Document LpnDDgn65qx26R7VMNKaqVVJg

r * "Ur.*.. is * *r`-. uri io a j on the p. That. hlordanc Steroids 5cnal and insure to omlnated osierone, Steroids W'-HCBJ of 4-,4C lormueof. biphenyl* mp!ated OPMI Of action of ] levels In {PCB} on Bnviron. t sleeping ino'mice. aw York: ebiohenyl biphenyls ,f adrenal -53*. f estradiol rone {PJ- 77, )9?9 79, J979 INTERSPECIES COMPARISON OF CARCINOGENIC POTENCY RECEIVED FEB 15 1980 Edmund Crouds, Richard Wilson J V. BmHR Energy and Environmental Policy Center, Jefferson Physics Laboratory, Harvard University, Cambridge, Massachusetts For guidance In decisions on how to safeguard humans from carcinogens, ft h necessary to use data on carcinogenesis In an!mo's. This paper discusses how such data, combined with human experience, may Pt used quantitatively in such decisions. It is demonstrated efiipir/ca/iyi that gvod correlations exist between different species for suitably defined carcinogenic potencies for various chemicals. This allows sufficient accuracy In extrapolating from animal data to human riskjo support a logical scheme for the evaluation of such risks. Some recommendations for future research art given. INTRODUCTION A prudent policy for cancer prevention requires that we use data on animal carcinogenesis as a quantitative predictor of human risk. No available theory' allows such quantitative predictions, although they should be possible if the animals' metabolism is similar to that of humans. In this paper we outline and assess experimental data on which such comparisons could be based. in the next section wc give the definitions and assumptions used in this paper: the basic assumptions concerning the relation between cancer incidence and dose of carcinogen {a linear, no-threshold theory), and the definition of carcinogenic potency {for a given chemical and species of animal). The third section shows the results of computing the potency of various chemicals for various species. In particular, the potencies for -- 70 chemicals are evaluated in 2 species from a set of experiments with nearly identical designs; there arc good intcrspecics correlations between the potencies, allowing intcrspecics extrapolation within about an order of magnitude. Most important is a set of comparisons of carcinogenic potencies in animals and humans derived from the limited studies avail able. Despite the shortcomings of the "experimental design" of this data set, the results are consistent with the possibility of extrapolating between species, in particular between animals and humans, within a factor of 10. We thank many colltagues lor Interfiling discussions end Mr. Mike Sylvanus and Ms. Laura Hcdil for their help in locating tht references for this paper. Requests for reprints should be sent so Edmund Crouch, Energy and Environmental Policy Center, Jefferson Physics Laboratory, Harvard University, Cambridge, Massachusetts 02138. 109J Journal of Toxicology and Environmental Health, 3:1095-1118. 1979 Copyright 1979 by Hemisphere Publishing Corporation 0028-4108/79/051095-2412.25 AP00018450 f j t09fi C. CROUCH AND ft. WILSON r i: On tills basis, a scheme is discussed in the fifth section for quantita) v `Ov* tivcly estimating carcinogenic risk to humans. The accuracy of such . }CC' ;; 9 j^stimates depends on the number of carcinogens included. Before this, a rew of the possible confusing effects resuiting from metabolism and f-'ro*'* synergisms are mentioned. Possible exceptions to the proposed scheme are ' discussed in the sixth section. Finally, possible improvements, modifica- ; tions, and extensions are discussed, as well as ways to further verify the . scheme. ft r 1- - -: ASSUMPTIONS AND DEFINITIONS We make the following assumptions and test the results against experiment: 1. The lifetime probability Pof an animat getting cancer depends on the Integrated lifetime dose of the carcinogen. (We may restrict ourselves to cancer of a particular type or at a specific site.) 2. The relevant measure of dose is the ratio of the weight of carcinogen to the body weight of the animal. 3. Assumptions 1 and 2 hold for lifetime ingestion at constant dose rate, and to first approximation for nonconstant dose rates if the dose is spread over an appreciable fraction of a lifetime. In what follows, we always quote integrated lifetime dose as an equivalent lifetime average dose rate (mg/kg*d). ' 4. At low doses the dose-response curve is linear. We define the potency of a carci.-togcn as the gradient of the line P = a + 0d oc>0 0>O (7) where of is the probability of getting cancer In the absence of carcinogen, d is the measure of dose of carcinogen (mg/kg'd averaged over a lifetime), and $ is the potency of the carcinogen (kg*d/mg). 5. At high doses, because the probability cannot exceed 1, we assume the dose-response curve saturates to. 1 exponentially. Tins assumption is needed for some of the animal experiments, where large fractions of the experimental populations got cancer. The form of the dose-response curve we use is /*-l -(!-) exp (-1^-) (2) which reduces to Eq. (1) for low doses, d < (1 -- )/0. 6. In animal experiments, the number getting cancer is binomially distributed with probability P, i I s 6 y w w q n Jwn * AP00018451 iH IN7ER5PECIE5 COMPARISON OF CARCINOGENIC POTENCY 1097 The question of linearity at low doses has been argued at length, but 1 Eq, (2) has the merits of simplicity and of providing a plausible upper bound for a wide range of possible dose-response relationships (Guess et i a!., 1977), and hence is generally considered conservative. The absence of e a threshold level for carcinogenic action is ensured by the restriction t- a> 0, again to obtain conservative results for low doses. There is good e evidence that, assumption 5 above is incorrect in at least one ease [vinyl chloride; see Maltoni et ah, (1974}} and that it is possible for P to saturate to some value less than 1. This possibility makes little difference to the results reported here. The parameters or and p are estimated from * experimental data, using maximum likelihood techniques (Appendix A). In st ^ most cases, experiments were performed at no more than two nonzero dose levels, and for many, especially studies of humans (Appendix B), only one dose level is available and the estimate of dose may have a large error. 