点击显示 收起
1 From the Childrens Hospital, Department of Medicine, Boston (SKO); the Department of Nutrition, Harvard School of Public Health, Boston (MJS, ER, and WCW); the Department of Epidemiology, Harvard School of Public Health, Boston (MJS, ER, DS, JEM, and WCW); the Channing Laboratory, Department of Medicine, Brigham and Womens Hospital and Harvard Medical School, Boston (MJS, ER, DS, JEM, and WCW); the Department of Biostatistics, Harvard School of Public Health, Boston (DS); and the Division of Preventive Medicine, Department of Medicine, Brigham and Womens Hospital and Harvard Medical School, Boston (JEM).
2 Supported by research grants HL24074, HL34594, HL60712, and CA87969 from the National Institutes of Health. 3 Address reprint requests to SK Osganian, Clinical Research Program, Childrens Hospital, 300 Longwood Avenue, Boston, MA 02115. E-mail: stavroula.osganian{at}tch.harvard.edu.
ABSTRACT
Background: Numerous studies have shown that higher intakes or higher blood concentrations of carotenes are associated with a lower risk of coronary artery disease (CAD). Given the null results in trials of ß-carotene supplementation, considerable attention has focused on the potential role of other dietary carotenoids in the prevention of CAD.
Objective: Our objective was to prospectively examine the relation between dietary intakes of specific carotenoids and risk of CAD in women.
Design: In 1984, 73 286 female nurses completed a semiquantitative food-frequency questionnaire that assessed their consumption of carotenoids and various other nutrients. The women were followed for 12 y for the development of incident CAD (nonfatal myocardial infarction and fatal CAD), and dietary information was updated in 1986, 1990, and 1994.
Results: During 12 y of follow-up (803 590 person-years), we identified 998 incident cases of CAD. After adjustment for age, smoking, and other CAD risk factors, we observed modest but significant inverse associations between the highest quintiles of intake of ß-carotene and -carotene and risk of CAD but no significant relation with intakes of lutein/zeaxanthin, lycopene, or ß-cryptoxanthin. For women in the highest compared with the respective lowest quintile of intake, the relative risks for ß-carotene and -carotene were 0.74 (95% CI: 0.59, 0.93) and 0.80 (95% CI: 0.65, 0.99), respectively. The association between the specific carotenoids and CAD risk did not vary significantly by current smoking status.
Conclusion: Higher intakes of foods rich in -carotene or ß-carotene are associated with a reduction in risk of CAD.
Key Words: Carotenoids antioxidants coronary artery disease diet Nurses Health Study women
INTRODUCTION
Considerable attention has focused on the role of dietary carotenoids in the prevention of coronary artery disease (CAD) (1). The carotenoids consist of > 500 plant-derived compounds, with lutein, ß-cryptoxanthin, lycopene, -carotene, and ß-carotene accounting for most of those found in plasma (24). As lipid-soluble compounds, carotenoids are incorporated into lipoprotein particles during transport in plasma and are stored primarily in the liver or adipose tissue (3). In vitro studies have shown that carotenoids are efficient quenchers of singlet oxygen (5), directly scavenge free radicals (4, 6), and inhibit lipid peroxidation (79). They exhibit considerable differences in the rate of quenching singlet oxygen, with lycopene having the strongest quenching ability of those studied, followed by -carotene, ß-carotene, zeaxanthin, lutein, and ß-cryptoxanthin (10, 11).
Carotenoids may prevent the development of atherosclerosis by directly inhibiting the oxidation of LDL (1216). In vitro, the direct addition of ß-carotene to LDL inhibits agonist-initiated oxidation of LDL (9, 17, 18) and decreases LDL degradation by macrophages (9, 17). Some studies have reported significant decreases in markers of lipid peroxidation or LDL oxidation after supplementation with lutein in mice (19) and after supplementation with ß-carotene (2023) or dietary lycopene (24) in humans. However, most studies examining the in vitro effects of ß-carotene on LDL oxidation after supplementation of human volunteers showed no effect (25).
Alternatively, carotenoids may protect vascular cells from oxidative injury and preserve vascular function through tissue-specific actions that are independent of the direct inhibition of LDL oxidation (12, 25). Recent data provide support for a protective role of lutein in the development or progression of atherosclerosis. Dwyer et al (19) found the formation of atherosclerotic lesions in the aortic arch to be significantly lower in lutein-supplemented apolipoprotein E null mice than in controls. Additionally, in a cohort of men and women, plasma lutein concentrations at baseline were inversely associated with carotid intima-media thickness progression over 18 mo (19).
