A physiologically based toxicokinetic model of inhalation exposure to xylenes in Caucasian men J.C. Adams 1 , R.L. Dills 2 , M.S. Morgan 2 , D.A. Kalman 2 , C.H. Pierce * Department of Environmental and Occupational Health Sciences, University of Washington, Seattle, WA, USA Received 29 March 2005 Abstract Widespread exposure to the volatile aromatic hydrocarbons, ortho-, meta-, and para-xylene occurs in many industries including the manufacture of plastics, pharmaceuticals, and synthetic fibers. This paper describes the development of a physiologically based toxicokinetic model using biomonitoring data to quantify the kinetics of ortho-, meta-, and para-xylenes. Serial blood concentrations of deuterium-labeled xylene isomers were obtained over 4 days after 37 controlled, 2 h inhalation exposures to different concentra- tions of the isomers. Peak toxicant concentrations in blood occurred in all subjects at the termination of exposure. Systemic clear- ance averaged 116 L/h ± 34 L/h, 117 L/h ± 23 L/h, and 129 L/h ± 33 L/h for ortho-, para-, and meta-xylene, respectively. The half-life of each toxicant in the terminal phase (>90 h post-exposure) was fit by the model, yielding values of 30.3 ± 10.2 h for para-xylene, 33.0 ± 11.7 h for meta-xylene and 38.5 ± 18.2 h for ortho-xylene. Significant isomeric differences were found (p < 0.05) for toxicant half-life, clearance and extrahepatic metabolism. Inter-individual variability seen in this study suggests that airborne concentration guidelines may not protect all workers. A Biological Exposure Index is preferred for this purpose since it is integrative and reflective of inter-individual kinetic variability. ȑ 2005 Elsevier Inc. All rights reserved. Keywords: Physiologically based toxicokinetic model; PBTK; Xylene; Biological monitoring; Blood; Human exposure 1. Introduction Xylene, or dimethylbenzene, is a common lipophilic aromatic hydrocarbon that exists in three isomeric forms: meta-, para-, and ortho-xylene. Xylene is widely used as starting material in the manufacturing of chem- icals, pharmaceuticals, plastics, and synthetic fibers and has extensive uses as a solvent and paint thinner (ATS- DR, 1998). Xylene is highly volatile with a vapor pres- sure at room temperature of about 8 Torr and thus enters the body primarily through respiration. Vapors in the lung alveoli diffuse into the blood and are trans- ported throughout the body by the circulatory system and distributed among the tissues of the body (Sato, 1988). Depression of the central nervous system is the major toxicity associated with acute high doses of xylene (Low et al., 1989). Symptoms of chemical toxicity in- clude disturbances in manual coordination, prolonged reaction time, and irritation of the eyes and airways. Chronic occupational exposure has been associated with subjective findings of depression, fatigue, headache, anxiety, feeling of drunkenness, and sleep disorders. 0273-2300/$ - see front matter ȑ 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.yrtph.2005.07.005 * Corresponding author. Present address: Department of Public Health Professions, UW-Eau Claire, 105 Garfield Avenue, P.O. Box 4004, Eau Claire, WI 54702-4004, USA. Fax: +1 715 836 3379. E-mail addresses: [email protected](J.C. Adams), [email protected](R.L. Dills), [email protected](M.S. Morgan), [email protected](D.A. Kalman), [email protected](C.H. Pierce). 1 Present address: Department of Natural Science, St. Petersburg College—Gibbs Campus, P.O. Box 13489, St. Petersburg, FL 33733- 3489, USA. 2 Present address: Box 357234, University of Washington, Seattle, WA 98195-7234, USA. www.elsevier.com/locate/yrtph Regulatory Toxicology and Pharmacology 43 (2005) 203–214 Regulatory Toxicology and Pharmacology
Transcript
Regulatory
www.elsevier.com/locate/yrtph
Regulatory Toxicology and Pharmacology 43 (2005) 203–214
Toxicology andPharmacology
A physiologically based toxicokinetic model of inhalation exposure toxylenes in Caucasian men
J.C. Adams 1, R.L. Dills 2, M.S. Morgan 2, D.A. Kalman 2, C.H. Pierce *
Department of Environmental and Occupational Health Sciences, University of Washington, Seattle, WA, USA
Received 29 March 2005
Abstract
Widespread exposure to the volatile aromatic hydrocarbons, ortho-, meta-, and para-xylene occurs in many industries includingthe manufacture of plastics, pharmaceuticals, and synthetic fibers. This paper describes the development of a physiologically basedtoxicokinetic model using biomonitoring data to quantify the kinetics of ortho-, meta-, and para-xylenes. Serial blood concentrationsof deuterium-labeled xylene isomers were obtained over 4 days after 37 controlled, 2 h inhalation exposures to different concentra-tions of the isomers. Peak toxicant concentrations in blood occurred in all subjects at the termination of exposure. Systemic clear-ance averaged 116 L/h ± 34 L/h, 117 L/h ± 23 L/h, and 129 L/h ± 33 L/h for ortho-, para-, and meta-xylene, respectively. Thehalf-life of each toxicant in the terminal phase (>90 h post-exposure) was fit by the model, yielding values of 30.3 ± 10.2 h forpara-xylene, 33.0 ± 11.7 h for meta-xylene and 38.5 ± 18.2 h for ortho-xylene. Significant isomeric differences were found(p < 0.05) for toxicant half-life, clearance and extrahepatic metabolism. Inter-individual variability seen in this study suggests thatairborne concentration guidelines may not protect all workers. A Biological Exposure Index is preferred for this purpose since it isintegrative and reflective of inter-individual kinetic variability.� 2005 Elsevier Inc. All rights reserved.
