Document Lg4LezEE98xJRqvYpvwvwRBVg

Quantitative Aspects of Risk Assessment in Chemical Carcinogenesis Arch. Toxicol., Suppl. 3, 79--94 (1980) by Springer-Verlag 1980 Copyrighted Material Do Not Reproduce Without Publisher's Permission Non-Linear Pharmacokinetic Parameters Need to be Considered in High Dose/Low Dose Extrapolation R. H. Reitz, J. F. Quast, A. M. Schumann, P. G. Watanabe, and P. J. Gehring Health and Environmental Sciences, USA, Dow Chemical USA, 1803 Building, Midland, Michigan 48640 USA A. Introduction Numerous recent studies, both with short-term in vitro tests and longterm animal bioassays, have led to the designation of many materials encountered in our daily lives as potential carcinogens. However, the classification of a material as a potential carcinogen does not necessarily mean that this substance, at the concentrations present in the environment, is actually a danger to human health. In order to conclude that a po tential carcinogen does constitute a public health hazard, we must first perform a quantitative risk estimation for that material In doing so, we must remember that any estimation process will always indicate some finite degree ofrisk, no matter how small. Thus, our aim must be to identify acceptable levels ofrisk, rather than to attempt to er roneously eliminate all risk. Only with realistic risk estimates can we properly evaluate the consequences ofreg ulatory action. For example, would it be wise to stop chlorinating drinking water to re duce cancer risk from chloroform if this action might lead to a typhus epidemic? Or, should we discontinue use of all insecticides with even the slightest suggestion of carcinogenic risk if this action would lead to crop failures and famine? In view ofthe fact that literally hundreds ofcommon industrial or natural chemicals have been indicted on some basis or another as potential carcinogens, we must insist on a careful evaluation of all the factors involved in carcinogenesis. Reasonable esti mations of carcinogenic risk to man can be obtained only when we combine carefully conducted animal bioassays with a fundamental understanding of pharmacokinetic processes and biochemical mechanisms of tumorigenesis. 00 fi C/3 rloa. -J 00 B. Significant Factors in Risk Estimation I. Mechanisms of Tumorigenesis It is becoming increasingly clear that there are at least two types of processes which can contribute to the observation of an increased incidence of tumors in an animal bioassay; genetic and epigenetic. The first of these involves reaction of the test R&S 142749 80 R. H. Reitz et ai. substance (or a metabolically activated form ofthe test substance) with the genetic ma terial ofthe cell. This genetic mechanism is illustrated in the pioneering work ofthe Mil lers (1966, 1970) showing that many potent carcinogens are electrophiles which react with sites on the bases of DNA to disturb normal base pairing. If the altered bases are not removed by DNA repair systems before replication, they may become permanent genetic changes or mutations. Support for the somatic mutation hypothesis ofchemical carcinogenesis includes the following observations: many chemical carcinogens induce mutations; the long-latent period between chemical exposure and the manifestation of cancer is consistent with effects in a macromolecule such as DNA which is capable of retaining information over a long period of time; genetic diseases with associated chromosomal abnormalities are associated often with an increased incidence of can cer; and many chemical carcinogens interact with DNA. In addition to genotoxic effects, chemicals can also produce tumors through cytotoxic effects. In this case, tissue damage occurs to such an extent as to cause cell death and subsequent cell regeneration. During chronic administration of cytotoxic agents, DNA replication is stimulated greatly in the affected tissues. Since there is a tiny but still finite chance for error in each replication cycle, the net effect is to increase the spontaneous mutation rate. Another consequence ofincreasing cell division is that the relative rates ofDNA re pair and DNA replication are altered. This may be very important because there is evi dence that the existing DNA repair systems may fail to recognize DNA alterations af ter replication. Consequently, repair becomes less effective in protecting cells from the consequences of endogenous genetic damage. This is shown by the work of Berman et al. (1978) and Weymouth and Loeb (1978) who found that increased division of normal cells increases their susceptibility to mutagenic events. This is also demon strated by the studies ofMcCormick (1979) who found that momentarily arresting cell division to give added time for DNA repair greatly diminished the mutational action of uv light. Further indication that increased cell division can be significant comes from the ob servations of Laroye (1974) who found that many cancers develop in chronically inflamed or scarred tissue (colonic cancer in patients with ulcerative colitis or squa mous cell carcinomas in ulcers of bum scars). Berenblum (1929) observed that repeated freezing of the skin with dry-ice induced tumors, and Peraino et al. (1973) re ported the enhancement of "spontaneous" tumors in mice by dietary phenobarbital, a treatment also known to stimulate DNA synthesis. Similarly, the tumor yield from a known genetic carcinogen is enhanced greatly after partial hepatectomy (Date et al., 1976) or by physical trauma at the site of initiation (Berenblum, 1944). Although the tumors that develop from each type of mechanism cannot be distinguished, there are fundamental differences in the type of safety precautions appropriate to each type of agent. Consequently, these will be discussed separately in the sections to follow. II. Factors Influencing Genetic Events Some individuals have invoked the somatic mutation theory as grounds for postulating that since a single molecule of a chemical could conceivably react with DNA to cause a R. H. Reitz