Reevaluation of the Cancer Potency Factor of Toxaphene: Recommendations from a Peer Review Panel

Jay I. Goodman*,1, David J. Brusick{dagger}, William M. Busey{ddagger}, Samuel M. Cohen§, James C. Lamb,2 and Thomas B. Starr||

* Department of Pharmacology and Toxicology, Michigan State University, E. Lansing, Michigan; {dagger} Covance Laboratories, Inc., Vienna, Virginia; {ddagger} Experimental Pathology Laboratories, Inc., Herndon, Virginia; § Department of Pathology and Microbiology, University of Nebraska, Omaha, Nebraska; Jellinek, Schwartz, & Connolly, Inc., Arlington, Virginia; and || Environ, Raleigh, North Carolina

Received August 16, 1999; accepted December 1, 1999

ABSTRACT

This reevaluation of the current U.S. EPA cancer potency factor for toxaphene is based upon a review of toxaphene carcinogenesis bioassays in mice conducted by Litton Bionetics (unpublished report, 1978) and the National Cancer Institute (NCI) (Technical Report Series No. 37, conducted by Gulf South Research Institute, 1979). The mechanistic data available for toxaphene, including consideration of the potential of the compound to induce genotoxicity, was examined with an emphasis on whether this information supports a change in the cancer potency factor. If a quantitative dose-response assessment for toxaphene is to be performed, the data from both the NCI and Litton cancer bioassays should be used. Additionally, liver tumor results from female mice, rather than male mice, should be used for estimating potential human cancer risk because the background rate of liver tumors in females is lower and less variable than that exhibited by males. An ED10 was estimated as the point of departure. The mechanistic data were not sufficient to fully support a margin of exposure approach. Therefore, we believe that applying a linear extrapolation from the ED10 to the origin is an appropriate means to estimate risk at low doses. This is a highly conservative approach and, when it is applied, we conclude that the current EPA cancer potency factor should be reduced from 1.1 (mg/kg/day)–1 to 0.1 (mg/kg/day)–1.

Key Words: toxaphene; cancer potency factor; margin of exposure; pathology working group (PWG).

Toxaphene is an organochlorine pesticide belonging to a class of compounds known as polychlorinated bicyclic terpenes and was used as an insecticide on many types of fruit, vegetable, and field crops in the United States before 1982. U.S. Environmental Protection Agency (EPA) major restrictions on toxaphene were canceled in 1982 because of EPA concerns about human and environmental safety. Final cancellations occurred in 1990. Because toxaphene is persistent, potential soil and water contamination by the compound continue to be the subject of regulatory and judicial attention at many sites around the country.

Toxaphene is presently considered a category B2 carcinogen (probable human carcinogen) by EPA. For its classification, EPA relied on the results of three lifetime feeding studies, two in mice and one in rats. Toxaphene administration was reported to increase the incidence of liver tumors (adenomas or neoplastic nodules and carcinomas) in male and female mice, and thyroid tumors (primarily adenomas) in male and female rats. EPA last revised its estimates of potential human cancer risk from exposure to toxaphene in January 1991 (IRIS listing). EPA`s current potency, or unit lifetime cancer risk, posed by toxaphene exposure of 1.1 (mg/kg/day)–1 is based on a linearized multistage model fit to the male mouse liver tumor incidence data taken from a Litton Bionetics bioassay (Litton Bionetics, unpublished report, 1978).

EPA's approach to cancer risk assessment is in the process of changing from its 1991 assessment, when the toxaphene potency factor was established. EPA has proposed new guidelines, which encourage a more science-based approach. For example, the proposed guidelines eliminate the use of the linearized multistage modeling procedure, and its associated Q1* values, from consideration (EPA, 1996). The linearized multistage modeling methodology is being replaced with biologically-based mechanistic dose-response modeling, a simple straight line extrapolation, empirical nonlinear extrapolation, or application of threshold-based uncertainty factors. An estimate of the dose causing a given tumor response in the vicinity of the low end of the observable dose-response range (e.g., the ED10 [dose estimated to cause a 10 percent increase in response above background tumor frequency] or an LED10 [the lower 95 percent confidence boundary of the ED10 value]) may be used as a point of departure for low dose extrapolation in margin of exposure calculations. The proposed guidelines also recommend that human equivalent doses be calculated using three-fourths rather than two-thirds power body weight scaling. The proposed cancer risk assessment guidelines provide an opportunity to revisit the toxaphene cancer potency factor. Because the production of toxaphene was discontinued in the 1980s, the panel relied upon historical information within the context of current literature.

The scientific basis for a change in the potency factor for toxaphene was reviewed with emphasis on the following issues: (1) is the 1978 Litton Bionetics rodent bioassay an appropriate study upon which to base the cancer potency factor, or are there other bioassays that may be more appropriate? (2) what is the overall genotoxic potential of toxaphene and what implications does this potential have concerning the dose-response relationship? and (3) do the available mechanistic data support a change in the cancer potency factor?

Toxaphene Carcinogenesis Bioassays

The Litton 1978 cancer bioassay provided the basis for the current EPA potency factor. A second study, sponsored by the NCI, was also considered in this re-evaluation (NCI 1979). Histopathological assessment criteria have changed since the original Litton and NCI studies were conducted and it is possible that at least some nodules classified as adenomas or carcinomas in 1978 and 1979 would today be dismissed as non-tumorogenic hyperplastic lesions (Frith and Ward, 1979Go; Vesselinovitch and Mihailovitch, 1983Go; Ward et al., 1979Go, 1996Go; Tarone et al., 1981Go). Liver tumor slides from the NCI mouse study were available. It should be noted that Good Laboratory Practice (GLP) was not in existence at the time the Litton and NCI bioassays were performed. The NCI study's liver tumor pathology data, which had been reanalyzed by an expert pathology working group (PWG) according to contemporary diagnostic criteria, was also considered. This peer review was conducted according to GLP (Experimental Pathology Laboratories, unpublished report, 1996). The original liver tumor slides from the Litton mouse study were unavailable, rendering reanalysis of the pathology data from this study impossible.

