|
Items in cart [4] |
Questions? Call 888-624-8373 |
| Copyright © 2009. National Academy of Sciences. All rights reserved. Terms of Use and Privacy Statement |
Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter.
Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.
OCR for page 21
Issues in Risk Assessment
2
Correlations Between Carcinogenic Potency and Other Measures of Toxicity
DEFINITIONS AND BACKGROUND
McConnell (1989) has provided a definition of the maximum tolerated dose (MTD) and explained how it is determined as follows:
Sontag et al. (1976) had defined the MTD as "the highest dose of the test agent during the chronic study that can be predicted not to alter the animals' longevity [through] effects other than carcinogenicity" and stated that it should cause "no more than a 10% weight decrement, as compared to the appropriate control groups, and … not produce mortality, clinical signs of toxicity, or pathologic lesions (other than those that may be related to a neoplastic response) that would be predicted to shorten an animal's natural lifespan." That definition has been modified, but is still essentially the same. However, the main characteristic now used in selecting the MTD is histopathologic appearance; weight is a secondary consideration.
The estimated maximum tolerated dose (EMTD) is based on a 90 day or other subchronic test, and its determination involves scientific judgment applied to the information available at the end of the test period. How well the EMTD approximates the true MTD can be evaluated only after the bioassay. The highest dose tested HDT in a long-term rodent bioassay is usually used as the EMTD. However, that was not always the
OCR for page 22
Issues in Risk Assessment
case, especially in bioassays conducted before testing at the EMTD became the standard practice.
The MTD, by definition, is an inverse measure of the potency of an agent in causing chronic toxic effects, specifically those manifested as premature death, weight loss, or histopathologic changes after near-lifetime exposure. Potency refers to the range of doses over which a chemical produces increasing responses. Chemical A is considered more potent than chemical B if more of B than of A is required to elicit an identical response.
The LD50 (dose that is lethal to 50% of animals tested) is an inverse measure of the acute toxicity of an agent. It is defined as the dose (in milligrams per kilogram of body weight) that is expected to kill half a set of animals after a single administration.
The TD50 is an inverse measure of the carcinogenic potency of an agent and was defined by Peto et al. (1984) as follows:
For any particular sex, strain, species and set of experimental conditions, the TD50 is the dose rate (in mg/kg body weight/day) that, if administered chronically for a standard period—the "standard lifespan" of the species—will halve the mortality-corrected estimate of the probability of remaining tumorless throughout that period.
Gold et al. (1984, 1986a,b,c, 1987a, 1989a,b, 1990) have tabulated estimates of the TD50 for individual tumor sites and (in some cases) total tumors from more than 4,000 sets of tumor data on 1,050 chemicals. The criteria used by Gold et al. (1984) in deciding what chemicals to include in their Cancer Potency Data Base (CPDB) were as follows:
National Cancer Institute (NCI)/NTP bioassay, or
Bioassay in the published literature meeting all the following criteria:
Animals tested were mammals,
Administration was begun early in life (100 days of age or less for rats, mice, and hamsters),
Route of administration was diet, water, gavage, inhalation, or intravenous or intraperitoneal injection (i.e., where the whole body was
OCR for page 23
Issues in Risk Assessment
more likely to have been exposed than only a specific site, as with subcutaneous injection or skin painting),
Test agent was administered alone, rather than in combination with other chemicals,
Exposure was chronic, with not more than 7 days between administrations,
Duration of exposure was at least one-fourth the standard lifespan of the test species,
Duration of experiment was at least half the standard lifespan of the test species,
Research design included a control group,
Research design included at least five animals per group,
Surgical intervention was not performed,
Pathology data were reported as the number of animals with tumors, rather than the total number of tumors,
Results reported were original data, rather than secondary analyses of experiments already reported by other authors.
Bioassays of particulate or fibrous matter and of mixtures of chemicals were not included (except some commercial preparations to which humans are often exposed).
The CPDB or the computerized National Toxicology Program/National Cancer Institute (NTP/NCI) database served as the data sources for the statistical analyses of correlations between carcinogenic potency and other measures of toxicity conducted by a number of investigators (Zeise et al., 1984, 1985, 1986; Bernstein et al., 1985; Crouch et al., 1987, and Rieth and Starr, 1989a,b). In particular, in a paper specifically prepared for the present committee's workshop on the MTD, Krewski et al. (Appendix F) calculated estimates of the TD50 for a subset of 191 chemicals listed in the CPDB; they used three models of the dose-response relationship: the single-stage models used by Peto et al. (1984), a multistage model, and a Weibull (in dose) model.
