Professor Kjell Asplund Department of Medicine, University Hospital, SE-901 85 Umeå, Sweden (fax: +46 90 14 5921; e-mail: firstname.lastname@example.org).
Abstract. Asplund K (Umeå and Stockholm, Sweden). Antioxidant vitamins in the prevention of cardiovascular disease: a systematic review (Review). J Intern Med 2002; 251: 372–392.
Do antioxidant vitamins, in regular food or as food supplements, protect against myocardial infarction and stroke? In this systematic literature review on the effects of antioxidant vitamins in the primary prevention of cardiovascular disorders, studies with ischaemic heart disease, stroke or combined cardiovascular events as end-points have been included. Studies on the effects of antioxidant vitamins on intermediary end-points (such as blood lipids and blood pressure) and as secondary prevention in patients with manifest cardiovascular disease are reviewed in a conventional manner.
In observational studies (case–control or cohort design), people with high intake of antioxidant vitamins by regular diet or as food supplements generally have a lower risk of myocardial infarction and stroke than people who are low-consumers of antioxidant vitamins. The associations in observation studies have been shown for carotene, ascorbic acid as well as tocopherol. In randomized controlled trials, however, antioxidant vitamins as food supplements have no beneficial effects in the primary prevention of myocardial infarction and stroke. Serious adverse events have been reported. After an initial enthusiasm for antioxidants in the secondary prevention of cardiovascular disease, recent reports from of several large randomized trials have failed to show any beneficial effects.
Thus, the apparent beneficial results of high intake of antioxidant vitamins reported in observational studies have not been confirmed in large randomized trials. The discrepancy between different types of studies is probably explained by the fact that supplement use is a component in a cluster of healthy behaviour. Antioxidant vitamins as food supplements cannot be recommended in the primary or secondary prevention against cardiovascular disease.
During the 1930s, studies in Australia showed that Seventh-Day Adventists who had a diet rich in ascorbic acid and tocopherol had lower blood pressure than control persons on a conventional diet (Mormons). A low risk of cardiovascular death in subjects with a high intake of ascorbic acid was first reported in the 1950s (cited in Ref. ). Ecological studies performed in the United Kingdom later reported a high mortality in myocardial infarction  and stroke  in areas with a low consumption of fresh fruit and vegetables.
Seasonal variations with a particularly high incidence of cardiovascular events during winter-time was proposed to be related to low intake of fresh fruit and vegetables during this period of the year . In the United States, the decline in cardiovascular mortality from the end of the 1950s and onwards occurred at the same time as the consumption of fruit and vegetables increased . A high intake of fruits and vegetables, dark green vegetables in particular, has been associated with a reduced risk of coronary heart disease [6, 7], stroke [6, 8] and total cardiovascular morbidity  or mortality [6, 8, 9]. Patients with myocardial infarction have been observed to have low tissue content of lycopene, considered as a biomarker for intake of vegetables . A meta-analysis has indicated that the relative risk reduction of ischaemic heart disease in high-consumers of fruits and vegetables may be in the order of 15% . There have also been reports on a particularly low risk of cardiovascular disease amongst high-consumers of nuts (which are rich in tocopherol) [12-14]. Furthermore, high intake of food items rich in flavonoids, that are powerful scavengers of free radicals (red wine, tea, onions and apples), have been associated a low risk of cardiovascular disease [15-17].
In a large number of ecological studies, high intake or high plasma levels of antioxidant vitamins have been associated with low risk for cardiovascular morbidity or mortality (see for instance Refs [10, 18-23]). Also in individual-based cross-sectional studies, inverse relationships between plasma levels of antioxidant vitamins and degree of atherosclerosis detected by ultrasound [24-26] or clinical signs of ischaemic heart disease [27, 28] have been reported. Obviously, any correlation in ecological studies may depend on confounding factors. Such possible confounders could be the intake of unsaturated fats or wine. In Europe, a high intake of ascorbic acid and tocopherol is common in countries with a high consumption of unsaturated fats and wine , which also are dietary components that have been associated with a low risk for myocardial and brain infarction.
