Validation of Human Physiologically Based Pharmacokinetic Model for Vinyl Acetate Against Human Nasal Dosimetry Data
http://www.100md.com
《毒物学科学杂志》
Haskell Laboratory for Health and Environmental Sciences, E. I. du Pont de Nemours and Co., Newark, Delaware19714
Battelle Pacific Northwest Laboratories, Biological Sciences Division, Richland, Washington 99352
Dow Benelux NV, Epidemiology, Health Services, Terneuzen, The Netherlands
ABSTRACT
Vinyl acetate has been shown to induce nasal lesions in rodents in inhalation bioassays. A physiologically based pharmacokinetic (PBPK) model for vinyl acetate has been used in human risk assessment, but previous in vivo validation was conducted only in rats. Controlled human exposures to vinyl acetate were conducted to provide validation data for the application of the model in humans. Five volunteers were exposed to 1, 5, and 10 ppm 13C1,13C2 vinyl acetate via inhalation. A probe inserted into the nasopharyngeal region sampled both 13C1,13C2 vinyl acetate and the major metabolite 13C1,13C2 acetaldehyde during rest and light exercise. Nasopharyngeal air concentrations were analyzed in real time by ion trap mass spectrometry (MS/MS). Experimental concentrations of both vinyl acetate and acetaldehyde were then compared to predicted concentrations calculated from the previously published human model. Model predictions of vinyl acetate nasal extraction compared favorably with measured values of vinyl acetate, as did predictions of nasopharyngeal acetaldehyde when compared to measured acetaldehyde. The results showed that the current PBPK model structure and parameterization are appropriate for vinyl acetate. These analyses were conducted from 1 to 10 ppm vinyl acetate, a range relevant to workplace exposure standards but which would not be expected to saturate vinyl acetate metabolism. Risk assessment based on this model further concluded that 24 h per day exposures up to 1 ppm do not present concern regarding cancer or non-cancer toxicity. Validation of the vinyl acetate human PBPK model provides support for these conclusions.
Key Words: vinyl acetate; nasal dosimetry; PBPK modeling, human.
INTRODUCTION
Inhalation bioassays have shown that vinyl acetate induces degeneration of the nasal olfactory epithelium of rats and mice and nasal tumors in rats (Bogdanffy et al., 1994). Because of the differences in the nasal anatomy, physiology, and pharmacokinetic determinants of risk between humans and rodents, approaches to quantitative extrapolation of risk to humans suggest the need for biologically based modeling (Bogdanffy and Sargangapani, 2003). Physiologically based pharmacokinetic (PBPK) models reflect the biology and physiology of various species of interest and the manner in which chemicals are absorbed, distributed, metabolized, and excreted by the body. Previously, PBPK models of vinyl acetate dosimetry were developed for the rat and human (Bogdanffy et al., 1999). These models describe the major pharmacokinetic and sentinel pharmacodynamic steps involved in the pathogenesis of vinyl acetate–induced nasal toxicity.
The mode of action of vinyl acetate has recently been reviewed by Bogdanffy and Valentine (2003), who proposed a series of critical mechanistic steps for vinyl acetate–induced neoplasia. Vinyl acetate is metabolized in nasal tissues to acetaldehyde and acetic acid (Bogdanffy and Taylor, 1993). Ionization of acetic acid at physiological pH and further oxidation of acetaldehyde generate a total of three protons per molecule of vinyl acetate. At sufficient vinyl acetate concentrations, the increased proton levels cause intracellular acidification of nasal epithelial cells, which is believed to be the sentinel pharmacodynamic response. Intracellular acidification enables a subsequent sequence of events that lead to neoplasia (Bogdanffy, 2002; Lantz et al., 2003). The sequence of tissue responses includes acidification-induced mitogenic cellular proliferation of nasal respiratory epithelium and acidification-induced cytotoxic (i.e., compensatory) cellular proliferation in olfactory epithelium.
Organic acids and cellular acidification have been shown to induce clastogenicity and topoisomerase II–mediated DNA strand breaks (Morita, 1995; Xiao et al., 2003). Under conditions of enhanced cellular replication, this genetic damage is converted to mutation that may ultimately be expressed as cancer. At high airborne vinyl acetate concentrations, tissue exposure to the intracellularly formed acetaldehyde may induce DNA–protein crosslinks and clastogenic effects. Alternatively, cellular acidification alone may produce the clastogenic responses that characterize vinyl acetate genotoxicity.
Physiologically based pharmacokinetic models can be applied to test mechanistic hypotheses of mode of action and interspecies extrapolation of chemical dosimetry in biologically based quantitative risk assessments. The PBPK models developed for vinyl acetate described the major steps in the mode of action, and the model predictions of cellular acidification were subsequently demonstrated experimentally. The PBPK model for the rat adequately predicts the measurable pharmacokinetic behavior of vinyl acetate (Bogdanffy et al., 1999), including the rates of deposition of vinyl acetate and production of acetaldehyde over a range of doses and under several different inspired airflow rates.
The structure of the vinyl acetate PBPK model is fundamentally similar to other PBPK models developed to describe upper airway dosimetry of inhaled vapors of various compounds (Andersen et al., 1999, 2002; Frederick et al., 2001; Morris et al., 1993). These models include an inspired air stream that splits in two major directions; a dorsal-medial path and a lateral-ventral path. The airstreams then pass over tissue compartments representative of the anatomical distribution of respiratory and olfactory mucosa. The substructure of each mucosal compartment accounts for diffusion of chemical from the apical surface of the epithelium to the blood-exchange region and includes appropriate kinetic parameters that describe metabolic removal of the chemical from within the tissue compartment.
Although these models have been demonstrated to predict dosimetry of inhaled organic vapors effectively in rats, the human dosimetry models have yet to be validated against data obtained from controlled human studies. A major obstacle was the development of experimental methodology to measure the constituents of air from the nasal cavity in real time. Recently, however, a methodology employing ion trap mass spectrometry has been developed, thereby enabling the study of isotopically mass labeled materials, as applied to 13C-acetone (Thrall et al., 2003). This offers the advantage of distinguishing xenobiotic-derived metabolites from chemically identical endogenous compounds (e.g., acetaldehyde derived from vinyl acetate metabolism). The present work applies this experimental technique to vinyl acetate (using 13C1,13C2 vinyl acetate) to validate PBPK model–predicted human pharmacokinetics (Bogdanffy et al., 1999).
Physiologically based pharmacokinetic model development is typically based on certain defined data sets, with validation conducted on an additional data set not included in the model development. If the validation data set is within the desired range of the model predictions, the model validation is considered successful. Validation is most appropriate and useful when conducted on a data set similar to the final application of the model. For example, many model parameters are developed from in vitro data or extrapolated from other species, but the final model is used for human risk assessment. Thus for true model validation, human in vivo data are most appropriate for validation. The Bogdanffy et al. (1999) model was parameterized by rat in vivo and human in vitro data and validated in vivo in the rat. The present study compares the uptake of inhaled vinyl acetate in the nasal cavity and the concentration of acetaldehyde released into the nasal cavity with the model predictions. In testing the predicted air mass-transfer and metabolite production in the nasal cavity, this investigation addresses key components of the PBPK model.
For this project it was desired to provide additional statistical support to the model validation. A literature search for statistical methods in this area found only one published study (Krishnan et al., 1995), and the described method is not applicable here because it involves combining data from separate studies. Consequently, a more applicable method for validating the human PBPK model for vinyl acetate was developed by the authors and is presented herein. The objective of this work is to provide a human data set for model validation and to provide a statistical measure of model validation.
METHODS AND MATERIALS
Human subjects.
Human studies were conducted under approval from the Battelle Institutional Review Board (IRB) in accordance with the terms and conditions of Federal Regulation 45 CFR 46 under the authority of Multiple Project Assurance M1221, and were conducted in the spirit of the Guidelines for Good Clinical Practice (ICH, 1996; WHO, 1995). Prospective volunteers were given information regarding the study protocol and potential risks associated with vinyl acetate exposure. All volunteers provided signed informed consent. Five healthy non-smoking volunteers (2 women, 3 men) between the ages of 18 years and 58 years, were recruited. They were asked to answer a short questionnaire on their health history, medication used, and dietary habits. They were submitted to a medical examination. Exclusion criteria included nasal congestion or having trouble with nose breathing, a significant medical disease, pregnancy or the possibility for pregnancy, prior adverse reactions to oxymetazoline and/or lidocaine or other chemical sensitivities, or claustrophobia.
