CLIN. CHEM. 41/12, 1835-1 840 (1995) #{149}European Beckman Conference

CLINICAL CHEMISTRY, Vol. 41, No. 12, 1995 1835

Macromolecular Adducts Caused by Environmental Chemicals

Hans-G#{252}nterNeumann,’ Olaf Albrecht, Corinne van Dorp, and Iris Zwirner-Baier

We describe three biomonitoring studies in which hemo-globin (Hb) adducts were used as biochemical markers toassess indirectly the target dose of genotoxic chemicals.We monitored the exposure to 1,3-butadiene in occupa-tionally exposed workers and in two control groups byanalyzing the adducts formed by the reaction of the firstactivation product, butadiene monoepoxide, with theterminal valine of Hb; we also measured hydrolyzableadducts formed by the reaction of metabolically formednitroso derivatives with Hb from five selected nitropoly-cyclic aromatic hydrocarbons (1-nitropyrene, 2-nitroflu-orene, 3-nitrofluoranthrene, 6-nitrochrysene, and 9-nitro-phenanthrene) in coke oven workers of different jobcategories and control workers of the same geographicalarea. We detected hydrolyzable adducts from monocy-clic nitroarenes in blood from individuals living in acontaminated area where explosives had been producedand from controls. The contaminants considered were2,4,6-trinitrotoluene; 2,4- and 2,6-dinitrotoluene; and 1,3-dinitrobenzene. Differences between groups were signif-icant, but interindividual variation was great and back-ground exposures must be considered.

Indexing Terms: genotoxins/biomonitoring/butadiene/nitroarenes

The exposure to genotoxic carcinogens from the en-vironment or at the workplace is of great concern to the

public. Methods are therefore required to identify truehazards and to quantify exposure-related risks. A dif-ferentiated concept of dose (1) was instrumental for thedevelopment of new strategies. Before this develop-ment, dose-response relations had been based on theexternal dose as measured by environmental concen-trations. The actual uptake of environmental contam-inants is much more relevant and can be determined bymeasuring blood concentrations of the parent chemicalor major metabolites. However, genotoxic chemicalsusually act via reactive metabolites not readily avail-able from blood or urine samples. These chemicals canbe analyzed by measuring their reaction products, e.g.,with DNA in lymphocytes or with hemoglobin (Hb) inerythrocytes.2 Such reaction products are currentlyconsidered to be the best equivalent for the biologicallyactive or target dose; this conception should correlatemuch better with biological endpoints than does that of

Department of Toxicology, University of Wurzburg, VersbacherStr. 9, 97078 Wurzburg, Germany.

1 Author for correspondence. Fax Int + 49-931-2013988.2 Nonstandard abbreviations: Hb, hemoglobin; GC-MS, gas

chromatography-mass spectrometry; BD, 1,3-butadiene; PAH,polycyclic aromatic hydrocarbon; SIM, single-ion monitoring; NT,mononitrotoluene; DNT, dinitrotoluene; TNT, trinitrotoluene;and 1,3-DNB, 1,3-dinitrobenzene.

Received June 12, 1995; accepted July 24, 1995.

the external dose. Various methods have been devel-oped recently to quantify such biochemical markers,and they have been applied for biomonitoring in hu-man populations (2). The results so far indicate thatthese biochemical markers are more specific and moresensitive than early biological markers such as cytoge-netic alterations. The use of biomonitoring procedureshas improved exposure control considerably, and weexpect that the accumulating information will eventu-ally help to assess the risks associated with specificexposures.

Covalently, DNA-bound metabolites are mostly ana-lyzed by immunoassays and 32P-postlabeling, with sen-sitivities down to one adduct in 107_lOb normal bases.Gas chromatography-mass spectrometry (GC-MS) isgenerally used for protein adducts, with the sensitivitydown to 0.001 pmol/g protein. Target tissues are usu-ally not available to analyze either one of the two typesof adducts, but lymphocytes and blood proteins areaccepted as reasonable substitutes (2). Although reac-tions with DNA are considered more indicative ofgenotoxic effects, the analysis of protein adducts hasadvantages that make it attractive for use in exposurecontrol (3), particularly because any use of adductmeasurements for risk assessment has yet to be devel-oped.

