点击显示 收起
1 From the Center for Nutrition and Health, National Institute for Public Health and the Environment, Bilthoven, Netherlands (BB, EJMF, and DK); the Division of Human Nutrition, Wageningen University, Wageningen, Netherlands (BB, FJK, DK, and LCPGMG); the Medical Faculty, Basel University, Basel, Switzerland (DS-G); and the Service de Geriatrie, Centre Hospitalier, Valence, France (MF)
2 The SENECA study is supported by the European Union. 3 Reprints not available. Address correspondence to B Buijsse, Center for Nutrition and Health, National Institute for Public Health and the Environment, PO Box 1, 3720 BA Bilthoven, Netherlands. E-mail: brian.buijsse{at}rivm.nl.
ABSTRACT
Background: Only a few observational studies have related plasma carotene and -tocopherol to mortality in elderly subjects.
Objective: The objective was to study the association of plasma carotene (-and ß-carotene) and -tocopherol with all-cause and cause-specific mortality in elderly subjects who participated in a European prospective study.
Design: Plasma concentrations of carotene and -tocopherol were measured in 1168 elderly men and women. After a follow-up period of 10 y, 388 persons had died. The association between plasma antioxidants and mortality was analyzed by using Cox proportional hazard models. To put our results in context, we performed a meta-analysis of 5 studies on plasma antioxidants and all-cause mortality in elderly populations.
Results: Plasma carotene concentrations were associated with a lower mortality risk [adjusted rate ratio (RR) for an increment of 0.39 µmol/L: 0.79; 95% CI: 0.70, 0.89]. This lower mortality risk was observed for both cancer (RR: 0.59; 95% CI: 0.44, 0.79) and cardiovascular disease (RR: 0.83; 95% CI: 0.70, 1.00). The lower risk of cardiovascular death was confined to those with a body mass index (in kg/m2) <25 (RR: 0.67; 95% CI: 0.49, 0.94). Plasma concentrations of -tocopherol were not associated with all-cause or cause-specific mortality. The results for both plasma antioxidants and all-cause mortality were confirmed by the meta-analysis.
Conclusions: This prospective study suggests that high plasma concentrations of carotene are associated both with lower mortality from all causes and with cancer in the elderly. For cardiovascular mortality, the inverse association was confined to elderly with body mass indexes <25.
Key Words: Antioxidants • carotene • -tocopherol • elderly • mortality • cancer • cardiovascular disease • cohort study
INTRODUCTION
Aging is associated with higher levels of oxidative stress (1), which are likely to play an important role in the development of cardiovascular disease (CVD) (2) and cancer (3). Oxidative stress is lower among subjects with a high antioxidant status (4). Therefore, elderly subjects with a high antioxidant status may have a lower risk of these chronic diseases.
Carotene and vitamin E are 2 fat-soluble antioxidants that have received much scientific attention during the past decade. Observational studies relating plasma or serum concentrations of ß-carotene to CVD and cancer and relating vitamin E to CVD are mainly conducted in middle-aged populations. A high status of ß-carotene was related to a lower risk of CVD in most (5–8) but not all (9) observational studies. A few studies of vitamin E status found an inverse association with CVD (10, 11), but most others did not (5, 7, 11, 12, 13). A high ß-carotene status was strongly related to a low risk of cancer, especially lung cancer (14).
Studies of ß-carotene and vitamin E status in elderly populations are limited to a few observational studies that related plasma concentrations of these antioxidants to mortality (15–19). Most of these studies focused on all-cause mortality and had a small sample size. Although several studies found an inverse association of serum ß-carotene or total carotenoid concentrations with all-cause (15, 18, 19) and cardiovascular (15, 18) mortality, this association was in most cases not significant. One study did not find a relation between plasma ß-carotene and all-cause mortality (17). Plasma -tocopherol was not associated with either all-cause (15–18) or cardiovascular (15, 16, 18) mortality in elderly subjects. Only one study investigated the relation of plasma carotenoids and vitamin E to cancer mortality in an elderly population, and it found no association (15).
