cancer research - genetic errors, cell proliferation, and ......genetic errors, cell proliferation,...

(CANCER RESEARCH 51, 6493-6505, December 15, 1991]

Perspectivesin CancerResearch

Genetic Errors, Cell Proliferation, and Carcinogenesis1

Samuel M. Cohen2 and Leon B. Ellwein

Department of Pathology and Microbiology and the Eppley Institute for Cancer Research, University of Nebraska Medical Center, Omaha, Nebraska 68198

Chemicals have been recognized as etiological factors inhuman carcinogenesis since the 18th century observations ofthe nasal carcinogenicity of snuff by Hill and the scrotal car-cinogenicity of soot in chimney sweeps by Sir Percival Pott (1).Numerous other specific chemicals and chemical mixtures havesubsequently been identified as carcinogenic for humans andare considered to be responsible for a large proportion of humancancers (2, 3). A major contributor to the induction of humancancers is the chemical mixture resulting from smoking cigarettes, which leads to an increased incidence of cancers of thelung, upper respiratory tract, urinary bladder, and other tissues(4).

It was not until the 20th century, however, that animal modelswere developed for further investigation of chemical carcinogenesis, both for identifying carcinogenic chemicals and fordelineating the mechanisms by which cancer is caused (1). Inthis century considerable progress has been made in identifyingetiological factors of human carcinogenesis and also in understanding specific mechanisms by which these agents act.

Most chemicals identified as carcinogenic for humans havealso been established as carcinogenic in various animal models(1-3). Most of these chemicals have been shown to be genotoxicin various short-term assays, and most have been shown to bemetabolically activated to reactive intermediates that bind directly to DNA, presumably leading to the genetic mistakes thatresult in cancer.

It is not surprising that most of the chemicals initially identified as carcinogens in animal models and in humans werepotent chemicals that were also shown to be potent mutagensfollowing metabolic activation. Radiation was also found to becarcinogenic in humans and animals (5). The results of variousepidemiological investigations, animal bioassays, and DNAdamage studies strongly suggested that radiation and carcinogenic chemicals increase the risk of developing cancer, even atlow doses (2, 3, 5, 6). At high doses, there was a high incidenceof cancer resulting from exposure to these agents, and lowerincidences occurred as the dose was decreased. Although observations in the very low exposure dose ranges could not becharacterized experimentally, the fact that radiation and chemicals produced DNA damage even at extremely low doses andthat this damage was considered to lead ultimately to theproduction of cancer led to the paradigm that carcinogenicchemicals and radiation had no threshold with respect to theproduction of cancer (7). Thus, even a very small exposure wasconsidered to be associated with an increased risk of developingcancer.

This nonthreshold paradigm gave rise in the 1950s and 1960s

Received 5/23/91 ; accepted 10/7/91.' Supported in part by USPHS Grants CA32513, CA44886, and CA36727

from the National Cancer Institute and funding from the International LifeSciences Institute-Nutrition Foundation. Presented in part at the November 8,1990 meeting of the Committee on Risk Assessment Methodology, NationalResearch Council, Washington. DC 20418.

2Odeth E. Wall-Sylvia Havlik Professor of Oncology. To whom requests for

reprints should be addressed, at Department of Pathology and Microbiology,University of Nebraska Medical Center, 600 South 42nd Street, Omaha, NE68198-3135.

to the concept that prevention of cancer required completeelimination of all cancer causing chemicals from the environment, without regard to exposure levels (8). The likelihood thata chemical that produces cancer in animals also poses a risk tohumans led to the enactment of laws and regulations restrictinghuman exposure to chemicals in the environment and in food.

These laws and regulations are based on two assumptions:(a) a chemical that causes cancer in animals poses a cancer riskto humans; and (b) cancer risk at high doses implies a cancerrisk at low doses. These two assumptions still seem appropriatefor radiation and for most, if not all, genotoxic chemicals, whilerecognizing that there will be some variation in dose-responseoccurring between species and even among individuals within agiven species. However, considerable evidence is accumulatingwhich suggests that these two assumptions may not be appropriate for all chemicals whose risk for humans is based onevaluation of the results of animal bioassays. With numerousadvances in cell and molecular biology, genetics, chemistry, andpharmacokinetics increasing our understanding of mechanismsof carcinogenesis, we are in a position to reevaluate theseassumptions and to more rationally assess the cancer risk ofchemicals, particularly those that do not have direct genotoxiceffects. Increased cell proliferation appears to be a critical factorfor this approach.

Carcinogenesis: What We Know

There are certain aspects of carcinogenesis, regardless of theetiological agent, that are strongly supported by experimentalobservations. Any model of carcinogenesis and extrapolationsbetween species must take these basic principles into account.These are summarized in Table 1.

The first principle is that cancer results from genetic errorsin normal cells that have the potential to become a cancer. Thegenetic basis of carcinogenesis was originally stated by Boveriin 1914 (9) with his somatic mutation theory, and recentadvances in the molecular biology of cancer, with the discoveryof oncogenes and suppressor genes, have provided definitiveevidence (10-12).

The second basic principle of carcinogenesis is that morethan one genetic mistake must occur for cancer to develop. Itis unclear what specific genetic errors must occur, or how theyoccur, but the multievent requirement for carcinogenesis is wellestablished in both animal models and human tumors. In the1940s, several investigators began to present evidence thatcarcinogenesis could occur through multiple steps. These reports, which culminated in the publication by Berenblum andShubik (13) in 1947, defined the model referred to as initiationand promotion. These investigators, and subsequently others,have delineated characteristics of each of these two processes,although the specific mechanisms involved have yet to be defined (14-16). Many of the chemicals with promoting activitywere found to be nongenotoxic in a variety of short-termscreens, whereas most of the chemicals shown to have initiatingactivity were shown to be genotoxic (17, 18).

6493

Research. on January 9, 2021. © 1991 American Association for Cancercancerres.aacrjournals.org Downloaded from

http://cancerres.aacrjournals.org/

GENETIC ERRORS, CELL PROLIFERATION, AND CARCINOGENESIS

Table 1 Basic principles of carcinogenesis

1. Cancer arises due to genetic alterations.2. More than one genetic alteration is required.3. DNA replication is not 100% precise.

Although the initiation-promotion model was the first specific multistage model of carcinogenesis (13), a more definitive,genetically based, multievent model of carcinogenesis was defined by Knudson (19) in 1971 regarding retinoblastoma. Thismodel, including identification of the specific gene involved,has been described in great detail in numerous publications (20,21). Defects in both alÃelesof the Rb gene must be altered forcancer to arise in retinoblasts. In the inherited form of thedisease, a defect in one of the alÃelesis present from thegermline; the mistake in the second alÃeleoccurs during fetaldevelopment or childhood. In the sporadic form of the disease,both alÃelesbecome affected during development of the eye. Inthe inherited cases, since only one genetic error remains tohappen, patients usually develop multiple lesions which arefrequently bilateral. In contrast, the sporadic form of the disease, requiring both mistakes to occur during eye development,usually occurs singly and unilaterally.

The third basic principle of carcinogenesis is based on anobservation well known to geneticists for decades. Every timeDNA replicates, it does so without 100% fidelity (22, 23). Inother words, every time a cell goes through the process ofdivision, there is an opportunity, albeit extremely small, for amistake to be made in any gene, including those responsible forcancer development. DNA damage and replication errors thatgo unrepaired during the mitotic cycle are fixed and appear insubsequent cell progeny. Thus, cell replication contributes tocarcinogenesis, not by a new, ultimate, unknown mechanismbut by indirectly providing opportunities for somatic mutationsto occur.

A Biological Model of Carcinogenesis

The above three principles provide a framework within whicha general model of carcinogenesis can be constructed (24-27).The key feature is that an agent can increase the incidence ofcancer in a given tissue in one of two ways: either by specifically

damaging the DNA of the cell or by increasing the number ofcell divisions, thereby providing greater opportunity for a spontaneous genetic error to occur during normal cellular DNAreplication. It is assumed that genetic errors occur randomlyduring cell replication (recognizing that different portions ofthe DNA in different cells might be more susceptible to error),where the spontaneous error rate in a given gene is extremelylow (for example, 1 error/IO6 to IO8cell divisions). Further, the

genetic error must occur in a cell with the potential for developing into a cancer. This is generally the stem cell (or equivalent) population within a given tissue. Errors that occur in cellsthat have already begun the process of differentiation will notdevelop into a cancer since differentiated cells are committedto eventual death. Obviously, rates of DNA repair and othercellular repair mechanisms will affect the rate at which specificirreversible genetic alterations occur. Numerous mechanismswill enhance or inhibit these direct genetic and cell kineticeffects and thereby influence the carcinogenic potential of theagent (Fig. 1). Also, different tissues, whether slowly or rapidlyproliferating, are likely to have differing susceptibilities andrepair capacities, influencing responsiveness to various agents.

Using this framework as the starting point, a probabilisticbiologically based model of carcinogenesis has been formulated(24-27) (Fig. 2). This model assumes that carcinogenesis occursas a result of two genetic errors, but it could be extended toinclude settings that require an additional number of geneticevents (28, 29) (Fig. 3). It should be noted, however, that all ofthe experimental data sets examined thus far can be interpretedin terms of two irreversible genetic events.

The normal cell population in a given tissue is composed ofstem cells, cells committed to differentiation, and fully differentiated cells (30). In a layered epithelium, such as that of theurinary bladder, these cell categories can be readily visualized;stem cells are within the basal cell layer, committed cells are inthe basal and intermediate layers, and fully differentiated cellsform the superficial layer. In other tissues, distinction betweenstem cells and non-stem cells remains controversial. Nevertheless, under normal circumstances, the usual stem cell or itsequivalent (we will use the term stem cell for simplicity, realizing its limitations) divides into a stem cell to replace itselfand into a cell committed to differentiation to replace fully

Exposure

Fig. 1. Numerous factors can enhance orinhibit the sequence from normal to intermediate to malignant cells. These can relate directly to the chemical and its metabolism or tocellular responses to the chemical. +. effectsthat increase the likelihood of cancer development; â€”¿�,those decreasing it; Â±,those whichcan do either. A chemical can affect one ormore of the identified factors.

Differentiation

PhysiologicMilieu

MetabolicActivation

Excretion

PharmacokineticVariables

Cytotoxicity Â±

6494




Fig. 2. Diagrammatic representation of the two-stage model of carcinogenesis (from Ref. 24, reproduced with permission from IRL Press). Circles on left, stemcell states which lead to development of malignancy beginning in the lower left-hand corner with normal cells, which can undergo a change to intermediate cells,which in turn can undergo a change to transformed (malignant) cells. Circles to right of stem cell states, cells committed to differentiation and ultimately death.Downward arrows represent cell death directly.

