Oxidative stress biomarkers and the role of lipid oxidation in cardiovascular disease

A general problem in studing oxidative stress in biological systems and in the evaluation of the effects of AO in vivo, i.e. in patients, concerns the strategies for reliable measurements of oxidative parameters. Several markers and methods have been used for the assessment of the generation of oxidation products (markers of oxidation) of various biomolecules in vitro, in ex vivo systems and in vivo. The in vitro measurements, although quite effective in the assessment of the antioxidant potential of a given compound in a controlled system, are not greatly predictive of the possible activities in vivo. It should also be added that since different antioxidants act through different mechanisms and different oxidative substrates may yield different types of products, assays should be aimed at measuring various oxidative products using different substrates (Halliwell, 1995).

Ex vivo measurements are often also used in connection with the evaluation of oxidative processes in pathological states, but again in some cases some artefactual modification may occur during the collection of the samples (e.g. cells, plasma preparation). The in vivo assays are made directly on samples collected without any manipulation, e.g. urines, but although they reflect processes occurring in the organism, they do not imitate the site(s) of these events. Measurement of isoprostanes, non-enzymatically produced oxidative metabolites of arachidonic acid, is considered, with the above-mentioned limitations, a valid indicator (biomarker) of lipid peroxidation. Increments of this marker have been observed in conditions in which enhanced lipid peroxidation may be predicted (in people who smoke, or have diabetes or hyper-cholesterolemia) (Pratico et al., 2001).

Concerning specifically the measurements of lipid peroxidation markers, ideal assays should have the following features (Halliwell, 1999):

In vitro (susceptibility of substrates to oxidation under controlled conditions) Substrates/markers:

Substrates = lipids: fats, oils, lipids in membranes and lipoproteins Markers: TBARS, conjugated dienes, lipid peroxides, oxygen uptake, fall of PUFA and vitamin E, isoprostanesSubstrates = Proteins: ±SH groups, amino-acid residues, etc.

Markers: electrophoretic mobility. adduct formation, carbonyl content, etc.

A. Substrates = Nucleic acids: DNA bases, deooxyguanosine

Markers: mass spectrometry (MS) of high performance liquid chromatography

(HPLC) of modified bases, electrophoresis of damaged 5'-GG-3' doublets, `comet

assay’ for DNA bases.

A. Substrates = sugars: ribose and deoxyribose in DNA

Markers: oxidation products

Ex vivo (evaluations on samples, e.g. blood, or cells, obtained from animals/humans without further treatments, except those made in vivo)

Antioxidant/oxidant status, antioxidant capacity, antioxidant levels and activities of AO enzymes, levels of negatively charged LDL (a fraction with different chromatographic behaviour in HPLC systems), antibodies against modified LDL, ex vivo assays of DNA oxidation, ex vivo assays of protein oxidation

In vivo (determinations in biological samples collected non-invasively)

Lipids/lipoprotein oxidation): urinary levels of isoprostanes, hydrocarbons in expired air

DNA damage: urinary levels of modified DNA bases

Quantitation of major products of the peroxidation process.Low coefficients of variation of analyses.No interference by other biomolecules.Methods: Chemically reliable (e.g. mass spectroscopy, MS or high performance liquid chromatography, HPLC) or validated.Possibly not confounded by oxidized lipids ingested with the diet.Assess steady-state levels of peroxidation products and total rates of ongoing lipid peroxidation.Parameters measured should be stable on storage and not produced artefactually.

Measurement of valid biomarkers of oxidative processes should be promoted before conducting studies on the effects of antioxidants in human studies (Mayne, 2003).

A vast literature over the past two decades has been produced, devoted to the possible involvement of oxidative stress and of ROS-derived products in various

pathological states. To some extent the published information is speculative, owing to major conceptual and analytical difficulties in the assessment of oxidative processes in vivo and in the evaluation of their real contribution to pathologies. Uncontrolled free radical production has indeed been advocated as a factor in a number of diseases: atherosclerosis, arthritis, diabetes, pulmonary diseases, cancers, Alzheimer’s disease, lateral amyothrophic sclerosis, neuritis, hepatitis and senile cataracts, but most of the attention has been devoted to the possible involvement of lipid/lipoprotein oxidation in atherogenesis and in cardiovascular disease, CVD (Steinberg, 1997, Berliner and Heinecke, 1996).

As to the issue of oxidative stress and atherosclerotic CVD, certainly rather convincing evidence has been produced in in vitro studies, showing that LDL that have been exposed to oxidative stress (oxLDL) through various mechanisms (exposure to chemicals, to physical factors or to cellular processes) are highly atherogenic. Atherogenesis induced by oxLDL has been shown to activate a sequence of events, involving several types of circulating cells (monocytes, platelets) and cellular components (e.g. smooth muscle cells, SMC) and present within the vessel walls (macrophages).

