Summary of carcinogenic potency and positivity for 492 rodent carcinogens in the carcinogenic potency database.

A tabulation of carcinogenic potency (TD50) by species for 492 chemicals that induce tumors in rats or mice is presented. With the use of the Carcinogenic Potency Database, experimental results are summarized by indicating in which sex-species groups the chemical was tested and the respective evaluations of carcinogenicity. A comparison of three summary measures of TD50 for chemicals with more than one positive experiment per species shows that the most potent TD50 value is similar to measures that average values or functions of values. This tabulation can be used to investigate associations between rodent potency and other factors such as mutagenicity, teratogenicity, chemical structure, and human exposure.


Introduction
For a variety of purposes it is desirable to have a summary measure of the carcinogenic potency of a chemical in rodents. A single value is needed, for example, to summarize the chronic toxicity of a chemical, to estimate carcinogenic hazards to humans by a comparison of rodent potency and human exposure (1,2), or to compare results of short-term tests with those of rodent bioassays (3). For several years we have been developing the Carcinogenic Potency Database (CPDB), a compilation of the results of chronic rodent cancer tests and the potency values derived from those results (4)(5)(6). The CPDB contains data on approximately 3700 experiments of 975 compounds, including tests from the National Cancer Institute and National Toxicology Program (NCI/ NTP), as well as from the general published literature. For a given compound, the database may include experiments in both rats and mice, males and females, a variety of strains or routes of administration, and a variety of doses and experimental conditions; alternatively, for a different substance there may be only one experiment conducted in a single sex-species group. In order to construct a chemical-by-chemical list of carcinogenic potency in each species, some *Biology and Medicine Division, Lawrence Berkeley Laboratory, Berkeley, CA 94720. tDepartment of Preventive Medicine, University of Southern California, School of Medicine, Los Angeles, CA 90033. method must be selected to summarize the potency of a carcinogen when there is more than one potency estimate for it. Additional information to summarize the experimental results includes the number of sexspecies groups tested and the evaluations of carcinogenicity in each group.
In this paper we present a table summarizing carcinogenic potency in rats and mice for the 492 chemicals that have positive results in the CPDB; we also indicate in which sex-species groups the chemical was tested and the respective evaluations of carcinogenicity. Our intent is that this summary compilation will be a useful reference source for the scientific and regulatory communities, and that it will facilitate the use of our larger published plots of the CPDB. The larger plots provide detailed information on each experiment including the species, sex, strain, route of administration, duration of exposure and of experiment, dose levels, target sites, shape of the dose response, estimates of carcinogenic potency and the confidence limits surrounding it, statistical significance of the carcinogenic dose response, tumor incidences, and bibliographic citation to the published paper or to the NCI/NTP Technical Report.

Methods
Our analyses are based on the chemicals reported in the CPDB (4)(5)(6), which has been fully described in Gold et al. (4) as to bioassay inclusion criteria, protocol characteristics and derived variables. The database is organized by chemical using a plot format and includes bioassay results from two sources: papers published in the general literature through 1984, and NCI/NTP Technical Reports published through May 1986. All experiments in the database meet a specific set of inclusion criteria that were designed to permit the estimation of carcinogenic potency; therefore, reasonable consistency of experimental protocols is assured. Rodent bioassays are included in the database only if the test agent was administered alone rather than in combination with other substances; if the bioassay included a control group; if the route of administration was either diet, water, gavage, inhalation, IV injection or IP injection; and if the length of experiment in rodents was at least 1 year with dosing for at least 6 months. For the CPDB, we do not evaluate the evidence for carcinogenicity in an experiment; rather, we report the evaluation of the published author and calculate the statistical significance of the tumorigenic dose response in the experiment.

Carcinogenicity
Our tabular compilation of results by sex-species group and of carcinogenic potency by species is restricted to chemicals identified as carcinogens in our database. We classify the results of an experiment as either positive or negative on the basis of the author's opinion in the published paper, and we include in the present publication only those chemicals that have been evaluated as positive by the author of at least one experiment. In some cases authors do not clearly state their evaluation, and in some NCI/NTP Technical Reports the evidence for carcinogenicity is considered only suggestive; in our analyses we consider these cases as lacking clear evidence of carcinogenicity and do not use them to identify a chemical as positive. We use the author's opinion to determine positivity because it often takes into account more information than statistical significance alone, such as historical control rates for particular sites, survival and latency, and/or dose response. Generally, this designation by author's opinion corresponds well with the results of statistical tests for the significance of the dose-response effect (4,7).
In our tabular compilation of positivity by sexspecies group for each of the 492 chemicals classified as positive, we indicate whether the compound was tested in each group and list the strongest level of evidence for carcinogenicity based upon any author's evaluation in either the general literature or the NCI/NTP. The strongest evaluation is clear evidence of carcinogenicity (+). When there was no such evaluation in one of the sex-species groups, but the compound was tested by NCI/NTP and their evaluation was stronger than "no evidence of carcinogenicity" (-), we indicate whether that NCI/NTP evaluation was "some evidence of carcinogenicity" (A), "equivocal" (E) or "in-adequate bioassay" (I). These evaluations correspond to the opinions reported in our published plots; we note that in a few borderline cases our interpretation of the Technical Reports differs from those recently tabulated by Haseman et al. (8).

