Document 93MVzqrX1x9dpnEZ8EaYnEL55

f V V c c ASSESSMENT OF ESTIMATED RISK " RESULTING FROM AFLATOXINS IN CONSUMER PEANUT PRODUCTS AND OTHER FOOD COMMODITIES BUREAU OF FOODS FOOD AND DRUG ADMINISTRATION' January 19, 1978 BFG17208 `y A*/ * f *v '009 r C< SUMMARY Although direct proof that aflatoxins (B^ B2> G^, G) induce cancer in humans does not exist, laboratory tests have demonstrated that they are animal carcinogens. Further, epidemiology studies in Thailand and parts of Africa show a significant relationship between liver cancer incidence and estimated levels of aflatoxin intake. Aflatoxins cannot be completely eliminated from food products without eliminating the- products themselves. FDA has been enforcing a limit of 20 parts per Dillion (ppb) for aflatoxins in certain foods and is considering reducing that limit to 15 ppb for peanuts and peanut products (39 FR 42748). Unfortunately;- the direct risks, if any, to humans from ingestion of aflatoxins cannot be measured. Therefore, risk assessments must rely on mathematical treatment of animal toxicology and epidemiology studies. Such risk assessments have been made by FDA and are described herein. Aflatoxins can occur in a variety of grains and nuts and can also occur in the meat, milk, and eggs of food animals that have consumed aflatoxin-containing feeds. Corn products and peanut products only have been considered in the risk calculations since inclusion of other products, because of contamination levels and/or frequency of contamination or consumption does not significantly affect the numerical results. Emphasis has been placed on corn products in the Southeastern United States because both consumption and aflatoxin contamination levels are relatively high in that region. N iK* Q fO u o O BFG17209 2 Utilizing FDA and USDA aflatoxin survey and food consumption data, it is estimated that aflatoxin intake ranges from an average of 2.7 to a maximum of 9.0 nanograms/kilogram of body weight/day . (0.1 to 0.3 parts per billion in the diet) in the Southeast. _ "" An extensive body of data exists on aflatoxin toxicology in ' .-laboratory--animals. -Because it would be impractical to make risk calculations for each such study, five studies, involving long-term feeding of aflatoxins to three species of rats, were selected. (Rats in general, and male rats of the Fisher strain in particular, are considered the most aflatoxin-susceptible mammals.) Using the mathematical procedure of Mantel and Bryan, estimates of lifetime liver cancer incidence rates were derived for each study, and for the five studies combined. These estimates range from 30 to 1400 per 100,000 for the individual rat studies and from 240 to 1100 per 100,000 for their combination. By way of comparison, risk estimates derived from studies on rats and other species combined by Cornfield and co-workers yield a range from 17 to 126 per 100,000. The actual lifetime liver cancer incidence rate from all causes in the U.S. is approximately 161 per 100,000. Actual incidence rates in Georgia and Alabama are approximately 103 and 100 per 100,000 respectively. (Rates for other Southeastern states are not readily available.) The fact that the risk estimates for rats in general exceed the JO human incidence rate for liver cancer from all causes can be attributed,^ in part, to the conservative mathematical procedures employed. In O il BFG17210 r3 addition, available in vitro data would indicate that humans are biologically closer to relatively resistant animals such as mice and monkeys than to aflatoxin-susceptible rats. Epidemiology studies conducted in Thailand and several parts of Africa show a positive correlation between actual liver cancer incidence and estimated intake of aflatoxins. Utilizing the relation ship derived by Peers and Linsell, risk estimates were calculated to range-from' 20.'2 tro 67.0 per 100,000 for estimated average and maximum aflatoxin intakes. How applicable such epidemiological studies are to the U.S. is not known, particularly since the liver cancer incidence rates in these areas outside the U.S. may be greatly influenced by the presence of pyrrolizidine alkaloids in herbal medicines and the existence of hepatitis-3 virus. Risk estimates calculated from the epidemiology data for peanut products only yield a range of 1.1 to 2.7 per 100,000. Assuming that ;'v a proportional reduction in aflatoxin levels would result from the ; proposed regulatory enforcement limit of 15 ppb yields a projected range of 0.8 to 2.1 per 100,000. For the lowest enforcement limit currently possible, 5 ppb, the theoretical range would be 0.3 to 0.7 per 100,000. Although the ranges are broader, the animal data yield a similar comparison between a limit of 15 ppb and that of 5 ppb. Neither the estimated incidence rates calculated from the animal data nor chose from the epidemiology studies can be construed as representing the actual human risk from aflatoxins. It is reassuring, Jk % . O J'j Co o BFG17211 c* however, that the two approaches yield similar