Professor R. Walker
Food Safety Group, Division of Toxicology, School of Biological Sciences, University of Surrey, GUILDFORD, GU2 5XH
COMMENTS ON THE STATEMENT BY DR. ERIK MILLSTONE
The historical background provided in the General Introduction is incorrect in a number of particulars. The Joint FAO/WHO Expert Committee on Food Additives (JECFA) was established in 1985, not the mid 1960's, and preceded the Establishment of the Codex Committee on Food Additives and Contaminants. It certainly was not set up by the Codex Alimentarius but by the Directors General of FAO and WHO following recommendations from the Joint FAO/WHO Expert Committee on Nutrition.
It is incorrect to state (page 2 last para.) that the SCF defers to the JECFA; it has a totally independent existence and role. It is perhaps not surprising that they frequently, but not always, allocate the same Acceptable Daily Intakes to food additives since they may review identical data. Furthermore, the Secretary of the SCF usually attends the JECFA as a WHO Temporary Adviser. Therefore the two committess would be aware of any divergence of views and may seek to reconcile them. However, sometimes the two Committees, may take somewhat differing views. On other occasions, the timetables for reviewing and evaluating particular additives may differ between the two Committees and there may be additional data generated between the review by one and the other. The relationship between the SCF and the COT is alno no tquite as stated. The COT deliverations take place independently from, though cognisant of, the SCF evaluations in progress and again the two Committees may arrive at somewhat different ADIs (as occurred with cyclamate). Recently, however, the SCF has actually begun to delegate first consideration of sme additives/contaminants to national committees of member states of the Union, like the COT, who will make the first evaluation for endorsement or modification in the SCF. It is at the stage of promulgating directives relating to the acceptability and conditions of use of food additives that the European Commission claims precedence and they would be more likely to be guided by the SCF rather than committees of individual states.
The statement is toxicologically naive and does not display a level of expertise which would enable a meaningful critique to be made of the evaluation by the national and international authorities. There are many factual inaccuracies as well as misinterpretations of the data and those relating to specific compounds are discussed later. In particular, Dr. Millstone does not mention nor appear to recognise that toxicity is dependent on dose and many other factors; there are eve recorded cases of human fatalities from drinking large amounts of water. Furthermore, a number of essential nutrients are toxic at doses only a few fold higher than their essential intakes for health (e.g. vitamin A; vitamin D) and others (like selenium) are carcinogenic at eleveated intakes. It was recognised by Paracelsus as long ago as the 16th century that toxicity is generally thresholded and "it is the dose that distinguishes the poison from the remedy: "sola dosis facit venenum".
By the very nature of the safety evaluation process, in order to establish with confidence the No Observed Adverse Effect Level, the test material is given at increasing doses until adverse effects are seen In many cases this leads to gross overdoses of no relevance to the circumstances of human exposure. To then cite the manifestations of toxicity as if they were of significance is akin to saying that water is toxic to the heart because loss of minerals associated with pathological intake of water ultimately lead to cardiac arrest. The intakes of water required to produce this effect are only about 15-20 times normal daily requirements whereas much larger safety factors would be applied to food additives. It is therefore naive or deliberately misleading not to aknlowledge dose-dependence of toxicity.
There may be some exceptions to the "Paracelsus Principle" in that it is not generally accepted that the effects of genotoxic carcinogens are thresholded - but such compounds would not be approved as food additives anyway. Further, the simple dose-response relationship may not hold for certain idiosyncratic intolerances and allergies, but such reactions to food additives are relatively rare and less frequent than allergies to natural components of foods like egg or milk proteins.
The statement does not address the crucial role of comparative studies (including humans, on metabolism and pharmacokinetics (i.e. the ways in which compounds are absorbed, distributed, metabolised and excreted, and the temporal aspects of these processes, including the magnitude and duration of exposure of target tissues). Such studies are pivotal in elucidating mechanisms of toxicity explaining interspecies differences in the qualitative or quantitative aspets of toxicity and determining the relevance to humans of toxicity generated in animal studies. The neglect of this crucial component of the toxicological evaluation leads to a failure fully to appreciate why some effects seen in animals may be considered of no significance for man. Furthermore, comparative pharmacokinetic data are beginning to be used to determine the magnitude of the safety factor required to derive the ADI.
Throughout the statement, Dr. Millstone seems to be of the opinion that the ADI represents a threshold of toxicity which is a misunderstanding of the concept. The ADI is:
1. a mechanism for ensuring safety-in-use of food additives (or safe levels of exposure to contaminants;
2. a time weighted average covering all ages except the neonate (defined by the WHO as up to 12 weeks of age);
3. derived using (generally) conservative assumptions.
