The redundancy of mouse carcinogenicity bioassays

The redundancy of mouse carcinogenicity bioassays

FUNDAMENTAL AND APPLIED TOXICOLOGY 3:631-039 (1983) Issues The Redundancy of Mouse Carcinogenicity Bioassays MANFRED SCHACH yon WITTENAU and PAUL C. ...

NAN Sizes 1 Downloads 56 Views

FUNDAMENTAL AND APPLIED TOXICOLOGY 3:631-039 (1983)

Issues The Redundancy of Mouse Carcinogenicity Bioassays MANFRED SCHACH yon WITTENAU and PAUL C. ESTES Department of Drug Safety Evaluation, Pfizer Central Research, Groton, CT 06340 ABSTRACT

T h e R e d u n d a n c y o f M o u s e C a r c i n o g e n i c i t y Bioassays. S c h a c h von W i t t e n a u , lVl. a n d Estes, P.C. (1983). Fundam. Appl. ToxicoL 3:631-639. Testing o f chemicals for carcinogenic potential usually involves studies in rats and mice. The approaches

followed in recent years often were limited to assessing tumor incidences and rarely viewed such information from the perspective of other kinds of toxicity. This practice supports the notion that certain chemicals possess the inherent characteristic of carcinogenic activity, which once identified in one species must be regarded as potential oneogenic hazard to all, including man. As long as public policy is based upon thai premise, the classification of chemicals depends upon the worst results obtained in any species. It is not obvious why substances must be ti~sled in both rats and mice when confirmatory or contradictory responses have little impact. Recent experience was examined to build a basis for an alternate approach. Review of o n c o g e n i d t y information obtained for 273 chemicals fed to rats and mice shows that both species responded similarly to the majority ef the substances tested. This "redundancy" probably would be higher had modern protocols been used. In view of the high cost in scientific resources of chronic studies, tests in only one rodent would be more cost effective. Reasons are presented for favoring the rat as the species to be used. S o m e of the savings thus achieved should be expended to improve the design of the experiments to yield toxicology data more comprehensive than mere tumor counts, so that the proper perspective can be obtained on the results.

I N T R O D UC TION Evaluation of substances for their potential to represent a carcinogenic hazard to man relies mainly on lifetime rodent studies. Rigid protocols have evolved w h i c h must be followed in order to satisfy expectations and requirements. Despite occasional questioning of the validity of the current approach, there appears to be little critical analysis of the body of experience, and fundamental changes have not been proposed. For example, the view is widely held and reflected in regulations and Government-directed research programs, that the "bioassay ''1 for carcinogenicity is not complete unless substances were administered to two rodent species, usually rats and mice (NCI, 1976; CSM, 1979; OECD, 1981; FSC, 1982; EPA, 1979). The justification for the ltwo rodent concept is illdefined and in fact constitutes a paradox; if "carcinogens" possess the inherent property of causing cancer, one species should suffice for detecting this quality, but if "carcinogenic'" activity is species-dependent then the validity of the models as surrogates for man is questionable.In view of the considerable cost in scientific resources, it is justified to inquire whether the two-species "bioassay" is a reasonable approach towards safeguarding man or w h e t h e r research efforts could be expended more wisely. Experience in our laboratories suggested to us that little additional information relevant to human risk assessment was obtained from mouse studies if tests in rats also had been performed. Due in part to protocol design, w h i c h mandates the collection of additional data such as clinical chemistry values, rat experiments were more informative as to the overall toxic-

I'rhe term bioassay implies a reproducible qua ntita rive measuren~en| of inherent biological activity. Such concept appears to be too simplis|ic when applied to the e w l u a t i o n o[ carcino~eniC potential (Schach yon WHtenau, I979).

ity of the substance studied. Although there is a long history of using the mouse in carcinogenicity testing (Tomatis et aL, 1973; Berenblum, 1974,~, several well-recognized problems are encountered in the interpretatidn of results from such studies. The tenet that an apparent carcinogenic response eo ipso signals human hazard cannot be reconciled w i t h all accumulated experience. For example, there is evidence indicating stress alone can lead to a high number of tumors in the mouse (Riley et aL, 1981). High and variable incidences of spontaneous lymphoreticular, lung, liver and mammary gland tumors present interpretational difficulties which have caused some scientists to question the validity of the mouse as a test system (Grasso and Crampton, 1972; Grasso et aL, 1977; Butler, 1981 ). Consequently we questioned whether inclusion of the mouse in addition to the rat in the "bioassay" for carcinogenicity had contributed significantly to risk assessment decisions; how many and w h a t kind of substances purportedly constituting a carcinoge'nic hazard to man were recognized in mouse bioassays but missed by rat bioassays? A REVIEW OF O N C O G E N I C I T Y D A T A A recent compilation (Soderman, 1982) of chemicals evaluated by NCI and IARC provided the opportunity to examine the performance of the currently prescribed approach. Specifically, we were interested in evaluating the wisdom of testing each chemical in two species, rather than twice the number of substances in one species. As many widely-used materials have not yet been evaluated for carcinogenic pot,~P~'lial, a change in testing requirements might be purposeful. The survey cited above includes 614 bioassay results and catalogues data for 273 substances w h i c h were orally administered to both rats and mice. 2 It accepts at face value the

Copyright 1983, 5ociety of Toxicology

Fundamenlal and Applied Toxicology

(3) 11-12/~3

631

S C H A C H yon W I T T E N A U A N D ESTES

TABLE 1 The Effect of Mouse "Bioassay" Data on the Oncogenic Classification of 273 Substances Noncontributory Mouse

Data (NMD)

Rat and Mouse "Bioa.ssay'" Concordance Carcinogens (Table 2) Noncarcinogens (Table 3) "'1nconclusives" (Table 4)

SublotalofConcordance NMD

86 90 30

31% 33% 11%

20o

75%

20 1 9

7.3% 0.4% 3.3%

Rat and Mott.,,e "Bioaxsay'" Nonconcordance Rat Carcinogens - Mouse Noncarcinogens (Table 5) - Mouse " l n c o n c l u s i v e s " ( T a b l e 6) Rat Noncarcinogens - Mouse "lnconclusives'" (Table 7)

Subtotal of Nonconcordance N M D Total of Noncontributory Mouse Data

30

11%

236

86%

28 3 0

10% 1.1% 2.2%

37

13.6%

Contributory M o u s e D a t a Rat and Mouse "'Bioassay" Nonconcordance Mouse Carcinogens - Rat Noncarcinogens (Table 8) - Rat "'lnconclusivcs'" (Table 9) Mouse Noncarcinogcns - Rat "'lnconclusives" (Table 10) Total of Contributo ry Mouse Data

published conclusions w h i c h were reached by panels of experts who reviewed the data. These conclusions can be broken down as shown in Table 1.

