Document omgN7ZQ3yNn4g5KdBoz2zz1G8

n DDMreated rats. Toxicol. *9. GrEim, H. (1984). ^^wroxine on 2,3,7,8-teti^^DD)-induced toxicity. 2,372-376. -CHEUFLER, E., PAZDER985). Thyroid hormones 7,8-tetrachlorodibenzo-pEnviron Health 16,481-- M., Moore, R. a., and of 2,3,7,8-tetrachlorodim body weight and lipid 'ppl. Pharmacol. 81,356- ND DORMANDY, T. L. veen erythrocyte superoxythrocyte copper levels in ; with rheumatoid arthri-403. lotmone action at the cell 17-123, 173-177. R, Hassan, M. Q., Muraadpour, H. A. (1986). reactive oxygen species in idation. Adv. Exp. Med. and Murray, W. J. roxidation and inhibition TCDD. In Banbury Re ims ofDioxin Action (A. ugh, Eds.), pp. 241-253. MBL', Cold Spring Harbor, (1978). Determinajcursor in tissues by thiotchem. 86,271-278. n of cardiotoxic hazards. 187. ore, J. A,, and Gupta, and clinical chemistry Jibenzo-p-dioxin in laboerspect. 5,111-118. IOXICOLOGY AND APPLIED PHARMACOLOGY 95, 185-199 ( 1988) Physiologically Based Pharmacokinetic Modeling with Methylchloroform: Implications for Interspecies, High Dose/Low Dose, and Dose Route Extrapolations R. H. Reitz, * t'` J- N. McDougal^ M. W. Himmelstein4 R. J. NOLAN,t AND A. M. SCHUMANNf *Toxicology Research Laboratory, 1803 Building, Dow Chemical Company, Midland, Michigan 48674; tMammalian & Environmental Toxicology Research Laboratory, Health & Environmental Sciences, Dow Chemical U.S.A., Midland, Michigan 48674; and %Harry G. Armstrong Aerospace Medical Research Laboratory, Toxic Hazards Division (AAMRL/TH), Wright Patterson AFB, Ohio 45433 ReceivedJune 29,1987; acceptedMay 22,1988 Physiologically Based Pharmacokinetic Modeling with Methylchloroform: Implications for Interspecies, High Dose/Low Dose, and Dose Route Extrapolations. Reitz, R. H., McDougal, J. N., Himmelstein, M. W., Nolan, R. J., and Schumann, A. M. (1988). Toxicol. Appl. Pharmacol. 95, 185-199. A unified physiologically based pharmacokinetic (PB-PK) model was developed and used to describe the disposition ofmethylchloroform (1,1,1 -trichloroethane, MQ in three different species (rats, mice, and humans) after four different routes ofexposure (inhala tion, intravenous injection, bolus gavage, and drinking water administration). Metabolism of MC followed Michaelis-Menten kinetics in each species. Rux's were calculated from the allo- metric equation: = 0.419 BW0,7, and Km appeared to be identical in each species (5.75 mg equivalents/liter). Once the PB-PK. model had been developed for young adult animals (1-3 months ofage), it was used to study the disposition ofMC in older rats and mice (approximately 18.5 months ofage). Most of the changes in the pharmacokinetic behavior of MC in older rats could be simulated by increasing the size ofthe fat compartment in the PB-PK model from 7 to 18% of body weight. However, the pharmacokinetic behavior in older mice was more complex; increasing the size of the fat compartment in this species from 4 to 18% only accounted for part ofthe observed differences between old and young animals. An appropriate dose surrogate (average area under the liver concentration/time curve) was selected and the PB-PK model was used to make quantitative comparisons between "internal doses" of MC in long term animal studies and "internal doses" associated with human exposures to MC. Values ofthe dose surro gate in humans consuming 2 liters/day ofwater with typical levels of MC contamination (1-10 ppb) were four to six orders of magnitude lower than the dose surrogates in the rodent studies at levels of MC exposure which failed to produce adverse effects on the liver (875-1500 ppm, 6 hr/day, 5 days/week). C 19S8 Academic Press, Inc. Lifetime bioassays in rodents are conducted to assess the potential of various substances to produce chronic toxicity, including carci nogenicity. When these studies are com pleted, they are used to estimate the risk that 1 To whom correspondence should be addressed. similar toxicity will be manifested in human populations exposed to the same agents. However, there are significant differences be tween the conditions in the lifetime rodent bioassay and conditions existing in the hu man environment, and these differences need to be considered in hazard evaluations. Physiologically based pharmacokinetic 0041-0G8X/88 $3.00 Copyright 1968 by Academic Pros. Inc, Ail rights ofreproduction in any form reserved. 186 REITZ ET AL. (PB-PK) models provide a technique for eliminating some of the uncertainty in risk estimations (NAS, 1986; Andersen et al,, 1987). These models calculate the "internal dose" of relevant chemical species in target organs during specific exposure scenarios, and are useful for high dose/low dose extrap olations, dose route extrapolations, and interspecies extrapolations. The utility of the PB-PK models stems from the fact that they realistically describe the sizes oforgan compartments, partitioning of test material between blood and tissues, flow of blood and gases through the organs, and rates of metabolic transformation of the test chemical. The concept of incorporating this type of information into pharmacoki netic models was proposed 50 years ago by Teorell (1937), but computer hardware and software to implement his ideas were not available at that time. Examples of PB-PK models implemented with modem technol ogy may be found in the works of Himmelstein and Lutz (1979), Dedrick (1973), and Fiserova-Bergerova (1975). In order to demonstrate the versatility of this technique with a specific compound, we have developed a PB-PK model for 1,1,1-trichloroethane (methylchloroform, MC). This model was based on the model used byRam sey and Andersen (1984) to describe the dis position of styrene, and details of the mathe matical formulation of such models may be found in that publication. Our objectives in this project were three fold: (1) To demonstrate the ability of the PBPK model to perform high dose/low dose, dose route, and interspecies extrapolations. (2) To determine whether PB-PK models could be used to predict previously reported changes in MC disposition in older rats and mice (Schumann etal., 1982b). (3) To develop a mechanism for making quantitative comparisons between chronic animal inhalation studies and human expo sures. MATERIALS AND METHODS Test materials. A sample of 1,1,1-trichloroethane (MC) was obtained from the Inorganic Chemicals De partment of the Dow Chemical Co. This material was analyzed by gas chromatography (7.5% Oronite NIW plus 2.5% Carbowax 20 M TPA on 80/100 mesh Chromosorb W.H.P.) using a programmed temperature gradi ent of 4'C/min, starting at 70*C and finishing at 130"C. The test material was found to be more than 99.8% MC. Radioactive MC [2-l4C], Lot No. 1119-293, (sp act 2.51 mCi/mmol) was obtained from New England Nuclear (Boston, MA) This material was repurified be fore use by preparative gas chromatography (10% SP 1000 on Chromosorb WAW, isothermal 100'C). The ra diochemical purity after purification was found to be 97.9%. For administration to the animals, a saturated so lution was prepared by stirring [UC]MC in water for 8 hr (excess MC present). The sp act of the MC used in these experiments was 4.53 x 10~3 mCi/mmol and the concen tration was 3.57 mg/ml (2.66 x 10+i dpm/ml). Animals. Male Fischer 344 rats (weight range 240-260 g) were obtained from Charles River Breeding Labora tories (Kingston, NY) and acclimated to the laboratory for at least 7 days before being placed on test The ani mals were housed in stainless-steel cages in rooms de signed to maintain 22'C, with relative humidity con trolled between 40 and 60%, and a 12 hr/day lighting cy cle from 7 am to 7 pm. Animals were fed Certified Purina Chow (Ralston Purina, St Louis, MO) and municipal water ad libitum. Intravenous injections. Male F344 rats were anesthe tized with ketamine/xylene and cannulas were im planted the day before the exposures (minimum 18 hr recovery time). MC was dissolved in heparinized rat plasma at a concentration sufficient to give doses of8.84, 25.6, and 47.0 mg/kg when volumes of 1.0 ml plasma/ kg were injected into an indwelling femoral cannula of rats. Blood samples (0.1 ml) were collected from an in dwellingjugularcannula and immediately transferred to a sealed vial containing 1.0 ml hexane. After 45 min ex traction, aliquots of the hexane layer were analyzed for MC by gas chromatography with an electron capture de tector. Oral gavage (MC in water). Solutions of MC in water were prepared by stirring overnight in a closed glass ves sel. The concentration of MC in the water was deter mined by gas chromatography before administration to each rat, and was in the range of0.8-1.0 mg/ml. MC was administered to rats (mean weight 250 g) in a volume of 4.0 ml to give a dose of 14.2 1.9 mg/kg (n = 6). Blood samples were collected from indwellingjugular cannulas and analyzed for MC by gas chromatography as outlined above. Drinking-water exposure. This exposure was carried out essentially as described by Frantz and Watanabe (1983). Animals were placed in all-glass metabolism cagesc pirev the* tawn satura night, substi; had fr 48 hr Th. wate the si tion i drink matel absor amin< collet water end c (2.70 x 10 Sai sure, homi anah Ph logic. Andt MC bloo. are li parti diffe -95 toth pria Tt take asa livei mo< inte dat neti rofc rep* the. 5 tr 1 tnp (i.e "si; no. SUf D METHODS 1,1,1 -trichloroethane ^Rrganic Chemicals De al Co. This material was phy (7.5% Oronite NIW A on 80/100 mesh Chrommed temperature gradiC and finishing at I30*C. be more than 99.8% MC. rt No. 1119-293, (sp act >ed from New England aferial was repurified beiromatography (10% SP thermal 100'C). The raication was found to be e animals, a saturated sol4C]MC in water for 8 hr of the MC used in these 7i/mmol and the concen10+5 dpra/ml), ts (weight range 240-260 River Breeding Labora'mated to the laboratory placed on test The aniteel cages in rooms de relative humidity conla 12 hr/day lighting cyvere fed Certified Purina i is, MO) and municipal F344 rats were anesthecannulas were imes (minimum 18 hr in heparinized rat ent to give doses of8.84, umes of 1.0 ml plasma/ iiiagfemoral cannula of re collected from an in' mediately transferred to exane. After 45 min ex layer were analyzed for < an electron capture de- dutions of MC in water 4ht in a closed glass vesn the water was deterefore administration to J.8-1,0 mg/ml. MC was at 250 g) in a volume of 9 mg/kg (n = 6). Blood veilingjugular cannulas matography as outlined 3 exposure was carried Frantz and Watanabe n all-glass metabolism METHYLCHLOROFORM PHARMACOKINETICS 187 cages designed to allow separate collection ofurine, feces, expired air, and CO;. Animals were allowed to acclimate to the cages for 32 hr and then water and food were with drawn for 8 hr (from 4 pm to 12 midnight). Food and a saturated solution of [l4C]MC were presented at 12 mid night. At 8 AM the following morning, fresh water was substituted for the solution of [l4C]MC and the animals had free access to food and water until they were killed 48 hr later (56 hr after beginning the exposure). The solution of (l4C]MC was administered with a glass water bottle which contained two stainless-steel balls in the sipper tube in order to eliminate leakage or evapora tion of MC during periods when the animal was not drinking. Air was drawn through the cages at approxi mately 500 ml/min. Activated charcoal traps served to absorb (l4C]MC and ethanoiamine traps (3 parts ethanolamine to 7 parts propylene glycol monomethyl ether) collected [,4C]C02. The activity ofthe [ l4C]MC drinking water (dpm/ml) was determined at the beginning and end of the experiment and remained relatively constant (2.70 0.23 X 10+s dpm/ml at the start; 2.62 0.24 X 10+J dpm/ml at the end; n = 4 in both cases). Samples ofurine and feces were collected during expo sure, and samples of liver, kidney, fat, skin, and carcass homogenate were collected at the end of exposure for analysis ofradioactivity. Physiological modeling. A four-compartment physio logical model similar to that developed by Ramsey and Andersen (1984) was used to describe the behavior of MC in rats, mice, and humans. Tissue volumes and blood and airflow rates employed in these simulations are listed in Table I. The three species used in these stud ies differ widely in the percentage of fat in their carcass (Mouse = 4%, Rat = 7%, Human = 23%). Consequently, the percentage ofcardiac output directed to the fat com partments was adjusted between species to reflect the different amounts offat: Mouse - 2%, Rat = 5%, Human = 9%. To maintain mass balance in the model, the flows to the richly perfused compartment were adjusted appro priately (Table 1). To adapt this model for drinking-water exposures, up take ofMC from the gastrointestinal tract was simulated as a zero-order process depositing MC directly into the liver compartment This mathematical form of the model assumes rapid absorption of MC from the gastro intestinal tract. This assumption is consistent with the data ofWithey et al. (1982) who studied the pharmacoki netics of four similar solvents (methylene chloride, chlo roform, 1,2-dichloroethane, and trichloroethylene) and reported that following gavage with aqueous solutions of these materials, peak blood levels were reached in about 5 min. This form of the model also assumes that zero-order input is a good approximation of the drinking process (i.e., that the animals consume a large number of small "sips" ofthe dosing solution throughout the exposure pe riod). Pharmacokinetic analyses conducted by others suggest that this is a reasonable assumption (NAS, 1986). Simulation of bolus gavage results was conducted as described by Ramsey and Andersen (1984). Absorption from the gastrointestinal tract was assumed to be a firstorder process, with a rate constant of 1.25 hr'1, and all material absorbed from the gastrointestinal tract was de posited directly into the liver compartment. Partition coefficients. Tissue/air partition coefficients were determined for rat blood, liver, fat, and muscle us ing the vial equilibration technique of Sato and Nakajima (1979). Measurements were made at Wright Patter son AFB and were generously supplied by M. L. Gargas (personal communication). Tissue/btood partition co efficients for rat fat, muscle, and liver were then calcu lated by dividing the measured tissue/air partition co efficients by the blood/air coefficient for rat blood. Tis sue/blood partition coefficients for the slowly perfused and rapidly perfused groups oftissues were assumed to be equal to the tissue/blood partition coefficients for muscle and liver respectively. Blood/air partition coefficients for human and mouse blood were also determined by M. L. Gargas at Wright Patterson AFB (personal communication). Tissue/blood partition coefficients for human and mouse tissues were calculated by dividing the measured or estimated rat tissue/air partition coefficient by the blood/air partition coefficient for humans and mice respectively. All parti tion coefficients used in these simulations are listed in Table 1. Metabolic rale constants (rat). Schumann et al. (1982a) exposed rats to 150 and 1500 ppm of [,4C]MC for 6 hr. Samples ofexpired air and excreta were collected for 66 hr postexposure in an apparatus which separated [,4C]MC from radioactive metabolites. After 66 hr the rats were killed and the levels of radioactivity remaining in the carcass (assumed to be all metabolites) were deter mined by combustion analysis. This procedure provides a quantitative measure ofthe total amounts of MC me tabolites formed from [14C]MC and eliminated postex posure. However, since samples of urine and expired air were not collected during the inhalation exposure, any radioactive metabolites eliminated during this period would have been lost This would lead to an underesti mation of the overall rate of metabolism. In order to compensate for this underestimation, data from the drinking water studies were used to derive a correction factor. In the drinking-water study, watercontaining [l4C]MC was provided to the animals for 8 hr and samples were collected during this period and for an additional 48 hr postexposure. Animals were observed to drink regularly throughout this period. Forty-one percent of the total metabolites were collected during the exposure period, and 59% of the total metabolites were collected after ex posure. Since the experimental conditions were similar to those used in the inhalation exposures, it was assumed that this procedure would give a reasonable approxima tion of the percentage of total metabolites recovered 188 REITZ ET AL. postexposure in the inhalation experiments. Conse quently, the postexposure levels of metabolism reported by Schumann et al. (1982a) in inhalation experiments were divided by 0.S9 to give an estimate of total metabo lism in the inhalation studies. At this point, the PB-PK model was fully defined ex cept for the metabolic rate constants and Km. Thus it was possible to estimate and Km by a computer ized least-squares optimization procedure which varied the rate constants until the model correctly predicted the corrected amounts of metabolites formed during the ex posures to 150 and 1500 ppm MC. Final values for the enzymatic rate constants were V^C = 0.419 0.095; K,, = 5.75 2.49 (mean and standard deviation of esti mates). Details ofthis type ofcomputerized optimization have been reported elsewhere (Agin and Blau, 1982). Computer simulation. Simultaneous differential and algebraic equations describing the movement of MC through the body were formulated as a computer pro gram. Simulations were conducted with the SimuSolv software package which contains routines for numerical integration, graphics display, and derivation ofmodel pa rameters from experimental data. SimuSolv was devel oped by Agin and Blau (1982) and is commercially avail able from Mitchell & Gauthier Associates.2 * RESULTS Rat Simulations Inhalation simulations. After incorpora tion of the appropriate physiological and physical chemical parameters (Table 1), the PB-PK model was used to predict the time course of the venous blood concentrations of MC in rats after exposure to 150 or 1500 ppm MC. Experimental data previously collected by Schumann et al. (1982a) were used to check the predictions, and the results are shown in Fig, 1. The data showed the best agreement with model predictions for blood level measurements made postexposure. Blood levels measured during the exposure were somewhat lower than model predictions at 4, 5, and 6 hr, but the predictions were still within a factor of 2 of the experimental data (Fig. 1). 2 Mitchell & Gauthier Associates, 73 Junction Square Drive, Concord, MA 01742. TABLE 1 Parameters Used in the Physiologically Based Pharmacokinetic Model for MethylchloroFORM* Human Rat Mouse Weights Body wt (kg) Liver Rapidly perfused Slowly perfused Fat Rows (liters/hr) Alveolar vent. Cardiac output 83.0 3.1% 3.7% 61.1% 23.1% 348.0 348.0 0.215 4,0% 5.0% 75.0% 7.0% 0.029 4.0% 5.0% 78.0% 4.0% 5.11 1.26 5.11 1.26 Percentage ofcardiac output Liver Rapidly perfused Slowly perfused Fat Partition coefficients Blood/air Liver/air Rapidly perfused/air Slowly perfused/air Fat/air Biochemical constants V__ C (allometric) Kh (mg/liter) K,(hr-') 24.0 49.0 18.0 9.0 2.53 8.6 8.6 3.15 263. 0.419 5.75 -- 24.0 53.0 18.0 5.0 5.76 8.6 8.6 3.15 263. 0.419 5.75 1.25 24.0 56.0 18.0 2.0 10.8 8.6 8.6 3.15 263. 