Keywords: Physiologically based toxicokinetic model; PBTK; Xylene; Biological monitoring; Blood; Human exposure
1. Introduction
Xylene, or dimethylbenzene, is a common lipophilicaromatic hydrocarbon that exists in three isomericforms: meta-, para-, and ortho-xylene. Xylene is widely
0273-2300/$ - see front matter � 2005 Elsevier Inc. All rights reserved.
doi:10.1016/j.yrtph.2005.07.005
* Corresponding author. Present address: Department of PublicHealth Professions, UW-Eau Claire, 105 Garfield Avenue, P.O. Box4004, Eau Claire, WI 54702-4004, USA. Fax: +1 715 836 3379.
College—Gibbs Campus, P.O. Box 13489, St. Petersburg, FL 33733-3489, USA.2 Present address: Box 357234, University of Washington, Seattle,
WA 98195-7234, USA.
used as starting material in the manufacturing of chem-icals, pharmaceuticals, plastics, and synthetic fibers andhas extensive uses as a solvent and paint thinner (ATS-DR, 1998). Xylene is highly volatile with a vapor pres-sure at room temperature of about 8 Torr and thusenters the body primarily through respiration. Vaporsin the lung alveoli diffuse into the blood and are trans-ported throughout the body by the circulatory systemand distributed among the tissues of the body (Sato,1988). Depression of the central nervous system is themajor toxicity associated with acute high doses of xylene(Low et al., 1989). Symptoms of chemical toxicity in-clude disturbances in manual coordination, prolongedreaction time, and irritation of the eyes and airways.Chronic occupational exposure has been associated withsubjective findings of depression, fatigue, headache,anxiety, feeling of drunkenness, and sleep disorders.
204 J.C. Adams et al. / Regulatory Toxicology and Pharmacology 43 (2005) 203–214
Langman (1994) found that chronic exposure was alsolinked to anemia, leukopenia, electrocardiogram abnor-malities, dyspnea, and cyanosis.
To protect workers from toxicity, airborne xyleneconcentrations are measured and then compared withguidelines such as the Time-Weighted Average Thresh-old Limit Value (TWA-TLV) and Permissible ExposureLimit (PEL). However, actual body burden may varygreatly among equivalently exposed workers due to indi-vidual differences in the dose received (caused by differ-ences in ventilation rates), as well as differences in theabsorption, metabolism, and distribution of the toxi-cant. These differences in toxicokinetics are related tomany population variables including genotype, bodysize and composition, gender, life style factors such asalcohol consumption and smoking, and nutritionaland health status (Langman, 1994; Pierce et al., 1998).Biological indicators of exposure, such as toxicant levelsin the blood or metabolite levels in the urine, reflectthese individual differences (Pierce et al., 1998). In fact,biological indicators provide a better basis for estima-tion of the toxicologically active dose as compared tomeasured airborne levels.
For many years, researchers have made great effortsto understand and predict solvent toxicokinetics in hu-mans (Holm and Nordqvist, 1985; Sato et al., 1989;and Sedivec and Flek, 1976). Recent advancement incomputational hardware gives us greater capability toutilize more complex and realistic physiologically basedtoxicokinetic (PBTK) models. These predictive modelsaim to account for person-specific factors that influencetoxicokinetics, which in turn allow for a better assess-ment of biological levels of the chemical following expo-sure (Jang and Droz, 1997; Tardif et al., 1995).However, since they attempt to closely reproduce ana-tomical structures and physiological events, PBTK mod-els often have a high number of parameters, some ofwhich are difficult to measure. Past studies have simplysubstituted or scaled, where possible, values for param-eters into a PBTK model obtained from animal studies(Tardif et al., 1995). Such interspecies extrapolation ofphysiologic parameters includes large uncertainty (He-trick et al., 1991; Mordenti, 1986). In contrast, modelsbased on biological monitoring from controlled humanexposures have presumably greater utility in the humanpopulation. PBTK models have an enormous potentialfor providing a sound theoretical basis for the extrapo-lation and generalization of controlled exposures toworker and general populations.
While PBTK modeling enables us to make predic-tions with limited experimental data using educatedassumptions, it is important to challenge the validityof the model with experimentation. Our researchdevelops a PBTK model using biomonitoring data toquantify the human body kinetics of ortho, meta, andpara-xylenes. Samples of exhaled breath and blood
concentrations of deuterium-labeled xylene isomers wereobtained for each of 27 Caucasian men who underwentrepeated controlled inhalation exposure to xylene iso-mers. The isomers were isotopically labeled to distin-guish administered dose from background exposures(Baines et al., 2004). Individual measurements of age,height, body weight, and adipose tissue fraction wererecorded, and the latter two used in the model whichwas intended to closely predict the kinetics of the xyleneisomers in adult men, and was tested using experimentaldata. Results from previous sensitivity analyses (Hetricket al., 1991; Pierce et al., 1996a) were used to identifyinfluential parameters, as well as to guide model simpli-fication. A Bayesian approach was used to help deter-mine a best-fit model by accounting for the meanvalue and error for each influential model parameter.Our goals were to test for structural problems in themodel and to provide more realistic estimates of fittedor measured parameter precision, using subject-specificdata instead of population averages when possible. Wehave previously used this approach with toluene (Pierceet al., 1996a). We also aimed to quantify inter-individualdifferences in the toxicokinetics of ortho-, meta-, andpara-xylene.
2. Subjects and methods
2.1. Subjects
Twenty-seven men ages 21–49 were recruited byadvertisement at the University of Washington and giv-en a self-administered questionnaire to screen for vola-tile organic chemical exposure. Only Caucasiansvolunteered for the study. Eight subjects participatedin two to four replicate exposures (with at least 2 weekselapsing between exposures) for a total of 37 experi-ments. Smokers, those with chronic illness or on chronicmedication, and those with occupational solvent expo-sure were excluded from this study. All subjects deniedhistory or current symptoms of anemia, acute respirato-ry infection , rhinitis or asthma, which would have lim-ited their participation. This study was approved by theUniversity of Washington Human Subjects Division.Weight (59.7–109.1 kg) and height (168–196 cm) foreach subject were measured. The fraction of the bodythat was adipose tissue (11–39%) was estimated by askin fold method using Lange calipers (Durnin andWomersley, 1974). Inhaled xylene isomer-specific con-centrations were also measured (Morgan et al., 1993).