results, at least when inter-species differences are taken into account. Assuming that these approaches at least provide conservative indications of potential human risk, there appears to be little, if any, gain in public health protection associated with a tolerance for aflatoxins in peanut products at 10 ppb or 5 ppb as opposed to the 15 ppb level being considered. Again assuming that these approaches have some validity, it appears prudent to continue efforts to control the largely intra-state problem with aflatoxins in corn in the Southeast--notwith standing the apparently lower actual incidence of liver cancer in the Southeast. The several other reasons why a 15 ppb tolerance for aflatoxins in consumer peanut products is deemed appropriate were described in thel preamble to the proposal to establish the tolerance. BFG17212 rr INTRODUCTION The Food and Drug Administration (FDAj is considering a regulation that establishes a tolerance for aflatoxins in shelled peanuts and peanut products used as human food.' The regulation would set a 15 parts per billion (ppb)" tolerance for total"aflatoxins*+"3^ + G^~ + G^) in these product's.- Aflatoxins are mold-produced toxins which are unavoidable contaminants in peanuts and a number of other food commodities (1)'. Although the human health effects of chronic exposure to aflatoxins are not known with certainty there is evidence which associates dhronic dietary exposure to aflatoxins with the induction of primary liver cancer. In particular, several studies -have...indicted aflatoxins as animal , carcinogens (2). ___ ...,,_A.t present,.it.is not possible, to. det.rmine..a level of dietary exposure ,to carcinogenic substances that would.not lead to a finite risk of human cancer. Therefore, assurance of absolute safety for the consumer (i.e., no risk of cancer) is achievable only through complete elimination of carcino genic substances from the diet and from other sources. However, unlike chemicals intentionally added to food, unavoidable food contaminants cannot be prohibited from the diet, unless the food susceptible to contamination is itself prohibited. This approach to the assurance of absolute safety for food unavoidably contaminated with aflatoxin would mean the complete elimination of those food commodities that have been shown to be susceptible to such contamination These food commodities include com, peanuts, sorghum, rice, wheat, soybeans, walnuts, almonds, pecans, pistachio nuts, 3razil nuts, figs, and the edible BFG17213 y-t o/'or\v)iJ!L' & r -2- c tissue (meat, eggs, and milk) from food-producing animals that consume animal feed containing these compounds (3,4). Because not all lots of these foods are expected to contain measurable aflatoxins, an alternative approach to indiscriminately eliminating from the diet both contaminated and non-contaminated food is to prevent from entering consumer channels only those lots demonstrated to contain aflatoxin. In the case of consumer peanut products, aflatoxins at or above a level of 5 parts per billion (ppb) can be measured with the degree of certainty required of FDA to initiate regulatory action (e.g., seizure) to prevent the marketing of aflatoxin-contaminated peanuts. In short, the goal of attaining a "zero" tolerance for aflatoxin would translate to an enforcement level of 5 ppb, i.e., the lowest level at which it can be determined with certainty whether aflatoxin is or is not present in peanuts. (It should be noted that aflatoxins at levels less than 5 ppb can be detected in peanuts, but the validity of measurements at such levels is questionable.) While a level of 5 ppb may represent the lowest enforceable toler ance consistent with FDA's consumer protection objectives of minimizing dietary exposures to aflatoxins, such a tolerance would be beyond the capa bility of manufacturers to meet consistently. This is the case because of the high variability of aflatoxin levels within and among peanut lots. Such variability requires that quality control limits in manufacturing be set much lower than the regulatory tolerance (5). For example, consistent production of lots conforming to the proposed tolerance of 15 ppb would require quality ^ control limits on the order of 5-7 ppb. Lower tolerances would require pro- portionately lower quality control limits. T' A tolerance of 5 ppb would thus C.0 O BFG17214 r -3- c require a quality control limit below the level of reliable detection and would result in the inadvertent release of illegal lots into commercial channels, there to be seized by FDA. In brief, if peanut products are not to be entirely prohibited, consumer exposure to aflatoxins from these products can only be kept below detectable levels (i.e., 5 ppb) by seizure (and a significant loss of product). For a higher tolerance (e.g., 15 or 10 ppb), the emphasis would be on manufacturing quality control (and there would be less loss of product). Because of the foregoing characteristics of the control situation, information is needed to assess the human risks