ToxSci Advance Access originally published online on September 25, 2007
Toxicological Sciences 2008 101(1):32-50; doi:10.1093/toxsci/kfm251
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Physiologically Based Pharmacokinetic Modeling of 1,4-Dioxane in Rats, Mice, and Humans
* The Sapphire Group, Dayton, Ohio 45431
Battelle, Richland, Washington, 99352
ARCADIS, Novi, Michigan 48377
1 To whom correspondence should be addressed at 2661 Commons Blvd., Suite 240, Dayton, OH 45431. Fax: (937) 458-0050. E-mail: LMS29{at}cwru.edu; LMS{at}thesapphiregroup.com.
Received April 16, 2007; accepted September 18, 2007
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAFUNDINGREFERENCES |
|---|
1,4-Dioxane (CAS No. 123-91-1) is used primarily as a solvent or as a solvent stabilizer. It can cause lung, liver, and kidney damage at sufficiently high exposure levels. Two physiologically based pharmacokinetic (PBPK) models of 1,4-dioxane and its major metabolite, hydroxyethoxyacetic acid (HEAA), were published in 1990. These models have uncertainties and deficiencies that could be addressed and the model strengthened for use in a contemporary cancer risk assessment for 1,4-dioxane. Studies were performed to fill data gaps and reduce uncertainties pertaining to the pharmacokinetics of 1,4-dioxane and HEAA in rats, mice, and humans. Three types of studies were performed: partition coefficient measurements, blood time course in mice, and in vitro pharmacokinetics using rat, mouse, and human hepatocytes. Updated PBPK models were developed based on these new data and previously available data. The optimized rate of metabolism for the mouse was significantly higher than the value previously estimated. The optimized rat kinetic parameters were similar to those in the 1990 models. Only two human studies were identified. Model predictions were consistent with one study, but did not fit the second as well. In addition, a rat nasal exposure was completed. The results confirmed water directly contacts rat nasal tissues during drinking water under bioassay conditions. Consistent with previous PBPK models, nasal tissues were not specifically included in the model. Use of these models will reduce the uncertainty in future 1,4-dioxane risk assessments.
Key Words: biological modeling; risk assessment; hepatocytes; in vitro and alternatives; physiologically based pharmacokinetics; biotransformation and toxicokinetics; toxicokinetics; volatile organic compounds; agents.
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAFUNDINGREFERENCES |
|---|
1,4-Dioxane (CAS No. 123-91-1) is used primarily as a solvent or as a stabilizer for solvents. It can be irritating to eyes, skin, and the respiratory tract, and can cause lung, liver, and kidney damage depending upon the extent and duration of exposure (De Rosa et al., 1996
). In chronic drinking water studies, exposure to 1,4-dioxane caused increased incidences in liver tumors in rats and mice, and nasal tumors in rats (De Rosa et al., 1996; Leung and Paustenbach, 1990; Reitz et al., 1990). An extensive database exists describing the toxicity, modes of action, and carcinogenic potential of 1,4-dioxane (reviewed in Stickney et al., 2003; Agency for Toxic Substances and Disease Registry, 2006). 1,4-Dioxane is believed to be primarily metabolized by cytochrome P450s to β-hydroxyethoxyacetic acid (HEAA) or its interconversion product p-dioxane-2-one, which is formed under acidic conditions. Results have been influenced by the pH of the analytical methods (Braun and Young, 1977; Woo et al., 1977).
Two physiologically based pharmacokinetic (PBPK) models of inhaled, injected (iv), or ingested 1,4-dioxane and HEAA were published in 1990 (Leung and Paustenbach, 1990; Reitz et al., 1990). Tissue compartments in these models included liver, fat, slowly perfused tissues, and richly perfused tissue (both) and blood (Reitz only) for 1,4-dioxane and a single compartment for HEAA (Reitz only). Both models were used to derive cancer potency estimates for 1,4-dioxane based on occurrence of liver tumors as well as toxicity values based on a threshold approach. These liver cancer potency estimates were many orders of magnitude less potent than those derived by the U.S. Environment Protection Agency (EPA) during their last evaluation of 1,4-dioxane carcinogenicity in 1990 using standard default approaches (U.S. EPA, 1990). Neither PBPK modeling study addressed rat nasal tumors as nasal tissues were assumed to have been exposed through direct contact with inspired drinking water as well as through internal exposure, based on unpublished data.
Although each PBPK model had attributes that allowed risk assessments to be performed in 1990, each model has uncertainties and deficiencies that could be addressed and the model strengthened for use in a contemporary cancer risk assessment for 1,4-dioxane. Deficiencies in the Reitz model include the manipulation of breathing parameters and measured partition coefficients to achieve model fits and the estimation of certain mouse parameters (partition coefficients, metabolic rate constants) by interpolation of rat and human values. In addition, there was no validation data available for mice and no provision was made to account for enzyme induction (Nannelli et al., 2005; Young et al., 1978) in the Reitz model. The Leung model is limited in application because no mouse model is described. The Leung model also used an estimated human muscle:blood PC by optimizing model fits to in vivo human data, instead of using the rat muscle:air partition coefficient directly. Although the Leung approach to estimating the human muscle:blood partition coefficient is not uncommon, use of a measured value (e.g., the rat muscle:air partition coefficient) might be preferred. Both models also suffer because there are no independent measurements of the rate constants or relative rates of metabolism for any of the three species of interest.
Studies were performed to fill these data gaps and reduce uncertainties pertaining to the pharmacokinetics of 1,4-dioxane and HEAA in rats, mice, and humans. In addition, a study was completed to confirm rat nasal tissues are exposed by direct contact under drinking water study conditions. Three types of studies were performed to inform the development of updated PBPK models: partition coefficient measurements, blood time course in mice, and in vitro pharmacokinetics.
- Partition coefficients: The partition coefficient measurements included new measurements for mouse blood and tissues (liver, kidney, fat, and muscle) and confirmatory measurements for human blood and rat blood and muscle.
- Mouse blood concentrations over time: The blood time course measurements in mice were conducted for gavage administration of nominal single doses (20, 200, or 2000 mg/kg) of 1,4-dioxane administered in water.
- Metabolic rate constants: Vial incubations of 1,4-dioxane with rat liver microsomes failed to produce detectable declines in headspace concentration of 1,4-dioxane or increases in HEAA in buffer. However, incubations of 1,4-dioxane with rat, mouse, and human hepatocytes did produce measurable amounts of HEAA, and estimates of rate constants for metabolism of 1,4-dioxane by rat, mouse, and human liver were thus derived.
New data and previous kinetic studies in rats, workers, and human volunteers reported by Young et al. (1976, 1977, 1978) were used to develop updated PBPK models for the rat, mouse, and human. The updated models are described in this paper.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAFUNDINGREFERENCES |
|---|
Animals.
Male Sprague–Dawley rats (300–350 g, approximately 9–11 weeks of age) and male B6C3F1 mice (25–35 g, at least 45 days of age) were obtained from Charles River Breeding Laboratory (Raleigh, NC). Animals were housed in solid-bottom cages with hardwood chips and were acclimated in a humidity- and temperature-controlled room with a 12-h light/dark cycle for at least 5 days prior to use. Certified Purina rodent chow (Ralston Purina Co., St. Louis, MO) and water were provided ad libitum throughout the acclimation period. Water to rats in the nasal exposure study was provided using water bottles and sipper tubes similar to those used at the time of the chronic drinking water studies. All animal protocols were approved by the Institutional Animal Care and Use Committee at Pacific Northwest National Laboratory and studies were performed according to the "Guide for the Care and Use of Laboratory Animals" (National Research Council, 1996, Washington, DC).
Isolated hepatocytes.
