NATURAL TOXINS I:100-1 05 (I992)
Aflatoxins Isolated by lmmunoaffinity Chromatography From Foods Consumed in The Gambia, West Africa Geoffrey J. Hudson, Christopher P. Wild, Audrey Zarba, and John D. Groopman MRC Dunn Nutrition Unit, Cambridge, England (G.J.H.); International Agency for Research on Cancer, Lyon, France (C.P.W.); Department of Environmental Health Sciences, The Johns Hopkins University School of Hygiene and Public Health, Baltimore, Maryland (A.Z., J.D.G.)
ABSTRACT An aflatoxin-specific, monoclonal antibody-based immunoaffinity chromatography method has been develowd for the rapid isolation of aflatoxins from human foods. Aflatoxins were isolated by immunoaffinity chromatography from a variety of cooked foods, including maize, rice, millets, groundnut sauces, and leaf sauces, collected in The Gambia, West Africa. The aflatoxins were measured by direct fluorescence or high-pressure liquid chromatography. The highest levels were found for groundnut sauces, mean 162 ppb (range 18 to 943 ppb) for 18 positive samples, but aflatoxins were found in other foods; e.g., maize, mean 9.7 ppb (range 2 to 35 ppb) for nine positive samples. The food analysis results were used with records of the amounts of cooked food to estimate a mean daily intake for an individual of the order of 3.5 pg of aflatoxins per day. This approach for exposure assessment is considered in relation to other biornarkers of aflatoxin exposure using biological fluids. 1999 wiley-Liss, Inc. @
Key Words: Chemical carcinogens, Liver cancer, Mycotoxins
INTRODUCTION
Human populations are exposed to aflatoxins by consuming foods contaminated by toxin-producing strains of Aspergillus jlavus and Aspergillus parasiticus during growth, harvest, or storage [Busby and Wogan, 198.51. Many factors affect the levels of aflatoxins produced by these fungi, including the strain of fungus, the intrinsic susceptibility of the host plant, post-harvest storage conditions, and the chemical composition of the commodity. In addition, environmental factors such as drought and insect damage increase the likelihood of aflatoxin production [CAST Report, 19891. The requirements for aflatoxin contamination by these molds are relatively non-specific and the toxins have been found in nearly every agricultural commodity. Visible contamination of a food is not proof of the presence of the toxin, only chemical analysis is able to detect aflatoxins positively in food. Many analytical methods exist for the measurement of aflatoxins in food. These techniques include chromatography following extraction with solvent, such as the minicolumn, thin-layer chromatography and high-pressure liquid chromatography (HPLC) [Trucksess et al., 1984; Groopman and Donahue, 19881, and immunological assays using specific antibodies or antisera, such as enzyme-linked immunosorbent assays and radioimmunoassays [Trucksess et al., 1989; Chu et al., 19871. 0
1992 Wiley-Liss, Inc
Each method has unique characteristics with respect to specificity and sensitivity; some methods are adequate only for qualitative analysis and some are quantitative. A major problem in measuring trace amounts of aflatoxins in a food is the lack of selectivity of extraction with solvents. Such extracts may contain hundreds of chemical compounds, and the ability to separate interfering substances from the material of interest will specify the eventual sensitivity of the method. We have utilized the inherent strengths of monoclonal antibody selectivity in an immunoaffinity column to isolate low levels of aflatoxins from complex mixtures. This technique has been used to analyse samples of urine and serum [Groopman et a]., 1984, 1992a,b,c,d], and here we describe its use for the analysis of diverse human foods. The specificity of the procedure has permitted the direct measurement of aflatoxins by fluorescence and the separation and measurement of individua1 aflatoxins by HPLC. The measurements have been used in conjunction with records of weighed amounts of food to estimate the levels of exposure to dietary aflatoxins in a West African village. The difficulty of assessing individual human exposure to Received February 17, 1992, accepted for publication August 24, 1992 Address reprint requests to Dr John D. Grooprnan, Department of Environmental Health Sciences, The Johns Hopkins Unwersity School of Hygiene and Public Health, 615 North Wolfe Street, Baltimore, MD 21205
101
AFlATOXlNS IN HUMAN FOODS
aflatoxins using measurements of dietary intake is discussed in relation to the biomarkers now available for quantification of this exposure. MATERIALS AND METHODS Sample Collection
Samples of cooked food were collected as part of two separate studies in the village of Keneba in The Gambia, West Africa [Groopman et al., 1992c, Wild et al., 1992; Hudson and Day, 19891. Usually, only two meals are cooked each day, consisting of a staple dish and a sauce. The most frequently eaten staple is rice, but the diet includes Sanyo millet (Pennisefum typhoideum), sorghum (Sorghum margaritiferum), and maize (Zea mays). Sauces are usually based on roast groundnuts (Arachis hypogaea) or on leaves when groundnuts are in short supply. Fieldworkers recorded the identity and weighed the quantity of each ingredient used in the preparation of a meal. The cooked food is placed into one or more bowls from which it will be eaten. These bowls were weighed empty, again after the addition of the staple, and again after the addition of the sauce. Samples were taken immediately before the subjects were to eat, and a record was made of the people who were to eat from each bowl. In one study, over 40 different food items were collected [Hudson and Day, 19891 and in another study, 400 food samples were taken over a 7-day monitoring period [Groopman et al., 1992~;Wild et al., 19921. Samples were freeze-dried in the MRC laboratory at Keneba. People in this community eat directly from shared bowls of cooked food. The food is taken a handful at a time and transferred directly to the mouth. Therefore, it is impossible to measure the amount of food eaten by an individual. In these studies, the amount of cooked food in each bowl was known and the identity of the people who ate from each bowl was recorded. Estimates of food intake were made by dividing the amount of cooked food mathematically amongst the people, weighting for sex and age as: woman, 1.0; man, 1.25; girl, 0.5; boy, 0.625. In other words, men were assumed to eat 25% more than women, and children (up to 16 years) were assumed to eat half the amount of food assigned to an adult of the same sex. These weighting factors and the measured levels of aflatoxins in the food were used in one study [Wild et al., 19921 to estimate the exposures of 20 individuals matched, as far as possible for age and sex. Isolation and Measurement of Aflatoxins