associated with low levels of aflatoxin exposure. Unfortunately, such information is not readily avail able. However, evaluation of that information which is available from toxi cology and epidemiology studies does help to put the risk question in per spective. In the following, risk estimates are derived from animal tests and from epidemiology studies. Such estimates cannot be construed as representing actual human risk but they do provide an indication of the relative potential toxicity of aflatoxins. INFORMATION NEEDED FOR RISK ASSESSMENT Existing daca indicate that the most likely potential effect from chronic aflatoxin exposure is primary liver cancer (1,2). In all risk estimates to follow, it is this disease that is being considered. An estimate of the risk of primary liver cancer in humans due to exposure to dietary aflatoxins requires two types of information. First, some measure of the toxicity of aflatoxin to humans must be available. For purposes of BFG17215 rC -4- this assessment, the most relevant toxicity data would be those relating, long-term, low-level aflatoxin exposure to primary liver cancer in the exposed group. Ideally, what is sought is a dose-response curve expressing the relationship between the level of aflatoxin exposure (the dose) and the incidence of liver cancer (the response) in the exposed group. The most relevant and least suspect of such information would be that obtained by direct experimentation in human beings. Clearly no such data are available nor, for ethical reasons, could they ever be developed. Therefore, reliance must be placed on data resulting from: 1) epidemiological studies in groups of humans known to have been' inadvertently exposed to af latoxins; or 2) direct experimentation in laboratory animals. Epidemiological studies, which should provide the most relevant information obtainable, nonetheless '-' have several disadvantages, perhaps the most important of which are their inherent limitations in establishing clear causal relationships and their relative insensitivity in detecting such relationships. Studies in experi mental animals are usually more sensitive than epidemiological studies and, when properly conducted, are capable of establishing direct causation and dose-response relationships of relatively high certainty. But the'/ have been criticized on the grounds of being irrelevant to humans. It must also be kept in mind that epidemiological and animal data obtained from relatively small, homogenous population groups requires, for estimation of risks in an entire population, extrapolation to very large, heterogeneous groups of humans. It is not known how to estimate the uncertainty associated with extrapolations of these types. BFG17216 Z3 f r -o- In addition to toxicity information, risk assessment depends on data concerning the level and duration of human exposure to the toxic material. For purposes of this assessment, the relevant data are the food commodities in which aflatoxins occur, the levels of aflatoxins in those foods, the frequency and amounts of human consumption of those foods, and the length of time of exposure. As with the toxicity information, the human exposure factor in the risk assessment equation is also subject to much uncertainty. DIETARY EXPOSURE TO AFLATOXINS a. Occurrence of Aflatoxins in Food: The food classes chat contribute most significantly to dietary exposure to aflatoxins in the United States are dry milled corn products (e.g.^com meal and grits) and-peanuts and peanut products (3). Exposure to aflatoxins through contaminated com occurs mainly in the southeastern states (i.e.j Worth Carolina, South Carolina, Georgia, Alabama, Florida, Mississippi, and Louisiana), where the contamination of corn is most severe. In this area corn is a dietary staple and the com consumed is primarily of local origin. On the basis of FDA and USDA survey data, it is estimated that the average level of total aflatoxins in dry milled corn pro ducts has been in the range of 5-10 ppb (3,6,7). Com produced and consumed in other regions of the nation has a substantially lower aflatoxin content. Therefore, the United States population groups that appear to be at highest risk, those of the southeastern states, have been selected for emphasis in this risk assessment. To be conservative, the upper level of 10 ppb is used in the risk calculations. Peanuts, while grown in the Southern states, are processed and consumed throughout the country. Data from surveys conducted by FDA indicate an asti-^j mated average level of 2 ppb total aflatoxins in consumer peanut products >> v` tw (3,6,7). This es timated level is equally applicable to population groups on ^O4 a national or regional basis. BFG17217 r c-6- - r As "previously mentioned,--aflatoxins -may also 'occur -in ocher food com modities Consumed in ifHe United-Sc-ates? *'^However , survey and experimental' data indicate that--'either-the levels-and/or-iri-eiderice : of aflatoxin in these foods arid/or ^Heir "consumption pafte-rris-are 'riot 'domparable C-o - those - f-ouhd in peanuts arid corn (3,6,7)." Calculations used riri'-the estimates of total risk -' would-not ^be 'significantly^af fected by ignoring these Mother food-items. The relevant data on the aflatoxin occurrence in food is presented in Table Jf. ,; TABLE I Major Food Classes With Aflatoxins (3 + + G^) Food Class . . . . ........... ............ Estimated Average Level (total aflatoxins, ppb) Peanuts and Peanut Products 2 Com Products 5-10 * * Average is for Southeastern States; average in all other areas is lower (see text). b. Consumption of Aflatoxin - Containing Food Products: A survey conducted by the U.S. Department of Agriculture (USDA) in 1965 provides the best available data for estimating human consumption of the significant food sources of aflatoxins (8). The USDA survey data were collected through personal interviews which required the respondent to recall all foods and the amounts actually consumed during the 24-hour period prior to the intervie The data collected, which covered the entire United States, are available for four major regions--Northeast, North Central, West, and South. BFG17218 . ..:>yW>`k rr -7- For purposes of this risk assessment, the data collected for the Southern region were used. This survey region included the states of Alabama, Florida, Georgia, Louisiana, Mississippi, North Carolina, South Carolina, Arkansas, Kentucky, Delaware, District of Columbia, Maryland, Oklahoma, Texas, Virginia, and West Virginia. Although it would have been more appropriate to base estimates of intake on the southeastern states alone (i.e., the first seven states listed), since dry milled corn products which may contain aflatoxins are mainly produced and consumed in the'se states, the food intake estimates used in the risk calculations assume that dietary patterns of the South and Southeast are identical. Food intake estimates were used'for the consumer peanut products and dry milled com products listed in Table II. TABLE II: Food Items From USDA Survey Used To Estimate Intake Of Aflatoxin-Susceptible Food (3). Consumer Peanut Products peanuts, shelled and in-shell peanut butter peanut butter sandwich peanut butter & jelly sandwich peanut butter & banana sandwich peanut butter & jelly Dm Milled Corn Products com bread & muffins com pone spoonbread johnnycake hush puppies cornmeal pancakes corn grits, hominy grits cornmeal, cornmeal mush grits with fat fried cornmeal mush BFG17219 The estimated average levels of aflatoxins in consumer peanut products and dry milled corn products (Table I) are derived from surveys of these food items as marketed as opposed to food items prepared for consumption (i.e., as eaten). Therefore, conversion factors were applied, where appropriate, to the amount of intake of food items listed in Table II to obtain the amount of intake expressed in terms of the food items listed in Table I. Calculation of the average and maximum daily intakes of foods susceptibl to aflatoxin contamination also requires knowledge of the frequency at which these foods are consumed. Direct knowledge of the frequency of consumption of these foods is not available. To estimate daily consumption, it was necessary to make use of the information that the total population in the South surveyed by USDA consisted of 6,268 persons, 560 of whom had eaten peanut products and 1,580 of whom had eaten corn products during the 24-hour period preceding the questioning. Since it would be expected that these individuals would not consume these foods every day, the ratio of "eaters" for one day versus "non-eaters" for one day was used, to establish the esti mated frequency at which these foods were eaten. This ratio, or frequency factor, was applied to both the average amount and to the amount consumed by the 90th percentile of the "eaters" surveyed to derive an estimate of average and maximum daily intakes. These estimates are presented in Table III. TA3LE III: Estimated Consumption Rates For The Southeastern United States Of Food Products Which May Contain Aflatoxins. One day average intake (grams) Peanut Products 45 Corn Products 53 One day maximum (90th percentile) intake (grams) 116 186 Factor for frequency of consumption Estimated average daily intake (grams) Estimated maximum daily intake (grams) .090 .252 4 14 10 47 BFG17220 5 i. w T-.O O 'v -V -9- fc c. Human Exposure to Aflatoxins: The estimates of consumption contained in Table III can be combined with the levels of aflatoxin contamina tion (Table I) to estimate average and maximum human exposure to aflatoxins in the southeastern part of the country (the exposure in all other regions being lower). The results of these calculations are presented in Table IV. Also shown are the estimated average and maximum human intakes of dietary aflatoxin expressed in ng/kg body weight/day and in parts per billion (ppb) in the diet. TABLE IV: Estimated Average and Maximum Daily Aflatoxin Intake In The Southeastern U.S. FOOD CLASS Peanuts and Peanut Products DAILY AFLATOXIN INTAKE Average