Interaction between dose and susceptibility to environmental cancer: a short review.

Increased risk of environmentally induced cancer is associated with various types of exposures and host factors, including differences in carcinogen metabolism. Since many carcinogenic compounds require metabolic activation to enable them to react with cellular macromolecules, individual features of carcinogen metabolism may play an essential role in the development of environmental cancer. In this context, cigarette smoking has often been the main type of carcinogenic exposure examined in human studies. Increasing attention has recently been paid to the dose level at which individual susceptibility may be observed. Present studies on increased risk of smoking-related lung cancer associated with phenotypic or genotypic variation of the genes encoding for CYP1A1 or CYP2D6 enzymes are summarized. Similarly, higher risks of lung or bladder cancer seen at various levels of smoking in association with polymorphism of the glutathione S-transferase gene GSTM1 or NAT1 and NAT2 genes involved in N-acetylation are reviewed. Finally, the influence of CYP2E1, GSTM1, or the combined at-risk genotype on the risk of hepatocellular carcinoma in smokers is briefly discussed.


Introduction
Increased risk of cancer and other environmentally induced diseases is associated with various types of exposures and host factors, including differences in carcinogen metabolism. Since many carcinogenic compounds require metabolic activation before they can react with cellular macromolecules, individual features of carcinogen metabolism may play an essential role in the development of environmental cancer (1). Many compounds are converted to reactive electrophilic metabolites by the This paper was prepared as background for the Workshop on Susceptibility to Environmental Hazards convened by the Scientific Group on Methodologies for the Safety Evaluation of Chemicals (SGOMSEC) oxidative, mainly cytochrome P450-related enzymes (CYPs). Secondary metabolism, mainly involving epoxide hydrolase and another subset of activating CYP isoforms, leads to the formation of the highly reactive metabolites that can bind to genomic DNA (2). Other enzyme systems, e.g., many transferases such as glutathione S-transferases, UDP-glucuronosyltransferases, sulfotransferases, and acetyltransferases, take part in detoxification of oxidated forms of carcinogens. Thus, the concerted action of these enzymes may be crucial in determining the final biological effect(s) of a xenobiotic chemical. A number of genes that encode carcinogen-metabolizing enzymes is presently known. Individual variation in enzymes activating or detoxifying carcinogens and other xenobiotics have subsequently been related to discovered genetic polymorphisms for these genes (1).
Recently, increasing attention has been paid to the dose level of exposure at which individual susceptibility may be observed (3). In this context, cigarette smoking is the main type of carcinogenic exposure examined in many human studies. Although it is well known that the risks for tobaccorelated cancers increase with increased smoking (4) and that the composition and filter type of cigarettes modifies the risk (4), few studies have focused on possible dose dependence of cancer susceptibility.
The possible role of genetic factors in the regulation of the dose-response relationship after exposure to carcinogens has been seen in animal studies. A difference related to genotype in formation of 2-aminofluorene-hemoglobin adducts was found in hamsters between slow and rapid acetylators; adduct formation increased more with dose in slow than in rapid acetylators (5). Similarly, formation of 4-aminobiphenyl-DNA adducts in mice was increased with dose in the liver and urinary bladder, but the effect appeared to be sex-and organ-specific (6).