Observations such as these, together with increased knowledge of the antioxidative properties of carotene, ascorbic acid and tocopherol, have served as an impetus for a wide range of studies exploring possible preventive effects of antioxidant vitamins against cardiovascular diseases.
The primary aim of this review was to answer the question: Does high intake of antioxidant vitamins, by regular food or as food supplements, protect against ischaemic heart disease and stroke?
The key part of the article is a systematic review on the associations between intake of antioxidant vitamins in the primary prevention of cardiovascular disease. This means that all literature published that meets the inclusion criteria has been included. The first part of the article – effects on intermediary variables (blood lipids, blood pressure, haemostatic variables, etc.) – is a conventional review without the intent to include all scientific material available. As the focus of the review is on primary prevention by high intake of antioxidant vitamins, the use of antioxidant vitamins in the treatment or secondary prevention in subjects with manifest cardiovascular disease is only briefly reviewed.
The results of cross-sectional studies, whether based on comparisons of populations (ecological studies) or on comparisons between individuals, are difficult to interpret because there is usually an abundance of possible confounding factors. It is difficult to infer any cause-effect relationships, and the results should be regarded as hypothesis-generating.
In the systematic review, Medline database searches have been performed, using all combinations of antioxidant(s)/carotene/ascorbic acid/tocopherol(s)/vitamin(s) and cardiovascular/heart/cardiac/coronary/cerebrovascular/stroke/atherosclerosis. Additional articles have been retrieved by following landmark articles in Science Citation Index and by scrutinizing reference lists of published articles.
The following inclusion criteria have been used in the systematic reviews:
•Studies in humans only.
•Studies published from 1989 onwards. Last literature search for observational studies was performed in March 2001 and for randomized trials in November 2001. One large cohort study published late during 2001  has also been included.
•Only studies reporting on original data. Review articles, editorials, textbook chapters, etc., have been used only to identify articles with original data.
•Only articles with case–control, cohort or randomized controlled trial (RCT) study designs.
•Studies of antioxidant vitamins (carotene, ascorbic acid, tocopherol) only; other antioxidant substances such as selenium, Q10 and drugs with antioxidant properties have not been included.
•Studies reporting on morbidity and mortality of clinically meaningful manifestations of ischaemic heart disease or stroke. Effects on other manifestations of cardiovascular disease have not been subjected to a systematic review but are summarized.
•Studies on the effects of antioxidants as primary prevention against cardiovascular disease. Effects of antioxidant vitamins to prevent recurrences or disease progression (secondary prevention) have only been summarized, and studies performed in subgroups such as diabetic subjects have not been reviewed.
•Articles on the relationship between consumption of food components and cardiovascular disease without calculations of content of antioxidant vitamins have not been included.
Based on strict quality assessment criteria, all studies have been ranked into one of three categories: low, intermediate and high scientific quality . Results from studies with low quality have been excluded when conclusions have been drawn from the systematic reviews.
Data from cohort studies and RCTs were entered into the Review Manager computer program, version 4.1 , which permits uniform handling and display of the data and calculations of the Peto fixed odds ratio for the outcome of individual and combined studies. To achieve uniformity between studies, this odds ratio is given throughout in the present review rather than odds ratios and relative risks reported in the original articles.
Lipid peroxidation, atherosclerosis and antioxidant vitamins
Lipid peroxidation plays a central role in the atherosclerotic process. Polyunsaturated fatty acids in the cell membranes are subject to oxidation, and peroxyl radicals are formed. A chain reaction is started, in which peroxyl radicals, lipid hydroperoxyls and aldehydes are responsible for the toxic effects in the vessel wall (for review, see Ref. ).
Oxidized low-density lipoprotein (LDL) is prone to be absorbed by macrophages via scavenger receptors . In in vitro experiments, LDL oxidation is prevented when antioxidant vitamins are added. Conversely, during oxidation of LDL, there is a consumption of antioxidants, starting with ubiquinone, continuing with α-tocopherol and being completed with β-carotene depletion. When the antioxidants have been consumed, there is a rapid phase with destruction of polyunsaturated fatty acids in LDL and formation of reactive aldehydes .
