Document Rj90RNq8G3awbpYBeRE6xeXZE

Molecular Epidemiology and the Genetics of Environmental Cancer Peter G. Shields, MD, Curtis C.Harris, MD e' Environmental, occupational, and recreational exposures to carcinogens contribute to cancer risk in humans. Cancer formation is a multistage process involvingtumor initiation,promotion,conversion,and progression. Carcinogens canaffect anyof these stagesthroughgenetic andepigeneticmechanisms, The association of a suspected carcinogenic exposure and cancer risk can be studied in populations with classic epidemiologic techniques. However, these techniques are not applicable to the assessment of riskin individuals. Molecular epidemiology, in contrast, is a field that integrates molecular biology, invitro and in vivo laboratory models, biochemistry, and epidemiology to infer individual cancer risk. Carcinogen-macromolecularadduct levels, and somatic cell mutations can be measuredto determinethe biologicallyeffective dose of a carcho- Sen. Molecularepidemiologyalso exploreshost cancer susceptibilities, such as carcinogen metabolic activation, DNA repair, sndogenous mutation rates, and inheritanceof mutatedtumor suppressorgenes. Substantial interindividualvariation for each of these biologic end points has been shown and, therefore, highlights the need for assessing cancer risk on an individual basis. Given the pace of the last decade, it is feasible that the next 10years will allow molecular epidemiologists to develop a cancer-risk profile for an individual that includes assessmentof a numberoffactors. This will helpfocus preventivestrategies and strengthenquantitative risk assessments. (JAMA. 1991;266:681-687) THE ~ T U D Yof environmental carcinogenesis finds its roots in the epidemiology of life-styleirelated and occupationally related cancers.' In 1769, John Hill reported that tobacco snuff could cause oral cavity cancers. This was followed by a publication from Perci.val Pott who concluded that working as a chimney sweep, exposure to soot, and poor hygiene led to scrotal cancer. Other associations followed, such as lung cancer and asbestos, arsenic, or radium: bladder cancer and aromatic amines; bone sarcoma and radium; skin cancer and ionizing radiation; and leukemia and radiation or benzene. Equally important to the study of environmental carcinogenesis are .studies based on laboratory animal and in vitro cell and tissue studies. As From the Laboralory of HumanCarcinogenests,Nalional Cancer Institute. National Institutes of Health. Bethesda. Md. Reprint requesls 13 Laboratory of Human Carcinogenesis. NalionalCancer Institule.Nationallnstllutesof Health.Bldg37,RoomXD1,Bethesda.M D 20892(Dr Harris). early as 1918, and in relationship to Pott's findings, Yamigawa and Ichikawa reported that coal tar can cause skin cancer in laboratory animals. Later studies reported that individual chemi- cals can cause cancer. "he assessment of cancer risk is a multidisciplinary process, Methods and resources to assist physicians for estimating risk in individuals have been proposed.' However, quantitative risk assessments, used by regulatory agencies to assess risks in populations, are not applicable to risk in individuals. Mo- lecular epidemiology, in contrast, is a field that uses molecular genetics and biochemical techniques to assess an in- dividual's cancer risks. Central to mo- lecular epidemiologyis the study of car- cinogen exposure, metabolism, and DNA repair. These factors determine the biologically effective dose of a carcinogen. Equally important, an individual's cancer risk is related to endogenous rates of mutation, genetic instability, alterations in proto-onco- genes (naturally occurring genes that, when altered, contribute as oncogenes to cancer formation) and tumor s u p pressor genes (naturally occurring genes that are involved in controlling normal cell growth, differentiathn, and development). This article will review the current understanding of the multistage process of carcinogenesis, its role in molecular epidemiology, and the a p plication of molecular epidemiology to assessing carcinogenic risk. (Due to the limitations of citing multiple references herein, review articles are frequently cited and individual studies are offered as examples of the available literature. The reference list is not intended to be all-inclusive.) MULTISTAGECARCINOGENESIS Carcinogenesis is a multistage process of normal growth, diiTerentiation, and development gone awry. It is driven by spontaneous and carcinogen-induced genetic and epigenetic events. Figure 1presents a simpued scheme. Tumor initiation involves the direet ef- fects of carcinogenic agents on DNA, mutations, and altered gene expres- sion. The attendant defects are involved in tumor promotion, whereby cellshave selective reproductive and clonal expansion capabilities through altered growth, resistance to cytotoxicity, and dysregulation of terminal differentiation.' Tumor promotion further involves an "initiated"cellular clone that may also be affected by growth factors that control signal transduction.' Dur- ing this p m e s s , progressive phenotyp- ic changes and genomicinstabilityoccur (aneuploidy,mutations, or geneamplification).'.' These genetfc changes enhance the probability of initiated cells transforming into a malignant neoplasm, the odds of which are increased during repeated rounds of cell replication.' During tumor progression, angiogenesis allows for a tumor to grow beyond l or 2 mm in size.' Ultimately, JAMA, August 7,1991-Vol266, No.5 MolecularEpidemiology-ShieldsL Harris 681 Carcinogen Exposure Initiation Promotion Conversion Progression S Fig1.--Multistageprocessof carcinogenesis. 0 Activation of Proto-oncogenes ' 0 lnadivation of Tumor Suppressor Genes 0 Inactivation of Antimetastasis Genes tumor cells can disseminate through tivated by mutagenic mechanisms. For Table 1.