Aflatoxins were isolated from freeze-dried food samples by monoclonal antibody immunoaffinity chromatography and quantified by direct fluorescence or by HPLC as described [Groopman and Donahue, 19881. To extract the aflatoxins, 10 ml of 70% (v/v) methanol per gram of dried food was added and NaCl to 2% (w/v)
final concentration. The mixture was blended at high speed for 1 min then filtered through filter-paper. Distilled water was added to the filtrate to give 20% (v/v) methanol final concentration. A sample equivalent to 1 g of food was applied onto the affinity column [Groopman et al., 19841 or onto an Aflatest column (Vicam Inc., Somerville, MA). The column was washed twice with 10 ml of water each time and then the aflatoxins were eluted with methanol. The fluorescence of the eluate was measured directly (360 nm excitation and 450 nm emission wavelength). Positive samples were dried under reduced pressure and redissolved in 10 mM triethylammonium formate (TEAF), pH 3.0, for HPLC analysis. HPLC Analysis
HPLC separation and measurement of individual aflatoxins was done with a Beckman model 324M Gradient Liquid Chromatograph equipped with a HewlettPackard model 1040 A Diode Array detector. A 5pm C I 8 Ultrasphere reverse column (Rainin Instrument Co., Woburn, MA) was used with isocratic elution with 23% (v/v) ethanol, 10 mM TEAF, pH 3.0 at a flow-rate of 1 ml/min. The column temperature was maintained at 35°C or 55°C. Peak areas on the chromatograms were integrated and converted to nanogram values using a standard curve of authentic aflatoxins. The detection limit of the method usi,ng ultraviolet light absorption is 0.5 ng for all aflatoxins. RESULTS
Most of the food samples analysed were collected as part of a study of seasonal changes in the diet in The Gambia [Hudson and Day, 19891 in September, at the end of the rainy season, with the exception of the millet samples, which were obtained in June, at the end of the dry season. All samples were freeze-dried in The Gambia before shipment. Only cooked foods were sampled, so the intake of aflatoxins from any contaminated uncooked food that the subjects may have eaten is not known. Samples of maize, rice and millets, and of groundnut sauce and leaf sauce were analysed by fluorescence or HPLC after isolation by monoclonal antibody immunoaffinity chromatography. The highest levels of contamination were found in the groundnut sauces. Of 20 samples analysed, 18 were positive, with levels ranging from 19 to 943 ppb, mean 162 ppb. Figure 1 shows the HPLC profile of the most heavily contaminated sample, revealing the presence of all four of the major aflatoxins: aflatoxin B , (AFB,), aflatoxin B, (AFB,), aflatoxin G I (AFG,), and aflatoxin G2 (AFG,). This is typical of most groundnut samples, which are often contaminated by A. parasiticus which produces all four of the aflatoxins [CAST Report, 19891. Nine .of the ten maize samples examined contained de-
102
HUDSON El AL.
210
250 300 350 Wavelength ( n m ?
I
I
403
Reference
spectrum
AFB 1 8.1
4
2
e
6
Time
I0
12
14
16
19
(mfn.)
Fig. 1. HPLC prohle of the aflatoxins found in a groundnut sauce sample prepared using a monoclonal antibody immunoaffinity column Details of the HPLC method are described in the Materials and Methods section, and the identity of the indlvldual aflatoxins was assigned by co-chromatography with authentic standards HPLC column temperature was 35'C
I
RFBl
AT 21.8
MIN
3 00 350 Wavelength ( n m )
5
10 Time
A
4 00
15 (min.)
11
B
25 1 Refer e n ce
20
4Ql s p e c t r urn
25
Fig. 2. HPLC prohle of the aflatoxins found in a maize sample prepared using a monoclonal antibody immunoafhnity column Details of the HPLC method are described in the Materials and Methods section, and the identity of the indhtidualaflatoxins was assigned by co-chromatography with authentic standards HPLC column temperature was 35'C
tectable levels of aflatoxins; mean 9.7 ppb, range 2 to 35 ppb. Figure 2 shows the HPLC profile of the most heavily contaminated sample of maize, revealing the presence of only AFB, and AFB2. Nine samples of Sanyo millet were examined; all contained aflatoxins, mean 9.8 ppb, range 1 to 27 ppb. One of these was examined by HPLC and only AFB, was detected. Eight samples of sorghum were examined, two were positive for aflatoxins with levels of 2 ppb and 16 ppb. Rice is the major dietary staple in this community; 20 samples were analysed and 14 contained detectable levels of aflatoxins, mean 7.9 ppb, range 2 to 19 ppb. Finally, three leaf-sauces were analysed; they contained aflatoxins at levels of 21 ppb, 26 ppb, and 34 ppb. It is clear that aflatoxin contamination is widespread in the diet and is not restricted to ground-
nuts and maize. The results of these analyses identified which foods might be important in a study of aflatoxin intakes. In a separate study [Wild et al., 19921, 20 adults were matched, as far as possible, in pairs for age and sex, with one member of each pair positive and the other negative for evidence of exposure to the hepatitis B virus. The diets of the subjects consisted predominantly of boiled rice eaten with either a leaf sauce or a groundnut sauce. A total of 87 leaf sauces, 22 flour sauces, and 47 groundnut sauces was analysed; 64%, 59%, and 87%, respectively, contained detectable aflatoxins. The highest level of aflatoxins found was 774 ppb in a leaf sauce that contained raw groundnuts and Figure 3 shows the HPLC profile for this sample.
