Multiplex Genotyping as a Biomarker for Susceptibility to Carcinogenic Exposure in the FLEHS Biomonitoring Study

It has been estimated that the incidence of cancer can be reduced by as much as 85% upon preventing exposure to carcinogenic environmental components (1). Therefore, continued efforts are needed to investigate the effect of hazardous environmental exposures on environmental cancer risk and, eventually, to translate this knowledge into preventive policy measures. Molecular epidemiologic approaches have the advantage of applying methodologies that enable investigating the preclinical effects of exposure by using biological markers (biomarkers) present in human cells, tissues, and body fluids (2). These biomarkers comprise measurements of the so-called internal dose of an environmental carcinogen (e.g., urinary metabolites resulting from carcinogen metabolism), biologically effective dose (e.g., DNA damage reflecting certain genotoxic exposures), and early biological effects (e.g., deleterious mutations). The EU Environmental Health Action plan stimulates biomarker studies among relevant human populations (http://www.eu-humanbiomonitoring.org/sub/back/ehap.htm).

The Flanders Environment and Health Survey (FLEHS) is a large-scale biomonitoring study conducted in Flanders from 2002 to 2006 and assigned by the Flemish Government (www.milieu-en-gezondheid.be). Its aim is to obtain a better perspective on health risks associated with exposure to harmful environmental compounds among inhabitants of different regions of Flanders to eventually underpin new policy measures.

Conventional risk assessment methods may underestimate the individual's risk for exposures to environmental carcinogens because of the default assumption that all individuals within a certain population possess equal susceptibility to a specific carcinogen dose (1). Although the established risk assessment procedure for setting exposure limits to environmental carcinogens is considered sufficiently safe because of the application of the relatively rigid linear, nonthreshold, extrapolation model and the inclusion additional safety factors, environmental health policy making may benefit from more information on the degree of interindividual variations in environmental cancer risk. Investigating genetic polymorphisms predisposing for increased susceptibility in monitored populations may thus be of relevance.

In the FLEHS study, individuals were monitored for exposure to organochlorides, heavy metals, polycyclic aromatic hydrocarbons (PAH), and benzene. PAHs and benzene are established environmental genotoxins, whereas organochlorides and heavy metals are known to produce free radicals and thereby lead to oxidative DNA damage, and thus represent a cancer risk (3-7). To elucidate the genotoxic effects of these exposures among the Flemish populations, DNA damage levels as measured by the Comet assay and as micronuclei in lymphocytes, urinary 8-hydroxydeoxyguanosine and several serum tumor markers were analyzed.

In the current study, for the purpose of determining more susceptible subpopulations, a random selection of 790 individuals from the FLESH study population was genotyped for 36 polymorphisms in 23 critical genes involved in biotransformation, oxidative stress, and DNA repair, with the aim to investigate the impact of these genotypes plus gender on various exposure-effect relationships.

Volunteers were all recruited upon advertisement by the FLESH Biomonitoring and Surveillance team. The campaign was approved by the ethical committee of the University of Antwerp, and every individual signed an informed consent before collecting blood. Blood samples were donated by all individuals for analysis of exposure (internal dose) and effect biomarkers. Information on age, sex, and smoking habits was obtained by questionnaire. Two age groups were defined, coming from across Flanders and consisting of 429 adolescents and 361 adults; this represented a selection from a grand total of 4,800 subjects who participated in the FLESH biomonitoring study, which has run from 2002 to 2006.

The adolescents were sampled between October 2003 and July 2004 and consisted of 256 males and 173 females, with an averaged age of 15 ± 0.5 years. Fifty-four individuals reported smoking 7 ± 7 (range, 1-41) cigarettes per day. These 429 adolescents are a representative subset of the larger source population, enrolled by the FLEHS via 50 schools located in Flanders, and sampled between October 2003 and July 2004. Inclusion criteria were the following: being born in 1988 or 1989, studying in the third year of secondary school, living for at least 5 years in the same area, and giving informed consent (both adolescent and parents).