Pulmonary Carcinogenesis
Inducibility of CYPlA-related phenotypic activity can be regarded as an indirect quantitative measurement of effects related to dose and host factors. In the lungs an implication of this is the decreased level of inducibility found in lung cancer dependent on the time period patients refrained from smoking, while in noncancer patients no cigarette smoke-related inducibility could be demonstrated (7). Inducibility of CYPIAJ phenotype was first demonstrated by Kellermann et al. (8) to be higher in lung cancer patients than in controls. This association has also been shown using CYPIAI activity in cryopreserved lymphocyte as the marker (9). Since these studies, association of CYPIAJ genotype with inducibility and lung cancer risk has been extensively studied (10)(11)(12)(13)(14)(15)(16)(17)(18)(19). Anttila et al. (20) demonstrated the presence of CYPlAl isozyme in the lung tissue of smokers with peripheral lung cancer. In the Japanese population, associations between risk of lung cancer and certain CYPIAI alleles have been demonstrated in smokers (14)(15)(16); the association has not been confirmed in Caucasian populations, probably due to more infrequent occurrence of the at-risk allele in Caucasians (17,18,21,22). Possible dose dependence of CYPlAl-associated metabolic susceptibility has not been much examined. One of the first reports was a Japanese study (23) that observed that patients with the at-risk MspI CYPIAI genotype (m2m2) contracted cancer after fewer cigarettes (mean lifetime consumption 31 ± 12 x 104 cigarettes) than those with other genotypes (mean lifetime consumption 43 ± 18 x 104 cigarettes). The individuals with the susceptible genotype were then found to have a high Environmental Health Perspectives * Vol 105, Supplement 4 * June 1997 risk (OR 7.3, 95% CI 2.1-25) at a low dose level and the difference in susceptibility between the genotypes was reduced at high dose levels (23). In a later report (24), the finding of increased lung cancer risk associated with the MspI CYPlAI susceptible genotype was confirmed; an excess risk was seen particularly in light smokers (less than 30 x 104 cigarettes over lifetime) with squamous cell carcinoma as contrasted to those who had smoked more. In adenocarcinoma, no such relationship was found (Table 1). A recent study among African Americans (25) observed no overall association between lung cancer and a rare race-specific CYPIAI polymorphism, although a slightly increased but nonsignificant risk was associated with the presence of the variant CYPJAl allele at higher smoking level (1-35 pack-years, OR 1.3, 95% CI 0.6-3.2; > 35, OR 2.2, 95% CI 0.6-7.8). Taken together these results appear to emphasize the influence of different allele frequencies in different human populations on risk experienced.
In genotyping studies using polymerase chain reaction assays (26,27), a tendency for an association between the GSTMI genotype and squamous cell carcinoma has been reported (26,28,29). A Japanese study combining the risk genotypes for CYPIAI and glutathione S-transferase 1 (GSTMI) demonstrated an enhanced risk of lung cancer 30). Patients with the atrisk MspI or Ile-Val genotypes of CYPIAI contracted lung cancer after smoking fewer cigarettes than those with the other CYPIAI genotypes. In combined genotyping, the individuals with the susceptible  (30). Similarly in another study, the GSTMI null genotype was associated with an overall increased lung cancer risk (OR 1.87) (31). The study found a higher risk at the highest cumulative dose of cigarette smoke exposure in squamous cell carcinoma patients with GSTMI null genotype; the odds ratio however, remained nonsignificant or at borderline significance (Table 2). Association between some of the biomarkers related to initiation of carcinogenesis and susceptible genotypes has been noticed. Formation of smoking-related DNA adducts in human lung tissue can be used as a marker of internal dose, and smokers are known to have significantly elevated levels of aromatic and/or hydrophobic adducts compared with nonsmokers (32,33). In recent studies, up to 6-fold increase in the amount of pulmonary DNA adducts has been found in smokers as compared to ex-smokers or nonsmokers, with a tendency of GSTMJ null genotype individuals to have higher levels of adducts than those with at least one allele present (34).
Several studies have suggested that the capability to metabolize debrisoquine (a test drug) extensively (extensive metabolizers, EM) is associated with increased risk of lung cancer, as compared to poor metabolizers (PM) (35). A weak association between CYP2D6 genotype predictive of the EM phenotype and increased risk of lung cancer has since been reported by some studies, but other studies found no association (36)(37)(38)(39)(40)(41). Recently, a positive association between the frequency of poor metabolizers and the risk of adenocarcinoma of the lungs was published (42). The association of PM genotype of CYP2D6 with increased risk of lung adenocarcinoma was speculated to be due to metabolic activity by this isozyme, a tobacco-specific nitrosamine 4-(methylnitrosamino)-1-(3pyridyl)-1-butanone (NNK), as also suggested by others (43). Present evidence for interaction between the risk related to CYP2D6genotype/phenotype and the extent of smoking is difficult to evaluate because most studies lack quantitative data on smoking. Bouchardy et al. (44) observed an increased risk of lung cancer for extensive metabolizers. The excess risk in EMs was dependent on the level of smoking, while in individuals who were poor or intermediate metabolizers, no such increase was found even at high level ofsmoking ( Table 3). The increased risk (OR 4.95) in extensive metabolizers became evident at the tobacco consumption level of > 30 g/day as compared to smokers who consumed < 20 g/day. This was again speculated to be associated with the increasing levels of metabolites of the tobacco-specific carcinogen NNK present in extensive metabolizers (44).
