Iron intake and cardiovascular disease

Apical iron uptake

The apical iron uptake is the absorption of iron from the intestinal lumen across the apical membrane to the enterocytes. The mechanisms of apical iron uptake from haeme and non-haeme iron sources are different.

Haeme iron derived from digested hemoproteins is able to interact with the membrane of the intestinal epithelial cells. This haeme is taken up by the enterocytes through a distinct receptor-mediated transport (Grasbeck et al. 1979). Inside the enterocytes, the haeme is broken down by haemeoxygenase-1. The released iron then enters the intracellular iron pool.

Non-haeme iron complexes can exist in two oxidation states: divalent iron, Fe(II), and trivalent iron, Fe(III). The soluble Fe(II) complexes can be taken up more easily than the almost insoluble Fe(III) derivatives by the cell membrane. However, in physiological condition, Fe(II) is rapidly oxidised to Fe(III). The presence of either endogenous chelators, such as mucin, lactoferrin or bile acids, or exogenous chelators, such as fatty acids, amino acids (e.g. asparagine, glycine, histidine) or organic acids (e.g. lactic, citric, succinic, ascorbic, malic, pyruvic) may increase the solubility and hence the uptake of iron.

Secretion of gastric acid lowers the pH of the gut. This increases the solubility and enhances the uptake of iron as well. When gastric acid production is impaired, for instance by the consumption of antacids and in the conditions of achlorhydria, iron absorption is reduced.

The soluble Fe(II) is taken up by the enterocytes through a specialised divalent metal transporter called Nramp-2, DCT-1 (divalent cation transporter) or DMT-1 (divalent metal transporter) (Gunshin et al. 1997). DMT-1 is highly expressed in the duodenum of normal individuals, and even strongly upregulated in anaemic conditions (Gunshin et al. 2001). This protein, however, is also capable of transporting other metal ions, such as zinc, lead, cadmium, manganese and copper. This lack of iron specificity may cause an increase in the absorption of toxic metals in iron deficiency.

The insoluble Fe(III) can only be taken up by the enterocytes after a reduction step catalysed by an iron-reductase. The iron-reductase, called Dcytb (duodenal cytochrome-b) has highest activity in the duodenum and lowest in the ileum (McKie et al. 2001), which is well-matched with the profile of iron absorption along the gut.

Mucin, a stomach glycoprotein, may also help the uptake of Fe(III) complexes in the duodenum. The mucin±Fe(III) complex, termed gastroferrin, readily traverses the mucus layer and acid microclimate at the mucosal surface. Within the enterocytes, Fe(III) is dissociated from the mucin. This released iron is readily reduced by several reductases including /l-integrin, mobilferrin and flavin monooxygenase (Conrad et al. 1999), contributing to the intracellular iron pool.

Basolateral iron transfer

The basolateral iron transfer is the transport of iron from the enterocytes across the basolateral membrane to the circulation. Ireg-1, an iron-regulated protein, also called Ferroportin-1 or MTP-1, is responsible for this transfer (McKie et al. 2000). The expression of the protein is localised in the duodenum and also in several other organs, such as the macrophages and the placenta where iron transfers between maternal and fetal circulations.

The extent of iron absorption is mainly affected by the level of body iron, the degree of erythropoiesis, the amount of iron in the diet, and the composition of the diet itself. Other conditions, such as hypoxia, pregnancy and inflammation, may also alter the absorption. Furthermore, iron absorption is inappropriately increased in primary haemochromatosis.

The iron stores regulator induces a moderate increase in iron absorption as the body iron stores fall, and vice versa. It is still an unresolved question in how the duodenal mucosa is able to sense the level and changes in demand for iron. Iron content of the enterocytes is likely to be an indicating factor for this regulator. A central role of this regulatory process is assigned to the recently discovered hepcidin, a protein that is secreted into the plasma from the liver (Nicolas et al. 2001).

