 20 +2.5 ± 4.0 163 ± 21 +18.2 ± 5.8C
Postprandial TG peak (mg/dl) 202.4 ± 24.6 +9.8 ± 5.1 211.1 ± 28.3 +38.1 ± 7.9C
Fasting cholesterol (mg/dl) 186 ± 8 +3.9 ± 2.0 186 ± 8 +10.1 ± 1.3D
Fasting LDL-C (mg/dl) 123.4 ± 5.9 +3.6 ± 3.0 115.3 ± 8.0 +13.9 ± 2.3E
Fasting HDL-C (mg/dl) 39 ± 2 –2.4 ± 2.1F 40 ± 3 +3.5 ± 1.8E
Fasting apoB (mg/dl) 86 ± 6 +3.0 ± 3.6 79 ± 6 +27.2 ± 4.3B
Postprandial apoB (mg/dl) 81 ± 6 +6.9 ± 3.7 74 ± 6 +25.0 ± 4.9B
apoB/apoA1 (mg/dl) 0.75 ± 0.07 +1.8 ± 3.4 0.63 ± 0.06 +22.4 ± 4.3G
Fasting sdLDL-C (mg/dl) 29.9 ± 3.5 +13.3 ± 5.8E 24.7 ± 2.7 +44.9 ± 9.7G
Fasting oxLDL-C (U/l) 53.3 ± 3.0 +0.7 ± 3.0 50.8 ± 3.9 +12.8 ± 2.6E
Postprandial RLP-TG (mg/dl) 70.7 ± 11.4 +15.2 ± 6.3 82.6 ± 16.5 +78.6 ± 19.8C
Postprandial RLP-C (mg/dl) 10.1 ± 1.4 +3.7 ± 5.4 10.9 ± 1.6 +33.9 ± 11.8C
Mean 24-hour FFA (mEq/l) 0.27 ± 0.01 +9.0 ± 2.6F 0.28 ± 0.02 +0.9 ± 3.8
Fasting fractional DNL (%) 8.8 ± 1.8 +12.3 ± 10.3 9.9 ± 1.3 –6.3 ± 16.6
Postprandial fractional DNL (%) 13.4 ± 2.8 +27.3 ± 13.6 11.4 ± 1.4 +75.4 ± 25.6H
Fasting postheparin LPL activity (U/l) 79.0 ± 5.8 +0.2 ± 4.4 101.4 ± 10.3 +0.7 ± 7.1
Postprandial postheparin LPL activity (U/l) 77.0 ± 5.8 +20.3 ± 8.7 106.9 ± 11.1 –5.4 ± 8.9
AP < 0.05; BP < 0.001; CP < 0.0001, PROC MIXED 3-factor (time, sexual phenotype, MSRF) RM ANOVA with previous day’s energy intake as time-level
covariable, effect of time. DP < 0.0001; EP < 0.01; FP < 0.05; GP < 0.001, PROC MIXED 3-factor (time, sexual phenotype, MSRF) RM ANOVA, effect of
time. HP < 0.05, paired Student’s t test, 10 weeks vs. 0 weeks. Data represent mean ± SEM.

increase fasting TG concentrations in humans (6–12). However, tion but not glucose consumption increases hepatic fractional DNL
increases of fasting TG concentrations have been reported after 2 or in humans when measured during energy-balanced feeding.
more weeks of fructose consumption at 15%–20% of energy require- The increased rate of fructose-induced DNL generates fatty acids
ments (13–18). The reason for these conflicting results is unclear but for production of hepatic TG. Additionally, hepatic DNL limits fatty
may be related to the marked within-group and within-individual acid oxidation in the liver via production of malonyl-CoA, which
variability we observed in fasting TG responses to dietary fructose. reduces the entry of fatty acids into the mitochondria (28). Thus,
Bantle et al. previously reported that a 6-week diet providing fructose-induced DNL may increase hepatic lipid not only by sup-
17% of energy from fructose increased postprandial TG concen- plying endogenous fatty acids but also by increasing the intrahepatic
trations compared with an isocaloric glucose diet in healthy men availability of fatty acids derived from the circulation (28). Increased
but not in healthy women (13). The increases of postprandial TG hepatic lipid levels are associated with increased VLDL synthesis and
in men in the present study confirm those reported by Bantle et secretion, specifically that of VLDL1 (29). apoB is essential for the
al.; however, the women consuming fructose from this study and intracellular assembly of TG into VLDL, and apoB degradation is
our previous study (11) also had significantly increased TG AUCs reduced when hepatic lipid is increased (30). The positive correlations
compared with women consuming glucose. The women studied between the previous day’s energy intake and postprandial apoB and
by Bantle et al. were leaner and younger than the women in the TG concentrations in subjects consuming fructose suggest that posi-
present study and of mixed menopausal status. Body fat (19), age tive energy balance also increases hepatic lipid availability.
(20), and menopausal status (21, 22) have all been shown to affect While we propose that increased VLDL synthesis/secretion is
postprandial TG responses in women. the main contributor (31), the significantly different postpran-
We have demonstrated that a mechanism by which fructose induc- dial postheparin LPL responses between the 2 treatment groups
es postprandial hypertriglyceridemia is through increased hepatic suggest that reduced TG clearance might also contribute to fruc-
DNL. It has long been established that in contrast to the metabolism tose-induced postprandial hypertriglyceridemia. Both reduced
of glucose, fructose metabolism is independent of phosphofructose postmeal exposure to insulin (32) and decreased insulin sensitiv-
kinase regulation; thus, its uptake by the liver and its metabolism ity (33) may have contributed to lowered postprandial LPL activity
to DNL substrate is not limited by energy status (cytosolic ATP and in subjects consuming fructose compared with those consuming
citrate levels) (23). In addition, fructose may activate sterol receptor glucose. It has been demonstrated that SAT is more sensitive to
element–binding protein-1c independently of insulin, which acti- the effects of insulin in activating LPL than VAT (34); thus, the
vates genes involved in DNL (24, 25). However, demonstrations that differential LPL responses may contribute to the increased fat
sustained fructose consumption increases DNL in humans are lim- deposition in SAT in subjects consuming glucose and increased
ited to an abstract (26) and an overfeeding study (800–1000 kcal/d fat deposition in VAT in subjects consuming fructose.
fructose in excess of energy requirement) (27). This is the first study, There is growing evidence linking increases of postprandial TG
to our knowledge, to demonstrate that prolonged fructose consump- concentrations with proatherogenic conditions (35–40). This link

1326 The Journal of Clinical Investigation    http://www.jci.org    Volume 119    Number 5    May 2009
 research article

Figure 2
Plasma TG. 24-hour circulating TG concentrations in subjects before and after 2, 8, and 10 weeks of consuming glucose-sweetened beverages
(A) or fructose-sweetened beverages (B). ++P < 0.01 PROC MIXED 3-factor RM ANOVA with prior day’s energy intake covariable for 23-hour
TG AUC. Glucose, n = 14; fructose, n = 17. Data represent mean ± SEM.