Carcinogenic Potency Values
In our analyses of carcinogenic potency we use our numerical index, the TD50, which has been fully described in Sawyer et al. (9) and in Peto et al. (10), and which is reported for each target site in our published plots. Briefly, TD50 may be defined as follows: For a given target site(s), TD50 is the chronic dose rate in milligrams per kilogram body weight per day that would induce tumors in half the test animals at the end of a standard lifespan for the species in the absence of tumors in control animals. Since the tumor(s) of interest often occurs in control animals, TD50 is more precisely defined as the chronic dose rate that will halve the probability of remaining tumor-free throughout the standard lifespan of the species (9,10).
For bioassays from the NCI/NTP program, TD50 values are estimated from li-fetable data (11) and are adjusted for the differential effects of toxicity among dose groups and for differences in the time pattern of tumor incidence. For experiments from the general literature, TD50 values are based on summary incidence data [for a comparison of methods and TD50 values see (9,11)]. The range of TD50 values for carcinogens in the CPDB is more than 10 million-fold.
For the purposes of this summary, in any given positive experiment we select the lowest TD50 value from among positively evaluated target sites with a statistically significant dose response (two-tailed p < 0.1). If no positive sites have a significant dose response, then we select the most potent (lowest TD50) from among positively evaluated sites with p . 0.1. This method provides a single TD50 to represent an experiment. For approximately one-half of the carcinogens in rats and one-third of the carcinogens in mice, the CPDB includes only one positive experiment. For chemicals with more than one positive test, we summarize potency in a species by selecting the lowest significant TD50 value from among those representing each experiment. If none is significant, the lowest is chosen from among these nonsignificant values. In some experiments, no TD50 could be estimated because all animals in the only dose group had the tumor of interest, and the only data available were for crude percentages of animals with a tumor (4). For these cases we use the 99% upper confidence limit of TD50 as a replacement for the TD50.

Results
In Table 1                  For each species, the reported TD50 value is the most potent in the CPDB from among sites that a published author evaluated as positive. This TD-0 was selected from those with a statistically significant dose response effect (two-tailed p<O.1). If no site evaluated as positive was significant, then the reported TDr() was the most potent among those positive sites, and the footnote "b" indicates that p(t).l1 .
Abbre% iations: CAS = Chemical Abstracts Serxvice registry number NT = No Test in the CPDB in this group + = The CPDB contains at least one experiment in which the compound was evaluated as a carcinogen by the published author. Foi NCI/NTP tests, the evaluation was "clear evidence of carcinogenicity." I = No tests in the CPDB in this sex-species group were evaluated as positive, however, the NCI/NTP test was evaluated as inadequate. A = 'I'he chemical was evaluated as a carcinogen in at least one test in the CPDB, but not in this sex-species group. However, the NCI evaluated the compounId in this sex-species group as "associated with carcinogenicity" in their test, or the NTP evaluated it as having "some evidence of carcinogenicity." E The chemical was evaluated as a carcinogen in at least one test in the CPDB, but not in this sex-species group. The only evidence for carcinogenicity in the CPDB for this sex-species group was an "equivocal" evaluation by the NTP. -= The chemical was evaluated as a carcinogen in at least one test in the CPDB but all test.s in this group were negative. B+ = In the only positive test in the sex-species, results were reported only for males and f'emales combined. B-= In the only test in the sex-species, results were reported only for males and females combined, and the test was negative.
Footnotes: a = The ClPDB contains more than one positive test in the species. h = The reported TD50 is nlOt statistically significant (i.e. p>O.l), and all positivelv evaluated results in the species are not signif'icanit. c = Gnly an upper bound and no TD50 could be estimated because all dosed animals had the tumor of interest and only summary data were available. The reported value is the 99%O upper confidence limit. d = All positive results in the CPDB are from tests in which the compound was administered by either intraperitoneal or intravenous injection. e = The reported TD50 is from a test in which the compound was administered by intraperitoneal or intravenous injection; however, the CPDB also contaiins a positive test with a less potent TD50 value from a test with an oral or inhalation rout e. = TD5() values from different significant, positive experiments in this species vary by more than ten-fold f'rom one another.
The most potent TD-0 value is reported here.
26 '9 GOLD, SLONE, AND BERNSTEIN chemicals among the 955 chemicals in the CPDB that were tested in rats or mice. The table provides information for each substance on the most potent TD50 value in each species, the strength of evidence for carcinogenicity in each sex-species group, and the Chemical Abstracts Service (CAS) registry number. Forty-six percent of these chemicals were tested in both rats and mice, 35% in rats only, and 19% in mice only. The positivity results for the 492 compounds can be summarized as follows: 342 are positive in rats, and 278 are positive in mice. Among the 228 carcinogens tested in both rats and mice, 100 (44%) are positive in only one species. One hundred sixty-two chemicals were tested in all four sex-species groups, and 52 (32%) of these are positive in all four. Only 133 (27%) of the carcinogens listed in Table 1 were tested in the NCI/NTP Bioassay Program. A detailed analysis of positivity and target sites in the CPDB can be found in Gold et al. (7).
The distribution of TD50 values in Table 1 is summarized in Figure 1  TP . TH < TG < TA Table 2 shows the distribution of values of the ratios TH/Tp, TG/Tp, and TA/Tp for rats and mice for chemicals that have more than one positive test in a species; Table  2 therefore includes one-half of the rat carcinogens in Table 1 and two-thirds of those in mice. The remaining chemicals in Table 1 have only one positive test. Those compounds for which we obtained a 10-fold or greater discrepancy in these ratios are listed in the footnote to

