Ann. occup. Hyg., Vol. 47, No. 5, pp. 349-378, 2003
© 2003 British Occupational Hygiene Society
Published by Oxford University Press
Review |
Monitoring Human Occupational and Environmental Exposures to Polycyclic Aromatic Compounds
1 Z. Jansestraat 46 II, 1097 CN, Amsterdam, The Netherlands; 2 Syngenta Central Toxicology Laboratory, Alderley Park, Macclesfield, Cheshire SK10 4TJ, UK
Received 12 June 2002; in final form 10 February 2003
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONPAC ADDUCT DETERMINATIONS IN...EXAMPLES OF MONITORING HUMAN...ANALYSIS OF THE DATAINHALATION VERSUS DERMAL...DISCUSSIONCONCLUSIONSAPPENDICESREFERENCES |
|---|
Many research groups have been carrying out studies to develop biomarkers of exposure to polycyclic aromatic compounds (PACs) and apply these for human biomonitoring. The main objectives of the use of biomarkers are to determine specific occupational and environmental exposures to monitor the effectiveness of exposure controls and prediction of the risk of disease. This article presents a review of the literature in the field of biomarkers of human exposure to PACs and an evaluation of the relevant biomarkers for monitoring exposure to PACs in a range of exposure situations from coke ovens to bitumen handling and environmental exposures. For this evaluation, the relationships between external PAC exposures and the corresponding biomarker levels have been studied. The literature data indicate that urinary excretion of 1-hydroxypyrene correlates well with external PAC exposure and this compound appears to be a suitable marker for internal exposure to PACs. DNA adducts, mostly measured in white blood cells, do not show satisfactory correlations with exposure to PACs in a variety of workplace and exposure situations. It is not clear which factors are mainly responsible for this poor correlation. Micronuclei and sister chromatid exchanges measured in peripheral white blood cells are also unsatisfactory as biomarkers for PAC exposure. From the relatively limited data available, chromosome aberrations appear to show considerable promise as indicators of exposure to PACs. Because of their strong association with cancer, chromosome aberrations are considered suitable indicators of increased cancer risk arising from exposure to PACs.
Keywords: environment; human exposure; monitoring; occupation; polycyclic aromatic compounds
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONPAC ADDUCT DETERMINATIONS IN...EXAMPLES OF MONITORING HUMAN...ANALYSIS OF THE DATAINHALATION VERSUS DERMAL...DISCUSSIONCONCLUSIONSAPPENDICESREFERENCES |
|---|
Polycyclic aromatic compounds (PACs) are classes of compounds found in crude oils, mineral oils, bitumens and tars (Gelboin and Ts'o, 1978; IARC, 1985, 1987). They are also formed during the incomplete combustion of fossil fuels and oil products. As a result, PACs are compounds that are widely distributed in the environment, principally through combustion processes and spillages of oil-derived materials that contain these compounds. They are highly lipophilic non-polar persistent substances and have been detected in aquatic environments and in many wild organisms. Many industrial processes, such as coke oven operations, aluminium smelting, asphalt paving and roofing operations, give rise to environmental releases and occupational exposures to these compounds (reviewed in Brandt et al., 1985; Herbert et al., 1990; Hemminki et al., 1990a,b, 1994; Dor et al., 1999). Oil well fires, automotive exhaust gases, tobacco smoking and dietary intake also represent sources of human exposure (Darcey et al., 1992; Hemminki et al., 1994). The main concern about such human exposures is that many PACs have mutagenic and carcinogenic properties (Lijinsky, 1991). Although in occupational settings people are exposed to mixtures of PACs that also contain heteroaromatic compounds, for monitoring purposes mostly one, often benzo[a]pyrene (BaP), or a small subset of unsubstituted polycyclic aromatic hydrocarbons (PAH) are used to represent the whole set of PACs. The term PAH refers specifically to those PACs that consist only of hydrogen and carbon, unsubstituted and substituted by alkyl (mostly methyl) side chains.
Analytical techniques for detecting PACs continue to increase in sensitivity, thus progressively lowering the detection limits for these compounds in the environment, animals and humans. Because of the very low levels detected, it is difficult to assess with certainty the possible increased cancer and health risks associated with these low level exposures. Until recently, the detection and identification of human carcinogens were based exclusively on the analysis of tumour incidences in human epidemiological studies. However, for exposures to PACs, it is only in instances of specific and high exposures that these methods have identified the causative agents, for example, coal tar pitch volatiles (CTPV). Because of the long latency for the onset of human cancer, the occurrence of mixed exposures and the low resolving power and retrospective nature of epidemiology, alternative methods are needed to make prospective risk assessments. A key factor in the risk assessment process is the determination of the dose. Biomonitoring offers an approach to determine an individual’s received dose. This article reviews the principal methods that have been used to monitoring exposures to PACs and discusses how well the measurements correlate with the external PAC exposures and the potential for estimating cancer risk.
Exposures to PACs
Human exposure to foreign compounds (xenobiotics), such as PACs in bitumen fumes, can occur via several routes, e.g. inhalation, skin deposition/contamination or orally. In general, exposures are proportional to the concentration of the compound in the working environment. The environmental concentration can be measured by taking static samples of the ambient air. Methods for measuring skin deposition involve taking skin wipes or the determination of the deposition on a skin pad, and for inhalation taking breathing air samples with a personal sampler near the breathing zone. Because numerous different sets of PAHs have been used to report external PAC exposures, which are very difficult to compare, in this article we use, whenever possible, the sum of eight (carcinogenic) 4–6 ring PAHs (
8PAHs: benz[a]anthracene, chrysene, benzo[b]-fluoranthene, benzo[k]fluoranthene, benzo[a]pyrene, dibenz[ah]anthracene, benzo[ghi]perylene and indeno[1,2,3-cd]pyrene) as a way of presenting PAC exposures. In addition, the often reported exposure to benzo[a]pyrene (BaP) is given where possible. 8PAHs is only a subset of the group of potentially carcinogenic PACs (4–6 ring PAHs, their sulfur- and nitrogen-containing analogues and their alkylated, mostly methylated, homologues). Nevertheless, 8PAHs can be used as a marker for the whole set because generally the concentrations of heterocyclic PACs show proportionality to the concentrations of the corresponding PAHs (Williams et al., 1986; Grimmer et al., 1987; Heinrich et al., 1994; Sivak et al., 1997; Binet et al., 2002).
The above mentioned measurements only give estimates of the external concentration that can potentially enter the body. In order to exert an effect, the compound has to enter the body, e.g. by passage through the skin or the lung tissue. Once in the body there are several possibilities for its detection. A first possibility is to look for the unchanged compound or its metabolites in urine, blood, faeces, exfoliated cells, sweat, nails, etc. For biomonitoring in the work environment, mainly urine and blood samples are used. The concentrations of the unchanged parent PACs in urine are generally so low that a concentration determination is seldom possible (Grimmer et al., 1994). Mostly, the concentration of one or more metabolites is high enough to be determined (Jongeneelen et al., 1988; Grimmer et al., 1994). Both phenols, dihydrodiols and tetrahydrotetrols can be found in this way. These compounds are called specific biomarkers of exposure, because they record that exposure to a particular compound has taken place. Sometimes the part of the molecule to which the metabolite is conjugated, e.g. thioethers (van Doorn and Henderson, 1979) or D-glucaric acid (Pasquini et al., 1989), is looked for. These conjugate fragments provide evidence of exposure to any electrophilic compound that the worker has been exposed to and are therefore called aspecific biomarkers of exposure. The most important value of these measurements is to signal that exposure to an electrophilic compound has taken place, regardless of the route(s) of exposure.
Bioactivation of PACs
Many xenobiotics are subject to mammalian metabolism (biochemical alteration), which is divided into two distinct phases.
