Physiologically Based Toxicokinetic Modeling of Inhaled Ethyl tertiary-butyl Ether in Humans

Transcription

1 TOXICOLOGICAL SCIENCES 51, (1999) Copyright 1999 by the Society of Toxicology Physiologically Based Toxicokinetic Modeling of Inhaled Ethyl tertiary-butyl Ether in Humans Annsofi Nihlén*,1 and Gunnar Johanson*, *Department of Toxicology and Risk Assessment, National Institute for Working Life, SE Solna, Sweden; and Department of Medical Sciences University Hospital, Uppsala, Sweden Received September 29, 1998; accepted June 5, 1999 A physiologically based toxicokinetic (PBTK) model was developed for evaluation of inhalation exposure in humans to the gasoline additive, ethyl tertiary-butyl ether (ETBE). PBTK models are useful tools to relate external exposure to internal doses and biological markers of exposure in humans. To describe the kinetics of ETBE, the following compartments were used: lungs (including arterial blood), liver, fat, rapidly perfused tissues, resting muscles, and working muscles. The same set of compartments and, in addition, a urinary excretion compartment were used for the metabolite tertiary-butyl alcohol (TBA). First order metabolism was assumed in the model, since linear kinetics has been shown experimentally in humans after inhalation exposure up to 50 ppm ETBE. Organ volumes and blood flows were calculated from individual body composition based on published equations, and tissue/blood partition coefficients were calculated from liquid/air partition coefficients and tissue composition. Estimates of individual metabolite parameters of 8 subjects were obtained by fitting the PBTK model to experimental data from humans (5, 25, 50 ppm ETBE, 2-h exposure; Nihlén et al., Toxicol. Sci., 1998; 46, 1 10). The PBTK model was then used to predict levels of the biomarkers ETBE and TBA in blood, urine, and exhaled air after various scenarios, such as prolonged exposure, fluctuating exposure, and exposure during physical activity. In addition, the interindividual variability in biomarker levels was predicted, in the eight experimentally exposed subjects after a working week. According to the model, raising the work load from rest to heavy exercise increases all biomarker levels by approximately 2-fold at the end of the work shift, and by 3-fold the next morning. A small accumulation of all biomarkers was seen during one week of simulated exposure. Further predictions suggested that the interindividual variability in biomarker levels would be higher the next morning than at the end of the work shift, and higher for TBA than for ETBE. Monte Carlo simulations were used to describe fluctuating exposure scenarios. These simulations suggest that ETBE levels in blood and exhaled air at the end of the working day are highly sensitive to exposure fluctuations, whereas ETBE levels the next morning and TBA in urine and blood are less sensitive. Considering these simulations, data from the previous toxicokinetic study and practical issues, we suggest that TBA in urine is a suitable biomarker for exposure to ETBE and gasoline vapor. Key Words: biological monitoring; ethyl tertiary-butyl ether 1 To whom correspondence should be addressed. Fax: Annsofi.Nihlen@niwl.se. (ETBE); gasoline; inhalation exposure; methyl tertiary-butyl ether (MTBE); oxygenates; physiologically based toxicokinetic (PBTK) model; risk assessment; tertiary-butyl alcohol (TBA); toxicokinetics. Toxicokinetics is defined as the mathematical description of the time-course of disposition (absorption, distribution, metabolism, and excretion) of xenobiotics in living organisms. Knowledge about a compound s toxicokinetic properties is important for toxicological risk assessment. One of the rationales for a toxicokinetic study is that the response in a target tissue should be viewed in relation to the level of the parent compound or metabolites in that particular tissue. In animals, it is possible to measure the concentration of a compound in different tissues, but this is not feasible in humans. Instead mathematical models have been developed and used to simulate tissue doses and to predict the toxicokinetics in situations when experimental data cannot be obtained. Physiologically based toxicokinetic (PBTK) models, which consist of several compartments where organs are grouped together according to blood flow and fat content, are especially useful. Data on physiological tissue blood flow, organ volumes, and tissue/ blood partition coefficients are used in PBTK models. Organ volumes and blood flows are calculated from individual body composition, based on published equations and tissue/blood partition coefficients from liquid/air partition coefficients and tissue composition. The whole body is simulated and the fate of a chemical in different tissues in the human body can then be described (Andersen et al., 1995; Johanson, 1997). Biological exposure monitoring should be considered as a complement to air monitoring. During air monitoring, the exposure level is measured rather than the absorbed dose. Biological exposure monitoring takes dermal absorption and factors shown to modify the disposition of solvent into account, including physical activity (work load), body composition, age, sex, variations in metabolic capacity, ethnicity, diet, smoking, drug treatment, and coexposure to ethanol and other solvents (Clewell and Anderson, 1996; Löf and Johanson, 184