There are, however, still several issues to be defined. First, LDL are rather etherogenous molecular complexes, with significant individual differences in macro- and micro-components, including a number of lipophilic compounds that are associated to them, and it is difficult to identify and quantify all the products generated after exposure to oxidative stress, which may contribute to atherogenesis. Second, in vitro LDL oxidation is generally carried out in conditions that maximize the oxidative process, e.g. removal or depletion of hydrophilic and amphiphilic antioxidant compounds that are normally present in plasma, exposure to strong pro-oxidant factors that are difficult to compare quantitatively with in vivo free radical generating systems. Therefore the final products, i.e. oxidized LDL, cannot be easily compared with oxLDL possibly

generated in vivo. In vitro studies have also convincingly shown that several types of antioxidants are able to prevent LDL oxidation induced by various agents, but the use of AO, mainly in the form of supplements, in clinical studies has not shown significant protection against CVD. Although some of these issues are considered in detail in other posts, it is worth underlining some the strong and the weak points in the overall relationships between oxidative stress and CVD.

There is evidence that lipoproteins (LP) with some of the general features of oxLP produced in vitro, evaluated with the use of the typical markers of oxidation (see further), are present in atherosclerotic plaques. On the other side, it is not completely clear whether oxLDL are generated within the vessel wall exposed to high oxygen fluxes, from previously accumulated particles, or whether they are deposited in the vessel walls after being produced in the circulation, i.e. whether the presence of oxLDL is a secondary or an associated process, rather than a causative event.

For monocytes, again, the accumulated reactive material could be produced in a secondary process. In addition, the recognition by antibodies has several limitations: poor characterization of the oxLDL used as antigens for the preparation of the antibody, and eventual (epitope) differences between the artificially produced oxLDL and those generated in vivo. In addition there may be some lack of specificity and poor quantitative responses in the reaction.

Some of the previously mentioned limitations may apply to the presence of autoantibodies against oxLDL in sera of atherosclerotic patients. There is also some evidence that antioxidant consumption may slow the progression of the disease. This, however, is a rather controversial aspect. In essence, the difficulties in the evaluation of the outcome of the studies concern the form and doses of administration of the AO and in the selection of the people to be treated.

In addition to the role of oxidized LDL in the atherogenetic process, a number of studies have been devoted to assess the involvement of oxidative stress in several CV conditions and functions, as discussed in the following reviews: endothelial functions (Cai and Harrison, 2000; Lum and Roebuck, 2001; Matsuoka, 2001; Terada, 2002), neutrophil activation (Kaminski et al., 2002), macrophage involvement (Jessup et al., 2002), smooth muscle cell function (Bomzon and Ljubuncic, 2001), vascular ageing (Yu and Chung, 2001), congestive heart failure (Mak and Newton, 2001), arterial hypertension (Zalba et al., 2001) and diabetes (Bayraktutan, 2002). However, as already discussed, most of the evidence is derived from in vitro models, animal studies or ex vivo situations, i.e. in somewhat artefactual conditions where some of the processes may be amplified. It is therefore rather problematic to assess and quantify the actual role and relevance of oxidative stress in CVD.

Based on all the direct and indirect evidence in support of the hypothesis that free radical-mediated processes and specific products arising from them may play a role in CVD, great interest has been devoted to the possible protective effects of AO in the diet, or as pure compounds, on biomarkers and on clinical endpoints in population studies.

A vast number of studies have been carried out since 1990 on various aspects of the issue of AO protection: they range from epidemiological investigations to controlled trials and have involved a great number of participants. In reality, early observations on the relationships between dietary antioxidant vitamins and disease date back to the 1930s (Seventh-Day Adventists) and the 1950s (Mormons) (reported by Enstrom et al., 1992), and the whole area has been recently reviewed systematically (Asplund, 2002). This review is based on the following inclusion criteria: human studies only, published after 1989, reporting only original data, obtained in case-control, cohort or randomized controlled trials; related to AO vitamins only; mainly reporting on morbidity and mortality of clinically meaningful manifestations of ischaemic heart disease or stroke. The following contexts have been considered: primary prevention of various endpoints (ischaemic heart disease, stroke or combined cardiovascular events), the effects on intermediary endpoints (e.g. blood lipids and blood pressure), studies on secondary prevention in patients with manifest CV disease.