may be due to lipoprotein remodeling induced by increased lev- during the outpatient intervention, it is not known whether fruc-
els of VLDL1 and mediated by cholesteryl ester transfer protein tose consumption decreases insulin sensitivity to the same degree in
(CETP) and hepatic lipase, which results in increased concentra- the absence of BW and fat gain. Interestingly, the changes of BW or
tions of sdLDL and RLP (31, 41–44). sdLDL is more easily oxidized body fat and the change of insulin sensitivity were not correlated.
than larger LDL particles (45), and accordingly, subjects consuming Recently, it was reported that the inclusion of fructose with an
fructose also had significantly increased concentrations of oxidized energy-balanced diet in men increased fasting glucose levels, but
LDL. The doubling of fructose-induced increases of both fasting other indices of insulin sensitivity were unaffected (17). Factors
and postprandial sdLDL concentrations in subjects with metabol- that may contribute to the differences between these results and
ic syndrome (MSRF = 3) compared with subjects with 0–2 MSRF those of the present study include study duration, fructose expo-
was striking. In all cases, the additional risk factor in the 5 subjects sure, subject weight gain, age, and baseline insulin sensitivity.
with MSRF 3 compared with the 5 subjects with MSRF 2 was the Effect of sexual phenotype. Bantle et al. (13) previously reported that
presence of fasting TG over 150 mg/dl, suggesting that preexisting fructose-induce postprandial TG responses were greater in men
hypertriglyceridemia can exacerbate lipoprotein remodeling associ- than in women; thus, the higher 24-hour TG exposure, postprandial
ated with fructose-induced increases of postprandial TG. TG peaks, and RLP-C concentrations we report in men compared
The unchanged FFA exposure in subjects consuming fructose is an with women were not unexpected. The finding that women had
important finding. It has been suggested that fructose consumption sdLDL increases comparable to those of men despite having lower
promotes development of the metabolic syndrome through increased TG responses to fructose consumption suggests that the TG thresh-
adiposity and adipose insulin resistance, which leads to increased cir- old required to increase the production of sdLDL may be lower in
culating and portal levels of FFA (46). The resulting increase in hepat- women than men. The greater decrease of insulin sensitivity noted
ic FFA uptake increases hepatic lipid availability and hepatic insulin in women compared with men in response to sugar consumption
resistance (47). However, the absence of an effect of fructose on sys- was unexpected and contrasts with the sexual phenotype effect noted
temic FFA suggests that fructose may promote insulin resistance by during a fructose overfeeding study (55). Indices of insulin sensitivity
providing a more direct source of intrahepatic lipid via DNL (48). were decreased in healthy young men but were unchanged in healthy
Insulin sensitivity and glucose tolerance. We propose that the increased young women (55). Hepatic lipid accumulates when TG production
hepatic lipid resulting from fructose-induced DNL leads to hepatic exceeds FFA oxidation and VLDL production and secretion (56). It is
insulin resistance (49, 50), possibly by increasing levels of diacylglyc- possible that women in the current study exhibited larger decreases
erol (49). Diacylglycerol is a known activator of novel PKC (51), and in insulin sensitivity than men due to decreased rates of VLDL pro-
increases of both diacylglycerol and novel PKC activity are associated duction and secretion, which resulted in greater increases of hepatic
with lipid-induced insulin resistance (52, 53). It has been previously lipid content. The younger women (55) may have accumulated less
reported that consumption of 1,000 extra kcal/d fructose along hepatic lipid than the older women due to their having increased
with ad libitum diet reduced insulin sensitivity in healthy subjects, rates of FFA oxidation (57). The overall effect of sexual phenotype on
whereas insulin sensitivity was unchanged in subjects consuming an the changes of insulin sensitivity also contrasts with the sexual phe-
extra 1,000 kcal/d glucose (54). Our results confirm this difference notype effect for the changes of intraabdominal fat, which increased
with a smaller quantity of fructose (617 ± 24 kcal/d) consumed with more in men than in women. These opposing sexual phenotype
an energy-balanced diet in a controlled metabolic setting for 3 days effects also suggest that fructose decreases insulin sensitivity inde-
prior to testing. However, because the subjects gained BW and fat pendently of visceral adiposity and FFA levels (48).

The Journal of Clinical Investigation    http://www.jci.org    Volume 119    Number 5    May 2009 1327
research article

Figure 3
apoB, sdLDL, and RLP-TG.
Fasting apoB concentrations
in subjects before and after 2,
8, and 10 weeks of consuming
glucose-sweetened beverag-
es (A) or fructose-sweetened
beverages (B). Fasting sdLDL
concentrations in subjects
before and after 2, 8, and 10
weeks of consuming glucose-
sweetened beverages (C) or
fructose-sweetened beverag-
es (D). Postprandial RLP-TG
concentrations in subjects
before and after 2, 8, and 10
weeks of consuming glucose-
sweetened beverages (E) or
fructose-sweetened beverag-
es (F). ++P < 0.01; +++P < 0.001;
++++P < 0.0001, PROC MIXED

3-factor RM ANOVA (C and

D) with prior day’s energy
intake covariable (A, B, E,
and F). *P < 0.05; **P < 0.01;
***P < 0.001; ****P < 0.0001,
Tukey’s multiple comparison
test vs. 0 weeks. Glucose,
n = 15; fructose, n = 17. Data
represent mean ± SEM.

Model. Figure 6 presents a proposed model for the divergent met- from beverages alone approaches or exceeds 15% of energy in ado-
abolic effects of glucose and fructose consumption. lescents and adults up to 40 years of age. The large SDs in several
Fructose and public health. While this study was designed to compare of these reports suggest that at least 16% of the studied popula-
the biological effects of glucose and fructose consumption on lipid tions was consuming over 25% of daily energy requirements from
and carbohydrate metabolism, the potential implications of the sugar-sweetened beverages (59, 62, 63).
results on public health is of interest. Foods and beverages in the US Conclusions. We reached the following conclusions: (a) The increase
are typically sweetened with sucrose (50% glucose and 50% fructose) in VAT in subjects consuming fructose and the increase in the expres-
or high-fructose corn syrup (HFCS), which is usually 45%–58% glu- sion of lipogenic genes in SAT in subjects consuming glucose suggest
cose and 42%–55% fructose, rather than pure glucose or fructose. We that fructose and glucose have differential effects on regional adipose
have reported in a short-term study that the 23-hour postprandial distribution. We believe that these results are novel and warrant fur-
TG profiles in male subjects consuming 25% energy as HFCS (55% ther investigation. (b) In addition to increases of postprandial TG
fructose) or sucrose were elevated to a degree similar to that observed and fasting and postprandial apoB, we show for what we believe is
when pure fructose–sweetened beverages were consumed (19). There- the first time that fructose consumption increases plasma concen-
fore, it is uncertain whether the adverse effects of sucrose and HFCS trations of fasting sdLDL, oxidized LDL, and postprandial RLP-C
consumption are “diluted” by their lower fructose content relative and RLP-TG in older, overweight/obese men and women, whereas
to pure fructose. Additional studies are needed to compare the long- glucose consumption does not. These changes may be associated
term effects of consuming HFCS and/or sucrose with 100% fructose. with increased risk of cardiovascular disease (30, 36, 45, 64–66). (c)
The amount of sugar consumed by the subjects in this study, Fructose consumption increased hepatic fractional DNL, and post-
25% of energy requirements, is considerably higher than 15.8%, the prandial LPL activity was lower in subjects consuming fructose com-
current estimate for the mean intake of added sugars by Americans pared with those consuming glucose. These results suggest that both
(58). However, recent reports (59–63) suggest that the sugar intake increased DNL and decreased LPL-mediated clearance contribute to

