Document e18dYRBMEyMwZJw9dOwRRpvwp

:llular response .Idez, John J. Miglio, ;ESE HAMSTER V79 COUS CLEARANCE R. K. Wolff, R. F. HANCES INDUCED .Y A 9-WEEK RECOV- APPLICATION OF THE "FILTER MODEL" TO A RISK ASSESSMENT FOR VINYL CHLORIDE Clark S. Olson Illinois Environmental Protection Agency, Springfield, Illinois David J. Schaeffer U.S, Army Construction Engineering Research Laboratory, Champaign, Illinois The filter model was used to estimate thresholds for the induction of cancer from many dose-response sets for inhalation and ingestion exposure to vinyl chloride for rat and inhalation exposure for mouse. Estimates fora variety of end-point combina tions were log-normally distributed over about 2 decades from about 7 to 700 ppm for inhalation exposure to rat and 0,1 to 30 ppm for mouse. When the data is transformed to "dose" (milligrams per kilogram body weight per dayl, the estimates for inhalation and Ingestion exposure and also for rat and mouse are similar. Estimates for different experiments carried out for different durations of time (single exposure to 7 yr) are comparable. Since the threshold is an intrinsic property of the biological system, the estimate, even from a protocol for short exposure and less than lifetime observation, can be used directly in a risk assessment as the maximum safe dose. INTRODUCTION A "threshold" is defined in Webster's Seventh New Collegiate Dic tionary (1961) as the "point at which a physiological or psychological effect begins to be produced." Druckrey (1959), for example, found that hepatomas were induced by 1012 molecules, but not 4 x 1011 mol ecules, of dimethylaminoazobenzene. Most models of carcinogenesis use a risk-level approach involving one-to-one Interactions ("hits") be tween the insult and target molecules. These theories assume that cells ..receiving even a single direct hit are affected, and do not consider the possibility of a threshold. However, up to 80% of cells do survive direct hits (by radiation) (Thompson, 1985). There appear to be two popula tions of ceils, however: the sensitive, for which one hit is enough to cause a biological effect, and the nonsensitive. Victor P. Bond (Brookhaven), as quoted by Thompson (1985), notes that "A single hit by a charged particle can cause a single cell effect." The result is that "it dies or does not die." If it lives, "it is mutated or not." We thank the reviewers for stimulating and thoughtful suggestions. Requests for reprints should be sent to David |. Schaetfer. CERL IEN), P.O. Box 4t)(J5. Cham paign, Illinois 61B20. 1 Journal ol Toxicology and Environmental Health, 17:1-22, 1986 Copyright s 1986 by Hemisphere Publishing Corporation AP00020018 7 C. S. OLSON AND D. J. SCHAEFFER If it is mutated, "it is carcinogenic or not." There is no middle ground for any of these choices. "The response is a statistical phenomenon." Because both the "hit" and the response are statistical events, even the one-hit mode) of chemical carcinogenesis allows for a threshold. At another level, the existence of real-life thresholds is defined by two factors. One is the stochastic process of molecular interactions in a complex chemical milieu. The other aspect of real-life thresholds is the "no observable effect level" (NOEL) determined by the background noise of any detection system. (This detection requirement is similar to Von Borstal's (1982) concept of threshold.) These factors govern whether or not an effect can be observed in a population exposed to a carcinogenic insult, and not with the probability of a single molecule causing cancer in an individual. One reason why most models used for extrapolation of carcinogen assays to low exposures (Hoel et al., 1983; Fishbein, 1980; Thompson and Funderlich, 1981) do not consider a threshold is the irreversibility of the initial biochemical reactions in carcinogenesis. This ignores the fact that reactions that are irreversible at one level may, through ho meostatic mechanisms, be reversible at further steps in carcinogenesis and at higher levels of biological organization, such as through detox ication of activated products, repair of DNA, or control by the immune system. Another reason why a threshold is not incorporated into extrapo lation models is that it is simpler to ignore the large background chal lenge from thousands of xenobiotics and radiation in the environment through the diet and other modes of exposure (Ames, 1983). A re sponse threshold level must be considered against the uncertainty of the magnitude of both the background challenge and the natural re sponse to it. Thus, relatively few enzyme types of the mixed-function oxidase system (MFO) and other detoxication systems process both natural and human-made carcinogens without distinction (jakoby, 1980; Jenner et al., 1981). When exposure to a substance of interest is low, these systems must handle the individual substance of interest along with the natural carcinogens already challenging the organism. Risk from exposures to industrial carcinogens must therefore include the magnitude of the added exposure to the background exposure and the added response to the background response. A realistic extrapolation model would incorporate the homeostatic concept of challenge and response and use a threshold based on de parture of the exposed population's response from the population's background response. The idea of approaching exposure-response re lations from a homeostatic point of view is similar to that used in the formalism of the steady-state rate of mutation iStover and Eyring, 1970a,b; Eyring and Stover, 1970; Eyring et al., 1971). These authors considered populations heterogeneous for genotypes and response AP00020019 X 1 i i | extripound < iialtfivironment \%l\. \ re`ertaintv of natural re ed-function yprocess both on [)akoby, >of interest is of interest -`'e organism. Jfore include "tposure and homeostatic 'ised on de population's response rei used in the ' and Eyring, hose authors !nd response APPLICATION OF THE FILTER MODEL TO VC 3 phenotypes. "Since biological processes are close to equilibrium most of the time," (Eyring and Stover, 1970), a steady-state theory of muta tions was derived on the basis that a biological response (death, mu tation, cancer) "ensues when the rate of insult has exceeded the rate of recovery to a sufficient extent that reserves are depleted and the steady-state can no longer be maintained" (Eyring and Stover 1970). Although the inbred colonies of rats and mice used in carcinogenesis studies have more homogeneous genotypes than do the wild popula tions considered by Stover and Eyring, these colonies are not isogenic. Further, they appear to evolve rapidly (Fitch and Atchley, 1985; Lewin, 1985). One threshold model incorporating homeostatic mechanisms is the filter model (Schaeffer et a)., 1980). The thermodynamic, first-order kinetic, and statistical theory of this model has been given (Janardan and Schaeffer, 1977; Schaeffer et al., 1983). The mathematical formula for the filter model is iR-R0) = a (In D) + b (In D)2 (1) where R is the response at the Oth level of exposure, RQ is the response of the controls, and a and b are constants obtained by regression or maximum likelihood estimation. The threshold (0r) for the filter model is estimated either as the point exp (1-a/2b) where the exposure-re sponse curve changes slope or from the confidence limit on the extra polated value (Schaeffer et al., 1981). Threshold estimates from this model have been compared with "virtual safe dose" (VSD) estimates from risk-level models (Schaeffer et al., 1982; Schaeffer, 1983). We apply the filter model to over 100 data sets reported for vinyl chloride (VC) by several authors for various organisms and end-points. VC is an especially appropriate substance for evaluating extrapolation models because of its importance as an industrial compound and as an environmental contaminant (U.S. Environmental Protection Agency, 1975,1980; Waxweiler et al., 1981). Furthermore, there is abundant data on the carcinogenicity, general toxicology, and pharmacology of VC in animals and data from epidemiology studies on humans. Other esti mates of risk from VC permit comparisons between many models. VC is representative of a number of organic compounds that are low in chemical reactivity and acute toxicity but that are activated to a poten tially carcinogenic form by the MFO system. - METHODS Sources of Data Data on frequencies of tumors resulting from VC exposure were obtained from published studies on several strains of rat and mouse involving exposure by inhalation and ingestion, short- and long-term AP00020020 4 C. S. OLSON AND D. |. SCHAEFFER experiments, and different sacrifice schedules. These studies encom passed many end-points, including incidence of animals with various specific tumors or all tumors, or number of specific tumors or ail tu mors per animal. Latency data was not used. Each set used showed a well-defined exposure-response relationship, contained at least three exposures and a control, and usually was the complete set given by the author. In addition, some sets of Maltoni were combined. The protocols used in different experiments are summarized in Table 1. Units Used in Exposures Inhalation exposure is generally expressed as the concentration (ppm) in air. For VC, 1 ppm is equal to 2.56 mg/m3 or 0.00256 mg/! (U.S. Environmental Protection Agency, 1975). The inhalation exposure con centration was transformed into an estimated dose (milligrams per ki logram body weight per day) by multiplying the concentration in the air by daily inhalation (liters) and by the fraction of a day used for exposure. For the rat, the conversion factor is obtained as (200 1/24 h x 0.00256 mg/I ppm)/0.25 kg = 0.086 mg/kg body weight * ppm h, in good agreement with the value 0.095 mg/kg ppm h obtained by an integrative method (Withey and Collins, 1976). The factor for the mouse is (40 1/24 h x 0.00256 mg/I ppm)/0.025 kg = 0.171 mg/kg body weight ppm h. Statistical Methods After subtracting off control responses, data were fitted to the filter model using ordinary least squares constrained to pass through the origin. Some data sets, those with very shallow slopes or showing poor exposure-response behavior, were fitted to a model that constrained as* 0 and b 0. The Rosen (1960) algorithm for a nonlinear gradient search, as implemented by Kuester and Mize (1973), was used. The al gorithm proceeds by the user selecting a feasible starting point for the coefficients a and b and for the step size. The derivative of the objec tive function fEq. (1)] with respect to the independent variable (In D) is evaluated