1C CO 1NTERSPECIES COMPARISONS of Mouse and Rat se The National Cancer Institute (NCI) series of Carcinogenesis Bioassay Is Reports describe experiments performed with similar experimental designs t/e ' on rats and mice, containing sufficient data to compute values of the 6* carcinogenic potency* Maximum likelihood estimates for the parameters et and j3 in Eq. (1) were computed for those results in the carcinogenesis he reports* that were considered statistically significant therein, together with some that were not statistically significant. Tables 1-3 list the potencies computed for the site of tumor incidence giving the greatest potency, even ,1) though in some eases the incidence at this site was not considered statistically significant, or other sites had a tumor incidence of greater significance. (Testicular tumors in male Fischer 344 rats were ignored.) in, . These results are plotted in Figs. 1-5. Figures 1-3 compare potencies e). in males and females of three species. In all cases there Is.good correlation (dashed lines are lines of equal potency), the potency in males being a me good predictor of the potency in females and vice versa. Furthermore, the. is potencies arc approximately equal over a wide range of values; best-fit the straight lines of unit slope in Figs. 1-3 indicate slightly higher potencies in rve male mice and Fischer rats by factors of 1.1 and 1.7, respectively. Figures 4 and 5 show intcrspecics correlations in potency. The geometric mean potency in males and females of one species is plotted against the same U) mean for the other species. For the Osbornc-Mcndel rat versus the B6C3F1 mouse (Fig. 4) the correlation is excellent, and for the Fischer 344 rat versus the B6C3F1 mouse (Fig. 5) it is still good. In both cases, best-fit lines of unit slope (dashed lines) lie within an order of magnitude lily of most points and, with outlying points omitted, give ratios of potencies 1 We had analyzed 90 such reports at Die time this paper was reviled. j f3 r I I J \ f t * A .'' i 1 ' T '\ * '., * ! . ii i i 109ft t. CROUCH ANU R. WILSON TABLE 1. Potencies of Oiemicati by Ingestion Potency (kg-d/mg) Osborne-Mcndi-I rat B6C3H mouse Chemfeal Male Female Male Female Chloroform Chfordeeone Trrchloretlhylrne CMordane HcpUchlor Oichlervos Tetrachloroethylene Aidrln Dieldrln Picloram Chiorambcn Nitrofen Tetrachloroethane Tetrichlorvinpho* Trifluratln MetftoxyeJor 'Carbon tetrachloride Endrln Oiforottulonil 1,2-Oichloroethane 1,4-Dioxane 1,2- Dibro mocthane" Dioxathion Pifithion 2E-3? SE-2 <t-3 2E-2 IE-2 c,d IE^ jE-a* 7E-5f 2E*4 IE-3f 1E-3C 2E-4C 4E-4 4C-3 2E-3'C 9E-I 2E-4 3E-3 1 E-3 2E-1'* 3E-2 IE-1 2E-3 ie-i . t,d IE-2 !? 1E.3` e,rf 6E-2 SE*2 3E-4 IE-4 IE-3 4E*3 36*4 2E-4 2E-3fc IE-3 ?E-1 2E-4 6E-3 6E-4 IE.,*.* 2E-2f IE-I SC-3b EE-I 7E-4 2E-1 SE-1* IE-3 SE-1* 4E-t* e IE-4* SE-3* 6E-3 IE* 3 2E*4f 4 E-3 3E-1 2E-3* 7E-4* 9E-2*'e 2 -3 XE-3C IE-2 2E-1 2E-4 9E-2* 3E-1 . IE-3 9E-2C 1EIe e 9E-5 8E-3 IE-2 2l\ IE-3 3E-3 2E-3 2E-3 IE-1'* IE-3 4E-4C value AE-B means A X 10. *Significant effect noted in oriflinsl study. ^Computed potency is less than twice the sensitivity. Study lasted only about half the lifetime because of high mortality. * High-dose group omitted from analysis because of high mortality. in Osborne-Mendel rats and Fischer rats to potency in B6C3F1 mice as 0.40 and -- 4.5, respectively. There is no a priori reason why any correlation should correspond to direct proportionality (unit slope on the figures). It is, however, the simplest hypothesis and it gives good agreement with the data. Further details of the derivation of the potencies from the experimental results are given in Appendix A. j With these data, the experimental evidence on carcinogenesis in one species can be used to estimate the carcinogenic potency of the same chemical in another species by simply multiplying by an intcrspccics relative sensitivity factor. However, this argument applies only to potency summed over all tumor sites, or to potency at the most dominant site. The data show that $ i V '* INTC > : ' APOOOt8453 `IN as iO te ef re te ie es all at INTERSPECIES COMPARISON OF CARCINOGENIC POTENCY 1099 TABLE 2. Potencies of Chemicals by Ingestion Potency (kg'd/mg) ! Fischer 344 rat flC3Fl mouse ' Chemical Male Female Male Female Proflavine i&a SM U-2 * Nitrilewiaccttc acid (NTA) 2E-4 SE-4 4E-4 6E-5 Na,NTA 4E-4 IE-4* Mgj NTA 7E-4* CE-4* _ Ptienformin 9E-3 6E*3 6E-4 2E-4 J EDTA 4E*4 2E-4* 2E-4 IE-4 Oapsone IE-2 2Mf IE-4 4 2<MethyM-niTOanthfiqutaone 7E-3 4E-3 Arocior Lasiocarplne 5E* 7E-1 6E'f,e __ Tris (2,3-dibromopropyl) phosphate (TBP) 2,4-Diamlnoanlsole 2E-1 2E-3^ 3E-2? E*4 5E-36 BE-4 6E-3 6E-4 Ethionamide Acetohexamide >Nitroproprionie acid 2-Amlno-S-nluothiacole 2J4-0inltrotoluenefi' 4-Chloro-o-phenylenediamine 3E-4 1E*2. 1E-2* 3E-r 2E-3 IE-3 2E-S 2t* 6J JE-aJ 2E-3 IE-3 9 6E-3 2E-3 a 5E-46 IE-3 9 SE-3 IE-2 2E-4 APC mixture 2E-4 2E-4 IE-4 3E-5* Tfimethylphosphate 4E-3 2E-3 6E-S 2E-3* 4-Chloro-m-phenyIenediamine df-Mer.thol 9E-4* 4 7E-4 (E-S 2E-4 2E-46 2E-4 Phcnatopyrldine hydrochloride IE-3 6E-4 IE-3 7E-3 Anllialne 3EO 4E-3^ 9 0 5-NItro-o-toluidine 2E-3 2E-3 4 a 8-Amino-4>ethoxyacetamide S*Nitroaecnaphthcne a \Z-*`e a IC-j'* 3E-4 9 9E-5* IE-2 Piperonyl sulfoxide 8E-S 4E-4 f E-2 8E-4 2,5-ToIucnediamine sulfate tE-3? IE-3. 2E-3t 2E-3 Nitro-o-an|$idine 6E-3* 9E-4* 3E-4 !E-4 Trimethyl thiourea 4E-3* 2E-2* IE-3 4E-41' Aniline hydrochloride 2E-3 3E-4 4E-S? 7E-S 1 Cl. Vat Yellow 4 SE-5 3 E-S ^Computed potency less than twice the sensitivity (see Appendix A). Significant effect noted in original study. *High-dose group omitted from analysis because of high mortality. Benign tumors In rats (skin in male, mammary in female). - *1 $ I t i f* t: AP00018454 noo E. CROUCH AND R. WILSON TABLE J. Potencies of Chemicals Potency (kg-d/mg) ft a Sprague-Dawley rat B6C3F1 mouse Chemical Mate Female Male Female Uophosphamide IE*2 ' IE-1 5E-2 7E-2 S'Aaacytidlne Emetine Acronycine 4E-1* 9E-1* JE->e 7E*1tf 4E-1 IE-26 *1 b 6 0*2*->oxy-6*thloguanoiine 14 monohydratc (jS-TGdR) 9L-f 2-le Tris (Wziridinyl) phosphine sulfide 2E-Oc 260* 3E-Oc stoF Phenoxybenzamine hydrochloride 6E-lf 2E-IC 1E-1C 9E-2* Estradiol mustard* $1 46*1 9E-2 9E*2C Phene*terin* IE-2 3E-1e 1 E*1c 2E*tc *High-do$e group(j) omitted from analysis because of high mortality. High early mortality precluded useful analysts. Significant effect noted in original study. "Gavage. All others by ip injection. 10-I To* E Wio-2 < 2 UJ _ r' ira" B6C3F1 MOUSE TfMMlilfltMwIUtt ^ TnllutMiii V + ' Tf + *. 4if*- + '+ +' 4- +*jponyl Switauf* s '+ 10* itf5 icT4 Kf to.'2 .10 POTENCY IN MALE (mg'1 kg d ) FIGURE 1. Carcinogenic potency in 86C3FJ mouse: male versus female. > IN7 ' APOOOf8455 . UON INTERSPECJES COMPARISON OF CARCINOGENIC POTENCY ; y `* +w j +4 * 3 < .o2- zUi 1101 + >'+ ^+ OSBORNEMENDEL RAT .i !0'4 t'6"3 10 2 I0'1 t POTENCY IN MALE (mg*1 kg d) FIGURE 2. Carcinogenic potency In Osborne-Mendel rat: male versus female. } r. i APOOO18456 It02 . CROUCH AND R. WILSON H510-t r z i- 10 DimitHW + + v .4 to itfS iff4 Kf* Iff2 Iff1 POTENCY IN B6C3F1 MOUSE (mg'kgd) .FIGURE 4 Geometric meant of carcinogenic potencies in mate* and females: Oiborne-Mundei rat versus B6C3FI mouse. J "D 5 ffio1 < (C -2 5 |-3 10 CO E : 10 + 4- * ' '+ 4'FMiuxeetT<d<ne MftfrocMenda SuMmMe -5 to 1a"5 tcf4 icf3 iff2 to'1 1 POTENCY IN B6C3F1 MOUSE(mgkg d) FIGURE S. Geometric means of carcinogenic potencies in males and females; Fischer 344 rat versus B6C3F1 mouse. INT the bet alto 3 the be 1 WOT une: com than a di' 1 .of 1 gooc strait card conti dnoj Hfcti rate the. c of tl age, incid such Its si: when is a : exam be ccorre range proce for a tumor At report and l review aS*aMin ' *-T-* ","fV h 'v.s " '*4: *- AP00018457 WILSON id rai INTERSPECiES COMPARISON OF CARCINOGENIC POTENCY 1103 the dominant site in these mice is often the liver! whereas the site could he thyroid, kidney! or mammary gland in the rat (and perhaps another site altogether In humans). Because our concern is for cancer, no matter what the site, this simplification is important. Humans and Other Animals There are few good data on human cancers from which the cause can be even probably assigned. The available data are primarily for a few workers industrially exposed to large concentrations of a substance and unexposed workers with whom their mortality or morbidity can be compared. The potencies derived from these data will be even less precise than those obtained from controlled experiments. The animai data with which we compare the human cases comes from a diverse set of experiments with different protocols and different strains of the species considered. We cannot, therefore, expect correlations as good as those obtained in the last section (e.g.f difference between 2 rat strains of a factor of -- 10). It is necessary in some cases to estimate carcinogenic potencies from experiments in which observation was not continued for the lifetime of the experimental animals and/or the car* cinogcn was administered to the animals for only a fraction of their lifetime. In the latter case it was assumed that the lifetime average dose rate was the controlling variable (assumption 3), while in the former case the observed numbers of cancers were corrected for those missed because of the limited observation period. Because cancer incidence varies with age, this correction could have been made by using typical data on cancer incidence, but was in fact performed by using a simple theory that fits Such cancer incidence. We used the theory of Armitage and Doll (1954) in its simplest form, in which the rate of appearance of cancers is given by ^ = X/* dt (3) ' where dN{dl is the rate of appearance cancers, t is the elapsed time, and k Is a function of the site of the cancer; empirically. 2<A<8. Thus, for example, if the animals are observed only half their lifetime, the data can be corrected to a lifetime incidence by multiplying by 2**1. The correction will be uncertain to the extent that k is uncertain within the range 2 </?<8. Various authors have gone through the details of the procedure (Gchan, 1969; Peto, 1974), which can describe the cancer hazard for a wide range of variations from the linear theory (lengthening of time to tumors, thresholds, repair mechanisms). An attempt has been made in a National Academy of Sciences (NAS) report (1975) to perform the comparison of this section, and Meselson and Russell (1977) compare carcinogenic and mutagenic potency. We review and suggest changes in these results (converting them to car- AP00018458 1104 E. CROUCH AND X. WILSON cinogentc potencies) and add some new ones, mostly upper limits. The resulting comparisons are shown in Tabic 4 and their derivations are discussed in Appendix 6. The same results are shown graphically in Figs. 6 and 7, where it is evident that they arc consistent with correlations between human and animal carcinogenic potencies similar to those demon strated above between mouse and rat (dashed lines on the figures arc lines of equal potency). As expected, these data are not so well correlated, but the intcrspccics sensitivities appear to be 5:1 for both human:mouse and humantrat. Various agencies, in attempting to derive carcinogenic potencies in humans from potencies in animals, have included multiplicative factors to account for differing sensitivities in different species. The NAS report { assumes that animals and humans arc equally sensitive when they have the I same total intake (as a fraction of body weight) during a lifetime. The | Food and Drug Administration (FDA) assumes that they are equally . sensitive when they have the same fraction of pollutant in their food or i jwater intake, whereas the Environmental Protection Agency (EPA) makes a correction for surface area by the factor (Afhumin/Manimii)''3 where M is the mass. The procedure used here is to derive these intcrspccics relative TABLE 4. Comparison of Potencies: Animal and Human Chemical Mouse Potency^ Rat Dog Human^ AN AHatoxin 8, A* Benzene Btnildin* Chtprnaphazine Chloroform DC6 Oiethylstilbestrol EDB Lead acetate Saccharin Vinyl chloride Radiation^, (rem/yr)-1 Smoking*, (no./d'lg)"1 130e 1.5-30 - 0.0003 o.oa 20 o.oi 0.006 14 6 0.001 -- 0,004 0.01 0.06 0.06 $00-1300 <0.01 -0.0008 130-2500* -- 0.002 0.025 -- 6 0.007 0.0003c 0.01 -- - -- -- -- 0.2 -- -- O.M - -- _ -- _ - <0.3 200(3) 15 (3) 0.001 (3) 34 001 2(t0) <0.001 <5 1r (10) 0.8 (10) < 2.S <0.04 04)2(3) 0.02 (3) 0.06 (3) ^Values are kgd/mg except where noted. Number in pjrenthesct next to human potency it our estimate of the accuracy of the number. ^includes Intrauterine exposure. "Oral administration. Value for sc injection b 0.04 in rat and 0.03 in mouse (see text and Appendix B). `Women ingesting pills In pregnancy, resulting in cancer in their dauKhters. *Net included in Pig. 7. *v to I-1 eT & | s 2 10-3 - +' td* T Kf2 POTENCi FIGURE 7. Carcinog AP00018459 09*81000dV !r I* :; % !: 'if .i Ii .r fc: I r. i f. -4 I ti % -t:; r V\ i: >/ \ V :> 1106 C. CROUCH AND R. WILSON sensitivities from experiment. Table 5 lists the relative sensitivities (using our definition of potency). METABOLIC EFFECTS insofar as possible, we compare data from experiments in which the same method of application of the chemical was used for both animal and human--and the exposure is for the same fraction of the lifetime in both animal and human. In some eases we have to compare animal exposure by ingestion .with human exposure by inhalation or dermal contact, and the comparison is then (css accurate. Ideally, we would like to use data on animal and human metabolism to describe how the carcinogenic chemical reaches the organ of interest from the various sites of administration. The rat experiments with vinyl chloride show equal potencies for liver cancers by both ingestion and Inhalation. For chemicals metabolized in the lung, such as benzo[ffjpyrene, we might expect a bigger difference. For benzofoj pyrene applied on the skin, the sensitivity of mice apparently depends on the solvent system used (Bingham and Faik, 1969). In this experiment, the concentra tions of benzoftfj pyrene in 2 different solvent systems required to elicit cancers differed by a factor of 1000, although the difference in numbers affected in each case reduces any apparent potency variation to 500. This difference may result from the relative efficacy of the two solvent systems in transporting the bcnzo[ojpyrcnc into or through the skin. By using the potency for ingestion given in Table 4 and the doses used by Bingham and Falk, the remarkable observation is the lack of sensitivity when the benzo{a]pyrene was applied in the "inert" solvent rather than the sensitivity when it was applied with the "cocarcinogcn." In some cases, giving a small number of large doses may result in fewer cancers than giving the same total amount in small daily doses, because metabolic activity may be saturated in the first ease. For this rcasun, animal studies utilizing sc injection may be less sensitive than continuous ingestion or Inhalation studies. We note here that the correlations may be upset by synergistic effects. TABLE 5. Relative Sensitivities between Species Agency National Academy of Science* (1975) Food and Drug Administration (* in diet) Environmental Protection Agency (corrected for surface area) This paper (experiment) Rat -1.5 ! 