Fig. 3. Multiple genetic alterations are frequently observed in intermediatepopulations. These alterations could occur sequentially and all be required forcancer development (28, 29) as illustrated in the top sequence. Alternatively, asillustrated in the bottom sequence, there might be two critical and essential geneticalterations which are required for cancer development, with other genetic andnongenetic alterations, such as changes in susceptibility to gene damage, cellproliferation, or differentiation which affect the normal and intermediate cellpopulations and enhance cancer development but are not required for cancer tooccur (from Ref. 71, reproduced with permission from the U.S. and CanadianAcademy of Pathology, Inc.).

differentiated cells that have died and have been removed. Thisreplacement process is necessary for maintaining the steady-state size of a given organ.

Rarely, when a stem cell divides, one of its daughter cellsundergoes one of the genetic errors required for cancer development. This cell then enters an intermediate population, solabeled because it has progressed partially along the pathwayto the full cancer development. If more than two genetic errorsare required for the development of cancer, there will be aprogression of intermediate populations, each with an increasing number of genetic errors (see Fig. 3). Ultimately, the final

genetic mistake occurs, producing a cell which will give rise toa malignancy. It must be emphasized that this cell is not abenign cell, such as those comprising papillomas or hyperplasticnodules.

Even though cell division rates can be very high, the actualtransitions between cell populations are low frequency eventsunder normal circumstances. The probability of a genetic transition event, once cell division is already in process, is represented in Fig. 2 as PÂ¡and P2 for transitions from normal tointermediate and from intermediate to malignant cellular states,respectively. To increase the number of normal cells that become intermediate cells, an agent can increase the number ofcell divisions in the normal stem cell population (by increasingthe number of cells in the normal stem cell population and/orby increasing the mitotic rate of these cells), or it can increasethe conditional transition probability P\. Similarly, to increasethe number of intermediate cells that become malignant, anagent can increase the number of cell divisions in the intermediate stem cell population (by increasing the number of intermediate stem cells and/or their mitotic rate), or it can increasethe conditional transition probability PI. Of course, an agentcan influence more than one of these variables.

Genotoxic Chemicals and Cancer

Most of the chemicals identified as human carcinogens or ascarcinogens in animal models prior to 1960 are genotoxic, i.e.,chemicals that directly interact with DNA and cause DNAdamage (2, 3). For most of these chemicals, there is metabolicactivation to reactive intermediates which bind to DNA, forming DNA adducts, which lead to DNA damage and ultimatelycancer. Radiation is similarly genotoxic (5).

6495



GENETIC ERRORS, CELL PROLIFERATION. AND CARC1NOGENESIS

Optimally, extrapolation of effects to low doses and especiallyextrapolations between species require considerable information, including that pertaining to differences in metabolic pathways and pharmacokinetic parameters, such as absorption,distribution, and excretion (31, 32). These parameters are considered in physiologically based pharmacokinetic models (31),which are useful for deriving species-specific information relating administered dose to target organ dose. However, in examining dose-response relationships, the response side of theequation is also important. Response at the cellular level, suchas metabolic activation, DNA repair, cellular toxicity, andregenerative proliferation, will greatly affect the carcinogenicresponse of a target tissue to a given target dose of a chemical.

Genotoxicity and Cell Proliferation: 2-AAF3

Dose-response differences for genotoxic chemicals due todifferences in cellular response have been most clearly delineated in an experiment involving more than 24,000 female micetreated with various doses of 2-AAF, with enough animals pergroup at the low doses to detect an increased tumor incidenceof 1% (33, 34). In that experiment, mice were given 2-AAF at30 to 150 ppm of the diet and sacrificed after 18, 24, or 33months of treatment. The resulting dose responses for the twotarget organs, liver and bladder, are shown in Fig. 4.

The obvious differences in the dose-response relationship forthese two target organs are not due to differences in pharmacokinetic factors, since Beland et al. (35) have demonstratedthat the dose relationships of DNA adduct formation in bladderand liver are linear to doses as low as 5 ppm. The bladderactually had higher levels of adducts than the liver. The differences in dose-response relationships can, however, be readilyexplained in terms of differences in cellular response.

In the liver, 2-AAF is metabolically activated by normalhepatocytes but not by those in the intermediate cell population.Also, at the doses utilized in this experiment, there was noeffect on hepatocyte proliferation. Thus, the only parameter inthe liver affected by 2-AAF at administered doses was Pt. Incontrast, in the bladder the reactive metabolite (different fromthe one in the liver) interacts with both normal and intermediatebladder cells; in addition, there is increased cell proliferation atdoses of 60 ppm and above (34). Correspondingly, at doses of60 ppm and above, there is a sharp increase in the tumorincidence in the bladder, whereas at the lower doses there isapparently no detectable increase in incidence. These dose-response relationships have been analyzed in detail previously(34). Although the lower 2-AAF doses did not produce asignificant increase in bladder tumor incidence, in contrast tothe liver, the fact that DNA adducts were formed at the lowerdoses in both tissues clearly indicates that there is an effect atthese doses. The shape of the dose-response curve for thebladder reflects the interaction between cell proliferation anddirect genetic damage. The lack of a detectable effect in thebladder at doses below 60 ppm, despite DNA adduct formation,indicates that the cancer effect was below the level of detection(1%) of the study. On the basis of these investigations, thischemical fits all of the principles that we have described abovefor a genotoxic carcinogen, including the absence of a no-effectdose threshold.

Different dose-response relationships are anticipated for different chemicals depending on the dose response of DNA

'The abbreviations used are: 2-AAF, 2-acetylaminofluorene; NTP, NationalToxicology Program; TCDD, 2,3,7,8-tetrachlorodibenzo-p-dioxin.

0 30 45 60 75 100

Dose (ppm)

Fig. 4. Model results for effects of 2-AAF dose on liver (â€¢)and bladder tumor(A) prevalences after p.o. administration to mice for 24 months. Symbols representdata from the ED0, study and lines represent the results of the modeling analyses.These analytical results are consistent with actual data from the ED0i study (33,34).

adduct formation (or other types of DNA damage), cell proliferation, and various repair mechanisms. A small increase intumor incidence at low doses, where genetic interactions occur,with much steeper increases in tumor incidences at higher doses,where there is genetic interaction and increased cell proliferation, is observed with a variety of chemicals, such as thenitrosamines dimethylnitrosamine and diethylnitrosamine, inrat liver (36-39).

In summary, for genotoxic chemicals, the classical paradigmof a nonthreshold response is likely to be correct, but the actualdose-response relationship for tumor incidence is greatly influenced by an effect of a chemical on cell proliferation.

Nongenotoxic Agents and Cancer

An increasing number of chemicals are being discoveredwhich are not metabolically activated to reactive intermediates,do not interact with DNA directly, are not genotoxic in mostshort-term screens, and yet increase the incidence of cancer inexperimental animals (40, 41). In terms of the model presentedabove, "nongenotoxic" means that there is no increase in either

of the conditional genetic transition probabilities, P\ and P2.Thus, for these chemicals the only remaining mechanism forincreasing the incidence of cancer is through increasing thenumber of cell divisions (25). Even with a replication error rateper cell division that is the same as that in control, untreatedanimals, an increased number of opportunities for spontaneousgenetic mistakes will translate into an increased risk of cancer.

Foreign Bodies. Foreign bodies represent the simplest examples of such nongenotoxic agents (42). Thin films of varioussubstances implanted into the subcutaneous tissue of rodentscan result in an increased incidence of sarcomas. The specificmechanisms are unclear but appear to be dependent entirely ona continuous increase in the proliferative rate of the fibroblastssurrounding the implanted material. There is an obvious threshold effect; either the material is present or not, although anobject of at least some minimum size may be required. Puttingholes in the object also permits repair of the process, ratherthan sustained proliferation, and tumors form to a much lesserextent.

Implantation of pellets of various materials, such as glass,stainless steel, cholesterol, or paraffin wax, into the bladderlumen of rats or mice also produces cancer (43-45). Depending

6496




on the smoothness of the pellet surface, variable degrees ofurothelial abrasion occur with consequent regenerative hyper-plasia. The hyperplastic response continues as long as the pelletremains in the bladder lumen, and eventually carcinomas develop. The incidence of carcinoma increases with time. Forexample, Jull (46) implanted paraffin wax pellets into 3-month-old mice and observed cancer incidences of 10.6% after 40-50weeks, 26.8% after 70-80 weeks, and 53.8% after 100-110weeks. Again, there is an obvious threshold response: either thepellet is present or it is not. Several variables appear to influencethe carcinogenic response to such pellets, including smoothnessof the pellet, differences between species (the mouse tends tobe less susceptible than the rat), rates of cell death as well asproliferation rates, and the presence of urine (probably relatedto the high levels of growth factors in urine). Thus, despite highrates of cell proliferation, not every animal with a pellet (or, asdescribed below, a calculus) will develop a cancer in a 2-year orshorter bioassay. It is, remember, a probabilistic phenomenon.

Calculi. In contrast to the implantation of foreign materialsdirectly into the bladder, foreign material can be generated byadministering chemicals which produce urinary calculi (43,47-49). The calculi can form from the ingested chemical itself,such as melamine or uracil; a metabolite of the ingested chemical, such as from glycine; or a substance derived from physiological alterations secondary to the administered chemical, suchas increased phosphate, calcium, or oxalate excretion (Table 2).For example, feeding uracil as 3% of the diet to rats (50, 51) ormice (52) leads to the formation of uracil calculi, a rapidincrease in the proliferative rate of the bladder epithelium, andthe formation of nodular and papillary hyperplasia. This represents a marked increase in the number of urothelial cellswhich, when combined with their increased proliferative rate,produces an enormous increase in the total number of celldivisions occurring over time. In contrast, if uracil is administered in the diet at a level of only 1%, still a high dose, there isno formation of urinary calculi, no increase in cell proliferation,and no formation of tumors.