The ADI is not:
1. an estimate of the lower bound of toxicity;
2. the No Observed Effect Level ¸ 100;
3. immutable (it may be revised up or down in the light of experience or new data)
In this context, the JECFA have explained: "Because...data are extrapolated from lifetime animal studies, the ADI relates to lifetime use and provides a margin of safety large enough for toxicologists not to be concerned about short-term use at exposure levels exceeding the ADI providing the average intake over longer periods of time does not exceed it". This Committee has further explained that a number of factors may affect the size of the safety factor applied to derive the ADI and these include:
adequacy of the data base; reversibility of observed effects; age-related effects; carcinogenicity; reasons for applying a lower safety factor, such as the availability of toxicity and dose response data in humans, evidence that the compound is metabolised to normal body constituents, or lack of overt toxicity.
It is not true to say (page 4 of the Millstone statement) that "JECFA and the SCF consider specific chemicals in abstraction from their uses". The whole role of the FAO component of the JECFA is to provide up-to-date information on the specifications, actual and proposed technological applications of food additives, and actual or predicted exposure of consumers. These are always important considerations, but absolutely imperative when considering the possible allocation of "ADI not specified".
Referring to page 5, para. 1, it is not surprising that the numbers [ADIs] change through time as the Committees would regularly reevaluate the situation in the light of new data or knowledge and ADIs have not infrequently been revised upwards e.g. when the original NOEL was the highest level tested but subsequent studies have shown it to be even higher; or when a safety factor larger than 100 has been used to allocate a Temporary ADI and further work has removed the need for this additional precaution. The reasons why the Committees do not always agree has been mentioned above.
The statement (Page 5 para.2) that the COT lists chemicals in five groups is out of date; this no longer the case since the COT, in line with other similar Committees, has started to allocate ADIs. Further, the implication that the consumer is not given the benefit of the doubt is misguided. Firstly, the various regulatory committees would require the sponsor of a new food additive to demonstrate to their satisfaction that it would be safe in use; in the absence of sufficient information to give this assurance, no ADI would be allocated. Secondly, there are representatives of consumer interest groups on the FAC who can ensure that indeed the consumers are given the benefit of the doubt.
Dr. Millstone's glossary of toxicological terms is flawed in a number of particulars and in others paraphrases generally accepted definitions. The definition of a carcinogen may be questioned in relation to causation. Modern understanding of carcinogenesis indicates that this is a multistage process involving several (at least three) steps: initiation, promotion and progression. In some carcinogenicity studies, an increase in tumours may be seen because the first, obligate stage has already occurred (e.g. due to oncogenic viruses) and the test compound may be acting only as a promoter which alone would be unable to cause cancer.
The definition of mutation is imprecise since not all mutations are heritable; only those arising in germ cells and not those in somatic cells may be passed on to subsequent generations. Further, dominant lethal mutations in germ cells would not be transmissible beyond one generation.
The definition of toxic is cyclical and does not relate toxicity to dose. By this definition all substances would be toxic.
The term tumour is used to describe proliferative growths but not all tumours are cancerous as appears to be implied; some tumours (like warts) are benign. In this context, and relevant to the interpretation of the Newberne study on nitrite (see below) it is necessary to distinguish between proliferative foci, adenomas (benign) and adenocarcinomas (malignant). This is a task for specialist histopathologists and the diagnoses lay at the root of the controversy about whether or not nitrite is a carcinogen (see below)
The attached glossary (Appendix 1) is reproduced from the WHO Environmental Health Criteria No.70, Principles for the Safety Assessment of Food Additives and Contaminants in Food, and is authoritative.
In discussing the food colours, Sunset Yellow FCF and Amaranth, the emotive and archaic term "coal-tar dye" is used. This term originated last century when the feedstock for the chemical and dyestuffs industries was largely derived from coal tar. However, this is no longer the case as the petrochemical industry is the major source of chemical raw materials. It may be that the association between coal-tar and cancer is being used implicitly to create a suspicion with regard to these colours which is quite unjustified. These compounds are no more related to the carcinogens in coal tar than mushrooms are to the pig excrement which forms the basis of the compost on which they are grown!
Sunset Yellow FCF As indicated above, Millstone's use of the term "coal-tar dye" is inaccurate and inappropriate.
The first JECFA review of sunset yellow FCF was conducted in 1964 (not l966 as indicated by Millstone) although the toxicological monographs were not published until 1966. It is not too surprising that data generated in the interim led the SCF to arrive at a somewhat lower ADI at their review 13 years later.
Millstone mentions the CoT comment about the colour being associated with hypersensitivity reactions in man but the incidence is not mentioned In the report. In fact, hypersensitivity reactions to this colour are relatively rare, largely in atopic individuals with multiple food intolerances. Atopic individuals would need to monitor the composition of their food to ensure the absence of components to which they are intolerant and regulations on labelling provide the opportunity to identify the presence and nature of additives in foods. In this specific case, McDonalds have made available such information to consumers and the colour is used only in Orange Flavour Drink and Strawberry Flavour Milkshake. Many of the reports of intolerance to sunset yellow are anecdotal and unsupported by double-blind tolerance tests so the true incidence of intolerance is probably significantly lower than is claimed.