1 -1-3 Mouse \ 0.3S 0.43 1 Human -- 35 4 4.7 SS P. fa lai fir (F of ex. dc: hu ant she APOOO18461 . t\,SON our the and aoth e by the ilism erest vinyl and >[a]- plied ivent ntraclicit ibers This .terns g the i and i the r the fewer cause ason, tuous fects. tNTERSPECJES COMPARISON OF CARCINOGENIC POTENCY 1107 In the animal experiments these effects should be small or absent, because the environment is controlled. However, our estimates of potency in humans necessarily include any synergistic effects resulting from other environmental carcinogens, so that the potency of the chemical acting alone on humans may differ from that given here. From a practical point of view this is irrelevant, provided the background of other carcinogens or cocarcinogens to which humans arc exposed does not vary too much. POLICV IMPLICATIONS The results of the third section confirm that a logical policy can be followed, independent of the particular material involved. 1. Animal experiments at high doses determine the carcinogenic potency of the material. 2. Interspecies relative sensitivities allow comparisons of the animal experi ments and extrapolation to humans, to obtain a good estimate of the (high-dose) carcinogenic potency in humans. 3. Use of the linear, no-threshold model gives a conservative estimate of the low-dose carcinogenic potency (l.e., that it is equal to the high-dose potency).' 4. Evaluation of the expected human exposures (doses) allows computa tion of the expected human risk: for exposure to (lifetime average) dose d (mg/kg*d)f the probability or cancer caused by the carcinogen is {3d, where 0 is the potency in kg-d/mg (provided (3d < I). 5. Any available data on metabolic differences between animal and human may be used to modify this preliminary risk estimate. 6. A decision is then required as to whether the resultant individual risk or the total societal risk is acceptable. EXCEPTIONS Some materials do not fit Into the scheme we have outlined, and such possible exceptions are Identified by name on the figures. In Fig. 1, piperonyl sulfoxide, trfmethylphosphate, and trifluralin lie substantially farther from the equal potency line than the other materials. Whereas the last two may simply represent the tails of a distribution, the position of the first in Fig. 5 confirms its exceptional nature. Similarly, acetohexamidc (Fig. 3) and phcnazopyridinc hydrochloride (Fig. 5) may simply be cases of large random error, but parathion and dioxaihion (Fig..4) stand out as. exceptions. The latk-of-xomrollcd experiments and different experimental designs makes the identification of such exceptions more difficult in the human-animal comparisons, but there is a clear difference between Wistar and Sprague-Dawlcy rats in the ease of benzidine (Appendix B), and As shows exceptional behavior in Fig. 6. AP00018462 .\ r1 L -j f v r- \*_* * -4 ' fj h i~- I* ti . C. CROUCH AND fc. W< ISON tNTt 1 These exceptions may represent differences in the metabolism.of these particular chemicals in these particular species (or even in one sex of a species) and are obvious candidates for metabolic studies and testing in other species. As (and possibly Pb) may have to be treated slightly differently* because in the case of such materials (as opposed to the organic chemicals otherwise included) we might expect the carcinogenic fron curv aver cons com. ' potency to be largely independent -of the (inorganic) chemical form, except insofar as those forms are differently mctaholi/cd. level of a , label IMPROVEMENTS AND EXTENSIONS The interspecies relative sensititMes, especially between animal and likeli are t human, need further work. It would be useful to obtain them between the animal strains currently used for carcinogenicity testing and humans, by retesting on those strains the chemicals for which the potency in humans is known, using tests with exposure routes and times similar to those in the human exposure. There is evidently a wide variation in sensitivity between strains of a given species (e.g., the Fischer and Osbornc-Mcndei rats discussed in the third section have relative sensitivities differing by a factor of ~ 10). Correlations between mutagenic potency (in In vitro tests) and car cinogenic potency should be searched for using quantitative measures of both. Mcsclson and Russell (1977), made a quantitative comparison of animal potency and mutagenesis for several chemicals. Direct correlation between mutagenic potency and human carcinogenic potency is even more desirable. We would welcome details of evaluations of mutagenic potency for the chemicals in Tables 1-4. T in Ta statis The . indui type APPENDIX A i sensit cancc We give here details of the derivation of Tables 1-3. All arc based on the NCI carcinogenesis bioassays, reported in their technical report series (NCI-CG-TR). Each report gave experimental design, together with results consisting of the number of animals with tumors of various types (sites) N under ignore from compared with the number of animals examined at those sites. Doses were computed from the experimental designs. Dose was given as (1) milligrams per kilogram or (2) fraction of diet. In ease 1, it was straightforward to compute total lifetime dose (mg/kg) and average over the assumed lifetime Ai (728 d in rats and 637 d in mice). In case 2 curves of food consumption TJ (in body weights) as a function of age and sex were constructed from and it control animal feeding data in NCI-CG-TR-2 for B6C3F1 mice and comp; Osborne-Mendcl rats. These were assumed to be typical for all the mice needei and rats in all the experiments, except for a correction for body weight: Intern food consumption (in body weights) was assumed to vary as (full-grown used i weight)"1'3. The full-grown weight was estimated in each case by eye Table < APOOOt8463 INTERSPECIES COMPARISON OF CARCINOGENIC POTENCY .1109 from the growth curves in the technical reports. The weight-adjusted curves allowed consumption of lifetime doses (mg/kg), which were averaged as in ease 1. (Better weight corrections could be made, but it was considered that errors resulting from the procedures adopted are small compared with statistical fluctuations.) - Thus the experiments gave, for each tumor type or tumor site, a dose level (d{)t the number of animals getting cancer (r,)t and the total number of animals examined (n;) for various different doses (usually 3 or 5) labeled { [fm 1, 2, .N). For the particular turnon the maximum likelihood estimates of the parameters a and in Eq. (2) (a>0, 0>O) are the values that maximize 1 - (I -- a)4?