Vitamin C (ascorbic acid) is a water-soluble vitamin that must be consumed from external sources (primarily fruits and vegetables) because humans cannot synthesize it endogenously. In addition to being an anti-oxidant , vitamin C improves endothelial-dependent vascular function, reduces monocyte adhesion to endothelium, reduces vascular smooth muscle cell apoptosis, and impacts oxidative modification of low-density lipoprotein [1]. However, while diets high in fruits and vegetables are associated with reduced cardiovascular disease (CVD) risk [2-5], anti-oxidant vitamin supplements, including vitamin C, have failed to reduce CVD events in randomized trials [6,7]. Additional evidence suggests there may be a differential impact on CVD between vitamin intake from supplements vs food [8-10]; however mechanistic pathways are not understood.

Previous studies have shown that carotid artery intimal medial thickness (IMT), measured by B-mode ultrasound, is a measure of pre-clinical atherosclerosis, and is a correlate of angiographically determined coronary atherosclerosis [11]. Progression of IMT is associated with conventional atherosclerotic risk factors [12] and CV events [12,13]. We evaluated the relations between intake of vitamin C from supplements vs. food with progression of carotid artery IMT.

Statistical analyses were performed using the maximum likelihood SAS procedure for mixed models (repeated measures with unconstrained covariance structure) [21]. Change in carotid IMT from baseline (μm/year) was regressed on quintiles of dietary intake and five groups of supplement intake (non-users and quartiles among users). Tests of linear trend were performed by placing the constraint β'L=0, where β is a vector of coefficients for quintiles and L' = (-1 -0.5 0 0.5 1). For purposes of graphic depiction and a posteriori inspection for non-linear dose-response patterns, least squares means and standard errors were also estimated.

Three models were estimated. Model 1 included age in years, sex, sex by age interaction, logarithm of total energy intake, two categorical variables for supplement intake (vitamin C and low-dose multi-vitamins), categorical variables for dietary intake (vitamin C from food), and two continuous variables (tablets per week) for stress and therapeutic multiple vitamins. Model 2 added the covariates of smoking habit (current, former smoker), ethnicity (Hispanic, Black, and Asian, non-Hispanic white and other), alcohol intake (quintiles), physical activity (quintiles of number of strenuous activity sessions per week), use of anti-hypertensive medication, use of lipid lowering medication, diagnosis of diabetes, and number of cold/flu episodes in the previous year. Model 3 added the covariates of serum cholesterol level, systolic blood pressure and body mass index (kg/m²) as continuous variables. If dietary and supplement intake mediate their effects via pathways that include an impact on the biological variables in Model 3, then Model 3 may underestimate effects.

Characteristics of the study subjects by supplement intake of vitamin C, low-dose multi-vitamins, and vitamin C intake from foodare shown in Tables 1 and 2.Vitamin C intake from food was strongly associated with male sex (Table 2), possibly due to increased food intake among men. Non-smokers had a higher intake of vitamin C from food than smokers (p=0.006).Intake of low-dose multi-vitamins was moderately positively correlated with intake of vitamin C supplements, but did not reach statistical significance (p=0.21). Follow-up was conducted among 87% (n=500) of the cohort (480 persons at 1.5-years and 447 at 3-years). No significant differences in demographic characteristics (p>0.10) between those who completed the study and subjects that were lost to follow-up.

Plasma AA at baseline increased with both dietary and supplemental vitamin C intake. Among persons with no intake of supplement containing vitamin C, plasma AA increased monotonically from) 26.3±21 to 39.1±19 (mean±SD) μmol/L across quintiles of vitamin C from food (P-trend<0.001). After adjustment for intake of vitamin C from food, plasma AA among those who took vitamin C supplements was further elevated by 3.5±20, 14.8±16, 10.2±16, and 19.7±15 μmol/L across quartiles of vitamin C intake from vitamin C supplements (P-trend<0.001).