103
AFIATOXINS IN HUMAN FOODS
250 300 3S0 W a v e length t n m )
210 4 03 Reference spectrum
J
3
a
4:
RFGl 10 Time
AFB2
20
30
(min.1
Fig. 3. HPLC profile of the aflatoxins found in a leaf sauce containing groundnuts sample prepared using a monoclonal anttbOdy immunoaffinity column Details of the HPLC method are described in the Materials and Methods section, and the identity of the individual aflatoxins was assigned by co-chromatography with authentic standards HPLC column temperature was 55'C
The vast majority of the toxin present is AFB,, which is in contrast to the presence of all four aflatoxins found in the groundnut sauces, suggesting that the groundnuts were not the source of this contamination. Of the positive groundnut sauce samples, six contained aflatoxins at levels in excess of 100 ppb. HPLC analysis of these samples yielded chromatograms (not shown) that were very similar to that shown in Figure 1, with up to 35% of the total aflatoxin content as AFG,. Thirty samples of other sauces were examined; 22 contained no detectable aflatoxins and only one of the eight positive samples had a level as high as 20 ppb. None of the 15 samples of boiled rice examined from this study contained detectable aflatoxin. Determination of Aflatoxin Intakes
The analytical values for aflatoxins in the samples of cooked foods were used with records of the amounts of food eaten to estimate the intakes of aflatoxins from those sources as described in Materials and Methods. Figure 4 shows a box and whisker plot of the intake values for individuals and on a per kg basis. DISCUSSION
The development of methods to assess human exposure to environmental carcinogens requires techniques that are sensitive, specific, and amenable to large numbers of samples. For example, the chromatographic methods [Trucksess et al., 1984; Groopman and Donahue, 19881have the power to resolve mixtures of aflatoxins into the individual components, providing that the extraction procedure does not introduce high levels of interfering compounds. The antibody-based methods [e.g., Chu et al., 19871are more specific and selective than straightforward chromatography but immunoassays are
compromised by the intrinsic cross-reactivity of an antibody for multiple aflatoxins. The application of immunoaffinity chromatography to the analysis of environmental carcinogens has increased steadily and its use in aflatoxin analysis [Groopman et al., 1992a,b]. benzo(a)pyrene metabolite detection [Garner et al., 19881, oxidative DNA damage products [Degan et al., 19911; and heterocyclic amine measurements [Vanderlaan et al., 1991; Turesky et al., 19891 has been reported. The major advantage of the approach is the single-step removal of interfering substances from complex mixtures as diverse as food, serum, milk, and urine. The chromatography is non-destructive, allowing subsequent analysis by mass spectroscopy, HPLC or other specific methods. We have taken advantage of the inherent strengths of both antibody selectivity and chromatographic separation to develop an immunoaffinity chromatography/HPLC procedure for the isoiation and measurement of aflatoxins in food and biofluids. Immunoaffinity chromatography provides a means to isolate authentic AF molecules from the complex milieu of food samples. This greatly facilitates subsequent qualitative and quantitative measurements. The major problems encountered in field studies aimed at measuring individual intakes of aflatoxins are not associated with these sensitive and specific assays but with extrapolating from the aflatoxin concentration in foods to the individual intakes. Measurement of the habitual food intake of a free-living individual is notoriously difficult [Bingham, 1987, 199I], and there is a vast and largely negative literature devoted to the subject [Borelli, 19901. Two particular problems are relevant to the studies referred to here. The food sampling process can be highly variable due to the heterogeneity of aflatoxin contamination [Campbell et al., 19861, and we were constrained to take small samples
104
HUDSON ET AL.
looooo W
z
f 10000
w
z X
a
0
A
II i
1000
100
contamination below the detection threshold of the analytical method could contribute significantly to total intakes of AF. It is clear that the development of methods using molecular dosimetry markers such as DNA adducts and protein adducts [Groopman et al., 19911will be crucial for the proper interpretation of field studies of human exposure to carcinogens.
25 I 0
P 4 2
r
0
TOTAL INTAKE
0
INTAKE
/
KG
Fig. 4. Total dietary intake of aflatoxins during 6 to 7 consecutwe days The box and whisker plot represents a description of the data showing the 10th and 90th centiles (ends of the whiskers), the ‘25th and 75th centiles (ends of the box), and the 50th centile (line within the box) Outlier points are shown as indlvldual values A Total intake B Intake per kg body weight
from large amounts of cooked food. Small samples were used because the removal of large samples unduly influences a subject’s food intake. I t was necessary to take samples from larger quantities of food than are commonly encountered in studies of food intake. This is because it is normal in the study community for people to eat in groups, taking food a handful at a time, directly from a large communal bowl. This leads directly to the second problem; how to determine the distribution of a bowl of food between the people who eat from it. The stratagem used here is crude but unlikely to lead to serious errors; the amounts of food eaten are probably underestimated for older children and overestimated for the younger, smaller children but this will tend to cancel out somewhat when children are considered as a group. Similar calculations have been made for energy intakes in this community and the results are compatible with independent values for energy expenditure (Hudson, unpublished