Measures of overweight and obesity, such as body mass index, are consistently found to be inversely associated with plasma concentrations of ß-carotene (20–22), whereas their association with plasma -tocopherol is less clear (20, 22, 23). Obesity has been shown to be associated with systemic low-grade inflammation (24) and oxidative stress (25). Taken together, these observations could indicate that plasma ß-carotene in particular is depleted because of oxidative stress in obese subjects.
We have studied whether plasma concentrations of carotene and -tocopherol were associated with all-cause and cause-specific mortality in apparently healthy elderly subjects who participated in a prospective European study. We examined whether these associations were modified by body mass. Finally, to put our results in context, we performed a meta-analysis of observational studies of both plasma antioxidants in relation to all-cause mortality in elderly populations.
SUBJECTS AND METHODS
Study population
The Survey in Europe on Nutrition and the Elderly, a Concerted Action (SENECA) is a prospective study investigating whether diet and lifestyle influence the health of elderly people in various European countries (26). Subjects were excluded if they were psychogeriatric patients living in nursing homes, did not speak the country’s language fluently, or were not able to answer questions independently. Between December 1988 and March 1989, baseline measurements took place in a random age- and sex-stratified sample of European elderly aged 70–75 y (27). Nine study centers, in which 2038 subjects were examined, collected blood samples for antioxidant measurements and provided information about vital status and causes of death. Participation rates varied between 34% and 62% (27). Plasma antioxidants were actually measured in 1682 subjects. We excluded 102 participants with missing information on important variables and another 392 with a history of myocardial infarction, stroke, diabetes, or cancer. In addition, 16 participants were excluded because information on follow-up time was lacking, and 4 participants were excluded because their plasma antioxidant concentrations exceeded the mean by >9 times the SD. This study is therefore based on 1168 European elderly.
All participants gave written informed consent. Approval of the study was obtained from the participating SENECA centers.
Antioxidant and lipid measurements
Blood samples were obtained by venipuncture after an overnight fast. Plasma and sera were stored at –80°C before transport to the Division of Human Nutrition at Wageningen University for measurement of serum total and HDL cholesterol. For plasma antioxidant measurements, plasma samples were then sent to the laboratories of Hoffmann-La Roche Ltd (Basel, Switzerland). Plasma concentrations of carotene and -tocopherol were measured by HPLC methods (28, 29). For the current study, the sum of -and ß-carotene was reported by the laboratory. CVs for the day-to-day variance of duplicate assays of one sample were 3.0% for carotene and 1.8% for -tocopherol. Total and HDL cholesterol and triacylglycerols were measured in sera by using enzymatic calorimetric methods in the standardized lipid laboratory of Wageningen University (30).
Lifestyle assessment and health status
Baseline information on smoking, years spent in full-time education, the use of vitamin supplements, alcohol consumption, and the prevalence of diabetes, myocardial infarction, stroke, and cancer was obtained by using questionnaires (31). Alcohol consumption was categorized as use or nonuse. Household, sports, and leisure-time physical activities were estimated by using a validated questionnaire (32). Physical activity was classified in the total study population by sex-specific tertiles.
Body weight was measured to the nearest 0.5 kg on a calibrated scale. Subjects were weighed in the morning after breakfast and after emptying the bladder; they were wearing light underclothing. Height was measured to the nearest 0.1 cm while the subjects were standing erect and wearing no shoes. Body mass index (BMI) was calculated by dividing body weight (in kg) by the square of height (m2).
Case ascertainment
Information on vital status was collected in 1999–2000. One experienced clinical epidemiologist coded the causes of death by using the 9th revision of the World Health Organization International Classification of Diseases (ICD9) (33). CVD is covered by ICD9 codes 390–459, and cancer is covered by codes 140–209. The cause of death was unknown for 104 subjects. For the analysis of cause-specific mortality, the follow-up time of these subjects was censored. Only 4 subjects were lost to follow-up; their follow-up time was censored after 5 y. After 10 y of follow-up, 388 had died—38% of CVD and 25% of cancer.