Cryopreserved isolated hepatocytes from male Sprague–Dawley rats, B6C3F1 mice, and human donors were purchased from In Vitro Technologies (Baltimore, MD). Male human donor demographics, as provided by the supplier, are given in Table 1. Cell viability was assessed based on plasma membrane integrity by trypan blue dye exclusion; rat cell preparations had > 89% viability and mouse had 81% viability. Viability of human cell preparations ranged from 75% to 88% (Table 1). Pilot studies were conducted to determine time linearity and the optimal number of hepatocytes and substrate concentrations. For the final incubations, targets of 1–2.5 x 106 cells/ml in 0.25 ml Krebs–Henseleit bicarbonate buffer (pH 7.4) with 12.5mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid were incubated with five to seven concentrations of 1,4-dioxane at 37°C for 60 min in a shaking water bath. The reaction was stopped by the addition of 0.25 ml of a 1.0N HCl/internal standard (glycolic acid) solution. The production of HEAA in these incubations was measured (see "Analytical Methods" below) and evaluated based on Michaelis–Menten kinetics using the GraphPad Prism (San Diego, CA) analyze function. Some additional human samples were incubated with 0–25 mg/ml 1,4-dioxane for 30 min to determine cell count/viability after incubation.
|
Chemicals.
1,4-Dioxane was obtained from Aldrich Chemical Co. (St Louis, MO). HEAA was a gift from Dow Chemical (Midland, MI).
Analytical methods development.
A method was developed for the simultaneous analysis of 1,4-dioxane and its major metabolite, HEAA. The additional metabolite, p-dioxane-2-one, is converted to HEAA under acidic conditions. Heated gas chromatography (GC) headspace methods have been reported for 1,4-dioxane determination in blood or urine (Royal Society of Chemistry, 1988) or for cosmetics (Wala-Jerzykiewicz and Szymanowski, 1998). These methods were utilized for partition coefficient studies. The analysis of in vivo samples and samples from in vitro hepatocyte incubations was conducted using a GC with a mass selective detector based on a modification of the Young et al. (1976) method. Each analytical method is described below.
Gas chromatographic analysis.
Headspace 1,4-dioxane concentrations were determined using a GC method on an Agilent Model 6890 system (Avondale, PA) equipped with a flame ionization detector. The column was a J&W Scientific (Folsom, CA) DB 624 (30 m x 0.53 mm id x 3.0-µm film thickness). The detector was operated at 275°C, the inlet at 250°C, and the final oven temperature was 200°C. Under these conditions, 1,4-dioxane had a retention time of approximately 2.1 min and a limit of reliable quantitation of 26 ppm and was linear to 4000 ppm.
Gas chromatography/mass spectrometry analysis.
Mouse blood 1,4-dioxane and HEAA concentrations were determined using an Agilent Model 6890 GC (Avondale, PA) with a mass selective detector (GC/MS [mass spectrometry], model 5973N). The column was a Restek (Bellefonte, PA) Rtx-5MS (30 m x 0.25 mm id x 0.25 µm df). The GC oven was programmed to ramp from an initial 40°C to a final temperature of 280°C at a rate of 18°C/min. The temperatures of the injection port and MS interface were 200°C and 280°C, respectively. For 1,4-dioxane, 0.1 g blood samples were treated with 0.9 ml of a 1.0N HCl/internal standard (1,4-dioxane-d-8) solution and extracted using 0.5 ml ethyl acetate. The supernatant was analyzed with quantitation achieved using m/z 88 and 96, for 1,4-dioxane and 1,4-dioxane-d-8 (internal standard), respectively, both with retention times of approximately 2.4 min. For HEAA, 0.1 g blood samples were treated with 0.9 ml of a 1.0N HCl/internal standard (glycolic acid) solution and were twice extracted using 0.5 ml of tri-n-octylphosphine oxide. To measure HEAA in hepatocyte incubation medium, the samples were extracted twice using 0.5 ml of methyl t-butyl ether containing tri-n-octylphosphine oxide. The final supernatants were evaporated to dryness under an ultra high purity nitrogen stream and reconstituted with 0.45 ml of toluene and 50 µl of n-methyl-n-tert-butyldimethylsilyl-trifluoroacetimide and incubated for 1 h at 60°C. Quantitation of the supernatant was achieved using m/z 247 for glycolic acid as the internal standard with a retention time of approximately 8 min and m/z 291 for HEAA, with a retention time of approximately 9.9 min. All ions were acquired in scan mode. The limits of reliable quantitation were 0.41 and 0.39 µg/g for 1,4-dioxane and HEAA, respectively.
1,4-Dioxane partition coefficients.
Male Sprague–Dawley rats and B6C3F1 mice were euthanized by CO2 asphyxiation, and heparinized blood was collected by cardiac puncture. Tissues were removed, weighed, and stored at –80°C until use. Heparinized human blood samples from single donors (n = 6) were purchased from Golden West Biologicals, Inc. (Temecula, CA). Partition coefficients for blood were determined immediately following collection of the sample using the vial equilibration method described by Gargas et al. (1989). Partition coefficients were determined using 0.5-ml samples of heparinized blood or homogenized tissue (without the addition of saline) weighed into 25-ml glass scintillation vials and sealed air-tight with caps modified by drilling a 4-mm-diameter hole in the center and replacing the liner with a Teflon-coated rubber septum (Supelco, Inc., Bellefonte, PA). 1,4-Dioxane was introduced into the sealed vial as a vapor from a Tedlar gas-sampling bag (SKC-West, Fullerton, CA) containing an air concentration of 21,600 ppm. Vials were incubated at 37°C with shaking in a vortex evaporator for 3 h. Preliminary studies indicated no difference in partition coefficient values with 1 h compared with 3-h incubations. Headspace 1,4-dioxane concentrations were determined as described under the "Analytical Methods" section.
In vivo mouse toxicokinetics.
Male B6C3F1 mice (25–35 g BW, at least 45 days of age) were assigned to one of three groups (27 mice per group) and administered a single low (20 mg/kg), mid (200 mg/kg) or high (2000 mg/kg) nominal dose of 1,4-dioxane by oral gavage in a water vehicle. Fresh 1,4-dioxane dosing solutions were prepared on exposure days and concentrations were verified by triplicate analysis of dosing solution aliquots using GC/MS. Each animal received an approximate 0.25 ml volume of the appropriate dosing solution by oral gavage; actual administered volumes were measured as the difference between syringe weight before and after dosing. At time intervals of 0-, 0.5-, 1-, 2-, 3-, 6-, 9-, 12-, and 24-h postdosing, subgroups of three mice/dose group/time point were sacrificed by CO2 asphyxiation and blood collected by cardiac puncture for analysis of 1,4-dioxane and HEAA concentrations. Blood samples were stored at –80°C until analysis; extraction and analysis were conducted according to the GC/MS method described above. The dose levels and sacrifice intervals were selected based on simulations using the preliminary mouse PBPK model for 1,4-dioxane developed by Reitz et al. (1990).
Rat nasal tissue exposure: preparation and administration of drinking water.
For the two groups of five animals exposed to the fluorescent dye mixture, each individual 45-ml drinking water bottle contained 500 µg Cell Tracker Red CMTPX, 380 µl of dimethylsulfoxide (used to transfer the contents of Cell Tracker Red CMTPX from 10 x 50 µg vials), and 2 ml of FluoSpheres carboxylate-modified microspheres (0.2 µm). The drinking water for one group of five animals exposed to the Cell Tracker Red/FluoSpheres mixture also contained 0.5% 1,4-dioxane, which is within the range of doses used in previous chronic studies (Stickney et al., 2003). Control animals (five) received tap water alone. All drinking water solutions were prepared in the afternoon, weighed, covered in foil to prevent loss of fluorescence due to room lighting, and provided to the animals overnight. The next morning, the water bottles were weighed to estimate the amounts of water consumed and each animal sacrificed under CO2 asphyxiation followed by exsanguination. The heads were removed, skinned, and split along the midline for evaluation by fluorescence microscopy. One additional rat was dosed twice by gavage with 2 ml of drinking water containing fluorescent dye (the second dose was 30 min after the first dose; total of 4 ml administered) and sacrificed 5 h later to evaluate the potential for systemic delivery of fluorescent dye to the nasal tissues.