-*W,-"y/ (4) These values were, computed, and the resulting values of 0 arc shown in Tables 1-4 for the eases and tumor types or sites that were considered statistically significant in the original reports, together with some others. The rationale adopted for selection of the nonsignificant results was to include all those in which the potency at some site or for some tumor type was greater than twice the sensitivity of the experiment, defining the sensitivity as the 95?S confidence upper bound on the potency if no cancers had been observed at all. No account was taken of possible antitumor effects of the chemical under test, so any tumors or tumor sites showing such effects were ignored. The potency values shown arc the largest that can be obtained from the data in the NCI reports. APPENDIX B The procedures used to calculate the potencies in Table 4 are simpler and less accurate than those described in Appendix A, the accuracy of comparison is low, and improved precision would be irrelevant. Wherever needed, the values shown in Table 6, which were taken from an International Agency for Research on Cancer (IARC) monograph, were used in the derivations. The following summaries outline the derivation of Table 4. *.*- 1110 E. CROUCH AND R. WILSON INI TABLE 6. Vilun Used in Calculating Potencies fn Appendix B 3 lx end fou; Weight Lifetime Food Water consumption . consumption Air breathed .1 wai .1. Species M. (ye) m (ml/dj 0/d| An; Mouse 0.025 1.75 s 5 . 40 '1 of Rat - 0.25 Do*. 10 3 10 \5 35 200 250 500 IS,000. and Human 70 70 1500 2500 15,000 -1 inh: Acrylonitrile bee. dust A continuing Dow Chemical Co. lifetime feeding study in rats {Food and <5 Drug Administration Hearing, 1977) shows at 13 mo a risk of 4 X 10~4 per ppm acrylonitrile (AN) in the water supply (Wilson, 1977a). Extrapolation to a full lifetime gives a risk of 6 X 10~3 per ppm AN in the water supply, although subsequent (still incomplete) analysis ot this experiment suggests that this may be an overestimate. This is equivalent to 0.1 mg/kg*d, giving a potency of 0.06 kg'd/mg. The data of Maltoni cl al. (1974) arc for lower doses and are not statistically significant, although they correspond to a potency of 0.05 kg*d/mg for ingestion and -- 0.2 kg'd/mg for inhalation. The data for humans come from a study of 470 DuPont employees who worked in AN concentrations averaging 20 ppm over the working day. Although only a few of these workers have developed cancer, the incidence is twice normal. This is not very significant and certainly not proved to be caused by AN; the preponderance of lung cancer suggests that cigarette smoking is the cause. The upper limit to the lifetime cancer risk is taken by assuming that they are aH caused by AN; the lifetime expexp and moi our cent mea 5-K actu Frat. rept used The cancer incidence is 0.15, which was used lo obtain the potency given in Table 4, using the human inhalation figures of Table 6. Aflatoxin 8t This Is a considerable literature on aflatoxin, and the animal data arc not entirely consistent. For this comparison the average outlined by Mesefson and Russell (1977) is taken. For the human data the risk was evaluated on a linear basis by Piers and Linscll (1973) and used by the FDA in their analysis of aflatoxin In milk, peanut butter, and corn (Bureau of Foods, 1978). Arsenic The carcinogenicity of As has been reviewed elsewhere (International Agency for Research on Cancer, 1973; National Institute of Occupational Safety and Health, 1975; American Industrial Health Council, 1978). In poti men et a arse i ppm in fi Jifet. the expc limit This Ther no c in ti I APOOO18465 "WILSON d and ** per lation ippjy, ggcsts vinga lower to a n, oyees rking , the y not jgests ancer ctimc cn in a are d by : was / the corn donai ional 0* In ''If INTEKSPECtCS COMPARISON OF CARCINOGENIC POTENCY nit endeavoring to derive a potency from human epidemiology, we note here four fatrty clear-cut cases. The prevalence of skin cancers attributed to (high) levels of As in the water supply in an area of Taiwan is described in Tseng et al. (1968). Analysis of their data leads to a potency for skin cancer for As ingestion ' of 4-8.kg*d/mg. No search was made for other cancers. Investigations of cancer incidence in a British sheep dip factory (Hill v.arvd Faning, 1948; Perry et al., 1948) allow an estimate of potency of ^ 110 kg`d/mg for all cancers and -- 35 kg'd/mg for lung cancer only for inhalation of arsenical dusts. These may be up to a factor of 5 too high, because we have assumed that only 20% of the dust could pass the crude dust marks worn, although >20% of the dust was in the si/e range <5 pm. Pinto et al. (1977) related mortality in a group of 527 men to an exposure index that could be directly correlated with thd current As exposure (by inhalation), using a measured correlation between urinary As and airborne As concentrations. The observed linear trend of standardized mortality ratio for respiratory cancer with exposure index translates, in our terms, to a potency of 2Q kg*d/mg for lung cancer if air con centrations in the past were the same as current concentrations. Scattered measurements suggest that past concentrations may have been a factor of 5-10 higher, so 20 kg*d/mg may be regarded as an upper bound, the actual value being 5-10 times lower. The figures reported by Lee and Fraumeni (1969) agree with this, if the data on airborne As concentrations reported in the Federal Register (1978) for the period of this study are used. There have been at least 40 attempts to produce cancer ?n animals. The inhalation studies were too short to provide useful information at potencies comparable to those estimated above. Three long-term experi ments on As Ingestion in animals provided evidence of low potency. Byron et al. (1967) reported no carcinogenic effect of sodium arsenite or sodium arsenate fed for 2 yr to Osborne-Mendet rats at levels up to 250 and 400 ppm, respectively, the latter corresponding to ~ 7.6 and *^6.1 mg/kg*d As in females and males, respectively. Krocs et al. (1974) reported a similar lifetime experiment on Wistar rats, feeding 416 ppm sodium arsenate in tne diet; again, no carcinogenic effect was noted. If we assume that these experiments could detect any effect in 10?$ of the rats, we get an upper limit on the potency of AS of '-'lO'"2 kg*d/mg (for ingestion by rats). This dearly disagrees with the potencies estimated above for humans. There is a similar disagreement with Schrocder et al. (1958), who found ( no carcinogenic effect in rats fed 0.46 ppm As in their diet and 5 ppm As in their water. In mice, carcinogenic activity of As has been observable, and at least AP00018466 1 r '- U !