R E D U N D A N C Y OF M O U S E B I O A S S A Y D A T A According to this tabulation, the mouse experiments were redundant in 75% of all cases (Tables 2, 3 and 4) and in another 3.7% (Tables 6 and 7) neither confirmed nor refuted the conclusions based on rat data. For 7.3% (Table 5) the negative results in mouse studies did not ameliorate the impact of tumorigenicity observed in rats. Thus for 236 of the 273 chemicals tested by the oral route in rats and mice, the mouse studies retrospectively can be considered superfluous, because they did not contribute to risk assessment decisions. 3 For nine chemicals (butyl benzyl phthalate, 3-nitropropionic acid, parathion, phosphamidon, picloram, TDE, heptachlor, toxaphene and trimethyl phosphate), rat experiments did not provide definitive data but mouse studies did (Tables 9 and 10). For seven of these chemicals, a verdict of "inconclusive" in rats was reached by the panel. Only benign tumors had been observed (Table 9 (mouse carcinogens) heptachlor and toxaaPart of these data have been surveyed by others as well. Purchase (1980) provides tabulations similar to, but not identical to ours. His paper also includes data obtained by parenteral administration, but does not include all chemicals listed by us. Also, in a few instances our classifications vary. Griesemer and Cueto, Jr. (1980) have surveyed a large number of chemicals tested by NCI, and formed their o,,vn judgments as to the appropriate classifications, which differ in many instances from the assignments made by us, or by Purchase. See also Chu et aL (1981) and Ward et al. (1979). alt must be emphasized that the terms "redundant" and "superfluous" ,,-,,'ere chosen only because of public policy decisions which apparently were made without adequately considering whether or not apparent oncogenic activity observed in one species was corroborated in another. As long as the scientific value of confirmatory or contradictory information is not acknowledged in decision making, there is little point in expending large resources for obtaining such evidence. 632

phene, thyroid follicular celt adenomas and trimethylphosphate, benign subcutaneous fibromas; Table 10 (mouse noncarcinogens) 3-nitropropionic acid and picloram, benign liver tumors; parathion, cortical adrenal adenomas; TDE, thyroid follicular cell adenomas). That only three of these seven chemicals were concluded to be carcinogenic in the mouse emphasizes the need for caution that must be exercised in interpreting benign tumors as indicators of a carcinogenic hazard for another species. The approach we followad to this point identifies 31 compounds (Tables 8 and 9) or 11.1% of 273 tested w h i c h were identified by the mouse, but not by the rat, as putative carcinogens. Risk assessment decisions might be based on these results. Assuming past experience to be a valid guide for the future, a rough estimate of the cost effectiveness 't of the current testing strategy can be obtained. For this purpose let use assume that resources equivalent to $100 million are available for testing, and that these are sufficient to conduct 400 rodent studies w h i c h yield results in the same proportions as those observed for the 273 tested chemicals. Among the numerous options are the current approach, w h i c h evaluates 200 materials and identifies 101 as "carcinogenic", or a plan w h i c h evaluates 400 substances in one species only. This latter strategy is expected to result in the detection of 156 "'carcinogens" if rats are used, and 172 "carcinogens" if the mouse is chosen. A hybrid strategy w h i c h also uses only one species, but selects either the rat or the mouse based on considerations of chemi*Within the context of this paragraph, cost effectiveness is measured in number of substances recognized as "carcinogens." As long as "ca rcinogenicily'"is thought Io be a property a chemical does or does not possess, such measure is logical. One would seek to identify with the resources availableas many carci21ogens as possible so as to avoid human contact with them. We do not agree with lhis stance, however, and believe that carcinoge:dcity should be viewed like other kinds of toxicity, whlch must be interpreted in the context of mechanisnl, dosage, metabolic pathways, ,and species differences, all of which will contribute to a rational ,assessment of hazard. The statement that a chemical"causesca,acer" may not be more informative than the conclusion that a substance is "lethal".

F u n d a m . Appi. ToxicoL (3)

N o v e m b e r / D e c e m b e r , i 9&]

MOUSE CARCINOGENICI'I'Y BIOASSAYS TABLE 2 Rat and Mouse Carcinogens 1. 2. 3. 4. 5. b. 7. 8, 9. 10, 11. t2. 13. 14. 15. 16. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27. 28. 29. 30. 31. 32. 33. 34. 35. 36. 37. 38. 39. 40. 41. 42. 43.