In phase I, the major reaction is usually hydroxylation, catalysed by members of a class of enzymes referred to as monooxygenases or cytochromes P-450 (CYP). Other types of reactions in phase I include reduction and hydrolysis. In phase II, the hydroxylated or other metabolites produced in phase I are converted by specific enzymes to various polar metabolites by conjugation (binding) to, for example, glucuronic acid, sulphate, glutathione or certain amino acids (Murray, 1990). The overall purpose of these two phases of metabolism of xenobiotics is to increase their water solubility (polarity) and thus to facilitate their excretion from the body. If this did not occur, very water-insoluble (hydrophobic) compounds, such as PACs or PCBs, would remain in the tissues for very long times.
In certain cases, phase I metabolism converts chemically unreactive compounds into chemically and biologically highly reactive species. For example, PACs can be oxidized to electrophilic compounds that can react with macromolecules such as DNA or proteins to give adducts, before phase II conjugation can take place. If DNA adducts are not repaired prior to cell replication or are misrepaired, a permanent gene mutation can occur, which can lead to cancer development. Those compounds that can lead to DNA adduct formation are referred to as being genotoxic.
Biomarkers of exposure and of effect
Some biomarkers are indicative of the internal dose received, but do not persist in the body. These are, for example, hydroxylated metabolites in urine or faeces and those that indicate an effect in a specific target organ, e.g. DNA adducts or sister chromatid exchanges (SCE). Other biomarkers are indicative of irreversible effects, for example chromosomal aberrations (CA), micronuclei (MN) and mutations (van Delft et al., 1998).
Hydroxylated metabolites
These are compounds that are formed in phase I metabolism and are excreted, mainly as conjugates, in urine. They can be analysed after hydrolysis of the conjugates. Currently, 1-hydroxypyrene (1OHPy) is the metabolite that is used most frequently. Other metabolites that are occasionally used are the hydroxylated metabolites of phenanthrene, benz[a]anthracene and BaP. The ratio of phenanthrene metabolites to 1OHPy might be a useful indicator of the relative contribution of smoking in the presence of other PAH sources (Jacob et al., 1999). To monitor smoking itself, urinary cotinine is, however, a reliable and specific biomarker of exposure (Yang et al., 2001). Linear correlations have been found between the total amount of PAHs inhaled during a working day and the amount of the main metabolite(s) for phenanthrene, pyrene and BaP excreted in the corresponding 24 h urine (Grimmer et al., 1994). Correlations between atmospheric pyrene and urinary 1OHPy concentrations improved greatly (r = 0.74) if the amount of pyrene inhaled over the shift and the corresponding amount of urinary 1OHPy excreted were considered (Lafontaine et al., 2000). The fraction of unchanged PAH excreted was found to be much lower than the amount inhaled (Grimmer et al., 1994). For phenanthrene and pyrene the total amount of hydroxylated metabolites was between 10 and 15% of the amount inhaled, but for BaP this amount was much lower (0.6%) (see Table 1). In a study where hydroxylated metabolites in urine were used to monitor personal exposures to phenanthrene, pyrene, benz[a]anthracene and BaP, a correlation was not observed between the external exposure to PAHs and the internal exposure (Gundel et al., 2000).
|
1-Hydroxypyrene
Jongeneelen et al. (1987) introduced 1OHPy as a biomarker of internal dose arising from exposures to PACs. Subsequently Jongeneelen (2001) proposed three benchmark levels for measurements based on 1OHPy. These are as follows.
A no observed effect level equivalent to a measurement of 1.4 µmol 1OHPy/mol creatinine. This is the level below which non-smokers showed no increased level of high frequency (HF)-SCEs (Buchet et al., 1995).
The lowest observed level of genotoxic effects indicated by 1.9 µmol 1OHPy/mol creatinine for coke oven workers and 3.8 µmol/mol for aluminium plant workers.
A level that equates to the present occupational exposure limits for PACs (0.2 mg/m3 benzene soluble matter and/or 2 µg/m3 BaP). The value used is dependent on industry type and pyrene content of the exposure and is equivalent to 2.3 µmol 1OHPy/mol for coke oven workers and 4.9 µmol 1OHPy/mol for aluminium industry workers.
Further to this, Jongeneelen defined the different benchmark levels for coke oven workers and workers in the aluminium industry on the basis of the occurrence of different PAC profiles in fumes from the two industries. The average PAC exposure profile in the aluminium industry has more of the lower molecular weight PACs, including pyrene, than that in coke oven emissions (Dr F. J. Jongeneelen, personal communication, 2001).
In actual human monitoring studies, half-lives for urinary 1OHPy elimination of 4–48 h have been reported (Buckley and Lioy, 1992; Boogaard and van Sittert, 1994, 1995; Jongeneelen et al., 1988; Vu Duc and Lafontaine, 1999). In rodent studies, however, Jacob et al. (1989) showed that the elimination of 1OHPy is dependent upon the mode of administration. These factors suggest that when spot samples are taken, the time of sampling relative to the exposure time is important. Grimmer’s (Grimmer et al., 1994) and Lafontaine’s data (Lafontaine et al., 2000) suggest that a sampling strategy designed to collect 24 h urine samples gives the best relation between external PAC dose and 1OHPy excretion.
Urinary thioethers or D-glucaric acid
These polar fragments, to which hydroxylated metabolites may be conjugated, are aspecific biomarkers that provide evidence of exposure to all electrophilic compounds, not only PACs. They have been reported in only a few studies on occupational exposure to PACs, possibly because these are very difficult to interpret due to confounding factors.
DNA adducts
PAC–DNA adducts arise from the reactions of reactive oxidation products of PACs with DNA in various target organs, such as skin, lungs and liver. During occupational monitoring of workers only blood and exfoliated cells (cells cast off from the tissue surface) are normally available as surrogate tissues. DNA adducts can be determined by various methods, amongst which the 32P-post-labelling assay is most sensitive. Details of this method of analysis, which has found widespread use for PAC–DNA adduct determination, can be found in the Appendix (for a review see Beach and Gupta, 1992). Immunochemical methods for detecting DNA–carcinogen adducts are generally less sensitive than 32P-post-labelling, but are simple and are relatively inexpensive (Appendix). Both polyclonal and monoclonal antibodies have been generated against a variety of PAH carcinogen adducts, although some show considerable cross-reactivity (Booth et al., 1997). The most common procedures include enzyme-linked immunoabsorbent assay (ELISA), competitive radioimmunoassay (RIA) and ultra-sensitive enzymatic radioimmunoassay (USERIA). Details of the use of immunoassays and 32P-post-labelling methods for detecting carcinogen–DNA adducts have been reviewed by Poirier (1991) and Poirier and Weston (1991). Because the immunochemical methods often show cross-reactivity for a range of PAHs and related compounds, the end points measured are generally described as ‘PAH–DNA adducts’. In studies with 32P-post-labelling, however, which measures a much wider range of adducts, these are normally simply referred to as ‘aromatic DNA adducts’. The cross-reactivity of immunoassays reduces their value for the quantitative measurement of adducts, but nevertheless antibody based immuno-enrichment devices are useful in the pre-concentration of adducts (Booth et al., 1997; Bartsch, 1996). The post-labelling methodology provides more reliable quantitative information when used with reference standards. Phillips et al. (2000) have reviewed the strengths, limitations and potential for inter-laboratory variation of the various assays for DNA adducts. Inter-laboratory variation in the results from the post-labelling assay can occur due to differences in reagents, enzymes and the particular protocol used.
