- Kay-Tee Khaw, professora,
- Peter Woodhouse
- a Clinical Gerontology Unit, University of Cambridge School of Clinical Medicine, Addenbrooke's Hospital, Cambridge CB2 2QQ
- Correspondence to: Professor Khaw.
- Accepted 15 May 1995
Objective: To examine the hypothesis that the increase in fibrinogen concentration and respiratory infections in winter is related to seasonal variations in vitamin C status (assessed with serum ascorbate concentration).
Design: Longitudinal study of individuals seen at intervals of two months over one year.
Subjects: 96 men and women aged 65-74 years living in their own homes.
Main outcome measures: Haemostatic factors fibrinogen and factor VIIC; acute phase proteins; respiratory symptoms; respiratory function.
Results: Mean dietary intake of vitamin C varied from about 65 mg/24 h in winter to 90 mg/24 h in summer; mean serum ascorbate concentration ranged from 50 μmol/l in winter to 60 μmol/l in summer. Serum ascorbate concentration was strongly inversely related to haemostatic factors fibrinogen and factor VIIC as well as to acute phase proteins but not to self reported respiratory symptoms or neutrophil count. Serum ascorbate concentration was also related positively to forced expiratory volume in one second. An increase in dietary vitamin C of 60 mg daily (about one orange) was associated with a decrease in fibrinogen concentrations of 0.15 g/l, equivalent (according to prospective studies) to a decline of approximately 10% in risk of ischaemic heart disease.
Conclusion: High intake of vitamin C has been suggested as being protective both for respiratory infection and for cardiovascular disease. These findings support the hypothesis that vitamin C may protect against cardiovascular disease through an effect on haemostatic factors at least partly through the response to infection; this may have implications both for our understanding of the pathogenetic mechanisms in respiratory and cardiovascular disease and for the prevention of such conditions.
A winter increase in fibrinogen concentration has been associated with a winter increase in respiratory infection
Vitamin C status is related to fibrinogen and factor VIIC concentrations as well as markers of infections including acute phase proteins and to respiratory function
Maintaining adequate vitamin C intake—for example, by eating an extra orange daily—may reduce some of the seasonal variation in mortality
Differences in vitamin C status, infection, and fibrinogen concentration might explain some of the observed socioeconomic differential in respiratory and cardiovascular disease
Scurvy is a well recognised consequence of vitamin C deficiency; the role of vitamin C in prevention of other diseases, however, is more controversial. Pauling suggested that a high intake of vitamin C might prevent colds and influenza1; more recently, studies have suggested that vitamin C protects against cardiovascular disease and cancer.2 3 4
Most countries experience seasonal variations in mortality. Mortality in Britain is 20-30% higher in winter than during the rest of the year. Of the estimated 40000 excess winter deaths annually, a third are attributed to respiratory disease and over half to cardiovascular disease, predominantly heart attacks and strokes.5 The cardiovascular risk factors blood pressure, cholesterol, and fibrinogen vary seasonally.6 7 8 We recently reported that in men and women living in their own homes the observed winter rise in fibrinogen concentration was related to a winter increase in infection measured both by self reported respiratory symptoms and by biological markers including neutrophil count and (alpha)1 antichymotrypsin.8 This provided a biologically plausible mechanism for the observed association between respiratory and cardiovascular disease in cross sectional and prospective studies. Why infections should occur more often in winter than in other seasons is not understood. We tested the hypothesis that vitamin C may influence the biological response to infection, concentration of haemostatic factors, and hence, risk of cardiovascular disease.
