Monitoring occupational exposure to styrene from hemoglobin adducts and metabolites in blood.

OSTERMAN-GOLKAR S. Monitoring occupational exposure to styrene from hemoglobin adducts and metabolites in blood. Scand J Work Environ Health 1993;19:255-63. Monitoring occupational ex posure to styrene was achieved through quantification of adducts of styrene 7,8-oxide to N-terminal valine in hemoglobin (Hb) on the basis of the enrichment of adducted globin chains by ion-exchange chromatography and gas chromatographic-mass spectrometric analysis by the use of the N-alkyl Ed man method. Application to blood samples from reinforced plastics workers exposed to styrene and from referents showed Hb adduct levels correlating with the blood styrene glycol and urinary man delic acid concentrations. The blood styrene glycol and styrene 7,8-oxide levels of the exposed work ers averaged 2.5 !Lmol· I-I (17 subjects) and 0.09 umol . I-I (7 subjects), respectively. The blood styrene glycol and urinary mandelic acid content (mean 9.5 mmol· I-I, 17 subjects) suggested a styrene con centration of about 300 mg . m' (75 ppm) in the workplace air. The Hb adduct levels were low (mean 28 pmol . s"). indicating rapid detoxification of styrene 7,8-oxide in humans.

Styrene is used in the production of various plastics, synthetic rubber, and polyester resins. The most extensive human exposure to styrene occurs in the reinforced plastics industry. The occupational exposure limit is at present 25 ppm in Sweden and 20 ppm in Finland .
Th e metabolism of styrene has been extensively studied in experimental animal s ( I, 2). Th e fate of styrene in humans with respect to uptake and di sposition is also well understood (3)(4)(5). The fir st step in the major metabolic pathway is the formation of styrene 7,S-oxide (styrene oxid e), wh ich is hydrated to styrene gly col or conj ugated with glutathione in enzyme-catalyzed reactions. The major urin ary excretion products, mandelic acid, phenylgl yoxylic acid and hippuric acid, are rel ated to styrene gl ycol, indicating the intermediate formation of styrene oxide. Styrene oxide has been detected in the blood of animal s experime nta lly expo sed to styre ne (2) . Low levels of this epoxide have also been indicated in human vo luntee rs exposed to styrene and in occ upation ally exposed workers (3).
Hemoglobin (Hb) is suitable as a monitor of genotoxic agents becau se of its ready availability and its relati vel y long life spa n, which permits cumulative do ses to be determined (6). Th e measur em ent of Hb adducts ha s been used for monitoring en vironmental factors such as occupational exposures to ethylene oxide (7)(8)(9)(10)(11), propylen e oxid e (9,12), and various components of tobacco smoke (13,14). In the case of ethy lene oxide the sensitiv ity of the analysis of adducts to N-terminal valine in Hb by the Nalkyl Edman procedure (8,15,16) ha s been shown to be high enough to monitor exposures down to levels that may be acce ptable wi th respect to the ensuing can cer risk ( 15).
In a pre viou s study ( 17) using the N-alk yl Edman procedure to monitor Hb adducts in sty rene-expo sed workers, deut erated 2-hydroxyethy lvaline was used as the internal sta ndard. Due to the uncertainty with respe ct to the rel ati ve yie lds of the free amino ac id ver sus the modified N-terminal valine residue in Hb in the Edman rea cti on , the results were express ed as proportion al values . The expo sures of the wor kers ranged from 0.6 to 44 ppm . There was a large, though not sign ificant, difference in the level s of Hb adducts of styrene oxide among the exp osed workers and referent s, primarily as a consequence of a sing le hea vily exposed individual with an extremely high level of adducts .
In order to make Hb adducts useful for risk estimation, quantitative data on adduct levels are required. We have developed a procedure to det ermine Hb addu cts of styrene oxid e quantitativel y in workers occupationally exposed to styrene. CZHs)Styrene oxide was synthesized and used for the preparation of an alkylated globin, which served as the internal standard. Adducted Hb chains were isolated by ionexchange chromatography according to Bergmark et al (I8), and adducts of styrene oxide to N-terminal valine were determined by the N-alkyl Edman method. The procedure was applied to styrene-exposed workers and referents. In the same subjects, free styrene oxide and styrene glycol in blood and mandelic acid in urine were monitored. These metabolites give a short-term measure of styrene uptake, whereas the adducts to N-terminal valine in Hb, which according to animal experiments are chemically stable (19), give a measure of the dose (defined as the time integral of the concentration of the alkylating agent) accumulated during the four months preceding the blood sampling.
The results presented in this paper are a part of a comprehensive project including measurements and comparisons of various chemical and biological end points for the biomonitoring of styrene expo sure (20 , 21).  (22), was a gift from Dr A LOf (National Institute of Occupational Health, Solna, Sweden). Pentafluorophenyl isothiocyanate (PFPITC) (purum) and pentafluorobenzoyl chloride (puri ss) were obtained from Fluka AG, Buchs, Switzerland. Formamide (analytical grade; Merck, Darmstadt, Germany) was extracted with npentane before use . The other reagents and solvents were of analytical grade and used without further purification. The C, g-Sep-Pak cartridges were from Waters Associates, Milford, Massachu setts, United States; and the CM-Sepharose CL-6B was from Pharmacia, Uppsala, Sweden.