Subsequently, 361 adults aged 58 ± 4 years at the time of inclusion were sampled between September 2004 and June 2005. This study population was composed of 185 males and 176 females, of whom 65 individuals reported to smoke 17 ± 9 (range, 1-40) cigarettes per day. These 361 adults are also a representative selection of the larger source population, randomly sampled by the FLEHS via communal address databases between September 2004 and June 2005. Inclusion criteria were age between 50 and 65 years, living in Flanders for more than 5 years, and being able to complete questionnaires in Dutch.

Approximately 200 mL of urine and 40 mL of blood were sampled from each participant to carry out various analyses. The creatinine content in urine was determined spectrophotometrically by Algemeen Medisch Laboratorium in Antwerp. Na₂EDTA (10% volume for volume) was added to whole blood, whereas serum was prepared by immediate centrifugation of the coagulated blood. Samples were stored at −20°C until analysis.

WBCs were isolated by destroying RBCs using lysis buffer containing 155 mmol/L NH₄Cl, 10 mmol/L KHCO₃, and 10 mmol/L EDTA of 0°C (blood-buffer ratio is 1:2). After two rounds of (a) incubation with lysis buffer on ice for 30 minutes and (b) centrifugation (10 minutes at 1,000 rpm at 4°C), the cells were washed with 1 mL 150 mmol/L NaCl and stored at −80°C before DNA isolation.

DNA isolation was done in a 96-well format using the Invisorb Blood Midi HTS 96 Kit/C (Invitek) and stored in 96-well plates at −4°C until genotyping.

Oxidative stress, 36 single-nucleotide polymorphisms in 23 genes involved in biotransformation, and DNA repair were analyzed in this study. Selection of these polymorphisms was based on their expected contribution to the formation and repair of DNA damage induced by environmental (carcinogenic) chemicals (Table 1). Moreover, they had to occur at a frequency in the population of >5% and should likely affect the enzyme activity. An exception has been made for polymorphisms in the phase I biotransformation gene CYP1A1 (because their frequency of occurrence is <5% in most populations) because of the prominent role of this enzyme in the bioactivation or detoxification of numerous xenobiotics.

All DNA sequences and allele frequencies were obtained from the SNP 500 cancer database (http://snp500cancer.nci.NIH.gov).

PCR and single-base extension primers were designed as described previously. PCR was carried out in four different multiplex PCRs (Table 1) on a Tgradient 96-well thermal cycler (Biometra), each in a 10-μL volume using 96-well plates (described previously in ref. 8). Values for Tm were optimized for every multiplex PCR individually: 56°C for multiplex 1, 60°C for multiplex 2, 57°C for multiplex 3, and 60°C for multiplex 4 (Table 1). PCR products were subsequently incubated (37°C for 45 minutes) with 4 μL Exo-SAP-IT (Amersham) to digest contaminating deoxynucleotide triphosphates and PCR primers. Enzymes were deactivated at 75°C (15 minutes).

Genotyping was done by primer extension using the Snapshot methodology (Applied Biosystems). Four different multiplex single-base extension reactions were conducted on the corresponding multiplex PCR products using 96-well plates. The protocol used for single-base extension was described previously (8). After genotyping, single-base extension products were mixed with 13 μL deionized formamide containing 0.4 μL Genescan 120 LIZ size standard, denatured at 95°C for 5 minutes, and thereafter analyzed on an ABI Prism 3100 genetic analyzer using Genscan Analysis software (version 3.7).

Lead (Pb) and cadmium (Cd) concentrations in whole blood were determined after an acid digestion pretreatment destroying the organic matrix, followed by high-resolution inductively coupled plasma–mass spectrometry detection.

Analysis of serum polychlorinated biphenyls and chlorinated pesticides (hexachlorobenzene and dichlorodiphenyl dichloroethane, a metabolite of dichlorodiphenyltrichloroethane) was based on methods as previously described by Gomara et al. (9) and Covaci and Schepens (10). Shortly, blood serum was mixed with formic acid, internal standards were added, and the mixture was equilibrated in an ultrasonic bath. Subsequent elution was done using a solid-phase extraction cartridge, which was washed, dried, and placed on top of a multilayer column filled with anhydrous sodium sulfate and silica with sulfuric acid. Polychlorinated biphenyls and chlorinated pesticides were eluted and concentrated, and subsequently analyzed using gas chromatography combined with an electron capture detector. The detection limit of all chlorinated compounds in serum was 0.02 μg/L.