Significantly, for all calculus generating chemicals, a certainthreshold level of the chemical must be administered to generatethe calculi that ultimately lead to the production of tumors.This threshold can be determined through measurement ofpertinent chemical and physiological factors. For such chemicals, statistical extrapolations to low dose exposures from highdose responses are inappropriate and misleading. Thresholdexposure levels for humans can be estimated by taking intoaccount the specific factors involved in the formation of thecalculi. A qualitative evaluation of this nature has recently beenused by the United States Environmental Protection Agency inruling that melamine does not pose a carcinogenic risk forhumans, despite its urinary bladder carcinogenicity in rats (53-

55).Sodium Saccharin and Crystalluria. Sodium saccharin poses

a similar, albeit somewhat more subtle mechanism (56). Sodiumsaccharin has been shown to increase the incidence of bladdercancer in rats when administered beginning at conception or atbirth and continuing through the lifetime of the offspring, withconsiderably less activity if administration is begun after 6weeks of age. The lowest dose with which there has been anincreased effect on carcinogenesis is approximately 2.5%; noeffects are observed at 1% of the diet. In addition, if saccharinis administered at doses as high as 7.5% of the diet, but as theacid rather than the sodium salt, no effect is seen. Similarly,any treatment which results in acidification of the urine while

Table 2 Chemicals that lead to urinary calculi"

Uracil

MelamineUric acidHomocysteineCysteine oxalatesCalcium oxalateCalcium phosphateDiethylene glycolBiphenyl

4-Ethylsulfonylnaphthalene-1-sulfonamide

OxamideAcetazolamideTerephlhalic acidDimethyl terephthalateNitrilotriacetatePolyoxyethylene-8-stearateGlycineOrotic acid

Â°This list is not meant to be complete but indicates some of the wide range of

chemicals capable of leading to urinary calculi.

sodium saccharin is administered, e.g., by coadministration ofNH4C1, inhibits the effects on the urothelium (57). The malerat is considerably more susceptible than the female rat, andthe mouse, hamster, and monkey appear to be resistant to thebladder effects of sodium saccharin (56). Multiple large epidemiolÃ³gica!studies suggest that humans are also resistant to theeffect of saccharin on the urinary bladder (56, 58).

Saccharin is not metabolically activated to a reactive electro-phile (it is actually nucleophilic with a pKa of approximately2.8), it does not react with DNA, and it is not mutagenic inmultiple short-term screens (56). However, it does increase theproliferative rate of the adult rat bladder epithelium approximately 5- to 10-fold. It is also capable of increasing the proliferative rate during the first 3 weeks of the rats' life, a time

during which the bladder normally proliferates rapidly, as itdoes ;'/; utero; in the adult rat, the epithelium normally is

mitotically quiescent. As described in detail elsewhere (56, 59),P, and P2 can be kept at control levels in modeling analysesand the additional numbers of cell divisions during the lifetimeof the rat given high doses of sodium saccharin are sufficientto account fully for the increased incidence of bladder cancerin the rat if only one biologically plausible assumption is made:the proliferation of intermediate cells is enhanced more thanthat experienced by the normal cell population.

Although multiple variables which increase or decrease theeffects of sodium saccharin on the rat bladder have been identified, the formation of silicate precipitates and/or microcrystalsin the urine appears to be central to the production of increasedurothelial proliferation (25, 56, 60). These precipitates andcrystals form if the urinary pH is above approximately 6.5;their formation is enhanced by increasing concentrations ofsodium (and possibly potassium), and large amounts of proteinare required for their formation. It is well known that the malerat has large amounts of protein compared to the female andalso that rats have considerably greater amounts of urinaryprotein than humans (61), even more than patients with thenephrotic syndrome. It appears that not only is the amount ofprotein important, but the type of protein is critical as well.Low molecular weight proteins, such as a2u-globulin, appear tobe more effective in producing silicate precipitate and crystalsthan higher molecular weight proteins, such as albumin. Takingthese multiple factors into account, including silicates, pH,protein, and sodium, it is likely that sodium saccharin is not carcinogenic for humans, even if it could be administered at thehigh doses fed to rats. The human appears to be like the mousewith respect to a lack of urothelial response to sodium saccharin(58, 62). The mouse is resistant to the effects of sodium saccharin even at doses as high as 7.5% of the diet. Humans alsodo not appear to have either a bladder proliferative (62) orcarcinogenic (56, 58) response to sodium saccharin.

Using mechanistic information such as this, one can estimatethe risk to humans more realistically than by extrapolating high

6497



GENETIC ERRORS. CELL PROLIFERATION. AND CARC1NOGENESIS

dose rat data using some mathematical-statistical formulation(56, 59, 63-65). Because, in fact, the human is unlikely torespond to sodium saccharin with the formation of bladdercancer, any dose could be tolerated. However, even if we wereto assume that humans are like the male rat, the most susceptible sex and species, with a no-effect threshold at approximately 1% of the diet, an estimated threshold dose for humanswould be on the order of 25,000 mg/day for a 70-kg person. Incontrast, application of the linear extrapolation approach forestimating risk results in a maximum allowable human dose ofapproximately 3 mg/day.

Vitamin C. The difficulties that can be presented if thresholdsand mechanisms are not taken into account when dealing withnongenotoxic chemicals are illustrated by considering the substance vitamin C. Sodium ascorbate (and for that matter, mostsodium salts of organic acids) behaves like sodium saccharin inthe male rat (66-68). Ascorbic acid itself enhances rat bladdercancer formation if coadministered with sodium bicarbonate tokeep the urinary pH elevated (69). The doses at which sodiumascorbate is effective are comparable to those for sodium saccharin (57, 66). If mechanistic considerations and thresholdeffects are not taken into account in extrapolating results fromthe male rat to humans, a maximum allowable dose for a riskof one new bladder cancer/IO6 population is approximately 3

mg/day, as with sodium saccharin. Yet, for this essential vitamin, the minimum recommended daily allowance to avoidscurvy is 60 mg/day (70). Obviously, there is no increased riskof bladder cancer by taking vitamin C at a level of 3 mg/day,60 mg/day, or even higher doses. Biological considerationsmust be taken into account in extrapolating animal data tohuman exposures.

Chemicals Involving Tissues Other Than Bladder. Althoughwe have focused our examples on chemicals related to bladdercarcinogenesis, similar mechanistic considerations and analysespertain to essentially any organ system that has been examinedin experimental models and in human carcinogenesis (24-27,71). For example, several chemicals increase incidences ofkidney cancer in male but not female rats secondary to interactions with Â«^-globulin (72). Differences in the types of urinary proteins between sexes and between species make humancancer risk from these substances unlikely. Butylated hydroxy-anisole at high doses in the diet increases proliferation of thesquamous epithelium of the rat and hamster forestomach, eventually producing carcinomas. This chemical poses two problemsfor extrapolation to humans (73, 74). In addition to extrapolating from high to low doses, there is no analogue in humans forthe rodent forestomach. The closest analogue is the esophagus,but butylated hydroxyanisole has no effect on the rodent esophagus. A large number of chemicals increase the incidence ofthyroid neoplasms in rats, but the mechanisms are not directlypertinent to the human situation (75, 76). PhÃ©nobarbital increases rat thyroid cancer by enhancing the clearance of thyrox-ine, with consequent increased thyroid stimulating hormoneproduction and subsequent increased proliferation in the targettissue, the thyroid. PhÃ©nobarbitalhas also been shown to be aliver carcinogen in rodents, even though it is not genotoxic (77,78). Again, its effect is through increasing proliferation in thetarget tissue, predominantly by acting on those cells that havealready entered into the intermediate cell population, and athreshold dose must be reached to produce an effect. There isno evidence of human thyroid or liver carcinogenicity by phÃ©nobarbital in several large epidemiolÃ³gica! investigations ofhuman populations (79).

Classification of Chemicals

The foregoing discussion of mechanisms has important implications for the terms scientists and regulators use to characterize and categorize chemicals associated experimentallywith cancer induction. Chemicals that increase the risk ofdeveloping cancer have traditionally and somewhat loosely beenreferred to as carcinogens. However, with our increased understanding and dissection of the mechanisms involved in carcinogenesis, this becomes a particularly difficult term to define.Adding complexity to the semantics is the use of terms such ascomplete carcinogen, promoter, initiator, progressor, cocarcin-ogen, complÃ©ter,converter, etc. The difficulty with all of theseterms is that they rely on specific experimental protocols fortheir definition rather than being defined from the perspectiveof a biological process. We prefer to refer to chemicals thatenhance the carcinogenic process in terms indicative of themechanism by which they can act (25): either direct DNAdamage (genotoxic) or increased cell proliferation (nongenotoxic). Of course, chemicals can do both. This classificationscheme is illustrated in Fig. 5.

Chemicals are initially evaluated in animal bioassays to determine whether they influence the carcinogenic process. Therehas been considerable argument as to the appropriate doses,species, strains, and several other details of the bioassay (80-84). There is general agreement, however, on the necessity forlong-term bioassays to evaluate whether a chemical can increasethe risk of developing cancer under specific circumstances,because short-term screens, whether for mutagenicity or cellproliferation, are inadequate for definitively predicting whetherchemicals will or will not increase the risk of cancer (41, 85-87). Once a chemical has been shown to increase the risk ofcancer in a bioassay, it can be placed into either the genotoxiccategory or the nongenotoxic one. Although exceptions canoccur, the present battery utilized by the NTP for evaluatinggenotoxicity should be able to serve as an initial screen for thisdetermination (88, 89). Unique circumstances for specificchemicals can be ascertained as appropriate. The distinctionbetween genotoxic and nongenotoxic is critical for a properdose and species extrapolation.

CHEMICAL CARCINOGEN

GENOTOXIC

Threshold unlikelyDose-response may be affectedby cell proliferation(usually toxicity relatedat high doses)

NON-GENOTOXIC

Reaction withcell receptor7

PROLIFERATIVE

1. Threshold questionable2. Usually eftective at

low doses

PROLIFERATE

Threshold likelyUsually relatedto toxicity andregeneration

Fig. 5. Proposed classification scheme for carcinogens (from Ref. 25; Copyright1990 by the American Association for the Advancement of Science). The effectof genotoxic chemicals can be accentuated if cell proliferative effects are alsopresent. Nongenotoxic chemicals act by increasing cell proliferation directly orindirectly, either through interaction with a specific cell receptor or nonspecificallyby (a) a direct mitogenic stimulus, (ft) causing toxicity with consequent regeneration, or (c) interrupting physiological process. Examples of the latter mechanisminclude thyroid stimulating hormone stimulation of thyroid cell proliferation aftertoxic damage to the thyroid and viral stimulation of proliferation afterimmunosuppression.

