441

Vol.

7. 441-447,

May

1998

Detectable United

Cancer

Levels of Serum Aflatoxin -Albumin Kingdom Population: Implications for Exposure in the United Kingdom’

Paul C. Turner,2 Karen H. Dingley, Serina Russell, and Cohn R. Garner The Jack Biology,

Birch Unit for Environmental University of York, Heslington,

Jonathan

Carcinogenesis, York YOl

5DD,

Coxhead,

Department United

of Kingdom

Abstract This study aimed to estimate aflatoxin B1 (AFB1) exposure in the United Kingdom population by measuring levels of serum AFB1-albumin (alb), using immunoassay and high-performance liquid chromatography (HPLC) with fluorescence detection. A self-questionnaire on dietary habits from 104 volunteers (47 men and 57 women) in York was completed, and blood samples were collected. Serum alb was extracted, and AFB1-lysine (lys), the digest product of AFB1-alb, was isolated and measured. A sensitive ELISA (detection limit, 1.4 pg of AFB1.lys) was developed. A good correlation was found between calibration of ELISA results and scintillation counting, for rats dosed with [3H]AFBJ (r 0.972; P < 0.001). This ELISA was subsequently used to analyze human serum alb. For United Kingdom human sera, the mean adduct levels were 29.3 ± 14.8 pg AFB1-lys equivalents (eq) mg albumin (males) and 26.9 ± 14.4 pg AFB1-lys eq/mg nib (females). Confirmation of the ELISA data was sought using reversed-phase HPLC with fluorescence detection. HPLC chromatograms of digested York serum alb were compared to digested serum alb for humans from Qidong County, People’s Republic of China, and from AFB1dosed rats. These all gave similar HPLC profiles. Each sample contained fluorescent material that coeluted with and just before the AFB1.lys standard. Fluorescent fractions were found to be inhibitory in a separate antiAFB1-lys ELISA, indicating that these earlier fluorescent peaks contained AFB1 residues. Our results suggest that measurable internal AFB1 exposure may be occurring in some United Kingdom individuals, albeit at lower levels than those seen for areas with high AFB1 exposure. The source of this exposure may reflect the known difficulties in accurately

Received 1/27/97; revised 2/2/98; accepted 2/I 1/98. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. I The funding for this research was provided by the Ministry of Agriculture, Fisheries and Food (United Kingdom). 2 To whom requests for reprints should be addressed. Present address: Molecular Epidemiology Unit, 3rd floor, Algenon Firth Building. University of Leeds, Leeds LS2 9JT. United Kingdom.

Epidemiology,

Biomarkers

Adducts

& Prevention

441

in the

monitoring regulated imported foodstuffs and/or the lack of regulations on other potentially contaminated imports. However, no positive correlations were found between our AFB1-lys measurements and any dietary questionnaire information. Animal studies, as well as human studies, have been important in developing exposure and internal adduct relationships in humans. Based on this literature, our AFB1-alb data indicate a mean daily exposure of 3 .tg of AFB1 and a mean internal dose in liver DNA of 5.9 adducts/107 nucleotides. We believe this may be an overestimate of the AFB1 exposure level in the United Kingdom, and further studies are needed to accurately relate external dose and internal AFB1 biomarkers in humans.

Introduction AFB1,3 a human class 1 carcinogen (I), is a potential contaminant of a wide range of human foodstuffs (2). Current United Kingdom regulations (3) regarding the mandatory analysis of AFB1 in imported foodstuffs are restricted to nuts, nut products, dried figs, and dried fig products and do not include all potentially contaminated foodstuffs (4). Bulk food surveillance is not an effective way to control human exposure to the aflatoxins because contamination can occur in isolated food pockets (5). Accurate and sensitive biological markers (biomarkers) of carcinogen exposure are now used to estimate cancer risks (6). Biomarkers can be metabolites, DNA adducts, or protein adducts (7, 8). Knowledge of the metabolic fate of has led to the use of several biomarkers in blood, urine, feces, and tissues that can be used to monitor exposure. I is among the few compounds for which the relationship between biomarkers and exposure has been extensively studied (9). Once ingested. I is metabolized by the cytochrome P-450 system, primarily in the liver ( I 0). The resulting reactive AFB1-8,9-epoxide binds to DNA, forming AFB1-guanine adducts (11, 12). The persistent AFB1-formamidopyrimidine DNA adduct may be a useful biomarker of I exposure but is limited by availability of tissue (13). A proportion of AFBguanine adducts released from DNA or RNA by repair, turnover, or chemical hydrolysis are excreted in the urine (14). Metabolism of I can also lead to metabolites such as AFB1, which is excreted in the urine (15). Urinalysis can be used for acute exposure studies but does not reflect chronic exposure (16). AFB1-8,9-epoxide formation also leads to AFB1-protein binding via AFB1-8,9-dihyrodiol (17). The major AFB1-ad-

The abbreviations used are: AFB,, aflatoxin albumin: lys. lysine: HPLC. high-performance Tween 20: eq. equivalent(s): IAC. immunoaflinity 3

B1: AFM1. aflatoxin liquid chromatography: chromatography.

M1:

alb, Tw.