1-amino-2-met hylanthraquinone 3-amino-9-ethylcarbazole hydrochloride 2-aminoanthraquinone o-aminoazotoluene amitrol o-anisidine hydrochloride aramite auramine benzo(a)pyrene bis-dimet hylaminodiphenylmet hane carbon tetrachloride chlordecone 4-chloro-m-phenylenedia mine 4-chloro-o-phenylenedia mine chloroform 3-(chloromethyl)pyridine hydrochloride cinnamyl anthranilate citrus red 2 p-cresidine cupferron cyeasin 2,4-diaminoanisole dibromochloropropa ne 1,2-dichloroethane di(2-ethylhexyl)phthalate diethylstilbestrol dihydrosa frole 1,2-dimethylhydrazine 1,4-dioxane ethinylestradiol ethylene dibromide ethynodiol diacetate 2-(form ylhyd razino)-4-(5-nit ro-2-furyl) t hiazole hexachlorodibenzo-o-dioxins hydrazine sulfate hydrazobenzene isosafrole lead acetate mestra nol 4,4'-methylene bis(2-chloroaniline) N-methyI-N'-nit ro-N-nit rosogua nidine methylthiouracil metronidazole

cal structure and species sensitivities, may provide the opport u n i t y t o approach the maximum cost effectiveness, i.e. identification of 203 "carcinogens". This analysis obviously cannot be acceptable without ful:ther scrutiny as a guide for future action although it reinforces the suspicion that the two-species-approach may not be the best. Evaluations of the options mentioned have to consider factors other than just the number of "carcinogens" identified. For example, although the mouse more often than the rat was judged to show a positive response, one could conclude that the mouse is more useful only if the chemicals "missed" by the mouse are thought to be less of a human hazard than those substances "missed" by the rat. It is unlikely that an unequivocal answer can be derived from our data base, because to do so one would have to assume that the data are reproducible, that the 273 chemicals tested are an adequate representation of all substances, and that sufficient information exists to a l l o w an assessment of the validity of the rodent data as predictors of human hazard. Despite these reservations, we made these assumptions and continued our analysis. Fundamental and Applied Toxicology

(3) 11-12/83

44. 45. 4o. 47. 48. 49. 50. 51. 52. 53. 54. ,55. 56. 57. 58. 59. 60. 61. 62. 63. 64. 65. 66. 67. 68. 69. 70. 71. 72. 73. 74. 75. 76. 77. 78. 79. 80. 81. 82. 83. 84. 85. 86.

Michler's ketone mirex rnonuron 1,5-naphthalenediamine niLhiazide nitrilotriacetic acid 5-nitro-o-anisidine N-(4- f5-nitro-2-furyl)-2-thiazolyl)acetamide 5-niu oacenaphthene N-nitroso-N-methylurethane N-nit rosodi-n-butylamine N-nit rosodiethyla mine N-nit rosodimethyla mine p-nitrosodiphen ylamine N-nitrosomorpholine N-nitrosopiperidine N-nitrosopyrrolidine N-nitrososarcosine norethisterone

norethynodrel 4A'-oxydianiline phenazopyridine hydrochloride phenesterin phenobarbital sodium polychlorinated biphenyls Ponceau MX propyhhiouracil reserpine sa frole selenium sulfide sterigmatocystin sulfallate 2,3,7,8-tet racblorodibenzo-p-dioxin tetrachlorvinphos thioacetamide 4,4'-thiodianiline thiouracil 2,4-toluenediamine o-toluidine hydrochloride 2,4,6-trichlorophenol 2,4,5-t rimethyla niline tris(2,3-dibromopropyl) phosphate urethane

From Tables 5 and 6 which list the chemicals identified as "carcinogens" by the rat only, one gains the general impression that a test system which does not detect these chemicals as carcinogens, among them, aflatoxin B1 ~ and azobenzene, is inadequate as a tool for human risk assessment. On the other hand, among the 31 chemicals carcinogenic in mice only (Tables 8 and 9) are many chlorinated hydrocarbons, which are not accepted universally as posing an oncogenic hazard to man (Deichmann, 1981; Deichmann and MacDonald, 1977, 1976).

S Q U I R E SCORES FOR THE RA T A N D MOUSE CARCINOGENS To gain further perspective on the carcinogenic hazard to man each of the 52 chemicals of Tables 5, 6, 8 and 9 might represent, we applied the concepts and scheme proposed by Squire (1981). He suggested that apparent animal carcino-

~AflatoxlnB] is capableof causingtumorsin miceby int raperitonea]administration (Vesselinovitch el al., 1972). 633

SCHACH yon WITTENAU AND ESTES TABLE 3 Rat a n d M o u s e N o n c a r c i n o g e n s l. acetohexamide 2. aldicarb 3. p-aminodiphenylamine 4. anilazine 5. p-anisidine o. anthranilic acid 7. arsenic trioxide 8. aspirinlphenacetinlca ffeine 9. azinphosmethyl 10. 1H-benzot riazole 11. benzoin 12. BHT 13. bisphenol A 14. gamma-butyrolactone 15. calcium cyanamide 1o. caprolactam 17. carbrumal 18. carmoisine to. chlormadinone acetate 20. 4-chloroacetylaceta nilide 21. 2-chloroethyl-trimethylammonium chloride 22. 2-(chloromethyl)pyridine hydrochloride 23. 2-chloro-p-phenylenediamine sulfate 24. chloropropham 25. 3-chloro-p-toluidine 20. chlorpropamide 27. cou,naphos 28. cyclohexylamine 29. DDT 30. dexon 31. diarylanilide yellow 32. diazinon 33. dibenzo-p-dioxin 34. l,l-dichloroethane 3,5. dichlorvos 30. N,N'-dicyclohexylthiourea 37. sodium diethyldithiocarbamate 38. dimethoate 39. 2,4-dimethoxyaniline hydrochloride 40. dimethyl terephthalate 41. dioxathion 42. disulfiram ,13. endrin 44. ethionamide 45. sodium ethylenediaminetetraacetate

gens could be assigned to one of five possible classes depending on their accumulative scores from 6 factors: number of species affected, number of histogenetically different types of neoplasms, background incidence of observed neoplasms, dose-response relationship, degree of malignancy of induced neoplasms and genotoxicity. Classes I (scores 86-100) and II (71-85) w o u l d receive the highest priority for regulatory action w h i c h could result in a total ban; w h i l e classes III (56-70), IV (41-55), and V ( < 4 1 ) w o u l d reflect decreasing concern and could "permit many options such as no action, approval for limited uses, labeling or public education programs." Although neither Squire nor we are firlhly committed to the magnitude of e a c h w e i g h t i n g proposed, we followed the concept, and have performed the calculations for each of the 52 chemicals. As the available information did not allow us to assign mutagenicity scores to each, the calculations were performed assuming the substances are n o t mutagenic. 634