2.2. Exposure and sampling
Subjects inhaled a humidified atmosphere containing0–40 ppm of each of 2H10-ortho,
2H10-meta, and 2H10-para-xylene through a gated mouthpiece for two hours
J.C. Adams et al. / Regulatory Toxicology and Pharmacology 43 (2005) 203–214 205
at rest. Of the total 37 exposures, four were to 2H10-ortho,and four were to 2H10-ortho and 2H10-para isomers only.A continuous sample of the exposure atmosphere fromeach exposure was collected using charcoal tube samplesfor subsequent gas chromatography–mass spectrometrymeasurement of exact xylene isomer concentrations. In-haled and exhaled concentrations of total xylenes weremeasured continuously during the exposures by a photo-ionization detector. Exhaled breath flow rates were mon-itored continuously by a pneumotachograph whosesignal was integrated to produce exhaled volumes as10 min averages by Labview software (National Instru-ments, Austin, TX). The dose was measured as the differ-
Table 1Values and distributions of parameters used in the PBTK Model
a Tissue:blood partition coefficients were calculated as (Tissue:air partition
ence of the inhaled and exhaled concentrations times theexhaled volume. Doses for each isomer were also estimat-ed for all subjects by multiplying exposure concentrationby exposure duration (2 h) and the model-fitted subject-specific alveolar ventilation rate.
One antecubital venous blood sample was taken priorto exposure for determination of pre-exposure concen-trations of xylenes. Fifteen to 20 blood samples weretaken after the end of exposure over the following 4days, with a frequency that varied from every 15 minimmediately following exposure to every 12 h at timesgreater than 24 h after exposure. Blood samples werecollected in Vacutainer tubes containing citrate.
el assignment Mean (Range) SD Source
ct-specific (59.7–109.1) Measured
* (BW)0.74 (190–458) Pierce et al. (1996a)sian 13.94 2.79 Pierce et al. (1996a) (mean)
Thomas et al. (1996) (SD)
* (BW)0.74 (266–416) Pierce et al. (1996a)12.92 Thomas et al. (1996)
* QC (36–104) Pierce et al. (1996a)sian 0.25 0.055 Thomas et al. (1996)
* QC (10–33) Pierce et al. (1996a)sian 0.05 0.0075 Thomas et al. (1996)C � QL (135–248) Pierce et al. (1996a)C � QF (47–73) Pierce et al. (1996a)
* BW0.023 Pierce et al. (1996a)
ct-specific (11.3–39.0) MeasuredBW � VF Pierce et al. (1996a)BW � VL Pierce et al. (1996a)
1.89 Tardif et al. (1995)
c * (BW)0.7 Pierce et al. (1996a)
79.24 Tardif et al. (1995)
pc * (BW)0.7 Pierce et al. (1996a)
stable (0–9.72)
sian 35.1 2.48 Pierce et al. (1996b)sian 68.4 17.8 Pierce et al. (1996b)
3.09 Gargas et al. (1989)3.09 Gargas et al. (1989)
stable (0.01–20) Gargas et al. (1989)
sian 31.6 3.08 Pierce et al. (1996b)sian 56.5 14.6 Pierce et al. (1996b)
2.8 Gargas et al. (1989)2.8 Gargas et al. (1989)
stable (0.01–20) Gargas et al. (1989)
sian 38.4 4.26 Pierce et al. (1996b)sian 47.6 11.7 Pierce et al. (1996b)
2.01 Gargas et al. (1989)2.01 Gargas et al. (1989)
stable (0.01–20) Gargas et al. (1989)
coefficient)/(Blood:air partition coefficient).
Fig. 1. Diagram of the physiologically based toxicokinetic (PBTK)model for xylene distribution, where Cinh represents the inhaled xyleneconcentration, Calv represents the concentration in alveolar air, Cven
represents the venous blood concentration, and Cart represents thearterial blood concentration. Blood flows and partition coefficients aredefined in Table 1.
206 J.C. Adams et al. / Regulatory Toxicology and Pharmacology 43 (2005) 203–214
2.3. Chemical analysis
Standard curves (0.2–200 ng/ml [1.9–1900 nmol/L]blood) were generated by weighted (1/x) linear regres-sion for each isomer using each subject�s baseline bloodsample. Blood samples were analyzed in triplicate foreach of the xylenes by a headspace method (Dillset al., 1991), using gas chromatography with mass spec-trometry detection in selected ion mode, which quanti-fied each isomer with the signal m/z 98. The internalstandard was 1,3,5-trimethylbenzene (m/z 120). In allsubjects, background levels of proteo-xylenes could bedetected. At the lower calibration points, the CV wasapproximately 10–15%. The limit of quantification wasthe lowest standard (0.2 ng/ml [1.9 nmol/L]). The sam-ples for which we could measure blood concentrationsbeyond 100 h were not from subjects exposed to the low-est concentrations of xylene.
2.4. Physiologically based model
A semi-empirical PBTK model was constructed withSAAM II software (SAAM Institute, Seattle, WA) touse subject-specific values of body weight, adipose tissuefraction, and exposure concentration. Values of frac-tional tissue compartment volumes (except measuredadipose tissue), fractional blood flows, tissue/blood par-tition coefficients, and hepatic metabolic constants weretaken from the literature sources (Table 1). Cardiac out-put, and therefore all blood flows, were scaled to (bodyweight)0.74. A minimum number of compartments werechosen to define the model: blood, liver, fat, rapidly per-fused tissue, and slowly perfused tissue (Fig. 1). Anassumption of 1 kg/L density for all tissues was used.As a small lipophilic molecule, xylene was expected torapidly diffuse across tissue membranes; the exchangeof xylene between arterial blood and tissue groups wastherefore treated as flow-limited. Mass transfers withinthe PBTK model were defined by a series of simulta-neous differential equations (Table 2). Changes ofamount within the liver included a Michaelis–Mententerm for toxicant metabolism, likely due to P450 2E1activity. Metabolism in the lung was also allowed, if itwas found that apparent systematic clearance of xyleneswas well in excess of hepatic blood flow. Previous studieshave found that CYP450 enzymes in the rabbit lung canmetabolize a substantial portion of inhaled xylenes(Smith et al., 1982). The clearance, volume of distribu-tion at steady-state, and volume of distribution andhalf-life in the terminal phase of each isomer was calcu-lated for each subject�s exposure (Table 3).