Occupational exposure to asbestos increases the risk of lung cancer and mesothelioma. Indications of increased risk of asbestos-related mesothelioma and other pulmonary disorders have been obtained recently. Both GSTMI and NA T2 (Nacetylation) genotypes were associated with malignant mesothelioma in workers who were highly exposed to asbestos (45). In another study, GSTMI null genotype was found to be related to risk of asbestosis (pulmonary fibrosis) among workers exposed to high levels of asbestos (46), whereas no association between radiographical or lung function changes and GSTMI genotype was seen in a group of asbestos workers with mostly low or moderate exposure (47). Glutathione S-transferase   (44). bNumbe studied in parentheses (cases/controls).  (51). "Determined by urinary nicotine + cotinine concentration. bNumber studied in parentheses. isozymes take part in reactions detoxifying reactive oxygen species (48), known to be generated by asbestos exposure and smoking. The excess risk associated with GSTMI null genotype in the asbestos-exposed patients in these studies appeared, however, to be exerted at a high level of exposure only. The possible mechanistic role of Nacetylation genotype in modifying risk for asbestos-related malignant mesothelioma is more difficult to see.

Bladder Carcinogens and Bladder Cancer
Malaveille et al. (49) found a dose-dependent relationship between urinary mutagenicity, amount of thioethers, and the number of cigarettes smoked per day as well as the amount of nicotine and cotinine in urine. Furthermore, the mutagenicity was nearly twice as high in dose-corrected black tobacco smokers than in blond tobacco smokers. Bartsch et al. (50) showed a dosedependence of ABP aminobiphenylhemoglobin (ABP-H6) adducts and the concentration of cotinine + nicotinine in urine, but no clear-cut quantitative association could be shown between the amount of adducts and the amount of smoking. Later, a dose-dependent increase in ABP-Hb adducts was demonstrated (51).
Furthermore, the amount of ad( higher in slow than in rapid N-a (determined both genotypically notypically) at low dose levels ( by urinary cotinine + nicotine le at higher dose levels the differeni to level off ( Table 4). The diff acetylation phenotype was also re the presence of DNA adducts t biphenyl metabolites in urothe and the amount of adducts incre increasing urinary mutagenicity.
within the same genotype no cc between the acetylation rate as a caffeine metabolites was found (5 Aspects related to dose might I tant in evaluating the postulated tion of bladder cancer risk by r polymorphisms, e.g., those for GSTM1 (52)(53)(54)(55)(56). In the study al. (57), an increase in risk o0 cancer was found for patients GSTM1 null genotype than for t were homozygous or heterozygoi GSTMI wild-type allele. No si relationship with dose could although some difference betweer levels was observed (< 50 pack > 50 pack-years;

43
with GSTMI null genotype, did not find dependence of the risk on increasing amount -7 of smoking; in contrast, the risk appeared to be associated with lower dose level. In keepr of subjects ing with other studies, both NATI and NAT2 enzymes catalyze the activation (0acetylation) and inactivation (N-acetylation) of arylamine carcinogens. Polymorphism in ducts was acetylation rate was first associated with cetylators NAT2 gene locus, NAT2 (59). In a study on and phe-occupational exposure to bladder carcinoas judged gens, low acetylation rate was observed to vels), but offer moderate protection from bladder Lce tended cancer in benzidine-exposed workers (54). erence in Cigarette smoking appeared to influence the .flected in risk, with a somewhat higher estimate for :o amino-higher level of smoking (< 20 pack-years, lial cells, OR 1.4; > 20 pack-years, OR 1.6).