0.419 5.75 -- 'Metabolic parameters for the rat (V^C, K,,) were obtained from the data of Schumann et at. (1982a) by computer optimization. KmC is an allometric measure of the maximum velocity of metabolism such that the maximum enzyme rate (Pam) may be calculated for any size animal according to the equation = VmC X (body wt)*-7. Several other types of data were available from the studies of Schumann et al, (1982a) to check the accuracy of the PB-PK model. These were (1) the total body burden of MC and metabolites after 6 hr ofexposure, (2) the concentration of radioactivity in fat tissue at the end of the exposure, and (3) the concen tration of radioactivity in liver tissue at the end of the exposure. Comparisons between the model predic tions and these additional data are summa- SL 036331J OLOGICALLY BASED Methylchloro- in Rat Mouse 0.215 0.029 *> 4.0% 4.0% % 5.0% 5.0% 75.0% 78.0% fc 7.0% 4.0% 5.11 1.26 5.11 1.26 ntage ofcardiac output 24.0 53.0 18.0 5.0 24.0 56.0 18.0 2.0 9 0.419 0.419 5.75 5.75 rat (VauiC, Km) were nann et al. (1982a) by an allometric measure tabolism such that the iy be calculated for any luation ^ = V^C ata were available lann et al. (1982a) -he PB-PK model. >dy burden of MC )fexposure, (2) the vity in fat tissue at nd (3) the concenliver tissue at the he model predicdata are summa- METHYLCHLOROFORM PHARMACOKINETICS 189 Fic 1. Blood levels of MC in rats during and following a 6-hr inhalation exposure to 150 ppm (closed circles) or 1500 ppm (open circles) of MC. Values predicted by computer simulations are shown as a solid line(s). rized in Table 2. Overall, the ratio of pre dicted to actual data for these two exposures ranged from a low of 0.72 to a maximum of 1.92 and the mean ratio was 1.28 0.48 (standard deviation; Table 2). Intravenous administration. The PB-PK model was then used to attempt a dose route extrapolation: inhalation exposure to intra venous injection. Venous blood concentra tions were measured in groups ofanimals (six per group) injected with MC dissolved in hep arinized plasma. Intravenous injection was simulated in the model as a rapid infusion into pooled venous blood (about 1 min in duration). Three doses were studied: 8.8, 26, and 47 mg/kg. The only changes made to the PB-PK model for this simulation were in the equations describing entry ofMC into the body. Experimental data and model pre dictions for 4 hr postinjection are shown in Fig. 2. Bolus gavage (MC in water). A third route of administration (bolus gavage) was investi gated. Six rats were administered a dose of 14.2 mg/kg MC dissolved in water, and serial blood samples were taken for analysis at vari ous times afterwards. The equations in the PB-PK model were modified to reflect the new route of exposure. Uptake of MC was simulated as a first-order process with a rate constant (Ka) equal to 1.25 hr-1. The model predictions and experimental data for this route are shown in Fig. 3. Drinking-water exposure. A variant of the bolus gavage route was consumption of MC in drinking water over an extended period of time. Animals were presented with water containing radioactive MC for 8 hr and sam ples of urine, expired air, and selected tissues were analyzed for radioactivity at various times thereafter. The average water con sumption of the rats (mean wt = 250 g) dur ing the 8 hr in which they had access to the treated water was 8.1 3.8 ml, corresponding to an average dose of 116 mg/kg. Overall recovery of radioactivity in these experi ments (based on water consumption) was 95.2 4.33% (standard deviation). Entry of MC into the animal was simulated as a zeroorder process depositing MC in the liver com partment at a constant rate for 8 hr. Other than the changes to the input equations, no parameters were changed in the PB-PK model for drinking-water simulation. Simu lated rates of MC elimination in exhaled air and experimental data (radioactivity recov ered from charcoal traps) are shown in Fig. 4. In addition, the use of [l4C]MC made it possible to predict the extent of MC metabo lism for this route of metabolism. The model predicted that 5.49 /tniol of MC (about 2% the ingested MC) would be metabolized, 8.19 Atmol (about 3% ofthe ingested MC) was actually recovered in urine and C02 (Ta ble 2). Mouse Simulations Inhalation simulations. The PB-PK model was then used for interspecies extrapolation by predicting the pharmacokinetic behavior of MC in male B6C3F1 mice after inhalation exposure. Model predictions were compared to experimental data gathered by Schumann et al. (1982a). Other than adjusting the ap propriate physiological and biochemical pa rameters, no changes were made in the PBPK model for simulation of the mouse expo sures. 190 REITZ ET AL. TABLE 2 Comparison of Predicted and Observed Values for Selected Parameters from Inhalation Exposures and Drinking Water Exposures in Young Rats and Young Mice (2-3 months of age) and Human Volunteers' Observed Predicted Ratio (pred/obs) End exposure blood level (mg/liter) Body burden at 6 hr (/imol) Cone in fat (/imol/liter) Cone in liver (/imol/liter) Young rat-- 150 ppm 2.64 33.0 724.0 68.2 4.37 25.5 1264.0 48.9 1.66 0.77 1.75 0.72 End exposure blood level (mg/liter) Body burden at 6 hr (/imol) Cone in fat (/imol/liter) Cone in liver (/imol/liter) Young rat--1500 ppm 23.1 263 8400 503 44.4 237 12800 500 1.92 0.90 1.52 0.99 Young rat--drinking water (116 mg/kg) Amount metabolized (/imol) 8.19 5.49 0.67 End exposure blood level (mg/liter) Body burden at 6 hr (/imol) Amount metabolized (/imol) Cone in fat (/imol/liter) Cone in liver (/imol/liter) Young mouse-- 150 ppm 9.27 4.97 1.10 1330 75.7 8.54 3.30 1.02 1570 50.6 0.92 0.66 0.93 1.19 0.67 Young mouse--1500 ppm End exposure blood level (mg/liter) Body burden at 6 hr (/imol) Amount metabolized (/imol) Cone in fat (/imol/liter) Cone in liver (/imol/liter) 111.0 39.5 2.01 16200 631 88.1 25.8 1.90 16100 525 0.79 0.65 0.94 0.99 0.83 Amount metab. 35 ppm (/imol) Amount metab. 350 ppm (/imol) Human 35 and 350 ppm 32.4 246 32.2 236 * Inhalation data are from Schumann etal.(\982a) and Nolan et al. (1984). 0.99 0.96 The predicted and observed venous blood concentrations (collected at the orbital sinus) in mice exposed to either 150 or 1500 ppm of MC are shown in Fig. 5. The PB-PK model predicted that MC would be eliminated from the mouse much more rapidly than the rat, and this is consistent with the observed data (Figs. 1, 5). This is also consistent with the observation of Schumann that elimination half-lives for MC in mice were 5- to 10-fold less than the corresponding half-lives in rats (Schumann et al. 1982a). Several additional sources of experimental data were available for assessing the consis- ioooo ooo o to 00! Fig. 2. nous inje (closed cii rat plasm concentre while ma plasma/k; simulatio tencyo! and pr body bt tration. and tht ered an predict 01 Fig. . with as 1.9 r proxim by com sorptio afirst-c lation Expo^Pg;e) and Human Ratio (pred/obs) 1.66 0.77 1.75 0.72 1.92 0.90 1.52 0.99 0.67 0.92 0.66 0.93 1.19 0.67 0.79 0.65 0.94 0.99 0.83 0.99 0.96 iistent with the tat elimination re 5- to 10-fold lalf-lives in rats >f experimental sing the consis METHYLCHLOROFORM PHARMACOKINETICS 191 Time (hr) Fig. 2. Blood levels of MC in rats following intrave nous injection of 8.8 mg/kg (open circles), 26 mg/kg (closed circles), or 47 mg/kg (crosses) of MC dissolved in rat plasma. MC was dissolved in heparinized plasma at concentrations calculated to give the indicated doses while maintaining a constant vehicle volume of I ml plasma/kg of body wt. Values predicted by computer simulations are shown as a solid line($). Time (hr) Fig. 4. Rate of elimination of l4C[MC] in exhaled air (mg equivalents of MC/hr) during and following ad libi tum exposure ofrats to a solution of(l4C]MC in drinking water. The average water consumption ofthe rats (mean wt = 250 g) during the 8 hr in which they had access to the treated water was 8.1 3.8 ml, corresponding to an average dose of 116 mg/kg. Values predicted by com puter simulations are shown as a solid line(s). tency ofthe PB-PK model for mice. Observed and predicted values of the end exposure body burden, the end exposure liver concen tration, the end exposure fat concentration, and the total amount of metabolites recov ered are summarized in Table 2. The ratio of predicted to actual data ranged from a low of 0.65 to a high of 1.19 and the mean ratio was 0.86 0.16 (standard deviation; Table 2). Human Exposures The PB-PK model was used to predict the disposition ofinhaled MC in humans. Model predictions were compared to data gathered by Nolan et al. (1984) at 35 and 350 ppm MC. Experimentally observed and simulated concentrations of MC in venous blood are shown in Fig. 6a. Concentrations of MC into blood was somewhat lower than predicted Fig. 3. Blood levels ofMC in rats following oral gavage with a solution ofMC in water. The dose ofMC was 14.2 1.9 mg/kg (standard deviation), administered in ap proximately 4.0 ml water/kg body wt. Values predicted by computer simulations are shown as a solid line(s). Ab sorption from the gastrointestinal tract was simulated as a first-order process with a rate constant of 1.25 hr"1. Time (hr) Fig. 5. Blood levels of MC in mice during and follow ing a 6-hr inhalation exposure to 150 ppm (closed circles) or 1500 ppm (open circles) of MC. Values predicted by computer simulations are shown as a solid line(s). SL 036334 192 REITZ ET AL. Fig, 6a. Concentration of MC in exhaled air (mg/liter) during and following a 6-hr inhalation exposure of hu man volunteers to 35 ppm (dosed circles) or 350 ppm (open circles) ofMC. Values predicted by computer sim ulations arc shown as a solid line(s). during exposure (insert) but was well de scribed for the post exposure period (main figure). Predicted and observed concentrations of MC in expired air are shown in Fig. 6b. Con centrations of MC in expired air were well simulated by the model during exposure (in sert). The model predicted a slightly more rapid decline in the concentration of MC in expired air than was observed during the pe riod of 10-30 hr, but the final slow elimina tion phase (30-240 hr) was well simulated by the model (main figure). Predicted and ob served values ofexpired air and venous blood concentrations were both proportional to ex posure concentration throughout the range of 35-350 ppm MC. The PB-PK model predicted that 4.30 mg equivalents of MC would be metabolized during the 240 hr following a 6-hr exposure to 35 ppm MC, and 4,32 mg equivalents of MC metabolites were actually recovered by Nolan et al. (1984). Similarly, the model pre dicted that 31.5 mg equivalents of MC tabolites would be formed during the 240 hr following a 6-hr exposure to 350 ppm MC, and 32.9 mg were actually recovered (Ta ble 2). Older Animals Old rats, inhalation simulations. Schu mann et al. (1982b) also studied the pharma cokinetics of MC in control and exposed (1500 ppm, 6 hr/day, 5 days/week) male rats obtained from a study of the chronic inhala tion toxicity of MC. The age of these animals was approximately 18.5 months, and the av erage body weight was 481 g. They reported that the disposition of a single dose of [l4C]MC in the treated and control animals was nearly identical (i.e., that chronic preex posure to unlabeled MC had not altered phar macokinetic behavior through induction of enzymes, alteration of renal function, etc.). However, the disposition of MC in old ani mals differed significantly from the disposi tion previously observed in younger animals (Schumann etal., 1982a). The PB-PK model was used to predict the rate of elimination of MC in exhaled air in untreated (control) 18.5-month-old male rats following a 6-hr exposure to 1500 ppm MC. Experimental data and the initial simulation are shown in Fig. 7a. The PB-PK model pre dicted a much more rapid decline in the rate of MC exhalation when the physiological pa rameters chosen for young rats were used (heavy line. Fig. 7a). Lutz et al. (1977) improved the accuracy of their simulation of polybrominated biphenyl compounds in rats by increasing the relative FIG. 6b. Concentration of MC in venous blood (mg/ liter) during and following a 6-hr inhalation exposure of human volunteers to 35 ppm (closed circles) or 350 ppm (open circles) ofMC. Values predicted by computer sim ulations are shown as a solid Iine(s). Fig. 7a. Rate (mg equivalent months) rats tc experimental d. Simulated valu> body weight (i.e young rats) arc light solid line compartment ei proportion o compartmen a second co mine whethe give ^feter 18.5-^^th-< lowed to var\ ment (VF) fir* ing simulatic the fraction slowly perfu; constant perc tissue (91% o improved fits tained after line. Fig. 7a). PB-PK mode exhaled air d from 7% of weight. The model was then uset of experimen ies of Schuir eluded end ex sure concent SL 036335 rntrol and exposed ^^s/week) male rats chronic inhala age of these animals months, and the av31 g. They reported a single dose of nd control animals that chronic preexiad not altered phar. rough induction of enal function, etc.)1 of MC in old aniy from the disposiin younger animals s used to predict the C in exhaled air in month-old male rats e to 1500 ppm MC. le initial simulation ; PB-PK model pred decline in the rate he physiological paang rats were used jved the accuracy of ^Mminated biphenyl ^Pasing the relative 1 160 ZOO 240 (hr) -C in venous blood (mg/ ir inhalation exposure of losed circles) or 350 ppm ^dieted by computer sim>e(s). METHYLCHLOROFORM PHARMACOKINETICS 193 Q 1 i ' 1 " i i* i ............ . ......... 0 6 12 ta 24 30 36 Time (hr) 0 001 ` * ........................ * --* 0 6 12 16 24 30 36 Time (hr) Fig. 7a. Rate ofelimination of [l4C]MC in exhaled air (mg equivalents of MC/hr) following exposure ofold (18 months) rats to 1500 ppm of [l4C]MC for 6 hr. Actual experimental data are shown as open circles on the plot Simulated values with a fat compartment equal to 7% of body weight (i.e., simulation based on the parameters for young rats) are shown by the heavy solid line, while the light solid line shows the model predictions with a fat compartment equal to 18% ofbody weight. Fig. 7b. Rate ofelimination of[ l4C]MC in exhaled air (mg equivalents of MC/hr) following exposure ofold (18 months) mice to 1500 ppm of [,4C]MC for 6 hr. Actual experimental data ate shown as open circles on the plot. Simulated values with a fat compartment equal to 4% of body weight (i.e., simulation based on the parameters for young mice) are shown by the heavy solid line, while the light solid line shows the model predictions with a fat compartment equal to 18% ofbody weight proportion of body weight allocated to the fat compartment. Consequently, we conducted a second computer optimization to deter mine whether a similar modification would give a better simulation of the data from the 18.5-month-old rats. The computer was al lowed to vary the volume of the fat compart ment (VF) from 0 to 30% ofbody weight dur ing simulations. Corresponding decreases in the fraction of body weight assigned to the slowly perfused compartment maintained a constant percentage ofthe carcass as perfused tissue (91% ofbody weight perfused). Greatly improved fits to the exhaled air data were ob tained after the second optimization (light line, Fig. 7a). This optimization indicated the PB-PK model was most consistent with the exhaled air data when the VF was increased from 7% of body weight to 18% of body weight. The model with the larger fat compartment was then used to predict several other types of experimental data available from the stud ies of Schumann et al. (1982b). These in cluded end exposure body burden, end expo sure concentration of MC-derived radioac tivity in the liver, end exposure concentration ofMC-derived radioactivity in the fat, and to tal amount of radioactive metabolites col lected in 66 hr subsequent to the 6-hr expo sure. The comparisons between the PB-PK model prediction and the actual data are summarized in Table 3. The predictions of the PB-PK model for old rats with the original fat compartment (7% of body weight) are summarized in col umn 2 ofTable 3. The model underpredicted the end exposure body burden and amount metabolized, but overpredicted the concen tration of MC in fat tissue. After increasing the size of the fat compartment to 18% of body weight, the PB-PK model predicted lower concentrations of MC in fat tissue, and increased body burdens and amounts metab olized, giving a better simulation of the rat data. Old mice inhalation simulations. Schu mann et al. (1982b) also collected data in old male B6C3F1 mice (18.5 months of age) ob tained from a study of the chronic inhalation toxicity of MC in mice. As was observed with the rats, the pharmacokinetics of MC in old SL 036336 194 REITZ ET AL. TABLE 3 Comparison of Observed and Predicted Values for Selected Parameters from Inhalation Exposures in Old Rats and Old Mice (18.5 months of age)0 Actual Before reoptimization After reoptimization Body burden at 6 hr (jimol) (Ratio) Amount metabolized (//mol) (Ratio) Cone in fat (pmol/liter) (Ratio) Cone in liver (Mmol/liter) (Ratio) Old rat--1500 ppm 744 31.8 5,685 606 490 (0.66) 17.5 (0.55) 11,700 (2.06) 494 (0.82) 638 (0.86) 22.1 (0.69) 6200 (1.09) 454 (0.75) Body burden at 6 hr (//mol) (Ratio) Amount metabolized (//mol) (Ratio) Cone in fat (//mol/liter) (Ratio) Cone in liver (Mmol/liter) (Ratio) " Experimental data are from Schum " optimization are listed in Table 1, exi Alter optimization, the size ofthe fat c Old mouse--1500 ppm 149 13.6 10,600 1,790 35.1 (0.24) 2.61 (0.19) 15,900 (1.50) 532 (0.30) 71.4 (0.48) 4.11 (0.30) 8790 (0.83) 491 (0.27) - al. (1982b). Physiological parameters for rats and mice before computer body weight which was 481 g for the rats and 39.8 g for the old mice. ,ment was 18.6 and 18.1% ofbody weight in rats and mice, respectively. unexposed (control) mice differed signifi cantly from the pharmacokinetics of MC in young control mice. When the PB-PK model utilized to de scribe MC disposition in young mice was used to predict the rate of elimination of MC in exhaled air, a poor description of the ex perimental data was obtained (heavy line. Fig. 7b). However, when the size of the fat compartment was increased from 4% ofbody weight to 18% of body weight, a much better description of the experimental data was ob tained (light line. Fig. 7b). When the model with 4% fat was used to simulate additional parameters measured in older mice, it underestimated the body bur den, the amount metabolized, and the con centration of radioactivity in liver tissue and overestimated the concentration of radioac tivity in fat (Table 3). When the size of the fat compartment was increased to 18% of body weight, the description of the experimental data was improved, but the model still underpredicted the amount metabolized and the concentration of radioactivity in liver tissue. DISCUSSION Versatility ofthe PB-PK Model The first objective of these studies was to demonstrate the extent to which incorpora tion of physiological and biochemical princi ples into pharmacokinetic modeling would allow extrapolation between different routes of administration and different species. Ac cordingly, we have not conducted extensive "curve fitting" procedures to try to improve i Inhalation After reoptimization 638 (0.86) 22.1 (0.69) 6200 (1.09) 454 (0.75) 71.4 (0.48) 4.11 (0.30) 8790 (0.83) 491 (0.27) 1 mice before computer 19.8 g for the old mice, and mice, respectively. ^Fto 18% of body ' the experimental : model still undertabolized and the rity in liver tissue. ON rodel iese studies was to which incorpora* iochemical princi- ; modeling would en different routes rerent species. Acnducted extensive to try to improve METHYLCHLOROFORM PHARMACOKINETICS 195 the appearance of the data presented here, and it is obvious that the extrapolations