Four bioassays were conducted on toxaphene in rodents – one in rats, one in hamsters, and two in mice (Litton Bionetics, unpublished report, 1978Litton Bionetics, unpublished report, 1979; NCI, 1979).

In the NCI rat study (1979), Osborne-Mendel rats (50/sex/group) were fed diets containing toxaphene for a period of 80 weeks and were returned to a control diet for an additional 28 or 30 weeks. Time-weighted average concentrations were 0 ppm, 556 ppm, and 1112 ppm for males; and 0 ppm, 540 ppm, and 1080 ppm for females. Because of clinical signs of toxicity in high-dose males at week 2, both dietary levels were dropped by half. At 52 weeks, both high-dose males and high-dose females developed generalized body tremors, and doses were again halved. The dietary levels are summarized in Table 1Go.


View this table:
[in this window]
[in a new window]
 
TABLE 1 Dosing Regimen in the 1979 NCI Rat Oncogenicity Study
 
In the hamster study (Litton Bionetics, 1979) ARS Golden Syrian hamsters (51/sex/group) were placed on diets containing either 0 ppm, 100 ppm, 300 ppm, or 1000 ppm toxaphene. Males were maintained on a treated diet for 21.5 months and females for 18 months. Treatment-related effects seen in this study consisted of decreased body weight in males at both treatment levels, and in megahepatocytes upon histologic examination in the liver at the high-dose level in males. The administration of toxaphene was not associated with an increase in any tumor type in hamsters. Therefore, it can be concluded that toxaphene is not oncogenic to hamsters.

In the Litton mouse study (Litton Bionetics, 1978), B6C3F1 mice (55/sex/group) were fed diets containing toxaphene for a period of 18 months, then placed on untreated diets for an additional 6 months. Dosage levels were 0 ppm, 7 ppm, 20 ppm, and 50 ppm. No treatment-related effects were seen on survival or clinical signs. Body weight and food consumption were not monitored in this study.

The rodent cancer bioassays conducted by NCI used the same batch of toxaphene (Lot No. X-16189–49) as the Litton study (NCI, 1979). The mice were obtained from either the Charles River Laboratories (NCI study) or the Frederick Cancer Research Institute (Litton study) and were bred from the same original parental generations. The doses used in the NCI study, however, were considerably higher than in the Litton study.

In the NCI mouse study, B6C3F1 mice (50/sex/group, except the matched control of 10/sex/group) were fed toxaphene for a period of 80 weeks, then placed on an untreated diet for 10 to 11 weeks. The initial nominal dietary concentrations were 160 ppm and 320 ppm; however, due to several deaths, the nominal concentrations were lowered at 19 weeks to 80 ppm and 160 ppm, respectively. This is a clear indication that, at least for the first 19 weeks of dosing, the maximum tolerated dose had been exceeded. Based on time-weighted averages, the concentrations were 0 ppm, 99 ppm, and 198 ppm (Table 2Go), over the course of the study. Liver tumor incidence data for the Litton and NCI studies are included in Table 3Go.


View this table:
[in this window]
[in a new window]
 
TABLE 2 Experimental Design for the NCI Study
 

View this table:
[in this window]
[in a new window]
 
TABLE 3 Incidence of Relevant Tumors Seen in Mouse and Rat Toxaphene Studies
 
The major difficulty in utilizing the Litton mouse study for a quantitative risk assessment is that the only significant differences between groups were in the total number of liver tumors at the highest dose in males. Control and all dose level females showed no significant differences in total tumors. There was no statistically significant difference for adenomas or carcinomas evaluated separately in males or females between groups, even at the highest dose level. In the original report submitted by Litton, statistical analyses were done utilizing a 0.05 level of significance. The combined incidence of hepatocellular adenoma and carcinoma was significant in the 50 ppm male mice with p = 0.048. However, utilizing a p value of 0.01, which is recommended for commonly occurring neoplasms (Haseman et al., 1986Go), the incidence of hepatocellular adenomas, carcinomas, or combined adenomas and carcinomas was not significant in any of the treatment groups.

If the Litton liver tumor slides were available for reexamination, it is likely that the use of modern diagnostic criteria would result in a reclassification of some or all of the carcinomas as adenomas, and adenomas as hyperplasia. Even using the original criteria for liver pathology, the results in the Litton study alone do not provide convincing positive evidence of the carcinogenicity of toxaphene under current standards. Thus, it is inappropriate to base a risk assessment for toxaphene solely on the data from the Litton study.

The original study pathologist's reading is presented in Table 3Go and the subsequent PWG reevaluation is presented in Table 4Go. Statistical analysis of the incidence of hepatocellular carcinoma and combined carcinoma and adenoma, as classified by the original study pathologist, reveals significance when compared with either the pooled or matched controls in both treatment levels of male and female mice. However, the PWG reevaluation of the liver pathology, utilizing current criteria for the classification of hepatocellular adenomas and carcinomas (Maronpot et al., 1986Go) changed the classification of the majority of the carcinomas to adenomas and, in some instances, adenomas were reclassified as foci of cellular alteration. The PWG analysis indicated no statistically significant increase in carcinomas at any dose level, and an absence of dose-response or sex differences in carcinoma rates. However, the incidence of adenomas and total tumors remained highly significant at both doses in both male and female mice. Given this positive result using current evaluation criteria, the panel recommends that any quantitative risk assessment be based on results from this study, if only one study is to be used.


View this table:
[in this window]
[in a new window]
 
TABLE 4 Liver Tumor Incidence from the NCI Mouse Study
 
The following are weaknesses in the NCI study:

The substantial incidence (20 percent) of liver tumors among male control mice also could introduce considerable uncertainty into risk estimates based upon the male mouse data.