Determination of 1-hydroxypyrene (a metabolite of pyrene) and tt-muconic acid (a metabolite of benzene) in urine are based on those of Angerer and Schaller (11, 12). The determination of 1-hydroxypyrene was done with high-performance liquid chromatography with fluorescence detector. The detection limit was 0.060 μg/L.

tt-Muconic acid was determined in urine by means of ion chromatography using solid-phase extraction–SAX columns. High-performance liquid chromatography was used to separate the extract from other compounds, and quantification was done by a UV detector. The detection limit was 8.6 μg/L.

Loss of DNA integrity was evaluated by the alkaline Comet assay as described by Singh et al (13). Median percentages of DNA migration in the tail areas were determined and used as a measure of DNA damage.

Micronuclei frequencies were investigated using the cytokinesis block micronucleus assay on whole blood cultures, based on standard procedures (14). For each individual, 1,000 to 2,200 cells were evaluated for the presence of micronuclei using the metafer automatic program from Metasystems.

The competitive immunosorbent assay (Gentaur) based on ELISA was used to measure urinary 8-hydroxydeoxyguanosine levels according to the manufacturer instructions.

The serum protein level of the tumor marker p53 was analyzed using the enzyme immunometric assay and titerzyme enzyme immunometric assay p53 (Assay Designs). Prostate-specific antigen and carcinoembryonic antigen levels were measured using solid-phase chemiluminescent immunometric assay and Immulite 2000 (DPC, Los Angeles, CA). Samples were analyzed for prostate-specific antigen within 24 hours after collection.

First, the putative effect of smoking on exposure was tested by comparing levels of the exposure markers between smokers and nonsmokers using a Mann-Whitney U test. In case that smoking significantly affected biomarker levels, subsequent analyses were conducted among nonsmokers because primary focus was on the exposure to environmental compounds. In case that no effect of smoking was observed, further statistical analyses were done on the total population, cigarette smoking being regarded as an independent variable.

The subsequent analysis has been adapted from our statistical methods described earlier (15) and summarized in Fig. 1. Shortly, after elimination of smokers, if necessary, relationships between markers for endogenous exposure and genotoxic effect were investigated by linear regression analysis. For each individual, the deviation from this relationship was determined by dividing the level of response that was actually observed (O) by the level that was expected (E) according to the regression line. Using this approach, the population was divided into two subgroups: (a) so-called low responders, thus having an O/E<1, and (b) high responders, that is, the more sensitive subjects, showing an O/E>1 (an O/E = 1 was not observed). Subsequently, to identify the most important factors for discrimination of the different subgroups, binary logistic regression analysis has been conducted to classify each individual into one of the two subgroups based on the putative variables cigarettes per day (only if smokers were included), gender, and all single-nucleotide polymorphisms. The investigated genotypes were coded 2 in case of double variants, 0 for wild type, and 1 for heterozygous individuals (15). Results are presented as mean ± SE.

A total of 28,440 genotypes was generated in the overall study population. Statistical analyses assessing the impact of these genotypes on various exposure-effect relationships were conducted in the adolescent and adult population separately, and obtained results will therefore be described for each subpopulation. If the distribution of the biological markers was not normal, a log transformation has been applied to the data (Table 2).

Because blood Cd (P = 0.000) and urinary 1-hydroxypyrene (P = 0.013) were found to be significantly elevated in smokers, subsequent analyses of these markers were conducted solely in nonsmokers. Of all possible exposure-effect relationships, only chlorobenzene (P = 0.046) and ethylbenzene (P = 0.022) exposures were significantly related to levels of DNA damage levels as determined by Comet assay (Table 2A). Dosimeter data for the different benzene exposure markers were only available in 54 individuals.