In addition to the TD50, carcinogenic potency can be measured on the basis of the slope of the dose-response curve in the low dose region, expressed by the parameter q1. The parameter q1 is the coefficient of the linear term in the multistage model of Armitage and Doll (1961) as adapted for risk assessment by Crump (1984). When the model is applied to experimental data on tumor frequencies, q1 is an estimate of the
OCR for page 24
Issues in Risk Assessment
carcinogenic potency of an agent at low doses. The statistical upper confidence limit on q1, denoted q1*, can be determined by the methods of Crump (1984), which are sometimes referred to as the linearized multistage (LMS) model. Although q1 is sometimes zero, q1* is always positive, and the upper limit on the extra risk of cancer (above the spontaneous incidence) associated with a small dose, d, has approximately the linear form q1* d.
Krewski et al. (1991) proposed a ''model-free" estimate (MFX) of low dose carcinogenic potency based on a series of secant approximations to the slope of the dose-response curve between points in the low dose region and controls. Because their derivations are similar, MFX and q1* generally give similar estimates of low dose carcinogenic potency (Krewski et al., 1991).
CORRELATIONS
Several authors have reported a high correlation between the TD50 (or carcinogenic potency) and the HDT within various selected subsets of data in the CPDB (Bernstein et al., 1985; Crouch et al., 1987; Rieth and Starr, 1989a,b). In most cases, the HDT was also the EMTD; that implies a high correlation between high dose carcinogenic potency and potency in causing other chronic toxic effects. Zeise et al. (1984, 1985, 1986) and Metzger et al. (1989) have reported high correlations between the TD50 and the LD50 (i.e., between carcinogenic potency and acute toxicity).
The committee decided to review those correlations in an effort to investigate the relationship between toxicity and carcinogenic potency in MTD bioassays. Krewski et al. (Appendix F) performed the review and extended the reported correlations in several ways on the basis of data from their subset of 191 chemicals in the CPDB. They included in their analysis all studies in the CPDB that met the following criteria:
Rodents were used.
Chemicals were given orally.
Results specified organ or tumor type, not total tumor-bearing animals.
OCR for page 25
Issues in Risk Assessment
Exposure to the test chemical did not notably reduce survival of the test animals in comparison with unexposed controls.
The dose-response trend was significant at p < 0.01.
Authors stated that results were positive for carcinogenicity.
The study included at least two doses and controls.
In their analysis, Krewski et al. omitted data at the highest dose if the dose-response curve turned downward and used the smallest TD50 if data were available from multiple sites or experiments. Krewski et al. estimated the TD50 with three models of the dose-response relationship: single-stage, multistage, and Weibull models. The correlation coefficients between estimates of the TD50 and HDT were 0.924, 0.952, and 0.821, respectively. Krewski et al. attributed the differences in correlation coefficients to the fact that the multistage model provides for upward curvature of the dose-response relationship, whereas the Weibull model provides for both upward and downward curvature and so is likely to permit a greater range of TD50 values. Krewski et al. also calculated correlations between the HDT and estimates of low dose carcinogenic potency; they reported a correlation coefficient of -0.941 between the HDT and q1* and a correlation coefficient of -0.960 between the HDT and the estimate of low dose potency based on the MFX. Finally, Krewski et al. explored how estimates of low dose and high dose carcinogenic potency could be predicted from the HDT on the basis of the observed correlations. Using the method of Gaylor (1989), they showed that a preliminary estimate of the upper-bound dose corresponding to the 95% upper confidence limit for an increased cancer risk of 1 × 10-6 based on the LMS model could be made in the absence of a standard bioassay by dividing the MTD by 380,000.
The main issue that has arisen in interpreting the observed correlations, both in previous publications (Bernstein et al., 1985; Crouch et al., 1987; and Rieth and Starr, 1989a,b) and in the MTD workshop discussions, is the extent to which the correlations are tautologous, that is, determined by features of the experimental designs and by the ways in which the experimental data are selected and analyzed, rather than by the underlying biologic mechanisms.
Figure 2-1 shows the TD50s calculated by applying the one-stage dose-response model to data in the CPDB (Krewski et al., Appendix F) plotted against MTDs on a log-log scale. (The horizontal axis is the
OCR for page 26
Issues in Risk Assessment
highest dose tested, which is assumed to correspond to an estimate of the MTD.) As Figure 2-1 indicates, the data on the 191 chemicals are tightly grouped about the best-fitting linear regression line; none of the TD50s appears to differ from that predicted by the regression line by more than a factor of about 10.
Bernstein et al. (1985) point out that this relationship can be explained
FIGURE 2-1 Association between carcinogenic potency and maximum tolerated dose. Relationship between the highest dose tested for 191 chemicals that tested positive for carcinogenicity in chronic rodent bioassays and their TD50, an inverse measure of carcinogenic potency defined dose rate that halves likelihood of remaining tumor-free. Region A corresponds to chemicals of low carcinogenic potency (high TD50) relative to MTD; Region C corresponds to chemicals of high carcinogenic potency (low TD50) relative to MTD. Most chemicals fall into Region B, their toxicities and carcinogenic potencies are correlated.