2.5. Bayesian approach
A Bayesian strategy was employed during fitting tooptimize parameters with well-characterized distribu-
tions (Vicini et al., 1999). This strategy took into ac-count the mean value and the error (standarddeviation) associated with each parameter, and penal-ized fit values farther from the mean. Alveolar ventila-tion rates, blood flows, and blood/air partitioncoefficients were assigned as Bayesian parameters (Table1). Parameters with poorly understood distributions,such as extrahepatic metabolism, were assigned asadjustable, allowing the model to freely fit any valuewithin the specified range without penalty. The rangefor extrahepatic metabolism included zero and allowedfor up to 25% of hepatic metabolism. The slowly per-fused tissue/blood partition coefficient was allowed tovary between 0.01 and 20, allowing for the variability(10% C.V.) seen by Gargas et al. (1989) using rat muscleand a variety of solvents, with additional factors repre-senting uncertainty for species extrapolation and inher-ently greater variability in humans compared to inbredstrains of rats.
2.6. Chemicals
2H10-ortho-,2H10-meta-, and 2H10-para-xylene (98%
purity) were obtained from Cambridge Isotope Labora-tories (Woburn, MA).
Table 3Equations used for calculation of toxicant clearance, volume of distribution at steady-state, and volume of distribution and half-life in the terminalphase
Clearance = Inhaled dose (lmol)/area under blood concentration curve (AUC) (lmol h/L)Volume of distribution at steady-state (Vss) = BV + V1Kp,1 + V2Kp,2 + V3Kp,3 + V4Kp,4
where V is the volume of each tissue group, and Kp is the tissue/blood partition coefficient for each isomer of xyleneVolume of distribution (Vd) = (BV * s1 + q1 + q2 + q3 + q4)/s1where BV is the total blood volume (L), s1 is the toxicant concentration in model-fitted venous blood (lmol/L), and q1, q2, q3, and q4 are the amounts(lmol) of toxicant in the liver, adipose tissue, slowly perfused tissue and rapidly perfused tissue, respectively.
Toxicant half-life ðt1=2Þ ¼ln 2 � Amount xylenes in body
ðRate of hepatic metabolismþRate of exhalationþRate of lung metabolismÞ
where (s1 + CA) is the toxicant concentration in model-fitted venous and arterial blood (lmol/L), BV is the total blood volume, q1, q2, q3, and q4 arethe amounts (lmol) of toxicant in the liver, adipose tissue, slowly perfused tissue and rapidly perfused tissue, respectively. PL is the toxicant liver/blood partition coefficient. QP is the alveolar ventilation rate (L/h) and s7 is the toxicant concentration in the model-fitted alveolar breath (lmol/m3).
Table 2Differential equations defining mass transfers in the PBTK Model
Changes in the amount of xylenes in nonmetabolizing tissues (slowly perfused, rapidly perfused, and adipose) were given by
dAt=dt ¼ QtCa � ðQtAt=KpV tÞ;
where At is the amount of xylenes in the tissue (lmol); Qt is the blood flow to tissue (L/h); Ca is the concentration of xylenes in arterial blood (lmol/L); Kp is the tissue/blood partition coefficient for each isomer of xylene; and Vt is the tissue volume (L)Changes in the amount of xylenes in the liver included a Michaelis–Menton term for metabolism by P450 2E1 enzymes and were determined by
where Vmax is the maximum rate of metabolism (lmol/h); and Km is the affinity constant for metabolism (lmol/L)Mass balance in the lung was defined as the rate of change between venous and arterial blood across the lung, equaling the rate of absorption less therate of metabolism (Pierce et al., 1996a):
where Cv is the concentration of xylenes in venous blood (lmol/L); Cinh is the inhaled concentration of xylenes (lmol/L); Calv is the alveolarconcentration of xylenes (lmol/L); and CLint-p is the intrinsic clearance of xylenes in pulmonary tissues, defined as Vmaxp/Km (lmol/h).
J.C. Adams et al. / Regulatory Toxicology and Pharmacology 43 (2005) 203–214 207
3. Results
Fitted alveolar ventilation rates across all exposuresand isomers ranged from 142 to 469 L/h with an aver-age rate of 307 L/h and standard deviation of 74 L/h.Fitted blood flow to the liver varied from 35.9 to104.3 L/h, representing an average of 23.7% of cardiacoutput (Table 4). Corresponding blood flow to otherrapidly perfused tissues averaged 52.3% of cardiac out-put, with an average of 175 ± 27 L/h. Fitted blood flowto adipose tissue varied from 10.1 to 32.9 L/h, repre-senting an average 6.2% of cardiac output. Corre-sponding blood flow to slowly perfused tissueaveraged 17.8% of cardiac output, with an average of58.8 ± 6.6 L/h.
The fitted values for extrahepatic metabolism weregreater than zero for ortho-xylene in 38% of subjects ex-posed and ranged from 0.18 to 9.72 lmol/h kg. Theaverage for all subjects (including values of zero) was16.8 lmol/h for a 70 kg man (Table 5, 0.861 · 700.7).
para-Xylene had similar fitted values of extrahepaticmetabolism, ranging from 0.24 to 6.76 lmol/h kg in48% of subjects exposed, with an average of17.9 lmol/h for a 70 kg man (Table 6, 0.913 · 700.7).Fitted values for extrahepatic metabolism for meta-xy-lene ranged from 0.93 to 6.02 lmol/h kg (Table 7) andwere found in 58% of the subjects exposed. Extrahepaticmetabolism for meta-xylene was highest amongst theisomers with an average of 22.0 lmol/h for a 70 kgman (1.122 · 700.7).
