Biological markers in PAH exposed workers and controls

Mutation Research, 300 (1993) 231-239 © 1993 Elsevier Science Publishers B.V. All rights reserved 0165-1218/93/$06.00

231

MUTGEN 01900

Biological markers in PAH exposed workers and controls P. V a n H u m m e l e n

a, j . p . G e n n a r t b, j . p . B u c h e t h, R. L a u w e r y s b a n d M. K i r s c h - V o l d e r s a

a Laboratory of Human Genetics, Vrije Universiteit Brussel, Brussels, Belgium and b Industrial Toxicology and Occupational Medicine Unit, Universit# Catholique de Louvain, Brussels, Belgium (Received 13 January 1993) (Revision received 3 March 1993) (Accepted 3 March 1993)

Keywords: Biomonitoring; Polycyclic aromatic hydrocarbon; Cytogenetics; Hydroxypyrene; Thiocyanate; Micronuclei; Sister-chromatid exchange; High frequency cells

Summary Workers employed in a graphite electrode producing plant (n = 16) and a coke oven (n = 33) were compared with a control population of maintenance workers in a blast furnace (n = 54). The following parameters were analyzed: concentration of 13 different PAHs in the work environment measured by personal air samplers, concentration of hydroxypyrene in the urine, smoking habits (via urinary thiocyanate levels and a questionnaire) and cytogenetic aberrations in lymphocytes (SCE, HFC and MN). On the basis of PAH levels in the work environment and hydroxypyrene concentrations in the urine, the workers from the graphite electrode producing plant were the most exposed. However, statistically significant differences in SCE and HFC and positive correlations between the cytogenetic markers and airborne PAH levels on the one hand, and urinary hydroxypyrene concentrations on the other hand were only detectable in the workers from the coke oven with a lower exposure. No statistically significant effect of smoking was observed. As to the inter-comparison of the different cytogenetic markers, one may consider that SCE and HFC are more sensitive than MN frequencies for the biomonitoring of exposure to PAHs. Whether MN or SCE are the best biomarker for risk assessment of cancer and whether the presence of PAHs in the work environment is really responsible for the cytogenetic effects found in this study could not be ascertained.

Exposure to coal tars, coke oven emissions and similar products of coal combustion or carbonization processes have been shown to increase the

Correspondence: P. Van Hummelen, Labo voor Antropogenetica - Fac. Wetenschappen, Vrije Universiteit Brussel, Pleinlaan 2, 1050 Brussels, Belgium.

risk of lung cancer and possibly kidney, bladder and other cancers (Henry, 1937; Kawai et al., 1967; Kennaway and Kennaway, 1936; Doll, 1952; Lloyd, 1971; Redmond et al., 1976; NIOSH, 1973). A variety of mutagenic and/or carcinogenic polycyclic aromatic hydrocarbons (PAHs), amines, heterocyclic compounds, phenols and trace metals have been identified in coal liquefaction and gasification emissions.

232 As far as genotoxic effects from occupational exposure to PAHs are concerned, only few data are available. Miner et al. (1983) have reported an increased sister-chromatid exchange (SCE) frequency in lymphocytes from steel-workers employed on coke ovens. Bender et al. (1988) have confirmed that coke oven workers have statistically significantly elevated levels of SCE and chromatid aberrations. Recently, a new methodology for screening micronuclei (MN) in peripheral lymphocytes was developed (Van Hummelen and Kirsch-Volders, 1990); it allows the identification of first mitotic cells as binucleates after treatment with cytochalasin-B (Fenech and Morley, 1985). This study reports on the frequencies of SCE and of MN in PAH exposed workers and controls in relationship with the intensity of exposure to 13 different airborne PAHs and the urinary excretion of 1-hydroxypyrene, which is considered a good indicator of the internal level of exposure to PAHs (Buchet et al., 1992). Materials and methods

Studied population One hundred and three workers employed in a graphite electrode producing plant (plant A; n = 16), a coke oven (plant B; n -- 33) and the maintenance work place of a blast furnace (plant C; n = 54) volunteered to participate in the study. In plant B seven subjects, working at the top of the coke oven, are specially monitored because of the expected higher level of exposure. All workers belonged to the same social class.

