Apparent correlation between structure and carcinogenicity of phenylenediamines and related compounds.

The carcinogenicity of 23 phenylenediamines and related compounds was reviewed. An extensive analysis of the methods used indicated that the bioassays were conducted well. The data suggest that the carcinogenicity of 4-substituted 1,3-phenylenediamines is reduced substantially or eliminated completely by oxidation of one or both amine groups or by N-substitution. Oxidation of a methyl substituent on nitroaniline to a carboxyl group eliminated all carcinogenic activity. It required dichlorination to make ring-substtuted 1,4-phenylenediamine carcinogenic whereas only one chlorine atom was needed to make 1,2- and 1,3-phenylenediamine carcinogenic. While the available data suggest that as a class, 4-substituted 1,3-phenylenediamines are carcinogenic more often than ring-substituted 1,4-phenylenediamines, the type of added substituent and its position on the benzene ring also are important in exerting carcinogenic activity.


Introduction
Phenylenediamines comprise one of several classes of chemicals which are thought to contribute to the increase of cancer risk observed among workers in the dye manufacturing industry (1). The National Institute for Occupational Safety and Health has estimated that more than 64,000 people are potentially exposed in the workplace to a group of seven phenylenediamines, the production of which is over 50 million pounds. Exposure of the general public is estimated at more than 15 million individuals, since phenylenediamines are used in dyes, either directly as color-yielding compounds which include hair and fabric dyes or as intermediates and photographic development fluids (1).
Phenylenediamines were defined by the Interagency Testing Committee (ITC) as "all nitrogen-unsubstituted phenylenediamines with zero to two substituents on the ring selected from the same or different members of the group of halo, nitro, hydroxy, hydroxy-lower alkoxy, lower alkyl, and lower alkoxy" (2). For this purpose, the term "lower" is defined as a group between one and four carbons. The ITC listed 50 phenylenediamines as occurring in the Toxic Substances Control Act (TSCA) public inventory (2).
Long-term testing for carcinogenicity of all phenylenediamines and related compounds (i.e., those compounds in which -NH2 is replaced by -NO2) would *Health and Environmental Review Division, Office of Toxic Substances, U.S. Environmental Protection Agency, Washington, DC 20460. require a significant economic burden on industry. It would be useful, therefore, to be able to determine, from structural characteristics, which chemicals are most likely to be carcinogenic. It was the aim of this investigation to examine those published carcinogenicity studies for which there is adequate information for analysis and to attempt to identify correlations between chemical structure of phenylenediamines and their carcinogenic activity.

Analysis of Methodologies
The methods used for the conduct of the bioassays were analyzed according to the following criteria in order to ensure that meaningful conclusions could be made from the results obtained.
Chemical Purity. A plus (+) was designated when analytical tests indicated a chemical of high (99%) purity. A minus (-) indicated the presence of impurities ( Table 1).
Number of Animals. A plus (+) designated that there were 49 to 50 animals in chemical-treated and control groups. A minus (-) indicated that only 20 to 25 control animals and 49 to 50 chemical-treated animals were used, except in the studies by Weisburger et al. (24), in which only 25 chemical-treated animals per group were used (Table 1). Dose Schedule. A plus (+) indicated no change in dose scheduling throughout the bioassay. A minus (-) indicated that some modifications in the dose schedule occurred during the study ( Table 1).
Length of Dosing. The length of dosing was noted in weeks since it affects the total dose administered ( Table 1).
Weight. A plus (+) designated a decrease in weight gain relative to control which is not greater than 10% at the high dose. A minus (-) indicated a greater than 10% weight decrement relative to control at the high dose. "NC" indicated no noticeable change in weight gain relative to control (Table 1). Survival. A plus (+) designated that sufficient numbers of animals in all groups were at risk for the development of late-appearing tumors. A minus (-) designated poor survival, generally less than 20% at 103 weeks (Table 1). Amount ofDosing. The concentration of test material in the feed was noted in ppm/day since it provides a comparison of amounts of different test substances presented to test animals ( Table 1).