The main conclusions are: in observational studies (case-control or cohort design) people with high intake of AO vitamins by regular diet or as food supplements generally have a lower risk of myocardial infarction and stroke than low consumers. In randomized controlled trials, however, AO vitamins as food supplements have no beneficial effects in the primary prevention of myocardial infarction and stroke, with some report also of adverse events. In addition, in contrast with the initial favourable reports on AO in the secondary prevention of CVD, recent reports apparently failed to show beneficial effects. Some of the negative findings on the effects of AO vitamins, however, may be attributed to pitfalls in the design of the experiments: inadequate characterization of subjects under investigation in terms of ongoing oxidative stress, inappropriate formulations and dosages, especially in comparison with the situation in natural sources: single compounds rather than mixtures, concentrations too high (possibly pro-oxidant) or too low (ineffective), administered as a bolus (capsules or tablets) rather than in the context of foods (better absorption, protection vs. oxidation of dietary components, balance between various ingredients with maintenance of natural structural and functional relationships).

In summary, some relationship exists between intakes/plasma levels of some risk factor for vitamin C (reduction of cholesterol and blood pressure with high intakes/levels), for vitamin E (reduced platelet adhesiveness with high intakes) and for multivitamin supplementation (reduced platelet aggregation), but correlations are generally weak and the area has not been investigated in detail. For case-control studies there is some support for low plasma concentrations of beta-carotene, and possibly of vitamin E, being linked to increased risk of myocardial infarction. The same does not apply to vitamin C. Altogether, owing to rapid changes in plasma AO vitamins during CV events, the data must be interpreted with caution. Concerning cohort studies, people with high intakes of AO vitamins (regular food or food supplements) have a modest reduction of risk for CV events. Plasma levels of carotene and vitamin C are stronger predictors of future CV events than dietary intakes.

Primary prevention in healthy subjects: 1 out of 8 studies has shown protective effects with beta-carotene vs. retinol on a limited number (1203) of subjects. 1 study with beta-carotene show enhanced risk of lung cancer in smokersSecondary prevention of CVD in patients with manifestations of the diseaseOut of 14 studiesIn 5, reduction of CV eventsIn 9, no effectIn 1 increase of CV events (beta-carotene).

The effects of dietary supplements of AO in the primary and secondary preventions of CVD in randomized controlled trials are summarized above. The general conclusions from these studies are as follows:

People affected by ischaemic heart disease and stroke, and populations with high occurrence of CVD often have low intakes/plasma levels of AO vitamins (causal or unfavourable lifestyle factors?).In case-control or cohort studies, people with high intakes of AO vitamins (food or supplements) have a low risk of myocardial infarction and stroke.In randomized controlled trials, AO vitamins as supplements have no beneficial effect on risk for MI or stroke (not recommendable for prevention).Some support from observational studies that low intakes of fresh fruits/ vegetables may confer a high risk for CVD.

Diets, however, especially those rich in fruits and vegetables, contain several factors or mechanisms other than AO or AO other than vitamins, exerting protective effects on various systems (Halliwell, 1999). The issue of the effects of bioactive compounds in foods and their role in the prevention of CV disease is therefore quite complex, since a large number of potentially health beneficial substances have been described (Kris-Etherton et al., 2002).

Flavonoids in particular have been investigated in relation to possible health benefits (Ross and Kasum, 2002), owing to their potential antioxidant and free-radical scavenging activities observed in vitro. Human feeding studies have shown that their absorption and bioavailability are higher than originally believed, but their overall function in vivo has yet to be clarified, whether antioxidant, anti-inflammatory, enzyme inhibitor, enzyme inducer, inhibitor of cell division, or some other function (Rice-Evans, 2001). Epidemiological studies exploring the role of flavonoids in human health have been inconclusive: some studies support a protective effect of their consumption on CVD and cancer, other studies demonstrate no effect and a few studies suggest potential harm (Ross and Kasum, 2002). Additional selected classes of bioactive compounds with antioxidant and other types of potentially healthful activities are the large groups of phenolics that are present in edible fluids — obtained from fruits of plants exposed to stressful conditions, such as grapes and olives ­ which, since the beginning of recorded history, have been part of the diet of populations living in certain areas, such as the Mediterranean basin, i.e. wine and olive oil. A vast literature is available on the properties of these compounds (German and Walzem, 2000; Visioli et al., 2002), although the impact of their consumption on health through the diet has not yet been fully assessed.