In the intact organism, it seems that oxidation of the LDL does not occur in plasma but rather in the arterial wall, mainly in the intima. Oxidized LDL is bound to scavenger receptors of macrophages and foam cells are generated, a primary event in the atherosclerotic process. Oxidized LDL also stimulates the migration of monocytes into the intima and their transformation to macrophages.
In animal experiments, supplementation of synthetic antioxidants may retard the atherosclerotic process. In humans, intake of large amounts of α-tocopherol, but not β-carotene, may retard LDL oxidation  and this effect seems to be dose-dependent .
The scientific information available is summarized in Table 1.
Table 1. Summary of studies on the relationship between antioxidant vitamins and cardiovascular risk factors. Details and references are given in the text
When the diet is changed from a regular mixed one to vegetarian, serum levels of total cholesterol and LDL cholesterol are reduced, whereas high-density lipoprotein (HDL) cholesterol levels increase . This has usually been ascribed to quantitative and qualitative changes of lipid content in the diet, but it is also possible that antioxidant substances have direct effects of lipid levels.
α-Tocopherol given as food supplements in a dose of 800 U day–1 during 3 months to hypercholesterolemic individuals did not affect the levels of total cholesterol, LDL cholesterol, HDL cholesterol or triglycerides despite a reduction in TBARS, a marker of oxidative stress . β-Carotene, ascorbic acid, α-tocopherol and multivitamins as food supplements may modify the oxidation of LDL cholesterol although this protective effect does not occur under all conditions [35, 38-41]. It is interesting to note that individuals with a genetically determined tocopherol deficiency do not have increased LDL oxidation .
In intervention studies, intake of ascorbic acid in doses between 0.3 and 3 g day–1 has been shown to reduce serum levels of total cholesterol in subjects with low plasma levels of ascorbic acid and hypercholesterolemia at the start of the intervention [43, 44]. The reduction in total cholesterol may be as great as 1 mmol L–1 (a 10–15% reduction) and HDL cholesterol may be increased by 10–15% . However, in subjects with normal levels of ascorbic acid and cholesterol, there seems to be little effect of ascorbic acid [44, 45]. In a randomized placebo-controlled trial, supplementation of a high dose of ascorbic acid (4.5 g day–1) during 12 weeks did not effect lipoprotein(a) levels .
The scientific information available is summarized in Table 1.
Transition from a regular mixed diet to a vegetarian diet is usually accompanied by a reduction of blood pressure levels [36, 47]. In a randomized, cross-over study, a change from a carnivorous to a vegetarian diet resulted in a mean reduction of systolic blood pressure of 6–8 mmHg and of diastolic blood pressure of 3 mmHg . Another randomized trial showed that it was sufficient to make an ordinary diet rich in fruit and vegetables to reduce the blood pressure . To a large extent, the blood pressure reduction is resulting from weight loss and reduction of sodium salts and increased intake of potassium and magnesium. This has not been adjusted for in studies of the relationship between ascorbic acid and blood pressure.
Plasma levels of ascorbic acid have also been inversely correlated with blood pressure levels with a 5-mmHg difference in systolic as well as diastolic blood pressure levels between people with high and low ascorbic acid levels . In an RCT, large doses of ascorbic acid were found to lower blood pressure levels . There seems to be no relationship between plasma levels of tocopherol and blood pressure [50, 52], and little is known about the relationship between carotene and blood pressure.
The scientific information available is summarized in Table 1.
Although one Swedish study has shown elevated levels of plasminogen activator inhibitor (PAI-1) in subjects with low intake of fruit and vegetables , most of the scientific evidence indicates that there is no relationship between intake of ascorbic acid  or tocopherol  and various components of the fibrinolytic system.