Test material and breathing atmosphere.
13C1,13C2 vinyl acetate was obtained from Icon Isotopes (Summit, NJ) and had an isotopic and chemical purity of 99%. 13C1,13C2 acetaldehyde was obtained from Cambridge Isotope Laboratories (Andover, MA), and had an isotopic and chemical purity of 99% and 98%, respectively. Unlabeled vinyl acetate and acetaldehyde were obtained from Sigma-Aldrich (Milwaukee, WI), and had a chemical purity of 99% and 99.5%, respectively. The last material was used to test for possible interference in the analysis. The test atmosphere was prepared by injecting a known volume of 13C1,13C2 vinyl acetate into a 500 l foil-lined bag (Hans Rudolph, Inc., Kansas City, MO) into which hospital grade (certified grade D) purified breathing air was metered (Cole-Parmer, Inc., Vernon Hill, IL) to achieve a target air concentration of 1, 5, or 10 ppm. Exposure atmosphere was provided to the volunteer on demand from the exposure bag equipped with a two-way stopcock to switch from bag atmosphere to room air.
Gas chromatographic analyses of test atmospheres.
To confirm the concentration of 13C1,13C2 vinyl acetate test material in the test atmosphere, samples were collected in triplicate from the exposure bag before and after exposure. These samples were analyzed by a gas chromatograph (GC) method in an Agilent model 6890 system (Palo Alto, CA) equipped with a hydrogen flame ionization detector (FID) and a DB-Wax column, 30 m x 0.5 mm, 1.0 μm film thickness (J&W Scientific, Folsom, CA). The detector was operated at 250°C and the inlet at 240°C. Under these conditions, 13C1,13C2 vinyl acetate had a retention time of approximately 1.4 min.
Nasopharyngeal probe.
The nasopharyngeal probe consisted of a sterilized small-diameter (10-French) control suction catheter (Baxter Healthcare Corp., Deerfield, IL) modified by removal of the suction control. Pilot studies verified no loss of the compounds in the nasal probe. The nasopharyngeal probe was used to sample air directly from the nasopharyngeal cavity, allowing the determination of the concentration of the test material and metabolites.
Human exposure conditions.
Before each experiment, the volunteer was examined by the attending physician for any new conditions that might interfere with the experiment or result in unnecessary discomfort for the volunteer, such as a cough. Under the supervision of the study physician, the volunteer was administered a nasal aerosol spray consisting of 2% lidocaine and 0.025% oxymetazoline hydrochloride. Lidocaine, a local anesthetic, is routinely injected, but it can also be used topically in the nose, mouth, or throat. Oxymetazoline hydrochloride is a commercially available topical decongestant normally used for over-the-counter treatment of nasal congestion associated with rhinitis. The aerosol spray was used to facilitate placement of the probe with minimal discomfort to the subject.
The nasopharyngeal probe was inserted approximately 9 cm into one nostril such that the tip of the catheter was positioned in the nasopharyngeal region (Fig. 1). Placement of the catheter was verified visually with the use of a fiber-optic flexible nasopharyngoscope. The exterior portion of the catheter was threaded through a port on a Hans Rudolph (Kansas City, MO) facemask containing two one-way non-rebreathing valves; the port was sealed with polytetrafluoroethylene (ptfe) tape to prevent leakage. The exterior portion of the probe outside the facemask was joined with a ptfe transfer line, which fed directly into an ion-trap mass spectrometer (Fig. 2).
Volunteers were instructed to inhale and exhale through the nose. During the experiment, the volunteer was instructed to maintain a regular breathing rhythm under the testing regime of rest and light exercise according to a predetermined protocol outlined in Table 1. Light exercise was achieved by pedaling a Schwinn Model 217p magnetic resistance stationary bicycle at a rate sufficient to achieve a workload of 50 W, as displayed by the bicycle computer. Light exercise was included in the protocol to investigate whether dispositional changes are caused by the exercise state.
The three exposure levels of 1, 5, and 10 ppm were administered in a random fashion for each volunteer to minimize the effect of learning on the results. All exposure levels were below the ACGIH TLV of 10 ppm (ACGIH, 2003). For practical reasons, but also to reduce variability, all experiments were conducted early in the afternoon. Each exposure for a given volunteer was conducted on a separate day.
Real-time breath-analysis system.
The system used to monitor nasopharyngeal vapor concentrations during the human studies has been described elsewhere(Thrall et al., 2003). In brief, a Discovery II (LGC, Inc., Los Gatos, CA) ion-trap tandem mass spectrometer (MS/MS) equipped with an atmospheric sampling glow discharge ionization (ASGDI) source was used to sample directly from the nasopharyngeal probe by connecting the probe to a ptfe tube attached to the inlet orifice of the MS/MS system. Nasopharyngeal-region air was thus sampled at a calibrated flow rate of 12 l/h. This high flow in comparison with the dead volume of the connective tubing minimized the lag in analytical response as much as possible. A new data point was provided every 0.8 s for both 13C1,13C2 vinyl acetate and 13C1,13C2 acetaldehyde. Intensity data from the MS/MS systems were converted to concentration (ppb) using external gas standards prepared in Tedlar bags (Supelco, Inc., Bellefonte, PA) and a calibration curve. New standards and calibration curves were generated each day of operation.
The mass fragmentation of 13C1,13C2 vinyl acetate was evaluated with methane gas introduced through the glow discharge cornea as the collision gas. Blank methane gas produced minor background mass fragments at mass to charge (m/z) ratios 68, 72, 103, 122, and 134. In comparison, 13C1,13C2 vinyl acetate produced major mass fragments at m/z ratios 61 and 91 and a unique fragment at m/z ratio 74. Evaluation of the 13C1,13C2 vinyl acetate standards at the m/z ratio of 74 showed good linearity (R2 = 0.999) from the level of quantification of 3 ppb up to 66,000 ppb. Under the same analytical method, 13C1,13C2 acetaldehyde produced major mass fragments at m/z ratio 61 and 95, and a unique fragment at m/z ratio 63. Evaluation of the 13C1,13C2 acetaldehyde standards at the m/z ratio of 63 showed good linearity (R2 = 1) from the level of quantification of 8 ppb up to 16,000 ppb. In contrast, 12C-acetaldehyde produced major peaks at m/z ratio 61 and 91, and a unique fragment at m/z ratio 59. Nave exhaled breath showed minor endogenous mass fragments at m/z ratios of 59, 61, 62, and 134, which did not correspond to nor interfere with the analyses of any of the 13C-compounds. Analyses indicated that one compound had no impact on the analysis of the other over the range of concentrations tested.
Physiological monitoring.
Breathing amplitude was monitored using a plethysmograph (Qubit Systems, Inc., Kingston, Ontario). It is made up of a breathing monitor belt with an inflatable rubber bladder, worn around the lower chest and upper abdomen and connected to an absolute pressure sensor. The sensor monitored the change in air pressure within the bladder as the lung expanded and contracted, and transmitted a voltage oscillation signal to a portable computer via an interface. Breathing frequency was calculated by dividing the number of breaths by the time, in minutes, over which the breaths were taken.
An acoustic rhinometry system (RhinoMetrics A/S, Lynge, Denmark) was used to measure the cross-sectional area and volume of the nasal cavity and provide a two-dimensional graphic display. The rhinometer measures the nasal airway dimensions by emitting wideband noise into the nose and digitally analyzing the incident and reflected waves recorded by a microphone (Hytnen et al., 1996). The mean of five measurements is displayed as a curve and is updated four times per second. The nasal cavity of each volunteer was measured both before and after application of the nasal spray to evaluate mucosal changes and to validate the volume for the human nasal cavity used in the model.
Preparation for model evaluation.