This study is described in detail elsewhere.8 Forty seven men and 49 women aged 65 to 74 years living in their own homes were recruited from a general practice in Cambridge. Each subject was visited at home by one investigator (PW) at intervals of two months from January 1991 to February 1992; visits were made on the same weekday and at the same time of day for a given individual to minimise the influence of circadian and weekday variation. At each visit the investigator administered a standard questionnaire that included a 24 hour dietary recall and questions on drug treatment, intake of vitamin supplements, and respiratory symptoms in the past fortnight. Forced expiratory volume in one second was determined with a pocket spirometer (Micro Medical, Kent). Blood was obtained without venostasis; subjects were not fasting. Full blood counts were measured on the same day (Coulter model STKS). Acute phase reactants C reactive protein and (alpha)1 antichymotrypsin were estimated by immunoturbidimetry.9 Concentration of thrombin-clottable fibrinogen was measured with the method of Clauss10 and factor VIIC with a single stage assay with bovine adsorbed deficient plasma and rabbit brain thromboplastin that is sensitive to the more active, two chain form of factor VII.11 A separate serum sample for ascorbate concentration was deproteinised with metaphosphoric acid. Total ascorbic acid concentration was measured with fluorometry. The intraassay coefficient of variation was 2.7% and interassay coefficient of variation 3.8%.12 The 24 hour dietary recall data were coded for specific foods and portion sizes, and the nutrient content was calculated with standard tables on food composition.13
Seasonal variation was modelled on the assumption that the outcome variable would follow a sinusoidal curve with a period of one year (fig 1). The curve can be described mathematically in terms of three variables: the annual mean (a), increase in annual mean (b), and the decrease in annual mean (c), with the seasonal difference, being a function of b and c. The null hypothesis of no seasonal variation corresponds to both b and c being zero and may be tested with a Wald test.14 The expected value for the outcome variable on visit (v) is given by: a+b.sin(2(pi)(v-1)/6)+c.cos(2(pi)(v-1)/6).
Measurements of the outcome variable at each visit would vary randomly about the expected curve. Repeated measurements throughout the year, however, would be correlated with each other and be accounted for by generalised estimating equations15 when estimating a, b, and c. The relation between serum ascorbate, haemostatic variables, and markers of infection was analysed by linear regression with generalised estimating equations, with adjustment for age, sex, and cigarette smoking. Analyses were repeated after exclusion of subjects who reported taking vitamin C supplements at any time during the year (n=19) and of cigarette smokers (n=14).
Trends for men and women were similar, so pooled data are reported here. Table I shows the mean values for the 24 hour dietary intake of vitamin C, serum ascorbate concentration, haemostatic factors, markers of the acute phase response (alpha)1 antichymotrypsin and C reactive protein, neutrophil count, self reported colds, and forced expiratory volume in one second, all of which varied significantly seasonally. While the mean values of carotene and vitamin E showed significant seasonal variations, the pattern of variation between winter (regarded as January to February) and summer (July to August) was not consistent. Other dietary variables—notably, total caloric and alcohol intake—did not show significant seasonal changes. Behavioural variables, including physical activity, cigarette smoking, and drug treatment, did not vary seasonally (not shown here).8 Selected variables are plotted in figure 2. Serum ascorbate concentration was related to dietary intake of vitamin C calculated from the 24 hour dietary recall (r=0.40, P<0.0001).
Table II shows regression coefficients of biological variables on serum ascorbate concentration after the effects of age, sex, and smoking status were adjusted for. Serum ascorbate concentration was strongly inversely related to haemostatic factors fibrinogen and factor VIIC as well as to acute phase proteins but not to self reported respiratory symptoms or neutrophil count. Serum ascorbate concentration also related positively to forced expiratory volume in one second. Dietary vitamin E and carotene intake was not significantly related to either haemostatic factors or acute phase proteins, and the significant associations of serum ascorbate concentration with these factors remained after dietary vitamin E and carotene were adjusted for in multivariate analysis.
The 14 subjects who were cigarette smokers had significantly lower serum ascorbate concentrations than non-smokers (38 μmol/l and 56.1 μmol/l respectively, P=0.004). Mean plasma fibrinogen concentrations in smokers and non-smokers were 3.02 g/l and 2.80 g/l respectively (P=0.006); and mean factor VIIC concentrations were 1.30 and 1.16 (P=0.10). The 19 subjects who reported taking vitamin C, or multivitamins that probably contained vitamin C, at any visit during the study had significantly higher mean serum ascorbate concentrations than subjects who did not take vitamin C (70.6 μmol/l v 49.1 μmol/l, P<0.001); their mean plasma fibrinogen concentrations were 2.68 g/l and 2.87 g/l respectively (P=0.06), and factor VIIC concentrations were 1.07 and 1.20 respectively (P=0.10).
