Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
1 THE SOMATIC-MUTAT ION THEORY OF CANCER CHROMOSOMAL ABERRATION . In 1914, Boveri26 proposed a link between mutagenesis and carcinogenesis. But the molecular biology of cancer is still elusive, and, although mutation is thought to be intimately connected with cancer, its exact role in the disease cannot be definitively described. Recent reviews have highlighted the uncertainty.32~34,5l,67,105,114 Boveri's hypothesis stemmed from his observations on chromosomal aberrations in tumor cells. Chromosomal rearrangements continue to be found in some cancers. but no general pattern has emerged. Klein61 has discussed the apparent tissue-specific involvement of different chromosomes in tumor-associated nonrandom karyotype changes. For example, human B-cell lymphoma is associated with the 14q+ band, Burkitt's lymphoma with a reciprocal 8:14 translocation, and myeloid differentiation with changes in chromosomes 22 and 9. Wiener et al. 119 have described a trisomy-~5 in marine B-cell lymphoma that suggests a gene duplication on chromosome 15, and the chromosomal instability associated with Bloom's syndrome is thought to predispose those afflicted to leukemia and other cancers. 110 Other examples of chromosomal aberrations assay elated with neoplasia include a chromosome 3:S translocation with hereditary renal cell carcinoma, 40 a chromosome 13 de le t i on wi th re t inob las tome, 64 and a de let ion in chromes ome 11 with Wilms' tumor. In a report from the National Wilms' Tumor Study, Breslow and Beckwith29 suggested that the pro- portion of Wilms' tumors due to inherited mutation may be sub- stantially smaller than previously estimated. None of the 20 familial cases had the features associated with genetic tumors . Al though cytogenet ic analyses were not pe rformed in this large study, that many familial cases were unilateral and did not depend on the sub Sects' ages suggested that mechanisms other than mutation may be responsible for most cases of Wilms' tumor.
These findings led Klein,62 Cairns, 34 and Nevers and Saedler86 to suggest that chromosoma1 rearrangements may be a gross manifestation of a critical gene-dosage effect. By chro- mosomal rearrangement, a set of endogenous cancer genes may come under the influence of more active transcriptional con- trol . Recent ly , insertion of viral promoters has been shown to cause transformation in cell cultures by enhanced transcrip- tion of viral or endogenous oncogenes.3b,91 EndogenOgs5 onc5O- genes have been mapped on specific human chromosomes, ~ and there is evidence that reciprocal translocations may bring these genes under different promotional control.62 Thus, chromosomal rearrangements (observed by using cytogenetic techniques) and small-ecale transposition events (that can be detected only genetically) may both produce the same effect. MOST ULTIMATE CARCINOGENS ARE MUTAGENS Many chemical carcinogens have been tested for mutagenicity in the last decade. The development of the Salmonella/micro- some reverse~mutation assay byAmes and colleagues71~73 sig- naled the beginning of large-~cale efforts. In combination with results from other in vitro mutagenicity tests, the Salmo- nella/microsome assay detects almost all carcinogens with muta- genic properties. The Committee expressed great confidence in a small test battery for detecting chemicals with mutagenic properties.84 The Committee also agreed with other estimates (e.g., Brusick3~) that approximately 90: of currently known animal carcinogens are bacterial mutagens. However, the causal relationship between mutation and cancer is not as direct as these statistics imply, and the Committee was reluctant to assert mechanistic links in its initial report.84 One reason is that some confirmed animal carcinogens have no mutagenic activity or are mutagenic in only a minority of the systems in which they are tested. For example, benzene and arsenic compounds are not mutagenic in the Salmonella/microsome assay, but do display activity in some other short-term assays. 2 Phenobarbital, succinic anhy- dride, and dieldrin are carcinogens that are negative in most mutagenicity assays.54 The mechanisms by which these came pounds and some others, such as the steroid hormones (e.g., 175-estradiol), cause cancer are not known. Rubinl05 has stressed the difficulty in ascribing mutational mechanisms to cancers caused by a variety of nongenetic agents. Cairns34 has proposed that transposons may constitute major mechanism in the development of cancer and has urged examination of agents that increase the frequency of trans- positions. Many of these agents would not be detected in conventional in vitro mutagenicity assays, especially if eukaryotic and bacterial transposons behaved differently. 2
Fahmy and Fahmy45,46 have begun studies on the chemical induction of transposition with the white (w+) eye-color gene complex of Drosophila. Stable and nonstable genetic features of the ~ omplex are used to assess germinal mutation and somatic gene expression that results from transposition of a duplicate regulatory locus in w+. Preliminary evidence derived from the few carcinogens tested suggests that mutagenic carcinogens have an influence on supposed transpositions, but this phenol enon is unrelated to a carcinogen's mutagenic potency. Further work is required to confirm and extend these findings. Eeken and Sobels44 have studied chemically induced transposition at the singed bristle (en) locus of Drosophila. Powerful muta- gens, such as ethyl nitrosourea and methyl methanesulfonate, did not greatly influence reversion of insertion mutation, but DNA-repair defects (measured in the repair-deficient strain me i-9 /me i-4 1) did inte rfere . Although not all