2 PBTK MODELING OF INHALED ETBE IN HUMANS ). The levels of parent compound or metabolites in blood, urine, or expired air are commonly used as biomarkers. Ethers or other oxygen-containing compounds have come to replace lead in gasoline (U.S. EPA 1993; HEI 1996). Oxygenates enhance the octane number, and in addition, improve combustion efficiency and thereby reduce emissions. In particular, methyl tertiary-butyl ether (MTBE) is now used world wide and is produced in large quantities. Other aliphatic ethers, such as ethyl tertiary-butyl ether (ETBE), tertiary-amyl ethyl ether, and alcohols are alternatives to MTBE. One reason for the interest in ETBE is that it may serve to increase the market for biofuels: ETBE can be manufactured from ethanol produced from agricultural and forestry feedstock. The ethers or their metabolites in blood, urine, or exhaled air may be useful biomarkers of gasoline-vapor exposure. Less volatile compounds predominantly present in liquid gasoline, that have been proposed as biomarkers, include metabolites of benzene, e.g., trans, trans-muconic acid (Ong and Lee, 1994) and of trimethylbenzenes, e.g., dimethylhippuric acids (Järnberg et al., 1997). The aim of the present study was to develop a PBTK model for inhalation exposure to ETBE in humans. The model was validated against experimental exposures (5, 25, 50 ppm ETBE, 2 h, 50 W, n 8) previously performed in our laboratory (Nihlén et al., 1998a). A further goal was to use the PBTK model to study how various exposure scenarios, such as prolonged exposure (weeks), fluctuations in the exposure levels, and physical activity influence the levels of ETBE and of its metabolite tertiary-butyl alcohol (TBA), in blood, urine and exhaled air. Further, the suitability of using the levels of ETBE and its metabolite TBA in exhaled air or in arterial and venous blood, or TBA in urine as potential biomarkers for exposure to ETBE and gasoline vapor was examined, after predictions with this present PBTK model. PBTK Model MATERIALS AND METHODS Compartments. Organs were grouped together into compartments according to blood flow and fat content. This flow-limited model consists of 13 compartments, 6 similar compartments each for ETBE and TBA, and in addition, a urinary excretion compartment for TBA (Fig. 1). Equations as well as abbreviations are presented in the Appendix. The lung compartment comprises lung tissue and arterial blood. Both uptake and excretion of ETBE by ventilation were included in this compartment. The experimental ETBE exposures (Nihlén et al., 1998a) were used for validation of the model. The ETBE used during the experimental exposures contained approximately 0.4% TBA. This was accounted for in the model with uptake of TBA in the lung-blood compartment for TBA. During the experimental ETBE exposures (Nihlén et al., 1998a) the subjects performed physical exercise on a bicycle ergometer, involving work done almost exclusively by the legs, or approximately half the body muscle mass. The present PBTK model, therefore, comprises two muscle compartments: resting and working muscles, and the latter reflects the increased blood flow through the leg muscles during bicycle exercise. Two muscle compartments have previously been used successfully to describe experimental data obtained FIG. 1. Physiologically based toxicokinetic model used in the simulations of ethyl tertiary-butyl ether (ETBE) and of a metabolite, tertiary-butyl alcohol (TBA). Symbols are explained in the Appendixes. during bicycle exercise (Järnberg and Johanson, 1999; Johanson and Näslund, 1988). The skin is included in the muscle compartments, since only combined perfusion data are available for skin and muscle tissues during physical exercise. A compartment for fat was incorporated in the model, since the ether is lipophilic (Nihlén et al., 1995) and may therefore accumulate in adipose tissue. First order metabolism was assumed to occur in our model, since linear kinetics has been shown experimentally for humans up to 50 ppm ETBE (Nihlén et al., 1998a). Further, metabolism was postulated to occur only in the liver. The related ether, MTBE, has been shown to be metabolized, e.g., to TBA, 2-methyl 1,2-propanediol, and -hydroxyisobutyric acid, in both humans (Nihlén et al., 1998b) and animals (Bernauer et al., 1998), and it is assumed that ETBE has a similar metabolite pattern (Bernauer et al., 1998; HEI 1996; White et al., 1995). The rapidly perfused tissue compartment was set as the sum of brain tissue, kidneys, and remaining tissues with high blood perfusion (Droz 1992). The clearance of TBA to urine was calculated from the rapidly perfused tissues (CLiT, see Fig. 1) and TBA was excreted to urine by a first order elimination process. Previously published equations, given in the Appendix, were used to calculate the physiological characteristics of the PBTK model (Droz 1992). Tissue volumes and blood flows were calculated from individual data on body weight and body height. Blood flows were also expressed as functions of physical activity. The tissue/air partition coefficients (P) were calculated from water/air and olive oil/air partition coefficients measured in vitro (Nihlén et al., 1995) and the average fat and water content in different tissues (Baker, 1969; Fiserova- Bergerova 1983; Pelekis et al., 1995) as follows: Ptissue/air (water content in tissue Pwater/air) (fat content in tissue Poil/air) The tissue/blood partition coefficients were calculated as the quotient of the calculated tissue/air partition coefficient and the measured blood/air partition coefficient (Nihlén et al., 1995). The tissue/blood partition coefficient for the rapidly perfused tissue was set to that obtained for the brain tissue. The partition coefficients used in the model are given in Table 1.