-Examples 01 Functions of Tumor Sup- vessels, invading distant tissues and es- example, the p53 tumor suppressor pressorGenes' . tablishing metastatic colonies.' gene, located on chromoaome 17,is the The role of proto-oncognes and their most commonly altered suppressor activation have become increasingly a p gene among all tumors studied so far.." parent in the multistage model of carci- Germ line mutations of p53 have been nogenesis. These genes are important found in familial cancer syndromes (Li- to the regulatory mechanisms of Fraumeni syndrome).I6This gene nor- growth, cell cycle, and terminal differ- mally encodes for a phosphoprotein inentiation.' Activation of proto-oncwrf volved in control of cell proliferation. *FnnnHanis.' . genes enhances the probability of neo- 'Single-base substitutions can result in plastictransformation, which can either loss of function or production of proteins be'an early or late event. Carcinogens that either interfere with normal func- ma, leukemia, and thyroid.' Base sub- can cause mutations in proto-oncogene tion or otherwise directly enhance neo- stitutions a t codons 12, 13, and 61 on DNA sequences, or they can act as tu- plastic transformation." exposure to such agents as nitrosa- mor promoters enhancing the activity of ,,,memethod by which carcinogenesis mines, polycyclic aromatic hydrocar- oncogeneprotein products. affects DNA and Droduces mutational bons, and radiation causes activation in Tumor iuppresjlor genes also play an spectra is varied. dhemical carcinogens animal models."" important role in carcinogenesis.'.' Like interact directly with DNA by covalent- Some mutations may reflect specific proto-oncogenes, these genes regulate ly binding to nucleotides and forming carcinogen exposures and may exhibit cell growth and terminal differentia- adducts (Fig2). If present at the time of target organ specificity. For example, tion. However, they have the opposite DNA synthesis, these adducts can p63 mutations at codon 249, frequently effect by limiting growth and stimulat- cause mutations. Nitrosamines, for ex- observed in hepatocellular carcinoma ing terminal differentiation (Table 1).If ample, which are present in the diet and from China*' or southern Africa," are inactivated, this type of dysregulation arelinked to esophagealand gastric car- consistent with the type of damage increases the probability of neoplastic transformation and is a dominant function. Tumor suppressor genes are re- cessively inherited and generally re- cinoma, result in base substitutionsdue to mispairing at sites where adducts are formed. h r t h e r , site specificity may be dependent on the type of nitrosocom- dcaused by ailatoxin B,exposure, a com- mon dietary carcinogen linked o this tumor. In contrast, several t of p63 mutations have been observ in lung quire lossofboth genomicalleles. This is pound affecting specificoncogene or tu- cancer,"which is consistent w t h a mul- exemplified by inheritance of a predis- mor suppressor gene function."Among tiple carcinogen exposure from tobacco position to retinoblastoma and/or osteo- the best studied is the m8 fsmily of smoke. sarcoma, where patients possess a ho- proto-oncogenes. The rus protein prod- Some carcinogens can damage DNA mozygous loss of the Rbl gene on ucts are involved in signal transduction directly, eg, bis(chloromethyl)ether, chromosome 13."" In the familialform, pathways initiated by growth factors but most require metabolic activation loss of one allele is inherited and the and hormones at cell membrane recep by cytochrome P-450 metabolic en- other is lost through later mutation. Loss of suppressor and antimetastasis genes can be an early or late event involving several steps, including angio- tors. In several experimental systems, activ&tionis associated with tumor for- mation," angiogenesis,' and metastasis." Mutation of ~ a p9roto-oncogenes zymes.8 The primary role of cyto- chrome P-450enzymes is detoxification. During metabolism, howeber, functional groups can be added or exposed that genesis and metastasis.'S has been observed in human cancers, result in the formation of reactive elec- Tumor suppressor genes can be inac- includinglung, colon, pancreas, melano- trophilic intermediates that bind cova- 682 JAMA,August 7,1991-Vol266, No.5 Molecular Epidemiology-Shields 8 Harris k Adenine R Guanine A Cytosine R Thymine Fi2.-Sites of alkylationof DNA under physiologi- CBIconditions. NIrosamines are metabolizedto re- activeintermediates t a t canbindto nucleotidesala variety of sites (indicated by asterisks). These lesions can be promutagenic (data adapted from PeggW). lently with nucleotides and form DNA adducts (Fig2). Among the best studied are the epoxidation reactions in the metabolism-of polycyclic aromatic hydro-, carbons and- aflatoxin B,. Cytochrome P-450enzymes can be either induced or constitutively expressed. Acetyltransferases and glutathione transferases are other types of metabolic changes. Decreased function of these enzymes also results in increased accumulation of toxic intermecXatesof mutation. Independent of carcinogen exposure, human cells are continuously undergoingspontaneousmutations at a low rate, eg, oxidative damage, polymerase infidelity, chromosomal rearrangement, recombinaseinfidelity, and telomere reduction. The process of cell and DNA replication can increase the rate of mutation.' When one considers that the human body contains 10" cells and that these cells undergo 10'' divisions over a person's life span, it is possible that genomic instability plays an important role in Oxidative DNA damage following free radical activation is the result of both endogenous and exogenous factors. Endogenously, oxidative damage results from the production of active oxygen species mediated through hydrogen peroxide, superoxide anions (0,9 and hydroxyl radicals (OH-) commonly found in cells associated with inflammation. Lipid peroxidation also can cause oxidative damage. Ionizing radiation, benzo[alpyrene, benzene, cigarette smoke (due to the presence of catechols, quinones, nitrogen oxides, and other agents), and asbestos all cause JAMA, August 7,1991 -VOl266, NO,5 oxyradical damage.-Qxidative DNA combined events are more important damage can be prevented chemically than the actual order in which they (vitamin E, glutathione, uric acid) or OCCW. enzymatically {superoxide dismutase, catalase, peroxidase)."