In discussing the relation between the two types ofmarkers, one should keep in mind two aspects. If theadduct-forming metabolite is the same, or if a reactiveprecursor is common for both the reactions with DNAand with protein, a direct correlation between DNAand protein binding can be assumed and has beendemonstrated in several cases (4). The difference inelimination kinetics of the different adducts is also notvery confounding, because with environmental expo-sures and long-lived adducts, steady-state situationsusually prevail. The analysis of Hb adducts proved tobe quite suitable to control exposure in many cases.

1 ,3-Butadiene (BD)

BD is listed in the 1990 US Clean Air Act amend-ments as one of 189 hazardous air pollutants (5) andhas been classified as a group 2A carcinogen (6). BD is

metabolically activated to 1,2-epoxybutene-3, whichcan be further oxidized to diepoxybutane or hydrolyzedand oxidized to yield a diolepoxide (Fig. 1). Reaction ofthe epoxides with glutathione is considered to be aninactivating pathway competing for the bioavailabilityof reactive metabolites. Although the relative contribu-tion of the mono- and diepoxide for genotoxic effects isnot clear (the bifunctional diepoxide could be the moreeffective mutagen), analyzing the reaction products ofthe monoepoxide provides information on the balance

CH2-CH-I1- CH2- OH DNAAdduct

/

I’CH2-CH-CH- CH2

LJGlutathione

25

20C

.2 15C,

10

E

5

0

Fig. 2. Biomonitoring of workers exposed to BD (-1 ppm, males) bymeasuring adducts formed by reaction of BD monoepoxide with theterminal valine of globin.Box plots areshown indicating meanand median(waist)valuesas well as the95 percentile. Exposed group and one control group are individuals from a BDproduction plant in Kralupy near Prague (Kr), and one control group is fromWurzburg ONu).

Exposed C Kr. C WU.(N=24) (N-19) (N=10)

‘‘- controls Kr. (N19) Exposed Kr. (N25)LI” controis WU. (N10)

Reference Value (95 Percentile of controls)

III II

1836 CLINICAL CHEMISTRY, Vol. 41, No. 12, 1995

- o,__. ,o..,,cH2-cH-cH-cH2--.cH2-cH-cp-cH2-- CH2-CH-CH-CH2

1.3-Butadiene Butadiene monoepoxide Butathene diepoxide

OH /GSH \H2CCH\CH -Valine--Globjn

HOCH’

Globin adduct

Fig. 1. Metabolic activation of BD and formation of DNA and Hbadducts.

between activating and inactivating pathways in theindividual studied. BD-monoepoxide reacts with,among other amino acids,the terminal valine in Hb (7),and adduct concentrationsinblood ofrats and mice aresignificantlydifferent,which correlateswith the differ-ent susceptibility of these species for tumor formation(8, 9). Mice are more susceptible and show three toeight times higher Hb adduct concentrations than dorats when exposed to the same environmental BDconcentrations.

As part of a coordinated project (Commission of theEuropean Community project STEP-CT91-0152), weanalyzed human blood samples from male workersexposed to BD in a manufacturing unit near Praguewith external exposures averaging 1 ppm (n = 24). Two

control groups were set up, one consisting of employeesof the same company but working outside the produc-tion area (n = 19) and one with volunteers from ourinstitutein Wurzburg (n = 10). The method wasessentiallyas describedin Albrecht et al.(8), followingthe modified Edman degradation procedure developedby Tornqvist et a!. (10). Although an interlaboratorycomparison on the absolute adduct concentrations be-tween our results and those obtained by S. Osterman-Golkar has not yet been completed, our data shouldreflect the relative adduct concentrations in individu-als and the different groups. A wide and overlappingrange of globin adduct concentrations was found inexposed and control individuals (Fig. 2). The adductconcentrations of control subjects from the same areaappeared rather high; further studies are necessary toexplain this finding. The adduct concentrations of con-trols from Wurzburg were lower and significantly dif-ferent from those of the exposed individuals (P = 0.04).The distribution of adduct concentrations (Fig. 3)shows that several workers had values greater thanthe upper limit (95th percentile) of controls, whichindicates that exposures at the workplace led to anincreased body burden (target dose) in these cases. Thesmoking status was considered, but no difference wasfound between smokers and nonsmokers.

