点击显示 收起
1 From the Toxicology Unit, Chemical Safety Division, Food Standards Agency, London, United Kingdom
2 Presented at the conference "Multivitamin/Mineral Supplements and Chronic Disease Prevention," held at the National Institutes of Health, Bethesda, MD, May 15–17, 2006. 3 The views in this paper are those of the authors and not necessarily those of the Food Standards Agency. 4 Address reprint requests to DJ Benford, Toxicology Unit, Chemical Safety Division, Food Standards Agency, 125 Kingsway, London, WC2B 6NH, United Kingdom. E-mail: diane.benford{at}foodstandards.gsi.gov.uk.
ABSTRACT
Assessment of the safety of nutrients presents a challenge different from that posed by the assessment of other chemicals in food such as additives or contaminants. Because nutrients are essential, a dose-response relation exists at both ends of the intake range, separated by a safe range of intake that reflects normal homeostatic processes. The safe intake may not be the same for all population groups and life stages. The size of the safe intake range for each nutrient will vary and in a few cases may be very small. Certain nutrients such as vitamin A and manganese have known and potentially serious adverse effects at high intakes, whereas others such as iron or vitamin C may have more minor adverse effects that are readily reversible and may only be associated with supplement intake. The risk of harm occurring from taking dietary supplements will depend on the safe intake range of the nutrient concerned, the susceptibility of the individual, and the likely intake of the same nutrient from other supplements or the rest of the diet. In many cases, the available database for the safety of nutrients is very limited because the studies, where available, were not designed to assess adverse effects but may have detected problems when they occurred. Further information on the safety of nutrients could be obtained through careful experimental design.
Key Words: Minerals • vitamins • dietary supplements • nutrients • risk assessment
INTRODUCTION
The safety assessment of nutrients is an issue that presents a challenge different from that posed by the assessment of other chemicals in food, such as additives or contaminants. For example, because nutrients are essential, a dose-response relation exists at both ends of the intake range and in many cases the available database is limited. The topic of nutrient risk assessment has been considered by several authoritative bodies in recent years. Most recently, the Food and Agriculture Organization (FAO) and the World Health Organization (WHO) (1) convened and reported on a technical workshop on nutrient risk assessment, but other authoritative bodies in the European Union, United Kingdom, and United States have also addressed this subject. This article attempts to give an overview of nutrient risk assessment and draws on these reports (2-4).
DOSE-RESPONSE RELATIONS FOR NUTRIENTS
It is widely accepted that risks for nutrient substances form separate dose-response relations. At the lower end of the intake range, the risk of compromised health due to deficiency increases with decreasing dose. At higher intakes, the risk of ill health due to toxicity increases with increasing dose. Interindividual variability applies at both ends of the range. The 2 dose-response curves overlie, producing an apparent U-shaped curve (Figure 1). The nature of the dose-response curve (steep or shallow) will differ for nutrients and by whether deficiency or toxicity is being considered, and the curve will not necessarily be symmetrical. The base of the U is a range of intakes not associated with adverse effects, which reflects the normal homeostatic range. In this article, this range will generally be referred to as either the homeostatic or safe intake range. It is also sometimes referred to as the range of acceptable intake (2). The width of this range is highly variable depending on the nutrient. Some examples of this are given in Table 1. For example, the adult reference intake for vitamin C is in the range of 45–90 mg/d, an amount expected to prevent scurvy in a normal population. Much larger quantities of vitamin C are associated with gastrointestinal effects such as osmotic diarrhea, which occurs at intakes of several grams. Maximum safe intakes are set at 1 g, a difference of >10-fold from the reference intake. The margin is comparable or greater for many of the water-soluble vitamins, although few data are available for many of these nutrients. In contrast, the recommended intake of vitamin A is 600–900 µg retinol equivalents (RE)/d, and evidence exists for adverse effects on bone health at intakes >1500 µg RE/d, <2-fold higher.
FIGURE 1.. Dual curves of the dose-response relations between intake and risk of adverse effects.