Fluorescent microscopy.
Fluorescence microscopy was performed with a Nikon Eclipse TE300 inverted microscope equipped for differential interference contrast (DIC) imaging with Plan Fluor/DIC objectives (4–60x), epifluorescence optics for various fluorophores, including 4,6-diamidino-2-phenylindole hydrochloride, fluorescein, rhodamine, and Texas Red, and a Nikon Super High Pressure Mercury Lamp. A charge-coupled device camera (Quantix, Photometrics, Ltd, AR) was used to obtain the microscopic images. In a subset of images, equal exposure times (200 ms) were used to capture dye-specific images of the stained tissue, relative to unstained tissue from controls. Images of tissue autofluorescence were obtained using the filter set for fluorescein (green channel) for comparison with the red-fluorescent dye-specific signature captured using the Texas Red filter set (red channel). Therefore, for each image of red fluorescent dye (red channel), a corresponding image was taken using the green channel that can be viewed separately or merged with the red channels. This process was used to determine whether information acquired in the red channel (dye-specific) was qualitatively different or similar to information acquired in the green channel (autofluorescence).
To further illustrate the ability to distinguish the dye-specific signature from autofluorescence, a separate subset of images was captured using the dye-specific filter set and overexposing the images to demonstrate that autofluorescence was qualitatively comparable in both fluorescein and Texas Red filter sets for comparison with the dye-specific images captured using the Texas Red filter set at 200-ms exposure times.
Sources of preexisting data for modeling.
Some human and rat pharmacokinetic data were available in numerical form from Dr Richard H. Reitz (retired, Dow Chemical) and from Young et al. (1976, 1977, 1978). Additional human and rat pharmacokinetic data were available in graphical form from Young et al. (1977, 1978). Scanned images were converted into numerical data using Plot Digitizer (version 2.4.0), with minor adjustments made to match reported sampling times. Copies of worksheets reporting blood 1,4-dioxane and HEAA concentrations for the four individuals in Young et al. (1977) were graciously provided by Dr Bill Stott, Dow Chemical Company, Midland, MI.
Model description.
The model structure was similar to those used by Reitz et al. (1990) and Leung and Paustenbach (1990) and is depicted in Fig. 1. Model parameter values are summarized in Table 2. Tissue volumes and fractional blood flow rates were taken from Brown et al. (1997). Partition coefficients were generally those derived in this study. The measured mouse kidney:air partition coefficient was used for all three species for lumped richly perfused tissue group, and muscle:air partition coefficients used for slowly perfused tissues. The rat fat:air value was reported by Reitz et al. (1990). Human liver:air, fat:air, and slowly perfused tissue:air partition coefficients were estimated as the average of measured mouse and rat values.
|
|
Estimated/optimized model parameters.
The determination of certain model parameters by estimation/optimization is described in greater detail under "Results," but described briefly below.
The metabolic rate constants VmaxC (maximum rate of metabolism, normalized to scaled body weight, BW0.7) and Km (Michaelis constant, or apparent enzyme affinity) for rats were derived by fit to the iv data of Young et al. (1978). Young et al. (1978) had noted that administration of a dose of 1000 mg/kg, but not 10 mg/kg 1,4-dioxane appeared to induce metabolism of 1,4-dioxane. Nannelli et al. (2005) also reported the induction of cytochrome P450 2B1/2- and 2E1-dependent metabolic activities in rat liver due to oral exposure to 2 g/kg/day 1,4-dioxane by gavage for 2 days. The appropriateness of dose-specific VmaxC values was tested by optimizing the fit to high or low iv doses separately.
Based on the similarity of Km values derived in vitro for metabolism of 1,4-dioxane by rats and mice, (as reported under "Results"), the Km value derived by optimization for rats was also used for mice. Estimates of the oral absorption rate constant and VmaxC values for mice were made based on fit to blood 1,4-dioxane concentrations collected in this study. Because the analytical method measured background/artifactual levels of 1,4-dioxane (approximately 1.6 mg/l) and HEAA (approximately 0.8 mg/l) in blood of unexposed mice, only values that were greater than threefold higher than the background level were used in modeling. The oral absorption rate constant for mice was also applied to simulations of oral dosing in rats.
Human VmaxC estimates were made using the parallelogram approach, relying on the "best fit" in vivo values derived for rats and mice and the in vitro rates determined using rat, mouse, and human hepatocytes. Hepatocyte yields of 128, 110, or 137 x 106 hepatocytes per gram of mouse, rat, and human liver (Arias et al., 1982; Carlile et al., 1997; Seglen, 1976), respectively, and the default tissue volumes and BWs in Table 1 were used to scale in vitro data.
The first order elimination rate for metabolite in urine (Kme) of rats was estimated by best fit to amounts excreted when rats were dosed by single iv or gavage (Young et al., 1978). Kme and the volume of distribution of the metabolite (VDMC) of mice was estimated by best fit to blood concentrations of HEAA measured in mice dosed by gavage.
Model validation.
The model was further tested against additional data of Young et al. (1976, 1977, 1978) and low-dose mouse data derived in this study and evaluated by visual inspection of the plots, as described under "Results."
Sensitivity analysis.
Sensitivity analyses were conducted to evaluate the effect of small changes (1%) in the value of a single parameter on certain model outputs for specified exposure scenarios. Results were computed as normalized sensitivity coefficients (SCs, % change in output/% change in input) for parameters with |SC| > 0.1. Although many different exposure scenarios and model outputs could be analyzed, the analyses presented herein were limited to scenarios corresponding to experimentally evaluated regimes and dose metrics that could be compared with experimental data. Thus, the scope of the sensitivity analyses was focused on the potential for the given data set to reliably identify/validate parameter values and provide insights on potential reasons for lack of model fit.
Software, algorithms, and model code.
All simulations and parameter fitting were conducted using ACSL Sim 11.4 and ACSL Math, Version 2.5.4 (Aegis Technologies, Hunstville, AL) on a Dell Optiplex GX260 computer with a Pentium 4 processor. The Gear algorithm was used for integration of double precision variables. Parameter fitting was performed using the relative error model (variance is assumed to be proportional to the measured value across the range of measured values, or heteroscedasticity = 2) and the Nelder–Mead algorithm. The fitting criterion was maximization of the log-likelihood function. Starting values for parameter fitting in ACSL Math were determined from parameter estimates derived by visual best fit in ACSL Sim. Goodness of fit is described as the "percentage of variation explained," which is similar to the r2 value derived for linear regression. Model code is available from the corresponding author upon request.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAFUNDINGREFERENCES |
|---|
Partition Coefficients
Blood, saline, and tissue to air partition coefficient values for 1,4-dioxane are provided in Table 3, along with comparative data reported by Reitz et al. (1990)
, and Leung and Paustenbach (1990). For tissues where comparative data are available, including human blood, the values determined in the current study compare well with those reported by Leung and Paustenbach (1990). Both Leung and Paustenbach (1990) and Reitz et al. (1990) used the same partition coefficient data set for model development; although Reitz et al. (1990) optimized several values to best describe the experimentally observed pharmacokinetics using their model. In particular, Reitz et al. (1990) doubled the experimentally determined human blood to air partition coefficient value given by Leung and Paustenbach (1990) from 1825 (± 94) to 3650 to better describe the data using their model. In comparison, the human blood to air partition coefficient value determined here (1666 ± 287) correlates well with the experimental value reported by Leung and Paustenbach (1990) (1825 ± 94). In addition, Reitz et al. (1990) set the kidney (richly perfused tissues) partition coefficient equal to the liver value (1557). However, the mouse kidney to air partition coefficient value determined here (560 ± 175) suggests that the liver might not be a good surrogate for the kidney.
|
Mouse In Vivo Study
For both the high nominal dose (2000 mg/kg BW) and mid-nominal dose (200 mg/kg BW) groups, peak concentrations occurred within 1 h following oral dosing and declined thereafter. In contrast, blood 1,4-dioxane concentrations for the low nominal dose (20 mg/kg BW) group were at or near background levels throughout the collection time course. However, no efforts were made to confirm whether the 1,4-dioxane peak at, or near the background concentration was a true 1,4-dioxane peak or an interfering peak at the same mass and retention time.