Exposure assessment Historical data, engineering diagrams, work place, concurrent personal monitoring, worker records and questionnaire data were used to select the exposed workers. Each worker provided a urine sample before and after the work shift during which the airborne concentration of PAHs was monitored with a personal sampling system.

Personal monitoring Light PAHs with two or three rings are mainly present in the vapor phase and can be retained on an adsorbent such as Chromosorb 102 while

heavier compounds are mainly retained on a filter as particulates. Each worker was thus equipped with a personal air sampler. Air was aspirated at a flow rate of 2 l/min by a battery powered air sampling pump (model 224 PCEXR3, SKC Inc., Eighty Four, PA, USA). Particles were retained on a glass microfiber filter (GFF from Whatman, diameter 3.7 cm) and vapors were adsorbed on Chromosorb 102, contained in a tube (SKC Inc.) placed between the filter and the pump. Tubes and filters were kept in a refrigerated dark room until analysis which occurred within 2 weeks. The air sampling lasted up to 6 h but in case of heavy pollution, several 1 or 2 h successive samples were collected. Blood samples were taken at the end of the work shift for cytogenetic analyses.

Questionnaire Interviews were conducted by us just before blood samples were obtained. A pre-tested questionnaire was administered to each participant in the study. In addition to demographic information, the questionnaire contained items about use of protective equipment, non-routine events in the work place, specific descriptions of tasks and work areas, duration of current and past employment and prior occupational exposures to potential carcinogens and mutagens. Because cigarette smoke is a source of PAHs, a detailed smoking history was obtained from each participant. This included smoking status (current smoker, former smoker, non-smoker), years smoked, packs/day, pipe smoking, cigar smoking and passive smoking. Dietary history focused on caffeine, chips, vegetables, fruit, grilled meat and alcohol. Specific questions were asked about medications and health status (viral infections), vaccinations, chemotherapy, phototherapy, and diagnostic or therapeutic X-rays.

Sampling A venous blood sample (45 ml) was collected at the end of the work shift. Blood samples were collected in calparine containing tubes and delivered to the collaborating laboratory within 4 h of collection. All assays were run on coded samples with laboratories 'blind' as to subject identity or status (control/exposed).

233

Measurements of PAils in air samples The adsorbent sections and the filter were extracted with methanol. Methanol was chosen because it is readily miscible with the chromatographic mobile phase and does not interfere with the detection of PAH (Lankmajr and Miiller, 1979). Thirteen PAHs (naphthalene, fluorene, phenanthrene, anthracene, fluoranthene, pyrene, benz[a]anthracene, chrysene, benzo[e]pyrene, perylene, benzo[a]pyrene, dibenz[a,h]anthracene and benzo[ g,h,i]perylene) were measured by high pressure liquid chromatography and fluorescence detection. A ChromSpher PAll column (200 × 3 mm, L x ID) from Chrompack was used. Elution was performed at a flow rate of 0.5 ml/min with a gradient of water/acetonitrile delivered by two high pressure pumps model 302 and 305 from Gilson Medical Electronics (Villiers le Bel, France). At the start, the eluent was a 50/50 mixture; it changed to pure acetonitrile in 15 min; pure acetonitrile was delivered for 12 additional minutes and its concentration was then reduced to the 50/50 mixture in 2 min; a re-equilibration of the column was allowed during 6 min before the next analysis. Samples were injected with an autosampler Marathon (Spark, Holland). The detection of PAHs was performed by fluorimetry (821-FP Jasco spectrofluorometer, Gynkotek, Germering bei Miinchen, Germany); for naphthalene, fluorene and phenanthrene the excitation and emission wavelengths were 280 and 350 nm, for the other compounds 305 and 430 nm (Das, 1984). Standard PAH mixtures (Alltech Associates, Deerfield, IL, USA) were used as reference materials. The tube breakthrough was considered to have occurred when the second sotbent section of the Chromosorb tube contained more than 10% of the amount collected in the first section; in this case, the data were discarded.