OCR for page 27
Issues in Risk Assessment
on the basis that, given the MTD, the TD50 is constrained to lie between two bounds. The lower bound is determined by the number of animals tested at the MTD, the spontaneous tumor rate, and the level of statistical significance required to label a chemical as a carcinogen. For example, if there is a 10% spontaneous rate and 50 animals are tested at the MTD, at least 10 animals must respond at the MTD if an effect is to be declared statistically significant at the 5% probability level. That minimum corresponds to a maximum TD50 value of 5.9 times the MTD. The upper bound is determined by the fact that it is very rare for 100% of the animals tested at the MTD to get tumors. If 49 of 50 animals get cancer at the MTD, compared with five of 50 control animals, the TD50 is estimated as 0.18 MTD. Thus, within those two bounds, the TD50 differs from the MTD only by, at most, a factor of about 6—i.e., 5.9 or 5.6 (1/0.18). Bernstein et al. (1985) showed that similar bounds apply to more general experimental designs involving two or three dosed groups. Assuming that the TD50 is uniformly distributed within those limits, Krewski et al. (Appendix F) showed that the theoretical correlation between log MTD and log TD50 would be 0.965, which is very close to the 0.918 obtained by Krewski et al. from the data shown in Figure 2-1. Thus, given the bounds established by Bernstein et al. (1985), a high correlation between MTD and TD 50 is inevitable.
The committee's discussion of the possible interpretations and implications of those findings centered around the three regions shown in Figure 2-1. Region B (the region between the two broken lines) corresponds to the region determined by the bounds introduced by Bernstein et al. (1985); Region A corresponds to chemicals of low carcinogenic potency (high TD50) relative to their MTD; and Region C corresponds to chemicals of high carcinogenic potency (low TD50) relative to their MTD. The salient feature of this plot is that most of the chemicals in the analysis fall in Region B, whereas Regions A and C are virtually empty. Because the existence of a correlation is implied by the absence of chemicals in Regions A and C, the committee undertook to understand more fully why Regions A and C are nearly empty.
Krewski et al. (Appendix F) restricted their analysis to all chemicals in the CPDB that were clearly carcinogenic. Chemicals that were not identified as carcinogenic in any animal bioassay might have included both true noncarcinogens and chemicals with a carcinogenic potency that was too low to cause statistically significant increases in tumors in ani-
OCR for page 28
Issues in Risk Assessment
mals exposed at the MTD for a lifetime. Although the latter chemicals cannot be positioned with precision in Figure 2-1, it is clear from the definition of the three regions that they are Region A chemicals; if they could be positioned in Region A, the observed correlations would probably be reduced. Thus, the committee concludes that the correlation between the MTD and the TD50 might apply not to all chemicals, but only to those with carcinogenic potency high enough to cause statistically significant increases in tumors in animals exposed at the MTD for a lifetime. Therefore, the observed correlation is partially tautologous, to the extent that it might result partially from our inability to position Region A chemicals in graphs like Figure 2-1.
However, the absence of chemicals from Region C is not obviously tautologous. If a chemical tested in a standard long-term bioassay is a true Region C chemical, it should be identifiable as such. If a chemical caused cancer in all animals tested at the MTD, it still might not cause cancer in all animals exposed at lower doses or for shorter periods, in which case it could be positioned in Figure 2-1. If a chemical caused tumors in all the dosed animals in a bioassay, it could be identified as belonging in Region C, although it would not be possible to position it at a specific location in Region C. However, it is possible that chemicals belonging to Region C have been systematically excluded from the CPDB or, if present in the CPDB, excluded from the analyses conducted by Krewski et al. Although the inclusion criteria used by Krewski et al. (e.g., inclusion only of studies that used the oral route) resulted in the inclusion of only a fraction of the chemicals in the CPDB, it does not appear that there would be a bias toward excluding chemicals with low TD 50s relative to their HDTs. But the inclusion criteria for the CPDB could have resulted in the omission of some Region C chemicals. For instance, the exclusion of studies that lasted for less than half the normal lifespan of the animals could have resulted in excluding studies that were terminated when tumors were detected very early. And some chemicals might have been identified as potent carcinogens long ago and consequently not tested in a bioassay that satisfied standards used by Gold et al. (1984) as criteria for inclusion; e.g., some early bioassays might not have used control groups or might not have reported results in terms of the numbers of animals with tumors.