Chemicals
Synthesis of (2H s) styrene 7,8-oxide. Deuterium-substituted styrene 7,8-oxide was prepared according to a modification of the procedure described by Kologrivova et al (23). CZHg)Styrene (0 .5 g) wa s dissolved in 2.5 ml of ethanol in a 25-ml round-bottomed reaction vessel. Sodium tungstate, 75 mg , wa s added, and the reaction mixture was heated under vigorous stirring at 60-65°C (oil bath) for 30 min. The pH of the reaction mixture was then adjusted to 7.5 with 0 .1 M sodium hydroxide (NaOH) in ethanol, and 256 2 ml of 30 % hydrogen peroxide was added through a dropping funnel during 40 min . The reaction mixture was kept at 65°C under stirring for 12 h. Three milliliters of water (pH 7.5 ) was added, and the reaction mixture was shaken twice with 5 ml of methylene chloride. The organic fractions were pooled and dried with anhydrous sodium sulfate and evaporated. The residue was purified on a column (80 x 0.6 em) packed with silica gel 60 and eluted with toluene:ethyl acetate (8:2). The yield was 130 mg of material, of which 30 mg was dissolved in 15 ml of hexane and washed three times with 2 ml of 0.5 M Tris-HCl [tris(hydroxymethyl)aminoethanehydrochloride acid] buffer, pH 7.5. The CZHg)styrene oxide was then extracted from the hexane into 5 ml of acetonitrile, and this procedure was repeated twice. The sol vent was evaporated. The purity and identity of the compound was assessed by thin-layer chromatography, high pressure liquid chromatography, quantification of the epoxide through reaction with nicotinamide according to an improved version of the method described by Nelis & Sinsheimer (24) , and mass spectrometry. The mass spectrometric analysis was performed by use of a gas chromatograph (HP 5890), a mass-selective detector (HP5970), and a microprocessor workstation (HP 59970C).
Preparation of alkylated globins . Internal standard globin and reference globin were prepared as follows: CZHg)styrene oxide or unsubstituted styrene oxide (6 mg) were mixed with 3H-labeled styrene oxide (1.7 /-lCi) and added to hemolysates from 10ml samples of human whole blood. The mixtures were incubated for 2 h at 37°C and then left at room temperature overnight. Globin was isolated according to Mowrer et al (25). The level of adducts to Nterm inal valine was determined on the basis of radioactivity extractable after derivatization with PFPITC. (Compare with the following text; in thi s case the extraction with hexane was omitted.) A parallel globin sample, treated in the same way but without the addition of PFPITC, was used as a control. The internal standard globin was also used in a study of Hb adduct levels in mice and rats exposed to styrene or styrene oxide (19 ).

