点击显示 收起
1 From the Nutritional Epidemiology Unit, French Institute of Health and Medical Research (INSERM U557), UMR (INSERM/INRA/CNAM), CRNH Ile-de-France, Bobigny, France (SC, AC, SB, A-CV, LD, PG, and SH); and the Department of Nutrition, Hôtel-Dieu Hospital, Paris, France (AP-HP)
2 Supported by a research grant from the Association de Langue Française pour l'Etude du Diabète et des Maladies Métaboliques (ALFEDIAM), France (to SC). 3 Reprints not available. Address correspondence to S Czernichow, Service de Nutrition, Hôpital Hôtel-Dieu, 1 place du Parvis Notre-Dame, 75004 Paris, France. E-mail: sebastien.czernichow{at}htd.aphp.fr.
ABSTRACT
Background: Observational data suggest a protective effect of several antioxidants on fasting plasma glucose (FPG) and type 2 diabetes. However, randomized trials have yielded inconsistent results.
Objectives: The first objective was to assess the effect of 7.5 y of antioxidant supplementation on FPG at 7.5 y. The second objective was to examine the epidemiologic association of baseline dietary intakes or plasma antioxidants and FPG (at baseline and at 7.5 y).
Design: Subjects (n = 3146) from the Supplementation en Vitamines et Minéraux Antioxydants (SU.VI.MAX) primary prevention trial in France were randomly assigned to receive a daily capsule containing 120 mg vitamin C, 30 mg vitamin E, 6 mg ß-carotene, 100 µg Se, and 20 mg Zn or a placebo.
Results: After 7.5 y, no significant difference was observed between age-adjusted mean FPG in men (P = 0.78) and women (P = 0.89) in either group. Baseline ß-carotene dietary intakes and plasma concentrations were inversely associated with FPG in multivariate mixed models (P = 0.0045 and P < 0.0001, respectively). Baseline plasma vitamin C and selenium were negatively (P = 0.0455) and positively (P < 0.0001) associated, respectively, with FPG.
Conclusions: Supplementation with antioxidants at nutritional doses for 7.5 y had no effect on FPG in men or women who followed a balanced diet. An inverse association of baseline ß-carotene dietary intake and plasma concentrations with FPG was found, probably because ß-carotene is an indirect marker of fruit and vegetable intakes.
Key Words: Antioxidant supplementation • ß-carotene • fasting plasma glucose • cohort study
INTRODUCTION
The pathogenic effect of hyperglycemia, with or without an interaction with fatty acids, may be mediated via higher production of reactive oxygen species (ROS) and reactive nitrogen species (RNS) and subsequent oxidative stress (1). However, some studies have shown an increase in oxidative stress production and an impairment of antioxidant status in diabetic patients (2-7).
Antioxidants such as ß-carotene and vitamin C may diminish the oxidative process by inactivation of free radicals (8). Several observational, nested, case-control and prospective studies have suggested a protective effect of dietary antioxidant or fruit intake or both (2, 9-12) or of plasma antioxidants (2, 10, 13-15) on type 2 diabetes and associated complications. In contrast, negative evidence has been found in large clinical trials of the effect, at pharmacologic doses, of a single antioxidant such as vitamin E (16, 17) or ß-carotene (18) on the risk of type 2 diabetes or associated complications. The apparent discrepancy between observational studies and randomized trials may be explained by method-related issues (eg, dose, duration of follow-up, or type and source of antioxidant) or by the fact that dietary intake or plasma antioxidant biomarkers are indirect surrogate markers of fruit and vegetable intakes that are related to a healthier lifestyle.
Thus far, no definitive conclusion about a protective effect of multiple antioxidant long-term supplementation, at nutritional doses, on fasting plasma glucose has been drawn. The randomized placebo-controlled Supplementation en Vitamines et Minéraux Antioxydants (SU.VI.MAX) trial (19-21) provides a unique opportunity for assessing, in middle-aged subjects, the effect of supplementation with a combination of antioxidant vitamins and minerals on fasting plasma glucose (FPG). This study also elucidates the prospective association of baseline dietary intake and plasma antioxidants with FPG.