3 186 NIHLÉN AND JOHANSON TABLE 1 Calculated and Measured Partition Coefficients (P) of Ethyl tertiary-butyl Ether (ETBE) and tertiary-butyl Alcohol (TBA) Blood/air a Water/blood a Liver/blood b Fat/blood b Muscle/blood b Rpt/blood c ETBE TBA Note. Mean values are given. a Measured P blood/air in vitro and calculated P water/blood from measured P blood/air and P water/air (Nihlén et al., 1995). b Partition coefficients were calculated from tissue composition (water and oil content) (Baker 1969; Fiserova-Bergerova 1983; Pelekis et al., 1995) and measured liquid/air partition coefficients in vitro (Nihlén et al., 1995). c The rapidly perfused tissue partition coefficient (P rpt/blood ) was calculated from tissue composition (water and oil content) of the brain (Fiserova-Bergerova, 1983) and measured liquid/air partition coefficients in vitro (Nihlén et al., 1995). Software packages. The model was written and compiled in Fortran version 5.1 (Microsoft, Redmond, WA). The individual estimates of the metabolism parameters (CLiE, CLiT, CLiM, Kel; see Appendix for symbols) were determined for all 8 subjects experimentally exposed to ETBE with the software package WinNonlin version 1.5 (Scientific Consulting Inc., Cary, NC) using a personal computer. The residuals of the metabolism parameters were minimized by the iteratively reweighted least squares method (at least 1000 iterations) with the use of the Gauss-Newton (Levenberg-Heartly) algorithm. The sensitivity analysis was performed with the software package Simnon (version 2.0, SSPA Maritime Consulting AB, Gothenburg, Sweden) with the Dorman-Prince algorithm, using a personal computer. Metabolic parameters. Individual body composition (Table 2) and exposure levels of ETBE and TBA (ETBE contained 0.4% TBA) were used as model constants. Previously obtained data (individual ETBE and TBA concentrations in blood, ETBE and TBA exhalation rates, and rates of urinary excretion of TBA) in experimental ETBE exposures (2-h chamber exposure at the nominal levels of 5, 25, and 50 ppm ETBE, 50 W (work load, watt)) (Nihlén and Johanson, 1998a) were used to validate the model and estimate parameters associated with metabolism. All data from each subject (5, 25, and 50 ppm) were modeled in the same run. Thus, the model was written so that the subject was exposed for 2hat2-week intervals, first to 5 ppm ETBE and ppm TBA, then to 25 ppm ETBE and 0.14 ppm TBA, and finally to 50 ppm ETBE and 0.28 ppm TBA (individual exposure levels were used in the estimations). In the model, the subjects performed physical exercise (50 W) during the 2-h exposure in accordance with the experimental design. After exposure, light physical activity (25 W) was assumed during the day (until 11 PM) corresponding to normal indoor activities (Åstrand and Rodahl, 1986). At night (11 PM to 7 AM) work load was set to zero. Sensitivity analysis. A sensitivity analysis was performed, according to a previously published method (Pierce et al., 1996), to determine the influence of all parameters used in the PBTK model. Sensitivity coefficients are defined as the ratio between the percentage change in biomarker and the percentage change in parameter value. The magnitude of this sensitivity coefficient is proportional to the influence of the parameters on for example ETBE and TBA in arterial blood and its sign indicates the direction of dependence. Thus, a sensitivity coefficient of 0.3 indicates a 3% decrease in blood levels of ETBE when the parameter value increases 10%. The sensitivity coefficients were calculated for both ETBE and TBA in blood and exhaled air and along a time course of one day of exposure (8 h) followed by 16 h of no exposure. The physical activity was set to 8 h each of 50 W (during exposure), 25 W, and 0 W (at night). A dummy variable (Pierce et al., 1996), which had no effect on either blood or exhaled air levels in this model, was used to ensure the absence of false relationships in the analysis. The parameters that influenced the levels of ETBE in blood and exhaled air were mainly alveolar ventilation, the blood/air partition coefficient of ETBE, and parameters associated with the fat, such as body weight (Fig. 2). The sensitivity analysis indicated that those parameters which had the largest TABLE 2 Metabolic Parameter Values Estimated by Fitting the Physiological Model to Individual Experimental Data from Exposure to Ethyl tertiary-butyl Ether (ETBE) Metabolic parameters Subject BHt (m) BWt (kg) CLiE (L/min) CLiT (L/min) CLiM (L/min) Kel (min 1 ) A (0.6%) (0.8%) 0.10 (0.6%) 75 (0.8%) B (0.7%) (0.7%) (1.0%) 80 (0.7%) C (0.8%) (0.6%) (0.9%) 113 (0.7%) D (0.7%) (0.8%) (0.8%) 93 (1.1%) E (0.8%) (0.9%) 0.11 (0.8%) 68 (1.1%) F (0.7%) (0.8%) (1.1%) 104 (0.6%) G (0.6%) (0.5%) (0.6%) 97 (0.7%) H (0.7%) (0.7%) 0.10 (0.8%) 76 (0.8%) Average % CI Note. Exposures to 5, 25, and 50 ppm of Ethyl tertiary-butyl Ether (ETBE) (Nihlén et al., 1998). Individual average estimate with coefficient of variation (%, in parentheses) and average values and 95% confidence intervals (CI) are given. Symbols are explained in the Appendixes; n 8.