Approximately 70 per cent of body iron is incorporated into haemoglobin. In average, an adult person produces 2 x 10 red blood cells daily containing 2 x 10 atoms (20 mg) of iron. To meet this daily requirement, the body develops regulatory mechanisms whereby erythropoiesis profoundly influences iron absorption. This regulator would balance the rate of erythropoiesis in the bone marrow with the duodenal iron absorption.

Iron absorption is also modulated by the amount of iron in the diet and the composition of the diet itself. When increasing amounts of iron are ingested, the relative amount of iron absorbed decreases owing to the feedback mechanism of the absorption machinery; however, the absolute amount may still increase. Several chelators present in the diet, such as citrate from citrus fruits, can promote an increase in iron absorption, by increasing the solubility of iron in the duodenum. In contrast, phytates in wheat and some other cereals, as well as tannins in teas, chelate iron but prevent its uptake. Several metal ions, such as lead, cobalt, manganese and zinc, which are taken up by the same absorption machinery, may also block the iron uptake through competitive inhibition.

Haeme iron found in meats is more readily absorbed than non-haeme iron. The absorption is independent of duodenal pH. Experimental data indicate that haeme iron absorption is less responsive to the store regulator than that of non­haeme iron. Consequently, meat is an excellent nutrient source of iron. Lack of meat in the diet can be a cause of iron deficiency.

Iron absorption, plasma iron transport, iron incorporation into cells and iron storage are meticulously regulated in the body to maintain iron homeostasis. Only a small fraction of body iron actually circulates, while most of body iron is prominently represented in haemoglobin, ferritin and haemosiderin.

Organs and cells communicate their needs for iron via the plasma as the central compartment of iron metabolism. Essentially, the body contains three types of cells: (1) those that need to obtain iron from the plasma (iron-requiring cells), (2) those that need to export iron towards the plasma (iron-donor cells), and (3) those that are able to take up and release iron for the protection of other more vulnerable cells in the body (hepatocytes). Transport and storage of iron in these cells are modified in situations of iron deficiency or overload.

The iron-donor cells, mainly the macrophages and the intestinal mucosal cells, release iron to the plasma as Fe(II). The majority of this iron is rapidly oxidised by hephaestin or ceruloplasmin, then bound to transferrin. Hephaestin is found in the basolateral membrane of the mature villus enterocytes along the gut, while ceruloplasmin is a humoral protein produced and secreted by the liver to the plasma. Transferrin-bound iron is offered to iron-requiring cells, with majority going to the erythroblasts in the bone marrow for haemoglobin synthesis.

Transferrin is a 80 kDa single-chain glycoprotein containing two structurally similar subunits, each with one iron binding site. Therefore, one transferrin molecule can bind two Fe(III) atoms. Upon binding to iron the subunit undergoes a rigid rotation to enclose the iron atom. A distinctive feature of transferrin is its dependence on a synergistic anion, normally carbonate or bicarbonate for Fe(III) binding. When this anion is protonated, iron will be expelled from this harbouring protein.

Normally, all the non-haeme iron in the circulation is bound to transferrin. The liver synthesises transferrin and secretes it to the plasma. Transferrins are also produced locally in the testes and the central nervous system. Only around 20±45 per cent of transferrin binding sites are occupied in the circulation, so that most of available transferrins are free from iron. Nevertheless, non-transferrin­bound iron (NTBI) can be detected in some iron-overload conditions (de Valk et al. 2000) and may be attached to a variety of ligands. Most NTBI is taken up by the hepatocytes via the portal venous system. Furthermore, NTBI may also enter many other cell types promoting tissue damage.

Both monoferric and diferric transferrins are internalised by receptor-mediated endocytosis. Differic transferrin binds with higher affinity than monoferric transferrin. Two transferrin receptors have been described, i.e. TfR-1 and TfR-2. TfR-1 is expressed far more abundantly in iron-requiring cells than TfR-2, while TfR-2 is constitutively expressed in the liver (Gatter et al. 1983).