." On a molecular basis, cells also pos- CARCINOGEN EXPOSURE ASSESSMENT sess the ability to repair DNA damage. Genetic damage from carcinogens re- An extensively studied repair enzyme sults from a number of detmminants, is 06-alkylguanine-DNA-alkyltransfer- including exposure, absorption, metab ase. This enzyme repairs damage from o h m , and DNA repair, Each of these ~ alkylating agents such as nitrosa- can be affected by host factors. The mines." It is a suicide protein in that it amount of carcinogen that reaches the transfers the alkyl group to itself and target macromolecule (eg, DNA) in a becomes inactivated. Cell cytotoxicity sensitive organ \1sthe biologically effec- and tumor cell resistance are negatively tive dose. Measurements of DNA and correlated with the levels of this en- protein adducts in humans have been zyme," and levels vary within organs the focus of research in numerous lab- and among people.2g" oratories. Central to these studies, In almost every step of the multistage however, is the application of sensitive process of carcinogenesis, person-to- and specific assays that are required to person differences in cancer susceptibil- detect femtomole and attomole levels of ity can be found." Interindividual dif- adductsin microgram amounts of DNA. ferences for particular traits can be Current methods are challenged be- acquired or inherited. Inheritance of causeof the complexity and multitude of the ability to metabolize debrisoquin possible exposures in human tissues. sulfate, an antihypertensive medica- Previous research has demonstrated tion, is correlated with lung cancer the need to develop assays using au- risk."." Another example is the aryl hy- thentically synthesized standards and drocarbon hydroxylase enzyme in- micropreparative techniques. More- volved in the metabolism of w- ls-cy-clic over, corroborative techniques are aromatic hydrocarbons. Activity varies needed for validation. in lung tissue" but can also be inducible A variety of assays are available to on exposure to agents such as tobacco identify carcinogen DNA adducts in hu- smoke."Induction can even vary among man tissues. Theseincludethe phospho- individuals and is notably higher in pa- rus SZpostlabeliig assay nucleotide tients with lung cancer than in non- chromatography, immunoassays, fluo- cmcer controls.' Levels of aryl hydro- rescence spectroscopy, gas chromato- carbon hydroxylase activity have also graphy-mass spectroscopy, and electri- been correlated with prognosis in lung cal chemical detection (reviewed in a cancer." study by Shields et al"). The choice of The multiple stages of carcir'ogenesis tissue to be studied (eg, blood, lung, are best exemplified by a model of hu- buccal mucosa, or urine) can be critical man colorectal tumorigenesis, as re- for determination of exposure assess- cently described by Vogelstein and oth- ments because of the multiple factors ers (summarized in a study by Fearon that determine the degree of adduction. and Vogelstein"). In the early stages, Expo~turesto polycyclic aromatic hy- and apparently more common in pa- drocarbon (PAH) compounds am asso- tients with familial polyposis, loss or ciated with an increased risk of lung and inactivation of a candidate tumor s u p skin cancer. Industrial pollution, fo rsil pressor gene, MCC located on chromo- fuels, and tobacco smoke account forthe some 5q, is associated with cellular hy- major environmental sources. Dietary perproliferation. Later, a phenotypic exposures also occur due to overcooked appearance of adenoma occurs that is or charcoal-broiledmeats. Carcinogenic accompanied by hypomethylation with PAHs are metabr.hxl by cytochrome genomic instability. More advanced tu- P-460 monooxygtllases, and reactive mors involveoncogenes and tumor s u p epoxide intermediates readily form ad- pressor genes not observed in early ade- ducts in DNA. Adduct levels have been nomas. These include E-rasmutations correlated with exposure in cokesven on chromosome 12; mutation of p53 tu- workers,- tobacco consumption,u and mor suppressor genes on chromosome urban vs rural residences," but de- 17;and a deletion of DCC, the putative creases during vacation from occupa- tumor suppressor gene on chromosome tional s o w e s have been observed.' 18q and that may be involved in cell-to- Seasonal variation in adduct levels has cell adhesion and pssibly metastasis. also been observed.* Somestudies have Other allelic losses can occur in other not found correlations with tobaccocon- chromosomes. Thus, it appears that at sumption, but this may be due to other least six genetic events occur in the de- sources of exposure. For example, di- velopment of colorectal carcinoma. The etary exposures to PAHs can contrib- Molecular Epidemiology-Shields8 Harris (u)3 . ute significantly to adduct levels in i ~ p h o c ~47-(s8 . Nitrosamnes and other nitrosocom- pounds are potential human carcino- gens. These compoun& readily alkylate DNA and form adducts but most re- quire metabolic activation. Exposure can occur through endogenous forma- tion of nitrosamines or directly from di- etary sources, cosmetics, drugs, house- hold commodities, and tobacco smoke. Endogenous formation occurs in the .. stomach from the reaction of nitrosata- ble amines and nitrate, which are con- verted to nitrites by bacteria. Endoge- nous nitrosamine formation can be . estimated by administration of proline, a nontoxic precursor that reacts with nitrites and is excreted in the urine. .. Excretion of nitrosoproline varies among individuals' and is associated with risk of stomach"*@and esophageal cancer."*"Whether this is related to en- dogenous host factors or coexisting di- etary exposures remains unknown. I t has also been shown that coadministra- tion of vitamin C can reduce the rate of endogenous nitrosation. The reaction of nitrosamine metabo- lites and DNA resultsin several types of adducts, only some of which are promu- tagenic. These adductscan be measured in human tissues (Fig 3) and have been correlated with cancer incidence.