442

Serum

Aflatosdn

B1-Albumin

in the

United

Kingdom

ducted protein found in the blood of AFB 1-dosed rats is alb; this binding occurs specifically at lys residues (17). It has been shown that formation of , -alb occurs in a dose-dependent manner. with an accumulation to steady-state levels that correlate with AFB1-DNA adduct levels in internal organs. Hence, the measurement of AFB 1-alb adducts in the blood may be used as a surrogate dose monitor for target organ DNA adducts (18). Rat alb has been shown to bind 1-3% of a dose of AFB1 (19-22). Humans are reported to form I adducts in alb and DNA in a manner and rate similar to those observed in the rat (23). Therefore, because human blood alb has a half-life of approximately 20 days, AFB1-alb adduct detection should provide a useful biomarker of cumulative AFB1 exposure (20). Also, it is less invasive to collect blood samples for analysis than to surgically remove tissue samples from individuals. I biomarkers have been extensively validated using animal studies. For example, the dose-dependent excretion of I -guanine and in urine has been shown in rats acutely dosed with (24), and the usefulness of DNA and protein adducts in chronic dosing studies has been demonstrated (20, 25). These biomarkers have now been examined in humans and appear to be useful in biomonitoring studies (14, 19, 26). It is widely considered that AFB1-DNA and AFB1-alb are among the most suitable biomarkers for the examination of human population exposure. They provide a measure of the amount of AFBJ activated to AFB1-8,9-epoxide that is available to bind to macromolecular targets with deleterious consequences (27). Previous preliminary studies from our laboratory reported measurable levels of AFB1-DNA adducts in autopsy tissue collected from United Kingdom individuals (28). In this study, we sought verification of this reported I exposure by measuring human serum AFB1-alb levels. To do this, we have established a sensitive ELISA method for monitoring human exposure. Confirmation of our ELISA results was sought using HPLC with fluorescence detection and ELISA of individual HPLC fractions. ELISA and HPLC data from AFB -dosed rats and from humans in known -contaminated areas of the world, were compared to those from the York study. Materials

and

Methods

H2O2 (30%, v/v), PBS tables, tetramethlybenzidine, Tw, NaH,PO4 and Na2HPO4 (SigmaUltra for HPLC) were supplied by Sigma Chemical Co. Poole, Dorset, United Kingdom. All solvents were HPLC grade, supplied by Aldrich Chemical Co. (Gillingham, Dorset, United Kingdom.) Other reagents (Analar grade) were supplied by Merck Ltd. (Lutterworth, Leicester, United Kingdom). AFB1 Rat Dosing Study. Male Wistar rats (Harlan Olac Ltd., Bicester, Oxon, United Kingdom), weighing 200-250 g, were housed in wire mesh floor cages (two per cage) and fed expanded SQC diet and water ad libitum. Two animals per dose were i.p. injected with 0. 10, 50, or 200 pg/kg body weight [3HJAFB (28 mCi/mmol; Moravek Chemicals, Brea, CA). Twenty-four h after dosing, bleeds were obtained from each animal. Bleeds were collected into serum vacutainers (Sarstedt Ltd., Leicester, United Kingdom), and serum was allowed to separate overnight at 4#{176}C. Collection of Human Blood Samples. Blood samples were collected from 104 individuals aged 18-65 years, following advertising for volunteers within the University of York. Informed consent was obtained regarding the nature and purpose of the study. All volunteers agreed that they would have no right of access to any individual datum obtained from the study

and that blood samples would be used for no other purpose. Questionnaires were completed by all persons taking part in the study concerning health status, recent travel outside of Europe, and dietary habits [amounts and types of nuts/nut products, dried fruit (particularly figs and dates) spices, cereals, and beers], including vegan or vegetarian status. Volunteers were healthy and of mixed ethnicity (although they were predominantly Caucasian), and there were no volunteers with a known high risk of liver cancer. Serum was isolated from whole blood as described previously and stored at -70#{176}Cuntil analysis. In addition, several alb samples from Kenya (kindly donated by Prof. C. Wild, University of Leeds, Leeds, United Kingdom) and Qidong County, People’s Republic of China (kindly donated by Prof. Zongtang Sun, Cancer Institute, Chinese Academy of Medical Science, Beijing, People’s Republic of China) were used. Extraction and Purification of AFB1 Serum nib Adducts. alb was extracted from rat and human scm according to Chapot and Wild (29), using HPLC-grade solvents. alb was quantified using the Coomassie Protein Kit (Pierce and Warriner UK Ltd., Chester, United Kingdom), according to the manufacturer’s instructions. Methods for the isolation of AFB residues from alb were modified from Chapot and Wild (29). Briefly, 2 mg of alb were digested with 610 g of pronase (Calbiochem-Novabiochem, Ltd., Nottingham, United Kingdom) in 800 l of 0.02 M phosphate buffer for 16 h at 37#{176}C with shaking (250 rpm). Samples were stored on ice for 1 h. Pronase and undigested alb were precipitated with 2 volumes of acetone (-20#{176}C) on ice for 2 h. Samples were centrifuged (2000 X g, 15 mm, 0#{176}C), the supernatants were retained, and the pellet was washed with 1 ml of cold acetone. The pellet was centrifuged (as above), and the two supernatants were combined. These were dried in vacuo and dissolved in 10 ml of water. Sep-Pak cartridges (Waters Ltd., Watford, United Kingdom) were used to purify digest material according to manufacturer’ S instructions, except that the numbers of passages of test material were altered to investigate AFB 1-lys recovery. AFB ,-lys was synthesized according to the method of Chapot and Wild (29). AFB1-lys standards (0, 2.86, 1 1.4, and 45.7 pg) were purified by one passage and repeated passages (3). Percentage recoveries for the two processes were then determined using ELISA. Samples were dried in vacuo and dissolved in 0.5 ml of PBS for quantitation by ELISA. AFB1-lys ELISA. Competitive ELISAs were carried out using 7.13-1830 pg/mI AFB1-lys standard and a monoclonal antiAFB,-BSA antibody, 6e9 (Biocode Ltd., York, United Kingdom). All incubations were carried out in the dark for 2 h, and between each antibody incubation, the plates were washed five times with PBS containing 0.1% 1w. Negative controls were also included to verify that nonspecific binding of the various antibodies did not confound the assay. Immulon 4 microtiter plates (Dynatech Laboratories Ltd., Billinghurst, United Kingdom) were coated with 2.5 ng of AFB1-ovalbumin conjugate (Biocode Ltd, York, United Kingdom) in 0.05 M NaH2CO3Na2HCO3 buffer, pH 9.6 (4 h in the dark). HPLC was carried out using an analytical Dynamax-300A (5 jim) 25 cm X 4.6 mm reversed-phase C18 HPLC column (Waters Ltd., Watford, United Kingdom). Solvents were filtered using 0.2 m nylon membrane filters (Whatman Instruments Ltd., Maidstone, Kent, United Kingdom) and degassed with helium. Solvent flow rate was 0.75 mI/mm, with a 40100% (v/v) methanol in 0.02 M sodium phosphate buffer (pH 7.2) 20-mm linear gradient. Fluorescence detection parameters were set to excitation at 399 nm and emission at 457 nm.