6498



GENETIC ERRORS. CELL PROLIFERATION, AND CARC1NOGENESIS

No-Effect Thresholds and Mechanism. For genotoxic compounds it can be assumed that a no-effect threshold does notexist. A threshold is likely for any accompanying proliferativeeffects, which will significantly influence the dose response ofthe chemical, as illustrated above for 2-AAF (33-35) and as istrue for numerous other genotoxic chemicals (2, 3, 5, 7, 36-

38).Nongenotoxic chemicals act at specific doses and with spe

cific mechanisms to increase cell proliferation, which in turnincreases the opportunities for mutagenic events in dividingcells and, ultimately, increases the risk of cancer (25,40). Thesecompounds can be divided (25) into those that interact throughspecific cell receptors, such as the phorbol esters and TCCD,and chemicals that do not, such as sodium saccharin, sodiumascorbate, melamine, phÃ©nobarbital,and rf-limonene. A chemical which acts nonspecifically at the onset can ultimately triggera specific hormone or other substance which acts through a cellreceptor. For example, many of the chemicals that affect thy-roxine metabolism ultimately lead to an increase in thyroidstimulating hormone production, which clearly acts through aspecific cell receptor in the thyroid to produce increased cellproliferation in that organ. Hormones associated with increasedcancer risks, such as estrogen, act through specific cellreceptors.

Despite the fact that agents acting through specific cell receptors do so at low doses, they nonetheless are likely to havea threshold for a response. This has been suggested for over 30years for substances such as the phorbol esters (initially administered as croton oil), for which a threshold phenomenon hasbeen well described (15, 16). Like other physiological responsesinvolving specific cell receptors (90), a certain percentage of thereceptors on a cell must be occupied for the cellular responseto occur; it takes more than a single molecule interacting witha single cell receptor to produce a cellular response.

For chemicals not acting through a specific cell receptor,threshold mechanisms certainly apply.

TCDD: A Nongenotoxic Experimental Carcinogen. TCDD isan example of an active compound where current classificationterminology (e.g., promoter, secondary carcinogen, indirectcarcinogen) is found lacking, if not misleading (91-93). TCDD

is not genotoxic, it increases cell proliferation in the targettissue, and it is one of the most potent cancer producingchemicals in rodents yet discovered (94). Nevertheless, since itdoes not act through direct DNA damage, several investigatorsare hesitant to call it a carcinogen, instead relating its action tosome unspecified promoting mechanism. Such compounds donot act through some nebulous mechanism which cannot bedefined further than calling it promotion; TCDD is a chemicalthat increases the risk of developing cancer in a specific animalby increasing cell proliferation in the target tissue. By itself, itcan induce liver cancer and tumors of other organs at dosesthat are substantially lower than classical liver carcinogens,such as the azo dyes and 2-AAF. There is suggestive evidencethat TCDD might increase the risk of developing certain cancers, such as soft tissue sarcomas, in humans (95), especially athigher exposure levels. Is it reasonable to continue to refer to2-AAF and the azo dyes as carcinogens and to refer to TCDDas a promoting substance? They all increase cancer risk, although they do so through different mechanisms. On the onehand, 2-AAF and the azo dyes interact directly with DNA andalso have cell proliferative effects, while TCDD only has thelatter capability.

And yet, because metabolites of 2-AAF and azo dyes damage

DNA directly and TCDD does not, they are biologically quitedifferent. Risk extrapolation from rodent bioassays to humansrequires different approaches for nongenotoxic compared togenotoxic chemicals (25). Calculus forming chemicals are obvious examples of chemicals that can produce cancer in experimental animals at high doses, but at doses to which humansare exposed, there is no expected potential for risk of cancerdevelopment. At present, the classification system used by theInternational Agency for Research on Cancer does not specifically designate such a category (96). If a chemical producescancer in an experimental animal, it is assumed to have potential for producing tumors in humans. A new category is neededfor chemicals that have probable or possible carcinogenic activity in experimental animals but are not expected to be carcinogenic in humans. To be sure, only nongenotoxic chemicalsshould be considered to be classified for inclusion in such acategory.

Cell Proliferation:Phenomenon

A Quantitative and Qualitative

Considerable controversy has evolved concerning the evidence that increased cell proliferation can be a mechanism bywhich an agent increases the risk of developing cancer (25, 80-84, 97-99). In examining the connection between cell proliferation and cancer, there has been some attempt to equate generalcytotoxicity with cell proliferation. Finding little correlationbetween toxicity and carcinogenicity based on analysis of chemicals examined in NTP bioassays, Hoel et al. (100) concludedthat toxicity is not a pertinent mechanism for carcinogenicity.What was not appreciated, however, is that cell proliferation isnot increased by all types of toxicity, including several forms ofneurotoxicity, cardiotoxicity, immunotoxicity, and nutritionaltoxicity, to mention but a few. Also, a chemical can producesuch severe toxicity that the target organ is severely reduced insize, resulting in fewer cells. For example, monuron producessevere growth retardation and a liver approximately one-halfthe size of that in control mice ( 101). In this case, the compoundwould have to produce a doubling of mitotic rates to offset the50% loss in cell numbers just to maintain the same number ofcell divisions as in controls. Examining only cell division ratesin estimating cancer risk would be very misleading.

There are several reasons why simply measuring cell proliferation is inadequate for judging whether a chemical will ultimately increase the risk of cancer. First, the relevant proliferation is only that which occurs in the stem cell (pluripotential)population of the tissue, not in the differentiated cells. This ismost convincingly apparent in comparing adenomatous polypsin the human colon to hyperplastic polyps (71, 102). Adenomatous polyps involve proliferation of the crypt (stem) cells andhave a significant risk of eventually evolving into cancer. Hyperplastic polyps, in contrast, are proliferations of the mucuscontaining differentiated cells and do not have any increasedcancer risk, even if an individual has a large number of them.

Second, to develop a detectable level of cancer incidence,some aspect of the cellular proliferation must usually be sustained (25). For example, di(2-ethylhexyl)phthalate, a chemicalknown to stimulate peroxisome proliferation and produce cancer in the rat liver, produces an initial burst of mitotic activityin normal hepatocytes, but the mitotic rate then rapidly returnsto normal (103). However, by the time the labeling indexmeasure of mitotic activity has returned to normal levels, theliver is 30% larger than normal and it remains larger. This

6499




increase might be due partially to an increase in cell size(hypertrophy) but also includes an increase in cell number(hyperplasia). Thus, the cell population to which the labelingindex applies has increased, and although the rate is the sameas that in control animals, the total number of cell divisionsoccurring in hepatocytes is greatly increased. In evaluating cellproliferation studies, it is essential to take into account the totalnumber of stem cell divisions; not only mitotic rate but thenumber of cells present must also be ascertained.

Third, the mitotic rate may not be markedly increased in thenormal cell population, but that does not mean that there isnot a proliferation stimulus preferential to cells already in theintermediate population, such as in hepatocellular foci in therat liver. This holds true for another peroxisome proliferator,Wyeth-14,643 (103), and for phÃ©nobarbital (104). With phÃ©nobarbital there is actually little stimulation of normal hepatocytes, but a marked increase in cell division rates occur withinthe foci. The number of cells in these foci also increases dramatically, possibly due not only to the increase in the prolifer-ative stimulus but also to a decrease in cell death rates.

Inequalities in Carcinogenesis. Other examples have beencited of increased cell proliferation without cancer and, conversely, increases in cancer production without an apparentincrease in cell proliferation (97-100). These examples are citedto refute the relationship of cell proliferation to carcinogenesis.However, this is analogous to the observation that there arechemicals that are positive in short-term genotoxicity screensthat do not produce cancer in 2-year animal bioassays and thereare chemicals which produce cancer in 2-year bioassays that arenot mutagenic in short-term screens (41). To conclude thatmutagenesis is unrelated to carcinogenicity is obviously incorrect given the genetic basis of cancer. The same simplisticthinking with respect to cell proliferation is equally illogicaland noninformative. Three essential inequalities mut be considered in carcinogenesis risk assessment:

Mutagenesis ^ Carcinogenesis

Cell proliferation ^ Carcinogenesis

Toxicity ;* Cell proliferation

These inequalities do not mean that cell toxicity, cell proliferation, or mutagenicity are unrelated to the carcinogenic process, they only reinforce the point that the carcinogenic processis complex and involves considerable interaction between various biological variables.

This complexity extends to mechanistic considerations,whether related to genetic damage or to proliferation. Forexample, a chemical that is mutagenic in a short-term in vitroscreening assay, such as the Ames test, does not necessarilyproduce mutagenic consequences in a whole organism, possiblydue to differences in metabolic activation, inactivation, or DNArepair, among other possibilities. However, a chemical which ismutagenic in a short-term screen should be considered to havethe potential for cancer production until demonstratedotherwise.

Similarly, any chemical that increases cell proliferationshould be viewed as suspicious with regard to increasing cancerrisk. Of critical importance for chemicals with only proliferativeeffects, as emphasized above, is that mechanism as well as dosemust be considered in extrapolation between animals and humans. It must be realized that the effects of a chemical at highdoses do not necessarily imply similar changes at low doses,

and the changes in a rat or a mouse do not necessarily implysimilar changes in humans. That is not to say that the rodentbioassay system is unpredictive. It only means that greater caremust be taken in making the extrapolation from results ofrodent bioassays to human risk than simply assuming that apositive result in the bioassay implies a risk to humans, nomatter what the dose.

Because of dose and/or cellular response differences betweenspecies, utilizing high doses of nongenotoxic chemicals in rodent bioassays will produce a large number of false positiveswith respect to human cancer risk (25). Like any test, specificityis sacrificed to increase sensitivity (105). To assess human risk,the result of the bioassay is only a beginning which must befollowed by mechanistic considerations, including questions ofgenotoxoicity versus nongenotoxicity, threshold, dose effects,pharmacokinetics, and cell proliferative effects.

Illustrative Examples

We present several theoretical modeling exercises [using amathematical simulation model described elsewhere (24, 25,34, 59)] to illustrate the dynamic nature of the interactionsbetween cell proliferation, genetic events, and the developmentof cancer. To begin with, every tissue in the organism, whetherrat, mouse, or human, has a certain background ("spontaneous") incidence of tumors. Utilizing information from the

NTP, we estimate a background incidence of liver cancer ofabout 0.5% by the end of a 2-year bioassay in male F344 rats.With a background probability of genomic error in each of twocritical alÃelesof 10~7and utilizing information from the liter

ature regarding normal liver growth and development, thecumulative cancer incidence over time can be simulated, asillustrated in Table 3.