Blood/air partitioning for meta-, para-, and ortho-xy-lene had fitted values of 31.6 ± 0.1, 38.5 ± 0.1, and35.2 ± 0.1, respectively. Adipose tissue/blood partition-ing was 43.4 ± 15.1 for para-xylene, 52.7 ± 19.7 forortho-xylene, and 48.8 ± 16.4 for meta-xylene. Fittedvalues for partitioning between slowly perfused tissueand blood had higher variability, and ranged from0.65 to 17.44 for ortho-xylene (2.88 ± 3.37), 0.55–12.27for para-xylene (2.25 ± 2.58), and 0.68–13.78 for meta-xylene (3.05 ± 3.77).
Table 4Subject anthropometric and model-fitted characteristics
a Value given is the mean (standard deviation) for fitted values from the model fit to the data for each isomer.b Subject was exposed to all three isomers.c Subject only exposed to d10-ortho-xylene.d Subject exposed to d10-ortho- and d10-para-xylene.e For this subject, QLC was set to 0.27 for all three isomers to provide a reasonable model fit.
208 J.C. Adams et al. / Regulatory Toxicology and Pharmacology 43 (2005) 203–214
Steady-state volumes of distribution (Vss) were as fol-lows: para-xylene (887 ± 372 l) < meta-xylene(1010 ± 433 L) < ortho-xylene (1067 ± 447 L) (Tables5–7). Terminal volumes of distribution (Vd) werepara-xylene (3681 ± 1401 L) < ortho-xylene (4695 ±3449 L) < meta-xylene (6170 ± 7954 L). Systemic clear-ance of ortho-xylene was 116 L/h ± 34 L/h (Table 5),for para-Xylene was 117 L/h ± 23 L/h (Table 6), andwas highest for meta-xylene at 129 ± 33 L/h (Table 7).The half-life of each isomer was estimated in the modeledterminal phase (>90 h post-exposure) for each subject�sexposure. The terminal half-life of para-xylene was short-
est, at 30.3 ± 10.2 h (Table 6), of meta-xylene was inter-mediate, at 33.0 ± 11.7 h (Table 7), and of ortho-xylenewas longest, at 38.5 ± 18.2 h, (Table 5). Significant(p < 0.05) isomeric differences were found for toxicanthalf-life, clearance and extrahepatic metabolism using apaired two-sided t test.
The modeled and observed peak blood concentrationsfollowing the two-hour exposure were highest for everysubject exactly at the end of exposure. A plot of a ‘‘typi-cal’’ model fit is shown in Fig. 2. In most cases, residualswere less than 10% of measured blood concentrationlevels. Measured blood concentrations of 2H10-meta-,
Table 5Descriptive kinetic parameters for subjects exposed to 2H10-ortho-xylene
Subject Volume of distribution (L) Volume at steady state (L) Dose (lmol) t1/2 (h) CL (L/h) Peak level (lmol/L) Vmaxpca (lmol/h kg)
J.C. Adams et al. / Regulatory Toxicology and Pharmacology 43 (2005) 203–214 209
-ortho-, and -para-xylene for all subjects can be seen inFigs. 3–5. Instrumental problems precluded the measure-ment of some of the inhaled–exhaled dose estimation formeta- and para-xylene isomers. In general, the dose esti-mates from this calculation tended to be higher thanthose estimated from the fitted ventilation rate approach(Fig. 6).
4. Discussion
Toxicant elimination from the body can be describedin physiological terms as clearance, the volume of bloodfrom which the chemical is completely removed per unittime (Nies et al., 1976). Model-fitted values of systemicclearance of the xylenes had inter-individual coefficients
of variation of 29, 23, and 25% for d10-ortho-, d10-para-,and d10-meta-xylene, respectively. Average fitted valuesfor systemic clearance of ortho-xylene (116 L/h), para-xylene (119 L/h), and meta-xylene (131 L/h) were veryclose to an average metabolic clearance found previous-ly by Sato (1988) for meta-xylene (116 L/h). Fitted clear-ance values obtained in this study also agreed withvalues obtained in work done by Wallen et al. (1985),which averaged about 2.1 L/h kg depending on subject�sexposure concentration. The higher blood clearance ofthe meta isomer was consistent with findings of higherurinary clearance of meta-xylene by Miller and Edwards(1999), for reasons not fully understood.
The steady-state volumes of distribution for the threeisomers were lower than corresponding terminal vol-umes, as expected for a toxicant that is stored in a tissue
Table 6Descriptive kinetic parameters for subjects exposed to 2H10-para-xylene
Subject Volume of distribution (L) Volume at steady state (L) Dose (lmol) t1/2 (h) CL (L/h) Peak level (lmol/L) Vmaxpca (lmol/h kg)
210 J.C. Adams et al. / Regulatory Toxicology and Pharmacology 43 (2005) 203–214
such as fat. The greater variance in terminal volumes(C.V. values of 38–129%) compared to steady-state vol-umes (all about 42%) may have reflected greater analyt-ical variability in the measurement of terminal samples,which were used to model the terminal phase ofdisposition.
Our study found that the terminal half-life, a goodindicator of the total duration of internal exposure ofthe xylene isomers, had high inter-individual variabilityamong subjects, ranging from 12 to 108 h. The averageterminal half-life for the three isomers among all subjects(34 h) was similar to a value of 29 h found for toluene(methylbenzene), an aromatic hydrocarbon with similarstructure and the same metabolic pathway, by Pierceet al. (1996a). Using an approach from Nies et al.(1976) (t1/2 = 0.693 * Vd/CL), an average terminal half-life was calculated for each isomer, using mean clearanceand volume of distribution values for ortho-, meta-, andpara-xylene, giving values of 28, 33, and 22 h, respective-ly. The relatively high inter-subject variability of the
terminal half-life (C.V. = 37%) was also seen in the bio-monitoring work of toluene (C.V. = 33%, Pierce et al.,1996a).