In case of the exposure-effect relationship between chlorobenzene exposure and Comet DNA damage, no significant predictors for the observed interindividual variation were found among evaluated single-nucleotide polymorphisms (Table 3). In contrast, CAT C-262T (P = 0.027) and GSTT1*0 (P = 0.035) polymorphisms seemed the most important genetic variations affecting Comet DNA damage levels in association with exposure to ethylbenzene (Table 3). For the CAT C-262T polymorphism, individuals carrying the double-variant allele had 1.4 times higher DNA damage levels than CAT C-262T wild-type individuals (P = 0.018; Fig. 2A). Individuals lacking the GSTT1 gene had 1.2 times higher DNA damage levels as compared with subjects carrying the wild-type GSST1 gene variant in response to ethylbenzene exposure (Fig. 2B), although not statistically significant (P = 0.139). No significant role for gender was observed.

Investigated exposure biomarkers in adults, associations of smoking with these biomarkers, and exposure-effect relationships are shown in Table 2B. Cd, in blood and in urine, and urinary 1-hydroxypyrene levels were significantly elevated by smoking (P = 0.000 for each biomarker; Table 2B). Therefore, all subsequent statistical analyses of these biomarkers were conducted in nonsmokers. Significant exposure-effect relationships were established between urinary levels of the early effect marker 8-hydroxydeoxyguanosine and four of the investigated exposure markers, namely, blood hexachlorobenzene (P = 0.003), urinary Cd (P = 0.007 in nonsmokers), urinary 1-hydroxypyrene (P = 0.029 in nonsmokers), and urinary trans-trans muconic acid (P = 0.001; Table 2B). Moreover, two tumor markers in blood, namely, carcinoembryonic antigen and p53, were significantly associated with Pb in blood (P = 0.004) and in urinary Cd (P = 0.003, in nonsmokers), respectively (Table 2B). Finally, a positive correlation between hexachlorobenzene exposure and micronuclei levels was observed (P = 0.000; Table 2B).

The observed associations will be detailed below; it is noted that because of limited space, error bars corresponding to these results are not shown, and data are presented as numbers.

Hexachlorobenzene versus 8-Hydroxydeoxyguanosine. A positive exposure-effect relationship was found between hexachlorobenzene exposure and 8-hydroxydeoxyguanosine DNA damage levels. Females have 1.1 times higher 8-hydroxydeoxyguanosine DNA damage levels upon hexachlorobenzene exposure than males (P = 0.01). In addition, interindividual variations in developing DNA damage after exposure to hexachlorobenzene significantly depend on the mEH genotype (Table 3). Hexachlorobenzene-related 8-hydroxydeoxyguanosine levels were 0.9 times lower in carriers of one or more mEH H¹³⁹R variant alleles compared with wild-type individuals; however, this difference was not statistically significant (P = 0.055).

Hexachlorobenzene versus Micronuclei. Hexachlorobenzene exposure was found to have a positive effect on micronuclei levels. Within this relationship, 1.3 times higher micronuclei levels were observed in females compared with males (P < 0.001). Moreover, 0.9 times lower micronuclei levels were found in individuals with two variants for the mnSOD V¹⁶A genotype (P = 0.032) and for individuals carrying at least one variant for the CYP1A1 T³⁸⁰¹C genotype (0.86 times for heterozygous individuals and 0.65 times for double variants; P = 0.001) compared with the respective wild-type genotypes.

Urinary Cd versus 8-Hydroxydeoxyguanosine. Exposures to Cd as measured in urine was found to be associated with levels of 8-hydroxydeoxyguanosine in nonsmokers. The deletion of the GSTT1 gene was found to be a significant predictor of interindividual differences within this relationship (Table 3). For individuals carrying the wild-type allele for GSTT1, 0.9 times lower DNA damage levels were observed compared with individuals lacking the GSTT1 gene (P = 0.039). No role of gender was found.

Urinary Cd versus p53. A positive exposure-effect relationship between urinary Cd and serum p53 was observed in nonsmokers. GSTM1*0 (P = 0.004) and BRCA2 N³⁷²H (P = 0.024) were found to contribute significantly to interindividual variations within this relationship (Table 3). Wild-type GSTM1 individuals had 1.1 times higher p53 levels in association with urinary Cd excretion than individuals lacking the GSTM1 gene, although not statistically significant (P = 0.106). Moreover, serum p53 levels in individuals carrying the double-variant allele for the BRCA2 N³⁷²H polymorphism were observed to be 1.3 and 1.4 times higher as compared with wild-type and heterozygous individuals (P = 0.024), respectively. No role of gender was observed.

