点击显示 收起
1 From the Departments of Medicine and Community Medicine, Lund University, Malmö University Hospital, Malmö, Sweden.
2 Supported by the Swedish Cancer Society (2684-B93-05XAA), the Swedish Medical Research Council (B93-39X-09534-03C), the Ernhold Lundström Foundation, and the Swedish Nutrition Foundation. 3 Address reprint requests to P Wallström, The Malmö Diet and Cancer Study, Medical Research Centre, Malmö University Hospital, Malmö SE-205 02, Sweden. E-mail: peter.wallstrom{at}smi.mas.lu.se.
ABSTRACT
Background: Previous studies of associations between diet, obesity, and blood concentrations of -tocopherol and ß-carotene have been equivocal. Furthermore, most studies used only body mass index (BMI) as an obesity measure.
Objectives: Our objectives were to examine the associations between energy and nutrient intakes, alcohol consumption, tobacco use, and serum cholesterol and serum concentrations of -tocopherol and ß-carotene, and to examine the associations between different measures of general and central adiposity and serum concentrations of -tocopherol and ß-carotene.
Design: This was a cross-sectional, population-based study of 253 men and 276 women aged 4667 y. Nutrient data were collected by a modified diet history method. Measures of obesity included BMI, percentage of body fat (impedance analysis), waist-to-hip ratio, and waist circumference. The associations between serum nutrient concentrations and the other factors were examined by multiple linear regression.
Results: Twenty-one percent of men and 34% of women used antioxidant supplements. The mean BMI was 26.1 in men and 25.4 in women. Serum ß-carotene concentration was positively associated with serum cholesterol concentration, fiber intake, and ß-carotene intake, and negatively associated with smoking and all measures of obesity. In men, serum ß-carotene concentration was not significantly associated with central adiposity after adjustment for body fat. Serum -tocopherol concentration was positively correlated with serum cholesterol, obesity, and vitamin E intake. In women, serum -tocopherol concentration was also positively associated with intakes of ascorbic acid and selenium. Serum -tocopherol concentration was associated with central adiposity after adjustment for body fat.
Conclusion: Serum ß-carotene and -tocopherol concentrations have different associations with diet, smoking, general adiposity, and central adiposity.
Key Words: Vitamin E ß-carotene -tocopherol diet obesity body composition body mass index waist-to-hip ratio smoking alcohol cross-sectional studies
INTRODUCTION
The antioxidants -tocopherol, the dominant component of vitamin E, and ß-carotene contribute to the body's defense against reactive oxygen species (15). High blood concentrations of these nutrients were prospectively shown to be associated with low incidences of cardiovascular disease and several cancers (3, 68). Similar results, although less conclusive, were obtained in prospective studies of the dietary intake of these nutrients and the risk of cancer and cardiovascular disease (911).
It has been argued that low ß-carotene or -tocopherol serum concentrations may not indicate a causal link between these nutrients and disease but rather a relative absence of other protective dietary factors. This was suggested, particularly for ß-carotene, with the results of 3 randomized controlled trials in which relatively long-term ß-carotene supplementation in high-risk (7, 12) and low-risk (13) populations failed to produce any protective effects. Still, low blood concentrations of -tocopherol and ß-carotene are markers of disease risk. Moreover, carotenoids, such as ß-carotene, in plasma or serum may be regarded as markers of fruit and vegetable intake (14, 15) and there is evidence that a high consumption of fruits and vegetables protects against cancer and cardiovascular disease (16).
There are no straightforward relations between the measured average intake of ß-carotene and -tocopherol and ß-carotene and -tocopherol blood concentrations. First, the daily variation in blood concentrations is considerable (17, 18) and may be due in part to a variation in dietary intake. Second, all instruments used for measuring diet in large groups of individuals entail substantial measurement errors (19). Third, there are interindividual differences in biological properties, eg, differences in absorption, metabolism, and tissue distribution of nutrients (5, 2022). Other factors that may influence these associations include the intake of other nutrients, smoking (2330), alcohol use (2429, 31, 32), plasma lipid profile (23, 24, 27), and obesity.