1328 The Journal of Clinical Investigation    http://www.jci.org    Volume 119    Number 5    May 2009
 research article

tient intervention period during which subjects consumed fructose- or glu-

cose-sweetened beverages providing 25% of daily energy requirements with
an energy-balanced diet. The inpatient periods allowed comparisons of the
high-fructose and glucose diets under well-controlled metabolic conditions.
However, sugar-sweetened beverages are typically consumed as part of an ad
libitum diet that is likely to contain more energy than the inpatient diet and
have the potential to promote weight gain. Therefore, the purpose of the
8-week outpatient period was to compare the effects of consumption of fruc-
tose or glucose along with ad libitum diet on BW gain and composition.
Subjects. Participants were recruited through newspaper advertisements
and underwent a telephone and an in-person interview with medical history,
a complete blood count, and a serum biochemistry panel to assess eligibility.
Inclusion criteria included age from 40 to 72 years and BMI of 25–35 kg/m2
with a self report of stable BW during the prior 6 months. Women were con-
sidered postmenopausal based on a self report of no menstruation for at
least 1 year. Exclusion criteria included evidence of diabetes, renal disease,
or hepatic disease; fasting serum TG concentrations greater than 400 mg/dl;
Figure 4 hypertension (>140/90 mmHg); and history of surgery for weight loss. Indi-
Hepatic fractional DNL. Change of fractional DNL before and during viduals who smoked, reported exercise of more than 3.5 hours/week at a level
steady-state feeding of meals with glucose- or fructose-sweetened more vigorous than walking, or reported having used thyroid, lipid-lowering,
beverages (9 weeks) compared with high–complex carbohydrate
glucose-lowering, antihypertensive, antidepressant, or weight-loss medica-
meals (0 weeks). *P = 0.016, GLM ANOVA, effect of sugar on Δ of
16-hour fractional DNL AUC at 9 weeks vs. 0 weeks. Glucose, n = 8; tions were also excluded. Diet-related exclusion criteria included habitual
fructose, n = 10. Data represent mean ± SEM. ingestion of more than 1 sugar-sweetened beverage per day or more than 2
alcoholic beverages per day. The UCD Institutional Review Board approved
the experimental protocol, and subjects provided informed consent for par-
fructose-induced postprandial hypertriglyceridemia. (d) Consump- ticipation in the study. Thirty-nine subjects enrolled in the study, and experi-
tion of fructose at 25% of energy requirements with an ad libitum mental groups were matched for sexual phenotype, BMI, and fasting TG and
diet decreased glucose tolerance and insulin sensitivity in older over- insulin concentrations. Seven subjects (3 in the glucose group, 4 in the fruc-
weight/obese adults compared with glucose consumption. (e) VAT tose group) did not complete the study because of inability/unwillingness to
accumulation and increases of 24-hour TG exposure, peak postpran- comply with protocol or due to personal or work-related conflicts.
dial TG concentrations, and postprandial RLP-C concentrations in Diets — inpatient baseline. During the 2-week baseline and 2-week interven-
response to fructose consumption were more pronounced in men tion inpatient metabolic phases of the study in the CCRC, subjects con-
than in women. Consumption of sugar-sweetened beverages resulted sumed energy-balanced diets providing 15% of energy as protein, 30% as
in greater decreases in insulin sensitivity in women than in men. fat, and 55% as carbohydrate (Supplemental Table 1). During the baseline
Dose-response studies are needed to determine what levels of dietary period, the carbohydrate content consisted primarily of complex carbo-
fructose and HFCS and/or sucrose are associated with adverse chang- hydrates. The diet was designed as a 4-day rotating menu composed of
es of lipids and decreased insulin sensitivity in different populations. conventional foods served in 3 meals, with 25% of the energy provided
at breakfast (0900 hours), 35% at lunch (1300 hours), and 40% at dinner
Methods (1800 hours). During the inpatient periods when the diets were controlled
Study design. This was a double-blinded parallel arm study that used matched and monitored, the subjects were required to consume all of the food and
subjects and consisted of 3 phases (Table 1): (a) a 2-week inpatient baseline were limited to only the food provided. Daily energy requirements were
period during which subjects consumed an energy-balanced diet; (b) an calculated by the Mifflin equation (67), with an adjustment of 1.3 for activ-
8-week outpatient intervention period during which subjects consumed ity on the days of the 26-hour isotope infusion and 24-hour serial blood
either fructose- or glucose-sweetened beverages providing 25% of daily ener- collections and an adjustment of 1.5 for the other inpatient days. BW was
gy requirements along with their usual ad libitum diet; and (c) a 2-week inpa- monitored daily, and energy intake was adjusted when the slope of the BW

Table 5
Baseline levels and percentage changes in fasting glucose and insulin and indices of insulin sensitivity after consumption of glucose- or
fructose-sweetened beverages for 10 weeks

Outcome variable Glucose Glucose Fructose Fructose

(0 weeks) (% change 10 weeks) (0 weeks) (% change 10 weeks)
Fasting glucose (mg/dl) 87.6 ± 1.5 –1.4 ± 0.6A 88.7 ± 1.0 +5.3 ± 1.0B
Fasting insulin (μU/ml) 15.0 ± 1.9 +2.9 ± 4.0 14.0 ± 1.5 +10.2 ± 4.2C
Glucose 3-h AUC OGTT (mg/dl × 3 h) 129.4 ± 16.2 +31.4 ± 16.5D 107.7 ± 18.5 +60.2 ± 23.8E
Insulin 3-h AUC OGTT (μU/ml × 3 h) 232.9 ± 33.0 +13.9 ± 9.2 273.1 ± 44.4 +26.9 ± 5.9D
Insulin sensitivity index (mmoles 2H20/4-h insulin AUC) 0.236 ± 0.036 +1.1 ± 8.6 0.254 ± 0.049 –17.3 ± 3.8E
AP < 0.05; BP < 0.001; CP < 0.01, PROC MIXED 3-factor (time, sexual phenotype, MSRF) RM ANOVA, effect of time. DP < 0.05; EP < 0.01, paired Student’s
t test, 10 weeks vs. 0 weeks. Data represent mean ± SEM.