Human samples. Venous blood and urine samples
were collected at the end of a workshift (Thursday and Friday) from workers exposed to styrene (17 subjects) in a plant manufacturing containers from unsaturated polyester resins. The exposed workers were laminators with a mean age of 33 years and 6.7 years of average exposure to styrene. Eleven of the workers were smokers. Office workers (originally 16 subje cts outside the plant and one subject from the plant) were chosen as referents. A part of each blood sample was immediately extracted for the analysis of styrene oxide. (See the following text.) Analysis of hemaglobin adducts Enrichment of adducted globin chains. The blood samples were collected in heparinized tubes and were immediately put into an icebath to prevent in vitro formation of styrene 7,8-oxide. Globin was isolated from red blood cells according to Mowrer et al (25). Alpha and beta globin chains containing hydroxyphenethylvaline [here used as a collective name for diastereomers of~-hydroxyphenethyland, possibly, a-(hydroxymethyl)benzyl-valine) were enriched by ion-exchange chromatography using samples of 500 mg of globin as described by Bergmark et al (18). The material eluting immediately before the unmodi-fied~and a chains was collected, dialyzed against distilled water, lyophilized (18), and derivatized as described in the following text for the quantification of Hb adducts by gas chromatography-mass spectrometry (GC-MS).
Derivatization of globin samples. The preparation and derivatization of the globin samples were performed essentially as described by Tornqvist et al (16). The conditions for the derivatization were optimized on the basis of exploratory experiments using different concentrations of reagent (PFPITC) and different incubation times. The following standard procedure was chosen for the analysis: samples of about 20--30 mg of material were dissolved in formamide (0.75 ml) and the internal standard globin (50 ug) was added . Pyridine (6 Ill) and PFPITC (6 Ill) were added, and the samples were incubated at 45°C for 2 h with occasional stirring. In the original method of Tornqvist et al (16) freshly prepared I M NaOH was used to neutralize the globin samples. The use of pyridine for this purpose was considered handier. It was shown that neutralization with pyridine or NaOH gave equivalent results in the analysis of hydroxyphenethyl adducts . The samples were extracted three times with 1.5 volumes of diethyl ether, and the extracts were evaporated to dryness under nitrogen, redissolved in 2 ml of toluene, and washed twice with water, 0.1 M disodium carbonate, and again with water. Finally, the toluene was carefully evaporated , and the samples were dissolved in I ml of methanol:water (60:40) and extracted with 2 x 2 ml of hexane . Hexane was evaporated under nitrogen, and the samples were dissolved in 50 III of toluene for analysis by GC-MS.
Calibration. Standard curves for pentafluorophenylthiohydantoins (PFPTH) of hydroxyphenethylvaline (see also reference 19) were prepared through the addition of various amounts of globin alkylated with unlabeled styrene oxide (reference globin) to a constant amount (50 ug) of globin alkylated with CZH 8 )styrene oxide (internal standard) to 25 mg of reference globin as a biological matrix . The mixtures were worked up according to the method already described . Gas chromatography-mass spectrometry. PFPTH derivatives were analyzed by the use of a Finnigan 4500 gas chromatograph-mass spectrometer. The gas chromatograph was equipped with an on-column injector (OCI-3; SGE Scientific Pty Ldt, Melbourne, Australia) and a DB-5 fused silica capillary column (30 m, 0.32 mm inner diameter, 1 um phase thickness; J&W Scientific Inc, California, United States) coupled to a deactivated precolumn (1 m, 0.3 mm inner diameter) with a glass-lined stainless-steel union (SGE). (See reference 15.) The oven of the gas chromatograph was kept at 100°C for 1 min; the temperature was then programmed 15°C/min to 200°C and kept at this temperature for 1 min, then programmed 10°C/min to 300°C and kept at this temperature for 3 min. The mass spectrometer was operated in the negative ion chemical ionization mode, with an ion source temperature of 100°C, and an ionization energy of 100 eV. The ion source pressure was 0.45 torr (60 Pa), and the helium carrier gas pressure was 9 psi (60 kPa). The mass spectrometer was focused at mlz 424 and 432, the base peaks (M-HF)for the PFPTH derivatives of the valine-styrene oxide products and their deuterated variants, respectively. Quantification was based on peak areas relative to the internal standard. To correct for a possible shift in the ratio of recorded fragments, one of the samples from the calibration curve was reanalyzed on each analytical occasion.
Selection of samples. The determination of adducts to N-terminal valine in Hb involved a time-consuming preisolation of adducted globin chains, and the analysis was therefore limited to samples from three referents and seven exposed workers. The exposed workers were selected on the basis of data on urinary mandelic acid concentrations in order to cover the whole range of exposures .