As observed with 1,4-dioxane, HEAA blood concentrations peaked rapidly (0.5–2 h) and declined thereafter. With the exception of the high dose group, no HEAA was detected in blood samples collected beyond the 9-h time point (i.e., 12 and 24 h). At the high-dose group, HEAA was detected in blood samples collected through 24 h. Although HEAA was measured in blood samples from all three dose groups, the highest levels, as a percent of administered 1,4-dioxane, were observed with the lowest (20 mg/kg BW) dose group. A comparison of the blood concentration–time curves (AUCs) for 1,4-dioxane and HEAA suggests nonlinear 1,4-dioxane metabolism, as illustrated by a disproportional increase in the 1,4-dioxane AUC with increased administered dose, along with a decrease in the HEAA blood AUC/1,4-dioxane dose ratio with increased 1,4-dioxane dose, as shown in Table 4.
|
In Vitro Metabolism
Originally, an attempt was made to measure 1,4-dioxane metabolism in microsomes prepared from the livers of rats. Neither loss of 1,4-dioxane nor production of HEAA was observed above the background of 0.4 µg/ml in incubations with microsomes up to an hour in duration (data not shown). Typically, microsomes are incubated for less than 30 min with relatively high substrate concentrations. In our microsome studies, the high 1,4-dioxane substrate concentrations may have resulted in solubilization of the artificial (endoplasmic reticulum) microsomal membranes, alternatively, the limited incubation time was not sufficient to produce HEAA concentrations at levels above our analytical detection limit. The ability of isolated hepatocytes, with their intact cellular physiology and membranes, to metabolize 1,4-dioxane was investigated. Pilot studies in cryopreserved rat liver hepatocytes were conducted to determine optimal incubation conditions, as described above under "Methods." The Vmax for the production of HEAA from 1,4-dioxane was greater in mouse than in rat hepatocytes, but the Km values were similar (Table 5; Fig. 2a). In human hepatocytes, the Vmax for the production of HEAA from 1,4-dioxane ranged from 2.4 to 8.7 µg/h/106 cells, and the Km ranged from 3.8 to 17.6 mg/ml (Table 6; Fig. 2b). The rate constants and metabolite profile from three of the human donor hepatocytes samples are very similar, and the results from donor EQB, indicate a possible outlier with a higher Km (Table 6). These in vitro Michaelis–Menten constants are remarkably similar to the values obtained in rat and mouse hepatocytes. The Michaelis constant in rat and mouse hepatocytes was 2.5 and 2.6 mg/ml, respectively. The Vmax in these species were nominally lower than in human hepatocytes (1.92 and 3.74 µg/h x 106 cells for rats and mice, respectively).
|
|
|
Selected samples from donors CEC and EQB were incubated with 1,4-dioxane for 60 min and trypan blue dye exclusion used as a measure of cell membrane integrity. Cells incubated with 25 mg/ml showed appreciably more leakage of the die through the membrane. The pilot time linearity studies were conducted using 25 mg/ml 1,4-dioxane and indicate linear HEAA production in these samples for at least 90 min, suggesting that the apparent affects on cell membrane do not adversely affect 1,4-dioxane metabolism at least through 90 min. However, HEAA production in cells incubated with 50 mg/ml 1,4-dioxane tended to be lower than in samples incubated with 25 mg/ml 1,4-dioxane. The 50 mg/ml 1,4-dioxane samples were therefore not included in the determination of 1,4-dioxane Michaelis–Menten rates.
The supplier (In Vitro Technologies) provided marker substrate metabolism data, and these values were compared with the Vmax and pseudo-first order rate (Vmax/Km) for 1,4-dioxane. Correlations greater than 50% were only obtained between the reported 7-hydroxylation of Coumarin (Fig. 3) and O-demethylation of dextromethorphan (data not shown). The correlation between the pseudo-first order rate for 1,4-dioxane metabolism and the metabolism of coumarin was good (r2 = 0.83), the correlation for dextromethorphan was not nearly as good (r2 = 0.63). Coumarin is metabolized by a number of CYP450 enzymes, but the principal P450 is 2A6. Dextromethorphan is primarily metabolized by CYP450s from the 2D family. Adding (or averaging) the metabolism of coumarin and dextromethorphan results in a slight increase in the correlation with the pseudo-first order rate constant to r2 = 0.86. This suggests a possible relationship between the combined activities P450s 2A and 2D and the metabolism of 1,4-dioxane.
|
Rat Nasal Exposure Study
The presence of the fluorescent dye mixture had no measurable impact on water consumption. However, 1,4-dioxane present at 0.5% reduced the amount of water consumed by an average of 62% (range: 42–77%) of controls following a single, overnight exposure. The BW and water consumption for each rat is summarized in Table 7.
|
Fluorescent dye was readily detected in the oral cavity and nasal airways of each animal exposed to the Cell Tracker Red/FluoSpheres mixture in their drinking water (Fig. 4; Supplementary Data, Figs. 1–8). The presence of 1,4-dioxane had no effect on the detection of the dye (Fig. 4, Supplementary Data, Figs. 1–3). Little to no tissue background autofluorescence was detected in areas where the fluorescent dye was detected using a 200-ms exposure time. Fluorescent dye was readily observed in numerous areas of the anterior third of the nose of each rat along the nasal vestibule, maxillary turbinates, and dorsal nasoturbinates where previous bioassays have identified the presence of nasal irritation and tumors (Goldsworthy et al., 1991
; National Cancer Institute [NCI], 1978; Yamazaki et al., 1994). Fluorescent dye was occasionally detected in the ethmoid turbinate region and nasopharynx. No attempts were made to dissect the tissues further to evaluate the extent of water deposition in nasal airways beyond that observed along the midline (septum).
|
Little or no fluorescence at the wavelength associated with the dye mixture was detected in control animals using a 200-ms exposure time (Supplementary Data, Figs. 9–13) or in the animal that received the dye mixture by oral gavage (Supplementary Data, Fig. 14) at the same image settings used for detecting the dye in exposed animals. In addition, several images of control tissue were overexposed using the Texas Red filter set to demonstrate that autofluorescence present in the red channel was qualitatively similar to autofluorescence observed in the green channel (nonspecific autofluorescence). These results directly contrast the qualitatively different tissue staining patterns observed in the red and green channels associated with animals exposed to the dye mixture.
These results indicate rat nasal tissues are exposed by direct contact with drinking water under study conditions and, consistent with previous PBPK models developed for 1,4-dioxane, were not specifically included in the model.
PBPK Modeling
The studies used in PBPK model parameter derivation and validation are summarized in Table 8.
|
Determination of VmaxC and Km for the rat.