The wording of the paragraph (page 9) beginning "In 1982 JECFA...." seems deliberately obscurantist by linking the JECFA and SCF comments. In fact, a full toxicological monograph was published following the JECFA review in which 29 references are cited (WHO 1982); these included special studies on reproduction, teratogenicity, mutagenicity and chronic toxicity/carcinogenicity At that time, four long-term studies in rats, three long-term studies in mice and additional chronic toxicity studies in hamsters and dogs were reviewed and were negative with regard to carcinogenicity. Since then, a further carcinogenicity study carried out at the maximum tolerated do se and 509; of this dose also revealed no treatment-related increase in tumour incidence.
With regard to the observations on adrenal phaeochromocytomas, these were observed only at the very high levels in the diet of 3% (females) or 5% (both sexes). At these levels, many relatively inert substances, such as hydrogenated carbohydrates (polyols) cause changes in the gastrointestinal tract leading to secondary effects on calcium absorption and the adrenal gland. In addition to the detailed explanation provided by the COT as to why they discounted the relevance of these effects in relation to human exposure to sunset yellow FCF, an earlier meeting of the JECFA discussed this issue of rat adrenal phaeochromocytoma and reached essentially the same conclusions (WHO 1983) This report points out that Gilman et al. (1953) observed a high spontaneous incidence of these lesions in untreated rats (50% in females and 82% in males); since then the United Nations International Agency for Research on Cancer has found spontaneous incidences in different rat strains of from 0 to 80%. Furthermore, the International Programme on Chemical Safety (IPCS 1987) noted that the incidence of phaeochromocytomas in rats depends on overnutrition. They stated: "Rats fed ad libitum on highly nutritious diets tend to develop a wide variety of neoplasms, particularly of the endocrine glands, in much higher incidence than animals provided with enough food to meet their nutritional needs but not enough to render them obese. The adrenal medulla is just one of the sites affected by overfeeding. Controlled feeding especially early in life, reduces the life-time expectation of developing either hyperplasia or neoplasia of the adrenal medulla in rats . In other words, this lesion is a disease of over nutrition and the forced sedentary lifestyle of caged animals and not a toxic effect of sunset yellow FCF. IPCS (1987) Environmental Health Criteria 70, Principles for the Safety Assessment of Food Additives and Contaminants in Food, Geneva: WHO WHO (1982) Toxicological Evaluation of Certain Food Additives. WHO Food Additives Series No.17. Geneva: WHO WHO (1983) Evaluation of Food Additives and Contaminants, 27th Report of the Joint FAD/WHO Expert Committee on Food Additives. Technical Report Series 696. Geneva: WHO Amaranth
The controversy which arose over the safety evaluation of amaranth dates back to Russian reports of studies on carcinogenicity and teratogenicity which were claimed to be positive. These studies did not meet the criteria for well conducted toxicity tests on a number of scores and Millstone is wrong to assert that the low cancer rate in controls was the so]e grounds for rejection Importantly, there was no specification for the material tested and it was not clear that it was of food (as opposed to industrial) grade; this information has never been forthcoming nor have samples of the material tested been made available for independent evaluation. The numbers of animals involved appeared to be small. The reporting of the study was incomplete and inadequate No detailed records were provided of the histopathological data and there was a highly unusual report that no tumours were observed in control animals; this is extremely rare in long term carcinogenicity studies where normally there would be a background incidence of spontaneous tumours. Millstone misread the actual incidence of tumours stating in his report that 13 out 15 animals had tumours. In fact, out of SO test animals a total of 13 animals had altogether 15 tumours (i.e. some animals had more than one tumour). However, this is not crucial to the interpretation; of more significance was that the tumours were of a variety of types affecting different tissues and several of a type not uncommon in old rats. Several national and international regulatory committees reviewing these data, including the JECFA found the study flawed and of inadequate quality to demonstrate that food grade amaranth was carcinogenic.
A number of earlier studies an carcinogenicity had been negative but as these did not meet the criteria of the time, the Russian study could not be totally discounted, despite its Many shortcomings For this reason the FDA embarked on studies of reproductive toxicity and carcinogenicity. While the teratogenicity study refuted the claims of the Russian tests, the carcinogenicity assay was animals totally ruined by errors in its conduct. The principal problems arose because the animals in different dose groups were mixed up and received the wrong doses; the putative controls were also involved in this mix up. Accordingly it was impossible to relate dose and effect. There were other shortcomings in that many animals which: died on test were not recovered quickly enough to prevent tissue decomposition (autolysis) and could not meaningfully be examined histologically. This reduced the effective number of animals involved, All these shortcomings point to a very badly conducted study and had this study originated in a sponsor's laboratory there is no doubt that it would have been rejected as of inadequate quality. However, as it had been carried out "in house" the FDA tried to salvage something from a very expensive fiasco, reaching an unsupportable conclusion that the data demonstrated that amaranth was carcinogenic, This has not been accepted by most other regulatory committees at national or international levels.