et al. e) with the genetic maeering work ofthe Milctrophiles which react If the altered bases are lay become permanent hypothesis ofchemical cal carcinogens induce id the manifestation of 4A which is capable of .eases with associated ased incidence of con duce tumors through extent as to cause cell :plication is stimulated fiance for error in each us mutation rate, lative rates ofDNA re sit because there is evi:e DNA alterations a/otecting cells from the theuadc of Berman et inAHed division of s. This is also demonmentarily arresting cell re mutational action of ant comes from the obJevelop in chronically srative colitis or squa(1929) observed that 'eraino et al. (1973) reietary phenobarbital, a the tumor yield from a aatectomy (Date et al., lblum, 1944). mechanism cannot be of safety precautions discussed separately in grounds for postulating t with DNA to cause a I i Non-Linear Pharmacokinetic Parameters Need 81 mutation, there can be no threshold in carcinogenesis. Although the subject of thresh olds can never be experimentally resolved, it is clear that there are a number of factors which can dramatically affect the rate of tumor production. To illustrate why this is true, refer to Figure 1 which shows the typical chemical pro cesses that determine the fate of a chemical (C) in the body. Each arrow represents a first-order rate process. The dominant process governing the fate of the chemical may be different for different species, different chemicals, and frequently, for different doses of the same chemical. Interferring with excretion or detoxification will enhance cova lent binding of a reactive, electrophilic ("electron-seeking") metabolite (RM) to macromolecules, both genetic (CBG) and nongenetic (CBN). For example, furosemide, a diuretic, is excreted predominantly intact in the urine when low doses are given to patients. When high doses are administered, renal clearance is overwhelmed causing a disproportionate increase in the formation of toxic metabolites which react covalently with macromolecules. As another example, bromobenzene is transformed in the liver to the chemically reactive bromobenzene-3,4epoxide. This molecule is detoxified enzymatically by conjugation with glutathione. However, if large doses are administered, glutathione is depleted and the reactive metabolite reacts instead with macromolecules. Actual metabolic reactions are often more complex than the foregoing illustrations. For many chemicals, absorption, distribution, biotransformation, and excretion occur via active transport or enzymatic reactions. Rather than being best portrayed by first-order or linear rate processes in ac cordance with the equation at these reactions are described more accurately by Michaelis-Menten kinetics or dosedependent kinetics in accordance with the equation ac vmc at ~ Km + c' This suggests that at some-level these reactions may "saturate". iC. BN Activation (a) RM- -CBG RBPai,(f) PCBGR Excretion (e) Detoxifi^ation (d) Replication (k,} Ce IM RC8G c- RM C Ce IM B CBN B CBG ac CBGR * RCBG Chemical Reactive Metabolite Excreted Chemical Inactive Metabolite Covalent Binding, Nongenetic Covalent Binding, Genetic Repaired Covalently Bound Genetic Material Retained Genetic Program, Critical & Noncritical Fig. 1. Chemical processes determining the fate of a chemical (C) in the body R&S 142750 R&S 142751 82 R. H. Reitz et al. It is likely that saturation of detoxification is the rule rather than the exception for chemicals possessing a low order of toxicity, because high doses can be administered without killing the animal. The implication ofsaturating protective mechanisms such as DNA repair can be illustrated by modelling the processes shown in Figure 1. In this fig ure the processes referred to previously (absorption, excretion, metabolic activation) are considered to occur in accordance with Michaelis-Menten kinetics. Reaction ofthe reactive metabolite (RM) with macromolecules some of which are nongenetic (CBN) and genetic (CBG) is perceived to occur as an apparent first-order rate process. Replication of genetic material (CBG) having bound covalently to it the reactive metabolite is represented, naively, as an apparent first-order rate process. This process gives rise to replicated genetic material possessing a finite probability, albeit very small, of programming the cell to become cancerous. Repair of genetic material having reactive metabolite bound covalently to it does occur and there exists evidence that the repair process can be overwhelmed (Pegg et al., 1976). Hence this process is envisioned to occur in accordance with MichaelisMenten kinetics. Using the model in Figure 1 as discussed above, a series of differential equations were ascribed to the model and realistic values for the various constants selected (Gehring and Blau, 1977). Using different doses (Co), the equations were numerically integrated using a computer. As a function of dose, values were determined for nongenetic material bound covalently with reactive metabolite (CBN/Co), genetic material bound covalently with reactive metabolite and repaired (CBGR/Co), and ge netic material bound covalently with reactive metabolite but not repaired (RCBG/Co). The values were normalized for dose. The results are shown, in Figure 2. Figure 2 shows that with increasing doses there occurs a disproportionate increase in the amount of reactive metabolite bound to genetic material which escapes repair (RCBG/C,,). Since formation of such material has been associated with chemical carcinogenesis (Goth and Rajewsky, 1974), this highly conservative model illustrates why excessive doses of some chemicals cause discernible increases in cancer while smaller doses may be without effect It also illustrates clearly why data generated from such studies cannot be extrapolated as a simple linear function of dose to predict the in cidence of cancer which may be incurred by exposure to smaller amounts of a chemical. Data developed for two potent carcinogens, N-ethyl-N-nitrosourea (ENU) and