Fruit and vegetable consumption has been associated with a reduced risk of several diseases including CVD(Reference Bazzano, He and Ogden1). Studies have suggested that chronic low-grade inflammation and free radicals in excess may play a role in the development of atherosclerotic plaques subsequently leading to diseases in the cardiovascular system(Reference Ross2, Reference Steinberg, Parthasarathy and Carew3). Antioxidants, usually found in fruits and vegetables, may be beneficial by reducing the levels of inflammation and free radicals in vitro (Reference Ames, Shigenaga and Hagen4) but the role of antioxidants in vivo in atherosclerosis and CVD is still largely unknown and under constant debate. Epidemiological studies have shown that humans with either a high dietary intake of or a high blood concentration of α-tocopherol, β-carotene and ascorbic acid have a lower risk of developing CVD(Reference Asplund5, Reference Clarke and Armitage6). On the other hand, large placebo-controlled intervention studies have not shown any beneficial cardiovascular effects of α-tocopherol, β-carotene or ascorbic acid supplementation(Reference Clarke and Armitage6–Reference Eidelman, Hollar and Hebert8). From these contradictory data it may be concluded that the regulation of antioxidants, inflammation and free radicals in atherosclerosis is a complex process and yet not fully understood. One approach to acquire more knowledge of the role of antioxidants in atherogenesis may be to study the impact of dietary vitamin intake and to relate these vitamins to the levels of inflammation and free radicals in vivo.
Inflammatory processes can be studied by the quantification of cyclo-oxygenase (COX)-induced or cytokine-mediated products(Reference Helmersson, Vessby and Larsson9). Prostaglandins and thromboxanes are derived from arachidonic acid; COX catalyses their formation endogenously. These prostanoids are important mediators of vasoconstrictive and inflammatory processes(Reference Bergström10, Reference Vane11). PGF2α is one of the major prostaglandins formed at sites of inflammation(Reference Willoughby12, Reference Mutschler, Eriksson and Wikström13) and its formation in vivo is preferably quantified by the measurement of 15-keto-dihydro-PGF2α(Reference Samuelsson14), which is a major metabolite of PGF2α in plasma. 15-Keto-dihydro-PGF2α has been shown to be a potent indicator of PGF2α formation and COX-mediated inflammatory processes in vivo (Reference Basu15, Reference Basu and Eriksson16).
F2-isoprostanes, structural isomers of PGF2α, are formed during free radical-catalysed (non-enzymic) peroxidation of arachidonic acid(Reference Morrow, Hill and Burk17). 8-Iso-PGF2α, a major F2-isoprostane, is a reliable indicator of oxidative stress in vivo. Enhanced formation of F2-isoprostanes has been associated with coronary artery disease(Reference Shishehbor, Zhang and Medina18) and several risk factors for atherosclerosis including diabetes, obesity and hypercholesterolaemia, as reviewed earlier(Reference Basu and Helmersson19).
In a recent study we showed that high levels of the antioxidant Se in serum were associated with reduced formation of the inflammatory mediator PGF2α and the oxidative stress indicator F2-isoprostanes after a long follow-up(Reference Helmersson, Ärnlöv and Vessby20). The aim of the present study was to study the relationships of registered dietary intake of other antioxidants, namely ascorbic acid, α-tocopherol and β-carotene, with levels of inflammation (PGF2α, high-sensitive C-reactive protein (CRP) and IL-6 formation) and free radicals (F2-isoprostane formation) in a 7-year follow-up study.
Materials and methods
Study population
The present study is based on 704 study participants from the re-investigations of the Uppsala Longitudinal Study of Adult Men (ULSAM) in 1990–3 and 1997–2001. This population-based cohort originally started in 1970–3, when all men born 1920–4 and living in Uppsala (Sweden) were invited to participate in a health screening at age 50 years as previously described(Reference Hedstrand21). The cohort was re-investigated after 20 and 27 years. Out of the re-investigated men, 704 were eligible for evaluation of inflammation and oxidative stress at age 77 years and had dietary data from the re-investigation at age 70 years. The Ethics Committee at Uppsala University approved the study and all participants gave informed consent.