View this table:
TABLE 1. Width of the safe intake range for selected nutrients1
The lower end of the homeostatic range is usually defined by the prevention of deficiency, because this is the basis for the establishment of reference intakes. It does not usually address the promotion of beneficial effects (such as a reduction in the risk of chronic diseases), a concept sometimes known as optimal nutrition.
The safe range may not be applicable to all groups. It may vary with life stage if, for example, nutritional requirements are higher or lower because of growth or increased or decreased susceptibility to adverse effects, for example, altered renal function. A dose that is beneficial for some subgroups in the population may be harmful to others. Folate supplementation reduces the incidence of neural tube defects in the fetus but may mask the anemia associated with vitamin B-12 deficiency in older persons, allowing the neuropathy also associated with the deficiency to progress undiagnosed. Other risks and benefits associated with folate have been proposed, but the evidence is less convincing (5).
The WHO/FAO workshop defined an upper level of intake as the maximum habitual intake from all sources of a nutrient or related substance judged to be unlikely to lead to adverse health effects in humans (1). The other authoritative bodies have comparable definitions of the same concept, such as the safe upper level or the tolerable upper intake level (2, 4). Because some individuals will be less susceptible to toxicity than others, the upper level is an indicator of potential for increased risk rather than being a clearly defined threshold, as is also the case for health-based guidance values for other chemical exposures. Occasionally exceeding an upper level by a small amount is unlikely to be harmful, but as the amount and frequency of intakes above the upper level increases, the more likely it becomes that some persons will suffer harm. Similarly, responses to low intakes vary, and individuals may adapt to either low or high intakes.
ADVERSE EFFECTS AND NUTRIENTS
The severity of any adverse effect varies with different nutrients and will depend on the dose-response relation. The possible changes range from biochemical changes within the normal homeostatic range through minor reversible changes to reversible and then irreversible organ damage (6; Figure 2).
FIGURE 2.. Sequence of adverse effects in increasing order of severity. Adapted from Renwick et al (6).
The nature of the possible adverse effects may also depend on whether the nutrient is consumed in a concentrated form, that is, as a supplement, or within a normal food matrix. Large, single doses of many nutrients can result in gastrointestinal effects, for example, vitamin C in large doses is associated with diarrhea, as is magnesium, whereas iron is generally associated with constipation but also with nausea, vomiting, and epigastric pain (2). Effects such as these are reversible when the dose is lowered or can be lessened in other ways, for example, by taking the supplement with food.
Some nutrients, however, may have more serious and irreversible effects. For example, the oxidized metabolites of retinol are known to be teratogenic in both animals and humans. Children exposed in utero to the anti-acne drug isotretinoin (13-cis-retinoic acid) have a pattern of congenital abnormalities known as the retinoic acid syndrome (4). This includes defects of the craniofacies (small or absent external ears and auditory canals, cleft palate micrognathia, low-set ears), central nervous system (micro- or anophthalmia, cerebellar or cortical defects, microcephaly), thymus, and cardiovascular system (transposition of the heart vessels, aortic arch hypoplasia, ventricular septal defects). The abnormalities are thought to be associated with alterations in the migration of cells from the neural crest, with the most critical period of exposure possibly being weeks 2–5 of pregnancy. Comparable effects were reported in women consuming large amounts of vitamin A during pregnancy. The link between dietary vitamin A and congenital abnormalities has been explored in several epidemiology gstudies. The threshold and dose response for the effect is poorly defined, with supplemental or total intakes of 3000 or 4500 µg RE, respectively, being the lowest intakes linked to these effects (7).
The effects of vitamin A in pregnancy had been considered a clear example of an effect specific to a particular population subgroup, occurring at lower exposures than would be of concern for other population groups. However, more recent evidence suggests that dietary vitamin A is associated with adverse effects on bone health, such as decreased bone mineral density and increased fracture risk (8-10). The data suggest that the association may be linear, and although it is clearly apparent at intakes >1500 µg RE/d in those susceptible to osteoporosis, such as postmenopausal women (8, 9), other data suggest that it is more widespread (10). At present, it is unclear whether there is a critical exposure period or whether total long-term exposure is associated with the adverse effects on bone health. Overall, the data suggest that for at least some population groups the adverse effects on bone occur at exposure levels lower than those associated with the teratogenic effects of vitamin A.