In this coordinated study biochemical and biologicalmarkers were applied. Chromosomal aberrations, mi-cronuclei, sister chromatid exchange, and a ras geneproduct were not significantly increased (11). The 32P-postlabeling method for DNA adducts was developed

but has not yet been applied to the human samples.The results shown with Hb adducts look promising:The method is sensitive enough to measure adductconcentrations in humans. A difference between occu-pational and environmental exposures in controls,however, is masked by the large interindividual varia-tion. One may conclude from these data that occupa-tional exposures on top of background environmentalexposures take place, but their contribution to thebiologicallyactivedose is limited.

PolycyclicNitroarenes

If organic material is heated or combusted, complexmixtures are produced whose compositions depend onthe starting material, oxygen supply, and reactionconditions. Various kinds of tars as well as coke ovenemissions have been found to be carcinogenicto hu-mans; this carcinogenicity has been predominantlyattributed to the content of polycyclic aromatic hydro-carbons (PAHs). Numerous biomonitoring studiesinvolving biochemical and biological markers have

40S

C

‘5

C

e 30

a

C 20

a0.E‘I

10-

CaS0.- 0

0 5 10 15 20 25pmol/g Nb

Fig. 3. Biomonitoring of workers exposed to BD: distribution ofglobin adducts.Abbreviations as in Fig. 2

DNA Adduct

Aryl-N02

Exposed(n = 102)

Controls(n = 19)

.0

0E0.

2

Controls Bottom Bench Top Side(N19) (N=18) (N-71) (N-12)

Fig. 5. Biomonitoring of coke oven workers by measuring hydrolyz-able Hb adducts of five selected nitro-PAHs (see text for details).The exposed subjects, all males, were assigned to three different jobcategories: those working in the bottom area, those working in the middle(bench), and those working on the top side of the coke oven.

CLINICAL CHEMISTRY, Vol. 41, No. 12, 1995 1837

//

eO OH

Aryl-NO Aryl-NHOHH..9

NS-Cys-Hb

Hb Adduct

Fig. 4. Metabolic activation of 1 -nitropyrene and the formation ofDNA and Nb adducts.Only adduct-forming pathways are shown.

demonstrated an enhanced exposure in certain occupa-tional and lifestyle situations (12, 13). Benzo(a)pyrenehas often been used as a representative marker ofexposure, and its DNA adducts were measured byimmunological methods or by 32P-postlabeling. Nitro-PAHs, many of which are strong mutagens, may alsobe present in pyrolysis products (14). Nitro-PAHs arealso carcinogenic, but their contribution to the carcino-genic potency of such mixtures is not clear.

We have previously shown that amino- and ni-

troarenes produce Hb adducts suitableforbiomonitor-ing (15, 16). Nitroarenes are metabolically reduced tonitroso and N-hydroxy derivatives, reactive precursorsfor both the reaction with DNA and with proteins. Hbadducts are generated by the reaction of nitroso deriv-atives with the SH group of cysteine in Hb followed byan intramolecular rearrangement that results in theformation of a sulfinic acid amide. This Hb adduct isusually stable in vivo but can readily be hydrolyzed invitro (Fig. 4). The resulting amino arene can be ex-tracted and quantified by HPLC or GC-MS. Otheradducts may be formed as well. With this method onlythe hydrolyzable sulfinamide-type adducts are mea-sured, but they account usually for 80-90% of the Hbadducts.

We proposed analyzing blood samples from coke ovenworkers for Hb adducts from several nitro-PAHs aspart of a coordinated project (European CommunitySteel and Coal Commission, contract 7280-01-014) inwhich several biochemical and biological markers werestudied. The following questions were addressed: (a)

Can Hb adducts from nitro-PAHs be detected in hu-mans, and is the method suitable for exposure control?(b) How does the biologically active dose compare withthat of PAH? (c) How do the biochemical markerscompare with the biological markers?

We selected five representative mononitro-PAHs foradduct analysis; blood samples were processed, and therespective amino arenes were derivatized with pen-tafluoropropionic acid anhydride and analyzed in the

Table 1.Biomonitoringof coke oven workers:Hb adducts.