Compounds           Examples                        Sources

Flavonoids

Flavones             Apigenin, luteolin             Parsley, thyme, celery

Flavonols            Quercetin, myricetin         Onions, broccoli, apples, cherries,

berries, tea

Flavanones         Naringenin, hesperedin     Cirtus foods, prunes

Catechins            Epicatechin, gallocatechin Tea, apples, cocoa

Anthocyanidins    Pelargonin, malvadin        Cherries, grapes

Isoflavones         Genistein, daidzein           Soya beans, legumes

Phytoestrogens

Lignans,                    Enterolatone, coumestrolk         Flaxseed oil, clover
coumestran

Resveratrol                                               Grapes, red wine, peanuts

Lycopene                                                 Tomatoes, tomato products

Organosulphur              Allicin, diallyl sulphide                 Garlic, onion, leek
compounds

Isothiocyanates            Phenethyl benzyl,                       Cruciferous vegetables
sulphoranes

Monoterpenes        d-Limonene, perillic acid   Essential oils of citrus fruit, rice

bran oil, cherries, mint

Plant sterols          Sitostanol, stigmasterol    Tall oil, soybean oil, rice bran oil

Olive oil                Hydroxytyrosol, oleuropein Olives, virgin olive oil

Dietary fat consumption and lipid oxidation and cardiovascular disease

Human fat consumption has certainly changed drastically from the hunter-gatherer conditions, through the beginning of agriculture to modern times.

Changes concerned both the amounts and the quality (Simopoulos, 1999), from the low amounts of fats, especially of vegetable origin, with relative abundance of long-chain polyunsaturated fatty acids (LC-PUFA), components of structural lipids in lean meat of wild animals and fish, in prehistoric conditions, to the progressive increment in the consumption of fats from farmed animals and cultivated vegetables.

The introduction and development of agriculture have changed fat intake markedly, although for a long time changes concerned mainly the continuity of fat supply after agriculture development as opposed to the sporadic intake in hunter-gatherers. Following the progressive depletion of food obtained from small mammals, fish, fowls and gathered plants, associated with the increase in human population numbers, cereal grains became the dominant caloric and protein source of most early cultures.

Drastic changes in fat intakes have occurred however, especially in recent times, i.e. in the period after the Second World War, for a number of reasons: fats represented in the past the most expensive part of the diet, since fat/oil productions in developing countries were limited by climatic and economic reasons, and importation from fat-producing countries was expensive. Fat consumption was therefore strictly correlated with national per capita incomes (FAO, 1977). With the introduction of extensive cereal grains and seed oil-raising crops, the availability of fats for human consumption and animal feeding increased dramatically. Fats became recently rather inexpensive, even used as fuel, and available on a global scale to most populations, which in several situations appear to be exposed to hypercaloric and yet deficient (in several essential micronutrients) dietary conditions. In addition, increments in seed oil consumption brought about marked increments in the consumption of PUFA, especially of the omega-6 series (i.e. linoleic acid, 18:2 omega-6). However, differences in fat intakes among populations are still present, with generally lower intakes (7±15 energy per cent) in countries from the East and Far East, e.g. Bangladesh, Korea, China, India, Philippines, and from Africa, e.g. Tanzania, Nigeria, Ethiopia (FAO, 1994), vs. around 32±38 energy per cent (en per cent) in several countries on Western diets. A relatively recent study carried out in Tanzania dealt with populations on diets with 8±13 en per cent from fats (Pauletto et al., 1996), i.e. still much lower than the levels in Western countries. High fat intakes are generally associated with high saturated fatty acids (SFA), and also relatively high intakes of PUFA, especially of the omega-6 series.

There are also still appreciable differences in fat intakes among Western populations as indicated by a cross-evaluation in 14 European countries (Hulshof et al., 1999). Variations concerned both the absolute intakes with values ranging from around 31 en per cent, in Finland, Italy, Norway and Portugal, up to over 40 en per cent in Germany, Iceland, Spain and Belgium. As to the qualitative differences, SFA range between around 10 en per cent, in most Mediterranean Countries to about 19 per cent, monounsaturated fatty acids (MUFA) contribute to about 9±12 en per cent, with higher values in Greece and the southern parts of Italy and Spain (high olive oil intake), and PUFA ranging between 3 and 7 en per cent. Trans FA range between 0.5 en per cent in Greece up to round 2 en per cent in Iceland, and are therefore not considered to be a major problem. As to the trends in nutrient intakes over time, it is of interest that a study carried out in 10-year-old children over two decades (1973±94) in Louisiana, revealed that total energy intake remained unchanged during that time period (although it declined as Kcal/body weight), but there was a trend toward weight gain. There was a significant increase in percentage energy from proteins and carbohydrates and a decrease in percentage energy from fat (mainly SFA and MUFA). In general, although more children met the recommendations for total fat, SFA and dietary cholesterol, the vast majority continued to exceed prudent diet recommendations.