Most tellingly, despite the pressures of cross-border trade, Canada did not follow the US ban on amaranth and took the unusual step of issuing a press release explaining their reasons and which provides a detailed account why they (and other regulatory Committees) rejected the study and the conclusions reached therefrom by the FDA (see Annex). Further evidence supporting the conclusion that food grade amaranth is non-carcinogenic has been provided by a subsequent, well conducted long-term study conducted at BIBRA and referred to by Millstone on page 15 of his report With regard to the "renal pelvic calcification and epithelial hyperplasia" the COT were not alone in discounting the toxicological relevance of these observations; the JECFA and the COT have leached a similar conclusion. The principal reasons for this are that renal disease (glomerulonephrosis and nephrocalcinosis) is not uncommon in ageing rats, and secondly the effect is exacerbated by mechanisms related to those which caused the adrenal hyperplasia with sunset yellow FCF The presence of high concentrations of poorly absorbed material in the gut of the rat causes enlargement of the caecum, facilitated calcium absorption and derangement of mineral metabolism leading to increased calcium deposition in the kidney. This is an effect which is exaggerated by overnutrition, imbalance of minerals in the diet, particularly imbalance of calcium, phosphate and magnesium, and is seen on administration of high doses of a variety of relatively inert materials such as lactose. As with adrenal medullary hyperplasia mentioned above, restricting food intake of rats very much, reduced this effect, As the long-term BIBRA study employed very high doses of amaranth in order to test the carcinogenicity "to destruction", it is not surprising that these secondary effects of nutritional derangement in senile animals were seen and they are of no relevance to the levels of human exposure to the colour Furthermore, Millstone fails to mention a special study carried out at BIBRA to investigate this aspect of the effects of high doses of amaranth This study further led to the conclusion that renal calcification occurs only in animals with developing senile nephrosis and that a no-effect-level of 80mg amaranth/kg body weight was established in this study (Ford et al. 1983). This was the 90-day study requested by the SCF to which Millstone refers on page 14 of his report but the results of which he seems unaware, The statement that the ADI for amaranth was derived from data which did not demonstrate a no-effect-level is erroneous as far as the evaluation by the JECFA is concerned. With reference, to the long-term study, it is made quite explicit in the JECFA toxicological monograph (WHO 1984) that "Statistical analysis (one-tailed Fisher exact test) showed the low dose (50mg/kg) incidences of pelvic calcification and hyperplasia not to be significantly different from control. " This formed the basis for the ADI of 0-0.5mg/kg b.w.. allocated by the JECFA.
In the Canadian press release, reference is also made to a chick-embryo study which Millstone mentions on p.l2 of his report. In fact, most regulatory toxicologists would discount the chick embryo test as a surrogate for teratogenicity in humans for the reasons indicated in the press release. With regard to the reproductive effects which Millstone mentions (p. l 3 of his report) the evidence of resorption was not ignored by the regulatory authorities. What he seems to overlook is the relationship between dose and effect. Accepted guidelines for the conduct and interpretation of animal test for teratogenicity/fetotoxicity indicate that evidence of fetotoxicity derived at dose levels such that the dam was intoxicated or nutritionally compromised is not valid. No fetotoxicity or teratogenicity is observed at more appropriate dose levels, This appears to be another example of Dr. Millstone's failure to recognise that substances which may produce adverse effects at very high doses may he innocuous at lower doses.
Finally, the comments about the studies carried out in the Industrial Bio Test Laboratories are irrelevant since they played no part in the latest evaluations by the JECFA, SCF and COT. The evaluation was based principally on the more recent studies carried out to modern standards of Good Laboratory Practice.
Ford, G.P., Butler W.H. & Gaunt, K.F. (1983) Report of the effects of amaranth on renal calcification in the mature rat -a 28 and 90-day study BIBRA Report 140.453/2/83
WHO (1984) Toxicological evaluation of certain food additives and contaminants WHO Food Additives Series No.19. Geneva WHO
Nitrates and nitrites
There is a vast literature on the uses and biological properties of nitrates and nitrites. The review document prepared for the SCF review in 1990 (Walker 1990) cites more than 360 primary sources and Millstone has been highly selective in citing from these. In addition to the data obtained experimentally in animals, there is a great amount of information derived from human exposure including clinical use where adverse reactions would have been recorded. No mention is made of the use of nitrate as a diuretic where daily doses of gram quantities have been without adverse effect. The reduction of nitrate to nitrite appears to be very limited and no rnethaemoglobinaemia resulted even from these very high intakes. At more modest oral doses, approximately 5% of the amount ingested may be converted to nitrite.
In discussing the difference in sensitivity to nitrites with age, Millstone asserts that children are vulnerable to smaller doses than adults. In making this assertion he mixes a number of different issues. Firstly, if the dose is expressed on a body weight basis, only neonatal babies are more sensitive. This is partly due to the fact that fetal type haemoglobin, which represents a large percentage of total circulating haemoglobin at birth, is more readily converted to me/haemoglobin by nitrite and levels of methaemoglobin reductase activity have not yet fully developer!. By the age of 12 weeks and certainly by the age that they have been weaned and might eat McDonalds products, these additional risk factors have disappeared. Thus increased sensitivity would counter indicate the use of nitrate or nitrite in infant formulae but hardly in a baconburger!