In general, observational studies have found inverse associations between blood concentrations or dietary intakes of carotenes, primarily ß-carotene, and risk of CAD (2633). Null results from trials of ß-carotene supplementation (30, 3436) suggest that other nutrients in ß-carotenerich foods, perhaps other carotenoids, may be responsible for the apparent benefit. However, few studies have examined the association of other carotenoids and risk of CAD (28, 37), and none have assessed dietary intakes of the specific carotenoids. We therefore prospectively examined the relation between dietary intakes of specific carotenoids and risk of nonfatal and fatal CAD in a large cohort of US women during 12 y of follow-up.
SUBJECTS AND METHODS
Study population
The Nurses Health Study was initiated in 1976 when 121 700 female registered nurses aged 3055 y and residing in 11 large US states completed a mailed questionnaire concerning their medical history and lifestyle (38). Follow-up questionnaires have been sent every 2 y thereafter to ascertain information on potential risk factors and to identify newly diagnosed cases of CAD and other diseases. In 1980 a 61-item semiquantitative food-frequency questionnaire (SFFQ) was developed to assess intakes of micronutrients and other components of diet. In 1984 the SFFQ was expanded to include 116 items, and similar questionnaires were used to measure diet in 1986 and 1990. Because the expanded version of the SFFQ contained additional food items that are important for the assessment of specific carotenoid intakes, we considered 1984 as the baseline for this analysis. The Nurses Health Study was approved by all applicable institutional review boards.
A total of 81 757 women returned the 1984 diet questionnaire. A priori, we excluded respondents who had implausibly high (> 3500 kcal/d) or low (< 600 kcal/d) total energy intakes or those who left ≥ 11 food items blank. We further excluded women with cancer (except nonmelanoma skin cancer; n = 4313) or cardiovascular disease [ie, angina, myocardial infarction (MI), stroke, or other cardiovascular diseases; n = 3127] diagnosed before 1984. The final baseline population consisted of 73 286 women. More than 90% responded to the subsequent biennial questionnaires, and 80% completed the diet questionnaires at each follow-up assessment.
Ascertainment of diet and other exposures
Dietary intakes of carotenoids and other aspects of diet were measured by using the SFFQs administered in 1984, 1986, 1990, and 1994. For each food, a commonly used unit or portion size (eg, 1 egg or 1 slice of bread) was specified, and the participant was asked how often, on average, during the previous year she had consumed that amount. Nine responses were possible, ranging from "never or less than one per month" to "six or more times per day." The intake of specific nutrients was computed by multiplying the frequency of consumption of each unit of food by the nutrient content of the specified portion. Food-composition values were obtained from the Harvard University Food Composition Database derived from US Department of Agriculture (USDA) sources (39) and supplemented with information from manufacturers and other published values. The measurement of glycemic load in the Nurses Health study was reported elsewhere (40). Briefly, we calculated the glycemic load by multiplying the carbohydrate content of each food by its glycemic index and then multiplied this value by the frequency of consumption and summed the values for all foods. Each unit of dietary glycemic load represents the equivalent of 1 g carbohydrate from white bread.
We computed nutrient intakes for the individual carotenoids by using the USDAs new carotenoid database of > 2400 fruit, vegetables, and selected multicomponent foods (33, 41). Lutein and zeaxanthin intakes are presented together because the analytic procedures did not permit separate quantification of these carotenoids in foods. The carotenoid content of tomato-based food products was updated with values from the USDA, which were recently derived from reversed-phase HPLC (42). In our data, the foods providing the greatest contribution to the total absolute nutrient intake of the specific carotenoids were as follows: carrots (96%) for -carotene; carrots (36%), spinach (37%), and tomato products (6%) for ß-carotene; tomato products (100%) for lycopene; oranges (56%), orange juice (28%), and peaches (10%) for ß-cryptoxanthin; and spinach (79%), broccoli (5%), and peas (4%) for lutein/zeaxanthin. The intake of ß-carotene included both dietary and supplemental sources. Spearman correlations between the individual carotenoids ranged from 0.16 to 0.78. Moderate to strong correlations were observed between -carotene and ß-carotene (0.78), lutein and ß-carotene (0.70), lutein and -carotene (0.37), and lycopene and ß-carotene (0.33). Weaker correlations were observed for lycopene with -carotene (0.17), ß-cryptoxanthin (0.16), and lutein/ zeaxanthin (0.30) and for ß-cryptoxanthin with -carotene (0.22), ß-carotene (0.28), and lutein/zeazanthin (0.25).