1-Hydroxypyrene versus 8-Hydroxydeoxyguanosine. An association between urinary 1-hydroxypyrene and 8-hydroxydeoxyguanosine DNA damage levels was found in nonsmokers. Interindividual variations within this relationship was affected by gender (P = 0.012), mEH H¹³⁹R (P = 0.023), and CYP1A1 T⁴⁶¹N (P = 0.05; Table 3). According to their urinary 1-hydroxypyrene levels, females had 1.1 times higher DNA damage levels as compared with males (P = 0.008). Moreover, individuals carrying the mEH H¹³⁹R polymorphism have 0.9 times lower 1-hydroxypyrene related DNA damage levels than wild-type individuals (P = 0.044). Lastly, the level of 8-hydroxydeoxyguanosine DNA damage, based on urinary 1-hydroxypyrene dose, was 1.2 times higher in individuals carrying one variant allele for the CYP1A1 T⁴⁶¹N polymorphism as compared with wild-type individuals (P = 0.033). No CYP1A1 T⁴⁶¹N double-variant individuals were detected in this population.

trans-trans Muconic acid versus 8-Hydroxydeoxyguanosine. Only gender had a significant effect on the positive relationship between urinary trans-trans muconic acid and 8-hydroxydeoxyguanosine DNA damage levels (P = 0.002; Table 3). Females had 1.1 times higher DNA damage levels compared with males (P = 0.002) based on the exposure-effect curve between urinary trans-trans muconic acid and 8-hydroxydeoxyguanosine.

Pb versus Carcinoembryonic Antigen. The BRCA2 N³⁷²H polymorphism (P = 0.046), together with smoking (P = 0.000), significantly affected the positive relationship between Pb exposure and carcinoembryonic antigen levels (Table 3). Although smoking was not found to elevate the level of Pb exposure (P = 0.510), smokers had 2.7 times higher carcinoembryonic antigen levels as compared with nonsmokers (P < 0.001) in response to Pb exposure. In addition, variations around the exposure-effect relationship were also found to be predicted by the BRCA2 N³⁷²H polymorphism. Observed Pb-associated carcinoembryonic antigen levels were 1.9 and 1.6 times higher in individuals carrying the double-variant allele as in heterozygous and wild-type individuals, respectively (P = 0.003). No effect of gender has been observed.

In this study, a random selection taken from the FLEHS study population, comprising 429 adolescents and 361 adults, was genotyped for 36 single-nucleotide polymorphisms in 23 genes involved in relevant pathways with respect to environmental cancer risk assessment. In both subpopulations, several important and, in some cases, novel exposure-effect relationships between environmental toxicants and effect markers were established. These will be described separately below. With one exception, more susceptible subgroups could be defined within all significant relationships by analysis of genetic variants in relevant genes. Furthermore, gender seemed to influence susceptibility to various environmental carcinogens. This stresses the significant role that genetic-based interindividual differences play in biomonitoring projects studying environmental health effects.

DNA damage levels measured by Comet analysis were used as a biomarker for genotoxic effects in association with several environmental exposures in the adolescent subpopulation. Of the investigated toxicants, only chlorobenzene and ethylbenzene were significantly associated with this biomarker; both associations have not yet been described in the literature. The main pathway of benzene biotransformation involves formation of metabolites that are known to generate reactive oxygen species through redox cycling (16, 17). These reactive oxygen species eventually induce oxidative DNA damage; exposure to benzene in mice has been shown to induce oxidative DNA damage measured by the Comet assay (18, 19). Interindividual variations around the regression line were observed. In case of chlorobenzene exposure, however, no genetic susceptible subgroups could be defined. However, regarding ethylbenzene exposure, catalase and GSTT1 genotypes significantly affected the relationship between ethylbenzene exposure and DNA damage. Increased levels compared with the expected response based on the regression line were observed in carriers of the CAT C^-262T polymorphism and in individuals lacking GSTT1 activity. Catalase and GSTT1 are involved in the primary defense against (environmentally induced) oxidative stress. Both polymorphisms have been associated with reduced enzyme activity, which leads to higher levels of reactive metabolites and increased levels of reactive oxygen species and, thus, higher DNA damage levels (20, 21). Therefore, GSTT1-null individuals and/or individuals carrying the CAT C^-262T polymorphism may be more susceptible for developing health effects induced by reactive oxygen species resulting from environmental exposure to ethylbenzene, which is in agreement with our observations; GSTT1-null individuals were found to have 1.2 times higher DNA damage levels compared with GSTT1 wild-types, whereas the CAT C^-262T polymorphism was observed to be associated with a 1.4 times higher DNA damage levels compared with wild-type individuals.

