Urinary mandelic acid and hemoglobin adducts in fiberglass-reinforced plastics workers exposed to styrene.

mandelic acid and hemoglobin adducts in fiberglass-reinforced plastics workers exposed to styrene. Scand J Work Environ Health 1994;20:451-8. OBJECTIVES- A field study was undertaken to investigate the effects of occupational styrene expo sure on mandelic acid excretion and the formation of styrene-7,8-oxide hemoglobin adducts. Espe cially the sensitivity of a gas chromatography-mass spectrometry method for determining hemoglobin adducts was evaluated. METHODS - Over a four-week period, each individual of a group of 52 fiberglass-reinforced plastics workers was monitored once a week by the simultaneous measurement of styrene in the air and uri nary postshift mandelic acid. In addition mandelic acid and hemoglobin adducts were monitored in a group of 24 unexposed referents. At the end of the monitoring period styrene-7,8-oxide adduct for mation on N-terminal valine in hemoglobin was examined by gas chromatography-mass spectrometry according to the modified Edman degradation technique. RESULTS - Personal air samples showed an average styrene exposure of 31 mg · m', The average postshift mandelic acid was 98 mg . g creatinine" . For workers not wearing respirators and not show ing breath ethanol, the correlation coefficient between styrene and mandelic acid was 0.78. The blood samples were analyzed for styrene-7,8-oxide adducts on hemoglobin. With a detection limit of 10pmol . no styrene-7,S-oxide adducts were found under these exposure conditions. CONCLUStON- Adduct formation in humans is less effective than in mice. In comparison with ethyl ene, styrene is at least 70 times less effective in forming hemoglobin adducts. Investigating adduct formation in humans at or below the exposure levels reported in this study would require a detection limit of about one order of magnitude better.

As an industriall y important che mical, styrene is widely used in th e production of plastics and re sin s. Because of manual applica tio n te chniques, rel atively high ex pos ure occurs in the product ion of fi be rglass-re inforc ed styrene-po lyes ter resin s.
T he fate of sty rene in humans wi th re spect to uptake is well under stood ( I, 2). Th e fir st step in the m ajor met abolic pathway of styre ne in volves bio tra ns forma tion to sty re ne-7,8-oxide via microsomal cytochro me P-450 monoo xygenase. Th e maj or metab olic excretion products are urinary mand el ic ac id an d phenyl gl yoxyl ic aci d.
The mutagenic and carcinogenic potency of styrene is act ivated by the transformation of sty rene to , Laboratory of Occupational Hygiene and Toxicology, Occupational and Insurance Medicine, Katholieke Universiteit Leuven (Catholic University Louvain), Leuven, Belgium. 2 Laboratory for Antropogenetics, Vrije Universiteit Brussel (Free University Brussels), Brussel, Belgium. 3 Fund of Occupational Diseases, Brussel, Belgium.
Reprints request to: Dr H Veulemans, Laboratorium voor Arbeidshygiene en toxicologie, Kapucijnenvoer 35, 6de verdiep, B-3000 Leuven, Belgium. styrene-7, S-oxide. This epoxide is abl e to b ind covalently to nu cleophilic sites in proteins and DNA (deoxyri bo nucleic ac id) . Th e measurem ent of DN A and hem ogl obin adducts is a useful tool for monitor ing ex posur e to elec trophi lic age nts, and it gives some informa tion abo ut the ir po te ntia l mutagenic ris k. Th e determi nation of DNA adducts is the most suitable parameter for me asuring the biologically effec tive dose of a pot ential carci nogen. Ho we ver , the lack of available D NA in ti ssue sa mp les and the require d detecti on lim it makes it very diff icult to determine them ro uti nely. In co ntras t, hem ogl obin is suffi cient ly available in blood samples and has a life spa n of 120 d. Th erefore it is possible to measure a cum ulative do se . Fo r determining sty re ne-7,8-oxide adduc ts to N-ter mi nal va line in hemogl ob in , the N-a lky l Edman degradat ion technique was used (3). Styrene-7,8 -oxide adduc t for ma tio n on hemoglobin and DNA foll owing exposure to styren e has been demonstrated in ani mals (4,5). In hum an s, styre ne-7 ,S-oxide adduc ts were fo und in three recent stu dies. Brenn er et al (6) fou nd high er adduc t levels in styrene-ex posed wo rkers than in un exposed refere nts. C hris takopou los et al (7) fo und low bu t detectable 45 1 Scand J Work Environ Health 1994. vo l 20 , no 6 styrene-7,S-ox ide adducts on hemoglobin in styreneexpos ed workers. In lamination workers 0 6-guanine adducts were detected by Vod icka et al (S) at a level ove r five times the level in the referents.