Recently, it has been proposed that the role of the diet, and particularly of dietary fats in vascular disease and in its protection, have been vastly underestimated, owing to failure to understand the importance of postprandial events (Spencer, 2002). The compounds that repeatedly enter the circulation every day during our lifespan certainly result in the exposure of vessel walls to a large variety of nutrients, and also of potentially stressful factors. These include postprandial oxidative stress, consequent to the consumption of meals containing oxidized and oxidizable lipids. This results in the postprandial elevation of plasma lipid peroxides (Ursini and Sevanian, 2002), while on the other hand, AO in meals may minimize postprandial oxidative stress.

Based on the above considerations, it appears that the consumption of heated/ fried fats may be a contributing factor in the impact of dietary fat on health. Alterations of fats and oils and of the lipid components of meals are induced by various factors (heat, light, irradiation, pH, oxygen, moisture, pro-oxidizing agents, storage at room temperature) through several processes. Refining of vegetable fats and oils has no deleterious effects upon their composition as far as deliming, neutralization or bleaching are concerned, but during deodorization or physical refining, small amounts of dimeric triglycerides and of trans fatty acids are formed depending upon temperature and duration (Billek, 1992). In general, boiling and baking have no effects, and short-term shallow frying shows only minor changes in quality. The situation is different for deep-fat frying, which can cause serious alterations, especially if the oil is used for too long. The chemical reactions involved are predominantly isomerizations, polymerization and oxidation processes. There are certainly differences related not only to the cooking/frying conditions, e.g. conventional cooking methods vs. microwave cooking (Regulska-Ilow and Ilow, 2002), but also to the type of fat, the oils containing more unsaturated fatty acids, e.g. several seed oils rather than olive oil, being more susceptible to oxidative changes. In addition to alterations in the chemistry of fats, changes can also occur in antioxidant levels and antioxidant activity of the oils (Warner, 1999).

The addition of antioxidants to the oils may protect the fatty acids from oxidation, and during heating/frying the loss of lipid-soluble vitamins (e.g. the tocopherols) from the oil precedes that of more polar compounds, e.g. phenolics in the case of olive oil (Gomez-Alonso et al., 2003), suggesting that they may act as protecting agents against AO vitamins. Fried foods are generally considered detrimental to our health especially in relation to lipid oxidation, but not all food components are equally affected, since there is, for example, little or no effect on the protein or mineral content of fried food. In addition, it should be considered that when fat/oils are used in frying food, e.g. potatoes, the temperature reached at the surfaces between food and oils is markedly lower than the temperature of the boiling oily phase, owing to the extensive evaporation of the water in the food, and that the formation of a `crusty’ surface prevents a significant penetration of the oxidized products into the food.

Among the oxidative products generated from lipids in foods, attention has been expressly paid to cholesterol oxides. Several cholesterol oxides are commonly found in foods with high cholesterol contents, such as meat, egg yolk and egg-based products (cakes, sweet biscuits, mayonnaise) if fresh materials are not used in their manufacture, and in full-fat dairy products (Savage et al., 2002). Fresh foods generally contain very low levels of cholesterol oxides, while their levels are increased by storage, cooking and processing. Dietary cholesterol oxides appear to be well absorbed, and to influence postprandial lipoprotein particle size and composition. These changes may have effects on the clearance of chylomicrons from plasma, arterial delivery of oxysterols and possible deposition in arterial lesions (Vine et al., 1997). In general, lipid peroxides from the diet may contribute significantly to the whole process of lipid peroxidation, especially during the postprandial phase, in addition to the peroxides produced through endogenous processes.

A number of studies have been devoted to investigate the health effects of thermoxidized oils and fats (Billek, 2000). While early studies using extremely overheated fats showed toxic effects in animals, the administration of fats and oils heated in equipment for deep-fat frying under the conditions of good commercial practice did not show detrimental effects on classical parameters (e.g. growth, toxicity tests) even when fed in high amounts for long time periods. However, human studies specifically related to the postprandial effects of unheated or heated oils showed increments of markers of lipid oxidation in serum related to the type of oil (e.g. safflower had greater effects than olive oil, both when uncooked and especially when cooked) (Sutherland et al., 2002). The effects of the administration of both cooked oils on major functional parameters, e.g. endothelium-dependent dilatation were, however, minimal (Williams et al., 2001). Other parameters not directly related to lipid peroxidation and to changes in plasma antioxidants have also been shown in animal studies after the administration of thermally oxidized fats: an increase in plasma thyroxine concentrations irrespective of the vitamin E and selenium status (Eder et al., 2002). In general, this type of study needs to be substantially extended and applied to more practically relevant conditions. The experimental design is crucial in this respect since the type of oxidized fats to be administered and their actual chemical composition, the context of the other components (macro- and micro-nutrients) of the diet, the doses and duration of the experiments and selection of subjects are major determinants in the outcomes.