were not perfect in every case. However, since we are proposing that these techniques may be useful in the area of risk assessment, it is im portant to keep in mind the level of precision which currently exists in this process. Cancer risk estimations produced from the same sets of data often differ by several orders of magnitude, depending upon the assump tions used about the shape of the dose-re sponse curve at low concentrations. Further more, it is not uncommon for bioassays of the same material in the same species to differ by as much as an order of magnitude in their end points (tumor incidences). Consequently, we felt that if our integrated PB-PK model could predict relevant pharmacokinetic parameters for MC in different species and dose routes within a factor of 2 or 3, we would have dem onstrated its potential for increasing the accu racy of current risk estimation procedures. The construction ofthe PB-PK model used here was relatively straightforward. Physio logical parameters for the various species have been published in the scientific litera ture (Davis and Mapleson, 1981; Caster et ai, 1956; ICRP, 1975), and partition coefficients were measured according to published proce dures (Sato and Nakajima, 1979). The most difficult task was the estimation of the rate constants for in vivo metabolism, MC is not extensively metabolized in any of the species studied (Schumann et a!., 1982a,b; Nolan et al., 1984). Consequently the gas uptake technique used by Gargas et al. (1988) for estimation of in vivo metabo lism with other compounds was not suitable for this material, since uptake depends upon metabolism once tissue loading has occurred. Furthermore, a variety of metabolites are formed from biotransformation of MC (trichloroethanol, the glucuronide oftrichloroethanol, and trichloroacetic acid). The pro duction of multiple metabolites, each of which is present at very low levels, makes it difficult to estimate total metabolism by chemical analysis. However, radioactive metabolites of MC are easily separated from [14C]MC and may be readily quantified. Consequently, we have used available data on the total levels of ra dioactive metabolites obtained in balance studies with rats exposed to [14C]MC to esti mate the rates of metabolism in this species. Levels of metabolites produced during expo sure and postexposure were calculated with the PB-PK model, which contains the struc ture necessary to deal with the complexities of postexposure metabolism. We then em ployed allometric scaling procedures to esti mate metabolic rate constants in the other species. Since the metabolic rate constants were obtained from a limited data set (two ex periments in rats with four animals/experi ment), there is obviously room for error in these parameters. The magnitude of the pos sible error may be roughly estimated from the standard deviations for the estimates of LmaxCand Km, which were 23 and 43% ofthe mean, respectively. Nevertheless, it is encour aging that once the metabolic rate constants were obtained in this manner, they were able to accurately predict metabolism in two other species exposed to MC vapor (mice and hu mans) as well as in rats exposed to MC in drinking water (Table 2). It must be emphasized that since metabo lism plays a minor role in the elimination of MC in the various species, the behavior of MC itself (as distinct from MC metabolites) is almost entirely dependent upon the solu bility in various tissues (partition coefficients) and the physiology of the various species. Small errors in the values chosen for the met abolic rate constants will have little effect on the overall disposition of MC. Schumann et al. (1982a) utilized a conven tional, data-based pharmacokinetic model to describe the time course ofMC in rats. In this approach, the venous blood time/concentra tion curve for rats was actually described bet ter by a simple two-compartment open model with elimination from the central compartment (Schumann et al., 1982a) than the present PB-PK modeL However, when SL 036338 196 REITZ ET AL. we attempted to use this model to describe the time course of MC in the venous blood of mice, a very poor fit was obtained (data not shown), pointing out the difficulties of using conventional models for extrapolating to hu mans. In contrast, once a PB-PK model had been developed for rats, it immediately pro vided a good description of the blood level data obtained in mice (Fig. 5) as well as hu mans (Fig. 6a). Furthermore, the development ofa PB-PK model allowed us to predict other types of data of potential toxicological significance. For example, the concentration of MC in liver tissue at the end of inhalation exposures to MC was well described in both rats and mice (Table 2). Since the liver has been iden tified as the target organ in both rats and mice during long-term animal bioassays (Rampy et al., 1978; Quast et al., 1984), this finding has obvious significance. Another advantage of PB-PK models is their ability to provide quantitative descrip tions of material administered by a variety of dose routes. In this particular case, the phar macokinetics of MC in rats after inhalation (Fig. 1), intravenous injection (Fig. 2), bolus gavage (Fig. 3), and drinking-water admin istration (Fig. 4) could be described with a single, integrated model. Conventional databased pharmacokinetic models lack the ca pacity for this type of route to route extrapo lation. tissue are 20-100 times higher than in other tissues. Consequently, the increased size of the fat compartment in older, sedentary ani mals would be expected to increase the amount of MC taken up by older animals ex posed to a given concentration of MC, and the elimination of MC from these older ani mals should occur more slowly. These predic tions of the PB-PK have been experimentally verified in the data of Schumann et al. for both rats and mice (Figs. 7a, 7b; Tables 2, 3). Most ofthe changes in the pharmacokinet ics of MC in older rats were correctly pre dicted when the size of the body fat compart ment was increased (Table 3). However, the situation appears to be somewhat more com plex with B6C3F1 mice. The increased halflife of MC and lower concentration of radio activity in the fat compartment of older mice were correctly predicted by a PB-PK model with a larger fat compartment. However, the predicted body burden and amount of MC metabolized in these mice were still much lower than actually observed by Schumann et al. (1982b) (Table 3). It appears that addi tional factors, such as increased metabolic ca pacity, may have to be considered in evaluat ing the kinetics of MC in older mice. The na ture ofthese factors remains to be elucidated. Relating Animals Studies to Human Expo sures Pharmacokinetics in Older Animals A variety of physiological and biochemical changes may occur in test animals during the course of a lifetime bioassay, where animals are started on test a few weeks after weaning and continue on test until nearly the end of their natural lives. Development of a PB-PK model offers a mechanism whereby the con sequences of these changes may be quantita tively evaluated. For example, the PB-PK developed for MC indicates that the concentrations of MC in fat MC has been studied in two long-term in halation bioassays at the Toxicology Re search Laboratory of the Dow Chemical Co. Animals were exposed to MC for 6 hr/day, 5 days/week for up to 2 years. MC exposure was not associated with increases in the inci dences of either benign or malignant tumors in these studies. Reversible microscopic alter ations in liver tissue were noted in both rats and mice, but cytotoxicity and necrosis were not seen. The no observed adverse effect lev els (NOAEL) for the cellular changes estab lished in these studies were 875 ppm in rats and 1500 ppm in mice (Rampy et al., 1978; ^Acom ef Value AMOU' Rat Mou$< Huma "Tt mice i in uni huma Qua port pose weel Si |inin wou wou ofN (i.e. tern chei seqi con tim> gate sun cus: ceei NA 5 MC tioi che ma ofi ligher than in other increased size of :r, sedentary anito increase the oy older animals ex cretion of MC, and am these older aniowly. These predic>een experimentally -humann et ai for 7a, 7b; Tables 2, 3). he pharmacokinetivere correctly pre; body fat comparte 3). However, the mewhat more comrhe increased halfentration of radioment of older mice >y a PB-PK model lent. However, the id amount of MC e were still much jd by Schumann et appears that addiased metabolic caisidered in evaluatmice. The na^^0 be elucidated. to Human Expo- two long-term in; Toxicology Re tow Chemical Co. AC for 6 hr/day, 5 irs. MC exposure creases in the incimalignant tumors microscopic alternoted in both rats and necrosis were adverse effect levar changes estab- 875 ppm in rats impy et al,, 1978; METHYLCHLOROFORM PHARMACOKINETICS 197 TABLE4 Comparison of the Lifetime Average Concentration of MC in the Liver (ACL, Dose Surrogate for MC Effects on Liver) for Rats and Mice at the No Observed Adverse Effect Levels (NOAEL) with Values of the Same Dose Surrogate in Humans Consuming 2 liters/day of Water Containing Small amounts of MC" Rat Mouse Human Cone (ppm) 875 1500 0.001 0.003 0.010 0.030 0.100 0.300 Route Inhalation Inhalation Water Water Water Water Water Water Average ACL (Mmol/liter) 28 95 3.4 x 10"5 1.0 x 10"* 3.4 X 10~* 1.0 x 10'3 3.4 X 10`3 1.0 X 10"J Safety factor relative To mouse __ -- 2.8 x 104* 9.5 x 1043 2.8 X 1043 9.5 x 1044 2.8 X 1044 9.5 x 1043 To rat _ -- 8.2 x 1043 2.8 x I045 8.2 X 1044 2.8 x 1044 8.2 X 1043 2.8 x 1043 * The NOAEL for the rat was 875 ppm, 6 hr/day, 5 days/week for I year (Rampy etal., 1978) and the NOAEL for mice exposed 6 hr/day, 5 days/week for 2 years was 1500 ppm (Quasi et at., 1984). All dose surrogates are reported in units ofumol/liter. "Safety factors" are calculated by dividing the calculated animal dose surrogate by the predicted human dose surrogate. Quast et ai, 1984). McNutt et ai (1975) re ported similar effects in the liver of mice ex posed to 250 ppm continuously for several weeks. Since the liver is the target organ in both animal species, it is assumed that the liver would be the site where any adverse effects would occur in humans. Because the effects of MC on liver tissue were mild and reversible (i.e., without necrosis), it is not possible to de termine whether they were caused by parent chemical or more reactive metabolites. Con sequently we have chosen to use the average concentration of MC in the liver over the life time of the animal (ACL) as a "dose surro gate" for MC. Rationale for selection of dose surrogates with PB-PK models has been dis cussed in more detail by Andersen in the pro ceedings of a workshop conducted at the NAS (1987). Selection of this type of dose surrogate for MC is probably a health protective assump tion. Reactive metabolites, rather than parent chemical, are responsible for the toxicity of many other halogenated solvents, and levels of the enzyme which produces metabolites of MC appear to be lower in humans than in the small rodent species. For example, we can use the allometric equation to calculate that the concentration of MC-metabolizing enzyme (units of enzyme per liter of tissue) is about 13-fold lower in human liver than it is in mouse liver. Consequently, if we had chosen to use production of reactive metabolites as the dose surrogate rather than ACL, the esti mated risk to humans would be considerably lower than with the present procedure. Once the dose surrogate was chosen, the PB-PK model was used to calculate the ACL for rats and mice at the NOAEL in the long term inhalation studies at Dow Chemical. The ACL for rats was about 55.2 ^mol/liter during the 1-year exposure period, or about 27.6 //mol/liter averaged over the 2-year life span. The corresponding value for B6C3F1 mice was 95.1 ^mol/liter (Table 4). Small quantities of MC (typically 1-10 ppb, highest observed 300 ppb) have been de tected in some finished drinking-water sup plies. However, there are no long-term stud ies in animals in which MChas been admin istered by this route, and even if there were, 198 REITZ ET AL. the limited solubility of MC would limit the dose that could be given to the animals. Con sequently, there is a need for a mathematical procedure to quantitatively relate the "inter nal dose" in humans drinking such water to the "internal dose" in animal inhalation studies. This was readily accomplished with the PB-PK model developed in these studies. ACLs associated with human consump tion of water containing the indicated con centrations of MC are summarized in Table 4. It is obvious from this table that the ACLs in humans consuming this water are much lower than the ACLs in animals during the chronic inhalation studies. For example, the ACL in humans exposed to 10 ppb of MC in drinking water are 81,000-fold lower than the ACL at the NOAEL in the rat, and 270,000fold lower than the ACL in the mouse (Table 4, columns 4 and 5). Simulations with the PB-PK model re vealed that "steady-state" levels of MC in the human liver are achieved within a relatively short period. The ACL after 100 days of drinking water containing 10 ppb MC was only 2.7 times higher than the ACL after a single day of consuming such water. This in dicates that the risk to humans will not in crease disproportionately after long periods of consumption, as might be the case with compounds such as lead or mercury. There are bound to be many uncertainties in human risk estimations derived from ani mal studies. Although they cannot address all of the uncertainties inherent in this process, PB-PK models offer promise for improving the basis for high dose/low dose, interspecies, and dose route extrapolations. It should be noted, however, that PB-PK models are most useful when they are combined with indepen dent studies elucidating the mechanism of toxicity so that appropriate dose surrogates may be chosen. REFERENCES agin, G. L., and Blau, G. E. (1982). Application of DACSL (Dow Advanced Continuous Simulation Lan guage) to the design and application of chemical reac tor systems. Amer. Inst. Chern. Eng. Symp. Ser. No. 214,78, 108-118. Andersen, M. E., Clewell, H. J., Gargas, M. L., Smith, F. a., and Reitz, R. H., (1987) Physiologi cally based pharmacokinetics and the risk assessment process for methylene chloride. Toxicol Appl. Phar macol. 87,185-205. Andersen, M. e., Gargas, M. L., and Ramsey, J. C. (1984). Inhalation pharmacokinetics: Evaluating sys tem extraction, total in vivo metabolism, and the time course ofenzyme induction for inhaled styrene in rats based on steady-state blood: air concentration ratios. Toxicol. Appl. Pharmacol. 73, 176-187. Caster, W. O., Poncelet, J., Simon, a. B,, and Arm strong, W. D, (1956). Tissue weights of the Rat. I. Normal values determined by dissection and chemical methods. Proc. Soc. Exp. Biol. Med. 91, 122-126. Davis, N. R., and Mapleson, W. W. (1981). Structure and quantification of a physiological model ofthe dis tribution of injected agents and inhaled anaesthetics. Brit. J. Anaesth. S3,399-404. Dedrick, R. L,, (1973) Physiological pharmacokinetics. J. Dyn. Syst. Meas. Cont. Trans. ASME, September, pp. 255-257. FRantz, S. W., and Watanabe, P. G. (1983). Tetrachloroethylene: Balance and tissue distribution in Male Sprague-Dawley rats by drinking-water admin istration. Toxicol. Appl. Pharmacol. 69,66-72. Fiserova-Bergerova, V. (1975) Mathematical model ing of inhalation exposure. J. Combust. Toxicol. 3, 201-210. Gargas, M. L., and Andersen, M. E., (1988) A gas phase technique for determining the kinetic constants of chemical metabolism in the rat Toxicologist, 8, 112. Gehring, P. J. (1968). Hepatotoxic potency of various chlorinated hydrocarbon vapors relative to their nar cotic and lethal potencies in mice. Toxicol. Appl. Phar macol. 69,66-72. Himmelstein, K. J,, and Lutz, R. J., (1979). A review ofthe applications ofphysiologically based pharmaco kinetic modeling. J. Pharmacokinet. Biopharm. 7, 127-145. International Commission on Radiation Protection (1975). Report of the task group on Reference Man, ICRP Publication 23 (W. S. Snyder et al.. Eds). Pergamon. New York. Lutz, R. J., Dedrick, R. L., Matthews, H. B,, Eling, T. E., and Anderson, M. W. (1977). A preliminary pharmacokinetic model for several chlorinated biphe nyls in the rat Drug. Metab. Dispos. 5,386-396. McNutt, N. S., amster, R. L., McConnell, E, E., and Morris, F. (1975) Hepatic lesions in mice after continuous exposure to 1,1,1-trichloroethane. Lab. Invest. 32,642-654. National Aca trwojatiom gw. WPS, Ch Press, Was! National Acai Pharmacol Washington emy Press, Nolan, R. . Carty, L. and metab trichloroet Toxicol. 4, Quast, J. F. (1984). Ch cologic am Results of I RaMPY, L. ' Gehring, tion studie chloroethy Intematiot W. A. M. r Ramsey, J. F logically b cokinetics Appl. Phar Sato, a., a: SL :ation ofchemical reac- ,7ng. Symp. Ser, No. rj., gargas, m. l., H,, (1987) Physiologimd the risk assessment Toxicol. Appl. Phar- ... and Ramsey, J. C. netics: Evaluating systabolism, and the time inhaled styrene in rats r concentration ratios. 76-187. son, A. B,, and Armweights of the Rat. I, ssection and chemical fed. 91, 122-126. . W. (1981). Structure 4cal model ofthe disinhaled anaesthetics. cal pharmacokinetics, v. ASME, September, P. G. (1983). Tetraissue distribution in -inking-water adminol. 69,66-72. vfathematical modelombust. Toxicol. 3, E., (1988) A gas tkinetic constants rat Toxicologist. 8, c potency of various relative to their narToxicol. Appl. Phar- (1979). A review 'ly based pharmacoinet. Biopharm. 7, diation Protection )n Reference Man, r et al., Eds). Perga- iews, H. B., Eling, 77), A preliminary chlorinated biphe.5,386-396. * cConnell, E. E., sions in mice after hloroethane. Lab. METHYLCHLOROFORM PHARMACOKINETICS 199 National Academy of Sciences (1986). Dose route ex trapolations: Using inhalation toxicity data to set drinking water limits. In Drinking Water and Health Vol. 6. Chap. 6, pp. 185-236. National Academy Press, Washington, DC. National Academy ofSciences (1987). Proceedings ofthe Pharmacokinetics in Risk Assessment Workshop, Washington DC, October 7-9, 1986. National Acad emy Press, Washington, DC. Nolan, R. J,, Freshour, N. L., Rjck, D. L., Mc Carty, L P., and Saunders, J. H. (1984). Kinetics and metabolism of inhaled methylchloroform (1,1,1trichloroethane) in male volunteers. Fundam. Appl Toxicol. 4,654-662. Quast, J. F,, Calhoun, L. L., and McKenna, M. J. (1984). Chloroethene VG. A chronic inhalation toxi cologic and oncogenic study in rats and mice. Part I. Results of findings in mice. Toxicologist, 5,14. Rampy, L. W., Quast, J. F., Leong, B. K. J., and Gehring, P. I., (1978). Results of long-term inhala tion studies on rats of 1.1.1-trichloroethylene and perchloroethylene formulation. In Proceedings ofthe 1st International Congress on Toxicology (G. L. Plaa and W. A. M. Duncan, Eds.). Academic Press, New York, Ramsey, J. R., and Andersen, M. E. (1984). A physio logically based description of the inhalation pharma cokinetics of styrene in rats and humans. Toxicol. Appl. Pharmacol. 73,159-175. Sato, A., and Nakajima, T., (1979) Partition coeffi cients of some aromatic hydrocarbons and ketones in water, blood, and oil. Brit. J. Ind. Med. 36,231 -234. Schumann, a. M., Fox, T. R,, and Watanabe, P. G. (1982a). [HC]Methylchloroform (1,1,1-trichloroethane): Pharmacokinetics in rats and mice following in halation exposure. Toxicol. Appl. Pharmacol. 62,390401. Schumann, a. M., Fox, T. R., and Watanabe, P. G. (1982b). A comparison of the fate of inhaled methyl chloroform (1,1,1 -trichloroetbane) following single or repeated exposure in rats and mice. Fundam. Appl. Toxicol. 