Samples with detectableconcentrations of adducts, %

1 -Aminopyrene 95 89

2-Aminofluorene 91 58

9-Aminophenanthrene 89 53

3-Aminofluoranthrene 56 32

6-Aminochrysene 51 21

single-ion monitoring (SIM) mode, negative chemicalionization, with a Hewlett-Packard 5988 GC-MS. Thedetection limit was 0.1-0.4 fg. Deuterated 9-aminoan-thracene served as an internalstandard. The recoveryof each of the five amino arenes was 78-89%.

The coke oven workers were assigned to three differ-ent job categories: (a) working in the bottom area (n =

18); (b) working in the middle, i.e., the bench area (n =

72); and (c) working at the top side (n = 12) of the cokeoven. Controls were selected from the same city, someof them from the same company but working outsidethe production area (n = 19).

All fiveanalyzed amino-PAHs were detected to var-ious degrees in exposed and control subjects (Table 1).Comparing the sum of the hydrolyzable Hb adductsbetween controls and the three job categories showedsignificantdifferences for the bottom and top-sideworkers but not for those working in the middle (Fig.5). There was no difference between smokers andnonsmokers. The interindividual variation was againgreat. The distribution of Hb adduct concentrationsamong groups clearlyshows that many exposed indi-viduals had adduct concentrations above a cutoff set atthe 95th percentile of the controls. As an example, Fig.6 shows the distribution of Hb adducts from the mostabundant individual nitro-PAH, 1-nitropyrene. Al-though the pattern of Hb adduct concentrations from

0 2pmollg Nb

3

70aC

60

. 50 controls (N19) Bottom (N18)

4o

30 Reference Value (95 Percentile of controls)20

U . . .1,

Fig. 6. Biomonitoring of coke oven workers: distribution of hydro-lyzable Hb adduct concentrations generated from 1-nitropyrene.

the five nitro-PAHs varied to some extent, a job-specificpattern could not be identified.

In this coordinated study biological markers such as

chromosomal aberrations, micronuclei, and sister chro-matid exchanges were not significantly increased inexposed individuals (17). Leukocyte DNA adducts de-tected with an antiserum directed against benzo-(a )pyrene-modified DNA correlated with environmen-tal PA.H concentrations and were significantlyincreased in maintenance workers (18). DNA and pro-tein adduct concentrations are compared primarilywith regard to the analytical sensitivity required forbiomonitoring. Assenato et al. (18) gave mean concen-trations of total PAH-DNA adducts as 0.085 ± 0.012

fmol4tg DNA (85 pmol/g DNA). This is to be comparedwith the sum of five specified Hb adducts from nitro-PAH with a mean concentrationof1 pmol/g Hb in those

subjects working in the bottom area.

Monocyclic Nitroarenes

Areas that were sites of the production and use ofexplosives during World Wars I and II are still consid-erably contaminated with waste. Some of these areasare inhabited; the question arose whether this contam-ination poses a hazard to the people who live there.Moreover, such contaminants appear in groundwater,creating a more general problem of environmentalpollution.

2,4,6-Trinitrotoluene (2,4,6-TNT) was the predomi-nant explosive used, but the waste contained a mixtureof related compounds that were byproducts of its syn-thesis, particularly isomeric mononitrotoluenes (NTs),dinitrotoluenes (DNTs), and TNTs. In the area of one ofthese former munition factories 3907 analyses of sys-tematically selected soil specimens were carried out,and the following five chemicals were found most often

(19): 2,4,6-TNT (38% of the samples); 2,4-DNT (16%);4-amino-2,6-DNT (16%); 2-amino-4,6-DNT (14%); and2,6-DNT (13%). Among the constituents less frequentlydetected was 1,3-dinitrobenzene (1,3-DNB, 1.3%).

Researchers also found the following chemicals inmost of the 232 groundwater samples of this area (20):

2,4,6-TNT (82% of the samples); 2,4-DNT (84%); 4-ami-

1838 CLINICAL CHEMISTRY, Vol. 41, No. 12, 1995

no-2,6-DNT (73%); 2-amino-4,6-DNT (73%); and 2,6-DNT (13%).

The content in soil, of course, varies widely, and thesum of various nitrotoluenes ranged from a detectionconcentration of 10-50 pg to 1 g/kg dry weight.