A detailed description of the dietary questionnaires and documentation of their reproducibility and validity are published elsewhere (4346). The questionnaires provided a reasonable measure of dietary intake of macronutrients and micronutrients when compared with multiple 1-wk dietary records. In 1986 a biochemical validation study among a random sample of cohort members compared the correlation between average nutrient intake of specific carotenoids from 2 FFQs administered 1 y apart and plasma concentrations of carotenoids from a blood specimen collected before the second FFQ (47). Among nonsmoking women (n = 162), the adjusted diet-plasma correlations were 0.48 for -carotene, 0.27 for ß-carotene or lutein, 0.32 for ß-cryptoxanthin, and 0.21 for lycopene; all correlations were statistically significant (P < 0.05).
The biennial mailed questionnaire collected information on several demographic, behavioral, and clinical variables proven or suspected to be associated with CAD. These included self-reported height and weight; current cigarette smoking; menopausal status; postmenopausal hormone use; history of hypertension, diabetes, or high cholesterol; parental history of MI before the age of 60 y; current aspirin use; and average number of hours per week spent in various types of recreational physical activity. During each cycle, we asked the participants to provide information on the use of multiple vitamins, the specific brand, and the usual number of tablets taken per week. Information was also collected on the use of specific supplements including ß-carotene, vitamin E, and vitamin C, together with the dose of tablets and the usual number of tablets taken per week.
Ascertainment of endpoint
The primary endpoint for this analysis was incident CAD, which included any nonfatal MI or fatal coronary disease that occurred after the return of the 1984 questionnaire but before 1 June 1996. For each woman, only the first event was considered. Women who reported a diagnosis of CAD on a biennial follow-up questionnaire were asked for permission to examine their medical records. Study physicians with no knowledge of the womens self-reported risk factor status reviewed the records to verify the occurrence of an event. An MI was considered confirmed if it met the World Health Organization criteria of symptoms and either typical electrocardiographic changes or elevated cardiac enzymes (48). Infarctions of indeterminate age were excluded. Infarctions that required hospital admission and for which confirmatory information was obtained by interview or letter but for which no medical records were available were designated as probable. These events constituted 13% of all nonfatal MI cases. We included all confirmed and probable cases in the analyses because results were not substantially different after excluding probable cases (data not shown).
Deaths were identified from state vital records and the National Death Index or were reported by the womens next of kin and the postal system. We estimate that the follow-up for deaths was > 98% complete (49). Fatal CAD was confirmed by hospital records or autopsy or if CAD was listed as the cause of death on the death certificate and evidence of previous CAD was available. We designated as presumed fatal CAD those cases in which CAD was the underlying cause on the death certificate but for which no records were available. These cases constituted 25% of fatal CAD cases. We also included sudden death within 1 h of onset of symptoms in women with no other plausible cause (other than coronary disease), which constituted 17% of fatal CAD cases. Analyses limited to confirmed cases yielded similar, although less precise, results (data not shown).
Statistical analyses
The person-time for each participant was calculated from the date of return of the 1984 diet questionnaire to the date of the first CAD event, death, or 1 June 1996. For dietary exposures, women were categorized into 5 equal groups according to energy-adjusted nutrient intakes. Nutrients were energy-adjusted by using the residual method (46). We used the cumulative average of repeated measures of diet to reduce the within-person variation in diet and to more precisely represent long-term intakes (50). Specifically, the cumulative average intake of each nutrient was calculated from all dietary questionnaires available up to the start of every 2-y follow-up interval. For example, the intake reported on the 1984 questionnaire was related to the incidence of CAD from 1984 through 1986, and the average intake reported on the 1984 and 1986 questionnaires was related to the incidence of CAD from 1986 to 1988. When dietary data were missing at the start of a new interval, we carried forward the last available dietary data. Incidence rates were calculated by dividing the number of events by the person-time of follow-up in each quintile. The relative risk of a coronary event was computed as the incidence rate in a specific category of energy-adjusted intake of specific carotenoids divided by the rate in the lowest category of intake, with adjustment for 5-y age categories.