2,27-32. Teorell, T. (1937) Kinetics distribution of substances administered to the body. Arch. Im. Pharmacodyn. Ther. 57,205-240. Torkelson, T. R., and Rowe, V. K. (1981). Halogenated aliphatic hydrocarbons containing chlorine, bromine, and iodine. In Patty's Industrial Hygiene and Toxicology (G. D. Clayton and E. Clayton, Eds.), 3rd ed., pp. 3502-3510, Wiley, New York. U.S. Environmental Protection Agency (1983). Draft HealthAssessment Documentfor 1,1,1-Trichloroethyl ene (Methylchloroform). Section 3, p, 29. Office of Health and Environmental Assessment, Research Tri angle Park, NC 27711. Withey, J. R., Collins, B. T,, and Collins, P. G., (1982) Effect of vehicle on the pharmacokinetics and uptake of four halogenated hydrocarbons from the gastrointestinal tract of the rat J. Appl. Toxicol. 3, 313-332. SL 036342 TOXICOLOGY AND APPLIED PHARMACOLOGY 95, 230-240(1988) Physiologically Based Pharmacokinetic Model for Vinylidene Chloride R. W. D'SOUZA AND M. E. ANDERSEN* The Procter & Gamble Company. Miami Valley Laboratories. Box 398707, Cincinnati, Ohio 45239-8707, and *ArmstrongAerospace Medical Research Laboratories, Biochemical Toxicology Branch, Wright Patterson AFB, Ohio 45433 Received January 15.1988; accepted May 22.1988 Physiologically Based Pharmacokinetic Model for Vinylidene Chloride. D'Souza, R. W., and Andersen, M. E, (1988). Toxicol. Appl. Pharmacol. 95, 230-240. Vinylidene chloride (VDQ, a potent hepatotoxin and suspected carcinogen, is metabolized by mixed-function oxi dases into a reactive metabolite(s) which is responsible for its toxicity. The metabolite is detoxi fied by glutathione (GSH), and liver GSH status is an important factor in the expression ofVDC toxicity. A physiologically based pharmacokinetic (PB-PK) model has been developed for VDC in the tat based on oxidative metabolism ofVDC and subsequent GSH detoxification ofmetabo lite. The model offers insight into the complex interrelationship between the processes ofabsorp tion, metabolism, and GSH conjugation, and simulates the manner in which these factors oper ate in regulating VDC toxicity. The PB-PK model successfully predicts blood, tissue, and ex haled air concentrations of VDC, and liver GSH levels as a function of dose and route of administration. The model also explains the complex dose-response mortality curves seen with VDC. Because of the low bloodtair partition coefficient of VDC and its saturable metabolism, the amount of VDC dose that is metabolized is sensitive to the rate of absorption. After an intravenous bolus dose, most ofthe administered VDC is exhaled unchanged within a few min utes. Blood VDC half-life is not representative of metabolism rates but to reequilibration of VDC from fat. Rats with greater fat content, therefore, display longer VDC blood half-lives. Simulations are shown to demonstrate the strength of PB-PK modeling techniques in under standing the kinetic behavior of VDC in the rat under a variety of experimental conditions. IMS Academic Press, Inc. Mathematical modeling of physiologic sys tems to determine the kinetic behavior of compounds, commonly known as physiolog ically based pharmacokinetic (PB-PK) mod eling, has recently been introduced in the area of cancer risk assessment. The utility of this type of modeling has been demonstrated in explaining differences in cancer bioassay data after oral and inhalation exposure to methylene chloride (Andersen et ai, 1987) and in improving the process of estimating cancer risk for low level exposure to methyl ene chloride (Andersen et ai, 1987) and eth ylene dichloride (D'Souza et ai, 1987). This modeling can also be used to explain or pre dict potential for toxicides other than cancer. To illustrate this utility a PB-PK model has been developed for vinylidene chloride (VDC; 1,1-dichloroethylene) in the rat. The model demonstrates the manner in which the interaction of various, often nonlinear pro cesses, can affect both the pharmacokinetics and toxic potential of VDC. VDC is a large volume chemical that is an air and drinking-water contaminant (U.S. EPA, 1982, 1986). VDC has been shown to be a potent hepatotoxin (Jenkins et ai, 1972; Reynolds et ai, 1975; Andersen et ai, 1979) and is also a suspected carcinogen (Maltoni et ai, 1977). The metabolism of VDC has been well studied (Henschler, 1977; Leibman and Ortiz, 1977). VDC is metabolized by mixed- 0041-008X/88 $3.00 Copyright 1988 by Academic Press, Inc, All rights of reproduction in any form reserved. 230 function j^ds ur, ^ve t^chlor ther deti macrom toadihi .i expin dose is a reactive ent VD hepatoti 1977, 1` lism, ai affected level, ro dosing y VDC sponse i plex, int ppm in ing virti for a 4ai, 197` kinetics been of dose af intravei periton of expc (1977, Of COVE VDC-n ter oral crease i inhalat are det< tive to therefo (Jaeger 1977)e tration ery veh atotoxi dosing, with V PB-PK MODEL FOR VINYLIDINE CHLORIDE 231 function oxidases into an epoxide. This epox ide is unstable and can decompose to highly reactive intermediates, primarily chloroacetyl chloride (CAC). The intermediates are ei ther detoxified by glutathione (GSH), bind to macromolecules, or are further metabolized to a dihydrodiol, resulting in C02 elimination in expired air, A portion of unchanged VDC dose is also eliminated in exhaled air. It is the reactive intermediates like CAC, and not par ent VDC, that is responsible for the acute hepatotoxic effects of VDC (Andersen et al., 1977, 1978; Henschler, 1977). The metabo lism, and hence the toxicity, of VDC is affected by several factors. These include dose level, route ofexposure, GSH status, and oral dosing vehicle employed. VDC metabolism is saturable and dose-re sponse curves for mortality in rats are com plex, increasing sharply between 100 and 200 ppm inhalation concentration, but remain ing virtually flat between 200 and 1000 ppm for a 4-hr inhalation exposures (Andersen et al., 1979). Large differences in the pharmaco kinetics and metabolic fate of VDC have been observed with an increase in exposure dose after oral (Jones and Hathway, 1978; McKenna et al., 1978a), inhalation (Mc Kenna et at., 1978b; Andersen et al., 1979), intravenous (Putcha et al., 1986), and intraperitoneal (Jones and Hathway, 1978) routes of exposure. Furthermore, McKenna et al. (1977, I978a,b) have shown that the amount of covalent binding, VDC exhalation, and VDC-related 14C02 exhalation is different af ter oral and inhalation exposures, and an in crease in dose has different effects for oral and inhalation exposures. Since VDC metabolites are detoxified by GSH, VDC toxicity is sensi tive to liver GSH concentration. Toxicity is, therefore, affected by such factors as fasting (Jaeger et al., 1974; Andersen and Jenkins, 1977) and diurnal variation in GSH concen trations (Jaeger et al., 1973). Effects of deliv ery vehicle on the pharmacokinetics and hepatotoxicity have also been observed after oral dosing. Cheico et al. (1981) dosed rats orally with VDC in com oil, mineral oil, and an TABLE I Parameters Used in the VDC PB-PK Model Partition coefficients Liverblood Richly perfused;blood Slowly perfusedtblood Fat:blood Blood.air 1.1 1.1 0.6 18.4 5.0 Kinetic constants ^(mg hr1) Km (mg liter-1) K^OiM^hr-1) /^(hr'1) ^(hr1) Koo^Im 'hr-1) HjCMm) 2.6 0.25 0.33 50 9000 1.82 X l0-: 55 aqueous Tween vehicle and observed that hepatotoxicity was greater with vehicles like com oil and mineral oil from which absorp tion is expected to be slow compared with Tween from which absorption ofVDC would be more rapid. They also observed that the half-life ofexhaled VDC was more prolonged after com oil and mineral oil, compared to Tween vehicle. Our purpose was to develop a physiologi cally based pharmacokinetic model for VDC in the rat that could be used to explain the changes in VDC behavior seen with the above factors and could also predict the phar macokinetic behavior and toxic potential of VDC in untested conditions. METHODS PB-PK Model Development The basic structure of the PB-PK model is similar to that developed by Ramsey and Andersen (1984) for sty rene, utilizing a lumped-tissue group approach. Com partments are the "richly perfused" group comprising such tissues as the lung, kidney, and spleen; the "slowly perfused" group, comprising muscle and skin; and a fat compartment. Metabolism was assumed to take place solely in the liver. The various parameters that were used in developing the PB-PK model are shown below. Partition coefficients. Partition coefficients for blood: air and tissue-.blood (Table 1) were determined by the 232 D'SOUZA AND ANDERSEN method of Sato and Nakajima (1979). Briefly, tissues were homogenized with normal saline using a Polytron homogenizes Homogenates were spiked with low levels of VDC and allowed to equilibrate in sealed reaction ves sels at 37*C in a water bath for 1-3 hr. A sample of the headspace vapor above the homogenate was then ana lyzed for VDC content. By appropriate mathematical corrections, the tissue:air partition coeflicienl was com puted. Tissue:blood partition coefficient was obtained by dividing the tissueiair partition coefficient by the blood: air partition coefficient. Experience with partition co efficient measurements of over 50 chemicals in one of our (M.E.A.) laboratories has indicated that reaction of chemical with tissue homogenate does not typically take place. This may be due to the fact that cofactors, oxygen, etc., are not supplied, and also because the tissues are not handled in a particularly gentle manner. As a rule, two to three measurements are made a few hours apart. Ifthe results are identical, this indicates to us that metabolism is not occurring. Physiologic parameters. Values for organ volume and blood flow were taken from Gaigas etal.( 1986). Modeling VDC metabolism. The pathway ofVDC me tabolism (Scheme I) contained a single activation step. Metabolism rate constants for conversion ofVDC to the epoxide (f'oui and K,,, Table 1) were determined by gas uptake studies (Gaigas et al., 1986). Metabolism was found to be saturable, as previously reported in the litera ture (Andersen et al., 1979). It was assumed that VDC epoxide converted almost instantaneously to CAC or re acted with water to form C02 as the end product. The reaction ofepoxide with water resulting in C02 was mod eled similar to that reported by Dekant et al. (1984) for trichloroethylene oxide. Based on the data of McKenna et al. (1978), where 10% of radiolabel was recovered as C02 in expired air after a 1 mg/kg oral dose ofradiolabel VDC, these two reactions were split in a 9:1 ratio (epox ide to CAC or C02, respectively). In order to simulate virtually instantaneous reactions while maintaining the 9:1 ratio, the first-order rate constant for conversion of epoxide to CAC (Anm) was arbitrarily set to 9000 and the second-order reaction rate constant of epoxide with water (Aixjj* 55 M water in liver) was therefore 1000, re sulting in a value of 1.82 * I0-5 for Koo2 (Table 1). CI*C=CH, A K*. oJ=,] CljC-CH HjO \ MAT GCM Precursors Breakdown .Products ` GSH Synthetase' Precursors Ka GSH - . Breakdown Products [cchSci] IU / Detoxified Products Scheme II. GSH synthesis and depletion by chloroacetyl chloride. For simplicity, it was assumed that the reaction with GSH was only with CAC and not with VDC epoxide or other intermediates. This assumption is justified based on the data of Liebler et al. (1985) where CAC reacted with model thiols (in vitro) I03 times faster than VDC epoxide and 10s times faster than minor metabolites formed from the epoxide. The relative contribution of the GSH pathway in detoxifying CAC, compared to the ability ofthis intermediate to elicit toxicity by binding to macromolecules (metabolite available for toxicity, MAT), was modeled as two first-order processes: reaction ofCAC with GSH (rate constant and reaction with "everything else" (rate constant A'**). Values for and A'*, (Table 1) were obtained by using the PB-PK model for VDC developed until this level and varying A*, (trial and error) until GSH depletion model predictions were a visual fit to GSH depletion data reported by McKenna et al. (1977), for inhalation exposures to increasing VDC concentrations. Starting values for ,, and Aw were ap proximated based on previous experience with modeling GSH conjugation with reactive metabolites (Andersen et al., 1986; D'Souza et al., 1987,1988). Modeling GSH turnover. The model for GSH turnover and reaction with CAC (Scheme II) is similar to that re ported for GSH depletion by ethylene dichloride (D'Souza et al., 1988). It is somewhat simpler in that, unlike ethylene dichloride, parent VDC does not conju gate with GSH. Except for and A**, the other rate constants are similar to those utilized by D'Souza et al. (1988). Mathematical construction of model. Mass-balance differential equations were written for the various com partments in the model (Appendix A). Briefly, for non metabolizing compartments differential equations were written to describe the influx, binding, and efflux of VDC. The lung compartment was described similarly, but included a description of gas exchange between the lung and the ambient air. The equations for the liver compartment also accounted for VDC metabolism and GSH turnover. These equations were solved simulta neously by numerical integration, using Gear's algo rithm for stiff systems. The ACSL1 computer program CO, Scheme 1. Model for VDC metabolism pathways. 1 Advanced Continuous Simulations Language, Mitchell and Gauthier, Concord, MA. 10J 8 10 10"' Fig. VDC ci ingasi centrat a single "richly (Data s was et these e Anii werep and to he da: from t tabolit 1986) mentt tions. souro taken thetk SL 03634` f iSH , Breakdown Products oxifwd iducts depletion by chloro- that the reaction with with VDC epoxide or tion is justified based >) where CAC reacted mes faster than VDC in minor metabolites lative contribution of .'AC, compared to the toxicity by binding to ailable for toxicity, ier processes: reaction and reaction with J. Values for and ing the PB-PK model and varying K,a (trial odel predictions were eported by McKenna res to increasing VDC JC__and Kin were ap^Bvice with modeling ^Brolites (Andersen et 8). del for GSH turnover ) is similar to that re ethylene dichloride vhat simpler in that, VDC does not conjud /Tte, the other rate sed by D'Souza a ai. nodel. Mass-balance for the various comA). Briefly, for nonntial equations were nding, and efflux of . described similarly, xchange between the 'uations for the liver DC metabolism and were solved sitpulta- using Gear's algo1 computer program ulations Language, IA, PB-PK MODEL FOR VINYLIDINE CHLORIDE 233 Fig, 1. Model predicted (--) and observed (symbols) VDC concentrations in different tissues ofthe rat follow ing a single 50 mg/kg iv bolus dose. Predicted VDC con centrations for blood, liver, lung, and brain are shown as a single curve, as tissue:blood partition coefficient for the "richly perfused" tissue group was approximately one. (Data source: D'Souza, 1984). was employed on an IBM PC-AT computer to solve these equations. Animal studies. Although some experimental studies were performed to validate various portions ofthe model and to demonstrate predictability ofthe model, much of the data utilized to confirm model predictions were taken from the literature. Because the literature for VDC me tabolism and acute toxicity is quite extensive (U.S. EPA, 1986) it was unnecessary to repeat many of the experi ments. In all the figures, curves are PB-PK model predic tions and symbols are experimental data. When the data source is acknowledged and referenced, the data were taken from the literature. When no reference is made, the data were collected by the authors for this study. difference in VDC concentrations was ex pected between this group of tissues and blood. Therefore, model prediction is shown as a single curve for the richly perfused tis sues. Simulated and experimentally deter mined fat concentrations are also shown. As predicted, VDC concentrations in fat are considerably higher than concentrations in other tissues. The PB-PK model appears to predict VDC biodistribution in these iv dos ing studies. Dose level and route-of-exposure effects. The model was used to gain insight into the disposition of VDC as a function ofdose level and route ofexposure (Fig. 2). Because ofsat uration of VDC metabolism, there is a de crease in the percentage of dose that is con verted to the epoxide as dose is increased. The percentage ofVDC dose that reacts with GSH (glutathione-conjugated metabolite, GCM), or is eventually exhaled as C02, decreases with increasing dose (Fig. 2). This decrease begins at about a 50 mg/kg dose of VDC, in dicating the dose where saturation ofthe oxi dative pathway occurs. Although a decrease is seen in the parameters noted above, be cause ofthe complex interaction between the processes of absorption, metabolism and GSH conjugation, the relative decreases in these parameters do not run parallel. At the RESULTS Once the PB-PK model was constructed, simulations could be generated to predict the disposition of VDC, its metabolic products, or liver GSH. In order to validate the model, predictions were generated for blood and tis sue VDC concentration-time profiles after an iv bolus dose. These predictions were con firmed with experimental data (Fig. 1). Since the tissueiblood partition coefficient for the richly perfused group of tissues was 1.1, no Oral Dose (mg/kg) Fig. 2. Model simulations (--) and observed data for the effect of oral dose level on percentage of VDC dose that is exhaled unchanged (O), exhaled as C02 (), or GCM (glutathione-conjugated metabolite (no observed data)). Ka of 1.0 hr'1 used in simulations. (Data source: 1 and 50 mg/kg doses, McKenna et al., 1978a; 350 mg/ kg dose, Jones and Hathway, 1978.) SL 036346 234 D'SOUZA AND ANDERSEN ioo.ooh 10.00 Body Burden Exhaled CO: 0.01 10 100 1000 Inhalation Concentration (ppm) Fig. 3. Model simulations (--) for the effect of inhala tion dose level (6-hr exposures) on body burden () and amount of dose exhaled as C02 (O). (Data source; Mc Kenna etal.. 1978b.) point where metabolism is saturated, an in creasing percentage of absorbed VDC dose is also exhaled in the breath (Fig. 2). Experi mental data for exhaled VDC and C02 are also shown to confirm model predictions. Once again, model predictions agree quite well with reported data. A series of simulations were performed to study the effects of inhalation exposures (6 hr) on the disposition of VDC (Fig. 3). After inhalation exposures an apparent linear rela tionship exists between exposure concentra tion and the parameters shown, up to about 200 ppm concentration. At concentrations exceeding 200 ppm, VDC metabolism is sat urated and the model predicts a less than pro portional increase in the parameters shown with an increase in exposure concentration. Data from the literature (McKenna et al., 1978) for body burden and C02 exhalation (Fig. 3) are shown along with the simulations. Saturation of VDC metabolism at about 200 ppm concentration has also been confirmed by Dallas et al. (1983). OSH depletion effects. Figure 4 illustrates the effects of inhalation (Figs. 4a and 4b) and oral (Fig. 4c) VDC exposures on liver GSH levels. Experimentally measured liver GSH concentrations for inhalation (McKenna et at.. 1977; Reynolds et al., 1980) and oral ex posures agree quite well with model predicted values. As GSH levels are inversely related to toxicity, prediction ofthe time course ofGSH depletion indirectly allows for the estimation i of toxic