Toxicology Letters, 52 (1990) 141-152 Elsevier
141
TOXLET 02348
In vivo metabolic interactions of benzene and toluene
Karen J. Purcell’, Gregory H. CasoS, Michael L. Gargas3, Melvin E. Andersen3 and Curtis C. Travis1 ‘Ofice of Risk Analysis, Health and Safety Research Division, Oak Ridge National Laboratory, TN 37831, 2Biochemical Toxicology Branch, Armstrong Wright-Patterson
Oak Ridge,
Aerospace Medical Research Laboratories,
Air Force Base, OH 45433 and ‘Chemical Industry Institute of Toxicology,
Research
Triangle Park, NC 27709 (U.S.A.)
(Received 2 November 1989) (Revision received 4 December 1989) (Accepted 18 February 1990) Key worris: Physiologically based pharmacokinetic
modeling; Benzene; Toluence; Co-exposure;
Gas uptake
SUMMARY The metabolic interactions of benzene and toluene co-exposure were investigated in male Fischer rats. A closed recirculated exposure system was used to obtain inhalation uptake curves for individual chemicals as well as for a mixture of the two compounds. Pharmacokinetic parameters for benzene and toluene individually were determined in previous experimental studies. These values were incorporated into a physiologically based phannacokinetic model which simulated the inhalation uptake process for both chemcials simultaneously. An optimal fit to the uptake curves for simultaneous exposure was obtained by adjusting the metabolic interaction terms for each chemical. Mutual suppression of metabolism was apparent. Toluene more effectively inhibited benzene metabolism than the reverse. This simulation approach for analyzing gas uptake data provided a method to determine the metabolic interactions occurring upon inhalation exposure to two different chemicals. Such analyses will prove useful in improving predictive toxicokinetic models.
Curtis C. Travis, Office of Risk Analysis, Health and Safety Research Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831-6109, U.S.A.
Address for correspondence:
0378-4274/90/$3.50 @ 1990 Elsevier Science Publishers B.V. (Biomedical Division)
142 INTRODUCTION
Benzene and toluene are two of the most commonly used solvents in industry and manufacturing. Both of these aromatic compounds have deleterious effects of human health. Benzene exposure is linked to many different types of blood disorders, such as leukopenia and aplastic anemia [I]. It is also a known animal and human carcinogen [2,3]. Toluene, while not a known carcinogen, is more acutely toxic than benzene and has been shown to depress the central nervous system upon long-term exposure. Due to these health risks, the pharmacokinetics of the individual compounds have been studied extensively in humans and rats. However, since benzene and toluene are often present simultaneously in industrial environments, it is worthwhile to determine the pha~acokineti~s of these chemicals upon co-exposure. Recently, physiologically based pharmacokinetic (PBPK) models have been shown to accurately describe the distribution of volatile organics such as styrene [4], methylene chloride [5], tetrachloroethylene [6] and benzene [7,8] among bodily tissues. When used in conjunction with gas-uptake data, these models have also proven to be valuable analytical tools for deriving the in vivo metabolic and interactive constants [9,10]. Thus, we used this approach to characterize simultaneous exposure to benzene and toluene. A model was developed and used in conjunction with closed-chamber studies to investigate the nature and magnitude of the metabolic interactions between these two compounds. Developing such a model to include both solvents provides a preliminary step in determining realistic estimates of human risk resulting from co-exposure. METHODS
Chamber exposure The gas-uptake exposure system and physiologically based simulation analysis of the uptake data have been previously documented [IO]. For this experiment, male Fischer 344 rats*, weighing between 190 and 220 g, were kept on a 12-h on/lZh off artificial light cycle in portable laminar flow containment units, and were given food and water ad libitum at all times except during exposure. All exposures were begun between 8:00 and 9:00 a.m. and lasted 6 h. Groups of 3 rats were acclimated for 30 min in the 9-liter desiccator jar chamber before measured amounts of benzene and toluene were injected into the recirculating system. Co-exposure concentrations were 200 ppm benzene with 1000 ppm toluene and then 1000 ppm benzene with 200 ppm toluene. The disappearance of both chemicals was monitored every 10 min by au-
*The animals used in this study were handled in accordance with the principles stated in the &i&for the Cure and Use of Laboratory Animals, prepared by the Committee on Care and Uses of Laboratory Animals of the Institute of Laboratory Animal Resources, National Research Council, DHHA, National Institute of Health Publication 85-23, 1985, and the Animal Welfare Act of 1966, as amended.
143
tomated gas chromatographic* analysis of the chamber atmosphere. Both chemicals were obtained commercially** with greater than 99% purity. Computer model and data analysis
PBPK models simulate the uptake, distribution and elimination of compounds by an animal species using a series of mass-balance differential equations. In this experiment, two sets - one for each chemical - of 5 equations were used to quantify the concentration of chemical in 5 different compartments. One equation represented the chamber compartment, while the other 4 described the amount of chemical present in various bodily tissues, grouped by similar blood flows, chemical solubility characteristics, and metabolic capabilities. These compartments represented richly perfused organs, such as the kidney and spleen, moderately perfused tissues, such as skin and muscle, and fat, which is slowly perfused, but has a large capacity for storing most organic compounds. The 5th compartment, describing the liver, incorporated saturable Michaelis-Menten metabolism, as described by the biochemical constants V,,, and K,,,. Three types of metabolic interaction were considered in this study: competitive, non-competitive and uncompetitive. Equations for these 3 cases can be derived from a general interactive equation described by Andersen et al. [9]. When two chemicals compete as substrates for the same site on an enzyme, the inhibition is termed ‘competitive’. In this instance, the rate of change of benzene in the liver is described by the equation:
dAMTB ~ = dt
(Q
CaB!
- (Q
CvB)
-
v t?lClXB CvB Kn~(l
•I-CvdKiTB)
•I-CUB'
where AMTB is the amount of benzene present in the liver, Q is the cardiac output, and C,B is the concentration of benzene in arterial blood. C,B and CVrare the concentrations of benzene and toluene in venous blood, and were assumed to be the liver concentration divided by the liver/blood partition coefficient of the particular chemical. For competitive inhibition, K,rTBis the dissociation constant for the enzymeinhibitor complex, which is equal to the dissociation constant of enzyme-substrate complex, or Km~ [ 111. Inhibition may also occur even though the inhibitor and the substrate have different enzymatic binding sites. In non-competitive inhibition, the inhibitor binds to the enzyme, causing a change in the stereochemical arrangement of the enzyme such that the substrate cannot bind. Here, KiTB represents the dissociation of the enzymesubstrateinhibitor complex, and is equal to the enzyme-substrate dissociation constant, Km~ [ 111.This form of metabolic interaction is characterized by Equation (2): *Hewlett **Aldrich
Packard Chemical
Co., Palo Alto, CA 94304-l 181. Co., Inc., Milwaukee,
WI.
