Comparative results of 327 chemical carcinogenicity studies.

The National Cancer Institute (NCI) and the National Toxicology Program (NTP) have carried out a number of laboratory animal carcinogenicity studies and presented the results of these experiments in a series of Technical Reports. This paper tabulates the results of the 327 NCI/NTP studies carried out to date on 308 distinct chemicals, and discusses certain issues relevant to the evaluation of carcinogenicity in these experiments. This compilation of results from NCI/NTP carcinogenicity experiments provides a large database that can be used to study structure-activity correlations, interspecies concordance, and associations between laboratory animal carcinogenicity and other toxicological effects.


Introduction
The National Cancer Institute (NCI) and the National Toxicology Program (NTP) have designed, carried out, and evaluated more than 300 long-term chemical car-cinogenicity<studies in laboratory rodents. The majority of these studies involve four sex-species experiments: male and female rats and mice. The strains most commonly used are Fischer 344 rats and B6C3F1 mice; other animals occasionally used include Osborne-Mendel and Sprague-Dawley rats, Syrian golden hamsters, and ICR Swiss, Swiss-Webster, and Swiss CD-1 mice. In most of these studies the chemical was administered for 2 years, although certain NCI mouse experiments were of a shorter duration. The results of these NCI/NTP studies have been presented in a series of Technical Reports (TRs), and in addition are often published in the scientific literature. These data are utilized by the international scientific community and by various government agencies in making regulatory decisions affecting public health. The objective of this paper is to provide a tabulated compilation of the results of these NCI/NTP laboratory animal carcinogenicity studies.

Materials and Methods
This compilation of results covers ali chemicals studied for long-term toxicity and carcinogenicity by the NCI or by the NTP and reported in the Technical Report series. Included are all studies that have been approved by the NCI Clearinghouse or by the NTP Board of Scientific Counselors' Peer Review Panel (established in June, 1980) through June 1, 1987. A total of 327 studies (1237 individual sex-species experiments) have been evaluated, involving 308 distinct chemicals. Seventeen chemicals were studied twice (trichloroethylene three times) by different laboratories, in different species or strains, and/or by different routes of administration.
The numbering of the Technical Reports contains gaps that correspond to chemicals for which Technical Reports were originally intended but never issued. Also, certain Technical Report numbers were assigned to guidelines and other documents that do not report results of specific studies. Technical Reports that were not printed or  For experiments evaluated by the NCI or the NTP prior to in June, 1983, results are reported in this paper as "positive," "negative," "equivocal," or "inadequate." In June, 1983, the NTP adopted the use of "categories of evidence" (12)(13)(14), which were used to classify study results evaluated after that date. This approach places the result of each individual sex-species experiment into one of five categories: two correspond to positive results ("clear evidence" or "some evidence" of carcinogenicity), one is for uncertain findings ("equivocal evidence"), one is for negative studies ("no evidence"), and one is for studies that cannot be evaluated because of major flaws ("inadequate studies").
Nitrilotriacetic acid (NTA) Nitrilotriacetic acid, Na3 H20 (study 1) Nitrilotriacetic acid, Na3 H20 (study 2) 5-Nitroacenaphthene 3-Nitro-p-acetophenetide   We did not attempt to reclassify the earlier experiments in terms of these categories. A chemical was considered to be a carcinogen if it produced a carcinogenic response in at least one sex-species group. Carcinogenicity results are reported separately for the 202 studies evaluated by the NCI (Table 1) and the 125 evaluated by NTP (Table 2).
The distribution of carcinogenicity results was similar in NCI and NTP studies. Among the NCI carcinogenicity studies ( Of the 120 studies showing no chemically induced neoplasia, 95 were judged negative in all four sex-species groups, 3 were negative in three groups (the fourth being inadequate), 4 were negative in two groups (with two inadequate experiments), and 18 were studied in one species and found to be noncarcinogenic in both sexes.
Of the 159 studies with carcinogenic effects, 38 were positive in all four sex-species groups, 24 were carcinogenic in three, 60 were carcinogenic in two (including 9 that were studied in one species), and 37 were positive in one sex-species group. Only 14 of the 37 "one-sexspecies-positive" chemicals were negative in the other three sex species groups: 17 showed equivocal effects in at least one of the other groups; 4 were inadequately studied in at least one of the other groups; and the remaining 2 were studied in one species.  Rats and mice showed a high concordance with regard to carcinogenicity outcome. This association is summarized in Table 3 for the 266 chemicals that were adequately studied in both sexes of both species. The concordance in response between rats and mice (with equivocal results considered negative) was 75% (138/ 183) for feeding studies, 66% (41/62) for gavage studies and 90% (19/21) for all other routes of administration.
If equivocal studies are considered negative, then 67 chemicals showed carcinogenic effects in a least one sex of both species; 131 chemicals showed no carcinogenic effects in any of the four sex-species groups; 32 chemicals were carcinogenic in rats (males, females, or both) but not in mice; and 36 were carcinogenic in mice but not in rats. As shown in Table 3, the concordance among species is similar, regardless of how the equivocal study results are considered.
If individual sex-species groups are compared, then the overall concordance in carcinogenic response between sexes is quite high for both rats (255/292, 87%) and mice (255/288, 89%). For the four interspecies comparisons, the lowest concordance is that observed between male rats and male mice (191/270, 71%), the highest concordance is between female rats and female mice (213/275, 77%).
The interspecies concordance in carcinogenic response for the NCI/NTP studies is similar to that reported by Purchase (15) in a literature-based evaluation of 250 carcinogenicity experiments (which included some of the NCI studies considered in our analysis). He reported 82% concordance among rats and mice: 38% of the chemicals were not carcinogenic in either species, and 44% were carcinogenic in both species.