Several health organizations over the last few years have provided recommendations on fat intake with the aim to improve our health status especially with respect of CV disease and cancer. The Scientific Conference on Dietary Fatty Acids and Cardiovascular Health (AHA, 2001) and the Executive Summary of the NCEP Expert Panel (NCEP, 2001) have provided the following recommendations: total fat 25±35 en per cent, SFA < 7 en per cent, MUFA up to 20 en per cent and PUFA up to 10 en per cent. On the other side, since evidence has been accumulating on the differential and somewhat contrasting biological roles of the omega-6 and omega-3 fatty acids, it has also been proposed by the board of the International Society for the Study of Fatty Acids and Lipids (ISSFAL) (NIH Workshop 7±9 April 1999) that individual PUFA should be considered separately and that, in addition to a value not exceeding 7 en per cent for total PUFA, LA, the major omega-6, should not exceed 4-5 en per cent, while the omega-3 ALA (alpha linolenic acid) should be at least 1 en per cent and EPA + DHA in a range of at least 0.3 g up to 1 g/day. The omega-6/omega-3 ratio is also considered an important parameter and a ratio of about 4 or 5/1 has been recommended, i.e. a ratio in the range of that apparently present in the diet before the explosion of modern agriculture, and lower than the ratio greater than 10/1 in our diets. Practical approaches to the definition of a diet with an adequate FA composition and can be based on the use of food composition data, such as those in the web site of the USDA (http://www.usda .gov). Although information from databases may not be totally adequate with reference to some minor FA components, e.g. some omega-3 FA, their use is valuable.

As to the recommendations concerning the consumption/intake of AO, although randomized controlled trials of AO vitamins as supplements have shown that they have no beneficial effect on risk for myocardial infarction or stroke, increments in the consumption of vegetables and fruits should be highly recommended. As an example, the list of 10 foods recommended as very healthy by Time magazine (2002), on the basis of generally accepted scientific evidence, and selected also for the content in AO in addition to other bioactive compounds, include the following vegetables and fruits: tomatoes (rich in the carotenoid lycopene and vitamin C), spinach (rich in the AO phytochemicals lutein and zeaxanthine, in addition to providing iron and folate), broccoli (rich in beta-carotene and vitamin C, in addition to some phytochemicals, e.g. indole-3­ carbinol with detoxifying activity), nuts (rich in vitamin E, as well as in the omega-3 FA alpha-linolenic acid, and in ellagic acid, with potential anticancer activities), red wine (polyphenolic AO derived from the skin of the grapes), oats (rich in tocotrienols, AO with vitamin E-like activities, and in fibres, e.g. beta­glucan), (green) tea (rich in the AO phenols, the catechins), blueberries, very rich in several types of AO (especially the antocyanins). Dietary AO, in addition to providing precious protective and health-promoting agents, may play a special role by acting at the gastrointestinal tract, possibly a major site of production of toxic oxidized products (Halliwell et al., 2000).

Factors affecting lipid oxidation in complex food systems

When n-3 LC PUFA are incorporated into food systems the oxidation mechanisms and thereby the oxidation rates may change dramatically. This is due to the fact the oxidisability depends on the physical structure and form of the lipid. Moreover, food systems contain a wide range of different ingredients that may influence lipid oxidation, as will be discussed in the following.

Metals can catalyse oxidation by two different mechanisms:

1 by electron transfer:

M(n+1)+ + RH-~Mn+ + R’ + H+                      15.1

2 by catalysing the decomposition of hydroperoxides:

Mn + ROOH-~Mn+1 + RO’ + OH-                      15.2

Mn+1 + ROOH-~Mn + ROO’ + H+                      15.3

Fe and Cu are the most active metals, but Mn, Zn, Co and Ni can also act as pro-oxidants. Ferrous iron (Fe2+) is more effective in decomposing peroxides than the ferric ion (Fe3+) and both ferrous and ferric iron are more effective than copper.

Trace metals are present in most foods. Even after refining and deodorisation most oils will contain trace levels of lipid hydroperoxides. Therefore, metal-catalysed decomposition of lipid hydroperoxides is probably the reaction responsible for the initiation of lipid oxidation in most foods. The reactions in equations 15.2 and 15.3 not only generate free radicals, which may initiate further oxidation reactions, but will also give rise to the formation of secondary volatile oxidation compounds as previously described. In fish oil-enriched mayonnaise, the iron present in the egg yolk, which is used as an emulsifier, was suggested to be the most important catalyst of oxidation. The mechanism by which iron promotes lipid oxidation in mayonnaise was suggested to be as follows: the low pH in mayonnaise (pH < 4.2) is responsible for releasing small amounts of iron ions from the oil±water interface where iron is bound to the egg yolk protein phosvitin. Subsequently, the released iron promotes the decomposition of pre-existing lipid hydroperoxides located at the oil±water interface and perhaps also in the aqueous phase. The effect of pH on iron release will be discussed further below. Recent results have indicated that metals from certain milk proteins are also important oxidation catalysts in fish oil-enriched milk.

