Professional Documents
Culture Documents
INTRODUCTION
Some (4, 5), but not all (6), dietary intervention studies
involving either lycopene-containing foods or lycopene supplementation have shown potential short-term improvements in
LDL oxidation. However, whether these apparent short-term
benefits translate into long-term improvements in health, manifested by a reduction in the risk of CVD, remains unknown.
Limited data suggest an inverse association between lycopene
and CVD (713). Data concerning the relation between plasma
lycopene concentrations and CVD risk are limited, particularly
in women. Lingering questions remain about whether all carotenoids (9), specific carotenoids such as lutein/zeaxanthin (14),
or carotenoids in general (15) are associated with a reduced risk
of CVD.
Therefore, we evaluated baseline concentrations of plasma
lycopene, other carotenoids, and retinol in middle-aged and
elderly women who were initially free of CVD in the Womens
Health Study (WHS) to examine their relation with the subsequent risk of CVD in female health professionals in the United
States.
Study population
The WHS is an ongoing randomized, double-blind, placebocontrolled 2 2 factorial trial of the relation between low-dose
aspirin and vitamin E and the primary prevention of CVD and
cancer (16). The -carotene component of the trial was terminated in 1996 after a median follow-up of 2.1 y, in part because
of the lack of effect of -carotene on cancer incidence in the
1
From the Division of Preventive Medicine, Department of Medicine,
Brigham and Womens Hospital and Harvard Medical School, Boston
(HDS, JEB, and JMG); the Department of Epidemiology, Harvard School
of Public Health, Boston (HDS and JEB); the Massachusetts Veterans
Epidemiology Research and Information Center, VA Boston Healthcare
System, Boston (HDS and JMG); the Department of Ambulatory Care and
Prevention, Harvard Medical School, Boston (JEB); and the Department of
Medical Research, Our Lady of Mercy Medical Center, and Community
and Preventive Medicine, New York Medical College, Bronx, NY (EPN).
2
Supported by grants CA-47988 and HL-43851 from the National
Institutes of Health, Bethesda, MD, and a grant from DSM Nutritional
Products, Inc.
3
Address reprint requests to HD Sesso, Brigham and Womens Hospital, 900 Commonwealth Avenue East, Boston, MA 02215-1204. E-mail:
hsesso@hsph.harvard.edu.
Received March 31, 2003.
Accepted for publication July 7, 2003.
Am J Clin Nutr 2004;79:4753. Printed in USA. 2004 American Society for Clinical Nutrition
47
48
SESSO ET AL
49
Of the 483 cases identified with CVD, there were 109 cases
of MI, 112 cases of stroke (86 ischemic, 24 hemorrhagic, and
2 other), 85 cases of revascularization, 33 cases of CVD death,
and 144 cases of angina pectoris. The mean (SD) follow-up
time was 4.8 2.4 y for cases and controls matched for
follow-up time. Baseline lifestyle, clinical, and dietary risk
factors for cases and controls at the start of the study are
compared in Table 1. As expected, women who developed
CVD tended to weigh more, and more of these women had a
history of hypertension, diabetes mellitus, and hypercholesterolemia than did those women who remained free of CVD.
Women who developed CVD also had higher total cholesterol
concentrations and were more likely to have a parent with a
history of MI before the age of 60 y. Age and smoking status
were identical in the cases and controls because of our matching criteria. Small differences in dietary factors between the
cases and controls were observed.
The mean plasma lycopene concentration in women who
developed CVD was 16.2 7.6 g/dL and in the controls was
16.9 8.2 g/dL. Plasma lycopene concentrations ranged
from 2.3 to 81.4 g/dL in the 483 controls free of CVD and
were significantly correlated with dietary lycopene concentrations (r 0.14, P 0.004), although the magnitude of this
age-adjusted Spearman correlation coefficient was low and suggested little or no correlation. Of the major lycopene food sources,
tomato sauce was the only food source with a significant ageadjusted Spearman correlation coefficient (r 0.16, P 0.001)
with plasma lycopene in the controls. Intake of other dietary
factorscheese, refined grains, sodium, and red meatwere not
correlated with plasma lycopene concentrations.