4 PBTK MODELING OF INHALED ETBE IN HUMANS 187 FIG. 2. The influence of changes in parameter values on arterial blood levels of ethyl tertiary-butyl ether (ETBE) and tertiary-butyl alcohol (TBA) is given as sensitivity coefficients (range). The sensitivity coefficients were calculated for a time course of exposure for 8 h followed by 16 h of no exposure. The physical activity was set to 8 h each of 50 W, 25 W, and 0 W. Symbols are explained in the Appendixes. (exactly at the termination of an 8-h exposure) and the next morning (24 h after the beginning of exposure and immediately prior to the next work shift). Branchial venous blood was assumed to correspond to the blood leaving the resting-muscle compartment. This assumption, used in a previous PBTK model (Johanson and Näslund, 1988), has been shown to agree well with experimental branchial venous-blood data obtained after acetone and methylene-chloride exposure during bicycling up to 150 W. Regarding the other biomarker levels (exhaled air, arterial blood levels of ETBE and TBA, and TBA in urine), the model has been validated against experimental ETBE and TBA data (Nihlén et al., 1998). Several different exposure scenarios were described using the present PBTK model and the estimated metabolite parameters. The partition coefficients in Table 1 and data obtained from one of the 8 volunteers in the chamber exposure (Subject A) were used in the 3 following exposure scenarios: 2 working weeks, work load, and fluctuating exposure. In order to approximate the interindividual variability in 8 subjects after a working week, each subject s individual metabolic parameters, body composition, and exposure levels were used. Software package. The simulations, 2 working weeks, work load, fluctuating exposure, and interindividual variability were all made with the software package Simnon (version 2.0, SSPA Maritime Consulting AB, Gothenburg, Sweden) with the Dorman-Prince algorithm. To ensure that the PBTK model written in Simnon was consistent with the WinNonlin model, simulations were performed in Simnon at 5, 25, and 50 ppm ETBE exposure (2 h, 50 W) and the results compared to the WinNonlin simulations. Two working weeks. In these simulations, the subject was exposed daily (5 days/week) to 50 ppm ETBE from 8 AM to noon and 1 PM to 5 PM. A physical activity of 50 W was assumed during the exposure. During the lunch break (noon to 1 PM), before the exposure (7 to 8 AM), after the exposure (until 11 PM) and during the weekend (7 AM to 11 PM) light physical activity (25 W) was assumed. At night (11 PM to 7 AM) work load was set to zero. Two weeks of 8-h daily exposure for 5 consecutive days/week was simulated. Work load. The effect of physical exercise was studied by varying the work load from rest (0 W) to heavy exercise (150 W) during an 8-h constant exposure (8 AM to 4 PM) to 50 ppm ETBE. Biomarker levels were calculated at the end of the work shift and the following morning prior to the next work shift. Fluctuating exposure. Fluctuating exposure during 8h(8AM to 4 PM) to 50 ppm ETBE (time-weighted average, TWA) was generated as previously influence on TBA levels in blood and exhaled air were associated with the body weight and the liver, including metabolic parameters and partition coefficients. Figure 3 illustrates the complex time-dependent behavior of the sensitivity coefficients. For example, during exposure, ETBE is transferred from blood to fat tissues, and hence the sensitivity coefficients of all fat tissue parameters (such as fat blood flow) are negative. However, after exposure, the sensitivity coefficients become positive as ETBE is transferred back from fat tissues to blood. Predictions in Humans Exhaled air, arterial, and venous blood levels of ETBE and TBA, and in addition, TBA in urine were all considered as potential biomarkers of ETBE and gasoline-vapor exposure. The present PBTK model was used to predict biomarker levels (blood, urine, and exhaled air) at the end of the work shift FIG. 3. Representative sensitivity analysis plots of the effects of parameters on predicted arterial blood levels of ethyl tertiary-butyl ether (ETBE) and tertiary-butyl alcohol (TBA). The magnitude of the sensitivity coefficients is proportional to the influence of each parameter on the blood concentrations. The sensitivity coefficients were calculated for a time course of exposure for 8 h followed by 16 h of no exposure. The physical activity was set to 8 h each of 50 W (during exposure), 25 W, and 0 W.

5 188 NIHLÉN AND JOHANSON FIG. 4. Experimental (dots) (Nihlén et al., 1998a) and predicted values (lines) for one individual (subject A) after exposure to 5, 25, and 50 ppm ethyl tertiary-butyl ether (ETBE) for 2hat50W.Symbols: E, ETBE in blood ( M),;, ETBE in exhaled air ( mol/min); F, tertiary-butyl alcohol (TBA, a metabolite) in blood ( M);, TBA in exhaled air ( mol/min); and, urinary excretion rate of TBA ( mol/min). described by Petreas et al. (1995). Two different exposure scenarios were simulated corresponding to continuous emission of ETBE and rapid air exchange (Scenario 1), and intermittent emission of ETBE and slow air exchange (Scenario 2) in the workplace. Scenario 1 was created with the exposure pattern of a log-normal distribution corresponding to a TWA of 50 ppm ETBE and a standard deviation of 25 ppm. The auto-correlation coefficient was set to 0.1. This coefficient accounts for a small dependence between air levels over time. In Scenario 2, the TWA was equal (50 ppm) but the standard deviation was set to 100 ppm and the auto-correlation coefficient to 0.9 (i.e., large dependence between air levels and time). The concentrations of ETBE in air were created in intervals of 2.5 min. The simulations were performed using a random generator in Simnon. The biomarker levels were obtained from 500 Monte Carlo simulations and reported at the end of the work shift and the following morning prior to next work shift. Interindividual variability. The interindividual variability of biomarker levels was predicted after exposure to 50 ppm ETBE (50 W) during one working week (8 PM to noon and 1 AM to 5 PM, for 5 days). The physical activity was assumed as described above (in the Two working weeks section). Individual data such as body composition, exposure levels, and metabolic parameters (estimated earlier with WinNonlin) were included for each subject in the PBTK model. The predictions were performed for all 8 subjects exposed experimentally to ETBE (Nihlén et al., 1998) using the Simnon software. PBTK Modeling RESULTS The PBTK model could adequately describe the previously published experimental data (Nihlén et al., 1998). As an example, the experimental and simulated results for one subject after ETBE exposure at 5, 25, and 50 ppm are given in Figure 4. Similar fits were obtained for the other 7 subjects (not shown). Estimated metabolism-parameter values for the 8 subjects are given