Statistical analysis
Descriptive analyses were performed with stratification for cohort and tertiles of plasma carotene and -tocopherol. Because the distribution of plasma carotene was skewed to the right, values were log transformed, and, therefore, geometric means and geometric SDs are presented.
Cox proportional hazards models were used to obtain rate ratios (RRs) and 95% CIs for the association between a 1-SD increment in plasma antioxidant concentration and mortality. Although women had a higher average concentration of both plasma carotene and -tocopherol, all analyses were done for men and women combined because the associations with mortality did not differ between men and women. Analyses were adjusted for age (continuous) and sex (model 1); in additional multivariate analyses, they were also adjusted for other covariates [model 2: BMI (continuous), serum total cholesterol (continuous), serum HDL cholesterol (continuous), current smoking (yes or no), alcohol consumption (yes or no), physical activity (low versus medium and high tertiles), and SENECA center (indicator variables)]. All analyses were further adjusted for either plasma carotene or plasma -tocopherol (model 3).
Plasma carotene and -tocopherol were measured 5 y later in a subsample of 644 participants. Using Pearson’s coefficient for correlation between the measurements at baseline and 5 y later, we adjusted our risk estimates for measurement error and within-person variance (regression dilution bias) (34).
To examine the roles of body mass, smoking, and sex, stratified analyses were performed. Possible interactions of plasma concentrations of antioxidants with body mass or smoking for mortality were studied by entering product terms in the Cox regression models.
We performed a meta-analysis of observational studies of the association of plasma ß-carotene or -tocopherol with all-cause mortality in subjects aged 60 y, including the current study. We performed a systematic, computerized literature search by using a combination of the Medical Subject Heading terms "aged," "ß-carotene," "carotenoids," "vitamin E," "antioxidants," and "mortality." Reference lists of selected articles were inspected. In addition to 3 studies that measured plasma ß-carotene (17–19), we included one study that measured plasma carotenoids (15). Reported adjusted RRs were used to calculate new mortality rates in the lowest and highest category of plasma antioxidant. On the basis of these event rates, RRs were calculated in which the lowest plasma category was taken as the reference. We used a random-effects model to combine the results of all studies (35). Heterogeneity was assessed by inconsistency squared (I2), which describes the percentage of total variation between studies that is due to heterogeneity rather than to chance; higher values indicated greater heterogeneity (36).
All data were analyzed by using SAS software (version 9.1; SAS Institute, Cary NC). Statistical tests were 2-sided, and P < 0.05 was considered significant.
RESULTS
Plasma carotene varied from 0.28 µmol/L in Belgian subjects to 0.69 µmol/L in Swiss subjects. Plasma -tocopherol ranged from 25.3 µmol/L in the Belgians to 35.6 µmol/L in the Swiss. Selected characteristics of the participants from the SENECA centers are shown in Table 1. Plasma concentrations of carotene and -tocopherol were positively associated with serum total and serum HDL cholesterol and inversely associated with smoking tobacco (Table 2 and Table 3). Plasma carotene was inversely related to serum triacylglycerols and physical activity, whereas plasma -tocopherol was positively associated with serum triacylglycerols and years spent in full-time education.
View this table:
TABLE 1. Baseline characteristics of apparently healthy participants by SENECA (Survey in Europe on Nutrition and the Elderly, a Concented Action) study center1
View this table:
TABLE 2. Baseline characteristics of the study population by tertiles of plasma carotene1
View this table:
TABLE 3. Baseline characteristics of the study population by tertiles of plasma -tocopherol1
After multivariate adjustment, subjects with BMIs 25 had lower mean plasma carotene concentrations than did subjects with BMIs < 25 (0.47 µmol/L and 0.60 µmol/L, respectively; P < 0.0001). Plasma -tocopherol was not associated with BMI. Women had higher concentrations of both plasma -tocopherol and plasma carotene than did men (P < 0.0001 for both after multivariate adjustment).
Plasma antioxidants and mortality
As shown in Table 4, plasma carotene concentrations were inversely associated with mortality due to all causes, CVD, and cancer. The strongest inverse association was observed for mortality due to cancer. Additional adjustment for plasma -tocopherol yielded similar results.