Baseline characteristics of the subjects according to quartiles
of plasma lycopene in the 483 women who remained free of
CVD are shown in Table 2. Women with higher plasma
lycopene concentrations tended to be younger, consume more
alcohol, and weigh less. In contrast, only women in the highest
quartile of plasma lycopene were more likely to be former
smokers. Total cholesterol increased with increasing quartiles
of plasma lycopene (P 0.001), as did the proportion of
women with a history of hypercholesterolemia (P 0.04). No
appreciable differences in dietary factors were observed, other
than lycopene intake (P 0.03), which was higher in higher
quartiles of plasma lycopene. The proportion of women randomly assigned to active treatment were equally distributed
within each quartile of plasma lycopene (all P 0.05).
In the analyses matched for age and smoking and with
adjustment for randomized treatment assignments and plasma
total cholesterol concentrations, the risk of total CVD appeared
to decrease with higher concentrations of plasma lycopene
(Table 3). The RRs of total CVD for women in the lowest to
highest quartiles of plasma lycopene were 1.00 (referent), 0.78,
TABLE 1
Baseline characteristics of 483 women who subsequently developed
cardiovascular disease (cases) and an equal number of women who
remained free of cardiovascular disease (controls)
Age (y)1
Smoking status (%)1
Never
Former
Current
BMI (kg/m2)
History of hypertension (%)
History of diabetes mellitus (%)
History of hypercholesterolemia (%)
Plasma total cholesterol (mg/dL)
Parental history of myocardial infarction
at 60 y of age (%)
Exercise (%)
Rarely or never
1 time/wk
13 times/wk
4 times/wk
Alcohol consumption (%)
Rarely or never
13 drinks/mo
16 drinks/wk
1 drink/d
Postmenopausal hormone use (%)
Never
Former
Current
Fruit and vegetable intake (servings/d)
Other dietary factors6
Lycopene intake (g/d)
Total fiber intake (g/d)
Folate intake (g/d)
Saturated fat intake (g/d)
Plasma lycopene (g/dL)
Cholesterol-adjusted plasma lycopene
(g/dL)
Plasma -cryptoxanthin (g/dL)
Plasma lutein/zeaxanthin (g/dL)
Plasma retinol (g/dL)
Cases
(n 483)
Controls
(n 483)
58.8 8.42
58.8 8.4
42.2
35.4
22.4
27.3 5.2
51.1
12.6
47.6
207 40
42.2
35.4
22.4
25.8 4.73
32.53
2.73
34.23
199 364
21.4
14.15
45.8
22.0
23.6
8.7
42.7
17.8
29.8
9.7
46.0
15.5
28.4
10.1
51.4
11.8
26.7
10.1
39.5
18.7
41.8
6.0 3.4
41.7
14.9
43.4
6.1 3.5
8705 5874
19.0 6.0
435 234
20.0 5.2
16.2 7.6
9078 5564
19.2 5.8
414 214
20.0 4.9
16.9 8.2
16.0 7.3
10.6 7.9
16.4 7.0
56.2 16.7
17.1 8.05
11.2 7.7
17.6 9.55
54.1 15.45
Matching variable.
x SD.
35
Significantly different from cases: 3P 0.0001, 4P 0.001, 5P
2
0.03.