Dietary assessment of vitamins
The intake of dietary ascorbic acid, α-tocopherol and β-carotene was estimated with the use of a 7 d pre-coded food record as part of the investigation at age 70 years. The validity and design of the dietary assessment has previously been described in detail(Reference Smedman, Gustafsson and Berglund22). In brief, a dietitian instructed all subjects to record their dietary intake using a 7 d pre-coded food diary. Daily intakes of the antioxidants were calculated using a computerised program (MATs; Rudans Lättdata, Västerås, Sweden) and the Swedish National Food Administration database.
Biochemical analysis of indicators of inflammation and oxidative stress
Urine was collected over 24 h. Blood samples were drawn in heparinised tubes in the morning after an overnight fast and plasma was separated. Both urine and plasma were stored at − 70°C until analysis. Free 8-iso-PGF2α in urine was analysed by a RIA as previously described by Basu(Reference Basu23). The intra-assay CV was 14·5 % at low and 12·2 % at high concentrations. Urinary 15-keto-dihydro-PGF2α was analysed by a RIA developed as previously described by Basu(Reference Basu15). The intra-assay CV was 12·2 % at low and 14·0 % at high concentrations. Levels of 8-iso-PGF2α and 15-keto-dihydro-PGF2α were corrected for urinary creatinine analysed as described previously(Reference Helmersson and Basu24). IL-6 was analysed by a high-sensitivity ELISA kit (IL-6 HS; R&D Systems, Minneapolis, MN, USA). The total CV of the method was 7 % and inter-assay CV was 5 %. High-sensitive CRP measurements were performed by latex-enhanced reagent (Dade Behring, Deerfield, IL, USA) with the use of a Behring BN ProSpec analyser (Dade Behring). Intra-assay CV of the high-sensitive CRP method was 1·4 % at both 1·23 and 5·49 mg/l. One high-sensitive CRP outlier (116 mg/l) was excluded from the statistical analysis.
Statistical analysis
Variables with skewed distribution, according to the Shapiro–Wilk test (W < 0·95), were log-transformed to reach normal distribution. Associations between dietary vitamins and eicosanoids and cytokines, respectively, were tested in linear regression models with the respective vitamin as the dependent variable and eicosanoids and cytokines as independent variables. BMI as a continuous variable, diabetes (plasma glucose ≥ 7·0 mmol/l or anti-diabetic medication), hyperlipidaemia (serum cholesterol >6·5 mmol/l or TAG >2·3 mmol/l or lipid-lowering medication), hypertension (office blood pressure >140/90 mmHg or anti-hypertensive medication) and smoking at baseline were considered as potential confounders based on associations earlier reported from this cohort or from other studies(Reference Helmersson, Vessby and Larsson9, Reference Basu and Helmersson19, Reference Basu25). Dietary vitamins were entered into the models as continuous variables and in quartiles. P values < 0·05 were regarded as statistically significant. Calculations were performed with Stata 8.2 (StataCorp LP, College Station, TX, USA).
Results
Baseline characteristics of the study population and dietary vitamin intake
In the cohort of men aged 70 years, 9 % had diabetes mellitus, 29 % hypertension, 36 % hyperlipidaemia and 16 % were active smokers. Mean BMI was 26·3 (sd 3·27) kg/m2. The mean dietary intake of ascorbic acid was 54·2 (sd 35·4) mg/d. Quartile 1 corresponds to 4·73–29·8 mg/d; 2 to 29·9–46·5 mg/d; 3 to 46·6–68·1 mg/d; 4 to 68·2–285 mg/d of ascorbic acid intake. The mean dietary intake of α-tocopherol was 5·80 (sd 2·03) mg/d. Quartile 1 corresponds to 1·62–4·33 mg/d; 2 to 4·34–5·40 mg/d; 3 to 5·41–6·66 mg/d; 4 to 6·68–34·9 mg/d of α-tocopherol intake. The mean dietary intake of β-carotene was 1·74 (sd 1·41) mg/d. Quartile 1 corresponds to 0·08–0·79 mg/d; 2 to 0·80–1·38 mg/d; 3 to 1·39–2·20 mg/d; 4 to 2·21–11·9 mg/d of β-carotene intake.