View this table:
TABLE 4. Relative risks (RRs) and 95% CIs for the association of plasma antioxidants with all-cause and cause-specific mortality among participants in the Survey in Europe on Nutrition and the Elderly, a Concented Action (SENECA) study1
Excluding subjects with subclinical CVD and cancer by omitting the first 2 y of follow-up resulted in similar associations. The inverse association with cancer mortality became stronger (RR for an increment of 0.39 µmol/L plasma carotene: 0.50; 95% CI: 0.35, 0.72), though the association with cardiovascular mortality was slightly attenuated (RR: 0.89; 95% CI: 0.74, 1.07).
The inverse association between plasma carotene and risk of death due to CVD was confined to subjects with a low body mass. Persons with BMIs <25 had RRs of 0.67 (95% CI: 0.49, 0.94), whereas those with BMIs 25 had RRs of 0.97 (95% CI: 0.79, 1.20) (P for interaction = 0.033; based on model 3 without adjustment for BMI). The association between plasma carotene and CVD death did not differ between men and women (P for interaction = 0.65), and nor did that between smokers and nonsmokers (P for interaction = 0.63).
After adjustment for age, sex, and other potential confounders, plasma -tocopherol was not significantly associated with mortality due to all causes, CVD, or cancer. These results did not differ significantly by strata of BMI, smoking, or sex (data not shown).
With the use of Pearson’s correlation coefficients between baseline and follow-up measurements of plasma carotene (r = 0.73) and -tocopherol (r = 0.68; P < 0.0001 for both), we adjusted for measurement error and within-person variation of both plasma antioxidants. For associations between plasma carotene and mortality, this resulted in RRs in age- and sex-adjusted RRs that were 7–16% lower than those obtained by using only baseline information for plasma carotene.
Meta-analysis of plasma antioxidants and all-cause mortality
We identified 5 observational studies, including the current study, that reported associations of plasma (ß)-carotene, plasma vitamin E, or both with all-cause mortality in elderly populations. For the current study, multivariate-adjusted RRs for the highest versus the lowest antioxidant quartile were used. After the combination of all studies, elderly subjects with high concentrations of plasma carotene had an RR of 0.72 (95% CI: 0.59, 0.87; Figure 1) compared with those with low plasma carotene concentrations, whereas high plasma vitamin E concentrations were not associated with all-cause mortality (RR: 1.07; 95% CI: 0.94, 1.22; Figure 2). Heterogeneity between studies was moderate, varying from 36% for studies of vitamin E status to 50% for studies of carotene status.
FIGURE 1.. Meta-analysis of the association of high and low carotene status with all-cause mortality in elderly subjects. USA, United States; NL, Netherlands; UK, United Kingdom; EU, Europe; I, inconsistency. *This study investigated plasma carotenoids. **This study investigated plasma - and ß-carotene. The P value represents the result of the chi-square test for heterogeneity.
FIGURE 2.. Meta-analysis of the association of high and low vitamin E status with all-cause mortality in elderly subjects. USA, United States; Fin, Finland; NL, Netherlands; UK, United Kingdom; EU, Europe; I, inconsistency. The P value represents the result of the chi-square test for heterogeneity.
DISCUSSION
In this prospective study conducted in elderly European subjects, plasma carotene concentrations were associated with lower mortality risks due to all causes, cancer, and CVD. The inverse relation with CVD mortality was, however, confined to subjects with a BMI < 25. Plasma concentrations of -tocopherol were not associated with all-cause or cause-specific mortality.
The strengths of our study are the performance of all assays of plasma antioxidant concentrations in one laboratory, the wide range in plasma concentrations of carotene and -tocopherol, the longitudinal study design, and the standardized and validated methods to assess lifestyle factors and anthropometric variables (27). The limitations of this study also must be considered. First, although vital status follow-up was nearly complete, there were 104 deceased subjects for whom the cause of death could not be ascertained, which limited our power to study cause-specific associations. Second, baseline plasma values of carotene and -tocopherol are assumed to be stable in persons during the years of follow-up, but those concentrations could change over time. However, concentrations of both plasma antioxidants at baseline correlated strongly with those found 5 y later.