Preliminary values of VmaxC and Km in the rat were derived by optimizing the fit to the 1000 mg/kg iv data (Young et al., 1978
) ("induced" rat VmaxC) and iv doses of 3, 10, 30, and 100 mg/kg ("uninduced" rat VmaxC). The 300 mg/kg iv data were initially omitted as a likely border-line case which could distort the optimization of fit to "high" and "low" data. Preliminary best-fit values of VmaxC = 12.8 ± 0.036 mg/h·kg0.7 and Km = 22.0 ± 1 mg/l were derived for the induced rat, and values of VmaxC = 7.4 ± 0.05 mg/h·kg0.7 and Km = 20.5 ± 1.4 mg/l for the uninduced rat. Because of the similarity of the Kms for the uninduced and induced rats, an average value of Km = 21 mg/l was selected as being applicable to all doses. With the Km value set at 21 mg/l, the best-fit value of VmaxC for induced rats, 12.7 ± 0.08 mg/h·kg0.7 (80.4% of variation explained) was determined from fit of the 1000 mg/kg data. Likewise, a best fit value of VmaxC = 7.5 ± 0.2 mg/h·kg0.7 was derived for uninduced rats (66.6% of variation explained). The best-fit VmaxC for the 300 mg/kg iv data was found to be 10.8 ± 0.2 mg/h·kg0.7, indicating that these data would be more appropriately described by the "induced" VmaxC rather than the uninduced VmaxC. The model fit to the iv data is shown in Fig. 5.
|
Determination of VmaxC, Km, and KA for the mouse.
The in vivo Km value for the mouse was estimated as being equal to the best-fit rat value of 21 mg/l. The basis for this selection was that the in vitro Kms for production of HEAA from 1,4-dioxane from incubated rat and mouse hepatocytes (2.51 ± 0.88 and 2.63 ± 0.68 mg/ml) are statistically indistinguishable. Thus, it is expected that the in vivo Kms will also be similar. The in vivo mouse data have insufficient samples where the blood concentration of 1,4-dioxane was at or below the likely Km, so it was not possible to identify the in vivo Km on the basis of fit to the in vivo data.
Mouse VmaxC and KA values were derived by optimizing fit to the blood 1,4-dioxane concentrations in mice administered nominal doses of 200 and 2000 mg/kg 1,4-dioxane by gavage in a water vehicle. 1,4-Dioxane measurements in blood of the animals in the 20 mg/kg group were indistinguishable from the background for the analytical method, and thus could not be used for pharmacokinetic analysis. Because doses > 300 mg/kg have been found to induce 1,4-dioxane metabolism in rats, the possibility of dose dependency of VmaxC was also assumed for mice. Preliminary VmaxC and KA values for potentially induced mice (2000 mg/kg dose) were 46.6 ± 1.1 mg/h·kg0.7 and 0.73 ± 0.09/h, whereas the preliminary values for presumptive uninduced mice (200 mg/kg) were 39.1 ± 0.3 mg/h·kg0.7 and 0.94 ± 0.009/h. Because the absorption rate would be expected to be similar across doses, a single value of 0.8/h was assumed for both doses. With KA fixed, dose-dependent VmaxC values were then optimized as 46 ± 1 and 39 ± 1 mg/h·kg0.7 for 2000 and 200 mg/kg mice, respectively (91.8% and 91.5% of variation explained, respectively). The model fit to the mouse oral data is shown in Fig. 6.
|
Scaling of in vitro metabolism data/estimation of Human VmaxC and Km.
The in vitro Vmax values for rats and mice were scaled to estimated in vivo rates, which were compared with the optimized values. The scaled and optimized rat VmaxC values were very similar. The discrepancy between the scaled and optimized mouse values was larger, which was attributed to possible induction in mice at the lowest dose tested (200 mg/kg). The ratio of optimized/scaled values for the rat was used to adjust the scaled human VmaxC values to projected in vivo values (Table 9).
|
The Km value derived for the rat in vitro (2510 mg/l) differs substantially from the Km estimated from the in vivo data (21 mg/l). This difference may be related to unexpected difficulty with measuring 1,4-dioxane metabolism in vitro (i.e., the inability to detect 1,4-dioxane disappearance or HEAA appearance using microsomes). Human in vivo Km values were estimated by multiplying the in vitro values by the in vivo/in vitro ratio for the rat. Km values for representative, minimum, and maximum cases were 32, 29, and 147 mg/l.
Estimation of K me for the Rat
The first order rate constant for the urinary elimination of the 1,4-dioxane metabolite HEAA by rats was estimated based on fit to the time course for total amount of HEAA eliminated in urine by rats dosed with 1,4-dioxane by iv (10 or 1000 mg/kg) or gavage administration (10, 100, or 1000 mg/kg) (Young et al., 1978). Dose-specific VmaxC values (derived as described above) were used. The oral absorption rate constant for the rat was assumed to be equal to the best-fit value derived for the mouse (KA = 0.8). The optimized value of Kme for the rat was 0.48 ± 0.049/h (93.0% of variation explained). The model fit to the rat urinary metabolite data is shown in Figs. 7a, 7c, 7e.
|
Kme values were also estimated for each of data set individually. The optimal Kme values ± the SDs generally encompassed the optimal values for all five data sets considered together. The single exception was the low dose (10 mg/kg) iv data, where an optimal fit was found with Kme = 0.16 ± 0.02/h. Because the optimal Kme for an equal oral dose was more in line with the group Kme value (0.62 ± 0.11/h), a dose-dependence in Kme did not seem to be indicated. The Kme value derived for the rat using all five data sets was used in the modeling.
Estimation of K me and VDMC for the Mouse
The volume of distribution of the 1,4-dioxane metabolite HEAA (VDMC) and the rate constant for urinary elimination of HEAA were optimized based on the fit to the time course of HEAA in blood of mice dosed with 200 or 2000 mg/kg 1,4-dioxane by gavage. The resulting values were VDMC = 0.83 ± 0.12 l/kg and Kme = 0.35 ± 0.02/h (56.7% of variation explained). If the low-dose HEAA data were included, a similar Kme value resulted (0.40/h), but VDMC was significantly reduced (0.56 l/kg), and the fit deteriorated substantially (41.7% of variation explained). The VDMC and Kme values from the mid- and high doses (with the low dose omitted) were used in modeling (see Fig. 6). The poor fit may result from an overly simplistic model structure for HEAA.
Model Validation/Fit to Other Rodent Data
Model outputs were compared with other data not used in fitting model parameters by visual inspection. The model predictions gave an excellent match to the 1,4-dioxane exhalation data after a 1000 mg/kg iv dose. 1,4-Dioxane exhalation was overpredicted by a factor of 
3 for 10 mg/kg iv dose. Similarly, the simulations of exhaled 1,4-dioxane after oral dosing were excellent at 1000 mg/kg, very good at 100 mg/kg (within 50%), but poor at 10 mg/kg (model overpredicted by a factor of five). The prediction of the 1,4-dioxane exhalation data is shown in Figs. 7b, 7d, 7f.
The simulation of blood 1,4-dioxane concentrations in rats exposed to 50 ppm 1,4-dioxane was excellent (Fig. 8a), but total excretion in urine was under predicted by a factor of 3 (data not shown). In order to match the model prediction to the data for HEAA excretion, the inhalation rate had to be increased by factor of almost 4, and blood concentrations were no longer accurately predicted. Although restraint in a head-only chamber (Young et al., 1978) might be expected to cause some stress, a fourfold increase in ventilation rate seems unlikely.
|
Predictions of blood concentrations of 1,4-dioxane and HEAA were made for mice exposed to a low dose (20 mg/kg) of 1,4-dioxane by gavage. Predictions were consistent with the measured levels of 1,4-dioxane in blood not being distinguishable from the background of the method (
1.6 mg/l). The model dramatically underpredicted the blood concentrations of HEAA 0.5 and 1 h after dosing, whereas overpredicting at 2 h (Fig. 8b). The model predicted that HEAA levels would be above the background of the method (1.1 mg/l) at the 3- and 6-h sample points, but they were not.