6
50
SESSO ET AL
TABLE 2
Comparison of baseline characteristics of 483 controls according to quartile of plasma lycopene
Quartile of plasma lycopene
Age (y)
Smoking status (%)
Never
Former
Current
BMI (kg/m2)
History of hypertension (%)
History of diabetes mellitus (%)
History of hypercholesterolemia (%)
Plasma total cholesterol (mg/dL)
Parental history of myocardial infarction
at 60 y of age (%)
Exercise (%)
Rarely or never
1 time/wk
13 times/wk
4 times/wk
Alcohol consumption (%)
Rarely or never
13 drinks/mo
16 drinks/wk
1 drinks/d
Postmenopausal hormone use (%)
Never
Former
Current
Fruit and vegetable intake (servings/d)
Other dietary factors3
Lycopene intake (g/d)
Total fiber intake (g/d)
Folate intake (g/d)
Saturated fat intake (g/d)
1 (n 121)
11.7 g/dL
2 (n 121)
11.716.4 g/dL
3 (n 121)
16.4 21.0 g/dL
4 (n 120)
21.0 g/dL
60.0 8.92
59.0 8.0
58.7 8.7
57.5 7.9
43.0
30.6
26.4
26.4 5.1
34.2
5.0
30.6
188 34
43.0
35.5
21.5
25.5 4.3
32.2
2.5
32.2
197 38
45.5
34.7
19.8
26.0 4.9
32.8
1.7
32.5
203 34
37.5
40.8
21.7
25.2 4.2
30.8
1.7
41.7
209 35
17.0
12.7
13.2
13.4
46.3
17.3
27.3
9.1
35.5
24.0
28.9
11.6
43.8
14.9
30.6
10.7
45.0
15.0
32.5
7.5
57.0
16.5
20.7
5.8
50.4
10.7
24.8
14.1
52.1
11.6
25.6
10.7
45.8
8.3
35.8
10.0
P1
0.16
0.66
0.17
0.31
0.05
0.04
0.001
0.24
0.58
0.09
0.55
45.0
15.8
39.2
6.3 4.0
42.2
14.1
43.8
6.3 3.8
40.5
10.7
48.8
5.8 3.0
39.2
19.2
41.7
6.0 3.1
0.73
8117 5654
18.9 5.5
395 183
20.1 5.3
8563 4652
19.8 6.6
447 263
19.1 4.6
9683 5440
19.1 5.4
392 194
20.2 4.8
9925 6233
19.1 5.8
423 207
20.6 4.9
0.03
0.65
0.16
0.15
ANOVA was used for continuous variables, Cochran-Armitage trend tests were used for dichotomous variables, and chi-square tests were used for
categorical variables.
2
x SD.
3
Energy adjusted by using the residual method (25).
Our prospective data indicate an apparent L-shaped association between increasing quartiles of plasma lycopene and a
lower risk of CVD in middle-aged and elderly women that
appears to be independent of major lifestyle, clinical, and
51
8.4
2
11.716.4 g/dL
3
16.4 21.0 g/dL
14.1
18.6
4
21.0 g/dL
P (linear
trend)
RR1
24.2
1.00 (referent)
1.00 (referent)
1.00 (referent)
1.00 (referent)
1.00 (referent)
1.00 (referent)
1.00 (referent)
1.00 (referent)
0.004
0.18
0.05
1.00 (referent)
1.00 (referent)
1.00 (referent)
0.006
0.36
0.15
1
For total CVD, post hoc tests compared quartiles 3 and 4 versus quartiles 1 and 2 of plasma lycopene; for important vascular events, post hoc tests
compared quartiles 2 4 versus quartile 1 of plasma lycopene.
2
n values for cases and controls, respectively, were as follows: quartiles 1 (151 and 121), 2 (124 and 121), 3 (97 and 121), and 4 (111 and 120).
3
Matched for age and smoking status and adjusted for randomized aspirin treatment, randomized vitamin E treatment, randomized -carotene
treatment, and plasma cholesterol concentration.
4
Adjusted for the covariates in footnote 3 plus BMI, exercise, postmenopausal hormone use, parental history of MI at 60 y of age, diabetes,
hypertension, and high cholesterol.
5
Adjusted for the covariates in footnote 4 plus alcohol, fiber, folate, saturated fat, and fruit and vegetable intakes.
6
Other carotenoids and retinol were added separately into the model.
7
n values for cases and controls, respectively, were as follows: quartiles 1 (109 and 79), 2 (85 and 97), 3 (69 and 77), and 4 (76 and 86).