- Us- ing a radioimmunoassay for O'-methyl- deoxyguanosine, adduct levels were found to be higher in persons with esophageal cancer in China." In'a sepa- rate study, levels of O'ethylthymine were found to be higher in liver samples from persons with liver and other cancers." Among the best studied potential di- etary carcinogens are aflatoxins pro- duced by Aepergillusjlavw, and Asper- gillus parasiticus. These molds are contaminants of corn, peanuts, sor- ghum, rice, and other agricultural prod- ucts. Levels of aflatoxin vary between regions. Aflatoxin B, consumption has been linked to primary hepatocellular carcinomain areaswhere liver cancer is common, eg, the Orient and Africa. Metabolicactivation of aflatoxin B, var- ies amonein'dividuals, resulting in dif- 1 ferent levels of adducts." Exposure and adduct levels are inversely correlated with residence in industrialized coun- tries."'."Exposure also has been linked to a mutation in the p53 tumor suppres- sor gene in hepatocellular carci- nomas."" Aromatic amines have been implicat- ed in bladder carcinogenesis, especially in occupationally exposed cohorts and tobacco smokers. Metabolic activation occurs via multiple cytochrome PA50 ' and acetyltransferase." Internal dosim- 1.1.1.1, 5 10 15 20 25 30 35 40 45 Time, rnin Exp. = lo' 0 Fig 3.-Representathn, analysis of human lung DNA for the detwtbn and quantitcltknof N7methyWoxy. guanodne (N7methyidGp) by combining high-portonnance liquid chhatography and phosphorue 32-postlabelIng assay. DNA was enzymatkxllydigested to lndhddudnucleotklesthatwenthen Wated by high-performance chromatography (top). Radbactb orthophosphateh enzymaticallytransfend to the nacleotideo and then subjected to two-dhnmskmal thin-laye-rhc (boltom Wl and right). Autoradiography identitlesthe unmodifiednudeotlda(bottomleft)and the adduct (bottomright) Thio person was a 57-yearold male painter"hghad a 4o.padcyear smokinghistory. The h lof adduct to unmodified nucleotides was one In 7 . 2 ~ 1 0(dCp lndlcates deonycylldine 3'-phosphate; dGR deoxyguanosine 3'-phosphate:dTp,deoxythymidine3'-phosphate:and d& deoxyadenoslne3'-phosphate). etry has generally focused on hemoglobin rather than DNA adducts. Levels are higher in smokers than nonsmokers and decrease with cessation.q Among the most important life-style risk factors in carcinogenesis is tobacco smoke exposure. Due to widespread aggressive advertising and the addictive nature of cigarettes, tobacco smoke has become the major cause of cancer. Of growing concern is the documentation that passive exposure to tobacco smoke also increases the risk of lung cancer." Because of the causal relationship to lung and other cancers, tobacco smoke exposure is useful for developing expo- sure assessments. Putative adducts have been correlated with consumption by the say in phosphorus human lunf 3a2npdosptllaacbeenltianiBbaus-t not in peripheral lymphocyteswor oral mucosa.u 'lbbacco-specific nitrosa- mines are potent carcinogens in labora- tory animals." Levels are higher in sec- ondary rather than mainstream smoke, raising the question of their role in pas- sive smoke exposure. Hemoglobin ad- ducts are increased in smokers more than nonsmokers, while higher levels are reported in snuff dippers." Inhala- 884 JAMA, August 7,1991 -Vol266, No.5 Molecular Epidemiology-Shields8 Harris Tab I- E vira -my Ck - *) - .Vim I P-4 lw tia Po an & an su ra, ag se. re: vi ce' an ha lul re: inc tio to: alc is SN nil CB1 lat de thc pa: I to1 tic chc str bil; sm tY! ox1 ag' int tos a CI INt I hos CBn I00 viet JAN T i l e 2.-Interactive Effects of Carcinoaen Exposure and host Factors. ~ Chemicalchemical Viral-chemical Physicalchemical Chemical-host Physical-host Viral-host ~~~~ Tobaccosmoke and alcoholic beverages HP;' and tobacco smoke EBV and nitrosamine HBV and aflatoxin8. Asbestosand tobacco smoke Radonand tobacco smoke PAHand CYP2D6 Tobaccos h e and CYP2D6 Chemotherapy Asbestos and CYP2D6 Sunlight and xeroderma pigmentosum Radiationand RBdeficlentgenotype EBV and X-linked immunodeficiency syndrome Otolaryngeal,esophageal &Nhl Nasopharyngeal Liver Lung Lung Lung Lung Leukemia Lung Skin Osteosarcoma Lymphoma 'HPV indicates human papillomavirus: EBV, Epstein-Barr virus: HBV, hepatitis B virus: CYP2D6, c y t o c h m P-450CYP2D6metabolicphenotypedeterminedbydebrisoquinsulfate administrationand measurementof urinary malatoliies:PAH. polycyJic aromatichydrocarton:and RB. retinoblastoma. tion of tobacco smoke also results in ex- posure to ethylene oxide, aromatic amines, and agents that cause oxyradical damage. Adducts from these agents are also correlated with active consumption and passive exposure.us68." Interactive effects of chemicaland viral agents, chemical and physical agents, and host factors have been observed (Table Z).'" One example is the relationship of aflatoxin B, and hepatitis viruses to the develo::..nent of hepatocellular carcinoma. Hepatitis B virus and, more recently, hepatitis C virus have been linked to primary hepatocellular carcinoma. However, geographic relocation of hepatitis B v i m carriers indicates that factors other than infection axe important. These include aflatoxin B, exposure and consumption of alcoholic beverages. Another example is the combined exposure of tobacco smokeand radon in uranium miners significantly acts to increase risk of lung cancer. Cigarette smoke in unventilated rooms increases levels of radon decay products, theoretically because the products attach to tobacco smoke particulates. Occupational asbestos exposure and tobacco consumption also act synergistically to increase the incidence of bronchogenic carcinoma. In vitro DNA strand breaks increase with the combined exposure." Moreover, tobacco smokeand asbestos, through their ability to induce inflammation and lipid peroxidation, cause oxidative DNA dam- Another consideration for the interactive effect is the ability of asbestos to absorb benzo[a]pyrehe and act as a carrier for the cwcinogen.".7s INHERITED PREDISPOSITIONS Inheritance plays an important role in host susceptibilities to cancer. Family cancer syndromes can lead to up to a 1000-fold increased risk of cancer (reviewed in a study be Li''). Patients with the Li-Fraumeni syndrome, at risk for tumors at,multiple sites, inherit mutant p53 genes." Retinoblastoma, another inheritable tumor, is associated with loss of a suppressor gene on chromosome 13q. These individuals are also at risk of other tumors (soft tissue sarco- mas and osteosarcomas) within radiation therapy p o r k n Host susceptibility can be manifested through inheritable interindividual differences in metabo- lism, DNA repair, genomic instability, or altered proto-oncogene or suppres- sor-gene expression. The study of the cytochrome P-450 CYP2D6 enzyme is among the best ex- amples of inheritable interindividual differences in metabolism. CYF'2D6 is responsible for metabolism of several medications, including tricyclic antide- pressants, p-blockers, and debrisoquin. Poor metabolizers are at risk of adverse drug reactions. However, in an English cohort, a fourfold-higher risk of lung cancer was associated with the exten- sive metabolic phenotype.= This associ- ation has been confirmed in a second study of lung cancer, with an odds ratio (OR) of 6.1." The extensive metabolic phenotype has also been shown to have an interactive effect with occupational exposures to asbestos and PAHs." A method for genotyping individuals from a small blood sample using the polymerase chain reaction and amplification of mutants and wild-type alleles targeting specific base mismatches is now available.'" This method, in conjunction with a restriction fragment length polymor- phism analysis, can correctly characterize over 95% of metabolic phenotypes. Thus far, the only carcinogenic substrate identified for CYP2D6 is methyl nitrosaminopyridylbutanone but not other tobacco-specific nitrosaminesthat have been tested (F. Gonzalez, PhD, oral communication,January 1991).Unless there are other unidentified carcinogenic substrates, an alternative ex- planation is that the gene is in linkage disequilibrium with another gene re- lated to cancer risk. The cytochrome P-450 CYPlAl en- zyme metabolically activates PAHs. A restriction fragment length polymorphism using the polymerase chain reac- tion has been described that is reported to correlate with lung cancer risk (OR of 3) inJapanesecancer patients compared with historical controls." Other ethnic populations will also need to be investigated. For example, a striking differ- ence in allelicfrequencybetween whites and blacks in the United States was found." An investigation involving the m e tabolism of isoniazid, commonly used in the treatment of tuberculosis, led to the identification of acetylation phenotypes for the enzyme acetyltransferase. This enzyme, besides metabolizing caffeine, also metabolizes certain carcinogenic aromatic amines. The slow acetylation phenotype has been associated with increased risk of bladder cancer.p*" In contrast, rapid acetylation is correlated with tba risk of coloncancer." A correlation of aromatic amine adducts and ace- tylator phenotype in tobacco smokers highlights the relationship of acetylation and DNA damage.' Detoxification of carcinogenic reac- tive intermediates also can vary among individuals. Glutathione transferase ac- tivity is inherited in an autosomal dominant fashion. It isresponsible for detoxifying metabolites of aromaticand other compounds. Specific m p class isoen- zymes that have high activity toward trans-stilbene oxide have been correlated with lung cancer risk.mAn OR of 3.25 for the absence of such activity in adenocarcinoma patients was reported. In a separate study, the absence of this transferase activity wascorrelated with the risk of stomach and colon cancer." Genetic differences in protwnco- gene sequences can also be predictive of risk. Using Southern blot analysis and restriction fragment length polymor- phism, it has been possible to demonstrate that rare alleles of Ha-ras are detectable primarily in cancer patients but not in controls.' Thes alleles are the result of reiterated DNX sequences arranged in tandem array. These re- gions may provide an unstable site where recombination or amplification occurs. An association of rare alleles with lung cancer risk was, tested in a case-control study, and the association was found to be strongest for black perA t c t d of 16studies have tested the association with lung and other cancers, including bladder, breast, leukemia, brain, and colorectal, but have found conflicting results." This may be JAMA, August 7,1991-Val 266, NO 5 Molecular Epidemiology-Shields & Harris 685 due to differences in characterization of rarealleles or study designs. Analysis of the L-myc proto-oncogene also demonstrates a polymorphic restriction enzyme site. The site-present allele has been detected more frequently in males with soft tissue sarcoma"and alsoin patients with gastric and lung cancer.""The association also sug gests a worse prognosis at the time of diagnosis for patients with renal cell cancer." A recent analysis of persons with lung cancer failed to reveal the predictive value of this polymorphism." Interestingly, racial differences were again found, whereby the homozygote site-present genotypes were tietected more frequently in American blacks. PERSPECTIVE The assessment of cancer risk in an individual and on a molecular level is currently not possible because no single genetic factor is sufficiently predictive. I t islikelythat further elucidation ofthe multiple stages of cancer and their in- 1 ,I ractions will need to be identified. Carcinogen exposure and exposure in target tissues, mutational spectra of proto-oncogenes and tumor suppressor genes, and DNA repair all play a role. Given the pace of the last decade, however, it is feasible that the next 10years will allow molecular epidemiologists to develop a profile for an individual that includes assessment of multiple DNA damage (adducts and mutations), geno- typing of metabolism, DNA repair, and susceptibilities based on germ line and altered proto-oncogenes and tumor suppressor genes. The development of these techniques, however, is incumbent on their trials in well-designed case-control and