Urinary excretion of the DNA repair product 8-hydroxydeoxyguanosine is the most commonly used biomarker for assessing oxidative DNA damage as induced by (environmental) genotoxicants in combination with the efficiency of repair of such damage (22). Several studies have shown an effect of environmental exposures to various agents on urinary 8-hydroxydeoxyguanosine levels (23, 24). In the current study, in adults, an association of hexachlorobenzene in blood, urinary Cd, urinary 1-hydroxypyrene, and urinary trans-trans muconic acid with 8-hydroxydeoxyguanosine levels was observed.

Within the relationships between 1-hydroxypyrene and 8-hydroxydeoxyguanosine, and also between trans-trans muconic acid and 8-hydroxydeoxyguanosine, an effect of gender was found. Higher exposure-related DNA damage levels were observed in females than in males. Gender also affected the relationship between hexachlorobenzene exposure and micronuclei levels; again, higher micronuclei levels upon hexachlorobenzene exposure were found in females compared with males. Although these results are difficult to interpret because little is known about the mechanisms relating gender differences and carcinogen susceptibility, they are consistent with earlier reports that women have a higher susceptibility toward various carcinogens than men (25, 26).

In addition to this effect of gender, an effect of CYP1A1 T⁴⁶¹N and mEH H¹³⁹R on the exposure-effect relationship between 1-hydroxypyrene and urinary 8-hydroxydeoxyguanosine was found. CYP1A1 is a phase I biotransformation enzyme that activates certain carcinogenic compounds such as several PAHs into highly reactive metabolites. A key role for CYP1A1 in pyrene metabolism has been described (27), and variants of the CYP1A1 gene have been associated with increased enzyme activity (28-31). The CYP1A1 T⁴⁶¹N polymorphism described here results in a greater capacity of metabolizing PAHs, and, thus, higher levels of reactive metabolites are formed. These reactive metabolites are known to induce oxidative stress and, eventually, oxidative DNA damage (32). Therefore, within the described exposure-effect relationship between 1-hydroxypyrene and 8-hydroxydeoxyguanosine levels, it may be expected that individuals carrying the CYP1A1 T⁴⁶¹N polymorphism are more susceptible in developing oxidative DNA damage upon PAH exposure compared with wild-type individuals. Indeed, the observed 1-hydroxypyrene-associated urinary 8-hydroxydeoxyguanosine levels were 1.2 times higher in individuals carrying the CYP1A1 T⁴⁶¹N polymorphism than in CYP1A1 T⁴⁶¹N wild-type individuals. The role of mEH is more complex because it may catalyze phases I and II biotransformation reactions (33). In general, mEH has been described as a very important phase II enzyme (33). At least two functional polymorphisms have been determined in mEH, of which a high activity variant of the enzyme is coded by the mEH H¹³⁹R polymorphism in exon 4 (34). In the current study, individuals carrying the mEH H¹³⁹R polymorphism had 0.9 times lower 8-hydroxydeoxyguanosine levels as compared with mEH H¹³⁹R wild types. Although the level of this effect is very low, the observed pattern is as expected because the high activity variant of the enzyme (mEH H¹³⁹R, mEH R¹³⁹R) leads to more detoxification and, thus, to lower DNA damage levels.