After binding to its receptor on the cell surface, transferrin is rapidly internalised through the formation of a clathrin-coated pit, which further develops into an endocytotic vesicle. This endosome undergoes acidification to pH 5.5 weakening the association between iron and transferrin. A membrane iron­reductase may help to completely dissociate iron from transferrin (McKie et al. 2001). Iron is then transported to the cytosol by DMT-1. The intact receptor­apotransferrin then recycles to the cell surface, where neutral pH promotes

detachment of apotransferrin into the circulation. Exported apotransferrin can undergo further cycles of iron delivery into cells. The average transferrin molecule with a half-life of 8 days may be used up to one hundred times for iron delivery.

In iron overload, because of excessive iron intake, genetic defects, or repeated blood transfusions, considerable amounts of NTBI may be present in plasma. This iron can be weakly complexed to citrate, albumin, amino acids or sugars (Loreal et al. 2000). Most of NTBI is found in the complex form of Fe(III) to citrate, as shown by nuclear magnetic resonance (NMR) spectroscopy of serum from patients with iron overload (Grootveld et al. 1989). Non-haematopoietic tissues, mainly the liver, and also endocrine organs, kidneys, heart and the endothelium lining the blood vessels, preferentially take up NTBI through a transferrin-receptor independent mechanism. This mechanism may explain the continuous uptake of iron by the hepatocytes, in which iron overload has suppressed TfR1 expression beyond detectability. Furthermore, NTBI may generate toxic oxygen radicals and promote tissue damage in these organs. Normal amount of iron, obtained from dietary iron or recycled body iron released by macrophages, is required for normal body functioning, especially for haemoglobin formation in the bone marrow. In the case of increased total body iron, either from increased dietary iron absorption, regular blood transfusion or intravenous iron injection, excess toxic iron may enter the circulation. Toxic iron is readily taken up by several organs, including the liver, the heart and the endothelium lining the blood vessels, and may cause further tissue damage in these organs.

Molecular metabolism and storage of iron

Inside the cell, iron first enters the intracellular iron pool in the form of Fe(II), which is soluble and biologically available. This iron is able to enter various intracellular locations, including mitochondria (especially for haeme biosynthesis, and ferritin (for storage).

Haeme biosynthesis occurs in all tissues, although the principal sites of synthesis are erythroid cells (~85 per cent) and hepatocytes (accounting for nearly all the rest of haeme synthesis). In hepatocytes, haeme is incorporated into cytochromes, in particular the P450 class which is important for detoxification. In erythroid cells, almost all of the haeme is synthesised for incorporation into haemoglobin. When the red cells mature, both haeme and haemoglobin synthesis cease. Normally after 120 days, senescent red blood cells are engulfed by macrophages. The globin is recycled or converted into amino acids, which in turn are recycled or catabolised as required. The haeme is oxidised by haemoxygenase, which results in the production of linear tetrapyrrole biliverdin, iron and carbon monoxide (CO). Most of the CO is excreted through the lungs, while the erythrocytic iron is then either stored as ferritin or released into the plasma via the iron export protein, ferroportin-1. The released iron is oxidised to Fe(III) by ceruloplasmin and is bound to circulating transferrin.

Sequestering of iron is necessary in all cells to avoid its tendency to form oxygen radicals that may damage cells. Ferritin and hemosiderin (reviewed by Harrison & Arosio 1996) are iron storage proteins that store iron within cells. Ferritin forms a hollow, spherical particle, in which 2000—4500 iron atoms can be stored as Fe(III). All ferritins are composed of 24 subunits associating to form a spherical particle. In animals, ferritin is found not only inside cells, but also circulating in the plasma. Plasma levels of ferritin are routinely been used as a measure for body iron.

Haemosiderin is another iron-storage complex. Its molecular nature is less defined than ferritin, but it is always found within cells and appears to be a complex of ferritin, denatured ferritin and other materials. Haemosiderin is most commonly found in macrophages and is especially abundant in tissues following internal haemorrhage, suggesting that its formation may be related to phagocytosis of red blood cells and haemoglobin.