-ased with ased withOther Malignancies However, Drrelation Risk of hepatocellular carcinoma (HCC) ssayed by was studied in a Taiwanese population, 1). most of whom were also positive for hepatibe importis-B antigen (60). In this study, the signifimodificacance of cytochrome P450E and GSTMI metabolic genotypes was examined. An increase in NAT2 or the HCC risk was reported to occur by Bell et depending on the dose of cigarette smoke f bladder exposure in patients who were homozygous with the for the cl allele (cllcl) of the CYP2EI those who gene. No association with GSTMI null us for the genotype was found (Tables 6, 7). Defective ignificant alleles of the CYP2D6 gene may also be be seen, involved in increased risk of liver cancer; smoking subjects having functional gene (extensive -years vs. metabolizer genotypes) seemed to have 6.4-m6ller et fold higher risk of primary liver cancer than 1.40-fold those who had defective alleles (61).   (1.14-3.77) Abbreviations: OR, odds ratio; Cl, confidence interval. "Modified from Lin et al. (64). metabolized by this cytochrome P450 isozyme are not known.
Another malignancy for v i modification of risk by metabolic p. orphisms have been implied is color. .1 cancer. Rapid acetylators (NAT2) may oe at risk for colorectal cancers (62), although a study by Bell et al. (63) in a British population could not confirm this. Instead, an increased risk (OR 1.9) was found to be associated with NAT*J 0 allele of the NAT] gene, which encodes a different acetyltransferase isozyme (59). However, the increased risk associated with the NAT] variant allele was most apparent in NA T2 rapid acetylators (OR 2.8), proposing a gene-to-gene interaction. Lin et al. (64), on the other hand, studied possible influence of GSTMI on the risk of colorectal adenoma. They found that the risk was increased with smoking, but no associationi with GSTMI genotype was found (Table 8).

Conclusions
Susceptibility to certain environmental cancers due to variation in xenobiotic metabolism has been demonstrated in many studies using either phenotyping or genotyping assays. Some of the studies, e.g., those on lung cancer, have reported conflicting results. It is only recently, and so far in relatively few studies, that influence of exposure levels or internal dose has been examined as a potential modifier of increased cancer risk linked to genetic susceptibility. Relationship between dose and increased metabolic susceptibility to cancer appears to vary from gene to gene, and with respect to combinations of genotypes as well as the organ under study (i.e., tissue specific expression of the isozymes). On the basis of present studies it can be speculated that in one tissue, increased susceptibility may become evident at low exposure doses while in the other this is seen at a high exposure dose only. Present knowledge does not allow a conclusion as to whether this hypothesized variation could be related to the overall capacity of the tissue to metabolize xenobiotics. In such a case, high exposure levels might saturate metabolic capacity by producing large quantities of reactive intermediates, while at low doses individual variations in metabolism might be more significant. Where the increased risk associated with susceptible genotypes is observed only at high doses, it could be that tissue-specific repair mechanisms would be efficient in protecting against carcinogenicity at low doses, thus masking possible influence of genetic susceptibility factors. It is also possible that other than host-related protective factors might be involved. For example, different dietary habits between populations could result in such a situation.
Another major point to be taken into consideration is the remarkable variation in metabolic phenotypes and genotypes reported for different ethnic or geographic populations (65)(66)(67)(68)(69). When comparing the influence of genetic susceptibility to lung cancer between ethnically different populations, e.g., Caucasians and Japanese populations, and taking smoking habits also into account, the results obtained vary greatly; in large part, this can be attributed to differences in distribution of the at-risk alleles in these populations. In a study by Caporaso et al. (35) in which extensive metabolizers of debrisoquine were found to be at 4.5to 10.2-fold risk of lung cancer, a 2-fold difference between whites and blacks was observed. Epidemiologic studies have pointed to differences in smokingassociated lung cancer risk between various ethnic groups. In a Hawaiian study in which many confounding factors-e.g., smoking habits, pack-years of smoking, cancer histology, occupation, education, and dietary factors-were adjusted for, researchers observed a greater than 2-fold difference in cancer risk between Japanese and Hawaiian men (70).
In conclusion, the present data appear to support suggestions that increased susceptibility to environmental cancer due to phenotypic or genotypic variation in carcinogen metabolism may be different at different levels of exposure, although other factors, internal or external, are likely to be involved also.