More complete information about dose-response is provided by combining the results from both the Litton and NCI studies. These data consist primarily of tumorigenicity counts (benign lesions), not carcinogenicity. The Litton study spanned a seven-fold dose range and the NCI study spanned a two-fold dose range. In combination, these studies spanned a 33-fold dose range, and tumor incidence in male mice ranged from about 20 percent up to 96 percent, while tumor responses in female mice ranged from 0 percent to 83 percent (39 to 47).

Prior to the PWG reclassification of the liver tumors observed in the NCI study, it would not have been appropriate to combine the findings of both studies in a single analysis because of the apparent marked discrepancy between the studies in the incidence of benign and malignant tumors. In the Litton study, the most compelling evidence of a treatment-related effect was the presence of adenomas in high-dose males, whereas the major effect in the NCI study was a dramatic treatment-related increase in carcinoma incidence. However, after reclassification according to contemporary diagnostic criteria, the pattern of tumor types in the NCI study appears to be in line with the pattern in the Litton Bionetics study. Therefore, a quantitative risk assessment for toxaphene should be based on the combined female mouse liver data from both the Litton and NCI studies. The female mouse data are more appropriate than that of the males for the scientific reasons described below.

Combining the Litton and NCI data sets will result in an improved risk assessment for toxaphene because the combined data would better define the dose-response curve with respect to tumor incidence. Although there may be some question about the diagnostic legitimacy of the lesions in the Litton study, it is likely that a recalculation would result in a lower total number of tumors. The combined data would be particularly useful for better defining an ED10, which has been suggested as a point of departure for extrapolation to possible effects at lower doses.

Genotoxic Potential of Toxaphene

The genotoxicity of toxaphene has been studied in a wide variety of test systems. The results of these studies are summarized in Table 5Go.


View this table:
[in this window]
[in a new window]
 
TABLE 5 Results of Genotoxicity Testing of Toxaphene
 
The mutagenicity test results were mixed. Toxaphene was negative in the non-plasmid containing strains of salmonella (TA 1535, TA 1537, and TA 1538), but was positive (albeit weakly) in two other strains (TA 98 and TA 100) containing the pKm101 plasmid. Several sources of metabolic activation were used in the Ames assays, including preparations from hamster, mouse, rat, and human livers. Results using the human liver preparation were negative in both TA 98 and TA 100, as were the liver preparations from other sources when corn oil was used as the vehicle in the assay. Toxaphene was positive in the presence and absence of metabolic activation in TA 100 and TA 98 (positive in the presence of rat S9 only); however, the presence of S9 reduced the mutagenic activity. The results of the various Ames salmonella assays conducted on toxaphene (with and without metabolic activation) are summarized in Table 6Go. These results indicate that toxaphene is mutagenic toward salmonella strains TA 100 and TA 98, but the presence of metabolic activation in the test system reduces toxaphene's mutagenic activity toward these two salmonella strains.


View this table:
[in this window]
[in a new window]
 
TABLE 6 Ames Test Results with Toxaphene
 
Toxaphene was negative for mutation induction in Saccharomyces cerevisiae strain D4, dominant lethals in mice, DNA breaks in Escherichiacoli K-12 DNA, and chromosomal aberrations in human leukocytes (Table 6Go). Toxaphene was tested for mutagenic potential in Saccharomyces in conjunction with the Ames assay by Litton Bionetics (D. Brusick, unpublished report, 1977). The results with and without activation were negative.

In the mouse dominant lethal study (Epstein et al., 1972Go), toxaphene was one of 174 test agents. The purpose of the study was to develop standard criteria for the evaluation of the mouse dominant lethal assay. Toxaphene was tested by two routes of exposure, intraperitoneal and oral, at dose levels of 36 mg/kg and 180 mg/kg (single), and 40 mg/kg and 80 mg/kg/day (five daily doses), respectively. Deaths were seen at all dose levels except 36 mg/kg, indicating that the dose levels tested were sufficient, or even excessive, for testing for dominant lethal mutations. Toxaphene was clearly negative by the most stringent criteria.

In the Escherichia coli K-12 DNA assay (Griffin and Hill, 1978Go), a purified and radiolabeled circular plasmid was incubated with the test material for up to 28 days, and breaks in the DNA were assessed using density gradient separation of fragments. Three known alkylating agents and mutagens were used to validate the method (MMS, EMS, and MNNG). The results using toxaphene (1 mg/ml) were negative.

The third negative study was sponsored by EPA (1978) and involved monitoring workers who applied toxaphene to cotton (Texas), cattle (Colorado), and onions (Colorado) for chromosomal aberrations in blood (leukocytes). Blood was drawn from workers immediately after exposure and from appropriate unexposed individuals. The analyses for chromosomal aberrations were performed blind to avoid investigator bias. The results indicated no increase in the incidence of aberrations in the blood of individuals exposed to toxaphene.

A study by Samosh (1974) shows that accidental exposure of eight women to toxaphene resulted in an increase in the number of lymphocyte chromosomal aberrations (acentric fragments and chromatid exchanges compared to unexposed individuals). These women were exposed to an area treated with toxaphene by air.

The four remaining studies listed in Table 5Go were positive. These included assays for sister chromatid exchange (SCE) in human lymphoid cells, SCE in Chinese hamster lung cells (DON), induction of prophage lambda, and chromosomal aberrations in human lymphocyte cultures. In the human lymphoid cell study (Sobti et al., 1983Go), cells of the LAZ-007 line were incubated with eight different organochlorine pesticides. A statistically significant increase in SCE frequency was seen without the addition of metabolic activation in cells exposed to 10–5 and 10–4 M toxaphene, but not at 10–6 M. The increases were dose-related, but small (i.e., they did not reach a doubling of the solvent control). The biological relevance of SCE data is questionable. In the presence of metabolic activation, SCE frequencies were reduced. In another assay for SCE in Chinese hamster lung cells (DON) (Steinel et al., 1990Go), toxaphene was also considered to induce a positive response. In this study, toxaphene statistically increased the frequency of SCE in DON cells in a dose- and time-dependent manner. Again, a doubling over the background rate was not observed, although the rate nearly reached doubling at the longest incubation time period (28.4 h). All of the toxaphene doses tested in this study caused cell cycle delay.