potential. Absorption rate effects. While simulating various aspects of the behavior of VDC, sev eral observations were made that would not be readily noticed without the PB-PK model. For instance, because of the low blood:air partition coefficient and saturable metabo lism of VDC, its disposition is sensitive to the rate of entry into the body. Figure 5 depicts the effects of increasing absorption rates on the percentage of VDC dose that will be ex haled unchanged at different oral dose levels. These simulations were performed for doses ranging from 1 to 200 mg/kg. Experience in studying the oral absorption rates of halogenated hydrocarbons (data not shown) sug gests that the absorption rate constant for these compounds from aqueous vehicles, like Tween/water mixtures, is typically 5.0 hr-1, while the rate constant from com oil vehicle is typically 1.0 hr-*. As absorption rate in creases, particularly at higher doses, a greater percentage of VDC dose is exhaled un changed. The toxicological implications of this observation are obvious. That is, greater toxicity will be seen ifcom oil is used as the oral delivery vehicle compared to aqueous vehicles. Data collected by Cheico et al. (1981), where greater hepatotoxicity was seen with com oil and mineral oil vehicles com pared to a Tween vehicle, are consistent with this finding. Pharmacokinetic data in the lit erature also confirm this observation. Mc Kenna et al. (1978a) noted that from a com oil vehicle approximately 3 and 20% of the oral doses of 1 and 50 mg/kg, respectively, were exhaled. Our model predicts similar re sults ifan absorption rate constant of0.5-1.0 hr-1 is used. Similarly, the model predicts data reported by Jones and Hathway (1978) where 0.6 and 62% were exhaled after oral and 11 and 90% were exhaled after intraperitoneal doses of 0.5 and 350 mg/kg, respec tively, (k,, of 10 hr-* was used to simulate rapid absorption for the intraperitoneal doses; simulations not shown). Fig. * (c)expo- Intraven predicts th; VDC, beca the body, n changed w tioas not si finned by s and Hathw venous do; kg intrave metabolisr ft) J- 01 Fig. 5. Sii stant on the i at various dc negligible at doses, and n all of the dc rate. This pr< mg/kg dose (absorption aqueous veh SL 036367 i for the estimation -ts. While simulating ehavior of VDC, sevmade that would not >ut the PB-PK model, of the low bloodrair d saturable metabolion is sensitive to the )dy. Figure 5 depicts . absorption rates on dose that will be ex rent oral dose levels, performed for doses ig/kg. Experience in tion rates of halogeta not shown) sugn rate constant for queous vehicles, like s typically 5.0 hr-1, om com oil vehicle absorption rate indier doses, a greater se is exhaled unjal implications of That is, greater ^roil is used as the ipared to aqueous by Cheico et al totoxicity was seen oil vehicles com- are consistent with etic data in the litobservation. McJ that from a com 3 and 20% of the /kg, respectively, >redicts similar re- onstantofO.5-1.0 e model predicts ! Hathway (1978) .xhaled after oral ed after intraperi0 mg/kg, respecused to simulate intraperitoneal n). PB-PK MODEL FOR VINYLIDINE CHLORIDE 6-Hour tnhaMkm Exposures at DHfefVftt CortQantvatkNM 4*Hour Inhalation Exposure at 200 ppm Conc*nation _ 100 1 600 200 mg/kg Oral Exposure In Com Oti Vehicle 235 0.00 1 10 100 1000 0 2 4 6 8 10 12 inhalation Concentration (ppm) Time (hour) 024 6 Time (hour) Fig. 4. Predicted (--) and observed () liver GSH concentrations following inhalation (a and b) and oral (c) exposure to VDC. (Data source: (a) McKenna et at., 1977; (b) Reynolds et al., 1980.) Intravenous dosing studies. The model also predicts that after a bolus intravenous dose of VDC, because ofthe rapid input of VDC into the body, most ofthe dose will be exhaled un changed within the first few minutes (simula tions not shown). This prediction can be con firmed by studying the data reported by Jones and Hathway (1978), where 80% of an intra venous dose was exhaled after a low 0.5 mg/ kg intravenous dose. Furthermore, if VDC metabolism was partially blocked or induced Fig. 5. Simulated effect of oral absorption rate con stant on the percentage of VDC dose exhaled unchanged at various dose levels. It can be observed that the effect is negligible at low doses, very dramatic at the intermediate doses, and minor at doses of200 mg/kg, where virtually all of the dose is exhaled, regardless of the absorption rate. This prediction suggests that VDC toxicity at say 50 mg/kg dose will be much greater with com oil vehicle (absorption rate constant 0.5-2 hr-1) compared to an aqueous vehicle (absorption rate constant 3-6 hr"`). (Fig. 6a) no change in blood VDC half-life would be seen after an iv bolus dose. A sim ilar simulation was performed for oral doses with absorption rate constants of 1 and 5 hr*1 to illustrate the effect of absorption rate on blood VDC kinetics. As absorption rate decreases, metabolism differences due to changes in become more apparent. These observations suggest that unlike stud ies with nonvolatile compounds, intravenous bolus dosing has limited utility when study ing the pharmacokinetic behavior or metabo lism of volatile materials for which exposure from the environment is not intravenous. Body-fat effects. PB-PK model simulations suggested that VDC blood half-life was due to reequilibration of VDC from fat depots back into the blood. Therefore, an increase in body fat would be expected to cause an increase in blood VDC half-life. Simulations were per formed to determine blood VDC half-life in rats ofthree body-weight levels, 180,300, and 600 g, and the simulations compared to liter ature data (D'Souza, 1984) to confirm predic tions (Fig. 7). The only parameter that was changed when performing these simulations was percentage of body fat. For the three body-weight groups, body fat was set at 7% for the 180-g group, 12% for the 300-g group, and 22% for the 600-g group (Zucker and Zucker, 1963). Effect ofdose level on mortality. Andersen et al. (1979) showed an interesting dose-re sponse relationship between inhaled VDC dose level and mortality in rats. They noted a very sharp increase in mortality, from 0 to SL 036348 236 D'SOUZA AND ANDERSEN c Time (hour) Time (hour) Fig. 6. Simulated blood VDC concentrations following intravenous bolus (a), and oral (b and c) doses. These simulations demonstrate that after intravenous bolus doses effects ofenzyme inhibition or induction will not be observed from studying blood VDC levels. These effects will manifest themselves in blood VDC levels after oral dosing. The slower the absorption rate constant,,the more pronounced the effect. 50%, when inhalation exposure concentra tion for a 4-hr exposure was increased from 100 to 200 ppm. However, from 200 to 1000 ppm there was virtually no increase in per centage mortality. The PB-PK model was used to study the relationship between inha lation exposure concentration and MAT. This relationship was then plotted on the dose-response plot of Andersen et al. (1979). The result is shown in Fig. 8. It is clear from this plot that the dose-response relationship for VDC is similar to the model predicted dose-MAT relationship. It can be seen that 50% mortality is achieved when about 12 /tmol MAT is produced. Using this kind of information, mortality can be predicted for, say, other routes of intake. That is, regardless of the VDC dose and exposure route, 50% mortality is predicted when 12 /tmol MAT is formed. Preliminary extrapolation to humans. Al though none of the parameters for construct ing a PB-PK model were obtained in hu mans, a preliminary attempt was made to ex trapolate the model to humans. The PB-PK model for the rat was scaled to the human by correcting for known physiologic differences between the rat and human, and by estimat ing VDC metabolism rates in humans based on allometric scaling principles (Dedrick et al., 1973). In extrapolation to the human, Time (hour) Fig. 7. Predicted (--) and observed (symbols) effects of body weight (body fat) on VDC blood concentrations in rats of different body-weight levels following a single 50 mg/kg iv dose. (Data source: D'Souza 1984). Concentration (ppm) Fig. 8. Predicted (--) metabolite available for toxicity (MAT, chloroacetyl chloride not detoxified by glutathi one (see text for more detailed explanation) and mortal ity dose response for 4-hr VDC inhalation exposures. (Data source: Andersen etai. 1979.) I 2.67 4.00 fie (hour) al (b and c) doses, lition or induction ives in blood VDC Te effect. Using this kind of i be predicted for. That is, regardless x)surc route, 50% i 12 ixmol MAT is >n to humans. Al ters for construct- obtained in huit was made to exnans. The PB-PK the human by ^^gic differences , and by estimatin humans based ;iples (Dedrick et a to the human. -25 20 * -15 oe 10 5 2 >50 1000 m) available for toxicity ictoxified by glutathianation) and mortalnhalation exposures. .) i PB-PK MODEL FOR V1NYLID1NE CHLORIDE 237 cardiac output and pulmonary ventilation was scaled as a function of body weight (BW) to the 0.7 power, as BW to the 0.74 power, and body fat was changed from 7% for a 200 g rat to 20% for a 70-kg human. This type of scaling has been successfully used in extrapolating animal data to the human for halogenated hydrocarbons using PB-PK models (Gaigas et al., 1986; Andersen et al., 1987; D'Souza et al., 1987). Figure 9 displays the relationship between VDC exposure dose level and the amount of epoxide formed for the rat and human. This simulation was con ducted for both oral (Fig. 9a) and inhalation (Fig. 9b) exposures. Although in previous simulations MAT was used as the "internal dose," in this case, because humans are un likely to be exposed to VDC concentrations that would decrease liver GSH, the shape of the curves for MAT and epoxide (versus VDC dose) are expected to be similar, and therefore epoxide is being used. Additionally, the use of epoxide serves to minimize the number of rate constants to be scaled from rat to human. It can be seen that saturation effects in VDC metabolism begin at about 10 mg/kg for oral, and 200 ppm for inhalation exposures for both the rat and human. Also, the amount ofepoxide formed is predicted to be lower in the human compared to the rat at equivalent mg/kg oral or ppm inhalation concentrations. DISCUSSION The PB-PK model developed for VDC demonstrates the complex interplay of the many factors that contribute to the kinetic behavior and subsequently the toxicity of the compound. In terms of metabolism and pharmacokinetics at least, data from the liter ature appear to be different from one study to the other and at times appear to contradict each other when only minor differences in ex perimental design exist. The PB-PK model, however, demonstrates that given the physi cal, chemical, and biochemical properties of F!g. 9. Simulated amount of VDC epoxide produced at different (a) oral and (b) inhalation dose levels in the rat and human. After oral exposure saturation effects in VDC metabolism are expected in both rat and human at doses of 10 mg/kg, while after inhalation exposure satu ration is predicted after about 200 ppm (6-hr) exposures. VDC, such differences are expected and can be predicted. For example, VDC metabolism profile, in terms of percentage VDC exhaled, percentage metabolized and percentage con jugated with glutathione is different for different routes of exposure and dose levels as reported by McKenna et al. (1978a,b) and Jones and Hathway (1978). It appears that slightly different experimental conditions can have large effects on the kinetic behavior of VDC. For example, Putcha et al. (1986) have shown that although ether anesthetic used in their experiments may have had potential to inhibit VDC metabolism, no differences were seen in the kinetic behavior of VDC when ether-treated and untreated rats were com pared after intravenous bolus injections of 10 mg/kg. The PB-PK model was used to study such a situation and simulations indicate that with intravenous bolus dosing no differences will be seen in VDC blood kinetics even if metabolism is blocked. Also, Putcha et al. (1986) noted a slight increase in the blood clearance ofVDC with an increase in intrave nous dose, while most of the other studies suggest a decrease in clearance because ofsat uration of metabolism. Putcha et al. (1986) calculated clearance from the area under the blood VDC time plot from the first time point (2-3 min postdose) to infinity. The PB-PK model demonstrates that for an intravenous dose most ofthe injected dose will be exhaled within the first few minutes, and, therefore. SL 036350 m 238 D'SOUZA AND ANDERSEN under these experimental conditions VDC blood clearance will not show a decrease in clearance with increasing dose, but may actu ally show an increase, depending on when the first sample was taken (simulations not shown). Similarly, the effect of com oil vehi cle on hepatotoxicity (Cheico et al,, 1981) can also be readily explained on the basis of different rates of absorption of VDC from various dosing vehicles. The purpose of obtaining animal kinetic data is to help in predicting the risk ofchemi cals to humans under realistic exposure situa tions. Because of the extreme sensitivity of VDC kinetics to these various factors, any model that is used to predict risk of VDC ex posure to humans must be sensitive to these factors in order to meaningfully extrapolate the animal data to humans. A model that is restricted to extrapolation within a narrow data base cannot offer insight in untested conditions and, therefore, would have lim ited utility in extrapolation of animal data to the human. Scaling ofthe PB-PK model from the rat to the human presented here serves mainly the purpose ofillustrating the manner in which these models can be used to extrapo late dose and toxicity information from ani mal to human for risk assessment. The PBPK model for the human has parameters (and Km, for example) that were not ob tained or verified in the human, nor have the final blood VDC concentrations in the hu man been validated by experimental data. In summary, the metabolism and pharma cokinetics of VDC is complex and is affected by many factors. A composite model, physio logically realistic in design, has been devel oped to explain the kinetic behavior of VDC, predict untested situations, and help assess human risk to VDC exposure. APPENDIX A Mass Balance Equationsfor VDC Model Mass balance differential equations for lung uptake of compound and for influx and efflux in noneliminating compartments were ] similar to those for methylene chloride (An dersen et al,, 1987), while equations fori glutathione synthesis and depletion can be found in D'Souza et al. (1988). Differen tial equations describing VDC metabolism, which are unique to this model, are shown below. <W4 = (K**CVL)/Km + CVL) ^ammA4 = dAM/d, -- ^acoj/d, ~ d^cAc/d, d^coi/d, = Kcot'CMM*H20*VL d\cAcidt = Kjnm * AMM dACACm/d, = ^ACAC ~ ^AMAt/dt -- ^gcmM ^amat/d, = Kfcc * ACACM daaJd, = KgimGSH * CCACM * VL APPENDIX B Nomenclature ACAC ACACM aco2 AM AMAT d/d, GCM Amount of chloroacetyl chloride formed. Amount of chloroacetyl chloride present at time, t. Amount of C02 formed. Amount of epoxide formed. Amount of metabolite available for toxicity. Amount of epoxide present at time, t. Amount of glutathione-conju gated metabolite. Concentration of chloroacetyl chloride at time, t. Concentration ofepoxide at time, t. Differential. Glutathione-conjugated metabo lite. Glutathione concentration. Water concentration. Rate constant for reaction of ep oxide with water. ANDERSEN, M aNdCONOL kinetic modi by inhaled h AND^^L M SMrrH, F. A cally based ] process for macol. 87,1 ANDERSEN, N Jenkins, L acute toxici Pharmacol Andersen, N toxicity of sex, age ar 157-167. ANDERSEN, 1> (1978). Thi dichloroetl tion and i> on mortali Cheico, P., (1981). Efi chloroethy 146-155. Dallas, C. 1 and Brlk tion of 1,1 exposure. SL Q3635lJ PB-PK. MODEL FOR VINYLIDINE CHLORIDE 239 K(a A,nm Km Ko Kd K> K,MAT Vr mu First-order rate constant for for mation of MAT. First-order rate constant for for mation ofGCM. First-order rate constant for chloroacetyl chloride formation. Michaelis constant for oxidative pathway. Zero-order glutathione synthesis, time and GSH dependent. Glutathione synthetase forma tion. First-order rate constant for gluta thione breakdown. First-order rate constant forgluta thione synthetase breakdown. Metabolite available for toxicity. Maximum velocity of oxidative pathway. REFERENCES Andersen, M. e., Clewell, H. J., Ill, Garoas, M. L., andConolly, R. B. (1986). A physiologic pharmaco kinetic model for hepatic glutathione (GSH) depletion by inhaled halogenated hydrocarbons. Toxicologist 6, 148. Andersen, M. E,, Clewell, H. J,, III, Gargas, M. L., Smith, F. a., and Reitz, R. H. (1987). Physiologi cally based pharmacokinetics and the risk assessment process for methylene chloride. Toxicol. Appl. Phar macol. 87, 185-205. Andersen, M. E,, French, J. E,, Jones, R. A., and Jenkins, L. J., Jr. (1979). Saturable metabolism and acute toxicity of 1,1-dichloroethylene. Toxicol. Appl Pharmacol. 47,385-393. Andersen, M. E., and Jenkins, L. J,, Jr. (1977). Oral toxicity of 1,1-dichloroethylene in the rat: Effects of sex, age and fasting. Environ. Health Perspect. 21, 157-167. Andersen, M. E., Jones, R. a., and Jenkins, L. J., Jr. (1978), The acute toxicity of single, oral doses of 1,1dichloroethylene in the fasted male raf Effect ofinduc tion and inhibition of microsomal enzyme activities on mortality. Toxicol. Appl Pharmacol. 46,227-234. Cheico, P,, Moslen, M. T., and Reynolds, E. S. (1981). Effect ofadministration vehicle on oral 1,1 -dichloroethylene toxicity. Toxicol. Appl. Pharmacol. 57, 146-155. Dallas, C. E,, Weir, F. W., Feldman, S., Putcha, L., and Bruckner, J. V. (1983). The uptake and disposi tion of 1,1-dichloroethylene in rats during inhalation exposure. Toxicol Appl. Pharmacol. 68,140-151. Dedrick, R. L. (1973). Animal scale-up. J. Pharmacokinet. Biopharm. 1(5), 435-461. Dekant, W., Metzler, M., and Henschler, D. (1984), Novel metabolites of trichloroethylene through dichlorination reactions in rats, mice and hu mans. Biochem. Pharmacol. 33,2021-2027. D'Souza, R. W. (1984). Pharmacokinetics of haloge nated hydrocarbons in the raL Ph.D. dissertation, Uni versity ofHouston. D'Souza, R. W,, Francis, W. R,, and Andersen, M. E. (1988). Physiological pharmacokinetic model for tissue glutathione depletion and increased resyn thesis following ethylene dichloride exposure. J. Phar macol. Exp. Ther. 245,563-568. D'Souza, R. W,, Francis, W. R., Bruce, R. D,, and Andersen, M. E. (1987). Physiologically-based phar macokinetic model for ethylene dichloride and its ap plication in risk assessment In Pharmacokinetics in Risk Assessment: Drinking Water and Health. Vol. 8, pp. 286-301. Natl. Acad. Press, Washington, DC. Gargas, M. L., Clewell, H. J., Ill, and Andersen, M. E. (1986). A physiologically based simulation ap proach for determining metabolic constants from gas uptake data. Toxicol. Appl. Pharmacol. 86,341-352. Henschler, D. (1977). Metabolism and mutagenicity of halogenated olefins--A comparison of structure and activity. Environ. Health Perspect. 21,61-64. Jaeger, r. T., Conolly, R. B., and Murphy, S. D. (1973). Diurnal variation of hepatic glutathione con centration and its correlation with 1,1-dichIoroethylene inhalation toxicity in rats. Common. Chem. Pa thol. Pharmacol. 6,465-471. Jaeger, R. J., Conolly, r. B., and Murphy, S. D. (1974). Effect of 18 hr fast and glutathione depletion on 1,1-dichloroethylene induced hepatotoxicity and lethality in rats. Exp. Mol. Pathol. 