Because of lim ited data co nce rning styre ne-7,8oxide add uct formation in hum ans, a field study was unde rtak en to monitor simultaneously hem oglob in adducts, mandelic acid excre tio n, and cy toge netic end points (unpublished results).

Chemicals
Pentafluoroph enyl isothiocyanate (PFPITC) was obtained from Fluka (Buchs, Switzerland) and was used witho ut further purification. Styrene-d , and m-chloroperbenz oic acid were purch ased from Janssen, Beerse, Belgium. Hem oglobin (human) was obtain ed from Sigma (St Loui s, Mi ssouri , United States). Forrnarnide (analytical grade) was ex tracted with pent ane before use. All other chemicals and solvents were of analytical gra de and used without further puri fication .
Synthesis of 2-hydroxy-2-phenylethylvaline. Valin e ( 1 g) was suspended in 0.1 M sodium hydroxide (5 ml), and styrene-7,S-oxide (2 g) wa s added. The suspension was stirred for 2 1 d at SO°C, and then the solid was filtered off. The solid was was hed seve ral times with methylene chloride, the yield bein g 2-hydroxy-2-ph en yleth ylval ine (HOS tVal) as a white powder. Th e nuclear magnetic resonance spectrum (DP , K 2CO ) and mass spectrum show ed that the reaction occurred onl y through the chiral beta-carbon of sty rene-7,S-oxi de and not throu gh the alphacarbon that would result in the for mation of 2-hydroxy-l-phen ylethylvaline.

Synthesis of deuterated styrene-d s-7,8-oxide.
Styrened s-7 ,8-oxide was prepared acco rding to the meth od of Imut a & Ziffer (9). m-Chl orope rben zoic acid (1.73 g) was added in sma ll portions over 10 min at O°C to a stirred solut ion of styrene-d, (5 g) in dichl oromethane ( 100 ml) and a pho sphate buffer (l00 ml, NazHPO. 0. 1 M, NaHzPO. 0.1 M, pH 8). The mixture was stirred for 5 h at room tempe rature, then cooled to O°C, and aga in m-chloroperbenzoic acid ( 1.73 g) was add ed in small port ions ove r 10 min . The mixture was stirred for another 5 h at roo m temperature, and then the orga nic ph ase was separated and washed with a saturated sodium thiosulfate solution and water and dried with disodium sulfate. The so lvent was evaporated unde r vacuum and styrened g -7,8-oxide was purified with flash chro matography (silica ge l-c hloroform).
Preparation of globin alkylated with sty rene oxide. Hemoglobin (l g) was dissolved in saline (7 m!), styrene-7 ,8-oxide (0.7 ml) was added, and the mixture was stirred for 60 h at 37°C. The alkylated globi n l2-h ydroxy-2-phenylethylvalin e globin (HOStGb)l was then isolated and washed with ethyl acetate and pentane and dried by a ge ntle stream of air led ove r the globin. Globin alky lated with styrene-d,-7,8oxi de (HOStGb-d s) was prep ared in the same way and used as an internal standard. Both the deut erated and nondeuterated globins were used to establish calibration curves.