Millstone also claims that a further risk factor is that infants "absorb nitrate more rapidly from their stomachs". There is no evidence to support this statement and it appears that he is confused about the possible increased conversion of nitrate to nitrite which it has been postulated might occur. It has been suggested that neonatal infants are hypochlorhydric (i.e. have low levels of hydrochloric acid in the stomach) and thus the stomach may be colonised by nitrate-reducing bacteria which convert nitrate to nitrite. In fact, gastric acid secretion in the baby commences shortly after birth and Agunod et al. (1969) found that, of 12 babies aged 12 hours to 3 months, only one had achlorhydria and other studies indicate that babies of 7 months have normal, mature stomach pH levels (a measure of acidity). Millstone meanders from talking about babies to infants to children in an indiscriminate way and then reaches the conclusion that children should never be given foods containing these additives this is totally unjustified by the facts. Children of an age to eat at McDonalds are no more sensitive than adults.
A further issue is that children eat food as a greater proportion of their body weight daily than do adults and therefore may be at greater risk of exceeding the ADI. In this regard, the proportion of nitrate in the diet contributed by cured meats is very small compared to vegetables. Removing cured meats from a child's diet, therefore, will make little impact on total intake. In fact, a consumer is likely to get more nitrate from the french fries than from the portion of bacon or ham in a McDonald's meal, and the lettuce garnish is likely to contain more than 10 times the concentration of nitrate found in bacon.
With regard to the association between nitrate or nitrite intake and increased cancer risk, this is hypothetical and unsupported by the facts. Consider:
1.Nitrate was non-carcinogenic at very high dietary dose levels of 50,000mg sodium nitrate/kg diet in mice and rats i.e. equivalent to intakes of 2500mg/kg body weight per day in rats which would correspond to 150,000mg/day for an adult human!
2.If it is argued that the risk arises from conversion of nitrate to nitrite in vivo this conversion occurs to the extent of only 5% of the dose in humans. In properly controlled studies in which artefactual formation of carcinogenic nitrosarnimes in the diet was avoided, nitrite too was non-carcinogenic in rats, mice and hamsters. The Newberne study to which Millstone refers was re-evaluated by a panel of expert histopathologists who concluded that there was no increased incidence of tumours due to ingestion of nitrite in this study; misdiagnosis of adenomas (benign) and adenocarcinomas (malignant) was at the root of the controversy which has now been resolved even to Newberne's satisfaction More recent studies have confirmed that nitrite per se in drinking water of al levels of up to 0.25% (Maekawa,, 1982) or in food or drinking water at levels of 2000mg/kg 0 was not carcinogenic in rats. In the first of these studies' survival rates were higher in treated males while in the latter study the incidence of monocytic leukaemias was actually decreased.
3. Epidemiological evidence has failed to demonstrate a link between nitrate intake and cancer incidence, even among workers in the nitrate fertiliser industry exposed occupationally to much higher levels than normal. While some studies have indicated a possible correlation between nitrate and cancer, others have equally shown a negative correlation. Furthermore the highest consumers of nitrate in the diet are vegetarians and here epidemiological evidence indicates a reduced risk of cancer in this group.
Thus, despite many attempts to show that dietary nitrate or nitrite are risk factors for cancer, this has not been demonstrated experimentally or epidemiologically.
When the JECFA reviewed the toxicology of nitrates in 1974, the ADI allocated of 0 -5mg/kg b.w./day (expressed as sodium nitrate) was based on a NOEL of 500mg/kg b.w./day in a two-year rat study to which a safety factor of 100 was applied. In fact the NOEL was the highest dose tested in this study. The more recent long term studies mentioned above enabled a higher NOEL of 2500mg/kg b.w./day to be established. The results of this study were available to the SCF when they reviewed nitrate in 1990 but the Committee retained the original AD I, thus effectively applying an unusually large safety factor of 500. It is clear that, with such a large safety factor, there is little concern about short-term excursions of intakes above the ADI. Indeed, at their latest review in 1992, the COT explicitly states that "there was no health risk to the general population from nitrate and nitrite in food" and "there was no health risk to groups such as vegetarians who may occasionally exceed the ADI for nitrate.
With regard to nitrite, although the question of possible carcinogenicity has been resolved, there does remain another issue. This relates to reports that dose levels of potassium nitrite of 100mg/litre or above in drinking water of rats for 13 weeks caused hypertrophy of the adrenal zone glomerulosa; the incidence and degree of hypertrophy were dose related (Til et al. 1988). No such effect has previously been reported in either short term or long-term studies and the toxicological significance is unclear. Further work to clarify this issue and determine a NOEL for adrenal hypertrophy is currently nearing completion. It is understood that this new work indicates a NOEL of 50mg/kg b.w. for potassium nitrite.