Control mechanism for iron homeostasis in cells

Iron levels may regulate the expression of iron-related proteins, such as DMT-1, Ireg-1, ferritin, transferrin and TfR-1. This is shown by the presence of an iron responsive element (IRE) in the transcription product of the gene, which allows an cytoplasmic iron responsive protein (IRP) to bind in response to the level of intracellular iron (Hentze et al. 1987). In the case of ferritin production, for example, low intracellular iron conditions allow the binding of IRP to IRE, stopping protein synthesis. High intracellular iron level, on the other hand, prevents IRP—IRE binding, which results in increased ferritin synthesis. This IRP—IRE interaction provides control mechanism for intracellular iron homeostasis.

Disorders in the iron homeostasis may lead to either iron deficiency or iron overload. Iron deficiency is a condition where the iron intake does not meet the body’s demands. Its manifestations are paleness, lethargy, palpitations and shortness of breath. Iron overload, also termed haemochromatosis, on the other hand, is characterised by a progressive increase in the total amount of body iron followed by an abnormal iron deposition in multiple organs. In advanced cases, it also causes a bronze colour of the skin because of the deposition of iron-containing pigments in various tissues. The disease was once thought to be a singular disease with varying degrees of severity. Nowadays, it is known to be heterogeneous, resulting from defects in various genes.

Several types of primary hereditary haemochromatosis have been described. Type-1 hereditary haemochromatosis (HH) is a common autosomal recessive disorder affecting mostly Caucasians. One in 200 (about 3.5 million) Europeans is homozygous for this, initially symptomless, chronic disease (Powell et al. 2000). Most individuals with primary haemochromatosis absorb excessive amount of dietary iron irrespective of the level of body iron, suggesting that the iron store regulator is dysfunctional. The excess iron accumulates over time, leading to tissue damage and organ failure. Clinical consequences include hepatic failure, liver carcinoma, arthritis, diabetes, impotence and cardiac failure.

This type-1 HH is associated with mutations in the HFE gene (Feder et al. 1996). The progression of iron overloading for this type of HH is quite slow, and affected individuals often start to have clinical symptoms only after the fifth or sixth decade of life. The initial symptoms include fatigue and joint complaints. As iron loading is progressing, patients develop skin hyperpigmentation and liver disease, which deteriorates gradually from fibrosis to cirrhosis. Cardiomyopathy and arrhythmias may develop from deposition of iron in the heart. Endocrine abnormalities, such as hypogonadism and diabetes mellitus, are also common.

HFE is strongly expressed by intestinal crypt cells and liver macrophages. The function of HFE protein itself is poorly understood. It appears to be a regulatory molecule that influences the efficiency of intestinal iron absorption, and may play an important role in iron homeostasis through its interaction with the transferrin receptor, TfR1. HFE facilitates TfR-1±mediated iron uptake from plasma into crypt cells, and its action is abrogated in HFE-linked HH in which there is functional loss of HFE protein. In the gut of HH, the cells behave as though they are relatively iron-deficient, causing an increase in intestinal iron absorption (Moura et al. 1998). HH patients have no iron-loading of macrophages, since wild type HFE also functions to inhibit iron release from these cells (Drakesmith et al. 2002), which results in increased release of low molecular weight iron as Fe(II) from the cells to the circulation, and may further promote NTBI formation. The majority of type-1 HH patients carry a missense mutation (C282Y) in HFE. Other mutations and polymorphisms (H63D, S65C, I105T, G93R) have been identified, but their contributions to HH are not clearly understood. Treatment for type-1 HH is by phlebotomy, in order to keep serum ferritin levels below 50 µg/L. Initial treatment is 500 ml phlebotomy per week, followed by continuous treatment of one to four times a year. Cirrhosis usually occurs in HH patients when hepatic iron concentrations exceed 400 µmol/g dry weight liver (22.4 mg/g).