A different type of interaction can also occur when the two compounds have different enzymatic binding sites. If the substrate must first bind to the enzyme before the inhibitor can, the interaction is termed ‘uncom~titive inhibition’. In this case, no inhibition is apparent at low concentrations and the representive equation has the form of Equation (3):
A second set of equations, corresponding to the rate of change in the amount of toluene (r) is formed by interchanging all B and T terms in the above equations. The PBPK simulation used for these studies assumed that the chemicals were freely distributed to all tissues and eliminated only by expiration and hepatic metabolism. The model was completely defined, lacking only the inhibition constants. Partition coefflcients, the ratio of chemical concentration in blood to air and blood to tissue, were determined by vial-equilibration head-space analysis [ 123. Physiological parameters, such as tissue volume, blood flows, and cardiac output, were found in the literature. The metabolic constants, V,,, and Km, were determined for both benzene and toluene in previous gas-uptake studies. To determine the metabolic inhibition constants, values were incorporated into the model and varied during multiple simulations until good visual agreement was found between the model prediction and the observed experimental data, These parameters were then used as starting values for optimization using the SIMUSOLV@* software package. RESULTS
All physiological and metabolic values used in the present experiment are listed in Tables I and II. The Michaelis-Menten metabolic constants, K,,, and Vmx, were determined with gas-uptake techniques by Gargas et al. [lo]; the partition coefficients were determined using Cal-equilibration head-space analysis 1121.Physiological parameters were determined from the literature. A set of uptake curves was obtained from closed-chamber rat inhalation exposures. Benzene and toluene were tested individually at 200 ppm. Two subsequent runs were made with co-exposures of 200 ppm benzene with 1000 ppm toluene and then 1000 ppm benzene with 200 ppm tofuene. The set of 4 uptake curves was analyzed
*SIMUSOLV@ is a registered trademark of the Dow Chemical Company, Midland, MI.
14.5 TABLE I PHYSIOLOGICAL
PARAMETERS
FOR MALE FISCHER RAT@ Notation
Parameter Alveolar ventilation (l/h) Cardiac output (l/h) Blood flow fractions Liver Fat Richly perfused group (RPG) Moderately perfused group (MPG) Tissue group volume fractions Liver Fat Richly perfused group Moderately perfused group
Value 15.00 15.00
QL QF
QRPG Q MPG
VL VF VRPG VMPG
0.25 0.09 0.51 0.11
0.04
0.07 0.05 0.72
Partition coefficients for benzene Blood/air Liver/blood Fat/blood RPG/blood MPG/blood
17.8 0.96 28.0 0.96 0.58
Partition coefficients for toluene Blood/air Liver/blood Fat/blood RPG/blood MPG/blood
18.0 4.64 56.7 4.64 1.54
Michaelis-Menten constants 3.3 7.5
Maximum reaction rate (mg/h): for benzene for toluene Michaelis constant (mg/l):
for benzene for toluene
f&B KMT
0.3 0.3
“Note: The listed constants Qulvc,Qendc and V,,,~c are scaled to a 1 kg rat using the allometric equation Q,. = Qdvc x BW0.‘4.
using pharmacokinetic model simulations the metabolic interactions of benzene and To determine the inhibition constants, above (competitive, non-competitive, and
to determine the nature and magnitude of toluene during simultaneous exposure. each of the metabolic possibilities listed uncompetitive inhibition) were considered
146 TABLE II OPTIMIZED INTERACTION
PARAMETERS AND MODEL DISCRIMINATION
Model
K,B, (mg/l)
K,,,, (mg/l)
LLFa
Competitiveb Uncompetitive Non-competitive
0.3 1.69+0.036’ 2.47kO.033’
0.3 0.26+0.021c 0.28k0.018c
-698 -520 -431
BLog of the Likelihood Function as determined by SIMUSOLV@ optimization. The larger (or more positive) the LLF is, the better the fit to the data. A difference of 10 units is considered significant [23]. bThe competitive model was not optimized by varying the interaction parameters. See Results section. ‘These are the optimized constants and standard deviations as determined by SIMUSOLV@.
in turn. Since in vitro work had shown the two compounds to interact competitively [13], a model incorporating Equation (1) was used first to simulate the uptake. The variables were set to the following restrictions: Kisr = K,B and KiTB = KmT. Contrasting the in vitro work, the competitive inhibition model did not produce predictions which agreed well with the experimental data (Fig. 1). Next, a choice of model parameters consistent with non-competitive inhibition was tried. In Equation (2), Kisr and Kirs were varied using SIMUSOLV@ until a best least-squares fit was obtained (Fig. 2). Successful characterization of both sets of exposures was accomplished with KirB = 2.47 and KirB = 0.28. Uncompetitive inhibition was also investigated using Equation (3). SIMUSOLV@ optimization resulted in adequate fits to benzene data, but inconsistent results were obtained with the toluene predictions (Fig. 3). The optimized constants (Table II) overestimated the toluene curves starting at 1000 ppm and underestimated the toluene curves starting at 200 ppm. The goodness of fit for the 3 models was determined using the estimated log likelihood functions (LLF) ascertained from SIMUSOLV@. The inhibition constants for the competitive model were not optimized with this software package before determining the LLF. Instead, those values obtained from visual optimization were used in estimating the goodness-of-fit criteria; by definition, the Ki terms must equal the Km terms and therefore cannot be varied as is done with SIMUSOLV@ optimization. For the other two models, however, optimizations were performed by varying the interactive constants before estimating the LLF. Based on the results of this statistical analysis, the order of goodness of fit was non-competitive > uncompetitive > > competitive (Table II). Figures 4 and 5 show the chamber concentration difference between a single exposure and co-exposure. It is noticeable from these figures that toluene more effectively inhibits benzene than does the reverse. The inhibition constants listed in Table II also depict this relationship; KiTB at 0.28 mg/l is much smaller than KieT at 2.47 mg/l, showing, therefore, that toluene is a much better inhibitor of benzene than benzene of toluene.