The Journal of Clinical Investigation    http://www.jci.org    Volume 119    Number 5    May 2009 1329
research article

Figure 5
OGTT and glucose disposal test. Glucose concentrations during an OGTT in subjects before and after 9 weeks of consuming (A) glucose-sweet-
ened beverages or (B) fructose-sweetened beverages. Insulin concentrations during an OGTT in subjects before and after 9 weeks of consuming
glucose-sweetened beverages (C) or fructose-sweetened beverages (D). *P < 0.05; **P < 0.01; ***P < 0.001, paired Student’s t test, 9 weeks vs.
0 weeks. Glucose, n = 15; fructose, n = 17. Insulin sensitivity index during glucose disposal test as percentage of baseline in subjects before and
after 9 weeks of consuming glucose- or fructose-sweetened beverages (E). **P < 0.01, paired Student’s t test, 9 weeks vs. 0 weeks. Glucose:
n = 14; fructose: n = 17. Data represent mean ± SEM.

chart trended up or down (68). For 31 of the 32 subjects who completed beverage supply twice weekly at the CCRC. The beverages contained a
the study, no adjustments were required. biomarker (riboflavin), which was measured fluorometrically in urine
Diets — outpatient intervention. Subjects were instructed to consume samples collected at the time of beverage pickup to monitor compliance.
their usual diets. Sugars were provided to the subjects as 3 daily servings Subjects were informed that they were being monitored for compliance.
of glucose- or fructose-sweetened beverages flavored with an unsweet- Based on fluorescein counts, urinary riboflavin levels were 14.6 ± 1.3
ened drink mix (Kool-Aid; Kraft). Subjects were instructed to drink 3 times higher in subjects consuming glucose and 12.4 ± 0.8 times higher
servings per day, 1 with each meal, and not to consume other sugar- in subjects consuming fructose during the intervention weeks than dur-
containing beverages including fruit juice during the study protocol. ing baseline (P = 0.16), which suggests the 2 groups were comparably
Beverages were prepared under the supervision of the study supervisor compliant (Supplemental Table 14).
at the UCD Department of Nutrition Ragle Human Nutrition Research Estimates of food intake during the outpatient phase of the study were
Facility. The subjects, CCRC personnel, and technicians who performed collected by 24-hour recall (via telephone) using the USDA 5-step multiple-
analyses were blinded to the sugar assignments. Subjects obtained their pass method as described by Conway (69) (Supplemental Table 2). The recalls

1330 The Journal of Clinical Investigation    http://www.jci.org    Volume 119    Number 5    May 2009
 research article

Figure 6
Proposed mechanisms underlying the differential effects of fructose and glucose consumption. Hepatic glucose metabolism is regulated by phos-
phofructokinase, which is inhibited by ATP and citrate when energy status is high, thus limiting hepatic uptake of dietary glucose and production
of DNL substrates. The hepatic metabolism of dietary fructose is independent of energy status, resulting in unregulated hepatic fructose uptake
and increased lipogenesis. The resulting increased hepatic lipid decreases apoB degradation and increases production/secretion of VLDL-TG,
mainly as TG-rich VLDL1 (29). This, along with chylomicron competition for LPL-mediated TG hydrolysis and reduced LPL activation by insulin,
results in longer VLDL residence time, allowing for augmented cholesteryl ester transfer protein–mediated (CETP-mediated) lipid exchanges with
LDL and increased LDL-TG and RLP levels. Hydrolysis of LDL-TG by hepatic lipase increases plasma sdLDL concentrations. After an overnight
fast, DNL is no longer elevated and VLDL and chylomicrons remnants have been cleared; thus, plasma TG levels are normal. Postprandially,
the increment of plasma apoB levels is associated with VLDL particles; in the fasting state, it is presumably associated with sdLDL, which turns
over more slowly. As SAT is more sensitive to insulin activation of LPL activity than VAT, reduced postmeal insulin exposure may lead to less
TG uptake in SAT and thus increased TG uptake/accumulation in VAT. Increased hepatic lipid supply may also induce hepatic insulin resistance,
possibly through increased levels of diacylglycerol, which activates novel PKC (85). Novel PKC decreases tyrosine phosphorylation of the insu-
lin receptor/insulin receptor substrate 1, resulting in increased hepatic glucose production, impaired glucose tolerance, and increased fasting
glucose and insulin concentrations. oxLDL, oxidized LDL.

were conducted on 6 random days at 2-week and 7-week intervention. The that were judged by the dietitian to be outliers to the usual dietary pattern
same registered dietitian administered the recall to all subjects. Recalls were due to illness or other circumstances.
analyzed with Nutrition Data System for Research (version 2005, University Diets — inpatient intervention. Following the 8-week outpatient interven-
of Minnesota). The results from all 6 recalls were averaged, except for reports tion period, subjects returned to the CCRC for a 2-week inpatient inter-

The Journal of Clinical Investigation    http://www.jci.org    Volume 119    Number 5    May 2009 1331
research article