Discussion
Although we present a list of conclusions regarding the carcinogenicity of a series of chemicals in rats and mice as positive, negative, equivocal, or inadequate, we recognize that these are categories whose boundaries are not clearly defined. These categories are designed to encompass a spectrum of responses because each carcinogenesis study produces a unique set of results. While these categories are useful in providing a general indication of a chemical's carcinogenicity, as well as providing a certain comparability across studies, they should never be used as a substitute for a more detailed evaluation ofthe study design, data analysis, and results as presented in the Technical Reports. While the reader is encouraged to consult the full Technical Report for more detailed evaluations, it must also be kept in mind that there have been significant advances in chemical carcinogenesis in recent years. These include increased knowledge about specific organ or tissue tumor responses, more refined and uniform histopathologic diagnoses, use of survival-adjusted statistical methods, extensive data on historical control tumor incidences, and increased understanding of biological/toxicological mechanisms of chemically induced neoplasia. These factors all have an impact on current evaluations of experimental results.
For the majority of the NCI/NTP studies, the summary conclusions given in the Technical Reports were unambiguous. In some cases the particular wording used for the conclusion made it difficult to place a result in the most appropriate classification in terms of "positive,"' "negative," or "equivocal." Perhaps for these reasons, previous summaries (1)(2)(3) of certain of these carcinogenicity studies are not in complete agreement regarding the overall interpretation of experimental results.
We have indicated by a superscript in Tables 1 and 2 those experimental results that were considered to be particularly difficult to classify based on the wording of the Technical Report summaries. For example, for some chemicals the conclusion in the Technical Report was "not carcinogenic," but to this evaluation was added the notation that increased incidences of certain tumors "may have been related to" or "may have been associated with" chemical exposure. One interpretation could be that the intended conclusion was "negative," and that the additional information was provided to indicate effects that had been considered, but perhaps discounted as not being biologically important. Another interpretation could be that the intended conclusion was "not positive," and that the additional information was provided to convey findings that were considered "less than positive," but not fully negative, i.e., "equivocal." In our evaluation, the latter interpretation was adopted. Chemicals for which this type of language was used in the Technical Reports included 11-aminoundecanoic acid (MM), D & C red 9 (FR), dieldrin (MM), ethyl tellurac (MR, MM, FM), propyl gallate (MR, MM), stannous chloride (MR), 1,1,1,2-tetrachloroethane (MR), and the dermal study of 2,3,7,8-tetrachlorodibenzo-p-dioxin (MM).
For example, previously published evaluations of dieldrin exposure to male mice (based on the data in NCI Technical Report 21) range from no evidence of carcinogenicity (3) to evidence suggestive of a carcinogenic effect (1) to carcinogenic (2). NCI Technical Report 21 concluded that an increased incidence of hepatocellular carcinoma "may be associated with treatment," and we regarded this as an equivocal response (Table 1). For the tabulated results given in Tables 1 and 2, we relied upon the conclusions given in the individual Technical Reports, but we recognize that in some instances alternative interpretations of these conclusions are possible. Approximately 50% of all chemicals evaluated for carcinogenicity in rodents by the NCI/NTP gave positive results in at least one sex-species group. This agrees with earlier findings (3,4). However, this percentage may be misleading, as it does not differentiate between a chemical producing a single-site carcinogenic response in only one sex-species group and a chemical showing multiple organ effects in all four sex-species experiments.
Thus, a weight of the evidence approach must be used when considering potential hazards to humans. For example, the 38 chemicals that were positive in all four sex-species groups should perhaps receive the highest priority with regard to comprehensive epidemiologic studies, as well as increased public health consciousness. Again, the full experimental results on a chemical must be considered and evaluated before deciding on a course of public health action.
The concordance in carcinogenic response found between rats and mice in the NCI/NTP data was 74%. Despite this high concordance, however, we believe that both sexes of two rodent species should continue to be used, in most studies, to determine the long-term toxicology and carcinogenesis effects of chemical exposures. Although some investigators feel that this high concordance implies that the mouse is redundant and should not be used in determining the carcinogenicity of chemicals (16), most national and international scientific guidelines for laboratory animal carcinogenicity studies (17)(18)(19) recommend that at least two species be used. Further, for the NCI/NTP studies the similarity in carcinogenic response between sexes within a species was greater than the redundancy across species.
We are hopeful that this tabulation of chemical carcinogenesis results from all NCI/NTP studies carried out to-date will stimulate more in depth review of the actual data in the NCI/NTP Technical Reports that led to the abbreviated results shown in Tables 1 and 2.