147
10000 aMM ca3J3
-
-.
E :
seem - -
i
1st exp’tol exPos”re exposure Benrena. 2nd e~osure loluene. 2nd erp’tol exp*sure Benzene. model predictmn Toluene. model prediction
Benzene.
toiuene, 1st
exp’hl exp’tol
Fig, 1. Optimal model simulation of gas-uptake data using competitive inhibition model. The first experimental exposure was 200 ppm benzene with IO00 ppm toluene. The second experimental exposure was 1000 ppm benzene with 200 ppm toluene. The competitive model simulation output, shown by lines, used values of I& = 0.3 mg/l and KrB = 0.3 mg/l (Table II).
Benzene. camTcluene. MA&A Banzene. ,W.. k,luene. Benzene. - toiuene.
Le4fa
A
Eb
iriO
150 Time
1st expw exposure 1st exwol ewosure 2nd e*p%Jl erposwe 2nd exp’tll ,er~osure model prodlctm model jxediction
240
300
360
(min)
Fig. 2. Optimal model simulation of gas-uptake data using a non-competitive inhibition model. The first experimental exposure was 200 ppm benzene with 1000 ppm toluene. The second experimental exposure was 1000 ppm benzene with 200 ppm toluene. The non-competitive model simulation output, shown by lines, used values of I& = 2.47 mgjl and KirB = 0.28 m&l (Table II).
148
20
a - ,:o * Time
1
i
18G
240
/ 300
i 360
(min)
Fig. 3. Optimai model simulation of gas-uptake data using uncompetitive inhibition model. The first experimental exposure was 200 ppm benzene with 1000 ppm toluene. The second experimental exposure was IO00 ppm benzene with 200 ppm tcluene. The uncompetitive model simulation output, shown by tines, used values of KFYIBT = 1.69 mgjl and Krs = 0.26 mgp cfable 11).
Fig. 4. Closed-chamber concentrations of benzene are depicted for an exposure of 200 ppm benzene alone and for a co-exposure of ZOOppm benzene with IO@3ppm toluene. Modef predictions of benzene chamber concentration are shown by lines.
149
-200 olm200 -
200 200
ppm ppm ppm ppm
totuene toluene toluene totuene
with 1000 ppm benzene. with 1000 ppm benzene, atone. experimental data alone. model predtctlon
180
Time
experimentql data model predvztlon
240
(min)
Fig. 5. Closed-chamber concentrations of toluene are depicted for an exposure of 200 ppm toluene alone and for a co-exposure of 200 ppm toluene with 1000 ppm benzene. Model predictions of toluene chamber concentration are shown by lines.
DISCUSSION
The major route of exposure for both benzene and toluene, as with most volatile organics, is via inhalation. To exert their toxic effect, however, they must first be metabolized, which occurs primarily in the liver via the cytochrome P-450 proteins. Benzene is initially metabolized to benzene oxide, which is quickly converted to its endproducts by at least 3 different pathways. The majority is transformed to phenol and then to hydroquinone. The second pathway produces benzene glycol and subsequently catechol. The third pathway requires conjugation of benzene epoxide with glutathione S-transferase to produce phenyl mercapturic acid [ 141. The primary metabolic pathway for toluene is oxidization to benzoic acid with subsequent reaction with glycine to yield hippuric acid. Minor pathways include conjugation of bezoic acid with glucuronic acid, and ring hydroxylation of toluene to form o-cresol and p-cresol [ 151. Previous studies have suggested that benzene and toluene compete as substrates for the microsomal mixed-function oxidase system [ 13,1618]. In vivo studies in rats and mice found mutual suppression of benzene and toluene metabolism upon co-administration [ 13,161. Sato and Nakajima [ 171extended this work and found the metabolic antagonism to be dose-dependent, appearing only at high concentrations. This mutual suppression was also apparent in occupational settings. Inoue et al. [18] analyzed urinary metabolite levels at the end of a shift where workers had been exposed to both compounds. The biotransformation of benzene to phenolic compounds (excluding catechol) was inhibited by co-exposure to toluene with a compensatory increase in the pulmonary excretion of unmetabolized benzene. Metabolism
150
of toluene to hippuric acid was also suppressed, but not as effectively as toluene inhibited benzene metabolism [18]. In the present paper, a PBPK model was utilized in order to characterize the nature of the metabolic interaction between benzene and toluene. These techniques have been shown to be valuable analytical tools for deriving metabolic interactions in vivo [9]. The kinetic patterns of uptake, distribution, metabolism and elimination of inhaled gases and vapors are determined by physiological and biochemical factors. PBPK models incorporate these factors into a mathematical reconstruction of the physiological system, utilizing differential equations to represent compartments which are biologically well-defined; the animal is described in terms of particular organs with associated blood flows, volumes and partition coefficients. This mathematical reconstruction can be fitted to pha~acokinetic time-course data to indirectly determine metabolic terms in vivo [4,6,7,9]. Results obtained with these techniques have shown benzene and toluene to inhibit one another in a non-competitive fashion, in contrast to previous in vitro results which indicated competitive inhibition [ 13,171. The model could not adequately simulate our in vivo data on the assumption of competitive or uncom~titive inhibition. Only by incorporating an approach mathematically consistent with non-competitive inhibition could the model successfully describe the closed-chamber concentration data. We recognize that these results are preliminary in nature and inconclusive. Mathematical models cannot be used to establish mechanisms of action; they can only show that a given mathematical description is consistent with empirical data. Further biological studies are needed to determine the exact type of inhibition and to insure that the concentrations used in the present study did not compromise the cell’s energy status; the electron transfer co-factors may have been depleted or altered with such large concentrations. Other factors, such as metabolite interactions, may also have influenced the results. Validation studies of toxicity or enzyme interaction analyses need to be performed before a true biomedical mechanism of inhibition can be concluded from these present results. In any case, it is important to note the effect to toluene co-exposure on the toxicity of benzene. Several studies have indicated that suppression of benzene metabolism protects against the development of leukopenia [16,19]. Other studies have shown a reduction of chromosomal aberrations in bone marrow cells when toluene was administered simultaneously 1201.In addition, co-exposure to toluene has been shown to protect against benzene-induced depression of 59Fe utilization by red cells, which is commonly used by many investigators as a measure of erythropoiesis [13]. It does appear that by inhibiting benzene metabolism, co-exposure to toluene would lessen the danger associated with benzene. This is important in assessing the risk incurred by workers who handle automobile and industrial gasoline, which are known to contain both toluene and benzene [21,22]. Exposure to benzene vapors from such mixed sources may be less effective in reducing peripheral leukocyte counts than studies of benzene exposure alone would suggest.