Analysis of Test Protocols
Before attempting to correlate carcinogenic activity of phenylenediamines and related compounds and changes in chemical structure, it was important to examine first the methods used in the bioassays in order to ensure that the conclusions were based on adequate test procedures. These findings can be seen in Table 1.
With the exception of mice given N2-phenyl-1,4,phenylenediamine for only 51 weeks because of sudden markedly reduced survival, all test animals were fed the appropriate diet containing the chemical for 78 to 103 weeks.
Only 20 to 25 control or chemical-treated rats and mice were used in some experiments. However, statistical analyses of results from all studies except those with 1,2-phenylenediamine dihydrochloride (I), 1,3-phenylenediamine dihydrochloride (IV) and 1-chloro-2,4dinitro-benzene (XIX) compared findings in chemicaltreated groups with concurrent and with historical controls.
While some dose changes occurred during several chronic studies (Table 1), survival was adequate at terminal sacrifice in all groups except in rats receiving the high dose of 4-methyl-1,3-phenylenediamine (VI), male rats receiving the high dose of 2-methoxy-5-nitroaniline (XXI), and male mice receiving the high dose of clonitralid.
A greater than 10% weight gain depression relative to control was seen in some high dose groups in the various bioassays (Table 1) indicating that the maximal tolerated dose (MTD) may have been exceeded in these groups. This finding, however, was not sufficient to negate the studies since, with the exception of 2-methyl-5-nitroaniline (XVI) and 3-nitro-4-ethoxy-N-acetylaniline (XXII), all carcinogenic chemicals in which the high dose caused a greater than 10% depression of mean body weight gain relative to control were also carcinogenic in at least one site of one sex of one species at the lower dose which approximated the MTD (3)(4)(5)(6)(7)(8)(9)(10)(11)(12)(13)(14)(15)(16)(17)(18)(19)(20)(21)(22). In the experiment with 3-nitro-4-hydroxyaniline (XX), mice had less than 10% weight gain depression relative to control, indicating that the MTD may not have been obtained.
Of the nine chemicals that affected the liver of the rodent ( Table 2), eight were carcinogenic for mouse liver, whereas only two significantly increased the incidence of liver tumors in the male rat [1,2-phenylenediamine dihydrochloride (I) and 4-methyl-1, 3-phenylenediamine (VI)]. 4-Methyl-1,3-phenylenediamine was also carcinogenic for the skin of both sexes of the rat and for the mammary gland of female Fischer 344 rats. Of the mouse hepatocarcinogens, only4-methoxy-1,3-phenylenediamine was also carcinogenic for other sites (i.e., hematopoietic system). While the high incidence of liver tumors (22%) in untreated male B6C3Fl mice might suggest that this site should not be considered in the evaluation of the carcinogenicity of these chemicals, it should be noted that, with the exception of 3-nitro-4-ethoxy-N-acetylaniline, all chemicals which were carcinogenic for the liver of male mice were also carcinogenic for the liver of female B6C3Fl mice where the historical control incidence of these neoplasms is only 4.0%. These results indicate a major difference in the tissue sensitivity of mice and rats to this group of chemical carcinogens.
In general, ring-substituted 1,4-phenylenediamines and related compounds were not carcinogenic (4 of 6 of the 1,4-phenylenediamines and related compounds were not carcinogenic), ring-substituted 1,3-phenylenediamines and related compounds were carcinogenic (8 of 12 were carcinogenic), and not enough information was available to conclude on the carcinogenic potential of ring-substituted 1,2-phenylenediamines (1 of 2 were carcinogenic).

False Positives and False Negatives
Before making structure-activity correlations, the possibility of false positives and false negatives must be considered. False positives in carcinogenicity studies may be caused by the use of impure chemicals, the placement of several chemicals in the room, the presence of abnormally low tumor rates in control animals, etc., as well as random variation in results. As can be seen from Table 1, several test criteria were not met by a number of the agents tested so that the possibility of a false positive in any test does exist.
The possibility of false negatives is likely if effects are small enough to escape detection in 50 animals, if the MTD was not reached in the studies, if impure chemicals were used, if nonsusceptible species or strains were employed, the use of low numbers of animals, poor survival, etc. With some chemicals, several tissue sites did show a positive dose-response trend, suggesting that if the doses were slightly higher then the chemicals might have been found to be carcinogenic at those sites. For example, 2-chloro-1,4-phenylenediamine sulfate was concluded not to be carcinogenic. However, there was a significant (p < 0.038) positive association between dosage and the combined incidence of hepatocellular carcinoma or hepatocellular adenoma in male mice. This finding suggests that 2-chloro-1,4-phenylenediamine sulfate may be found to be a carcinogen if retested under more vigorous conditions. Similar findings were seen with 2-carboxyl-5-nitroaniline (XVII) [positive trend (p < 0.035) in circulatory system neoplasms in the male mouse], and with clonitralid [positive trend in thyroid (p < 0.040) and uterus (p < 0.036) neoplasms in female rats].
Other limitations of this evaluation which should be noted include (1) limited data set; (2) choice of species; (3) choice of doses; (4) variations in the incidence of tumors in control groups; (5) biological variations including interspecies differences in pharmacokinetics; (6) effect of housing (i.e., cage rotation, etc.) on results; and (7) appropriateness of statistical methods employed.