Toxaphene also has been reported to induce prophage lambda in Escherichia coli, an assay method that was developed specifically for chlorinated compounds (Houk and De Marini, 1987Go). Toxaphene was positive both with and without metabolic activation; however, the response was dramatically reduced in the presence of metabolic activation. Chlordane and dichlorvos, two other presumed non-genotoxic agents, were also reported positive in this assay.

The response patterns for toxaphene are similar to those reported for phenobarbital (McClain, 1990Go), a drug that produces a similar tumor pattern and which appears to produce tumors through a similar mode of action. Phenobarbital has given mixed results in an extensive battery of mutagenicity tests. For example, it has been reported both positive and negative for mutagenic effects in the Ames salmonella assay in strains TA 1535 and TA100. The positive findings were seen only in the absence of metabolic activation. Phenobarbital has been reported to be weakly positive for gene mutations in human lymphoblast cells, and both positive and negative in the mouse lymphoma and Chinese hamster V79 cell assays for gene mutations. In the latter assay, the positive findings were seen only in the absence of metabolic activation. Results of unscheduled DNA synthesis and DNA binding assays have been negative. Phenobarbital has also been reported to induce SCE in Chinese hamster cells, but not in in vivo studies. The weight of evidence indicates clearly that phenobarbital is not a genotoxic compound (Whysner et al., 1996Go).

Additionally, both phenobarbital (McClain, 1990Go) and toxaphene (Trosko et al., 1987Go) have been shown to inhibit cell-to-cell communication. Phenobarbital inhibits intercellular communication at non-toxic dose levels. Toxaphene inhibits gap junctional-mediated intercellular communication in Chinese hamster V79 cells in the absence of cytotoxicity. Trosko et al. (1987) have concluded that these findings could explain toxaphene's tumor-promoting and neurotoxic effects.

In a study conducted by Drs. C. C. Hedli and Robert Snyder of Rutgers University, livers from toxaphene-treated mice were frozen and DNA was extracted to determine whether toxaphene induced DNA adduct formation (Hedli et al., 1998Go). DNA adduct formation was tested using nuclease P1-enhanced DNA 32P postlabeling (Reddy and Randerath, 1986Go). Three to four male mice per group were treated with either 0 mg/kg/day, 10 mg/kg/day, 50 mg/kg/day, or 100 mg/kg/day toxaphene by corn oil gavage for seven consecutive days. No evidence for DNA adduct formation was found in the livers of toxaphene-treated mice.

Based on the battery of genotoxicity studies described above and the mouse and rat oncogenicity studies, the panel supports the interpretation that any genotoxicity would be eliminated in vivo and is not responsible for the carcinogenic activity of toxaphene. Although toxaphene demonstrated genotoxicity in a limited number of tests, the expression of this property appears to be limited to in vitro test conditions. Toxaphene seems to be readily inactivated in vivo (as evidenced by the fact that DNA reactivity observed in the Ames test and in the SCE studies was strongly suppressed with the addition of S9 mix). No adducts were identified in liver cell DNA from mice exposed to relatively high doses of toxaphene. Genotoxic carcinogens typically form DNA adducts in the target tissue for tumor expression. The panel concluded that the weight of evidence suggests that toxaphene is not genotoxic in vivo and that its tumorigenic activity stems from a nongenotoxic mode of action. This conclusion is consistent with that reached by a review of the toxicology of toxaphene, published after the panel's deliberations were completed (de Geus et al., 1999Go).

Two other observations contribute, indirectly, to the support of this interpretation. First, toxaphene induces a pattern of hepatic cytochrome P-450s and liver enlargement similar to that produced by the classic non-genotoxic rodent liver tumorigen, phenobarbital. The production of mouse liver tumors and rat thyroid tumors is consistent with a nongenotoxic mode of action. Second, genotoxic carcinogens are typically characterized by the induction of multi-site tumors (e.g., in both sexes of one or more species). The tumor patterns in the toxaphene studies do not fit a genotoxic carcinogen model. The current potency factor published by EPA is based on a linearized multistage model. Although there remains some disagreement on whether linear extrapolation is appropriate for genotoxic chemicals, most non-genotoxic modes of action are compatible with a nonlinear dose-response relationship.

Review of Mechanistic Data

The mechanistic data suggest that the dose-response curve for toxaphene-induced tumors in rodents is nonlinear and, perhaps, exhibits a threshold. Additionally, the tumors produced by this compound (mouse liver, rat thyroid, and equivocal kidney tumors) are, at most, of dubious relevance in regard to human risk. Therefore, a change permitting toxaphene to be viewed as a less potent carcinogen is appropriate. Indeed, a change to a lower potency factor that includes the assumption of a linear dose-response relationship should be viewed as a conservative and appropriate approach.

Genotoxicity.
The genotoxicity data, including the study performed at Rutgers University indicating no evidence of 32P post-labeling of hepatic DNA following administration of toxaphene to mice, suggest that toxaphene has little, if any, capacity to interact directly with DNA in vivo to produce mutations (Hedli et al., 1998Go). The current potency factor published by EPA is based on a linearized multistage model; however, a nongenotoxic mode of action is compatible with a nonlinear dose-response relationship.