ng/,k.g2-/ ppbj3--/ Maximum ng--1/ ng/kg 21 ppb. 3/ 8 0.15 0.005 20 0.37 0.013 Com Products 140 2.58 0.093 470 8.66 0.313 TOTAL 148 2.73 0.098 490 9.03 0.326 1/ ng = nanograms (1 x 10 -- 9 grams) 2/ ng/kg body weight; average body weight of 54.3 kg 3/ ppb in diet; average daily food intake of 1500 g BFG17221 >- O <'-v r,i Xo\. w -10- RISK" ASSESSMENT a." ~ Animal Toxicity' Data: The toxic' effects" of aflatoxin have been examined'in'several studies In'a~variety of species and'strains of experi mental' animals. These inClude"several-Straifis' Of"ritS` as' well as mice; marmosets', tree' shrews , trout; ' ducks ,' rhe'stis'monkeys;"and'hamsters (2). The'literature is much too extensive"to cover exhaustively in this report. Therefore, selected studies are employed in the following to illustrate the range of results obtained in animal studies. Generally, the rat, and in particular the male Fisher strain rat, is considered the mammalian species that is most sensitive to aflatoxin' carcinogenesis (2). For this assessment, five studies involving three strains of rats were examined and the statisti cal procedure developed by Mantel and Bryan (9) as modified (10) was applied to the experimental results---This statistical pro^e4ur-e-is one of several available which allow the estimation of carcinogenic risks associated with dose levels of the carcinogen below those necessary to yield a positive carcinogenic response in the animal experimentation. In the present example, the levels of interest are the average and maximum intakes, 0.1 and 0.3 pp'o aflatoxins, respectively (see Table IV). No experiments have been conducted in rats at these low dose levels, so no direct measure of risk can be made; but, as noted, the Mantel-Bryan procedure can be applied to the available animal data to estimate the risks associated with the 0.1 and 0.3 ppb dose levels. It must be kept in mind that the Mantel-Bryan procedure estimates animal, not human, risks. The relative sensitivities of humans and, in the N k bV > present case, rats, remains to be considered. The results of the risk assess ( J y -) ment in rats are presented in Table V. OA. \U u BFG17222 As can be seen from Table V, the estimated lifetime liver cancer incidefice_rat.es vary widely among the rat studies, ranging from a low of 30 to' a high-of-1400 per 10Q-,60Q.- Another indication of-the inherent Variabilit3> in these-' experiments is''provided'by-the upper" 99% confidence limits 6r-risk shown" in-parentheses- in Table V. - TABLE V: Estimated Lifetime Cancer Risks In Laboratory Animals (Rats)-- Incidence Rate Per 100,000^ DATA SOURCE (Reference No.) AFLATOXIN INTAKE (0.1 opb) (0.3 pob) ----' (12)............. 7 (13) ............... (14) (15) ' 2 f( 60)30( 220) 70( 600) 140 ( 620) 320(1400) ' COMBINED RAT STUDIES COMBINED ANIMAL STUDIES^ 240( 470) 17 ( -- ) .... ..... 2/( 320)2 i60( 990) 360(2300) 650(2400) 1400(4600) 1100(1900) 126( -- ) 1/ Incidence rates are based on the procedure of Mantel and Bryan (9), modified by Mantel, et al. (10). 2J Cannot be estimated. 3/ Numbers in parentheses are upper 99% confidence limits. 4/ Includes rats, rainbow trout, ducks, and tree shrews. See text. Risks are graphically estimated. BFG17223 rc -12- The upper 99% confidence limits on the lifetime risks derived from the studies in Table V are not only conservative risk estimates (i.e., they represent a "worst case" estimate) , but also are useful for comparing the several studies since the confidence limits take into account differences in sample size and other variables in experimental design. Although the results of 'these five rat studies are too variable to justify statistically their pooling, they have been combined in Table V to indicate an overall level of response in rat studies involving different strains. Another approach to aflatoxin risk assessment based on animal studies is that taken by Cornfield, et al_. (16). Combining data from experiments on the rainbow trout, duck, tree shrew, and six different studies on five different strains of rats, these authors fitted various mathematical models. A probit model with a slope of 1.2 probits per log^g dose appeared to give the best fit. Estimates from graphs in this paper yield lifetime incidence rates of approximately 17 and 126 per 100,000 for doses of 0.1 and 0.3 ppb, respectively. As shown in Table IV, the average and maximum dietary intakes of afla toxin contributed by peanuts are approximately 0.005 and 0.01 ppb, respec tively. For the combined rat studies, the risk estimates for these aflatoxin levels are 2 and 7 per 100,000, respectively. Utilizing the Cornfield approach yields risk estimates on the order of less than 0.1 per 100,000 (graphically estimated) for the 0.005 and 0.01 ppb levels. >. I O:v . BFGl 7224 c r-13- To assess the reasonableness of these animal derived risk, estimates, it is of interest to compare them to the overall human liver cancer incidence rates in the United Stake's'.' The "tcTtal crude -Tnciaence''per year of 'primary liver'cancer in the" total' Unit eh"'S takes"'pophTatiohy'estimated' in'19`69 , was 2.2/10^ (17). 