at this base point, and the normalized direction vector com ponents are determined. If the derivative is less than or equal to the tolerance limit, the procedure is stopped because the required point has been found. Otherwise, a new point is located according to rules specified in the algorithm and the objective function is evaluated. RESULTS Rat Threshold estimates for chronic inhalation experiments are 1-100 ppm <0.1-10 mg/kg body weight d) (Table 2 and Fig. 1). Table 2 is irr.iru'* tn visccnding order of Ot. The table gives our identifying set AP00020021 m AND D. /. SCHAEFFER AP00020022 sjoujuj Xjeuouqry sjouin) AiPUDiuinj SJOUim AjpuOuijdj joiuei snojjBA sjcuuni Aicuoutind Jd^uep snoucA jODuei snopeA a OUJ a + + out zi + OUJ ZL + s a euicnjesoiSue inoi eiuoDJcsofSue |P|oi eujoiiesotSue jooj. euxojesotSue |eioi a a a a j3uea snoucA J33UC3snou*A jjkjucdsnoMfA jO)upj snoprA ;a3us3 snopeA J93UB3 snoilPA Jd3UPD SHOIJPA JSDUC5 snopCA OUI 11 + a a a a a a 0 tutod-puj ,1? uoijPAiasqo apuofip i^uia 411M apijqtp ^uiAXjod psj ,v uojirqnjujp 9Ui)l uoneAjasqo jeuomppB ' + 'ainsodxa Jaqc paitjtj^K '5 JqjFap 01 Q , |0J|U03 e papn|3Ui ||y q 'M :Aa|MB<]-an8eid5 'Q.5 f R3u|S) q 1 ouj t 'JfAVP S'fVM 9 )fM j, 'VVP S 'PA) S 1* a *5fAVP S 'P/M 9 93 `>t*VP S 'fVM 9 ** zs-% ,5l*vp s 'ivq 9 t|M Ot 't|AVp S 'pfl) t- ** m-SCL 'JJAVP/1 'FVq * >|M 6S-ZS >VXS >(6S-ZS>VxS SZS 5(M st- *>|*VP S'WS sjm ZS 'WP S 'Pfl\ V js ^M/p 5 *pA| tj|m rs ''I'w/p i* 'pai v ** zs ,5i*vp s 'ivq ZS '5|>u/p s `pAt fr |M ZS 'wp S `p/q fr >)M Z -5|AVp S `PAt t- ooo'os-w oooi-os 001- i oooi-os oooi-os 0001-05 ooo'oi-os (ujdd) w-zi os-coo l"O0 os-rt ip. 3i(/Suj) oooi-os ooo'oz-os DOO'Ol-OS ooo'ot-t OOO'K-l SZ-L DOZ-OOL ooo'oi-os (uidd) l 8 91 91 Z i 9 S Z S H V S LI S Zl fV/H3! 03 03 03 03 03 SSJMS M a-s o-s a-s o-s M a-s M a-s a-s a-s as ijq-'H Suoh i^nzn; Suoh 8uoh a1 t^ia iuoijcw (uoi)C|equj| asnow jUOJOJ ZZ'U-J.9 zz-ia uia plUOIJCVV fuoiisaSui) )r Suoh ojnde3 10 a'L-iA St.'6'9'z'l-J.0 aiisoduro^ St-IB Z18 t-19 juoirew (uoiipjeijui) icy to^otojd ajnsodkj ^aiiuBj asoQ Sias pjep )o Jdquirt^j .ujeiis aoudjapy uoiiPiuiisj p|oqs?jq^ joj paspi SJuawjjadxg p Ajeuiuins 1 Jiavi "3 Oi > 0> g S-o c = -r- Z S~ SxJ j; OX ^ *5b^ >* >o- -5rj % oVIi -Q C J* .E ,p .n --c cto <u ">D~ Z3 cOU) oVC -ua r- * ^X> VI Eo 25 ooi: <4/i>1 T3 "ooc---*-*-n-'*-*-- ?- * mO 00 ? c 5 j "a P'iSw'OC^o) w 3 -xoS SP s SS "K" S> cr; So*1~5 -Tcw2 -E>3 Mauo >oC c- a OO --m 4> 0) 00 c sn)-- *= c --a 1 AP00020023 i I\Ul 2. Threshold Estimates <0T) for Rat Inhalation and Ingestion Data Obtained with the Fitter Model VI lumber 624 6M 424 452 42J S71 453 454 018 440 600 6tU 549 020 574 626 62ft 6V 4 T9 610 591 553 609 588 414 629 586 54U 699 42fl 442 End-point* total #, F nod. hyp. LAS IAS other total # total LAS total # LA LC, F Kidney total mal., # total #, M&F total #, M&f, 6 mo total #, M&F NP LAS total LAS LAS total total #, M&F, 6 mo total mal.. # LC. M LAS, F total # other Number of d-r pairs 4 7 7 5 7 7 5 5 7 11 16 15 4 7 7 4 4 4 11 7 4 7 7 4 4 4 4 4 4 7 11 Input units ppm mg mg mg mg ppm mg rag ppm mg ppm ppm mg ppm ppm ppm ppm mg mg mg mg ppm mg mg mg rag mg mg ppm mfc mg 0t estimale* ppm mg/Wg - d 0.0038 -- (0.024) -- (1.3) 1.3 (2.2) 2.6 12.7) 3.1 3.6 -- 5.3W 5.8 6.8 fi.0rf (6.2) (9.41 -- -- 11 -- -- -- -- -- -- 12 (16) (16) (0.0014) o.oos 0.006 0.04d 0.32 (0.32) 0.53 Q.55J (0.64) n.w, (0.76J (0.8`J) 1.2 -- (1.4) (2.5) -- .1.0 2-3 2-3 2.5 (2J) 2.9 2.9 3.3 3.7J 3.9 3.9 (4.4) 4.0 3.9 Correlation R* 0.89 0.91 0.93 0.99 0.99 0.88 0.97 0.9*) 0.99 0.94 0.93 0.92 0.99 0.91 0.97 0.99 0.99 0.99 0.94 0.99 0.99 0.98 0.99 0.99 0.99 0.95 0.99 0.96 0.99 0.95 Reference' 4 8; 11,27 7;1 6:15 7;1 8;1 b;l5 6:15 7;1 6;(. 8;C 2 7;1 4 6;C 8; 11.27 8;27 7;1 8;11,27 6; 11 6; 11 4 ft; 11 2 4 7;1 6;C Related set number 625 424 -- -- -- -- -- 426 415 628 _ '' sin 555 609 588 414 rot Z 699 428 W2 ^ total laS lA5 total total maU ^ LC.M Lfi&' F total * Other 7 < t(>I ppm It 7 mg 4 mg 4 mg 4 mg 4 "6 4 4 7 VI mg ppm mg mg 12 (16) (1b) 2.9 2.9 3.3 3?' 3.9 3.9 (4.4) 4.0 3.9 0.98 0.99 0.99 0.99 0.95 0.99 0.96 0.99 0.95 0: H-- ft.U 6; 11 4 8:11 2 4 7;1 6;C 4S 626 553 1 i 416 total # 546 INP, M 412 LAS 607 LAS 533 Las, f 583 LAS 443 total 4 551 LAS, F 550 IAS, M 578 total mal., * 576 total 4 592 LAS 62S total 4, F 700 LAS,M4F 698 LA5, M 622 total #, M 581 total mal., # 417 tolal 427 other 413 total # 605 Zymbai 419 total # 022 other 580 2ymbal 021 Zymbal 606 total mal., # 438 total # 589 tolal mal., 4 623 total #, M 532 LAS 701 LAS, M&F (See footnotes on page 8) t4 4 7 8 4 7 11 4 4 7 4 7 V 4 4 4 7 7 7 7 15 7 7 7 7 8 4 4 4 4 4 mg mS mg ppm mg ppm mg mg mg ppm ppm ppm mg ppm ppm ppm ppm mg mg mg ppm ppm ppm ppm ppm ppm mg mg mg mg mg -- -- (17) 16 (18) 19 (19) -- -- 21 21 23 -- 23 24 25 27 (27) (32) (32) 33 33 34 35 36 38 (41) -- (41) (41) (41) 4.1 4.1 4.3 (4.4) 6.8 (4.7) 4.6 4.9 5.0 <5.21 (5.2) (5.7) R.2* (8.5) (8.8) 19.2) (6.6) 6.6 7.9 7.8 (B.0) (8.1) (8.3) (3.61 (9.0) (9.5) 10 12 15 15 15 0.98 0.99 0.80 0.90 0.96 0.90 0.99 0.99 0.99 0.93 0.97 0.99 0.89 0.98 0.97 0.99 0.92 0.96 0.92 0.97 0.81 0.96 0.93 0.98 0.84 0.87 0.99 0.58 0.99 0.98 0.96 6;t1 2 6;7 8;7 4 8;7 6;C 2 2 8:3 8;2 1 4 4 4 4 8;7,17 6; 3 7;1 6;7,17 8rC 6:3 7;1 8; 3 7;1 8:7,17 b;2 0;27 4 4 4 -- -- 533 -- 699 412 -- -- -- -- -- -- 624 701 S32 623 -- -- 022 -- -- -- 427 418 426 -- -- -- 622 698 700 AP00020024 TABLE 2. Threshold Estimates (Df) lor Rat Inhalation and Ingestion Data Obtained with the Filler Model (Continued) Set number End-point* Number of d-r pairs Input units 0j estimate* ppm mg/kg d Correlation te Reference1 Related set number 595 total # 593 lung ad. 4% LAS 437 other 594 skin 418 Zyrrbai 426 Zymbal 441 Zymbal 7 ppm S3 7 ppm 54 4 mg (57) 4 mg (57) 7 ppm 61 7 mg (73) 7 mg (77) 11 mg (93) (13) (13) 14 14 (15) 18 19 23 0.89 0.89 0.99 0.99 0.85 0.99 0.85 0.89 1-- 1-- 6;2 -- 6;2 -- 1-- 6;3 580 7;1 021 6;C -- ' IAS, liver angiosarcoma; NP. neoplasia; 1C, liver carcinoma; total trial., total malignant; lung ad., lung adenoma; #, number ol tumors per individual, all others are incidence; M, male; f, female; nod, hyp., nodular hyperplasia of liver. 6 Those marked with dash--from ingestion experiment. ` First number is reference as follows; (1) Caputo et al. (19741; (2) Feron el a(. (1981); (it Hehir et at. (1981); 141 Hang et at. (1981); (5) lee el al. (1973); (6) Maltoni (1979); (7) Maltoni and Lelemine (1975); (8) Maltoni 0981); (9) Suzuki (1983). Second number gives Maltoni experiment numbers)- C, composite, S-D rat, inhalation. * 0| very large; threshold estimate obtained by confidence interval method. O, very large for original data set; only first Ihree values of dose-response sel used. AP00020025 1 APPLICATION OF THE FILTER MODEL TO VC y 3 T= XT ll 3 5 a> ~** ~i : 5~ e> s- *3 3 :-zSEE-*f :'!5 >31 v =c a /J* W3 2s .5.. --* 3 ii li J- s^c ^ 7 t * %`il 5 a sf *04 Thriho)d ittimat* tmg/kg-diy) FIGURE 1. Frequency distribution of threshold estimates obtained for rat and mouse data. number, the biological end-point(s) for the study, the number of ex posure-response pairs used in calculations, the input units of exposure (ppm or milligrams per kilogram body weight per day), the estimated threshold (in ppm and milligrams per kilogram body weight per day), *" the coefficient of determination (/?*), the literature reference used, and a modeled set based on the same or similar data. For example, the data from Maltoni BT-1 for incidence of animals with any type of tumor was entered into the computation as milligrams per kilogram body weight per day in set 423 and as ppm in set 018. Exposure values given as ppm in the original reference were also transformed to estimated dose (mil ligrams per kilogram body weight per day) prior to the modeling. Values in parentheses in this table are unit conversions made post modeling. The rationale and significance of these unit conversions are given in the discussion section. Figure 1 shows that similar estimates of the threshold are obtained overall using either metric in the mod eling. AP00020026 '^9 10 C S. OLSON AND D. J. SCHAEFFER The protocols for Maltoni's (Maltoni and Lefemine, 1975; Maltoni, 1979; Maltoni et al., 1981) inhalation experiments BT-1# -2, -6, -9, and -15 using Sprague-Dawley (S-D) rats and BT-7 using Wistar (W) rats, and Caputo's (Caputo et al., 1974) study using Wistar rats (4 h/d x 5 d/wk for 12 mo) were virtually identical. The Hong et al. (1981) experiment differed only in having 1 h more per day and a shorter (10 mo) total exposure period. Threshold estimates for all of these studies are similar irrespective of strain used and exposure duration (17 to >52 wk). DT estimates for Zymbal-gland tumors are consistently at the high end of the range. Estimates of Or for other tumors and neoplasia are distrib uted throughout the range, which implies that the threshold level must be about the same for ail affected organs. Thresholds were estimated for ingestion experiments using gavage [BT-11 and BT-27 (Maltoni, 1979; Maltoni et al., 1981)3 and feeding (Feron et al., 1981). Values of DT for several end-points are randomly distributed within the range found for inhalation experiments (Table 2 and Fig. 1). Results for BT-11 (S-D rat) and Feron (W rat) are similar in spite of the somewhat different protocols and strains of rat. Mouse Most of the threshold estimates for various inhalation experiments fall in a range of 0.1 to 10 mg/kg body weight d (Table 3, Figs. 1 and 2). The arrangement of data in Table 3 follows the arrangement in Table 2. Pre- or postmodeling conversion of units (ppm or milligrams per kilogram body weight per day) made no difference in the distributions of estimates expressed in the same units (Fig. 2). Two of the studies were for long-term chronic exposure. Experi ment BT-4 of Maltoni and Lefemine (1975) exposed Swiss mice using almost the same protocol as they used with the rat (4 h/d, 5 d/wk, 30 wk, and observed until death). The other experiment (Hong et al., 1981) exposed CD mice for 6 h/d, 5 d/wk for 26 wk and observed them for an additional 12 mo. DT values for these two studies overlap com pletely. In addition to the long-term chronic experiments, there are several other experiments for the mouse with exposure periods from one ses sion to several months and with the observation period terminated by sacrifice or lasting for varying times after exposure. Figure 2 shows that a similar range of threshold estimates was obtained from data for ex periments of varying protocols. Threshold estimates for pulmonary tumor (adenoma) are represented in each line of each experiment of Fig. 2. The estimates for this specific end-point fall in the narrow range of 1-10 mg/kg body weight d with the exceptions of two sets (527, 642) of Hong et al. (1981) 13-wk inhalation exposure and the Suzuki (1983) 4-wk experiment (in sets 640, 642). AP00020027 '5; Maltoni, , -6. -9, and A"> rats, and d x 5 d/wk experiment ;0 mo) total > are similar >52 vvk). Dt high end of i are distribU level must .sing gavage ind feeding . l' randomly _*nts (Table 2 e similar in .