Analysis of mandelic acid
Mandelic acid in urine was measured according to Engstrom & Rantanen (26).

Analysis of styrene oxide and styrene glycol
Styrene oxide and styrene glycol in blood were analyzed essentially as described by Wigaeus et al (3).
Blood samples of 2 ml from four referents and seven exposed subjects were extracted for the analysis of styrene oxide. Allylbenzeneglycol, 5.67 ng, was used as the internal standard. With the analytical procedure used (3), contribution from certain conjugates of styrene glycol , such as the 2-acetate (27) -which can be extracted into hexane , and also, although certainly less susceptible than the epoxide, is hydrolyzable in acid solution -cannot at present be excluded. Therefore, the reported styrene oxide concentrations have to be considered as upper estimates of the true concentrations. Only 11 of the 34 samples were considered suitable for ga s chromatographic anal ysis because of contaminating material of an unknown nature in the samp les.
Free styrene gl ycol was measured in 0.2-ml blood samples from all of the exposed workers and eight of the referents. For the se analyses 22.7 ng of the internal standard, allylbenzeneglycol, was added.
An essentially quantitative recovery of styrene oxide through extraction with hexane wa s ensured through experiments in which the blood was spiked with radiolabeled styrene 7,S-oxide. Extractions of blood spiked with styrene glycol showed that no detectable amounts of the gl ycol « 0.1 %) were extracted with hexane (3). Calibration curves were obtained with mixtures of all ylbenzene glycol (22 .7 ng) with styrene gl ycol (2-100 ng ), the samples being pro cessed as already described.
A Packard model 436 gas chromatograph equipped with an electron-capture detector and a fused silica capillary column (Oribond SE-54, 25 m, 0.25 mm inner diameter, 0.20 urn phase thickness, Nordion Instruments Oy Ltd , Espoo, Finland) wa s used for the ga s chromatography. The operating condition s were as follows: temperature pro gramming I O°C/min from 150°C to 280°C ; injector temperature I70°C and detector temperature 320°C ; carrier gas nitrogen.

Chara cterization ofeH8)styrene oxide
The chem ical structure of the (ZHs)styrene oxide was confirmed by mass spe ctrometry. Ma ss spectra of the commercially available (unsubstituted) and synthetic Thin-layer chromatography, high-pressure liquid chroma tog raphy, and measurements of alky lating act ivity according to Neli s & Sinsheimer (24) show ed that the purity of the sy nthetic (ZHg)sty rene oxide corresponded to that of the commercially available unsubstituted compound.