Occupational exposure to manganese by inhalation of fumes in miners and smelters is associated with manganism, a neurotoxic condition resembling Parkinson disease. Chronic studies in animals also show neurotoxic effects associated with exposure to high levels of manganese (11). The dose response is unclear, but limited data suggest that lower manganese exposure may be associated with more subtle neurologic effects in humans (12); this has not been reproduced in other epidemiology studies but the database is small. Consequently, the potential effect of manganese exposure from the diet or from supplements is uncertain.
Diets rich in fruit and vegetables are reported to be associated with a reduced risk of various types of cancer. However, 2 intervention trials, the Alpha-Tocopherol, Beta-Carotene Cancer Prevention Study (ATBC) (13) and the β-Carotene and Retinol Efficacy Trial (CARET) (14), which were intended to investigate the hypothesis that β-carotene could reduce cancer risk, were halted when supplementation at 20–30 mg/d showed an association with increased incidence of lung cancer in smokers and asbestos-exposed individuals. An explanation for this observation has still not been determined. The association is confined to supplemental β-carotene, implying that overwhelming of the normal homeostatic processes is involved.
For some vitamins and minerals, such as vitamin B-12, pantothenic acid, and silicon, harmful intakes have not been reported, but this may be related to a lack of evidence rather than any fundamental difference that renders these vitamins and minerals nontoxic regardless of intake. As intakes increase, it seems likely that the possibility of overwhelming normal metabolic processes also increases, resulting in unexpected and potentially adverse effects. This is particularly likely where exposure is in a bolus form, because this can lead to higher blood concentrations than when a similar amount is ingested in food throughout the day. Even when foods are rich sources of a particular nutrient, the nutrient intake is likely to be spread over several meals and would not be comparable with the intake from a single supplement. For example, the European Union labeling Recommended Daily Amount (RDA) (15) for iron, 14 mg, may be obtained from a single tablet or from 670 g roast beef, 875 g spinach, or 209 g cornflakes (16). This is even more apparent when megadoses are consumed. For example, the safe upper level for vitamin C, 1000 mg, can be provided by a single tablet, 1.69 kg kiwi fruit, or 2.56 L orange juice. In addition, absorption is likely to be slowed because the nutrient is present in a food matrix.
Whether there is a possibility of harm from taking multivitamin-multimineral supplements at the RDA levels will depend on the total amount taken in from the rest of the diet as well as the width of the homeostatic range. Thus, persons who regularly consume liver and liver products have a high intake of vitamin A, and taking multivitamin-multimineral supplements could result in exceeding the upper levels, which could be harmful to some individuals. Similarly, concomitant use of several different supplement products, individually containing RDA levels of nutrients, could lead to a total intake above upper levels. For example, someone taking a multivitamin-multimineral supplement for general health as well as a fish liver oil supplement for joint health, both containing the European Union RDA for vitamin A as is common for many supplements, would be taking in 800 x 2 = 1600 µg RE of vitamin A in addition to any dietary intake. As noted above, this is a level of vitamin A intake at which the risk of adverse effects on bone starts to increase even before dietary intakes are considered. This is reflected in recent advice given to consumers by the UK Food Standards Agency (17) that persons who eat liver regularly should avoid taking vitamin A supplements in the form of retinol. This advice was based on a report by the UK Scientific Advisory Committee on Nutrition (18).
Different sources of a single nutrient may be chemically different, which could lead to different biological properties with respect to both function and toxicity. Cholecalciferol (vitamin D3) and ergocalciferol (vitamin D2) are the natural and synthetic forms of vitamin D (the latter being derived from ergosterol found in plants). The 2 forms had been thought to be equivalent, but recent data suggest that due to differences in their effects on calcium metabolism, ergocalciferol may be less likely to result in toxicity (19). Both forms of the vitamin may occur in supplements. The differences may affect the interpretation of the available database because in some studies the form of the nutrient used was not specified. Vitamins E and K and niacin have several forms, often with significantly different biological effects. The chemical form of minerals is also important because it may affect solubility and absorption, thus potentially increasing toxicity. For example, the trace element selenium has a narrow range of safe intake and using it in a more soluble form such as selenomethionine will increase uptake but could also increase the risk of toxicity. Hexavalent chromium compounds are carcinogenic, whereas trivalent compounds are not, although it is unclear to what extent this is due to the insolubility of trivalent chromium. The chemical form and supplement formulation of iron salts affects the potential for gastrointestinal disturbance, with ferrous sulfate being associated with a higher incidence of effects than the same dose of iron in a chelated bis-glycino iron (20). Similarly, ferrous sulfate in a wax matrix form is associated with a lower incidence of gastrointestinal effects than is ferrous sulfate in the conventional form (21).