Dietary vitamins and PGF2α (indicator of cyclo-oxygenase-mediated inflammation)
The registered intake of dietary ascorbic acid was negatively associated with PGF2α formation linearly (regression coefficient (b) − 46·9; P < 0·05) and in quartiles (Fig. 1) at follow-up. The linear association did not substantially change when adjusted for potential confounders at baseline, but the association between the highest and lowest quartile of ascorbic acid intake was attenuated (P = 0·085).
PGF2α formation (pmol/mmol creatinine) in quartiles of dietary intake of ascorbic acid (a), α-tocopherol (b) and β-carotene (c), where quartile 1 is low intake. PGF2α formation is estimated as urinary 15-keto-dihydro-PGF2α. Note that the y axis is cut at 200 pmol/mmol. Values are means, with standard errors represented by vertical bars. *Mean value is significantly different from that for quartile 1 (P < 0·05). †Mean value is marginally significantly different from that for quartile 1 (P = 0·06). For ascorbic acid, P for trend = 0·007; for α-tocopherol P for trend = 0·06; for β-carotene P for trend = 0·21.
Intake of α-tocopherol was negatively associated with PGF2α formation at a borderline significance level linearly (b = − 35·1; P = 0·06) and in quartiles (Fig. 1). These results were substantially attenuated when adjusted for potential confounders (P = 0·15).
Intake of β-carotene in the diet was not significantly associated with PGF2α formation (Fig. 1).
Dietary vitamins and high-sensitive C-reactive protein (indicator of cytokine-related inflammation)
The registered intakes of dietary ascorbic acid and α-tocopherol were negatively associated with CRP linearly (b = − 0·14, P < 0·001 and b = − 0·15, P < 0·001, respectively) and in quartiles (Fig. 2) at follow-up. The results did not substantially change when adjusted for potential confounders at baseline.
C-reactive protein (mg/l) in quartiles of dietary intake of ascorbic acid (a), α-tocopherol (b) and β-carotene (c), where quartile 1 is low intake. Values are means, with standard errors represented by vertical bars. Mean value is significantly different from that for quartile 1: *P < 0·05, **P < 0·01, ***P < 0·001. For ascorbic acid, P for trend < 0·001; for α-tocopherol P for trend < 0·001; for β-carotene P for trend = 0·14.
The estimated intake of β-carotene was not linearly associated with high-sensitive CRP (data not shown), but quartiles with high intake (quartiles 4 and 2) of β-carotene were associated with lower levels of high-sensitive CRP compared with the quartile with the lowest intake of β-carotene (Fig. 2). However, adjusting for potential confounders at baseline removed these associations.
Dietary vitamins and interleukin-6 (indicator of cytokine-related inflammation)
Intakes of dietary ascorbic acid and α-tocopherol were linearly negatively associated with IL-6 at follow-up (b = − 0·06, P < 0·05 and b = − 0·07, P < 0·01, respectively). The quartile with the estimated highest intake of ascorbic acid and α-tocopherol had the lowest levels of IL-6 compared with the quartile with the lowest intake (Fig. 3). The results were attenuated when adjusted for potential confounders: P = 0·066 (ascorbic acid) and P < 0·05 (α-tocopherol).
IL-6 (ng/l) in quartiles of dietary intake of ascorbic acid (a), α-tocopherol (b) and β-carotene (c), where quartile 1 is low intake. Values are means, with standard errors represented by vertical bars. Mean value is significantly different from that for quartile 1: *P < 0·05, **P < 0·01. For ascorbic acid, P for trend = 0·009; for α-tocopherol P for trend = 0·002; for β-carotene P for trend = 0·18.
The estimated intake of β-carotene was not linearly associated with IL-6 (data not shown), but quartiles with high intake (quartiles 3 and 2) of β-carotene were associated with lower levels of IL-6 compared with the quartile with the lowest intake of β-carotene (Fig. 3). However, adjusting for potential confounders at baseline removed these associations.