Cancer

Table

I

Amount

of AFB

-lys

obtained

Scintillation

from

counting

and

alb digests

as measured

by

ELISA

Male Wistar rats (2 per dose) were dosed with a range of [3H]AFB, (200-0 mg/kg). Sera were isolated, and alb was extracted. Scintillation counting and ELISA analysis were carried out. These data showed a good statistical correlation (r 0.972; P < 0.001) between the methods. Dose (.sg/kg)

ND,

ELISA injected jection

equivalent counting

200

1003 ± 89.8

So

439 ± 58.0

10

132

0 “

I -lys

Scintillation

± 20.5 0

pg/mg

alb ELISA 1051 ± 128

326

± 121

59 ± 6.4 ND”

not detected.

inhibitory material (750 onto the HPLC column, with AFB1-lys standard.

pg) from the test samples both with and without

was coin-

ELISA Analysis of HPLC Fractions. For each HPLC run, 0.5-mm fractions were collected. Each fraction was dried in vacuo and then reconstituted in 200 d of PBS for ELISA. Ml07, a mouse anti-aflatoxin-BSA monoclonal antibody (an in-house monoclonal antibody produced during the study) was used as an alternative to 6e9 in the ELISA. Assay conditions were identical, except M107 was diluted 1:2 X l0 in PBS, Tw, and FCS. Results 6e9 Anti-AFB1-lys ELISA. The coefficient of variation of the AFB 1-lys competitive ELISA within runs was <0.1, and between runs, it was <0.2, indicating that the assay was reproducible. The detection limit was - 1 .4 pg of AFB 1-lys. Investigation of the recovery of AFB 1-lys from Sep-Pak cartridges, used to purify the rat and human alb digests, revealed that losses of adduct were reduced by passage of the samples three times through the cartridge before elution. By this method, recoveries were increased from 61.1 ± 56.2 to 85.3 ± 9.2%. AFB1-lys was used to generate the standard curve in ELISA, and results are expressed as pg AFB 1-lys eq/mg alb. For the ELISA of all digested alb samples, the sensitivity of the assay was restricted to the linear portion of standard curves. Positive results were only designated for samples giving greater than 20% and less than 80% inhibition in competitive ELISA. Comparison of AFB1.lys Adduct Levels, from [3H]AFB1. dosed Rats, by Competitive ELISA and Scintillation Counting. The amounts of AFB 1-lys, obtained from alb digests, for rats dosed with J at several concentrations, were measured by ELISA and scintillation counting (Table 1). There was a good statistical correlation between the ELISA and the scmtillation counting data (r = 0.972; P < 0.001). ELISA Analysis of Human Serum nib from York, Kenya, and Qidong County, People’s Republic of China. The amount of AFB ,-lys eq pg/mg of albumin from human blood samples for individuals from York, Kenya, and Qidong County were determined (Fig. 1). All samples (except the Kenyan samples) were analyzed in triplicate on at least two occasions. A cross-section of coded samples from Kenya were supplied by Prof. C. Wild (University of Leeds, Leeds, United Kingdom) to validate our methodology. Analysis of the Kenyan alb samples was also carried out independently by Wild’s group (IARC, Lyon, France). Comparison of the two sets of data indicated a good statistical correlation (r = 0.952). The York blood samples generally had

Epidemiology,

Biomarkers

& Prevention

low levels of AFB1-alb, -45% in the range of 15-45 pg AFB1-lys eq/mg albumin. The Qidong County and York samples showed some overlap; however, only -4% of the York samples had >55 pg of AFB 1-lys eq/mg albumin, compared to -45% of the Qidong County samples. The mean values from York were 29.3 ± 14.8 pg AFB1-lys eq/mg alb (males) and 26.9 ± 14.4 pg AFB1-lys eq/mg alb (females). For each human sample, a coefficient of variation was determined. The coefficient of variation for the human samples in competitive ELISA were <0.1 within assays and a mean of 0.53 between assays. Verification of AFB1-alb Digest ELISA Data by HPLC Fluorescence Analysis. Standard AFB 1-lys had a retention time of approximately 8 mm on the gradient system used and a fluorescence detection limit of 22.9 pg. For each test sample, 750 pg AFB1-lys eq were injected. The injected test samples produced chromatograms with several fluorescent peaks followed by a minor peak (approximately 5% of the total integrated area) at 8 mm. This minor peak coeluted with standard AFB1-lys. HPLC fluorescence analysis of alb digests from dosed rats and human samples revealed similar profiles, with a large number of fluorescent peaks eluting prior to the retention time of AFB1-lys. To determine if any of these peaks contained I residues, 30-s HPLC fractions were collected, dried, and measured by ELISA. These samples had been obtained by immunoaffinity purification of alb digests using immobilized 6e9 antibody; therefore, an alternative antibody was required for the ELISA of fractions. This would increase the probability that ELISA inhibition was due to FB1 residues. The antibody (mouse monoclonal antibody, M 107) used for this purpose was raised in-house against AFB1-ovalbumin. Ml07 had a slightly higher IC50 with AFB1-lys, AFB1, aflatoxin B2, aflatoxin G, aflatoxin G2 and AFM1 (data not shown) and, thus, produced a less sensitive assay. For -lys, the limit of detection in the M107 ELISA was -3.2 pg. HPLC fluorescence chromatograms and data obtained from ELISA of the HPLC fractions were overlaid for each test sample (Fig. 2). In each case, the data overlapped. The total amounts of fluorescent material obtained (area under the chromatogram) were calculated by comparison to AFB 1-lys standards for each tesL The total AFB 1-lys eq from the ELISA of HPLC fractions for each test were also calculated. These values were expressed as a percentage of AFB1-lys eq. measured by 6e9 ELISA, prior to HPLC (Table 2). For HPLC fluorescence detection, just over 70% of the amount of material obtained by the 6e9 ELISA was detected for all test samples. For ELISA of HPLC fractions, approximately 50% of the ELISA injected material was detected. Assumptions used for this quantitation are discussed below.