Spontaneous Tumor Incidence
The ability to detect chemicals which are carcinogenic for specific sites of the rodent is directly dependent on the normal, background incidence of tumors at those sites in unexposed animals. For example, the high incidence (22%) of hepatocellular neoplasms in untreated male B6C3Fl mice substantially reduces the sensitivity of the assay method for detecting hepatocarcinogens in this species. Likewise, comparing results from carcinogenicity studies conducted in different strains of rats having differing spontaneous tumor incidences may be difficult.
Most of the studies reported herein were conducted in both sexes of B6C3Fl mice and Fischer 344 rats. Osborne-Mendel rats were used only in the study of clonitralid (XXIII) and that chemical was not found to be carcinogenic.
The spontaneous incidence of neoplasms in untreated Fischer 344 rats is nearly comparable to that of Osborne-Mendel rats, with a few exceptions (25). For example, the incidence of mammary gland tumors in untreated female Osborne-Mendel rats is nearly 37%, whereas in female Fischer 344 rats it is only 18%. Also, kidney tumors appear in greater numbers in unexposed Osborne-Mendel rats of both sexes (approximately 3.5%) compared to Fischer 344 rats (0.4%) as are thyroid tumors in female Osborne-Mendel rats (11% versus 7%, respectively). Thmors of the uterus and pituitary gland, however, are significantly reduced in untreated female Osborne-Mendel rats compared to female Fischer 344 rats (4% versus 16% and 21% versus 30%, respectively). Differences in spontaneous tumor incidence can also be seen between sexes of the same strain and species. For example, whereas liver and lung neoplasms are more abundant in untreated male B6C3F1 mice than in the corresponding females, the incidence of pituitary gland tumors and leukemia/lymphoma is greater in female B6C3Fl mice than in the males (26). Similarly, the incidence of mammary gland and pituitary gland tumors is increased in female Fischer 344 or Osborne-Mendel rats over the corresponding males, whereas adrenal gland tumors appear at two to three times greater frequency in untreated males than in female rats (25).
Structure-Activity Relationships 1,2-Phenylenediamines. Only three 1,2-phenylenediamines were examined for carcinogenicity in this review, and these gave conflicting findings so that no definitive conclusion could be made on the carcinogenic potential of this class of compounds. Compounding further the interpretation of the results from this class of compounds is the fact that 1,2-phenylenediamine dihydrochloride was tested in male Charles River CD rats and both sexes of random-bred albino CD-l mice derived from HaM/ICR whereas 4-chloro-1,2-phenylenediamine and 4-nitro-1,2-phenylenediamine (III) were tested in Fischer 344 rats and B6C3F1 mice of both sexes.
Both 1,2-phenylenediamine dihydrochloride and 4chloro-1,2-phenylenediamine were carcinogenic. However, changing the 4-chloroto a 4-nitrogroup eliminated all carcinogenic activity. Additional studies with varied analogs are needed to analyze more fully the carcinogenic potential of this group of compounds. 1,3-Phenylenediamines. While 1 ,3-phenylenediamine dihydrochloride was not carcinogenic in rats or mice, several of its ring-substituted analogs were. For example, chlorination, methylation, or hydroxylation at a position para to an amine group produced a compound with carcinogenic activity. In comparison, methylation at a position ortho to the amine function, i.e., 2-methyl-1,3-phenylenediamine dihydrochloride (VII) yielded a compound which was completely inactive.
As the substituent at the fourth position of 1,3phenylenediamine was changed from -Cl to -CH3 to -OCH3, the compound became carcinogenic for more species, more sexes, and more target sites. However, N-substitution (i.e., 4-ethoxy-N1-acetyl-1,3-phenylene-diamine and 3-nitro-4-ethoxy-N-acetylaniline reduced the carcinogenicity to only one site, of one sex, of one species (male mouse thyroid or liver). Oxidation of one of the -NH2 groups to -NO2 function substantially reduced the carcinogenicity of the compounds that fell into this class. For example, 4-methyl-1,3-phenylenediamine was carcinogenic for several sites of male and female rats and male mice, while the oxidation of the -NH2 function at position -1 to -NO2 group (i.e., 2-methyl-5-nitro-aniline) limited the carcinogenicity of the chemical to the liver of both sexes of mice only. In a similar manner, 4-methoxy-1,3-phenylenediamine sulfate was carcinogenic for several sites of rats and mice of both sexes whereas 2-methoxy-5-nitroaniline, the N1-oxidized analog, was carcinogenic only for female rats and female mice.