Fit of the Model to Human Volunteer Data
The fit of the model to the human data (Young et al., 1977) (Fig. 9) was problematic. Using physiological parameters of Brown et al. (1997) and measured partitioning parameters (this study and Leung and Paustenbach, 1990) with no metabolism, measured blood 1,4-dioxane concentrations reported by Young et al. could not be achieved unless the estimated exposure concentration was increased from 53 to 100 ppm. Inclusion of any metabolism necessarily decreased predicted blood concentrations. If estimated metabolism rates were used (Tables 1 and 9) with the reported exposure concentration, urinary metabolite excretion was underpredicted. Urinary metabolite excretion rates could be matched if either exposure concentration was increased to 62 ppm or alveolar ventilation (QPC) was increased to 17 l/h·kg0.74. Both of these adjustments are plausible. Because the volunteers were given "bottled water, coffee, and a sandwich on demand" (Young et al., 1977) it is possible that additional 1,4-dioxane partitioned into food and beverages, increasing the total dose. The QPC estimate taken from Brown et al. (1997) (QPC assumed equal to cardiac output), 13 l/h·kg0.74 is on the low side; the average value reported by Price et al. (2003) is 18 l/h·kg0.74. The ventilation rate used by Reitz et al. (1990) equates to a QPC of 30 l/h·kg0.74, which seems inconsistent with the low activity levels (volunteers were seated in an exposure chamber, Young et al., 1977). With the ventilation rate or concentration adjusted to match urinary excretion, the human model predicts significantly lower blood concentrations of 1,4-dioxane (sixfold) than reported by Young et al. (1977). Conversely, if the estimated exposure concentration is increased by a factor of 6, model predictions are consistent with measured blood 1,4-dioxane concentrations of individuals P, T, and G, but urinary excretion of HEAA is overestimated by a factor of 6.
|
To increase the predicted level of 1,4-dioxane in human blood, both Reitz et al. (1990)
and Leung and Paustenbach decreased the effective volume of distribution for the parent compound. The effective volume of distribution is the sum of the blood volume and the sum of the tissue volume multiplied by the ratio of the tissue:air and blood:air partition coefficient for all the tissues. Reitz et al. (1990) decreased the effective volume of distribution by doubling the blood:air partition coefficient, whereas Leung and Paustenbach (1990) reduced the tissue:air partition coefficient of the largest compartment, the slowly perfused tissues, by a factor of 2.5. The validity of these adjustments needed to be considered. The original human blood:air partition coefficient (Leung and Paustenbach, 1990; Reitz et al., 1990) was confirmed in this current study. The measured rat muscle:air partition coefficient was 997 ± 254 (Leung and Paustenbach), but Reitz et al. (1990) used the liver:air partition coefficient (1557) in place of the measured muscle:air value. The current measurements (slowly perfused tissue:air partition coefficient [PSA] = 1348 for rat, 1705 for mouse) confirm that the originally measured value of the rat muscle:air partition coefficient was too low. Thus, the manipulation of the slowly perfused tissue partitioning by Leung and Paustenbach does not seem justified. 1,4-Dioxane appears to be rapidly distributed into tissues (brain, liver, kidney, and testes), with peak concentrations of radiolabel achieved within 15 min of ip injection (Mikheev et al., 1990). The tissue/blood ratios of radiolabel (Mikheev et al., 1990) were consistent with the ratio of the richly perfused tissues:air and blood:air partition coefficients of 1,4-dioxane measured in vitro. Overall, the information on 1,4-dioxane partitioning does not support the alterations Reitz et al. (1990) and Leung and Paustenbach (1990) made in their attempts to fit the human data of Young et al. (1977).
Fit of the Model to Human Occupational Exposure Data
In contrast to the fit to the volunteer blood concentrations, the fit to the urinary concentrations of 1,4-dioxane and HEAA in occupationally exposed workers (Young et al., 1976), the fit was excellent (Table 10). Because there is no "urine compartment" per se, some assumptions were made to convert the Young et al. (1976) urinary concentration data into estimated body burden. It was assumed that the urinary concentration x urine production rate = body burden x elimination rate into urine. The urine production rate was assumed to be 1 ml/min (Young et al., 1977). The elimination rate of 1,4-dioxane into urine by humans (0.0033/h) was taken from Young et al. (1977). The elimination rate of HEAA into urine was the value derived from the mouse model (0.35/h). The group average values of estimated body burden of 1,4-dioxane and HEAA are within 10% of the modeled group average value.
|
Sensitivity Analyses
Mouse PBPK model.
Time course determinations for blood 1,4-dioxane and HEAA SC were computed for the high and low doses of 1,4-dioxane (nominal doses of 2000 and 20 mg/kg). Moderately and highly sensitive parameters for blood 1,4-dioxane and HEAA simulations are summarized in Table 11, and include the optimized parameters KA, VMAXC, VDMC, and KME. The time course of the SC for these parameters is shown in Fig. 10. At the high dose, blood 1,4-dioxane concentrations were extremely sensitive to the value of VMAXC (Fig. 10c).
|
|
Rat PBPK model.
Time course determinations for blood 1,4-dioxane SC were computed a 6-h rat inhalation exposure to 52.1 ppm 1,4-dioxane. Moderately and highly sensitive parameters for blood 1,4-dioxane simulations are summarized in Table 12, and include the optimized parameters KA, VMAXC, VDMC, and KME. The time course of the SC for KM and VMAXC is shown in Fig. 11. During exposure, the blood concentration of 1,4-dioxane is most dependent on ventilation rate and exposure concentration, but in the postexposure period is increasingly dependent on the VMAXC and KM values (Fig. 11). The |SC| values for VMAXC and KM are approximately equal, indicating that the 1,4-dioxane metabolism kinetics are approximately linear.
|
|
Human PBPK model.
Time course determinations for blood 1,4-dioxane and HEAA SC were computed a 6-h human inhalation exposure to 50 ppm 1,4-dioxane. Moderately and highly sensitive parameters for blood 1,4-dioxane and HEAA simulations are summarized in Table 13, and include parameters that have previously been optimized or adjusted by previous modelers (QPC, PB, PSA, KM, VMAXC) (Leung and Paustenbach, 1990
; Reitz et al., 1990). Similar to the findings for rats, the concentration of 1,4-dioxane in the blood during inhalation exposure was most sensitive to the ventilation rate and exposure concentration, with additional parameters, including metabolic parameters, influencing the postexposure blood 1,4-dioxane concentrations (Fig. 12). The human model adjustments reported above ("Fit to Human Volunteer Data") was also a form of sensitivity analysis with larger changes in certain parameter values, all of which were determined to be "highly sensitive" as defined in this analysis.
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAFUNDINGREFERENCES |
|---|
Comparison with Previous PBPK Models
The new data discussed above reduced the uncertainty associated with the PBPK models for 1,4-dioxane. The partition coefficients developed here were consistent with previously measured values. In addition, newly generated values were consistent with many of the estimates used in the models published in 1990. Perhaps the most important is that these results confirm the measured human blood:air partition coefficient values reported by Reitz et al. (1990)
, but not used in the modeling in that paper.
The VmaxC, Km, and Kme derivations for the rat for this modeling effort and the previous efforts (Leung and Paustenbach, 1990; Reitz et al., 1990) drew on the same experimental data sets (Young et al., 1978). The rat VmaxC values derived in this effort (7.5 and 12.7 mg/h·kg0.7, for uninduced and induced rats, respectively) were intermediate between the values determined by Leung and Paustenbach (1990) (normalized values of 5.0 and 9.2 mg/h·kg0.7 calculated from reported Vmax values) and Reitz et al. (1990) (13.7 mg/h·kg0.7) and were similar to the value derived from scaling the in vitro data. The ratio of induced VmaxC to uninduced VmaxC determined by Leung and Paustenbach (1990) was similar to the ratio from the current effort (current: 1.7, previous: 1.8). The in vivo rat Km for the current effort (21 mg/l) was intermediate between the Reitz et al. (1990) and Leung and Paustenbach (1990) values of 29.4 and 7.5 mg/l, respectively. The VmaxC/Km ratios for the current effort (0.36 and 0.60 l/h·kg0.7, uninduced and induced) were closer to the VmaxC/Km ratio of Reitz et al. (1990) (0.47 l/h·kg0.7) than Leung and Paustenbach (0.67 and 1.2 l/h·kg0.7, uninduced and induced). The Kme value of 0.28/h used by Reitz et al. (1990) appeared to have been derived only from the iv data. In contrast, the current evaluation (Kme = 0.48/h) used both iv and oral data, and one of the iv data sets was found to best fit a much lower Kme than the other data sets, as discussed above.