Oxygen is required for oxidation to occur. Therefore, lipid oxidation may be reduced by reflushing the food product with nitrogen as observed in fish oil-enriched mayonnaise and by packaging in an air-tight container. The total amount of oxygen in the headspace above the product and the oxygen dissolved in the product will limit the extent of oxidation. However, usually oxidation can proceed for a relatively long time, because plenty of oxygen will be available even in a closed container, unless oxygen has been completely removed. Importantly, the oxygen is not required for the decomposition of lipid hydroperoxides. Therefore, off-flavour products may be formed even after all oxygen is consumed.

The large interfacial area in emulsions increases the potential contact area between the oil droplet and trace metals in the continuous, aqueous phase. The interfacial area is governed by the size of the droplets in emulsions. In the literature, contradicting reports are available on the effect of the droplet size on oxidation. In fish oil-enriched mayonnaise, lipid oxidation was faster in mayonnaises with small droplet sizes in the initial part of the storage period, whereas no effect of droplet size was observed in the later part of the storage period. The following mechanism was suggested to explain these findings: in the initial phase of the oxidation period a small droplet size, i.e. a large interfacial area, would increase the contact area between iron located in the aqueous phase and lipid hydroperoxides located at the interface and this would increase oxidation. In the later stage, oxidation proceeds inside the oil droplet and therefore the droplet size is less important. Further studies are required to elucidate this matter.

Lipid oxidation is significantly affected by the water activity in foods, especially in powders. Water may act as a solvent for metal ions, and metal salt hydrates may be formed. These hydrates are less lipid soluble and less active than the metal ions themselves. Lipid oxidation may decrease, owing to the formation of hydrogen bonds between water and lipid hydroperoxides, which prevent their decomposition into initiating free radicals Water may facilitate thenbreakdown of alkoxyl radicals formed by the decomposition of hydro­peroxides. Moreover, a decrease in water activity will also decrease the so-called glass transition temperature, which is the temperature at which the food matrix changes from the glassy state to the rubbery state. It has been suggested that lipids are more susceptible to oxidation in the rubbery state, because they can react more readily with oxygen in this state. In the glassy state, the lipids are encapsulated because there is less free volume that is not taken by the macromolecules and therefore diffusion of oxygen is also limited. Thus, bringing the powder into the glassy state by optimising the recipe or by decreasing the storage temperature/water activity may reduce oxidation. It is important to take these phenomena into consideration in relation to the incorporation of fish oil into powders such as infant formula.

Temperature affects oxidation rates in an exponential manner. The mechanism of oxidation changes with temperature, especially above 60ëC, and the lipid hydroperoxides from different fatty acids decompose into secondary volatile oxidation products at different temperatures. Therefore, it is difficult to mathematically predict the effect of temperature on shelf-life and sensory properties of foods. Nevertheless, Presa-Owens et a1. attempted to predict the shelf-life of fish oil-enriched infant formula using an accelerated stability test (Rancimat). They reported that shelf-life predicted by long-term studies at 25 ëC and 60 ëC based on peroxide values and sensory evaluation were in accordance with the shelf-life predicted by repeated Rancimat measurements at temperatures ranging from 60 to 130ëC.

Mei et a1. showed that the effect of NaCl on oxidation depended on the charge of the emulsifier, the concentration of NaCl and the concentration of ferrous in corn and salmon oil emulsions. In traditional mayonnaise, salt was shown to increase anisidine values and the effect of salt was shown to depend on the salt concentration and salt type. In fish oil-enriched mayonnaise, NaCl did not, however, promote free radical formation, which indicated that NaCl was not a pro-oxidant. Taken together, these data suggest that the effect of NaCl on lipid oxidation should be investigated in each individual food system.

Metal ions are generally more soluble at low pH than at high pH. This may explain why lipid oxidation generally is slowest at high pH values. Furthermore, pH influences the emulsifier charge and this may significantly affect oxidation as will be discussed later. In fish oil-enriched mayonnaise, lipid oxidation increased with decreasing pH. The following hypothesis was suggested to explain this phenomenon: the egg yolk used as an emulsifier in mayonnaise contains large amounts of iron, which is bound to phosvitin. At the natural pH of egg yolk (pH 6.0) the iron also forms iron bridges between phosvitin and other components in egg yolk, namely LDL and lipovitellin. These components are located at the oil—water interface in mayonnaise. When pH is decreased to 4.0, which is the pH in mayonnaise, the iron bridges between the egg yolk components are broken and iron becomes dissociated from LDL and lipovitellin. Thereby, iron becomes more active as a catalyst of oxidation.