The human exposures provide 13C1,13C2 vinyl acetate and 13C1,13C2 acetaldehyde concentrations in the nasopharyngeal region during inhalation exposure for model validation. The human nasal PBPK model (Bogdanffy et al., 1999) predicts the corresponding concentrations of vinyl acetate and acetaldehyde in the nasopharyngeal region at given inspired concentrations of vinyl acetate. The rat PBPK model has previously been validated using a constant unidirectional (inhaled) air flow through the nasopharyngeal cavity (Plowchalk et al., 1997). In the experiments described here, the flow, by necessity, is bidirectional; alternatively inhaled and exhaled air passes the probe at the back of the nasal cavity. The experimental set-up did not allow for linking nasopharyngeal sample collection with the inhalation phase of the breathing cycle. Consequently, air concentration measurements were taken every 0.8 s, resulting in recording data during all phases of the breathing cycle in an uncontrolled pattern.
The PBPK model predicts rapid equilibration between the tissue and nasal air phase during vinyl acetate exposure (dotted line in upper panel of Figure 3), with air phase concentrations coming to a constant value in less than 0.01 s. Thus the model estimates essentially reflect steady state concentrations in the nasopharyngeal region (per breath). Due to the rapid kinetics, the experimental design cannot measure the intermediate concentrations in each breath but rather provide a sampling that represents a concentration versus time at various phases of the breathing cycle. This given, the measurements of vinyl acetate concentrations under experimental conditions result in concentration curves as shown in Figure 3. Taking into consideration the experimental factors, the average of peak concentrations during the inhalation phases in a specific test period (e.g., resting, light exercise) were selected as the indicator most closely mirroring the animal validation set-up.
From the plethysmograph data, the number of breaths in a given exposure period (2–5 min) was estimated. The exposure period was then split into a corresponding number of time segments and the peak concentration in each segment was identified. Segments with no clear peak concentration were not included in the analysis. Once the peak concentrations were identified, the mean and standard deviation were calculated for each resting and exercise subgroup of an exposure experiment (see Table 1) for use in the subsequent steps.
Model evaluation.
The previously published human nasal model (Bogdanffy et al., 1999) was exercised at the actual exposure concentrations (Table 2). Briefly, the Bogdanffy model is a PBPK model of the respiratory and olfactory compartments of the human nose. The nasal tissue is split into four compartments (three respiratory and one olfactory), each containing compartments representing the mucus, epithelial cells, basal cells, and submucosa. Inhaled vinyl acetate is extracted into the nasal tissue, diffuses through the various layers, and is subject to metabolism. The model traces the parent and metabolite concentrations in the various compartments, along with the pH change induced by the vinyl acetate and acetaldehyde metabolism. For the current validation, the model predictions of the air concentrations leaving nasal compartment are compared to the experimental data. Full details of the model are presented in Bogdanffy et al. (1999) and the model code is presented in the Supplementary Material online.
Model parameterization was accomplished by means of a combination of measured human values, in vitro extrapolated values, and scaled rat values (Bogdanffy et al., 1999). All of the metabolic and physiological parameters were unchanged from the original published model parameterization (see Supplementary Material online). There were two main reasons for not changing the published model parameterization. The first was that from a strict modeling standpoint, validation is an exercise to determine the quality of the model and model predictions not to make adjustments to the model. Second, the primary application of this model is to risk assessment that models an entire population, not individuals, thus the generic human parameters were retained.
A sensitivity analysis of the model parameters on the output values was also conducted. A prior sensitivity analysis (Bogdanffy et al., 1999) only used intracellular pH as the end point, the current sensitivity analysis was conducted using air phase concentrations as the sensitivity variable. All modeling work was conducted in acslXtreme (version 1.4, AEgis Technologies Group, Huntsville, AL).
Statistical comparison.
To compare the experimental data and the model simulations, a statistical comparison was needed. A statistical evaluation of the goodness of fit of the line describing the simulated maximal vinyl acetate and acetaldehyde concentrations to the mean experimental peak concentrations was conducted. Because the model predicts linear relationships between exposure concentration and both nasopharyngeal vinyl acetate and acetaldehyde concentrations, Pearson's r was used as a statistical measure of the quality of the linear model fit. The resulting statistics described the quality of the model fit and provided a measure of the validation of the model.
RESULTS
The methods development conducted in this project resulted in an analytical method for the detection of 13C1,13C2 acetaldehyde and 13C1,13C2 vinyl acetate in air samples from the nasal cavity with instantaneous readout.. The method is free of interference from endogenous compounds and its detection limit is in the single-digit ppb level. For some volunteers, the acetaldehyde data were rejected because the raw mass/charge ratio intensity decreased during the course of the study resulting in the acetaldehyde intensity signal drifting downward over time. As a result, there are fewer acetaldehyde data sets in the model validation (Table 1).
General Observations
No volunteer reported pain or discomfort associated with the nasopharyngeal probe. All volunteers were tolerant of the lidocaine/oxymetazoline hydrochloride spray, nasal probe, facemask, exercise bicycle, and physiological monitoring equipment. All volunteers reported being able to smell the compound during exposure; no volunteer reported the sweet odor or other aspects of the experiments as unpleasant. The plethysmography results indicate an increase in breathing frequency with increasing exposure level, so that the breathing frequency had doubled at 10 ppm. This reflexive change in breathing frequency is a reaction on the stronger odor at higher exposure levels and also results in more shallow breathing.
Model Evaluation
The complete 13C1,13C2 vinyl acetate and 13C1,13C2 acetaldehyde data sets were treated as described in the methods section. An example of the resulting graph for vinyl acetate is pictured in Figure 3. Data representing peak values are shown in Figures 4 and 5 for vinyl acetate and acetaldehyde, respectively. For each individual volunteer, the data from each exposure sequence (i.e., the three resting and three light exercise periods) were characterized by a mean and standard deviation (Figs. 6 and 7), along with the model simulations. All model parameters used to simulate the experiments, including nasal dimensions, are unchanged from Bogdanffy et al. (1999). The model only simulates the upper respiratory tract. It does not cover the lower respiratory tract and little contribution is expected from the exhaled breath due to expected absorption in this region. This assumption also implies that the bi-directional flow in the human nose will only contain vinyl acetate from inhalation, not from release from the lower respiratory tract.
For both the vinyl acetate and acetaldehyde data sets, nearly all points are close to the model predicted peak concentrations (Pearson's r of 0.9 and 0.6 for VA and AA, respectively). The quality of the acetaldehyde fit is less than the vinyl acetate fit but this is largely due to the small number of experimental data points. A large portion of the remaining variation is not due to limitations in the model but rather to scatter in the experimental data. The vinyl acetate data may suggest a leveling off at concentrations above 9 ppm. However, if this indicated an increased deposition if vinyl acetate, an increase in acetaldehyde levels should also have appeared. This is not seen (Fig. 5) and a better explanation is an increase in breathing frequency, resulting in more difficulty of the analytical set up to accurately measure peak concentrations.
Sensitivity analysis
Because the purpose of this work was to validate the human model of Bogdanffy et al. (1999), model parameters were not adjusted but are the same as presented in the original work. Volunteer body weights varied between 47 kg and 97 kg, but because the model only includes the nasal region instead of the whole body, body weight is not a direct parameter in the model. Individual variation was measured by rhinometer measures of the nasal cavity volume (mean 12.1 ± 2.4 cm3). The default model value was 9.96 cm3. However, a 50% increase or decrease in the nasal cavity volume, and corresponding tissue surface areas and volumes, produced less than a 3% change in the nasopharyngeal vinyl acetate deposition. Sensitivity analysis indicates that the air phase concentrations are most sensitive to the tissue:air partition coefficients, nasal air flow, and the metabolic capacity of the mucus layer (results not shown). Because tissue metabolism of vinyl acetate is not a sensitive model parameter, interindividual variation in metabolic capacity is not expected to provide significant changes in the nasopharyngeal concentrations. Figures 6 and 7 also include model simulations with the inspired air flow at half and twice the default values to show that the majority of the data points fall within the range of data expected by normal breathing variation.
DISCUSSION
The complete set of human exposures resulted in a data set with over 22,000 vinyl acetate and acetaldehyde data points. To aid in the interpretation of this large data set, it was necessary to identify those data that were relevant to the vinyl acetate model predictions. A 1 min sample of the vinyl acetate and data sets is given in Figure 3. As can be seen, the rate of sampling was rapid enough to capture much of the behavior of vinyl acetate in the human nasal cavity, including inhalation and exhalation. The sampling frequency is not short enough, however, to accurately capture the peak concentration in every breath. This experimental limitation resulted in increased variability in the dataset.