Dietary vitamins and F2-isoprostanes (indicator of oxidative stress)
Intakes of dietary ascorbic acid, α-tocopherol and β-carotene were linearly negatively associated with F2-isoprostane formation at follow-up (b = − 41·3, P < 0·05; b = − 34·1, P < 0·05; b = − 48·0, P < 0·05, respectively). The quartile with the estimated highest intake of ascorbic acid as well as the highest intake of β-carotene had the lowest levels of F2-isoprostane formation compared with the quartile with the lowest intake (Fig. 4). The results were unchanged (ascorbic acid, α-tocopherol) or even stronger (β-carotene) when adjusted for potential confounders.
F2-isoprostane formation (pmol/mmol creatinine) in quartiles of dietary intake of ascorbic acid (a), α-tocopherol (b) and β-carotene (c), where quartile 1 is low intake. F2-formation is estimated as urinary 8-iso-PGF2α. Note that the y axis is cut at 100 pmol/mmol. Values are means, with standard errors represented by vertical bars. *Mean value is significantly different from that for quartile 1 (P < 0·05). †Mean value is marginally significantly different from that for quartile 1 (P = 0·05). For ascorbic acid, P for trend = 0·01; for α-tocopherol P for trend = 0·13; for β-carotene P for trend = 0·03.
Discussion
The novel finding in the present study is that the registered dietary intake of ascorbic acid and α-tocopherol was inversely associated with PGF2α formation, which may be interpreted as the higher the intake of food containing ascorbic acid and α-tocopherol the lower the level of COX-mediated inflammation. The association with α-tocopherol was, however, somewhat weaker, especially when adjusting for potential confounders. Studies evaluating the relationship between ascorbic acid and α-tocopherol, respectively, and PGF2α are very sparse and this may be due to the earlier difficulties in quantification and estimation of PGF2α formation. Serum α-tocopherol and PGF2α formation was not associated cross-sectionally in this cohort as previously reported(Reference Helmersson, Vessby and Larsson9). On the other hand, there was a positive correlation found between serum γ-tocopherol and PGF2α formation(Reference Helmersson, Vessby and Larsson26), although γ-tocopherol was not studied in the present study. An earlier published study showed that supplementation with α-tocopherol at relatively low doses (200 mg α-tocopherol/d) for 2 weeks or gradually increased doses (16–400 IU (10·7–266·7 mg) α-tocopherol/d) for 4 weeks did not change the level of PGF2α formation(Reference Smedman, Vessby and Basu27, Reference Basu, Lee and Dunster28).
A higher intake of food containing ascorbic acid, α-tocopherol and to some extent also β-carotene was also associated with lower levels of CRP and IL-6, indicating a lower level of cytokine-mediated inflammation related to higher vitamin intake. These associations are supported by a number of cross-sectional studies showing an inverse correlation between ascorbic acid in serum, blood, plasma and diet, and CRP levels(Reference Szeto, Kwok and Benzie29–Reference Wannamethee, Lowe and Rumley32). The relationship between α-tocopherol and CRP seems to be far more complicated in the literature. Plasma α-tocopherol has been shown to be inversely correlated with CRP levels in one study(Reference McMillan, Talwar and Sattar33) but serum α-tocopherol and CRP levels were actually positively correlated in an earlier study published from this present cohort(Reference Helmersson, Vessby and Larsson9). Some studies have shown a lack of association between α-tocopherol in plasma and serum and CRP(Reference Szeto, Kwok and Benzie29, Reference Chang, Chen and Ke34). However, to our knowledge no study has evaluated the dietary intake of α-tocopherol and CRP or IL-6. Intake of ascorbic acid via supplements has revealed conflicting results; some studies have shown a reduced level of CRP(Reference Block, Jensen and Dietrich35) while others have failed to show CRP-lowering effects of ascorbic acid supplementation(Reference Upritchard, Sutherland and Mann36, Reference Tahir, Foley and Pate37). The intake of α-tocopherol supplements has been shown to lead to reduced levels of CRP in one study(Reference Upritchard, Sutherland and Mann36) while a combination of α-tocopherol and ascorbic acid did not alter CRP(Reference Tahir, Foley and Pate37, Reference Bruunsgaard, Poulsen and Pedersen38) or IL-6(Reference Bruunsgaard, Poulsen and Pedersen38). A number of earlier published studies of β-carotene estimated in blood, serum and plasma and CRP confirm the weak inverse association between dietary β-carotene and CRP seen in the present study(Reference van Herpen-Broekmans, Klopping-Ketelaars and Bots30, Reference Ford, Liu and Mannino31, Reference McMillan, Talwar and Sattar33, Reference Chang, Chen and Ke34, Reference Erlinger, Guallar and Miller39). There is also a study that shows an inverse association between β-carotene and IL-6(Reference Hu, Reuben and Crimmins40), a correlation that we could not find in the present study.