For type-2 HH, the juvenile haemochromatosis (Perkins et al. 1965) the responsible gene has not been identified. This type of HH is more severe than type-1 and it is characterised by rapid iron loading and clinical manifestations within the second decade of life. Cardiac and endocrine abnormalities dominate the clinical picture, although liver problems are also significant. Type-3 HH is associated with mutations in TfR-2 (Camaschella et al. 2000) and is phenotypically similar to type-1 HH. Type-4 (Montosi et al. 2001; Njajou et al. 2001) and type-5 HH (Kato et al. 2001) are inherited in an autosomal dominant pattern. Type-4 is caused by missense mutations altering ferroportin-1. Patients accumulate large amounts of iron in the liver macrophages and have less transferrin-bound iron. However, they eventually also develop liver, heart and pancreatic complications. Type-5 HH, which has so far affected one Japanese family, affects the ferritin molecule, which in turn causes a defect in the iron-storing process.

Secondary haemochromatosis may be caused by several other conditions leading to iron overload. These include excess of dietary iron intake, chronic haemolysis and frequent blood transfusions. Phlebotomy is mostly impossible in these cases. The treatment most commonly used is a continuous administration of an iron­chelating agent.

Chronic anaemia such as aplastic anaemia, sickle cell anaemia, and thalassaemia cause iron overload mostly because of frequent blood transfusions. Each 250 ml transfused red cells adds about 250 mg elemental iron to the body. Frequent transfusions may promote diabetes mellitus and cardiac failure when iron concentrations exceed 268 µmol/g dry weight liver (15 mg/g). The end-organ manifestations of iron overload, such as cirrhosis, cardiac failure, hepatocellular carcinoma, diabetes mellitus and hypopituitarism resemble the manifestations in hereditary haemochromatosis patients.

excess are gaining more attention. Excessive iron may promote cardiomyopathy, arthropathy, infection, liver fibrosis, diabetes mellitus and malignancy, as well as endocrine and neurodegenerative disorders. A relatively new hypothesis has been postulated by Jerome Sullivan in 1981 (Sullivan 1981) that iron may play an important role in atherosclerosis and related cardiovascular diseases. Despite significant controversy and the negative results of several studies, de Valk and Marx (1999) concluded a strong epidemiological evidence for this iron hypothesis.

Serum ferritin concentration as a measure of body iron has been shown to significantly correlate to the risk of myocardial infarction or carotid artherosclerosis (Haidari et al. 2001; Kiechl et al. 1994; Kiechl et al. 1997; Klipstein-Grobusch et al. 1999b; Salonen et al. 1992; Salonen et al. 1998; Tuomainen et al. 1997b, 1998). Ultimately, Lauffer (1991) showed significant correlation between iron stores, measured by liver biopsy, and cardiovascular mortality. Additionally, low prevalence of CHD has been observed in areas with high prevalence of iron deficiency (Sullivan 1981).

The lower incidence of coronary heart disease in premenopausal women compared with men of the same ages and with postmenopausal women was shown to be due to the lower total body iron caused by menstrual blood loss (Sullivan 1989). In men, body iron assessed by ferritin concentration, rose after adolescence, while in women, ferritin began to rise only after the age of 45 years (Burt et al. 1993). The Framingham study showed that the risk of heart disease in women increased equally by natural or surgical menopause (Gordon et al. 1978; Hjortland et al. 1976; Kannel et al. 1976). In heterozygotes of familial hyperlipoproteinaemia, the premenopausal women had a lower risk of coronary heart disease than men (Ascherio & Hunter 1994; Slack 1969; Stone et al. 1974).