\periments Tigs. 1 and ujnt in Table tiligrams per distributions iure. Expert mice using !. 5 d/wk, 30 get ah, 1981) ved them for jverlap com- v are several rom one ses* rminated by - shows that ' data lor ex"iT pulmonary "'Poriment of *^rrow range <327. *'he Suzuki ao .D e H3 52311358882185935258 (S23So o 2 -I dn d^ $oSo'-oStdS r'ood^oo t-^-oo> j fe"s. onc-. o<o--- dL5--e-H, s_~ Oe rS--, io--n mo *- v4 CQOC003C5eo008 --. o, eN gN gN,gNdnD*1 oNe1/1d*"WNu ^ul RRR -0080gS0080fe0:g080|S0 S= S02fi.^0S0f0iRo)d:Sd 3o' Rd :n eo O rJ rr ci ^ d aaEfi.Ea.fi.tsEEaaaaEtaESESt E 1c.."E t 3t 8. 5 I f u. 5 - - . -a ^ n r* 2 .fi 2 g.2 u. 5 * 2 %J35 ^ "?5 ^ % i.ig.2 88.g.=s a< 2 aa. "oa. "63. ' a 2 222 % t/)Tk?"55<l^k 7S73J_?^`"<5 .3^2 SSS!I!SSSDi?RDjoRfl5!0? 8 I -- + + + + + + + + + +u+ + S^+ Ot+nnf+ iw+Hw+Nm+noN+in 5 JS 8 8. >V SI -- r- ^ r- r* n ^ wo n^v Pm So S,I ^a. . 2 2 s? 5* ^4,1 "s 3 K3 I3S 3O' 351a5S5g2g|g|f5|g2| AP00020028 TABLE 3. Threshold Estimates (D,) for Mouse Inhalation Data Obtained with tiltor Model ICnnfinui-d) Set Weeks of number exposure Weeks oil observation*1 End-point1' Number of dfr |Uirs Input units t>, estimate ppm rngykg d Correlation K' Kelrrenre* Related set number 434 30 + 43 all 7 mg (2.4) 1.2 0.97 7 552 552 30 +43 all 7 ppm 2.4 (1.2) 0.97 7 434 527 13 +52 pul, F 4 mg -- 2.r 1 4 640 !I 668 31-39 m 31-39 0 total #, F 4 ppm 3.1 0 total #, M&F 4 ppm 3.2 (2.3) (2.4) 1 1 5 689 5 691 646 13 + 52 total #, F 4 mg -- 2.4J 0.9*) 4 645 691 31-39 0 total #, M&F 4 mg 0-3) 2.4 1 S 690 645 13 + 52 total #, F 4 ppm 3.4rf <2.51* 0.99 4 646 670 26 + 52 total #, F 4 ppm 3.6 (2.7) 0.93 4 671 671 26 + 52 total #, F 4 mg (3.6) 2.6 0.93 4 670 463 31-39 Q pul 4 ms 14.4) 32 0.97 5 675 666 31-39 0 total #, M 4 ppm 5.0 (3.7) 1 5 687 632 4 + 52 total #, F 4 ppm (3.S)1 -- 1 4 bJJ 687 31-39 0 total #, M 4 mg (5.3) 3.9 1 5 686 657 18-26 0 pul, M&F 4 ppm 5.4 (4.0) 0.99 S 658 6S5 18-26 0 pul, F 4 ppm 5.6 (4.1) 1 5 656 676 31-39 0 pul, F 4 ppm 5.6 (4.1) 1 5 464 650 10-26 0 pul, M&F 4 mg (6.0) 4.4 0.99 5 656 18-26 0 pul, F 4 mg (6.1) 4.5 1.0 5 655 464 31-39 0 pul 4 mg (6.1) 4,5 0.99 5 676 684 31-39 0 AS #, F 4 ppm 6.5 (4.8) (1.94 5 685 685 31-39 0 AS #, f 4 mg (6.9) 5.1 0.94 5 684 663 26 + 52 LAS, M&F 4 ppm 7.0 15.2) 0.94 4 664 664 26 + 52 LAS, M&F 4 mg (7.5) 5.5 0.94 4 663 679 31-39 0 LAS, M 4 ppm 7.7 (5.7) 1 5 465 677 31-39 0 pul, M&F 4 ppm 8.0 IS.9) 1 5 025 30 43 pul 7 ppm 8.1 (4.0) 0.95 7 432 027 30 43 other 7 ppm 8.3 (4.1) 0.85 7 433 465 31-39 0 IAS 4 mg (8.6) 6.3 0.99 5 679 665 2b + 52 pul, M&F 4 ppm 8.7 (6.4) 0.88 4 667 616 30 D ELAS 5 ppm 9.0 (4.4) 0.95 8 619 678 31-39 0 pul, M&F 4 mg (9.0) 6.6 1 5 677 674 4 + 52 pui, M&F 4 ppm 9.3 (6.9) 0.97 4 524 667 26 + 52 pul, M&F 4 mg (9.4) 6.9 0.86 4 665 432 30 43 pul 7 mg (9.4) 4.6 0.95 7 025 AP00020029 inwKnw 027 JO 46S 31-39 665 26 616 30 676 31-39 674 4 667 26 432 30 43 0 + 52 D 0 + 52 + S2 43 other LAS pul, M&F ELAS pul, M&F pul, M&F pul, M&F pul 7 (("" i 4 mg (.(.) 4 ppm 8.7 5 ppnt 9.0 4 mg <9.(11 4 ppm 9.1 4 ntfi 19.4) 7 mg 19.4) (-t.lt I...I ir.4t <4.4| (.,6 (6.9) r>.9 4.6 <i n, (i 'W 0. 88 0.95 l 0.9? O.ftft 0.95 I 8 5 4 4 7 (.70 (4.7 (.19 (.77 524 6bf> 025 630 4 + 52 pul, M&F 4 PPm 9.6 (7.1) 0.91 4 631 631 4 +S2 1 pul, M&F 4 mg (10.0) 7.7 0.91 4 630 524 4 +S2 pul 4 mg (10) 7.5 0.97 4 674 619 30 ElAS 5 mg (11) 5.3 0.95 8 616 433 30 675 31-39 43 other 0 pul, M 7 mg (12) 6.0 0.85 7 027 4 ppm 13 (9.6) 0.99 5 463 651 4-13 0 pul, M&F 4 ppm 15 (111 0.98 5 652 666 26 + 52 IAS, F 4 ppm 16 (12) 0.98 4 528 635 4 + 52 total #, M&F 4 mg -- 12* 1 4 634 652 4-13 0 pul, M&F 4 mg (18) 13 0.98 5 651 523 26 + 52 LAS 4 mg (18) 13 0.96 4 666 617 30 597 --* D pul D pul 4 ppm 18 5 ppm 18 (8.7) (3.1) 0.99 0.96 8 3 6_20 636 13 +52 LAS, F 4 ppm 19 (14) 1 4 52S 660 31-39 0 IAS, M&F 4 ppm 19 (14) 0.99 5 681 631 31-39 0 LAS, M&F 4 mg (22) 16 0.99 5 680 52S 13 + S2 LAS 4 mg (22) 16 0.99 4 636 637 13 +52 LAS, M&F A ppm 23 (17) 0.9B 4 638 620 30 D pul A mg (25) 12 0.99 8 617 703 26 + 52 pul. F 4 ppm 25 (18) 0.88 4 530 430 30 43 LAS 7 mg -- 18rf 0.88 7 638 13 + 52 LAS, M&F A (27) 20 0.98 4 637 633 4 +52 total #, F A mg -- 21d 0.9S 4 632 530 26 + 52 pul A mg (30) 22 0.88 4 703 618 30 D skin 6 ppm 38 (18) 0,99 8 621 621 30 O skin 6 mg (55) 27 0.99 8 618 * +, Alter exposure; +0 sacrificed after exposure; D, to death. 6 pul. Pulmonary; AS, total angiosarcoma, liver and elsewhere; ElAS, exluliver angiosarcoma; other abbreviations as in Table 2. ' Literature citation given in Table 2, footnote c. u Dt very large; threshold estimate obtained by confidence interval method. * 0| = <10'*; threshold estimate obtained by Rosen technique (see methods section). ' Dt very large with original dose-response-set; this O, obtained with set of three d-r pairs. 8 Single exposure. * Sets 661 and 662 Include spontaneous tumors, while sets 659 and 660 do not. Data from Suzuki (1983). AP00020030 14 C. S. OLSON AND D. J. SCHAEFFER 5,001 0.01 0.01 o.l OJ If 1.0 10.0 Thrtslioltf oiumiu (mg/lg 0V> FIGURE 2. Frequency distribution of threshold estimates lor mouse experiments Of different ex posure duration. All data except that Of Hehir et al. (19811 were obtained by input in milligrams per kilogram per day; + indicates data obtained at 41 wk (Maitoni and Lefemine, 1975). When compared on the basis of exposure (ppm), Dr values for the mouse are dearly less than those for the rat. The median DT is 22 ppm for the rat and 8.2 ppm for the mouse, and 92% of the mouse values are less than the rat median. However, when compared on the basis of estimated dose (milligrams per kilogram body weight per day), the distribution of DT values are similar and the median for each is about 4 mg/kg body weight d. DISCUSSION The filter model has been used to make a large number of threshold estimates from the extensive literature on VC. In general, these esti mates are plausible in terms of the original data set. For instance, the estimate is clearly less than the first obviously positive result in the set. Also, a conservative estimate of Dr of 1 mg/kg body weight d, given as the 10-percentile of the distribution, is small compared with the intake of about a hundred mg/kg body weight d of natural carcinogens estimated by Ames (1983). Furthermore, the potency, and thus the hazard, of VC is not unusually large compared to potencies of other AP00020031 > D. |. SCHAEFFER J______L i i* i* Ml I -- 1 j ic. nts of different exinput in milligrams une. T975). values for the Dr is 22 ppm mouse values d on the basis t per day), the each is about erof threshold ral, these esti>r instance, the ?sult in the set. eight d, given jared with the iral carcinogens , and thus the jncies of other APPLICATION OF THE FILTER MODEL TO VC 15 human-made and natural carcinogens (Crouch and Wilson, 1979; Crouch et al., 1983; Sugimura, 1982; Gold et al. 1984). Threshold estimates are also plausible when compared with mea sured and estimated levels of exposure leading to binding of VC with macromolecules. Thus there is measurable protein binding at an ex posure level of 1 ppm for 6 h (= 0.5 mg/kg body weight d) in the rat liver (Watanabe et al., 1978). From their dose-response data, we esti mate with the filter model that the threshold for binding to protein is 0.1 mg/kg body weight d. Since the level of binding to DNA is about one-hundredth that for cytoplasmic protein for a given exposure (Bolt et a!., 1980), a threshold for DNA binding would presumably only occur at a higher exposure level. However, binding to DNA has been mea sured in rats exposed to 10 ppm for 6 h (= 5 mg/kg body weight d) (Guengerich and Watanabe, 1979) and is approximately linear from 0 to 250 ppm. An interpolation of this data to a dose of 1 mg/kg body weight d gave an estimate of 40 adducts per cell. Estimates were sim ilarly made from two other studies that had used single dose exposures of VC to rat {Bolt et al., 1980) and mouse (Bergman, 1982). Binding levels of 13,000 and 260 adducts per cell, respectively, were estimated for doses of 1 mg/kg body weight d. (This calculation used the con version factor from the methods section and assumed 10 e for the weight of liver in rat, 109 cells per liver in rat, and 10 x 10-^2 g DNA per cell for rat and mouse.) These calculations Indicate that the evi dence is mixed on whether there will be a significant level of adduct formation at 1 mg/kg body weight d in comparison with the hundreds to thousands of lesions formed as "background" events in a cell each day (Stott and Watanabe, 1982). Such adducts as there are could be further reduced by repair before the possibility of fixation by replicative DNA synthesis. Also, the threshold estimate we give may refiect re versal of the process at later stages in a multistep process of carcino genesis. Estimates of Dr for rats or mice cover a range of two to three de cades. The large number of estimated values for each species defines the distribution of DT in detail and brackets the portion of the infinite range between the lowest exposure and zero, which should be used for risk assessment. The variability in the threshold estimates should also be compared to that for biological data in general. Incidence for cancer in different concurrent control groups can vary from 0 to 28% (Society of Toxicology, Task Force of Past Presidents, 1982), while dif ferences in enzyme activity can vary by a factor of 150 (Harris et al., 1982). Tumor rates in replicate groups in the ED01 study differed by up to 3- to 6-fold for controls and treatment groups, and Hartley-Sielkin risk estimates for "virtually safe dose" at 30 mo differed by 0.04-2 ppm for replicates (Society of Toxicology, 1981). The results for the mouse experiments of various durations of ex- AP00020032 16 C S. OLSON AND 0. J. SCHAEFFER posure and/or observation, along with the rat experiment BT-3, suggest that the instantaneous level of exposure is more important than the cumulative effects over time. A practical conclusion from these results is that in some cases a carcinogen assay need be carried out only until there is a noticeable exposure-response relationship over several ex posures. This is in contrast to the recommendations of the ED01 group (Gaylor, 1980). In all the experiments used for estimating threshold, exposure started when the animals were young. Groth