In Table 3, as in the ones to follow, we have indicated thenumber of normal hepatocytes in the liver during the rats'

lifetime, the number of newly intermediate cells (reflective ofthe number of foci), the total number of intermediate cells(reflective of the total number of cells in these foci), and thecumulative incidence of tumors. We have made the assumptionthat a malignant tumor can be recognized if it contains IO'1cells(approximately 1 mm3) and the animal is then identified as

positive for malignancy. The development of one malignant celldoes not result in the diagnosis of a malignancy, but rather,there is a time lag for the development of the tumor to adetectable size, either clinically or histopathologically.

As shown in Table 3, the liver during fetal and neonatalgrowth is a relatively rapidly proliferating tissue, but in olderanimals it is mitotically quiescent and grows very slowly, essentially replacing dying hepatocytes. It is also evident that, even

Table 3 Simulation of control rat liver

Age(wk)-1

036

2652

104130No.

ofnormal

hepatocytes3.0x 10Â«

9.2 x IO74.0 x 10"8.1 x 10"1.6X 10'1.8 x 10'1.9 x 10"1.9 x 10'No.

of newintermediatecells(foci)"10(0)

103(0)244 (0.2)764 (0.6)

1141 (1.1)1805(2.0)2148(3.5)No.

of totalintermediate

cells2.8

x 10'2.9 x IO27.8 x IO22.3 x IO33.4 x IO38.0 x IO31.2 x 10*Cumulative

probabilityof visibletumor*0.00013

0.000970.003770.00633

Â°Foci assumed to be detectable with 250 cell clone.4 Malignant tumors with 10' cells.

6500



GENETIC ERRORS, CELL PROLIFERATION. AND CARCINOGENESIS

Table 4 Simulation of the effects of a single dose of a genotoxic chemical at 6weeks of age, producing immediate hepatocyte death followed by a short-termspike in both mitotic activity and the genetic error probability for normal cells

Age (wk)6

78

2652

104130No.

ofnormal

hepatocytes8.1x 10"

5.8 X 10"6.6 x 10"9.7 x 10"1.2 x IO91.3 x IO91.3 x IO9No.

of newintermediatecells(foci)"244

(0.2)5371 (0.2)5406 (0.2)5650 (0.6)5945 (2.0)6394 (3.5)6631 (5.9)No.

of totalintermediate

cells7.8x IO2

6.3 x IO37.2 x IO31.3 x IO42.5 x IO45.2 x IO47.5 x IO4Cumulative

probabilityof visibletumor*0.00016

0.005630.025800.04200

Â°Foci assumed to be detectable with 250 cell clone.4 Malignant tumors with 10' cells.

Table 5 Simulation of the effects of continuous administration beginning at 8weeks of a nongenotoxic chemical, producing a preferential proliferative

stimulation of intermediate cells

CumulativeNo. of total probabilityintermediate of visible

Age (wk)

No. ofnormal

hepatocytes

No. of newintermediatecells (foci)0 cells tumor

6g26521041308.1Â±10"1.1x10'1.7x10'2.0x10'2.1xIO92.2x IO9244

(0.2)361(0.2)879

(3.5)1399(27)2295(518)2762

(967)7.8

xIO21.2xIO31.0xIO44.6x10"8.2x10!3.4x IO60.000210.009200.223000.66100

1Foci assumed to be detectable with 250 cell clone.*Malignant tumors with 10' cells.

with a background rate for PÂ¡and P2 of 10 7, all rats by thebeginning of a 2-year bioassay (6 weeks of age) already haveinitiated cells, usually several hundred. This is referred to byseveral investigators as the background incidence of liver cellfoci (39). This is the natural consequence of replication errorsthat occur with cell division during normal growth and development. It is also apparent that the rate of growth within theinitiated (and normal) cell population slows during the remaining lifetime of the animal; there is no exogenous stimulus forcells to proliferate at a high rate. Despite the rapid proliferationoccurring during development of the liver, there are insufficientcell divisions in the intermediate population to expect that amalignant cell will be produced by the time of birth. Eventuallywith continued proliferation in hepatocellular foci, the development of a malignant tumor is possible, and by 2 years thereis a cumulative incidence of 0.377%.

It is unclear what the stimuli are for background ("spontaneous") genetic errors and proliferation rates. It has been sug

gested that oxidative damage is responsible for many of these,but other mechanisms might also be applicable (80-82).

If a genotoxic chemical (e.g., diethylnitrosamine) is administered which is active toward the liver and also has a cellproliferative effect, it is easy to see how the incidence of livertumors will be increased. In Table 4, such a chemical is administered as a single dose at 6 weeks. There is a rapid increase inthe number of initiated cells following administration of thechemical, but not much of an effect on eventual cancer development because of the absence of any sustained increase in cellproliferation and a return to background levels of genetic errorrates shortly after the chemical is removed. Many intermediatecell foci are created but very few evolve into foci of any appreciable (detectable) size.

If a nongenotoxic chemical (e.g., phÃ©nobarbital)is administered which causes increased cell proliferation, especially of thecells in the intermediate population, there is a gradual increasein the number of normal hepatocytes entering the intermediate

cell population. In particular, there is a noticeable increase inthe number of initiated foci of detectable size and of total cellsin the intermediate population (Table 5). Intermediate cells arecontinuously proliferating. It is obvious that tumors can evolvein these animals by 2 years of age. If the dose is such that itproduces a proliferative stimulus greater than that modeled, theincidence can reach 100%. On the other hand, if the dose ofthe nongenotoxic chemical is below the threshold required toincrease cell proliferation, there will be no increase in cancerincidence over control levels.

Combining these two sceneries, the administration of a singledose of a genotoxic stimulus followed by a proliferative stimulus(Table 6), results in a high incidence of liver tumors by 2 yearsof age. There is a slight decrease in the number of detectablefoci compared to the number seen with only the proliferativeagent alone because of the disruptive effect of the prior short-term administration of the genotoxic compound.

Table 7 presents a simulation where genotoxic effects onnormal cells are added to the proliferative compound previouslyconsidered in Table 5. This could correspond, for example, tolong-term administration of the genotoxic chemical illustrated

in Table 4, but at a lower, less toxic dose. There is a dramaticincrease in the number of new intermediate cells produced fromthe normal cell pool. So, after allowing time for the aggressiveclonal expansion to manifest itself, a large number of the focireach a detectable size. It should be noted that if genotoxicitywere directed toward already initiated cells (i.e., P-Â¿is elevatedinstead of/>i), the appearance of malignant tumors is acceler

ated: 59% by 1 year and 100% by 1.5 years.No matter when the stimulus is given, if there is an increase

in genotoxicity and/or increase in cell proliferation, there isultimately a chance of an increased risk of developing cancer.However, as is evident from these analyses, if the stimulusoccurs sufficiently early in the animal's life, spontaneous genetic

error rates and normal cell division rates often will be adequateto carry the process to a detectable incidence of malignancywithout the need for a continuous exogenous stimulus.

Table 6 Simulation of the effects of a single dose of a genotoxic chemical at 6weeks of age followed by continuous administration beginning at 8 weeks of age

of a nongenotoxic chemical which stimulates cell proliferation

Age(wk)682652104No.

ofnormalhepatocytes8.1

xIO86.6x10"1.1xIO91.2X

10'1.3x 10'No.

ofnewintermediatecells

(foci)0244

(0.2)5406(0.2)5720(3.5)6036(23)6579

(507)No.

oftotalintermediatecells7.8

xIO27.2xIO35.6x10"2.4x10'4.3x IO6Cumulativeprobabilityof

visibletumor*0.000620.048800.73400

1Foci assumed to be detectable with 250 cell clone.' Malignant tumors with IO6cells.

Table 7 Simulation of the effects of continuous administration beginning at 8weeks of a chemical which produces a preferential proliferation stimulation of

intermediate cells and is also genotoxic for normal cells

Age(wk)682652104No.

ofnormalhepatocytes8.1

x10"1.1xIO91.7x10'2.0x10'2.1x 10'No.

ofnewintermediatecells(foci)0244

(0.2)361(0.2)52,121

(3.5)104,113(27)193,685

(>10,000)No.

oftotalintermediatecells7.8

xIO21.2xIO31.5x10'7.7x10'1.4x IO7Cumulativeprobabilityofvisibletumor*0.000210.110000.98600

" Foci assumed to be detectable with 250 cell clone.4 Malignant tumors with 10' cells.

6501




Initiation-Promotion-Progression. In these examples, the effects on cancer incidence over time by agents that damage DNAdirectly and/or increase cell proliferation can be seen. Eachscenario leads to an increased incidence of cancer, but theapproach avoids the confusion of the terminology of initiation,promotion, and progression. Theoretically, the model of initiation-promotion-progression first involves the short-term administration of a genotoxic agent for initiation, which accelerates the transition of normal cells to the intermediate population. Promotion then follows and involves long-termadministration of a nongenotoxic agent, causing clonal expansion of the intermediate cell population. Progression is simplythe genetic transition of a benign intermediate cell to a fullymalignant cell. In our model, a pure initiating chemical wouldaffect only Pt, a promoter would affect the proliferation ofintermediate cells, and a chemical acting primarily in the progression mode would affect PÃŒ.However, the genetic transitionof cells from the intermediate population to the malignant stateis also influenced by increasing the mitotic rate of intermediatecells and does not require a direct effect on P2. Because of this,it is not always easy to separate out the difference betweenpromotion and progression in experimental models. Difficultiesalso arise in the real world with this terminology becausechemicals usually affect more than one variable. Even with asufficiently large effect on only one variable, the backgroundlevels of the other variables are not zero and will, therefore,have some contribution to the production of cancer.

Many agents clearly do not fit into the model of initiation-promotion-progression. For example, a single dose of aflatoxincan produce liver cancer in rats without administration of asubsequent "promoter" (106). UlcÃ©rationof the rat bladder

mucosa behaves as a classical initiating stimulus, yet there isno genotoxic stimulus during the ulcÃ©rationand regenerativehyperplasia (107-110). UlcÃ©rationinduces changes in the bladder that are morphologically reversible, but with an irreversibleinfluence that is remembered by the cells. They can then beenhanced subsequently by a chemical, such as sodium saccharin,to produce tumors even though sodium saccharin by itself atthose doses does not produce an increased risk of cancer.UlcÃ©rationproduces an enormous burst in proliferation over a1-week period in the animal's life (110). This provides enough

new cell divisions to produce a small but significant increase inthe number of cells reaching the intermediate population, ineffect an initiating action (59). If these intermediate cells arenot further stimulated to continue to proliferate rapidly, therewill not be an adequate number of further intermediate celldivisions to significantly affect the likelihood that the secondgenetic event may lead to cancer. However, if they are continuously stimulated to divide rapidly, such as following administration of sodium saccharin at high doses in the diet, the chanceof a malignant cell being produced rises with a correspondingincreased risk of cancer development.