151
ACKNOWLEDGEMENT
Research at the Oak Ridge National Laboratory was sponsored by the American Petroleum Institute under Interagency Agreement applicable under Martin Marietta Energy Systems, Inc. Contract No. DE-AC05-840R21400 with the U.S. Department of Energy. REFERENCES I Roh, J., Moon, Y.H. and Kim, K. (1987) The cytogenetic
effects of benzene and toluene on bone mar-
row cells in rats. Yonsei Med. J. 28,297-309. 2 Rinsky,
T.A., Alexander,
B.S., Hornung,
P.J. (1981) Benzene and leukemia: 3 Cronkite,
R., Filloon,
an epidemiologic
T.G., Young,
R.J., Okun,
risk assessment.
A.H. and Landrigan,
N. Engl. J. Med. 316, 10441049.
E.P., Drew, R.T., Inoue, T. and Bullis, J.E. (1985) Benzene hematoxicity
and leukemogene-
sis. Am. J. Ind. Med. 7,447456. 4 Ramsey,
J.C. and Andersen,
macokinetics 5 Andersen,
M.E. (1984) A physiologically
of styrene in rats and humans.
Toxicol.
M.E., Clewell, III, H.J., Gargas,
based pharmacokinetics
based description
Appl. Pharmacol.
of the inhalation
phar-
73, 1599175.
M.L., Smith, F.A. and Reitz, T.H. (1987) Physiologically
and the risk assessment
process
for methylene
chloride.
Toxicol.
Appl. Phar-
macol. 87, 185-205. 6 Ward,
R.C., Travis, C.C., Hetrick,
of tetrachloroethylene. 7 Travis, macol.
Toxicol.
C.C., Quillen,
D.M., Andersen,
Appl. Toxicol.
M.E. and Gargas,
M.L. (1988) Pharmacokinetics
93, 108-l 17.
J.L. and Arms, A.D. (1989) Pharmacokinetics
of benzene.
Toxicol.
Appl. Phar-
102,40&420.
8 Medinsky,
M.A., Sabourin,
ical model for simulation
P.J., Lucier, G., Bernbaum, of benzene metabolism
L.S. and Henderson,
by rats and mice. Toxicol.
R.F. (1989) A physiologAppl. Pharmacol.
99, 193-
206. 9 Andersen,
M.E., Gargas,
of the metabolic take methods. 10 Gargas, proach
M.L., Clewell, III, H.J. and Severyn,
interactions Toxicol.
between
trichloroethylene
Appl. Pharmacol.
M.L., Andersen,
metabolic
evaluation
in vivo using gas up-
89, 149157.
M.E. and Clewell,
for determining
K.M. (1987) Quantitative
and 1, I-dichloroethylene
constants
III, H.J. (1986) A physiologically from gas uptake
data. Toxicol.
based simulation
Appl. Pharmacol.
ap-
86,341-
342.
11 Fersht, A. (1985) Enzyme Structure 12 Gargas,
M.L., Burgess,
and Mechanism.
R.J., Voisard,
cients of low-molecular-weight
volatile
D.E., Cason, chemicals
W.H. Freeman
and Company,
G.H. and Andersen,
in various
New York.
M.E. (1989) Partition
liquids and tissues. Toxicol.
coeffi-
Appl. Phar-
macol. 98, 87-99. 13 Andrews,
L.S., Lee, E.W., Witmer,
metabolism, 14 Sabourin,
disposition P.J., Bechtold,
in the metabolism Appl. Pharmacol. 15 United Toluene.
CM.,
and hemopoietic
Kocsis, J.J. and Snyder, toxicity
W.E., Birnbaum,
and dispostion
of [3H]benzene.
L.S., Lucier,
of inhaled
[‘HIbenzene
R. (1977) Effects of toluene Biochem.
G. and Henderson, by F344/N
Pharmacol.
on the
26,293-300.
R.F. (1988) Differences
rats and B6C3Fi
mice. Toxicol.
94, 1288140.
States Environmental Final Report.
Protection
Prepared
Agency
(US EPA) (1983) Health
for U.S. EPA, Research
Triangle
Assessment
Document
16 Ikeda, M., Ohtsuji, H. and Imamura, T. (1972) In vivo suppression of benzene and styrene by co-administered toluene in the rats and effects of phenobarbital. Xenobiotica 2, 101-106. 17 Sato, A. and Nakajima,
T.
in vivo and in vitro. Toxicol.
(1979)Dose-dependent Appl. Pharmacol.
metabolic 48,249-256.
for
Park. NTIS PB#84-100056.
interaction
oxidation
between benzene and toluene
152 18 Inoue, O., Seiji, K., Watanabe, T., Kasahara, M., Nakatsuka, H., Songnian, Y., Li, G., Cai, S., Jin, C. and Ikeda, M. (1988) Mutual metabolic suppression between benzene and toluene in man. Occup. Environ. Health 60, 15-20. 19 Abromova, Z. and Gadaskina, I. (1966) Inhibition of oxidation of benzene by certain anti-oxidants. Fed. Proc. 25 (trans. Suppl. T91), T91-T92. 20 Gad-El-Karim, M.M., Harper, B.L. and Legator, M.S. (1984) Modifications in the myeloclastogenic effect of benzene in mice with toluene, phenobarbital, 3-methylcholanthrene, Aroclo 1254 and SKF525A. Mutat. Res. 135,225-243. 21 Ikeda, M., Kumai, M., Watanabe, T. and Fujita, H. (1984) Aromatic and other contents in automobile gasoline in Japan. Ind. Health 22,235-241. 22 Ikeda, M. and Kasahara, M. (1986) n-Hexane and benzene contents in gasoline for industrial purpose. Ind. Health 24,6366. 23 Bard, Y. (1974) Nonlinear Parameter Estimation. Academic Press, Inc., Orlando, Fl.