To investigate further the extent to which Region C carcinogens exist, the committee compiled a list of 18 chemicals (Table 2-1) that it judged as potential Region C carcinogens and conducted a detailed study of
OCR for page 29
Issues in Risk Assessment
TABLE 2-1 Chemicals Considered as Potential Region C Carcinogens
2-acetylaminofluorenea
Dimethyl sulfatec
Acrylonitrilea
Ethylene dibromidea
Benzidinea
Ethylene oxidea
Benzo[a]pyreneb
Ethylnitrosoureaa
1,3-Butadieneb
Methyl bromide
Carbon tetrachloridea
4,4'-Methylene-Bis(2-chloroaniline) (MOCA)a
C.I. Direct Black 38a
C.I. Direct Blue 6a
Plutoniumc
C.I. Direct Brown 95a
Vinyl chloridec
Dibenz[a,h]anthracenec
aReported in CPDB.
bAdditional data obtained.
cNot analyzed.
them. The detailed results of the study are included as Appendix G to this report. Twelve of the 18 chemicals were already represented in the CPDB. Suitable quantitative data were found on benzo[a]pyrene and 1,3-butadiene that permitted estimation of TD50s for them, and an additional ingestion study of vinyl chloride was identified that was not included in the CPDB; data on those three chemicals were provided to Krewski et al., who calculated TD50s with the same procedure as used in the workshop paper. TD50s could not be estimated for dibenz[a,h]anthracene, dimethyl sulfate, and methyl bromide, because no suitable quantitative data were located. Plutonium was not included in the analysis, because the dose measure used for it was not commensurate with that used for chemical carcinogens. Thus, of the 18 chemicals identified by the committee as potential Region C carcinogens, 14 were investigated. Five of these—the three benzidine dyes (C.I. direct black 38, C.I. direct blue 6, and C.I. direct brown 95), carbon tetrachloride, and 1,3-butadiene—belong in Region C (Figure 2-2). However, none of the TD50s for these five chemicals was more than a factor of 10 or so smaller than the bound that divides Region C from Region B. Thus, the committee did not uncover any chemicals that appear to be positioned substantially far out into Region C.
The methods used by Krewski et al. (Appendix F) to estimate TD50s
OCR for page 30
Issues in Risk Assessment
FIGURE 2 Association between carcinogenic potency and maximum tolerated dose. This graph is essentially the same as Figure 1 with 14 potential Region C chemicals explicitly identified. Included are data on three chemicals that were examined by the committee but were not present in the CPDB.
are similar to those used in the CPDB. As in the CPDB, the TD50s in Figures 2-1 and 2-2 are based on an essentially linear one-stage model applied to the crude proportions of animals that developed tumors during the course of the study. For studies that lasted less than the standard lifespan of the test species, the TD50 is adjusted to a standard rodent lifetime by multiplying it by a correction factor, f2, where f is the ratio of the length of the experiment to the normal lifespan of the test species. That correction is based on the assumption that, if experimental animals had lived longer, a greater percentage of them would have developed
OCR for page 31
Issues in Risk Assessment
tumors as a result of exposure to the test agent, or agent-related tumors would have been discovered at lower doses.
The studies of the three benzidine dyes involved rats and lasted only 3-months, so a correction factor of (3/24)2 = 0.016 was applied. Although the criteria for inclusion of rodent experiments in the CPDB stipulate that they lasted for a minimum of 12-months, all NTP studies are included in the CPDB, regardless of duration. The three experiments with the benzidine dyes were actually subchronic toxicity studies in which neoplastic lesions were observed. For ethylene dibromide, most of the rats died of tumors within 6- months, and a correction factor of (6/24)2 = 0.06 was used. The carbon tetrachloride study lasted 84 weeks and involved a correction factor of (84/104)2 = 0.65. Because the bioassays of each of the chemicals lasted less than the lifespan, the exact positioning of the five chemicals in Region C is uncertain (although more nearly certain for carbon tetrachloride).
The correction used in the CPDB is ad hoc and does not have a strong experimental or theoretical basis. The Environmental Protection Agency typically uses a correction factor of f3 (Anderson et al., 1983), which would make estimated TD50s even smaller. Portier et al. (1980) found that a factor of f3 was consistent with many bioassays conducted by the NTP. Doll (1971) observed that a factor of f2 to f6 is needed to describe the age-incidence curves for human cancer. It is not clear what is an appropriate correction factor, so TD50s the basis of studies lasting less than the standard lifespan of the test species are more uncertain than those estimated on the basis of whole-lifespan studies. The effect of this adjustment factor on the correlation between the TD50 and the MTD warrants investigation, although ultimate resolution of this uncertainty would require retesting of these chemicals at lower doses for a full lifespan.