SUBJECTS AND METHODS
Subjects and measurements
The French SU.VI.MAX study is a randomized double-blind, placebo-controlled, primary prevention trial that was initially designed to test whether supplementation with antioxidant vitamins and minerals at nutritional doses would reduce the incidence of cardiovascular disease and cancer. The study's methods and results were detailed previously (19, 20). In brief, from 1994 to 2002, all 13 017 recruited participants (women aged 35–60 y and men aged 45–60 y) underwent a yearly visit that alternated blood sampling (from baseline) and clinical examination (from 1995–1996).
All participants in the SU.VI.MAX. Study provided written informed consent. The SU.VI.MAX Study was approved by the ethics committee for studies in human subjects, the Comité Consultatif pour la Protection des Personnes se prêtant la Recherche Biomédicale (CCPPRB no. 706) of Paris-Cochin, and the Comité National Informatique et Liberté (CNIL no. 334641), the latter of which ensures that all medical information is kept confidential and anonymous (19).
Supplementation included either a combination of antioxidants at nutritional doses [120 mg vitamin C, 30 mg vitamin E, 6 mg ß-carotene, 100 µg Se (in the form of selenium-enriched yeast] and 20 mg Zn (as gluconate) or a matching placebo, in a single daily capsule. The mean (± SD) follow-up time was 7.5 ± 0.3 y. Results of the SU.VI.MAX. trial (20) showed a high adherence rate (74% of the participants reported having taken at least two-thirds of the capsules). Compliance was confirmed in the intervention group by a significant increase in plasma antioxidant markers at the end of the study (20).
We performed post hoc analyses in subjects for whom baseline data—ie, FPG, supplementation status (intervention or placebo), antioxidant plasma and dietary intakes, age, sex, BMI (in kg/m2), physical activity level, and education level—and data on FPG after 7.5 y of follow-up were available. For the current analysis, we excluded 154 subjects who had FPG > 7 mmol/L or who reported the use of antidiabetic medication or both. Thus, the final sample in this analysis included 3146 men and women.
Blood samples were obtained after a 12-h fast, and all biochemical measurements were centralized in a single laboratory. FBG and total cholesterol were measured by using an enzymatic method (Advia 1650; Bayer Diagnostic, New York, NY). For vitamin C, before freezing, whole venous blood was collected in heparinized tubes and immediately centrifuged for 20 min at 3000 x g and 4 °C, and 0.5 mL plasma was diluted 1-in-10 with 4.5 mL of an aqueous 5% metaphosphoric acid solution. Vitamin C status was evaluated by serum ascorbic acid measurement according to an automated method based on the principle of continuous flow segmented by air bubbles (22). Serum concentrations of retinol, ß-carotene, and tocopherol were measured by HPLC using the Biotek-Kontron HPLC system (Biotek-Kontron, Montigny-le-Bretonneux, France; 23); serum concentrations of zinc and selenium were measured by using flame atomic absorption spectrometers [Perkin Elmer 3110 (for zinc and selenium) and Perkin Elmer 4100 ZL (for selenium); Perkin Elmer, Norwalk, CT] (24, 25).
Blood pressure (BP) was measured with a standard mercury sphygmomanometer. Hypertension was defined as systolic BP 140 mm Hg or diastolic BP 90 mm Hg (or both) or antihypertensive drug use. Age was defined at baseline. The level of education was obtained from a questionnaire at baseline and was coded into 3 categories (primary school, high school, or university or equivalent). Data on smoking status (current smokers, previous smokers, or nonsmokers) and physical activity were collected via the same questionnaire. The habitual physical activity level was coded into 3 categories (ie, inactivity, < 1 h/d, or 1 h/d). Measured body mass index (BMI; in kg/m2) was collected at the first clinical examination.