46. 47, 48. 40. 50.

ethyl tellurac iodoform ledate lithocholic acid lynestrenol

51. malaoxon 52. malathion 53. maleic hydrazide 54. O,L-menthot 55. methoxychlor

56. methyl parathion 57. mexacarbate 58. N-(1-naphthyl)ethylenediamine HCI 59. 4-nitroanthranilic acid o0. l - n i t r o n a p h t h a l e n e Ol. 4-nitro-o-phenylenediamine 02. beta-nitrosytrenelsytrene (30%170%) solution ,53. norgestrel 04. Orange G o5. pentachlorophenol 66. phenformin 67. phenol o8. p-phenylenediamine hydrochloride 09. l-phenyl-3-methyl-5-pyrazolone 70. 1-phenyl-2-thiourea 71. photodieldrin 72. phthalamide 73. phthalic anhydride 74. piperonyl butoxide 78. Ponceau SX 76. quintozene 77. styrene 78. sulfisoxazole 79. 3-sulfolene 80. Sunset Yellow FCF 8 I. 2,3,5,o-tetrachloro-4-nitroanisole 82. titanium dioxide 83. tolazamide 84. tolbutamide 85. 2,5-toluenediamine 86. 2,6-toluenediamine 87. triphenyltin hydroxide 88. L-tryptophan 89. Yellow AB 90. Yellow OB

TABLE 4 Rat and Mouse "Inconclusives" 1. 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15.

amaranth arsenic cadmium acetate earbary] carrageenans p-chloroa niline chloropicrin chloroprene o-dichlorobenzene p-dichlorobenzene 2,4-dichlorophenoxyacetic acid diglycidyl resorcinol ether endosulfan ferbam heptachlor epoxide

Fundam. AppL Toxicol. (3)

16. 8-hydroxyquinoline 27. maneb 18. megestrol acetate

19. N-nitrosoproline 20. ochratoxin A 2.1. orange l 22. proflavine 23. propham 24. 25. 26. 27. 28. 29.

selenium sudan I sudan II sudan Ill 1,1,1-trichloroetha ne zineb

30. z i r a m

November/December, 1983

MOUSE'CARCINOGENICITY BIOASSAYS TABLE 8 Rat N o n c a r c i n o g e n s w h i c h are C a r c i n o g e n i c in Mice

TABLE 5 Rat C a r c i n o g e n s w h i c h a r e Noncarcinogenic in M i c e

Squire Score

Squire

Score

1. aflatoxin Bl 2. 3. 4. 5. 6. 7. 8. 9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20.

05

4-a mino-2-nit rophenol 2-a mino-5-nit rot hiazole 11-aminoundecanoic acid aniline hydrochloride azobenzene chlorothalonil m-cresidine dapsone N,N'-diethylthiourea 3-dimethoxybenzidine-4,4'-diisocyanate 2,4-dinit rotoluene direct black 38 direct blue 6 direct brown 95 N-nit rosodiphenyla mine pivalolactone p-quinone dioxime thiourea trimethyhhiourea

45 45 40 55 50 45 45 50

45 51 4] 43 45 45 45 40 45 55 45

TABLE 6 Rat C a r c i n o g e n s w h i c h are

"Inconclusive" in Mice" Squire

Score 1. daminozide

I. 2. 3. 4. 5. o. 7. 8. 9. 10. I I. 12. 13. 14. 15. lo. 17. 18. 19. 20. 21. 22. 23. 24. 25. 26. 27.

aldrin 3-a mino-4-ethoxyaceta nilide Aroclor 1254 bis(2-chloroet hyl)et her captan chloramben chlordane 4-chloro-o-toluidine 5-chloro-o-loluidine chlorobenzilate C.I. vat yellow 4 p,p'-DDE di(2-et hy/hexyl)adipa t e 2,6-dichloro-p-phenylenediamine dicofol dieldrin estradiol mustard hexachloroethane 5-nitro-o-toluidine 3-nitro-p-acetophenetide 2-nit ro-p-phenylenedia mine O-nit robenzimidazole nitrofen piperony] sulfoxide semicarbazide hydrochloride l,l,2,2-tetrachloroetha ne 1,1,2-t richloroelhane 28. trifluraline

40 45 21 35 41 30 40 45 45 40 3o 40 30 35 35 40 50 40 45 20 30 40 35 35 31 40 45

50

60

"The compendium lists Ponceau 3R as carcinogenic in rats, but "inconclusive" in mice. The "inconclusive" in this instance results from inadequate data.

TABLE 9 Rat " l n c o n c l u s i v e s " Which are C a r c i n o g e n i c in Mice"

Squire Score TABLE 7 Rat N o n c a r c i n o g e n s w h i c h a r e

"Inconclusive" in Mice I. 2. 3. 4. 5. 6. 7. 8. 9.

allyl chloride clonitralid 2,7-dichlorodibenzo-p-dioxin dithiobiurea p,p'-ethyI-DDD fenthion fluometuron pyrazinamide pyrimethamine

It is evident, that the 21 "rat carcinogens" (Tables 5 and 6) received generally higher scores (average 48) than the 31 "mouse carcinogens" (Tables 8 and 9 average 38.8). It is also noteworthy, that among the former there is only one chemical (5%) (pivalolactone) w i t h a score of 40 or less, w h i l e there are 22 (71%) among the latter. Conversely, 6 substances (30%) (aflatoxin B1, aniline hydrochloride, azobenzene, dapsone, 3-dimethoxybenzidine-4,4'-diisocyanate and thiourea) among the "rat carcinogens" had a score of 50 or more, but only 2 (6%) (estradiol mustard and trifluraline) of those in Tables 8 and 9 did so. A l t h o u g h valid mutagenicity information could, and in some instances w i l l change this ranking, this scheme Fundamental and Applied Toxicology