What Is a Carcinogen?

Forty years ago and more, it was relatively easy to define achemical as a carcinogen. A carcinogen was simply a chemicalthat increased the incidence of cancer in a given tissue in agiven period of time in a bioassay. However, during the past 4decades, it has become increasingly difficult to live with thisbioassay driven definition. There are too many qualifiers nowto continue to accept this simple definition. It is still easy tovisualize 2-AAF or numerous other aromatic amines as carcin

ogens in the classical sense. At both high and low doses thesecompounds produce cancer development in rodents and humans. They fit our classical picture of what a carcinogenicchemical should be and are congruent with the astute clinicalobservations by Hill on snuff, by Pott on soot, and by Rehn onthe development of bladder cancer in dye workers in Germany(1).

However, this definition falls far short in delineating thecancer risk resulting from chemicals such as melamine, uracil,sodium saccharin, vitamin C, phÃ©nobarbital,TCDD, or phorbolesters. These chemicals produce cancer only under certaincircumstances; the circumstances may be dose related, speciesspecific, and/or mechanistically related.

The issue has become even more complex with the availabilityof newer molecular technologies such as transgenic mice (111).The rodents used in 2-year bioassays have generally been bredfor susceptibility to cancer to increase sensitivity. Mice are nowavailable with certain genes inserted into their genome to produce an increased risk of cancer development, which can beaccelerated even more if specific types of chemicals are administered. If a chemical causes cancer in a transgenic mouse, is itto be considered a carcinogen? The carcinogenicity of a chemical must take into consideration the host being exposed.

Implication for Risk Assessment

We suggest that there be changes in both the conceptualization of carcinogenesis and in the regulatory approach to chemicals. Of greatest importance is the realization that genotoxicchemicals are different from chemicals that are nongenotoxic(25). The difference is particularly pertinent in assessing thepotential for producing cancer in humans. Utilizing moderntoxicological and biological methodologies, considerable mechanistic information can be obtained to provide a rational approach to assessment of cancer risk.

Can we definitively divide chemicals into genotoxic and nongenotoxic categories? At the present time this is difficult, but itis usually possible and our methods are improving (40, 41).There are a number of short-term screens that give us predictiveinformation regarding the potential mutagenic capabilities ofchemicals. In these short-term screens it is important that doseeffects and other idiosyncrasies be taken into account, so thatthey are not interpreted in a vacuum. Ultimately, there must bean assessment of mutagenic effect in whole organisms andultimately in humans. The screens that are available today aregenerally adequate for defining chemicals as either genotoxicor nongenotoxic, especially if chemical and metabolic information is incorporated into the evaluation.

For genotoxic compounds, the presumption must be madethat there is no threshold effect. A bioassay is still required,however, to ascertain whether a mutagenic chemical actuallycauses cancer in a whole organism. In interpreting the doseresponse, however, it is mandatory to take into separate accountthe effects of the interaction of the chemical with DNA, theeffects of the chemical on cell proliferation, and any pharma-cokinetic differences between species. These dose effects can beascertained to some degree for many of these chemicals utilizingtechnologies which are available today, such as 32P-postlabeling(112) and pharmacokinetic measurements (31). Cell prolifera-tive effects can be ascertained by close examination of short-term studies for proliferative effects in the target tissue. Withthe development of new immunohistochemical markers of celldivision, such as Ki-67 (113) and proliferating cell nuclear

6502




antigen (114), as well as others, we should be able to actuallymeasure proliferative rates in tissues already being utilized foracute or subacute toxicity studies. Ultimately it would be bestto include cell proliferative studies in the testing and development phase for new chemicals, but this is not always practical.Also, time effects are critical in addition to dose effects for ourunderstanding of carcinogenesis. Obviously, the more information available the more biologically based the extrapolationcan be. If detailed dose response information is lacking, conservative approaches can be implemented, accounting forknown genotoxicity, chemistry, and metabolism of the chemicaland also incorporating information from other known relatedchemicals.

For nongenotoxic compounds, some investigation of whetheror not the chemical acts through a specific cell receptor ishelpful in delineating potential mechanisms and in ascertainingdosage levels that might pose a risk for humans. A variety ofmechanisms have already been investigated (e.g., a2u-globulinin the male rat kidney, calculi in the bladder, peroxisomeproliferation in the liver, and thyroxine clearance for the thyroid), and ascertainment of whether a chemical fits into analready familiar mechanistic model should be helpful. To determine thresholds, biological, biochemical, and physiologicalexperiments will be required.

It must be emphasized that for nongenotoxic compounds, allof the current mathematical-statistical models of extrapolationto low dose effects are inappropriate. The central task is notone of statistical curve fitting, but a judgmental one relating tobiological mechanism. For nongenotoxic chemicals, some estimate of a no-effect threshold should be possible.

Acknowledgments

We gratefully acknowledge comments from Richard A. Merrill,University of Virginia, and Emily M. Garland, University of NebraskaMedical Center, and the clerical assistance of Ginni Philbrick in thepreparation of this manuscript.

References

1. Ruddon, R. W. (ed.). Cancer Biology, Ed. 2. New York: Oxford UniversityPress, 1987.

2. Tomatis, L., Aitio, A., Wilbourn, J.. and Shuker, L. Human carcinogens sofar identified. Jpn. J. Cancer Res., 80: 795-807, 1989.

3. Chemicals and industrial processes associated with cancer in humans. IARCMonographs, Suppl. 1. Lyon, France: International Agency for Researchon Cancer, 1979.

4. Loeb, L. A., Ernster, V. L., Warner, K. E., Abbotts, J., and Laszio, J.Smoking and lung cancer: an overview. Cancer Res., 44: 5940-5958, 1984.

5. Upton, A. C. Radiation effects. In: H. H. Hiatt, J. D. Watson, and J. A.Winsten (eds.). Origins of Human Cancer. Cold Spring Harbor Conferenceson Cell Proliferation, pp. 477-500. Cold Spring Harbor, NY: Cold SpringHarbor Laboratory, 1977.

6. Board on Radiation Effects Research Commission on Life Sciences, National Research Council. Health Effects of Exposure to Low Levels ofIonizing Radiation. V. Washington. DC: National Academy Press, 1990.

7. Upton, A. C. The question of threshold for radiation and chemical carcinogenesis. Cancer Invest., 7: 267-276, 1989.

8. Hoel, D. G., Merrill, R. A., and Perera, F. P. (eds.). Risk Quantitation andRegulatory Policy. Banbury Report, Vol. 19. Cold Spring Harbor, NY: ColdSpring Harbor Laboratory, 1985.

9. Boveri, T. Zur Trage der entstehung maligner Tumoren. Jena, Germany:G. Fisher, 1914.

10. Pimentel, E. Oncogenes. Boca Raton, FL: CRC Press, Inc., 1986.11. Bishop, J. M. Molecular themes in oncogenesis. Cell, 64: 235-248, 1991.12. Stanbridge, E. J. Human tumor suppressor genes. Annu. Rev. Genet., 24:

615-657, 1990.13. Berenblum, I., and Shubik, P. A new quantitative approach to the study of

the stages of chemical carcinogenesis in the mouse's skin. Br. J. Cancer, I:384-391, 1947.

14. Berenblum, I. Carcinogenesis and tumor pathogenesis. Adv. Cancer Res.,2: 129-175, 1954.

15. Boutwell, R. K. Some biological aspects of skin carcinogenesis. Progr. Exp.Tumor Res., 4: 207-250, 1964.

16. Weinstein, I. B. The origins of human cancer: molecular mechanisms ofcarcingoenesis and their implications for cancer prevention and treatment.Cancer Res., 48: 4135-4143, 1988.

17. O'Connell, J. F., Rotstein, J. B., and Slaga, T. J. Tumor promotion andprogression in the mouse skin model. Banbury Rep. 25: 11-23, 1987.

18. Pilot, H. C., Grosso, L., and Dunn, T. The role of the cellular genome inthe stages of carcinogenesis. In: J. M. Bishop, J. D. Rowley, and M. Greaves(eds.), Genes and Cancer, pp. 81-98. New York: Alan R. Liss, Inc., 1984.

19. k muIsoii, A. G. Mutation and cancer: statistical study of retinoblastoma.Proc. Nati. Acad. Sci. USA, 68: 820-823. 1971.

20. Knudson, A. G. Two-event carcinogenesis: roles of oncogenes and antion-cogenes. In: S. H. Moolgavkar (ed.), Scientific Issues in Quantitative CancerRisk Assessment, pp. 32-48. Boston: BirkhÃ¤user,1990.

21. Gallic, B. L., Squire, J. A., Goddard, A., Dunn, J. M., Canton, M., Hinton.D., Zhu, X., and Phillips, R. A. Biology of disease: mechanism of oncogenesis in retinoblastoma. Lab. Invest., 62: 394-408, 1990.

22. Mendelsohn, M. L. The somatic mutational component of human carcinogenesis. In: S. H. Moolgavkar (ed.), Scientific Issues in Quantitative CancerRisk Assessment, pp. 22-31. New York: Birkhaeuser, Boston, Inc., 1990.

23. Lewin, B. The apparatus for DNA replication. In: Genes III, pp. 312-334.New York: John Wiley & Sons, Inc., 1987.

24. Greenfield, R. E., Ellwein, L. B., and Cohen, S. M. A general probabilisticmodel of carcinogenesis: analysis of experimental urinary bladder cancer.Carcinogenesis (Lond.), 5: 437-445, 1984.

25. Cohen, S. M., and Ellwein, L. B. Cell proliferation in carcinogenesis. Science(Washington, DC), 249: 1007-1011.

26. Moolgavkar, S. H., and Knudson, A. G., Jr. Mutation and cancer: a modelfor human carcinogenesis. J. Nati. Cancer Inst., 66: 1037-1052, 1981.

27. Moolgavkar, S. H., Luebeck, G., and deGunst, M. Two mutation model forcarcinogenesis: relative roles of somatic mutations and cell proliferation indetermining risk. In: S. H. Moolgavkar (ed.), Scientific Issues in Quantitative Cancer Risk Assessment, pp. 136-152. New York: Birkhaeuser Boston,Inc., 1990.