in Table 2. Predictions in Humans Two working weeks. Biomarker levels are given after exposure to ETBE for 2 working weeks (Fig. 5). According to these simulations, there was a small accumulation of all biomarkers during a week of exposure. Low levels of TBA in urine and exhaled air were seen, whereas the level of ETBE in exhaled air was, as expected, higher. A slightly higher concentration in arterial blood compared to venous blood was seen (for both ETBE and TBA). Work load. Physical exercise markedly influenced the uptake of ETBE (Fig. 6). The work load increase from rest (0 W) to heavy exercise (100 W) increased all biomarker levels, approximately 2-fold at the end of the work shift and 3-fold the next morning. Fluctuating exposure. The effects of fluctuations in exposure levels during an 8 h exposure were estimated according to two different scenarios. The first scenario, which assumed continuous emission and rapid air exchange, predicted low variation in biomarker levels. In the second scenario, intermittent emission and slow air exchange were assumed, resulting in a substantial variability in blood levels of ETBE and TBA. These two scenarios are illustrated with three Monte Carlo simulations (Fig. 7). Results of the variability in biomarker levels calculated from 500 Monte Carlo simulations are given in Table 3. These simulations showed that the levels of ETBE in blood and exhaled air at the end of the work shift are highly

6 PBTK MODELING OF INHALED ETBE IN HUMANS 189 FIG. 5. Predicted levels of parent ether and a metabolite (tertiary-butyl alcohol, TBA) in blood, exhaled air and urine after two weeks (8 AM to noon and 1 PM to 5 PM, 5 consecutive days/week) of exposure to 50 ppm ethyl tertiary-butyl ether (ETBE) at a work load of 50 W. The work load was set to zero at night (11 PM to 7 AM), whereas light physical activity (25 W) was assumed for the remaining time. sensitive to fluctuations in exposure levels. However, ETBE levels the next morning and TBA levels in general are less sensitive to fluctuations. Interindividual variability. Predicted average values of biomarker levels in blood, urine, and exhaled air of 8 subjects are given after one work week of exposure (Table 4). The coefficient of variation (CV) illustrates the interindividual variability. The CV was low for ETBE in venous blood and exhaled air at the end of the work shift and prior to the next work shift the following morning. Low variabil- FIG. 6. The effect of physical exercise was studied by varying the work load from rest (0 W) to hard work (150 W) during a simulated 8-h constant exposure to 50 ppm ethyl tertiary-butyl ether (ETBE). After the exposure, the physical activity was set to 8hoflight physical activity (25 W) and 8hofrest (0 W). Biomarker levels of ETBE and a metabolite (tertiary-butyl alcohol, TBA) are shown at the end of the work shift (line) and next morning prior to the following shift (bold line).

7 190 NIHLÉN AND JOHANSON FIG. 7. Fluctuations in exposure levels (time-weight average of 50 ppm ETBE, 8 h, 50 W; after exposure 25 W) were predicted with 2 different exposure scenarios: (1) continuous emission and rapid air exchange and (2) intermittent emission and slow air exchange. The 2 scenarios are illustrated here by 3 Monte Carlo simulations. Fluctuations in exposure levels (upper graphs) and arterial blood levels of ethyl tertiary-butyl ether (ETBE) and tertiary-butyl alcohol (TBA) (lower graphs) are shown. ity was seen between subjects for TBA in venous blood, in exhaled air, and in the concentration of TBA in urine at the end of the work shift, but the variability was 2 3-fold higher the next morning. Further, the variability for TBA in urinary excretion rate was fold higher compared to urinary concentration (Table 4). Predictions from these 8 subjects show high variability of biomarker levels on Monday morning after one working week of exposure (30 40% for ETBE and % for TBA). DISCUSSION The aim of this study was to develop a PBTK model for inhalation exposure to ETBE in humans. The model was further used to predict biomarker levels under various exposure scenarios and to determine biomarker levels, in blood, exhaled air, or urine of the parent compound or the first metabolite, TBA. To our knowledge, this is the first PBTK model for ETBE, and there is no rodent study that describes the kinetics

8 PBTK MODELING OF INHALED ETBE IN HUMANS 191 TABLE 3 Variability in Biomarker Levels Resulting from 500 Monte Carlo Simulations of Random Fluctuating Exposure to 50 ppm Ethyl tertiary-butyl Ether (ETBE) at 50 W ETBE TBA Venous blood a Exhaled air b Venous blood a Exhaled air b Urine b Urine a Scenario 1 End of work shift Next morning Scenario 2 End of work shift Next morning Note. Scenario 1 corresponds to continuous ETBE emission with rapid air exchange, and Scenario 2 to intermittent emission with slow air exchange. The coefficient of variation (CV; %) illustrating the variability in biomarker levels is given for ETBE and tertiary-butyl alcohol (TBA) at the end of an 8-h work shift and the next morning prior to the next work shift (16 h after exposure). a M. b mol/min. for ETBE thoroughly enough that the data could be used in a PBTK model for dosimetry studies. Previously, one model has been developed for a related substance, MTBE, in order to predict the dosimetry between rodents and humans (Borghoff et al., 1996). The model by Borghoff et al. has been further refined to predict the toxicokinetics of human exposure to MTBE from contaminated water (Rao and Ginsberg, 1997). In developing the present PBTK model for human inhalation exposure, we used levels of ETBE and TBA in blood and exhaled air and TBA in urine. These levels were obtained during and after 3 short-term ETBE exposures (Nihlén et al., 1998a). Our PBTK model could adequately describe the experimental data (Fig. 4). One advantage of using experimental data is that all exposure conditions are known and well controlled, and usually several samples from one subject are available in the validation process. First order kinetics was assumed in this model, since linear kinetics for both ETBE and TBA was seen up to 50 ppm ETBE in the previous experimental study (Nihlén et al., 1998). For the ether analogue MTBE, linear kinetics have been reported in TABLE 4 Predicted Levels Using the PBTK Model for (ETBE) and (TBA) in Blood, Exhaled Air, and Urine after Exposure to 50 ppm ETBE 8 h Daily for Five Consecutive Days ETBE TBA Venous blood a Exhaled air b Venous blood a Exhaled air b Urine b Urine a End of shift Monday Tuesday Wednesday Thursday Friday CV (%) Next morning Tuesday Wednesday Thursday Friday Saturday Monday CV (%) Note. Mean values 95% confidence interval, n 8. Exposure was at a workload of 50 W. The variability (coefficient of variation, CV) is given as the average CV (Monday to Friday, end of work shift and Tuesday to Saturday, next morning). a M. b mol/min.