dimethylnitrosamine (DMN) demonstrate the appropriateness of certain aspects of the model, specifically the existence of a saturable repair mechanism. Administration of rather high doses of ENU to newborn rats results in a high incidence of malignant tu mors of the central and the peripheral nervous systems. At such dose levels, liver tumors do not develop. Goth and Rajewsky (1974) correlated these events with ethylation of the 0--6 position of guanine in DNA, (i.e., alkylation of DNA bases, as discussed earlier). Excision repair by enzymatic removal ofthe ethylated guanine from DNA occurs readily in liver but only poorly in nervous tissue. It has been shown by Gerchman and Ludlum (1973) that 0--6 alkylation of guanine results in misincorporation ofbases in a bacterial nucleic acid polymerase system. Thus, in nervous tissue the 0--6-ethylguanine in DNA has a greater opportunity to miscode DNA replicated for daughter cells. The ethylated guanine may pair with thymine rather than cytosine. As a consequence, cancer occurs in nervous tissue but not in liver. R. H. Reitz et al. ir than the exception for ses can be administered live mechanisms such as /n in Figure 1. In this fig>n, metabolic activation) kinetics. Reaction of the :h are nongenetic (CBN) first-order rate process, lently to it the reactive ate process. This process lability, albeit very small, und covalently to it does tverwhelmed (Pegg et al., irdance with Michaelis- of differential equations rious constants selected rations were numerically les were determined for >olite (CBN/Co), genetic red (CBGR/Co), and geaot r^red (RCBG/Co). Jwi^^Figure 2. lisproportionate increase rial which escapes repair asodated with chemical ervative model illustrates ncreases in cancer while why data generated from a ofdose to predict the ino smaller amounts of a itrosourea (ENU) and dis of certain aspects of the anism. Administration of ncidence of malignant tu\t such dose levels, liver elated these events with /lation of DNA bases, as re ethylated guanine from ue. It has been shown by mine results in misincorm. Thus, in nervous tissue miscode DNA replicated nine rather than cytosine. iot in liver. Non-Linear Pharmacokinetic Parameters Need 83 Fig. 2. Effect of increasing initial dose (CJ upon fraction of chemical bound to nongenetic macromolecules (CBN), fraction bound to genetic macromolecules and subsequently repaired (CBCR), and fraction bound to genetic macromolecules and fixed by replication (RCBG) DMN given as a large single dose produces kidney tumors in rats, but does not pro duce liver cancer. Recently, Pegg et al. (1976, 1977) have demonstrated that excision repair ofDNA having guanine methylated in the 0--6 position occurs readily in the liv er of rats given both a high and low dose of DMN. In the kidney; however, repair of DNA having 0--6 methylated guanine occurs readily only after administration oflow, noncarcinogenic doses, Figure 3. Thus, we must conclude that alkylation of DNA may lead to cancer induction. Further, we must conclude that administration ofhigh doses inhibits or exceeds the ca pacity ofthe repair mechanism. Indeed, the data presented on ENU and DMN indicate that cancer occurs if the target tissue has a poor capacity for repair of DNA of if the doses are of sufficient magnitude to inhibit repair or exceed the capacity of the repair process. Another important consideration is the role of the various biotransformation reactions in the expression of the toxicity of any given chemical. For example, it is clear that for many ofthe small halogenated hydrocarbons, meta bolic activation is required for toxicity. This may be indicated in a variety of ways; for example, by finding increased toxicity after induction of mixed function oxidases (MFO) or decreased toxicity after inhibition ofMFO (Ilett et al., 1973). Alternatively, a requirement for S-9 liver extract in the Ames bacterial mutagenicity test indicates that metabolic activation is occurring. R&S 142752 84 R. H. Reitz et aL This information is very important when attempting extrapolation of results from animal tests to man. As pointed out by Rail (1969), it is often possible to estimate roughly the relative sensitivity of several species to a material by using the approxima tion that the basal metabolic rate is roughly proportional to the body surface area. What this means is that, ifwe neglect factors other than metabolism, a large animal spe cies is more sensitive than a small one to a directiy toxic agent. Conversely, a large ani mal species is less sensitive than a small one to toxicity mediated through a reactive me tabolite (Reitz et al., 1978). The contribution from oxidative metabolism to relative risk factors in various species is summarized in Table 1. Application of this rule to a series of six halogenated hydrocarbons giving a positive bioassay in at least one species at the National Cancer Institute of the United States correctly predicts the most sensitive species in every case (Table 2). However, it must be emphasized that for some types of metabolic activation the relative rates may R&S 142753 Fig. 3. DNA repair in liver and kidney of rats after 20 mg/kg or 2.5 mg/kg of DMN Table 1. Predicted relative cancer risk from equivalent doses (mg/kg) calculated on the basis of (body weight)10* A Directly toxic agents B Metabolicaily activated agents Man (70 kg) Dog (20 kg) Rabbit (3 kg) Rat (0.5 kg) Mouse (0.03 kg) 1.00 0.66 0.35 0.18 0.08 1.00 1.52 2.85 5.58 13.2 After Rail (1969) R. H. Reitz et aL ipolation of results from ten possible to estimate by using the approxima3 the body surface area, olism, a large animal spe. Conversely, a large aniid through a reactive meletabolism to relative risk hydrocarbons giving a er Institute of the United .se (Table 2). However, it on the relative rates may e V ykg j mg/kg 72 96 our) : mg/kg of DMN Iculated on the basis of (body etabolically activated agents .00 .52 :.85 .58 :,2 Non-Linear Pharmacokinetic Parameters Need 85 not be a simple function of body size. For