The data treatment described above was needed to identify the relevant experimental data—i.e., the peak concentrations—for comparison to the model output. The PBPK model predicts a very rapid equilibration of both vinyl acetate and acetaldehyde in the nasal airway. Because of this, the only data of use for validating the model are the peak concentrations of each breath. Breaths with peak levels far below the peaks seen in most breaths are likely experimental artifacts from volunteer breathing pattern and experimental response time and were not included in the analysis. Samples collected while breathing room air are also shown in Figures 4 through 7 to demonstrate the rapid vinyl acetate and acetaldehyde wash out and low carryover between experimental segments.
The model output is shown in Figures 4 and 5, along with the mean nasopharyngeal concentrations of vinyl acetate and acetaldehyde. The airway concentrations of both compounds appear to be linear as the model predicts. Figures 6 and 7 show the mean peak concentrations by volunteer and resting/exercise state. It can be seen that given the variations in the experimental data there is no significant inter-individual variation and no significant difference in nasopharyngeal concentrations of vinyl acetate and acetaldehyde based on exercise state. Because of this, the standard model parameters, such as ventilation rate and nasal dimensions, were not adjusted for the validation data set. These results demonstrate that the model reasonably predicts the experimental observations with regards to nasal disposition of inhaled vinyl acetate and wash-out of acetaldehyde in a concentration range close to the TLV of 10 ppm.
One inopportune aspect of the study design was the necessary use of the nasal spray containing oxymetazoline hydrochloride and lidocaine to decongest and anesthetize the nasal cavity. Although lidocaine is a short-acting anesthetic, vasoconstriction induced by oxymetazoline hydrochloride can persist for 5 to 6 h. Although the impact of this vasoconstriction nasal uptake efficiencies compared to normal physiological conditions is not known, sensitivity analyses of the rat and human vinyl acetate PBPK model suggest little role, if any, for nasal blood perfusion in the clearance kinetics of vinyl acetate (Bogdanffy et al., 1999). Removal of vinyl acetate diffusing through the epithelial cell layers toward the blood-exchange region through very fast metabolic steps reduces the amount reaching the vasculature substantially (Bogdanffy and Valentine, 2003).
CONCLUSIONS
These experiments present the first attempts to validate a human nasal dosimetry PBPK model. Model predictions of vinyl acetate nasal extraction data compared favorably with measured values obtained using 13C1,13C2 vinyl acetate as did predictions of nasopharyngeal acetaldehyde when compared to measured 13C1,13C2 acetaldehyde. The results show that the current PBPK model structure is appropriate for inhaled vinyl acetate and, by extension, would likely seem appropriate for inhalation of similar organic esters. These analyses were conducted over a range of concentrations relevant to workplace exposure standards. The current ACGIH 8 h time-weighted average workplace exposure standards for vinyl acetate is 10 ppm (ACGIH, 2003). The PBPK model validated here has been used in a risk assessment that suggested these current workplace standards are adequately protective. A risk assessment based on this model further concluded that 24 h per day exposures up to 1 ppm do not present concern regarding cancer or non-cancer toxicity. Validation of the vinyl acetate human PBPK model provides support to these conclusions.
NOTES
The authors certify that all research involving human subjects was done under full compliance with all government policies and the Helsinki Declaration.
ACKNOWLEDGMENTS
The authors wish to thank the members of the CEFIC Acetyls Technical Committee for their review and comment on the study design. This work was funded entirely by the CEFIC Acetyls Sector Group, Brussels, Belgium.
REFERENCES
ACGIH (American Conference of Governmental Industrial Hygienists). (2003). Threshold limit values for chemical substances and physical agents & biological exposure indices. SCGIH, Cincinnati, OH.
Andersen, M. E., Green, T., Frederick, C. B., and Bogdanffy, M. S. (2002). Physiologically based pharmacokinetic (PBPK) models for nasal tissue dosimetry of organic esters: Assessing the state-of-the-knowledge and risk assessment applications with methyl methacrylate and vinyl acetate. Reg. Toxicol. Pharmacol. 36, 234–245.
Andersen, M. E., Sarangapani, R., Frederick, C. B., and Kimbell, J. S. (1999). Dosimetric adjustment factors for methyl methacrylate derived from a steady-state analysis of a physiologically based clearance-extraction model. Inhal. Toxicol. 11, 899–926.
Bogdanffy, M. S., and Valentine, R. (2003). Differentiating between local cytotoxicity, mitogenesis, and genotoxicity in carcinogen risk assessments: the case of vinyl acetate. Toxicol. Lett. 140/141, 83–98.
Bogdanffy, M. S., and Sarangapani, R. (2003). Physiologically-based kinetic modeling of vapours toxic to the respiratory tract. Toxicol. Lett. 138, 103–117.
Bogdanffy, M. S. (2002). Vinyl acetate-induced intracellular acidification: implications for risk assessment. Toxicol. Sci. 66, 320–326.
Bogdanffy, M. S., and Taylor, M. L. (1993). Kinetics of carboxylesterase-mediated hydrolysis of the nasal carcinogen vinyl acetate. Drug Metab. Dispos. 21, 1107–1111.
Bogdanffy, M. S., Sarangapani, R., Plowchalk, D. R., Jarabek, A., and Andersen, M. E. (1999). A biologically based risk assessment for vinyl acetate-induced cancer and noncancer inhalation toxicity. Toxicol. Sci. 51, 19–35.
Bogdanffy, M. S., Dreef-van der Meulen, H. C., Beems, R. B., Feron, V. J., Cascieri, T. C., Tyler, T. R., Vinegar, M. B., and Rickard, R. W. (1994). Chronic toxicity and oncogenicity inhalation study with vinyl acetate in the rat and mouse. Fund. Appl. Tox. 23, 215–229.
Frederick, C. B., Gentry, P. R., Bush, M. L., Lomax, L. G., Black, K. A., Finch, L., Kimbell, J. S., Morgan, K. T., Subramaniam, R., Morris, J. B., and Ultman, J. S. (2001). A hybrid computational fluid dynamics and physiologically based pharmacokinetic model for comparison of predicted tissue concentrations of acrylic acid and other vapors in the rat and human nasal cavities following inhalation exposure. Inhal. Toxicol. 13, 359–376.
Hytnen, M. L., Sala, E. L., Malmberg, H. O., and Nordman, H. (1996). Acoustic rhinometry in the diagnosis of occupational rhinitis. Am. J. Rhinology 10, 393–397.
ICH (ICH Harmonised Tripartite Guideline). (1996). "Guidelines for Good Clinical Practice." International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use, International Federation of Pharmaceutical Manufacturers and Associations, Switzerland.
Krishnan, K., Haddad, S., and Pelekis, M. (1995). A simple index for representing the discrepancy between simulations of physiological pharmacokinetic models and experimental data. Toxicol. Indust. Health 11, 413–421.
Lantz, R. C., Orozco J., and Bogdanffy M. S. (2003). Vinyl acetate decreases intracellular pH in rat nasal epithelial cells. Toxicol. Sci. 75, 423–431.
Morita, T. (1995). Low pH leads to sister-chromatid exchanges and chromosomal aberrations, and its clastogenicity is S-dependent. Mutation Res. 334, 301–308.
Morris, J. B., Hassett, D. N., and Blanchard, K. T. (1993). A physiologically based pharmacokinetic model for nasal uptake and metabolism of nonreactive vapors. Toxicol. Appl. Pharmacol. 123, 120–129.
Plowchalk, D. R., Andersen, M. E., and Bogdanffy, M. S. (1997). Physiologically based modeling of vinyl acetate uptake, metabolism, and intracellular pH changes in the rat nasal cavity. Toxicol. Appl. Pharmacol. 142, 386–400.
Thrall, K. D., Schwartz, R. E., Weitz, K. K., Soelberg, J. J., Foureman, G. L., Prah, J. D., and Timchalk, C. (2003). A real-time method to evaluate the nasal deposition and clearance of acetone in the human volunteer. Inhal. Toxicol. 15, 523–538.
Xiao, H., Li, T.-K., Yang, J.-M., and Liu, L. F. (2003). Acidic pH induces topoisomerase II-mediated DNA damage. Proc. Natl. Acad. Sci. U.S.A. 100, 5205–5210.