The five most frequently occurring monocyclic ni-troarenes are mutagenic and either carcinogenic inexperimental animals (2,4,6-TNT,2,4-DNT, 2,6-DNT)or suspected of being carcinogenic. In the rat 2,6-DNTis a comparatively strong liver carcinogen. Seven and14 mg/kg in the feed produced hepatocarcinoma inpractically all animals within 1 year (21). An exposure-

4 related formation of cancer in humans is thereforewidely discussed. A report of an increased incidenceoflivercancers among munitions workers has recentlybeen published (22). Aside from the problem ofassess-ing such complex mixtures, two questions can be ad-dressed: (a) Are these chemicals indeed taken up byinhabitants living in the contaminated area? (b) Arethe same critical metabolites formed in humans as in

experimental animals?The question of internal exposure cannot be an-

swered satisfactorily even by the most sophisticatedcalculations, but relevant information can be expectedfrom biomonitoring. On the basis of animal experi-ments, we have previously proposed the use of Hbadducts as markers for this purpose (23). In a firststudy we have now analyzed blood samples from peopleliving in one of the contaminated areas covering Hbadducts from the five most frequently occurring ni-troarenes mentioned above and from 1,3-DNB. Morethan one reaction product could be formed from thetoluene derivatives containing two and three nitrogroups, depending on which or how many nitro groupswere metabolicallyreduced and further metabolized(23). In this study we concentrated on the analysis ofonly those adducts resulting from the reaction of me-tabolites in which only one nitro group was reduced. Inthe case of 2,4,6-TNT, hydrolysis of the Hb adductcould release2-amino-4,6-DNT and 4-amino-2,6-DNT.The two monoreduction products of 2,4,6-TNT werealso present in soil and groundwater, presumably be-cause of bacterial reduction of TNT. (The amines couldgive rise to the same Hb adducts after N-oxidation;therefore,on the basis of adduct cleavage products wecould not determine whether the absorbed compoundwas an amino or a nitroarene.)Single reduction of2,6-DNT gave only one cleavage product, that of 2,4-DNT, either 2-amino-4-NT or 4-amino-2-NT.

Analysis of blood samples from 34 potentially ex-posed and 34 controls for six different cleavage prod-ucts from Hb adducts by GC-MS (C. van Dorp andH.-G. Neumann, manuscript in preparation) gave thefollowingpreliminary result(Table 2): 1-Amino-3-NBcould not be detected in any one of the blood samples.Both amino-DNTs were present in some of the samplesof both groups, the mean and the median concentra-tions being higher in potentially exposed individuals(Fig. 7). The most surprising finding was that all threeamino-NT isomers were found in practically all sam-

1000

900

800

700.0

600

500

400300

200

100

0

Fig. 7. Biomonitoring of monocyclic nitroarenes: hydrolyzable Hbadducts of three isomeric amino-NTs (ANTs) and two amino-DNTs(ADNTs) (see text) from residents living in an area contaminated withwastes from explosives (WS; males and females) and from controls(C) living outside this area.

C ANTs WS ANTs C ADNTs WS ADNTs

CLINICAL CHEMISTRY, Vol. 41, No. 12, 1995 1839

Table 2. Biomonitoring of monocyclic nitroarenes fromcontaminations with explosives.

Detected, %

Cleavage products Cases ControlsNitroarene from Hb adducts (n = 34) (n = 34)

2,4,6-TNT 2-Amino-4,6-DNT4-Amino-2,6-DNT

5035

5327

2,4-DNT 2-Amino-4-NT 35 47

4-Amino-2-NT 76 762,6-DNT 2-Amino-6-NT 91 1001,3-DNB 1-Amino-3-NB 0 0

ples from both groups. There was no difference betweensmokers and nonsmokers.

The interindividual variation was extremely great.Hb adduct concentrations of 2,6-DNT, probably themost hazardous component, were from 1 to 120 ng/g Hbwith a detection limit of 0.1 nglg Hb. The distributionbetween exposed and control subjects was not different.

We identified the cleavage products on the basis ofcochromatography with authentic material in GC andHPLC on differentcolumns and by the appropriatemass peak in the SIM mode of the mass spectrometer.In the case of the amino-NTs we checked representa-tive samples for a second mass peak.