141
TOXLET 02348
In vivo metabolic interactions of benzene and toluene
Karen J. Purcell’, Gregory H. CasoS, Michael L. Gargas3, Melvin E. Andersen3 and Curtis C. Travis1 ‘Ofice of Risk Analysis, Health and Safety Research Division, Oak Ridge National Laboratory, TN 37831, 2Biochemical Toxicology Branch, Armstrong Wright-Patterson
Oak Ridge,
Aerospace Medical Research Laboratories,
Air Force Base, OH 45433 and ‘Chemical Industry Institute of Toxicology,
Research
Triangle Park, NC 27709 (U.S.A.)
(Received 2 November 1989) (Revision received 4 December 1989) (Accepted 18 February 1990) Key worris: Physiologically based pharmacokinetic
modeling; Benzene; Toluence; Co-exposure;
Gas uptake
SUMMARY The metabolic interactions of benzene and toluene co-exposure were investigated in male Fischer rats. A closed recirculated exposure system was used to obtain inhalation uptake curves for individual chemicals as well as for a mixture of the two compounds. Pharmacokinetic parameters for benzene and toluene individually were determined in previous experimental studies. These values were incorporated into a physiologically based phannacokinetic model which simulated the inhalation uptake process for both chemcials simultaneously. An optimal fit to the uptake curves for simultaneous exposure was obtained by adjusting the metabolic interaction terms for each chemical. Mutual suppression of metabolism was apparent. Toluene more effectively inhibited benzene metabolism than the reverse. This simulation approach for analyzing gas uptake data provided a method to determine the metabolic interactions occurring upon inhalation exposure to two different chemicals. Such analyses will prove useful in improving predictive toxicokinetic models.
Curtis C. Travis, Office of Risk Analysis, Health and Safety Research Division, Oak Ridge National Laboratory, Oak Ridge, TN 37831-6109, U.S.A.
Address for correspondence:
0378-4274/90/$3.50 @ 1990 Elsevier Science Publishers B.V. (Biomedical Division)
142 INTRODUCTION
Benzene and toluene are two of the most commonly used solvents in industry and manufacturing. Both of these aromatic compounds have deleterious effects of human health. Benzene exposure is linked to many different types of blood disorders, such as leukopenia and aplastic anemia [I]. It is also a known animal and human carcinogen [2,3]. Toluene, while not a known carcinogen, is more acutely toxic than benzene and has been shown to depress the central nervous system upon long-term exposure. Due to these health risks, the pharmacokinetics of the individual compounds have been studied extensively in humans and rats. However, since benzene and toluene are often present simultaneously in industrial environments, it is worthwhile to determine the pha~acokineti~s of these chemicals upon co-exposure. Recently, physiologically based pharmacokinetic (PBPK) models have been shown to accurately describe the distribution of volatile organics such as styrene [4], methylene chloride [5], tetrachloroethylene [6] and benzene [7,8] among bodily tissues. When used in conjunction with gas-uptake data, these models have also proven to be valuable analytical tools for deriving the in vivo metabolic and interactive constants [9,10]. Thus, we used this approach to characterize simultaneous exposure to benzene and toluene. A model was developed and used in conjunction with closed-chamber studies to investigate the nature and magnitude of the metabolic interactions between these two compounds. Developing such a model to include both solvents provides a preliminary step in determining realistic estimates of human risk resulting from co-exposure. METHODS
Chamber exposure The gas-uptake exposure system and physiologically based simulation analysis of the uptake data have been previously documented [IO]. For this experiment, male Fischer 344 rats*, weighing between 190 and 220 g, were kept on a 12-h on/lZh off artificial light cycle in portable laminar flow containment units, and were given food and water ad libitum at all times except during exposure. All exposures were begun between 8:00 and 9:00 a.m. and lasted 6 h. Groups of 3 rats were acclimated for 30 min in the 9-liter desiccator jar chamber before measured amounts of benzene and toluene were injected into the recirculating system. Co-exposure concentrations were 200 ppm benzene with 1000 ppm toluene and then 1000 ppm benzene with 200 ppm toluene. The disappearance of both chemicals was monitored every 10 min by au-
*The animals used in this study were handled in accordance with the principles stated in the &i&for the Cure and Use of Laboratory Animals, prepared by the Committee on Care and Uses of Laboratory Animals of the Institute of Laboratory Animal Resources, National Research Council, DHHA, National Institute of Health Publication 85-23, 1985, and the Animal Welfare Act of 1966, as amended.
143
tomated gas chromatographic* analysis of the chamber atmosphere. Both chemicals were obtained commercially** with greater than 99% purity. Computer model and data analysis
PBPK models simulate the uptake, distribution and elimination of compounds by an animal species using a series of mass-balance differential equations. In this experiment, two sets - one for each chemical - of 5 equations were used to quantify the concentration of chemical in 5 different compartments. One equation represented the chamber compartment, while the other 4 described the amount of chemical present in various bodily tissues, grouped by similar blood flows, chemical solubility characteristics, and metabolic capabilities. These compartments represented richly perfused organs, such as the kidney and spleen, moderately perfused tissues, such as skin and muscle, and fat, which is slowly perfused, but has a large capacity for storing most organic compounds. The 5th compartment, describing the liver, incorporated saturable Michaelis-Menten metabolism, as described by the biochemical constants V,,, and K,,,. Three types of metabolic interaction were considered in this study: competitive, non-competitive and uncompetitive. Equations for these 3 cases can be derived from a general interactive equation described by Andersen et al. [9]. When two chemicals compete as substrates for the same site on an enzyme, the inhibition is termed ‘competitive’. In this instance, the rate of change of benzene in the liver is described by the equation:
dAMTB ~ = dt
(Q
CaB!
- (Q
CvB)
-
v t?lClXB CvB Kn~(l
•I-CvdKiTB)
•I-CUB'
where AMTB is the amount of benzene present in the liver, Q is the cardiac output, and C,B is the concentration of benzene in arterial blood. C,B and CVrare the concentrations of benzene and toluene in venous blood, and were assumed to be the liver concentration divided by the liver/blood partition coefficient of the particular chemical. For competitive inhibition, K,rTBis the dissociation constant for the enzymeinhibitor complex, which is equal to the dissociation constant of enzyme-substrate complex, or Km~ [ 111. Inhibition may also occur even though the inhibitor and the substrate have different enzymatic binding sites. In non-competitive inhibition, the inhibitor binds to the enzyme, causing a change in the stereochemical arrangement of the enzyme such that the substrate cannot bind. Here, KiTB represents the dissociation of the enzymesubstrateinhibitor complex, and is equal to the enzyme-substrate dissociation constant, Km~ [ 111.This form of metabolic interaction is characterized by Equation (2): *Hewlett **Aldrich
Packard Chemical
Co., Palo Alto, CA 94304-l 181. Co., Inc., Milwaukee,
WI.