Tocopherol in doses up to 1200 E daily does not affect platelet aggregation or bleeding time in healthy individuals . On the other hand, tocopherol deficiency may lead to increased platelet aggregation , which may be counteracted by addition of a tocopherol . In an RCT, an `antioxidant cocktail' with β-carotene, ascorbic acid, α-tocopherol and selenium reduced platelet aggregation . Platelet adhesiveness to the vessel wall is profoundly reduced also in healthy individuals when tocopherol is taken in doses around 400 mg daily . However, this seems not to be associated with the antioxidative properties of tocopherol but to other mechanisms (inhibition of protein kinases and membrane stabilization) [57, 59, 60].
Arterial wall structure and function
Within a large cohort study, carotid intima-media thickness did not correlate with plasma levels of β-carotene or α-tocopherol, but there was an inverse relationship with lutein and zeaxanthin, carotenoid compounds that are regarded as biomarkers of fruit and vegetable intake . In a small longitudinal study, low total antioxidant capacity of plasma was a strong predictor of the development of carotid atherosclerosis as evaluated by sonography .
Vasomotor reactivity reflects endothelial function. Patients with hypertension or spastic angina pectoris have impaired vasomotor responses, but this can be corrected by administration of ascorbic acid [63, 64] or α-tocopherol . Arterial endothelium-dependent vasodilator responsiveness has also been shown to improve by α-tocopherol treatment during 6 weeks in healthy postmenopausal women .
Systematic review – case–control studies
Three studies [66-68] were excluded because they did not meet the quality criteria for being included in the systematic review, two of them because of inappropriate storage of blood analysed many years after it was sampled. For one  only an abstract with few details was available and it was not included. Of the remaining 17 case–control studies, 14 were performed by a conventional technique, i.e. levels of antioxidant vitamins in plasma, serum or tissue have been analysed in subjects with and without a cardiovascular event. Three of the studies that met the quality criteria used a nested case–control design, in which intake or plasma levels have been measured before the event. The study characteristics and the results are summarized in 23Table 2.
Table 2. Case–control studies on intake or plasma/serum levels of antioxidants and the risk for cardiovascular events. When available, odds ratios (OR) or risk ratios (RR) adjusted for multiple confounders are presented
Significantly lower plasma, serum or tissue level of carotene/carotenoids was observed in five out of 11 case–control studies. In one of the negative studies, adipose tissue content of carotene was analysed  which minimizes the effects of the acute event (myocardial infarction or stroke) on circulating levels of antioxidant vitamins.
Five out of 16 case–control studies on tocopherol levels showed a difference between cases and controls, whereas none of the five studies on ascorbic acid did. No case–control studies have shown any connection between high levels of antioxidant vitamins and increased risk of acute cardiovascular events.
To summarize, there is some support from case–control studies for low plasma concentrations of β-carotene, possibly also of tocopherol, being linked to increased risk of myocardial infarction. On the other hand, the case–control studies have failed to establish a tight relationship between low plasma levels of ascorbic acid and increased risk of cardiovascular events. Plasma levels of antioxidant vitamins may change rapidly during an acute cardiovascular event, and comparisons between cases and controls should therefore be made with caution. It is notable that, in one of the largest case–control studies, tissue adipose content of carotene and tocopherol were similar in cases and controls .
Systematic review – cohort studies
In cohort studies, intake or plasma levels of antioxidant vitamins have been measured at baseline (and in a few studies during follow-up) and this has been related to the occurrence of cardiovascular events during long-term follow-up.
Of the 20 cohort studies where intake or plasma levels of antioxidant vitamins have been analysed, two have been excluded because of inferior quality [1, 72]. Of the remaining 18 studies, eight have been performed in the USA, eight in Europe, one in China, and one is a multinational study. In several of the studies, intake of antioxidant vitamins by the regular food and as food supplements have been reported together and therefore this is what has been used in the present review. The study characteristics are summarized in 35Table 3.
Table 3. Cohort studies on intake or plasma levels of antioxidants and the risk for cardiovascular events
In eight studies, a total of 943 cardiovascular events were recorded in 38 768 study subjects. When calculated as Peto's odds ratios, only one study  showed a significantly reduced risk of cardiovascular events in subjects with a high intake of carotene or carotenoids. Overall, there was a small risk reduction in high consumers of carotene/carotenoids, which did not reach statistical significance [–12%; 95% confidence interval (CI) –23 to +1%] (Fig. 1).