A similar effect of mEH on 8-hydroxydeoxyguanosine levels was found in association with hexachlorobenzene exposure. Hexachlorobenzene is a persistent organochloric pesticide that accumulates in humans and is widespread in the environment because it is not well degraded (35). The oxidative metabolites of hexachlorobenzene are able to bind to macromolecules such as DNA (36). Urinary levels of 8-hydroxydeoxyguanosine were observed to increase in association with hexachlorobenzene exposure, and mEH H¹³⁹R wild-type individuals were defined as the more susceptible subgroup. Similar to the effect observed in response to PAH exposure, the level of the effect was marginal (0.9 times lower levels in heterozygous and homozygous carriers of the mEH H¹³⁹R polymorphism compared with wild types); however, the observed pattern indicates again a protective effect of the mEH H¹³⁹R polymorphism, where the low activity of the wild-type enzyme may lead to less detoxification of reactive metabolites and, thus, to higher DNA damage levels.

In addition, exposure to hexachlorobenzene was also found to be associated with increased micronuclei levels. It has been described that hexachlorobenzene is a genotoxic carcinogen and is able to produce a significant increase in micronuclei frequencies (36). Within this exposure-effect relationship, lower hexachlorobenzene-related micronuclei levels were observed for individual carrying the mnSOD V¹⁶A variant and for individuals carrying the CYP1A1 T³⁸⁰¹C variant alleles. The variant allele of the CYP1A1 T³⁸⁰¹C genotype leads to an increase in CYP1A1 enzyme activity (37). However, the role of this enzyme in biotransformation is rather complex because it can both activate and detoxify certain compounds. In addition, little is known regarding its role in the biotransformation of hexachlorobenzene. Nevertheless, the observed results for CYP1A1 T³⁸⁰¹C might be explained by a higher activity of the CYP1A1 T³⁸⁰¹C double variant in the detoxification of hexachlorobenzene, which leads to lower hexachlorobenzene-related micronuclei levels. Moreover, it has been shown both in rats (38) and in human breast tumors (39) that organochlorine pesticides such as hexachlorobenzene are associated with an increased oxidative stress. Regarding mnSOD, the mnSOD V¹⁶A variant is associated with a decrease in enzyme activity and, hence, an impaired antioxidant capacity against the oxidative metabolites produced during the metabolism of hexachlorobenzene. This is in contrast to what would be expected because less detoxification would lead to a higher accumulation of oxidative metabolites and, thus, higher micronuclei levels. Therefore, these results regarding mnSOD remain inconclusive.

Furthermore, an important exposure-effect relationship between Cd and urinary 8-hydroxydeoxyguanosine levels was found, which is affected by the GSTT1 genotype. Cd is a highly toxic metal and a well-known occupational (e.g., heavy metal mining, waste incinerators, and farm fertilizers) and environmental pollutant (water, food, and air contaminations; ref. 40). Occupational exposure to Cd has been associated with lung, prostate, pancreas, and kidney cancer, and it has been correspondingly classified as a category 1 carcinogen by IARC (41). Cd is not directly genotoxic, but it may affect genome instability by inhibition of DNA repair, depletion of cellular antioxidants such as glutathione (reduced), and indirect formation of reactive oxygen species (through the replacement of iron, zinc, and copper from various proteins such as ferritin, thereby increasing the level of free metal ions that participate in oxidative stress via fenton reactions; refs. 42-44). This Cd-induced oxidative stress has been described to cause 8-hydroxydeoxyguanosine-related DNA strand breaks (45). In addition, in the current study, a positive exposure-effect relationship between Cd exposure and 8-hydroxydeoxyguanosine levels was found. Within this relationship, individuals lacking GSTT1 activity have 1.1 times higher oxidative DNA damage levels after Cd exposure as compared with GSTT1-positive individuals. GSTT1 is an enzyme involved in the reduced glutathione pathway, one of the most essential antioxidant defense mechanisms. Thus, GSTT1-null individuals have an impaired antioxidant defense mechanism that results in an enhanced effect of oxidative stress, induced indirectly by Cd exposure, and thus, to higher 8-hydroxydeoxyguanosine DNA damage levels. This is confirmed by our observations.