Discussion Regulations regarding the mandatory analysis of I in foodstuffs imported into the United Kingdom are restricted to nuts, nut products, dried figs, and dried fig products. However, other foodstuffs in the United Kingdom may be potentially contammated, e.g. , spices, breakfast cereals, corn and cornmeal, dairy products, and other processed foods (30-33). An earlier study from our laboratory reported the detection of AFB 1-DNA adducts in humans from the United Kingdom (4). The adduct levels reported in that study were not confirmed in subsequent studies in our laboratory. Problems in the development of a highly specific and sensitive AFB1 DNA immunoassay seemed to restrict the monitoring of aflatoxin exposure via DNA adducts. Nonspecific binding of antibody and control DNA occurred in several of the assays developed. In addition, postmortem samples from elderly persons were our source of material.

443

444

Serum

Aflatoxin

B,.Albumin

in the

United

.

Kingdom

iii:i:-iii

--.

Fig. 1. Measurement of human serum -alb in samples from York, Qidong County. and Kenya by ELISA. alb was isolated and digested by pronase, followed by cxtraction of aflatoxin-containing resdues. These were then measured by ELISA in triplicate on at least two occasions.

0 <5

to

$toZ5

2$to35

36to45

MB1-

46io55

IUVort

DG*ie

561065

65to35

S5ioO

150

lbumln

(eq.)pghybg

SKenye

Eating habits. including quantity and variety, of these persons would not reflect those in the general population. Therefore, alternative methods to measure aflatoxin exposure were sought. Monitoring human serum alb samples provides a useful, noninvasive measure of AFB1 exposure because AFB1-alb correlates with liver AFB,-DNA adduct levels (20), and AFB1 does not appreciably bind to any other macromolecule in the blood ( I 7). To estimate internal exposure to , we have measured AFB1-alb adduct levels from volunteer subjects primarily by ELISA. With the use of monoclonal antibodies, ELISA provides one of the most sensitive methods for the detection small molecules. A rat dosing study was carried out using I ‘ Serum alb was extracted and digested, and AFB1 residues isolated. By comparing scintillation counting data with that from the ELISA, the reliability of the assay was established (Table 1 ). We confirmed our ELISA results for York human serum alb using a physicochemical method and by comparison to serum alb from AFB1-dosed rats and human serum alb samples from areas where AFB1 exposure is known to occur. A highly sensitive AFB,-lys ELISA (detection limit, -1.4 pg) was developed. The good statistical correlation (r 0.972; P < 0.001) between ELISA and scintillation counting for AFB1-dosed rat, provides strong evidence that our ELISA methodology was suitable for measuring AFB -alb adduct 1evels. When comparing the analysis of a set of Kenyan samples by two separate groups (IARC and York), similar levels of adducts were observed (r = 0.952). Although the lower adducted samples were not in complete agreement (our results were generally slightly higher). this may be explained by the improved recovery ofAFB1-lys standard observed at low levels during our isolation procedure. ELISA of human serum samples from 104 individuals indicated that 95% contained AFB1-alb, albeit at lower levels than those from high-risk areas. The mean values for York samples were 29.3 ± 14.8 pg AFB1-lys eq/mg albumin (males) and 26.9 ± 14.4 pg AFB1-lys eq/mg albumin (females). Although no negative control serum was used for human studies, the rat study indicated that at no AFB1 dose were no adducts detectable by 6e9 competitive ELISA. To confirm our ELISA data. HPLC analysis was carried out on samples of digested human alb. AFB,-lys standard coeluted with the minor fluorescent peak at approximately 8 mm, indicating the presence ofAFB,-lys in the digested serum albumin. In addition, fluorescent material eluted from the column, forming several peaks, immediately prior to the AFB1-lys