Determination of sty rene oxide adducts Isolation of hemoglobin . Immedi atel y afte r the blood sampling, the red blood ce lls were separa ted by centrifugation at 300 0 revolutions/min for 10 min. The cells were washed three times with isotonic saline and then lysed through the addition of I volume of distill ed water. Cell membranes and debri s were sedimented by centrifugation at 5000 revolutions/min for 1.5 h. The lysed blood sa mples were then stored at -20°C until furth er wor k-up and analysis. Globin was isolated acco rding to Mowrer et al ( 10). After the addition of 0.05 M hydrochloric acid (HCI) in 2-propanol , the mixture was centrifuged at 3000 revolutions/min for 10 min , a dark red pellet being left. Th e globin was precipitated from the supernatant by the addition of eth yl acetate. The globin was filtered and washed twice with ethyl acetate and pentane and then dried by a ge ntle strea m of air passing ove r the glob in.
Deri vatization . Der ivatization of the globin samples was based on the Ed man degradati on technique (3), in which the N-term inal valine, alkylated by styrene-7,8-oxide, specifically is split off as a pentafluorophenyl thiohydant oin (HOStVal PFPTH).
About 50 mg of globin was dissolved in 1.5 ml ofform amide and HOStGb-d g , containin g 17.3 pmol of N-alkylated valin e (HOStVal-d g ) , was added as an internal sta ndard. To adju st the pH , 30 ul of sodium hydroxide (I M) was add ed, foll owed by 15 fl.l of PFPITC. The mi xture was shake n ove rnight at roo m temperature, and the de rivatization was co mpleted at 45°C for 90 min. The reaction mixture was extracted three times with ether (2 ml), and the combined extracts were evaporated to dryness under nitrogen. The residue was redis solved in toluene ( 1 rnl) and washed twice with 2 ml of sodium hydrogen carbonate (NaHCO]) (0. 1 M) and finall y with water . After eva poratio n to dryness, the res idue wa s redi ssolved in 50 fl.l of toluene for analysis by gas chromatograph y-m ass spec tro metry.
Determination by means of gas chromatographymass spectrometry. The analyses wer e ca rried out with a HP 5890 series II gas chro matograph coupled to a HP 5970 quadruple mass spectrometer and a flP 59940A workstation. The chroma tographic separation of the different components in the samples was made on a DB-5-ms fused silica capillary column (30 m x 0.32 mm, 0.l 2 urn phase thick ness). Helium was used as the carrier gas. The samples were injected (1-10 ul) by a solid injector. The oven temperature was programmed at 5°C per minute from 150 to 250°C, follo wed by an increase of 10°per minute to 300°C.
The mass spectrometer was opera ted in the electron impact mode with an electron energy of 70 eV. The interface temperature was 270°C, and the source pressure was 0.005 Pa. The analyses were carried out with selected ion monitoring.
Calibra tion. Calibration curves were established from mixtures of HOStGb (0-10 ug, containing 2.03 nmol HOStVal per milligram of hemoglobin), HOStGb-d s (10 ug, containing 1.73 nmol of HOStVal-d g per milligram of hemoglobin) and 50 mg of unalkylated hemoglobin. The samples were derivatized and analyzed according to the preceding description.
Calibration of reference globin. First, solutions of HOStV al and HOStVal-d g were prepared in a mixture of l-propanol and 0.5 M NaHCO ] (volume/volume 112). A calibration curve was established with the use of mixtures of HOStVal (5-500 JlI of a solut ion containing I nmol · rnl" ) and HOStV al-d g (50 JlI of a solution containing 1 nmol · ml' ). The mixtures were derivatized as described by Tornqvist et al (3). l -Propanol/Nal-lf.O , was added until the total volume was 1.5 ml, IS JlI of PFPITC was added, and the mixture was stirred at 45°C for 90 min. Then the reaction mixture was extracted twice with n-heptane (2 ml), and the combin ed extracts were evaporated to dryne ss under nitrogen. The residue was redissolved in toluene (I ml) and washed twice with 2 ml of NaHCO] (0.1 M) and finally with water. After evaporation -to dryness, the residue was redissolved in 50 JlI of toluene and analyzed by means of gas chromatography-mass spectrometry.