vention period. The energy-balanced intervention diet was the same as (55.6 ml/h) of 0.5 g/h sodium [1-13C]acetate (Cambridge Isotope Laboratories
described for the baseline diet, except that while the overall carbohydrate Inc.) was initiated. Fasting blood samples were collected the next morning at
content remained at 55% of energy requirements, 30% of the energy was 0700, 0730, and 0800 hours, followed by the initiation of steady-state feed-
from complex carbohydrates and 25% was provided by fructose- or glucose- ing using a feeding protocol previously validated to study VLDL kinetics (72).
sweetened beverages (Supplemental Table 1). Subjects consumed 1/16 of their energy requirement each hour from 0800 to
Meals consumed during and prior to the 24 hours in which blood was collected. 2300 hours as rice and chicken casseroles with bread during baseline and as
Meals served during the 24-hour serial blood collections were identical at rice and chicken casseroles with fructose or glucose beverages during inter-
all 3 intervention time points (2 weeks, 8 weeks, 10 weeks), and the inter- vention. Postprandial blood was collected hourly from 1300 to 2400 hours.
vention meals were matched as closely as possible to the baseline meals (0 Lipoprotein fractions were isolated using sequential ultracentrifugation (73).
weeks), except for the substitution of 25% of energy from sugars for the Plasma VLDL-13C palmitate enrichment and mass isotopomers were mea-
complex carbohydrates. The baseline (0 weeks) and final (10 weeks) inter- sured as described by Hellerstein et al. (74) and Faeh et al. (27) in the VLDL1
vention 24-hour serial blood collections were performed after subjects had fraction. Fasting and postprandial hepatic DNL were calculated as the mean of
consumed energy-balanced, weight-maintaining diets in the CCRC for 10 the samples collected before and during the final 10 hours of steady-state feed-
days. The 24-hour serial blood collections that occurred after 2 weeks and ing, respectively. Due to budget constraints, the stable isotope infusion was
8 weeks of intervention were preceded by 2- and 8-week periods of ad libi- conducted only on the first 23 subjects enrolled in the study, and of these tests,
tum food intake. On the day before the 2- and 8-week blood collections, 5 were unsuccessful due to nonpatent catheters. There was a disproportionate
subjects entered the CCRC at 0700 hours and consumed buffet meals ad number of men in the fructose subset (7 men, 3 women); however. the effect
libitum for breakfast, lunch, and dinner along with the sugar-sweetened of sexual phenotype on the general linear model (GLM) 2-factor ANOVA for
beverages. The buffet menu items and quantities provided were the same 16-hour AUC for fractional DNL was P = 0.89. Otherwise, there were no signifi-
on both days, and each subject’s ad libitum food intake was determined cant differences in baseline characteristics between the 2 groups of subjects in
without the subject’s being aware that their food intake was monitored. the DNL subset, and they were representative of the larger group.
Measurements of body composition and blood pressure. Subjects were weighed Postheparin LPL. During the baseline and intervention periods, a post-
daily in the morning before breakfast during the inpatient phases. Total prandial blood sample was collected at 2000 hours followed by i.v. injec-
body fat was determined by dual energy x-ray absorptiometry (DXA). CT tion of 50 units of heparin/kg BW and collection of another blood sample
scans of the abdomen were performed at the level of the umbilicus to quan- 10 minutes later. The procedure was repeated in the fasting state the fol-
tify SAT, VAT, and total abdominal fat areas. Total tissue area was com- lowing week. Blood samples were assayed for total lipase and hepatic lipase
puted as the area with an attenuation range of –250 to +1,500 Hounsfield activity by Asahi Kasei Pharma by the method of Imamura (75).
units; an attenuation range of –250 to –50 Hounsfield units was used to OGTT and disposal test. During baseline and intervention week 9, an OGTT
define fat areas. SAT and VAT areas were differentiated by delineating the was performed after an overnight fast. Blood samples were collected before
border of the peritoneal cavity. CT scans and fat quantifications were per- and 30, 60, 90, 120, 180, and 240 minutes after consumption of a 75-gram glu-
formed at the UCD Medical Center under the supervision of John McGa- cose solution (300 ml). The OGTT was expanded to include measurement of
han. Blood pressure was measured with an automatic blood pressure cuff glucose disposal and insulin sensitivity using the deuterated-glucose disposal
(Welch Allyn) twice daily during inpatient periods. test developed by Hellerstein and colleagues, as described previously (5). The
Gluteal adipose biopsy and RNA analyses. Needle biopsy samples of subcuta- 75 g oral glucose load contained 15 g of 6,6,D2 glucose (Cambridge Isotope
neous gluteal adipose tissue were obtained following lidocaine injection. Laboratories). Because the hydrogen atoms in C-H bonds of glucose at posi-
A 3-ml fat biopsy was obtained, transferred to RNAlater (Ambion; Applied tion C-6 are more than 90% lost to tissue water during glycolytic metabolism
Biosystems), and stored at 4°C. After 48 hours, adipose samples were but are retained in the glucose molecule in the absence of glycolytic metabo-
removed from RNAlater and stored at –80°C until analysis. RNA isolation, lism, 2H2O production provides a sensitive index of whole-body glycolytic
cDNA synthesis, and gene expression analysis using TaqMan Gene Expres- utilization of plasma glucose. In addition, glucose transport across capillary
sion Assays (Applied Biosystems) were performed as previously described endothelium, transport into cells, phosphorylation, and glycolytic metabo-
(70) in the laboratory of Ronald Krauss. The percentage changes of gene lism are all increased by insulin (76), and 2H2O production (μmoles)/AUC
expression were calculated as the natural log of the expression at 10 weeks/ insulin has been shown to be an insulin-sensitivity index that is highly cor-
natural log of the expression at baseline × 100. related to the M value from euglycemic-hyperinsulinemic glucose clamps (5).
24-hour fasting and postprandial blood profiles. 24-hour serial blood collec- The deuterium content of plasma samples was determined using a Thermo
tions occurred during baseline (0 weeks) and after 2, 8, and 10 weeks of Finnigan High Temperature Conversion/Elemental Analyzer coupled with a
intervention. At 0730 hours, an i.v. catheter was inserted into an arm vein Thermo Finnigan MAT 253 Isotope Ratio-Mass Spectrometer via a ConFlo
by a registered nurse and kept patent with slow saline infusion. Three fast- III Interface. The deuterium isotope abundance was calculated in δ2H values
ing blood samples were collected in EDTA at 0800, 0830, and 0900 hours. relative to the international Vienna Standard Mean Ocean Water standard,
Thirty-three postprandial blood samples were collected at 30- to 60-minute transformed to atom percentage excess by using a calibration curve of stan-
intervals from 0930 until 0800 hours the next morning (32, 71). Meals were dards, and converted to millimoles by multiplying the 2H2O enrichment by
served at 0900, 1300, and 1800 hours. An additional 3 to 6 ml of blood was the total body water pool size and dividing by 20 (the molecular weight of
collected at each of the following time-points: 0800, 0830, and 0900 and 2H O). Total body water was measured immediately prior to the OGTT by
2
2200, 2300, and 2330 hours. The plasma from the 3 fasting samples (0800, bioimpedance spectroscopy (Xitron Technologies).
0830, and 0900 hours) was pooled, as was the plasma from the 3 postpran- Analyses. Glucose concentrations were measured with an automated glu-
dial blood samples (2200, 2300, and 2330 hours); multiple aliquots of each cose analyzer (YSI) and insulin by radioimmunoassay (Millipore). Lipid and
pooled sample were stored at –80°C. lipoprotein concentrations (total cholesterol, HDL, TG, apoB, and apoA1)
Hepatic fractional DNL. Fractional DNL was studied via infusion of isotopic were determined using a Polychem Chemistry Analyzer (Polymedco Inc.).
acetate during baseline and 10-week intervention. At 2000 hours, i.v. catheters FFA concentrations were measured with an enzymatic colorimetric assay
were inserted into veins of both arms and kept patent with slow saline infu- (Wako) adapted to a microtiter plate. Oxidized LDL was measured with a com-
sion. Following a baseline blood collection, at 2200 hours, a 26-hour infusion mercially available ELISA (Mercodia). sdLDL (density = 1.044–1.063 g/ml)