A higher intake of food containing ascorbic acid, α-tocopherol and β-carotene was related to reduced levels of F2-isoprostanes. This would indicate that the higher the intake of these vitamins in the diet the less oxidative stress in vivo. We have previously reported an inverse association between α-tocopherol in serum and F2-isoprostane formation(Reference Helmersson, Vessby and Larsson9). A supplementation study with vegetable soup containing ascorbic acid showed that F2-isoprostane formation could be decreased by intake of this soup, but a control group was missing in this study(Reference Sanchez-Moreno, Cano and de Ancos41). There are a number of supplementation studies concerning α-tocopherol and ascorbic acid which evaluated the effects on F2-isoprostane formation as reviewed earlier(Reference Basu and Helmersson19). In brief, F2-isoprostane formation in healthy humans is generally not affected by α-tocopherol or ascorbic acid supplementation. In conditions where F2-isoprostane formation is increased, such as type 2 diabetes, homocysteinaemia and hypercholesterolaemia, α-tocopherol supplementation may be beneficial in reducing the F2-isoprostane level if given in appropriate doses and longer duration(Reference Roberts, Oates and Linton42). Ascorbic acid supplementation may decrease the elevated F2-isoprostane formation in smokers(Reference Dietrich, Block and Hudes43, Reference Reilly, Delanty and Lawson44). Supplementation with β-carotene alone did not alter F2-isoprostane formation(Reference Mayne, Walter and Cartmel45).
To summarise, a high intake of food rich in ascorbic acid seems to show the most obvious association with reduced levels of all investigated indicators of inflammation (PGF2α, CRP, IL-6) and oxidative stress (F2-isoprostanes) in the present study. A high intake of food rich in α-tocopherol was also related to reduced levels of PGF2α, CRP, IL-6 and F2-isoprostanes, but the associations were weaker and in some case were affected by confounding factors. On the other hand, a high intake of β-carotene from the diet was only related to reduced F2-isoprostane formation. It may be that these antioxidants by their different properties and mechanisms affect inflammation and oxidative stress differently. Ascorbic acid is a water-soluble chain-breaking antioxidant, tocopherols and tocotrienols (vitamin E) are lipid-soluble chain-breaking antioxidants and carotenoids (including β-carotene) are lipid-soluble antioxidants that effectively scavenge singlet oxygen. The earlier published inverse associations between Se (which has antioxidative properties mainly through the glutathione peroxidase system), and PGF2α and F2-isoprostanes, but not CRP, IL-6 or serum amyloid A protein, may support this theory(Reference Helmersson, Ärnlöv and Vessby20). It may also be true that the associations found in the present study merely reflect an association between a large intake of fruits and vegetables in general and a reduced level of inflammation and oxidative stress and that this association may be mediated through an unknown factor or mechanism not studied here.
A recommended daily intake of vitamins according to the Swedish National Food Administration today is 75 mg ascorbic acid, 10 mg α-tocopherol and 10·8 mg β-carotene. It may be noted that only those in the highest quartile of the respective vitamin intake achieved these health recommendations. A general well-known problem of food registrations is the constant and sometimes selective underestimation of the dietary intake. However, a possible underestimation of vitamin intake would rather tend to diminish the observed associations in the present study.