A subsample of subjects kept a 24-h dietary record every 2 mo during the 7.5 y. These records included 2 weekend days and 4 weekdays each year, and thus data on the mean intakes of all participants on each day of the week and in all 4 seasons were recorded during the course of the study. Information was collected in 1994 and 1995 with the use of the Minitel Telematic Network (France Télecom, Paris, France), a small terminal that was widely used in France as an adjunct to the telephone. This device enabled subjects to fill in computerized dietary records transmitted during telephone connections. An instruction manual for food codification contained validated photographs showing portions of most food preparations in 3 sizes. Subjects were trained to choose a portion from among the 7 pictures available, to indicate each portion consumed. For the current analysis, only subjects with 6 dietary 24-h records for the first 18 mo of the study were included. Individual records presenting < 100 kcal/d or > 6000 kcal/d were excluded from the analyses to limit reporting bias (16, 17, 26).
Statistical analysis
We excluded subjects with baseline FPG > 7 mmol/L or reporting the use of antidiabetic medication and those with both characteristics. Characteristics of men and women according to placebo and intervention groups were compared by Student's t test and chi-square test where appropriate. We assessed the age-adjusted effect of supplementation on FPG after 7.5 y by using Student's t test. Results are expressed as percentages or means ± SDs. We also performed multiple linear mixed-models analyses to assess the association of each independent baseline plasma antioxidant concentrations and dietary intake factor with combined baseline and 7.5 y FPG in the same model. Standardized correlation coefficients (ß), per 1-SD increase, are presented for each model after adjustment for potential confounders (ie, age, sex, BMI, smoking status, physical activity, educational level, supplementation group, and energy intake) for analysis with dietary antioxidant intake. The time of each FPG measurement (baseline and 7.5 y) was also included in the model. All statistical analyses were carried out by using standard procedures in SAS software (version 8.02; SAS Institute, Cary, NC). Statistical significance was judged at < 0.05.
RESULTS
Baseline characteristics
The baseline distribution of variables by supplementation group in men and women is shown in Table 1. The placebo and intervention groups did not differ significantly in terms of age, sex ratio, and usual cardiovascular disease risk factors. The distribution of energy intake and antioxidant dietary intake or plasma concentration, by supplementation group, in both sexes in shown in Table 2. Only the comparison for selenium differed significantly between groups (P < 0.05 for both sexes).
View this table:
TABLE 1. Baseline characteristics of the participants
View this table:
TABLE 2. Baseline antioxidants status of the participants1
Effect of supplementation
After an average of 7.5 y of supplementation with antioxidants, no significant difference in age-adjusted mean FPG was shown between the placebo and supplemented groups of men (5.18 ± 0.54 and 5.19 ± 0.72 mmol/L, respectively; P = 0.78) or women (4.89 ± 0.47 and 4.89 ± 0.47, respectively; P = 0.89).
Associations of dietary intake and plasma antioxidants with FPG
We performed linear mixed-model regressions to test the associations of baseline dietary intake and plasma concentrations of antioxidants with combined FPG at baseline and after 7.5 y of follow-up (Table 3). Baseline dietary intake and plasma concentrations of ß-carotene were inversely associated with FPG in multivariate mixed models (P = 0.0045 and P < 0.0001, respectively). Baseline plasma concentrations of vitamin C and selenium were negatively (P = 0.0455) and positively (P < 0.0001), respectively, associated with FPG. No association of dietary or plasma vitamin E, dietary vitamin C, or plasma zinc with FPG could be shown.
View this table:
TABLE 3. Standardized multiple regression coefficients (ß) from mixed models of the association of baseline antioxidant dietary intake and plasma concentrations with fasting plasma glucose (at baseline and at 7.5 y) in men and women
DISCUSSION
The current study is the first to show that a combination of low-dose antioxidant supplementation, including ß-carotene, does not influence FPG after 7.5 y of follow-up in a well nourished population. In contrast, we confirmed previous observational studies showing that baseline dietary and plasma ß-carotene concentrations were inversely correlated with FPG.