prospective studies. The ability to perform polymerase chain reactions in formalin-fixed, p a r a n embedded tissue sections and banked serum samples will allow molecular epidemiologists to study archival material. The current status of quantitative risk assessment, which concerns risks of populations rather than individuals, is hampered by inadequate human data. Assumptions about risk and safety factors required because of regulatory perspectives arenot based on sufficient human scientific data. The reliance on animal data leaves open the questions of interspecies and high- to lowdose extrapolations. However, the status of quantitative risk assessment also should improveasthe genetics of cancer and comparisons between species are better elucidated. Assessments of cancer risk in individuals have potential impact on society and its ethical considerations, eg, worker selection and potential discrimination. Separately, the impact of cancer risk assessmentsin individuals can best benefit those persons identified as being at risk and the focus of cancer prevention strategies, including education and chemopreventive therapies. We wish to thank Ainsley Weston, PhD, for his critical review of the manuscript and Bob Julia for his skillful editorial assistance. Reference8 1. Harris CC. Chemical and physical carcinogene- tumor-derived p53 mutants exhibit distinct biologi- sis: advances and perspectives. Cunmr Rea In cal activities. Sciewe. 1W,W:113-116. press. 17. Horsfall MJ,Goidon AJ, Burns PA, Zielenska 2. Shields PG, Harris CC. Environmental causes M, van der Vliet GM, Glickman BW. Mutational ofcancer. Med Clin North Am. 1990;74:263-277. specificity of alkylating agents and the influence of 3. Yuspa SH, P b i e r MC. Chemical carcinogene- DNA repair. Environ Mol Mufagen. 1990;15107- sis: from animal models to molecular models in one 122. decade. AdvCancer Rea 1988;50:2570. 18. Barbacid M. TMI Genes. Ann Rev Biochetn. 4. Weinstein IB. Growth factors, oncogenes, and 1987;56:7i9-627. multistage carcinogenesis. J Cell Biochem. 19. Brown K, Buchmann A, Balmain A. Carcino- 1987;33:213-224. gen-induced mutations in the mouse e-Hams gene 5. Rowley JD. Molecular cytogenetics: Rosettsl provide evidence of multiple pathways for tumor Stone for understanding cancer-twenty-ninth progression. PIVC Nut1 Acud Sei U S A. GHA Clowes Memorial Award lecture. Cancer 1990;87:639542. Rea 1990;50:316-3625. 20. Newcomb EW, Diamond LE, Sloan SR, Corn 6. Preston-Martin S, Pike MC, Ross RK, Jones minas M, Guerreiro I, Pellicer A. Radiation and PA, Henderson BE. Increased cell division as a chemical activation of rad oncogenes in different cause of human cancer. Cancer Res. 1990;60:7415- &use strains. Environ Health Perspect. 7421. 198981:3337. 7. Bouck N. Tumor angiogenesis: the role of onco- 21. Hsu IC, Metcalf RA, Sun T, Welsh J, Wang genes and tumor suppressor genes. Cancer Cells. NJ, Harris CC. p63 Gene mutational hotspot in 1990;2179-186. human hepatocellular carcinomas from Qidong, 8. Liotta LA. Tumor invasion and metastases- China. Nafare. 1991;350:427-428. role of the extracellular matrix: Rhoads Memorial 22. Bressac B, Kew M, Wands J, Ozturk M. Selec- Award lecture. CancerRea 1986;461-7. tive G to T mutations of p63 gene in hepatocellular 9. Bishop JM. Molecular themes in oncogenesis. carcinoma from southern Africa. Natttre. Cell. 1991;64:235248. 1991;350:429-431. 10. Hansen MF, Koufos A, Gallie BL. et al. Osteo- 23. Gonzalez FJ, Crespi CL, Gelboin HV. DNA- sarcoma and6tinoblastoma: a shared chromosom- expressed human cytochrome P450s:a new age of al mechanism revealing recessive predisposition: molecular toxicology and human risk assessment. h c Natl AcadSci USA. 1985$325216-6220. Mztfat Rea 1991;247:113-127. 11. Cavenee WK, Murphree AL, Shull MM, et al.. 24. Loeb LA. Endogenouscareinogenesis:molecuPrediction of familial predisposition to retinoblasy .C lar oncology into the twenty-first century-presi- toma. NEngl J Med. 1986314:1201-1207. dential address. CancerRea 1989:495489-6496, 12. Friend SH, Bernards R, Rogelj S, et al. A 26. Floyd RA. The role of &hydroxyguanine in human DNA eegment with properties of the gene carcinogenesis. Carcinogeneeia 1990,11:1447- that predisposes to retinoblastoma and osteosar- 1450. coma. Natzrre. 1986;323:643-646. 26. Ames BN. Mutagenesis and carcinogenesis: 13. Liotta LA, Steeg PS, StetlerStevenson WG. endogenous and exogenous factors. Ettviron Mol Cancer metastasis and angiogenesis: an imbalance Matageit. 1989;14(suppl16):66-77. of positiveband negative regulation. Cell. 27. Genion SL. Regeneration of Walkylguanine- 1991:64:327-336. DNA alkyltransfeme in human lymphocytes &er 14. Hollstein M,Sidransky D, Vogelstein B, Har- nitmourea euposure. Cancer Rea 1988,486368- ris CC. p63 Mutations in human cancers. Science. 6373. 1991;253:49-53. 28. Scudiero DA, Meyer SA, Clatterbuck BE, 16. Malkin D, Li FP, Strong LC, et al. Germ line Mattern MR, %iolkowskiCH, Day RS. Scnsitivity p63 mutations in a familial syndrome of breast can- of human cell strains having different abilities to cer, sarcomas, and other neoplasms. Science. repair Wmethylguanine in DNA to inactivation by 1990;250:1233-1238. alkylating agents including chloroethylnitrosour- 16. Halevy 0,Michalovitz D. Oren M. Different . eas. Cancer Rea. 1984:44:2467-2474. 29. Myrnes B, Ciercksky KE, Krokan H. Interin- dividual variation in the actlvity of O'-methyl gunnine-DNA methyltransferase and uracil-DNA glycosylase in human organs. Carcinogencsia 1983;415651568. 30. Grafstrom RC, Pegg AE, Trump BF. Harris CC. 06-alkylg~anincDNAalkyltransferase activi- ty in n o d human tissues and cells. Cancer &a 1984:44:2855-2857. 31. Harris CC. Interindividual variation among humans in carcinogen metabolism, DNA adduct formation and DNA repair. Carcinogenesia 1989;101563-1566. 32. Ayesh R, Idle J R , Ritchie JC. &others MJ, Hetzel MR. Metabolic oxidation phenotype8 u1 markers for susceptibility to lung cancer. Nature. 1984:312:169-170. 33. C a p o m NE, Tucker MA, Hoover R, et al. Lung cancer and the debrisoquinemetabolic phenatype. J N a f lCUncerhCsf. 1990;85:1264-1272. 