results). The results for A F intake are correlated with, but absolute values do not agree with, those obtained by extrapolation from the values for serum albumin-AF adduct in peripheral blood. The levels of AFalbumin adduct reflect long-term exposure to A F and levels of A F metabolites in urine reflect recent exposure. The levels found in the study referred to here [Wild et al., 19921suggest that the levels determined in the food samples are too low, and there are several possible reasons for this. There are the problems associated with sampling food and estimating food intakes discussed above, and the fact that food eaten at any time other than the two main cooked meals was not monitored. Further, none of the rice samples analysed in this study contained detectable levels of AF, yet 70% of the samples analysed in a previous study in this village [Hudson and Day, 19891 were shown to be contaminated with AF. Since considerable quantities of boiled rice are eaten, even a level of
ACKNOWLEDGMENTS
This work was supported by grant CA 48409 from the USPHS and by the Department of Cooperation and Development of the Ministry of Foreign Affairs of Italy. Salary support for A 2 was provided by grant ES 07141 from the NIEHS and for JDG by Research Career Development Award CA 01517. REFERENCES Bingham SA ( 1 987): The dietary assessment of individuals: Methods, accuracy, new tcchniqucs and rccommcndations. Nutr Abstr Rev [A] 57:705-742. Bingham SA (1991): Limitations of thc various methods for collecting dictary intake data. Ann Nutr Mctab 35:l 17-127. Borrclli R (1990): Collection of food intake data: A reappraisal of critcria for judging the mcthods. Br J N u t r 63:411-417. Busby WF. Wogan GN (1985): Aflatoxins. In Scarle CE (ed): “Chernical Carcinogens.” 2nd ed.Washington, DC: American Chemical Socicty, pp 945- 1 136. Campbcll AA, Whitakcr TB, Pohland AE, Dickcns JW. Park DL (1986): Sampling, sample prcparalion, and sampling plans for foodstufs for mycotoxin analysis. Purc Appl Chcm S8:305-314. CAST Report (1989): “Mycotoxins: Economic and Health Risks.” Council for Agricultural Science and Technology. Task Force Rcport 116. pp 1-91. Chu FS. Fan TSL, Zhang GS. Xu YC. Fdust S, McMahon PL (1987): Improved enzyme-linked immunoassay for aflatoxin B I in agricultural commodities. J Assoc Off Anal Chem 70:854-857. Dcgan P. Shigenagd MK. Park E-M, Alperin PE, Ames BN (1991): Immunoaffinity isolation of urinary 8-hydroxy-2‘-deoxyguanosine and 8-hydroxyguanine and quantitation of 8-hydroxy-2’-deoxyguanosine in DNA by polyclonal antibodies. Carcinogenesis I 2 865-871. Garner RC, Dvorackova I, Turi F (1988): Immunoassay procedures to detect exposure to aflatoxin Bi and benzo(a)pyrene in animals and man at the DNA level. Int Arch Occup Environ Health 601451so. Groopman JD, Donahue KF (1988): Aflatoxins, a human carcinogen: Analyzing foods and biological samples using monoclonal antibody affinity chromatography. J Assoc Off Anal Chem 71:861-867. Groopman JD, Trudel LJ, Donahue PR, Marshak-Rothstein A, Wogan G N (1984): High affinity monoclonal antibodies for aflatoxins and their application to solid phase irnmunoassays. Proc Natl Acad Sci USA 81:7728-7731. Groopman JD, Sabbioni G, Wild CP (1991): Molecular dosimetry of aflatoxin exposures. In Groopman, JD, Skipper P (eds): “Molecular Dosimetry of Human Cancer: Epidemiological, AndlytiCal and Social Considerations,” Boca Raton, FL: CRC Press, pp 302-324. Groopman JD, Hasler J, Trudel LJ, Pikul A, Donahue PR, Wogan GN (1992a): Molecular dosimetry in rat urine of aflatoxin-N ’-guanine and other aflatoxin metabolites by multiple monoclonal antibody affinity chromatography and HPLC. Cancer Res 52:267-274. Groopman JD, Zhu J, Chen J-S, Wogan GN (1992b): Molecular dosimetry of urinary aflatoxin DNA adducts in people living in
AFLATOXINS IN HUMAN FOODS
Guangxi Autonomous Region, People's Republic of China. Cancer Res 52:45-52. Groopman JD, Hall A, Whittle H, Hudson GJ, Wogan GN, Montesano R, Wild CP (1992~):Molecular dosimetry of aflatoxin-N -guanine in human urine obtained in The Gambia, West Africa. Cancer Epidemiol Biomarkers Prevention I :22 1-227. Groopman JD, Egner P, Love-Hunt A, DeMatos P, Kensler TW (1992d): Molecular dosimetry of aflatoxin DNA and serum albumin adducts in chemoprotection studies using I ,2-dithiole-3-thione. Carcinogenesis 13:101-106. Hudson GJ, Day KC (1989): Water content of the rural Gambian diet. Nutr Rep Int 40:335-339. Trucksess MW, Brumley WC, Nesheim S (1984): Rapid quantitation and confirmation of aflatoxins in corn and peanut butter, using a disposable silica gel column, thin-layer chromatography, and gas chromatography/mass spectrometry. J Assoc Off Anal Chem 67: 973-975. Trucksess MW, Stack ME. Nesheim S, Park DL, Pohland AE (1989): Enzyme-linked immunosorbent assay ofaflatoxins B, ,B,, and G , in
'-
105
corn, cottonseed, peanuts, peanut butter and poultry feed. J Assoc Off Anal Chem 72957-962. Turesky RJ. Forster CM, Aeschbacher HU, Wurzner HP, Skipper PL, Trudel LJ, Tannenbaum SR (1989): Purification of the food-borne carcinogens 2-amino-3-methylimidazo [4,5-flquinoline and 2-amino3.8-dimethylimidazo [4,5-flquinoxaline in heated meat products by immunoaffinity chromatography. Carcinogenesis 10: 151-1 56. Vanderlaan M, Alexander J, Thomas C, Djanegra T, et al. (1991): lmmunochemical detection of rodent hepatic and urinary metabolites of cooking-induced food mutagens. Carcinogenesis 12:349354. Wild CP, Jiang YZ,Allen SJ, Jansen LAM, Hall AJ, Montesano R (1990): Aflatoxin-albumin adducts in human sera from different regions of the world. Carcinogenesis I1:2271-2274. Wild CP, Hudson GJ, Sabbioni G, Chapot B, Hall A, Wogan GN, Whittle H, Montesano R, Groopman J D (1992): Dietary intake of aflatoxins and the level of albumin-bound aflatoxin in peripheral blood in The Gambia. West Africa. Cancer Epidemiol Biomarkers Prevention 1:229-234.