Several of the cohort studies used adjustments for differences in background variables at baseline between groups with different intake of antioxidant vitamins. After multiple adjustments, there was a trend towards lower risk for cardiovascular events in several of studies [74-78], but only in one instance  did this reach statistical significance.
Plasma or serum levels of β-carotene or carotenoids were measured at baseline in four cohort studies (Fig. 1). High levels predicted low risk for cardiovascular events during follow-up in three of the studies, and the overall odds ratio was 0.46 with a narrow 95% CI (0.37–0.58).
Nine cohort studies (a total of 1957 events in 82 379 participants) have assessed the relationship between high intake or high plasma levels of tocopherol (Table 3). In four of these studies, there was a statistically significant inverse relationship with the risk of cardiovascular events and, when all studies were taken together, high intake of tocopherol was associated with a highly significant reduction of cardiovascular events when compared with low intake, the odds ratio being 0.74 (95% CI 0.66–0.83) (Fig. 1).
The relationship between plasma tocopherol and future cardiovascular risk has been examined only in one study  with weak statistical power. Unexpectedly (and contrary to the findings for food intake of tocopherol in the same study), people with high plasma levels of tocopherol tended to have an increased risk of death from heart disease (Peto's odds ratio 1.61 (95% CI 0.78–3.33).
Data on intake or plasma levels of ascorbic acid have been reported in 11 cohort studies that have included a total of 50 000 participants who have experienced 2148 cardiovascular events during follow-up (Table 3). A statistically significant relationship between high intake of ascorbic acid and a low risk for cardiovascular disease was noted in two of the studies (Fig. 1). When analysed together, there was a somewhat reduced risk in high versus low consumers of ascorbic acid (odds ratio 0.89; 95% CI 0.79–0.99).
Plasma levels of ascorbic acid were measured in five cohort studies in which a total of 543 cardiovascular events occurred during follow-up of 13 018 individuals. In three of the five studies, people with high versus low plasma levels of ascorbic acid had a statistically significant reduction of the risk for cardiovascular events. As shown in Fig. 1, when taken together, the studies showed high ascorbic acid to be a powerful predictor of freedom cardiovascular disease during follow-up with an odds ratio of 0.58 (95% CI 0.47–0.72).
Longitudinal ecological study
In the Seven Countries Study, mortality from coronary heart disease was followed for 25 years in 16 cohorts in seven countries. Detailed data on number of cardiovascular events have not been given, but 25-year trends in coronary mortality did not correlate with trends in dietary intake of β-carotene, tocopherol or ascorbic acid .
Vitamins as dietary supplements
Six of the cohort studies have reported separately on the intake of antioxidant vitamins as dietary supplements [75, 76, 79, 81-82] of the six studies reporting on the use of β-carotene supplements was there an apparent beneficial effect on the risk for cardiovascular events. A statistically significant reduction of risk for at least some clinical manifestation(s) of cardiovascular disease was observed in three of the six studies on α-tocopherol [75, 81, 83], but in none of the four studies on ascorbic acid and in none of the three cohort studies reporting on the use of multivitamin preparations. A notable discrepancy between the studies was that in the Nurses' Health Study, tocopherol supplements but not high food intake of tocopherol was associated with a low risk for cardiovascular events , whereas the opposite was true for women in the Iowa cohort .
In summary, a general pattern emerges from the cohort studies, indicating that people who have a high intake of antioxidant vitamins, either by regular food or as food supplements, have a modest reduction of the risk for cardiovascular events. The risk reduction appears to be of similar magnitude for carotene, tocopherol and ascorbic acid. Plasma levels of carotene and ascorbic acid seem to be stronger predictors of future cardiovascular events than dietary intake, possibly because of poor precision in self-reported dietary intake .