3d. Petruzcelli S. Camus AM, C m z z i L, et al. Long-lasting effecta of tobarn smoking on pulmonary drug-metabolizing enzymes: a e a s e s o n h l study on lung cancer patients. Cancer Rea 1988;48:4696-4700. .%. McLemore TL, Adelberg S,Liu MC, et al. Expression of CYPlAl gene in patienb with lung cancer: evidence for cigarette smoke-inducedgene expression in normal lung tissue and for dtered gene regchtion in primary pulmonary earcinomcls. JNafl C a ~ ~ l m199t0.;82:1333-1339. 36. Kouri RE, McKinney CE, Slomiany Rl,Snodgnuur DR,Wray NP, McLemore TL. Fbitive cor- relation between high aryl hydrocarbon hydroxylase activityand primary lungcancerle analyzed in cryopreserved lymphocytes. Cancer Rea 1982;42:50305037. 37. Bartsch H. Hietanen E, Petnmelli S, et al. Possible prognostic value of pulmonary AH-locuslinked enzymes in patients with tobacco-related lung cancer. Int JCaiiccr. 1990;461&188. 38. Fearon ER. Vogelstein B. A genetic model for colorectal tumorigenesis. Cell. 1990;61:769-767. 39. Shields PC. Weston A. Sugimura H.et PI. Mo- lecular epidemiology:dosimetry, susceptibility and cancer risk. In: Vanderlaan M, Stanker LH. Watkin8 BE, Roberta DW. eds. Inrnticnwssays for ,.M c e Cheniical Analysis: Monitoring 7bric Cheniicals iir Hionana. W ,and !he Enriron-nten!. Washington. DC: America1 Chemical Soci- 886 JAMA, August 7,1991 -VOl266, NO.5 MolecularEpidemiology-Shields& Harris et: 40 D, ar ce 22 41 A, ea m fu 41 br di fn J. 43 h as eo 44 D tu la 45 m UJ in in 81: in 41 4t Bd in 1f 4i N m et u 15 48 2 8U 15 49 U tr 8b 19 w u tr; el 19 61 lei CI 52 ex inl ge ti0 C. 63 Hi Ch mc 661 64. Au ma 19 65. OS. M eer 56. Ku Yld out 97. 67. JAl ety; 1991:186-206. 40. Perera FP, Hemminki K, Young TL. Brenner D, Kelly G, Santella RM. Detection of polycyclic aromatichydrocarbowDNA adducts in white blood cellsof foundry workers. CancerRea 1988,482286- 2291. 41. Phillips DH, Hemminki K, Alhonen A, Hewer A, GroverPL. Monitoring oeeupationalexposure to carcinogens: detection by zP-postlabelling of aromatic DNA adducts in white blood cells from iron foundry workers. Mutat Rea 1988;204:531-541. 42. Van Schooten FJ, Van Leeuwen FE, Hillebrand MJ, et al. Determination of benzo[alpyrene diol epoxide-DNA adducts in white blood cell DNA from coke-oven workers: the impact of smoking. JKatl Cancerlnst. 1990;82:927-933. 43. Bryant MS, Wneis P, Skipper PL, Tannenbaum SR. Hemoglobin adducts of aromatic amines: assoeiations with smoking status and type of tobacco. Proc Natl Acad Sci U S A .1988;859788-9791. 44. HemminkiK, GnybowskaE, Chorazy M, etal. DNA adducts in humans environmentally exposed to aromatic compounds in an industrial area of Po- land. Cam'nogenesia 199&11:1229-1231. 45. Haugen A, Beeher G, Benestad C, et al. Determinationof polycyclicaromatichydrocarbons in the urine, be@a)pyrene diol epoxide-DNA adducts in lymphocyte DNA, and antibodies to the adducts in sera from coke oven workers exposed to measured amounts of polycyclic aromatic hydrocarbons in the workatmosphere. Cancer Rea 1986;46417& 4183. 46. Perera F, Mayer J, Jaretzki A. et al. Comparison of DNA adducts and sister chromatid exchange in lung cancer cases and controls. Cancer Rea 1989;49:4446-4451. 47. Liou SH, Jacobson-Kram D, Poirier MC, Nguyen D, Strickland PT, lbckman MS. Biological monitoring of fire fighters: sister chromatid ex- change and polycyclicaromatic hydrocarbowDNA adducts in peripheral blood cells. Cancer Res 1%49:4929-4935. 48. Rothman N, Poirier MC, Baser ME, et al. Formation of polycyclic aromatic hydrocarbon-DNA addiicts in peripheral white blood cells during consumption of charcoal-broiled beef. Carcinogenesis. 1990;11:1241-1243. 49. Zatonski W, Ohshima H, Pnewozniak K, et al. Urinary excretion of N-nitmsamino acids and nitrate by inhabitants of high- and low-risk areas for stomach cancer in Poland. Int J Canem 1989;44:823-827. 50. Kamiyama S, Ohshima H, Shimada A, e t al. Urinary excretion of N-nitmsamino acids and ni- trate by inhabitants in high- and low-risk areas for stomach cancer in northern Japan. IARC Sci Aibl. 1987;84:497-502, 61. Lu SH, Montesano R, Zhang MS, et 81. Prevalence of N-nitrosamines to esophageal cancer in China. JCell PhySiOl Suppl. 1986,4:51-58. 52. Lu SH, Ohshima H, AI HM, et al. Urinary excretion of N-nitrosamino acids and nitrate by inhabitants of high- and low-risk areas for esophageal cancer in Northern China: endogenous formation of nitrosoproline and its inhibition by vitamin C. CancerRea 1986,46:1485-1491. 63. Shields PG, Povey AC, Wilson VL, Weston A, Harris CC. Combined high-performance liquid ChrornatographyPlP-postlabeling assay of N7methyldeoxyguanosine. Cancer Res 1990;506Mu)- m. 54. Wilson VL, Weston A, Manchester DK, et al. Alkyl and aryl carcinogen adducts detected in human peripheral lung. ' Carcinogenenis. 1989;102149-2153. 55. Umbenhauer D, WildkP. Montesano R, et al. O'-methyIdeoxyguanosine in oesophageal DNA among individuals at high risk of oesophageal cancer. fnt JCancet. 1985;36:661-666. 56. Huh NH. Satoh MS, Shiga J, Rsjewski MF, Kuroki T. Immunoanalytical detection of O'gthylthymine in liver DNA in individuals with or without qi malignant tumors. Cancer Re& lca949:93- 97. 57. Grooprnan JD,Cain LG, Kenaler TW. Afln- ' - _ . .-I -. toxin exposure in human populations: measurements and relationship to cancer. Crit Rev Ibxicol. 1988;19113-145. 58. Wild CP, Jiang YZ. Sabbioni G. Chapot B. Montesano R. Evaluation of methods for quantitation of atlatoxin-albumin adducts and their application to human exposure assessment. Cancer'Rea. 1990;50:245.251. 59. Beland FA, Poirier MC. DNA adducts and carcinogenesis. In: Sirica AE, ed. The hthobiologyof Neoplasia. New York, N Y Plenum Publishing COT: 198957-80. 60. Maclure M, Bryant MS, Skipper PL, Tannenbaum SR. Decline of the hemoglobin adduct of 4aminobiphenyl during withdrawal from smoking. CancerRea 1990;50:181-184. 61..Fielding J E , Phenow KJ. Health effects of involuntary smoking. N Engl J Med. 1988;31914521460. 62. Phillips DH, Hewer A, Martin CN, Garner RC, King MM. Correlation of DNA adduct levels in human lung with cigarette smoking. Nature. 1988;336:790-792. 