Hydroxypyrene determination in urine The measurement of urinary hydroxypyrene excretion was carried out by the technique of Jongeneelen et al. (1987). This technique included a prior enzymatic hydrolysis of the conjugates with a mixture of glucuronidase and sulfatase; the next step was a sample purification on octadecyl silica cartridges (Bond-Elut 100 mg,

Analytichem, Harbor City, CA, USA) which was performed with an Aspec System from Gilson. Chromatography was performed with an ET 2 5 0 / 8 / 4 Nucleosil 10C18 column (MachereyNagel, Diiren, Germany); two Gilson pumps (model 302 and 305) delivered, at a flow rate of 1 ml/min, either pure acetonitrile (solvent A) or 10% acetonitrile in water containing 0.5 ml glacial acetic acid per liter (solvent B) to make the following elution gradient: from 100 to 20% solvent B in 25 min, from 20 to 0% solvent B in 2 min, isocratic (pure solvent A) elution during 3 min, back to 100% solvent B in 2 min and reequilibration of the column during 8 min before the next analysis. Hydroxypyrene was detected by fluorimetry (excitation: 240 nm; emission: 390 nm; 821-FP Jasco spectrofluorometer). The calibration was made with a control urine spiked with 10 mg hydroxypyrene/1 (Janssen, Beerse, Belgium); the detection limit was 0.2 mg/l. The results were standardized for 1 g urinary creatinine.

Measurements of creatinine and thiocyanate in urine Urinary creatinine was determined by the colorimetric method of Jaff6 using a Technicon RA1000 automate (Tarrytown, NY, USA). The urinary thiocyanate concentration was measured by the technique of Pettigrew and Fell (1972) and was expressed in m g / g creatinine.

Cytogenetic methods (a) Preparation of slides for analysis of SCE. Human peripheral lymphocytes were grown for 72 h in RPMI 1640 medium (with L-glutamine and 25 mM Hepes buffer: Gibco Laboratories) containing 15% fetal calf serum (FCS), phytohemagglutinin, 57 mM BrdU and, for the last 90 min, colcemid at 0.2 mg/ml. The cell suspension was then centrifuged at 1500 rpm for 10 min, resuspended and incubated at 37°C for 20 min in 0.075 M KCI. The pellet was fixed twice in methanol:acetic acid (3: 1). Fixed cells were dropped onto wet slides and air dried. All cultures containing BrdU were grown in the dark. Differential staining of chromatids was produced in the following manner: slides were stained with

234 Hoechst 33258 (0.5 / z g / m l ) for 10 min and exposed to U V light (366 nm) for 25 min at 13 cm lamp distance, then incubated for 15 min in 2 x SSC (60°C) and stained with 5% Giemsa (10 min), rinsed, and air dried.

(b) Preparations of slides for analysis of MN. Whole blood cultures (5 ml) were set up in H a m ' s F-10 containing 15% FCS and stimulated with phytohemagglutinin. After 44 h, cytochalasin-B in a final concentration of 6 t z g / m l was added. After harvest, the cultures were washed twice in R P M I 1640 containing 2% FCS. The pelleted cells were thoroughly re-suspended in 10 mi of hypotonic solution consisting of 1 part R P M I and 4 parts distilled water and 2% FCS, final concentration. After centrifugation, the supernatant was almost completely removed and smears were made. The slides were left to dry for 24 h. For the scoring of micronuclei, the smears were fixed in a m e t h a n o l - a c e t i c acid ( 3 / 1 ) solution and stained in 5% Giemsa (Van H u m m e l e n and KirschVoiders, 1990).