20,187-198. Jenkins, L. J., Jr., Trabulus, M. J., and Murphy, S. D. (1972). Biochemical effects of 1,1-dichloroethyl ene in rats: Comparison with carbon tetrachloride and 1,2-dichloroethylene. Toxicol. Appl. Pharmacol. 23, 501-510. Jones, B. K., and Hathway, D. E. (1978). The biologi cal fate of vinylidene chloride in rats. Chem. Biol. In teract. 20,27-41. Leibman, K. C, and Ortiz, E. (1977). Metabolism of halogenated ethylenes. Environ. Health Perspect. 21, 91-98. Liebler, d. C, Meredith, M. J., and Guengerich, F. P. (1985). Formation of glutathione conjugates by reactive metabolites of vinylidene chloride in microsoines and isolated hepatocytes. Cancer Res. 45,186-- 193. Maltoni, C., Corn, J., Morisi, J., and Cheico, P. (1977), Carcinogenicity bioassays of vinylidene chlo ride. Research plan and early results. Med. Lav. 68, 241-262. SL 036352 240 D'SOUZA AND ANDERSEN McKenna, M. J., Watanabe, P. G., and Gehring, P. J. (1977). Pharmacokinetics of vinylidene chloride in the rat Environ. Health Perspect. 21,99-105. McKenna, M. J,, Zemple, J. A., Madrid, E. O., Brown, W. J., and Gehring, P. J. (1978a). Metabo lism and pharmacokinetic profile of vinylidene chlo ride in rats following oral administration. Toxicol. Appi Pharmacol. 45,821-835. McKenna, M. J., Zemple, J. A., Madrid, E. O., and Gehring, P. J. (1978b). The pharmacokinetics of [14C]vinylidene chloride in rats following inhalation exposures. Toxicol. Appl. Pharmacol. 45,599-610. Putcha, L., Bruckner, J. V,, D'Souza, R. W,, Desai, F., and FEldman S. (1986). Toxicokinetics and bio availability of oral and intravenous 1,1-dichloroethylene. Fundam. Appl. Toxicol. 6,240-250. Ramsey, J. R., and Andersen, M. E. (1984). A physio logically based description of the inhalation pharma cokinetic ofstyrene in rats and humans. Toxicol. Appl. Pharmacol. 73,159-175. Reynolds, e. S., Moslen, m. T., Boor, P. J., and Jae ger, R. J. (1980). 1,1-dichioroethylene hepatotoxicity. Time course ofGSH changes and biochemical ab errations. Amer. J. Pathol. 101(2), 331-343. Reynolds, E. S,, Moslen, M. T,, Szabo, S,, Jaeger, R. J,, and Murphy, S. D. (1975). Hepatotoxicity of vinyl chloride and 1,1-dichloroethylene. Role of mixed function oxidase system. Amer. J. Pathol. 81, 219-232. Sato, A., and Nakajima, T. (1979). Partition coeffi cients of some aromatic hydrocarbons and ketones in water, blood and oil. Brit. J. Ind. Med. 36,231-234. U.S. EPA (1982). 1,1-Dichloroethylene. Occurrence in Drinking Water, Food andAir. Prepared by JRB Asso ciates, Inc., under Contract No. 68-01-6185 for the Office ofDrinking Water, U.S. EPA, Washington, DC. U.S. EPA (1986). Health Assessment Document for Vi nylidene Chloride. Final Report. U.S. EPA, Washing ton, DC. Zucker, T. F,, and Zucker, L. M. (1963). Fat accre tion and growth in the raL J. Nutr. 80,6-19. ! \ ICCLOG De, i Humar (DBE) i fruit fu knock be mut; et al., 1 tal anir metabc ceeds v tomi, 1 chromt 1 This Occupat 2 Tow DD)-treated rats, Toxicol. '9, Greim, H. (1984). ^roxine on 2,3,7,8-tetXDD)-induced toxicity. 372-376. CHEUFLER, E., PaZDER985). Thyroid hormones ^,8-tetrachlorodibenzo-pLnviron. Health 16,481- M., Moore, R. a., and of 2,3,7,8-tetrachlorodijn body weight and lipid tppl. Pharmacol. 81,356- ,nd Dormandy, T. L. veen erythrocyte superoxythrocyte copper levels in s with rheumatoid arthri-403. tormone action at the cell 17-123, 173-177. F., Hassan, M. Q., Mur4ADPOUR, H. A. (1986). reactive oxygen species in idation. Adv. Exp. Med ., and Murray, W. J. [oxidation and inhibition TCDD. In Banbury Re ims ofDioxin Action (A. -igh, Eds.), pp. 241-253. Cold Spring Harbor, ., M. (1978). Determinacutsor in tissues by thio>chem, 86,271-278. n of cardiotoxic hazards. 187. ore, J. A., and Gupta, and clinical chemistry libenzo-p-dioxin in laboerspect. 5,111-118. toxicology and applied pharmacology 95, 185-199 (1988) Physiologically Based Pharmacokinetic Modeling with Methylchloroform: Implications for Interspecies, High Dose/Low Dose, and Dose Route Extrapolations R. H. Reitz, * t'' J- N. McDougal,! M. W. Himmelstein,! R. J. Nolan,t and A. M. Schumann! `Toxicology Research Laboratory, 1803 Building, Dow Chemical Company, Midland, Michigan 48674; fMammalian & Environmental Toxicology Research Laboratory, Health <t Environmental Sciences. Dow Chemical U.S.A., Midland, Michigan 48674; and %Harry G. ArmstrongAerospace Medical Research Laboratory, Toxic Hazards Division (AAMRL/TH), Wright Patterson AFB, Ohio 45433 ReceivedJune 29,1987; acceptedMay 22,1988 Physiologically Based Pharmacokinetic Modeling with Methylchloroform: Implications for Interspecies, High Dose/Low Dose, and Dose Rouie Extrapolations. Reitz, R, H., McDougal, J. N., Himmelstein, M. W., Nolan, R. J., and Schumann, A. M. (1988). Toxicol. Appl. Pharmacol. 95, 185-199. A unified physiologically based pharmacokinetic (PB-PK) model was developed and used to describe the disposition ofmethylchloroform (1,1,1 -trichloroethane, MC) in three different species (rats, mice, and humans) after four different routes ofexposure (inhala tion, intravenous injection, bolus gavage, and drinking water administration). Metabolism of MC followed Michaelis-Menten kinetics in each species. K--.'s were calculated from the allometric equation: PTM, = 0.419 BW \ and Km appeared to be identical in each species (5.75 mg equivalents/liter). Once the PB-PK model had been developed for young adult animals (1-3 months ofage), it was used to study the disposition ofMC in older rats and mice (approximately 18.5 months ofage). Most of the changes in the pharmacokinetic behavior of MC in older rats could be simulated by increasing the size ofthe fat compartment in the PB-PK model from 7 to 18% of body weight. However, the pharmacokinetic behavior in older mice was more complex; increasing the size of the fat compartment in this species from 4 to 18% only accounted for part ofthe observed differences between old and young animals. An appropriate dose surrogate (average area under the liver concentration/time curve) was selected and the PB-PK model was used to make quantitative comparisons between "internal doses" of MC in long term animal studies and "internal doses" associated with human exposures to MC. Values ofthe dose surro gate in humans consuming 2 liters/day ofwater with typical levels of MC contamination (1-10 ppb) were four to six orders of magnitude lower than the dose surrogates in the rodent studies at levels of MC exposure which failed to produce adverse effects on the liver (875-1500 ppm, 6 hr/day, 5 days/week). i9S Academic Pmsj, Inc. Lifetime bioassays in rodents are conducted to assess the potential of various substances to produce chronic toxicity, including carci nogenicity. When these studies are com pleted, they are used to estimate the risk that 1 To whom correspondence should be addressed. similar toxicity will be manifested in human populations exposed to the same agents. However, there are significant differences be tween the conditions in the lifetime rodent bioassay and conditions existing in the hu man environment, and these differences need to be considered in hazard evaluations. Physiologically based pharmacokinetic ; 0041-0G8X/88 $3.00 Copyright Q 1988 by Academic Press, Inc. All rights of reproduction in any form reserved. 186 REITZ ET AL. (PB-PK) models provide a technique for eliminating some of the uncertainty in risk estimations (NAS, 1986; Andersen et al., 1987), These models calculate the "internal dose" of relevant chemical species in target organs during specific exposure scenarios, and are useful for high dose/low dose extrap olations, dose route extrapolations, and interspecies extrapolations. The utility of the PB-PK models stems from the fact that they realistically describe the sizes oforgan compartments, partitioning of test material between blood and tissues, flow of blood and gases through the organs, and rates of metabolic transformation of the test chemical. The concept of incorporating this type of information into pharmacoki netic models was proposed 50 years ago by Teorell (1937), but computer hardware and software to implement his ideas were not available at that time. Examples of PB-PK models implemented with modem technol ogy may be found in the works of Himmelstein and Lutz (1979), Dedrick (1973), and Fiserova-Bergerova (1975). In order to demonstrate the versatility of this technique with a specific compound, we have developed a PB-PK model for 1,1,1 -trichloroethane (methylchloroform, MC). This model was based on the model used by Ram sey and Andersen (1984) to describe the dis position of styrene, and details of the mathe matical formulation of such models may be found in that publication. Our objectives in this project were three fold: (1) To demonstrate the ability of the PBPK model to perform high dose/low dose, dose route, and interspecies extrapolations. (2) To determine whether PB-PK models could be used to predict previously reported changes in MC disposition in older rats and mice (Schumann et al., 1982b). (3) To develop a mechanism for making quantitative comparisons between chronic animal inhalation studies and human expo sures. MATERIALS AND METHODS Test materials. A sample of 1,1,1-trichloroethane (MC) was obtained from the Inorganic Chemicals De partment of the Dow Chemical Co. This material was analyzed by gas chromatography (7.5% Oronite NIW plus 2.5% Carbowax 20 m TPA on 80/100 mesh Chromosorb W.H.P.) using a programmed temperature gradi ent of 4*C/min, starting at 70'C and finishing at 130*C. The test material was found to be more than 99.8% MC. Radioactive MC [2-l4C], Lot No. 1119-293, (sp act 2.51 mCi/mmol) was obtained from New England Nuclear (Boston, MA) This material was repurified be fore use by preparative gas chromatography (10% SP 1000 on Chromosorb WAW, isothermal 100*C). The ra diochemical purity after purification was found to be 97.9%. For administration to the animals, a saturated so lution was prepared by Stirring [l4C]MC in water for 8 hr (excess MC present). The sp act of the MC used in these experiments was 4,53 X 10-3 mCi/mraoi and the concen tration was 3.57 mg/ml(2.66 x JO43dpm/ml). Animals. Male Fischer 344 rats (weight range 240-260 g) were obtained from Charles River Breeding Labora tories (Kingston, NY) and acclimated to the laboratory for at least 7 days before being placed on test The ani mals were housed in stainless-steel cages in rooms de signed to maintain 22*C, with relative humidity con trolled between 40 and 60%, and a 12 hr/day lighting cy cle from 7 am to 7 pm. Animals were fed Certified Purina Chow (Ralston Purina, St. Louis, MO) and municipal water ad libitum. Intravenous injections. Male F344 rats were anesthe tized with ketamine/xylene and cannulas were im planted the day before the exposures (minimum 18 hr recovery time). MC was dissolved in heparinized rat plasma at a concentration sufficient to give doses of8.84, 25.6, and 47.0 mg/kg when volumes of 1.0 ml plasma/ kg were injected into an indwelling femoral cannula of rats. Blood samples (0.1 ml) were collected from an in dwellingjugular cannula and immediately transferred to a sealed vial containing 1.0 ml hexane. After 45 min ex traction, aliquots of the hexane layer were analyzed for MC by gas chromatography with an electron capture de tector. Oral gavage (MC in water). Solutions of MC in water were prepared by stirring overnight in a closed glass ves sel. The concentration of MC in the water was deter mined by gas chromatography before administration to each rat, and was in the range of0.8-1.0 mg/ml. MC was administered to rats (mean weight 250 g) in a volume of 4.0 ml to give a dose of 14.2 1.9 mg/kg (n = 6). Blood samples were collected from indwelling/ugit/arcannulas and analyzed for MC by gas chromatography as outlined above. Drinking-water exposure. This exposure was carried out essentially as described by Frantz and Watanabe (1983). Animals were placed in all-glass metabolism cages* pi re* the drawn satura night, substi had b 48 hr Ti ll, at: the s tion drink mate! absor amin collet watei end i (2.70 X 10 Sa: sure. hom anal; Pl. logic And MC bloo arel . ies c Me the t part difff = 9- totl pria 1 take as a live mo inU dat net rof< rep the 5r int <u "si ric su  D METHODS | 1,1,1 -trichloroethane rganic Chemicals De al Co. This material was ahy (7.5% Oronite NIW \ on 80/100 mesh Chrommed temperaturegradiZ and finishing at 130*C. more than 99.8% MC. it No. 1119-293, (sp act ed from New England iterial was repurified beiromatography (10% SP jthermal 100'C). The racation was found to be ^ animals, asaturated sol4C]MC in water for 8 hr of the MC used in these 'i/mmol and the concen10+3 dpm/ml). s (weight range 240-260 River Breeding Laboraimated to the laboratory placed on test. The ani;teel cages in rooms de relative humidity conla 12 hr/day lighting cyvere fed Certified Purina iis, MO) and municipal F344 rats were anesthecannulas were imfcs (minimum 18 hr in heparinized rat ent to give doses of8.84, times of 1.0 ml plasma/ lingfemoral cannula of re collected from an inmediately transferred to exane. After 45 min ex layer were analyzed for i an electron capture de lations of MC in water ;ht in a closed glass vesn the water was deterefore administration to ).8-1.0 mg/ml. MC was it 250 g) in a volume of 9 mg/kg (n = 6). Blood veilingjugular cannulas matography as qutlined > exposure was carried Frantz and Watanabe i all-glass metabolism METHYLCHLOROFORM PHARMACOKINETICS 187 cages designed to allow separate collection of urine, feces, expired air, and CO;. Animals were allowed to acclimate to the cages for 32 hr and then water and food were with drawn for 8 hr (from 4 PM to 12 midnight). Food and a saturated solution of [l4C]MC were presented at 12 mid night. At 8 AM the following morning, fresh water was substituted for the solution of [14C]MC and the animals had free access to food and water until they were killed 48 hr later (56 hr after beginning the exposure). The solution of [ l4C]MC was administered with a glass water bottle which contained two stainless-steel balls in the sipper tube in order to eliminate leakage or evapora tion of MC during periods when the animal was not drinking. Air was drawn through the cages at approxi mately 500 ml/min. Activated charcoal traps served to absorb (,4C]MC and ethanolamine traps (3 parts ethanolamine to 7 parts propylene glycol monomethyl ether) collected [1 *C]C02. The activity ofthe [,4C]MC drinking water (dpm/ml) was determined at the beginning and end of the experiment and remained relatively constant (2.70 0.23 X 10+3 dpm/ml at the start; 2.62 0.24 X I0+3 dpm/ml at the end; n = 4 in both cases). Samples of urine and feces were collected during expo sure, and samples of liver, kidney, fat, skin, and carcass homogenate were collected at the end of exposure for analysis of radioactivity. Physiological modeling. A four-compartment physio logical model similar to that developed by Ramsey and Andersen (1984) was used to describe the behavior of MC in rats, mice, and humans. Tissue volumes and blood and airflow rates employed in these simulations are listed in Table 1. The three species used in these stud ies differ widely in the percentage of fat in their carcass (Mouse - 4%, Rat = 7%, Human - 23%). Consequently, the percentage ofcardiac output directed to the fat com partments was adjusted between species to reflect the different amounts offat: Mouse = 2%, Rat * 5%, Human = 9%. To maintain mass balance in the model, the flows to the richly perfused compartment were adjusted appro priately (Table 1). To adapt this model for drinking-water exposures, up take of MC from the gastrointestinal tract was simulated as a zero-order process depositing MC directly into the liver compartment. This mathematical form of the model assumes rapid absorption of MC from the gastro intestinal tract. This assumption is consistent with the data ofWithey et al. (1982) who studied the pharmacoki netics of four similar solvents (methylene chloride, chlo roform, 1,2-dichloroethane, and trichloroethylene) and reported that following gavage with aqueous solutions of these materials, peak blood levels were reached in about 5 min. This form of the model also assumes that zero-order input is a good approximation of the drinking process (i.e., that the animals consume a large number of small "sips" ofthe dosing solution throughout the exposure pe riod). Pharmacokinetic analyses conducted by others suggest that this is a reasonable assumption (NAS, 1986). Simulation of bolus gavage results was conducted as described by Ramsey and Andersen (1984). Absorption from the gastrointestinal tract was assumed to be a firstorder process, with a rate constant of 1.25 hr-1, and all material absorbed from the gastrointestinal tract was de posited directly into the liver compartment. Partition coefficients. Tissue/air partition coefficients were determined for rat blood, liver, fat, and muscle us ing the vial equilibration technique of Sato and Nakajima (1979). Measurements were made at Wright Patter son AFB and were generously supplied by M. L. Gargas (personal communication). Tissue/blood partition co efficients for rat fat, muscle, and liver were then calcu lated by dividing the measured tissue/air partition co efficients by the blood/air coefficient for rat blood. Tis sue/blood partition coefficients for the slowly perfused and rapidly perfused groups oftissues were assumed to be equal to the tissue/blood partition coefficients for muscle and liver respectively. Blood/air partition coefficients for human and mouse blood were also determined by M. L, Gargas at Wright Patterson AFB (personal communication). Tissue/blood partition coefficients for human and mouse tissues were calculated by dividing the measured or estimated rat tissue/air partition coefficient by the blood/air partition coefficient for humans and mice respectively. All parti tion coefficients used in these simulations are listed in Table 1. Metabolic rate constants (rat). Schumann et al. (1982a) exposed rats to 150 and 1500 ppm of [14C]MC for 6 hr. Samples ofexpired air and excreta were collected for 66 hr postexposure in an apparatus which separated [l4C]MC from radioactive metabolites. After 66 hr the rats were killed and the levels of radioactivity remaining in the carcass (assumed to be all metabolites) were deter mined by combustion analysis. This procedure provides a quantitative measure of the total amounts of MC me tabolites formed from [,4C]MC and eliminated postex posure. However, since samples of urine and expired air were not collected during the inhalation exposure, any radioactive metabolites eliminated during this period would have been lost This would lead to an underesti mation of the overall rate of metabolism. In order to compensate for this underestimation, data from the drinking water studies were used to derive a correction factor. In the drinking-water study, water containing [14C]MC was provided to the animals for 8 hr and samples were collected during this period and for an additional 48 hr postexposure. Animals were observed to drink regularly throughout this period. Forty-one percent of the total metabolites were collected during the exposure period, and 59% of the total metabolites were collected after ex posure. Since the experimental conditions were similar to those used in the inhalation exposures, it was assumed that this procedure would give a reasonable approxima tion of the percentage of total metabolites recovered 188 REITZ ET AL. postexposure in the inhalation experiments. Conse quently, the postexposure levels of metabolism reported by Schumann et al. (1982a) in inhalation experiments were divided by 0.59 to give an estimate of total metabo lism in the inhalation studies. At this point, the PB-PK model was fully defined ex cept for the metabolic rate constants and Km. Thus it was possible to estimate PTM and Km by a computer ized least-squares optimization procedure which varied the rate constants until the model correctly predicted the corrected amounts of metabolites formed during the ex posures to 150 and 1500 ppm MC. Final values for the enzymatic rate constants were Pm,,C = 0.419 0.095; Km = 5.75 2.49 (mean and standard deviation of esti mates). Details ofthis type ofcomputerized optimization have been reported elsewhere (Agin and Blau, 1982). Computer simulation. Simultaneous differential and algebraic equations describing the movement of MC through the body were formulated as a computer pro gram. Simulations were conducted with the SimuSolv software package which contains routines for numerical integration, graphicsdisplay, and derivation ofmodel pa rameters from experimental data. SimuSolv was devel oped by Agin and Blau (1982) and is commercially avail able from Mitchell & Gauthier Associates.2 RESULTS Rat Simulations Inhalation simulations. After incorpora tion of the appropriate physiological and physical chemical parameters (Table 1), the PB-PK model was used to predict the time course of the venous blood concentrations of MC in rats after exposure to 150 or 1500 ppm MC. Experimental data previously collected by Schumann et al. (1982a) were used to check the predictions, and the results are shown in Fig. 1. The data showed the best agreement with model predictions for blood level measurements made postexposure. Blood levels measured during the exposure were somewhat lower than model predictions at 4,5, and 6 hr, but the predictions were still within a factor of 2 of the experimental data (Fig. 1). 