standard. The profiles of these peaks were similar for AFB,dosed rat alb digests and human (York and Qidong County) alb digests. To determine if all fluorescent peaks from the HPLC contained I residues, HPLC fractions were collected every 30 5, and ELISA of fractions was carried out. The use of the same antibody in both ELISA of HPLC fractions and IAC may confound results. For this reason, it was prudent to use a distinct antibody for these two processes. The ELISA of HPLC fractions indicated that the fluorescent peaks contained AFB1 residues. This was not unexpected because incomplete digestion of I -alb, generating several fluorescent peaks that elute earher than AFB1-lys from reversed-phase HPLC, has been reported previously (34). AFB1-lys standards and AFB1-lys coinjected with alb digest eluted at the time indicated in Fig. 2. It was important to have quantitative comparisons of the data for the 6e9 ELISA, the HPLC, and the ELISA of HPLC fractions. For each test sample, 750 pg AFB1-lys eq (as determined by 6e9 ELISA) was IAC-purified and injected onto HPLC. The amounts of AFB 1-lys eq determined by HPLC fluorescence and ELISA of HPLC fractions were -70 and 50%, respectively, of the amount of material obtained by 6e9 ELISA. There may be some loss of material during the IAC of AFB1-alb digests carried out prior to injection on HPLC. Loss of some AFB1 residues during IAC purification of AFB1-alb digests has been reported previously (34, 35). The amounts of adduct detected by both fluorescence and ELISA of HPLC fractions correlate reasonably well with the original 6e9 ELISA data. Although incomplete digestion products containing residues will be fluorescent and inhibitory in ELISA, the exact nature of these adducts has not been established. Therefore, the absolute amount of non-AFB 1-lys fluorescence and ELISA inhibition remains uncertain. has been shown to bind to albumin specifically at lys residues ( I 7). Therefore, incomplete digestion products of AFB -alb will be short peptides linked to AFB1 via lysine. The fluorescent and ELISA properties of these residues are likely to be similar to those of AFB,-lys. The major finding from these studies is the implicated levels of internal AFB1 exposure and the estimated dose in target organ DNA. There appears to be a clear linear relationship between AFB, intake and AFB1-alb adducts in rats (20) and in humans ( I 9, 36). The average exposure of I for persons in York based on AFB1-alb levels would be -3 pg/day [based on 3% of ingested I being bound to serum albumin (19), a 30-fold accumulation of -alb (37), and total pe-

Cancer

Epidemiology,

Biomarkers

& Prevention

445

A 450

>

E

U)

a) 0

.o0 350

U) C

C.

a,

a)

250

C.)

C

a,

a) C Ca

20.00

C.) U)

150

a)

>

0

x o

50

U0.00

-50

Time

Fig. 2. Reversed-phase HPLC with fluorescence detection was carried out for digested alb from an AFB,-dosed rat (a), individuals from York (b), and individuals from Qidong County (c). Fractions were collected every 30 s of the HPLC run. Each of these was analyzed using an ELISA distinct from the original 6e9 ELISA. These are overlaid. Fluorescence detection (line graph; left) was measured in mV. The ELISA of HPLC fractions (bar chart; right) was measured in aflatoxin-lys eq (fmol). For ELISA of fractions, the AFB,-lys eq calcu-

I minutes

B > E Cl)

-450 0

u.oo 350

C

a) C.) C

a,

0 U)

250 20.00 150

a,

. C.

Q) a) C .

0

50

U0.00

-50

lated for a blank (phosphate buffer) injection at each time point was subtracted from the value for each test sample. Arrow, AFB,-lys standard coeluted.

10 .00

Time I minutes

C > .

450 woo

E 350

U) C

a)

250

a) 20.00 G)

a) 0

U-

C Cl)

150

C.) U)

C

500 00

-50

ripheral blood albumin = 96 g). On the basis of this level of AFB 1-alb, the expected mean internal dose at liver DNA would be 5.9 adducts/107 nucleotides (20). Our value for AFB1 exposure was lower than those estimated for parts of China (southern Guangxi), where up to 1 20 pg/day exposure has been reported (38), but was relatively high when compared to some African countries with exposure between 0.2 and 12 pg/day. No positive correlation was found between FB1 -alb 1evels and any of the responses from the dietary questionnaire. There was difficulty in obtaining reliable details of relative

amounts and types of foodstuffs consumed by people. Many people simply replied often or sometimes to specific food questions. The sample population used in this study was also fairly small. The relationship between AFB1-alb adduct levels and exposure was recently estimated for humans (23). The value was 1.56 pg AFB1-lys eq/mg albumin per jg AFB1Ikg body weight following acute exposure. This value was based on -25% recovery of AFB, residues and 1% of AFBJ being bound to peripheral blood alb. There are many stages prior

446

Serum

Table

Aflatoxin

2

B1-Albumin

Calculation

of

in the

I

-lys

United

eq/pg

HPLC AFB1-lys eq (750 is also represented

Total

for HPLC

___

fluorescence

AFB1-lys

eq (pg)

fluorescence

and

ELISA

was injected for each estimated material. ELISA

analyzed

of HPLC

test.

fractions

% of ELISA” estimated

Total

material

AFB1-lys eq (pg)

% of ELISAb estimated

material

York

549

73.4

402

53.6

555

74.0

415

55.3

Rat

537

360

48.0

Fluorescence-determined

eq (pg)l

71.6 AFB,-lys

eq (pg)/6e9

ELISA-determined

AFB1-lys

X 100.