Some preliminary conclusions can be drawn fromthese observations:(a) Human populationsare exposedto monocyclic nitroarenes; (b) these chemicals are ab-sorbed and metabolically activated; the same metabo-lites are formed as in experimental animals; (c) themajor components occurring in munitions wastes aredetectablein Hb adducts; (d) the results suggest anexplosive-relatedexposure to TNTs in some cases; (e)

exposure to DNTs is more extensive and not explosive-related; (f) other environmental sources must be con-sidered.

One can only speculate about the other sources atthis time. Drinking water and air contaminations fromatmospheric reactionsand automobile exhausts shouldbe considered. The DNTs, however, are not mentioned

in a compilation of data about diesel and gasoline

engine emissions (24).

Discussion

We used Hb adducts as biochemical markers tomonitor exposure in different situations. We detectedspecific adducts and, where a comparison with biolog-ical markers was possible, our method seemed to bemore sensitive than others; thus, it may be morespecific and more practical to use than DNA adductmeasurements. The method was sensitive enough todetect adduct concentrations in controls in all theexamples studied. Consequently, one should pay moreattention to the exposure sources that build up thesebackground concentrations.

It has been proposed that reference values be defined

as some upper limit (95th percentile)of backgroundexposures and that they be used to decide whethersituation-specific exposures occur (25). We used the95th percentile as upper limit. In the studies with BDand nitro-PAH, a sufficient number of individuals hadadduct concentrations above the reference value toindicate specific occupational exposures. In the case ofthe monocyclic nitroarenes, we observed only a trendfor exposure-related TNT adducts, and the adductconcentrations were much lower than those originatingfrom DNTs, which did not differ from the referencevalues.

Biomonitoring studies in general appear to have awide interindividual variation such that a sizable pro-portion of the exposed individuals have values compa-rable with those at the lower end of the nonexposedindividuals, or that adduct concentrations of somenonexposed individuals are in the range typical foroccupational exposures. This widely overlapping distri-bution of adduct concentrations in different exposuresituations, e.g., occupational vs environmental, ex-plains why it is so difficult to prove a cause-and-effectrelation in epidemiological studies unless specific ex-posures are extremely high. The next step in molecularepidemiology obviously is to find out what the reasonsare for high and low target doses under comparableexposures in the individual.The logicalprocedurewould be to phenotype and genotype individualsto-gether with adduct measurements.

An important conclusion from the present findingsregards the assessment of health hazards associatedwith specific exposures. We have demonstrated back-ground exposures from 10 different nitroarenes, andwe could add the contribution to the total load ofgenotoxic chemicals from several amino arenes knownto occur in the environment. One such example is4-aminobiphenyl, a human carcinogen that has been

found in smokers but also in nonsmoking controls (26).

Observing the “ALARA principle” (as low as reasonablyachievable) for genotoxic carcinogens, it seems unrea-sonable to ask that a specific situation be cleaned up tothe extent that exposures are lower than those leading

1840 CLINICAL CHEMISTRY, Vol. 41, No. 12, 1995

to the observed background adduct concentrations.This does not mean that we should not care aboutbackground exposures-they should also be loweredwhere possible- but to some extent they are unavoid-able. If set tolerance concentrations of exposure areneeded, it appears reasonable to use reference valuesas defined above as guidelines; wherever possible,reference values should be derived with the use ofbiologically effective doses as best represented by ad-duct concentrations.

Work described from our laboratory was supported by theDeutsche Forschungsgemeinschaft, by the Commission of theEuropean Community (STEP-CT91-0152), the European Com-munity Steel and Coal Commission (research contract 7260-01-014), and by the AAV Entsorgungsverband Nordrhein-Westfalen.