WHO (World Health Organization). (1995). Guidelines for Good Clinical practice (GCP) for trials on pharmaceutical products. Who Technical Report series, No. 850, Annex 3.(P. M. Hinderliter, K. D. )
Battelle Pacific Northwest Laboratories, Biological Sciences Division, Richland, Washington 99352
Dow Benelux NV, Epidemiology, Health Services, Terneuzen, The Netherlands
ABSTRACT
Vinyl acetate has been shown to induce nasal lesions in rodents in inhalation bioassays. A physiologically based pharmacokinetic (PBPK) model for vinyl acetate has been used in human risk assessment, but previous in vivo validation was conducted only in rats. Controlled human exposures to vinyl acetate were conducted to provide validation data for the application of the model in humans. Five volunteers were exposed to 1, 5, and 10 ppm 13C1,13C2 vinyl acetate via inhalation. A probe inserted into the nasopharyngeal region sampled both 13C1,13C2 vinyl acetate and the major metabolite 13C1,13C2 acetaldehyde during rest and light exercise. Nasopharyngeal air concentrations were analyzed in real time by ion trap mass spectrometry (MS/MS). Experimental concentrations of both vinyl acetate and acetaldehyde were then compared to predicted concentrations calculated from the previously published human model. Model predictions of vinyl acetate nasal extraction compared favorably with measured values of vinyl acetate, as did predictions of nasopharyngeal acetaldehyde when compared to measured acetaldehyde. The results showed that the current PBPK model structure and parameterization are appropriate for vinyl acetate. These analyses were conducted from 1 to 10 ppm vinyl acetate, a range relevant to workplace exposure standards but which would not be expected to saturate vinyl acetate metabolism. Risk assessment based on this model further concluded that 24 h per day exposures up to 1 ppm do not present concern regarding cancer or non-cancer toxicity. Validation of the vinyl acetate human PBPK model provides support for these conclusions.
Key Words: vinyl acetate; nasal dosimetry; PBPK modeling, human.
INTRODUCTION
Inhalation bioassays have shown that vinyl acetate induces degeneration of the nasal olfactory epithelium of rats and mice and nasal tumors in rats (Bogdanffy et al., 1994). Because of the differences in the nasal anatomy, physiology, and pharmacokinetic determinants of risk between humans and rodents, approaches to quantitative extrapolation of risk to humans suggest the need for biologically based modeling (Bogdanffy and Sargangapani, 2003). Physiologically based pharmacokinetic (PBPK) models reflect the biology and physiology of various species of interest and the manner in which chemicals are absorbed, distributed, metabolized, and excreted by the body. Previously, PBPK models of vinyl acetate dosimetry were developed for the rat and human (Bogdanffy et al., 1999). These models describe the major pharmacokinetic and sentinel pharmacodynamic steps involved in the pathogenesis of vinyl acetate–induced nasal toxicity.
The mode of action of vinyl acetate has recently been reviewed by Bogdanffy and Valentine (2003), who proposed a series of critical mechanistic steps for vinyl acetate–induced neoplasia. Vinyl acetate is metabolized in nasal tissues to acetaldehyde and acetic acid (Bogdanffy and Taylor, 1993). Ionization of acetic acid at physiological pH and further oxidation of acetaldehyde generate a total of three protons per molecule of vinyl acetate. At sufficient vinyl acetate concentrations, the increased proton levels cause intracellular acidification of nasal epithelial cells, which is believed to be the sentinel pharmacodynamic response. Intracellular acidification enables a subsequent sequence of events that lead to neoplasia (Bogdanffy, 2002; Lantz et al., 2003). The sequence of tissue responses includes acidification-induced mitogenic cellular proliferation of nasal respiratory epithelium and acidification-induced cytotoxic (i.e., compensatory) cellular proliferation in olfactory epithelium.
Organic acids and cellular acidification have been shown to induce clastogenicity and topoisomerase II–mediated DNA strand breaks (Morita, 1995; Xiao et al., 2003). Under conditions of enhanced cellular replication, this genetic damage is converted to mutation that may ultimately be expressed as cancer. At high airborne vinyl acetate concentrations, tissue exposure to the intracellularly formed acetaldehyde may induce DNA–protein crosslinks and clastogenic effects. Alternatively, cellular acidification alone may produce the clastogenic responses that characterize vinyl acetate genotoxicity.
Physiologically based pharmacokinetic models can be applied to test mechanistic hypotheses of mode of action and interspecies extrapolation of chemical dosimetry in biologically based quantitative risk assessments. The PBPK models developed for vinyl acetate described the major steps in the mode of action, and the model predictions of cellular acidification were subsequently demonstrated experimentally. The PBPK model for the rat adequately predicts the measurable pharmacokinetic behavior of vinyl acetate (Bogdanffy et al., 1999), including the rates of deposition of vinyl acetate and production of acetaldehyde over a range of doses and under several different inspired airflow rates.
The structure of the vinyl acetate PBPK model is fundamentally similar to other PBPK models developed to describe upper airway dosimetry of inhaled vapors of various compounds (Andersen et al., 1999, 2002; Frederick et al., 2001; Morris et al., 1993). These models include an inspired air stream that splits in two major directions; a dorsal-medial path and a lateral-ventral path. The airstreams then pass over tissue compartments representative of the anatomical distribution of respiratory and olfactory mucosa. The substructure of each mucosal compartment accounts for diffusion of chemical from the apical surface of the epithelium to the blood-exchange region and includes appropriate kinetic parameters that describe metabolic removal of the chemical from within the tissue compartment.
Although these models have been demonstrated to predict dosimetry of inhaled organic vapors effectively in rats, the human dosimetry models have yet to be validated against data obtained from controlled human studies. A major obstacle was the development of experimental methodology to measure the constituents of air from the nasal cavity in real time. Recently, however, a methodology employing ion trap mass spectrometry has been developed, thereby enabling the study of isotopically mass labeled materials, as applied to 13C-acetone (Thrall et al., 2003). This offers the advantage of distinguishing xenobiotic-derived metabolites from chemically identical endogenous compounds (e.g., acetaldehyde derived from vinyl acetate metabolism). The present work applies this experimental technique to vinyl acetate (using 13C1,13C2 vinyl acetate) to validate PBPK model–predicted human pharmacokinetics (Bogdanffy et al., 1999).
Physiologically based pharmacokinetic model development is typically based on certain defined data sets, with validation conducted on an additional data set not included in the model development. If the validation data set is within the desired range of the model predictions, the model validation is considered successful. Validation is most appropriate and useful when conducted on a data set similar to the final application of the model. For example, many model parameters are developed from in vitro data or extrapolated from other species, but the final model is used for human risk assessment. Thus for true model validation, human in vivo data are most appropriate for validation. The Bogdanffy et al. (1999) model was parameterized by rat in vivo and human in vitro data and validated in vivo in the rat. The present study compares the uptake of inhaled vinyl acetate in the nasal cavity and the concentration of acetaldehyde released into the nasal cavity with the model predictions. In testing the predicted air mass-transfer and metabolite production in the nasal cavity, this investigation addresses key components of the PBPK model.
For this project it was desired to provide additional statistical support to the model validation. A literature search for statistical methods in this area found only one published study (Krishnan et al., 1995), and the described method is not applicable here because it involves combining data from separate studies. Consequently, a more applicable method for validating the human PBPK model for vinyl acetate was developed by the authors and is presented herein. The objective of this work is to provide a human data set for model validation and to provide a statistical measure of model validation.
METHODS AND MATERIALS
Human subjects.
Human studies were conducted under approval from the Battelle Institutional Review Board (IRB) in accordance with the terms and conditions of Federal Regulation 45 CFR 46 under the authority of Multiple Project Assurance M1221, and were conducted in the spirit of the Guidelines for Good Clinical Practice (ICH, 1996; WHO, 1995). Prospective volunteers were given information regarding the study protocol and potential risks associated with vinyl acetate exposure. All volunteers provided signed informed consent. Five healthy non-smoking volunteers (2 women, 3 men) between the ages of 18 years and 58 years, were recruited. They were asked to answer a short questionnaire on their health history, medication used, and dietary habits. They were submitted to a medical examination. Exclusion criteria included nasal congestion or having trouble with nose breathing, a significant medical disease, pregnancy or the possibility for pregnancy, prior adverse reactions to oxymetazoline and/or lidocaine or other chemical sensitivities, or claustrophobia.