'For purposes' 'oT'comparison 'wittf tKe Tffatiae animal risk data,' it is" neces'sary to break 'down the 'above crude yearly incidence fate "in "terms o'f 'age-specific rates, ~to multiply "aach 'age-specific rate by a factor which reflects"the proportion of 'th'e 't'o'taT'liver cancer rate cont'ribiited hy members of"that 'age'group," 'and'"theri 'to' multiply each such agespecific rate by 'the expected human lifespan o'f'70 years. Recombining the 'age'-'specific rates yields' a calculated lifetime risk for liver cancer occurrence'of 161 for'every TOO,000 "in' the total' U.'S- 'population. Primary7 "liver''ca'hc'e'r lifetime _iricide'rice''fates in Alabama "and "Georgia, estimated in the same'way, are 100/10^ and 103/10^, respectively (18). The rates for these two states, which are the only ones readily available for the south- eas'terri states, 'are `most appropriate' 'for 'co'mp'aris'ori with the risks estimated 'from the'animal' data "since the'y 'repre'seri't`"'the "re'giori 'with th'e highest 'antici pated exposure to aflatoxins. The risks calculated from the rat data in Table V not only differ greatly among themselves, demonstrating an intraspecies difference in sus ceptibility, but, at the 0.3 ppb dose level, individually exceed the lifetime risks of primary liver cancer in the human population of the United States from all causes. Possible explanations for these differences are: 1) the level of human exposure to aflatoxin has been over-estimated; 2) the Mantel-Jv h Bryan extrapolation procedure is overly conservative in this case; and/or O 3) rats may not be an appropriate model for predicting aflatoxin-iniuced J. J (...) O BFG17225 r-14- primary liver cancer in humans. While data tc test the validity of the first or second explanation are not available, the third receives some cor roboration from studies showing differences in aflatoxin metabolism among l various animal species, most relevant of which are the differences between rats and humans observed _in vitro. Available in vitro data indicate that the ability of an animal species to metabolize aflatoxin to a metabolite called aflatoxicol (AFL) is related to the sensitivity of the species to acute aflatoxicosis, and that the ratio of reductive and oxidative activi ties, measured as the ratio of AF1 concentration to that of an oxidized metabolite known as Q^, is an index of species susceptibility to the carcino genic effects of B^. Although _in vitro data for humans are extremely limited, the data available indicate that, with respect to these modes of metabolic behavior, humans are closer to relatively resistant species such as mice and monkeys than they are to the highly susceptible rat strains (19). b. Epidemiology Data: Data reflecting actual consumption of aflatoxins over defined periods of time by particular individuals who develop liver cancer are not available. The data most closely representing that situation are those derived from studies in Thailand (20) , Kenya (21) , Mozambique (22), and Swaziland (23). For two or more districts in each of these countries, average aflatoxin content in samples of meals actually consumed were deter mined. These estimates of average aflatoxin content were then correlated with official annual liver cancer mortality statistics. As previously stated, there are limitations to these epidemiology data, not the least of which is their inherent inability to establish direct cause-effect relation- ships. For example, any comparison based on these epidemiological data tV too BFG17226 rr -15- rests on the assumption that sample meals are representative, that current contamination levels are indicative of past experience (i.e., 10-20 years ago), and that there is no significant exposure to other possible causes of liver cancer. With respect to other possible causes, there is evidence that pvrroli- zidine alkaloids consumed in herbal medicines and hepatitis-3 virus may play a role in the etiology of primary liver cancer in some of the areas studies (24,25). Thus, the appropriateness of applying the human epidemiology data derived from these studies to low-incidence western countries is question able. Several other interfering factors, well-described by van Rensburg (24), further cloud the epidemiological data. The calculations made below should be considered against this background of uncertainty. The results of the various epidemiology studies have been standardized by van Rensburg (24) and these data are presented in Table VI. There is a highly significant positive correlation between cancer rate and estimated level of aflatoxin intake in the geographical areas studied, and a linear relationship between log aflatoxin 3^ level and cancer incidence, expressed as follows: y - 6.449 log x - 2.4115 where y = total cancer rate/10^/year and x = aflatoxin concentration in ng/kg/body weight/day. BFG17227 i.