Aflatoxins Isolated by lmmunoaffinity Chromatography From Foods Consumed in The Gambia, West Africa Geoffrey J. Hudson, Christopher P. Wild, Audrey Zarba, and John D. Groopman MRC Dunn Nutrition Unit, Cambridge, England (G.J.H.); International Agency for Research on Cancer, Lyon, France (C.P.W.); Department of Environmental Health Sciences, The Johns Hopkins University School of Hygiene and Public Health, Baltimore, Maryland (A.Z., J.D.G.)
ABSTRACT An aflatoxin-specific, monoclonal antibody-based immunoaffinity chromatography method has been develowd for the rapid isolation of aflatoxins from human foods. Aflatoxins were isolated by immunoaffinity chromatography from a variety of cooked foods, including maize, rice, millets, groundnut sauces, and leaf sauces, collected in The Gambia, West Africa. The aflatoxins were measured by direct fluorescence or high-pressure liquid chromatography. The highest levels were found for groundnut sauces, mean 162 ppb (range 18 to 943 ppb) for 18 positive samples, but aflatoxins were found in other foods; e.g., maize, mean 9.7 ppb (range 2 to 35 ppb) for nine positive samples. The food analysis results were used with records of the amounts of cooked food to estimate a mean daily intake for an individual of the order of 3.5 pg of aflatoxins per day. This approach for exposure assessment is considered in relation to other biornarkers of aflatoxin exposure using biological fluids. 1999 wiley-Liss, Inc. @
Key Words: Chemical carcinogens, Liver cancer, Mycotoxins
INTRODUCTION
Human populations are exposed to aflatoxins by consuming foods contaminated by toxin-producing strains of Aspergillus jlavus and Aspergillus parasiticus during growth, harvest, or storage [Busby and Wogan, 198.51. Many factors affect the levels of aflatoxins produced by these fungi, including the strain of fungus, the intrinsic susceptibility of the host plant, post-harvest storage conditions, and the chemical composition of the commodity. In addition, environmental factors such as drought and insect damage increase the likelihood of aflatoxin production [CAST Report, 19891. The requirements for aflatoxin contamination by these molds are relatively non-specific and the toxins have been found in nearly every agricultural commodity. Visible contamination of a food is not proof of the presence of the toxin, only chemical analysis is able to detect aflatoxins positively in food. Many analytical methods exist for the measurement of aflatoxins in food. These techniques include chromatography following extraction with solvent, such as the minicolumn, thin-layer chromatography and high-pressure liquid chromatography (HPLC) [Trucksess et al., 1984; Groopman and Donahue, 19881, and immunological assays using specific antibodies or antisera, such as enzyme-linked immunosorbent assays and radioimmunoassays [Trucksess et al., 1989; Chu et al., 19871. 0
1992 Wiley-Liss, Inc
Each method has unique characteristics with respect to specificity and sensitivity; some methods are adequate only for qualitative analysis and some are quantitative. A major problem in measuring trace amounts of aflatoxins in a food is the lack of selectivity of extraction with solvents. Such extracts may contain hundreds of chemical compounds, and the ability to separate interfering substances from the material of interest will specify the eventual sensitivity of the method. We have utilized the inherent strengths of monoclonal antibody selectivity in an immunoaffinity column to isolate low levels of aflatoxins from complex mixtures. This technique has been used to analyse samples of urine and serum [Groopman et a]., 1984, 1992a,b,c,d], and here we describe its use for the analysis of diverse human foods. The specificity of the procedure has permitted the direct measurement of aflatoxins by fluorescence and the separation and measurement of individua1 aflatoxins by HPLC. The measurements have been used in conjunction with records of weighed amounts of food to estimate the levels of exposure to dietary aflatoxins in a West African village. The difficulty of assessing individual human exposure to Received February 17, 1992, accepted for publication August 24, 1992 Address reprint requests to Dr John D. Grooprnan, Department of Environmental Health Sciences, The Johns Hopkins Unwersity School of Hygiene and Public Health, 615 North Wolfe Street, Baltimore, MD 21205
101
AFlATOXlNS IN HUMAN FOODS
aflatoxins using measurements of dietary intake is discussed in relation to the biomarkers now available for quantification of this exposure. MATERIALS AND METHODS Sample Collection
Samples of cooked food were collected as part of two separate studies in the village of Keneba in The Gambia, West Africa [Groopman et al., 1992c, Wild et al., 1992; Hudson and Day, 19891. Usually, only two meals are cooked each day, consisting of a staple dish and a sauce. The most frequently eaten staple is rice, but the diet includes Sanyo millet (Pennisefum typhoideum), sorghum (Sorghum margaritiferum), and maize (Zea mays). Sauces are usually based on roast groundnuts (Arachis hypogaea) or on leaves when groundnuts are in short supply. Fieldworkers recorded the identity and weighed the quantity of each ingredient used in the preparation of a meal. The cooked food is placed into one or more bowls from which it will be eaten. These bowls were weighed empty, again after the addition of the staple, and again after the addition of the sauce. Samples were taken immediately before the subjects were to eat, and a record was made of the people who were to eat from each bowl. In one study, over 40 different food items were collected [Hudson and Day, 19891 and in another study, 400 food samples were taken over a 7-day monitoring period [Groopman et al., 1992~;Wild et al., 19921. Samples were freeze-dried in the MRC laboratory at Keneba. People in this community eat directly from shared bowls of cooked food. The food is taken a handful at a time and transferred directly to the mouth. Therefore, it is impossible to measure the amount of food eaten by an individual. In these studies, the amount of cooked food in each bowl was known and the identity of the people who ate from each bowl was recorded. Estimates of food intake were made by dividing the amount of cooked food mathematically amongst the people, weighting for sex and age as: woman, 1.0; man, 1.25; girl, 0.5; boy, 0.625. In other words, men were assumed to eat 25% more than women, and children (up to 16 years) were assumed to eat half the amount of food assigned to an adult of the same sex. These weighting factors and the measured levels of aflatoxins in the food were used in one study [Wild et al., 19921 to estimate the exposures of 20 individuals matched, as far as possible for age and sex. Isolation and Measurement of Aflatoxins