For the determination of the adduct level in the reference globin, 10 mg of HOStGb was dissolved in 0.5 ml of 6 M HCI, and 0.5 ml of a solution containing 100 pmol HOStVal-d g per milliliter of HCI (6 M) was added. The n the reaction flask was closed under nitrogen and stirred for 16 h at 120°C to hydrol yze the globin. Aliquots of 20 Jll of the hydrolysate were evaporated to dryness and redissolved in 1.5 ml of l-propanol/NalfCfr .. Derivatization and sample preparation were carried out as already described. With the use of the pre vious calibration curve the adduct level in HOStGb was determin ed.
The adduct level in HOStGb-d g was determined in a similar way, HOStVal being used as the internal standard. Determination of ethylene oxide adducts. The amount of hydroxyethylvaline (HOEtVal) was determined simultaneously with HOStVal using HOStGb-d M as an internal standard. Calibration curves were established with mixtures of hemoglobin alkylated with ethylene oxide (0-30 pmol of HOEtVal), HOStGbd, (10 ug, containing 1.73 nmol HOStV al-d s . mg hemoglobin" ) and 50 mg of unalk ylated hemoglobin. The hemoglobin alkylated with ethylene oxide was a gift from NJ Van Sittert (Shell, The Netherlands).

Analysis of air and urine samples
The air samples collec ted in a charcoa l tube were analyzed on a HP 5880 A gas chroma tograph equipped with a dual wall-coated open tubul ar (WCOT ) capillary column system (Carbowax 20M, 50 m x 0.2 mm, phas e thickness 0.2 urn and OV-I, 50 m x 0.2 mm, phase thickness 0.2 um), dual flame ionization detector and an automatic injector (0. 1 JlI) with an inlet splitter (ratio I: 10). Helium was used as the carrier gas. All samples were analyz ed simultaneou sly on both columns. The oven temperature was programmed at 40°C for 10 min followed by an increase of 10°per minute from 40 to 200°C. Sample preparation was done by chemi cal desorpt ion with I ml of carbo n disulfid e. Calibration curves were established by spiking charcoal tubes with different amounts of styrene.
The air samples collected with orga nic vapor monitors were analyzed on an HP 5890 series II gas chromatograph equipped with a WCOT capillary column (CP-Sil 5, 50 m x 0.32 mm, phase thickness 0.l 2 urn), flame ionization detector and an automatic injector ( I JlI) with an inlet splitter (ratio I :50). Helium was used as the carrier gas. The oven temperature increase was programmed at 5°per minute from 45 to 95°C. Sample preparation was done by chemical desorption with 1.5 ml of carbon disulfide. Calibration curves were obtained by spikin g the monitors with known amounts of styrene.
Mandelic acid in urine was determined by high pressure-liqu id chromatography. The analysis was perform ed on a Varian model 5000 high pressureliquid chrom atograph equipped with a C-18 reverse phase column (ET 250/8/4 Nucleo sil 120-7 CIS' 250 mm x 4 mm, particle size 7 urn), automatic injector (20 JlI), ultraviolet detector (220 nm) and an electronic integrator. A mixture of 80% water, containing 0.5% acetic acid (pH 4.6) and 20% methanol, was used as the eluent. Sample preparation was done through the extrac tion of mandeli c acid with ethyl acetate, evaporation of the extract, and the redisso lving of the dried extra ct in water. All of the mandelic acid values have been expressed in milligram s per gram of creatinine. Different concentrations of mandelic acid in water were analyzed to establish a calibrat ion curve. 4-Hydroxy-benzoic acid was used as an internal standard.