Test material and breathing atmosphere.
13C1,13C2 vinyl acetate was obtained from Icon Isotopes (Summit, NJ) and had an isotopic and chemical purity of 99%. 13C1,13C2 acetaldehyde was obtained from Cambridge Isotope Laboratories (Andover, MA), and had an isotopic and chemical purity of 99% and 98%, respectively. Unlabeled vinyl acetate and acetaldehyde were obtained from Sigma-Aldrich (Milwaukee, WI), and had a chemical purity of 99% and 99.5%, respectively. The last material was used to test for possible interference in the analysis. The test atmosphere was prepared by injecting a known volume of 13C1,13C2 vinyl acetate into a 500 l foil-lined bag (Hans Rudolph, Inc., Kansas City, MO) into which hospital grade (certified grade D) purified breathing air was metered (Cole-Parmer, Inc., Vernon Hill, IL) to achieve a target air concentration of 1, 5, or 10 ppm. Exposure atmosphere was provided to the volunteer on demand from the exposure bag equipped with a two-way stopcock to switch from bag atmosphere to room air.
Gas chromatographic analyses of test atmospheres.
To confirm the concentration of 13C1,13C2 vinyl acetate test material in the test atmosphere, samples were collected in triplicate from the exposure bag before and after exposure. These samples were analyzed by a gas chromatograph (GC) method in an Agilent model 6890 system (Palo Alto, CA) equipped with a hydrogen flame ionization detector (FID) and a DB-Wax column, 30 m x 0.5 mm, 1.0 μm film thickness (J&W Scientific, Folsom, CA). The detector was operated at 250°C and the inlet at 240°C. Under these conditions, 13C1,13C2 vinyl acetate had a retention time of approximately 1.4 min.
Nasopharyngeal probe.
The nasopharyngeal probe consisted of a sterilized small-diameter (10-French) control suction catheter (Baxter Healthcare Corp., Deerfield, IL) modified by removal of the suction control. Pilot studies verified no loss of the compounds in the nasal probe. The nasopharyngeal probe was used to sample air directly from the nasopharyngeal cavity, allowing the determination of the concentration of the test material and metabolites.
Human exposure conditions.
Before each experiment, the volunteer was examined by the attending physician for any new conditions that might interfere with the experiment or result in unnecessary discomfort for the volunteer, such as a cough. Under the supervision of the study physician, the volunteer was administered a nasal aerosol spray consisting of 2% lidocaine and 0.025% oxymetazoline hydrochloride. Lidocaine, a local anesthetic, is routinely injected, but it can also be used topically in the nose, mouth, or throat. Oxymetazoline hydrochloride is a commercially available topical decongestant normally used for over-the-counter treatment of nasal congestion associated with rhinitis. The aerosol spray was used to facilitate placement of the probe with minimal discomfort to the subject.
The nasopharyngeal probe was inserted approximately 9 cm into one nostril such that the tip of the catheter was positioned in the nasopharyngeal region (Fig. 1). Placement of the catheter was verified visually with the use of a fiber-optic flexible nasopharyngoscope. The exterior portion of the catheter was threaded through a port on a Hans Rudolph (Kansas City, MO) facemask containing two one-way non-rebreathing valves; the port was sealed with polytetrafluoroethylene (ptfe) tape to prevent leakage. The exterior portion of the probe outside the facemask was joined with a ptfe transfer line, which fed directly into an ion-trap mass spectrometer (Fig. 2).
Volunteers were instructed to inhale and exhale through the nose. During the experiment, the volunteer was instructed to maintain a regular breathing rhythm under the testing regime of rest and light exercise according to a predetermined protocol outlined in Table 1. Light exercise was achieved by pedaling a Schwinn Model 217p magnetic resistance stationary bicycle at a rate sufficient to achieve a workload of 50 W, as displayed by the bicycle computer. Light exercise was included in the protocol to investigate whether dispositional changes are caused by the exercise state.
The three exposure levels of 1, 5, and 10 ppm were administered in a random fashion for each volunteer to minimize the effect of learning on the results. All exposure levels were below the ACGIH TLV of 10 ppm (ACGIH, 2003). For practical reasons, but also to reduce variability, all experiments were conducted early in the afternoon. Each exposure for a given volunteer was conducted on a separate day.
Real-time breath-analysis system.
The system used to monitor nasopharyngeal vapor concentrations during the human studies has been described elsewhere(Thrall et al., 2003). In brief, a Discovery II (LGC, Inc., Los Gatos, CA) ion-trap tandem mass spectrometer (MS/MS) equipped with an atmospheric sampling glow discharge ionization (ASGDI) source was used to sample directly from the nasopharyngeal probe by connecting the probe to a ptfe tube attached to the inlet orifice of the MS/MS system. Nasopharyngeal-region air was thus sampled at a calibrated flow rate of 12 l/h. This high flow in comparison with the dead volume of the connective tubing minimized the lag in analytical response as much as possible. A new data point was provided every 0.8 s for both 13C1,13C2 vinyl acetate and 13C1,13C2 acetaldehyde. Intensity data from the MS/MS systems were converted to concentration (ppb) using external gas standards prepared in Tedlar bags (Supelco, Inc., Bellefonte, PA) and a calibration curve. New standards and calibration curves were generated each day of operation.
The mass fragmentation of 13C1,13C2 vinyl acetate was evaluated with methane gas introduced through the glow discharge cornea as the collision gas. Blank methane gas produced minor background mass fragments at mass to charge (m/z) ratios 68, 72, 103, 122, and 134. In comparison, 13C1,13C2 vinyl acetate produced major mass fragments at m/z ratios 61 and 91 and a unique fragment at m/z ratio 74. Evaluation of the 13C1,13C2 vinyl acetate standards at the m/z ratio of 74 showed good linearity (R2 = 0.999) from the level of quantification of 3 ppb up to 66,000 ppb. Under the same analytical method, 13C1,13C2 acetaldehyde produced major mass fragments at m/z ratio 61 and 95, and a unique fragment at m/z ratio 63. Evaluation of the 13C1,13C2 acetaldehyde standards at the m/z ratio of 63 showed good linearity (R2 = 1) from the level of quantification of 8 ppb up to 16,000 ppb. In contrast, 12C-acetaldehyde produced major peaks at m/z ratio 61 and 91, and a unique fragment at m/z ratio 59. Nave exhaled breath showed minor endogenous mass fragments at m/z ratios of 59, 61, 62, and 134, which did not correspond to nor interfere with the analyses of any of the 13C-compounds. Analyses indicated that one compound had no impact on the analysis of the other over the range of concentrations tested.
Physiological monitoring.
Breathing amplitude was monitored using a plethysmograph (Qubit Systems, Inc., Kingston, Ontario). It is made up of a breathing monitor belt with an inflatable rubber bladder, worn around the lower chest and upper abdomen and connected to an absolute pressure sensor. The sensor monitored the change in air pressure within the bladder as the lung expanded and contracted, and transmitted a voltage oscillation signal to a portable computer via an interface. Breathing frequency was calculated by dividing the number of breaths by the time, in minutes, over which the breaths were taken.
An acoustic rhinometry system (RhinoMetrics A/S, Lynge, Denmark) was used to measure the cross-sectional area and volume of the nasal cavity and provide a two-dimensional graphic display. The rhinometer measures the nasal airway dimensions by emitting wideband noise into the nose and digitally analyzing the incident and reflected waves recorded by a microphone (Hytnen et al., 1996). The mean of five measurements is displayed as a curve and is updated four times per second. The nasal cavity of each volunteer was measured both before and after application of the nasal spray to evaluate mucosal changes and to validate the volume for the human nasal cavity used in the model.
Preparation for model evaluation.
The human exposures provide 13C1,13C2 vinyl acetate and 13C1,13C2 acetaldehyde concentrations in the nasopharyngeal region during inhalation exposure for model validation. The human nasal PBPK model (Bogdanffy et al., 1999) predicts the corresponding concentrations of vinyl acetate and acetaldehyde in the nasopharyngeal region at given inspired concentrations of vinyl acetate. The rat PBPK model has previously been validated using a constant unidirectional (inhaled) air flow through the nasopharyngeal cavity (Plowchalk et al., 1997). In the experiments described here, the flow, by necessity, is bidirectional; alternatively inhaled and exhaled air passes the probe at the back of the nasal cavity. The experimental set-up did not allow for linking nasopharyngeal sample collection with the inhalation phase of the breathing cycle. Consequently, air concentration measurements were taken every 0.8 s, resulting in recording data during all phases of the breathing cycle in an uncontrolled pattern.