We used pooled logistic regression (51) with 2-y follow-up periods, which is asymptotically equivalent to the Cox-proportional-hazard regression, to model the incidence of CAD in relation to the cumulative average of carotenoid intake and to adjust simultaneously for potential confounding variables, including age (5-y intervals), body mass index (in kg/m2: < 21, 2122.9, 2324.9, 2528.9, or ≥ 29), smoking status (never, past, or current smoking of 114, 1524, or ≥ 25 cigarettes/d), alcohol consumption (0, 14, 514, or ≥ 15 g/d), menopausal status (premenopausal, postmenopausal without hormone replacement, postmenopausal with past hormone replacement, or postmenopausal with current hormone replacement), aspirin use (nonuser, 16/wk, ≥ 7/wk, or unknown dose), moderate to vigorous physical activity(< 1, 11.9, 23.9, 46.9, or ≥ 7 h/wk), parental history of MI before the age of 60 y, and history of diabetes, hypercholesterolemia, or hypertension. In multivariate models (46), dietary covariates included quintiles of energy-adjusted total intakes of folate, vitamin B-6, saturated fat, polyunsaturated fat, trans unsaturated fats, cereal fiber, vitamin E, and vitamin C as well as quintiles of total energy intake and dietary glycemic load. All nutrients were energy-adjusted by using the residual method (46). Most nondietary covariates were ascertained and updated every 2 y. Aspirin use was ascertained in 1984, 1988, and 1994. Parental history of MI was ascertained in 1976 and 1984, and physical activity was ascertained in 1982 and every 2 y from 1986 to 1992. Missing data on covariates were included as indicator variables. All models also included an indicator variable for each 2-y time period during the study (6 periods). Tests of linear trend across increasing quintiles of carotenoid intake were conducted by assigning the medians of intake to categories that were treated as a continuous variable in the regression. All P values are 2-sided. A P value < 0.05 was considered statistically significant.
To avoid a spurious finding due to the potential influence of disease or prodromal illness on womens diet, women with prior diagnoses of cardiovascular disease or cancer (except nonmelanoma skin cancer) were excluded at baseline, and women with new diagnoses were censored at the time of diagnosis during follow-up. The development of intermediate endpoints such as hypercholesterolemia, diabetes, and hypertension may also lead to changes in diet that confound the associations between diet and disease. Therefore, in secondary analyses, we excluded women with prior diagnoses of hypertension, hypercholesterolemia, or diabetes at baseline and stopped updating information on diet at the beginning of the interval during which those conditions were diagnosed. Stratified analyses were conducted to examine potential interactions between the intake of a specific carotenoid and smoking status (never, former, or current smoker), alcohol intake (nondrinker, 110 g/d, or ≥ 10 g/d), tertiles of intake of polyunsaturated fatty acids, or intake of other antioxidants (tertiles of total vitamin E intake or total vitamin C intake).
RESULTS
During 12 y of follow-up (803 590 person-years) from 1984 to 1996, we identified 998 incident cases of CAD: 718 nonfatal MIs (625 confirmed nonfatal cases) and 280 deaths from coronary disease (145 confirmed CAD deaths). The distribution of demographic, clinical, behavioral, and dietary risk factors for CAD according to quintile of ß-carotene intake at baseline is shown in Table 1. Participants with higher intakes of ß-carotene were less likely to smoke cigarettes and were more likely to exercise and take multivitamin, vitamin C, vitamin E, and ß-carotene supplements and postmenopausal hormones than were those with lower intakes. Those with higher intakes of ß-carotene also consumed more folate, vitamin B-6, cereal fiber, and vitamins E and C and somewhat less saturated and trans unsaturated fat. Intakes of each of the other carotenoids increased with increasing intake of ß-carotene.