example, the metabolic activation of 2acetaminofluorene involves formation of an active sulfate (Miller, 1978), and the species and organ sensitivity of the tumorigenicity of the agent correlate well with the level of sulfatransferase enzyme. However, in this case, the rat has a higher level ofsulfatransferase than the mouse and also develops more tumors when exposed to an equiv alent dose of 2-acetaminofluorene. Hence the reliability of interspecies extrapolation depends to a large degree upon how much is known of the details of metabolism. III. Integration of Concepts for Risk Extrapolation As a practical example ofthe application of these concepts, a risk extrapolation will be performed for vinyl chloride (VC). As with many other chemicals, the extrapolation will start with the results of a long-term animal bioassay; in this case the study ofMaltoni and Lefemine (1975). In this experiment, rats were exposed to different con centrations of VC (50 to 10,000 ppm) 4 h/day, 5 days/week, for 12 months and subsequently held for observation until death. The tumors observed in these animals are reported in Table 3, column 3. Before analysis of these data, two important points must be considered: 1) Numerous studies in rats indicate that it is not VC per se which is responsible for production of angiosarcoma but rather a reactive metabolite formed from it in the body (Bartsch et al., 1975; Barbin et al, 1975; Kappus et at, 1976; Malavielle et al., 1975; BoitetaL, 1975; Watanabeetal., 1978). Consequently, the risk of developing a tumor from VC will not be related to the exposure concentration directly, but rather to the rate of metabolism of VC. Table 2. Carcinogen bioassay results (National Cancer Institute, USA) for a series of materials in which metabolic activation is thought to precede toxicity Perchloroethylene Trichloroethylene Allyl chloride Trichloroethane (1,1,2) Carbon tetrachloride Chloroform Mice + + + 4* + 4**4- Rats _ -- + 4* Table 3. Predicted incidence ofangiosarcoma in rats exposed to vinyl chloride using various models versus dose (v, pg vc metabolized per day) compared to experimental results Exposure ppm Dose v, ng/day Experimental % Predicted % Model A Model B Model C Model D 10,000 6,000 2,500 500 250 50 5,521 5,403 5,030 3,413 2,435 739 14.8 (9/61) 21.7 (13/60) 22.0 (13/59) 11.9 (7/59) 6.8 (4/59) 1.7 (1/59) 19.8 19.3 17.9 12.1 8.1 1.4 20.0 20.5 18.1 11.8 8.0 1.4 19.7 19.3 17.9 12.2 8.7 2.6 18.9 18.6 17.4 12.2 8.8 2.8 R&S 142754 86 R. H. Reitz et al. 2) Biotransformation of VC in rats is a nonlinear process occurring in accordance with Michaelis-Menten kinetics: vms v --------------Kffl + S (Watanabe et al., 1976a~c; Bolt et al., 1976). In this equation v and Vm are velocity and maximum velocity, respectively, for the biotransformation of VC expressed as pg equivalents VC metabolized daily. S and Km are the concentration of VC being inhaled and the Michaelis constant expressed as pg VC/1 air, respectively. To extrapolate the risk to man, the first step is to empirically fit a model relating the quantity of VC metabolized and the observed tumor incidence. There are a number of possible models which could be used for this. Four of the most commonly employed were selected for evaluation. These are: A) Probit % = a + b(log v) = - 1.625 + 1.543(logv). B) Linear % = a + b(v) = - 1.48 + 0.389 x 1<T2(V). C) Linear Forced through Origin % * bv = 0.3565 x 10"2(v). D) One-hit Model (CAG)* % = [1 - exp-*)] x 100 * [1 - exp(-0.38 x 10"4 x v)] x 100 . The constants given for the preceding models were determined by linear regression analysis and represent the best fit ofthe data for the incidence ofhepatic angiosarcoma versus the amount of VC metabolized daily in rats exposed to different concentrations of VC (Gehring et al., 1978). The data used to derive the constants are given in Table 3. Using the equations de rived for the models, the predicted incidence of angiosarcoma for the groups ofrats ex posed to different concentrations of VC were calculated. These results are shown also in Table 3. All four models overpredict the incidence of hepatic angiosarcoma in rats exposed to 10,000 ppm VC. This is as expected, because this exposure resulted in ex cessive mortality unrelated to development of angiosarcoma. With respect to the predicted versus experimental incidence of hepatic angio sarcoma experienced by groups ofrats exposed to the remaining concentrations ofVC, none ofthe four models is unequivocally more reliable than any other. Thus, any one of the four models represents adequately the experimental data. Since no clear preference is discernible, all four models will be used to predict the incidence of hepatic an giosarcoma experienced or to be experienced by workers exposed to VC. In order to calculate the equivalent dose of VC metabolite in exposed workers, the following assumptions were made: CAG = Cancer Assessment Group, National Cancer Institute Noi ex[ lib ce dil V, V, wt m 1 73 (/) to *4 (71 (71 P' fc w o h e: s1 e e li 1 i i R. H. Reitz et al. s occurring in accordance 1 v and Vm are velocity and 1 of VC expressed as fig *ation of VC being inhaled spectively. illy fit a model relating the ce. There are a number of lost commonly employed 30. mined by linear regression e ofhepatic angiosarcoma o different concentrations 3. Using the equations dei* for the groups ofrats extese results are shown also satic angiosarcoma in rats us exposure resulted in exoma. ddence of hepatic angioling concentrations ofVC, my other. Thus, any one of - Since no clear preference incidence of hepatic ans exposed to VC. te in exposed workers, the Non-Linear Pharmacokinetic Parameters Need 87 1) The exposure period for rats (1 year = one-halflifetime) is equivalent to a 3 5-year exposure in man (also one half lifetime). 2) The rate of metabolism of VC in both species is low relative to the rate of equi libration between lungs and plasma. This means that inhalation ofequivalentvapor con centrations of VC would result in equal plasma concentrations of VC at steady state. 