(3) 11-12/83

1. heptachlor 2. toxaphene 3. trimethyl phosphate

40 40 45

"The compendium lists BHC, chrysoidine, hexachlorocyclohexane, beta naphthylamine, Oil Orange SS, phenytoin, pronetalol hydrochloride, tetrachloroethylene and trichloroethylene as carcinogenic in mice but"inconclusive" in rats. Tt~e "inconclusive" in these instances results from inadequately obtained andlor reported data. Beta naphthylamine is carcinogenic in several species, but rat studies according to modern protocols apparently have not been conducted. (Bonser et aL, 1952; Hadidian et aL, 1968).

suggests that if one were forced to choose between "'missing" the chemicals of Tables 5 and 6, or the substances of Tables 8 and 9, one might opt for the rat as the investigational tool. Such course of action indeed appears to be quite defensible, as a closer examination of the "'mouse carcinogens" (Tables 8 and 9)indicates.

SCRUTINY OF THE POSITIVE M O U S E BIOASSA V RESULTS An initial impression of the perplexity in assessing the oncogenic potential these chemicals might possess is gained by the 635

SCHACH yon WITTENAU AND ESTES TABLE 10

Rat "Inconclusives" which are Noncarcinogenic in Mice" 1. butyl benzyl phthalate 2. 3-nitropropionic acid 3. parathion

4. phosphamidon 5. picloram 0. TDE (DDD)

"The compendium lists dibutyhin diacetate, trich[orofluoromethane, and saccharin as rtoncarcinogenic in mice, but "inconclusive" in rats. The "inconclusive" in these instances results from inadequately obtained and/or reported data. A rat carcinogenicity bioassay retest has been scheduled for butyl benzyl phthalate, as early mortality i,m males in the first study precluded a conclusion of oncogenicity (PHSNTP, 1982).

realization that 20 of the 31 chemicals listed in Tables 8 and 9 elicited hepatocellular carcinomas in the mouse but no other malignancies. These lesions do not necessarily mark a substance as very hazardous to man. As alluded to previously, liver tumors in the mouse pose several interpretative problems w h i c h arise from the high background incidence in some strains, particularly in males, and the ill-defined differentiation of hyperplastic nodules from adenomas and the latter from malignant liver tumors (Frith and Ward, 1980). The biological behavior of these lesions does not correlate well w i t h their histologic features and their progression from the benign to the malignant state is not well established (CIIT, 1981; Frith and Ward, 1980; Ward and Vlahakis, 1978; Stewart, 1975). An illustration of the room for disagreement is given by the reevaluation of the histopathology slides of mouse liver lesions caused by heptachlor. NCI pathologists found cancer in 64°of 89 high-dose mice, but an NAS Committee only in 4 (O'Connor and Woodward, 1981 ). The differential diagnosis and the implications of lesions of this type have been debated for years (Grasso and Crampton, 1972; Tomatis et aL, 1973; Grasso et al., 1977; Ward and Vlahakis, 1 978; Frith and Ward, 1980; Buffer, 1981 ; Becket, 1982), and were the subjects of a recent CIIT Conference on Toxicology (CIIT, 1981 ). That forum served a useful purpose by focusing attention on the issues but ended with the problems unresolved. Nevertheless, opinions persist that mouse liver tumors may be unreliable indicators of carcinogenic potential. Ten of the 20 chemicals causing only mouse liver tumors are chlorinated hydrocarbons. Many of these have been in widespread use, and occupational exposure to some representatives of this group has been relatively large. Although epidemiology studies cannot prove a negative, it appears that efforts have failed so far to link these chemicals to the induction of cancer in man (Deichmann, 1981 ; Deichmann and MacDonald, 1977, 1976). Experiments with related compounds (Schumann eta/., 1980; Stott et aL, 1982) have led to the hypothesis that at high doses such chemicals become cytotoxic and that recurrent tissue damage w i t h consequent compensatory hyperplasia is a precondition of tumor formation. Apparently, the mouse is more prone to incur liver damage than is the rat a n d / o r presumably man. If that is so, then the finding of liver tumors in mice is of very limited relevance to assessment of risk resulting from man's exposure to such substances below cytotoxic concentrations (Stott et aL, 1981 ). The initial assumption that none of the substances listed in Tables 8 and 9 is mutagenic in a relevant battery of tests is not warranted. We consequently looked more closely at the chemicals with the highest scores, and perticularly at those two (estradiol mustard and trifluraline) w h i c h scored 50, because these would be considered highly hazardous if they were mutagenic. 636