View this table:
TABLE 1 . Characteristics of the cohort according to quintile of ß-carotene intake at baseline (1984)1
The relative risk of CAD according to quintile of intake of each specific carotenoid is shown in Table 2. Compared with women in the respective lowest quintile of intake and after adjustment for age, we observed a moderate-to-strong statistically significant reduction in the risk of total CAD among women in all categories of ß-carotene and lutein/zeaxanthin intake and in the 3 uppermost quintiles of -carotene intake as well as a significant inverse trend with increasing intakes of these carotenoids. The relative risks were substantially attenuated but significant after adjustment for smoking. In multivariate models adjusted for age, smoking, and other coronary disease risk factors, a modest but significantly lower risk remained for women in the highest quintile of intake of ß-carotene and the fourth and fifth quintile of intake of -carotene, whereas the relative risks for lutein/zeaxanthin were further attenuated and no longer significant. An inverse trend with increasing intakes of ß-carotene and -carotene remained significant (P values for trend = 0.05 and 0.04, respectively). For women in the highest compared with the respective lowest quintile of intake, the multivariate relative risks were 0.74 (95% CI: 0.59, 0.93) for ß-carotene, 0.80 (95% CI: 0.65, 0.99) for -carotene, and 0.90 (95% CI: 0.72, 1.12) for lutein/zeaxanthin. We were unable to determine the independent association between risk of CAD and intakes of -carotene and ß-carotene in simultaneous models because of to the strong correlation between these 2 nutrients (r = 0.78).
View this table:
TABLE 2 . Adjusted relative risks and 95% CIs for total coronary artery disease (CAD: nonfatal myocardial infarction and fatal CAD) according to quintile of specific carotenoid intake1
In age-adjusted analyses, the relative risks for CAD and intakes of lycopene or ß-cryptoxanthin were suggestive of an inverse association in some quintiles of intake; however, there was no significant inverse trend with increasing intakes across quintiles. The relative risks were attenuated and remained not significant in multivariate models that were further adjusted for a variety of other coronary disease risk factors, and there was a somewhat increased risk of CAD in some categories of intake of ß-cryptoxanthin. For women in the highest compared with the respective lowest quintile of intake, the multivariate relative risks were 0.93 (95% CI: 0.77, 1.14) for lycopene and 1.17 (95% CI: 0.94, 1.44) for ß-cryptoxanthin. The relative risks did not change appreciably in multivariate models also adjusted for vitamin C and vitamin E intakes for any of the carotenoids (data not shown). In addition, results from more parsimonious models, which included only the major physiologic risk factors (smoking, physical activity, and dietary fat intake) as covariates, did not differ appreciably from the multivariate model for any of the carotenoids (data not shown). In secondary analyses that excluded women with hypertension, hypercholesterolemia, or diabetes at baseline and that stopped updating diet for new diagnoses during follow-up, the results were not substantially different except for some attenuation of the increased relative risk associated with intake of ß-cryptoxanthin (data not shown).
To examine potential interactions between intakes of carotenoids and smoking, which increases oxidative stress, we conducted analyses stratified by smoking status and selected categories of alcohol intake. As shown in Tables 37, we found no significant variation in risk of CAD and intakes of any of the individual carotenoids according to smoking status (never, former, or current). Similarly, there was no significant variation in risk of CAD in analyses stratified by intake of alcohol (nondrinker, 110 g/d, or ≥ 10 g/d; data not shown). Others have shown stronger inverse associations between adipose tissue concentrations of ß-carotene and risk of CAD among those with higher concentrations of polyunsaturated fatty acids (37); however, we found no variation in risk for any of the individual carotenoids according to tertiles of polyunsaturated fat intake (data not shown).
View this table:
TABLE 3 . Multivariate relative risks and 95% CIs for risk of coronary artery disease (CAD: nonfatal myocardial infarction and fatal CAD) according to quintile of ß-carotene intake stratified by smoking status1
View this table:
TABLE 7 . Multivariate relative risks and 95% CIs for risk of coronary artery disease (CAD: nonfatal myocardial infarction and fatal CHD) according to quintile of ß-cryptoxanthin intake stratified by smoking status1
Carotenoids may act synergistically with other antioxidants (vitamin E or C), suggesting that greater beneficial effects may be observed at higher intakes of other antioxidants (9). Alternatively, carotenoids may serve as a secondary antioxidant defense of LDL, after depletion of vitamin E, suggesting that protective effects may only be observed among those with lower intakes of vitamin E (52). In analyses stratified according to tertiles of total vitamin C intake or vitamin E intake, we observed no effect modification for any of the specific carotenoids (data not shown).