Aflatoxins were isolated from freeze-dried food samples by monoclonal antibody immunoaffinity chromatography and quantified by direct fluorescence or by HPLC as described [Groopman and Donahue, 19881. To extract the aflatoxins, 10 ml of 70% (v/v) methanol per gram of dried food was added and NaCl to 2% (w/v)
final concentration. The mixture was blended at high speed for 1 min then filtered through filter-paper. Distilled water was added to the filtrate to give 20% (v/v) methanol final concentration. A sample equivalent to 1 g of food was applied onto the affinity column [Groopman et al., 19841 or onto an Aflatest column (Vicam Inc., Somerville, MA). The column was washed twice with 10 ml of water each time and then the aflatoxins were eluted with methanol. The fluorescence of the eluate was measured directly (360 nm excitation and 450 nm emission wavelength). Positive samples were dried under reduced pressure and redissolved in 10 mM triethylammonium formate (TEAF), pH 3.0, for HPLC analysis. HPLC Analysis
HPLC separation and measurement of individual aflatoxins was done with a Beckman model 324M Gradient Liquid Chromatograph equipped with a HewlettPackard model 1040 A Diode Array detector. A 5pm C I 8 Ultrasphere reverse column (Rainin Instrument Co., Woburn, MA) was used with isocratic elution with 23% (v/v) ethanol, 10 mM TEAF, pH 3.0 at a flow-rate of 1 ml/min. The column temperature was maintained at 35°C or 55°C. Peak areas on the chromatograms were integrated and converted to nanogram values using a standard curve of authentic aflatoxins. The detection limit of the method usi,ng ultraviolet light absorption is 0.5 ng for all aflatoxins. RESULTS
Most of the food samples analysed were collected as part of a study of seasonal changes in the diet in The Gambia [Hudson and Day, 19891 in September, at the end of the rainy season, with the exception of the millet samples, which were obtained in June, at the end of the dry season. All samples were freeze-dried in The Gambia before shipment. Only cooked foods were sampled, so the intake of aflatoxins from any contaminated uncooked food that the subjects may have eaten is not known. Samples of maize, rice and millets, and of groundnut sauce and leaf sauce were analysed by fluorescence or HPLC after isolation by monoclonal antibody immunoaffinity chromatography. The highest levels of contamination were found in the groundnut sauces. Of 20 samples analysed, 18 were positive, with levels ranging from 19 to 943 ppb, mean 162 ppb. Figure 1 shows the HPLC profile of the most heavily contaminated sample, revealing the presence of all four of the major aflatoxins: aflatoxin B , (AFB,), aflatoxin B, (AFB,), aflatoxin G I (AFG,), and aflatoxin G2 (AFG,). This is typical of most groundnut samples, which are often contaminated by A. parasiticus which produces all four of the aflatoxins [CAST Report, 19891. Nine .of the ten maize samples examined contained de-
102
HUDSON El AL.
210
250 300 350 Wavelength ( n m ?
I
I
403
Reference
spectrum
AFB 1 8.1
4
2
e
6
Time
I0
12
14
16
19
(mfn.)
Fig. 1. HPLC prohle of the aflatoxins found in a groundnut sauce sample prepared using a monoclonal antibody immunoaffinity column Details of the HPLC method are described in the Materials and Methods section, and the identity of the indlvldual aflatoxins was assigned by co-chromatography with authentic standards HPLC column temperature was 35'C
I
RFBl
AT 21.8
MIN
3 00 350 Wavelength ( n m )
5
10 Time
A
4 00
15 (min.)
11
B
25 1 Refer e n ce
20
4Ql s p e c t r urn
25
Fig. 2. HPLC prohle of the aflatoxins found in a maize sample prepared using a monoclonal antibody immunoafhnity column Details of the HPLC method are described in the Materials and Methods section, and the identity of the indhtidualaflatoxins was assigned by co-chromatography with authentic standards HPLC column temperature was 35'C
tectable levels of aflatoxins; mean 9.7 ppb, range 2 to 35 ppb. Figure 2 shows the HPLC profile of the most heavily contaminated sample of maize, revealing the presence of only AFB, and AFB2. Nine samples of Sanyo millet were examined; all contained aflatoxins, mean 9.8 ppb, range 1 to 27 ppb. One of these was examined by HPLC and only AFB, was detected. Eight samples of sorghum were examined, two were positive for aflatoxins with levels of 2 ppb and 16 ppb. Rice is the major dietary staple in this community; 20 samples were analysed and 14 contained detectable levels of aflatoxins, mean 7.9 ppb, range 2 to 19 ppb. Finally, three leaf-sauces were analysed; they contained aflatoxins at levels of 21 ppb, 26 ppb, and 34 ppb. It is clear that aflatoxin contamination is widespread in the diet and is not restricted to ground-
nuts and maize. The results of these analyses identified which foods might be important in a study of aflatoxin intakes. In a separate study [Wild et al., 19921, 20 adults were matched, as far as possible, in pairs for age and sex, with one member of each pair positive and the other negative for evidence of exposure to the hepatitis B virus. The diets of the subjects consisted predominantly of boiled rice eaten with either a leaf sauce or a groundnut sauce. A total of 87 leaf sauces, 22 flour sauces, and 47 groundnut sauces was analysed; 64%, 59%, and 87%, respectively, contained detectable aflatoxins. The highest level of aflatoxins found was 774 ppb in a leaf sauce that contained raw groundnuts and Figure 3 shows the HPLC profile for this sample.