Although most research on lycopene has focused on a potential reduction in the risk of prostate cancer (26), limited
epidemiologic data support a preventive effect of lycopene
against CVD. The broad range of plasma lycopene concentrations in the WHS is at the low end of the range of lycopene
measurements reported in previous observational studies (9,
10, 2735). This may reflect differences in assay methods, diet,
and baseline disease status or between serum and plasma
concentrations of lycopene across study populations. This
raises questions about the absence of any specific evidence of
TABLE 4
Multivariate relative risks (RRs) and 95% CIs of cardiovascular disease (CVD) in a comparison of quartiles of plasma lycopene with other plasma
carotenoids and retinol in separate models
Quartile of plasma biomarker
1
Plasma lycopene (g/dL)
Multivariate-adjusted RR2
Plasma -cryptoxanthin (g/dL)
Multivariate-adjusted RR2
Plasma lutein/zeaxanthin (g/dL)
Multivariate-adjusted RR2
Plasma retinol (g/dL)
Multivariate-adjusted RR2
1
8.4
1.00 (referent)
3.8
1.00 (referent)
9.4
1.00 (referent)
37.1
1.00 (referent)
14.1
0.94 (0.60, 1.49)
7.5
0.67 (0.41, 1.07)
14.0
0.82 (0.51, 1.33)
48.2
1.36 (0.83, 2.24)
18.6
0.62 (0.39, 1.00)
11.8
0.95 (0.58, 1.56)
18.5
0.89 (0.55, 1.44)
57.8
1.13 (0.65, 1.94)
24.2
0.67 (0.41, 1.11)
19.3
0.70 (0.41, 1.18)
26.5
0.81 (0.47, 1.41)
71.4
1.22 (0.69, 2.14)
P (linear trend)
0.05
0.34
0.58
0.69
52
SESSO ET AL
stability of plasma lycopene and other carotenoid measurements support the stability of the biochemical markers used in
this study (44, 45). Third, because we observed a potential
threshold effect in reducing the risk of CVD, we performed
post hoc tests that compared the upper with the lower half of
plasma lycopene concentrations for the risk of total CVD and
the upper 3 quartiles with the lowest quartile for CVD (exclusive of angina). Although such tests yielded results that support
the possibility of a threshold effect, they must be interpreted
with caution because we had not anticipated the observed
pattern of RRs that emerged in our results.
Next, we relied on a composite endpoint for total CVD that
included MI, stroke, revascularization, CVD death, and angina.
Heterogeneity in the results by specific CVD endpoints cannot
be ruled out because we had limited power to assess any single
outcome. The exclusion of angina case-control pairs strengthened the inverse association, which suggested that plasma
lycopene may prevent more serious CVD outcomes. Another
potential limitation was that we lacked information on whether
-carotene or vitamin E treatment affected carotenoid concentrations. However, because the randomized treatment status
was evenly distributed within each quartile of plasma lycopene,
potential confounding was minimized. Finally, residual confounding may also be of concern, as it would in any epidemiologic investigation. However, in addition to matching for age,
smoking, and follow-up time, we comprehensively controlled
for other coronary disease risk factors.
In conclusion, higher plasma lycopene concentrations may
be associated with a lower risk of CVD in middle-aged and
elderly women. Alternatively, plasma lycopene may simply be
a marker for another dietary or nutritional component that
travels with lycopene in foods. It remains unclear whether the
inverse association between plasma lycopene and CVD risk
follows a strict linear dose-response trend or is L-shaped, as
observed in the present study. Further research on the potential
mechanistic actions of lycopene in the prevention of CVD and
other chronic diseases is warranted.
We acknowledge the crucial contributions of the entire staff of the
WHS, including the leadership of David Gordon, Susan Burt, Mary Breen,
Marilyn Chown, Lisa Fields-Johnson, Georgina Friedenberg, Inge Judge,
Jean MacFadyen, Geneva McNair, David Potter, Claire Ridge, and Harriet
Samuelson. We are also indebted to the 39 876 dedicated and committed
participants of the WHS.
HDS helped conceive and design the study, analyze and interpret the
data, draft the article, and obtain funding. JEB helped critically revise the
article, collect and assemble the data, and obtain funding. EPN helped
critically revise the article and collect and assemble the data. JMG helped
conceive and design the study, interpret the data, critically revise the
article, and obtain funding. The authors had no financial or personal
interests related to this research.
REFERENCES
1. Arab L, Steck S. Lycopene and cardiovascular disease. Am J Clin Nutr
2000;71(suppl):1691S5S.
2. Clinton SK. Lycopene: chemistry, biology, and implications for human health and disease. Nutr Rev 1998;56:3551.
3. Fuhrman B, Elis A, Aviram M. Hypocholesterolemic effect of lycopene and beta-carotene is related to suppression of cholesterol synthesis and augmentation of LDL receptor activity in macrophages. Biochem Biophys Res Commun 1997;233:658 62.
4. Agarwal S, Rao AV. Tomato lycopene and low density lipoprotein
oxidation: a human dietary intervention study. Lipids 1998;33:981 4.
5. Bub A, Watzl B, Abrahamse L, et al. Moderate intervention with
6.
7.
8.
9.
10.
11.
12.
13.
14.
15.
16.
17.
18.
19.
20.
21.
22.
23.
24.
25.
53