The average ventilation rate among our subjects was307 L/h, lower than the average of 356 L/h measuredearlier by Pierce et al. (1996a), but 38% of fitted ventila-tion rates were within one standard deviation of themean found in this previous work. Our average ventila-tion rate did agree with values of 310 and 300 L/h usedin previous modeling work by Droz et al. (1989) andCsanady et al. (1994), respectively. A total of 33% of fit-ted QFC values and 85% of fitted QLC values werewithin one standard deviation from the mean used by Si-mon (1997). Average blood flow to adipose tissue (21 L/h) was similar to the value of 20.4 L/h used in modelingwork by Droz et al. (1989). Mean blood flow to slowlyperfused tissues (59 L/h) is very close to values of 60and 59 L/h used in modeling work by Johanson (1991)and in work by Perbellini et al. (1986), respectively.Average blood flow to rapidly perfused tissues was
Table 7Descriptive kinetic parameters for subjects exposed to 2H10-meta-xylene
Subject Volume of distribution (L) Volume at steady state (L) Dose (lmol) t1/2 (h) CL (L/h) Peak level (lmol/l) Vmaxpc (lmol/h kg)a
Fig. 2. ‘‘Typical’’ fit for labeled ortho-xylene exposure. Subjectexposed to 13.3 ppm 2H10-ortho-xylene for two hours. First bloodsample taken one minute after termination of exposure. CV and s1 aremeasured and model-fitted blood concentration levels of labeled ortho-xylene, respectively.
Fig. 3. Measured blood levels of 2H10-meta-xylene in 27 subjectsexposed at varying concentrations (4.8–25 ppm) for 2 h at rest.
J.C. Adams et al. / Regulatory Toxicology and Pharmacology 43 (2005) 203–214 211
175 L/h, close to a value of 183 L/h used by Koizumi(1989). Average blood flow to the liver was 78 L/h,which agreed with a value of 84 L/h used in modeling
work by Kumagai and Matsunaga (1995), and substan-tiated the potential of extrahepatic metabolism, sinceclearance in most subjects was well above hepatic bloodflow. In our study, extrahepatic metabolism was fitted asgreater than zero in 59% of exposures, and averaged1.3% of metabolism in the liver (including values ofzero), with a peak fitted value that was 12% of livermetabolism (Table 1). This parameter was found to belowest for ortho-xylene and highest for meta-xylene,
Fig. 4. Measured blood levels of 2H10-ortho-xylene in 35 subjectsexposed at varying concentrations (6.4–39 ppm) for 2 h at rest.
Fig. 5. Measured blood levels of 2H10-para-xylene in 33 subjectsexposed at varying concentrations (10–30 ppm) for 2 h at rest.
212 J.C. Adams et al. / Regulatory Toxicology and Pharmacology 43 (2005) 203–214
mirroring the trend for systemic clearance. Metabolismin rabbit lungs of para-xylene has been reported bySmith et al. (1982).
The metabolic parameters Vmax (mg/h) was convertedusing an allometric relationship, scaling to body weight(BW0.7), and Km (mg/L) for xylene metabolism was as-sumed to be the same among rats and humans (Tardifet al., 1995). Metabolic parameters obtained from hu-man liver studies were preferred for inclusion in thiswork; however, isomer-specific Vmax averages deter-
Fig. 6. Xylene isomer dose (¤, ortho-xylene; n, meta-xylene; m, para-xylemodel-fitted alveolar ventilation rate * concentration. A line of equivalence
mined from work by Tassaneeyakul et al. (1996) were15–20 times greater than those found in the rat studies.Inclusion of the human data resulted in our model fit-ting blood flows at minimal values not physiologicallypossible, suggesting the metabolic parameters were toohigh. It was then decided to use the rat data, whichhas been used in earlier human modeling work (Tardifet al., 1995) and allowed for more reasonable fits withour model.
Within-exposure coefficients of variation of about5–10% for fitted ventilation rate, hepatic blood flow,and adipose blood flow (Table 4) provided evidenceof model reliability. However, it is not clear why mea-sured dose tended to be twice as high as model-esti-mated dose (Fig. 6). While a lower optimized valueof alveolar ventilation rate could have actually reflect-ed greater metabolism and elimination, the mean valueof 307 L/h (Table 4) is close to measured values (i.e.,5 L/min alveolar ventilation compared to 7.4 L/minminute ventilation Diem, 1962). The monitor for ex-haled breath concentrations of xylene was operatingnear its limit of detection. Low values for exhaledbreath concentration would cause a positive bias inthe dose calculations.
Average partitioning (PB) between arterial blood andalveolar air fitted in our exposures were 38.5, 35.2, and31.6 for para-, ortho-, and meta-xylene, respectively.All values were within one standard deviation of meanvalues reported previously in human studies done byPierce et al. (1996b). In fact, inter-subject variabilityfor PB was very low, with coefficient of variations of0.38, 0.24, and 0.26% for para-, ortho-, and meta-xylene,respectively. Values for the adipose tissue/blood parti-
ne) measured as (inhaled � exhaled) * concentration, vs. estimated as(x = y) is included for reference.
J.C. Adams et al. / Regulatory Toxicology and Pharmacology 43 (2005) 203–214 213
tion coefficient (PF) showed more inter-subject variabil-ity with only 46% (38%, 45%) of subject-fitted PF valuesfalling within one standard deviation of a mean valuereported by Pierce et al. (1996b) for para-(ortho-,meta-) xylene. Human data were not found in the liter-ature for partitioning between slowly perfused tissueand blood (PS). We conducted a small experiment withfour pieces of human muscle obtained from pathologyand found partitioning in the muscle to be 10–15 timeshigher than what was determined from previous ratstudies by Gargas et al. (1989). Our modeling workfound average fitted PS values that were 2–3 times thosedetermined in the rat studies, but for each isomer, therewas high variability (C.V. > 100%). The increased vari-ability of partitioning between fat or muscle (slowly per-fused tissue) and blood (or air) seen in this work is notwell understood, but coefficients of variation up to40% have been seen before with some solvents (Gargaset al., 1989).