In summary, the committee's informal study suggested that documented Region C carcinogens are rare. The best candidates for Region C carcinogens are five chemicals that induce cancer in rodents after short exposures. No bioassay lasting the full lifespan of the test species has been conducted for any of the five chemicals. Consequently, estimates of the TD50s for the chemicals are uncertain, and their designation as Region C carcinogens is also uncertain.
In interpreting those results, we must keep two facts in mind. First, the TD50s are uncertain for chemicals that have not been tested in life-
OCR for page 32
Issues in Risk Assessment
time bioassays; it is theoretically possible that these chemicals could be positioned substantially further out into Region C than is indicated by Figure 2-2. Second, the analyses assume that the HDT in each study is a reasonable approximation of the MTD; this might not be the case for all studies in the CPDB, particularly some of the older studies.
On the basis of the evidence discussed above, the committee concludes that the chemicals tested to date in lifetime bioassays have been found generally not to have TD50s that are far less than their MTDs. The Bernstein lower bound of 0.18 MTD for the TD50 will not be universally applicable, however, because it is based on the use of an essentially linear one-hit model and the assumption that not all of the exposed animals will develop tumors. The lower bound also might not apply to TD50s adjusted for intercurrent mortality with the method described by Sawyer et al. (1984), as is done in the CPDB whenever individual animal survival times are available. Krewski et al. (Appendix F) provide bounds on the TD50 relative to the MTD by using a Weibull dose-response model, which allows for curvature in the dose-response relationship. Although the bounds are wider, the correlation between the TD50 and MTD remains high, regardless of the degree of curvature in the dose-response curve (Kodell et al., 1990).
The committee further concludes that comparisons between TD50s and MTDs indicate a relationship between measures of general toxicity (e.g., the MTD) and carcinogenic potency that can be expressed as follows: Animal carcinogens generally have a carcinogenic potency sufficient to cause just-detectable increases in cancer in standard bioassays at doses near the MTD.
The correlations considered here demonstrate that the carcinogenic potencies of materials found to be carcinogens are inversely related to MTDs; that is, if the MTD is high, the carcinogenic potency tends to be low. However, the MTD itself does not predict the likelihood that a material will be a carcinogen. The underlying cause of the observed relationship is not clear. General toxicity and cancer induction have a number of steps in common. A material must be absorbed, possibly metabolized, and transported to the site of action. Those common elements might be partially responsible for the observed relationship.
In addition, cell toxicity might result in increased cell division, which in turn could result in the permanent incorporation of spontaneous DNA damage that will eventually lead to cancer. The observed relationship
OCR for page 33
Issues in Risk Assessment
between toxicity (as quantified by the MTD) and carcinogenicity (as quantified by the TD50) is consistent with cell toxicity and the resulting cell proliferation's mediating of the carcinogenicity observed in some animal bioassays. However, the committee recognizes that other reasons for the observed relationship are possible. The committee suggests that experiments in which cell proliferation and carcinogenic or precarcinogenic responses are measured directly and compared will allow more definitive evaluation of relationships among toxicity, cell proliferation, and carcinogenicity.
RELATIONSHIP BETWEEN TOXICITY AND CARCINOGENICITY OBSERVED AT MTD
The practice of assessing risk associated with human exposures to chemicals on the basis of data from studies conducted in laboratory animals rests on a number of assumptions. Among them are the assumptions that the agents will produce qualitatively similar effects in animals and humans and that the relative potency in animals approximates the relative potency in humans. In general, assumptions about the relationships between animal and human data have proved fairly reliable. For instance, the application of toxicity, pharmacokinetic, and metabolic data derived from animal studies to human medicine has contributed to reducing the human risk associated with therapeutic agents.
The practice of classifying chemical substances as either carcinogenic or noncarcinogenic on the basis of animal tests conducted at the MTD involves a further assumption—that carcinogenesis is a specific response to exposure to specific chemical structures (agent specificity), rather than a nonspecific response of animals to induction of chronic toxicity. That assumption is necessary because chronic administration at the MTD often produces adverse effects in the tested animal populations. In fact, if no adverse effects have been observed in a chronic bioassay, the bioassay could be classified as inadequate, on the grounds that the MTD was not achieved and that the test had insufficient sensitivity to detect the carcinogenicity of the material tested. However, some researchers have argued that the observation of increased frequencies of tumors in animals receiving the MTD might not always be a chemical-specific
OCR for page 34
Issues in Risk Assessment
phenomenon, but might be a secondary response to the induction of chronic toxicity. That is, perhaps chronic toxicity itself or some other high dose phenomenon is capable of inducing cancer.