Proteins are commonly used as emulsifiers in foods to facilitate the formation and enhance the stability of oil-in-water emulsions. During homogenisation they are absorbed to the oil droplet surface where they lower surface tension and prevent coalescence of droplets by forming protective membranes around the droplets. Proteins also have a stabilising effect on the emulsion by providing the emulsion droplets with a positive or negative electrical charge at pH values below or above the pI of the proteins. It has been suggested that the electrical charge of the interfacial layer around the oil droplet significantly influences oxidation in emulsions in the presence of metal ions. Compared with a non­ionic emulsifier, an anionic emulsifier was shown to increase oxidation in corn oil model oil-in-water emulsions whereas a cationic emulsifier decreased oxidation. These results were explained by the ability of the emulsifier to attract and repel metal ions to the oil—water interface, respectively. Therefore, the charge and thereby the type of emulsifier may affect the oxidative stability of the emulsion.

pH will affect the charge of the emulsifier and this may in turn affect the oxidative stability of emulsions. Recently, it was reported that oxidation increased with increasing pH in salmon oil-in-water emulsions stabilised by whey proteins. More specifically, lipid oxidation rates were significantly lower at pH values below the pI of the whey protein isolate. This was suggested to be due the fact that the proteins would be positively charged at pH values below pI and therefore they would repel metal ions near the oil—water interface. However, the surface charge of the emulsifier does not seem to be the only factor influencing lipid oxidation. Thus, Hu et al. also reported that the order of lipid oxidation rates in salmon oil-in-water emulsions stabilised by either whey protein isolate, sweet whey or two of the proteins present in whey protein, namely a-lactalbumin or ,3-lactoglobulin, did not equal the order of the positive charge of the emulsion droplets. In another study, in corn oil-in-water emulsions it was observed that casein resulted in lower oxidation rates than whey protein isolate and soy protein isolate, even though all emulsions were cationic at low pH.

Based on these findings it was proposed that other factors responsible for the differences in oxidative stability of protein stabilised oil-in-water emulsions could be differences in how the proteins influences the thickness or packing of the emulsion droplet interface. Increasing the thickness of the interfacial layer could make it more difficult for aqueous iron to interact with lipid hydroperoxides located near the interface. Another factor affecting the antioxidative effectiveness of proteins could be their amino acid composition. The sulphydryl group of cysteine has thus been reported to have antioxidant activity because of its ability to scavenge free radicals. Antioxidative effects of tyrosine, phenylalanine, tryptophan, proline, methionine, lysine and histidine have previously been reported in the literature.

Surfactants are small lipophilic and hydrophilic molecules that are used to form emulsions. Normally, surfactants will be present in excess in emulsions and surfactants not associated with the emulsion droplets will form micelles in the continuous phase. It has been suggested that surfactant micelles are able to reduce lipid oxidation by altering the physical location of lipid hydroperoxides and/or iron in emulsions.

More research is required to completely understand the role of emulsifiers in the lipid oxidation in complex food systems.

Previous studies have indicated that some carbohydrates in high concentrations are capable of scavenging free radicals and thereby act as antioxidants. Sucrose addition has been suggested to be able to decrease oxidation by decreasing the concentration of oxygen in the aqueous phase, and sucrose may also decrease the diffusion coefficient of oxygen via its increasing effect on the viscosity of the emulsion. Apart from reducing the diffusion of oxygen, a high viscosity of the emulsion may also reduce the diffusion of metals and other reactants and reaction products, and this may slow down oxidation rates.

Addition of antioxidants to foods may delay the onset of oxidation or slow down the rate at which it proceeds. Antioxidants are usually classified as either primary or secondary antioxidants. The former are also referred to as free radical scavengers as they are chain-breaking antioxidants that delay or inhibit the propagation stage by donating a hydrogen atom to the lipid radical, the peroxyl radical or the alkoxyl radical. Primary antioxidants are often phenolic compounds such as the synthetic antioxidants BHA, BHT, propyl gallate or as naturally occurring compounds, such as tocopherol, and plant polyphenols, such as carnosic acid. The secondary antioxidants act by a number of different mechanisms such as metal chelation, oxygen scavenging and replenishing hydrogen to primary antioxidants. The secondary antioxidants often exert synergistic effects together with primary antioxidants. EDTA, lactoferrin and citric acid are examples of metal chelators that have been shown to reduce lipid oxidation in fish oil emulsions and fish oil.Ascorbic acid and the glucose oxidase±catalase enzyme system are examples of oxygen scavengers.Ascorbic acid is also able to regenerate tocopherol by replenishing hydrogen.