3) The K^, ofthe enzyme which activates VC is identical in the two species, but Vmajt differs in the two species according to the body surface area formula. That is, the V _(man) in pg VC metabolized per kg body weight = i 250/7.ooo (rat) or V _(man) = 0.153 Vm,, (rat), where 250 g and 70 kg are assumed to be the body weights of rat and man, respectively. After correction for the difference in hours of exposure (6 h for rat versus 8 h for man) the pg of VC metabolite produced per day in workers is given by the equation: 1675 (ig/h - S (xg/1 860 pg/1 + S fig/1 This equation may be used in conjunction with the four models and the rat data to predict the incidence of angiosarcoma in workers exposed to any concentration ofVC for a given fraction ofthe working life (35 years). Since a comprehensive survey ofVC workers has been published (Equitable Environmental Health, Inc. 1978), an excellent opportunity existed to assess the validity of the various models. The incidence of hepatic angiosarcoma in subgroups of 9677 workers with respect to their duration of exposure are given in Table 4. These workers include 95.1% ofthe 10,173 in the cohort study; the remainder could not be traced. The atmospheric concentrations ofvinyl chloride to which these workers had been exposed were not quantitated, except subjectively into high, medium, and low cat egories. Since the time-weighted-average (TWA) exposure recommended by the Amer ican Conference of Governmental Industrial Hygienists (ACGIH) prior to 1972 was 500 ppm and subsequently 200 ppm until adoption of the Occupational Safety and Health Act standard ofless than 1 ppm in 1974, it is assumed that even those exposures subjectively deemed low were high. Indeed, it is reasonable to expect the TWA exposures of200 ppm and greater for 8 h were common rather than the exceptions. The actual incidence ofangiosarcoma, as well as the predictions from each ofthe four mod els is summarized in Table 5 for workers exposed to 200 and 500 ppm VC. Ex- Table 4, Incidence of angiosarcoma in 9,677 workmen exposed to vinyl chloride Population Duration of exposure, years Mean exposure duration, years Fraction of working life Observed angiosarcomas 4,384 (0.31) 2,339 (0.27) 946 (0.58) 1,007 (1.0) 677 (1.0) 324 (1.0) <4 5-9 10-14 15-19 20-24 25+ 2 7 12 17 22 27 0.06 0.20 0.34 0.49 0.63 0.77 1 (18 years) 0 2 (15 and 23 years) 2 (18 and 19 years) 0 0 R&S 142756 88 R. H, Reitz et at. Table 5. Predicted number of angiosarcomas in workmen exposed to 500 or 200 ppm vinyl chloride 8 h/day, 5 day/week for various fractions ofa 35-year working life versus the numbers observed using bio transformation and angiosarcoma incidence data from rats and four models for extrapolation Population Angiosarcomas Model A observed 500 200 Model B 500 200 Model C 500 200 Model D 500 200 4,384 2,339 946 1,007 677 324 9,677 1 0 2 2 0 0 5 0.2 2.4 3.0 6.8 7.3 4.9 25 0.0 0.8 1.1 2.7 3.0 2.1 10 0 0 0 4.0 6.6 4.9 16 0 0 0 0 0.3 1.3 2 9.4 16.7 11.5 17.6 15.2 8.9 79 5.9 10.4 7.2 11.0 9.5 5.6 50 10.0 17.7 12.1 18.6 16.0 9.4 84 6.3 11.1 7.6 11.7 10.1 5.9 53 R&S 142757 animation ofthe data in this table indicates that Models C and D overestimate the inci dence of hepatic angiosarcoma experienced by these workers while Model B un derestimates the incidences. When a TWA of200 ppm is utilized to represent the atmospheric concentration to which these workers were exposed, the predicted incidence of angiosarcoma of 10 cases using Model A compares favorably with that experienced; five cases. For a significant number ofworkers the time elapsed since initial exposure has yet to reach 15 years, the shortest time needed for development ofangiosarcoma in any ofthe afflicted workers, Table 4. Thus, it may be anticipated that 3--4 additional cases might occur in this population. It is comforting that an epidemic of cancer cases, as has been envisioned by some, does not seem likely. The foregoing analysis using a probit percentage incidence model (Model A) has revealed that data collected in experiments on rats may be used to predict the incidence of hepatic angiosarcoma in humans with reasonable accuracy. It is of interest to calculate that, using the model which appears to be most reliable (Model A, Probit %) exposure ofworkers to 1 ppm VC for 35 years would be expected to result in 0.015 tu mors/million exposed population, clearly a rather minor risk. C. Nongenetic Events in Tumorigenesis As mentioned earlier, there appears to be at least two mechanisms which can result in tumor induction in animals: Genetic (involving direct DNA damage) and Nongenetic or Cytotoxic, involving prolonged stimulation of DNA replication. I. Implications for Risk Assessment Although tumors can be produced by either genetic or epigenetic mechanisms, there are fundamental differences in these mechanisms with important implications for risk extrapolation. Genetic damage may be latent for years and only be detectable with the actual appearance of tumors. Genetic damage may also be irreversible if the lesion es capes DNA repair. In contrast, tissue damage associated with epigenetic mechanisms R. H. Reitz et al. 90 or 200 ppm vinyl chloride le numbers observed using bionodels for extrapolation xiel C Model D ) 200 4 5.9 7 10.4 5 7.2 6 11.0 2 9.5 9 5.6 50 500 200 10.0 17.7 12.1 18.6 16.0 9.4 84 6.3 11.1 7.6 11.7 10.1 5.9 53 d D overestimate the incikers while Model B un- ospheric concentration to e of angiosarcoma of 10 lenced; five cases. For a posurehasyet to reach 15 ofthe afflicted ionarases might occur in ncer cases, as has been nee model (Model A) has ed to predictthe incidence tracy. It is of interest to iable (Model A, Probit %) ected to result in 0.015 turisk. anisms which can result in damage) and Nongenetic replication. genetic mechanisms, there >rtant implications for risk 3nly be detectable with the rreversible ifthe lesion esith epigenetic mechanisms Non-Linear Pharmacokinetic Parameters Need 89 should be clinically "visible" and is likely to be