Intra-individual differences were seen in both physio-logical and kinetic parameters in this study, and manyfactors could contribute to this variability. Any alter-ation of respiratory functions has a profound effect onthe uptake, distribution and elimination of volatile sub-stances (Sato, 1988). One subject had a coefficient ofvariation of 26% for all fitted ventilation rates acrossfour separate exposures. Similar variability(C.V. = 27%) was seen by Pierce et al. (1996a) in threemeasured ventilation rates of one of their subjects. In-tra-subject variability was also seen in the terminal vol-ume of distribution and toxicant terminal half-life, bothof which are related to intra-subject variability in sys-temic clearance (Nies et al., 1976). The observed vari-ability may be due to variations in blood flows to andpartitioning in tissue compartments, and/or to increasedanalytical variability seen in measuring low, terminalphase concentrations. Our small study of four humanmuscle tissue samples found significant (p < 0.05) in-tra-individual differences in solvent partitioning(C.V. 6 50%). Hepatic blood flow can vary over a four-fold range (from one-half to twice normal flow) due topathological and physiological variables, such as chang-es in posture, thermal stress, and exercise (Nies et al.,1976). Furthermore, daily differences in metabolism ofaromatic hydrocarbons can be explained by diet andalcohol consumption (Sato, 1988). Nakajima and Sato(1979) found that a one-day food deprivation acceleratesliver metabolism of xylene in rats. Nakajima et al. (1982)found that the in vitro rate of metabolism of toluene in-creased linearly with decreasing carbohydrate intake;such an observation may be relevant to solvent-exposedindividuals who have chosen a high- or low-carbohy-drate diet. Sato et al. (1981) found that rat liver metab-olism of volatile hydrocarbons was enhanced 16–18 hafter ethanol ingestion. Liver metabolism in our studywas constant and dependent on body weight. Thus, in-
tra-individual variability of extrahepatic metabolism,seen in this study (with one subject with a C.V. of65%), could be at least partially explained by subjectmetabolism enhancement due to diet or alcohol. Pierceet al. (1996a) found a range of intra-individual variabil-ity in fitted extrahepatic metabolism of toluene of 1.6–62% C.V. for five multiply-exposed subjects. However,these intra-individual differences in extrahepatic metab-olism are less than values of 220, 168, and 134% C.V.seen across all exposures for ortho-, para-, and meta-xy-lene, respectively (Tables 5–7).
The threshold limit values (TLVs) are developed asguidelines to assist in the control of health hazards (AC-GIH, 1998). The TLV-TWA for xylenes was set at100 ppm to protect nearly all workers exposed from ad-verse health effects. However, airborne concentrationsare not the best indicator of body burden, since physio-logical effects of organic solvents are more directly relat-ed to the amount of toxicant within the body (Langman,1994). This study, as well as many others, showed thateven among equivalently exposed subjects, actual bodyburden may vary up to 10-fold (Jang and Droz, 1997;Pierce et al., 1996a).
PBTK modeling helps to understand and predict thebehavior of toxicants in the body. By using measured,Bayesian-fit and adjustable model parameters, thismodeling work allowed for inter- and intra-humanvariability in influential model parameters, which ismore realistic than using mean values taken from theliterature. We found that there was large inter-individ-ual variability in these parameters, which would leadto large inter-subject differences in internal dose andduration of dose. This work further illustrated thatthe toxicokinetics and actual body burden of airbornesolvents varies among equivalently exposed individu-als. While our study was done with healthy Caucasianmen exposed at rest, additional inter-individual vari-ability is expected with women, different ethnic groups,working conditions, and those with pathological vari-ables (Langman, 1994; and Jang and Droz, 1997).For these reasons, airborne concentrations are impre-cise indicators of the actual toxicological dose andduration of internal exposure. An improvement is theuse of Biological Exposure Indexes (BEIs), which rep-resent the levels of toxicants in worker specimens suchas breath, urine, or blood. Such indexes should includeinter-individual kinetic variability, as assessed in ourstudy of xylenes, to best protect working and environ-mentally exposed populations.
Acknowledgments
This work was supported by the Superfund Basic Re-search program, NIEHS ES 04696, and by NIH GrantP41EB001975.
214 J.C. Adams et al. / Regulatory Toxicology and Pharmacology 43 (2005) 203–214
References
Agency for Toxic Substances and Disease Registry, 1998. Evaluationof health effects of chemicals. V. xylenes: health effects, toxicoki-netics, human exposure, and environmental fate. Toxicol. Ind.Health 14, 687–695.
American Conference of Governmental Industrial Hygienists, 1998.Threshold Limit Values for Chemical Substances and PhysicalAgents. Cincinnati, Ohio.
Csanady, G.A., Mendrala, A.L., Nolan, R.J., Filser, J.G., 1994. Aphysiologic pharmacokinetic model for styrene and styrene-7,8-oxide in mouse, rat and man. Arch. Toxicol. 68, 143–157.
Diem, K. (Ed.), 1962. Documenta Geigy: Scientific Tables, sixth ed.Geigy Pharmaceuticals, Ardsley, New York.
Dills, R.L., Kent, S.D., Checkoway, H., Kalman, D.A., 1991.Quantification of volatile solvents in the blood by static headspaceanalysis. Talanta 38, 365–374.
Droz, P.O., Wu, M.M., Cumberland, W.G., Berode, M., 1989.Variability in biological monitoring of solvent exposure. I. Devel-opment of a population physiological model. Br. J. Ind. Med. 46,447–460.
Durnin, J.V., Womersley, J., 1974. Body fat assessed from total bodydensity and its estimation from skin fold thickness: Measurementson 481 men and women aged 16 to 72 years. Br. J. Nutr. 32, 77–97.
Gargas, M.L., Burgess, R.J., Voisard, D.E., Cason, G.H., Anderson,M.E., 1989. Partition coefficients of low-molecular-weight volatilechemicals in various liquids and tissues. Toxicol. Appl. Pharamcol.98, 87–99.