Agunod M., Yamaguchi, N., Lopez, R., Luhby A.L. & Glass G.B.J. (1969) Correlative study Of hydrochloric acid, pepsin and intrinsic factor secretion in newborn and infants American Journal of Digestive Disorders, 14, 400414
Maekawa A., Ogiu, T., Onodera, H., Furutu, K., Matsuoka, C., Ohno, Y. & Odashima, S. (1982) Carcinogenicity studies of sodium nitrite and sodium nitrate in F-344 ruts. Food and Chemical Toxicology 20 25-33
Til, H.P., Falke H.E.., Kuper C.F.. & Willems, MJ. (1988) Evaluation of the oral toxicity of potassium nitrite in a 13-week drinking water study in ram. Food and Chemical Toxicology 26, 851-859 Walker, R. (1990) Nitrates Nitrites and N-nitroso compounds a review of the occurrence in food and diet and the toxicological implications. Food Additives and Contaminants 7, 717-768.
Butylated hydroxyanisole (BHA)
The assertion by Millstone that "BHA can be added [to food] to compensate for poor standards of manufacture and distribution" is at variance with the approved uses of food additives in the U.K. and the European Union where, in order to gain approval of the FAC or the SCF there has to be a demonstrable "case for need" and additives must be used in accordance with Good Manufacturing Practice.
Millstone discounts the health benefits of antioxidants, but in some circumstances, both BHA and BHT have shown beneficial effects, as have some other antioxidants. Far from being claims of "a vocal minority" and "receiving scant support from the scientific community" there is currently a vast research effort world-wide into the protective effects of antioxidants against cancer, atherogenesis and cardiovascular and cerebrovascular disease and various other degenerative and inflammatory diseases. A number of mayor symposia have been held on this topic recently, notably one supported by the Agriculture and Food Research Council at Norwich in 1992, and another in Stockholm in 1993, There is also a large multinational research programme in Europe funded by the European Commission under the FLAIR programme involving several countries and research institutions. There arc also large research programmes in this field in the USA and Japan. Hardly "scant support from the scientific community The evidence for beneficial effects of antioxidants is building and is quite compelling, not purely speculative.
The anecdotal report of effects of synthetic antioxidants on infants aged six to fifteen weeks cited by Millstone is quite irrelevant since neither BHA nor BHT are permitted for use in infant foods and children as young as 15 weeks are hardly likely to be eating at McDonald' s.
The review by Millstone of the toxicological position with respect to BHA is a naive interpretation which fails to appreciate that toxicity is dose dependent and that non-genotoxic carcinogenesis may arise by mechanisms which are thresholded. The statement that "several studies over the years have shown carcinogenic effects in some species at some doses while others have failed to demonstrate any effect is true as far as it goes but is incomplete in not dealing with the reasons.
The studies by Professor Ito, which Millstone acknowledges as "one of the best conducted... "showed that at high doses (greater than or equal to about l % of the diet) caused tumours in the forestomach of rats and hamsters. However, further studies, including work by Ito have demonstrated:
(a) the effect was clearly threshoIded and dose dependent e.g. in the rat at 0.5% of the diet BHA caused only hyperplasia, at 1% benign papillomas were observed but only at 2% of the diet were malignant squamous cell carcinomas seen;
(b) the specialist Committees on Mutagenicity in the U.K. concluded in 1988 that the tumours are produced by a non-genotoxic mechanism consequent on sustained irritation at high concentrations, leading to Increased rates of cell division (tumours may similarly be induced by repeated physical insult at the same site);
(c) other substances, unrelated to BHA which at high concentrations cause imitation can also induce forestomach tumours e.g. propionic acid at high dose levels although this compound is produced by bacteria in the stomach of ruminants without adverse effects because it is dose/concentration dependent;
(d) this is a local and not a systemic effect, and no tumours were seen at other sites;
(e) the dietary concentrations required to produce tumours are grossly higher than those present in human foods and, while BHA is anti-oxidant at levels of use in foods, it may become pro-oxidant at very high concentrations which are irrelevant to its food applications (it would be nonsense to use anti-oxidants at pro-oxidant concentrations;
(f) no tumours were seen in animals without a forestomach despite careful examination of the oesophagous, the most comparable organ and tissue type (i.e. squamous epithelium);
(g) numerous regulatory Committees, including the JECFA and the COT have reviewed the toxicological data and concluded that the mechanism of tumorigenesis is consequent on sustained hyperplasia and that there is no carcinogenic risk in the absence of hyperplasia;
(h) A NOEL can be established for forestomach hyperplasia in the rat of O. l % BRA in the diet.
The repeated calls for further work by regulatory Committees were to clarify the mechanisms of causation in the rat and the relevance for man and not to disregard the human implications of Ito's work as asserted by Millstone. The attribution of other motives to the SCFs comments (page 22 of the Millstone report) is cheap, unjustified, without foundation. There was evidence at a relatively early stage that the forestomach tumours produced in the rat only at very high doses were not relevant to the circumstances of human exposure but the Committees adopted a very cautious approach and required very comprehensive data providing reassurance in view of the nature of the effect in the rat.