Obesity, as assessed by body mass index (BMI), has been reported to be independently related to -tocopherol and ß-carotene concentrations, even when diet and lipid concentrations are adjusted for (23, 25, 2730, 32, 33). However, BMI, by definition, is a measure of weight adjusted for height, and thus can be assumed to be a good marker of obesity only when the variation between individuals is mostly because of differences in fat mass, not fat-free mass (34). Direct measures of body composition, such as body fat content or fat distribution [eg, waist-to-hip ratio (WHR)], may be useful in identifying the role of obesity in determining serum concentrations of -tocopherol and ß-carotene. Furthermore, measures of general and central adiposity are known to be associated differently with several chronic conditions, eg, myocardial infarction (35). One could hypothesize that the associations of central adiposity with blood nutrient concentrations may be different from those between general obesity and respective blood markers.
The first objective of this study from the Malmö Diet and Cancer Study (MDCS) cohort was to examine the association between energy and nutrient intakes, alcohol, tobacco use, and serum cholesterol and serum concentrations of -tocopherol and ß-carotene in a population-based sample of middle-aged and older men and women. Our second objective was to examine the associations between serum -tocopherol and ß-carotene and different measures of obesity, reflecting either general adiposity (BMI and percentage of body fat) or central adiposity (waist circumference and WHR).
SUBJECTS AND METHODS
Study population
The background population for this study consisted of all men and women born between 1926 and 1945 and living in Malmö (n = 53325 in 1991), Sweden's third largest city. The population was identified through the Swedish national population registries. A random sample of 1000 persons was selected from this population. The probands were invited by mail. Several attempts were made to reach each nonresponder by telephone. Five hundred forty-one persons completed the study. We obtained all relevant serum measurements from all but 11 subjects. One person was excluded because of an extremely high vitamin E intake, resulting in a serum -tocopherol value >150 µmol/L, which was 15 SDs beyond the mean of the other participants. Thus, the analytic sample for this study consisted of 529 subjects: 253 men and 276 women. Data were collected during 19911994. The design of the MDCS was approved by the Ethics Committee at Lund University.
Data collection
Study subjects visited the MDCS center twice. At the subject's first visit, project staff provided information on the background and aim of the project and detailed instructions about the lifestyle questionnaire and the dietary data collection procedure. Blood samples were drawn and processed. The subjects were not fasting. At the second visit, subjects were randomly assigned to interviewers who worked with each subject to complete the diet history and to check the lifestyle questionnaire for erroneous answers.
Dietary assessment
The MDCS dietary assessment method was described previously (36, 37). Briefly, it is a modified diet history method that combines quantitative and semiquantitative approaches and measures the entire diet, including cooking methods. It consists of 2 parts: a menu book to record descriptions of cooked meals, cold beverages (including juices and alcoholic beverages), and dietary supplements during 7 consecutive days, and a 168-item questionnaire on foods consumed regularly (other than cooked foods) during the past year. Data on the reproducibility (38) and validity (39, 40) of the method were published previously.
The amount of food and nutrients consumed was calculated with KOSTSVAR software (AIVO AB, Stockholm) with the use of the MDCS nutrient database, in which most of the nutrient information comes from the PC-KOST2-93 database (National Food Administration, Uppsala, Sweden). Names and contents of registered dietary supplements were provided by the Medical Products Agency (Uppsala, Sweden). Information on other dietary supplements was gathered from manufacturers, sales representatives, wholesale dealers, and retailers. Nutrient values of dietary supplements that could not be identified were replaced by the median intake from the type of supplement in question (3.8% of cases). If the supplement type could not be identified, the supplement was ignored (0.8% of cases). A person was considered a user of antioxidant supplements if he or she reported consuming a supplement that contained an antioxidant, ß-carotene, or selenium at least once during the registration.
The most common way of consuming vitamin E was in the form of multivitamin tablets. Such tablets typically contained 10 mg all-rac--tocopherol, which, according to current thinking, corresponds to 6.7 mg RRR--tocopherol (6.7 -tocopherol equivalents, -TE). Single-nutrient vitamin E preparations were uncommon but existed in amounts 100 mg all-rac--tocopherol (67 -TE). Use of ß-carotene supplements was uncommon during the time of this study.
Anthropometric examination and impedance analysis
Weight was measured to the nearest 0.1 kg on a balance-beam scale. Standing height was measured with a fixed stadiometer calibrated in centimeters. BMI was calculated as weight in kg divided by height in meters squared (kg/m2). Waist and hip circumferences of each participant were measured by a trained nurse (waist: midway between the lowest rib margin and the iliac crest; hip: horizontally at the level of the greatest lateral extension of the hips) and were used to calculate WHR. Both WHR and waist circumference were used as measures of central adiposity.