(c) Analysis. All cultures were made in duplicate. For SCE analysis, 50 second division metaphases, of good quality (from two cultures), per person were analyzed for SCE, and data are presented as the number of SCE per cell. The value of S C E / c e l l corresponding to the 95th percentile (according to Walsh, 1962) of the pooled data from the control population is used as the threshold value for the definition of high frequency cells (HFC). H F C data are then expressed as the percentage of high frequency cells per person. For the MN analysis, per person 2000 binucleated lymphocytes (from two cultures) were examined for the presence of one, two or more MN, and data are presented as per mill micronucleated binucleates. In addition the percentages of binucleated cells, polynucleated cells, metaphases and mononucleated cells with M N were recorded as a control for good growth. All slides were coded and analyzed with a Zeiss or Leitz microscope at a magnification of 1250 × .

Statistical analysis The distribution of all measured parameters showed significant departures from the normal distribution. Therefore, statistical analysis between biomarkers on the one hand, and between biomarkers and P A H measurements on the other hand was performed using non-parametric statistics. In the case of two independent samples, a two-tailed Mann-Whitney U-test was chosen and a Spearman rank correlation coefficient as a non-parametric measure of correlation. For analyzing multiple factors, in the case of the interaction of smoking habits on the occupational exposure, a two-way A N O V A was performed; the ANOVA-test, however, uses a testing hypothesis about means and variances and assumes that the data are normally distributed and the populations have equal variances. The SCE and MN data are not statistically significantly different from a simple Poisson distribution; this was tested by means of a dispersion test (Snedecor and Cochran, 1980). Therefore the analysis of variance was carried out on the square root of the values (Cooke et al., 1989). In order to give a convenient summary of the experimental results in the tables, means and standard errors (SE) of the non-transformed data are presented. Results Characteristics of the studied population are summarized in Table 1. The mean age of the workers was about 40 and was not significantly different between the three plants. Workers from plant A had a slightly higher state of duty com-

TABLE 1 CHARACTERISTICS OF THE POPULATION (MEAN + SD) Controls 54 42.9+6.8 -

Number of workers Age (years) Years of exposure Smoking habits: current + former smokers 40 never smokers 14

Plant A 16 40.35:11.2 13.0+ 4.7

PlantB 33 40.1+9.5 11.3+6.8

12 4

26 7

A, B: exposed workers; SD: standard deviation.

235 TABLE 2 PAH A I R B O R N E LEVELS ( / z g / m 3) A N D U R I N A R Y H Y D R O X Y P Y R E N E (HP) C O N C E N T R A T I O N S (PRE- A N D POSTSHIFT SAMPLING: I N / z g / g C R E A T I N I N E ) IN T H E D I F F E R E N T G R O U P S OF W O R K E R S Workers (mean _+SE) A+B

A

Mann-Whitney U test (p-values) B

Control

Total

HE

A+B

A

B

LE

Total

HE

LE

19.655:7.32 11.33_+2.29 23.70+10.80 90.105:44.12 5.575:1.30 1.33_+0.25 tl.0001 * 0.0001 * 0.0001 * 0.0001 * 0.0001 * 49 n = 16 n =33 n = 7 n = 26 n = 54 1.30-+0.47 3.18-+1.48 0.51_+0.08 0.60_+0.29 0.48-+0.08 0.49-+0.06 0.1376 0.0004 * 0.8940 0.6617 0.3336 HP Pre n = 47 n =14 n = 33 n = 7 n = 26 n = 53 HP post 2.47_+0.87 6.25_+2.55 0.75_+0.17 1.65_+0.65 0.51_+0.10 1.18_+0.34 0.0290 * 0.0001 * 0.7006 0.0771 0.3934 n = 48 n = 15 n = 33 n= 7 n = 26 n =48

PAH

n=

Mean ( + SE) values and statistical comparison (p-values) with control workers. A, B: exposed workers; n: number of subjects; HE: subjects working at the top of the coke oven (high PAH exposure); LE: low PAH exposure. * Statistically significantly different from the control population ( p < 0.05).