Chemicals with Widely Varying TD50 Estimates
To further indicate particular substances for which TD50 values from two or more experiments differ greatly, we denote in Table 1 those chemicals for which the minimum estimate differs from the maximum estimate by more than a factor of 10 (see footnote f to Table 1). There are 18 such substances in rats and 12 in mice. For these carcinogens, any summary measure of While noting that the number of such cases is small, we have investigated possible reasons for these widely differing potency estimates for a chemical within a species. We first compared the frequency of estimates varying more than 10-fold to the frequency observed for near-replicate experiments (Table 3). In an earlier paper (6), we examined reproducibility of results in 66 comparisons consisting of 2 or more bioassays of the same chemical administered by the same route and using the same sex and strain of rat or mouse. Here we update this analysis to include 35 additional comparisons from our more recent plots of the CPDB (5,6). Overall, there was good reproducibility of positivity: among the 101 comparisons, 51 were concordant and positive in all of the near-replicate tests, 35 were concordant and negative, and 15 were discordant. For each species, we took the ratio of the least potent TD50 and the most potent TD50 in each concordant-positive near-replicate case, and we compared the distribution of these ratios to the distribution of all chemicals having more than one positive experiment, i.e., those chemicals reported in Table 2. (The chemicals with near-replicate tests are also included in the larger distribution, but the TD50 values for those substances may be from different experiments.) Table 3 indicates that the distribution of the ratio of least to most potent TD50 values for all chemicals is similar to that for the near-replicate comparisons. This similarity suggests that the discrepant results for a chemical within a species are not an artifact of combining across strains, routes of administration, and sexes.
We also compared the extreme cases (TD50 values discrepant by > 10-fold) to all other chemicals in Table 1 with more than one positive experiment, in terms of how often they were tested. In mice, 92% of the extreme cases were tested more than twice compared to 33% of all other cases; in rats, 89% of extremes were tested more than twice compared to 42% of others. There are similar differences in the number of positive tests for the two groups. Thus, when there are extreme differences between TD50 values from different tests of the same chemical, the selection of the least and most potent Table 3. Ratio of least to most potent TD50 from different positive experiments for near-replicate comparisons and all chemicals with more than one positive experiment, by species. Legend: H = Ratio of harmonic mean (TH) to most potent TD50 (Tp) G = Ratio of geometric mean (TG) to most potent TD50 (Tp) A = Ratio of arithmetic mean (TA) to most potent TD50 (Tp) aChemicals with values of ratio greater than 10. Those followed by " were in the extreme category for both arithmetic and geometric means. Finally, we investigated whether differences between the dose levels administered in the comparison tests were greater for the extreme cases. Generally, within a species the doses in different tests of the same chemical are quite similar. We computed the ratio of maximum doses tested in the experiments that yielded the minimum and maximum TD50 values. The median of this ratio for the cases that were not extreme (i.e., differed by less than a factor of 10) was 1.09 for mice and 1.25 for rats. We found that three-quarters of the extreme cases with a ratio > 10 were above this median in each species. This result is not surprising since generally the TD50 value is restricted by the maximum dose tested in a bioassay, i.e., a statistically significant TD50 cannot be very far from the administered dose, given the usual experimental design (13).