Application of a Hybrid CFD-PBPK Nasal Dosimetry Model in an Inhalation Risk Assessment: An Example with Acrylic Acid

Melvin Andersen*, Ramesh Sarangapani{dagger}, Robinan Gentry{ddagger},1, Harvey Clewell{ddagger}, Tammie Covington{ddagger} and Clay B. Frederick§

* Colorado State University, CETT, Foothills Campus, Fort Collins, Colorado 80523; {dagger} The K. S. Crump Group, Inc., ICF Consulting, Research Triangle Park, North Carolina 27709; {ddagger} The K. S. Crump Group, Inc., ICF Consulting, Ruston, Louisiana 71270; and § Rohm & Haas Co., Spring House, Pennsylvania 19477

Received March 7, 2000; accepted June 28, 2000


    ABSTRACT
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
The available inhalation toxicity information for acrylic acid (AA) suggests that lesions to the nasal cavity, specifically olfactory degeneration, are the most sensitive end point for developing a reference concentration (RfC). Advances in physiologically based pharmacokinetic (PBPK) modeling, specifically the incorporation of computational fluid dynamic (CFD) models, now make it possible to estimate the flux of inhaled chemicals within the nasal cavity of experimental species, specifically rats. The focus of this investigation was to apply an existing CFD-PBPK hybrid model in the estimation of an RfC to determine the impact of incorporation of this new modeling technique into the risk assessment process. Information provided in the literature on the toxicity and mode of action for AA was used to determine the risk assessment approach. A comparison of the approach used for the current U.S. Environmental Protection Agency (U.S. EPA) RfC with the approach using the CFD-PBPK hybrid model was also conducted. The application of the CFD-PBPK hybrid model in a risk assessment for AA resulted in an RfC of 79 ppb, assuming a minute ventilation of 13.8 l/min (20 m3/day) in humans. This value differs substantially from the RfC of 0.37 ppb estimated for AA by the U.S. EPA before the PBPK modeling advances became available. The difference in these two RfCs arises from many factors, with the main difference being the species selected (mouse vs. rat). The choice to conduct the evaluation using the rat was based on the availability of dosimetry data in this species. Once these data are available in the mouse, an assessment should be conducted using this information. Additional differences included the methods used for estimating the target tissue concentration, the uncertainty factors (UFs) applied, and the application of duration and uncertainty adjustments to the internal target tissue dose rather than the external exposure concentration.

Key Words: acrylic acid toxicity; nasal inhalation; olfactory epithelium degeneration; interspecies dosimetry; rat; reference concentration; RfC.


    INTRODUCTION
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Acrylic acid (AA) is a clear, colorless, corrosive liquid used in the production of acrylic esters and polymers, with application in paints, coatings, and plastics. In studies conducted in laboratory animals exposed to AA via inhalation, lesions of the olfactory epithelium on the dorsomedial aspect of the nasal passage are the most sensitive end point that has been observed (Miller et al., 1979Go, 1981aGo; Nachreiner and Dodd, 1989Go). Similar lesions confined to the anterior region of the olfactory epithelium have also been observed in laboratory animals exposed to acrylate esters via inhalation (Miller et al., 1985Go; Reininghaus et al., 1991Go). The major route of metabolism of acrylate esters is by carboxylesterase-mediated hydrolysis, resulting in the rapid formation of AA and the corresponding alcohol in the nose (Frederick et al., 1994Go; McCarthy and Witz, 1997Go; Miller et al., 1981bGo; Sanders et al., 1988Go; Silver and Murphy, 1981Go). Hence, exposure to acrylate esters will lead to significant internal exposure to AA. The mechanism of action for the nasal toxicity of AA or the acrylate esters is unknown; however, a number of plausible mechanisms have been suggested, ranging from inhibition of mitochondrial metabolism by AA to the inherent irritant properties of the acid (Brass, 1994Go; Custodio et al., 1998Go; Keenan et al., 1990Go; Trela and Bogdanffy, 1991Go).