9 192 NIHLÉN AND JOHANSON humans up to 75 ppm MTBE (4-h exposure) (Pekari et al., 1996) and in rats up to 300 ppm MTBE (2 weeks of exposure) (Savolainen et al., 1985). These exposure levels are very high compared with measurements of MTBE in both occupational and non-occupational settings (Hakkola et al., 1996; HEI, 1996). This makes our assumption of first order kinetics plausible, and the model should be adequate for most exposure situations. Only a slight increase in biomarker levels was seen during a week of exposure to ETBE (Table 4) implying that biomarker levels mainly reflect the exposure during the same day. However, the biomarker levels were slightly higher after the second day compared with the first day of exposure, showing a small contribution from the previous exposure day. Physical exercise has a pronounced effect on ETBE and TBA levels as illustrated in Figure 6. An approximately 2-fold increase was seen in biomarker levels at the end of an 8-h work shift and a 3-fold increase the next morning when the work load increased from rest to heavy exercise. This illustrates the usefulness of biological exposure monitoring, since the internal exposure might be severely underestimated by simple air monitoring if physically demanding work tasks are performed. During physical exercise, pulmonary and alveolar ventilation increase via increases in respiratory rate and tidal volume. Cardiac output raises and is redistributed in that the perfusion of exercising muscles and adipose tissues enhances, while that of the liver and other splanchnic tissues and the kidney is reduced (Clewell and Andersen, 1996; Löf and Johanson, 1998). For chemicals with low partition coefficients, such as ETBE blood solubility, cardiac output, and metabolism, limit the uptake rate, whereas for chemicals with high partition coefficients, ventilation becomes rate limiting (Johanson 1997). During periods of no exposure, such as the evenings, lunches and weekends, all biomarker levels decrease (Fig. 5). However, enhanced physical exercise increases the exhalation of ETBE and TBA also when the subject is unexposed. This is best seen during the weekends when there is no exposure and the physical activity switches from light physical activity (25 W) during daytime to resting (0 W) during the night (Fig. 5). The two fluctuating exposure scenarios were arbitrarily chosen to correspond to conditions that may be expected in a work place. The model predicted that ETBE in blood and exhaled air was highly sensitive to exposure-level fluctuations at the end of the work shift. However, ETBE levels the next morning and TBA levels in the blood, exhaled air, and urine were less sensitive to fluctuations (Table 3). In these scenarios, only the exposure was varied and interindividual variability was not considered. The prediction of individual variation (data from 8 subjects) showed low variability in blood and exhaled air for ETBE and TBA at the end of the work shift, and in addition, for ETBE the next morning (Table 4). The interindividual variability of the urinary TBA excretion rates was 3 4-fold higher when compared with the variability of ETBE or TBA levels in blood and exhaled air, or in the urine concentration of TBA at the end of the work shift. Further, these predictions indicate that the interindividual variability of TBA levels was only slightly higher than the variability of ETBE levels in all biological samples. The excretion of a metabolite depends on several individual factors, e.g., uptake and disposition of the parent ether, metabolic capacity, and urine production, especially if TBA is excreted via passive diffusion. Urine sampling is generally preferable in biological exposure monitoring, since urine is easy to collect with a sampling technique far less invasive than blood sampling. In addition, urine samples are probably less sensitive to storage than blood samples. There may be both advantages and disadvantages to using TBA and the unmetabolized ethers as biological exposure markers for gasoline vapor. For example, TBA may be more appropriate than ETBE as a biological-exposure marker (for gasoline vapor) for the following reasons: (i) in a previous study it was shown that TBA is cleared very slowly from the body compared with the parent ether (Nihlén et al., 1998), and that TBA has one linear, single-elimination phase in blood and urine, whereas ETBE has several elimination phases, i.e. timing is less crucial for TBA than for the parent ether; (ii) TBA seemed to be less sensitive to fluctuations in exposure levels than the parent ether, especially at the end of the work shift, in the present Monte Carlo simulations; (iii) urinary recovery of TBA is higher than that of ETBE (Nihlén et al., 1998); and (iv) TBA is less volatile, i.e., less sensitive to sample losses. The timing of collecting the urine when TBA