Some nutrient sources present in supplements are poorly defined chemically (eg, components of mineral yeasts). It may be unclear whether the mineral has been chemically incorporated into amino acids and proteins within the yeast or is just part of the yeast mixture. There may also be concern about potentially harmful effects of contaminants and other nonnutrient components of some supplements.
THE DATABASE FOR NUTRIENTS
The available data on harmful levels of nutrients are generally limited. Most studies, whether in humans, experimental animals, or in vitro systems, are conducted not for characterizing adverse effects but for other purposes, such as investigating metabolism, nutrient balance, interactions, or possible benefits (1).
Some data may be obtained from individual case reports after accidental overdoses or from epidemiology studies in populations exposed to higher-than-usual amounts of a nutrient, for example, if the local geology results in high amounts of particular minerals in drinking water or soil. Inadvertent effects may become apparent in intervention studies (such as those for β-carotene) or other human volunteer studies. However, because such studies are not designed to specifically assess adverse effects, the designs are not ideal for this purpose. For example, only one supplemental dose level may have been used, providing no information on dose-response relations. There may also be no data on the nutrient content of the basal diet, making it difficult to draw conclusions about the effects of total intake. When reviewing a database for a particular nutrient, it may be necessary to consider whether a study would have detected a particular adverse effect had it occurred, that is, whether the absence of a particular effect can be inferred.
Animal studies may be useful for investigating mechanisms of effect and for determining the biological plausibility of observations from epidemiologic studies, but comprehensive packages of studies, such as those used to assess the toxicity of food additives or pesticides, are never available. Furthermore, studies conducted in animals that use high doses of a single nutrient may result in a nutrient imbalance not relevant to the assessment of effects at lower doses. For example, many nutrients such as copper–zinc and manganese–iron may affect each other's absorption.
The value of controlled human studies of nutrient metabolism could be improved in the future if such studies were to incorporate relevant markers of potential adverse effects, selected on the basis of biological plausibility or animal studies, for example, enzyme concentrations diagnostic of liver damage. Biomarkers of total nutrient status might help to provide data on dose-response relations.
Opportunities for using postmarket surveillance to establish the safety of multivitamin-multimineral supplements are limited because of the need to consider total nutrient intake and specific sources of nutrients. If an acute effect (eg, a gastrointestinal effect) occurs within a few hours of taking a supplement, the consumer might be able to make the link and report an adverse effect. Longer-term effects would be extremely difficult to link to a particular supplement. There would also be limitations related to diverse regulatory contexts in different countries. In the United Kingdom, adverse drug reaction reporting, although valuable, is known to reflect only a fraction of adverse events despite a well-publicized scheme that a variety of health practitioners as well as patients can contribute to. A reporting scheme for a food product would be likely to be even less successful.
CONCLUSION
Although the general principles of nutrient risk assessment have been established, enormous gaps exist in the information required to conduct a robust risk assessment for many of the nutrients. A systematic approach to filling these research gaps may not be feasible in terms of the costs and the ethics of conducting appropriate studies, and the nutrients of concern need to be prioritized. At the very least, it would be important for controlled trials of purported beneficial effects to include appropriate measures of harm.
ACKNOWLEDGMENTS
This manuscript was prepared by CAM and is based on a presentation written and delivered by DJB. CAM and DJB were members of the scientific secretariat to the UK Expert Group of Vitamins and Minerals (EVM), and the intellectual input of the EVM is acknowledged. CAM and DJB are employees of the United Kingdom Food Standards Agency, an independent agency of the UK government. They have no other relevant personal or financial interests.
REFERENCES