28. Vogelstein, B., Fearon, E. R., Hamilton, S. R., Kern, S. E. Preisinger, A.C., Leppert, M. Nakumura, Y., White, R., Smits, A. M. M., and Bos, J. L.Genetic alterations during colorectal-tumor development. N. Engl. J. Med.,319: 525-532, 1988.

29. Fearon. E. R., and Vogelstein, B. A genetic model for colorectal tumorigen-esis. Cell, 61: 759-767, 1990.

30. Wright, N., and Alison, M. (eds.). The Biology and Epithelial Cell Populations, Vols. 1 and 2. London: Clarendon Press, 1984.

31. Reitz, R. H., Mendrala, A. L., Park, C. N., Andersen, M. E., and Guenger-ich, F. P. Incorporation of in vitro enzyme data into the physiologically-based pharmacokinetic (PB-PK) model for mÃ©thylÃ¨nechloride: implicationsfor risk assessment. Toxicol. Lett., 43: 97-116, 1988.

32. Harris, C. C. Interindividual variation in human chemical carcinogenesis:implications for risk assessment. In: S. H. Moolgavkar (ed.), ScientificIssues in Quantitative Cancer Risk Assessment, pp. 235-251. BirkhaeuserBoston. Inc., 1990.

33. Staffa, J. A., and Mehlman. M. A. (eds.). Innovations in Cancer RiskAssessment (ED0, study). J. Environ. Pathol. Toxicol.. 3: 1-246, 1980.

34. Cohen, S. M., and Ellwein, L. B. Proliferative and genotoxic cellular effectsin 2-acetylaminofluorene bladder and liver carcinogenesis: biological modeling of the EDo, study. Toxicol. Appi. Pharmacol., 104: 79-93, 1990.

35. Beland, F. A., Fullerton, N. F., Kinouchi, T., and Poirier. M. C. DNAadduci formation during continuous feeding of 2-acetyl-aminofluorene atmultiple concentrations. IARC Sci. Pubi. 89: 175-180, 1988.

36. O'Neill, I. K., Von Borstel, R. C., Miller, C. T., Long, J., and Bartsch, H.A'-Nitroso compounds: occurrence, biological effects and relevance to humancancer. IARC Sci. Pubi. 57:627-665, 1983.

37. Boucheron, J. A., Richardson, F. C., Morgan, P. H., and Swenberg, J. A.Molecular dosimetry of O4-ethyldeoxythymidine in rats continuously exposed to diethylnitrosamine. Cancer Res., 47: 1577-1581, 1987.

38. Deal, F. H., Richardson, F. C., and Swenberg, J. A. Dose response ofhepatocyte replication in rats following continuous exposure to diethylnitrosamine. Cancer Res., 49: 6985-6988, 1989.

39. Pilot, H. C., Neveu, M. J.. Hully, J. R., Rizvi, T. A., and Campbell, H.Multistage carcinogenesis in the rat as a basis for models of risk assessmentof carcinogenesis. In: S. H. Moolgavkar (ed.). Scientific Issues in Quantitative Cancer Risk Assessment, pp. 69-95. New York: Birkhaeuser Boston,1990.

40. Butterworth, B. E., and Slaga, T. J. (eds.). Nongenotoxic Mechanisms inCarcinogenesis, Banbuary Report 25. Cold Spring Harbor, NY: Cold SpringHarbor Laboratory, 1987.

41. Ashby, J., and Tennant, R. W. Definitive relationships among chemicalstructure, carcinogenecity and mutagenicity for 301 chemicals tested by theU.S. NTP. MutÃ¢t.Res., 257: 229-306, 1991.

42. Brand, K. G. Solid state carcinogenesis. Banbury Rep. 25: 205-213, 1987.43. Clayson, D. B., and Cooper, E. H. Cancer of the urinary tract. Adv. Cancer

Res., 13: 271-381, 1970.44. Bryan, G. T. Pellet implantation studies of carcinogenic compounds. J.

Nail. Cancer Inst., 43: 255-261, 1969.45. DeSesso, I. M. Confounding factors in direct bladder exposure studies.

Comments Toxicol, 3: 317-334, 1989.46. Jull, J. W. The effect of time on the incidence of carcinomas obtained by

6503



GENETIC ERRORS, CELL PROLIFERATION. AND CARCINOGENESIS

the implantation of paraffin wax pellets into mouse bladder. Cancer Lett..6:21-25, 1979.

47. Clayson, D. B. Bladder carcinogenesis in rats and mice: possibility ofartifacts. J. Nati. Cancer Inst., 52: 1685-1689, 1974.

48. Frith, C. H., West, R. W., Stanley, J. W., and Jackson, C. D. Urotheliallesions in mice given 4-ethylsulfonylnaphthalene-l-sulfonamide. acetazol-amide and oxamide. J. Environ. Pathol. Toxicol.. 5: 25-38, 1984.

49. Cheng. L. Urinary tract calculi in man and laboratory animals: incidence,composition, and etiology. J. Environ. Pathol. Toxicol., 4: 317-349, 1980.

50. Shirai T., Ikawa, E., Fukushima, S., Masui, T., and Ito, N. Uracil-inducedurolithiasis and the development of reversible papillomatosis in the urinarybladder of F344 rats. Cancer Res., 46: 2062-2067, 1986.

51. Shirai, T., Fukushima, S., Tagawa, Y., Okumura, M., and Ito, N. Cellproliferation induced by uracilâ€”calculi and subsequent development ofreversible papillomatosis in the rat urinary bladder. Cancer Res., 49: 378-383, 1989.

52. Sakata, T., Masui, T., St. John, M., and Cohen, S. M. Uracil-induced calculiand proliferative lesions of the mouse urinary bladder. Carcinogenesis(Lond.), 9: 1271-1276. 1988.

53. Melnick, R. L., Boorman, G. A., Haseman, J. K., Montali, R. J., and Huff,J. Urolithiasis and bladder carcinogenicity of melamine in rodents. Toxicol.Appi. Pharmacol., 72: 292-303. 1984.

54. National Toxicology Program. Carcinogenesis bioassay of melamine inF344/N rats and B6C3F, Mice. NTP Report on Melamine, Vol. 245.Research Triangle Park, NC: NIH, U.S. Dept. of Health and HumanServices, 1983.

55. Melamine: Toxic chemical release reporting. Fed. Reg.. 53: 23128-23132,1988.

56. Ellwein, L. B., and Cohen, S. M. The health risks of saccharin revisited.Toxicology, 20: 311-326, 1990.

57. Cohen, S. M., Ellwein, L. B., Okamura, T., Masui, T., Johansson, S. L.,Smith, R. A., Wehner, J. M., Khachab, M., Chappel, C. I., Schoenig, G. P.,Emerson, J. L., and Garland, E. M. Comparative bladder tumor promotingactivity of sodium saccharin, sodium ascorbate, related acids, and calciumsalts in rats. Cancer Res., 51: 1766-1777, 1991.

58. Armstrong. B. K. Interpretation of negative epidemiological evidence forcarcinogenicity. IARC Sci. Pubi., 65: 129-143, 1985.

59. Ellwein, L. B., and Cohen, S. M. A cellular dynamics model of experimentalbladder cancer: analysis of the effect of sodium saccharin in the rat. RiskAnal., A: 215-221, 1988.

60. Cohen, S. M., Cano. M., Earl, R. A., Carson, S. D., and Garland, E. M. Aproposed role for silicates and protein in the proliferative effects of saccharinon the male rat urothelium. Carcinogenesis (Lond.), 12: 1551-1555. 1991.

61. Olson, M. J., Johnson, J. T., and Reidy, C. A. A comparison of male ratand human urinary proteins: implications for human resistance to hyalinedroplet nephropathy. Toxicol. Appi. Pharmacol., 102: 524-536, 1990.

62. Auerbach. O.. and Garfinkel, L. Histologie changes in the urinary bladderin relation to cigarette smoking and use of artificial sweeteners. Cancer(Phila.), 64: 983-987, 1989.

63. Van Ryzin, J. The assessment of low-dose carcinogenicity: discussion.Biometrics (Suppl.), 38: 130-139, 1982.

64. Lutz. W. K. Dose-response relationship and low-dose extrapolation inchemical carcinogenesis. Carcinogenesis (Lond.), //: 1243-1247, 1990.

65. Carlborg, F. W. A cancer risk assessment for saccharin. Food Chem.Toxicol., 23:499-506, 1985.

66. Fukushima, S., Imaida, K.. Sakata, T.. Okamura, T., Shibata, M., and Ito,N. Promoting effects of sodium L-ascorbate on 2-stage urinary bladdercarcinogenesis in rats. Cancer Res., 43: 4454-4457, 1983.

67. Fukushima. S., Shibata, M., Shirai, T., TamaÃ±o,S., and Ito, N. Roles ofurinary sodium ion concentration and pH in promotion of ascorbic acid ofurinary bladder carcinogenesis in rats. Cancer Res., 46: 1623-1626, 1986.

68. Ito, N., and Fukushima. S. Promotion of urinary bladder carcinogenesis inexperimental animals. Exp. Pathol., 36: 1-15, 1989.

69. Fukushima, S.. Imaida, K., Shibata, M-A., TamaÃ±o,S., Kurata, Y., andShirai, T. L-Ascorbic acid amplification of second-stage bladder carcinogenesis promotion by NaHCOj. Cancer Res., 48: 6317-6320, 1988.

70. West, R. C., Astle, M. J., and Beyer, W. H. Recommended daily dietaryallowances. In: CRC Handbook of Chemistry and Physics, p. D-279. BocaRaton, FL: CRC Press, Inc., 1986.

71. Cohen, S. M.. Purtilo, D. T., and Ellwein, L. B. Pivotal role of increasedcell proliferation in human carcinogenesis. Modern Pathol., 4: 371-382,1991.

72. Swenberg, J. A., Short, B.. Borghoff, S., Strasser. J., and Charbonneau, M.The comparative pathobiology of Â«Ju-globulinnephropathy. Toxicol. Appi.Pharmacol., 97: 35-46, 1989.

73. Ito, N., and Hirose, M. Antioxidantsâ€”carcinogenic and chemopreventiveproperties. Adv. Cancer Res., S3: 247-302. 1990.

74. Clayson, D. B.. Iverson, F., Nera, A., and Lok, E. The significance ofinduced forestomach tumors. Annu. Rev. Pharmacol. Toxicol. JO: 441-463.1990.