mutagenicity test data are in concordance with carcinogenicity data, most investigators believe that a relationship exists between mutagenesis and cancer. In addi- tion to the high correlation between short-term mutagenicity and animal chemical-carcinogenicity data, there is experimental support for the multistage model of carcinogenesis. Recent reviews53~65~80 have agreed that somatic mutation is likely to be the leading cause of initiation of the carcinogenic process. Berenblum and Shubik23 provided early evidence of a mutational mechanism in the initiation of cancer. Recently gathered evidence indicates that, in later stages of progres- sion to malignancy (i.e., promotion), DNA may be unaf- fected.22~69 However, as Cairns has pointed out,33 pro- motional activity that produces cell proliferation can be accommodated in a multistage cancer model in which the first stage derives from somatic mutation. Other new data also reflect the mutational etiology of cancer. Weinberg et al.ll6 and Barbacid et al.l°l have provided evidence that a single point mutation in the coding region of a human bladder proto-oncogene may be sufficient for its conversion to an on c' gene that elicits a carcinoma. Although this finding awaits additional documentation, 7~81 it offers some support for mutation as an early part of the care inogenic process . ULTIMATE CARCINOGENS THAT ARE NOT MUTAGENS From a regulatory standpoint, it is important to be aware of a set of potent human or animal carcinogens with no observed or only inconclus ive mutagenic activity. Human exposure to many of these chemicals may be quite high. This phenomenon has created concern over the protection afforded by mutagenicity tests in predicting carcinogenic potential, because many of 5 3
tines e substances are present in the environment at high concen- trations. Consequently, as Squirell2 emphasized, carcino- genicity testing in laboratory animals remains the primary basis for most regulatory decisions. Because no specific experimental factor accurately predicts human carcinogenicity, Squire proposed ranking potential carcinogens according to several factors, including genotoxicity tests. There are two principal reasons why some chemical carcin- ogens are not detected in mutagenicity tests. The first is obvious : several chemicals (e.g., steroid hormones) either cause cancer by nongenotoxic means or alter DNA indirectly; these chemicals do not appear to be mutagens under any con- ditions. The second reason is that mutagenicity tests are inef ficient because of inadequate permeability, insensitivity of genetic end points, metabolic inadequacy, or other factors. To counter weaknesses of individual tests, the phylogenetic range of test batteries is broad and the sources of metabolic activation are varied. Thus, the probability is low that, if a battery of short-term tests were used, a confirmed genotoxic carcinogen could be legitimately shown to be nonmutagenic. GENETICS OF CANCER Investigators have approached the genetic etiology of cancer in several ways. One approach is experimental. Another involves the study of the inheritance of cancer in human popu- lations and the ef fects of genetic disturbances in select populations. A third combines epidemiologic and experimental approaches in in vitro experiments with cells from human tumors that are known to be or suspected of being inherited geneti- cally. The outcome of such approaches will be a description of the inheritance of cancer and an understanding of the mechanism respons ible . Cancer occurs in a few genetic syndromes,63~65 and it is not unreasonable to assume a genetic basis for predisposition to the disease. However, a more obvious hereditary pattern of human carcinogenesis has not been found. Perhaps the lack of such a pattern is due to the multiple stages of the disease, each of which might require the action of separate genes. Perhaps cancer results from a polygenic trait in many cases. In lieu of strict simple Mendelian inheritance, the human popu- lation appears to be heterogeneous in its Susceptibility to neoplasia. "Susceptibility" may involve a genetically influ- enced lace-in-life change (i.e., like graying of hair) ; in sus- ceptible persons, this change occurs earlier in life than it does in less susceptible persons. Studies of human cell transformation in vitro reflect efforts to understand the mutational basis of carcinogenesis. Barrett and colleaguesi5~6 have shown that diethylstil- 4
bestro! (DES) induced neoplastic and morphologic transformation at a frequency similar to that of benzota~pyrene. However, DES failed to induce somatic mutation at two genetic loci, whereas benzota~pyrene was mutagenic. Thus, two independent mechanisms appeared to lead to the same effect. In a review of the liter- ature, Parodi and Brambilla 7 claimed that transformation occurs 100-1,000 times more frequently than Mutation at a defined concentration of chemical. Strausll asserted that the range of difference is one-tenth as great and pointed out the technical difficulties in comparing the two end points in divergent test systems. Using mammalian cells in culture, Iluberman57 studied mutagenesis at the ouabain focus and cell transformation and concluded that approximately 20 genes may be involved in one transformation event. Cell hybridomas produced by cell fusion provide another in vitro system for examining mutationally induced transfor- mation. Harris has discussed the heritability of malignancy on the basis of hybridoma data, 51 especially in the context of the phenomenon of chromosomal exclusion and tumorigenicity. He concluded that Mendelian segregation patterns of hybrid cells do not favor a truly dominant or recessive mode. Harpists conclusion was supported by the more recent