The implications of DNA adduct formation and detection
The reactions of chemicals with DNA in cells can cause the formation of DNA adducts, which can give rise to mutation. Mutated cells can sometimes lead to cancer. The number of adducts formed depends on the dose of the chemical and its reactivity or the reactivity of its metabolite(s). The formation of DNA adducts is the earliest critical event that can be observed in the multistage process of carcinogenesis caused by chemicals such as PACs. Single doses of potent chemical carcinogens when tested in animal studies typically give rise to levels of adducts in the range 1 adduct/103–105 normal DNA bases (see Steiner et al., 1992). This range corresponds to about 10 000–1 000 000 modified bases in the DNA per cell. The number of critical changes in a single gene required to fully initiate a cell is not known with certainty, but is considered to be in the range three to six, and these can occur in any sequence (McCormick and Maher, 1994; Hanahan and Weinberg, 2000). A level of DNA adduct formation of 
1 adduct/105 DNA bases corresponds to about 10 000 DNA base modifications per cell, as each cell contains about 109 DNA bases. At this level of adduct formation there is thus a reasonable statistical probability that mutation will be initiated. A level of 1 adduct/109 DNA bases, the typical detection limit of 32P-post-labelling, corresponds to only about one DNA base modification per cell. The statistical probability of this base modification occurring at critical sites in a single gene is clearly very low and therefore it is doubtful that this level of adduct formation has great significance in terms of cancer risk. Although it is doubtful that levels of DNA adducts below 1 adduct/108 DNA bases are significant, it is not possible to generalize (Mulholland et al., 1997). Different adducts may have different mutagenic and carcinogenic responses. Based on the premise that several DNA lesions are needed for the formation of a malignant cell, the linearity of the dose–risk relationship can be explained by the finding that it is unlikely that more than one mutation is induced by a specific agent, at the low dose rates that occur in human exposure environments. Therefore, the qualitative detection of low levels of adducts is viewed primarily as a positive indication of exposure.
Cytogenetic biomarkers
Sister chromatid exchanges (SCE)
SCE result from the interchange of DNA replication products within chromosomes between two sister chromatids at apparently identical loci, which are visualized in metaphase. The mechanism of SCE formation has not been established, but SCE seem to be a consequence of errors of DNA replication. The biomarker HFC (or HF-SCE) is defined as the percentage of cells with a high frequency of SCE. In most cases, the value of SCE/cell corresponding to the 95th percentile (according to Bender et al., 1992) of the pooled data from the control population is used as the threshold value for the definition of HFC.
Chromosome aberrations (CA)
CA are abnormalities of chromosome number or structure, visible in metaphase cells during cell division, and are, in general, a result of breakage of a chromosome or breakage and rejoining within or between chromosomes (also called exchanges). Because of their strong association with cancer, CA are viewed as indicators of increased cancer risk (see below).
Micronuclei (MN)
MN are small nuclear-like bodies in cells that have become separated from the main nucleus during cell division. In general, they contain a chromosomal fragment, due to breakage of a chromosome, or a complete chromosome that is left behind due to toxic effects on the spindle proteins involved in chromatid separation.
The European Study Group on Cytogenetic Biomarkers and Health has reported that CA have predictive value for cancer risk. People with high numbers of CA (a high CA frequency) appeared to be at elevated risk. No such elevated risk was observed for people with a low CA frequency. No association was seen between SCE or MN frequencies and subsequent cancer incidence/mortality (Bonassi et al., 1995; Hagmar et al., 1998). Typical values for the CA frequencies involved were: low, 
1%; high, >2.2 (median of values for the 10 laboratories in the Nordic countries) (Hagmar et al., 1994). For monitoring human exposure generally SCE, CA, MN and often also DNA adducts are determined in lymphocytes.
|
PAC ADDUCT DETERMINATIONS IN ANIMAL MODELS |
|---|
|
TOPABSTRACTINTRODUCTIONPAC ADDUCT DETERMINATIONS IN...EXAMPLES OF MONITORING HUMAN...ANALYSIS OF THE DATAINHALATION VERSUS DERMAL...DISCUSSIONCONCLUSIONSAPPENDICESREFERENCES |
|---|
From controlled animal studies one generally sees much better correlations between adduct levels and dose than in human exposure studies. Results from skin painting studies in rats with bitumen fume condensates (Genevois et al., 1996; Booth et al., 1997) show a positive dose–response effect (Fig. 1). However, the efficiency of DNA adduct formation (expressed as adducts per unit dose = adducts/108 nucleotides/mg PAC) decreases with increasing PAC dose (Fig. 2). The PAC fraction was expressed as the total 3–6 ring PAC determined by FIA-DMSO extraction (extraction with DMSO using flow injection analysis, coupled with LC/GC with FID detection for quantification) (Brandt et al., 1999). It has been shown for a wide range of oil products that the PAC fraction (for bitumen fume condensates 1 mg total 3–6 ring PACs roughly corresponds to 40 µg
8PAHs) is correlated with mutagenicity and carcinogenicity in rodent skin painting studies (Brandt et al., 1999).
|
|
In studies with bitumen fume condensates Booth et al. (1997) found that adduct levels in the lungs were five to seven times lower than those in the epidermis. Furthermore, the decrease in adduct formation efficiency with increasing dose was much less in the lungs than in the epidermis. An explanation for this effect may be that in the skin, the site of application, almost the full PAC dose is effective, whereas only a small fraction of the PACs reaches the lungs and is metabolized there.
In inhalation experiments with rats exposed by inhalation to coal tar aerosols, Lewtas et al. (1997) also found a decreasing tendency for lung DNA adduct formation with increasing PAC dose. In contrast, for some individual compounds (e.g. 2-acetylaminofluorene) linear correlations have been observed between chronically administered doses and the steady-state levels of DNA adducts in target organs. The steady-state levels of DNA adducts in such cases have been shown to be linearly related to the tumorigenic response (Beland and Poirier, 1993; Poirier and Beland, 1994). Because DNA adducts found in experimental animals can be detected in exposed humans, there is still the prospect of using DNA adduct data to estimate human cancer risk.
|
EXAMPLES OF MONITORING HUMAN PAC EXPOSURES |
|---|
|
TOPABSTRACTINTRODUCTIONPAC ADDUCT DETERMINATIONS IN...EXAMPLES OF MONITORING HUMAN...ANALYSIS OF THE DATAINHALATION VERSUS DERMAL...DISCUSSIONCONCLUSIONSAPPENDICESREFERENCES |
|---|
The main exposure groups
The main human exposure groups are as follows.
Workers having high exposure to coal tar fumes (CTPV) in coke ovens, the aluminium industry and foundries [average values for 8PAHs between 5 and 50 µg/m3, with high values up to 500 µg/m3; average values for BaP exposure 4 µg/m3 (1–50 µg/m3) with high values up to 90 µg/m3].
Workers exposed to diesel exhaust and bitumen fumes in road paving, roofing and bitumen manufacture, with exposure levels ranging from 3.5 to 397 ng/m3 8PAHs, 0.2–15 ng/m3 BaP. These levels are approximately three orders of magnitude lower than the high exposure group working in coke ovens and the aluminium industry.
Environmentally exposed persons. Here exposures can vary widely depending on the nature of the area (rural or urban, etc.).
The significance of airborne PAH concentration in exposure studies
In our opinion, the use of the concentration in air of one PAH or the sum of a limited number of individual PAHs as a marker of exposure should be treated with care. For example, when coke oven workers are compared with asphalt pavers, 8PAH and BaP exposures represent completely different burdens of exposure to potentially carcinogenic compounds. Coal tar fumes consist mainly of unsubstituted PAHs, and to a minor extent monomethyl-substituted PAHs. Tjoe Ny et al. (1993) showed that the 16 marker PAHs recommended by the US EPA (1986) constitute between 40 and 90% of the CTPV fraction. A similar observation has been made by Brandt (1994). In bitumen fumes, on the other hand, the 8PAHs constitute only a small fraction (<1%) of the total potentially carcinogenic PACs (Brandt et al., 1999). Hence, the 8PAH exposure represents ~50 times higher potency for bitumen fume exposures than for exposures to CTPVs. For environmental exposures to PACs the contribution of combustion sources and engine exhaust can be appreciable (Halsall et al., 1994; Schauer et al., 1996).