Bioelectrical impedance analysis was used to estimate body composition according to procedures provided by the manufacturer (BIA 103 single-frequency analyzer; RJL Systems, Detroit). The algorithm used to estimate body fat from impedance was supplied by the manufacturer and was used to calculate percentage of body fat.
Calculating the underreporting of energy intake
Basal metabolic rate (BMR) was calculated according to the formula of Garby et al (41):
RESULTS
Descriptive statistics
Participants are described in Table 1. More women than men had never smoked and more men were either ex- or current smokers. More women than men reported low or no alcohol intake, the women were more frequent users of antioxidant supplements, and the men were more likely to be overweight.
View this table:
TABLE 1. . Characteristics of the study population by sex
Dietary intakes and antioxidant serum concentrations of the participants are shown in Table 2. Women consumed less energy, dietary fiber, vitamin E, and selenium and more ascorbic acid and ß-carotene than did men. Women had 50% higher serum concentrations of ß-carotene than did men, but there were no significant differences in -tocopherol concentrations, neither crude nor when expressed as an -tocopherolcholesterol ratio. Women had slightly higher serum cholesterol concentrations than did men. There was no significant difference between men and women in underreporting energy intake (50% of men compared with 45% of women), but underreporting was strongly and positively associated with BMI (P <0.001) in both sexes (data not shown). The median intakes of vitamin E supplements were the same in men and women (6.7 -TE/d), although men consumed slightly higher doses overall. The range of intakes was 0.7182 -TE/d in men and 0.1108 -TE/d in women. Only 8 persons reported the use of ß-carotene supplements, 7 of whom were women.
View this table:
TABLE 2. . Average daily nutrient intakes and serum nutrient concentrations
Correlational analysis
The age-adjusted correlations between serum antioxidants, dietary intake, alcohol consumption, tobacco use, physical activity, obesity estimates, and serum cholesterol concentrations are shown in Table 3. The correlations between serum -tocopherol and ß-carotene concentrations were 0.34 in men and 0.25 in women. Serum ß-carotene concentration was positively associated with serum -tocopherol concentrations, serum cholesterol concentration, and the intakes of dietary fiber, ascorbic acid, and ß-carotene in both men and women. There was also a positive association with selenium intake in women. The amount of tobacco smoked was negatively associated with serum ß-carotene concentrations in men. All obesity measures were negatively associated with serum ß-carotene concentrations in men and women, but none, except for percentage of body fat, were statistically significant in men.
View this table:
TABLE 3. . Associations between serum ß-carotene, serum -tocopherol, and other selected factors (age-adjusted correlation coefficients)
Serum -tocopherol concentrations were strongly correlated with serum cholesterol concentrations in both men and women (Table 3). They were also associated with total intakes of vitamin E and selenium, and, in women, with ascorbic acid. All obesity measures were both significantly and positively associated with serum -tocopherol concentrations in both sexes, except for percentage of body fat in men.
Multivariate analyses
For each combination of sex and serum nutrient, 4 multiple linear regression models were specified, 1 for each obesity measure. The independent associations between serum nutrients and the independent variables were not substantially altered when any of the 4 obesity measures was included. Because BMI was used most commonly in previous studies, we decided to present the models that included BMI as the obesity measure. The description of the results, however, relates to all of the examined models.
The multivariate analyses of serum ß-carotene concentrations, illustrated in Table 4, indicated that cholesterol concentration and general and central adiposity were the strongest markers of serum ß-carotene in both men and women. Cholesterol was positively and obesity was negatively associated with serum ß-carotene concentrations. Intake of ß-carotene was a positive marker of serum ß-carotene concentration in women, but was not significant in men. Current smoking was associated with low ß-carotene concentrations in men, but not in women. In women, there was a negative relation between energy intake and ß-carotene concentration and a positive association between dietary fiber and serum ß-carotene concentration.
View this table:
TABLE 4. . Markers of serum ß-carotene in men and women (multiple linear regression)
Serum cholesterol concentration was positively associated with serum -tocopherol concentration and was the strongest marker of serum -tocopherol in both men and women, as illustrated in Table 5. Vitamin E intake was associated with serum -tocopherol concentrations, but not as strongly as was serum cholesterol concentrations. In women, serum -tocopherol concentrations were positively associated with the intakes of ascorbic acid and selenium, but negatively associated with the intake of ß-carotene. In contrast with serum ß-carotene, obesity measures were positively associated with serum -tocopherol, although the association was not significant in all analyses (see below).