Exposure to Cd was also found to be related to an increase in serum p53 levels, an association described for the first time in this study. It has been suggested by Xie and Shaikh (46) that Cd can activate p53 by the indirect generation of oxidative stress. For example, Cd has been reported to generate hydrogen peroxide (7), which in turn is known to trigger p53 transcriptional activity (47). Moreover, Cd is known to interfere with DNA repair, which in turn may also induce apoptosis (40). Analysis of the p53 tumor suppressor gene is proposed as a biomarker for effect in biomonitoring studies (48, 49). Within the observed exposure-effect relationship between Cd exposure and p53 levels, individuals homozygously carrying the BRCA2 N³⁷²H polymorphism were found to have significantly higher serum p53 levels than BRCA2 N³⁷²H wild-type and heterozygous individuals (1.3 and 1.4 times higher compared with wild-type and heterozygous individuals, respectively). In addition to the interference of Cd with the DNA repair system, carrying the BRCA2 N³⁷²H polymorphism, associated with a decrease in DNA repair, may lead to even higher DNA damage levels and, thus, an increase in Cd-induced serum p53 levels compared with average levels.

Another tumor marker, carcinoembryonic antigen, was also found to be increased upon exposure to a heavy metal, Pb. This is the first report that describes this relationship. Pb is an indirectly acting genotoxic metal, present in the environment as a pollutant of, for example, heavy metal mining, traffic waste by leaded fuel and as a food contaminant (e.g., drinking water by use of Pb pipelines). It is classified as carcinogenic by IARC, and (occupational) exposure to Pb has been reported to cause DNA damage as determined by Comet assay (50-52). In the current study, Pb-induced carcinoembryonic antigen levels were 1.9 and 1.6 times increased in carriers of the BRCA2 N³⁷²H polymorphism as compared with heterozygous and wild-type individuals, respectively. We cannot explain the functionality of elevated blood carcinoembryonic antigen levels in this case. However, it is striking that relatively higher levels of serum carcinoembryonic antigen have been reported in smokers compared with nonsmokers (53). In agreement to these findings, also in our study, smokers were observed to have 2.7 times higher carcinoembryonic antigen levels than nonsmokers upon Pb exposure.

In both subpopulations, the genotype was shown to significantly influence interindividual susceptibilities to environmental carcinogens. However, in most cases, observed impacts of these genotypes were relatively low. This may be the result of the rather low exposure levels of the Flemish population to the investigated compounds, which lead to weak exposure-effect relationships and, as a consequence, a low impact of genetic variations, including gender, predisposing for increased susceptibility. However, all observed relationships and effects of the genotype are biologically plausible and are statistically significant, which reduces the chance of bias.

Regarding the complexity of carcinogen exposure and the number of pathways that is involved in the modification of their effects, it is unlikely that one single-nucleotide polymorphism can be held responsible for these interindividual differences in genotoxic response. Investigating multiple polymorphisms in different pathways simultaneously should therefore be considered. Moreover, because these low-penetrance polymorphisms are risk factors for developing health effects in response to (environmental) exposures, future studies assessing genotyping as a biomarker for susceptibility must shift interest from exploring the effect of the genotype on the specific biomarker, toward the investigation of the effect of the genotype on the exposure-effect relationship between markers of interest. It is noted that in these types of study design investigating a large number of variables simultaneously, there is a chance of “multiple comparisons.” However, in accordance with the views formulated by Rothman (54), and because all observed findings are biologically plausible, in this study, P values are presented without adjustment for multiple testing.

In conclusion, this study has provided new insights in exposure-effect relationships in humans exposed to certain environmental contaminants and, secondly, has quantified the role of genetic polymorphisms in these relationships, enabling the identification of more susceptible subgroups. These findings suggest a role of interindividual differences in environmental cancer risk assessment. Although the results of this study should be interpreted carefully because overall effects were relatively modest and they comprise biomarker analysis instead of ultimate cancer incidence, environmental cancer risk assessment should challenge the assumption that all individuals are uniformly vulnerable to environmental pollutants.

No potential conflicts of interest were disclosed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

We thank the volunteers for participating and the nurses for the sample collections.