It has been suggested in particular that carcinogenic responses to exposures at high doses are in many cases either totally or partially caused indirectly by mitogenesis (Ames and Gold, 1990). The idea is that high doses (at or near the MTD) cause toxic responses, which can cause cell proliferation (mitogenesis). A dividing cell is at greater risk of mutating than a quiescent cell, so mitogenesis is indirectly mutagenic and consequently associated with an increased likelihood of carcinogenesis. That mechanism might be totally responsible for a carcinogenic response, as hypothesized for sodium saccharin (Cohen and Ellwein, 1990a), or might accentuate the carcinogenicity of genotoxic compounds, as hypothesized for 2-acetylaminofluorene (2-AAF) (Cohen and Ellwein, 1990b). In the former case, a threshold was hypothesized for saccharin on the basis of chemical evidence that silicate crystals responsible for cell proliferation in rats do not form at lower doses. In the latter case, a synergistic effect between genotoxicity and cell proliferation was hypothesized for 2-AAF at high doses in the bladders of female mice, but only a genotoxic effect at lower doses at which cell proliferation was not expected to occur. That observation suggests a dose-response relationship for bladder cancer that is nonlinear at high doses but linear at lower doses where cell proliferation is absent. 2-AAF does not induce cell proliferation in all target organ systems, however; the dose-response relationship for liver cancer in mice appears to be linear throughout the entire dose range.
The relationship between toxicity (including mitogenesis) and carcinogenesis has been studied recently. A direct relationship between toxicity and carcinogenesis has been suggested for a number of nongenotoxic chemicals, such as saccharin (noted above), the antioxidant butylated hydroxyanisole (BHA), di-(2-ethylhexyl) phthalate (DEHP), and polychlorinated biphenyls (PCBs). Chronic rodent bioassays of those chemicals have revealed tumor induction at doses that also are associated with toxicity and the presence of nonneoplastic proliferative lesions. For example, in two-generation studies in adults and weanling rats (Anderson et al., 1988; Williams, 1988), saccharin administered at 5% of the diet induces bladder tumors, cytotoxicity, and regenerative hyperplasia, increasing the labeling index (a measure of cell proliferation) of the
OCR for page 35
Issues in Risk Assessment
urothelium by a factor of 2-10. BHA induces forestomach carcinomas in rats and hamsters when administered as 2% of the diet; severe hyperplasia and cytotoxicity, as evidenced by erosion and ulceration, are also seen (Ito et al., 1991). DEHP is a peroxisome proliferator that has been shown to induce liver tumors, foci of hepatocellular alteration (previously described as neoplastic nodules), and an initial burst of mitosis in rats and mice when given at 3,000-12,000 ppm in the diet (Kluwe et al., 1982; Mitchell et al., 1985). Some PCB mixtures induce focal necrosis, fatty degeneration, and hyperplastic nodules in the livers of rats and mice at concentrations that also induce hepatic adenomas and carcinomas (Kimbrough and Linder, 1974; Kimbrough et al., 1975).
In addition to nongenotoxic carcinogens, genotoxic carcinogens induce toxicity, and consequent cell proliferation at higher, toxic doses might play a role in increasing tumor rates to beyond what would be expected from genotoxicity alone. For example, 2-AAF administration is associated linearly with DNANTP adduct formation in the mouse bladder; however, the tumor rate in that organ is consistent with the effects of an increased rate of cell proliferation at high doses that acts in combination with 2-AAF's genotoxicity to produce tumors (Cohen and Ellwein, 1990). A similar interactive effect between cell proliferation and tumorigenesis has been observed for benzo[a]pyrene applied to mouse skin (Albert et al., 1991). Epidemiologic evidence also supports an association between some kinds of chronic toxicity and cancer incidence, such as hepatitis and liver cancer, schistosomiasis and bladder cancer, tuberculosis and lung cancer, asbestosis and mesothelioma, and tropical ulcers and skin cancer (Preston-Martin et al., 1991). Explanations other than cellular proliferation (such as inflammation) are also possible.
Thus, there is evidence from various sources to support an association between toxicity and carcinogenesis. Several people have recently attempted to analyze the assumption that the phenomena are causally related. Hoel et al. (1988) and Tennant et al. (1991) have evaluated the relationships between mutagenicity, carcinogenicity, and toxicity in laboratory rodents with the NTP data base of chronic and, in some cases, subchronic bioassays performed on a total of 130 chemicals. In those bioassays, 50 rats and mice of each sex received the MTD, MTD/2, or MTD/4 for 2 years. Matched control groups were also used. Use of the data base provided an opportunity to compare the toxic properties of chemicals that were not carcinogenic with those of chemicals that
OCR for page 36
Issues in Risk Assessment
were, under similar experimental conditions. Results included subchronic toxicity, neoplastic and chronic toxic effects observed after a 2 year exposure, chemical structure, and mutagenicity in salmonellae. Toxicity was defined by the investigators as ''any deleterious change in the tissues of animals exposed to chemicals that was discerned by histopathology"; most, but not all, toxic lesions were found to be associated with a proliferative response. Qualitative structural descriptors of toxicity were used to evaluate the relationships between regenerative or hyperplastic responses and cancer or the absence of cancer; rates of induced mitogenesis or increased rates of cell proliferation were not measured quantitatively and would have remained undetected in the absence of any structural change.