Similarity to phenobarbital.
Toxaphene administration results in liver enlargement and induction of hepatic cytochrome P-450s in a fashion similar to that of phenobarbital, a classic, nongenotoxic rodent liver tumor promoter (Whysner et al., 1998Go). Furthermore, as with phenobarbital (McClain, 1990Go), toxaphene is capable of inhibiting intercellular communication (Trosko et al., 1987Go), a mechanism that appears to be involved with nongenotoxic carcinogenesis. Administration of phenobarbital alone can result in liver carcinomas (Whysner et al., 1996Go). However, the dose-response curve for promotion of carcinogenicity by phenobarbital exhibits a clear threshold (Goldsworthy et al., 1984Go; Maekawa et al., 1992Go). Additionally, phenobarbital administration can cause thyroid tumors in rats by a threshold-exhibiting secondary mechanism, discussed below, related to its effects on thyroxin metabolism. It is anticipated that toxaphene will have similar threshold and mode of action properties because of its many similarities to phenobarbital.

There is strong epidemiological data indicating that phenobarbital is not a human carcinogen (Clemmesen and Hjalgrim-Jensen, 1978Go and 1980Go; Olsen et al., 1993Go and 1995Go). Therefore, phenobarbital-like compounds that can produce tumors in rodents may not be capable of acting as human carcinogens. A study at the National Institute of Environmental Health Sciences compared another mouse liver tumorigen, oxazepam, with phenobarbital and concluded that "... if oxazepam-mediated carcinogenesis in mice occurs by the same mechanism as carcinogenesis mediated by phenobarbital as this study suggests may be a possibility, then oxazepam is not likely to be a human carcinogen." (Griffin et al., 1995Go)

Thyroid tumors.
Much like phenobarbital, toxaphene is capable of inducing thyroid follicular cell tumors in rats. The results of a study in which rats were treated with toxaphene for 28 days indicated that the compound caused an increase in thyroid-stimulating hormone (TSH) levels and follicular cell hyperplasia in the thyroid (Waritz et al., 1996Go). When administered at carcinogenic doses, many genotoxic compounds increase cell proliferation in their target organ (Cohen and Ellwein, 1990Go). Therefore, it is likely that toxaphene induces thyroid tumors like those produced by phenobarbital. The tumors arise through a disruption of the hypothalamic-pituitary-thyroid axis caused by increased metabolic inactivation of thyroid hormones (T3 and T4). The inactivation of thyroid hormones results in increased levels of TSH and sustained proliferation of cells in the thyroid and tumor formation (McClain, 1990Go). This is a threshold-exhibiting secondary mechanism, and humans are considerably less sensitive to this process than rats (McClain, 1994Go; EPA, 1996).

Mouse liver tumors.
Both the Litton and NCI bioassays utilized the mouse as their test system. However, this strain of mouse exhibits a particularly high sensitivity toward the development of liver tumors (Becker, 1982Go; Buchmann et al., 1991Go). A strong case has been made for viewing the mouse liver tumor response as irrelevant in human risk assessment (Counts et al., 1996Go; Counts and Goodman, 1995Go; Doull et al., 1983Go).

Kidney tumors.
An equivocal and statistically insignificant increase in kidney tumors in male rats exposed to toxaphene has been reported (NCI, 1979; W. M. Busey, unpublished report, 1994; J. A. Swenberg, unpublished report, 1995). However, the kidneys of the rats in this study also exhibited alpha-2u-globulin nephropathy (Swenberg, 1995). It is, therefore, likely that the rat kidney tumors resulted from a secondary mechanism that is irrelevant in human risk assessment.

DISCUSSION

We conclude that, based on current standards, the NCI bioassay of toxaphene in mice exceeded the maximum tolerated dose (MTD), and, even at those excessive doses, only a statistically significant increase in benign liver tumors was observed. Nevertheless, if a quantitative dose-response assessment for toxaphene is to be performed, the data from both the NCI and Litton cancer bioassays should be used in combination for that purpose. This will facilitate defining the dose-response curve over a broad dose range. While the diagnostic legitimacy of the lesions noted in the Litton study is questionable, it is likely that the results for total tumors (i.e., adenomas plus carcinomas) would be consistent with that originally reported, similar to the results from the expert PWG reevaluation of liver slides from the NCI study. Additionally, time-to-tumor data are available for both the NCI and Litton cancer bioassays and should be used in developing risk estimates. While an approach that incorporates time-to-tumor information ultimately may not materially alter the estimated risk, it allows a more efficient use of available information and should be applied whenever possible.

There is no scientifically defensible basis for combining the results of male and female mice by using a geometric mean. Large differences in background incidence of liver cancer between male and female mice further support selection of a sex-specific data set. The results from female mice, rather than male mice, are preferred as the basis for estimating potential human cancer risk because the background rate of liver cancer in the female mouse is lower and less variable than the rate seen in males.

The linear multistage model (LMS) for estimating potential human cancer risk is overly conservative, and consistent with EPA proposed revised guidelines (EPA, 1996), it is not appropriate for estimating potential human cancer risk associated with toxaphene. A benchmark dose approach consistent with EPA proposed cancer assessment guidelines (EPA, 1996) should be employed for a toxaphene risk assessment, and we have done this (Figs. 1–4GoGoGoGo). Applying this benchmark approach, the LED10 (or the lower 95% confidence limit on a dose associated with 10% extra risk) is calculated as 2.52 mg/kg/day for male mice and 5.05 mg/kg/day for female mice, when utilizing time-to-tumor data. The ED10 (or dose estimated to cause a 10% increase in response above the background frequency) is calculated as 3.32 mg/kg/day for male mice (Figure 3Go) and 6.44 mg/kg/day for females (Figure 4Go). The ED10 is the appropriate point of departure.