(ELISA of HPLC fractions-determined mined AFB1-lys eq (pg)l X 100. ,,

This

analyzed

Qidong County

“

of

fractions

pg: based on original ELISA) as a percentage of the ELISA

HPLC Test sample

Kingdom

AFB

-lys

eq

(pg)/6e9

ELISA-deter-

to immunoassay in which AFB1 adducts may be lost. We found that Sep-Pak purification was greatly improved (mean recovery. 85.3 ± 9.2% of the expected) when AFB1-lys standards were passed through the columns three times, as compared to once. In addition, we assumed 3% binding of AFB1 to serum alb. Taking these factors into consideration, our calculations are in good agreement with the value obtamed in Ref. 23. There are several anomalies with the relationship between AFB1-alb values and AFB1 exposure, e.g., control human serum albumin “obtained from persons unlikely to have been exposed to AFB1” were found to contain measurable levels of adducts, -50 ± 25 pg AFB1-lys eq/mg albumin (34). This would represent a daily exposures between 4 and 6 j.tg of I per day. A relationship between AFB1-alb and AFB1 exposure was recently reported for the F344 rat (23). However, discrepancies exist for studies in rats, e.g. , a study on F344 rats chronically dosed with 200 pg/kg I daily (35). The AFB alb value at steady states should have been approximately 300 pg AFB1-lys eq/mg albumin. The reported values were 400 pmol AFB1-lys eq/mg albumin (or >100,000 pg AF-lys eq/mg albumin). The above examples have used immunoassay to obtain AFB1-alb values. It is possible that nonspecific inhibition in immunoassay may be effecting the results. Nonspecific inhibition was determined using albumin from a Boston donor for a study measuring AFB1-alb values for residents of Guangxi Province, People’s Republic of China (19). This value was subtracted from the values obtained for Guangxi Province samples. Although there was a highly significant association between AFB1 intake and AFB1-alb values, the regression line did not pass through the origin, i.e. , no dose, no adducts. At no dose, the level of AFB1-alb adducts was -45 pg AFB1-lys eq/mg of alb. This would reflect a level of AFB1 consumption of -5 g per day. It was suggested that this may be a nonspecific response in the assay. Considering alb from a Boston donor had been used a negative control, this was surprising. It would appear that, on a population basis, there is a good relationship between consumption and -alb adduct level. However, there seems to be no absolute certainty that measurement of a quantity of adduct relates to a fixed level of AFB1 intake at the individual level. Correlation of estimated dietary intake of aflatoxin and measurement of AFB1-alb was carried out for persons in the People’s Republic of China (36). In this study, I exposure estimates of 100 p.g/day lead to

AFB 1-alb levels, ranging from 0 to 200 pg AFB 1-lys eq/mg albumin. This type of variation makes it difficult to relate I -alb adduct levels to exposure at the individual level. Obtaining quantitative data on dietary intake for individuals over extended periods even when plate food was analyzed may create difficulties with these types of correlation (39). Differences in the quantitative disposition of I metabolites related to ethnicity, age, sex, nutrition, or disease status of individuals may also hinder AFBJ exposure correlations. Levels of AFB1-alb found in populations chronically exposed to are related to the half-life of alb ( 19, 20). Much of the data on albumin turnover in humans are based on limited experiments in humans injected with radiolabeled albumin (40). However, alb turnover is more variable for persons with diseases such as cirrhosis or hepatitis (41). Temporal patterns of aflatoxin-alb adducts in residents of Daxin, Qidong County, People’s Republic of China, have been studied (35). Interestingly, it was found that the albumin levels (June-September) were considerably higher than albumin levels (SeptemberDecember), whereas I -alb levels decreased June-September and increased September-December. 1 exposure may or may not be significant in terms of modulation in serum albumin half-life; however, this issue was not discussed. Conclusions. The results from scintillation counting of albumm from dosed rats were in good agreement with ELISA data in our study. The same ELISA was used to quantify alb adducts in human serum samples from the York area. HPLC was carried out using fluorescence detection for -lys. The profiles of digest material from AFB 1-dosed rat and human York and human Qidong County alb using HPLC were all virtually identical, including a minor peak that coeluted with synthetic AFB 1-lys standard. A separate ELISA of HPLC fractions suggested that these fluorescent peaks were AF1 derived. Quantitative estimations of AFB 1-lys eq were in good agreement by all three methods. Although no extremely high levels of AFB1 exposure were detected, the detection of some -alb adducts in the blood suggests that exposure to has occurred, and there may be an associated health risk. Using AFB1-lys as a biomarker serves to estimate AFB1 intake and the effective dose at the critical target. There is clearly considerable evidence from the studies in rats relating 1 intake, binding at liver DNA, and surrogate dose monitors (blood and urine adducts). For alb adducts, there appears to be a linear relationship between I intake and AFB1-alb adducts. Despite this linear relationship, there are some noteworthy deviations in human studies, as outlined above. It would, thus, appear that absolute predictions from I biomarkers of exposure should be done cautiously. I -alb measurements serve as useful dose monitors in parts of the world where there is high contamination. AFB alb levels are the most accessible and useful means of monitoring the efforts to reduce I exposure (or alter AFB1 metabolism) at critical sites such as liver DNA. Levels of exposure to , extrapolated from our -lys determinations, would suggest that United Kingdom regulations for contamination may need to be investigated. Three pg/day may be an overestimate of the true level of exposure, but until more reliable human data exist, we must rely on the correlations already established for our estimates. Acknowledgments We thank Zongtang of China)

Prof. C. Wild Sun (Chinese for supplying

(University of Leeds, Leeds, United Kingdom) and Prof. Academy of Medical Science, Beijing, People’s Republic alb samples from Kenya and China, respectively.

Cancer

References I. IARC. IARC Monographs on the Evaluation of Carcinogenic mans. Overall Evaluations of Carcinogenicity: An Update of IARC 1 to 42, Suppl. 7. pp. 83-87. Lyon. France: IARC, 1987. 2. IARC Aflatoxins. IARC Monographs of Chemicals to Man: Some Naturally Lyon, France: IARC, 1976. 3. Great Figs and Great

on the Evaluation Occurring Substances,

Statutory

Instruments,

23.

Wild,

and

Montesano,

genesis

C. P., Hasegawa,

179-189,

of Carcinogenic Risk Vol. 10, pp. 5 1-71.

24.

Aflatoxins Food

in Nuts, Nut Safety Act No.

Products, Dried 3236, pp. 1-12.

1992.

animals

and

Groopman,

L., Chutimatawin, adducts:

humans.

J. D., Hasler,

G. N. Molecular

Cancer

by multiple

imrnunoaffinity/high 1992.

25.

R. G.,

Croy.

performance

Essigmann,

preparation Toxicology

and analysis. of Aflatoxins,

In: D. L. pp. 433-450.