- \4 ti ..' ,t :i 1 i :f 1i ..;* r-- 1112 _ E. CROUCH AND R. WILSON one study (Osswald and Gocrttlcr, 1971) allows estimates of the potency for sc injection of 1.5-30 mg/kg*d. However, the control animals did not receive any injections. We summarize the various results as follows: Location and reference Species .. > Application Tumor Potency (mg/kg-d) Error Taiwan . (Tseng trii., 1968) Human U.K. sheep dip factory (Hit! and Faring, 194ft) Human U.S. copper smelter (Pinto etal.,1977) Uj. copper smeller (Holland et al., 19S9) (Bryon et al., 1967: Kroes et al.,1974) (Osswald and Goemier, 1971) Human Human Rat Mouse imbibition inhalation Inhalation Inhalation Ingestion Sc injection Skin cancer /Total cancer (Lung cancer Lung cancer Lung cancer Any cancer Leukemia 4-8 < no) (-X3) Up to five times lower 2-20 IS <0.01 1,5-30 ("Xil I-X3) Benzene There is a great deal of human data on benzene, based mainly on European exposures (in Turkey and Italy) to very high levels in the workplace. Leukemia is produced at high doses but has only been observed together with a toxic cffcct--aplasia of the bone marrow. The leukemia could be a secondary consequence of the toxic effect, in which case there would be a threshold for the carcinogenic effect just as is believed to exist for the toxic effect. There arc some early data on animals exposed to benzene; until recently, the only lifetime inhalation studies were at exposure levels (10 ppm) where no carcinogenic effect would be expected with the sensitivity available. However, a preliminary report has been made available of a high-dose study on rats and mice (New York University Medical Center, 1977). Leukemias were observed, but only in conjunction with aplasia. Whether or not such a comparison means anything for the leukemia hazard of low-levci exposures to benzene, it is interesting to make the comparison at the high level. The animal data yield a leukemia incidence of ~ 2% for the exposure range 200-400 ppm; this must be corrected upward because the experi ment is ongoing and the rats and mice were not all dead at the time of the preliminary report. This might give a 16% lifetime risk for 200 ppm in the air or a risk of 6 X 10"4 for 1 ppm. This corresponds to a risk of 8 X 10"4 for 1 mg/kg daily or a potency of 0.0008 kg*d/mg. For humans, a total risk or 3.5 X 10~* yearly or a lifetime total risk of 2.5 X 10'4 for 1 ppm benzene in the air was estimated (Wilson, 1977b) i ; : f t 1t t I : 'L b n v, f< li f Ci ai e.' N .< d< V as Ti A APOOO18467 SON ncy not J w /r I J on the ten fhe ich . is mil (10 ity ra ler, sia. nia the ure eri- of i in of isk 7b) INTERSPECIES COMPARISON OF CARCINOGENIC POTENCY" 1113 by 3 proportional extrapolation to low levels of the Turkish and Italian experience. A level of 1 ppm in the air corresponds to a daily intake of 0.25 mg/kg, giving a lifetime risk of 10"3 per 1 mg/kg daily intake or a potency of 10 kg*d/mg. Benzidine - Data for sc injection in' the rat and mouse were summarized by Meselson and Russell (1977). Data for humans are exemplified by those of Zavon et ai. (1973), which show that 13 of 25 workers occupationally exposed to benzidine got bladder cancer; the workers were often exposed to dust levels of >5 mg/m3 In the air, and an average of 0.1 mg/m* seems likely over a lifetime. This gives a potency of 34 kgd/mg. In table 5 of the NAS (1975) report the lifetime accumulated dose is estimated to be 10 times lower, but they note that they may be underestimating the exposure. Ingestion data are available for animals. Boyland ct al. (1954) found tumors in female Wistar rats fed 0.017% benzidine (10 mg/kg*d) in the diet; there was no control experiment. This gives an upper bound on the potency of '-0.02. If we make a correction for full lifetime exposure (a factor of *w5-25), the bound becomes 0.1-0.5 kg*d/mg. On the other hand, Griswold ct al. (1968) administered benzidine by gavage to female Sprague-Dawiey rats. With total dose of only 12 mg, 50% of the rats developed mammary tumors; this leads to a potency of 7 kg*d/mg, or 130-2500 kg*d/mg if a correction is made for observation for 24 mo instead of the 9 mo they were actually observed before sacrifice. 3,3 '-Diehl orobenzidi ne Dichlorobcn2idioc (DCB) was shown to be carcinogenic when ingested by hamster (Sellakmmar ct al., 1969), rat {Pliss, 1959; Stula ct al., 1971), mouse (Pliss, 1959), and. dog (Stula, 1974). Although the rats and mice were fed DCB for only 12 and 11 mo, respectively, they were observed for a full lifetime. Wc therefore obtain the potency simply by using a lifetime average dose and the results of these experiments. The chemical similarities of DCB and benzidine and the many bladder cancers attributed to the latter prompted extensive epidemiologic studies at DCB plants (Gerardc and Gerardo, 1974; MacIntyre, 1975), but no excess bladder or other cancer was found among 175 workers in Haleden, N.J., and 225 workers in Britain. Wc therefore assume a lifetime risk of <1% in these cases. Although no extensive personnel monitoring was done, DCB levels of 70 pg/ma were observed at a DCB plant In Buffalo in 1939, corresponding to an average intake of --5 Mg/kg*d, or --2 f*g/kgd averaged over a lifetime if the whole working life was spent at the plant. This gives a potency, with very wide error margins, of <5 kg*d/mg. i i i i> i i ,i A r i ri * f*t-,4 APOOO18468 1114 E. CROUCH AND R. WILSON t- \ i i Saccharin There have been numerous animal tests with saccharin, many of which are rcvicwedJn a report by the Office of Technology Assessment (1977). From this we estimate potencies of X IQ"4 and S 5 X 10"s kg*d/mg for male and female rats, respectively. For humans, we use the results of Howe et al. (1977), even though the interpretation may be somewhat ambiguous. They reported a risk ratio for bladder cancer in males of 1.6, independent of length of time of consumption, for <2500 tablets per year. Assume that an average of 1500 tablets per year 200 mg/d *3 mg/kg*d gives a risk ratio of 1.6. (This implies a risk that increases more rapidly than linearly with dose.) The risk of bladder cancer averaged over the U.S. male population is *s' 1.9 X 10"4 yr"1, so the excess risk caused by saccharin consumption will be 1.1 X 10"4 yr"1 or ^8X10"* in a lifetime, giving a potency of *^3X10"* kg*d/mg. This estimate is subject to considerable uncertainty. Indeed, Miller and Howe (1978) present a fairly extreme model that corresponds to a potency of ''0.04 kg'd/mg, although that model has the lifetime risk increasing much more rapidly than linearly with lifetime saccharin consumption. We take this as an upper bound (50 mg per tablet is assumed throughout). Smoking (Cigarettes) We use directly the figures quoted In the NAS (1975) report. y { t 1 i 1 1 1 t c a a c: y n fo lci. re- sp re; rat cat kg- Vinyl Chloride We have the risk