Sampling procedure
In the ex pos ed gro up, air and urine sa mples were taken once a week dur ing four consec utive weeks. Air sa mples were collec ted on 3M 3500 organic vapor monitors and on standard charco al tube s ( 1001 50 mg, SKC Inc , Unit ed States, lot 120) with the aid of accuhaler pump s from MDA Scientific Inc, United States (type 808, 50 ml . min-I). Co nsider ing the type of work, it was judged that hal f-shift air samples would be representat ive for the entire workday. Urine samples were collec ted on the same day as the air samples at the end of the shift. At the end of the monitori ng peri od , blood samples were co llected in heparinized Vacut ainer" tubes for the analysis of hemoglobin addu cts and in Vacut ainer" tubes containing calparine for the determination of cytogenetic end points.

Subjects and worksites
A gro up of 52 men (ages 21-52 yea rs) daily exposed to styrene participated in the study. These workers were emplo yed in a plan t manufacturing fiberglass -reinforced plastic pipes and cis terns . As thi s study was incorp orated into the health policy of the factory, all workers of the fiberglass-rei nforced styrene -polyester resin plant participated in the study. The plant was divid ed into the fo llowi ng three units: pipe production (unit I), pipe fin ishing (unit II), cistern production and finishing (unit III). Work was orga nized into three shifts in unit I and into a single day shift in unit s II and III. The main activity in unit I co nsisted of operatin g the feede rs for the productio n of pipes (14 workers) , laminating standard co nnectors for the pipes (II workers), and cutting and transporting the pipes (8 workers). In unit II the main activi ty co nsisted of lam inating special connector s for the pipes (9 worke rs), prep ari ng different pieces for the connectors (2 workers), and transporting the pieces (I worker). Laminatin g the body of cisterns (5 workers) and spraying the end caps for the cisterns (2 workers) were the main activities in unit III.
A group of 24 workers from another factory and with a comparable socioeconomic background was 454 monitored as a referenc e group. These work ers were employed in the production and repair of woo de n pallets and were not exposed to styrene or org anic so lvents. Mandeli c acid was not dete cted in their urine samples .

Questionnaire
At the end of the monitoring period the subjects were asked to answe r a que stionnaire. In the questionnaire, inform ation was rec orded about such items as numb er of years wo rked in curren t occ upation, past employment, spec ific descr iption of task, and use of pro tective equipment. Other points of inte rest were life-style factors (drinking, eati ng, smoki ng) and medical history (X rays, history of ca ncer) . An over view of the data collec ted with the questionnaire is presented in table 1. The general features of the expo sed and reference groups were similar. The admitted alcohol intake was higher in the reference gro up. Because of the use of asbes tos in other units of the factory, radiography is part of the yearly medical check-up. This policy explains the difference betwee n the ex posed subjec ts and the referent s in yea rs since the last X ray was taken . Smaller differe nces were found for age , weight, and years of educa tio n.
Th e mean time in the curre nt jo b was 0.9 (range 0. 1-5.2) years. In unit s I and II, the time wor ked in the unit was alwa ys less than one year. In unit III the mean time was 3.0 years. Thirt y of the 52 subjects had worked earlier with asbestos , mostly in the asbes tos ce ment unit of the same plant .

Statis tical methods
Th e distr ibut ion of the exposure data on styrene and mandelic acid differed signific antly from the normal distribution. A log-transform ation of the data yielded a better approx imation of the norm al distribution. Both arithmetic and geometric mean s were calculated. All parameters of the different ex pos ure groups were co mpared by means of a two-w ay analysis of variance.
A simple and multiple regr ession was used to study the relation between the expos ure to styrene and the level of mandelic aci d found in the urine samples . Because of the dep artures fro m normal ity a Spearma nn rank corre lation was also calculated.
The statis tica l analysis was performed with SA S (statistical analysis system) using SA S/ASS IST software, version 6, fir st edi tion .

Results
Ana lysis of the air and urine samples of the exposed workers Th e analysis of the personal air sa mples for styrene showe d an expos ure range of 2.2-110. 1 mg · m-J and a mean of 3 1.0 mg . m? (mea n geom =22.4 mg . rrr"). Exposure to styrene was sig nifica ntly higher (P< O.OOI) in unit I. In units II and III the exposure levels were similar (table 2). Other products found in the air samples were mainly acetone and methyl ethyl ketone. In some of the air samples ethanol was detected in a range of 0.3-55 mg . m', indicating recent alcohol consumption by the subj ects.