View this table:
TABLE 5.. Markers of serum -tocopherol in men and women (multiple linear regression)
The proportion of the variation in serum ß-carotene values explained by the models (adjusted R2) ranged from 22.3% to 25.3% in men and from 22.0% to 23.6% in women. Corresponding ranges for serum -tocopherol were 45.848.5% in men and 52.453.4% in women. The difference between explanatory values for serum -tocopherol and ß-carotene concentrations was due mainly to the higher correlation of serum cholesterol with serum -tocopherol than with serum ß-carotene concentrations. No significant interactions between intake of ß-carotene or -tocopherol and smoking status were found.
Partial correlation coefficients of BMI, percentage of body fat, waist circumference, and WHR with serum nutrient concentrations, when each of the obesity measures was entered into the separate multivariate models, are shown in Table 6, as was shown previously for BMI in Tables 4 and 5. All obesity measures were negatively associated with serum ß-carotene concentration. Waist circumference and WHR were positively and significantly associated with serum -tocopherol concentration in both men and women, but measures of general obesity, ie, BMI or percentage of body fat, were not significant, with the exception of BMI in men.
View this table:
TABLE 6. . Summary of the independent associations of 4 obesity measures with serum ß-carotene and -tocopherol (partial correlation coefficients)1
There was little change in the associations between central adiposity and serum -tocopherol when adjusting for percentage of body fat, indicating an independent effect of central adiposity (Table 6). In contrast, the correlation between central adiposity and serum ß-carotene concentrations became nonsignificant in men, whereas it decreased in women when percentage of body fat was included in the model.
DISCUSSION
The present study indicated different associations between serum ß-carotene and -tocopherol concentrations and diet, central adiposity, general adiposity, and smoking.
Diet
Fiber intake was an equally consistent marker of elevated serum ß-carotene concentrations as was ß-carotene intake. Vegetables are important sources of fiber and carotenoids other than ß-carotene, and it was shown that carotenoids may interact in vivo (49). Thus, serum ß-carotene concentrations may be increased by the dietary intake of other carotenoids, as reflected by an independent association between high fiber intake and serum ß-carotene concentration.
Serum -tocopherol concentration was positively associated with total vitamin E intake (diet and supplements). In contrast with the relation in men, serum -tocopherol concentrations in women were positively correlated with the intake of ascorbic acid and selenium. These observations appear biologically plausible: selenium is a cofactor of intracellular glutathione peroxidase, which, with vitamin E, aids in the prevention of membrane lipid peroxidation. Thus, a synergistic effect between intracellular vitamin E and glutathione peroxidase is likely (50). Ascorbic acid can convert the oxidized -tocopherol molecule back into -tocopherol, although the importance of this mechanism in vivo is not clear (5). Another possible explanation for the observation is that many women in the MDCS consumed multivitamin supplements, which tend to contain large doses of vitamin E, selenium, and ascorbic acid (51).
Obesity
In the present study, serum ß-carotene concentration was negatively associated with both general and central adiposity, whereas serum -tocopherol concentration was positively associated with central adiposity (and with BMI in men). Some previous studies reported negative associations between -tocopherol concentrations and central adiposity, in both serum (52) and adipose tissue (53). It was suggested that part of the increased incidence of cardiovascular disease associated with central adiposity may be caused by low -tocopherol status. Our findings do not support this hypothesis. We observed a significant, positive association between serum -tocopherol concentration and central adiposity, both unadjusted and adjusted for diet and serum cholesterol concentration. In men, we also observed a positive association between BMI and serum -tocopherol concentration. To test whether these findings were caused by supplement use, we tested the multivariate models with terms for interaction between supplemental vitamin E use and the obesity measures. No significant interactions were found. In addition, we tested the original models without vitamin E supplement users. The results were not substantially altered (data not shown).
The independent negative association between general obesity and serum ß-carotene concentration observed in this study might have at least 2 explanations. The first is that ß-carotene is distributed between plasma and adipose tissue, adipose tissue being the dominant storage tissue in man (22). Accordingly, a person with high fat mass would have a larger proportion of ingested ß-carotene absorbed by fat tissue than would a lean person if all other metabolic factors were equal. A second explanation could be that obese persons consume less food that would increase serum ß-carotene concentrations, and this might not be adequately captured by fiber and ß-carotene intake estimates. Other causes of the negative association between general obesity and serum ß-carotene concentration were suggested by Hebert et al (23), who proposed, for example, a potential influence of differential bias in the reporting of dietary intake. There are few data to support this latter suggestion. Observations that are consistent with overreporting of nutrient-dense foods by obese persons, however, were made in the MDCS cohort (54).