View larger version (30K):
[in this window]
[in a new window]
 
FIG. 1. Dose-response characteristics based on the liver adenoma and/or carcinoma incidence data in the combination of the Litton Bionetics study (1978), the NCI study (1979), and the pathology working group (PWG) findings: male mice, ignoring observation time data. Upper panel, dose-response characteristics ED10 and LED10. Lower panel, dose-response characteristics ED10, LED10, q, and q*. Interspecies extrapolation of cancer potency from mice to humans is based on the default assumption of interspecies equivalence on the (body-weight) scale and is not adjusted for interspecies differences in background transition rates. Using the standard default body weight of 0.03 kg for mice and 60 kg for humans, the current default assumption for the interspecies extrapolation of the cancer potency (q or q*) from mice to humans would imply that the human cancer potency is approximately 6.69 times greater than the mouse cancer potency because (60 kg/0.03 kg) = 6.687403.

 


View larger version (30K):
[in this window]
[in a new window]
 
FIG. 2. Dose-response characteristics based on the liver adenoma and/or carcinoma incidence data in the combination of the Litton Bionetics study (1978), the NCI study (1979), and the pathology working group (PWG) findings: female mice, ignoring observation time data. Upper panel, dose-response characteristics ED10 and LED10. Lower panel, dose-response characteristics ED10, LED10, q, and q*. Interspecies extrapolation of cancer potency from mice to humans is based on the default assumption of interspecies equivalence on the (body-weight) scale and is not adjusted for interspecies differences in background transition rates. Using the standard default body weight of 0.03 kg for mice and 60 kg for humans, the current default assumption for the interspecies extrapolation of the cancer potency (q or q*) from mice to humans would imply that the human cancer potency is approximately 6.69 times greater than the mouse cancer potency because (60 kg/0.03 kg) = 6.687403.

 


View larger version (25K):
[in this window]
[in a new window]
 
FIG. 3. Dose-response characteristics based on the liver adenoma and/or carcinoma incidence data in the combination of the Litton Bionetics study (1978), the NCI study (1979), and the pathology working group (PWG) findings: male mice, incorporating observation time data. Upper panel, dose-response characteristics ED10 and LED10. Lower panel, dose-response characteristics ED10, LED10, q, and q*. Interspecies extrapolation of cancer potency from mice to humans is based on the default assumption of interspecies equivalence on the (body-weight) scale and is not adjusted for interspecies differences in background transition rates. Using the standard default body weight of 0.03 kg for mice and 60 kg for humans, the current default assumption for the interspecies extrapolation of the cancer potency (q or q*) from mice to humans would imply that the human cancer potency is approximately 6.69 times greater than the mouse cancer potency because (60 kg/0.03 kg) = 6.687403.

 


View larger version (25K):
[in this window]
[in a new window]
 
FIG. 4. Dose-response characteristics based on the liver adenoma and/or carcinoma incidence data in the combination of the Litton Bionetics study (1978), the NCI study (1979), and the pathology working group (PWG) findings: female mice, incorporating observation time data. Upper panel, dose-response characteristics ED10 and LED10. Lower panel, dose-response characteristics ED10, LED10, q, and q*. Interspecies extrapolation of cancer potency from mice to humans is based on the default assumption of interspecies equivalence on the (body-weight) scale and is not adjusted for interspecies differences in background transition rates. Using the standard default body weight of 0.03 kg for mice and 60 kg for humans, the current default assumption for the interspecies extrapolation of the cancer potency (q or q*) from mice to humans would imply that the human cancer potency is approximately 6.69 times greater than the mouse cancer potency because (60 kg/0.03 kg) = 6.687403.

 
It should be recognized that EPA has received conflicting advice on what an appropriate point of departure should be. Advice given early in the development of the new cancer risk assessment guidelines encouraged EPA to use a central estimate (e.g., an ED01, ED05, or ED10.) Subsequent advice encouraged use of a lower confidence bound estimate (e.g., an LED01, LED05, or LED10). In some circumstances, there can be marked differences between these two types of point of departure estimates. However, an LED is always smaller than its associated ED, and usually smaller than the NOAEL that could be established with the same data set. Thus, use of LEDs as points of departure tends to introduce additional and excessive conservatism into the process of establishing a safe human exposure level, and it does so without stating it explicitly. Therefore, the ED10 is the appropriate point of departure. This involves less mathematical manipulation of the data beyond the range of observation. It places the point of departure closer to the real data, providing a more transparent and credible risk assessment.

In light of the points discussed above, we conclude that the ED10 for female mice is the most appropriate point of departure, and either a linear extrapolation or a margin of exposure (MOE) calculation can be used. The qualitative data for toxaphene justify using an MOE approach based on questionable evidence of carcinogenicity; lack of genotoxicity in vivo; the dubious relevance of the end point, mouse liver tumors in human risk assessment; and the review of the toxaphene mechanistic data discussed above. However, the mechanistic database is not sufficient to fully support the use of an MOE approach that presumes a threshold for tumor response. Therefore, at this time, the panel concludes that applying a linear approach (i.e, extrapolation from the ED10 to the origin) is an acceptable means of estimating the dose-response relationship at low doses. This represents a highly conservative approach.

For the sake of comparison, a series of figures (Figs. 1–4GoGoGoGo) are presented. The use of a three-fourths power body weight scaling factor, as recommended in the proposed EPA cancer risk assessment guidelines (EPA, 1996), has been incorporated into these calculations. It should be noted that this represents a highly conservative approach, given the fact that the development of liver tumors is a very sensitive endpoint in B6C3F1 mice. It is likely that mice are more sensitive than humans and that an interspecies scaling factor proportional to body-weight would generate a very conservative estimate. Figures 1–4GoGoGoGo depict dose-response curves based on data from male and female mice both with and without adjustment for time to tumor data, and combining the results of the Litton and NCI studies. In our view, the dose-response curve for female mice, with the time to tumor adjustment is the most appropriate in estimating human risk in a conservative fashion. The slope of this curve (or Q1) is 0.10 (mg/kg/day)–1 based on an ED10 of 6.44 mg/kg/day (Figure 4Go) and the panel recommends that this be used as the cancer potency factor for toxaphene.