DISCUSSION
In this large prospective study of women, we observed a modest but significant inverse association between higher intakes of ß-carotene and -carotene and the incidence of nonfatal MI and fatal CAD. Compared with women in the respective lowest quintile of intake, women in the highest quintile of ß-carotene intake had a 26% lower risk of CAD and women in the highest quintile of -carotene intake had a 20% lower risk of CAD. We did not observe a significant inverse association between high intakes of lutein/zeaxanthin, lycopene, or ß-cryptoxanthin and risk of CAD. Furthermore, there were no marked differential associations by current smoking status for any of the specific carotenoids.
Similar to our findings, previous prospective investigations have observed an inverse association between plasma or dietary ß-carotene and risk of CAD (2832). Street et al (28) observed a 55% lower risk of MI for men and women in the highest compared with the lowest quintile of plasma ß-carotene (RR = 0.45; 95% CI: 0.22, 0.90). However, in stratified analyses, this was observed only among smokers. In the multicenter Skin Cancer Prevention Study, men and women in the highest versus the lowest quartile of plasma ß-carotene had a 43% lower risk of death from cardiovascular disease 10 y later (RR = 0.57; 95% CI: 0.34, 0.95) (30). With respect to dietary intakes, only one study observed a significantly lower risk of MI (RR = 0.55, 95% CI: 0.34, 0.83) for men and women in the highest (> 1.57 mg/d) compared with the lowest (< 1.13 mg/d) tertile of ß-carotene intake (29). Two studies observed nonsignificant, inverse associations between dietary intakes of ß-carotene and fatal or nonfatal events (31, 32). In the previously described studies and others, effect modification by smoking status has been inconsistent (28, 29, 31, 32, 37, 53, 54).
Despite favorable findings for ß-carotene in observational studies, randomized clinical trials of ß-carotene supplementation have shown no benefit on CAD. The Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study found a small increase in overall mortality among men randomly assigned to 20 mg ß-carotene/d compared with placebo (36). In subsequent subgroup analyses, Virtamo et al (55) found no effect on risk of fatal and nonfatal CAD among men without a prior MI at baseline, and Rapola et al (56) found no effect on the combined outcome of fatal and nonfatal events among men with a history of MI at baseline. However, in the latter subgroup, there was an increased risk of fatal events (RR = 1.75; 95% CI: 1.16, 2.64) and a nonsignificant protective effect for nonfatal events (RR = 0.67; 95% CI: 0.44, 1.02). Three other trials, the Beta-Carotene and Retinol Efficacy Trial (57), the Physicians Health Study (35), and the Skin Cancer Prevention Study (30) showed no benefit for cardiovascular events.
The lack of a protective effect of ß-carotene supplementation suggests that other nutrients found in foods rich in ß-carotene, perhaps other dietary carotenoids, may account for the apparent benefit observed in epidemiologic investigations (54). Studies of this theory have been few and inconclusive. Both the Multiple Risk Factor Intervention Trial (58) and the Lipid Research Clinics Coronary Primary Prevention Trial (59) observed a modestly lower risk of nonfatal MI and fatal CAD among men in the highest compared with the lowest quartile of total plasma carotenoids, although the difference was not statistically significant in the former study. Street et al (28) observed no significant inverse trend between risk of MI and serum lycopene, lutein, or zeaxanthin. However, low plasma concentrations of each carotenoid were associated with a significantly increased risk of MI among smokers. In the EURAMIC Study (37), adipose tissue concentrations of lycopene and ß-carotene were inversely and independently associated with risk of nonfatal MI [RR = 0.54 (95% CI: 0.29, 0.98) and RR = 0.47 (95% CI: 0.26, 0.99) for the fifth compared with the lowest quintile, respectively]. Only 3 studies examined dietary intakes of carotenoids. Sahyoun et al (60) found a nonsignificant inverse association between intakes of total carotenoids and risk of death from CAD (RR = 0.64; 95% CI: 0.33, 1.27) among men and women. Kushi et al (61) found no association between risk of fatal CAD and intake of total carotenes (RR = 1.03; 95% CI: 0.63, 1.70 for the highest compared with the lowest quintile) among women. In contrast, in the Health Professionals Cohort, men in the highest compared with the lowest quintile of total carotene intake had a 29% lower risk of nonfatal and fatal CAD (RR = 0.71; 95% CI: 0.55, 0.92) (62).