It can be concluded, therefore, that while it appears that 4-substituted 1,3-phenylenediamines tend to be carcinogenic, the carcinogenicity of chemicals in this small and nonrandom set is substantially reduced or eliminated completely by oxidation of one or both amine groups or by N-substitution. Oxidation of the methyl substituent on nitroaniline to a carboxyl group produced a compound which was devoid of carcinogenic activity.
While the monochloro analog of 1,2and 1,3-phenylenediamine and the monomethyl analog of 1,3-phenyllenediamine were carcinogenic for both species at several sites, the monochloro and monomethyl analog of 1,4-phenylenediamine were completely inactive. It required dichlorination (2,6-dichloro-1 ,4-phenylenediamine) to make 1,4-phenylenediamine carcinogenic, whereas only one chlorine atom was needed to make 1,2and 1,3-phenylenediamine carcinogenic. Therefore, while the data are limited, it appears that the position of the chlorine atom in relation to the diamines on the benzene ring may be important for carcinogenic activity. Both 1,2and 1,3-phenylenediamines were carcinogenic when the chlorine atom was para to an amine group. The chlorine atom on 1,4-phenylenediamine, on the other hand, can only be ortho or meta to the amines. This may explain the noncarcinogenicity of monochloro-1,4-phenylenediamine.
That the relationship of the chlorine atom to the amine groups on the benzene ring may be important for the carcinogenicity of the benzenediamines is further supported by the following examples. 4-Chloro-1,2phenylenediamine was carcinogenic for both sexes of mice and rats producing rare bladder tumors in both sexes of rats. On the other hand, 4-chloro-1,3-phenylenediamine was carcinogenic only for male rats and female mice and adrenal tumors rather than bladder tumors were observed in male rats treated with this agent. 2-Chloro-1,4-phenylenediamine was completely inactive in either species tested.
While 1,4-phenylenediamine dihydrochloride (IX) was inactive in mice and rats, so were the monochloro and monomethyl analogs of the parent compound. Addition ofa nitro group (2-nitro-1 ,4-phenylenediamine), however, elicited a carcinogenic response in the liver of female B6C3Fl mice.
Based on the available data, it can be concluded that ring-substituted 1,4-phenylenediamines tend to be noncarcinogens (four of six ofthe 1,4-phenylenediamines and related compounds were not carcinogenic). Also, there is one example (2,6-dichloro-1,4-phenylenediamine) in which twice as many electron-withdrawing groups (i.e., dichlorides) were needed on 1,4-phenylenediamine than on 1,2or 1,3-phenylenediamines in order to obtain nearly comparable carcinogenic activities. Sontag (27) suggested that "phenylenediamines appeared to be least active when the amine groups were para to one another, and gained activity as they became ortho to the substituted groups" Our evaluation, although limited by the number of compounds examined, is more extensive than Sontag's in that it includes several compounds not in his review, and it analyzes in detail the methods and results of the bioassays. Moreover, our evaluation partially supports Sontag's conclusion and indicates that 4-substituted 1,3-phenylenediamines are carcinogenic more often than ringsubstituted 1,4-phenylenediamines. Therefore, in setting priorities for consideration of phenylenediamines for testing for carcinogenicity, the likelihood that ringsubstituted 1,3-phenylenediamines will be carcinogenic in a long-term animal bioassay appears greater than that of ring-substituted 1,4-phenylenediamines. How-ever, the type of added substituent and its position on the benzene ring also are important in predicting carcinogenic potential. Since sufficient information is not available presently to draw conclusions related to 1,2-phenylenediamines, these chemicals should be considered on a case by case basis.

Conclusions
A qualitative assessment of the potential for carcinogenicity based on structural changes should provide the basis of additional definitive quantitative structure-activity relationship studies based on relevant molecular descriptors and reactivity indicators. Such studies, however, were beyond the scope of this review.