Eaton San

6. Perera, F. P. The significance of DNA toting studies. Mutat. Res., 205: 255-269, 7. Farmer, P. B., Neuman, man to substances reacting 251-260,

and J. D. Groopman Diego: Academic Press,

and

protein 1987.

adducts

in human

meassample

(eds.), Inc.,

The 1994.

biomoni-

H. G., and Henschler, D. Estimation of exposure covalently with macromolecules. Arch. Toxicol.,

27.

60:

Garner,

10. Forrester, L. M., Neal, G. E., Judah, D. J., Glancey, M. J. and Wolf, C. R. Evidence for involvement of multiple forms of cytochrome P-450 in aflatoxin B1 metabolism in human liver. Proc. NatI. Acad. Sci. USA. 87: 8306-8310, 1990.

28.

Harrison,

R. C. A high pressure liquid aflatoxin B C in rat liver and

H., Hudson, G. J., Wogan, G. N., dosimetry of aflatoxin-N7-guanine in Africa. Cancer Epidemiol. Biomark.

Sons

17. Sabbioni, characterization viva in rats. 18. Skipper, dosimetry 1990.

G., Skipper, P. L., Buchi, G., and Tannenbaum, of the major serum albumin adduct formed Carcinogenesis (Land.), 8: 819-824, 1987.

J. and MC

P. L., and of chemical

Tannenbaum, carcinogens.

S. R. Protein Carcinogenesis

G. N., and in urine as Perspect..

S. R. Isolation by aflatoxin

adducts in the (Lond.) 11:

and in

molecular 507-5 18,

19. Gan, L. S.. Skipper, P. L., Peng, X., Groopman. J. D., Chen, J., Wogan, G. N., and Tannenbaurn, S. R. Serum albumin adducts in the molecular epidemiology of aflatoxin carcinogenesis: correlation with aflatoxin intake and urinary excretion of aflatoxin M,. Carcinogenesis (Lond.), 9: 1323-1325, 1988. 20. Wild, C. P., Garner, R. C., Montesano, to plasma albumin and liver DNA upon genesis (Lond.), 7: 853-858, 1986.

R., and Tursi, F. Aflatoxin B, binding chronic administration to rats. Carcino-

21. Wong, Z. A., and Hsieh, D. P. H. The comparative metabolism kinetics of aflatoxin B C in the monkey, rat and mouse. Toxicol. AppI. 55: 115-125, 1980.

adduct

formed

of and

and other chromatography Cancer

in rat liver.

Wakhisi,

fluorescence

macromolecular and

Res.,

52:

G. N. Identifi-

in rico

A. K. M., with

R. G. Tardiff

Ltd.,

affinity

Proc.

J., and

Wa-

characteristics

BC in human (Lond.), 4: 1 193-I

iden-

urine collected 195, 1983.

adducts

indicators

of

B. Goldstein

(eds.),

Methods

for

Biota.

297-307.

New

Non-Human

pp.

as

in

York:

1991.

and toxicoPharmacol.,

22. Skipper, P. L., Hutchins, D. H., Turesky, R. J., Sabbioni, G., and Tannenbaum, S. R. Carcinogen binding to serum albumin. Proc. Am. Assoc. Cancer Res., 26: 90, 1985.

M., and Garner

constitute

a cancer

B.,

and

C. P. ELISA

and their

Wild,

application

risk?

R. C. Does

Environ.

aflatoxin

Health

exposure

Perspect.,

30.

Gamer,

R. C., Whattam,

exposure

in the

99: 99-105,

analysis

mide.

Diener,

Ayres

648:

122-150,

with

2,

M. W. Analysis

an

pyridinium

immunoaffinity liquid chromatobromide

perbro-

1993.

biological (eds.),

C. F. Distribution data, including programme.

on Mycotoxins. Wood,

substances

Impact

in foods:

of Toxicology

Thailand,

aflatoxins.

in Food

In: J. C.

Processing,

pp.

from Joint

an analysis

FAO, WHO, FAO/WHOIUNEP

of world

UNEP

wide

corn-

food contamination International Congress

1987. in domestic

72: 543-548,

Sheabar,

analysis

of mycotoxins:

data Second

G. E. Aflatoxin

Chem.,

34.

Stow,

Vol.

1981.

Jelinek,

33.

Pathology,

aflatoxins using high-performance

derivitisation

U. L. Unwanted

In: M. Van Worhol,

P. J. L., and

for by

485-490,

and J. C. Kirschman

32.

spices followed

and post-column

J. Chromatogr.,

of aflatoxin-albumin

assessment.

Techniques in Diagnostic Press, Inc., 1991.

M. M., Taylor,

of United Kingdom purchased column clean-up procedure graphic

for quantification

to human

D. Velzen, and G. R. Bullock (eds.), pp. 139-155. New York: Academic

and

imported

food

and

feeds.

J. Assoc.

1989.

F. Z., Groopman.

J. D., Qian,

of aflatoxin-albumin

adducts.

G. S., and Wogan,

Carcinogenesis

G. N. Quantitative

(Land.),

14:

1203-1208,

1993. 35.

Wang.

L.,

Gange,

patterns

i-S.,

Qian,

S. J.,

Cancer

Zarbe, A,

Biomark.

San

T.

W.,

Qidong

Prey.

Y-R.,

and

in hepatitis

of Daxin,

A. J., and Wild, of Aflatoxins:

223-253.

A.. He, X., Zhu,

Kenzler, adducts

residents

Epidemiol.

36. Hall, Toxicology pp.

G-S.,

Munoz,

of aflatoxin-albumin

antigen-negative 16. Groopman. J. D.. Wild, C. P., Hasler, J. A., Jun-shi, C., Wogan, Kensler, T. W. Molecular epidemiology of aflatoxin-N7-guanine levels a biomarker in experimental rat models and humans. Environ. Health 99: 107-113, 1993.