The structure of estradiol mustard suggests mutagenic activity, and oncogenic potential. As the molecule was designed to be an antineoplastic agent the compound poses an apparent enigma, not because it was found carcinogenic in mice, but because such activity was not established in rats. Possibly, the large discrepancy in doses offers an explanation; mice received 15 and 30 mg/kg, w h i l e rats ingested 0.62 and 1.25 mg/kg b.w. (NCI, 1978a). Although at these doses in rats no statistically significant increase of neoplasms was found in a single organ, it is of interest that there were 17 of 70 (24%) medicated females with malignancies, but no malignancy was observed in 20 control animals. In mice, 58% of all medicated animals (both sexes) showed malignant tumors w h i c h reached statistical significance in several organs, but only 2 animals of 60 controls did so. In view of these data we suspect that rats and mice will respond similarly to comparable doses of estradiol mustard, and that this chemical does not support the contention that the mouse is needed for the identification of a carcinogen likely to be hazardous to man if rat experiments follow modern protocols. The assessment of the trifluralin data is not purposeful because a technical grade product was used which contained measurable concentrations of a nitrosamine known to be carcinogenic in rats and mice (NCI, 197Bb). The data therefore allow the interpretation that mice are somewhat more but not uniquely sensitive to the contaminant. That particular material is no longer representative of the commercial product. In a repeat study the pure material did not show oncogenic activity in mice (Emmerson, 1982). If the proposal by Squire (1 981) were followed, none of the other chemicals would receive highest priority for regulatory action (Categories I and II - - scores 71 to 100) even if the maximum possible values for mutagenic activitywere added to their scores. Three of six chemicals with the score of 45 are toluidines (4-chloro-o-toluidine, 5-chloro-o-toluidine and 5-nitro-o-toluidine), which received this high score because hemangiosarcomas in addition to liver tumors were observed (NCI, 1979a; NCI, 1979b; NCI, 1978c). It is interesting that the record of the discussion of the review panel evaluating the data for NCI indicates that in two instances (4-chloro-otoluidine and 5-chloro-o-toluidine)(NCI, 1979a; NCI, 1979b) uncertainty was expressed as to the meaning of these mouse lesions for human risk assessment. A similar reservation was expressed concerning the significance of the thyroid tumors observed in mice following administration of 3-amino-~,ethoxyacetanilide (NCI, 1978d). These instances dramatically demonstrate that the mere recording of tumors provides an inadequate basis for human risk assessment. A more comprehensive toxicological evaluation is needed, as the tumors may be sequelae of primary insults (iron metabolism disturbances and hemosiderosis) not elucidated in these experiments. Table 11 shows the putative carcinogenicity of the aniline derivatives surveyed. It is difficult to discern a consistent structure-activity relationship and be confident that the data are reproducible. We submit that given a choice one would w i l l i n g l y trade the mouse oncogenicity data for more complete information on the effects these chemicals have in the rat. Trichloroethane received a score of 45 because it caused not only liver tumors, like other chlorinated hydrocarbons do, but in addition produced adrenal pheochromocytomas (NCI, 1978e). These latter tumors in this instance were benign, and we contend that trichloroethane should be viewed as other ch!orinated hydrocarbons are, i.e. as not a particularly hazardou~ chemical. Fundam. AppL Toxicol. (3)

November/December, 1983

M O U S E C A R C I N O G E N I C I T Y B I OA S S A Y S

"FABLE 1 1 Bioassay Data for Aniline Derivatives

anisidine,p-

Rat

Mouse

-

-

aniline (HCI)

NH," HCI

Rat

Mouse

+

-

OCH, chloro-p-toluidine,3-

a m i n o - 2 - n i t rophent)l,4 -

~

OH

NH,

NO,

NH#

CH, chloruaniline,p-

d i m e t h o x y a niline,2,4-

~

© NH,

NH,

CH~

cl

OCH,

0

a n t h r a n i l i c acid

cresidine,m-

II

COH

~

NH,

~

NH,

CH,

OCH= 0

n i t r o a n t h r a n i l i c acid,4-

+

anisidine,o-

II

COH

@

~

NH~

+

chloro-o, toluidine,5-

NH,

+

cresidine,p-

NH, C|.~

oCHz

NH, CH=

chloro-o-toluidine,p-

+

toluidine,o-

~

÷

t r i m e t hyla niline,2,4,6

+

+

+

+

+

+

CH,

NH, CH~

Cl

nilro-o-toluidine,5-

NH,

NH2

H~C~]~CH~

.~CH~ O,N~"~.,~

CH~ O

chloramben

CI

Fundamental and Applied Toxicology

+

II

COH

nitro-o-anisidine.5NH,

~ NH2

(3) 11-12/83

oCH3

o,N ~ " ' ~

637

SCHACH yon WITTENAU AND ESTES The alkylating agent trimethylphosphate with a score of 45 might be expected to be mutagenic and carcinogenic. The studies in rats and mice appear to establish that it is not a powerful carcinogen, as it seemed to have caused in rats subcutaneous fibromas only, and carcinomas only in female mice, but not in males (NCI, 1978f). As this reagent is alleged to preferrentially alkylate thiol groups under physiological conditions (Ross, 1962), it is plausible to postulate that carcinogenicity is observed only at doses which significantly deplete the thiol groups in the tissues.

CONCL US~ON The data base selected for this inquiry is less than firm. The underlying experimental results probably are not reproducible in all instances. Their interpretations at times must have posed considerable difficulties to the reviewing panels, and on occasion different verdicts might have been reached by other similarly qualified groups. Nevertheless, these data typify much of the foundation upon which the classification of a chemical as a "carcinogen" rests. In that such classification forces the taking of regulatory positions, these data constitute reality, flawed as they may be. Furthermore, there is no published information of similar comprehensiveness but of better quality, This is easily understandable when it is realized that the cost of testing 273 chemicals in mice and rats today would consume scientific resources which exceed in monetary value $100 million. Proponents and opponents of the two-species "bioassay" have to rest their cases on data such as these. The high degree of redundancy between the rat and mouse data is to be expected. Most, if not all, mechanisms causing or SUl3porting tumor emergence should be potentially available in both species. Consequently whenever a species difference is observed it most likely will reflect a quantitative (potency) rather than a qualitative difference in the response, unless a factor such as a virus is present in one but not in the other species. One must be careful, however, not to interpret the previous sentence as supporting the position that a chemical associated with apparent tumorigenesis poses a carcinogenic hazard to all species. Toxicity is a function of dose and susceptibility, and at some point a quantitative difference becomes a qualitative difference in practical terms. Eleven percent of the chemicals surveyed purportedly were carcinogenic in mice but not in rats. It appears reasonable to assume that a higher concordance would have been attained if the tests had been performed according to modern protocols. For example, if cytotoxicity is the cause of mouse liver tumors associated with the administration of some chlorinated hydrocarbons, it w o u l d not be surprising to find liver tumors in rats if cytotoxic doses are used. It seems the mouse is more susceptible to this mechanism of action, but it probably is not the only species which is vulnerable to such toxicity. However, because the mouse appears to be extremely sensitive, we would argue in the context of human risk assessment that the mouse data are not very useful, if at a reasonable multiple of the human exposure, only the mouse but not the rat responds with hepatocellular carcinoma to the administration of a chlorinated hydrocarbon. The question of human hazard should not focus on oncogenicity alone, but should also address the likelihood of hepatic cell cytotoxicity occurring. A few aromatic amines purportedly are carcinogenic in mice but not in rats. Perusal of Table 11, which lists data from a larger sampling of this chemical class, again leads us to suspect that some of the bioassay results may not be reproducible;