Smoking
Similar to other studies (55), male smokers had lower concentrations of serum ß-carotene than did male nonsmokers. Low ß-carotene concentrations in smokers may be caused either by low ß-carotene intake (55) or by increased metabolism of antioxidants (56). The quantitative contribution of ß-carotene to the serum antioxidant defense in smokers in vivo is not known (5, 56, 57). If this contribution were to be small, it would imply that smoking itself is not a major cause of low ß-carotene concentrations. On the other hand, there were not many heavy smokers in this studyonly 35% of male smokers and 15% of female smokers smoked >15 cigarettes/d. This may explain the lack of an association between the amount of tobacco smoked and low ß-carotene concentrations in female smokers.
Limitations
As expected, serum -tocopherol and ß-carotene concentrations, serum -tocopherol in particular, were strongly associated with serum cholesterol concentrations. Another lipid status variable of importance is triacylgycerol, which is usually correlated with serum cholesterol concentrations. Triacylglycerol was not measured in this study. Triacylglycerol was shown to be associated with both ß-carotene (27, 28, 32) and -tocopherol (23, 27, 28, 32, 33), but more strongly associated with serum -tocopherol than with serum ß-carotene concentrations (23, 27). Indeed, the association between blood -tocopherol and triacylglycerol appears to be almost of the same magnitude as that between serum -tocopherol and serum cholesterol concentrations (27). Because serum triacylglycerol is associated more strongly with central adiposity than with general obesity (58, 59), it is possible that the association we observed between serum -tocopherol concentrations and central adiposity may have been due to some extent to confounding by triacylglycerol. We attempted to partly correct for this by adjusting the multivariate models for use of lipid-lowering drugs (mostly fibrates). However, use of these drugs was small (n = 9) and the adjustment did not change the multivariate models (data not shown).
Furthermore, we had access to blood samples from nonfasting subjects only. Because triacylglycerol concentrations vary markedly with the composition and time of the last meal, it is likely that serum nutrient concentrations vary as well. This could result in an increased, seemingly random variation in -tocopherol and ß-carotene concentrations, implying that the true correlations between intakes and serum concentrations might be stronger than what was shown in this study.
Underreporting is a common problem in dietary analyses. Obese adults tend to underreport energy intake (60), which was observed in recent studies from the MDCS (54, 61, 62). Also, in a large, nationwide sample from the United States, Ma et al (63) noted that desserts were the most important source of vitamin E; desserts are likely to be underreported in dietary analyses (60). However, we observed significant associations between serum -tocopherol concentration and central adiposity after adjustment for both energy intake and general obesity. Furthermore, post hoc analyses of the multivariate models, eg, stepwise addition of independent variables in varying order, did not support an important role of underreporting in the present results (data not shown).
The MDCS participants constitute a fairly representative sample of the Malmö population in the age categories studied. Yet, the MDCS participants may be different from the nonparticipants (healthy cohort) (J Manjer, S Carlsson, S Elmståhl, et al, unpublished observations, 2000). A likely result of this is smaller ranges of nutrient, alcohol, and tobacco intakes in the study group than in the background population, which implies that we might have found stronger associations between serum values and the dietary and lifestyle variables had the participation rate been higher. Additionally, the subsample used in this study had a higher participation rate than did the rest of the MDCS (J Manjer et al, unpublished observations, 2000).
Conclusion
We conclude that serum ß-carotene and -tocopherol concentrations in this sample of middle-aged citizens of Malmö were primarily influenced by serum cholesterol concentrations and measures of obesity, but also by smoking and intake of several antioxidant nutrients. Serum -tocopherol concentration was significantly and positively associated with central adiposity in both men and women; however, the results of associations between general obesity and serum -tocopherol concentration were inconclusive. Serum ß-carotene concentration was significantly and negatively associated with both general and central adiposity, but there was no independent association between central adiposity and serum ß-carotene concentrations in men after adjustment for body fat. The differences between measures of central and general obesity in their associations with serum antioxidants may warrant some consideration in future studies.
ACKNOWLEDGMENTS
We thank Irene Mattisson and Ulla Johansson for helpful suggestions, the staff at the Department of Cariology at Umeå University for the serum nutrient analyses, and the staff and participants of the Malmö Diet and Cancer Study.
REFERENCES