The PBPK model predicts rapid equilibration between the tissue and nasal air phase during vinyl acetate exposure (dotted line in upper panel of Figure 3), with air phase concentrations coming to a constant value in less than 0.01 s. Thus the model estimates essentially reflect steady state concentrations in the nasopharyngeal region (per breath). Due to the rapid kinetics, the experimental design cannot measure the intermediate concentrations in each breath but rather provide a sampling that represents a concentration versus time at various phases of the breathing cycle. This given, the measurements of vinyl acetate concentrations under experimental conditions result in concentration curves as shown in Figure 3. Taking into consideration the experimental factors, the average of peak concentrations during the inhalation phases in a specific test period (e.g., resting, light exercise) were selected as the indicator most closely mirroring the animal validation set-up.
From the plethysmograph data, the number of breaths in a given exposure period (2–5 min) was estimated. The exposure period was then split into a corresponding number of time segments and the peak concentration in each segment was identified. Segments with no clear peak concentration were not included in the analysis. Once the peak concentrations were identified, the mean and standard deviation were calculated for each resting and exercise subgroup of an exposure experiment (see Table 1) for use in the subsequent steps.
Model evaluation.
The previously published human nasal model (Bogdanffy et al., 1999) was exercised at the actual exposure concentrations (Table 2). Briefly, the Bogdanffy model is a PBPK model of the respiratory and olfactory compartments of the human nose. The nasal tissue is split into four compartments (three respiratory and one olfactory), each containing compartments representing the mucus, epithelial cells, basal cells, and submucosa. Inhaled vinyl acetate is extracted into the nasal tissue, diffuses through the various layers, and is subject to metabolism. The model traces the parent and metabolite concentrations in the various compartments, along with the pH change induced by the vinyl acetate and acetaldehyde metabolism. For the current validation, the model predictions of the air concentrations leaving nasal compartment are compared to the experimental data. Full details of the model are presented in Bogdanffy et al. (1999) and the model code is presented in the Supplementary Material online.
Model parameterization was accomplished by means of a combination of measured human values, in vitro extrapolated values, and scaled rat values (Bogdanffy et al., 1999). All of the metabolic and physiological parameters were unchanged from the original published model parameterization (see Supplementary Material online). There were two main reasons for not changing the published model parameterization. The first was that from a strict modeling standpoint, validation is an exercise to determine the quality of the model and model predictions not to make adjustments to the model. Second, the primary application of this model is to risk assessment that models an entire population, not individuals, thus the generic human parameters were retained.
A sensitivity analysis of the model parameters on the output values was also conducted. A prior sensitivity analysis (Bogdanffy et al., 1999) only used intracellular pH as the end point, the current sensitivity analysis was conducted using air phase concentrations as the sensitivity variable. All modeling work was conducted in acslXtreme (version 1.4, AEgis Technologies Group, Huntsville, AL).
Statistical comparison.
To compare the experimental data and the model simulations, a statistical comparison was needed. A statistical evaluation of the goodness of fit of the line describing the simulated maximal vinyl acetate and acetaldehyde concentrations to the mean experimental peak concentrations was conducted. Because the model predicts linear relationships between exposure concentration and both nasopharyngeal vinyl acetate and acetaldehyde concentrations, Pearson's r was used as a statistical measure of the quality of the linear model fit. The resulting statistics described the quality of the model fit and provided a measure of the validation of the model.
RESULTS
The methods development conducted in this project resulted in an analytical method for the detection of 13C1,13C2 acetaldehyde and 13C1,13C2 vinyl acetate in air samples from the nasal cavity with instantaneous readout.. The method is free of interference from endogenous compounds and its detection limit is in the single-digit ppb level. For some volunteers, the acetaldehyde data were rejected because the raw mass/charge ratio intensity decreased during the course of the study resulting in the acetaldehyde intensity signal drifting downward over time. As a result, there are fewer acetaldehyde data sets in the model validation (Table 1).
General Observations
No volunteer reported pain or discomfort associated with the nasopharyngeal probe. All volunteers were tolerant of the lidocaine/oxymetazoline hydrochloride spray, nasal probe, facemask, exercise bicycle, and physiological monitoring equipment. All volunteers reported being able to smell the compound during exposure; no volunteer reported the sweet odor or other aspects of the experiments as unpleasant. The plethysmography results indicate an increase in breathing frequency with increasing exposure level, so that the breathing frequency had doubled at 10 ppm. This reflexive change in breathing frequency is a reaction on the stronger odor at higher exposure levels and also results in more shallow breathing.
Model Evaluation
The complete 13C1,13C2 vinyl acetate and 13C1,13C2 acetaldehyde data sets were treated as described in the methods section. An example of the resulting graph for vinyl acetate is pictured in Figure 3. Data representing peak values are shown in Figures 4 and 5 for vinyl acetate and acetaldehyde, respectively. For each individual volunteer, the data from each exposure sequence (i.e., the three resting and three light exercise periods) were characterized by a mean and standard deviation (Figs. 6 and 7), along with the model simulations. All model parameters used to simulate the experiments, including nasal dimensions, are unchanged from Bogdanffy et al. (1999). The model only simulates the upper respiratory tract. It does not cover the lower respiratory tract and little contribution is expected from the exhaled breath due to expected absorption in this region. This assumption also implies that the bi-directional flow in the human nose will only contain vinyl acetate from inhalation, not from release from the lower respiratory tract.
For both the vinyl acetate and acetaldehyde data sets, nearly all points are close to the model predicted peak concentrations (Pearson's r of 0.9 and 0.6 for VA and AA, respectively). The quality of the acetaldehyde fit is less than the vinyl acetate fit but this is largely due to the small number of experimental data points. A large portion of the remaining variation is not due to limitations in the model but rather to scatter in the experimental data. The vinyl acetate data may suggest a leveling off at concentrations above 9 ppm. However, if this indicated an increased deposition if vinyl acetate, an increase in acetaldehyde levels should also have appeared. This is not seen (Fig. 5) and a better explanation is an increase in breathing frequency, resulting in more difficulty of the analytical set up to accurately measure peak concentrations.
Sensitivity analysis
Because the purpose of this work was to validate the human model of Bogdanffy et al. (1999), model parameters were not adjusted but are the same as presented in the original work. Volunteer body weights varied between 47 kg and 97 kg, but because the model only includes the nasal region instead of the whole body, body weight is not a direct parameter in the model. Individual variation was measured by rhinometer measures of the nasal cavity volume (mean 12.1 ± 2.4 cm3). The default model value was 9.96 cm3. However, a 50% increase or decrease in the nasal cavity volume, and corresponding tissue surface areas and volumes, produced less than a 3% change in the nasopharyngeal vinyl acetate deposition. Sensitivity analysis indicates that the air phase concentrations are most sensitive to the tissue:air partition coefficients, nasal air flow, and the metabolic capacity of the mucus layer (results not shown). Because tissue metabolism of vinyl acetate is not a sensitive model parameter, interindividual variation in metabolic capacity is not expected to provide significant changes in the nasopharyngeal concentrations. Figures 6 and 7 also include model simulations with the inspired air flow at half and twice the default values to show that the majority of the data points fall within the range of data expected by normal breathing variation.
DISCUSSION
The complete set of human exposures resulted in a data set with over 22,000 vinyl acetate and acetaldehyde data points. To aid in the interpretation of this large data set, it was necessary to identify those data that were relevant to the vinyl acetate model predictions. A 1 min sample of the vinyl acetate and data sets is given in Figure 3. As can be seen, the rate of sampling was rapid enough to capture much of the behavior of vinyl acetate in the human nasal cavity, including inhalation and exhalation. The sampling frequency is not short enough, however, to accurately capture the peak concentration in every breath. This experimental limitation resulted in increased variability in the dataset.