--. b*! W -M O b ( -16- f TABLE VI:'"Summarized Results b'f~ Epidemiology Studies Measuring' ?ri mary ""`Liver' Cancer'Rate' and A'flatokiiv Intake* (ng /kg body weight/day). "(From Reference 24^'" - - - - : ' r -: -r -- ""il^er-"57" c:.::Locaiij:y; :: r:r rur.-: z: : . r':. i; Cancer- Rate- - (10^/year) vr ir. r; - -: :' ..." ' " ----- Estimat ed - Aflatoxin. Intake (ng/kg body weight/day Kenya - High Altitude Thailand - Songkhla Swaziland - Highveld Kenya - Middle Altitude Swaziland - Middleveld Kenya - Low Altitude Thailand - Ratburi Swaziland - Lowveld Mozambique - Inhambane * * *" ` T 0.7 2.0 " 2" V ' 2.9 4.0 4.2 6.0 9.7 13.0 .................... 3.5' 5.0 ,! j 1 5.1 , 7 < L 5.8 . z> tD 8". 9 , 22 V 10.0 45.0 43.1 /i't 222.4 BFG17228 f; c. bX a r.o P b r -17- C Using the estimated average and maximum chronic exposures for dietary levels of aflatoxins in the southeastern states (2-7 and 9.0 ng/kg body . weight/day,respectively, Table IV) and the lo'g relation between dietary aflatoxin and cancer incidence derived from the epidemiological studies, the calculated range in the expected crude cancer incidence/year is 0.37 to 3.74 per 100,000. The actual crude rates/lO'Vyear in Alabama and Georgia are 0.97 and 1.18, respectively (18). It should be noted that the above calculations involve a significant assumption that quite possibly is not the case. They assume that the back ground liver cancer rate (i.e., that rate due to unknown factor(s) other than aflatoxin present in the areas of the epidemiological studies) is equal to that of the United States. Of course, there is also the implicit assump tion that aflatoxia contributes to the rate of liver cancer in this country; there is no evidence that this is true. And certainly there are other sub stances in the environment, including the workplace, that are known to be hepatotoxic and could contribute to this disease rate. In an attempt to reduce the effects of this possible error, a different type of risk estimate has been made. Specifically, a linear, non-logarithmic relationship, derived from the human epidemiological data tabulated by Peers and Linsell (26) , allows estimation of the human cancer rate related to dietary aflatoxins (or to other interfering factors, since these epidem iologic studies cannot conclusively show that the increased rate of liver cancer above background is due only to dietary aflatoxin). The relationship y = .106 x + 2.2, where y and x have the same meaning as above and 2.2 is the background cancer incidence/10^/year when x = 0. BFG17229 -.4 o c -13- - , ..ESTIMATES OF, RISKS AT-ALTERNATIVE TOLERANCES':. . .. ...Using the-last derived relationship theoretical: risk estimates have made for various aflatoxin levels,that might result.from alternative toler ances... These are shown iniTable VII;..and are based.on. the-assumption that .the legal maximum level of -aflatoxins in peanut products is directly pro portional to .the average level.of aflatoxins actually-found in these products. For...example, . the current FDA action level of 20 ppb leads to survey estimates of average levels of-2 ppb in peanuts_and peanuts products and 10 ppb in dry ..milled com jpreduces. -This .information-is .based .on .data collected while the 20 ppb action level was in affect for all food commodities. If one assumes a direct proportional relationship, -a reduction of the 20 ppb level to 15 ppb .for. .peanuts .and peanut -products could produce .an average level of 1.5 ppb .aflatoxins .in this-type .of food.. Similarly, a reduction of the tolerance -ta d-Q .ppb and .5 ppb -could-produce.-average levels of 1.0 ppb and 0.5 ppb, respectively. Since, for purposes of this assessment, the legal limit for .corn is not under consideration, the estimated average level of aflatoxins .in this food is held constant. ............................ .. ^i i >V_> ; 4.# BFG17230 l t. b V* - r b-* c -20- As shown in Table VII, Che theoretical cancer rates over a lifetime of exposure (70 years) are estimated for the Southeast to range from 20.2 to 67.0/10"' at the current 20 ppb action level; 19.9 to 66.4/103 at a 15 ppb tolerance, 19.6 to 65.6/10^ at a 10 ppb tolerance and 19.4 to 65.0/10^ at a 5 ppb tolerance. Thus, based on the epidemiological data (26), assessment of the estimated risk of human exposure to varying levels of aflatoxins in peanuts and peanut products shows relatively no significant gain in the protection of the public health (i.e., reduction in liver cancer) by adopting a tolerance of 5 ppb (currently the lowest enforceable tolerance) as opposed to FDA's proposed tolerance of 15 ppb. This analysis applies to the population at highest risk, namely that of the southeastern United States. In other areas of the country, where aflatoxin contamination of corn is not a significant problem, the health risks from aflatoxins are due, for the most part, to exposure to contaminated peanuts. The estimated lifetime risk to humans from this source of exposure only, based on the same type of cal culations used to derive Table VII, ranges from 0.8 to 2.0/10"5, if it is assumed that all peanut products contain 1.5 ppb aflatoxins (under the pro posed 15 ppb tolerance). This theoretical lifetime risk of cancer would be reduced to 0.5-1.3/10^ at a 10 ppb tolerance and 0.3-0.7/103 at a 5 ppb tolerance. For the convenience of the reader, the lifetime liver cancer incidence rates listed in the foregoing tables are summarized in Table VIII. *c- |,v. o CO w ..