Systematic review – randomized controlled trials
Four large, one intermediate size and three smaller randomized trials of antioxidant vitamins as dietary supplements for primary prevention in healthy subjects have been published, in which a severe cardiovascular event has been the outcome variable (Table 4). Three of these have been conducted in the United States, three in Europe, one in China and one in Australia. Three of the trials have been performed amongst high-risk individuals, smoking middle-aged men in Finland , smokers and asbestosis workers in the United States , and people with at least one cardiovascular risk factor or age above 65 years in Italy . One of the trials was performed in a population at particularly low risk, health professionals in the United States  and one in a Chinese population in which deficiencies in intake of micronutrients is common . Dietary supplements of β-carotene were tested in six of the eight studies, α-tocopherol in four studies and ascorbic acid in two studies (combined with molybdenum in one of them).
Table 4. Randomized controlled trials of dietery supplements of antioxidant vitamins in the primary prevention of cardiovascular diseases
All of the eight studies failed to show a beneficial effect of antioxidant supplementation on cardiovascular disease, despite a large number of participants (together 104 512) and observation years (∼699 000 in total). Serious adverse cardiovascular effects were noted in one of the studies in which there was a significantly increased risk for fatal or nonfatal intracerebral and subarachnoid haemorrhage in subjects taking α-tocopherol [85, 90]. It is possible that the antiplatelet properties of tocopherol [57, 91, 92] have contributed to an increased risk of bleeding. In the same trial, there was also a significant increase in the risk for intracerebral haemorrhage in the group taking β-carotene, but no obvious mechanism to explain this is at hand . In two of the trials, intake of carotene supplements was associated with an increased risk of lung cancer in cigarette smokers [85, 86].
The results of the RCTs are shown in Fig. 1. The odds ratio for cardiovascular events is very close to one for all three types of antioxidant vitamins, which implies that there is no net beneficial or adverse effects. The 95% CIs are narrow for carotene and tocopherol, which reduces the risk of the results being spurious. It should be pointed out that ascorbic acid has been tested only in one large (and then only together with molybdenum with a hypothesized effect on cancer) and one very small randomized trial.
Antioxidant vitamins in the secondary prevention of cardiovascular disease
This part of the article is not a systematic review, but it summarizes the most important studies on antioxidant vitamins in the secondary prevention in people who already have clinical manifestations of cardiovascular disease.
In the Lyon Diet Heart Study , patients with myocardial infarction were randomized to intense dietary advice focused on Mediterranean diet or to regular diet advice. During a more than 2-year long follow-up period, the risk of severe cardiovascular complications was reduced by half in those randomized to Mediterranean diet. This diet has a high content of α-tocopherol and ascorbic acid but differs from the regular diet also in a number of other important aspects, including fat quantity and quality.
Important studies on the intake of specific antioxidant vitamins as dietary supplements are listed in 58Table 5. Some beneficial effects have been reported in at least one of the outcome variables in four of the 14 studies. One study also reported that α-tocopherol was cost-effective in the secondary prevention after myocardial infarction because of fewer hospital readmissions . The positive results have, however, not been possible to repeat in several other studies, or the beneficial effects have been counterbalanced by less favourable effects.
Table 5. Randomized controlled trials of dietary supplementation of antioxidant vitamins in the secondary prevention in patients with manifest cardiovascular diseases
In a nonrandomized substudy within a randomized trial of cholesterol-lowering strategies after coronary by-pass surgery, patients who choose to take tocopherol supplements had less coronary artery disease progression by angiography. No corresponding effect was seen in those who took ascorbic acid supplements .
There have also been attempts to prevent myocardial injury during coronary by-pass operations by pretreatment with high doses of antioxidant vitamins. In an RCT, pretreatment with α-tocopherol and ascorbic acid did not affect various indicators of myocardial damage .
The treatment of intermittent claudication with tocopherol has had its proponents since the 1940s. As summarized in a Cochrane review, there have been several small randomized trials of tocopherol supplementation in patients with intermittent claudication published in the years 1953–75 . In total, only 265 patients have been randomized. The overall results are slightly encouraging, but they do not conclusively confirm reduced symptoms in the intervention group. No beneficial effects of tocopherol treatment was observed in a more stringent but still small, randomized trial not included in the Cochrane review .