pared with plant B. The proportion of smokers (= current and former smokers)/never smokers was not significantly different between the three plants. The mean total PAH airborne levels and the mean urinary hydroxypyrene concentrations are presented in Table 2. PAH levels were statistically significantly higher in the exposed areas (plants A and B) compared to the control plant. The mean value and standard error of PAH concentration was high for plant B because of two very high measurements (148 and 333/zg/m 3) although workers from plant B had a significantly lower exposure than workers from plant A. Based

on the urinary hydroxypyrene measurements, only workers from plant A could statistically be distinguished from the controls, and this for pre- as well as for post-shift sampling. The mean number of SCE/cell and the frequency of HFC and MN are presented for the different groups of workers in Table 3. A statistically significant difference in the biomarkers between exposed persons and controls was only found in the frequency of SCE for the low exposed workers from plant B. The frequency of MN was statistically significantly lower for the high exposed workers from plant B than for the controls.

TABLE 3 C Y T O G E N E T I C E N D P O I N T S IN T H E D I F F E R E N T G R O U P S O F W O R K E R S BioM

Workers (mean + SE) A+B

SCE

4.27+0.15 n =36 H F C ( % ) 5.05+0.92 n = 36 M N ( % o ) 4.06+0.34 n=

44

A

Mann-Whitney U test (p-values) B

Control

A+B

Total

HE

LE

3.77+0.10 n = 10 2.66+0.73 n = 10 3.91+0.39

4.46+0.20 n =26 5.96+1.21 n = 26 3.915:0.43

4.37+0.54 n= 4 5.33+2.26 n =4 2.73+0.72

4.48+0.22 n = 22 6.08+1.38 n = 22 4.23+0.49

3.895:0.12 0.0681 n = 42 4.425:0.93 0.4957 n =42 4.805:0.35 0.3863

n = 16

n =33

n = 7

n = 26

n=53

A

B Total

HE

LE

0.8526

0.1~23"

0.3699

0.0182 *

0.5605

0.1624

0.4976

0.2855

0.2725

0.1340

0.0310"

0.4333

Mean (_+ SE) values and statistical comparison ( p - v a l u e s ) w i t h control workers. BioM: biomarkers; A, B: exposed workers; n: number of subjects; HE: subjects working at the top of the coke oven (high PAH exposure); LE: low PAH exposure. * Statistically significantly different from the control population ( p < 0.05).

3.89+0.19 n=9 2.00+1.05 n=9 3.414-0.53 n=11

4.404-0.19 n=27 6.064-1.12 n =27 4.054-0.37 n=38

3.644-0.18 n=4 0.504-0.50 n =4 4.00+0.35 n=4

3.86+0.12 n=6 4.104-0.68 n =6 3.884-0.52 n=12

smok 4.09+0.30 n=5 3.204-1.74 n =5 3.084-0.80 n=7

Nsmok

B

4.55+0.23 n=21 6.62+1.41 n = 21 4.134-0.49 n=26

smok 3.63+0.14 n=10 2.91+0.84 n =10 4.614-0.59 n=14

Nsmok

Control exp

smok

inter

exp

A smok

inter

B exp

smok

inter

3.97-1-0.14 0.1047 0.0569 0.7259 0.9176 0.2799 0.8429 0.0467 * 0.1406 0.8559 n=32 4.894-1.18 n =32 4.864-0.43 0.0598 0.4002 0.6295 0.4514 0.9498 0.7368 0 . 0 4 6 0 * 0.2541 0.3996 n=39

smok

Two-way A N O V A (p-values) A+B

M e a n ( + SE) values a n d statistical analysis. BioM: biomarkers; A , B: exposed workers; n: n u m b e r of subjects; exp: exposure; Nsmok: n e v e r smokers; smok: c u r r e n t a n d f o r m e r smokers; inter: interaction between smoking a n d exposure. * Statistically significantly different f r o m the control p o p u l a t i o n ( p < 0.05).