et al. (1981) exposed rats of different ages for a short time at a single dose. Based on lifetime observation, they concluded that older animals were more susceptible than younger animals. From their work, one would predict that animals exposed for longer periods of time would show a lower threshold simply due to aging. However, Drew et al. (1983) point out that Groth et al. (1981) did not really let the group of animals exposed at a younger age live long enough after the exposure period to develop the typical VC-induced cancer--liver angiosarcoma. In fact, Drewet al. (1983) con cluded from their own single-dose experiment that younger animals are more susceptible. They also noted that similar results have been obtained for other substances. Younger animals are known to have higher cell-proliferation rates and higher levels of endogenous pro moters. The mouse series shown in Fig. 2 probably shows the threshold for younger animals. It is unfortunate that these workers did not do a dose-response study at different ages, since the susceptibility (total incidence of cancer at an above-threshold dose) could be age-depen dent with a threshold that was age-independent. The scatter in the threshold estimates makes it difficult to decide if there is a lower threshold for older animals or animals exposed chronically for a longer period into old age. Since these estimates are averaged over age andother factors, they are population, not individual, estimates. Hence, if additional variables could be controlled, distinguishable estimates might be developed for different subgroups, although these must all be within the range developed here for the population. The conclusion Jhat the threshold is a function of exposure level, not duration, is con sistent with the meaning of "threshold" and contrasts sharply with du ration-dependent incidence rates (Druckrey, 1959) at higher exposures. Recent studies note that when rat and mouse (and also hamster) data are compared over a number of chemicals, there is rather good agreement in the carcinogenic potency of the chemicals (Schach von Wittenau and Estes, 1983; Crouch and Wilson, 1979; Gold et al., 1984). Dt estimates follow this pattern; the ranges of Dr estimates for rat and mouse overlap considerably when scaled by using standard respiratory constants. Because the method used to estimate Dr is nonlinear, there is no a priori reason why the distributions of Dr values estimated from data transformed pre- and postmodeling (Fig. 1) should be the same if : I AP00020033 D.|. SCHAEFFER ;T-3, suggest mt than the hese results ut only until r several ex- ED01 group dr exposure exposed rats J on lifetime i susceptible . that animals er threshold jt that Groth at a younger p the typical ;l. (1983) connger animals ts have been own to have lgenous prohe threshold did not do a jtibility (total e age-depenxatter in the re is a lower y for a longer over age and tes. Hence, if >le estimates hese must all conclusion -ation, is conarply with duer exposures, also hamster) 5 rather good > (Schach von d et al., 1984). :es for rat and rd respiratory >nlinear, there ''timated from ue the same if APPLICATION OF THE FILTER MODEL TO VC 17 the estimates only reflect mathematical relationships. If the estimates reflect biology, however, a DT value obtained from data expressed In one set of units must be linearly transformable to other units. The fact that this linear transformation is possible with Dr values strengthens our assertion that the fitter model is a valid mathematical expression of the net biochemical events that are scored as exposure-responses. Comparisons of DT With Other Estimates of Risk The Dt data base can be compared with the carcinogenicity potency (TD50) (Peto et al., 1984) values reported for vinyl chloride (Gold et al., 1984). The shapes of the distributions of Dr values given in Figs. 1 and 2 are very similar to the shapes of the distributions of TD50 reported by Gold et al. (1984) for vinyl chloride. The similarity between the dis tributions of the two measures is consistent with our assertion that De values are legitimate extrapolations from the observable range to very low exposures. Additional support for the assertion is given by the multiple correlation (r = 0.78; p = 0.0008) for the regression through the origin for log(TD50) and log(Dr + 1) values for nine sets of data. In evaluating the significance of our results, it was instructive to compare Dr values with other estimates of lifetime risk for VC, as we have done for other compounds (Schaeffer et al., 1982). One formal comparison was made previously by solving the model [Eq. (1)] as a Lagrange-gamma multihit dose-response model (lanardan and Schaeffer, 1984). The point estimate of 0.001 ppm for the "virtual safe dose" at an excess risk of 10"6 for the set used by Rai and Van Ryzin (1979) compares favorably with their estimate of 0.00056 ppm made using a generalized /(-hit dose-response model. Significance of DT es timates can also be judged by comparing them with the numerous published estimates of 0-0.1 for the excess lifetime risk to rodents and humans exposed to 1 ppm VC (OSHA 1980, p. 5200). Additional com parisons for specific data sets for most of the models commonly used for risk analysis can be made from the data in Tables 2-4. - Risk to Humans DT estimates for rat and mouse demonstrate that the procedure used here provides consistent estimates of the threshold, taken over all routes of exposure and end-points. Because the threshold is an intrinsic property of the biologic system with a basis in thermody namics, interspecies conversions of DT values should not be subject to the uncertainties in conversions of exposure-response data. While the delivered dose is dependent on metabolic and kinetic factors that differ with species and individuals, thermodynamic requirements for the reaction of a sufficient delivered dose are determined by the structures of the carcinogen and target molecules. Under conditions of homeostatic equilibrium (Eyring and Stover, AP00020034 AP00020035 TABLE 4. Low-Dose Extrapolations Using Data of Malloni* Model Experiment Linear Cayior and Kodeil (1980 Kuzmacfc and McGaughy 11975) Extreme value Krewski and Van Ryzin (1961) Logit Krewski and Van Ryzin (1961) Schneiderman et al. (1975) Probit Krewski and Van Ryzin (1961) Schneiderman el al. (1975) Weibull Carlborg (1961) Gamma muftihit Rai and Van Ryzin (1979) Cornfield et al. (1980) Van Ryzin and Rai (I960) Krewski and Van Ryzin (1981) Haseman, (1981) Food Safely Council (1978) Multistage Crump et al. (1977) Guess et al. (1977) Food Safety Council (1978) Gaylor and Kodeil (1960) Cornfield et al. (1990) Krewski and Van Ryzin (1961) Haseman et al. (1981) U.S. Environmental Protection Agency (1980) BT-I BT-1 BT-1 BT-1 BT-1 BT-1 BT-1 UT-1 UT-1 BT-1 BT-1 BT-1 BT-1 BT-1 BT-1 BT-1 BT-1 BT-1 BT-1 BT-1 BT-1 BT-1 Maltoni et al., 1961. * VSD, virtual safe dose. Data 50-6000 50-6000 50-6000 50 - 500 50-6000 so--soo 5O-WJ0G 50-500 50-6n0 50-WW0 50-6000 50-6000 50-6000 50-500 50-2500 50-6000 50-6000 50-6000 50-6000 50 - 6000 50-2500 Risk to-* 10'* io-* 10*s to-* io-s 10'* 10* 10-* 10'* 10'* 10** 10'* 10-* 10-" 10-* 10* io-* 10* 10 ' 10'* 10'* VSDh 7 x 10* 2 x 10 * 4 X IO-7 2 X 10** 0.119 5 X IO-3 0.073 6 x 10 8 5 X 10'* 3 x IO'' 0.12 8.4 x IO'1 3.9 x 10'* 3.9 x IO-1* 5 x 10'* IO'*- IO'1 0.02 5.2 x 10 * 2 0.2 0.02 0.005 J*. vINi <* Nei Se Se ?!2SSS S1 5S-n SSS8S 3 a Gfi SC CD (B l ra Xs c ? 1JS U'! APPLICATION OF THE FILTER MODEL. TO VC 19 1970), the net free energy for the reaction of a carcinogen with a target molecule in in vitro systems, hence the threshold, should not vary appreciably between species. This was found in the agreement of threshold estimates (in milligrams per kilogram body weight per day) for mice and rats, and we expect that thresholds for humans are also about the same. Thus, although the measured metabolic rate for VC in humans is about 20% that for the rat (Bolt et aL, 1981), which is about the same fraction for basal metabolism (Oser, 1981), the DNA repair rate for human cells is 5-9 times faster than rat (Hart and Setlow (1974); based on exposure to a different xenobiotic]. The National Academy of Sciences (1975) estimated that workers exposed to 280 mg/kg body weight d VC based on exposure of 150 ppm for 8 h/d, 5 d/wk, 50 wk/ yr for 5 yr, suffered a risk of 0.2% excess angiosarcoma. Using linear extrapolation, a risk of 10"* corresponds to 0.14 mg/kg body weight d (~20 ppm), in excellent agreement with this expectation. The question remains how large data sets such as ours or that of Cold et ai. (1984) should be used for risk assessment. One suggestion is to apply the same methods that are used to set safety exposures for toxic substances. These exposures are often estimated as one-hun dredth the median toxic dose (e.g., LD50, LC50, etc.). Ratios of the median of all DT estimates to the median of all TD50 values are: 0.010 (rat), 0.04 (mouse), 0.012 (rat + mouse). These ratios suggest thatTDSO values might be similarly adjusted using a factor of 0.01, at least as a first attempt. A different problem concerns the scatter observed in estimates made using various data. When only a few values are available, the median of the DT estimates could be used. When enough values to characterize the (log-normal) distribution of DT estimates are available, some reasonable probability point might be chosen. We suggest, for reasons based on the expected scatter in observations from a log normal distribution (Vysocanskii and Petunin, 1980), that this point be the lower three-standard-deviations bound from the mean of the log arithms of the values. In conclusion, we have used the filter model with a large number of dose-response sets for VC. This model considers the damage/re sponse aspects of homeostasis. Threshold estimates at the lower end of the range (1 mg/kg body weight - d) are at a level where there is an approximate threshold for DNA adduction in one rodent study but not in two others. The threshold estimate can be used in a risk assessment even when the experimental protocol is for short exposure and less than lifetime observation. REFERENCES Ames, 8. N. 7983. Dietary carcinogens and antieardnogens. Se/ence 227:1256-1264. Bergman, K. 1952. Reactions of vinyl chloride with RNA and DNA of various mouse tissues in vivo. Arch, Toxicol. 49:177-129, AP00020036 20 C. S. OLSON AND D. J. SCHAEFFER Bolt, H. M., Fllser, J. G., laib, R. and Ottenwalder, H. 1980. Binding kinetics of vinyl chloride and vinyl bromide at very low doses. Arch. Toxicol. Suppl. 