is analyzed is not crucial. For ETBE, however, the timing is most critical and therefore the unmetabolized ether should not be measured at the end of the work shift. A potential drawback to the use of TBA (or its metabolites) as a biomarker is that TBA itself might also be present in the environment (Te Koppele 1992): (i) TBA may be present in minor amounts in commercial oxygenates, as was the case in the ETBE study (0.4% TBA); (ii) TBA may be added as an oxygenate to gasoline; and (iii) TBA could be used for other purposes, e.g., as an industrial solvent. If there is reason to suspect exposure to TBA, then the unmetabolized ether may be considered more suitable as a biomarker. In this study, we have data only for the parent ether and the first metabolite. In a recent study from our laboratory, two additional urinary metabolites ( -hydroxyisobutyric acid and 2-methyl-1,2-propanediol) were characterized in humans after exposure to carbon13-labeled MTBE (Nihlén et al., 1999). These two metabolites occurred in markedly higher amounts in the urine than did TBA and MTBE. The same metabolites are most likely formed also after ETBE exposure (White et al., 1995) and further studies of other urinary metabolites may suggest even better candidates as biomarkers of gasoline vapor. In conclusion, a PBTK model was developed for inhalation exposure to ETBE in humans. Based on the predictions using the present PBTK model, previous toxicokinetic data (Nihlén

10 PBTK MODELING OF INHALED ETBE IN HUMANS 193 et al., 1998a) and practical considerations, we suggest the concentration of TBA in urine as an appropriate biomarker for exposure to ETBE and gasoline vapor. APPENDIX A Definition of Terms Used in the Model A Amount of TBA in urine ( mol) BHt Body height (m) BWt Body weight (kg) C Concentration of ETBE or TBA ( M) Cair Concentration in ambient air ( M). CLi Intrinsic hepatic clearance of ETBE or TBA (L/min) kel Excretion rate constant of TBA (min 1 ) Pblood Blood/air partition coefficient Pi Tissue/blood partition coefficient of compartment Qalv Alveolar ventilation (L/min) Qco Cardiac output (L/min) Qi Blood flow to compartment i (L/min) Vi Volume of compartment i (L) W Work load (watt) APPENDIX B Relationships between Physiological Variables (Adapted from Droz, 1992) Parameter (unit) Mathematical relationship Alveolar ventilation (L/min) Qalv 0.8 Qco(rest) 22 VOD Cardiac output (L/min) Qco Qh Qf Qr Qwm Qrm Free fat mass (kg) FFM TBW/0.72 Lean body volume (L) BV FFM/1.1 Oxygen uptake above rest (L/min) VOD 10.1 W/1000 Total body water (L) TBW BHt BWt Compartment volumes (L) Brain Vb BV Fat Vf (BWt - FFM)/0.92 Kidney Vk BV Liver Vh (TBW/0.72) Lungs and arterial blood Vl ( ) BV Muscle Vm BV Other tissues Vo BV Compartment blood flow (L/min) Rapidly perfused tissues (includes brain, kidneys, and other tissues) Qr 0.57 Vb Vk VOD 2.83 Vo Liver Qh ( ) Vh ( VOD) Resting muscle Qrm Vm Working muscle Qwm Qrm 7.38 VOD APPENDIX C Differential Equations for Rate of Change of ETBE or TBA in Model Compartments ETBE Compartments The concentration changes in the lung and arterial blood compartment (i indicates the following compartments: liver, fat, rapidly perfused tissues, working and resting muscle): dclung/dt Qalv (Cair-Clung/Pblood) (Qi Ci/Pi)-Qco Clung)/Vlung; the non-metabolizing compartments i (fat, rapidly perfused tissues, and working and resting muscle): and the liver compartment: dci/dt Qi (Clung - Ci/Pi)/Vi; dch/dt Qh (Clung Ch/Pliver) Ch (CLiETBE1 CLiETBE2))/Vh. TBA Compartments The concentration changes in the lung and arterial blood compartment (i indicates the following compartments; liver, fat, rapidly perfused tissues, working and resting muscle): dclung/dt (Qalv (CairTBA Clung/PbloodTBA) (Qi Ci/PiTBA) CLiETBE2 Ch - Qco Clung)/Vlung; the liver compartment: dch/dt (Qh (Clung Ch/PliverTBA) CLiTBA2 Ch)/Vh; the non-metabolizing compartments i (fat, working and resting muscle): Fat Qf Vf VOD dci/dt Qi (Clung Ci/PiTBA)/Vi;

11 194 NIHLÉN AND JOHANSON the rapidly perfused tissues compartment: dcr/dt (Qr (Clung Cr/PrTBA) (CLiTBA1 Cr))/Vr; and finally, the elimination compartment, the rate of change of TBA: da/dt CLiTBA1 Ch-kel A. ACKNOWLEDGMENTS We wish to acknowledge Fredrik Jonsson for valuable discussions. The project was financially supported by the National Board for Industrial and Technological Development (NUTEK), Sweden (Grant No. P3626 1). REFERENCES Andersen, M. E., Clewell, H. J., 3rd, and Frederick, C. B. (1995). Applying simulation modeling to problems in toxicology and risk assessment a short perspective. Toxicol. Appl. Pharmacol. 133, Åstrand, P.