analysis of Sabin.1 6 The suppression of malignancy by cell fusion and subsequent chromosomal loss also is stable and transmissible in some cases. However, the data are often inconsistent. For example, malignancy can be suppressed in natural ly occurring human cancers by fus ion of cancer cel Is to normal human fibroblasts, differentiating epithelial keratino- cytes, and cancer cells of different somatic cell origin. But suppression has not yet been specifically linked either with a cancer cell of a particular origin or with a type of non- malignant cell to which the cancer cell is fused. Attempts to correlate suppression with the segregation of specific chromo- somes, even for tissue-specific neoplasia, have also been inconsistent. Although the retention of specific chromosomes and the suppression of the malignant phenotypes imply a dominant mechanism, in vitro data cannot yet properly distinguish a definitive segregation pattern. Analysis of human populations indicates strongly that sus- ceptibility to cancer varies among individuals and that genetic predisposition may be more important for the development of ~ ome tumors than others .63 Moolga~kar and Knudson argued that the development of hereditary tumors should not be over- looked on the grounds that they represent only a small fraction of all cancers; rather, they offer an opportunity to study the multistage process of care inogenesis. Mendelian analysis shows that almost all heredi~cary cancers segregate in an autosomal dominant fashion. Assuming the relationship to be strict, Moolgavkar and Knudson8O consented that. "even though every cell in a susceptible tissue carries 5
the cancer gene, only a few of the cells go on to become malignant, which indicates that inheritance of the gene is not sufficient (at the cellular level) and that at least one other event is necessary fot malignant transformation. " In a two- stage cancer model, the other event i,7not necessarily mutation. Among many others, Clayton has discussed non- genetic mechanisms for the second stage of a two-stage cancer model. Enhanced cellular proliferation has been identified as a possible participant in producing tumors. Examples of autosomal dominant cancers are bilateral retinoblastoma, neuroblastoma Wilms' tumor, and some carci- nomas of the colon. McKusick}6 listed approximately 30 human cancers that may be inherited in a dominant fashion. These account for approximately I: of all human cancers in the United States.63 The total number of human cancer genes remains a matter of speculation; Moolgavkar and Knudson 0 suggested that there may be 100-200, whereas recent hybridization data between viral or human oncogenes and human DNA imply a range of only 10-20.25 However, the sensitivity of nucleic acid hybridization techniques is probably not so great that all cancer genes can be detected definitively. The even rarer recessively inherited cancers are important in unders tending mechanisms . Two disorders, xeroderma pigmentosum (XP) and ataxia telangiectasia, occur in persons severely deficient in DNA repair; the deficiency probably increases their disposition to cancer. Persone with two other genetic syndromes, Fanconi's anemia and Bloom's syndrome, exhibit several abnormalities, including chromosomal gaps and DNA strand-breaking. By implication, these recessive disorders are linked mechanistically to increased cancer inc idence . Again, however, the relationship demands scrutiny. Cairns34 argued that XP patients have a greater frequency of cancer only of the skin, where W-light-induced damage can occur. He asserted that exposure to other mutagens does not increase the incidence of other cancers beyond that in the general popula- tion. This anomaly cannot be easily explained, and a possible conclusion is that the cancers are not caused by the type of DNA damage that XP patients are unable to repair. However, the limited number of XP patients in these studies reduces the certainty of epidemiologic data. Indeed, some evidence sum Bests that an excess of cancers exists in XP patients in sites other than the skin .66 Garner and Hertzog48 proposed the counterargument that XP patients shoult not, a priori, be expected to have a higher incidence of internal cancers. Organ-specific metabolism, DNA repair, and resistance to dimes and other lesions like those induced by W may influence tumor formation. Peto9 has recently investigated the idea that there are dif ferent individual susceptibilities to cancer. From epidemiologic analyses, he set up three broad classes. The 6
first includes cases in which the increase in risk is more than a factor of 1,000. Cancers included in this group are bilateral retinoblastomas and the squamous cell skin cancers of XP patients. Although some pi rsone are at tremendous risk, only a minute fraction of human cancers are in this category. The second category includes cancers for which a genetic predispos ition leads to risks approximately 10-100 times higher than that in the general populat ion. Pe to argued that cancers in this group may account for a substantial proportion of all cancers, but their categorization by frequency may be very difficult to prove. He argued further that the relative risk in relatives of patients compared with the general population may be an order of magnitude lee ~ than the relative risk in suspectible persons compared with nonsusceptible persons. He gave three reasons for this: not all cancer patients are susceptible, even fewer relatives are susceptible, and the general population contains susceptible and nonsusceptible persons. Peto surmised that a third category is composed of persons whos e gene t ic risk is increased by les ~ than a factor of 10 and who may remain undetected in the population. 7