3:129-142. Bolt, H. M., Laib, R.J., Filser, ). G., Ottenwalder, H., and Buchter, A. 1981. Vinyl chloride and related compounds: Mechanisms of action on the liver. In frontiers in liver Disease, eds. P. D. Berk and T, C. Chalmers, pp. 80-92. New York: Thieme-Stratton. Caputo, A., Viola, P. L., and Bigotti, A. 1974, Oncogenicity of vinyl chloride at low concentrations in rats and rabbits. IRCS Med. Sci: Libr. Compend. 2:1582. Carlborg, F. W. 1981. Dose-response functions in carcinogenesis and the Weibull model. Food CosmeC. Toxicol. 19:25S-263. Cornfield, |., Rai, K., and Van Ryzln, J. 1980. Procedures for assessing risk at low levels of exposure. Arch, Toxicol. Suppl. 3:295-303, Crouch, E., and Wilson, R, 1979. Interspecies comparison of carcinogenic potency. J. Toxicol. Environ. Health 5:1095-1118. Crouch, ., Wilson, R., and Zeise, L. 1983. The risks of drinking water. Water Resour. Res. 19:13591375. Crump, K. S., Guess, H. A., and Deal, K. L. 1977. Confidence intervals and test of hypothesis concerning dose response relations inferred from animal carcinogenicity data. Biometrics 33:437-451. Drew, R. T., Boorman, G. A., Haseman, J. K., McConnell, E. E., Busey, W. M., and Moore. J. A. 1983. The effect of age and exposure duration on cancer induction by a known carcinogen in rats, mice and hamsters. Toxicol. Appl. Pharmacol. 68:120-130. Druckrey, H. 1959. Pharmacological approach to carcinogenesis. In C/ba Foundation Symposium on Carcinogenesis: Mechanisms of Action, eds. G. E. W. Wolstenholme and M, O'Connor, p. 110. Boston: Little, Brown. Eyrlng, H., and Stover, B. J. 1970. The dynamics of life. II. The steady-state theory of mutation rates. Proc. Natl. Acad. Sci. U.S.A. 66:441-444. Eyring, H-. Stover, B. J., and Brown, R. A. 1971. The dynamics of life. V. Applying the steady-state theory of mutations to human cancer, Proc. Natl. Acad. Sci. U.S.A. 63:1670-1672. Feron, V. )., Hendriksen, C. F. M., Speck, A. 5., Til, H. P., and Spit, B. ). 1981. Lifespan oral toxicity study of vinyl chloride in rats. Food Cosmet. Toxicol. 19:317-333. fishbein, L. 1980. Overview of some aspects of quantitative risk assessment. /. Toxicol. Environ. Health 6:1275-1296. Fitch, W. M., and Atchely, W. R. 1985. Evolution in inbred strains of mice appears rapid. Science 228:1169-1175. Food Safety Council. 1978. Quantitative risk assessment. Food Cosmet. Toxicol. 16:109-136. Gaylor, D. W. 1980. The ED0i study: Summary and conclusions. /. Environ. Pathol. Toxicol. 3(3>:179-183. Gaylor, D. W., and Kodell, R. L. 1960. Linear interpolation algorithm for low dose risk assessment of toxic substances. /. Environ. Pathol. Toxicol. 4:305-312. Gold, L. S., Sawyer, C. B., Magaw, R., Backman, C, M., de Veclana, M., Levinson, R., Hooper, N. K., Havender, W. R., Bernstein, L., Peto, R.., Pike, M. C., and Ames, B. N. 1984. A carcin ogenic potency database of the standardized results of animal bioassays. Environ. Health Perspect. 56:9-319. Groth, D. H., Coale, W. B., Ulland, B, M., and Homung, R. W. 1981. Effects of aging on the induction of angiosarcoma. Environ. Health Perspect. 41:53-57. Guengerich, F. P., and Watanabe, P. G. 1979. Metabolism of [14C) and [HCU-labeled vinyl chloride in vivo and in vitro, fi/ocbem. Pharmacol. 28:589-596. Guess, H., Crump, K., and Peto, R. 1977. Uncertainty estimates for !ow-dose-rate extrapolations of animal carcinogenicity data. Cancer Res. 37:3475-3483. Harris, C. C., Trump, B. F., Grafstrom, R., and Autrup, H. 1982. Differences in metabolism of chemical carcinogens in cultured human epithelial tissues and cells. /. Cell. Biochem. 18:285294. Hart, R. W., and Setlow, R. B. 1974. Correlation between DNA excission-iepair and life-span in a number of mammalian species. Proc. Natl. Acad. Sci. U.S.A. 76:2169--2173. AP00020037 \D D. |. SCHAEFFER tics of vinyl chloride Vinvl chloride and : i er Disease, eds. it tow concentrations Veibull model. Food >\v levels of exposure. : potency. J. Toxicol. Resour. Res. 19:1359- nd test of hypothesis licity data. Biometrics M., and Moore, f. A. a known carcinogen .lunaation Symposium ne and M. O'Connor, ite theory of mutation piving the steady-state 3:1670-1672. !. 1981. Lifespan oral -133. nt. i. Toxicol. Environ. vnears rapid. Science to/. 16:109-136. i iron. Pathol. Toxicol. w dose risk assessment Levinson, R., Hooper, .*. 8. N. 198a. A carcinassays. Environ. Health Effects of aging on the H-labeled vinyl chloride Jose-rate extrapolations enees in metabolism of , Ceil. Biochem. 18:28S- repair and life-span In a J-2173. APPLICATION OF THE FILTER MODEL TO VC 21 Haseman, J. K., Hoel, D. C., and Jennrich, R. I. 1981. Some practical problems arising from use of the gamma multihit model for risk estimation. /. Toxicol. Environ. Health 8:379-366. Hehir, ft. M., McNamara, B. P., McLaughlin, J., |r., Willigan, D. A,, Bierbower, G., and Hardisty, ). F. 1981. Cancer induction following single and multiple exposures to a constant amount ' of vinyl chloride monomer. Environ. Health Perspect. 41:63- 72. Hoei, O. G., Kaplan, N. 1., and Anderson, M. W. 1983. implication of nonlinear kinetics on risk estimation in carcinogenesis. Science 219:1032-1037. Hong, C. 8., Winston, |. M., Thornburg, L. P., and Lee, C. C. 1981. Follow-up study on the car cinogenicity of vinyl chloride and vinyildene chloride in rats and mice: Tumor incidence and mortality subsequent to exposure. /. Toxicol. Environ. Health 7:909-924. Jakoby, W. 8. 1980, Detoxication enzymes. In Enzymatic Basis of Detoxication, ed. W. 6. jakoby, pp. 1-6. New York: Academic. janardan, K. G., and Schaeffer, D. J. 1977. Models for the analysis of chromosomal aberrations in human leukocytes. Biom. }. 19:599-612. Janardan, K. G.. and Schaeffer, D. i. 1984. Risk assessment of toxic substances using a Lagrange* gamma multi-hit dose response model. Proc. Social Stat. Section Am. Statistical Assoc. 1984:244 -249, Philadelphia, Penna,, Aug 13-16,1984. Jenner, P., Testa, B., and DiCarlo, F. J. 1961. Kenobiotic and endobiotic metabolizing enzymes: An overstretched discrimination. TrendsPharmaceut. Sc/. 2:135-137. Krewski, D., and Van ftyzin, j. 1981. Dose response models for quantal response toxicity data. Iri Statistics and Related Topics. Proceedings of a Conference, Ottawa, May 5-7, 7980, eds. M. C. Sorgo, D. R. Dawson, J. N. K. Rao, and A. K. M. E. Saleh, pp. 201-231. New York: North Holland. Kuester. J. L.. and Mize. I. H. 1973. Optimization Techniques with FORTRAN. New York: McGrawHill. Kuzmack, A. M., and McGaughy, R. E. 1975. Qir*nt/fat/ve Risk Assessment for Community Expo sure to VC. U.S. Environmental Protection Agency report, December. Lee, C. C., Bhandari, |. C., Winston, J. M., House. W. B,, Dixon, R, L., and Woods. |. C. 1978. Carcinogenicity of vinyl chloride and vinyildene chloride. f. Toxicol. Environ. Health 4: 15-30. lewin, R. 1985. Why do inbred mice evolve so quickly/ Science 226:1187. Maltoni, Cesare. 1979. Vinyl chloride carcinogenicity: An experimental model for carcinogenesis studies. In Origins of Human Cancer, book A, eds. H. H. Hiatt, J. D. Watson, and J. A. Winston, pp. 119-146. Cold Spring Harbor, N.Y.: Cold Spring Harbor Laboratory. Maltoni, C., and Lefemine, G. 1975. Carcinogenicity bioassays of vinyl chloride: Current results. Ann. N.Y. Acad. Sci. 246:195-218. Maltoni, C., Lefemine, G., Cilibertl, A., Cotti, G., and Carrettl, D. 1981. Carcinogenicity bioassays of vinyl chloride monomer: A model of risk assessment on an experimental basis. Environ. Health Perspect. 41:3-29. National Academy of Sciences. 1975. Pest Control: An Assessment of Present and Alternative Technologies. VoI. I. Contemporary Pest Control Practices and Prospects. A Report of the Executive Committee, p. 77. Washington, D.C.: National Academy of Sciences. OSHA, Occupational Safety and Health Administration. 1980. Identification, Classification and Regulation of Potential Occupational Carcinogens. Fed. Reg. 45(15):5001-5296 (]an. 22). Oser, B. L. 1981. The rat as a model for human toxicological evaluation./. Toxicol. Environ. Health 8:521-542. Peto, R., Pike, M. C., Bernstein, L., Gold, L. S., and Ames, B. N. 1984. The TDM: A proposed general convention for the numerical description of the carcinogenic potency of chemicals in chronic-exposure animal experiments. Environ. Health Perspect. 58:1-8. Rai. K., and Van Ryzin, J. 1979. Risk assessment of toxic substances using a generalized multi-hit dose response model. In Energy and Health, eds. N. E. Breslow and A. S. Whittenmore. pp. 99-117. Philadelphia: SIAM institute for Mathematics and Society. Rosen, J. B. 1960. The gradient pro|ectlon method for nonlinear programming. Part I. Linear constraints. /. Soc. Ind. Appf. Math. 8:181-217. AP00020038 C. S. OLSON AND D. I. SCHAEFFER 5chach von Wittenau, M., and Estes, P. C. 1983. The redundancy of mouse carcinogenicity bioas says. Fundam. Appl. Toxicol. 3:631-639. Schaeffer, D- 1983. Thresholds for carcinogenesis and their significance to medical practice. Med. Hypotheses. 10:175-184. Schaeffer, D. ]., janardan, K. G., and Kerster, H, W. 1980. Threshold and dose response estimation using the "filter model." Environ. Int. 4:157-162. Schaeffer, D. J., Janardan, K. C., and Kerster, H, W. 1981. Threshold estimation from the linear dose-response model: Method and radiation data. Environ. Manage. 5:515-520. Schaeffer, D. }., Olson, C., Kerster, H. W., and Janardan, K. C. 1982. The low dose extrapolation problem: a review and a new model. Am J. Math. Manage. Sci. 2:223-251. Schaeffer, D. h, Janardan, K. G., Kerster, H. W., and Clarke, A. C. M. 1983. Statistically estimated free energies of chromosome aberration production from frequency data. Biom. f. 25:275292. Schneiderman, M. A., Mantel, N., and Brown, C, C. 1975. From mouse to man--Or how to get from the laboratory to Park Avenue and 59th St. Ann. N.Y. Acad. Sci. 246:237-248. Society of Toxicology. 1981. Re-examination of the ED^ study. Fundam. Appl. Toxicof. 1:26-l28. Society of Toxicology, Task Force of Past Presidents. 1982. Animal data in hazard evaluation: Paths and pitfalls. Fundam. Appl. Toxicol. 2:101-107. Stott. W. T., and Watanabe, P. G. 1982. Differentiation of genetic versus epigenetic mechanisms of toxicity and its application to risk assessment. Drug Metab. Rev. 13:653-873. Stover, B. )., and Eyring, H. 1970a. The dynamics of life. I. Death from internal irradiation by 231 Pu and 226 Ra, aging, cancer, and other diseases. Proc. Natl. Acad. Sci. U.S.A. 66:132-139. Stover. B. J., and Eyring, H. 1970b, The dynamics of life. III. Mechanisms of nonsurvival and the relation of dose size. Proc. Natl. Acad. Set. IAS.A. 66:672-676. Sugimura, T. 1982. A view of a cancer researcher on environmental mutagens. In Environmental Mutagens and Carcinogens, eds. T. Sugimura, S. Konda, and H. Takebe, pp. 737-741. New York: Us$. Suzuki, V. 1983. Neoplastic effect of vinyl chloride in mouse lung--Lower doses and short term exposure. Environ. Res. 32:91--103. Thompson, D. E. 1985. Heavy radiation and mammalian cells. Science News 127:394-395. Thompson, VV, A., Jr., and Funderlic, R. E. 1981. A simple threshold model for the classical bioassay problem. In Measurement of Risks, eds. G. C. Berg and H. D. Maitlle, pp. 521-533. New York: Plenum. U.S. Environmental Protection Agency. 