HFC (%) MN (%o)

SCE

Nsmok

Nsmok"

smok

A

A+B

BioM Workers ( m e a n + SE)

C Y T O G E N E T I C E N D P O I N T S IN T H E D I F F E R E N T G R O U P S O F W O R K E R S S T R A T I F I E D B Y S M O K I N G H A B I T S

TABLE 4

237

The cytogenetic endpoints (SCE, HFC and MN) did not show any correlation with the total PAH airborne levels (Spearman rank correlation), nor with the 13 different individual PAH measurements. Correlations between any of the cytogenetic endpoints and mean urinary hydroxypyrene concentrations before the work shift did show a statistically significant difference; however, the correlation was only significant for the workers of plant B and controls, or for groups containing those workers. End of shift sampling was only positively correlated with SCE for plant B and controls, and HFC frequencies in the control workers. Concerning the correlations between the different cytogenetic endpoints, SCE and HFC were highly significantly correlated but between SCE or HFC and MN no significant correlation was found. An effect of smoking has been found with a statistically significantly difference (p = 0.0350) in the frequency of HFC after grouping all workers in smokers and never smokers. However, after subdividing the workers in controls and exposed subjects (plant A + B), the significant difference was only observed again in the exposed workers but not in the controls. Therefore a two-way ANOVA was performed to evaluate the effect of exposure and smoking habits together (Table 4). The ANOVA was performed on the square root transformed data only for SCE and MN, and not on the HFC. HFC is a mathematical derivative from the frequency of SCE (___95 percentile) and cannot be normalized by any transformation. The same results were obtained as in Table 3: exposure was detected only for the workers in plant B for SCE, and the MN frequency of exposed workers from plant B was significantly lower than for the controls. An effect of smoking has not been detected. However, smokers and never smokers showed a statistically significant difference in urinary thiocyanate levels (pre-shift: p =0.0003; post-shift: p=0.0012). For the urinary thiocyanate levels, pre-shift as well as post-shift sampling were significantly correlated with SCE (pre-shift: p = 0.0497; post-shift: p = 0.0190) and HFC (pre-shift: p = 0.0064; post-shift: p = 0.0409). On the other hand the thiocyanate levels were also highly correlated with the concomitant

urinary hydroxypyrene levels (pre-shift: p = 0.0015; post-shift: p = 0.0001). Dietary history (e.g., caffeine, chips, vegetables, fruit, grilled meat and alcohol) did not induce a statistically significant effect for any of the cytogenetic markers. Discussion

Based on the air monitoring and metabolite detection in the urine workers from the graphite electrode producing plant (plant A) were by far the group with the highest exposure. In the coke oven plant (plant B) a few individuals, working on the top of the coke oven, showed peak values in airborne PAH levels, but no concomitant peak levels in urinary hydroxypyrene. Using these types of measurements, exposure to PAHs can effectively be monitored. Air measurements are indicative of the potential exposure, but screening for a metabolic marker in the urine is clearly an internal indicator of exposure, as was already concluded in a previous paper (Buchet et al., 1992). The aim of this work was to perform a cytogenetic survey to monitor the incidence of cytogenetic changes which is a real measure of the effect of internal exposure. The cytogenetic data indicated a significant difference between exposed and control groups; however, this is limited to SCE and to the comparison between control workers and low exposed workers from plant B. All three cytogenetic markers were positively correlated with the urinary hydroxypyrene data, but it was again limited to workers from plant B and controls. Correlations between cytogenetic data and the concentrations of the 13 different PAHs, separately or pooled, in the air have not been observed. As to the influence of smoking, no differences in the frequency of cytogenetic damage were found although smokers differed significantly from the non-smokers on the basis of thiocyanate levels in the urine. Two-way analysis of variance (exposure versus smoking) did not reveal any indication of the influence of smoking on the cytogenetic endpoints. It is clear that conflicting results were obtained in this study; not high, but low PAH exposed workers, on the basis of air or urine measure-