Occupational exposures to PAH
Coke oven workers
van Schooten et al. (1990) studied a group of 56 coke oven workers and 44 non-exposed controls for PAH–DNA adducts using an ELISA procedure. The average exposure to CTPV was 0.45 mg/m3, while the exposure to BaP was 0.98 µg/m3. The average levels of PAH–DNA adducts measured were 5.1 adducts/108 nucleotides for the exposed group and 2.7 adducts/108 nucleotides for the controls. No significant correlations were found between PAH–DNA adducts in blood and the air concentrations of total PAHs or BaP. In both groups smokers had significantly higher levels of adducts than non-smokers.
The specificity of biological monitoring probes, e.g. excretion of phenanthrene and pyrene metabolites in urine, and the usefulness of some other biomarker measurement techniques, e.g. alkaline filter elution, the 32P-post-labelling assay and measurement of SCE, in coke oven workers exposed to PACs were investigated by Popp et al. (1997). Only a weak correlation between urinary excretion of PAH metabolites and personal exposure (mean 8PAHs 15 µg/m3) was found. There were indications of metabolic differences between individuals and these were more pronounced for phenanthrene than for pyrene. Smoking also had an important influence. There was a significant difference in results of alkaline filter assays, indicating significantly more DNA single-strand breaks and also DNA protein cross-links in coke oven workers compared with controls. However, a significant difference in SCE between coke oven workers (6.2 SCE/cell) and controls (6.7 SCE/cell) was not observed. Smokers in both groups showed higher values than non-smokers. In addition, there was no difference in levels of PAH–DNA adducts in lymphocytes between smokers and non-smokers and these only correlated with exposure to PAHs in a semi-quantitative manner.
Hemminki et al. (1990a,b) measured DNA adducts in white blood cells of Polish coke oven workers and controls. Local controls living near the cokeries showed similar adduct profiles to the coke oven workers. The ‘topside’ workers had the highest BaP exposures. Values for rural control subjects were 2–3 times lower. In these studies there was a high variation in the estimated BaP between different plants. When DNA adduct measurements were made these were related to broad exposure groups, without precise definition of the actual personal exposure levels. For our estimate of 8PAH exposure we took an average of the reported estimates of BaP exposure.
Lewtas et al. (1997) found no statistically significant differences in DNA adduct levels in white blood cells between exposed coke oven workers and environmental controls (2.5–5.7 adducts/108 nucleotides for 8PAH exposures of 1.2–17.4 µg/m3 in the exposed group and 4.6–4.8 adducts/108 nucleotides for 8PAH exposures of 0.007–0.013 µg/m3 in the controls). This was attributed to a non-linear dose–response behaviour in adduct formation at the higher exposure levels. The ‘DNA adduct formation efficiency’ expressed as DNA adducts (relative adduct labelling)/PAH exposure ranged from 0.3 to 3.0 for the exposed groups and 368 to 700 for the control groups. A similar effect was found for the DNA binding potency in human bronchoalveolar lavage lung cells of another group of highly exposed subjects compared with controls. This was also reflected in values for cytogenetic markers.
Kalina et al. (1998) monitored cytogenetic markers in coke oven workers and found an increase in the percentages of aberrant cells (CA) and SCE/cell and the 8PAH exposure levels. Their data indicated a decreased formation efficiency of CA and SCE with increasing exposure. A 44 times higher average 8PAH exposure (18 versus 0.4 µg/m3) resulted in a 2 times higher CA (2.3 versus 1.1%) and in a 1.4 times higher SCE value (7.5 versus 5.5 SCE/cell). These observations correlate with the results found by Lewtas et al. (1997) for DNA adduct formation in white blood cells. Hemminki et al. (1997), however, reported, based on results of three studies, an indication of consistency in dose–response relationships for DNA adduct formation in foundry workers with BaP exposures ranging from 5 to 50–200 ng/m3. These authors note that others (Kriek et al., 1993; Ovrebo et al., 1994, 1995) reported only a small or no increase in DNA adducts in white blood cells and lymphocyte DNA for coke and aluminium workers. They suggest as an explanation that when the occupational exposure decreases, other sources, such as food, smoking, environmental air pollution and drinking water, become relatively more important. In our opinion, possibly at least a part of the difference in DNA adduct (or CA or SCE) formation efficiency is due to the unsuitability of 8PAH as a marker for potential carcinogenicity (see above), when comparing different exposure types.
Zhang et al. (2000) reported studies on aromatic DNA adducts in white blood cells of coke oven workers, in relation to exposure, lifestyle and genetic polymorphism of metabolic enzymes. The exposed workers were divided into high (estimated 8PAH 51 µg/m3) and low (estimated 8PAH 22 µg/m3) exposure groups. The DNA adduct level of the high exposure group (1.59 adducts/108 nucleotides) was not significantly higher than that of the low exposure group (1.33 adducts/108 nucleotides). Effects of variables, including smoking, genetic polymorphism of CYP1A1, NAT2 and the p53 gene and a family history of cancer, on DNA adduct levels were found, suggesting that these should be considered when evaluating the genotoxic effect of occupational exposure to PACs using white blood cell DNA adducts.
In a study by Reuterwall et al. (1991), in which the estimated exposure to 8PAHs (17 µg/m3) was similar to that measured in the studies by Kalina et al. (1998), there was no significant difference between exposed and control subjects for each of CA, SCE, frequency of MN, urinary thioether excretion or mutagenicity in urine. At this level of PAH exposure a factor of more than 10-fold difference in 8PAH exposure between exposed and controls did not give a significant difference in thioether excretion, CA, MN or SCE.
Binkova et al. (1998) evaluated the DNA adduct levels in total white blood cells and lymphocytes from the blood of the same individuals using the 32P-post-labelling assay. The DNA adduct pattern in lymphocytes and white blood cells were qualitatively the same and there was a reasonable quantitative relation between the two. A positive significant correlation between PAH (or BaP) exposure and DNA adduct level was observed. An order of magnitude difference in PAH exposure resulted in only about a 1.5-fold higher level of DNA adducts.
Kuljukka et al. (1996) studied the relation between airborne pyrene and BaP (mean 5.7 µg/m3, range 0.02–40) exposure and urinary 1OHPy excretion (mean 6 µmol/mol creatinine, range 0.2–70). The 1OHPy was linearly correlated with the airborne pyrene and BaP concentrations and was significantly higher than that of the controls. Skin wipe analyses also indicated a considerable skin contamination with pyrene and BaP.
van Delft et al. (2001) used urinary 1OHPy excretion as a biomarker of exposure for a group of coke oven workers. The urinary 1OHPy excretion of the coke oven workers was significantly increased relative to the controls. However, the mean value of 1.24 µmol/mol creatinine was below the no effect benchmark level of 1.4 µmol/mol creatinine proposed by Jongeneelen (2001). No significant induction of DNA adducts or cytogenetic markers (SCE, HFC and MN) was found. Smoking caused a significant increase in many of the biomarkers studied.
Wu et al. (2002) monitored coke oven workers for urinary 1OHPy and trans-anti-BaP-tetraol as biomarkers of exposure and benzene soluble matter (BSM, i.e. the fraction of the particulates collected on a filter that is soluble in benzene) as a surrogate for external PAH exposure. A significant correlation between external exposures and urinary excretion of the PAH metabolites was found.
Aluminium production workers
In studies on aluminium production workers, Carstensen et al. (1999a,b) measured urinary 1OHPy, DNA adducts and cytogenetic markers in peripheral lymphocytes in pot room workers exposed to 16 µg/m3 10 4–6 ring PAH (estimated 9.8 µg/m3 8PAHs). A 30 times higher level of 1OHPy excretion was found in pot room workers compared with unexposed controls. In contrast, no significant difference in DNA adduct levels, MN or DNA strand breaks were found. No correlations between any of the cytogenetic markers and 1OHPy in urine or PAH–DNA adducts were found. Age and smoking habits did not correlate with the occurrence of strand breaks.