1975. Scientific and Technical Assessment Report on VC and PVC. EPA-600/6-75-004 STAR series, U.S. Environmental Protection Agency, Office of Research and Development. Washington, D.C. U.S. Environmental Protection Agency. 1980. Ambient Water Quality Criteria for VC. EPA 440S080-078 Oct. 1980. U.S. Environmental Protection Agency. Office of Water Regulations and Standards. Criteria and Standards Division, Washington, D.C. Van Ryzin, and Rai, K. 1980. The use of quanta! response data to make predictions. In The Scientific Basis of Toxicity Assessment, ed. H. Witschi, pp. 273-290. New York: Elsevier. Von Borstal, R. C. 1982. Thresholds and negative slopes in dose-mutation curves. In Environmental Mutagens and Carcinogens, eds. T. Sugimura, 5. Konda, and H. Takebe, pp. 737-741. New York: Liss. Vysocanskli, D. F., and Petunia, (. I. 1980. Justification of the 3-sigma rule for unimodal distri butions. Theory of Probability and Mathematical Statistics (USSR) 21:25-35. Watanabe, P. G., Zempel, J. H., Pegg, D. G.f and Gehring, P. J. 1976. Hepatic macromolecular binding following exposure to vinyl chloride. Toxicol. Appl. Pharmacol. 44:571-579, Waxweiler, ft. )., Landrigan, P, J., Infante, P.f and Shapiro, R. 1981. Conference to reevaluate the toxicity of vinyl chloride, poly(vinyl chloride! and structural analogs. Environ. Health Perspect. 41:1-234. Withey, J. R.. and Collins, B. T. 1976. A statistical assessment of the quantitative uptake of VCM from aqueous solution. /. Toxicol. Environ. Health 2:311-321. AP00020039 , yjpf/!R9S^y &>0- ONCOGEN1C RESPONSE OF STRAIN A/J MICE TO INHALED CHEMICALS Bernard Adkins, Jr. Northrop Services, Inc., Research Triangle Park, North Carolina Ethard W. Van Stee, Jane Ellen Simmons, Scot L. Eustis National Toxicology Program. National Institute of Environmental Health Sciences, Research Triangle Park North Carolina Strain Atf mice were exposed by inhalation (or 6 hid, 5 d/wk, for 6 mo to carbon disulfide, 1,2`dibromoethane, ethylene oxide, naphthalene.nitrogen dioxide, or vinyl chloride. Significant increases in pulmonary adenoma formation were observed fol lowing exposure to 300 ppm carbon disulfide; 20 and 50ppm 1,2-dibromoethane; 70 and200ppm ethylene oxide; 10ppm nitrogen dioxide; and50,200, and500 ppm vinyl chloride comapred to control animals. Repeated studies with 1,2-dibromoethane, ethylene oxide, and vinyl chloride gave similarly significant results. Exposure of mice to 30 ppm naphthalene did not elicit a significant adenoma response. Histopathological examination of lungs from animals in these studies revealed multiple alveolar adenomas. Results from earlier studies with these chemicals, using strain A mice and Swiss mice, and bioassay information with tats and mice were compared with these data. These results provide further information for the va/idation of this in vivo model as a tool tor predicting oncogenic potential following chemical exposure. INTRODUCTION The strain A mouse lung tumor bioassay, originally described in 1940 (Shimkin, 1940; Andervont and Sbimkin, 1940) and refined over the last two decades, has been under extensive investigation as a short-term in vivo model for predicting the potential carcinogenicity of numerous chemicals (Shimkin and Stoner, 1975; Stoner and Shimkin, 1982; Stoner et al., 1984). With the exception of one report (Leong et al., 1971), testing-of chemicals from occupational exposures to date has employed noninhalation routes of administration of test chemicals. Several studies have compared tumor response in strain A mice fol lowing various routes of chemical treatment {Schut et al., 1983; Stoner This effort was supported by National Institute of Environmental Health Sciences fNIEHS) contract N01-E5-79-C009. The authors graterully acknowledge the excellent technical assistance of the Northrop Services, Inc., staff, including Robert W. O'Connor, Andrew K. Miller, Clinton R, Moorman, Rebecca C. Hamrick, Patricia A. Sams, Pamela Y. Seabrook, and Kathy S. Micor, and the NIEHS staff, including Richard A. Sloane and Michael P. Moorman. Requests for reprints should be sent to Bernard Adkins, )r., Northrop Services, Inc., P.O. Bo* 12313. Research Triangle Park. North Carolina 27709. 311 Journal of Toxicology and Environmental Health, 17:311-322, 1986 Copyright 9 I9B6 by Hemisphere Publishing Corporation AP00020040 312 8. ADKINS, JR., ET AL. et at., 1984), tumor response in 2-yr chronic bioassays (Maronpot et al., 1983), and interaction with other test chemicals (Witschi, 1983; Witschi et a!., 1981; Witschi and Doherty, 1984), Previous reports have also characterized the urethane-strain A mouse response (Dourson and Baxter, 1981; Shimkin and Stoner, 1975) and urethane-Swiss mouse pulmonary adenogenesis (Dourson and O'Flaherty, 1982; Sichak and O'Flaherty, 1984). The present study was designed to test the oncogenic response of the strain A mouse animal model with various chemicals administered via inhalation. All chemicals selected were tested at or below their threshold limit value (TLV) (American Conference of Governmental In dustrial Hygienists, 1983) and were either known to be carcinogens or potential carcinogens (genetic toxicity information), or were of un known carcinogenic potential due to an absence of testing data. The potent hepatocarcinogen, vinyl chloride, was tested by inhalation in this sytem and compared to previous results reported with strain CD1 mice (Hong etal., 1981; Lee et al., 1978; Suzuki, 1978, 1983). Similarly, chemicals of increasing occupational/environmenta! concern, 1,2-dibromoethane and ethylene oxide, were tested by inhalation in this system. Three other chemicals (carbon disulfide, naphthalene, and ni trogen dioxide) were also consequently tested during other inhalation studies with these chemicals. Representative lungs from treated and control groups of animals were also examined histopathologicallv to characterize the adenogenic response to the chemicals tested. METHODS Animals Six to 8-wk-old strain A/J male and female mice (15-25 g) obtained from jackson Laboratories, Bar Harbor, Me., were used throughout the study. Female mice were used predominantly in the study, except for one study conducted to compare the adenogenic response of both sexes of strain A/j mice following vinyl chloride treatment. The mice were maintained on hardwood bedding (Sani-chips, P.J. Murphy Forest Products Corp., Rochell, N.J.) in temperature-controlled (21-23C) and humiditytcontrolled (40-60%) exposure rooms with a 12-h light/dark cycle. NIH-31 certified rodent diet (Zeigler Brothers, Gardner, Pa.) or AIN-76 purified diet (Bio-Serv, Inc., Frenchtown, N.J.) and deionized water were provided ad libitum during nonexposure periods. The health status of all animals was observed twice daily, 7 d/wk. All healthy animals were randomly incorporated into treatment and control groups within 2 wk after receipt. Chemicals Reagent grade (98-99% purity) carbon disulfide, 1,2-dibromoethane, and naphthalene were obtained from Fisher Scientific Com- AP00020041 ONCOGENIC RESPONSE TO INHALED CHEMICALS t 313 pany, St. Louis, Mo. The ethylene oxide and nitrogen dipxide were,- supplied by National Welding Company, Raleigh, N.C., as cvrier-grade(>99.7% purity) cylinder gases. The purity of the vinyf chloride used, was not available. Exposure System I All chemical treatments, except urethane (positive Control), werp performed by inhalation for 6 h/d, 5 d/wk, for 6 mo in 1330*;! glass/stainless-steel exposure chambers. Operating conditions included ventila tion with conditioned air at 300-500 Ipm, static pres^bre pf -0.3 to -0.5 in of water, temperature of 20-26C, relative humidity of 40-60%, and 15- to 30-min exhausting prior to unloading animals* Each gas/ vapor phase chemical was metered into the main chamber.air supply through Koch mixing elements (Koch Engineering Company, New York, N.Y.) prior to introduction into the exposure chambers. Chamber chemical concentrations were monitored with Miran infrared spectrophotometers (Model 80, Foxboro Analytical Instruments, South Norwalk, Conn.) and computer-adjusted (DEC POP 11/34, Digital Equipment Corp.) to desired exposure levels via a software feedback loop arrangement similar to that described by O'Connor and Adkins (1984) and Van Stee and Moorman (1984a,b). Animals were exposed in stainless-steel wire-mesh exposure cages by treatment group. Animal cages were randomly rotated within exposure chambers each week both between and within the three or four tiers in the chambers. Chemicals were tested at the following inhalation exposure doses: carbon disulfide, 300 ppm; 1,2-dibromoethane, 50 and 20 ppm; eth ylene oxide, 200 and 70 ppm; naphthalene, 30 and 10'ppm; nitrogen dioxide, 10, 5, and 1 ppm; and vinyl chloride, 500, 200, and 50 ppm. Replicate studies were conducted with 1,2-dibromoethape, ethylene oxide, and vinyl chloride to test for consistency with this animal model. Pulmonary Tumor Bioassay The assay was performed according to the standardized protocol of Shimkin and Stffner (1975) with the exception that chemical treatment was via inhalation. Control groups consisted of conditioned air. chamber-exposure (sham) animals and urethane-treated animals (non* exposed, positive controls), which received single intraperitoneal in-; jections (1000 mg/kg). Lungs were removed from animals following 6 mo of treatment and fixed for 24 h in Tellyesniczky's fixative (Shimkin and Stoner, 1975). Enumeration of the typical pearly white nodules on individual lung lobes was performed in replicate by three, or more dif ferent technicians. Sections of paraffin-embedded lungs having nodules and stained with hematoxylin and eosin were examined histopathologically. \ AP00020042 314 B. ADKINS, |R.