1332 The Journal of Clinical Investigation    http://www.jci.org    Volume 119    Number 5    May 2009
 research article

was separated from plasma by precipitation (77). The LDL concentration sugar groups were analyzed in sugar-specific 2-factor GLM ANOVAs with
of the sdLDL extract and plasma was determined by direct homogenous Tukey’s multiple comparison tests, and the effects of the individual sugars
assay using detergents (LDL-EX; Denka Seiken Co.) (78). The accuracy of were analyzed by 2-tailed paired Student’s t test (10 vs. 0 weeks). P < 0.05
this homogenous method has been described (79, 80). RLP were quantified was considered significant. Data are presented as mean ± SEM.
with an immunoseparation assay (81, 82). Note added in proof. A new manuscript from our group appeared recently
Statistics. The AUC was calculated for TG and DNL using the trapezoidal (86) that extends our previous study from 2004 (32) of the effects of short-
method. The mean of the 3 baseline values was determined, and the net AUC term (1 day) fructose and glucose consumption in normal weight young
was calculated by subtracting the AUC values below baseline from the AUC women to young, obese men and women and examines the effects of base-
values above baseline. For TG, glucose, and insulin, fasting concentrations line insulin sensitivity (HOMA-IR) on the acute 24-hour TG response. The
represent the mean concentration of the samples collected at 0800, 0830, and current study provides data on the effects of sustained long-term fructose
0900 hours during the 24-hour serial blood collections. For all other lipid and glucose exposure on dyslipidemia, de novo lipogenesis, insulin sensi-
parameters, fasting concentrations were measured in pooled plasma samples tivity, and visceral adiposity.
collected during the 24-hour serial blood collections at 0800, 0830, and 0900
hours, and postprandial levels were measured in pooled plasma collected at Acknowledgments
2200, 2300, and 2330 hours. Statistical tests were performed with SAS 9.1. The authors thank Marinelle Nuñez, Theresa Tonjes, Brandi Bair,
Differences in the lead-in diet could potentially affect the responses to Rebecca Stewart, Sara Wuehler, Elaine Souza, and Patrick Lam for
fructose or glucose consumption at the 3 intervention time points (2, 8, their excellent technical support and Nicole Mullen and the nurs-
and 10 weeks) compared with consumption of complex carbohydrates at ing staff at CCRC for their dedicated nursing support. We thank
baseline (0 weeks). Specifically, the 0- and 10-week 24-hour serial blood Shigeyuki Imamura for analyses of postheparin LPL activity. We
collections were preceded by 10 days of consumption of an energy-bal- also thank Jorge Dubcovsky and Iago Lowe for guidance and Janet
anced diet, whereas the 2-week and 8-week collections were preceded by Peerson for expert advice on the statistical analysis of the data.
consumption of ad libitum diets. To differentiate the acute variations asso- This research was supported with funding from NIH grant R01
ciated with differences in the prior day’s energy intake from the effects of HL-075675. The project also received support from grant number
sugar intake, the energy intake of each subject during the 24 hours prior UL1 RR024146 from the National Center for Research Resources
to each 24-hour period in which blood was collected was entered as a time- (NCRR), a component of the NIH, and the NIH Roadmap for Med-
level covariable in an RM model using PROC MIXED with time, type of ical Research. P.J. Havel’s laboratory also receives support from
sugar, sexual phenotype, and MSRF as factors. The covariable was removed NIH grants HL-091333, AT-002599, AT-002993, and AT-003545
from the model when its F statistic was ≤ 1.0. Insignificant 3-way interac- and the American Diabetes Association. N.L. Keim’s research is
tions were removed if they decreased the precision of the model. MSRF supported by intramural USDA-ARS CRIS 5306-51530-016-00D.
was defined by the American Heart Association/National Heart, Lung, We are grateful to Janet King of Children’s Hospital Oakland
and Blood Institute (83, 84). Response variables were further analyzed Research Institute for her ongoing support of this research.
using the same model in separate RM ANOVAs for fructose or glucose
with time, sexual phenotype, and MSRF as factors. For both the 3- and Received for publication September 8, 2008, and accepted in
4-factor RM ANOVAs, Tukey’s multiple comparison post tests were used revised form February 11, 2009.
to compare effects of type of sugar, sexual phenotype, or MSRF within and
between groups. For response variables that were measured at only 2 time Address correspondence to: Peter J. Havel, Department of Molecular
points, 0 and 10 weeks, the Δ or percentage Δ between the 2 time points Biosciences, School of Veterinary Medicine, University of California,
was analyzed by GLM, with type of sugar, sexual phenotype, and MSRF as Davis, One Shields Avenue, Davis, California 95616, USA. Phone:
factors. The effects of sexual phenotype and MSRF within the individual (530) 752-6553; Fax: (530) 752-2474; E-mail: pjhavel@ucdavis.edu.

1. Bizeau, M.E., and Pagliassotti, M.J. 2005. Hepatic O.E., 4th, and Prather, E.S. 1981. Blood lipids and 16. Herman, R.H., Zakim, D., and Stifel, F.B. 1970.
adaptations to sucrose and fructose. Metabolism. their distribution in lipoproteins in hyperinsulin- Effect of diet on lipid metabolism in experimental
54:1189–1201. emic subjects fed three different levels of sucrose. animals and man. Fed. Proc. 29:1302–1307.
2. Havel, P.J. 2005. Dietary fructose: implications for J. Nutr. 111:1045–1057. 17. Le, K.A., et al. 2006. A 4-wk high-fructose diet alters
dysregulation of energy homeostasis and lipid/car- 10. Swanson, J.E., Laine, D.C., Thomas, W., and Bantle, lipid metabolism without affecting insulin sensi-
bohydrate metabolism. Nutr. Rev. 63:133–157. J.P. 1992. Metabolic effects of dietary fructose in tivity or ectopic lipids in healthy humans. Am. J.
3. Le, K.A., and Tappy, L. 2006. Metabolic effects of fruc- healthy subjects. Am. J. Clin. Nutr. 55:851–856. Clin. Nutr. 84:1374–1379.
tose. Curr. Opin. Clin. Nutr. Metab. Care. 9:469–475. 11. Swarbrick, M.M., et al. 2008. Consumption of fruc- 18. Reiser, S., et al. 1989. Day-long glucose, insulin, and
4. Wei, Y., Wang, D., Topczewski, F., and Pagliassotti, tose-sweetened beverages for 10 weeks increases fructose responses of hyperinsulinemic and non-
M.J. 2007. Fructose-mediated stress signaling in the postprandial triacylglycerol and apolipoprotein-B hyperinsulinemic men adapted to diets containing
liver: implications for hepatic insulin resistance. concentrations in overweight and obese women. either fructose or high-amylose cornstarch. Am. J.
J. Nutr. Biochem. 18:1–9. Br. J. Nutr. 100:947–952. Clin. Nutr. 50:1008–1014.
5. Beysen, C., et al. 2007. Whole-body glycolysis mea- 12. Turner, J.L., Bierman, E.L., Brunzell, J.D., and Chait, 19. Stanhope, K.L., et al. 2008. Twenty-four-hour
sured by the deuterated-glucose disposal test corre- A. 1979. Effect of dietary fructose on triglyceride endocrine and metabolic profiles following con-
lates highly with insulin resistance in vivo. Diabetes transport and glucoregulatory hormones in hyper- sumption of high-fructose corn syrup-, sucrose-,
Care. 30:1143–1149. triglyceridemic men. Am. J. Clin. Nutr. 32:1043–1050. fructose-, and glucose-sweetened beverages with
6. Bantle, J.P., Swanson, J.E., Thomas, W., and Laine, 13. Bantle, J.P., Raatz, S.K., Thomas, W., and Geor- meals. Am. J. Clin. Nutr. 87:1194–1203.
D.C. 1992. Metabolic effects of dietary fructose in gopoulos, A. 2000. Effects of dietary fructose on 20. Nabeno, Y., Fukuchi, Y., Matsutani, Y., and Naito,
diabetic subjects. Diabetes Care. 15:1468–1476. plasma lipids in healthy subjects. Am. J. Clin. Nutr. M. 2007. Influence of aging and menopause on
7. Crapo, P.A., and Kolterman, O.G. 1984. The meta- 72:1128–1134. postprandial lipoprotein responses in healthy
bolic effects of 2-week fructose feeding in normal 14. Crapo, P.A., Kolterman, O.G., and Henry, R.R. 1986. adult women. J. Atheroscler. Thromb. 14:142–150.
subjects. Am. J. Clin. Nutr. 39:525–534. Metabolic consequence of two-week fructose feed- 21. Kolovou, G.D., et al. 2004. The influence of natural
8. Koivisto, V.A., and Yki-Jarvinen, H. 1993. Fructose ing in diabetic subjects. Diabetes Care. 9:111–119. menopause on postprandial lipemia in heterozy-
and insulin sensitivity in patients with type 2 diabetes. 15. Hallfrisch, J., et al. 1983. Effects of dietary fructose on gotes for familial hypercholesterolemia. J. Womens
J. Intern. Med. 233:145–153. plasma glucose and hormone responses in normal Health. (Larchmt.) 13:1119–1126.
9. Reiser, S., Bickard, M.C., Hallfrisch, J., Michaelis, and hyperinsulinemic men. J. Nutr. 113:1819–1826. 22. Kolovou, G.D., and Bilianou, H.G. 2008. Influence