638

no consistent pattern emerges when comparing structures with the biological data. Comprehensive knowledge of the toxicity, as opposed to information confined to yes/no on oncogenicity, appears to be needed for human risk assessment for these chemicals. In our survey we find little or no justification for including on a routine basis the mouse, in addition to the rat, in the bioassay for carcinogenicity. In the few instances in which mouse studies may not be redundant, they do not appear to provide information of much value to human risk assessment. In consideration of the cost in scientific resources that lifetime mouse experiments consume, the two-species bioassay approach should be discontinued by eliminating the mouse as a test species. Admittedly, it could be argued that if only one species is used it could be the mouse instead of the rat. However, comprehensive toxicology is more commonly evaluated in the rat, and for that reason it is a more suitable species for long-term toxicology studies. In our opinion the rat "bioassay'" for oncogenicity should be redesigned to allow for an assessment of toxicity, not only for one aspect of it, namely oncogenicity. It may be defensible to limit the scope of inquiry to only one question if the investment of time and resources is limited, but such is not the case for life-time rodent studies. Furthermore, a single y e s / n o answer as to an increase in tumor count without additional information as to other biological effects and their dose relationships is no longer acceptable for intelligent assessment of human risk. Such assessment is complicated by the deplorable absence of systematic studies which seek to establish the validity of current models by defining uniform experimental conditions for the reliable identification of known human carcinogens. If such validation effort were made it would not be at all surprising to find that uniform conditions can not be delineated. All too often it is forgotten that cancer may have various causes, and credible extrapolations from animal to man and from high to low doses are not possible without some understanding of mechanisms.

A CKNO WLEDGMENT We thank Ms. J.V. Soderman for helping to prepare this manuscript and particularly for recasting data from her Handbook into the format of our tables. REFERENCES

Becker, F.F. (1982). Morphological Classification of Mouse Liver Tumors Based on Biological Characteristics. Cancer Res. 42:3918-3923. Berenblum, 1. (1974). Carcinogenesis as a Biological Problem. Ch. 1, 3. American Elsevier Publishing Co. Inc., New York, NY. Bonser, G.M., Clayson, D.B., Jull, J.W. and Pyrah, L.N. (1952). The Carcinogenic Properties of 2-Amino-l-naphthol Hydrochloride and Its Parent Amine 2-Naphthylamine. Brit. J. Cancer 6:412-424. Butler, W.H. (1981). Meeting Report: Current Perspectives in Mouse Liver Neoplasia. BIBRA 13ultetin, December, pp. 528-529. Chu, K.C., Cueto, Jr., C. and Ward, J.M. (1981). Factors in the Evaluation of 200 National Cancer Institute Carcinogen Bioassays. d. ToxicoL Environ. Iteahh 8:251-280. CIIT The Chemical Industry Institute of Toxicology (1981). Current Perspectives in Mouse Liver Neoplasia. Presented at the Fourth CIIT Conference on Toxicology, December 3 and 4, 1981, Raleigh, NC. CSM Committee on Safety of Medicines (1979). Carcinogenicio, Testing o f Medicinal Products~ C SM, London. Fundam. AppL Toxicoi. (3)

November/December, 1983

MOUSE CARCINOGENICITY BIOASSAYS Deichmann, W.B. (19811. Halogenated Cyclic Hydrocarbons. in Patt.vh" Industrial tlvg&ne and Toxicology, Clayton, G.D. and Clayton, F.E., eds., Third Ed., Vol. 3B, Ch. 49, pp. 3686-3687. lohn Wiley and Sons, New York, NY. Deichmann, W.B. and MacDonald, W.E. (1976). Liver Cancer Deaths in the Continental USA from 1930 to 1972...|t~1. htd. Ilyg. Assoc. d. 37:495-498. Deichmann, W.B. and MacDonald, W.E. (1977). Organochlorine Pesticides and Liver Cancer Deaths in the United States, 1930-1972. Eeoto.ric'oL I:.*ttviron. Safet.l' ] :89-110. Emmerson, J.L. (1982). Personal communication. EPA Environmental Protection Agency (1979). Chronic Effects Tesl S t a n d a r d s and Good Laboratory Practices. k~,d. Reg. 44:27334. Frith, C.H. and Ward, J.M. (1980). A Morphologic Classification of Proliferative and Neoplastic Hepatic Lesions in Mice. J. Ettviron. PathoL To.rieol. 3:329-351. l:SC Food Safety Council (I 982). A Proposed Food Safe U' Evaluation Process. Final Report o f the Board o/` Trustees. Food Safely Council, Washington, DC. Grasso, P. and Crampton, R.l:. (1972). The Value of the Mouse in Carcinogenicity Testing. Fd. Costlier. Tt~.vicoL 10:418-426. Grasso, P., Crampton, R.l:. and Hooson, J. (1977). 77w Mouse and Carcinogenieity Terlblg. The British Industrial Biological Research Association (BIBRA), Surrey. Griesemer, R.A. and Cueto, Jr., C. (1980). Toward a Classification Scheme for Degrees of Experimental Evidence for the Carcinogenicity of Chemicals for Animals. In 31olecldar arid Cclhdar Aspects o f Carcit~ogen Screening Tests, Montesano, R., Bartsch, H., Tomatis, L. and Davis, W., eds., IARC Scientific Publications No. 27, pp. 259-281. International Agency for Research t,n Cancer, Lyon. Hadidian, Z., Fredrickson, T.N., Weisburger, E.K., Weisburger, I.H., Glass, R.M. and Mantel, N. (1968). Tests for Chemical Carcinogens. Report on the Activity of Derivatives of Aromatic Amines, Nitrosamines, Quinolines, Nitroalkanes, Amides, Epoxides, Aziridines and Purine Antimetabolites. ,L Nat. Catlcer It:st. 41:98.5-1022. NCt National Cancer Institute (1978a). Bioassay o/`l'gstradiol Mustard f o r Possible Carcinogeniciu,. Technical Report Series NCI-CG-TR-59. DI-IEW Publication No. (Nil-l) 78-1309. NCI National Cancer Institute (1978b). Bioassa.vofTr([luralinJbr Possible Carcinogenieity. Technical Report Series NCI-CG-TR-34. DHEW Publication No. (NIH) 78-834. NCI National Cancer Institute (1978c). Bioassay of5-Nitro-otoluidine f o r Possible Carcinogeniciu,. Technical Report Series NCI-CG-TR-107. DHEW Publication No. (NIH) 78-1357. NCi National Cancer Institute (1978d). Bioassay ql'3-Amhlo-4etho.wacetanilide f o r Possible Carcinogenicity. Technical Report Series NCI-CG-TR-112. DHEW Publication No. (NIH) 78-1367. NCI Natiohal Cancer Institute (1978e). Bioassay ofl,l,2-Trichloroethane f o r Possible CarchtogeniciU'. Technical Report Series NCI-CG-TR-74. DHEW Publication No. (NIH) 78-1324. NCI National Cancer Institute (1978f). 13ioassay o f Trimethytphosphate ./'or Possible Carchtogeniciu'. Technical Report Series NCI-CG-TR-8I. DIqEW Publication No. (NIH) 78-1331.