Mandeli c acid in the urine samples was found in the range of 11-649 mg . g creatinine :' and a mean of 102 mg . g creatinine: ' (mean geo m = 76 mg . g ereatinine'"). The level of mandelic acid in urine was sig nificantly higher (P<O.OO I) in unit I. No significant difference was found between units II and III (table 2).
The relation between styrene exposure and mandelic acid was studied by multiple linear regression analysis for subjects not wearing respirators, the presence of ethanol in the breathing zone being used as a dummy variable. The highly sig nificant interaction between the presence of ethan ol and styrene expo-sure (P<O.OOO I) indicated different slopes for the regressio n lines.
Optimization of the gas chromatography-mass spec trometry procedure In vitro styrene-7,8-oxide adduct formation on the N-terminal valine in hemoglobin mainly occurs through the chiral beta-carbon of styrene-7,S-oxide, Table 2. Styrene exposure and urinary mandelic acid levels of the exposed workers in the different units.

Styrene in air (mg ' m-3 )
Mandelic acid in urine (mg ' g creat inine -2j Unit " Arithmet ic SOb Geometric   resulting in the formation of two diastereoisomers of the valine adduct (7 ). After derivatization of the adduct, both isomers were separated by means of gas chromatography on a DB-5-ms capillary column. The mass spectrum of HOStVal PFPTH is shown in figure 2; it was identi cal for both isome rs. The base peak, 325, corresponds with a loss of (C 6 H;)CHOCH. Besides the proton shift attend ing the McLafferty rearrangement, the fragmentation was accompanied by a second proton shift from the nearby carbon towards the positively charged nitrogen ( figure 2). In the case of the deuterated intern al standard, a deuterium shift replaced this second proton shift and resulted in the formation of a fragment with 326 amu. Becau se of their high abundanc e (sensitivity) and high mass number (selectivity), these ions were used for the selected ion monitoring analysis. The internal standard eluted a few seconds before the deri va- tized adduct. The parameters of the mass spectromete r were adj usted to allow si multaneo us detection of both styrene-7,8-o xide and eth ylene oxide adducts in hem oglobin. Th e fragment wi th 308 amu co rrespo ndi ng with the loss of NHCH 2CH PH, was monitore d fo r the ethy lene oxide adducts. W ith the se adj ustments , a det ection lim it of 10 pm ol . g-I wa s reached.

Discussion
The styrene expos ure was monitored by half-shift air sa mples . For the evaluation of th ese data as a me asure for the averag e daily exp osure, the styrene conce ntra tio n in the air samples was co mpare d with mandelic acid in the end-shift urine sa mples . Onl y data fro m individua ls not wearing respirator s during the sampling peri od were tak en into account ( I I) .
Since there was no ethanol used in the production process, the presence of ethanol in the breathin g zone (air sa mples) ind icated the int ake of a subs tantia l amo unt of alco hol by some workers be fore wor k or during the breaks. Wilson et al (12) and Berode et al ( 13) rep orted the kinetics of mandelic acid excretion to be substantially altered in hum an volunteers exposed to both styrene and ethanol. Also in this study, a highly significant interaction wa s found between alcoho l intake before or during work and styrene ex pos ure, a finding indicating that alco hol intake sig nificantly delays mand eli c acid excretion. Therefore the relation bet ween sty rene e xposure and urin ary mandelic acid was calculated for subjects not showi ng bre ath ethanol. The good correlatio n between both par am eters in thi s subg roup sup ports the assumption that half-shift air samples were represe ntative for the ave rage dail y ex pos ure. The 95 % co nfid ence int er val calculated for the level of mandelic aci d in ur ine corres ponding with a styrene ex pos ure of 2 13 mg · m? (occ upa tional ex pos ure limit) was bet ween 67 1 and 843 mg . g creatinine:'. Thi s interva l is co nsis tent with the presen t biological ex pos ure inde x of 800 mg . g creatinine:' of the Am eri can Co nference of Governmental Industrial Hygien ists. Styrene-7,8-oxide adducts were not found in any of the ex pos ed workers . Th is result implies that the level of addu ct form ation in the high est exp osed subgro up (unit I, 4 1.3 mg . m') could ha ve been at most 10 pm ol . g hem ogl obin:' , the det ecti on limit of our meth od. In co mpariso n, expos ure to I ppm of ethene gives appro xima te ly 70 pm ol hyd ro xyeth yl valin e (adduct of ethylene) . g hemoglobin" (14 ). Therefore styrene is at least 70 time s less effecti ve in co mparison with ethy lene with respect to adduct formation to hem oglobin.