It thus seems that the initial enthusiasm for antioxidants in the secondary prevention of new events in people with clinical manifest cardiovascular disease aroused by the positive results of the Cambridge Heart Antioxidant Study (CHAOS) trial  has weaned off as negative results of several large randomized trials have emerged in the last few years.
Ongoing RCTs are summarized in Table 6. There are three studies that have randomized (or are planning to randomize) close to 70 000 individuals together in primary prevention trials and there is one large trial on secondary prevention in women with clinical manifestations of cardiovascular disease.
Table 6. Ongoing randomized controlled trials of antioxidant vitamins to prevent cardiovascular disease
In a number of the observational studies, including some of the large cohort studies, there has been an association between high intake or high plasma levels of antioxidant vitamins and low risk of myocardial infarction and/or stroke. This relationship has been most clearly shown for tocopherol. On the other hand, the RCTs have all failed to confirm an effect of any of the antioxidant vitamins on the risk of cardiovascular events. Together, the eight RCTs of antioxidant vitamins as dietary supplements have covered close to 700 000 observation years, and more than 4000 fatal and nonfatal myocardial infarctions and strokes have been accumulated. This would give a sufficient statistical power to detect even a small effect (< 10% risk reduction) of antioxidant vitamins in the doses that have been used, if there were such an effect.
Our conclusion about ascorbic acid is based on only one trial performed in China. However, a similar conclusion was reached, when a meta-analysis of three relatively small randomized trials published in the years 1973–84 was conducted (trials published before 1989 were not included in our systematic review). These trials randomized a total of 1034 elderly, often sick people. There was no tendency for ascorbic acid supplementation to be of benefit (relative risk for all deaths was 1.08; 95% CI 0.93–1.23) .
•Most of them have not been designed to specifically test the protective effect of antioxidant vitamins on cardiovascular disease. The main hypothesis has most often been that antioxidant vitamins protect against cancer.
•Some of the largest trials have included only smokers and the relationship between smoking and oxidation is complex.
•In some of the studies only deaths from cardiovascular disease are reported, and data on effects of nonfatal myocardial infarction or stroke are missing.
•The duration of treatment may have been too short for any effect to be detected.
•Synthetic antioxidant vitamins have been used and it is not self-evident that these will have the same effects on cardiovascular disease as natural antioxidants.
•Doses of vitamins, especially tocopherol, may have been to low.
•Specific antioxidant vitamins have been used alone or in combinations of two. There are some theoretical advantages of using a cocktail of several antioxidant vitamins.
•Markers of oxidative stress or other intermediate end-points have not been followed in the large clinical trials to see if the vitamins have the expected antioxidative properties in the study populations.
•Physical exercise and low-fat diets with polyunsaturated fatty acids, often recommended for the prevention of cardiovascular disease, may actually increase oxidation and would therefore counteract any beneficial effects of antioxidant food supplements.
Some of this criticism may receive a response when the results on the ongoing RCTs become available.
Carotene and ascorbic acid have not only antioxidant but also prooxidant properties that are particularly evident at high concentrations of the vitamins [102, 103]. Ascorbic acid also enhances the absorption of dietary iron, a powerful prooxidant . A prooxidant activity of vitamins at high doses has been offered as an explanation not only for the lack of benefits on cardiovascular disease but for the apparent increased risk of cancer in two of the randomized trials [85, 86].
The scientific evidence emerging from ecological, cross-sectional, case–control and cohort studies is generally weaker than that of an RCT. A major problem with cohort studies is that people with high versus low intake of antioxidant vitamins differ in a number of aspects that can be only partly compensated for in the statistical analyses. Supplement use is one component in a cluster of healthy behaviour . Several studies have adjusted for differences in important prognostic variables (for instance smoking, alcohol consumption, elevated blood pressure, physical activity and socio-economic status). Other possible confounders like serum lipid levels have rarely been adjusted for.