A different type of interaction can also occur when the two compounds have different enzymatic binding sites. If the substrate must first bind to the enzyme before the inhibitor can, the interaction is termed ‘uncom~titive inhibition’. In this case, no inhibition is apparent at low concentrations and the representive equation has the form of Equation (3):
A second set of equations, corresponding to the rate of change in the amount of toluene (r) is formed by interchanging all B and T terms in the above equations. The PBPK simulation used for these studies assumed that the chemicals were freely distributed to all tissues and eliminated only by expiration and hepatic metabolism. The model was completely defined, lacking only the inhibition constants. Partition coefflcients, the ratio of chemical concentration in blood to air and blood to tissue, were determined by vial-equilibration head-space analysis [ 123. Physiological parameters, such as tissue volume, blood flows, and cardiac output, were found in the literature. The metabolic constants, V,,, and Km, were determined for both benzene and toluene in previous gas-uptake studies. To determine the metabolic inhibition constants, values were incorporated into the model and varied during multiple simulations until good visual agreement was found between the model prediction and the observed experimental data, These parameters were then used as starting values for optimization using the SIMUSOLV@* software package. RESULTS
All physiological and metabolic values used in the present experiment are listed in Tables I and II. The Michaelis-Menten metabolic constants, K,,, and Vmx, were determined with gas-uptake techniques by Gargas et al. [lo]; the partition coefficients were determined using Cal-equilibration head-space analysis 1121.Physiological parameters were determined from the literature. A set of uptake curves was obtained from closed-chamber rat inhalation exposures. Benzene and toluene were tested individually at 200 ppm. Two subsequent runs were made with co-exposures of 200 ppm benzene with 1000 ppm toluene and then 1000 ppm benzene with 200 ppm tofuene. The set of 4 uptake curves was analyzed
*SIMUSOLV@ is a registered trademark of the Dow Chemical Company, Midland, MI.
14.5 TABLE I PHYSIOLOGICAL
PARAMETERS
FOR MALE FISCHER RAT@ Notation
Parameter Alveolar ventilation (l/h) Cardiac output (l/h) Blood flow fractions Liver Fat Richly perfused group (RPG) Moderately perfused group (MPG) Tissue group volume fractions Liver Fat Richly perfused group Moderately perfused group
Value 15.00 15.00
QL QF
QRPG Q MPG
VL VF VRPG VMPG
0.25 0.09 0.51 0.11
0.04
0.07 0.05 0.72
Partition coefficients for benzene Blood/air Liver/blood Fat/blood RPG/blood MPG/blood
17.8 0.96 28.0 0.96 0.58
Partition coefficients for toluene Blood/air Liver/blood Fat/blood RPG/blood MPG/blood
18.0 4.64 56.7 4.64 1.54
Michaelis-Menten constants 3.3 7.5
Maximum reaction rate (mg/h): for benzene for toluene Michaelis constant (mg/l):
for benzene for toluene
f&B KMT
0.3 0.3
“Note: The listed constants Qulvc,Qendc and V,,,~c are scaled to a 1 kg rat using the allometric equation Q,. = Qdvc x BW0.‘4.
using pharmacokinetic model simulations the metabolic interactions of benzene and To determine the inhibition constants, above (competitive, non-competitive, and
to determine the nature and magnitude of toluene during simultaneous exposure. each of the metabolic possibilities listed uncompetitive inhibition) were considered
146 TABLE II OPTIMIZED INTERACTION
PARAMETERS AND MODEL DISCRIMINATION
Model
K,B, (mg/l)
K,,,, (mg/l)
LLFa
Competitiveb Uncompetitive Non-competitive
0.3 1.69+0.036’ 2.47kO.033’
0.3 0.26+0.021c 0.28k0.018c
-698 -520 -431
BLog of the Likelihood Function as determined by SIMUSOLV@ optimization. The larger (or more positive) the LLF is, the better the fit to the data. A difference of 10 units is considered significant [23]. bThe competitive model was not optimized by varying the interaction parameters. See Results section. ‘These are the optimized constants and standard deviations as determined by SIMUSOLV@.
in turn. Since in vitro work had shown the two compounds to interact competitively [13], a model incorporating Equation (1) was used first to simulate the uptake. The variables were set to the following restrictions: Kisr = K,B and KiTB = KmT. Contrasting the in vitro work, the competitive inhibition model did not produce predictions which agreed well with the experimental data (Fig. 1). Next, a choice of model parameters consistent with non-competitive inhibition was tried. In Equation (2), Kisr and Kirs were varied using SIMUSOLV@ until a best least-squares fit was obtained (Fig. 2). Successful characterization of both sets of exposures was accomplished with KirB = 2.47 and KirB = 0.28. Uncompetitive inhibition was also investigated using Equation (3). SIMUSOLV@ optimization resulted in adequate fits to benzene data, but inconsistent results were obtained with the toluene predictions (Fig. 3). The optimized constants (Table II) overestimated the toluene curves starting at 1000 ppm and underestimated the toluene curves starting at 200 ppm. The goodness of fit for the 3 models was determined using the estimated log likelihood functions (LLF) ascertained from SIMUSOLV@. The inhibition constants for the competitive model were not optimized with this software package before determining the LLF. Instead, those values obtained from visual optimization were used in estimating the goodness-of-fit criteria; by definition, the Ki terms must equal the Km terms and therefore cannot be varied as is done with SIMUSOLV@ optimization. For the other two models, however, optimizations were performed by varying the interactive constants before estimating the LLF. Based on the results of this statistical analysis, the order of goodness of fit was non-competitive > uncompetitive > > competitive (Table II). Figures 4 and 5 show the chamber concentration difference between a single exposure and co-exposure. It is noticeable from these figures that toluene more effectively inhibits benzene than does the reverse. The inhibition constants listed in Table II also depict this relationship; KiTB at 0.28 mg/l is much smaller than KieT at 2.47 mg/l, showing, therefore, that toluene is a much better inhibitor of benzene than benzene of toluene.