238

ments, were distinguished from controls using cytogenetic markers. Additionally, only for low and not for high exposure was a positive correlation observed between the urinary hydroxypyrene and the cytogenetic markers. One might therefore consider that the PAHs in the environment are not responsible for the observed cytogenetic effects, but that other confounding factors in the work environment (e.g., aromatic amines in the coke oven) may have caused this effect; or that inter-individual metabolic differences might be too high to define exposure on the basis of urine measurements accurately. Miner and co-workers (1983) also found higher SCE frequencies in coke oven workers compared with controls but this was only based on a comparison between groups and correlations with other external or internal PAH measurements were not performed. It would be useful to assess whether other markers (e.g., hemoglobin or DNA adducts) for the internal dose level of some genotoxic PAHs might be better correlated with the cytogenetic endpoints than hydroxypyrene in urine. Data from the literature (Harris et al., 1985; Perera et al., 1988; Savela et al., 1989; Szyfter and Hemminki, 1992) on aromatic adducts determined in white blood cell DNA of workers with the 32p-postlabeling technique, noted some instability of the adducts, upon storage at -20°C, large inter-individual variations in the levels of adducts and the role of host factors (e.g., metabolic capacities) in the control of adduct levels. The comparison between P450 genetic complement, cytogenetic endpoints and hemoglobin/DNA adducts is therefore a promising area to understand the inter-individual variation which troubles the biomonitoring studies. As to the inter-comparison of the different cytogenetic markers, one might consider that SCE and HFC are more sensitive than MN frequencies for biomonitoring exposure to PAHs. Whether MN or SCE are the best biomarker for risk assessment of cancer is, however, not established. The micronucleus incidence, which is a clastogenic and/or aneugenic endpoint, has of course a specificity different from SCE, and may therefore be important in other, or very specific, cases, e.g., occupational exposure to vinyl chloride monomer (Sinu6s et al., 1991).

Acknowledgements The authors wish to thank Christel Du Rang (VUB) and Ilse Goris (VUB) for their substantial contribution to the cytogenetic analysis, and also D. Bosmans (UCL) and B. Frik (UCL), for their skilful technical help. Special thanks also to Dr. Roland Hauspie (VUB) for the help and criticism concerning the statistical analysis. This research is supported by the Belgian Incentive Program Health Hazards initiated by the Belgian Science Policy Office.

References Bender, M.A, R.C. Leonard, O. White Jr., J,P. Costantino and C.K. Redmond (1988) Chromosomal aberrations and sister-chromatid exchanges in lymphocytes from coke oven workers, Mutation Res., 206, 11-16. Buchet, J.P., J.P. Gennart, F. Mercado-Calderen, J.P. Delavignette, L. Cupers and R. Lauwerys (1992) Evaluation of exposure to polycyclic hydrocarbons in a coke production and a graphite electrode manufacturing plant and assessment of the urinary excretion of 1-hydroxypyrene as a biological indicator of exposure, Br. J. Ind. Med. 49, 761-768. Cooke, D., J. Allen, M.G. Clare, C.J. Dor6 and L. Henderson (1989) Statistical methods for sister chromatid exchange experiments, in: D.J. Kirkland (Ed.), Statistical Evaluation of Mutagenicity Test Data, Cambridge University Press, Cambridge. Das, B.S. (1984) Applications of HPLC to the analysis of polycyclic aromatic hydrocarbons in environmental sampies, in: J.F. Laurence (Ed.), Liquid Chromatography in Environmental Analysis, Humana Press, Englewood Cliffs, NJ. Doll, R. (1952) The causes of death among gas-workers with special reference to cancer of the lung, Br. J. Ind. Med., 9, 180-185. Fenech, M., and A.A. Morley (1985) Solutions to the kinetic problem in the micronucleus assay, Cytobios, 43, 223-246. Harris, C.C., K. Vahakangas, M.J. Newman, G.E. Trivers, A. Shamsuddin, N. Sinapoli et al. (1985) Detection of benzo[a]pyrene diol epoxide-DNA adducts in peripheral blood lymphocytes and antibodies to the adducts in serum from coke oven workers, Proc. Natl. Acad. Sci. USA, 82, 6672-6676. Henry, S.A. (1937) The study of fatal cases of scrotal cancer from 1916-1935 in relation to occupation, with special reference to chimney sweeping and cotton mule spinning, Am. J. Cancer, 31, 28-57. Jongeneelenl F.J., R.B.M. Anzion and P.Th. Henderson (1987) Determination of hydroxylated metabolites of polycyclic aromatic hydrocarbons in urine, J. Chromatogr., 413, 227232.