The iron hypothesis may also explain the association between frequent blood donations and reduced risk of myocardial infarction (Meyers et al. 2002, Salonen et al. 1998, Sullivan 1991, Tuomainen et al. 1997b). Additionally, a community-based prospective cohort study showed that haem iron intake was positively associated with the total body iron and the risk of cardiovascular diseases (Ascherio et al. 1994; Klipstein-Grobusch et al. 1999a; Salonen et al.1992; Snowdon et al. 1984; Tzonou et al. 1998).

Recent studies have identified carriers of hereditary haemochromatosis (HH) gene to have significantly higher catalytically active iron than the normal population (de Valk et al. 2000). Some reports, moreover, demonstrated an association between heterozygous HFE gene mutation and the risk of cardiovascular events (Battiloro et al. 2000; Hetet et al. 2001; Rasmussen et al. 2001; Roest et al. 1999, Tuomainen et al. 1999). As for 3-thalassemia, the clinical evidence of vascular complications has been shown to match with higher levels of oxidative modification of LDL compared with healthy controls (Livrea et al. 1998).

Dietary prevention of chronic heart failure (CHF): the role of micronutrients, dietary fatty acids and reduced sodium intake

The incidence of chronic heart failure (CHF), the common end-result of most cardiac diseases, is increasing steadily in many countries despite (and probably because of) considerable improvements in the acute and chronic treatment of CHD, which is nowadays the main cause of CHF in most countries. In recent years, most research effort about CHF has focused on drug treatment, and little attention has been paid to nonpharmacological management. Some unidentified factors may indeed contribute to the rise in the prevalence of CHF and should be recognised and corrected if possible. For instance, CHF is now seen also as a metabolic problem, with endocrine and immunological disturbances potentially contributing to the progression of the disease. In particular, the role of the tumour necrosis factor (TNF) is discussed below. Recently it has also been recognised that increased oxidative stress may contribute to the pathogenesis of CHF. The intimate link between diet and oxidative stress is obvious, since the major antioxidant defences of our body are derived from essential nutrients. See below the section about the antioxidant nutrients.

While it is generally considered that a high sodium diet is detrimental (and may result in acute decompensation of heart failure through a volume overload mechanism), little is known about other aspects of diet in CHF in terms of both general nutrition and micronutrients such as vitamins and minerals. In these patients, it is important not only to take care of the diagnosis and treatment of the CHF syndrome itself, and for the identification and aggressive management of traditional risk factors of CHD such as high blood pressure and cholesterol (because they can aggravate the syndrome), but also for the recognition and correction of malnutrition and of deficiencies in specific micronutrients.

The vital importance of micronutrients for health and the fact that several micronutrients have antioxidant properties are now fully recognised. These may be as direct antioxidants, such as vitamins C and E, or as components of antioxidant enzymes: superoxide dismutase or glutathione peroxidase. It is now widely believed (but still not causally demonstrated) that diet-derived antioxidants may play a role in the development (and thus in the prevention) of CHF. For instance, clinical and experimental studies have suggested that CHF may be associated with increased free radical formation and reduced antioxidant defences and that vitamin C may improve endothelial function in patients with CHF.In the secondary prevention of CHD, in dietary trials in which the tested diet included high intakes of natural antioxidants, the incidence of new episodes of CHF was reduced in the experimental groups. Taken altogether, these data suggest (but do not demonstrate) that antioxidant nutrients may help prevent CHF in post-infarction patients.

Other nutrients, however, may be also involved in certain cases of CHF. While deficiency in certain micronutrients, whatever the reason, can cause CHF and should be corrected (see below), it is important to understand that patients suffering from CHF also have symptoms that can affect their food intake and result in deficiencies, for instance tiredness when strained, breathing difficulties and gastrointestinal symptoms such as nausea, loss of appetite and early feeling of satiety. Drug therapy can lead to loss of appetite and excess urinary losses in case of diuretic use. All of these are mainly consequences, not causative factors, of CHF. Thus the basic treatment of CHF should, in theory, improve these nutritional anomalies. However, since they can contribute to the development and severity of CHF, they should be recognised and corrected as early as possible.