ACKNOWLEDGMENTS

The authors would like to acknowledge the invaluable contributions that Dr. Judith W. Hauswirth and Dr. Robert L. Sielken, Jr., made to the development and completion of this manuscript.

NOTES

This manuscript is based on a report prepared for Hercules Incorporated, 1313 North Market Street, Wilmington, Delaware 19894.

1 To whom correspondence should be addressed at the Department of Pharmacology and Toxicology, Michigan State University, B-440 Life Sciences Building, E. Lansing, MI 48824. Fax: 517–353–8915. E–mail: goodman3{at}pilot.msu.edu. Back

2 Present address: Blasland, Bouck, and Lee, Inc., Reston, VA. Back

REFERENCES

Becker, F. F. (1982). Morphological classification of mouse liver tumors based on biological characteristics. Cancer Res. 42, 3918–3923.[Abstract]

Buchmann, A., Bauer-Hofmann, R., Mahr, J., Drinkwater, N. R., Luz, A., and Schwarz, M. (1991). Mutational activation of the c-Ha-ras gene in liver tumors of different rodent strains: correlation with susceptibility to hepatocarcinogenesis. Proc. Natl. Acad. Sci. U S A 88, 911–915.[Abstract]

Clemmesen, J., and Hjalgrim-Jensen, S. (1978). Is phenobarbital carcinogenic? A follow-up of 8076 epileptics. Ecotoxicol. Environ. Saf. 1, 457–470.[ISI][Medline]

Clemmesen, J., and Hjalgrim-Jensen, S. (1980). Epidemiological studies of medically used drugs. Arch. Toxicol. Suppl 3, 19–25.

Cohen, S. M., and Ellwein, L. B. (1990). Cell proliferation in carcinogenesis. Science 249, 1007–1011.[ISI][Medline]

Counts, J. L., and Goodman, J. L. (1995). Hypomethylation of DNA: A possible epigenetic mechanism involved in tumor promotion." In Growth Factors and Tumor Promotion: Implications for Risk Assessment (R. M. McClain, T. J. Slaga, R. LeBoeuf, and H. Pitot, Eds.), pp. 81–101. Wiley-Liss, Inc., New York.

Counts, J. L., Sarmiento, J. I., Harbison, M. L., Downing, J. C., McClain, R. M., and Goodman, J. L. (1996). Cell proliferation and global methylation status changes in mouse liver after phenobarbital and/or choline-devoid, methionine-deficient diet administration. Carcinogenesis 17, 1251–1257.[Abstract]

de Geus, H. J., Besselink, H., Brouwer, A., Klungsoyr, J., McHugh, B., Nixon, E., Rimkus, G. G., Wester, P. G., and de Boer, J. (1999). Environmental occurrence, analysis, and toxicology of toxaphene compounds. Environ. Health Perspect. 107(Suppl 1), 115–144.

Doull, J., Bridges, B. A., Kroes, R., Golberg, L., Munro, I. C., Paynter, O. E., Pitot, H. C., Squire, R., Williams, G. M., and Darby, W. J. (1983). The relevance of mouse liver hepatoma to human risk: A report of the International Expert Advisory Committee to the Nutrition Foundation. The Nutrition Foundation, Inc., Washington, DC.

Epstein, S. S., Arnold, E., Andrea, J., Bass, W., and Bishop, Y. (1972). Detection of chemical mutagens by the dominant lethal assay in the mouse. Toxicol. Appl. Pharmacol. 23, 288–325.[ISI][Medline]

Frith, C. H., and Ward, J. M. (1979). A morphologic classification of proliferative and neoplastic hepatic lesions in mice. J. Environ. Pathol. Toxicol. 3, 329–351.[ISI][Medline]

Goldsworthy, T., Campbell, H. A., and Pitot, H. C. (1984). The natural history and dose-response characteristics of enzyme-altered foci in rat liver following phenobarbital and diethylnitrosamine administration. Carcinogenesis 5, 67–71.[Abstract]

Griffin, D. E. III, and Hill, W. E. (1978). In vitro breakage of plasmid DNA by mutagens and pesticides. Mutat. Res. 52, 161–169.[ISI][Medline]

Griffin, R. J., Dudley, C. N., and Cunningham, M. L. (1995). Biochemical effects of the mouse hepatocarcinogen oxazepam: similarities to phenobarbital. Fundam. Appl. Toxicol. 29, 147–154.[ISI]

Haseman, J. K., Winbush, J. S., and O'Donnell, M. W., Jr. (1986). Use of dual control groups to estimate false positive rates in laboratory animal carcinogenicity studies. Fundam. Appl. Toxicol. 7, 573–584.[ISI][Medline]

Hedli, C. C, Snyder, R., Kinoshita, F. K., and Steinberg, M. (1998). Investigation of hepatic cytochrome P450 enzyme induction and DNA adduct formation in male CD/1 mice following oral administration of toxaphene. J. Appl. Toxicol. 18, 173–178.[ISI][Medline]

Hooper, N. K., Ames, B. N., Saleh, M. A., and Casida, J. E. (1979). Toxaphene, a complex mixture of polychloroterpenes and a major insecticide, is mutagenic. Science 205, 591–593.[ISI][Medline]

Houk, V. S., and DeMarini, D. M. (1987). Induction of prophage lambda by chlorinated pesticides. Mutat. Res. 182, 193–201.[ISI][Medline]

Maekawa, A., Onodera, H., Ogasawara, H., Matsushima, Y., Mitsumori, K., and Hayashi, Y. (1992). Threshold dose dependence in phenobarbital promotion of rat hepatocarcinogenesis initiated by diethylnitrosamine. Carcinogenesis 13, 501–503.[Abstract]

Maronpot, R. R., Montgomery, C. A., Jr., Boorman, G. A., and McConnell, E. E. (1986). National Toxicology Program nomenclature for hepatoproliferative lesions of rats. Toxicol. Pathol. 14, 263–73.[Medline]

McClain, R. M. (1990). Mouse liver tumors and microsomal enzyme-inducing drugs: experimental and clinical perspective with phenobarbital. Prog. Clin. Biol. Res. 331, 345–365.[Medline]

McClain, R. M. (1994). Mechanistic considerations in the regulation and classification of chemical carcinogens. In Nutritional Toxicology (F. N. Kotsonis, M. Mackey and J. Hjelle, Eds.), pp. 273–304. New York, Raven Press,

Mortelmans, K., Haworth, S., Lawlor, T., Speck, W., Tainer, B., and Zeiger, E. (1986). Salmonella mutagenicity tests: II. Results from the testing of 270 chemicals. Environ. Mutagen. 8(Suppl. 7), 1–119.