Reitz et al. (1990) estimated VmaxC and Km values for mice by averaging the values derived for rat and humans, but had no data against which to validate these parameters. In the current effort, in vitro data indicated that the mouse Km was similar to the rat value. The in vivo rat Km was identified as 21 mg/l by optimization. This value is similar to the value of 16.2 mg/l previously estimated by Reitz et al. (1990). The VmaxC estimated by Reitz et al. (10 mg/h·kg0.7) is significantly lower than the value estimated using fits to the 200 and 2000 mg/kg dosing data (39 and 45 mg/h·kg0.7, respectively). It is possible that the VmaxC identified for 200 mg/kg does not represent an "uninduced" value, but rather a value that is not induced to the same extent as the 2000 mg/kg dose. In rats, the transition from doses that do not induce 1,4-dioxane metabolism to doses that do induce metabolism is between 100 and 300 mg/kg. The larger discrepancy in mice, as compared with rats, between the in vivo best-fit value and scaled in vitro VmaxC also supports the theory that the 200 mg/kg dose induced 1,4-dioxane metabolism. The optimized mouse model predictions (heavy lines, Figs. 6 and 8b) differ substantially from the Reitz et al. (1990) predictions due to Reitz's apparent fourfold underestimation of the metabolic rate. Based on the mouse kinetic parameters estimated by Reitz et al. (1990), measurable levels of 1,4-dioxane were expected at the 0.5 and 1.0 h sacrifices (Fig. 8b, thin, dark line). Due to the unexpectedly rapid metabolism of 1,4-dioxane, the low dose selected did not yield measurable levels of 1,4-dioxane in blood at any data point (Fig. 8b, thick dark line).
The current 1,4-dioxane PBPK model more fully utilizes the available rat and human pharmacokinetic data than the previous models (Leung and Paustenbach, 1990; Reitz et al., 1990). Leung and Paustenbach (1990) used only the rat blood and exhaled 1,4-dioxane time course data from iv dosing, blood 1,4-dioxane after rat inhalation, and blood 1,4-dioxane and urinary HEAA excretion by human volunteers during and after inhalation (Young et al., 1977, 1978) to parameterize and validate the model. Reitz et al. (1990) did not use the rat 1,4-dioxane exhalation data from iv dosing, but did compare model predictions to cumulative 1,4-dioxane exhalation from rats exposed by gavage (Young et al., 1978), with an estimated rate of absorption from the stomach. In the current effort, we have used all of the data noted above plus the data from human occupational exposure (Young et al., 1976) and time course data for urinary excretion of HEAA and exhalation of 1,4-dioxane from rats dosed by gavage (Young et al., 1978). The current use of blood time course data from a gavage administration (to mice) to estimate the rate of uptake from stomach (new estimate: 0.8/h) is likely to be much more sensitive than total 1,4-dioxane exhalation.
Fit of the Model to Rat and Mouse Experimental Data
The optimized model parameters provide an acceptable fit to the blood measurements of 1,4-dioxane in mice and rats (Figs. 5, 6, and 8
) and exhaled breath 1,4-dioxane at mid- to high doses (Figs. 7b and 7f). The poorer fit to the low-dose exhaled breath 1,4-dioxane (Fig. 7d) may reflect limited metabolism in the upper respiratory tract that does not contribute significantly to whole body metabolism, but scrubs some 1,4-dioxane from exhaled breath. The fit to and prediction of the HEAA data was somewhat less successful than the prediction of the 1,4-dioxane data. The lack of fit to some of the HEAA data was likely due to an overly simplistic description of its distribution and elimination (single compartment, first order elimination). The modeling of HEAA kinetics could potentially be improved with additional data. Its likely distribution throughout the body (i.e., appropriate number of compartments and effective volume of distribution) could be inferred from measured tissue:blood partition coefficients. The relatively high blood HEAA concentrations observed in low-dose mice (Fig. 8b) and the delays in urinary excretion of HEAA in low-dose rats (Fig. 7c) suggest the possibility of saturable reabsorption of HEAA in the kidney. This possibility could be explored with further PBPK modeling and development of blood HEAA time course data in the rat.
Application of the Models in Risk Assessment
The updated PBPK model described in this paper will be useful in future human health risk assessments completed for 1,4-dioxane. The additional data and updated approach reduce the uncertainty associated with the previously available models, particularly with regard to 1,4-dioxane disposition in the mouse (Leung and Paustenbach, 1990; Reitz et al., 1990). The updated model also provides a more scientifically defensible basis for translating animal study data (administered dose to target tissue dose) to estimated human exposures than the default generic scaling used in the absence of chemical-specific information.
The results of the rat nasal tissue exposure study demonstrate nasal tissues are bathed in drinking water under study conditions (similar to tissue exposure in promotion studies), in addition to any systemic exposure. Levels of 1,4-dioxane in drinking water studies where tumors are reported are very high (e.g., NCI, 1978 at 0.5% and 1%, Kociba et al., 1974 at 1%, Yamazaki et al., 1994 at 0.5%). In addition, no tumors, including rat nasal tumors, were reported in the one rat inhalation study that has been published in the literature (Torkelson et al., 1974). Animals were only exposed at one level, 400 mg/m3. These data confirm the position taken in the Reitz et al. (1990), and Leung and Paustenbach (1990) papers that effects observed in the rat nasal tissues are due to direct contact with 1,4-dioxane in drinking water in addition to any systemic dose and not relevant to potential human exposures through drinking water.
There are limited human data that can be used to validate the model. Studies are limited to one study that measured the concentrations in the urine in five workers exposed to 1,4-dioxane (Young et al., 1976) and one study where blood levels were measured in four male volunteers exposed in an inhalation chamber for 6 h to 50 ppm 1,4-dioxane (Young et al., 1977). In the inhalation study, the four volunteers were allowed to eat and drink as desired.
The unadjusted model's predictions are reasonably consistent with the human excretion data (urine levels in workers; Young et al., 1976) (see Table 10), but predict blood levels that were not as consistent with the limited human blood level data. Blood levels reported in the study were approximately sixfold higher than predicted by the model. Thus, the model underpredicted levels in the blood based on this study. The blood concentrations of 1,4-dioxane in volunteer study (Young et al., 1977) were roughly twofold higher than the rat blood 1,4-dioxane concentrations of rats exposed to the same inhalation concentration of 1,4-dioxane (Young et al., 1978). This finding is surprising in the context of both a slightly lower human blood:air partition coefficient and faster metabolism in human hepatocytes. As indicated by the sensitivity analyses, both of these interspecies differences should yield lower blood concentrations in humans as compared with rats.
The discrepancies between the model predictions and experimental data could be addressed in several ways: (1) manipulate the human model parameters (measured parameters such as partition coefficients and default assumptions such as breathing rate) to match the available human in vivo data; (2) use the unadjusted human model as is; (3) use the unadjusted human model, but multiply 1,4-dioxane dose metrics by the sixfold discrepancy with the available experimental data; or (4) generate additional data that can be used to improve the model. The first and third options assume that the Young et al. (1977) human data are "correct" (reported blood levels can reasonably represent the range of values one may find in the general population) and the model predictions are adjusted accordingly. Under option 1, it is unclear which parameters should be altered. During inhalation exposure to 50 ppm 1,4-dioxane, blood 1,4-dioxane concentrations were found to be most sensitive to the value of the ventilation rate and the exposure concentration. Any proposed increase in the ventilation rate must remain within physiologically reasonable values. It is clear from the simulations (Fig. 9a) that if metabolism is slowed to increase predicted blood concentrations during exposure (improving the fit to the data collected during exposure), the fit to the postexposure clearance may suffer. Earlier modelers made substantial adjustments to in vitro partition coefficients (Leung and Paustenbach, 1990; Reitz et al., 1990), but additional experimentation supports the original values. Thus, option 1 is fraught with uncertainty and not attractive from a risk assessment perspective. Under the second option, the model is assumed to be "correct" and the data "wrong/not representative of the general population." If new human studies were conducted, based on the results of the human model sensitivity analyses, the blood concentrations of 1,4-dioxane in humans exposed via inhalation would be expected to be sensitive to ventilation rate, so if possible this parameter should be monitored.