239 Kawai, M., H. Amamoto and K. Harada (1967) Epidemiologic study of occupational lung cancer, Arch. Environ. Health, 14, 859-864. Kennaway, N.M., and E.L Kennaway (1936) A study of the incidence of cancer of the lung and larynx, J. Hyg., 36, 236-267. Lankmajr, E.P., and K. Miiller (1979) Polycyclic aromatic hydrocarbons in the environment, high-performance liquid chromatography using chemically modified columns, J. Chromatogr., 170, 139-146. Lloyd, J.W. (1971) Long-term mortality study of steelworkers: V. Respiratory cancer in coke plant workers, J. Occup. Med., 13, 53-68. Miner, K.J., W.N. Rom, G.K. Livingston and J.L. Lyon (1983) Lymphocyte sister chromatid exchange (SCE) frequencies in coke oven workers, J. Occup. Med., 25, 30-33. National Institute for Occupational Safety and Health (NIOSH) (1973) Criteria for a recommended standard: Occupational exposure to coke oven emissions, DHEW Publ. No. 73-11016, Washington, DC. Perera, F.P., K. Hemminki, T.L Young, D. Brenner, G. Kelly and R.M. Santella (1988) Detection of polycyclic aromatic hydrocarbon-DNA adducts in white blood cells of foundry workers, Cancer Res., 48, 2288-2291. Pettigrew, A.R., and G.S. Fell (1972) Simplified colorimetric determination of thiocyanate in biological fluids, and its application to investigation of the toxic amblyopias, Clin. Chem., 18, 996-1000.

Redmond, C.K., B.R. Strobino and R.H. Cypess (1976) Cancer experience among coke by-product workers, Ann. NY Acad. Sci., 271, 102-115. Savela, K., K. Hemminki, A. Hewer, D.H. Phillips, K.L. Putman and K. Randerath (1989) Interlaboratory comparison of the 32p-postlabelling assay for aromatic DNA adducts in white blood cells of iron foundry workers, Mutation Res., 224, 485-492. Sinu6s, B., A. Sanz, M.L Bernal, A. Tres, A. Alcala, J. Lanuza, C. Ceballos and M.A. Saenz (1991) Sister chromatid exchanges, prolifating rate index and micronuclei in biomonitoring of internal exposure to vinyl chloride monomer in plastic industry workers, Toxicol. Appl. Pharmacol., 108, 37-45. Snedecor, G.W., and W.G. Cochran (1980) Statistical Methods, 7th edn., Iowa State University Press, Ames, IA. Szyfter, K., and K. Hemminki (1992) Analysis of deoxyribonucleic acid adducts in workers, Scand. J. Work Environ. Health, 18, 22-26. Van Hummelen, P., and M. Kirsch-Volders (1990) An improved method for the in vitro micronucleus test on human lymphocytes, Mutagenesis, 5, 203-204. Walsh, J.E. (1962) Non-parametric confidence intervals and tolerance regions, in: A. Sarhan and B. Greenberg (Eds.), Contributions to Order Statistics, Wiley, New York, pp. 136-419.