Characterization and quantification of PFPTH derivatives of hydroxyphenethylvaline
Gas chromatographic analysis of the PFPTH derivatives obtained after derivatization of globin alkylated with styrene oxide showed two peaks. An identica l pattern was seen with the deuterated analogues. The mas s spec tra obtained from each of the se two peaks were alm ost identical. Figure 2 shows the spectra of the fir st peak. The mole cul ar ions of hydroxyphenethylvaline-PFPTH and the corres pondi ng deuterium-substituted derivati ves are mlz = 444 and mlz =452, res pectiv e ly. Th e dominating fragments are m/z =424 (or 432) and m/z = 318 (or 320) . The former fragments were formed by loss of HF from the molecular ions, and the latter by additional loss of C 6HsCHO (or C 6 zH sC 2HO) . This type of fragmentati on has been observed for adducts with a~-hy drox y group, such as 2-hydroxyethyl, 2-hydroxypropyl , and 2-hydrox y-3-buteny l (29 , 30 ), sug ge st ing that the initial attack of valine-Nlt, occurred at carbon-8 of sty rene oxide . Ho we ver , previous studies '" '" of styrene oxide [eg, in reactions with glutathione (3 1) and guanosine (32) ] showed that both carbons of the oxira ne ring are reactive . (See also reference 33.) Thi s possibility was not studied furthe r in the present investigatio n. The fragments 34 1 and 349 correspond to a loss of (CH) 2CHCH, CO, and F from the molecular ions. The fragments 424 and 432 were monitored for the quantification of adduct levels in globin samples.
Th e concentrations of adducts to N-terminal valine in the reference globin and the standa rd globin were both determined to be 1.2 nmol . mg' on the basis of radioactivity extracta ble after derivatization with PFPITC and the specific activity prepared from the normal and deuterium-substituted styrene oxide , respectively. The standard curve showed good linearity in the range studied (12.5-150 ug of reference globin, N =6, correlation coefficie nt =0.99845) . the seven exposed workers was 28 (range 15-52) pmol . g' .
Styrene oxide. The styrene oxide concentrat ion in the four referents was 0.02 urn ol . I-l or below. The values recor ded for the seven expose d worke rs, 0.04-0.13 umol . 1 1, agree with the preliminary data of Wigaeus et al (3) on styre ne oxide levels in styreneexposed volunteers.
Styrene glycol. The styrene glyco l concentrat ions in the reference samples were genera l1y below O.6llmol . I-I. (One value of 0.7 u rnol . I-I was recorded.) The average concentration of the 17 exposed workers, 2.5 (range 0.6-6.3) um ol . I I, suggeste d styrene exposure levels averaging about 300 mg . rrr' in the work environment (3,4) .

Analysis of human samples
The resu lts of the monitoring of Hb adduct levels, styrene oxide, and styrene glyco l in blood and mandelic acid in the urine of the exposed subjects and refere nts are presented in table I.
Hemaglobin adducts. Figure 3 shows the mass fragmentograms of adducts to N-terminal valine in globin fro m one individual worker exposed to high air concentrat ions of styrene. The level s of hydro xyphe nethy lvaline in the globin samples are given as an average for the a and~chains on the assumption that the proportions of adducts to the two chains were the same in the samples and in the internal standard globin. The Hb adduct levels in the refere nts were <13 pmol . g', and the avera ge Hb adduct level in Mandelic acid. The concentration of mandelic acid in urine was meas ured in the exposed workers and in the refere nt from the plant (19) . The average concentration in the exposed subjects, 9.5 mmol . I-t, suggested an average exposure level of about 300 mg · m-3 (34) . This findi ng is in agree ment with the exposure estimate based on the styrene glycol conce ntrations.

Regression analysis
The corre lation between the parameters studie d was analyzed by a linear regression. (See table 2.) The analysis showed a correlation between styrene glycol in blood and mandelic acid in urine. There was also a strong correlation betwee n the Hb adduct levels and these expos ure parameters. Scand J Work Environ Health 1993, v0119, no 4 Table 1. Mandelic acid in urine, styrene oxide and styrene glycol in blood, and adducts to N-terminal valine in hemoglobin from workers exposed to styrene and from referents. (