Hetrick, D.M., Jarabek, A.M., Travis, C.C., 1991. Sensitivity analysisfor physiologically based pharmacokinetic models. J. Pharm.Biopharm. 19, 1–19.
Holm, W.S., Nordqvist, M.B., 1985. Coexposure to toluene and p-xylene in man: uptake and elimination. Br. J. Ind. Med. 42, 111–116.
Jang, J., Droz, P.O., 1997. Ethnic differences in biological monitoringof several organic solvents II. A simulation study with a physio-logically based pharmacokinetic model. Int. Arch. Occup. Environ.Health 70, 41–50.
Johanson, G., 1991. Modeling of respiratory exchange of polarsolvents. Ann. Occup. Hyg. 35, 323–339.
Koizumi, A., 1989. Potential of physiologically based pharmacokineticmodeling for acetone. Occup. Environ. Med. 52, 344–352.
Kumagai, S., Matsunaga, I., 1995. Physiologically based pharmaco-kinetic model for acetone. Occup. Environ. Med. 52, 344–352.
Langman, J.M., 1994. Xylene: its toxicity, measurement of exposurelevels, absorption, metabolism and clearance. Pathology 26, 301–309.
Low, L.K., Meeks, J.R., Mackerer, C.R., 1989. Health effects ofalkylbenzenes II. Xylenes. Toxicol. Ind. Health 5 (1), 85–105.
Miller, M.J., Edwards, J.W., 1999. Possible preferential metabolism ofxylene isomers following occupational exposure to mixed xylenes.Int. Arch. Occup. Environ. Health 72, 89–97.
Mordenti, J., 1986. Man versus beast: pharmacokinetic scaling inmammals. J. Pharm. Sci. 75, 1028–1040.
Morgan, M.S., Dills, R.L., Kalman, D.A., 1993. Evaluation of stableisotope-labeled probes in the study of solvent pharmacokinetics inhuman subjects. Int. Arch. Occup. Environ. Health 65, S139–S142.
Nakajima, T., Sato, A., 1979. Enhanced activity of liver drug-metabolizing enzymes for aromatic and chlorinated hydrocarbonsfollowing food deprivation. Tox. Appl. Pharmacol. 50, 549–556.
Nakajima, T., Koyama, Y., Sato, A., 1982. Dietary modification ofmetabolism and toxicity of chemical substances—with specialreference to carbohydrate. Biochem. Pharmacol. 31, 1005–1011.
Nies, A.S., Shand, D.G., Wilkinson, G.R., 1976. Altered hepatic bloodflow and drug disposition. Clin. Pharm. 1, 135–155.
Perbellini, L., Mozzo, P., Brugnone, F., Zessee, A., 1986. Physiolog-icomathematical model for studying human exposure to organicsolvents: kinetics of blood/tissue n-hexane concentration and of 2,5-hexadione in urine. Br. J. Ind. Med. 43, 760–768.
Pierce, C.H., Dills, R.L., Morgan, M.S., Nothstein, G.L., Shen, D.D.,Kalman, D., 1996a. Interindividual differences in 2H8-toluenetoxicokinetics assessed by a semi-empirical physiologically basedmodel. Toxicol. Appl. Pharamcol. 139, 49–61.
Pierce, C.H., Dills, R.L., Silvey, G.W., Kalman, D.A., 1996b. Partitioncoefficients between human blood or adipose tissue and air foraromatic solvents. Scand. J. Work Environ. Health 22, 112–118.
Sato, A., Nakajima, T., Koyama, Y., 1981. Dose-related effects of asingle dose of ethanol on the metabolism in rat liver of somearomatic and chlorinated hydrocarbons. Toxicol. Appl. Pharma-col. 60, 8–15.
Sato, A., 1988. Toxicokinetics of benzene, toluene, and xylenes. IARCSci. Publ. 85, 47–64.
Sato, A., Endoh, K., Kaneko, T., 1989. Individual differences in thekinetic behavior of trichloroethylene. A simulation study using aphysiological pharmacokinetic model. Sangyo Igaku. 31, 348–354.
Sedivec, V., Flek, J., 1976. The absorption, metabolism, andexcretion of xylenes in man. Int. Arch. Occup. Environ. Health37, 205–217.
Simon, T.W., 1997. Combining physiologically based pharmacokineticmodeling with Monte Carlo simulation to derive an acuteinhalation guidance value for trichloroethylene. Reg. Toxicol.Pharm. 26, 257–270.
Smith, B.R., Plummer, J.L., Wolf, R., Philpot, R.M., Bend, J.R., 1982.o-Xylene metabolism by rabbit lung and liver and its relationshipto the selective destruction of pulmonary cytochrome P-450. J.Pharm. Exp. Ther. 223, 736–742.
Tardif, R., Lapare, S., Charest-Tardif, G., Brodeur, J., Krishnan, K.,1995. Physiologically-based pharmacokinetic modeling of a mix-ture of toluene and xylene in humans. Risk Anal. 15, 335–342.
Tassaneeyakul, W., Birkett, D.J., Edwards, J.W., Veronese, M.E.,Tassaneeyakul, W., Tukey, R.H., Miners, J.O., 1996. Humancytochrome P450 isoform specificity in the regioselective metabo-lism of toluene and o-, m-, and p-xylene. J. Pharm. Exp. Ther. 276,101–108.
Thomas, R., Bigelow, P., Keefe, T., Yang, R., 1996. Variability inbiological exposure indices using physiologically based pharmaco-kinetic modeling and Monte Carlo simulation. Am. Ind. Hyg.Assoc. J. 57, 23–32.
Vicini, P., Pierce, C.H., Dills, R.L., Morgan, M.S., Kalman, D.A.,1999. Individual prior information in a physiological model of 2H8-toluene kinetics: an empirical Bayes estimation strategy. Risk Anal.19, 1127–1134.
Wallen, M., Holm, S., Nordqvist, M.B., 1985. Coexposure to tolueneand p-xylene in man: uptake and elimination. Br. J. Ind. Med. 42,111–116.