In the current study, plasma carotene was strongly inversely associated with mortality from all causes, whereas no association was observed between plasma -tocopherol and all-cause mortality. Previous observational studies in elderly subjects related plasma carotene, carotenoids, or vitamin E to all-cause mortality (15–19), but the sample size of most of those studies was relatively small. Therefore, we performed a meta-analysis, including the current study, with all-cause mortality as the endpoint. Plasma carotene was strongly associated with a lower risk of all-cause mortality, whereas plasma -tocopherol was not associated with all-cause mortality.
We observed an inverse association of plasma carotene with cancer and CVD mortality. Together with results from other observational studies, these findings are in sharp contrast with those from large-scale supplementation trials (37–40), which makes it very unlikely that pharmacologic doses of ß-carotene are effective in preventing CVD and cancer. It was argued that the inverse associations between antioxidants and chronic disease in observational studies could be confounded by smoking or by socioeconomic status (41). We adjusted our analyses for current smoking. An alternative adjustment for smoking, the use of information on former smoking, did not result in different risk estimates. In our study, the number of years spent in full-time education was not significantly related to plasma carotene, and it did not confound the relation between plasma carotene and all-cause mortality. However, residual confounding by smoking, socioeconomic status, or other factors cannot be excluded.
Plasma carotene was inversely associated with cancer mortality in the current study. This relation became stronger after exclusion of subclinical cancer cases and adjustment for errors in measurement of and biological fluctuations in plasma carotene. Because large-scale trials did not show that pharmacologic doses of ß-carotene prevent cancer, the question remains whether lower doses of ß-carotene would reduce the incidence of cancer in susceptible populations. An alternative explanation is that other factors related to plasma carotene are responsible for the inverse associations found in observational studies (42). Plasma carotene may also be a marker of a concerted effect of multiple other nutrients in fruit and vegetables.
In the current study, the inverse association between plasma carotene and CVD mortality was confined to subjects with BMIs < 25. A possible explanation for this difference in association between plasma carotene and CVD mortality by BMI could be the relation with inflammation. It has been shown that BMI is positively associated with C-reactive protein (24), an indicator of inflammation. Our data suggest that, in persons with BMIs < 25, who are likely to have a low level of inflammation, plasma ß-carotene is inversely related to CVD mortality. However, in overweight and obese subjects, who are characterized by a high level of inflammation, this relation was not observed. These results therefore suggest that, in the presence of a low inflammation burden, carotene may be protective against CVD.
Results of observational studies on plasma vitamin E in relation to CVD mortality in elderly subjects are inconsistent. In elderly British subjects, a nonsignificantly lower risk of CVD death was observed for subjects in the 2 upper quintiles of plasma -tocopherol (18), whereas, in elderly living in the United States, plasma -tocopherol was positively associated with heart disease mortality (15). In the current study, plasma -tocopherol was not related to CVD mortality or to all-cause mortality. This observation was confirmed by our meta-analysis as well as by a meta-analysis of trials in which vitamin E supplementation did not reduce all-cause mortality and high doses of vitamin E were even found to increase the risk of all-cause mortality (43).
In conclusion, in the current study, conducted in European elderly, plasma concentrations of carotene were associated with a lower risk of death due to all causes and that due to cancer. Subjects with a high plasma carotene concentration and a BMI < 25 had also a lower risk of CVD death. The relation among plasma carotene, BMI, and CVD merits further study.
ACKNOWLEDGMENTS
The authors thank all principal investigators and collaborators who participated in the current study. This study used data from the European Union–funded Survey in Europe on Nutrition and the Elderly, a Concerted Action (SENECA) and the Healthy Aging: a Longitudinal Study in Europe (HALE) Project.
BB performed the statistical analysis and wrote the first draft of the manuscript. EJMF, FJK, and DK supervised and contributed to the interpretation of the data and revisions of the manuscript. DS-G, MF, and LCPGMG contributed to the initial study design, conducted the study, and gave critical comments on the manuscript. None of the authors had any personal or financial conflict of interest.
REFERENCES