In case–control studies, there are specific problems. One is that a relationship between low levels of antioxidants a cardiovascular event does not necessarily imply a causal relationship. It is possible that the ongoing atherosclerotic process is causing reduced levels of circulating antioxidant vitamins. In addition, in the acute phase of a myocardial infarction or a stroke there may be rapid changes in antioxidant vitamin levels and blood sampling after the event does not necessarily reflect the situation before onset of myocardial infarction or stroke. In attempts to overcome this problem, blood has been sampled very early or tissue content of lipid-soluble vitamins has been examined. Time series in the acute phase of myocardial infarction indicate, however, that the plasma levels of carotene and tocopherol do not change much during the first 24 h , so this may be a less serious problem than previously thought.
A weakness of most observational studies is also that intake or serum/plasma levels of antioxidant vitamins have been measured only on one occasion and that it has not been possible to account for variations occurring during long-term follow-up. This could possible reduce the possibilities of detecting a correlation. In two of the studies, dietary interviews have been repeated [76, 78]. There was a considerable discrepancy between what was reported on different occasions. This is partly because of the inaccuracies of various methods used to record dietary intakes, partly because of change in dietary habits during long-term follow-up. Many of the methods used to record dietary intake seem to have underestimated food intake. Therefore, it may be that it has been more difficult to show any relationship between antioxidant intake by the regular diet than by food supplementation . It must, however, be kept in mind that it is very hard to reach the same intake by regular food as by taking vitamin supplements – the content of tocopherol in multivitamin tablets is 3–4 times higher than that reached by normal food intake and the content of tocopherol may be 40–50 times higher .
Secular trends in cardiovascular disease epidemiology may add to the difficulties in interpreting the results. The first studies of antioxidants and cardiovascular disease where ecological, then followed case–control studies, cohort studies and finally randomized trials. As mortality from cardiovascular disease is declining rapidly in most affluent countries, it has become considerably more difficult to show effects of preventive interventions than when the first observational studies were performed. Much larger studies are needed today to reach the same statistical power.
With an increased intake of antioxidant vitamins in the general diet in those countries where the studies have been performed, the effects of high doses of antioxidant vitamins as dietary supplements may be more difficult to demonstrate. A larger share of subjects at high risk for cardiovascular disease may take other preventive medicines such as low-dose aspirin. It is also possible that, when indications of a beneficial effect of a certain preventive intervention are accumulating, more persons are excluded from RCTs and receive treatment already before its value has been fully documented. Together, all these factors may contribute to increase in difficulties in showing any effects of various cardiovascular interventions, at least as long as the effects are recorded as mortality from myocardial infarction and stroke.
•People affected by ischaemic heart disease and stroke and populations with a high occurrence of cardiovascular disease often have low intake or low plasma levels of antioxidant vitamins. This does not necessarily mean that low levels contribute to cause a cardiovascular event. Rather, it may reflect the fact that a low intake of antioxidant vitamins is one of several unfavourable lifestyle factors. Reduced plasma levels may also result from an emerging atherosclerotic process with consumption of antioxidant vitamins.
•In observational studies using case–control or cohort designs, people with high intake of antioxidant vitamins by regular diet or as food supplements usually have a low risk of myocardial infarction and stroke. The association is shown for carotene, ascorbic acid and tocopherol. The value of observational studies is limited by difficulties to control for all putative confounding factors. High intake of antioxidant vitamins is a component of a cluster of healthy behaviour.
•In RCTs, antioxidant vitamins as food supplements have convincingly been shown to have no beneficial effects on the risk for myocardial infarction or stroke. They cannot be recommended in the primary prevention against cardiovascular disease. This conclusion is based on solid scientific evidence for carotene and tocopherol. For ascorbic acid, the evidence is weaker. A report on increased risk of intracerebral and subarachnoid haemorrhage during treatment with carotene and tocopherol may caution against the use of antioxidant vitamins in healthy people.
•There is some support from observational studies that a low intake of fresh fruits and vegetables may confer an elevated risk for cardiovascular disease. This conclusion is not based on a systematic review of all scientific evidence available. Nevertheless, in view of other beneficial effects of a high and varied intake of fruits and vegetables, it seems reasonable to recommend it also in the protection against cardiovascular diseases.
•The results of several large ongoing randomized trials testing various combinations of antioxidant vitamins are awaited.