The Journal of Clinical Investigation    http://www.jci.org    Volume 119    Number 5    May 2009 1333
research article

of aging and menopause on lipids and lipoproteins protein. Biochem. Soc. Trans. 31:1066–1069. Clin. Chim. Acta. 367:36–47.
in women. Angiology. 59:54S–57S. 45. Packard, C.J. 2006. Small dense low-density lipopro- 66. Walldius, G., and Jungner, I. 2006. The apoB/apoA-I
23. Mayes, P.A. 1993. Intermediary metabolism of fructose. tein and its role as an independent predictor of car- ratio: a strong, new risk factor for cardiovascular
Am. J. Clin. Nutr. 58:754S–765S. diovascular disease. Curr. Opin. Lipidol. 17:412–417. disease and a target for lipid-lowering therapy--a
24. Matsuzaka, T., et al. 2004. Insulin-independent 46. Rutledge, A.C., and Adeli, K. 2007. Fructose and the review of the evidence. J. Intern. Med. 259:493–519.
induction of sterol regulatory element-binding metabolic syndrome: pathophysiology and molec- 67. Mifflin, M.D., et al. 1990. A new predictive equation
protein-1c expression in the livers of streptozoto- ular mechanisms. Nutr. Rev. 65:S13–S23. for resting energy expenditure in healthy individuals.
cin-treated mice. Diabetes. 53:560–569. 47. Bergman, R.N., et al. 2007. Abdominal obesity: role Am. J. Clin. Nutr. 51:241–247.
25. Nagai, Y., et al. 2002. Amelioration of high fruc- in the pathophysiology of metabolic disease and 68. Keim, N.L., Stern, J.S., and Havel, P.J. 1998. Rela-
tose-induced metabolic derangements by activa- cardiovascular risk. Am. J. Med. 120:S3–S8; discus- tion between circulating leptin concentrations and
tion of PPARalpha. Am. J. Physiol. Endocrinol. Metab. sion S29–S32. appetite during a prolonged, moderate energy defi-
282:E1180–E1190. 48. Stanhope, K.L., and Havel, P.J. 2008. Fructose con- cit in women. Am. J. Clin. Nutr. 68:794–801.
26. Schwarz, J.M., Neese, R.A., Schakleton, C., and Heller- sumption: potential mechanisms for its effects to 69. Conway, J.M., Ingwersen, L.A., Vinyard, B.T.,
stein, M.K. 1993. De novo lipogenesis during fasting increase visceral adiposity and induce dyslipidemia and Moshfegh, A.J. 2003. Effectiveness of the US
and oral fructose ingestion in lean and obese hyper- and insulin resistance. Curr. Opin. Lipidol. 19:16–24. Department of Agriculture 5-step multiple-pass
insulinemic subjects [abstract]. Diabetes. 42:A39. 49. Morino, K., Petersen, K.F., and Shulman, G.I. 2006. method in assessing food intake in obese and non-
27. Faeh, D., et al. 2005. Effect of fructose overfeed- Molecular mechanisms of insulin resistance in obese women. Am. J. Clin. Nutr. 77:1171–1178.
ing and fish oil administration on hepatic de novo humans and their potential links with mitochon- 70. Mangravite, L.M., Dawson, K., Davis, R.R., Gregg,
lipogenesis and insulin sensitivity in healthy men. drial dysfunction. Diabetes. 55(Suppl. 2):S9–S15. J.P., and Krauss, R.M. 2007. Fatty acid desaturase
Diabetes. 54:1907–1913. 50. Seppala-Lindroos, A., et al. 2002. Fat accumula- regulation in adipose tissue by dietary composition
28. McGarry, J.D. 1995. Malonyl-CoA and carnitine tion in the liver is associated with defects in insulin is independent of weight loss and is correlated with
palmitoyltransferase I: an expanding partnership. suppression of glucose production and serum free the plasma triacylglycerol response. Am. J. Clin.
Biochem. Soc. Trans. 23:481–485. fatty acids independent of obesity in normal men. Nutr. 86:759–767.
29. Adiels, M., et al. 2006. Overproduction of large J. Clin. Endocrinol. Metab. 87:3023–3028. 71. Havel, P.J., Townsend, R., Chaump, L., and Teff, K.
VLDL particles is driven by increased liver fat con- 51. Newton, A.C. 2003. Regulation of the ABC kinases 1999. High-fat meals reduce 24-h circulating leptin
tent in man. Diabetologia. 49:755–765. by phosphorylation: protein kinase C as a paradigm. concentrations in women. Diabetes. 48:334–341.
30. Olofsson, S.O., and Boren, J. 2005. Apolipoprotein B: a Biochem. J. 370:361–371. 72. Lichtenstein, A.H., et al. 1990. Comparison of
clinically important apolipoprotein which assembles 52. Itani, S.I., Ruderman, N.B., Schmieder, F., and deuterated leucine, valine, and lysine in the mea-
atherogenic lipoproteins and promotes the develop- Boden, G. 2002. Lipid-induced insulin resistance surement of human apolipoprotein A-I and B-100
ment of atherosclerosis. J. Intern. Med. 258:395–410. in human muscle is associated with changes in dia- kinetics. J. Lipid Res. 31:1693–1701.
31. Adiels, M., Olofsson, S.O., Taskinen, M.R., and cylglycerol, protein kinase C, and IkappaB-alpha. 73. Karpe, F., Steiner, G., Uffelman, K., Olivecrona,
Boren, J. 2008. Overproduction of very low-density Diabetes. 51:2005–2011. T., and Hamsten, A. 1994. Postprandial lipopro-
lipoproteins is the hallmark of the dyslipidemia in 53. Yu, C., et al. 2002. Mechanism by which fatty acids teins and progression of coronary atherosclerosis.
the metabolic syndrome. Arterioscler. Thromb. Vasc. inhibit insulin activation of insulin receptor sub- Atherosclerosis. 106:83–97.
Biol. 28:1225–1236. strate-1 (IRS-1)-associated phosphatidylino- 74. Hellerstein, M.K., et al. 1991. Measurement of de
32. Teff, K.L., et al. 2004. Dietary fructose reduces cir- sitol 3-kinase activity in muscle. J. Biol. Chem. novo hepatic lipogenesis in humans using stable
culating insulin and leptin, attenuates postprandial 277:50230–50236. isotopes. J. Clin. Invest. 87:1841–1852.
suppression of ghrelin, and increases triglycerides 54. Beck-Nielsen, H., Pedersen, O., and Lindskov, H.O. 75. Imamura, S., et al. 2007. A novel method for mea-
in women. J. Clin. Endocrinol. Metab. 89:2963–2972. 1980. Impaired cellular insulin binding and insulin suring human hepatic lipase activity in posthepa-