Regardless of how the uncertainty in the human PBPK model is approached in the context of a risk assessment, the updated models are based on more data and stronger science than the previously published models. As such, these models and the analyses or these models provide a better understanding of the behavior of 1,4-dioxane in mice, rats, and humans and provide a more scientifically sound basis for extrapolating study data to human equivalent doses of 1,4-dioxane than generic scaling factors or the previously available models.
|
SUPPLEMENTARY DATA |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAFUNDINGREFERENCES |
|---|
Supplementary data are available online at http://toxsci.oxfordjournals.org/.
|
FUNDING |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAFUNDINGREFERENCES |
|---|
ARCADIS on behalf of the Dioxane Risk Management Consortium.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONSUPPLEMENTARY DATAFUNDINGREFERENCES |
|---|
Agency for Toxic Substances and Disease Registry (ATSDR). Toxicological Profile for 1,4-Dioxane. U.S. Department of Health and Human Services (2006) Atlanta, GA: Public Health Service.
Arias IM, Popper H, Schacher D, Shafritz DA. In: The Liver: Biology and Pathobiology (1982) New York: Raven Press.
Braun WH, Young JD. Identification of b-hydroxyethoxyacetic acid as the major urinary metabolite of 1,4-dioxane in the rat. Toxicol. Appl. Pharmacol. (1977) 39:33–38.[CrossRef][Web of Science][Medline]
Brown RP, Delp MD, Lindstedt SL, Rhomberg LR, Beliles RP. Physiological parameter values for physiologically based pharmacokinetic models. Toxicol. Ind. Health (1997) 13:407–484.
Carlile DJ, Zomorodi K, Houston JB. Scaling factors to relate drug metabolic clearance in hepatic microsomes, isolated hepatocytes, and the intact liver: Studies with induced livers involving diazepam. Drug Metab. Dispos. (1997) 25:903–911.
De Rosa CT, Wilbur S, Holler J, Richter P, Stevens YW. Health evaluation of 1,4-dioxane. Toxicol. Ind. Health (1996) 12:1–43.
Gargas ML, Burgess RJ, Voisard DE, Cason GH, Andersen ME. Partition coefficients of low-molecular-weight volatile chemicals in various liquids and tissues. Toxicol. Appl. Pharmacol. (1989) 98:87–99.[CrossRef][Web of Science][Medline]
Goldsworthy TL, Monticello TM, Morgan KT, Bermudez E, Wilson DM, Jackh R, Butterworth BE. Examination of potential mechanisms of carcinogenicity of 1,4-dioxane in rat nasal epithelial cells and hepatocytes. Arch. Toxicol. (1991) 65:1–9.[CrossRef][Web of Science][Medline]
Kociba RJ, McCollister SB, Park CN, Torkelson TR, Gehring PJ. 1,4-Dioxane. I. Results of a 2-year ingestion study in rats. Toxicol. Appl. Pharmacol. (1974) 30:275–286.[CrossRef][Web of Science]
Leung HW, Paustenbach DJ. Cancer risk assessment for dioxane based upon a physiologically-based pharmacokinetic approach. Toxicol. Lett. (1990) 51:147–162.[CrossRef][Web of Science][Medline]
Mikheev MI, Gorlinskaya Ye.P, Solovyova TV. The body distribution and biological action of xenobiotics. J. Hyg. Epidemiol. Microbiol. Immunol. (1990) 34:329–336.[Medline]
Nannelli A, De Rubertis A, Longo V, Gervasi PG. Effects of dioxane on cytochrome P450 enzymes in liver, kidney, lung and nasal mucosa of rat. Arch. Toxicol. (2005) 79:74–82.[CrossRef][Web of Science][Medline]
National Cancer Institute (NCI). Bioassay of 1,4-dioxane for possible carcinogenicity. Technical Report Series No. 80, NCI-CG-TR-80. In: U.S. Department of Health, Education and Welfare (1978) Washington, DC: NIH.
Price PS, Conolly RB, Chaisson CF, Gross EA, Young JS, Mathis ET, Tedder DR. Modeling interindividual variation in physiological factors used in PBPK models of humans. Crit. Rev. Toxicol. (2003) 33:469–503.[Web of Science][Medline]
Reitz RH, McCroskey PS, Park CN, Andersen ME, Gargas ML. Development of a physiologically based pharmacokinetic model for risk assessment with 1,4-dioxane. Toxicol. Appl. Pharmacol. (1990) 105:37–54.[CrossRef][Web of Science][Medline]
Royal Society of Chemistry. 1,4-Dioxane. Solvents in common use: Health risk to workers. EUR, 11553. (1988) 105–134. (Cited in ATSDR, 2006).
Seglen PO. Preparation of isolated liver cells. Methods Cell Biol. (1976) 13:29–83.[Medline]
Stickney JA, Sager SL, Clarkson JR, Smith LA, Locey BJ, Bock MJ, Hartung R, Olp SF. An updated evaluation of the carcinogenic potential of 1,4-dioxane. Regul. Toxicol. Pharmacol. (2003) 38:183–195.[CrossRef][Web of Science][Medline]
Torkelson TR, Leong BKJ, Kociba RJ, Richter WA, Gehring PJ. 1,4-Dioxane. II. Results of a two year inhalations study in rats. Toxicol. Appl. Pharmacol. (1974) 30:287–298.[CrossRef][Web of Science]
U.S. EPA. 1,4-Dioxane (CASRN 123-91-1). Carcinogenicity assessment for lifetime exposure. (1990) Available at: http://www.epa.gov/iris/subst/0326.htm. Accessed July 31, 2006.
Wala-Jerzykiewicz A, Szymanowski J. Headspace gas chromatography analysis of toxic contaminants in ethoxylated alcohols and alkylamines. Chromatographia (1998) 48:299–304.[CrossRef][Web of Science]
Woo Y-T, Arcos JC, Argus MF, Griffin GW, Nishiyama K. Structural identification of p-dioxane-2-one as the major urinary metabolite of p-dioxane. Arch. Pharmacol. (1977) 299:283–287.[CrossRef]
Yamazaki K, Ohno H, Asakura M, Narumi A, Ohbayashi H, Fujita H, Ohnishi M, Katagiri T, Senoh H, Yamanouchi K, et al. Two-year toxicological and carcinogenesis studies of 1,4-dioxane in F344 rats and BDF1 mice. In: Drinking Studies. Proceedings of the 2nd Asia-Pacific Symposium on Environmental and Occupational Health Environmental and Occupational Hazards (2) (1994) Kobe, Japan: Kobe University. 193–198.
Young JD, Braun WH, Gehring PJ. Dose-dependent fate of 1,4-dioxane in rats. J. Toxicol. Environ. Health (1978) 4:709–726.[Web of Science][Medline]
Young JD, Braun WH, Gehring PJ, Horvath BS, Daniel RL. 1,4-Dioxane and beta-hydroxyethoxyacetic acid excretion in urine of humans exposed to dioxane vapors. Toxicol. Appl. Pharmacol. (1976) 38:643–646.[CrossRef][Web of Science][Medline]
Young JD, Braun WH, Rampy LW, Chenoweth MB, Blau GE. Pharmacokinetics of 1,4-dioxane in humans. J. Toxicol. Environ. Health (1977) 3:507–520.[Web of Science][Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||