Discussion
A method for analyzing hydroxyethyl and other lowmolecular weight adducts to N-terminal valine in Hb was developed by Tornqvist et al (15,16,25). We employed this method with a few modifications to allow quantification of the low adduct levels in styrene-exposed workers. An additional step for purification of the PFPTH derivative, involving extraction from methanol-water (60:40) to hexane, was introduced. This extraction removed a main fraction of the impurities (including PFPTH derivatives of more hydrophilic alkylvalines , such as hydroxyethyl-and hydroxypropyl-valine-PFPTH) still remaining in the PFPTH extracts washed according to Tornqvist et al (15,16). The enrichment of adducted a and~globin chain s on CM-Sepharose CL-6B prior to derivatization resulted in a substantial improvement in the analytic al sensitivity. The gain in sensitivity, if expressed in terms of the amount of protein in the parent chain compared with the amount of protein in the enriched fraction, was approximately tenfold . To compensate for recovery in all of the steps of the analytical procedure, a globin alkylated with e Hs)styrene oxide was used as an internal standard. The method was applied for the quantitative determination of Hb adducts in samples from workers exposed to styrene in the reinforced plastics industry and from matched referents . Free styrene glycol and styrene oxide (see footnote in table 1) in blood and mandeli c acid in urine were analyzed as measures of exposure. A correlation was found between the concentrations of blood metabolites and mandelic acid. The average values (2.5 umol . 1-1for styrene glycol and 9.0 mmol . I-I for mandelic acid) suggested an air concentration of about 300 mg . m-3 (about 75 ppm) of styrene in the work environment. The Hb adduct levels correlated with the concentration of styrene glycol in blood and mandelic acid in urine. The average Hb adduct level of hydroxyphenethylvaline was 28 pmol . s' for the seven exposed subjects.
The origin of the background levels of hemoglobin adduct s is not known , but these levels are probably produced as methodological artifacts rather than reflecting true adduct levels. Styrene has been measured in adipose tissue of nonoccupationally exposed persons (35). The average concentration in the tis-sue was estimated to correspond to an inhaled concentration of 0.476 ppm of styrene in the air. This finding indicates that styrene is present in unkno wingly exposed humans , a situation which may give rise to a background level of styrene oxide adducts to Hb. However, since 75 ppm gave a Hb adduct level of 30 pmol . g', the background level of 10 pmol . g' seems too high and may indeed reflect an artifact. This possibility will be studied furthe r.
The weekly uptake of styrene at 300 mg . rrr" is estimated to be 1.25 mmol . kg' I of body weight , on the assumption of an alveolar ventilation rate of 0.2 1 . min: ' (kg body weighty) and an absorpt ion of 90% of the compound from alveolar air. The steady state adduct level attained after prolonged exposure corresponds to nine weeks of cumulative uptake [ie, 11.2 mmol (kg body weight)"] and gives a value of 0.03 nmol . g-l for hydroxyphenethyl valine in Hb in humans . This adduct level is approximately 10 and 30 times lower than would have been obtained in rat and mouse, respectively, at a corresponding styrene uptake (19). This finding may demonstrate a relatively rapid detoxification of styrene oxide in humans.
Ethylene oxide is a known animal carcinogen (36). Studies at this laboratory (Hussain et al, unpublished) have demonstrated that styrene oxide has a somewhat (approximately twofold) higher chemical reactivity and a higher mutagenic effectiveness in Escherichia coli Sd-4 than ethylene oxide. If compared at equal adduct levels in deoxyribonucleic acid (DNA), the two compounds would give a similar mutagenic response. Thus a tentati ve risk estimation for styrene could be based on tissue doses of styrene oxide and the cancer-initi ating potency of ethylene oxide . Occupational exposure to ethylene oxide at 1 ppm (the occupat ional concentration limit in Sweden for new establishments) gives a steady state Hb adduct level of about 2.4 nmol . g:' for hydroxyethylvaline ( 10). According to the present study , occupational exposure to styrene at 100 ppm gives a value of approximately 0.04 nmol . g-). Thus exposure at 100 ppm of styrene and 0.02 ppm of ethylene oxide would give similar adduct levels .
Steady-state concentrations of epoxides during exposur e can also give a measure of dose. Data on environmental and instantaneous blood concentrations of ethylene oxide in exposed workers show that 26 1 Scand J Work Environ Health 1993. vol 19,no 4 e xposure at I ppm re sults in an average conce ntration of 0.13 umol . I-I for ethylene oxide in blood (37). Acco rd ing to this study exposure to styrene at 100 ppm gives about 0.09 umol . I-I for styrene oxide (upper estimate) (footnote in table I). Exposure at 100 ppm of styrene or 0.7 ppm of ethylene oxide would give the same epoxide blood concentrations. If an even di stribution of sty re ne oxide in blood is assumed (ie, similar concentrations inside and outside the erythrocytes), 100 ppm of styrene or 1.4 ppm of ethylene oxide would be expected to give sim ilar adduct levels. The discrepancy between expected and found adduct levels may be explained by a much lower concentration of styrene oxide in the erythrocyte s (where it is available for reaction with Hb) than in the plasma, where a main fraction of the compound could be noncovalently bound to proteins .