63. Everson RB, Randerath E, Santella RM, Cefalo RC, Avitts TA, Randerath K. Detection of smoking-related covalent DNA adducts in human placenta. Science. 1986;231:54-57. 64. Phillips DH, Schoket B, Hewer A, Bailey E, Kostic S, Wncze I. Influence of cigarette smoking on the levels of DNA adducts in human bronchial epithelium and white blood cells. Int J Cancer. 1990;46569-575. 0965. Chacko M. Gupta RC. Evaluation of DNA dam- age in the mucom of tobacco usern and nonusers by P-adduct assay. Carcinogenesis. 1988;92309-2313. 66. Hecht SS,Hoffmann D. Tobacco-specific nitrosamines, an important group of carcinogens in to- bacco and tobacco smoke. Carcinqenesis. 1988;9875-884. 67. Carmella SG,Kagan SS, Kagan M, e t al. Mass spectrometric analysis of tobacco-specific nitrosamine hemoglobin adducts in snuff dippers, smokers, and nonsmokers. Cancer Rea 1990;50:5438 5445. 68. Maclure M, Katz RB. Bryant MS, Skipper PL, Tannenbaum SR. Elevated blood levels of carcinogens in passive smokers. Am J Aibfic Health. 1989;79:1391-1384. 69. Leanderson P, Tagesson C. Cigarette smokeinduced DNAdamage: role of hydroquinone and catechol in the formation of the oxidative DNA- adduct, &hydroxydeoxyguanosine. Chcm Biol Internet. 1990;7671$1. 70. Haugen A, Harris CC. Interactive effectsbetween viruses and chemical carcinogens. In: Cooper CS, Gmver PL, eds. Carcinogenesis and Mutagenesis, Handbook of Experimental Pharmacology. New York, NY: Springer-Verlag, 1990249-268. 71. Jackson JH. %hraufstatter IU, Hyslop PA, et al. Role of oxidants in DNA damage: hydroxyl radical mediates the synergistic DNA damagingeffects of asbestos and cigarette smoke. J Clin Invest. 1987;80:1090-1095. 72. Mossman BT, Marsh JP. Evidence supporting a role for active oxygen species in asbestos-induced toxicity and lung disease. Enuiron Health k r sped. 1989,81:91-%. 73. Kamp DW, Dunne M. Weitzman SA, Dunn MM. The interaction of asbestos and neutrophils injures cultured human pulmonary epithelial cells: role of hydrogen peroxide. J Lab Clin Med. 1989;114604-612. 74. Cerde P, Scholander P. Adsorption of benzo(a) pyrene on to asbestos and manmade mineral fibres in an aqueous solution and in a biological model solution. BrJ Ind Med. 1988:45682-688. 75. Mossman BT, Enstman A, Landenman JM. Bresnick E. Effects of crocidolite aqd chrysotile asbestos on cellular uptake and metabolism of benzo(a)pyrene in hamster tracheal epithelial cells. Eirvimt Health Perspect. 1983;61:331-335. 76. Li FP. Familialcancer syndromes and clusters. C w r h b l Cancer 1990:14:73-114. 77. Weichselbaum RR, Beekett MA, Diamond AA. An important stepin radiition Carcinogenesis may be inactivation of cellular genes. Int J Radiat h 0 1 BiOl Phys. 1989;16Z77-282. 78. Caporaso N, Hayes RB, Dosemeci M. et al. Lung cancer risk, occupational exposure, and the debrisoquine metabolic phenotype. Cancer Res lW493675-3679. 79. Heim M, Meyer UA. Genotyping of paor metab o l i r s of debrisoquine by allele-specificPCR amplification.Lancet. 1990;336:529532. 80. Kawajiri K, Nakachi K, I& K, Yoahii A, Shinoda N, Watanabe J. Identification of g e n e t i d y high-risk i n d i v i d d to lung cancer by DNA polymorphisms of the cytochrome P450IA1 gene. FEBSLCtt. 1990;263:131-133, 81. Shields PG,Sugimura H, Caporaso NE, et al. Restriction fragment length polymorphism analysis of CYPlAl and relationship with lung cancer risk. .Biomunilor Suuceptibility Markera Hum Cancer In press. 82. Cartwright RA, Philip PA, Rogera HJ,Cla- shan RW. Genetically determined debrieoquine oxidation capacity in bladder cancer. Carcinogenenis 1984;51191-1192. 83. Mommsen S,Barfod NM, Angard J. N-acetyl- transferase phenotypes in the urinary bladder carcinogenesis of a low-risk population. Caninogene- sia 1985.6199201. 81. Hein DW. Acetylator genotype and arylamineinduced carcinogenesis. Biochim Biophys Acta. 1988;948:37-66. 85. Vineis P, Caporaso N, Tannenbaum SR, et al. Acetylation phenotype, rnreinogen-hemoglobin adducts. and cigarette smoking. Cancer Res 1990;50:30023006. 86. Seide@-& Pem RW, Markowitz MM, b u s h G,Miller D6,Beattie EJ.Isoenzymda)ofglutathione trawlerase (class mp) as a marker for the susceptibility to lung cancer: a follow-up study. carcinage7lcais.1990;11:33-36. 87. Strange RC, Matharoo B, Psulder GC. e t al. The human glutathione StmnsferPsea: a cnseeontrol study of the incidence of the GSTl 0 phenotype in patients with a d e n d n o m a . Carcinogemusis 1991;1225-28. 88. Krontiris TG,DiMartino NA, Colb M, Parkinson DR. Unique allelic restriction lrylments of the human Ha-ma locusin leukocyteandtumour DNAs of cancer patients. Natictc. 1985;313:869474. 89. Sugimura H. Caporno NE, Hoowr RN. et al. Association of rare alleles of the Harvey m protooncogene locus with lung cancer. Cancer Rca 1990;M):1857-1862. 90. Vieis P, Caporaso N. The analyeis of restriction fragment length polymorphism in human cancer a review from an epidemiological perspective. h i t JCancer. 1991;472630. 91. Kat0 M, ToguchidaJ, Honda K, et d.Elevated fmquency of a specific allele of the Lmye gene in male patients with bone and s&-tisaue sarcomas. Int J Cancer.1990;45:47-49. 92. Ishizaki K, Kat0 M. Ikenaga M, Hondn K, Ozawa K, ToguchidaJ. Correlation of L-myc genotypes to metPstnais of gastric cancer and breast cancer. J Natl Cancer Inat. 1990;82238-239. 93. K a w h i m a K, Shikama H, Imoto K. et al. Close correlation between restriction fragment length polymorphism ofthe L-mycgene and me& tasis of human lung cancer to the lymph nodes and other organs. Roe Natl Acad Sci U S A. 1988;85:2353-2356. 94. Kakehi Y, Yoshida 0.Restriction fragment length polymorphism of the L m y c gene and susceptibility to metsetasis in renal cancer patients. Int J Cancer 1989;43:391-394. 95. Tamai S,Sugimura H, C a p o m NE, e t al. Restriction fragment length polymorphism analysisofthe L-nrycgenelocusinac~controlstudyof lungcancer. fnt JCancct. 1990,46:411-416. 96. Pegg AE. Alkylation and subsequent repair of DNA atter exposure to dimethylnitrosamine and related carcinogens. Rei? Biochem 7bxicol. 1983:683-133. II JAMA, AUgusl7,1991 -VOl266, NO, 5 Molecular Epidemiology-Shields& Harris 687