33. Mead, J.R., Irvine, S.A., and Ramji, D.P. 2002. Lipo- sensitivity induced by high-fructose feeding in nor- rin plasma. J. Lipid Res. 48:453–457.
protein lipase: structure, function, regulation, and mal subjects. Am. J. Clin. Nutr. 33:273–278. 76. Williams, K.V., et al. 2003. Glucose transport and
role in disease. J. Mol. Med. 80:753–769. 55. Couchepin, C., et al. 2008. Markedly blunted meta- phosphorylation in skeletal muscle in obesity: insight
34. Fried, S.K., Russell, C.D., Grauso, N.L., and Brolin, bolic effects of fructose in healthy young female from a muscle-specific positron emission tomogra-
R.E. 1993. Lipoprotein lipase regulation by insulin subjects compared with male subjects. Diabetes phy model. J. Clin. Endocrinol. Metab. 88:1271–1279.
and glucocorticoid in subcutaneous and omental Care. 31:1254–1256. 77. Hirano, T., Ito, Y., and Yoshino, G. 2005. Measure-
adipose tissues of obese women and men. J. Clin. 56. Gibbons, G.F., Islam, K., and Pease, R.J. 2000. Mobil- ment of small dense low-density lipoprotein particles.
Invest. 92:2191–2198. isation of triacylglycerol stores. Biochim. Biophys. J. Atheroscler. Thromb. 12:67–72.
35. Bansal, S., et al. 2007. Fasting compared with non- Acta. 1483:37–57. 78. Hirano, T., Ito, Y., Saegusa, H., and Yoshino, G.
fasting triglycerides and risk of cardiovascular 57. Toth, M.J., and Tchernof, A. 2000. Lipid metabo- 2003. A novel and simple method for quantification
events in women. JAMA. 298:309–316. lism in the elderly. Eur. J. Clin. Nutr. 54(Suppl. of small, dense LDL. J. Lipid Res. 44:2193–2201.
36. Hyson, D., Rutledge, J.C., and Berglund, L. 2003. 3):S121–S125. 79. Okada, M., Matsui, H., Ito, Y., Fujiwara, A., and
Postprandial lipemia and cardiovascular disease. 58. Guthrie, J.F., and Morton, J.F. 2000. Food sources Inano, K. 1998. Low-density lipoprotein choles-
Curr. Atheroscler. Rep. 5:437–444. of added sweeteners in the diets of Americans. terol can be chemically measured: a new superior
37. Karpe, F. 1999. Postprandial lipoprotein metabolism J. Am. Diet Assoc. 100:43–51. method. J. Lab. Clin. Med. 132:195–201.
and atherosclerosis. J. Intern. Med. 246:341–355. 59. Mundt, C.A., et al. 2006. Relationships of activity 80. Sakaue, T., et al. 2000. Reactions of direct LDL-
38. Lopez-Miranda, J., et al. 2006. Postprandial lipo- and sugar drink intake on fat mass development in cholesterol assays with pure LDL fraction and IDL:
protein metabolism, genes and risk of cardiovascu- youths. Med. Sci. Sports Exerc. 38:1245–1254. comparison of three homogeneous methods. Clin.
lar disease. Curr. Opin. Lipidol. 17:132–138. 60. Storey, M.L., Forshee, R.A., and Anderson, P.A. Chim. Acta. 295:97–106.
39. Nordestgaard, B.G., Benn, M., Schnohr, P., and Tyb- 2006. Beverage consumption in the US population. 81. Marcoux, C., et al. 1998. Plasma remnant-like particle
jaerg-Hansen, A. 2007. Nonfasting triglycerides and J. Am. Diet Assoc. 106:1992–2000. lipid and apolipoprotein levels in normolipidemic and
risk of myocardial infarction, ischemic heart disease, 61. Striegel-Moore, R.H., et al. 2006. Correlates of bev- hyperlipidemic subjects. Atherosclerosis. 139:161–171.
and death in men and women. JAMA. 298:299–308. erage intake in adolescent girls: the National Heart, 82. Nakajima, K., et al. 1994. A new approach for the
40. Stalenhoef, A.F., and de Graaf, J. 2008. Association Lung, and Blood Institute Growth and Health detection of type III hyperlipoproteinemia by RLP-
of fasting and nonfasting serum triglycerides with Study. J. Pediatr. 148:183–187. cholesterol assay. J. Atheroscler. Thromb. 1:30–36.
cardiovascular disease and the role of remnant- 62. van der Horst, K., et al. 2007. Perceived parenting 83. Al-Daghri, N.M., Al-Attas, O.S., Al-Rubeaan, K.,
like lipoproteins and small dense LDL. Curr. Opin. style and practices and the consumption of sugar- and Sallam, R. 2008. Adipocytokine profile of type
Lipidol. 19:355–361. sweetened beverages by adolescents. Health Educ. 2 diabetics in metabolic syndrome as defined by
41. Berneis, K.K., and Krauss, R.M. 2002. Metabolic ori- Res. 22:295–304. various criteria. Diabetes Metab. Res. Rev. 24:52–58.
gins and clinical significance of LDL heterogeneity. 63. West, D.S., et al. 2006. Self-reported sugar-sweet- 84. Genuth, S., et al. 2003. Follow-up report on the diagno-
J. Lipid Res. 43:1363–1379. ened beverage intake among college students. sis of diabetes mellitus. Diabetes Care. 26:3160–3167.
42. Krauss, R.M. 2005. Dietary and genetic probes of Obesity (Silver Spring). 14:1825–1831. 85. Samuel, V.T., et al. 2007. Inhibition of protein kinase
atherogenic dyslipidemia. Arterioscler. Thromb. Vasc. 64. Matsuura, E., Hughes, G.R., and Khamashta, M.A. Cepsilon prevents hepatic insulin resistance in nonal-
Biol. 25:2265–2272. 2008. Oxidation of LDL and its clinical implication. coholic fatty liver disease. J. Clin. Invest. 117:739–745.
43. Marcovina, S., and Packard, C.J. 2006. Measurement Autoimmun. Rev. 7:558–566. 86. Teff, K.L., et al. 2009. Endocrine and metabolic effects
and meaning of apolipoprotein AI and apolipoprotein 65. Nakajima, K., Nakano, T., and Tanaka, A. 2006. of consuming fructose- and glucose-sweetened bever-
B plasma levels. J. Intern. Med. 259:437–446. The oxidative modification hypothesis of athero- ages with meals in obese men and women: influence
44. Packard, C.J. 2003. Triacylglycerol-rich lipoproteins sclerosis: the comparison of atherogenic effects on of insulin resistance on plasma triglyceride responses.
and the generation of small, dense low-density lipo- oxidized LDL and remnant lipoproteins in plasma. J. Clin. Endocrinol. Metab. doi:10.1210/jc.2008-2192.

1334 The Journal of Clinical Investigation    http://www.jci.org    Volume 119    Number 5    May 2009