analysis of Ku/mack and MeGaughy (197S) based on the work of Mutton! et al. (1974). The animal studies show a total lifetime I carcinogenic risk of 4.9 X 10"3</ {where d is the lifetime average dose in air (ppm)) for liver angiosarcomas and 10"Ji/ for ail tumors. (Ku/mack and MeGaughy actually' calculated for 4 h/d, 5 d/wk, for half a lifetime.) A rat breathes 0.14 l/min ** 200 1/d ** 220 g/d, so 1 ppm in air gives 1 mg/kg vinyl chloride daily. The best indication of human incidence is given in appendix D4 of Kuzmack and MeGaughy. This gives an average lifetime risk of 0.13 for 20 ppm or 6 X 10"3<7, whore d is the average dose (ppm). For a human who breathes 15 m* of air per day or 18 kg/d, 1 ppm in air is 0.25 mg/kg'd. This procedure gives a higher human risk than the procedure used by the NAS (1975) committee, who (quoting the IAKC) take a lower average angiosarcoma incidence in a larger population. This may indicate a nonlinear dose-response relationship, but could .be an overestimate of exposure for the law-exposure members of the group. a!., ad it ex pi a vc of c kfronsimii rats . of th E expor 1978 Radiation potor The doses here are expressed In terms of rads or rems. These arc doses per unit volume, so it is unnecessary to divide by body weight. Numbers area mafigi S.6.T AP00018469 INTERSPECIES COMPARISON OF CARCINOGENIC POTENCY 111J for humans come from the BEIR report (Committee on the Biological Effects, of Ionizing Radiation, 1972), which states that at an exposure of 0.1 rem/yr, 3000-4000 extra cancer deaths probably result out of a total of 2,000,000 deaths. The potency Is thus -- 0.020 yr/rem. The data for mice are taken from a United Nations (1977) report that 15% leukemia incidence arises from a 100-rad total dose'in mice. Allowing six times as many other cancers as leukemias (the factor for humans), the potency becomes 0.01 yr/rem. %, Chloroform For mouse and rat we use the NCI carcinogenesis bioassay results (Table 1). Human epidcmiolovy is limited and shows no effect. Daily ingestion for 10 yr of cough suppressant containing chloroform-codeine at dose levels estimated at 23-27 mg/kg*d for a 70-kg person showed only reversible hcpatoloxicity (Tardiff, 1977). The average dose over a 50-yr span was then 5 mg/kg`d. The 10,000 yearly deaths caused by liver cancer represent 0.005 of all deaths, and in a small population a doubling of this rate could probably have been seen, so the excess probability of death caused by liver'canccr in this group is 0.005, giving a potency 0.001 kg*d/mg. Chlornaphazine Figures arc abstracted directly from the NAS (1975) report. Ethylene Dibromide Ethylene dibromidc (EDB) has been shown (Olsen ct al., 1973; Powers ct ah, 1975) to cause an increased incidence-of gastric tumors in rats when administered by gavage. So many animals died, even at the low dose, that the experiment terminated at 61 wk, and attempting to directly fit Eq. (1) gives a very inaccurate determination of the potency. Using Eq. (2) and the times of death, the Carcinogenesis Assessment Group determined a value of k -- 5.9 and a potency of 6 kg*d/mg (our units and notation). Mouse data from these experiments averaged over females and males (see Table 1) similarly give 6 kg*d/mg. If a new experiment were performed in which the rats and male mice were dosed at lower levels, a more precise determination of the potency would be possible. Epidemiologic data from the Dow Chemical Co. on 161 employees exposed to EDB (Ott ct a!., 1977) were analyzed by Dow (Ramsey et al., 1978) and the EPA (Albert, 1977). On the basis of the EPAS estimate of potency, where the potency for rats (6 mg/kg*d) was multiplied by a surface area correction factor of 5, Ramsey et al. calculated an expected 85 malignant neoplasms where 8 were observed, and the expected background is 5.6. Thus we get 30 X 2.4/85 0.8 kg*d/mg for the potency. 1116 C. CROUCH ANO ft. WILSON Lead Acetate Basic lead acetate was fed to rats, mice, and hamsters in controlled lifetime experiments {Boyland et a!., 1962; van Esch et ah, 1962; van Esch and Krocs, 1969). Renal tumors were found in rats and hamsters, but were below the level of significance in mice. The results of these experiments were used for the estimates in Table 4. There have been extensive epidemiologic studies of mortality of Pb workers. Although in some eases excess canters were observed -(Cooper and Gaffey, 1974; Cooper, 1976}, they were not related to Pb intake and cannot therefore be attributed to Pb. However, we use this increase to obtain an upper limit on the carcinogenic potency of Pb (Table 4). REFERENCES Albert, R. E. 1977. The Carcinogenesis Assessments Group's preliminary report on ethylene dibromide, SPA RepL American Industrial Heaith Council. 1978. Reply to "Estimates of the fraction of cancer In the United States attributable to occupational factors. ArmJtage, P, and Doll, R. 1954. The age distribution of cancer and a multistage theory of carcinogenesis. Br. / Cancer 8:1. Bingham, E. and Falk, H. L 1969. Emironmental carcinogens-the modifying effect of co- carcinogens on the threshold response. Arch. Lmirott. Ihtilth 19:779. Boyland, E., Han is, j., and Horning, C. S. 1954. The induction of carcinoma of the bladder in rats with aceumidofiuorene. Br. ]. Cancer 8:647. Boyland, ,, Dulcet, C. Grover, P. L., and Mitchley, 6. C- V. 1962. The induction of renal tumors by feeding lead acetate to rats. Br. }, Cancer 16:283. Bureau of Foods, Food and Drug Administration. 1978. Assessment of estimated risks resulting from afiatoxins in consumer peanut products and other food commodities. Byron, W. R., BlerbOwer, G. W., Brouwer, ]. 8., and Hansen, W. H. 1967. Pathologic changes in rats and dogs from two year feeding of sodium arseniu or sodium arsenate. Toxicol. Appl. Pharmacol. 10:132. Committee on the dialogical Effect* or toniaing Radiation. 1972, The tfftcli Of} Populations of exposure to Low Levels of /enf/wy Radiation, p. 168. Washington, D.Ct National Academy of Sciences. Cooper, W. C. 1976, Cancer mortality patterns in the lead Industry. Atm. <V,Y, Scl. 271:250. Cooper, W. C. and Gaffcy, W. R. 1975. Mortality of lead workers. /. Occvp. Med. 17:100. Federal Register. 1978. Final standard for occupational exposure to Inorganic aisenlc. 43:19589 and 19591. Food and Drug Administration Hearing. 1977. Acrylonitrile copolymers used to fabricate beverage containers. Docket 76N 0070, exhibit GS7. Freiretch, . J., Gehan, E. A., Rail, D. P,, Schmidt, L. M,, and SMppcr. H. E. 1966. Quantitative comparison of toxicity of anticancer agents in mouse, rat, Hamster, dog, monkey and man. Cancer Chcmother. Rep. 50:219. Gehan, E. 1969. Estimating miittie functions from the life table, /. Chronic Dit. 21:629. Gerarde, H. W. and Gerarde, O. F. 1974. Industrial experience with 3,3'-dlchlorobenidinct an epidemiological study of a chemical manufacturing plant. /. Octup. Mtd. 16:323. Griswold, D. P,, Casey, A. E., WcFsburjcr, E, K., and Weisburger, i. H. 1968. The carcinogenicity of multiple inttagastrie doses of aromatic and heterocyclic nitto or amino derivatives in young female Sprague-Dawlcy rats. Cancer Ret. 28:924, i is t INTE Guess. a Heath, C Milt, A HoUan * Howe, Ft 1 ii: - Interna fo Kroes, cat dit Kutma. to lee, A. 42 Maclnt) Ot. Maltonl, rest Meselsor Hu: and Miller, A National Cor D.C National Stan 75-1 New Yor Office ot Olsen, , Indu Cant. Osswald, post* Ott, M. v empt Ferry, K,, Inorg . Peto, R. 1 Br.