Tjoe Ny et al. (1993) studied workers in an aluminium production plant reporting some very high personal exposures to BaP (0.9–48 µg/m3) and pyrene (3.5–130 µg/m3) and urinary 1OHPy at the beginning and the end of the working week (end of week concentration 1–43 µmol/mol creatinine). The airborne pyrene concentration during the working week was linearly related to the increase in urinary 1OHPy. An 8-fold increase in pyrene exposure gave rise to a doubling of the change in 1OHPy.
Heussner et al. (1985) studied workers in an anode pre-baking plant (BaP exposure > 0.2 µg/m3). A significant effect of CTPV exposure was only found for urine mutagenicity, in both smokers and non-smokers. No significant effect of exposure to CTPV on chromosome aberration rates or semen analysis was found.
van Hummelen et al. (1993) compared workers employed in a graphite electrode producing plant and a coke oven with a control population of maintenance workers in a blast furnace. Although the mean airborne PAH concentrations for the workers [8PAH estimated from total PAHs reported 
4 µg/m3, based on reported PAH profiles (Popp et al., 1997; Carstensen et al., 1999a,b; Zhang et al., 2000)] was significantly higher than that of the controls (0.25 µg/m3), statistically significant differences in SCE and HFC and positive correlations between the cytogenetic markers and airborne PAH levels or urinary hydroxypyrene concentrations were only detectable in the lower exposed workers from the coke oven. A statistically significant effect of smoking was not observed.
Vu Duc and Lafontaine (1996) measured 1OHPy in human urine as a biomarker of exposure to PAC in various processes in an aluminium production plant, such as relining, starting up of a pot, working in a paste plant and at bake ovens. They found a trend for 1OHPy excretion similar to that of pyrene and 
8PAHs.
Schoket et al. (1999) measured DNA adduct formation in blood and urinary 1OHPy excretion in aluminium production workers. DNA adduct levels in blood were significantly enhanced relative to controls. 1OHPy levels were 13- to 70-fold higher than in the controls. No PAH exposure measurements were made. A significant linear relation between DNA adduct levels and 1OHPy concentrations was not found nor was a significant influence of smoking on DNA adduct formation observed.
Mannschreck et al. (1996) measured the exposure to PAHs of workers in a graphite electrode production plant and found a significant correlation between urinary 1OHPy excretion and the BaP and pyrene concentrations.
Using 32P-post-labelling analysis coupled with HPLC, Tuominen et al. (2002) were able to resolve four DNA adducts from peripheral mononuclear blood cells of pot room workers and controls. For three of the DNA adducts there was no statistically significant difference between the exposed group and the controls. One adduct was, however, statistically increased in the exposed group. The levels of this adduct were shown to be related to the length of employment, genetic polymorphisms and use of respiratory protection.
Iron foundries
Phillips et al. (1988) measured DNA adducts in peripheral white blood cells of workers occupationally exposed to PAHs in a Finnish iron foundry and from control subjects not known to be occupationally exposed to PACs. Using exposures from previous industrial hygiene surveys, foundry workers were classified as belonging to high, medium or low exposure groups according to their exposure to airborne BaP (high, >0.2; medium, 0.05–0.2; low, <0.05 µg BaP/m3 air). The average DNA adduct level in the high and medium exposure groups was 1.8 adducts/108 nucleotides (range 0–10). DNA adduct levels of the low exposure group (average 0.06, range 0–0.6 adducts/108 nucleotides) were not different from those of the unexposed controls (average 0.2, range 0–1.9 adducts/108 nucleotides). A significant effect of smoking was not found. In the same foundry and using the same exposure classification, Reddy et al. (1991) investigated DNA adduct formation in white blood cells of 61 workers and 19 controls. There was a highly significant correlation between the estimated exposures and adduct levels as determined by analysis of variance. No effects due to age, sex or smoking habit of the subjects were observed. The results from this (larger) study confirmed the initial results reported by Phillips et al. (1988).
In separate studies at about a factor of 10 lower BaP exposure, Santella et al. (1993) used competitive ELISA to measure PAH–DNA adducts in white blood cells and urinary 1OHPy. Both the PAH–DNA and the 1OHPy levels showed an increasing trend with exposure. The inter-individual variability was high for both biomarkers.
The averaged results of three laboratories, measured on 53 foundry workers exposed to <0.05–>0.2 µg/m3 BaP and six unexposed controls, were 16 (range 9–26) adducts/108 nucleotides and 2.4 (range 1.7–3.1) adducts/108 nucleotides, respectively. No significant effect of smoking was found (Savela et al., 1989).
Hemminki et al. (1997) studied DNA adducts in leukocytes of workers exposed in a foundry over a period of 5 yr. The results indicated consistency in dose–response relationships for workplace BaP concentrations and DNA adducts. The post-labelling method was able to detect an increase in aromatic DNA adducts in leukocytes when exposure to BaP was 0.005 µg/m3. It was concluded that at such low levels, smoking and consumption of charcoal broiled food may be important contributors to adduct levels.
Forni et al. (1996) studied cytogenetic end points in leukocytes of a cohort of coke oven workers in a steel plant, engaged in jobs involving different levels of exposure. No significant difference was found in CA for the exposed subjects, relative to the controls. The average exposure to BaP was 0.96 µg/m3, versus <0.001 for the controls. The SCEs of the exposed groups were significantly increased only for smokers. The non-significance for CA is in line with the findings of Reuterwall et al. (1991) and Heussner et al. (1985).
Bitumen application workers
A considerable number of workers are involved in handling hot bitumen-derived products in, for example, road paving and roofing. In 1998 the US National Institute of Safety and Health (NIOSH) estimated the number of workers in the US asphalt and roofing industries to exceed 350 000 (NIOSH, 1998). On the basis of the volumes of bitumen/asphalt produced the number of workers in Europe is at least half of the US number.
Road pavers
Pasquini et al. (1989) carried out biological monitoring to assess the mutagenic/carcinogenic hazards associated with exposure to bitumen fumes during paving operations, analysing some biological parameters in the urine of a group of exposed workers. The urine samples were studied for mutagenicity by the Ames test and for levels of thioethers and D-glucaric acid. Urinary mutagenicity data for exposed workers were statistically higher than those of a group of unexposed subjects. Thioethers were higher only in subjects exposed simultaneously to bitumen fumes and cigarettes. D-Glucaric acid excretion was not increased significantly due to bitumen fume exposure.
Fuchs et al. (1996) studied road pavers, roofers and painters exposed to bitumen materials and reported DNA adducts for the pavers of the order of a few adducts/109 nucleotides; these were very low, but no values were reported for the controls. Alkaline strand breaks in pavers, roofers and painters were not significantly different from the controls.
Burgaz et al. (1988, 1991, 1992, 1998) surveyed asphalt workers, but did not report values for the levels of PAH exposure. In these studies the mean levels of urinary 1OHPy in bitumen-exposed workers were significantly higher relative to those of the controls. However, the average level of urinary 1OHPy was well below the 1.4 µmol/mol creatinine proposed by Jongeneelen (2001) as the no effect level. Mean SCE values of exposed subjects (all and non-smokers) were significantly higher than those of controls. There was a significant difference between means of the exposed and control groups in the frequency of MN. Urinary thioethers measured in the exposed groups were higher than in controls, however, this increase was only significant in the smoking group, possibly due to differences in smoking habits. A correlation between urinary 1OHPy and thioether levels was not found.
Hatjian et al. (1995, 1997) measured PAH exposures, SCE, urinary thioethers and urinary D-glucaric acid in pavers and roofers. No significant difference was found between exposed subjects and controls. An increase in urinary 1OHPy over a 3 day work period was found for the pavers and roofers, but not for the controls.