reversible. More importantly, although there may be some theoretical basis for arguing that genetic carcinogens may not have a threshold, all the cumulative experience of the science of toxicology suggests that there is a threshold for tissue damage and hence for nongenetic mechanisms of carcinogenesis. Thus, although we must continue to exercise great caution when dealing with genetic carcinogens, conventional techniques of industrial hygiene and toxicology can identify exposure levels which do not produce tissue damage and which will preclude the induction of any tumors through nongenetic mechanisms. II. Experimental Procedures for Assessing the Mechanism of Tumorigenesis In order to determine what types ofsafety precaution are appropriate for dealing with a carcinogen whose mechanism is not known, objective laboratory tests must be de veloped. Several procedures are under investigation currently at Dow's Toxicology Re search Laboratory in order to determine their utility in indicating the mechanism oftu morigenesis. Primary attention has been focused on in vivo procedures since these ap pear to have the greatest potential for understanding the tumorigenic process in the in tact animals. Four such indicators will be described, and then experimental results will be employed to classify three chemical agents, dimethylnitrosamine (DMN), vinylidene chloride (1,1-dichloroethylene, VDC) and perchloroethylene (1,1,2,2-tetrachloroethylene, PERC), as genotoxic or cytotoxic. 1. Genetic Indicators a) DNA Alkylation. In this procedure animals are exposed to a radioactive form ofthe chemical to be tested. Next, tissues are excised and DNA is isolated and purified by a modification ofthe method ofMarmur (1961). A stringent purification ofDNA is nec essary for these experiments because large amounts ofradioactivity become associated with macromolecules other than DNA. Consequently, contamination with even small amounts ofthese highly labelled materials would affect the estimation ofthe low levels of DNA bound radioactivity. For this reason, DNA isolated by the procedure of Marmur has been further purified by digestion with specific enzymes to remove protein, glycogen, and RNA (Reitz et al., 1980). The degree of alkylation (and hence damage) is determined by quantitating the DNA colorimetrically and measuring radioactivity by scintillation counting. For con venience, all alkylations are reported in the units ofalkylations/10* nucleotide residues. b) DNA Repair is an indicator ofpreceding DNA damage. This can be estimated in vivo by measuring the rate ofincorporation of 3H-thymidine into DNA in the presence of hydroxyurea (Cleaver, 1969; Craddock et al., 1976). While this procedure does not have the sensitivity of the DNA alkylation method, it can be performed with a nonradioactive form of the chemical to be tested. In utilizing this method, it is essential that correction for the effect ofthe test chemical on the amount ofnormal replication es caping hydroxyurea be performed as suggested by Arfellini et al. (1978). 2. Nongenetic (Cytotoxic) Indicators a) DNA Synthesis -- A common characteristic of the treatments which increase spontaneous or chemically-initiated tumor incidence is that they stimulate cell division R&S 142758 90 R. H. Reitz et al. (Marx, 1978). This can be measured directly by administration of 3H-thymidine to ani mals at a time when regeneration is occurring (48--96 h post exposure). DNA is isolated from the tissues of treated or control animals by the procedures already described (DNA Alkylation). Since DNA synthesis may vary with age, time of day, nutritional status, etc., a matched control group is always surveyed at the same time as the treated group, and the ratio of treated to control is determined. (A ratio of 1.00 indicates that treatment had no effect, while ratios greater than 1.00 reflect a stimulation of DNA synthe sis.) b) Histopathology -- The microscopic examination of tissue sections from treated animals has been one ofthe most useful techniques of toxicology. This technique yields information regarding the sensitivity of various organs, the persistence ofthe toxic ef fect, and a rough indication of the severity of cytotoxic lesions. III. Experimental Classification of Mechanism of Carcinogenesis The results of studies ofDNA alkylation with the three chemicals are shown in Table 6. In each case, the doses chosen for this study are nearly the maximum that can be given to these animals without causing some delayed toxicity, and consequently, approxi mate a "maximum tolerated dose" (MTD). In each case, the dose tested has been re ported also to induce tumors in the indicated species. Clearly there is a major difference between the ability of these agents to produce DNA alkylation. DMNproduces over 3,000 alkylations per 10* nucleotides in the sen sitive tissue (liver). However, the level of DNA damage induced by the MTD for PERC was not detectable (with a detection limit of 10 alkylations/106 nucleotides). DNA damage following exposure to VDC was detected (Table 6) but was very low. In the kidney ofmice (which is the site ofVDC-related tumors reported by Maltoni, 1977) the alkylation of DNA was more than one-hundred fold less than that observed with DMN. These studies thus classify DMN as an agent with considerable potential to in- Table 6. DNA alkylation after exposure to VDC, PERC, and DMN Treatment/tissue Alkylations/nucleotide x 104 VDC-50 ppm (CD-I mice, inhalation) Kidney Liver PERC-500 mg/kg (B6C3-F1 mice, p.o.) Liver DMN-10 mg/kg (rat, i.p.) Liver DMN-3 mg/kg (rat, i.p.) Liver 30 6.1 0* 3.5004 1.0004 * Detection limit 10 alkylations/IO4 nucleotides b Data from Pegg and Hui (1978) R. H, Reitz et al. of 3H-thymidine to ani>st exposure). DNA is he procedures already utritional status, etc., a 5 the treated group, and ndicates that treatment :ation of DNA synthe- te sections from treated ;y. This technique yields rsistence ofthe toxic efsions. genesis als are shown in Table 6. dmum that can be given consequently, approxidosc