By 1989 the JECFA felt that adequate evidence was available to discount the relevance for humans and stated "based on the dose dependence and reversibility of the lesions produced in the rat...it was concluded that an ADI can be established - 10 1989). This was based on the NOEL for hyperplasia of 50mg/kg body weight/day to which a safety factor of 100 was applied. Millstone quotes selectively from the JECFA monograph and carefully omits the statement of the Committee "Considering the absence of any significant adverse effects in two dog studies, it was concluded that further investigations in animals without forestomachs are not required . This conclusion was reached after the supposed evidence that BHA caused hyperplasia in the oesophagus of the pig was found to be equivocal.
The COT has been even more cautious than the JECFA in seeking firm reassurance that the forestomach tumours in rats were not indicative of a human risk but finally, in 1992, this Committee too concluded that an ADI could be established based on the NOEL for forestomach hyperplasia.
The amount of data generated in evaluating BHA has been very large indeed, as has the number of scientific review meetings of research scientists (not just regulatory Committees). The conclusions emerging have been subject to academic debate and challenge at every stage and one can therefore be confident that the conclusions of the JECFA and the COT are soundly based.
It is notable that, in common with other parts of the Millstone report, he makes reference to adverse effects without relating these to dose.
WHO (1989) Toxicological evaluation of certain food additives and contaminants WHO Food Additives Series No. 24, pp3-22. World Health Organisation. Geneva
Again, Millstone reiterates his comments on the protective effects of antioxidants and the reasons for refuting this are as given above for BHA
As stated by Millstone, there was indeed evidence that high doses of BHT (1 to 2% of the diet) could cause benign tumours (adenomas) in mice but these levels of intake are several thousand times higher than human intakes. In 1988, the JECFA also was presented with the results of a two-generation study in Tats which indicated that intakes of 100 or 250mg/kg body weight increased the incidence of hepatic tumours while earlier studies in rats had not shown BHT to be carcinogenic. However there was a feature of this study that confounded the interpretation (and which Millstone claims to find puzzling) in that animals receiving BHT survived longer than controls. This may he an aspect of the reduction in rates of degenerative disease which have been claimed for a number of anti oxidants i.e. the treated animals were protected against the age-related morality in controls The high mortality in controls meant that there were too few survivors to make a sound comparison of tumour rates with treated animals of the same age. In these circumstances it would have been both biologically and statistically unsound to refer to historical controls as suggested by Millstone.
These data have to considered against the following observations:
1. It is generally agreed by expert genotoxicologists that BHT is non-genotoxic;
2. high doses of BHT can cause hepatic toxicity which increases the rate of cell turnover as the tissue repairs and, as mentioned with BHA, repeated injury of this type can lead to increases in tumour incidence;
3. the non-genotoxic mechanism of tumorigenesis described is thresholded and is not observed at dose levels which are not toxic to the liver;
4. the dose levels needed to increase tumour incidence in rats are more than a thousand times the human intakes of BHT;
5. some studies have indicated that BHT at more realistic dose levels can protect against some known carcinogens and only at the high levels at which liver hyperplasia is produced by BHT is there potentiation.
Millstone comments that BHT (admittedly in this case at high dose levels) had an anti-vitamin K effect leading to impairment of blood clotting but this is not surprising. High doses of natural antioxidants including vitamin E have a similar effect in prolonging blood clotting times The JECFA in their latest report in 1991 stated "high doses of BHT are required to cause haemorrhage in vitamin K-deficient rats; it did not consider this effect to be critical with respect to the safety evaluation of BHT as a food additive in the human population".
The SCF have reviewed these data and concluded that BHT only produces tumours at dose levels which are frankly hepatotoxic and not at doses below the toxic threshold. This expert Committee then felt the evidence convincing enough to be able to allocate an ADI to BHT in the normal way and using normal safety factors.
At their latest review, the JECFA was cognisant of an ongoing study on BHT which would provide further evidence concerning the non-genotoxic mechanism of tumorigenesis and retained the existing temporary ADI pending the results of that study.
Millstone is correct in stating that degraded (low molecular weight) carrageenan is not used in food but then totally misrepresents the situation when he claims it produces similar problems to food grade carrageenan abut is thought to produce them more rapidly". He cites the JECFA report of 1984 in support of this statement but the report says nothing of the sort. Firstly, the JECFA monograph makes it quite clear that the toxicological effect of concern, namely ulceration of the colon in some species, was caused by degraded carrageenan and not by food grade material. Further, there was adequate evidence that the extent of degradation of food grade material after ingestion was very small indeed and this was discounted by the JECFA as not of toxicological significance. The reason for this was that the extent of degradation would not have led to the formation of significant amounts of low molecular weight material being formed. Studies on food grade material, including studies on primates (rhesus monkeys and baboons) given up to 1300mg/kg body weight, did not elicit any adverse effects on the gastrointestinal tract, The comment by Millstone in the last paragraph of page 31 that "one of the main problems concerns its ability to cause ulcers in the digestive tract..." relates to low molecular weight degraded carrageenan and not to food grade material.