In a study of 19 asphalt workers and 13 controls, Zhou (1997) reported airborne PAH exposures, PAHs in skin wipes, 1OHPy pre- and post-shift and total DNA adducts in exfoliated urothelial cells. Although airborne PAH concentrations and PAH in skin wipes were 20 and 4 times higher, respectively, for the exposed workers compared with the controls, no significant difference in 1OHPy or DNA adduct levels was found between controls and exposed workers. The post-shift 1OHPy levels were below the proposed benchmark no effect level of 1.4 µmol/mol creatinine.
No difference was found between pre- and post-shift urinary 1OHPy in a study by Järvholm et al. (1999), who monitored Swedish non-smoking asphalt workers and controls for PAHs in the breathing zone, urinary 1OHPy and SCE and MN in peripheral lymphocytes. For a PAH exposure in the ranges found in other studies (Brandt et al., 1993; Watts et al., 1998) the road pavers had no significant increase in SCE and MN.
In a long term (4 yr) monitoring study of a group of asphalt pavers, Major et al. (2001) measured CA, SCE and hypoxanthine phosphoribosyltransferase (HPRT) variation frequencies and compared these with the values for unexposed controls. During the study period a number of hygiene measures were instituted, such as improvements in personal protection, switching from crude oil to detergents to clean equipment and introducing proper ventilation in the cabins of the roller drivers. This resulted in a drop in CA in the study period from 3–3.6% to values between 1 and 1.2%, the level of the controls. A CA level <1 has been reported by the European Study Group on Cytogenetic Biomarkers and Health (Hagmar et al., 1994, 1998) as having no increased cancer risk. The SCE frequencies increased significantly relative to the controls in the first year of the study, but then decreased to the control level. HPRT values in the exposed groups did not differ in any of the study years. The authors concluded that the application of tar-free asphalt by itself does not increase the genotoxic risk when adequate personal protection is provided.
Szaniszlo and Ungvary (2001) monitored urinary PAH metabolites in asphalt pavers in the city of Budapest and measured atmospheric PAHs in the working environment. A group of non-PAH exposed health care workers was chosen as controls. In the same study, policemen working in the city were also monitored. Although total PAHs and pyrene were similar for policemen and asphalt pavers, the (5–6) ring PAHs was much lower for the asphalt pavers (7 ng/m3) than for the policemen (26 ng/m3). Urinary 1OHPy excretion was similar for the non-smokers in all groups and similar to the level of the controls. Levels for the smokers were about 5 times higher than those of the non-smokers. Hence, with this external PAH exposure level (7PAHs 5–54 ng/m3) smoking was the dominant factor affecting 1OHPy excretion. Other PAH metabolites (3-hydroxybenz[a]anthracene and 3-hydroxybenzo[a]pyrene) were not detected in the urine of any of the groups.
Roofers
Hatjian et al. (1995, 1997) monitored roofers, exposed only to bitumen fumes, for PAH exposure, urinary thioether excretion, urinary 1OHPy excretion and SCE. For a 8PAH exposure level of 0.2 µg/m3 no significant difference relative to the controls was found for any of the biomarkers.
Herbert et al. (1990) reported on a study of roofers involved, during the first half of the shift, in removal of an old coal tar roof and, during the second half, in application of a new bitumen-based roof. Personal air samples and skin wipes were taken and DNA adducts in white blood cells were determined by 32P-post-labelling. The levels of BaP relative to the total PAHs measured, in both personal air samples (BaP 0.8 µg/m3, 7 3–6 ring PAHs 8.3 µg/m3) and skin wipes, indicated that the main PAH exposure came from coal tar dust that the workers were exposed to during the removal of the old roof. The median level of DNA adducts was 1.3 per 108 nucleotides (83% above detection limit). The controls were 0.13 adducts/108 nucleotides (17% above detection limit). For the roofers the DNA adduct levels correlated with the post-shift levels of PAHs in the skin wipes.
In a separate study, the contribution of coal tar exposure was confirmed by Toraason et al. (2001), who measured DNA strand breaks and 1OHPy in roofers with coal tar pitch dust and/or bitumen fume exposure. The assessments of 1OHPy indicated that exposure to bitumen fumes might be a factor in elevated urinary 1OHPy, but that PAC exposures during removal of roofs containing coal tar was the primary contributor. 1OHPy levels in roofers with only bitumen fume exposure were <1 µmol/mol creatinine, whilst those in roofers co-exposed to coal tar were 6 times higher. Elevated strand breaks were only found in roofers with exposure to coal tar.
In a factory manufacturing clay pigeons, Lafontaine (2000) monitored five workers and two external observers by personal air and urinary 1OHPy sampling for a four shift working week. No respiratory protection was worn during the manufacture of the targets, which were moulded from coal tar pitch at 190°C. Atmospheric concentrations of pyrene and BaP ranged from 0.66 to 5.05 and 0.037 to 0.270 µg/m3, respectively. The correlation between atmospheric pyrene and urinary 1OHPy concentrations (increase over the shift) was poor (r = 0.37). It improved greatly (r = 0.74) if the amount of pyrene inhaled over the shift and the corresponding amount of urinary 1OHPy excreted were considered. This observation is in line with the findings of Grimmer et al. (1994) for coke plant workers.
Diesel exhaust
Diesel exhaust has been classified as a probable human carcinogen (Group 2A) by the International Agency for Research on Cancer (IARC, 1989). Epidemiological data suggest an increased risk of lung cancer for workers exposed to diesel exhaust (Lipsett and Campleman, 1999) at a level comparable with that for environmental tobacco smoke (Comstock, 1998). Using lung implantation of condensed exhaust fractions in rats, Grimmer et al. (1987) found that in a group of 35 rats the fraction of PAHs with four or more aromatic rings caused six out of seven total cancers, whereas the nitro-aromatic fraction contributed only one out of seven cancers.
Shu-Xin et al. (1997) studied coal miners over a period of weeks of extra high exposure to diesel exhaust that led to an increase in DNA adducts in white blood cells, relative to an already relatively high adduct level during normal work (non-exposed, 5.1 adducts/108 nucleotides, versus 10.4 adducts/108 nucleotides for normal exposure level and 13.9 adducts/108 nucleotides after extra high exposure).
In non-smoking bus garage workers exposed to an estimated 0.7 mg/m3 total particulate matter and 15 ng/m3 BaP, increased DNA adduct levels were found (Hemminki et al., 1994).
Nielsen et al. (1996a) studied DNA adduct formation, 1OHPy excretion and hydroxyethylvaline (HOEtVal) adducts in haemoglobin in a group of bus garage workers exposed to diesel exhaust. The exposed workers had significantly higher levels of all three biomarkers compared with the controls. The authors suggested that skin exposure to lubricating oil was a major confounder for biomarker formation. However, in our view the high level of HOEtVal in haemoglobin relative to the controls (33 versus 22 pmol/mol) suggested an important role for exhaust gas exposure. This was corroborated by the finding that the HOEtVal adduct levels correlated with the urinary 1OHPy levels.
DNA adduct formation and urinary 1OHPy excretion were measured in a group of garage mechanics exposed to 0.05–0.18 µg/m3 BaP (controls were exposed to 0.0006–0.0009 µg/m3). No difference in levels of DNA adducts was found between mechanics and controls. Significant differences were observed between smoking and non-smoking workers for both DNA adducts and 1OHPy excretion, but not between smoking and non-smoking controls. Approximately two to three orders of magnitude difference in ambient PAH levels gave rise to only a 2-fold increase in 1OHPy concentration (Schoket et al., 1999). In these types of studies it seems probable that skin co-exposures to diesel fuel and lubricating oil could have affected the results, a conclusion also made by Nielsen et al. (1996a).