-a C4 ^**5 BFG17232 f r. -21- TABLE VIII: Summary of Maximum Expected Lifetime Liver Cancer Incidence ""Rates Due to Aflatoxin Exposures Under Current FDA Guideline ... Of 20 PPB- _ . -; ; Actual Incidence Rates^ _: Total U: S. - 161 per 100,000 Alabama., - 100 .per .100,000 Georgia - 103 per 100,000 Estimated Incidence Rates Due to Aflatoxins 2..................... ................................ / -.....................- From Combined' Rat Studies ^ ...................... . . Total Consumption: 24O`to'il00 per 100,000 -. .. ..^-..Peanut Products-Only: 1.:: . 2 to..7;ger.100,000 :' 2/ " ' - From Combined Animal Studies -- 'Total' Consumption: ` 17 to 126 per 100,000 Peanut-Products Only:- :- Less chan .1 per 100,000 - 37 From Epidemiology Studies -- e r Total'Consumption: 20:2 to 67.0 per 100,000 : --Peanut Products Only:.; - -1,1_ to 2_: 7:per 100,000- . 1/ See Text pages 15, 16. 2/ See Text Table V. 3/ . See Text Table VII. BFG17233 fO M O too Cg km fr REFERENCES (1) Rcdricks, J. V., Hesseltine, C. W., and Mehlman (Eds), Mycotoxins in Human and Animal Health, Pathotox Publishers, Park Forest South, Illinois, 1977; pp. 6-186. e2) Wogan, G. N. (1973) Aflatoxin Carcinogenesis, in Methods in Cancer Research, Vol. VII,- H. Busch (Ed.), Academic Press, New York, New York, pp. 309-344. (3) Stoloff, L., Aflatoxins--An Overview, in Reference 1, pp. 7-28. (4) Rodricks, J. V., and Stoloff, L., Aflatoxin Residues From Contaminated Feed in Edible Tissues of Food-Producing Animals, in Reference 1, pp. 67-79. (5) Tiemstra, P. J., Aflatoxin Control During Food Processing of Peanuts, in Reference 1, pp. 121-138. (6) LilleHoj, G. B., and Hesseltine, C. W. , Aflatoxin Control During Plant Growth and Harvest of Com, in Reference 1, pp. 107-119. (7) Stoloff, L., Occurrence of Mycotoxins in Food and Feed, in Mycotoxins and Other Fungal Related Food Problems, J. V. Rodricks (Ed.), Advances in Chemistry Series, No. 149, Am. Chem. Soc., Washington, D.C., pp. 23-50. (8) U.S. Department of Agriculture Survey of Household Food Consumption in the United States, 1965-1966; Food Intake of Individuals, Spring 1965 (com puter tapes). (9) Mantel, N., and 3ryan, W. R. , "Safety" Testing of Carcinogenic Agents, ^J. Natl. Cancer Inst. , 27, 455-470 (1961). (10) Mantel, N., e_t _al. , An Improved Mantel-Bryan Procedure for "Safety" Testing of Carcinogens, Cancer Research, 35, 865-872 (1975). (ID Alfin-Stater, R. 3., et al., Studies of Long-Term Administration of Afla toxin to Rats as a Natural Food Contaminant, J. Amer. Oil Chem. Soc. , 46, 493-497 (1969). (12) Newberne, P. , and Butler, W. H., Acute and Chronic Effects of Aflatoxin on the Liver of Domestic and Laboratory Animals, Cancer Res., 29, 236-250 (1969). (13) Wogan, G. N. , et_ al. , Carcinogenic Effects of Low Dietary Levels of Afla toxin B]_ in Rats, Food Cosmet. Toxicol. , 12, 681-685 (1974). (14) Barnes, J. M., and Butler, Carcinogenic Activity of Aflatoxin to Rats, Nature, 202, 1016 (1964). BFG17234 y*. O tV C.O o ca \ / <r ( 2- - (15) Nawberne, P. , and Williams, G., Inhibition of Af-latoxin -Carcinogenesis by~D"iethylstilbestrol in Male Rats, Arch. Environ. Health, 19, 489-498 (1969) (16) .Carnfield, J., et al., Private Communication, 1977. (17) Biometry 3ranch, National Cancer Institute, Preliminary Report, Third National Cancer Survey, ,1969 Incidence. National Institutes of Health, DHEW, 1971. (18) Stoloff, L., and Friedman, L., Information Bearing on the Evaluation of the Hazard to Man From Aflatoxin Ingestion, PAG Bulletin, Vol. VI, No. 2, pjJ,;;21-32 .(1976). : u r;.................... .. (19) Hsieh, D., e_t al.. , Comparative Metabolism of Aflatoxin, in Reference 1, pp_. .37-50. _ -- (20) Shank, R., al., Dietary Aflatoxins and Human Liver Cancer IV, Food Cosmet. Toxicol., 10, 171-179 (1972). .- (21) (22) 'PTers, F7,*"ahd'Linsell, C., Dietary Aflatoxins and Liver Cancer--A Popu lation 3ased Study in Kenya, Brit. J. Cancer, 27, 473-484 (1973). < - van Rensburg, S. J., et^ ail. , Primary Liver Cancer and Aflatoxin in a High Cancer Area, S_. Afr. Med. J_., 2508a-2508d (1974). (23) Peers, F., et al., Dietary Aflatoxins ana Human Liver Cancer, A Study in Swaziland, Inst. J^. Cancer, 17, 167-176 (1976) (24) van Rensburg, S. J. , Role of Epidemiology in Elucidation of Mycotoxin Health Risks, in Reference 1, pp. 699-712. (25) Fakunie, Y. M., et al. , Primary Liver Cell- Carcinoma (PLCC) -in-the Northern Guinea Savanna of Nigeria, Trans. Royal Soc. Trot). Med. Hyg. , 71, 335-337 (1977). (26) Peers, F., and Linsell, C. A., Dietary Aflatoxins and Human Primary Liver Cancer, Presented at III IUPAC Symposium on Mycotoxins in Foodstuffs, Paris, Sept. 1976. BFG17235 N H' O (.0 o L-; O'