Table III shows regression analyses for fibrinogen and factor VIIC on serum ascorbate concentration after exclusion of cigarette smokers (n=14) and of subjects who took vitamin C supplements (n=19); results were similar to those for the whole cohort.
Our findings indicate a relation between vitamin C status (measured with serum ascorbate concentration), biological markers of infection, and haemostatic factors and support the hypothesis that vitamin C may protect against cardiovascular disease through an effect on haemostatic factors at least partly through the response to infection.
That such a relation could be shown was surprising given the potentially large measurement error in estimating intake of vitamin C and vitamin C status—determined on the basis of the 24 hour dietary recall and a single blood sample at each visit. Unlike with many other dietary variables, however, the variation among subjects in vitamin C status is greater than the variation within a subject. In any case, random measurement errors are likely to obscure any underlying relation rather than produce a spurious association.
Low social class and smoking, both of which are associated with lower ascorbate concentrations, and higher concentrations of haemostatic factors are possible confounders in cross sectional studies, but as they were constant in these subjects over the year they do not account for the longitudinal seasonal variations observed. As expected, smokers had lower ascorbate concentrations than non-smokers. The relation between ascorbate concentrations and haemostatic factors was consistent, however, when cigarette smokers were excluded from the analysis, indicating that this association is not explained by confounding by cigarette smoking.
The relation between fibrinogen, factor VIIC, and ascorbate concentrations was also consistent after subjects who took vitamin C supplements at any time during the year were excluded, indicating that haemostatic factors were related to variability in vitamin C status within the normal dietary range and that differences were not accounted for by supplementation.
The inverse association between haemostatic factors and serum ascorbate concentration was strong and consistent. However, only biological markers of infection—that is, C reactive protein and (alpha)1 antichymotrypsin concentrations, not respiratory symptoms—were significantly inversely related to serum ascorbate concentrations. Reduced ascorbate concentrations may result from, rather than cause, biological responses to infection. The strong relation between serum ascorbate concentration and dietary intake of vitamin C suggests, however, that serum concentrations largely reflect dietary intake. Our findings also accord with data from trials that indicate that vitamin C may not prevent respiratory infection but may modulate the biological response, resulting in a less severe infection.16 Our findings are also compatible with observations that dietary vitamin C has a protective effect on pulmonary function.17
PROTECTIVE EFFECT OF ANTIOXIDANTS ON CARDIOVASCULAR DISEASE
Our findings are of interest for several reasons. A protective role of antioxidants such as vitamins C and E, ß carotene, and selenium in cardiovascular disease has been suggested. Several prospective studies have documented an inverse relation between intake of vitamin C and cardiovascular disease3 18; however, the American male doctors' and the nurses' health studies, which found a strong protective effect of vitamin E supplementation for coronary heart disease, reported no independent relation between vitamin C and coronary heart disease after intake of vitamin E was adjusted for.19 20 These studies used self reported dietary intake, not biological determinations of vitamin C status, and measurement error—and hence lack of power—could account for the inconsistency between the studies. Two recent prospective studies from Switzerland and Finland that used blood ascorbate concentrations measured at baseline found that low vitamin C status predicted myocardial infarction.21 22 The Finnish study reported that low ascorbate concentration was associated with a 2.7-fold increase in risk of myocardial infarction independent of other risk factors.22 A secondary prevention trial of a Mediterranean diet showed a 70% reduction in myocardial infarction and mortality independent of an effect on blood pressure and lipids; though the intervention was primarily focused on oleic and (alpha) linolenic acid, intake of vitamin C, as well as plasma vitamin C concentrations, significantly increased in the active intervention group.23 The traditional model is that any protective effect on heart disease of antioxidants such as vitamins C and E is mediated through oxidation of low density lipoprotein cholesterol; our results, however, suggest an alternative mechanism for vitamin C involving inflammation and haemostasis.