( ET AL. Statistical Analysis Lung adenoma frequencies, expressed as the number of adenomas per mouse, were evaluated statistically by comparing experimental with control means of each data point using a one-way analysis of vari ance of Kruskal-Wallis test, as appropriate. When F values indicated significance <p < 0.05), comparisons were made of each group with the sham control using Duncan's new multiple-range test (Freund et al., 1960). RESULTS Pulmonary Tumor Bioassay Table 1 presents results on the prevalence of pulmonary adenomas in control and treated animals for all chemicals tested via inhalation exposure for 6 mo. Treatment-related mortality was observed in the 20 and 50 ppm 1,2-dibromoethane (first study) and 500 ppm vinyl chloride (first study, both sexes) groups of strain A/J animals. Significant in creases in the frequency (tumors per mouse) and incidence (tumors per tumor-bearing mouse lung) of adenoma formation during the testing period were observed with animals exposed to carbon disul fide, 1,2-dibromoethane (two studies), ethylene oxide (two studies), nitrogen dioxide, and vinyl chloride (two studies). Specifically, expo sure of mice to 300 pm carbon disulfide resulted in a small, but statisti cally significant (p < 0.03), increase in the frequency of adenoma for mation. The results of the two studies with 1,2-dibromoethane indi cated a concentration-related increase in the frequency of adenoma formation. These later results were significant and reproducible at 50 ppm. All animals in the first study responded with slightly more tumors than their counterparts In the second study. The results of the two studies with ethylene oxide Indicated a small, but statistically signifi cant (p < 0.05), increase in the frequency and incidence of adenoma formation. The results of the two studies with 1,2-dibromoethane indi cated a concentration-related increase in the frequency and incidence of adenoma formation. These latter results were significant and repro ducible at 50 ppm. All animals in the first study responded with slightly more tumors than their counterparts in the second study. The results of the two studies with ethylene oxide indicated a small, but statistically significant (p < 0.05), increase in the frequency and incidence of ade noma formation. The exposure of mice to naphthalene did not cause any significant change in tumor frequency, but did show a significant increase in tumor incidence following treatment at 10 and 30 ppm. Mice exposed to nitrogen dioxide showed a small, but statistically sig nificant (p < 0.05), increase in tumor frequency and incidence. The results of the two studies with vinyl chloride indicated a concentration- AP00020043 ONCOGENIC RESPONSE TO INHALED CHEMICALS 315 related increase in frequency and incidence of adenoma formation. The response was significant and reproducible and was observed in both sexes. The male mice exposed to 50 ppm in the first study devel oped significantly more tumors than the female mice (p < 0.001) ex posed to the same concentration in both studies. The female mice ex posed to 50 ppm in the second study developed slightly, but statisti cally significant (p < 0.05), more tumors than similar female mice in the first study (controls were not statistically different). The results of the studies showing little or no change in frequency and incidence of adenoma formation following the regimen of chem ical exposure (6 h/d, 5 d/wk, for 6 mo) were further compared with the pooled, negative control response from all experiments. Animals ex posed to carbon disulfide (300 ppm) or to ethylene oxide (70 to 200 ppm) showed a significant increase in a mean tumor frequency, even though the increases were relatively small compared to vinyl chloride and 1,2-dibromoethane. Similarly, 200 ppm ethylene oxide (first study) exposure resulted in a significant tumor incidence when compared to the pooled, negative control response. Exposure to naphthalene was without significant effect on tumor frequency, but was significant when the incidence of tumor formation was compared with the pooled, neg ative control response. The frequencies of adenoma formation from the negative control animals exposed to conditioned air are also presented in Table 1. The mean tumor count among the 30 female mice tested in the first study with 1,2-dibromoethane indicated a statistically significant (p < 0.05) increase from that observed in all studies. The mean tumor counts from the other studies did not differ significantly from each other. The incidence of tumor formation was different in the negative control mice in the naphthalene and first 1,2-dibromoethane studies. The inci dence of tumor formation from the other studies did not differ signifi cantly from each other. The frequency and incidence of adenoma formation from positive control animals given a single intraperitonea! injection of urethane (1000 mg/kg) is_summarized in Table 2. Data are presented by study group for both male and female mice surviving through the 26-wk holding period. The positive control mice from the first study with eth ylene oxide showed a significantly lower frequency of adenoma forma tion than the mice from studies with other chemicals, even though there was 100% tumor incidence. The tumor frequencies observed in positive control animals from other experiments were not significantly different from one another. Histopathologic Evaluation of Lung Adenomas Histopathological examination of lungs from animals used in these studies revealed multiple alveolar adenomas in treated animals and SfrOOZOOOdV ifIf Q. ? 91C s^ssiil^ggg5 2s 3 3 a |* |3|ao* b ei i, S j iSisS -- Cu C6 %3 f 2a. nS ; tS S w ^D kN x^j^Oa^O&4M'-iU4CihC-BtJWwi^^OviptDt^uO*sjjVOmf It i+ U it It it It It H It It It U eeoe?o-vi--wA 11 1 -B _ M -- --i -- O' ^ -* -* N#WUNN It It It It It It It It It It it It It ppppop -- ppopp t l ll S3 &a -w< SSi. *ow mi Naphthalene Nitrogen dioxide Vinyl chloride Vinyl chloride (study 2) 30 10 0 10 5 1 0 S00IM) SOD . 50<M) ' 50 0(M) 0 200 50 0 J* 50 17,51) 2*)- SO ML'50 JO'SO 20/511 2<KS0 5 3/70 47/70 72/72 69/70 66/70 70/70 30/30 28/30 JO/.SO II 1(1 (. <)I4 t S 12 2at>i ntw ur. ItKl 2ft 35 1324 5*) 11 SI) 0.37 t 0.55 1.25 0.07" 2*J (US 0.55 1.25 0.07" 21 0.21 0.39 i.oo o.ixy no 0.45 0.60" 1.55 0.00" ii 0.1.1 0.21 1.17 0.29 12 0.16 0.30 1.39 0.24 SO 0.40 0.02 1.39 0.10 1(X) 86.6 7.16"- 06.B 3.40"- MO 70.3 6.74"' 70.4 2.67"- 88 2.5S 0.73"- 2.92 0.23"- 74 1.59 0.57*'* 2.13 0.10"-- 34 0.39 0.28 1.16 0.03 .58 0.47 0.30 1.22 0.06 ino 43.9 7.83"- 44.1 1.26"- 87 1.97 1,32"- 2.42 0.23"- 26 1.45 a 0.29 * Inhalation exposure: 6 h/d, S df\v, 6 mu. Data presented for female mice, except (or three groups of male mice (M) lesled with vinyl chloride. 6 Survivors: number of survivors at end of stiidy/iiumlrer of animals at start of study. ' Mean standard deviation. " Significantly different (p < 0.05) from the corresponding negative control response within the treatment. r Significantly different fp < 0.05) from pooled negative control tumor response lor all treatments. ' Significantly different Ip < D.05) from the negative urntrirl tumor responses Irom all other treatment groups. AP00020046 316 B. ADKINS, JR., ET AL. TABLE 2. Strain A Mouse Positive Control Response to Urethane (1000 mg/kg)* Surviving animals Study group Total number Percent of of tumors animals with Sex Survivors6 (average) tumors Tumors per mouse' Carbon disulfide Ethylene oxide (study 1) Ethylene oxide (study 2) Vinyl chloride (study 1) Vinyl chloride (study 2) 1,2-Dibromoetbane (study 1,2-Dibromoethane (study 2) Naphthalene Nitrogen dioxide F F F M F F F F F F 20/20 19/20 19/20 20/20 17/20 20/20 19/20 29,'30 20/20 39/40 460 380 447 574 452 232 496 797 578 924 100 22.9 = 2.55 100 20-1 1.77" 100 23.5 6.49 100 28-7 = 3.40 100 26.6 - 2.40 100 24.1 = 5.36 100 26.1 = 4.4Q 100 27.5 z 6.46 100 26.9 z 7,10 100 23.7 = 3.42 ` Urethane was administered in a single intraperitoneal injection followed by 26 wk of observa tion. " Survivors: number of survivors at end of studvmumber of animals at start of studv. f Mean z standard deviation. Significantly different tp < 0.03) from other counts. single alveolar adenomas in untreated animals. The alveolar adenomas were morphologically similar in nearly all mice and consisted of large cuboidal or columnar epithelial cells supported by a sparse fibroblastic stroma and arranged in poorly defined acinar structures with papillary formations. An adenoma in a control mouse from the first vinyl chlo ride study had a primarily solid pattern of growth with uniform poly gonal-shaped cells arranged in small clusters separated by capillaries. There was no apparent difference in the occurrence of pat tern of anatomic distribution of alveolar adenomas occurring in ure thane-treated mice and treated mice of each study group examined. Alveolar epithelial hyperplasia was observed in lungs examined from treated and positive control (urethane) animals and consisted of focal proliferation of cubiodal cells comprising the alveolar epithelium but without effecement of normal alveolar architecture. This Jesion may precede the development of the alveolar adenoma; it was present in the lungs of most treated mice with adenomas and was not observed in examined control lungs. Several changes in the lungs of mice treated with 1,2-dibromoethane clearly distinguished them from others examined, A single bronchiolar adenoma with a prominent papillary pattern was observed, 1 APbbb26647 ONCOGENIC RESPONSE TO INHALED CHEMICALS 319 consisting of tall columnar epithelial cells. In each of the 1,2-dibromoethane-treated mice examined, multifocal hyperplasia and cellular atypia of the bronchiolar epithelium were observed. These changes consisted of focal stratification of bronchiolar epithelial cells or small papillary formations protruding Into the bronchiolar lumens and occa sional cells with enlarged hyperchromatic nuclei. Minimal chronic in flammation accompanied these changes and was characterized by a few widely scattered mononuclear cells in the bronchiolar lamina pro pria. In addition, alveolar epithelial hyperplasia In 1,2-dibromoethanetreated mice was frequently parabronchiolar in location and affected alveoli that abut the bronchiolar walls. In summary, there were no apparent differences in pattern of dis tribution or morphologic appearance of alveolar adenomas occurring in urethane-treated mice and mice treated with ethylene oxide, vinyl chloride, or 1,2-dibromoethane. Bronchiolar epithelial hyperplasia and cellular atypia occurred only in 1,2-dibromoethane-treated mice, and one mouse had a bronchiolar adenoma. No carcinomas were observed in the tissues examined. DISCUSSION This inhalation study was conducted to test first a variety of chem icals of occupational concern in the strain A/J lung adenoma bioassay system. Second, as has been reported earlier for different chemicals and testing regimen (Maronpot et al., 1983),this study was designed to provide additional information as a continuing effort to evaluate