103
AFIATOXINS IN HUMAN FOODS
250 300 3S0 W a v e length t n m )
210 4 03 Reference spectrum
J
3
a
4:
RFGl 10 Time
AFB2
20
30
(min.1
Fig. 3. HPLC profile of the aflatoxins found in a leaf sauce containing groundnuts sample prepared using a monoclonal anttbOdy immunoaffinity column Details of the HPLC method are described in the Materials and Methods section, and the identity of the individual aflatoxins was assigned by co-chromatography with authentic standards HPLC column temperature was 55'C
The vast majority of the toxin present is AFB,, which is in contrast to the presence of all four aflatoxins found in the groundnut sauces, suggesting that the groundnuts were not the source of this contamination. Of the positive groundnut sauce samples, six contained aflatoxins at levels in excess of 100 ppb. HPLC analysis of these samples yielded chromatograms (not shown) that were very similar to that shown in Figure 1, with up to 35% of the total aflatoxin content as AFG,. Thirty samples of other sauces were examined; 22 contained no detectable aflatoxins and only one of the eight positive samples had a level as high as 20 ppb. None of the 15 samples of boiled rice examined from this study contained detectable aflatoxin. Determination of Aflatoxin Intakes
The analytical values for aflatoxins in the samples of cooked foods were used with records of the amounts of food eaten to estimate the intakes of aflatoxins from those sources as described in Materials and Methods. Figure 4 shows a box and whisker plot of the intake values for individuals and on a per kg basis. DISCUSSION
The development of methods to assess human exposure to environmental carcinogens requires techniques that are sensitive, specific, and amenable to large numbers of samples. For example, the chromatographic methods [Trucksess et al., 1984; Groopman and Donahue, 19881have the power to resolve mixtures of aflatoxins into the individual components, providing that the extraction procedure does not introduce high levels of interfering compounds. The antibody-based methods [e.g., Chu et al., 19871are more specific and selective than straightforward chromatography but immunoassays are
compromised by the intrinsic cross-reactivity of an antibody for multiple aflatoxins. The application of immunoaffinity chromatography to the analysis of environmental carcinogens has increased steadily and its use in aflatoxin analysis [Groopman et al., 1992a,b]. benzo(a)pyrene metabolite detection [Garner et al., 19881, oxidative DNA damage products [Degan et al., 19911; and heterocyclic amine measurements [Vanderlaan et al., 1991; Turesky et al., 19891 has been reported. The major advantage of the approach is the single-step removal of interfering substances from complex mixtures as diverse as food, serum, milk, and urine. The chromatography is non-destructive, allowing subsequent analysis by mass spectroscopy, HPLC or other specific methods. We have taken advantage of the inherent strengths of both antibody selectivity and chromatographic separation to develop an immunoaffinity chromatography/HPLC procedure for the isoiation and measurement of aflatoxins in food and biofluids. Immunoaffinity chromatography provides a means to isolate authentic AF molecules from the complex milieu of food samples. This greatly facilitates subsequent qualitative and quantitative measurements. The major problems encountered in field studies aimed at measuring individual intakes of aflatoxins are not associated with these sensitive and specific assays but with extrapolating from the aflatoxin concentration in foods to the individual intakes. Measurement of the habitual food intake of a free-living individual is notoriously difficult [Bingham, 1987, 199I], and there is a vast and largely negative literature devoted to the subject [Borelli, 19901. Two particular problems are relevant to the studies referred to here. The food sampling process can be highly variable due to the heterogeneity of aflatoxin contamination [Campbell et al., 19861, and we were constrained to take small samples
104
HUDSON ET AL.
looooo W
z
f 10000
w
z X
a
0
A
II i
1000
100
contamination below the detection threshold of the analytical method could contribute significantly to total intakes of AF. It is clear that the development of methods using molecular dosimetry markers such as DNA adducts and protein adducts [Groopman et al., 19911will be crucial for the proper interpretation of field studies of human exposure to carcinogens.
25 I 0
P 4 2
r
0
TOTAL INTAKE
0
INTAKE
/
KG
Fig. 4. Total dietary intake of aflatoxins during 6 to 7 consecutwe days The box and whisker plot represents a description of the data showing the 10th and 90th centiles (ends of the whiskers), the ‘25th and 75th centiles (ends of the box), and the 50th centile (line within the box) Outlier points are shown as indlvldual values A Total intake B Intake per kg body weight
from large amounts of cooked food. Small samples were used because the removal of large samples unduly influences a subject’s food intake. I t was necessary to take samples from larger quantities of food than are commonly encountered in studies of food intake. This is because it is normal in the study community for people to eat in groups, taking food a handful at a time, directly from a large communal bowl. This leads directly to the second problem; how to determine the distribution of a bowl of food between the people who eat from it. The stratagem used here is crude but unlikely to lead to serious errors; the amounts of food eaten are probably underestimated for older children and overestimated for the younger, smaller children but this will tend to cancel out somewhat when children are considered as a group. Similar calculations have been made for energy intakes in this community and the results are compatible with independent values for energy expenditure (Hudson, unpublished