147
10000 aMM ca3J3
-
-.
E :
seem - -
i
1st exp’tol exPos”re exposure Benrena. 2nd e~osure loluene. 2nd erp’tol exp*sure Benzene. model predictmn Toluene. model prediction
Benzene.
toiuene, 1st
exp’hl exp’tol
Fig, 1. Optimal model simulation of gas-uptake data using competitive inhibition model. The first experimental exposure was 200 ppm benzene with IO00 ppm toluene. The second experimental exposure was 1000 ppm benzene with 200 ppm toluene. The competitive model simulation output, shown by lines, used values of I& = 0.3 mg/l and KrB = 0.3 mg/l (Table II).
Benzene. camTcluene. MA&A Banzene. ,W.. k,luene. Benzene. - toiuene.
Le4fa
A
Eb
iriO
150 Time
1st expw exposure 1st exwol ewosure 2nd e*p%Jl erposwe 2nd exp’tll ,er~osure model prodlctm model jxediction
240
300
360
(min)
Fig. 2. Optimal model simulation of gas-uptake data using a non-competitive inhibition model. The first experimental exposure was 200 ppm benzene with 1000 ppm toluene. The second experimental exposure was 1000 ppm benzene with 200 ppm toluene. The non-competitive model simulation output, shown by lines, used values of I& = 2.47 mgjl and KirB = 0.28 m&l (Table II).
148
20
a - ,:o * Time
1
i
18G
240
/ 300
i 360
(min)
Fig. 3. Optimai model simulation of gas-uptake data using uncompetitive inhibition model. The first experimental exposure was 200 ppm benzene with 1000 ppm toluene. The second experimental exposure was IO00 ppm benzene with 200 ppm tcluene. The uncompetitive model simulation output, shown by tines, used values of KFYIBT = 1.69 mgjl and Krs = 0.26 mgp cfable 11).
Fig. 4. Closed-chamber concentrations of benzene are depicted for an exposure of 200 ppm benzene alone and for a co-exposure of ZOOppm benzene with IO@3ppm toluene. Modef predictions of benzene chamber concentration are shown by lines.
149
-200 olm200 -
200 200
ppm ppm ppm ppm
totuene toluene toluene totuene
with 1000 ppm benzene. with 1000 ppm benzene, atone. experimental data alone. model predtctlon
180
Time
experimentql data model predvztlon
240
(min)
Fig. 5. Closed-chamber concentrations of toluene are depicted for an exposure of 200 ppm toluene alone and for a co-exposure of 200 ppm toluene with 1000 ppm benzene. Model predictions of toluene chamber concentration are shown by lines.
DISCUSSION
The major route of exposure for both benzene and toluene, as with most volatile organics, is via inhalation. To exert their toxic effect, however, they must first be metabolized, which occurs primarily in the liver via the cytochrome P-450 proteins. Benzene is initially metabolized to benzene oxide, which is quickly converted to its endproducts by at least 3 different pathways. The majority is transformed to phenol and then to hydroquinone. The second pathway produces benzene glycol and subsequently catechol. The third pathway requires conjugation of benzene epoxide with glutathione S-transferase to produce phenyl mercapturic acid [ 141. The primary metabolic pathway for toluene is oxidization to benzoic acid with subsequent reaction with glycine to yield hippuric acid. Minor pathways include conjugation of bezoic acid with glucuronic acid, and ring hydroxylation of toluene to form o-cresol and p-cresol [ 151. Previous studies have suggested that benzene and toluene compete as substrates for the microsomal mixed-function oxidase system [ 13,1618]. In vivo studies in rats and mice found mutual suppression of benzene and toluene metabolism upon co-administration [ 13,161. Sato and Nakajima [ 171extended this work and found the metabolic antagonism to be dose-dependent, appearing only at high concentrations. This mutual suppression was also apparent in occupational settings. Inoue et al. [18] analyzed urinary metabolite levels at the end of a shift where workers had been exposed to both compounds. The biotransformation of benzene to phenolic compounds (excluding catechol) was inhibited by co-exposure to toluene with a compensatory increase in the pulmonary excretion of unmetabolized benzene. Metabolism
150
of toluene to hippuric acid was also suppressed, but not as effectively as toluene inhibited benzene metabolism [18]. In the present paper, a PBPK model was utilized in order to characterize the nature of the metabolic interaction between benzene and toluene. These techniques have been shown to be valuable analytical tools for deriving metabolic interactions in vivo [9]. The kinetic patterns of uptake, distribution, metabolism and elimination of inhaled gases and vapors are determined by physiological and biochemical factors. PBPK models incorporate these factors into a mathematical reconstruction of the physiological system, utilizing differential equations to represent compartments which are biologically well-defined; the animal is described in terms of particular organs with associated blood flows, volumes and partition coefficients. This mathematical reconstruction can be fitted to pha~acokinetic time-course data to indirectly determine metabolic terms in vivo [4,6,7,9]. Results obtained with these techniques have shown benzene and toluene to inhibit one another in a non-competitive fashion, in contrast to previous in vitro results which indicated competitive inhibition [ 13,171. The model could not adequately simulate our in vivo data on the assumption of competitive or uncom~titive inhibition. Only by incorporating an approach mathematically consistent with non-competitive inhibition could the model successfully describe the closed-chamber concentration data. We recognize that these results are preliminary in nature and inconclusive. Mathematical models cannot be used to establish mechanisms of action; they can only show that a given mathematical description is consistent with empirical data. Further biological studies are needed to determine the exact type of inhibition and to insure that the concentrations used in the present study did not compromise the cell’s energy status; the electron transfer co-factors may have been depleted or altered with such large concentrations. Other factors, such as metabolite interactions, may also have influenced the results. Validation studies of toxicity or enzyme interaction analyses need to be performed before a true biomedical mechanism of inhibition can be concluded from these present results. In any case, it is important to note the effect to toluene co-exposure on the toxicity of benzene. Several studies have indicated that suppression of benzene metabolism protects against the development of leukopenia [16,19]. Other studies have shown a reduction of chromosomal aberrations in bone marrow cells when toluene was administered simultaneously 1201.In addition, co-exposure to toluene has been shown to protect against benzene-induced depression of 59Fe utilization by red cells, which is commonly used by many investigators as a measure of erythropoiesis [13]. It does appear that by inhibiting benzene metabolism, co-exposure to toluene would lessen the danger associated with benzene. This is important in assessing the risk incurred by workers who handle automobile and industrial gasoline, which are known to contain both toluene and benzene [21,22]. Exposure to benzene vapors from such mixed sources may be less effective in reducing peripheral leukocyte counts than studies of benzene exposure alone would suggest.