-O., and Rodahl, K. (1986). Textbook of Work Physiology: Physiological Bases of Exercise. McGraw-Hill, Singapore. Baker, G. L. (1969). Human adipose tissue composition and age. Am. J. Clin. Nutr. 22, Bernauer, U., Amberg, A., Scheutzow, D., and Dekant, W. (1998). Biotransformation of 12 C- and 2-13 C-labeled methyl tert-butyl ether, ethyl tert-butyl ether, and tert-butyl alcohol in rats: Identification of metabolites in urine by 13 C nuclear magnetic resonance and gas chromatography/mass spectrometry. Chem. Res. Toxicol. 11, Borghoff, S. J., Murphy, J. E., and Medinsky, M. A. (1996). Development of a physiologically based pharmacokinetic model for methyl tertiary-butyl ether and tertiary-butanol in male Fisher-344 rats. Fundam. Appl. Toxicol. 30, Clewell, H. J., 3rd, and Andersen, M. E. (1996). Use of physiologically based pharmacokinetic modeling to investigate individual versus population risk. Toxicology 111, Droz, P. O. (1992). Quantification of biological variability. Ann. Occup. Hyg. 36, Fiserova-Bergerova, V. (1983). Modeling of inhalation exposure to vapors: Uptake, distribution, and elimination. In Gases and Their Solubility: a Review of Fundamentals. (V. Fiserova-Bergerova, Ed.), Vol. 1, pp CRC Press, Boca Raton, Florida. Hakkola, M., Honkasalo, M. L., and Pulkkinen, P. (1996). Neuropsychological symptoms among tanker drivers exposed to gasoline. Occup. Med. 46, HEI (1996). The potential health effects of oxygenates added to gasoline: A review of the current literature. A special report of the institute s oxygenates evaluation committee. Health Effects Institute, Cambridge, MD. Johanson, G. (1997). Toxicokinetics: Modeling disposition. In Comprehensive Toxicology. (I. Sipes, A. Gandolfi, and C. McQueen, Eds.), Vol. 1, pp Amsterdam Elsevier, Amsterdam, The Netherlands. Johanson, G., and Näslund, P. H. (1988). Spreadsheet programming a new approach in physiologically based modeling of solvent toxicokinetics. Toxicol. Lett. 41, Järnberg, J., and Johanson, G. (1999). Physiologically based modeling of 1,2,4-trimethylbenzene inhalation toxicokinetics. Toxicol. Appl. Pharmacol. 155, Järnberg, J., Ståhlbom, B., Johanson, G., and Löf, A. (1997). Urinary excretion of dimethylhippuric acids in humans after exposure to trimethylbenzenes. Int. Arch. Occup. Environ. Health 69, Löf, A., and Johanson, G. (1998). Toxicokinetics of organic solvents: a review of modifying factors. Crit. Rev. Toxicol. 28, Nihlén, A., Löf, A., and Johanson, G. (1995). Liquid/air partition coefficients of methyl- and ethyl t-butyl ethers, t-amyl methyl ether, and t-butyl alcohol. J. Expos. Anal. Environ. Epidemiol. 5, Nihlén, A., Löf, A., and Johanson, G. (1998). Controlled ethyl tertiary-butyl ether (ETBE) exposure of male volunteers: I. Toxicokinetics. Toxicol. Sci. 46, Nihlén, A., Sumner, S., Löf, A., and Johanson, G. (1999). 13 C-labeled methyl t-butyl ether: Toxicokinetics and characterization of urinary metabolites in humans. Chem. Res. Toxicol., in press. Ong, C.-N., and Lee, B.-L. (1994). Determination of benzene and its metabolites: Application in biological monitoring of environmental and occupational exposure to benzene. J. Chromatogr. B Biomed. Appl. 660, Pekari, K., Riihimäki, V., Vainiotalo, S., Teräväinen, E., and Aitio, A. (1996). Experimental exposure to methyl-tert-butyl ether (MTBE) and methyl-tertamyl ether (MTAE). Proceedings from the International Symposium on Biological Monitoring in Occupational and Environmental Health, Hanasaari Cultural Centre, Espoo, Finland. Pelekis, M., Poulin, P., and Krishnan, K. (1995). An approach for incorporating tissue composition data into physiologically based pharmacokinetic models. Toxicol. Ind. Health. 11, Petreas, M. X., Woodlee, J., Becker, C. E., and Rappaport, S. M. (1995). Retention of styrene following controlled exposure to constant and fluctuating air concentrations. Int. Arch. Occup. Environ. Health 67, Pierce, C. H., Dills, R. L., Morgan, M. S., Nothstein, G. L., Shen, D. D., and Kalman, D. A. (1996). Interindividual differences in 2 H 8 -toluene toxicokinetics assessed by semiempirical physiologically based model. Toxicol. Appl. Pharmacol. 139, Rao, H. V., and Ginsberg, G. L. (1997). A physiologically based pharmacokinetic model assessment of methyl t-butyl ether in groundwater for a bathing and showering determination. Risk Anal. 17, Savolainen, H., Pfäffli, P., and Elovaara, E. (1985). Biochemical effects of methyl tertiary-butyl ether in extended vapour exposure of rats. Arch. Toxicol. 57, Te Koppele, J. M. (1992). 2-Methylpropanol-2. In Ethel Browning s Toxicology and Metabolism of Industrial Solvents (R. Thurman and F. Kauffman, Eds.), Vol. 3., pp Elsevier Science Publishers B. V., Amsterdam, The Netherlands. U.S. EPA (1993). Assessment of Potential Health Risks of Gasoline Oxygenated with Methyl tertiary Butyl Ether (MTBE). EPA/600/R-93/206, Office of Research and Development, U.S. Environmental Protection Agency, Washington, DC. White, R. D., Daughtrey, W. C., and Wells, M. S. (1995). Health effects of inhaled tertiary amyl methyl ether and ethyl tertiary butyl ether. Toxicol. Lett. 82/83,