75. Hill, R. N., Erdreich, L. S., Paynter, O., Roberts, P. A., Rosenthal, S. L.,and Wilkinson, C. F. Thyroid follicular cell carcinogenesis. Fundam. Appi.Toxicol., 72:629-697, 1989.

76. McClain, R. M., Posch, R. C., Bosakowski, T.. and Armstrong, J. M.Studies on the mode of action for thyroid gland tumor promotion in rats byphÃ©nobarbital.Toxicol. Appi. Pharmacol., 94: 254-265, 1988.

77. Ward, J. M. Increased susceptibility of livers of aged F344/NCr rats to the

effects of phÃ©nobarbitalon the incidence, morphology, and histochemistryof hepatocellular foci and neoplasms. J. Nail. Cancer Inst., 71: 815-823,1983.

78. Peraino, C. Initiation and promotion of liver tumorigenesis. Nail. CancerInst. Monogr., 5Â«:55-61, 1981.

79. PhÃ©nobarbitaland phÃ©nobarbitalsodium. IARC Monographs on the Evaluation of the Carcinogeneic Risk of Chemicals to Man. Vol. 13, pp. 157-181. IARC Scientific Publications. Lyon, France: International Agency forResearch on Cancer, 1976.

80. Ames, B. N., and Gold. L. S. Chemical carcinogenesis: too many rodentcarcinogens. Proc. Nati. Acad. Sci. USA, 87: 7772-7776, 1990.

81. Ames, B. N., Profet, M., and Gold, L. S. Dietary pesticides (99.99% allnatural). Proc. Nati. Acad. Sci. USA, 87: 7777-7781, 1990.

82. Ames, B. N., Profet, M., and Gold, L. S. Nature's chemicals and syntheticchemicals: comparative toxicology. Proc. Nati. Acad. Sci. USA, 87: 7782-7786, 1990.

83. McConnell, E. E. The maximum tolerated dose: the debate. J. Am. Coll.Toxicol., A: 1115-1120, 1989.

84. Long-term and short-term screening assays for carcinogens: a critical appraisal. IARC Monographs on the Evaluation of the Carcinogeneic Risk ofChemicals to Humans. IARC Scientific Publications (Suppl. 2). Lyon,France: International Agency for Research on Cancer, 1980.

85. Travis, C. C, Saulsbury, A. W., and Pack, S. A. R. Prediction of cancerpotency using a battery of mutation and toxicity data. Mutagenesis, 5: 213-219. 1990.

86. Haseman, J. K, and Clark, A-M. Carcinogenicity results for 114 laboratoryanimal studies used to assess the predictivity of four in vitro genetic toxicityassays for rodent carcinogenecity. Environ. Mol. Mutagen., 16 (Suppl. 18):15-31, 1990.

87. Rosenkranz, H. S., and Klopman, G. Prediction of the carcinogenicity inrodents of chemicals currently being tested by the U.S. National ToxicologyProgram: structure-activity correlations. Mutagenesis, 5: 425-432, 1990.

88. Auletta, A. Current status of short-term tests for carcinogenecity. Environ.Carcinog. Rev., 8: 1-43, 1990.

89. Williams, G. M. Methods for evaluating chemical genotoxicity. Annu. Rev.Pharmacol. Toxicol., 29: 189-211, 1989.

90. Demotz, S., Grey, H. M., and Sette, A. The minimal number of Class IIMHC-antigen complexes needed for T-cell activation. Science (Washington,DC): 249: 1028-1030, 1990.

91. Roberts, L. Dioxin risks revisited. Science (Washington, DC), 251: 624-626, 1991.

92. Kociba. R. J., Keyes, D. G., Beyer, J. E., Carreon, R. M., Wade, C. E.,Dittenberg, D. A., Kalnins, R. P., Frauson, L. E., Park, C. N., Barnard, S.D., Hummel, R. A., and Humiston, C. G. Results of a two-year chronictoxicity and oncogenicity study of 2,3,7,8-tetrachlorodibenzo-p-dioxin inrats. Toxicol. Appi. Pharmacol., 46: 279-303, 1978.

93. National Toxicology Program. Carcinogenesis bioassay of 2,3,7,8-tetra-chlorodibenzo-p-dioxin in Osborne-Mendel rats and B6C3F, mice (gavegestudy). Technical Report Series, No. 209, Research Triangle Park, NC,1982.

94. Delia Porta, G., Dragani, T. A., and Sozgi, G. Carcinogenic effects ofinfantile and long-term 2,3,7,8-tetrachlorodibenzo-p-dioxin treatment in themouse. Tumori, 73: 99-107, 1987.

95. Fingerhut, M. A., Halperin, W. E., Marlow, D. A., Piacitela, L. A.,Honchar, P. A.. Sweeney, M. H., Greife, A. L., Dill, P. A., Steenland, K.,and Suruda, A. J. Cancer mortality in workers exposed to 2,3,7,8-tetra-chlorodibenzo-p-dioxin. N. Engl. J. Med., 324: 212-218, 1991.

96. Ashby, J., Johannsen, F. R., Raabe, G. K., Doerrer, N. G., Lewis, S. C.,Reynolds, R. C., Flamm, W. G., Krivanek, N. D., Smith, J. M., Harris, J.E., McCarthy, J. F., Stevens, J. T., Teta, M. J., Hughes, D. H., Moolenaar,R. J., and Wilson, J. D. A scheme for classifying carcinogens. Reg. Toxicol.Pharmacol., 12: 270-295, 1990.

97. Perera, F. P. Carcinogens and human health: Part 1. Science (Washington,DC), 250: 1644-1646, 1990.

98. Rail. D. P. Carcinogens and human health: Part 2. Science (Washington,DC), 251: 10-13, 1991.

99. Weinstein, I. B. Mitogenesis is only one factor in carcinogenesis. Science(Washington, DC), 25/: 387-388, 1991.

100. Hoel, D. G., Haseman, J. K., Hogan, M. D., Huff, J., and McConnell, E.E. The impact of toxicity on carcinogenicity studies: implications for riskassessment. Carcinogenesis (Lond.), 9: 2045-2052, 1988.

101. National Toxicology Program. Monuron. NTP Technical Report Series,No. 266. Research Triangle Park, NC, 1988.

102. Rubin, E., and Farber, J. L. (eds.). The gastrointetinal tract. In: Pathology,pp. 686-714. Philadelphia: J. B. Lippincott Co., 1988.

103. Marsman, D. S.. Cattley. R. C., Conway, J. G., and Popp, J. A. Relationshipof hepatic peroxisome proliferation and replicative DNA synthesis to thehepatocarcinogenicity of the peroxisome proliferators di(2-ethyl-hexyl)phthalate and [4-chloro-6-(2.3-xylidino)-2-pyrimidinylthio] acetic acid(Wy-14,643) in rats. Cancer Res., 48: 6739-6744, 1988.

104. Kaufmann, W. K., Mackenzie, S. A., and Kaufmann, D. G. Quantitativerelationship between hepatocyte neoplasms and islands of cellular alterations during hepatocarcinogenesis in the male F344 rat. Am. J. Pathol.,119: 171-174, 1988.

105. Galen, R. S., and Gambino, S. R. Beyond Normality: The Predictive Valueand Efficiency of Medical Diagnoses, pp. 8-14. New York: John Wiley &Sons, Inc., 1975.

6504




106. Wogan, G. N. Aflatoxin carcinogenesis. Methods Cancer Res.. 7: 309-344, 110. Murasaki. G., and Cohen, S. M. Effect of sodium saccharin on urinary1973. bladder epithelial regenerative hyperplasia following freeze ulcÃ©ration.Can-

107. Cohen, S. M., Murasaki. G., Fukushima, S., and Greenfield, R. E. Effects cer Res., 43: 182-187, 1983.of regenerative hyperplasia on the urinary bladder: carcinogenicity of sodium 111. Hanahan, D. Transgenic mice as probes into complex systems. Sciencesaccharin and /V-[4-<5-nitro-2-furyl)-2-thiazolyl]formamide. Cancer Res., 42: (Washington, DC), 246: 1265-1275, 1989.65-71, 1982. 112. Watson, W. P. "P-Post-radiolabelling for detecting DNA damage. Muta-

108. Hasegawa, R., Greenfield, R. E., Murasaki, G.. Suzuki, T., and Cohen, S. genesis, 2: 319-331, 1987.M. Initiation of urinary bladder carcinogenesis in rats by freeze ulcÃ©ration 113. Gerdes, J., Lemke, H., Baisch, H., Wacker, H. H., Schwab, U., and Stein,with sodium saccharin promotion. Cancer Res., 45: 1469-1473, 1985. H. Cell cycle analysis of a cell proliferation associated human nuclear antigen

109. Hasegawa, R., St. John, M. K., Cano, M., Issenberg, P., Klein. D. A., defined by the monoclonal antibody Ki-67. J. Immunol.. 113: 1710-1715,Walker, B. A., Jones, J. W., Schnell, R. C., Merrick, B. A., Davies, M. H., 1988.McMillan, D. T., and Cohen, S. M. Bladder freeze ulcÃ©rationand sodium 114. van Dierendonck, J. H., Wijsman, J. H., Keijzer, R.. van de Velde, C. J.saccharin feeding in the rat: examination for urinary nitrosamines, mutagens H., and Cornelisse. C. J. Cell-cycle-related staining patterns of anti-prolif-and bacteria and effect on hepatic microsomal enzymes. Food Chem. Tox- crating cell nuclear antigen monoclonal antibodies. Am. J. Pathol.. 138:icol., 12: 935-942. 1984. 1165-1172, 1991.

6505



1991;51:6493-6505. Cancer Res Samuel M. Cohen and Leon B. Ellwein Genetic Errors, Cell Proliferation, and Carcinogenesis

Updated version

http://cancerres.aacrjournals.org/content/51/24/6493.citation

Access the most recent version of this article at:

E-mail alerts related to this article or journal.Sign up to receive free email-alerts

Subscriptions

Reprints and

[email protected] at

To order reprints of this article or to subscribe to the journal, contact the AACR Publications

Permissions

Rightslink site. Click on "Request Permissions" which will take you to the Copyright Clearance Center's (CCC)

.http://cancerres.aacrjournals.org/content/51/24/6493.citationTo request permission to re-use all or part of this article, use this link



http://cancerres.aacrjournals.org/cgi/alerts

mailto:[email protected]



cancer research - genetic errors, cell proliferation, and ......genetic errors, cell proliferation,...

Documents