It may also be discussed whether these studied vitamins have any direct anti-inflammatory effect or if the reduced levels of inflammation seen related to a high intake of vitamins are secondary to their well-documented antioxidative effect. There are several mechanisms described in the literature by which free radicals (oxidative stress) may be linked to the inflammatory pathway. A recent study showed that intravenous infusion of F2-isoprostanes, specifically 8-iso-PGF2α, in rabbits causes an increase in COX-catalysed PGF2α formation(Reference Basu46). Phospholipase A2, which hydrolyses phospholipids in the membrane allowing arachidonic acid and subsequently prostaglandins to form, can be activated by H2O2 (reactive oxygen species)(Reference Chan, Yurko and Fishman47, Reference Rao, Runge and Alexander48). Reactive oxygen species might also up-regulate phospholipase A2 by the IκB kinase (IKK)/NF-κB pathway(Reference Senftleben and Karin49). NF-κB is a nuclear transcription factor regulating cytokines (IL-6), cell-adhesion molecules, acute-phase proteins (CRP), phospholipase A2 and COX-2 in a complex series of biochemical processes.
The present study was carried out in a fairly large age-, sex- and ethnicity-matched population, which is an advantage when studying inflammation and free radicals, as these levels may vary due to the above-mentioned factors(Reference Basu and Helmersson19). On the other hand, the results may have limited generalisability to other age and ethnic groups and women. Further, the studied subgroup of the cohort may be healthier than the general population. The studied subgroup may also be healthier than the whole unselected cohort at age 70 years, as can be noted from slightly higher percentages of diabetes mellitus (15 %), smokers (21 %) and hypertension prevalence (32 %) in the whole unselected cohort at age 70 years (compared with the subgroup followed up with biomarkers). Hyperlipidaemia prevalence and BMI were the same in the whole cohort at age 70 years as in the subgroup followed up 7 years later. However, the healthy cohort effect would probably bias the results towards the null hypothesis and not lead to overestimation of the associations. Although the study has a longitudinal design with a follow-up of 7 years, it cannot be excluded that the associations found also would be found in a cross-sectional analysis. A limitation in the study is that we did not have data on vitamin supplements. However, it can be speculated that general Swedish men of age 70 years in 1990–3 did not take vitamin supplements to a large extent.
The dietary intake of vitamins was unfortunately not validated at follow-up. The Swedish National Food Administration concluded based on two food surveys (in 1989 and 1997) that the intake of tocopherols in the diet did not change, while the intake of ascorbic acid in the diet increased slightly and the intake of carotenes decreased slightly in men aged 18–74 years during the years 1989–97(Reference Becker and Pearson50). Changes in vitamin intake in the elderly are expected to be smaller. However, it is very unlikely that changes in nutrient intake during the years of follow-up would bias the results towards a type 1 error; rather, the associations would be more difficult to detect. Serum concentrations of α-tocopherol at baseline and after 7 years of follow-up correlated well (R 0·54; P < 0·0001), which also supports the assumption of small changes in consumption during follow-up.
In conclusion, the present study is the first to report that a registered large intake of a diet rich in several vitamins is associated with reduced inflammation, mediated by both COX and cytokines, and reduced oxidative stress 7 years later. Whether these effects seen are a consequence of a general effect of a healthy diet or a more direct consequence of intake of these specific vitamins can, however, not be concluded from the results of the present study. Yet, these anti-inflammatory and antioxidative effects of a large intake of food rich in vitamins may be linked to the beneficial effects of fruits and vegetables seen on CVD and cancer.
Acknowledgements
We thank Eva Sejby deceased, Barbro Simu and Siv Tengblad for technical assistance. J. H., J. Ä., B. V. and S. B. designed the study. J. H. and J. Ä. performed the statistical analysis. A. L. performed the analyses of inflammatory cytokines. J. H. and S. B. performed the eicosanoid analyses. All authors assisted in interpretation and presentation of the results. J. H. drafted the manuscript with S. B. and J. Ä. contributing.
None of the authors had any conflict of interest related to the present study.