Seidel et al. (2002) monitored underground salt miners exposed to diesel exhaust by air sampling for PAHs and nitroarenes and by measuring the urinary excretion of 1OHPy, hydroxylated phenanthrene metabolites and some aromatic amines as markers for exposure to nitro compounds. An increased urinary level of phenanthrene metabolites was found in diesel exhaust exposed workers that was further enhanced by smoking. The excreted amounts of aromatic amines found as metabolites of the nitroarenes were about 5- to 10-fold higher than expected from the levels determined by personal air sampling at the workplace of the individuals. The authors suggested that urinary 3-aminobenzanthrone be used as a specific biomarker of exposure to diesel exhaust.
Non-occupational exposures to PAH
Environmental
Nielsen et al. (1996b) monitored urban and suburban non-smoking bus drivers and rural controls (smokers and non-smokers) for DNA adducts. A significant influence of urban air pollution on the level of DNA adducts (40 adducts/108 nucleotides for urban and 18 adducts/108 nucleotides for suburban bus drivers versus 2.4 adducts/108 nucleotides for local controls) was found. No significant influence on adduct levels was associated with potential confounders, including smoking and diet.
Binkova et al. (1995) monitored non-smoking and non-occupationally exposed working women in an area in the Czech Republic heavily polluted by coal combustion. The authors found a significant correlation between PAH exposures (8PAHs between low <0.01 and high >0.02 µg/m3, BaP between low <0.002 and high >0.004 µg/m3) measured in the 24 h period prior to blood sampling and the DNA adducts in white blood cells (4–6.5 DNA adducts/108 nucleotides).
Lewtas et al. (1997) studied the effect of personal exposure to air pollution on DNA adducts in a group of women working outdoors in the Czech Republic. ‘Low’ levels of PAH exposures (8PAHs average 0.01 µg/m3) showed a significant correlation with DNA adducts in white blood cells.
Hemminki et al. (1990a,b) found that persons living in the vicinity of coke ovens in Poland (BaP exposure 0.015–0.057 µg/m3) had 2–3 times higher DNA adduct levels than persons living in the countryside. Furthermore, the results showed that the levels of aromatic adducts in white blood cell DNA did not linearly relate to ambient air levels of PAHs. It was concluded that other sources, such as food, might have been important contributors. In studies in the same geographical area and on similar groups, Perera et al. (1992) measured a number of biomarkers of exposure. Their results showed that exposure to environmental pollution (0.006 < BaP < 0.057 µg/m3) was associated with significant increases in DNA adducts, in SCE, including HFC, and in CA. Furthermore, the DNA adducts were significantly correlated with CA.
Poirier (1999) studied US army soldiers (61) normally stationed in Germany but deployed in Kuwait in the aftermath of the Gulf war. PAH exposures in Kuwait were low (BaP < 0.00023 µg/m3) compared with BaP concentrations (0.0015–0.007 µg/m3) in Germany. A significant difference in blood DNA adducts was found (1.7 adducts/108 nucleotides in Kuwait versus 2.9 adducts/108 nucleotides in Germany, before and after, respectively). The authors suggested that in addition to the environmental contribution, the difference in diet (e.g. consumption of charbroiled meat) might have played a role.
Darcey et al. (1992) reported no significant differences in DNA adduct levels (average 2.6 adducts/108 nucleotides) of nine US firefighters before and after 6 weeks of fighting oil fires in Kuwait in 1991. Their diet was almost exclusively imported.
Eder (1999) reviewed a number of the foregoing studies, with a particular focus on intra-individual variations, and found that environmental exposure can have a larger effect on PAH adduct levels than occupational exposure, food or smoking. The author indicated that the most important changes in PAH–DNA adduct levels were caused by environmental exposure due to heating fuels.
Effects of smoking and diet
Burgaz et al. (1988, 1992, 1998) surveyed groups of asphalt workers and controls, but reported no data for actual PAC exposures. Determinations of mean levels of urinary 1OHPy showed no significant difference between smokers and non-smokers in both groups. In both groups SCE frequencies of smokers were significantly higher than those of non-smokers. Smoking had no significant effect on MN. However, a significant effect of smoking on urinary thioether excretion was reported and a linear correlation between number of cigarettes smoked and thioether excretion was found. In contrast, at relatively low external PAH exposure levels (8PAHs < 0.06 µg/m3), Szaniszlo and Ungvary (2001) reported smoking to be the dominant factor influencing 1OHPy excretion. The exposures of the smokers were about 5 times higher than those of the non-smokers.
Savela and Hemminki (1991) compared a group of 11 otherwise unexposed male and female smokers with a similar group of non-smokers (10) and determined levels of white blood cell DNA adducts in T lymphocytes and in granulocytes. In T lymphocytes, with half-lives of several years, much greater (and significant) differences (31 versus 13 adducts/108 nucleotides) were found between smokers and non-smokers than in granulocytes (9.6 versus 7.6 adducts/108 nucleotides), with short half-lives (7–24 h). van Schooten et al. (1990) studied 56 coke oven workers and 44 non-exposed controls for PAH–DNA adducts using ELISA. In both groups smokers had significantly higher levels of adducts than non-smokers.
Phillips et al. (1990) analysed non-tumourous bronchial tissue of 37 cigarette smokers, eight former smokers and eight non-smokers for the presence of aromatic DNA adducts by 32P-post-labelling assay. Adduct levels detected in DNA from non-smokers, former smokers and current smokers were 3.45 ± 1.62, 3.93 ± 1.92 and 5.53 ± 2.13 adducts/108 nucleotides, respectively. The differences in adduct levels between current smokers and former or non-smokers were statistically significant. Among the smokers, significant correlations were found between adduct levels and both daily cigarette consumption and total cigarette consumption. DNA from peripheral blood leukocytes of 31 heavy smokers (>20 cigarettes/day) and 20 non-smokers was also analysed by 32P-post-labelling. Adduct levels in the samples from smokers were not significantly different from those of non-smokers. The authors concluded that measuring DNA adducts in peripheral white blood cells may not be a good procedure for monitoring inhalation exposure to carcinogens. In a study comparing 4-aminobiphenyl haemoglobin and aromatic adducts (detected by 32P-post-labelling) in lymphocytes of smokers it was found that the levels of both of these types of adducts saturated with high tobacco usage, 30 cigarettes/day (Dallinga et al., 1998). Eder (1999) reported that DNA adduct levels were not significantly influenced by smoking cessation, whereas SCEs significantly decreased after cessation.
Several authors (Hemminki et al., 1997; Poirier, 1999) have suggested that at relatively low exposure levels the influence of diet might be an important contributor to biomarker formation. Consumption of charbroiled meat has a marked influence on urinary 1OHPy excretion; increases in the range 10- to 80-fold have been observed (Kang et al., 1995). Other studies indicate that recent consumption of charbroiled food contributes to the PAH–DNA adduct load in peripheral white blood cells, albeit with large inter-individual variations (Rothman et al., 1990, 1993a,b; Poirier, 1999).
Human variability and effects of polymorphisms
Carcinogenesis may be modulated by host polymorphism in genes for PAC metabolism and DNA repair enzymes. This has become an important area of research. Studies relating phenotype/genotype to cancer have examined measurable early end points such as DNA adduct formation and/or cytogenetic damage. There is now substantial evidence emerging from numerous laboratories showing considerable inter-individual human variation in levels of adducts formed due to environmental and occupational exposures to PACs. It is evident that groups of predisposing polymorphic genes exist, for example, those involved in PAC metabolism and DNA repair (reviewed in Bartsch et al., 1998). These factors may increase the degree of adduct formation and thus the cancer risk in certain exposed subjects, even when only low level exposure has occurred. Biomonitoring and molecular epidemiology could thus play an important role in identifying susceptible individuals, particularly those suffering a combination of high risk factors, namely a high level of exposure to PACs, inherited cancer predisposing genes and a deficiency of protective factors such as occur in the diet. Once identified, cancer predisposing genes can be used as intermediate risk markers for cancer.