There is also an unsupported assertion on page 31 about concerns of possible toxic effects on embryos "while the most recent concerns relate to the possibility that carrageenan might be carcinogenic". In fact, food grade carrageenan has been found not to be teratogenic or fetotoxic in rats, mice, hamsters or rabbits and this is nor a current concern. Further, while degraded carrageenan causes squamous metaplasia and colorectal tumours in rats following administration of high doses by gavage, or in the diet, this was consequent on the chronic inflammation and ulceration of this organ. Food grade material was not carcinogen to rats and mice fed 5% in the diet for life. The International Agency for Research on Cancer (IARC) have concluded that the data did not provide evidence that food grade, undegraded carrageenan was carcinogenic to experimental animals.
In the light of these data, the JECFA allocated an ADI not specified to carrageenan (see glossary for definition) which means that levels of exposure associated with use according to good manufacturing practice are not of toxicological concern.
The COT in the U.K. have kept carrageenan under review, primarily for two reasons. Firstly, they wanted adequate reassurance that food grade carrageenan did not produce significant amounts of degraded carrageenan during food processing or after ingestion. secondly, they wanted to be sure that there were no effects of carrageenan on the immune system since this material can serve as an adjuvant on injection and potentiate the immune response to allergens.
Monosodium glutamate (MSG)
There is a vast literature on this substance and the latest JECFA review in 1988 (WHO 1988) cited 240 studies which had been considered al that meeting and which led the (committee to conclude that the evidence of safety was sufficient to allocate an ADI not specified (see glossary for explanation). Millstone's two page treatment trivialises this vast amount of research and is hopelessly superficial, confused and in some statements, quite wrong.
Millstone's statement about the effects of glutamic acid on the brain again does not relate the effect to dose, consequent blood concentrations, the effects of the blood brain barrier and the blood levels required to elicit effects on the brain (more specifically the hypothalamus). The changes in the hypothalamus of experimental animals have only been demonstrated after injection or following administration of large bolus doses by gavage. Lesions of the central nervous system have not been seen following voluntary administration of MSG except following dehydration by water deprivation in mice, the most sensitive species.
The JECFA report makes it clear that, as a consequence of metabolise, in the gut tissue, there is only a small rise in the concentration in portal blood (between gut and liver) and further metabolism occurs in the liver. Only overwhelmingly high doses of greater than 30mg/kg body weight (equivalent to 10g for an adult human) lead to a significant (not toxic) increase in circulating blood levels. Administration by injection may also lead to elevated blood levels but this is irrelevant to the circumstances of human consumption with food. The available data show that in the mouse, the threshold blood plasma levels of glutamate associated with neuronal damage are 100-130, 380 and greater than 630micromoles per decilitre in neonatal, weanling and mature animals respectively. In humans it has been demonstrated that plasma levels of this magnitude have not been recorded even after ingestion of a bolus dose of 150mg MSG/kg body weight in water (i.e. 10g MSG for an adult). Furthermore, the reasons for the increased sensitivity of the neonatal mouse relate to the fact that the blood/bruin barrier is not so well developed at birth. In humans, there is evidence that babies metabolise MSG similarly to adults and the brain development is at a more advanced stage at parturition than in the mouse.
Millstone's statement that there is "some evidence that glutamic acid concentrates in the placental fluid of women. is totally confused and downright wrong. What is meant by "placental fluid"? In fact, it appears that he has not read the JECFA toxicological monograph which states quite explicitly that "glutamate does not readily cross the placental barrier" i.e. the fetus is protected to some degree by the placental barrier and levels in fetal blood do not increase even when maternal blood levels are increased many fold. This is the very opposite of wheat Millstone appears to be implying by his statement.
The anecdotal reports of "Chinese Restaurant Syndrome" have not been confirmed in properly controlled studies. The claim that 20% to 30% were intolerant is demonstrably arrant nonsense and has certainly not been endorsed by regulatory Committees. In fact the JECFA monograph states "Controlled, double-blind crossover studies have failed to demonstrate an unequivocal relationship between "Chinese restaurant syndrome" and consumption of MSG. MSG has not been shown to provoke bronchoconstriction in asthmatics", Again, it appears that Millstone cannot have read the JECFA monograph when he asserts "JECFA did not comment on the frequency of these problems.
From the voluminous evidence available it is clear that MSG does not represent a health hazard to consumers. The statement by the JECFA concerning infant foods does not derive from the toxicological or metabolic studies since, as indicated above, babies are able to handle glutamate much as adults. However, the JECFA have as a general policy that, as a matter of prudence, food additives should not be used in infant foods for babies under 12 weeks of age; they did not wish this general position to be undermined by the statement in the report that babies metabolise glutamate like adults.
WHO(1938) Toxicological evaluation fo certain food additives WHO Food Additive Series No.22 pp97-161 (World Health Organisation: Geneva)
exhibit: Not applicable/ available