In:

J. C., Carvajal,

Chapot,

adducts

Anal.

15. Thu. J., Zhang, L. S., Hu, X., Xiao, Y., Chen, J. S., Xu, Y. C., Fremy, Chu, F. S. Correlation ofdietary aflatoxin BC levels with excretion ofaflatoxin in human urine. Cancer Res., 47: 1848-1852, 1987.

5:

1983. 29.

modities monitoring

R. C. Aflatoxin B1-oxide generated by chemical causes guanine substitution in nucleic acids. 1977.

14. Groopman, J. D., Hall, A. J., Whittle, Montesano, R., and Wild, C. P. Molecular human urine obtained in the Gambia, West Prey., 1: 221-227, 1992.

P. R. and

V. N. and Wogan,

Shamsuddin,

Measurement of Human

and

Kingdom

12.

13. Hertzog, P. J., Lindsay-Smith, J. R., and Garner, chromatography study on the removal of DNA-bound in vitro. Carcinogenesis (Land.). I: 787-793, 1980.

C.

Exposure

Wiley

31.

Nature

K. A.,

of a putative

exposure.

John

I I. Essigmann, J. M., Croy, R. G., Nadzan, A. M., Busby, W. F., Reinhold, V. N., Blichi, G., and Wogan, G. N. Structural identification of the major DNA adduct formed by aflatoxin B, in vitro. Proc. NatI. Acad. Sci. USA, 74: 1870-1874, 1977. C. N., and Garner, oxidation of aflatoxin (Lond.), 267: 863-865,

Bradley.

R.

carcinogen

United

J. D., Wogan, G. N., Roebuck, B. D., and Kensler, T. W. Molecfor aflatoxins and their application to human cancer prevention. 54(Suppl.): l907s-I9I Is, 1994.

Martin,

Donahue,

chromatography.

adduct 1978.

2,3-dihydro-2-[7’-guanyll-3-aflatoxin district, Kenya. Carcinogenesis

Assessing

8. Perera, F. P., Mayer, J., Santella, R. M., Brenner, G., Jeffrery, A., Latriano, L., Smith, S., Warburton, D., Young, T. L., Ted, W. Y., Hemminki, K., and Brandt-Rauf, P. Biological markers in risk assessment for environmental carcinogens. Environ. Health Perspect., 90: 247-254, 1991.

or enzymic

Prey.,

of

1987.

9. Groopman. ular biomarkers Cancer Res.,

H.,

A. Detection

A.,

antibody

J. M., Reinhold,

tical to Murang’s

with accurately with sampling,

carcino-

Biomark.

of aflatoxin-N7-guanine

liquid

5. Whitaker, T. B., and Park, uring aflatoxin in food and

Autrup,

sunna,

L. J., Pikul,

monoclonal

cation of the principal aflatoxin BC-DNA NatI. Acad. Sci. USA. 75: 1745-1749,

associated associated

447

B., Ito, N.,

for comparative

Epidemiol.

in rat urine

4. Harrison, J. C., Carvajal, M., and Garner R. C. Immunological detection of aflatoxin B1-DNA and aflatoxin B1-serum albumin adducts in rat and human tissues. In: R. C. Garner, P. B. Farmer, G. T. Steel, and A. S. Wright (eds.), Human Carcinogen Exposure. pp. 255-265. Oxford. United Kingdom: Oxford University Press, 1993. D. L. Problems feeds: errors

& Prevention

S., Chapot.

A basis

J. A., Trudel,

dosimetry

metabolites

267-274,

26.

Biomarkers

1996.

Wogan, and

R., Barraud,

R. Aflatoxin-albumin

between

aflatoxin

Britain Statutory Instruments. Dried Fig Products: Regulations,

Britain

Risks to HuMonographs

Epidemiology,

Zhang,

B surface

County,

5: 253-261,

B-C.,

Grooprnan

I.

antigen-positive

People’s

Republic

Academic

Press,

1996.

Inc.,

(eds.). The Significance

1994.

37. Egner, P. A., Gange, S. J., Dolan, P. M., Groopman, J. D.. Munoz, T. M. Levels ofaflatoxin-albumin biomarkers in rat plasma are modulated long-term

and

1769-1773, 38. Yeh, Hepatitis China. 39.

with

Oltipraz.

F-S., Yu, M. C., Mo, C. C., Luo, B virus, aflatoxins, and hepatocellular

Cancer Wild,

the world.

Res.,

C.

2506-2509,

49:

P., Jiang,

Y. Z.,

Carcinogenesis

(Land.),

16:

Carcinogenesis

Waldman,

Rosenoer,

noer.

M. Oratz,

Uses.

Oxford,

(Lond.),

Structure, Press,

V. M. Clinical United

M.

aspects

A. Rothchild

Kingdom:

in human

L. A.

M.,

sera from

2271-2274,

11:

catabolism.

Albumin Pergamon

and

S. J., Jansen,

adducts

T. Albumin

S., Tong, M. J., and Henderson, B. carcinoma in Southern Guangxi,

1989.

Allen,

R. Aflatoxin-albumin

Rothchild (eds.), United Kingdom: 41.

interventions

Kensler, by both

1995.

Montesano,

40.

transient

and of China.

C. P. In: D. L. Eaton and J. D. Groopman Human Health, Veterinary, Agricultural

Diego:

Jacobson.

D. Temporal

and

of albumin (eds.),

Pergamon

A. J., and regions

of

1990.

In: V. M. Rosenoer, Function 1977.

Hall,

different

Uses,

255-273.

metabolism.

Albumin Press,

M. Orate, pp.

1977.

Structure,

In:

and M. A. Oxford.

V. M. RoseFunction

and