A number of studies have investigated the relationship between occupational (Hemminki et al., 1997; Kalina et al., 1998; Pan et al., 1998; Rojas et al., 1998, 2000; Pavanello et al., 1999; Zhang et al., 2000) and environmental (Butkiewicz et al., 1998, 2000) exposure to PACs and genetic polymorphisms. Coke oven workers are a group that suffers high exposures to PACs, as detailed in several studies on this occupation (Kalina et al., 1998; Zhang et al., 2000). Kalina et al. (1998) studied 64 coke oven workers and monitored exposure to eight PAHs. The exposure levels varied widely from 0.6 to 547 µg/m3 for total PAH exposure, compared with 0.07–1.51 µg/m3 for the 34 controls studied. Measuring the cytogenetic markers CA, SCE, cells with high frequency of SCE, the heterogeneity index SCE and genetic polymorphisms of glutathione S-transferase (GST) M1 and N-acetyltransferase 2 (NAT2), it was found that all the cytogenetic markers were significantly increased in the exposed versus control groups. However, in this study no effect of GSTM1 and NAT2 genotypes were observed individually or in combination with the cytogenetic markers. The frequency of CA and SCE/cell were found to be related to exposure to carcinogenic PAHs. Using 32P-post-labelling for PAH–DNA adducts in white blood cells of coke oven workers, Zhang et al. (2000) found that subjects homozygous for the rare alle of CYP1A1 showed significantly higher DNA adduct levels than other CYP1A1 genotypes. This area has been the subject of an extensive review by the IARC, which related the observed adducts and other biomarkers to phenotype (IARC, 1999). This appears to be a very promising area of research with significant potential for a role in cancer prevention. However, in view of the overall variation in results observed by different laboratories, it is evident that more research is necessary for this potential to be fulfilled.
|
ANALYSIS OF THE DATA |
|---|
|
TOPABSTRACTINTRODUCTIONPAC ADDUCT DETERMINATIONS IN...EXAMPLES OF MONITORING HUMAN...ANALYSIS OF THE DATAINHALATION VERSUS DERMAL...DISCUSSIONCONCLUSIONSAPPENDICESREFERENCES |
|---|
The data reported in the articles reviewed are tabulated in the Appendix (Table A1), which lists the average values and the minima and maxima reported in those studies. In the discussion that follows we provide an overview of these data. Although we are aware of the difficulty of comparing results from different laboratories, we have nevertheless attempted to establish whether trends can be observed between the measured biomarker levels and the corresponding external PAC exposures (recorded in Table A1). As a marker for external PAC exposure we have selected the sum of eight carcinogenic PAHs (
8PAHs), since these values are reported in many studies. Where a value for 8PAHs was not available, we have estimated it from BaP exposure or other reported data, such as the sum of the reported PAHs or BSM.
Factors for the conversion of BaP concentrations into 8PAHs have been derived from publications where both BaP and 8PAHs were reported. Values for BaP can be translated into values for 8PAHs because the boiling point of BaP (496°C) is approximately in the middle of the relatively narrow boiling point range for all eight PAHs (425–534°C). This makes the ratio 8PAHs to BaP mainly dependent on the PAH profile for the specific industrial process, which is reflected in different factors for the different processes. We estimated the following factors for the conversion of BaP concentrations into 8PAH concentrations: for coke ovens, 10; for bitumen fume exposures, 12.5; for environmental exposures, 5. We then compared the results of this analysis of all studies with the conclusions of the authors of the original papers. Most of these papers only covered one type of occupational group, with its specific quantitative and qualitative exposure characteristics. In Table 2 an overview is given of the biomarker levels and external PAH exposures for each occupational group. The groups have been ranked according to their external PAH exposures.
|
The highest external PAH exposures were observed for workers in coke ovens and the aluminium industry, followed in order by those in foundries, in bitumen and asphalt applications, in those exposed to engine (diesel) exhaust and in those classified as being only environmentally exposed. Between the highest and the next exposure category there was about an order of magnitude difference in PAH exposure levels.
In this review we attempt to correlate each biomarker separately with the corresponding 8PAH exposures. Although the PAH exposures and urinary 1OHPy measurements were low in the environmentally exposed groups relative to all the occupationally exposed groups, the DNA adduct levels were remarkably similar in the environmentally exposed group and the coke oven workers.
However, a comment on the PAH exposures and biomarker levels of the different groups is first necessary. Although in the environmentally exposed group the PAH exposures and urinary 1OHPy excretions were very low, relative to the occupationally exposed groups, the DNA adduct levels reported were at a comparable level to those reported for coke oven workers. These workers have three orders of magnitude higher 8PAH exposures than the environmentally exposed group. The CA and SCE values of the environmentally exposed group were even higher than those of the coke oven workers. There are two possible explanations, which are not mutually exclusive. First, it is possible that exposures to other genotoxic compounds produced larger effects than those from PAHs and these other genotoxic agents were more abundant among environmentally exposed subjects. It is also possible that the exposure regimen played some role because environmentally exposed persons are exposed 24 h/day, for their lifetime, whereas occupationally exposed persons are exposed 8–10 h/working day, during their working life. Because of this difficultly in characterizing the actual exposures of the environmental group, the environmental data have not been included in the following analyses.
The increments in the biomarker levels for the exposed subjects over the corresponding control values (e.g. 1OHPyInc) have been correlated with 8PAH exposures. We have checked our simple linear regression analyses by two separate statistical approaches: a weighted regression analysis and an analysis whereby the exposed were divided into low and high exposure groups. Statistical analyses were conducted using the GLM procedure in the Statistical Analysis System (SAS), version 8.02. Group mean biomarker levels were considered by analysis of covariance on 8PAH and the study. Observations were weighted according to group size. A second approach was used whereby study groups were allocated to exposure levels of 0, >0–<2 and 
2 mg/m3. Group mean biomarker levels were considered by analysis of variance by exposure level, allowing for the study from which the data was taken. 1OHPy and DNA adduct data were log transformed prior to analysis. Observations were weighted according to group size. For the weighted regression analysis it was not possible to include standard deviations, because these were only reported for some studies.
Urinary 1-hydroxypyrene excretion (1OHPy)
For 1OHPy excretion at the end of shift there is an increasing trend with increasing PAH exposure in the range 0–150 µg/m3 (1OHPyInc = 0.12 x 8PAH + 0.65, R2 = 0.95, df = 11) (Fig. 3). Above 20 µg/m3 only 2 of 12 data points are found at high exposures. When the two highest exposure values are omitted and only the range 0–20 µg/m3 is taken, the slope increases 2-fold relative to that of the full range of exposures and the fit is less good. No differences were found with the weighted regression in which a statistically significant increase in 1OHPy with 8PAH at the 1% level was found.
|
In the seven publications on CTPV exposures (
8PAH exposure 0.5–150 µg/m3) that have been reviewed, a significant correlation between external PAH exposure and urinary 1OHPy excretion has been reported in four cases. In the three others a trend was observed. Also, in a roofing survey, where exposure to coal tar (0.01–8.5 µg/m3) occurred, a significant correlation was found. Furthermore, in a study in a plant manufacturing clay pigeons, where coal tar was used as the binder, high PAH exposures were recorded and a trend was found between pyrene exposure and urinary 1OHPy exposure.
In five cases where exposure to bitumen fumes only occurred in paving and roofing operations (8PAHs 0.01–9.6 µg/m3), no relation was found between the external PAH exposure and urinary 1OHPy excretion. In a study on garage workers exposed to diesel exhaust a trend between PAH exposure and urinary 1OHPy excretion was reported.
Chromosome aberrations (CA)
The only other biomarker that showed an i