Table 2-2 shows the chemicals from the Tennant et al. (1991) study whose subchronic and chronic administration induced toxicity at the same site; some of the chemicals were carcinogens and some were not, but none was carcinogenic at the site of obvious toxicity. In contrast, Table 2-3 lists the chemicals that were carcinogenic at sites where both subchronic toxicity and chronic toxicity were present; about 40% of these were mutagenic. For both the concordant and discordant chemicals, most of the toxic lesions observed were proliferative, although the presence of proliferative lesions clearly is not predictive of carcinogenesis. Results of the Tennant et al. (1991) analysis and the Hoel et al. (1988) analysis indicate that some sites of toxicity of both carcinogens and noncarcinogens were associated with neoplasia and many were not. Some chemicals induced tumors at sites where toxicity was not in evidence, and some induced toxicity in some organs without inducing carcinogenesis. However, the majority of both mutagenic and nonmutagenic carcinogens induced tumors that were associated with chronic toxicity, although many of the same chemicals caused chronic toxicity at other sites that was not associated with carcinogenesis. Tennant et al. (1991) conclude that, although their results do not dissociate toxicity from the neoplastic process, they "illustrate the high degree of complexity of neoplastic processes and imply that there may be multiple mechanisms of carcinogenesis associated with even potent mutagens. They also provide a clear demonstration that chronic- exposure of rodents to chemicals that exhibit toxic effects does not necessarily result in carcinogenic effects. Further, even when chronic-exposures resulted in overt tissue specific toxicity, neoplasia did not necessarily develop." A temporary
OCR for page 37
Issues in Risk Assessment
TABLE 2-2 Chemicals That Induced Both Subchronic and Chronic Toxicity but Not Carcinogenicity at Same Sitesa
Rats
Mice
Chemical
Site
Lesion
Site
Lesion
Noncarcinogens
2,4-Dichlorophenol
None
--
Liver
Syncytial alteration
Dimethoxane
Forestomach
Hyperplasia
Forestomach
Hyperplasia
Hydrochlorothiazide
Kidney
Nephropathy, mineralization
None
--
à -Methyldopa sesquihydrate
None
--
Kidney
Nephropathy, karyomegaly
Carcinogens
p-Chloroaniline HCl
Bone marrow, liver
Hyperplasia, hemosiderin
Kidney
Hemosiderin
Nitrofurantion
Testes
Degeneration
Testes
Degeneration
Tribromoethane
Liver
Inflammation, vacuolization
Liver
Inflammation, vacuolization
Malonaldehyde, Na salt
Glandular stomach; bone marrow
Ulcer, inflammation; hyperplasia
Pancreas
Atrophy
Furosemide
None
--
Kidney
Nephropathy
aFrom Tennant et al. (1991).
OCR for page 38
Issues in Risk Assessment
TABLE 2-3 Chemicals That Induced Toxicity and Carcinogenicity at Same Sitea
Chemical
Mutagen
Rats
Mice
Site
Lesion
Site
Lesion
Glycidol
+
Forestomach
Dysplasia, carcinoma
Forestomach
Hyperplasia, carcinoma
p-Chloroaniline HCl
+
Spleen;
adrenal
Fibrosis, metaplasia, sarcoma; hyperplasia, pheochromocytoma
Liver
Hemosiderosis,
hepatocellular tumors
N,N-Dimethylaniline
+
Spleen
Fibrosis, metaplasia, sarcoma
Forestomach
Hyperplasia, squamous cell tumors
Nitrofurantoin
+
Kidney
Nephropathy, tubular cell adenoma and carcinoma
Ovary
Atrophy, ovarian tumors
4-Vinyl-1-cyclohexene diepoxide
+
Skin
Hyperplasia, basal and squamous cell carcinoma
Skin, ovary
Hyperplasia, basal and
squamous cell tumors;
atrophy, ovarian tumor
N-Methylolacrylamide
-
None
--
Ovary
Atrophy, granulosa cell tumors
Benzofuran
-
Kidney
Nephropathy, tubular cell adenocarcinoma
Liver; forestomach; lung
Syncytial changes, liver tumors;
hyperplasia, carcinoma; hyperplasia,
OCR for page 39
Issues in Risk Assessment
Chemical
Mutagen
Rats
Mice
Site
Lesion
Site
Lesion
Ochratoxin A
-
Kidney
Degeneration, hyperplasia, tubular cell tumors
Not determined
Hexachloroethane
-
Kidney
Nephropathy, hyperplasia, tubular cell adenocarcinoma
Not determined
--
d-Limonene
-
Kidney
Mineralization, nephropathy, hyperplasia, tubular cell adenocarcinoma
None
--
Hydroquinone
-
Kidney
Nephropathy, tubular cell adenoma
Liver
Syncitial cell alteration, liver tumors
Phenylbutazone
-
Kidney
Papillary necrosis, nephropathy, transitional-cell carcinoma
Liver
Degeneration, hypertrophy, necrosis, liver tumors
aFrom Tennant et al. (1991).