Several large-scale primary or secondary prevention trials in persons with type 2 diabetes showed the lack of an effect of vitamin E supplementation on cardiovascular outcome (16, 17). Our study is in line with all primary and secondary prevention trials that did not find a beneficial effect of antioxidant supplementation on CVD risk factors (27-30) except the Cambridge Heart Antioxidant Study (31). Other results from the SU.VI.MAX.trial confirmed the absence of an effect of antioxidants on CVD risk and on intermediate CVD markers such as carotid atherosclerosis and arterial stiffness (21). Furthermore, a recent meta-analysis based on 135 967 participants showed that studies with high-dosage vitamin E supplementation may increase all-cause mortality (30). Those authors emphasized the need to assess the effect of low supplementation doses, whereas most of the available large-scale prevention trials tested high antioxidant doses (32).
Only one large trial of the specific effect of ß-carotene supplementation on the risk of diabetes mellitus was found in the literature (18). That trial reported that, in a cohort of 22 071 healthy US male physicians, ß-carotene supplementation had no effect on the risk of diabetes after 12 y of follow-up. Our results relating to dietary and plasma antioxidant and FPG also are in agreement with the literature. In cohort studies, dietary intakes of vitamins E and C and carotenoids (2) and fruit and vegetable consumption, were associated with a reduction in diabetes incidence and glucose tolerance (10, 12). Cross-sectional studies of the antioxidant status of diabetes patients found significantly lower plasma carotene (33-35), vitamin E (36), and vitamin C (37) concentrations than were seen in control subjects. Furthermore, with respect to dietary antioxidant intakes, several cohort studies showed an inverse association of plasma concentrations of vitamin E (12-15) or ß-carotene (14) with the risk of diabetes.
Several strengths of the current study should be underlined. It is based on a large sample of men and women recruited from the general population and includes a long follow-up (19). All serum antioxidants and FPG were measured and centralized in the same laboratory. In adition, the current study combined interventional and observational data, thereby enabling a better understanding of the contradictory results previously published, in trials and observational studies, concerning the association between antioxidants and diabetes. Potential limitations of our study should be mentioned. We cannot exclude a possible bias due to self-exclusion of subjects with a less healthy dietary pattern; indeed, our study subjects had volunteered to be enrolled in a nutritional intervention study and therefore may have had a healthier lifestyle than the general population. Because of clinical and biological screening of subjects during the trial, subjects were informed of abnormal fasting glucose and cholesterol values. The provision of this information may help to explain why FPG and total cholesterol improved during follow-up independently of any treatment. Another possible limitation was that of causal inference from post hoc analyses from a randomized trial that was not specifically designed to test the effect of antioxidants on FPG. However, because of the large study sample, the current study has high statistical power, and thus we consider our data to be conclusive.
Dietary and plasma antioxidants and the consumption of fruits and vegetables are associated in observational studies with a reduction in diabetes mellitus risk. Several studies (2, 9, 10) suggested that the possible protective effect of vegetables and fruits was due to the combined action of an antioxidant cocktail and other active compounds, such as fibers and polyphenols (38). The results of the current study reinforce this hypothesis, because we found both no effect of a combination of antioxidants in the trial data and a protective effect of dietary and plasma antioxidants found through observational data. These findings may increase our understanding of the discrepancy between data on the antioxidant effect on FPG in published antioxidant trials and in observational studies.
In conclusion, our randomized trial provides a unique opportunity to examine both observational associations and the effects of supplementation. Despite an inverse association of baseline dietary and plasma ß-carotene with FPG, we did not find a beneficial effect of low-dose antioxidant supplementation on this marker in a well-nourished population. Further post hoc analyses of large primary prevention trials are needed to better understand the discrepancy between randomized and observational data on diabetes or on prevention of diabetic complications or both (5).
ACKNOWLEDGMENTS
All authors participated in the design of the experiment. SC, AC, ACV, and SB were responsible for data collection and analysis. All authors contributed to writing the manuscript. None of the authors had a personal or financial conflict of interest.
REFERENCES