Finally, it has been shown that up to 50 per cent of patients suffering from CHF are malnourished to some degree, and CHF is often associated with weight loss. There may be multiple aetiologies to the weight loss, in particular lack of activity resulting in loss of muscle bulk and increased resting metabolic rate. There is also a shift towards catabolism with insulin resistance and increased catabolic relative to anabolic steroids. TNF, sometimes called cachectin (see above), is higher in many patients with CHF, which may explain weight loss in these patients. Interestingly, there is a positive correlation between TNF and markers of oxidative stress in the failing heart suggesting a link between TNF and antioxidant defences in CHF (the potential importance of TNF in CHF is discussed below in the section on dietary fatty acids and CHF). Finally, cardiac cachexia is a well-recognised complication of CHF, its prevalence increases as symptoms worsen and it is an independent predictor of mortality in CHF patients. However, the pathophysiological alteration leading to cachexia remains unclear and, at present, there is no specific treatment apart from the treatment of the basic illness and correction of the associated biological abnormalities.

As mentioned above, an important practical point is that deficiencies in specific micronutrients can both aggravate and cause CHF. The prevalence of these deficiencies among patients with CHF (and post-infarction patients) is unknown. Whether we should systematically search for them also remains unclear. In particular, we do not know whether the association of several borderline deficiencies that do not individually result in CHF may result in CHF, especially in the elderly. For certain authors, however, there is sufficient evidence to support a large-scale trial of dietary micronutrient supplementation in CHF.

There is not room here to fully explore the present knowledge in this field. Nevertheless, if we restrict our comments to human data, the situation can be summarised as follows. Cases of hypocalcaemia-induced cardiomyopathy (usually in children with a congenital cause for hypocalcaemia) that can respond dramatically to calcium supplementation have been reported. Hypomagnesaemia is often associated with a poor prognosis in CHF, and correction of the magnesium levels (in anorexia nervosa for instance) leads to an improvement in cardiac function. Low serum and high urinary zinc levels are found in CHF, possibly as a result of diuretic use, but there are no data regarding the clinical effect of zinc supplementation in that context. In a recent study, plasma copper was slightly higher and zinc slightly lower in CHF subjects than in healthy controls. As expected, dietary intakes were in the normal range and no significant relationship was found between dietary intakes and blood levels in the two groups. It is not possible to say whether these copper and zinc abnormalities may contribute to the development of CHF or are simple markers for the chronic inflammation known to be associated with CHF. Further studies are needed to address the point, since the implications for prevention are substantial.

Selenium deficiency has been identified as a major factor in the aetiology of certain nonischaemic CHF syndromes, especially in low-selenium soil areas such as eastern China and Western Africa. In Western countries, cases of congestive cardiomyopathy associated with low antioxidant nutrients (vitamins and trace elements) have been reported in malnourished HIV-infected patients and in subjects on chronic parenteral nutrition. Selenium deficiency is also a risk factor for peripartum cardiomyopathy.

In China, an endemic cardiomyopathy called Keshan disease seems to be a direct consequence of selenium deficiency. Whereas the question of the mechanism by which selenium deficiency results in CHF remains open, recent data suggest that selenium may be involved in skeletal (and cardiac) muscle deconditioning (and in CHF symptoms such as fatigue and low exercise tolerance) rather than in left ventricular dysfunction. Actually, in the Keshan area, the selenium status coincides with the clinical severity rather than with the degree of left ventricular dysfunction as assessed by echocardiographic studies. When the selenium levels of residents were raised to the typical levels in the non-endemic areas, the mortality rate declined significantly but clinically latent cases were still found and the echocardiographic prevalence of the disease remained high. What we learn from Keshan disease and other studies conducted elsewhere is therefore that in patients with a known cause of CHF, even a mild deficiency in selenium may influence the clinical severity of the disease (tolerance to exercise).