National Cancer Institute (1979). Bioassay of Toxaphene of Possible Carcinogenicity. Technical Report Series No. 37.

Olsen, J. H., Schulgen, G., Boice, J. D., Jr., Whysner, J., Travis, L. B., Williams, G. M., Johnson, F. B., and McGee, J. O. (1995). Antiepileptic treatment and risk for hepatobiliary cancer and malignant lymphoma. Cancer Res. 55, 294–297.[Abstract]

Olsen, J. H., Wallin, H., Boice, J. D., Jr., Rask, K., Schulgen, G., and Fraumeni, J. F., Jr. (1993). Phenobarbital, drug metabolism, and human cancer. Cancer Epidemiol. Biomarkers. Prev. 2, 449–452.[Abstract]

Reddy, M. V., and Randerath, K. (1986). Nuclease P1-mediated enhancement of sensitivity of 32P-postlabeling test for structurally diverse DNA adducts. Carcinogenesis 7, 1543–1551.[Abstract]

Samosh, L. V. (1974). Chromosome aberration and the character of satellite associations following accidental exposure of the human body to polychlorocamphene. Tsitol. Genet. 8, 24–27.[Medline]

Sobti, R. C., Krishan, A., and Davies, J. (1983). Cytokinetic and cytogenetic effect of agricultural chemicals on human lymphoid cells in vitro: II. Organochlorine pesticides. Arch. Toxicol. 52, 221–231.[ISI][Medline]

Steinel, H. H., Arlauskas, A., and Baker, R. S. (1990). SCE induction and cell-cycle delay by toxaphene. Mutat. Res. 230, 29–33.[ISI][Medline]

Tarone, R. E., Chu, K. C., and Ward, J. M. (1981). Variability in the rates of some common naturally occurring tumors in Fischer 344 rats and (C57BL/6NxC3H/HeN) F1 (B6C3F1) mice. J. Natl. Cancer Inst. 66, 1175–1181.[ISI][Medline]

Trosko, J. E., Jone, C., and Chang, C. C. (1987). Inhibition of gap junctional-mediated intercellular communication in vitro by aldrin, dieldrin and toxaphene: a possible cellular mechanism for their tumor-promoting and neurotoxic effects. Mol. Toxicol. 1, 83–93.[Medline]

U.S. Environmental Protection Agency (1978). Occupational exposure to toxaphene. A Final Report by the Epidemiologic Studies Program. Human Health Effects Monitoring Branch, Benefits and Field Studies Division, OPP, OTS, EPA.

U.S. Environmental Protection Agency (1996a). Proposed Guidelines for Carcinogen Risk Assessment. Fed. Register 61, 17960–18011.

U.S. Environmental Protection Agency (1996b). Assessment of thyroid follicular cell tumors. A draft report prepared by the Risk Assessment Forum.

Vesselinovitch, S. D., and Mihailovitch, N. (1983). Kinetics of diethylnitrosamine hepatocarcinogenesis in the infant mouse. Cancer Res. 43, 4253–4259.[Abstract]

Ward, J. M. (1996). Can so-called nongenotoxic hepatocarcinogens induce mouse liver neoplasms? Toxicol. Pathol. 24, 505.[ISI][Medline]

Ward, J. M., Goodman, D. G., Squire, R. A., Chu, K. C., and Linhart, M. S. (1979). Neoplastic and non-neoplastic lesions in aging (C57BL/6NxC3H/HeN) F1 (B6C3F1) mice. J. Natl. Cancer Inst., 63, 849–854.[ISI][Medline]

Ward, J. M., Shibata, M. A., and Devor, D. E. (1996). Emerging issues in mouse liver carcinogenesis. Toxicol. Pathol. 24, 129–137.[ISI][Medline]

Waritz, R. S., Steinberg, M., Kinoshita, F. K., Kelly, C. M., and Richter, W. R. (1996). Thyroid function and thyroid tumors in toxaphene-treated rats. Regul. Toxicol. Pharmacol. 24, 184–192.[ISI][Medline]

Whysner, J., Montandon, F., McClain, R. M., Downing, J., Verna, L. K., Steward, R. E. III, and Williams, G. M. (1998). Absence of DNA adduct formation by phenobarbital, polychlorinated biphenyls, and chlordane in mouse liver using the 32P-postlabeling assay. Toxicol. Appl. Pharmacol. 148, 14–23.[ISI][Medline]

Whysner, J., Ross, P. M., and Williams, G. M. (1996). Phenobarbital mechanistic data and risk assessment: enzyme induction, enhanced cell proliferation, and tumor promotion. Pharmacol. Ther. 71, 153–191.[ISI][Medline]





This Article
Abstract
FREE Full Text (PDF)
Alert me when this article is cited
Alert me if a correction is posted
Services
Email this article to a friend
Similar articles in this journal
Similar articles in PubMed
Alert me to new issues of the journal
Add to My Personal Archive
Download to citation manager
Search for citing articles in:
ISI Web of Science (2)
Disclaimer
Request Permissions
Google Scholar
Articles by Goodman, J. I.
Articles by Starr, T. B.
PubMed
PubMed Citation
Articles by Goodman, J. I.
Articles by Starr, T. B.