Some investigators have suggested that a benefit from ß-carotene may only be observed in populations with relatively low intakes (54, 63). Therefore, high doses of ß-carotene supplementation may have no additional benefit in well-nourished individuals, which may explain the lack of benefit observed in the clinical trials. In accordance with this, experimental studies of healthy volunteers placed on a short-term carotenoid-depleted diet have shown significant increases in markers of LDL oxidation or lipid peroxidation followed by a decrease in these markers after repletion with ß-carotene (20, 22). In our data, the risk of CAD did not decrease above the third quintile of ß-carotene intake. Our findings may be compatible with a benefit from ß-carotene supplementation among populations with low intakes.
Other dietary factors found in fruit or vegetables rich in carotenes may account for the apparent benefit we observed. Higher blood concentrations of carotenoids are found in persons who consume large amounts of vegetables and fruit (64), and greater consumption is associated with a decreased risk of CAD (65, 66). Gaziano et al (67) found a strong apparent protective effect on risk of cardiovascular disease mortality (RR = 0.59; 95% CI: 0.37, 0.94) for men and women in the upper compared with the lower quartile of intake of fruit and vegetables also high in carotenoids, including carrots or squash and salads or green leafy vegetables. Similarly, Sahyoun et al (60) found a significant inverse association between CAD mortality and higher intakes of dark-green vegetables (RR = 0.43; 95% CI: 0.20, 0.92), whereas Knekt et al (31) observed an apparent protective effect for higher intakes of vegetables that are important food sources of antioxidant vitamins. Other potentially protective dietary factors include vitamins C and E, folate, and fiber. However, in our data, after adjustment for several of these nutrients, a modest inverse association between CAD and ß-carotene or -carotene remained.
The strengths of our study include its prospective design and high rates of follow-up. The prospective collection of information eliminates the possibility of biased recall of diet. High rates of cohort participation lessen the potential for differential losses that threaten internal validity. Furthermore, the large number of events provide adequate statistical power to detect modest reductions in the relative risk CAD. Nevertheless, we cannot exclude the possibility of residual confounding by an unmeasured or imperfectly measured risk factors in this study. Women with higher intakes of carotenoids from diet had somewhat healthier risk profiles; however, we controlled for several known physiologic and lifestyle factors. Controlling for cigarette smoking produced a substantial attenuation in the observed age-adjusted relative risks. Cigarette smokers have been shown to eat fewer fruit and vegetables and have lower intakes and plasma concentrations of antioxidant vitamins than do nonsmokers (68, 69). The diagnoses of various self-reported conditions were validated by review of medical records and direct measurement (70).
Our findings suggest that higher intakes of foods rich in -carotene and ß-carotene may contribute to a reduction in risk of CAD. -Carotene may be responsible for the inverse association observed between risk of CAD and consumption of carotenoid-rich foods. Epidemiologic investigations of other individual carotenoids are necessary to further explore the role of these compounds in the prevention of CAD. At this time, greater consumption of fruit and vegetables remains the most appropriate public health policy recommendation (71, 72).
View this table:
TABLE 4 . Multivariate relative risks and 95% CIs for risk of coronary artery disease (CAD: nonfatal myocardial infarction and fatal CAD) according to quintile of -carotene intake stratified by smoking status1
View this table:
TABLE 5 . Multivariate relative risks and 95% CIs for risk of coronary artery disease (CAD: nonfatal myocardial infarction and fatal CAD) according to quintile of lutein/zeaxanthin intake stratified by smoking status1
View this table:
TABLE 6 . Multivariate relative risks and 95% CIs for risk of coronary artery disease (CAD: nonfatal myocardial infarction and fatal CAD) according to quintile of lycopene intake stratified by smoking status1
ACKNOWLEDGMENTS
We are indebted to the participants of the Nurses Health Study for their continuing, exceptional cooperation and to Frank Speizer, the founding principal investigator of the Nurses Health Study, for his invaluable support.
SKO contributed to the development of the analysis plan, conducted the statistical analyses, collaborated on the interpretation of the results, and wrote the manuscript. MJS, ER, and DS provided significant consultation on the statistical analysis plan, interpretation of results, and writing of the manuscript. JEM and WCW provided significant consultation on the interpretation of results and writing of the manuscript. None of the authors had any financial or personal interest in any company or organization sponsoring this research, including advisory board affiliations.
REFERENCES