These data should serve as a strong incentive for the initiation of studies testing the effects of natural antioxidants on the clinical severity of CHF. In the meantime, however, physicians would be well advised to measure selenium in patients with an exercise inability disproportionate to their cardiac dysfunction. Finally, low whole blood thiamine (vitamin B1) levels have been documented in patients with CHF on loop diuretics and hospitalised elderly patients, and thiamine supplementation induced a significant improvement in cardiac function and symptoms.

Beyond the well-known effect of high sodium intake in the clinical course of CHF (and the occurrence of acute episodes of decompensation), another important issue is the role of diet in the development of left ventricular hypertrophy (LVH), a major risk factor for CHF (and also SCD), as well as for cardiovascular and all-cause mortality and morbidity.

The cause of LVH is largely unknown. Whereas male gender, obesity, heredity and insulin resistance may explain some of the variance in LVH, hypertension (High blood pressure, HBP) is generally regarded as the primary culprit. Thus, the risks associated with LVH and HBP are intimately linked. Recent data have suggested that low dietary intake of polyunsaturated fatty acids and high intake of saturated fatty acids, as well as HBP and obesity, at age 50 predicted the prevalence of LVH 20 years later. Although the source of saturated fatty acids is usually animal fat, the source of unsaturated fatty acids in that specific Scandinavian population and at that time was less clear and there was no adjustment for other potential dietary confounders, such as magnesium, potassium, calcium and sodium. Thus this study did not provide conclusive data on the dietary lipid determinants of LVH. However, it does suggest that dietary fatty acids may be involved in the development of LVH and that this `diet—heart connection’ may partly explain the harmful effect of animal saturated fatty acids on the heart.

Another `diet—heart connection’ in the context of advanced CHF relates to the recent theory that CHF also is a low-grade chronic inflammatory disease with elevated circulating levels of cytokines and cytokine receptors that are otherwise independent predictors of mortality. High-dose angiotensin-converting enzyme (ACE)-inhibition with enalapril, a treatment that reduces mechanical overload and shear stress (two stimuli for cytokine production in patients with CHF), was recently shown to decrease both cytokine bioactivity and left ventricular wall thickness. Finally, various anti-cytokine and immuno­modulating agents were shown to have beneficial effects on heart function and clinical functional class in patients with advanced CHF suggesting a causal relationship between high cytokine production and CHF.

This also suggests that there is a potential for therapies altering cytokine production in CHF. In that regard, it has been shown that dietary supplementation with n-3 fatty acids (either fish oil or vegetable oil rich in n-3 fatty acid) reduces cytokine production at least in healthy volunteers. An inverse exponential relationship between leucocyte n-3 fatty acid content and cytokine production by these cells was found, most of the reduction in cytokine production being seen with eicosapentanoic acid in cell membrane lower than 1 per cent, a level obtained with rather moderate n-3 fatty acid supplementation. However, further studies are warranted to test whether (and at which dosage) dietary n-3 fatty acids may influence the clinical course of CHF through an anti-cytokine effect.

Sodium intake is the environmental factor that is currently most suspected of influencing blood pressure and the prevalence of HBP. However, the full damaging potential of high sodium intake for the heart (and also the kidney) seems to be largely independent of the blood pressure effect of sodium. Animal experiments and clinical studies have consistently shown that high sodium intake is a powerful and independent determinant of LVH and that such an effect of salt that is not related to arterial pressure is not confined to the heart. Whereas the long-term effect of a reduced sodium intake after a recent AMI is unknown, in particular on LVH, experts claim that even a 50 mmol reduction in the daily sodium intake would reduce the average systolic blood pressure by at least 5 mmHg (in patients aged over 50 years) and CHD mortality by about 16 per cent. Thus, as regards the damaging effect of high sodium intake on the heart, and despite the lack of strong data showing the beneficial effect of reducing sodium intake in that specific group of patients, we believe that cardiologists should extend their dietary counselling about sodium not only to the patients with HBP or CHF but also to all post-infarction patients.