151
ACKNOWLEDGEMENT
Research at the Oak Ridge National Laboratory was sponsored by the American Petroleum Institute under Interagency Agreement applicable under Martin Marietta Energy Systems, Inc. Contract No. DE-AC05-840R21400 with the U.S. Department of Energy. REFERENCES I Roh, J., Moon, Y.H. and Kim, K. (1987) The cytogenetic
effects of benzene and toluene on bone mar-
row cells in rats. Yonsei Med. J. 28,297-309. 2 Rinsky,
T.A., Alexander,
B.S., Hornung,
P.J. (1981) Benzene and leukemia: 3 Cronkite,
R., Filloon,
an epidemiologic
T.G., Young,
R.J., Okun,
risk assessment.
A.H. and Landrigan,
N. Engl. J. Med. 316, 10441049.
E.P., Drew, R.T., Inoue, T. and Bullis, J.E. (1985) Benzene hematoxicity
and leukemogene-
sis. Am. J. Ind. Med. 7,447456. 4 Ramsey,
J.C. and Andersen,
macokinetics 5 Andersen,
M.E. (1984) A physiologically
of styrene in rats and humans.
Toxicol.
M.E., Clewell, III, H.J., Gargas,
based pharmacokinetics
based description
Appl. Pharmacol.
of the inhalation
phar-
73, 1599175.
M.L., Smith, F.A. and Reitz, T.H. (1987) Physiologically
and the risk assessment
process
for methylene
chloride.
Toxicol.
Appl. Phar-
macol. 87, 185-205. 6 Ward,
R.C., Travis, C.C., Hetrick,
of tetrachloroethylene. 7 Travis, macol.
Toxicol.
C.C., Quillen,
D.M., Andersen,
Appl. Toxicol.
M.E. and Gargas,
M.L. (1988) Pharmacokinetics
93, 108-l 17.
J.L. and Arms, A.D. (1989) Pharmacokinetics
of benzene.
Toxicol.
Appl. Phar-
102,40&420.
8 Medinsky,
M.A., Sabourin,
ical model for simulation
P.J., Lucier, G., Bernbaum, of benzene metabolism
L.S. and Henderson,
by rats and mice. Toxicol.
R.F. (1989) A physiologAppl. Pharmacol.
99, 193-
206. 9 Andersen,
M.E., Gargas,
of the metabolic take methods. 10 Gargas, proach
M.L., Clewell, III, H.J. and Severyn,
interactions Toxicol.
between
trichloroethylene
Appl. Pharmacol.
M.L., Andersen,
metabolic
evaluation
in vivo using gas up-
89, 149157.
M.E. and Clewell,
for determining
K.M. (1987) Quantitative
and 1, I-dichloroethylene
constants
III, H.J. (1986) A physiologically from gas uptake
data. Toxicol.
based simulation
Appl. Pharmacol.
ap-
86,341-
342.
11 Fersht, A. (1985) Enzyme Structure 12 Gargas,
M.L., Burgess,
and Mechanism.
R.J., Voisard,
cients of low-molecular-weight
volatile
D.E., Cason, chemicals
W.H. Freeman
and Company,
G.H. and Andersen,
in various
New York.
M.E. (1989) Partition
liquids and tissues. Toxicol.
coeffi-
Appl. Phar-
macol. 98, 87-99. 13 Andrews,
L.S., Lee, E.W., Witmer,
metabolism, 14 Sabourin,
disposition P.J., Bechtold,
in the metabolism Appl. Pharmacol. 15 United Toluene.
CM.,
and hemopoietic
Kocsis, J.J. and Snyder, toxicity
W.E., Birnbaum,
and dispostion
of [3H]benzene.
L.S., Lucier,
of inhaled
[‘HIbenzene
R. (1977) Effects of toluene Biochem.
G. and Henderson, by F344/N
Pharmacol.
on the
26,293-300.
R.F. (1988) Differences
rats and B6C3Fi
mice. Toxicol.
94, 1288140.
States Environmental Final Report.
Protection
Prepared
Agency
(US EPA) (1983) Health
for U.S. EPA, Research
Triangle
Assessment
Document
16 Ikeda, M., Ohtsuji, H. and Imamura, T. (1972) In vivo suppression of benzene and styrene by co-administered toluene in the rats and effects of phenobarbital. Xenobiotica 2, 101-106. 17 Sato, A. and Nakajima,
T.
in vivo and in vitro. Toxicol.
(1979)Dose-dependent Appl. Pharmacol.
metabolic 48,249-256.
for
Park. NTIS PB#84-100056.
interaction
oxidation
between benzene and toluene
152 18 Inoue, O., Seiji, K., Watanabe, T., Kasahara, M., Nakatsuka, H., Songnian, Y., Li, G., Cai, S., Jin, C. and Ikeda, M. (1988) Mutual metabolic suppression between benzene and toluene in man. Occup. Environ. Health 60, 15-20. 19 Abromova, Z. and Gadaskina, I. (1966) Inhibition of oxidation of benzene by certain anti-oxidants. Fed. Proc. 25 (trans. Suppl. T91), T91-T92. 20 Gad-El-Karim, M.M., Harper, B.L. and Legator, M.S. (1984) Modifications in the myeloclastogenic effect of benzene in mice with toluene, phenobarbital, 3-methylcholanthrene, Aroclo 1254 and SKF525A. Mutat. Res. 135,225-243. 21 Ikeda, M., Kumai, M., Watanabe, T. and Fujita, H. (1984) Aromatic and other contents in automobile gasoline in Japan. Ind. Health 22,235-241. 22 Ikeda, M. and Kasahara, M. (1986) n-Hexane and benzene contents in gasoline for industrial purpose. Ind. Health 24,6366. 23 Bard, Y. (1974) Nonlinear Parameter Estimation. Academic Press, Inc., Orlando, Fl.
