Genetics of Smoking-Related Cancer and Mutagen Sensitivity

Tobacco has central etiologic importance in lung, head and neck, urinary tract, and other cancers. At the same time, risk of these cancers in relatives of affected patients is increased after taking smoking into account, implicating a hereditary component. High-penetrance genes likely explain only a small proportion of familial cases, and thus the specific gene(s) that account for excess risk in relatives for tobacco-related cancers are unknown. Given the evidence for both specific carcinogenic exposures and hereditary influence in the genesis of this class of malignancies, these cancers may offer the best opportunity to unravel a core problem of cancer biology: How do genes and environment act in concert to cause cancer? Early attempts to identify relevant genes focused on candidate genes that activate or eliminate tobacco carcinogens. Until the late 1980s, virtually all of these studies involved phenotyping, i.e., administering an in vivo or in vitro probe drug and establishing a relationship between the disposition of the probe drug and its metabolites (a “ phenotype”) reflecting an underlying genotype. The acetylation phenotype (NAT2), trans -stilbene oxide in erythrocytes (GSTM1), the debrisoquine metabolic phenotype (CYP2D6), and aryl hydrocarbon hydroxylase (CYP1A1) are some of the phenotypes (genotypes) that have been evaluated in relation to smoking-associated cancers. Based on current understanding, there is likely an association between the slow acetylation phenotype and bladder cancer ( 1 ), but the other associations are controversial (GSTM1 and lung cancer; CYP1A1 and lung cancer) or null (CYP2D6 and lung cancer) ( 2 - 4 ). Mutagen sensitivity as described by Cloos et al. ( 5 ) in this issue of the Journal is another putative cancer-associated phenotype. It is distinct from the other phenotypes in that it is cytogenetically based and likely involves a number of genes involved in DNA repair (rather than a single phase I or phase II gene), there is not an obvious bimodal distribution of the underlying phenotype, and a specific gene that accounts for the trait is unknown. Generally, studies support the idea that there is a higher rate of spontaneous chromosomal aberrations in lymphocytes in individuals with cancer ( 6 ) and an increased rate of cancer is documented in the chromosome instability syndromes. Mutagen sensitivity involves testing the more focused hypothesis of whether particular mutagenic agents specifically influence chromosome integrity. By use of in vitro exposure to the radiomimetic drug bleomycin (capable of inducing single- and double-strand chromosome breaks irrespective of cell cycle) as a test mutagen, the number of chromosome breaks per cell in standard lymphocyte cultures is quantitated. There have been some data to establish that this trait “ tracks” in individuals ( 7 , 8 ) and that potential confounders, such as smoking, sex of the individual, micronutrient consumption, and alcohol consumption, do not account for differences in mutagen sensitivity between individuals ( 9 - 11 ). This trait may, therefore, measure a component of constitutional capacity to accomplish DNA repair. Bleomycin-induced chromosome sensitivity has been studied in relation to a variety of cancers, and findings have been generally positive ( 9,12-15 ). A key advantage of studying mutagen sensitivity as a phenotype in relation to cancer risk is that it assesses an activity mechanistically important in tumorigenesis and integrates a number of host factors (i.e., diverse mechanisms of DNA repair) in one measurement. In view of the broad associations reported for mutagen sensitivity with these cancers, consideration of the methodologic issues involved in phenotype studies is warranted. There has been an historic trend to substitute complex phenotyping assays with genotype determinations. Genotyping depends on germline DNA that is invariant to the presence of disease in the host and increasingly easy to obtain by both blood-derived and noninvasive (e.g., buccal cells) methods. Both the extraction of DNA and the analysis of multiple putative candidates have become faster, cheaper, and less invasive and require less biologic material than just a few years ago. There are downsides. The character of studies has evolved from focused hypothesis testing of a single phenotype-cancer association to broad screening of multiple candidates of sometimes uncertain functional significance with attendant multiple comparison problems. However, phenotype determination in cancer patients also presents challenges. In the case-control design, by far the most common approach used in “molecular epidemiology” studies, the concern is that the phenotype, rather than identifying a putative host susceptibility state, merely reflects pathophysiologic alterations in the host secondary to disease and is therefore only a nonspecific tumor marker. Such effects include abnormal nutrition, secondary metabolic alterations of neoplastic disease, and effects of treatment, hospitalization, inactivity, or stress. Cultured cells obtained from patients with cancer and control subjects in a hospital setting can differ for all of these reasons allowing bias due to differential misclassification. A prospective cohort provides a theoretically ideal setting to avoid these problems because subjects are studied prior to onset of cancer; however, the number of assays required on healthy members of the cohort to obtain a sufficient number of affected patients with cancer would be very large. A nested case-control study performed on prospectively collected samples avoids this problem; however, the reliability of mutagen sensitivity using frozen samples has not been adequately studied, and exposure information may not be as broad as in a case-control study. Thus, the cancer associations reported with mutagen sensitivity have been cautiously interpreted mostly due to the concern that cells obtained from subjects with newly diagnosed cancer, even if untreated, may reflect the consequences of the cancer in the individual rather than the host's susceptibility. In this context, the potential importance of the finding of Cloos et al. ( 5 ) is clear. Inherited genes are invariant to all of these factors, so that the identification of a gene(s) that provides the same information as the phenotype would be a major methodologic improvement. The segregation analysis conducted by Cloos et al. ( 5 ) is generally convincing, but some cautions apply. The substantial estimate for the heritability of 0.75 accounted for by genetic factors reflects a pooling of subjects from head and neck squamous cell carcinoma kindreds and twins, groups who are somewhat heterogeneous. The lack of a clear bimodality in the trait distribution may suggest sensitivity to the threshold selected for breaks per cell and/or a polygenic origin of the trait. While bleomycin has been the mutagen best studied in this assay, polycyclic aromatic hydrocarbons, such as benzo[ a ]pyrene diol epoxide (BPDE), are more relevant to tobacco-associated cancer. Since BPDE-induced adducts would be repaired by the nucleotide excision pathway, the mix of genes involved will differ. Early findings suggest that BPDE sensitivity in lymphoblastoid cells is also associated with risk of smoking-related cancers ( 16 , 17 ). The apparent broad relevance of mutagen sensitivity to diverse cancers also suggests that multiple genes involved in different pathways of DNA repair may all be contributing to the integrated phenotype. The findings reported do not exclude that nongenetic factors, such as serum folate, play an important role in mutagen sensitivity. Further conducted studies of this phenotype are warranted, while efforts continue to sort out the specific gene mutations and polymorphic variants associated with different DNA repair pathways and to determine the degree to which lesions associated with particular carcinogenic exposures depend on each for repair ( 18 ). Of particular interest will be the examination of plausible candidates such as bleomycin hydroxylase and the genes in the various DNA repair pathways for associations with both mutagen sensitivity and cancer ( 19 ). The eventual identification of the genes that account for the mutagen sensitivity trait will enhance mechanistic understanding and allow better characterization of risk. This should lead to a more precise understanding of the contributions of exposure, heredity, and chance in these cancers, an advance that should lead to new insights for prevention and treatment.