IMT and Vitamin C Supplement Intake

Progression rate of IMT among subjects reporting no intake of vitamin C supplements and those who did report use of supplements are presented in (Table 3 and Fig. 1) by supplement dose quartiles. Among persons in the highest quartile (857-5000 mg/day) of vitamin C supplement intake compared to those not consuming any vitamin C supplements, carotid IMT progression increased three-fold (20.3±2.6 versus 7.6±1.8 μm/year (mean±SD); p<0.001).As intake of vitamin C supplements increased, IMT progression rate increased (p-trend=0.0009, Model 1). Adjustment for potential confounders in Models 2 and 3 did not alter this pattern (Table 3).

Fig. (1). IMT progression rate by quartiles of intake of vitamin C supplements and sex (A) and baseline smoking status (B). Means are adjusted for the covariates in Model 3 (intake of multiple vitamin supplements; intake of vitamins from food; age, ethnicity and cardiovascular risk factors). Error bars are standard errors. The p value is for trend test in progression rate of IMT across vitamin C supplement dose quartiles.

To our knowledge, this is the first study to report greater progression of carotid IMT in those who took vitamin C supplements, compared to a relatively less progression of carotid IMT among those with a greater vitamin C intake from food. The adverse association of vitamin C supplement intake was most evident in those subjects with elevated serum cholesterol level and among those with hypertension. The current results provide a potential mechanistic understanding for the differential effects of vitamin C in supplements compared to food.

The finding that vitamin C supplements had an adverse association with atherosclerosis, while vitamin C intake from food may be protective is consistent with studies that point to a differential impact between vitamin intake from food and vitamin intake from supplements [8-10]. For example, dietary vitamin E intake has been associated with beneficial effects [22, 23] and, at doses of 50 to 537 mg/day, no effect on CVD risk was seen [24-27], but a meta-analyses on high dose vitamin E supplementation has suggested harm [28]. In a meta-analysis of 14 cohort studies on vitamin C by Ye et al. showed that dietary intake of vitamin C (but not supplemental vitamin C) had an inverse association with coronary heart disease risk [4]. Several explanations for this difference between supplements vs. intake from food for vitamins exist. Studies have reported equal bioavailability of ascorbic acid from food vs. supplemental ascorbic acid from vitamins [29-31], and there is no distinction between the chemical structures of vitamin C from these two sources, but vitamin C from fruits and vegetables may be accompanied by interacting components and cofactors, such as carotenoids and flavonoids, that have beneficial effects compared to vitamin C supplements [32, 33]. This theory is supported by epidemiological studies that demonstrate low β-carotene concentrations increase the risk of cardiovascular disease mortality and that increased flavonoid intake from food is associated with decreased risk of heart disease and mortality [34-37]. These studies are mostly dietary observational trials therefore isolating the individual impact of each antioxidant is not possible. Furthermore, large prospective clinical trials evaluating carotenoids, flavonoids, and the other 8000 phenol derivatives that are found in fruits and vegetables that have antioxidant properties [32], independently are lacking.

The results of our study are consistent with several cardiovascular primary and secondary prevention studies that failed to show benefit with vitamin C supplementation, when used alone or as part of an anti-oxidant cocktail [6, 38-41]. Two large-scale randomized trials tested the individual effect of vitamin C supplements on primary CVD prevention and found no reduction in the incidence of major cardiovascular events [6, 7]. The Physicans’ Health Study II, a randomized controlled trial, reported a lack of effect of vitamin C supplement on major CVD events in 14,641 US male physicians over an 8 year follow-up period [7]. Similarly, no CVD benefits of vitamin C supplement were observed in the Women’s Antioxidant Cardiovascular Study in 8,171 high-risk women [6]. More recently, a large-scale, placebo-controlled, randomized trial reported a possible adverse interaction between vitamin C and the combination therapy with folic acid, vitamin B₆, and vitamin B₁₂ on the CVD events in a high-risk population of women over 7.3 years of follow-up [42]. The risk of CVD events for folic acid was higher among those receiving active vitamin C than those receiving placebo (p-value for interaction=0.03) [42]. Of note, vitamin supplementation (C, D, E) in general has not been shown to consistently reduce the risk of diseases, such as cancer, where theoretically the role of anti-oxidants vitamins should be beneficial [43-46].