References1. Ehrenberg L, Moustacchi E, Osterman-Golkar S. Dosimetry ofgenotoxic agents and dose-response relationships of their effects.Mutat Res 1983;123:121-82.2. Farmer PB. Carcinogen adducts: use in diagnosis and riskassessment. Clin Chem 1994;40:1438-43.3. Neumann H-G. Analysis of hemoglobin as a dose monitor foralkylating and arylating agents. Arch Toxico! 1984;56:1-6.4. Neumann H-G. Dosimetry and dose-response relationships.IARC Sci Pub! 1984;59:115-26.5. US Environmental Protection Agency. Preliminary draft list ofcategories and subcategories under Section 112 of the Clean AirAct. Fed Regist 1991;56:28548-57.6. 1,3-Butadiene. In: IARC Monogr Eval Carcinog Risks Hum1992;54:237-85.7. Osterman-Golkar S, Kautiainen A, Berkmark K, HakanssonK, Maki-Paakkanen J. Hemoglobin adducts and urinary mercap-tunc acids in rats as biological indicators of butadiene exposure.Chem Biol Interact 1991;80:291-302.8. Albrecht OE, Filser JG, Neumann H-G. Biological monitoringof 1,3-butadiene: species differences in haemoglobin binding inrat and mouse. IARC Sci Pub! 1993;127:135-43.9. Osterman-Golkar S, Bond JA, Ward JB Jr, Legator MS. Theuse of hemoglobin adducts as a tool for biomonitoring butadieneexposure. IARC Sci Pub! 1993;127:127-34.10. Tornqvist M, Mowrer J, Jensen S, Ehrenberg L. Monitoringof environmental cancer initiators through hemoglobin adductsby a modified Edman degradation method. Anal Biochem 1986;154:255-66.11. Sorsa M, Autio K, Demopoulos NA, Jarventaus H, Rdssner P,

Sram RJ, et a!. Human cytogenetic biomonitoring of occupationalexposure to 1,3-butadiene. Mutat Res 1994;309:321-6.12. Hemminki K, Perera FP, Phillips DH, Randerath K, ReddyMV, Santella RM. Aromatic DNA adducts in white blood cells offoundry workers. In: Bartsch H, Hemminki K, O’Neill IKO, eds.Methods for detecting DNA damaging agents in humans: appli-cations in cancer epidemiology and prevention. IARC Sci Pub!1988;89:190-5.13. Garner RC, Farmer PB, Steel GT, Wright AS, eds. Humancarcinogen exposure. Biomonitoring and risk assessment. Oxford:IRL Press, 1991.14. Howard PC, Hecht SS, Beland FA, eds. Nitroarenes: occur-rence, metabolism, biological impact. Environ Sci Res 1990;40.

15. Albrecht W, Neumann H-G. Biomonitoring of aniline andnitrobenzene. Hemoglobin binding in rats and analysis of ad-ducts. Arch Toxicol 1985;57:1-5.16. Neumann H-G, Birner G, Kowallik P, Schutze D, Zwirner-Baier I. Hemoglobin adducts of N-substituted aryl compounds inexposure control and risk assessment. Environ Health Perspect1993;99:65-9.17. Forni A, Guanti G, Bukvic N, Fern G, Foa V. Cytogeneticstudies in coke oven workers: preliminary results. Fifth EuropeanMeeting on Environmental Hygiene, Prague, 1995.18. Assenato G, Fern GM, Foa V, Strickland P. Poirier M, PozzoliL, Cottica D. Correlation between PAH airborne concentrationand PAH-DNA adducts levels in coke-oven workers. Int ArchOccup Environ Health 1993;65:143-5.19. Albers H. Expertengesprach Rustungsaltlasten, Marburg1992. Hessisches Ministenum f#{252}rUmwelt, Energie und Bundes-angelegenheiten, Wiesbaden, Contribution C-25.20. Stork G, Steinbach K. Untersuchungsergebnisse. PhillipsUniversitat Marburg, 1992.21. Leonhard TB, Graichen ME, Popp JA. Dinitrotoluene isomer-specific hepatocarcinogenesis in F344 rats. J Nat! Cancer Inst1987;79: 1313-9.22. Stayner LT, Dannenberg AL, Bloom T, Thun M. Excesshepatobiliary cancer mortality among munition workers exposedto dinitroto!uenes. J Occup Med 1993;35:291-6.23. Zwirner-Baier I, Kordowich FJ, Neumann H-G. Hydrolyzablehemoglobin adducts of po!yfunctiona! monocyclic N-substitutedarenes as dosimeters of exposure and markers of metabolism.Environ Health Perspect 1994;102(Suppl 6):43-5.24. Diesel and gasoline engine exhausts. In: IARC Monogr EvalCarcinog Risks Hum 1989;46:41-185.25. Aitio A. Reference limits in occupational toxicology. ClinChem 1994;40:1385-6.26. Bryant MS, Skipper PL, Tannenbaum SR, Maclure M. He-moglobin adducts of 4-aminobiphenyl in smokers and nonsmok-ers. Cancer Res 1987;47:602-8.