results). The results for A F intake are correlated with, but absolute values do not agree with, those obtained by extrapolation from the values for serum albumin-AF adduct in peripheral blood. The levels of AFalbumin adduct reflect long-term exposure to A F and levels of A F metabolites in urine reflect recent exposure. The levels found in the study referred to here [Wild et al., 19921suggest that the levels determined in the food samples are too low, and there are several possible reasons for this. There are the problems associated with sampling food and estimating food intakes discussed above, and the fact that food eaten at any time other than the two main cooked meals was not monitored. Further, none of the rice samples analysed in this study contained detectable levels of AF, yet 70% of the samples analysed in a previous study in this village [Hudson and Day, 19891 were shown to be contaminated with AF. Since considerable quantities of boiled rice are eaten, even a level of
ACKNOWLEDGMENTS
This work was supported by grant CA 48409 from the USPHS and by the Department of Cooperation and Development of the Ministry of Foreign Affairs of Italy. Salary support for A 2 was provided by grant ES 07141 from the NIEHS and for JDG by Research Career Development Award CA 01517. REFERENCES Bingham SA ( 1 987): The dietary assessment of individuals: Methods, accuracy, new tcchniqucs and rccommcndations. Nutr Abstr Rev [A] 57:705-742. Bingham SA (1991): Limitations of thc various methods for collecting dictary intake data. Ann Nutr Mctab 35:l 17-127. Borrclli R (1990): Collection of food intake data: A reappraisal of critcria for judging the mcthods. Br J N u t r 63:411-417. Busby WF. Wogan GN (1985): Aflatoxins. In Scarle CE (ed): “Chernical Carcinogens.” 2nd ed.Washington, DC: American Chemical Socicty, pp 945- 1 136. Campbcll AA, Whitakcr TB, Pohland AE, Dickcns JW. Park DL (1986): Sampling, sample prcparalion, and sampling plans for foodstufs for mycotoxin analysis. Purc Appl Chcm S8:305-314. CAST Report (1989): “Mycotoxins: Economic and Health Risks.” Council for Agricultural Science and Technology. Task Force Rcport 116. pp 1-91. Chu FS. Fan TSL, Zhang GS. Xu YC. Fdust S, McMahon PL (1987): Improved enzyme-linked immunoassay for aflatoxin B I in agricultural commodities. J Assoc Off Anal Chem 70:854-857. Dcgan P. Shigenagd MK. Park E-M, Alperin PE, Ames BN (1991): Immunoaffinity isolation of urinary 8-hydroxy-2‘-deoxyguanosine and 8-hydroxyguanine and quantitation of 8-hydroxy-2’-deoxyguanosine in DNA by polyclonal antibodies. Carcinogenesis I 2 865-871. Garner RC, Dvorackova I, Turi F (1988): Immunoassay procedures to detect exposure to aflatoxin Bi and benzo(a)pyrene in animals and man at the DNA level. Int Arch Occup Environ Health 601451so. Groopman JD, Donahue KF (1988): Aflatoxins, a human carcinogen: Analyzing foods and biological samples using monoclonal antibody affinity chromatography. J Assoc Off Anal Chem 71:861-867. Groopman JD, Trudel LJ, Donahue PR, Marshak-Rothstein A, Wogan G N (1984): High affinity monoclonal antibodies for aflatoxins and their application to solid phase irnmunoassays. Proc Natl Acad Sci USA 81:7728-7731. Groopman JD, Sabbioni G, Wild CP (1991): Molecular dosimetry of aflatoxin exposures. In Groopman, JD, Skipper P (eds): “Molecular Dosimetry of Human Cancer: Epidemiological, AndlytiCal and Social Considerations,” Boca Raton, FL: CRC Press, pp 302-324. Groopman JD, Hasler J, Trudel LJ, Pikul A, Donahue PR, Wogan GN (1992a): Molecular dosimetry in rat urine of aflatoxin-N ’-guanine and other aflatoxin metabolites by multiple monoclonal antibody affinity chromatography and HPLC. Cancer Res 52:267-274. Groopman JD, Zhu J, Chen J-S, Wogan GN (1992b): Molecular dosimetry of urinary aflatoxin DNA adducts in people living in
AFLATOXINS IN HUMAN FOODS
Guangxi Autonomous Region, People's Republic of China. Cancer Res 52:45-52. Groopman JD, Hall A, Whittle H, Hudson GJ, Wogan GN, Montesano R, Wild CP (1992~):Molecular dosimetry of aflatoxin-N -guanine in human urine obtained in The Gambia, West Africa. Cancer Epidemiol Biomarkers Prevention I :22 1-227. Groopman JD, Egner P, Love-Hunt A, DeMatos P, Kensler TW (1992d): Molecular dosimetry of aflatoxin DNA and serum albumin adducts in chemoprotection studies using I ,2-dithiole-3-thione. Carcinogenesis 13:101-106. Hudson GJ, Day KC (1989): Water content of the rural Gambian diet. Nutr Rep Int 40:335-339. Trucksess MW, Brumley WC, Nesheim S (1984): Rapid quantitation and confirmation of aflatoxins in corn and peanut butter, using a disposable silica gel column, thin-layer chromatography, and gas chromatography/mass spectrometry. J Assoc Off Anal Chem 67: 973-975. Trucksess MW, Stack ME. Nesheim S, Park DL, Pohland AE (1989): Enzyme-linked immunosorbent assay ofaflatoxins B, ,B,, and G , in
'-
105
corn, cottonseed, peanuts, peanut butter and poultry feed. J Assoc Off Anal Chem 72957-962. Turesky RJ. Forster CM, Aeschbacher HU, Wurzner HP, Skipper PL, Trudel LJ, Tannenbaum SR (1989): Purification of the food-borne carcinogens 2-amino-3-methylimidazo [4,5-flquinoline and 2-amino3.8-dimethylimidazo [4,5-flquinoxaline in heated meat products by immunoaffinity chromatography. Carcinogenesis 10: 151-1 56. Vanderlaan M, Alexander J, Thomas C, Djanegra T, et al. (1991): lmmunochemical detection of rodent hepatic and urinary metabolites of cooking-induced food mutagens. Carcinogenesis 12:349354. Wild CP, Jiang YZ,Allen SJ, Jansen LAM, Hall AJ, Montesano R (1990): Aflatoxin-albumin adducts in human sera from different regions of the world. Carcinogenesis I1:2271-2274. Wild CP, Hudson GJ, Sabbioni G, Chapot B, Hall A, Wogan GN, Whittle H, Montesano R, Groopman J D (1992): Dietary intake of aflatoxins and the level of albumin-bound aflatoxin in peripheral blood in The Gambia. West Africa. Cancer Epidemiol Biomarkers Prevention 1:229-234.