A field study on mice by Byfalt Nordqvist et al (4) indicated tha t the intraperitoneal injec tion of 1.1 mm ol styre ne -kg body weighr' lead s to the fo rmation of 3 nm ol alky la ted N-t erm inal valine -g hem oglobin". Extrapolation of the se results to humans would me an that adducts would still be detectable at the exp osure levels of our study. The fact that no adducts were found possibly reflects an important spe cies difference in styrene metabolism . Styrene is primarily eliminated by metabolism , and the initial and rate-limiting step in the metabolism of styrene is ca talyzed by microsomal monooxygenase. The ep oxide is detoxified by epoxide hydrol ase. Mendrala et al (15) reported that the ac tivity of the epoxi de hyd rolase relative to mon oo xygenase ac tivity was much greater in hum an liver than in mou se liver. Th e apparent half-time for styrene-7,8-oxi de in mouse and human wa s calculated to be 38 and 1.8 min, resp ectivel y.
Onl y a few studies report the presen ce of styrene-7,8-ox ide addu ct s on hemoglobin in hum an s. Brenner et al (6) found a higher adduct le vel in styre neex posed workers (mean gcom = 48.2 mg . rrr' ) versus referent s. The differ en ce bet ween both gro ups was not statistica lly significant and mainly due to one individual with a high adduct level. Farm er et al (16) have studie d occ upa tio nal ex pos ure to styre ne with a limi ted number of sa mples , but no addu cts have so far been detected. Recently Christ ak opoulos et al (7) found styrene-7,8-oxide addu cts on hem oglobin (mean 28 pm ol · g'") in styrene-exposed wor kers (mean = ± 300 mg . m'). This finding is not inconsistent with our results. The mean exposure in that study wa s about se ven times higher than th at of our most exposed subgroup (unit I) . Extrapolating the adduct levels fo und by Christakopoulos et al (7) to the exp osure level of our subgroup I would result in an expected adduct le vel of about 4 pm ol . g' . Th is value is c learly bel ow the de tection limit of the pre sent meth od. T o study adduct formation in humans at or below ex pos ure conce ntrations of 40 mg . rrr' would require a detection limit which is about one orde r of magnitude better.
In the study of Vodicka et al (8) DNA adducts we re foun d at exp osure le vel s comparable with the levels rep orted in the study of Christak op oul os et a l (7) . The sma ll difference s in adduct levels bet ween the exp osed person s (5 adducts per 10 8 nucle otides) and the refere nts (I adduct per 10 8 nucl eot ide s) in-dicate that, with the present techniques, DNA adducts are not more sensitive biomarkers for monitoring exposure to styrene at the low levels reported in this study.
Several studies report a significantly higher level of ethylene oxide adducts on hemoglobin in cigarette smokers than in nonsmokers (17)(18)(19)(20). The absence of a significant difference in ethy lene oxide adducts between smokers and nonsmokers is probably due to the safety policy of the factory. Since smoking was not allowed in the work area, the number of cigarettes smoked was rather low (50% of the smokers smoked 15 or fewer cigarettes per day).