}. Piers, F. a* Pinto, S. relate Hiss, G. B (USSI Powers, M Circle, admin ^4^ -vuSiivm, fc.Au- V/ AP00018471 WILSON trolled n Esch 5, but these of Pb ooper. -o and sc to i>lcne n the y of ' co rns enal om its pi. of ny 0. iif V c t I INTERSPECIES COMPARISON OF CARCINOGENIC POTENCY 1117 Gums, H., Dump, K., arid Pete, R. 1977, Uncertainty estimate* for low-dost-rate extrapolations of animal carcinogenicity data. Cancer Res. 37:3475. Heath, C, W. and Falls, H, 1975. Characteristics of cases of angiosarcoma of the liver among vinyl chloride workers in the U.5, Ann. N, Y, Acad. Set. 246:231. Hill, A. 8. and Fining, E, L. 1948. Mortality experience In the factory. Be. J. ind. Med. 5:2. Holland, R. H., McCall, M. 5., and Lena, H. C. 1959. A study of inhaled arsenic-74 In man. Cancer Res. 19:1154. Howe, G. R,, Burch, J. Miller, A. B., Morrison, B,, Cordon, P., Weldon, L., Chambers, L. W., Fodor, G., and Winsor, G. M. 1977. Artificial sweeteners and human bladder cancer. Lancet 11:578. 'International Agency for Research on Cancer. 1973. Atonog. 2, p. 58. Lyon: International Agency for Research on Cancer. Kroes, R., van Logtcn, M. J., Berkvens, ). M,, de Vries, T., and van Esch, G. J. 1974. Study on the carcinogenicity of lead arsenate and sodium arsenate and on the possible synergistic effect of dlethylnitrosamine. food Cosmei. Toxical. 12:67). Kuzmich, A. M. and McGaughy, R. E. 1975. Quantitative risk assessment for community exposure to vinyl chloride. EPA Rept. Lee, A. M. and Fraumeni, F. 1969. Arsenic and respiratory cancer in man, (. Nad. Cancer Inn. 42:1045. MacIntyre, I. 1975. Experience with tumors in a British plant handling 3,3'-divhIorobenzidlne. / Occup, Med. 17:23. Maltoni, C.. Lefemlne, G., Chicco, P., and CarretU, 0. 1974. Vinyl chloride carcinogenesis: Current results and perspectives. AUd. Lev. 65:421. Meseison, M. and Russell, K, 1977. Comparisons of carcinogenic and mutagenic potency, (n Origins of Human Cancer, Proceeding of the Cold Spring Harbor Conference, eds. H. Hiatt, |. O. Watson, and }. A, Winsien.'Cold Soring Harbor, N.Y.: Cold Spring Harbor Laboratory. Miller, A. B. and Howe, G. R. 1978. Saccharin-the risks and benefits. Nature (l.ond.) 273:8. National Academy of Sciences. )975. Health Effects of Oiemkul Pesticides, Report of the Consultative Panel on Health Hazards or Chemical Pesticides, vol. j, pp. 67-87. Washington, D.C.: National Academy of Sciences. National institute for Occupational Safety and Health. 1975. Criteria tor a recommended Standard--occupational exposure to Inorganic arsenic (new criteria 1975), HEW Pubt. {N(OSH) 75-149. Washington, D.C.: Environmental Protection Agency. New York University Medical Center. 1977, Letter to E. Gingham, 05HA. Office of Technology Assessment. 1977. Cancer testing technology and saccharin. Rcpr. OTA-H-5S. Otsefl, W. A.. Habetmann, R. }., Weisiberger, E. K., Ward, j. M., and Weissbsrger, H. 1973. Induction of stomach cancer In rats and mice by halogcnated aliphatic fumigants. /. Natl. Cancer hut. 51:1993. Osswald, H. and Goenrier, K. 1971. Arsenic-induced leukemia in mice after diaplacental and post-natal application. Dtsch. Get, PathPl. 55:289 (In German). Qtt. M. G., Schatnweber, ti. C., and Laugtwr, R. R. 1977. The mortality experience of 161 employees exposed to ethylene dibroinitle in two production units. Preprint. Perry, K., Bowler, R. G., Buckell, H. M., Pruett, H. A., and Schilling, R. S. F, 1948. Cancer from * Inorganic arsenic compounds. II. Clinical and environmental investigations. Sr. /. Ind. MtJ. 5:6. Peto, R. 1974. Guidelines on the analysis of tumor rates and death rates In experimental animals. Br. i. Cancer 29:101. Piers, F. and Unsell, C. 1973. Dietary aflatcxlns and liver cancer. Br. J. Cancer 27:473. Pinto, S. S., EnterUne, P. Henderson, V., and Varner, M. O. 1977. Mortality experience in relation to measured arsenic exposure. Environ. Health Pcnptct. 19:127. Biss* C- 8. 1959. Dlchlorobenzidine as a blastomogenic agent. Vopr. Onkof. 5{S}:524; Proh. Oncol. (VSSR) 5fS):11. Powers, M, 0., Voltker, R. W,, Page, N. P., WWssberger, E. K., and Krayblll, H. F. 1975. Carcinogenicity of ethylene dibromide (EDB) and 1,2-dlfcfDmo-3-dlcbloropropane after oral administration In rats and mice. Toxicol. Appl. Pharmacol, 33:171. .? n]i nit E. CROUCH AND K. WILSON Ramsey, J. C., Park, C, N., ou. M. G.. and Gobting, P. 1 197$. Carcinogenic risk assessmentethylene dibromidt. Doh Oumkol Ct. Kept. Scbroeder, H. A., Kanluwj, M., Frost, 0. v,, and Miichencr, M. 1968. Germanium, tin and arsenic in rati; effect* on growth, survival, pathological lesion*, and life span. /. Nuir. 96:37. SeHaVtimar, A. R., Montestno, R,, and Saffioti, U. 1969. Aromatic amines, carcinogenicity in hamsters, hoc. Am. 4*$oc. Cumer Rv% 10:7$. SluU, E. F. 1974. Long term oral study in dogs with 3,3*-dlchforobenHdIne. Medical research project 93#, ftoikeit Labanitury (DuPont) Rcpt. #23.74. Stula, E. F., Sherman, H,, and ?app, }. A. Experimental neoplasia in Ch R-CD rats with the oral administration of 3,3'-dici<lortjbnzk!if!e. Toxicol. AppL Pharmacol. 19:366. Tardiff, R. G. 1977. Hcaftn effccu of organics: Rlsfc and hazard assessment of ingested chloroform. Water Tccfttnsl. Qnct. f. riUTl'A: 658. Tsengj W. P., Chu, H. M, How, S. W., Fong, }. M., Lin, C. S., and Shu Veh. 1968. Prevalence of skin cancer in an endemic area cl chronic arscr.Sciim In Taiwan. /. Soil. Comer Inst. 40:453. United Nations. 1977. Sources and eftccix of ionizing radiation. Report of the U.N, Subcommittee on the Effects of Atomic Radiation (UNSCCAR), p. 5S3. van tsch, G. I. and Krocs, R. 1969. The induction of renal tumors by feeding basic lead acetate to mice and hamsters. Sr. ]. Cancer 23:765. van Esch, G. |., van Cendcren, H., and Vink, K. H. 1962. The Induction of renal tumor* by feeding of basic lead acetate to rats. Ur. f. Cancer 16:289, Wilson, R. 1977a. Testimony presented at the FDA hearings on acrylonitrile as a food additive; exhibit M90, p. 34, Wilson, R. 1977b. Testimony prevented at the OSHA hearing on benzene, Zavon, M. R., Hocgg, U., and Cingham, E, 1973. Benzidine exposure as a cause of bladder tumours. Arch. Environ. Health 27{l):l. * Rtceivcd r< bruury 14, 7979 AcavtritJttlv 21, 1979 i: V <* * . * * e . - r. i ' * 1. ' ) i '* -XT*t.w'*>V*v7V;C^<ViW*',.*W**lr%f-T'.-1 oum ior ;W-'T''T'VV**' tv*.-*. ',e~"VF7T*.'.s.y..g...r.v..e~wf-r..'- x-*^.-,. ... . . .r,-l * .-.V ^rfi iaa*4L 4 .** 6* ihiiif' newiiiiiiJ f'-ei'nrfciirj ill lit <>ivki<?ji 11 nfii AP00018473