2 Mitchell & Gauthier Associates, 73 Junction Square Drive, Concord, MA 01742. TABLE 1 Parameters Used in the Physiologically Based Pharmacokinetic Model for Methylchloroform" Human Rat Mouse Weights Body wt (kg) Liver Rapidly perfused Slowly perfused Fat Flows (liters/hr) Alveolar vent Cardiac output 83.0 3.1% 3.7% 61.1% 23.1% 348.0 348.0 0.215 4.0% 5.0% 75.0% 7.0% 0.029 4.0% 5.0% 78.0% 4.0% 5.11 1.26 5.11 1.26 Percentage ofcardiac output Liver Rapidly perfused Slowly perfused Fat Partition coefficients Blood/air Liver/air Rapidly perfused/air Slowly perfused/air Fat/air Biochemical constants vmi*C (ailometric) Km (mg/liter) JUhrl) 24.0 49.0 18.0 9.0 2.53 8.6 8.6 3.15 263. 0.419 5.75 -- 24.0 53.0 18.0 5.0 5.76 8.6 8.6 3.15 263. 0.419 5.75 1.25 24.0 56.0 18.0 2.0 10.8 8.6 8.6 3.15 263. 0.419 5.75 -- " Metabolic parameters for the rat (V__C. K,,) were obtained from the data of Schumann et al. (1982a) by computer optimization. ,,C is an ailometric measure of the maximum velocity of metabolism such that the maximum enzyme rate (fj may be calculated for any size animal according to the equation x (body wtr\ Several other types of data were available from the studies of Schumann et al. (1982a) to check the accuracy of the PB-PK model. These were (1) the total body burden of MC and metabolites after 6 hr ofexposure, (2) the concentration of radioactivity in fat tissue at the end of the exposure, and (3) the concen tration of radioactivity in liver tissue at the end of the exposure. Comparisons between the model predic tions and these additional data are summa- i ologically Based Methylchloro- an Rat Mouse 0.215 0.029 X 4.0% 4.0% % 5.0% 5.0% % 75.0% 78.0% & 7.0% 4.0% 5.11 1.26 5.11 1.26 ntage ofcardiac output 24.0 24.0 53.0 56.0 18.0 18.0 5.0 2.0 1 5.76 10.8 8.6 8.6 8.6 8.6 3.15 3.15 263. 263. 9 0.419 0.419 5 5.75 5.75 1.25 -- >e rat (V^C, Km) were maim et al. (1982a) by ; an allometric measure :tabolism such that the ay be calculated for any tnation ^ - Km,,C iata were available iann et al. (1982a) the PB-PK model, xly burden of MC ifexposure, (2) the vity in fat tissue at nd (3) the concenliver tissue at the '.he model predicdata are summa- METHYLCHLOROFORM PHARMACOKINETICS 189 Fig 1. Blood levels of MC in rats during and following a 6-hr inhalation exposure to ISO ppm (closed circles) or 1500 ppm (open circles) of MC. Values predicted by computer simulations are shown as a solid line(s). rized in Table 2. Overall, the ratio of pre dicted to actual data for these two exposures ranged from a low of 0.72 to a maximum of 1.92 and the mean ratio was 1.28 0.48 (standard deviation; Table 2). Intravenous administration. The PB-PK model was then used to attempt a dose route extrapolation: inhalation exposure to intra venous injection. Venous blood concentra tions were measured in groups ofanimals (six per group) injected with MC dissolved in hep arinized plasma. Intravenous injection was simulated in the model as a rapid infusion into pooled venous blood (about 1 min in duration). Three doses were studied: 8.8, 26, and 47 mg/kg. The only changes made to the PB-PK model for this simulation were in the equations describing entry of MC into the body. Experimental data and model pre dictions for 4 hr postinjection are shown in Fig. 2. Bolus gavage (MC in water). A third route of administration (bolus gavage) was investi gated. Six rats were administered a dose of 14.2 mg/kg MC dissolved in water, and serial blood samples were taken for analysis at vari ous times afterwards. The equations in the PB-PK model were modified to reflect the new route of exposure. Uptake of MC was simulated as a first-order process with a rate constant (Ka) equal to 1.25 hr-1. The model predictions and experimental data for this route are shown in Fig. 3. Drinking-water exposure. A variant of the bolus gavage route was consumption of MC in drinking water over an extended period of time. Animals were presented with water containing radioactive MC for 8 hr and sam ples of urine, expired air, and selected tissues were analyzed for radioactivity at various times thereafter. The average water con sumption of the rats (mean wt = 250 g) dur ing the 8 hr in which they had access to the treated water was 8.1 3.8 ml, corresponding to an average dose of 116 mg/kg. Overall recovery of radioactivity in these experi ments (based on water consumption) was 95.2 4.33% (standard deviation). Entry of MC into the animal was simulated as a zeroorder process depositing MC in the liver com partment at a constant rate for 8 hr. Other than the changes to the input equations, no parameters were changed in the PB-PK model for drinking-water simulation. Simu lated rates of MC elimination in exhaled air and experimental data (radioactivity recov ered from charcoal traps) are shown in Fig. 4. In addition, the use of [l4C]MC made it possible to predict the extent of MC metabo lism for this route of metabolism. The model predicted that 5.49 /xmol of MC (about 2% the ingested MC) would be metabolized, 8.19 jimol (about 3% ofthe ingested MC) was actually recovered in urine and C02 (Ta ble 2). Mouse Simulations Inhalation simulations. The PB-PK model was then used for interspecies extrapolation by predicting the pharmacokinetic behavior of MC in male B6C3FI mice after inhalation exposure. Model predictions were compared to experimental data gathered by Schumann et al. (1982a). Other than adjusting the ap propriate physiological and biochemical pa rameters, no changes were made in the PBPK model for simulation of the mouse expo sures. 190 REITZ ET AL. TABLE 2 Comparison of Predicted and Observed Values for Selected Parameters from Inhalation Expo sures and Drinking Water Exposures in Young Rats and Young Mice (2-3 months of Age) and Human Volunteers0 End exposure blood level (mg/liter) Body burden at 6 hr (ftmol) Cone in fat (Mmol/liter) Cone in liver Oimol/liter) Observed Young rat-- 130 ppm 2.64 33.0 724.0 68.2 Predicted 4.37 25.5 1264.0 48.9 Ratio (pred/obs) 1.66 0.77 1.75 0.72 End exposure blood level (mg/liter) Body burden at 6 hr (/imol) Cone in fat (/unol/liter) Cone in liver (jimol/liter) Young rat-- 1500 ppm 23.1 263 8400 503 44.4 237 12800 500 1.92 0.90 1.52 0.99 Young rat--drinking water (116 mg/kg) Amount metabolized Oimol) 8.19 5.49 0.67 End exposure blood level (mg/liter) Body burden at 6 hr (pmol) Amount metabolized (jimol) Cone in fat (pmol/liter) Cone in liver (Mmol/liter) Young mouse-- 150 ppm 9.27 4.97 1.10 1330 75.7 8.54 3.30 1.02 1570 50.6 0.92 0.66 0.93 1.19 0.67 Young mouse--1500 ppm End exposure blood level (mg/liter) Body burden at 6 hr Oimol) Amount metabolized (*imol) Cone in fat (/unol/liter) Cone in liver (nmol/liter) 111.0 39.5 2.01 16200 631 88.1 25.8 1.90 16100 525 0.79 0.65 0.94 0.99 0.83 Amount metab. 33 ppm (^mol) Amount metab. 330 ppm (/imol) Human 35 and 350 ppm 32.4 246 32.2 236 0 Inhalation data are from Schumann etal. (1982a) and Nolan et ai. (1984). 0.99 0.96 The predicted and observed venous blood concentrations (collected at the orbital sinus) in mice exposed to either 150 or 1500 ppm of MC are shown in Fig. 5. The PB-PK model predicted that MC would be eliminated from the mouse much more rapidly than the rat, and this is consistent with the observed data (Figs. I, 5). This is also consistent with the observation of Schumann that elimination half-lives for MC in mice were 5- to 10-fold less than the corresponding half-lives in rats (Schumann et al. 1982a). Several additional sources of experimental data were available for assessing the consis- lOfr O' o0 1 IK 0H oo Fig. 2 nous inj (closed c rat plasn concent! while m plasma/1 simulate tencyc and p< body b tration and th ered ai predic 5tc 6 ID O' g 1OO3 0! FlG.. with a s 1.9 r proxim by com sorptio a first-c ALATION EXPOe) and Human Ratio (pred/obs) 1.66 0.77 1.75 0.72 1.92 0.90 1.52 0.99 0.67 0.92 0.66 0.93 1.19 0.67 0.79 0.65 0.94 0.99 0.83 0.99 0.96 ;istent with the iat elimination re 5- to 10-fold lalf-lives in fats if experimental sing the consis- METHYLCHLOROFORM PHARMACOKINETICS 191 Time (hr) FIG. 2. Blood levels of MC in rats following intrave nous injection of 8.8 mg/kg (open circles), 26 mg/kg (closed circles), or 47 mg/kg (crosses) ofMC dissolved in rat plasma. MC was dissolved in heparinized plasma at concentrations calculated to give the indicated doses while maintaining a constant vehicle volume of 1 ml plasma/kg of body wt. Values predicted by computer simulations are shown as a solid line(s). Time (hr) FtG. 4. Rate of elimination of MC[MC] in exhaled air (mg equivalents of MC/hr) during and following ad libi tum exposure ofrats to a solution of[14C]MC in drinking water. The average water consumption ofthe rats (mean wt = 250 g) during the 8 hr in which they had access to the treated water was 8.1 3.8 ml, corresponding to an average dose of 116 mg/kg. Values predicted by com puter simulations are shown as a solid line(s). tency ofthe PB-PK model for mice. Observed and predicted values of the end exposure body burden, the end exposure liver concen tration, the end exposure fat concentration, and the total amount of metabolites recov ered are summarized in Table 2. The ratio of predicted to actual data ranged from a low of 0.65 to a high of 1.19 and the mean ratio was 0.86 0.16 (standard deviation; Table 2). Human Exposures The PB-PK model was used to predict the disposition of inhaled MC in humans. Model predictions were compared to data gathered by Nolan et al. (1984) at 35 and 350 ppm MC. Experimentally observed and simulated concentrations of MC in venous blood are shown in Fig. 6a. Concentrations of MC into blood was somewhat lower than predicted Flc. 3. Blood levels ofMC in rats following oral gavage with a solution ofMC in water. The dose ofMC was 14.2 1.9 mg/kg (standard deviation), administered in ap proximately 4.0 ml water/kg body wt. Values predicted by computer simulations are shown as a solid !ine(s). Ab sorption from the gastrointestinal tract was simulated as a first-order process with a rate constant of 1.25 hr"1. Time (hr) Fig. 5. Blood levels of MC in mice during and follow ing a 6-hr inhalation exposure to 150 ppm (closed circles) or 1500 ppm (open circles) of MC. Values predicted by computer simulations are shown as a solid line($). SL 036360 192 REITZ ET AL. Fig. 6a. Concentration of MC in exhaled air (mg/liter) during and following a 6-hr inhalation exposure of hu man volunteers to 35 ppm (closed circles) or 350 ppm (open circles) ofMC. Values predicted by computer sim ulations are shown as a solid line($). during exposure (insert) but was well de scribed for the post exposure period (main figure). Predicted and observed concentrations of MC in expired air are shown in Fig. 6b. Con centrations of MC in expired air were well simulated by the model during exposure (in sert). The model predicted a slightly more rapid decline in the concentration of MC in expired air than was observed during the pe riod of 10-30 hr, but the final slow elimina tion phase (30-240 hr) was well simulated by the model (main figure). Predicted and ob served values ofexpired air and venous blood concentrations were both proportional to ex posure concentration throughout the range of 35-350 ppm MC. The PB-PK model predicted that 4.30 mg equivalents of MC would be metabolized during the 240 hr following a 6-hr exposure to 35 ppm MC, and 4.32 mg equivalents of MC metabolites were actually recovered by Nolan et al. (1984). Similarly, the model pre dicted that 31.5 mg equivalents of MC tabolites would be formed during the 240 hr following a 6-hr exposure to 350 ppm MC, and 32.9 mg were actually recovered (Ta ble 2). Older Animals Old rats, inhalation simulations. Schu mann et al. (1982b) also studied the pharma cokinetics of MC in control and exposed (1500 ppm, 6 hr/day, 5 days/week) male rats obtained from a study of the chronic inhala tion toxicity of MC. The age of these animals was approximately 18.5 months, and the av erage body weight was 481 g. They reported that the disposition of a single dose of [14C]MC in the treated and control animals was nearly identical (i.e., that chronic preex posure to unlabeled MC had not altered phar macokinetic behavior through induction of enzymes, alteration of renal function, etc.). However, the disposition of MC in old ani mals differed significantly from the disposi tion previously observed in younger animals (Schumann et al., 1982a). The PB-PK model was used to predict the rate of elimination of MC in exhaled air in untreated (control) 18.5-month-old male rats following a 6-hr exposure to 1500 ppm MC. Experimental data and the initial simulation are shown in Fig. 7a. The PB-PK model pre dicted a much more rapid decline in the rate of MC exhalation when the physiological pa rameters chosen for young rats were used (heavy line. Fig. 7a). Lutz et al. (1977) improved the accuracy of their simulation of polybrominated biphenyl compounds in rats by increasing the relative Fig. 6b. Concentration of MC in venous blood (mg/ liter) during and following a 6-hr inhalation exposure of human volunteers to 35 ppm (closed circles) or 350 ppm (open circles) ofMC. Values predicted by computer sim ulations are shown as a solid line(s). Fig. 7a. Rate | (mg equivalent months) rats t< experimental d Simulated vain body weight (i.i | young rats) are light solid line compartment e I proportion o compartmer a second cc I mine whethc give^ktter l8.;^Bpnth-. lowed to var | ment (VF) fr ing simulatk j the fraction slowly perfu: constant pen tissue (91%o improved fib tained after line, Fig. 7a). PB-PK modi i exhaled air d from 7% of weight. The model was then use of experimen ies of Schun eluded end e; sure concent "03636^ rntrol and exposed j^s/week) male rats chronic inhala age of these animals months, and the av31 g. They reported a single dose of ind control animals that chronic preexlad not altered pharrough induction of inal function, etc.), i of MC in old aniy from the disposiin younger animals > used to predict the C in exhaled air in month-old male rats ; to 1500 ppm MC. le initial simulation PB-PK model prei decline in the rate ne physiological paing rats were used wed the accuracy of ^minated biphenyl ^^sing the relative D 1 tti 200 240 (hr) C in venous blood (mg/ it inhalation exposure of losed circles) or 350 ppm xlicted by computer sime(s). METHYLCHLOROFORM PHARMACOKINETICS 193 A Time (hr) Time (hr) Fig. 7a. Rate ofelimination of [14C]MC in exhaled air (mg equivalents of MC/hr) following exposure ofold (18 months) rats to 1500 ppm of [l4C]MC for 6 hr. Actual experimental data are shown as open circles on the plot Simulated values with a fat compartment equal to 7% of body weight (i.e., simulation based on the parameters for young rats) are shown by the heavy solid line, while the light solid line shows the model predictions with a fat compartment equal to 18% ofbody weight. Fig. 7b. Rate ofelimination of[l4C]MC in exhaled air (mg equivalents ofMC/hr) following exposure ofold (18 months) mice to 1500 ppm of [l4C]MC for 6 hr. Actual experimental data are shown as open circles on the plot Simulated values with a fat compartment equal to 4% of body weight (i.e., simulation based on the parameters for young mice) are shown by the heavy solid line, while the light solid line shows the model predictions with a fat compartment equal to 18% ofbody weight proportion of body weight allocated to the fat compartment. Consequently, we conducted a second computer optimization to deter mine whether a similar modification would give a better simulation of the data from the 18.5-month-old rats. The computer was al lowed to vary the volume of the fat compart ment (VF) from 0 to 30% ofbody weight dur ing simulations. Corresponding decreases in the fraction of body weight assigned to the slowly perfused compartment maintained a constant percentage ofthe carcass as perfused tissue (91% of body weight perfused). Greatly improved fits to the exhaled air data were ob tained after the second optimization (light line. Fig. 7a). This optimization indicated the PB-PK model was most consistent with the exhaled air data when the VF was increased from 7% of body weight to 18% of body weight. The model with the larger fat compartment was then used to predict several other types of experimental data available from the stud ies of Schumann et al. (1982b). These in cluded end exposure body burden, end expo sure concentration of MC-derived radioac tivity in the liver, end exposure concentration ofMC-derived radioactivity in the fat, and to tal amount of radioactive metabolites col lected in 66 hr subsequent to the 6-hr expo sure. The comparisons between the PB-PK model prediction and the actual data are summarized in Table 3. The predictions of the PB-PK model for old rats with the original fat compartment (7% of body weight) are summarized in col umn 2 ofTable 3. The model underpredicted the end exposure body burden and amount metabolized, but overpredicted the concen tration of MC in fat tissue. After increasing the size of the fat compartment to 18% of body weight, the PB-PK model predicted lower concentrations of MC in fat tissue, and increased body burdens and amounts metab olized, giving a better simulation of the rat data. Old mice inhalation simulations. Schu mann et al. (1982b) also collected data in old male B6C3F1 mice (18.5 months of age) ob tained from a study of the chronic inhalation toxicity of MC in mice. As was observed with the rats, the pharmacokinetics of MC in old 194 REITZ ET AL. TABLE 3 Comparison of Observed and Predicted Values for Selected Parameters from Inhalation Exposures in Old Rats and Old Mice (18.5 months of Age)" Actual Before reoptimization After reoptimization Body burden at 6 hr (jimol) (Ratio) Amount metabolized (pmol) (Ratio) Cone in fat Oimol/liter) (Ratio) Cone in liver (nmol/liter) (Ratio) Old rat-- 1500 ppm 744 31.8 5,685 606 490 (0,66) 17.5 (0.55) 11,700 (2.06) 494 (0.82) 638 (0.86) 22.1 (0.69) 6200 (1.09) 454 (0.75) Body burden at 6 hr (/imol) (Ratio) Amount metabolized (tmol) (Ratio) Cone in fat (Mmol/liter) (Ratio) Cone in liver (junol/liter) (Ratio) Old mouse-- 1500 ppm 149 13.6 10,600 1,790 35.1 (0.24) 2.61 (0.19) 15,900 (1.50) 532 (0.30) 71.4 (0.48) 4.11 (0.30) 8790 (0.83) 491 (0.27) Experimental data are from Schumann et al, < 1982b). Physiological parameters for rats and mice before computer optimization are listed in Table 1, except for body weight which was 481 g for the rats and 39.8 g for the old mice. After optimization, the size ofthe fat compartment was 18.6 and 18.1% ofbody weight in rats and mice, respectively. unexposed (control) mice differed signifi cantly from the pharmacokinetics of MC in young control mice. When the PB-PK model utilized to de scribe MC disposition in young mice was used to predict the rate ofelimination of MC in exhaled air, a poor description of the ex perimental data was obtained (heavy line. Fig. 7b). However, when the size of the fat compartment was increased from 4% of body weight to 18% of body weight, a much better description of the experimental data was ob tained (light line, Fig. 7b). When the model with 4% fat was used to simulate additional parameters measured in older mice, it underestimated the body bur den, the amount metabolized, and the con centration of radioactivity in liver tissue and overestimated the concentration of radioac tivity in fat (Table 3). When the size of the fat compartment was increased to 18% of body weight, the description of the experimental data was improved, but the model still underpredicted the amount metabolized and the concentration of radioactivity in liver tissue. DISCUSSION Versatility ofthe PB-PK Model The first objective of these studies was to demonstrate the extent to which incorpora tion of physiological and biochemical princi ples into pharmacokinetic modeling would allow extrapolation between different routes of administration and different species. Ac cordingly, we have not conducted extensive "curve fitting" procedures to try to improve Inhalation After reoptimization 638 (0.86) 22.1 (0.69) 6200 (1-09) 454 (0.75) 71.4 (0.48) 4.11 (0.30) 8790 (0.83) 491 (0.27) i mice before computer >9.8 g for the old mice, and mice, respectively. to 18% of body the experimental : model still under;tabolized and the /ity in liver tissue. ON odel esc studies was to which incorporaiochemical princi: modeling would en different routes :erent species. Acnducted extensive to try to improve METHYLCHLOROFORM PHARMACOKINETICS 195 the appearance of the data presented here, and it is obvious that the extrapolations were not perfect in every case. However, since we are proposing that these techniques may be useful in the area of risk assessment, it is im portant to keep in mind the level of precision which currently exists in this process. Cancer risk estimations produced from the same sets of data often differ by several orders of magnitude, depending upon the assump tions used about the shape of the dose-re sponse curve at low concentrations. Further more, it is not uncommon for bioassays ofthe same material in the same species to differ by as much as an order of magnitude in their end points (tumor incidences). Consequently, we felt that if our integrated PB-PK model could predict relevant pharmacokinetic parameters for MC in different species and dose routes within a factor of 2 or 3, we would have dem onstrated its potential for increasing the accu racy of current risk estimation procedures. The construction ofthe PB-PK model used here was relatively straightforward. Physio logical parameters for the various species have been published in the scientific litera ture (Davis and Mapleson, 1981; Caster et al., 1956; ICRP, 1975), and partition coefficients were measured according to published proce dures (Sato and Nakajima, 1979). The most difficult task was the estimation of the rate constants for in vivo metabolism. MC is not extensively metabolized in any of the species studied (Schumann et al., I982a,b; Nolan et al., 1984). Consequently the gas uptake technique used by Gargas et al. (1988) for estimation of in vivo metabo lism with other compounds was not suitable for this material, since uptake depends upon metabolism once tissue loading has occurred. Furthermore, a variety of metabolites are formed from biotransformation of MC (trichloroethanol, the glucuronide oftrichloroethanol, and trichloroacetic acid). The pro duction of multiple metabolites, each of which is present at very low levels, makes it difficult to estimate total metabolism by chemical analysis. However, radioactive metabolites of MC are easily separated from [,4C]MC and may be readily quantified. Consequently, we have used available data on the total levels of ra dioactive metabolites obtained in balance studies with rats exposed to [l4C]MC to esti mate the rates of metabolism in this species. Levels of metabolites produced during expo sure and postexposure were calculated with the PB-PK model, which contains the struc ture necessary to deal with the complexities of postexposure metabolism. We then em ployed allometric scaling procedures to esti mate metabolic rate constants in the other species. Since the metabolic rate constants were obtained from a limited data set (two ex periments in rats with four animals/experi ment), there is obviously room for error in these parameters. The magnitude of the pos sible error may be roughly estimated from the standard deviations for the estimates of V__Cand Km, which were 23 and 43% ofthe mean, respectively. Nevertheless, it is encour aging that once the metabolic rate constants were obtained in this manner, they were able to accurately predict metabolism in two other species exposed to MC vapor (mice and hu mans) as well as in rats exposed to MC in drinking water (Table 2). It must be emphasized that since metabo lism plays a minor role in the elimination of MC in the various species, the behavior of MC itself (as distinct from MC metabolites) is almost entirely dependent upon the solu bility in various tissues (partition coefficients) and the physiology of the various species. Small errors in the values chosen for the met abolic rate constants will have little effect on the overall disposition of MC. Schumann et al. (1982a) utilized a conven tional, data-based pharmacokinetic model to describe the time course ofMC in rats. In this approach, the venous blood time/concentra tion curve for rats was actually described bet ter by a simple two-compartment open model with elimination from the central compartment (Schumann et al., 1982a) than the present PB-PK model. However, when 196 REITZ ET AL. we attempted to use this model to describe the time course of MC in the venous blood of mice, a very poor fit was obtained (data not shown), pointing out the difficulties of using conventional models for extrapolating to hu mans. In contrast, once a PB-PK model had been developed for rats, it immediately pro vided a good description of the blood level data obtained in mice (Fig. 5) as well as hu mans (Fig. 6a). Furthermore, the development of a PB-PK model allowed us to predict other types of data of potential toxicological significance. For example, the concentration of MC in liver tissue at the end of inhalation exposures to MC was well described in both rats and mice (Table 2). Since the liver has been iden tified as the target organ in both rats and mice during long-term animal bioassays (Rampy et al., 1978; Quast et al., 1984), this finding has obvious significance. Another advantage of PB-PK models is their ability to provide quantitative descrip tions of material administered by a variety of dose routes. In this particular case, the phar macokinetics of MC in rats after inhalation (Fig. 1), intravenous injection (Fig. 2), bolus gavage (Fig. 3), and drinking-water admin istration (Fig. 4) could be described with a single, integrated model. Conventional databased pharmacokinetic models lack the ca pacity for this type of route to route extrapo lation. tissue are 20-100 times higher than in other tissues. Consequently, the increased size of the fat compartment in older, sedentary ani mals would be expected to increase the amount of MC taken up by older animals ex posed to a given concentration of MC, and the elimination of MC from these older ani mals should occur more slowly. These predic tions ofthe PB-PK have been experimentally verified in the data of Schumann et al. for both rats and mice (Figs. 7a, 7b; Tables 2,3). Most ofthe changes in the pharmacokinet ics of MC in older rats were correctly pre dicted when the size of the body fat compart ment was increased (Table 3). However, the situation appears to be somewhat more com plex with B6C3F1 mice. The increased halflife of MC and lower concentration of radio activity in the fat compartment of older mice were correctly predicted by a PB-PK model with a larger fat compartment. However, the predicted body burden and amount of MC metabolized in these mice were still much lower than actually observed by Schumann et al. (1982b) (Table 3). It appears that addi tional factors, such as increased metabolic ca pacity, may have to be considered in evaluat ing the kinetics of MC in older mice. The na ture ofthese factors remains to be elucidated. Relating Animals Studies to Human Expo sures Pharmacokinetics in Older Animals A variety of physiological and biochemical changes may occur in test animals during the course of a lifetime bioassay, where animals are started on test a few weeks after weaning and continue on test until nearly the end of their natural lives. Development of a PB-PK model offers a mechanism whereby the con sequences of these changes may be quantita tively evaluated. For example, the PB-PK developed for MC indicates that the concentrations of MC in fat MC has been studied in two long-term in halation bioassays at the Toxicology Re search Laboratory of the Dow Chemical Co. Animals were exposed to MC for 6 hr/day, 5 days/week for up to 2 years. MC exposure was not associated with increases in the inci dences of either benign or malignant tumors in these studies. Reversible microscopic alter ations in liver tissue were noted in both rats and mice, but cytotoxicity and necrosis were not seen. The no observed adverse effect lev els (NOAEL) for the cellular changes estab lished in these studies were 873 ppm in rats and 1500 ppm in mice (Rampy et al., 1978; #, E, VaLUI AMOl Rat Mous Hum. *T mice in uni hum- Qua port post wee wov ofN (i.e. terr che seq' cor tim gati sur cus cee Nt* t M( tio cht mi of ( ligher than in other le increased size of r, sedentary ani- #to increase the Ty older animals exiration of MC, and om these older aniiowly. These predic>een experimentally -humann et al. for 7a, 7b; Tables 2, 3). .he pharmacokinetwere correctly pre` body fat comparte 3). However, the mewhat more comfhe increased halfentration of radioment of older mice iy a PB-PK model lent However, the id amount of MC ;e were still much -xi by Schumann et appears that addi;ased metabolic caisidered in evaluat^Wer mice. The na^o be elucidated. to Human Expo- two long-term in; Toxicology Re>ow Chemical Co. AC for 6 hr/day, 5 ars. MC exposure :reases in the incimalignant tumors microscopic alternoted in both rats and necrosis were adverse effect levar changes estab- 875 ppm in rats -impy et al., 1978; methylchloroform pharmacokinetics 197 TABLE4 Comparison of the Lifetime Average Concentration of MC in the Liver (ACL, Dose Surrogate for MC Effects on Liver) for Rats and Mice at the No Observed Adverse Effect Levels (NOAEL) with Values of the Same Dose Surrogate in Humans Consuming 2 liters/day of Water Containing Small amounts of MC" Rat Mouse Human Cone (ppm) 875 1500 0.001 0.003 0.010 0.030 0.100 0.300 Route Inhalation Inhalation Water Water Water Water Water Water Average ACL (/imol/liter) 28 95 3.4 x I0'5 1.0 X (O'4 3.4 X 10-4 1.0 x lO'1 3.4 X I0*3 1.0 x lO"1 Safety factor relative To mouse -- 2.8 X 1046 9.5 X 10+5 2.8 X 1045 9.5 X 1044 2.8 X 1044 9.5 X 10+3 To rat _ -- 8.2 X 1045 2.8 X 1045 8.2 x 1044 2.8 x 1044 8.2 X 1043 2.8 X 1043 4 The NOAEL for the rat was 875 ppm, 6 hr/day, 5 days/week for I year (Rampy et al., 1978) and the NOAEL for mice exposed 6 hr/day, 5 days/week for 2 years was 1500 ppm (Quast et al., 1984). All dose surrogates ate reported in units of/imol/1iter. "Safety factors" are calculated by dividing the calculated animal dose surrogate by the predicted human dose surrogate. Quast et al., 1984). McNutt et al. (1975) re ported similar effects in the liver of mice ex posed to 250 ppm continuously for several weeks. Since the liver is the target organ in both animal species, it is assumed that the liver would be the site where any adverse effects would occur in humans. Because the effects ofMC on liver tissue were mild and reversible (i.e., without necrosis), it is not possible to de termine whether they were caused by parent chemical or more reactive metabolites. Con sequently we have chosen to use the average concentration of MC in the liver over the life time of the animal (ACL) as a "dose surro gate" for MC. Rationale for selection of dose surrogates with PB-PK models has been dis cussed in more detail by Andersen in the pro ceedings of a workshop conducted at the NAS (1987). Selection of this type of dose surrogate for MC is probably a health protective assump tion. Reactive metabolites, rather than parent chemical, are responsible for the toxicity of many other halogenated solvents, and levels of the enzyme which produces metabolites of MC appear to be lower in humans than in the small rodent species. For example, we can use the allometric equation to calculate that the concentration of MC-metabolizing enzyme (units of enzyme per liter of tissue) is about 13-fold lower in human liver than it is in mouse liver. Consequently, if we had chosen to use production of reactive metabolites as the dose surrogate rather than ACL, the esti mated risk to humans would be considerably lower than with the present procedure. Once the dose surrogate was chosen, the PB-PK model was used to calculate the ACL for rats and mice at the NOAEL in the long term inhalation studies at Dow Chemical. The ACL for rats was about 55.2 /tmol/liter during the 1-year exposure period, or about 27.6 nmol/liter averaged over the 2-year life span. The corresponding value for B6C3F1 mice was 95.1 /tmol/liter (Table 4). Small quantities of MC (typically 1-10 ppb, highest observed 300 ppb) have been de tected in some finished drinking-water sup plies. However, there are no long-term stud ies in animals in which MChas been admin istered by this route, and even if there were, 198 REITZ ET AL. the limited solubility of MC would limit the dose that could be given to the animals. Con sequently, there is a need for a mathematical procedure to quantitatively relate the "inter nal dose" in humans drinking such water to the "internal dose" in animal inhalation studies. This was readily accomplished with the PB-PK model developed in these studies. ACLs associated with human consump tion of water containing the indicated con centrations of MC are summarized in Table 4. It is obvious from this table that the ACLs in humans consuming this water are much lower than the ACLs in animals during the chronic inhalation studies. For example, the ACL in humans exposed to 10 ppb of MC in drinking water are 81,000-fold lower than the ACL at the NOAEL in the rat, and 270,000fold lower than the ACL in the mouse (Table 4, columns 4 and 5). Simulations with the PB-PK model re vealed that "steady-state" levels of MC in the human liver are achieved within a relatively short period. The ACL after 100 days of drinking water containing 10 ppb MC was only 2.7 times higher than the ACL after a single day of consuming such water. This in dicates that the risk to humans will not in crease disproportionately after long periods of consumption, as might be the case with compounds such as lead or mercury. There are bound to be many uncertainties in human risk estimations derived from ani mal studies. Although they cannot address all of the uncertainties inherent in this process, PB-PK models offer promise for improving the basis for high dose/low dose, interspecies, and dose route extrapolations. It should be noted, however, that PB-PK models are most useful when they are combined with indepen dent studies elucidating the mechanism of toxicity so that appropriate dose surrogates may be chosen. REFERENCES Agin, G. L., and Blau, G. E. (1982). Application of DACSL (Dow Advanced Continuous Simulation Lan guage) to the design and application of chemical reac tor systems. Amer. Inst. Chem. Eng. Symp. Set. No. 214,78, 108-118. Andersen, M. E., Clewell, H, J., Gargas, M. L., Smith, F. a., and Reitz, R. H., (1987) Physiologi cally based pharmacokinetics and the risk assessment process for methylene chloride. Toxicol. Appl. Phar macol 87, 185-205. Andersen, M. E., Gargas, M. L., and Ramsey, J. C. (1984). Inhalation pharmacokinetics: Evaluating sys tem extraction, total in vivo metabolism, and the time course ofenzyme induction for inhaled styrene in rats based on steady-state blood: air concentration ratios. Toxicol. Appl Pharmacol 73, 176-187. Caster, W. O., Poncelet, J., Simon, A. B., and Arm strong, W. D. (1956). Tissue weights of the Rat I. Normal values determined by dissection and chemical methods. Proc. Soc. Exp. Biol. Med. 91,122-126. Davis, N. R., and Mapleson, W. W. (1981). Structure and quantification ofa physiological model ofthe dis tribution of injected agents and inhaled anaesthetics. Brit. J. Anaesth. 53,399-404. Dedrick, R. L., (1973) Physiological pharmacokinetics. J. Dyn. Syst. Meas. Cant. Trans. ASME, September, pp. 255-257. Frantz, S. W,, and Watanabe, P. G. (1983). Tetrachloroethylene: Balance and tissue distribution in Male Sptague-Dawley rats by drinking-water admin istration. Toxicol Appl Pharmacol. 69,66-72. FIserova-Bergerova, V, (1975) Mathematical model ing of inhalation exposure. J. Combust. Toxicol 3, 201-210. Gargas, M. L,, and Andersen, M. E, (1988) A gas phase technique for determining the kinetic constants of chemical metabolism in the rat Toxicologist, 8, 112. Gehring, P. J. (1968). Hepatotoxic potency of various chlorinated hydrocarbon vapors relative to their nar cotic and lethal potencies in mice. Toxicol Appl. Phar macol. 69,66-72. Himmelstein, K. J., and Lutz, R. J., (1979). A review ofthe applications ofphysiologically based pharmaco kinetic modeling. J. Pharmacokinet. Biopharm. 7, 127-145. International Commission on Radiation Protection (1975). Report of the task group on Reference Man, ICRP Publication 23 (W. S. Snyder et al., Eds). Peigamon. New York. Lutz, R. Dedrick, R. L., Matthews, H. B., Eling, T. E., and Anderson, M. W. (1977). A preliminary pharmacokinetic model for several chlorinated biphe nyls in the rat Drug. Metab. Dispos. 5,386-396. McNutt, n. S., Amster, R. L., McConnell, E. E., and Morris, F. (1975) Hepatic lesions in mice after continuous exposure to 1,1,1-trichloroethane. Lab. Invest. 32,642-654. < National Aca. Press, Wasl National Acai Pharmacol Washingtoi emy Press, Nolan, R. J Carty, L. and metab trichloroeti Toxicol. 4, Quast, J. F.. (1984). Ch cologic ant Results of! Rampy, L. \ Gehring, tion studie chloroethy Intemalior W. A. M. L Ramsey, J.P logically b; cokinetics Appl. Phar Sato, A., At