The data treatment described above was needed to identify the relevant experimental data—i.e., the peak concentrations—for comparison to the model output. The PBPK model predicts a very rapid equilibration of both vinyl acetate and acetaldehyde in the nasal airway. Because of this, the only data of use for validating the model are the peak concentrations of each breath. Breaths with peak levels far below the peaks seen in most breaths are likely experimental artifacts from volunteer breathing pattern and experimental response time and were not included in the analysis. Samples collected while breathing room air are also shown in Figures 4 through 7 to demonstrate the rapid vinyl acetate and acetaldehyde wash out and low carryover between experimental segments.
The model output is shown in Figures 4 and 5, along with the mean nasopharyngeal concentrations of vinyl acetate and acetaldehyde. The airway concentrations of both compounds appear to be linear as the model predicts. Figures 6 and 7 show the mean peak concentrations by volunteer and resting/exercise state. It can be seen that given the variations in the experimental data there is no significant inter-individual variation and no significant difference in nasopharyngeal concentrations of vinyl acetate and acetaldehyde based on exercise state. Because of this, the standard model parameters, such as ventilation rate and nasal dimensions, were not adjusted for the validation data set. These results demonstrate that the model reasonably predicts the experimental observations with regards to nasal disposition of inhaled vinyl acetate and wash-out of acetaldehyde in a concentration range close to the TLV of 10 ppm.
One inopportune aspect of the study design was the necessary use of the nasal spray containing oxymetazoline hydrochloride and lidocaine to decongest and anesthetize the nasal cavity. Although lidocaine is a short-acting anesthetic, vasoconstriction induced by oxymetazoline hydrochloride can persist for 5 to 6 h. Although the impact of this vasoconstriction nasal uptake efficiencies compared to normal physiological conditions is not known, sensitivity analyses of the rat and human vinyl acetate PBPK model suggest little role, if any, for nasal blood perfusion in the clearance kinetics of vinyl acetate (Bogdanffy et al., 1999). Removal of vinyl acetate diffusing through the epithelial cell layers toward the blood-exchange region through very fast metabolic steps reduces the amount reaching the vasculature substantially (Bogdanffy and Valentine, 2003).
CONCLUSIONS
These experiments present the first attempts to validate a human nasal dosimetry PBPK model. Model predictions of vinyl acetate nasal extraction data compared favorably with measured values obtained using 13C1,13C2 vinyl acetate as did predictions of nasopharyngeal acetaldehyde when compared to measured 13C1,13C2 acetaldehyde. The results show that the current PBPK model structure is appropriate for inhaled vinyl acetate and, by extension, would likely seem appropriate for inhalation of similar organic esters. These analyses were conducted over a range of concentrations relevant to workplace exposure standards. The current ACGIH 8 h time-weighted average workplace exposure standards for vinyl acetate is 10 ppm (ACGIH, 2003). The PBPK model validated here has been used in a risk assessment that suggested these current workplace standards are adequately protective. A risk assessment based on this model further concluded that 24 h per day exposures up to 1 ppm do not present concern regarding cancer or non-cancer toxicity. Validation of the vinyl acetate human PBPK model provides support to these conclusions.
NOTES
The authors certify that all research involving human subjects was done under full compliance with all government policies and the Helsinki Declaration.
ACKNOWLEDGMENTS
The authors wish to thank the members of the CEFIC Acetyls Technical Committee for their review and comment on the study design. This work was funded entirely by the CEFIC Acetyls Sector Group, Brussels, Belgium.
REFERENCES
ACGIH (American Conference of Governmental Industrial Hygienists). (2003). Threshold limit values for chemical substances and physical agents & biological exposure indices. SCGIH, Cincinnati, OH.
Andersen, M. E., Green, T., Frederick, C. B., and Bogdanffy, M. S. (2002). Physiologically based pharmacokinetic (PBPK) models for nasal tissue dosimetry of organic esters: Assessing the state-of-the-knowledge and risk assessment applications with methyl methacrylate and vinyl acetate. Reg. Toxicol. Pharmacol. 36, 234–245.
Andersen, M. E., Sarangapani, R., Frederick, C. B., and Kimbell, J. S. (1999). Dosimetric adjustment factors for methyl methacrylate derived from a steady-state analysis of a physiologically based clearance-extraction model. Inhal. Toxicol. 11, 899–926.
Bogdanffy, M. S., and Valentine, R. (2003). Differentiating between local cytotoxicity, mitogenesis, and genotoxicity in carcinogen risk assessments: the case of vinyl acetate. Toxicol. Lett. 140/141, 83–98.
Bogdanffy, M. S., and Sarangapani, R. (2003). Physiologically-based kinetic modeling of vapours toxic to the respiratory tract. Toxicol. Lett. 138, 103–117.
Bogdanffy, M. S. (2002). Vinyl acetate-induced intracellular acidification: implications for risk assessment. Toxicol. Sci. 66, 320–326.
Bogdanffy, M. S., and Taylor, M. L. (1993). Kinetics of carboxylesterase-mediated hydrolysis of the nasal carcinogen vinyl acetate. Drug Metab. Dispos. 21, 1107–1111.
Bogdanffy, M. S., Sarangapani, R., Plowchalk, D. R., Jarabek, A., and Andersen, M. E. (1999). A biologically based risk assessment for vinyl acetate-induced cancer and noncancer inhalation toxicity. Toxicol. Sci. 51, 19–35.
Bogdanffy, M. S., Dreef-van der Meulen, H. C., Beems, R. B., Feron, V. J., Cascieri, T. C., Tyler, T. R., Vinegar, M. B., and Rickard, R. W. (1994). Chronic toxicity and oncogenicity inhalation study with vinyl acetate in the rat and mouse. Fund. Appl. Tox. 23, 215–229.
Frederick, C. B., Gentry, P. R., Bush, M. L., Lomax, L. G., Black, K. A., Finch, L., Kimbell, J. S., Morgan, K. T., Subramaniam, R., Morris, J. B., and Ultman, J. S. (2001). A hybrid computational fluid dynamics and physiologically based pharmacokinetic model for comparison of predicted tissue concentrations of acrylic acid and other vapors in the rat and human nasal cavities following inhalation exposure. Inhal. Toxicol. 13, 359–376.
Hytnen, M. L., Sala, E. L., Malmberg, H. O., and Nordman, H. (1996). Acoustic rhinometry in the diagnosis of occupational rhinitis. Am. J. Rhinology 10, 393–397.
ICH (ICH Harmonised Tripartite Guideline). (1996). "Guidelines for Good Clinical Practice." International Conference on Harmonisation of Technical Requirements for Registration of Pharmaceuticals for Human Use, International Federation of Pharmaceutical Manufacturers and Associations, Switzerland.
Krishnan, K., Haddad, S., and Pelekis, M. (1995). A simple index for representing the discrepancy between simulations of physiological pharmacokinetic models and experimental data. Toxicol. Indust. Health 11, 413–421.
Lantz, R. C., Orozco J., and Bogdanffy M. S. (2003). Vinyl acetate decreases intracellular pH in rat nasal epithelial cells. Toxicol. Sci. 75, 423–431.
Morita, T. (1995). Low pH leads to sister-chromatid exchanges and chromosomal aberrations, and its clastogenicity is S-dependent. Mutation Res. 334, 301–308.
Morris, J. B., Hassett, D. N., and Blanchard, K. T. (1993). A physiologically based pharmacokinetic model for nasal uptake and metabolism of nonreactive vapors. Toxicol. Appl. Pharmacol. 123, 120–129.
Plowchalk, D. R., Andersen, M. E., and Bogdanffy, M. S. (1997). Physiologically based modeling of vinyl acetate uptake, metabolism, and intracellular pH changes in the rat nasal cavity. Toxicol. Appl. Pharmacol. 142, 386–400.
Thrall, K. D., Schwartz, R. E., Weitz, K. K., Soelberg, J. J., Foureman, G. L., Prah, J. D., and Timchalk, C. (2003). A real-time method to evaluate the nasal deposition and clearance of acetone in the human volunteer. Inhal. Toxicol. 15, 523–538.
Xiao, H., Li, T.-K., Yang, J.-M., and Liu, L. F. (2003). Acidic pH induces topoisomerase II-mediated DNA damage. Proc. Natl. Acad. Sci. U.S.A. 100, 5205–5210.
WHO (World Health Organization). (1995). Guidelines for Good Clinical practice (GCP) for trials on pharmaceutical products. Who Technical Report series, No. 850, Annex 3.(P. M. Hinderliter, K. D. )