Fundamental and Applied Toxicology

(3) 11-12/83

NCI National Cancer Institute (1979a). Bioassay o f 4-Chloro-otO'luidine ld.vdrochloride for Possible Carcinogenicity. Technical Report'Ser[e~NCI-CG-TR-165. DHEW Publication No. (NIH) 79-1721. N ~ I National Cancer Institute (1979b). Bioassa.v of 5.Chloro-otoh~;i~tim'. Technical Report Series NCt-CG-TR-187. DHEW Publica|ion No. (NIH) 79-1743. NCI Sontag, J.A., Page, N.P. and Saffiotti, U. (1976). Guideline.~for Carcinogen 13ioassay in Small Rodents. Ca rcinogenesis Technical Report Series 1, DHEW Publication No. (NIH) 76-801. O'Connor, 111, C.A. and Woodward, S.C. (1981). Industry's Role in Cancer Research: Anticipating Regulatory Problems. Reg. 7bxicol. PharmaeoL 1:316-334. OECD Organization for Economic Cooperation and Development (1981 ). OECD Guhh, lim,.~'J'or Testh:g q/" Chemicals. Section 4, 8. OECD, Paris. PHSNTP Public Health Service National Toxicology Program (1982). Chemicals (16) Nominated for Toxicological Testing. Fed. Reg. 47:44884. Purchase, l,l:.H. (1980). Interspecies Comparisons of Carcinogenicity, lit. J. Cancer 41:454-468. Riley, V., Fitzmaurice, M.A., Spackman, D.I-I. (1981). Psychoneuroimmunologic Factors in Neoplasia: Studies in Animals. In Ps.vchoneuroimolunolog.|', Ader, R., ed., pp. 31-102. Academic Press, New York, NY. Ross, 'I,V.C.J. (1962). Biological Alkylatitlg Agents, pp. 7-8. Butlerworths, London. Schach yon Wittenau, M. (1979). Cancer Risk Assessment. Scictic'e 206:1258-1260. Schumann, A.M,, Quasi, J.F., Watanabe, P,G. (1980). The Pharmacokinetics and Macromolecular Interactions of Perchloroethylene in Mice and Rats as Related to Oncogenicity. Toxicol. Appl. Pharmacol. 55:207-219. Soderman, J.V. (1982). CRC Itandbook o f Identified Carcinogens anti Noncurch:ogens: Curchlogenh'it.v - Mtt/ageniciu' Database, Vol. I and 11. CRC Press, Inc., Boca Raton. Squire, R.A. (1981). Ranking Animal .Carcinogens: A Proposed Regulatory Approach. Science 214:877-880. Stewart, H.L. (1975). Comparative Aspects of Certain Cancers. Cancer-- A Comprehensive Treatise, Becker, F.F., ed., Vol. 4, pp. 320-329. Plenum Press, New York, NY. Stott, W.T., Quasi, J.l:., Watanabe, P.G. (1982). The Pharmacokinetics a nd Macr~.~molecular Interactions of Trichloroethylene in Mice and Rats. Toxicol. AI)pL Pharmacol. 62:137-151. Stott, W.T., Teitz, R.H., Schumann, A.M., Watanabe, P.G.'(1981). Genetic and Nongenetic Even ts in Neoplasia. Fd. Cosmel 7bxicol. 19:567-576. Tomatis, L., Partensky, C. and Montessane, R. (1973). The Predictive Value of Mouse Liver Tumor Induction in Carcinogenicity Testing - - A Literature Survey. hit. J. Cancer 12:1-20. Vesselinovitch, S.D., Mihailovich, N., Wogan, G.N., Lombard, L.S. and Rao, K.V.N. (1972). Aflatoxin B1, a Hepatocarcinogen in the Infant Mouse. Cancer Res. 32:2289-2291. Ward, I.M., Griesemer, R.A. and Weisburger, E.K. (1979). The Mouse Liver Tumor as an Endpoint in Carcinogenesis Tests. ToxicoL AppL PharmacoL 51:389-397. Ward, J.M. and Vlahakis, G. (1978). Evaluation of Hepatocellular Neoplasms in Mice. d. Natl. Cancer Inst. 61:807-811.

639