OCR for page 40
Issues in Risk Assessment
toxic condition's effect on carcinogenesis might not be detected with data from chronic or even subchronic bioassays; for example, it is possible that a chemical very early in the course of a bioassay induces toxicity that enhances its carcinogenic response, but that, because of an adaptive cellular response, no chronic proliferative lesions other than tumors develop. Nonetheless, the observations that have been reported after study of the NTP database support the existence of mechanisms of carcinogenesis more complex than simple mutation or induced cell proliferation; these mechanisms are yet to be identified.
Several other reports support the conclusion of an equivocal relationship between toxicity-induced proliferation and carcinogenesis discussed above. Wada et al. (1990) showed that p-methoxyphenol administered after initiation of rat forestomach tumors with N-methyl-N'-nitro-N-nitrosoguanidine caused epithelial damage and hyperplasia in a dose-dependent manner in the forestomach epithelium, but was not associated with any increase in tumors. In an investigation of the role of renal tubular cell hyperplasia in tumor promotion with barbital sodium (BBNa) after initiation with streptozotocin (STZ) in rats, STZ was found to reduce BBNa-induced nephropathy and cell proliferation without reducing renal tumor incidence (Konishi et al., 1990). The authors of the study note, however, that initiated cells might have a very different ability from noninitiated cells to respond to the mitogenic influences of a tumor promoter and that the reduction in overall DNA synthesis that was seen might be unrelated to the increased proliferation of preneoplastic or neoplastic cells. Ward et al. (1990) reached a similar conclusion in a study of the relationship between renal or hepatocellular hyperplasia and tumor promotion with di-(2-ethylhexyl)phthalate in mice initiated with N-nitrosoethylurea.
The observation that toxicity and carcinogenicity are not always detected simultaneously make it problematic to account for increased rates of cell proliferation that are associated with carcinogenesis when one performs risk assessments of either genotoxic or nongenotoxic chemicals. The greater-than-quadratic nature of many dose-response curves for mutagens tested at and below their MTDs and the observation of toxicity and proliferative lesions in the target organs of most mutagenic carcinogens suggest that mechanisms in addition to mutation are operative and, in particular, that enhanced cell proliferation is likely to be occurring and affecting the response. In addition, most nonmutagens
OCR for page 41
Issues in Risk Assessment
also induce toxicity and nonneoplastic proliferative lesions at doses that also are associated with neoplasia. Any information on the dose-response nature of these effects, especially cell proliferation, should be included in assessments of risk where possible, although, as Konishi et al. (1990) and Ward et al. (1990) emphasize, the target cells for proliferative activity associated with carcinogenesis might not be the total parenchymal tissue; identification of the affected target cells, such as stem cells, could be necessary. These problems are addressed in the second part of this report, Issues in Risk Assessment: Two-Stage Models of Carcinogenesis.
In summary, the committee evaluated the likelihood that observed correlations between cancer potencies and other measures of toxicity of chemicals tested at the MTD are tautologous and result from bioassay design or the statistical methods used for analysis or have a biologic basis. The committee performed its evaluation by determining the correlation between estimates of the TD50 and the HDT of clearly carcinogenic chemicals found in the CPDB. A strong correlation between those quantities was observed, with no chemicals classifiable as having either high toxicity and low potency or low toxicity and high potency. The committee concluded that the correlation is partly tautologous because it applies only to chemicals with cancer potencies high enough to be detected in an MTD bioassay. However, the relationship is not entirely tautologous, possibly because the phenomena of toxicity and carcinogenesis have several similarities. The dichotomy is reflected in the conflicting results of Tennant et al. (1991), who reported that an association between some measures of toxicity and positive carcinogenicity results in the same target organ in some, but not all, NTP bioassays. It is not yet possible to draw further conclusions about the relationship between toxicity and cancer potency.
OCR for page 42
Issues in Risk Assessment
This page in the original is blank.
Representative terms from entire chapter:
cell proliferation