this animal model system as a short-term test for predicting in vivo onco genesis. Our findings indicate that strain A/J mice develop lung ade nomas in a dose-related manner with 1,2-dibromoethane, ethylene oxide, and vinyl chloride. These data agree with previous reports in volving various laboratory animals. Including mice, that show positive carcinogenic responses for these same chemicals (National Cancer In stitute, 1982; Carman et a!., 1985; Snellings et al., 1984; Maltoni, 1977; Suzuki, 1983; Viola et al., 1971). Other chemicals tested in this study, including carbon disulfide and nitrogen dioxide, showed significant in creases in frequency and incidence of adenoma formation compared to corresponding control responses. These latter findings, if inter preted to indicate oncogenic potential, are supported in part by a pre viously reported in vitro carcinogenesis study [nitrogen dioxide: Sone et al. (1983)]. Carbon disulfide and napththalene are currently planned for testing in 2-yr National Toxicology Program carcinogenesis bioassays. Comparison of the control strain A/J mouse tumor response with previous reports showed some similarities with our data. The average frequency of adenomas found in our control mice was 41.7% (Table 1), 320 B. ADKINS, JR., ET-AL. with 30.4% of the animals having tumors after the 26*wk duration. These data agree with other 26-wk studies reported by Schut et al. (1983), Shimkin and Stoner (1975), and Stoner et al. (1984), and with a short study duration (4 mo) reported by Witschi et al. (1981). Positive control data expressed as the number of animals developing lung adenomas after a single intraperitoneal injection of urethane (1000 mg/kg) and the frequency of tumor formation in our studies (Table 2) also agree with previous reports (Schut et al., 1983; Shimkin and Stoner; 1975; Stoner et al., 1984). In addition, our data were similar to the con trol data reported by Maronpot et al. (1983). Histopathological charac terization of the multiplicity of the alveolar adenoma response in our studies agrees with the adenogenic response described by Shimkin and Stoner (1975). The lack of congruity with this model system reported by Maronpot et al. (1983) is evident in data presented in this effort. No reports have demonstrated significant oncogenic potential from animals studies for carbon disulfide or nitrogen dioxide, which did elicit significant ade noma formation in our study. Similarly, naphthalene did not display a significant tumor frequency response, but did, however,show a signifi cant tumor incidence at 10 and 30 ppm, which may have been expected considering the formation of unique lesions of nonciliated pulmonary bronchiolar epithelial cells (Clara ceils) of the mouse (Reid et al.,1973; Mahvi et al.,1977; Tong et al.. 1982). Ethylene oxide (70 ppm), 1,2-dibromoethane (20 ppm), and vinyl chloride (50 ppm) did elicit a positive tumor frequency and incidence, as shown in Table 1. These data are congruent with the positive carcinogenic findings that report the lowest effective levels for these chemicals to be 10 ppm (Carman et al., 1985), 10 ppm (National Cancer Institute, 1982), and 25 ppm (Mattoni, 1977), respectively. The lack of consistency of results from different testing laboratories cited by Maronpot et al. (1983) is somewhat evident in our findings. Nonreproducibility between duplicate sudies is shown in Table 1: treatment with 20 ppm 1,2-dibromoethane and treatment with 200 ppm ethylene oxide. Similarly, a lack of consistency in the negative control response is also evident in Table 1. The positive control response to urethane (Table 2) was consistent, except for the tumor response cited during the first ethylene oxide study. On the other hand, the positive findings for 1,2-dibromoethane, ethylene oxide, and vinyl chloride in Table 1 are consistent with positive carcinogenesis data from other an imal studies cited for these chemicals. These discrepancies in tumor response may be explained as follows. Our studies involved inhalation treatment, whereas other studies cited have typically involved other routes of exposure. Thus, the dose, metabolism, and mode of expo sure of target ceils were different from those of noninhalation studies; also, the strains, sources, diet, nutrition, etc. vary between studies in different laboratories. ONCOGENIC RESPONSE TO INHALED CHEMICALS 321 Further validation of this short-term In vivo model for predicting oncogenic potential of chemicals would be required to permit conclu sions relative to the validity and sensitivity of this test. This process should take into account the wealth of information being gathered through both prechronic and chronic testing of these and structurally related chemicals, as well as genetic toxicity data. Treatment of animals with an inert gas (i.e., nitrogen), should also be considered in further characterizing the negative control response. Only after a thorough ex amination and evaluation can the usefulness of this pulmonary ade noma bioassay be assessed. REFERENCES American Conference of Governmental Industrial Hygienist. 1983. Threshold Limit Values for Chemical Substances and Physical Agents in the Word Environment, firs! printing, pp. 10-43. Cincinnati, Ohio: ACGIH. Andervont, H, B., and Shlmkin, M. B. 1940, Biologic testing ol carcinogens, II. Pulmonary tumor induction technique./. Natl. Cancer Inst 1:225-239. Dourson, M. L., and Baxter, C, S. 1981. Reduced prevalence and growth rate of urethane-induced lung adenoma in aging adult strain A mice. Toxicology 20:164-172. Oourson, M. L.. and O'Flaherty. E. |. 1982. Relationship of lurtg adenoma prevalence and growth rate to acute urethane dose and target cell number. /. Natl. Cancer Inst. 89:851-857. Freund, J. E., Livermore. P. E., and Miller, I. 1980. Manual of Experimental Statistics, p. 74. Engle wood Cliffs. N.J.: Prentice-Hall. Carman, R. H., Snellings, W. M., and Maronpot, R. R. 1985- Brain tumors in F344 rats associated with chronic inhalation exposure to ethylene oxide. Xeurotoxicology 6:117-138. Hong. C. B.. Winston, J, M., Thornburg, l. P., Lee, C. C., and Woods, |. S. 1981. Follow-up studv on the carcinogenicity d< vinyl chloride and vinylidene chloride in rats and mice: Tumor Incidence and mortality subsequent to exposure. /. Toxicol. Environ. Health 7:909-924. Lee. C. C, Bhandari, J, C., Winston. J. M.. House, VV. B., Dixon, R, L., and Woods, J. S. 1978. Carcinogenicity of vinyl chloride and vinylidene chloride. /. Toxicol. Environ. Health 4:15-30. Leong, B. K, (., Macjarland, H. N,, and Reese, W. H., Jr. 1971. Induction of lung adenomas by chronic inhalation of bis(chloromethyi) ether. Arch. Environ. Health 22:663-666. Mahvi, D , Bank, H., and Harley, R. 1977. Morphology of a naphthalene-induced bronchioiar lesion. Am. J. Pathol. 86:559-572. Maltoni, C.-1977. Vinyl chloride carcinogenicity: An experimental model for carcinogenesis studies. In Origins of Human Cancer. Book A, eds. H. H. Hiatt, J. D. Watson, and |. A. Winsten. pp. 119-146, Cold Spring Harbor, N.Y.: Cold Spring Harbor Laboratory. Maronpot, Ri. R-, Witchshi, H. P., Smith, L. H., and McCoy, ]. L. 1983. Recent experience with the strain A mouse pulmonary tumor bioassay model. Environ. Sc/. Res. 27:341-349. National Cancer Institute. 1982. Carcinogenesis bioassay of 1,2-dibromoethane linhalation study). TR-210(CAS No. 106-93*4). Carcinogenesis Testing Program. Department of Health and Human Services publication (NIH) 81-1766. O'Connor, R. W., and Adkins, B., }r. 1984. A computer-asisted system for inhalation toxicology. Toxicologist 4(1):126. Reid, W. D., lieu, K. F., Click, J. M,, and Krishma, G. 1973. Metabolism and binding of aromatic hydrocarbons in the lung. Am. Rev. Respir. Dis. 107:539-551. Schut, H. A. J., Loeb, T. R., Crimes, L. A., and Stoner, G. D. 1983. Distribution, elimination, and test for carcinogenicity of 2,6-dinilrotoluene after inlraperitoneal and oral administration to strain A mice. /. Toxicol. Environ. Health 12:659-670. Shimkin, M. B. 1940. Induced pulmonary tumors in mice, II. Reaction of lungs ot strain A mice to carcinogenic hvdtocarbons. Arch. Pathol. 29:239 - 255. B. ADKINS, JR., ET AU Shimkin, M. B., and Stoner, C- D. 1975. Lung tumors in mice. Application to carcinogenesis bioassay. Adv. Cancer Res. 21:1-58. Sichak, S. P., and O'Flaherty, E. J. 19W. Consideration of the mechanism of pulmonary adenogenesis in urethane-treated Swiss mice. Toxicol. Appl. Pharmacol. 76:397-402. Snellings, W. M., Weil, C, S., and Maronpot, R. R. 1984. A two-year Inhalation study of the carci nogenic potential of ethylene oxide in Fischer 344 rats. Toxicol. Appl. Pharmacol. 75:105-117. Sone, 5., Brennan, L. M., and Creasia, D. A. 1983. In vivo and in vitro NO* exposures enhance phagocytic and tumoricidal activities of rat alveolar macrophages. /. Toxicol. Environ. Health 11:151-163. Stoner, G. D., and Shimkin, M. B. 1982. Strain A mouse lung tumor bioassay. J. Am. Coll. Toxicol1:145-169. Stoner, G. D., Greisinger, E. A., Schut, H. A. J., Pereiara, M. A., Loeb, T. R., KJaunig, J. E.. and Branstetter, D C, 1984. A comparison of the lung adenoma response In strain M mice after intraperitoneai id oral administration of carcinogens. Toxicol. Appl. Pharmacol. 72:313323. Suzuki, V. 1978. Pulmonary tumors induced in mice by vinyl chloride monomer. Environ. Res. 16:285-301. Suzuki, Y. 1983. Neoplastic effect of vinyl chloride in mouse lung--lower doses and short-term exposure. Environ. Res. 32:91-103. Tong, 5. 5., Lowe, M. C.. Trush, M. A., Mimnaugh, E. C., Cinsburg, E., Hirokata, Y., and Gram, T. E. 1982. Bronchiolar epithelial damage and impairment of pulmonary microsomal monoox ygenase activity in mice by naphthalene, fxp. Mol. Pathol. 37:358-369. Van Stee, E. W., and Moorman. M. P. 1984a. Overview of a system for the computer-assisted operation of a small animal inhalation facility. Environ. Health Perspect. 54:299- 310. Van Stee, E. W., and Moorman, M. P. 1984b. Calibration of a System for the computer-assisted operation of a small animal inhalation facility, Environ. Health Perspect. 54:311-320. Viola, P. 1., Bigotti, A., and Caputo. A. 1971. Oncogenic response oi rat skin, lungs and bones to vinyl chloride. Cancer Res. 31:516-522. Witschi, H. P. 1983. Promotion of lung tumors in mice. Environ. Health Perspect. 50:267-273. Witschi, H. P., and Doherty, D, G. 1954. Butylated hyroxyanisole and lung tumor development m A/J mice. Fundam. Appl. Toxicol. 4:795-801, Witschi, H. P., Hakkinen, P. J.. and Kehrer. J. P. 1981. Modification of lung tumor development in A/J mice. Toxicology 21 ;37-43. Received October 70, 1985 Accepted November 7, 1985 AP00020051