When vitamin C supplements are administered in conjunction with other anti-oxidant supplements, effects on early atherosclerosis progression are inconsistent [23, 47, 48]. One study reported that combined vitamin C and E supplementscan slow down the progression of IMT in men [23], however a larger trial conducted on 1162 apparently healthy subjects found that carotid IMT was not different between participants who received a combination of vitamin C, E, beta-carotene, and selenium versus placebo [47]. Waters et al. [48] confirmedthat a high-dose antioxidant vitamin supplement of 800 IU of vitamin E and 1000 mg of vitamin C per day compared to placebo in post-menopausal women provided no reduction in CVD events; however, they showed an increase in all-cause mortality in post-menopausal women with anti-oxidant vitamin consumption compared to women with placebo (HR, 2.8; 95% CI 1.1-7.2). Notably, the dose of vitamin C in this study associated with an increase in all-cause mortality was 2-fold higher than other studies that did not find similar results. These observations mirror our finding of greater IMT progression in subjects with increased supplementary vitamin C intake.

Additional data suggest important adverse interactions between vitamin supplements and prescribed medications. An angiographic trial found that an antioxidant cocktail (including 500 mg vitamin C and 400 IU vitamin E) reduced the beneficial effect of statins and niacin on coronary atherosclerosis [49]. It was postulated that this effect may have been due to an adverse effect of the vitamin supplement on serum high-density lipoprotein (HDL) [50]. We observed an adverse association of vitamin C supplementation with carotid IMT progression among subjects with relatively higher serum cholesterol levels (p-value for interaction=0.01), however, because only a few of our subjects were taking cholesterol lowering medications such as fibrates, niacin, and low-potency statins, we were unable to evaluate this interaction. Our results suggest that possible adverse association of vitamin C supplements may be even larger in coronary patients with elevated cholesterol similar to the results by Cheung et al. [50].

Strengths of our study include longitudinal measurements of early atherosclerosis concurrent with rigorously measured supplement and dietary vitamin C intake. The focus on early atherosclerosis avoids changes of intake in response to diagnosis, and concurrent measurement reduces the dilution of associations that occurs when baseline measurements of time-varying characteristics are related to future events. Confidence in the validity of measurements was enhanced by IMT reproducibility [15] and validation of self-reported dietary intake by plasma levels of AA. In addition, the high prevalence of supplement use and a wide range of intake levels in this Southern California cohort increased statistical power. The high participation rate after stratified random sampling from a population with uniform health insurance avoids numerous potential confounding factors. Confidence in the reliability of the observed associations between supplement intake and carotid IMT progression is supported by the consistency of findings across sexes and smoking status.

Due to the observational design of our study, there are inherent limitations. Unmeasured factors may be confounded with measured intake of supplements and food, or residual confounding by poorly measured risk factors may dilute covariate adjustments [21]. For example, the reduction in carotid IMT progression observed in the highest quintile of dietary vitamin C intake was attenuated by adjustment for smoking is suggestive of residual confounding. Previous studies have shown that smokers tend to have lower levels of Vitamin C, and therefore the recommended dietary allowance of Vitamin C for smokers is higher [51]. There was no clear pattern of association between supplement intake and lifestyle or biological CV risk factors, which argues against confounding by a general "healthy lifestyle" factor. In addition, data regarding supplement use and dietary patterns were obtained through memory recall which may be error prone.The present study was only an anatomic assessment of IMT; therefore, we cannot draw conclusions about plaque morphology or metabolism, which impacts stability of the plaque and causes CV events.

In summary, our findings suggest that intake of vitamin C supplements has an adverse association with early atherosclerosis while dietary vitamin C from food has a protective association. Our results further suggest that the adverse association of vitamin C supplements may be relatively greater among subjects with elevated cholesterol levels. The current results provide a potential mechanistic understanding for the observed differences between Vitamin C in supplements vs food in prior studies. Based on our observations, vitamin C supplementation does not appear to be currently advisable for the prevention or treatment of CVD.