INFECTION AND HAEMOSTASIS AND CARDIOVASCULAR DISEASE
Fibrinogen and factor VIIC are well recognised risk factors for myocardial infarction and stroke.24 25 26 Evidence also exists that chronic and acute infection and a raised white cell count are risk factors for cardiovascular disease.27 28 29 Infection may contribute towards inflammatory processes occurring in atherosclerosis.30 Fibrinogen, C reactive protein, and (alpha)1 antichymotrypsin are all acute phase proteins that are synthesised by hepatocytes in increased amounts during inflammatory processes. This synthesis is mediated mainly by interleukin 6, produced by monocytes and macrophages.31 Increased concentrations of fibrinogen provide a biologically plausible mechanism by which acute or chronic infection could increase cardiovascular risk.
HYPOTHESIS: ROLE OF VITAMIN C
While haemostatic factors, infections, and intake of vitamin C have separately been implicated as risk factors for cardiovascular disease, our results support the hypothesis that they may share a common biological pathway. This may give clues as to possible biological mechanisms involving antioxidants, inflammation, haemostasis, and atherosclerosis. Thus a winter reduction in dietary intake of vitamin C, leading to lower serum ascorbate concentrations, may be related to increased susceptibility to infection in winter and subsequently to an increase in levels of haemostatic factors and increased cardiovascular mortality.
This hypothesis may also provide an explanation for several disparate observations. While deaths from influenza contribute only a small proportion of the total excess winter mortality, the size of the winter excess in cardiovascular mortality in different years is closely related to influenza epidemics,5 suggesting that a high prevalence of infection is directly associated with increased cardiovascular risk in the general population. Additionally, respiratory and cardiovascular health are closely related: poor respiratory function is a strong predictor of subsequent cardiovascular mortality, independent of cigarette smoking.32 The pronounced social class gradient in coronary heart disease, independent of classic risk factors such as blood pressure, cigarette smoking, and lipid concentrations,33 may be better explained by the higher levels of chronic and acute infection and of fibrinogen,34 and lower intake of vitamin C35 in individuals of low socioeconomic status.
These observations also have clinical and public health implications. The important determinants of vitamin C status were dietary food sources, not supplements. The estimated average intake of vitamin C (mean 77 mg daily) and corresponding concentration of serum ascorbate (mean 54 μmol/l) were relatively low and the seasonal variation modest. Variations in haemostatic variables and infection markers were apparent within this range. Our findings suggest that large doses of vitamin C (several grams daily, as recommended by Pauling1) may not be necessary. Even increasing mean daily intake of vitamin C from food sources within the usual seasonal dietary range from 65 mg to 90 mg or an increase in mean serum ascorbate concentration from 50 μmol/l to 60 μmol/l had a measurable impact on risk factors. On the basis of estimates from our study, an increase in dietary vitamin C of 60 mg daily (about one orange) was associated with a decrease in fibrinogen concentration of 0.15 g/l, equivalent to a decline of about 10% in risk of ischaemic heart disease.23 This estimate is compatible with data from population studies that show that increasing daily intake of fruit or vegetables (the major source of vitamin C) by two to three servings is associated with a 20-40% decline in risk of stroke and 25% decline in risk of ischaemic heart disease.36 37 38
The hypothesis that vitamin C may protect against cardiovascular disease through an effect on haemostatic factors at least partly through the response to infection needs further investigation and may have implications both for our understanding of the pathogenetic mechanisms in respiratory and cardiovascular disease and for the prevention of such conditions. The limitations of observational studies are well recognised, and insufficient evidence exists to support widespread use of vitamin C supplements. Nevertheless, these findings further support the public health advice to increase intake of fruit and vegetables and suggest that maintaining intake in winter may be of particular value.
We thank the participants and the doctors of the general practice in this study, which was supported by the Medical Research Council (SPG 8913493).