tested has been rethe^^ents to produce 0* nucleotides in the sendbytheMTDforPERC 'lO* nucleotides). DNA sut was very low. In the ed by Maltoni, 1977) the ban that observed with siderable potential to in- 91 Mon-Linear Pharmacokinetic Parameters Need duce in vivo genetic damage, but do not indicate this potential for either VDC or PERACn.other indication of DNA damage is the presence of DNA repair following chemical exposure. In Figure 4, the relative amounts of DNA repair occurring after treatment with DMN or VDC are shown. The amount ofhydroxyurea-resistant DNA synthesis gives a good dose-response curve in the range of 3--20 mg/kg DMN, increasing 637% over the controlvalue after 20 mg/kgDMN. In contrast, the DNA re pair after50 ppm VDC was increased only 38%, barely above the experimental "noise" in the system. DNA repair was not studied after PERC administration, but the results for DMN and VDC again show a major difference; DMNclearly induced considerable geneInticdicdatmorasgeofwehpiliegeVneDtiCc adcidtivnitoyt.(DNA synthesis, histopathoiogy) were studied also forthese three chemicals. DNA replication, measured 48 h afterexposure, was ele vated 25-fold over controls inthe kidneys ofmice exposed to 50 ppm ofVDC. The mas sive increase in DNA synthesis observed in the kidney was not seen in livers from VDC-exposed mice(Table 7). This observation is consistent with the results ofMaltoni who reported tumors in the kidney, but not the liver, of VDC-exposed mice. Small increases in DNA synthesis were also seen in the target site (liver) of mice treated with 500 mg/kg ofPERC and after 10 mg/kg DMN. These are consistent with the presence ofsome cytotoxic activity for these agents also. It is significant, however that if the dose ofDMN is reduced to 3 mg/kg, considerable genotoxic activity is still present (Table 6, DNA alkylation. Figure 4, DNA Repair), but cytotoxic activity has disappeared (Table 7). The histopathoiogy of tissues from animals receiving DMN, VDC, and PERC confirms the cytotoxicity predicted by increased DNA synthesis (Table 8). Severe kidney damage, leading to nephrosis and subsequent regeneration. ONA REPAIR (Du* to DNA Demagel DMN 50 Liver R&S 142760 .Fig. 4 DNA repair in vivo after administration of DMN (mg/kg, tp.) or VDC (ppm, 6 h inhalation) to CD-I mice. Repair is calculated as hydroxyurea-resistant 3H thymidine uptake relative to control 92 R. H. Reitz et at. was apparent in the mice exposed to 50 ppm VDC for 6 hours, while VDC effects on the liver were minimal. Similarly, PERC produced moderate cellular damage in the liver of mice receiving 500 mg/kg by gavage. Since the administration ofeven the highest doses of VDC or PERC fails to give evidence for genotoxicity, but cytotoxicity is evident at Table 7. Percentage increase in DNA synthesis* after exposure to VDC, PERC, or DMN Treatment/tissue VDC-50 ppm (CD-I mice, inhalation) Kidney Liver VDC-10 ppm (CD-I mice, inhalation) Kidney Liver PERC-500 mg/kg (B6C3F1 mice, 12 doses, p.o.) Liver DMN-10 mgAg (tD-1 mice, i.p.) Liver DMN-3 mgAg (CD-I mice, i.p.) Liver 2,370* 145 672* 20 82* 97 14 * Significantly different than control, p < 0.05, r-test DNA synthesis is estimated by determining the specific radioactivity of DNA preparations after injection of 5H thymidine Lp. Table 8. Summary of histopathology findings in tissues from mice treated with VDC, DMN or PERC Exposure Hours Histologic findings postexposure Kidneys Liver VDC, 50 ppm, 6 h VDC, 10 ppm, 6 h 0 8 24 48 96 192 0 96 DMN, 10 mg/kg i.p. 4 52 DMN, 3 mg/kg i.p. PERC, 500 mg/kg p.o.* 4 52 18 Toxic nephrosis Progressing nephrosis Progressing nephrosis Increased mitotic figures Regeneration apparent Regeneration continuing Slight dilation, swelling Nephrosis-variable 0--20% affected No effect Slight centribular swelling No effect Centrilobular swelling Hyaline degeneration Accentuated lobular pattern, necrosis minimal or absent Centrilobular swelling No effect Moderate hepatic damage Sacrificed 18 h after the last in a series of 12 daily doses I f 51 R&S 142761 R. H, Reitz et al. s, while VDC effects on the lular damage in the liver of >n ofeven the highest doses it cytotoxicity is evident at PERC, ed with VDC, DMN or PERC Jver Slight centribular swelling So effect 'entrilobular swelling lyaline degeneration \cccntuated lobular pattern, iccrosis minimal or absent rentrilobular swelling lo effect <ioderate hepatic damage i Non-Linear Pharmacokinetic Parameters Need 93 the site(s) where tumors have been reported, these data provide a rational and objective basis for concluding that tumors related to VDC and PERC develop through primarily cytotoxic mechanisms. However, DMN clearly possesses a genotoxic mechanism, and may have a mild cytotoxic mechanism as well. This implies that although single exposures to DMN may be associated with some degree of excess cancer risk, exposures to levels of VDC or PERC which do not produce chronic tissue damage will not produce excess cancer risk. D. Summary It is impossible to prove that any chemical, natural or man-made, cannot cause cancer in man. However, it is possible to estimate the relative degrees of risk associated with various agents. The precision of these estimations increases as experimental pro cedures elucidate the basic type of mechanism associated with carcinogenesis, the role of absorption, metabolism and distribution, and excretion in increasing or decreasing activity, and the dose dependency ofmetabolic pathways. We must constantly strive to make the most accurate risk estimations possible so that the complex issues of risk/benefit may be properly considered. References Aifellini, G., Grilli, G., Prodi, G,: In vivo DNA repair after N-methyl-N-nitrosourea administration to rats of different ages. Z. Krebsforsch. 91, 157--164 (1978) Baririn, A., Brest!, HL, Croisy, A., Jacquignon, P., Malaviellc, C., Montesano, R-, Bartsch, H.: Livermicrosome-mediated formation of alkylating agents from vinyl bromide and vinyl chloride. Biochem. Biophys. Res. Commun, 67, 596--603 (1975) Bartsch, H., Malaviellc, C., Montesano, R.: Human, rat, and mouse liver mediated mutagenicity ofvinyl chloride in Salmonella typhimurium strains. Int J. Cancer 15, 429--437 (1975) Berenblum, I.: Irritation and carcinogenesis. Arch. Pathol 38, 233 (1944) Berenblum, I.: Tumor-formation following freezing with carbon dioxide snow. Br. J. Exp. Pathol. 10,179 (1929) Berman, J. J., Tong, C-, Williams, G. M.: Enhancement ofmutagenesis during cell replication ofcultured liver epithelial cells. Cancer Lett. 4, 277 (1978) Bolt, H. M, Kappus, H., Kaufmann, R., Appel, K. E., Buchter, A., Bolt, W.: Metabolism of1 'C-vinyl chlo ride in vitro and in vivo. Inserm. 52, 151--164 (1975) Bolt, H. M., Kappus, H., Buchter, A., Bolt, W.: Disposition of (l,2-MC) vinyl chloride in the rats. Arch. Toxicol. 35, 153-162 (1976) Cleaver, J. S.: Repair replication of mammalian cell DNA; Effects of compounds that inhibit DNA synthesis or dark repair. Radiat. Res. 37, 334--348 (1969) Craddock, V. M., Henderson, A. R., Ansley, C. M.: Repair replication ofDNA in the intact animal follow ing treatment with dimethylnitrosamine and with methylmethane sulfonate, studied by fractionation of nuclei in a zonal centrifuge. Biochem. Biophys. Acta 447, 53--64 (1976) Date, P. A., Gotoskar, S. V., Bhide, S. V.: Effect of partial hepatectomy on tumor incidence and metabolism of mice fed thioacetamide. J. Natl. Cancer Inst 56, 53 (1976) Equitable Environmental Health Inc. Epidemiological study of vinyl chloride workers: Final report Prepared for Manufacturing Chemists Association, Washington, DC, USA, January, 1978 Gehring, P. J., Blau, G. E.: Mechanisms of carcinogenesis: Dose response. Cancer Bull. 29, 152--160 (1977) Gehring, P. J., Watanabe, P. G., Park, C. N.: Risk ofangiosarcoma in workers exposed to vinyl chloride as predicted from studies in rats. Toxicol. Appl. Pharmacol. (In press) R&S 142762 94 R. H. Reitz et al. Gerchman, L. L., Ludlum, D. B.: The properties of0-6-methylguanine in templates for RNA polymerase. Biochem. Biophys. Acta 308, 310--316 (1973) Goth, R., Rajewsky, M. F.: Persistence of 0-6-ethylguanine in rat brain DNA: Correlation with nervous system specific carcinogenesis. Proc. Natl. Acad. Sci. USA 71, 639--643 (1974) Illett, K. F., Reid, W. D., Sipes, G., Krishna, G.: Chloroform toxicity in mice: Correlation ofrenal and he patic necrosis with covalent binding of metabolites to tissue macromolecules. Exp. Mol. Pathol. 19, 215-229 (1973) Kappus, H., Bolt, H. M., Buchter, A., Bolt, W.: Liver microsomal uptake of(14C) vinyl chloride and trans formation to protein alkylating metabolites in vitro. Toxicol. Appl. Pharmacol. 37, 461--471 (1976) Laroye, G. J.: How efficient is immunological surveillance against cancer and why does it fail? Lancet 1974, 1097-1100 Malavielle, C,, Bartsch, H-, Barbin, A., Camus, A. M., Montesano, R.: Mutagenicity of vinyl chloride, chloroethyleneoxide, chloroacetaldehyde and chloroethanol. Biochem. Biophys. Res. Commun. 63, 363-370 (1975) Maltoni, C., Lefemine, G.: Carcinogenicity assays ofvinyl chloride: Current results. Ann. N.Y. Acad. Sci. 246, 195-224 (1975) Maltoni, C., Cotti, G., Morisi, L., Chieco, P.: Carcinogenicity bioassays ofvinylidene chloride: Research plan and early results. Med. Lav. 68, 241--262 (1977) Martnur, J.: A procedure for the isolation of dioxyribonucleic acid from micro-organisms. J. Mol. Biol. 3, 208-218 (1961) Marx, J. L,: Tumor promoters: Carcinogenesis gets more complicated- Science 201, 515--518 (1978) McCormick, J. J,: Evidence that DNA excision repair processes in human fibroblasts can eliminate potentially cytotoxic and mutagenic lesions. Reported at the 10th Annual Meeting, Environmental Mutagen Society, New Orleans, Louisiana, USA, March 8--12, 1979 ,Miller, E. C., Miller, J. A.: Mechanisms ofchemical carcinogenesis: Nature ofproximate carcinogens and interactions with macromolecules. Pharmacol. Rev. 18 805 (1966) Miller, J. A.: Carcinogenesis by chemicals: An overview. Cancer Res. 30, 559 (1970) Miller, E. C.: Carcinogenesis by aromatic amines and amides. Reported at the symposium on envi ronmental carcinogenesis, Michigan State University, East Lansing, Michigan, May 21--23, 1978 Pegg, A. E., Nicholl, J. W., Magee, P. N., Swann, P, F.: Proc. Eur. Soc. Toxicol. 17, 39 (1976) Pegg, A. E.: Alkylation ofrat liver DNA by dimethylnitrosamine: Effect ofdosage on 0-6-methylguanine levels. J. Natl. Cancer Inst. 58, 681-687 (1977) Peraino, C-, Fry, R. J. M., Staffield, E.: Enhancement ofspontaneous hepatic tumorigenesis in C3H mice by dietary phenobarbital. J. Nad. Cancer Inst. 51, 1349 (1973) Rail, D. P.: Difficulties in extrapolating the results of toxicity studies in laboratory animals to man. Environ. Res. 2, 360-367 (1969) Reitz, R. H., Gehring, P. J., Park, C, N.: Carcinogenic risk estimation for chloroform: An alternative to EPA's procedures. Food CosmeL Toxicol, 16, 511--514 (1978) Reitz, R. H,, Watanabe, P. G., McKenna, M. J., Quast, J. F., Gehring, P. J,: Effects ofvinylidene chloride on DNA synthesis and DNA repair in the rat and mouse: A comparative study with dimethyl nitrosamine. Toxicol. Appl. Pharmacol. 53, in press (1980) Watanabe, P. G,, McGowan, G. R., Madrid, E. O., Gehring, P. J.: Fate of(,4C) vinyl chloride following in halation exposure in rats. Toxicol. Appl. Pharmacol. 37, 49--59 (1976a) Watanabe, P. G., McGowan, G. R-, Gehring, P. J.: Fate of (,4C) vinyl chloride after single oral administration in rats. Toxicol. Appl. Pharmacol. 36, 339--352 (1976b) Watanabe, P. G., Hefner, R. E., Jr., Gehring, P. J.: Vinyl chloride induced depression of hepatic non protein sulfhydryl content and effects on bromosulphthalein (BSP) clearance in rats. Toxicology 6, 1-8 (1976c) Watanabe, P. G., Zempel, J. H., Pegg, D. G., Gehring, P. J,: Hepatic macromolecular binding following exposure to vinyl chloride. Toxicol. Appl. Pharmacol. 44, 571--579 (1978) Weymouth, L. A., Loeb, L. A.: Mutagenesis during in vitro DNA synthesis. Proc. Natl. Acad. Sci. USA 75, 1924 (1978) R&S 142763