Before the early 1980s, the histological evaluation of the nose in inhalation toxicity studies was usually based on taking a single section or only a very limited number of sections. Work with formaldehyde in the early 1980s increased interest in nasal toxicity, and more comprehensive pathological evaluations of nasal tissues became commonplace. With these changes in standard pathology practices, a large number of compounds have been found to cause selective damage to the olfactory epithelium in rodents, e.g., styrene (Cruzan et al., 1997Go), methylmethacrylate (Lomax et al., 1997Go), vinyl acetate (Bogdanffy et al., 1997Go), chloroform (Mery et al., 1994Go), and ethyl acrylate (Miller et al., 1985Go). Among this large set of compounds are organic acids, organic esters, and several hydrocarbons and halocarbons that are metabolized by olfactory epithelial tissues. Increasingly, the RfCs for these compounds have been established using nasal effects in rodents as the critical end point for the evaluations.

Over the past several years, advances have occurred in several areas that could impact a risk assessment for a chemical that causes lesions of the nasal cavity. These include advances in histological reconstruction of the distribution of the squamous, olfactory, and respiratory epithelia lining the nasal cavity, in mechanistic understanding of the histopathology observed in vivo and in vitro with nasal tissue, and in computational fluid dynamic (CFD) models for airflow through rat and human nasal cavities. The nasal cavity has a complex anatomy and the nasal mucosa consists of a variety of cell types, each having a different activity towards the inhaled chemical. Hybrid CFD-PBPK modeling approaches provide an effective and accurate means to arrive at exposure-tissue dose measures in complex systems such as the nasal cavity.

These advances now provide a basis for developing more complex approaches for risk assessment by these compounds. The primary focus of this paper is the application of a hybrid CFD-PBPK model developed by Frederick et al. (1998) in the estimation of an RfC for AA. In addition to providing a brief description of the model, this article also provides a detailed evaluation of the model behavior and a brief overview of the toxicity and proposed mode of action for AA-induced toxicity. A comparison of the RfC developed using the CFD-PBPK model and toxicity data in the rat, with the current RfC based on end points in the mouse published by U.S. EPA (1998), is also presented.


    MATERIALS AND METHODS
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Description of the hybrid CFD-PBPK model.
The hybrid CFD-PBPK model estimates the tissue concentration of AA in various regions of the nasal cavity for a wide range of exposure scenarios. Frederick et al. (1998) presented the set of mass balance equations used to model the various transport processes of AA and acrylate in the nasal cavity. The various model parameters were either obtained from the available literature or were estimated from independent studies (Frederick et al., 1998Go, 2000Go) (Table 1Go). The model computes total tissue concentration (i.e., acid plus acrylate) under different exposure conditions in rats, mice, and humans. The model was coded using Advanced Continuous Simulation Language (ACSL) for Windows 95 (Mitchell and Gauthier Associates, Concord, MA /Pharsight, Palo Alto, CA) in such a way as to simulate both unidirectional and cyclic breathing conditions (Frederick et al., 1998Go). This model also simulates systemic distribution of AA by incorporating components such as the lung, liver, richly perfused tissue, poorly perfused tissue, and venous and arterial blood compartments into a generic whole-body PBPK model. A detailed description of the model is provided in Frederick et al. (1998).


View this table:
[in this window]
[in a new window]
 
TABLE 1 CFD-PBPK Model Parameter in Rat, Mouse, and Human
 
The nasal cavity forms the portal of entry for inhaled chemicals in both rodents and humans. The nasal cavity has a complex anatomy, with the inspired air following distinct paths, resulting in asymmetric ventilation to various regions of the nasal cavity. To describe the movement of inhaled AA in the nasal passage of the F344 rat, the nasal passage is partitioned into several regions in the model (Fig. 1Go). The anterior-most region of the nasal cavity is the nasal vestibule, which is lined with a squamous epithelium. The inspired airflow divides into multiple streams at the posterior end of the nasal vestibule. CFD models make it possible to aggregate the airflow data into two distinct pathways, a dorsal/medial pathway and a lateral/ventral pathway, in the nasal cavity of rats as well as humans (Kimbell et al., 1997Go; Subramaniam et al., 1998Go). The two air streams merge at the nasopharyngeal region before entering the lower respiratory tract. The compartmentalization of the nasal cavity in the hybrid CFD-PBPK model preserves this inherent division in the airflow into two distinct pathways.



View larger version (51K):
[in this window]
[in a new window]
 
FIG. 1. Schematic of the multicompartment description of the nasal cavity used in the hybrid CFD-PBPK model to compute tissue dose from acrylic acid exposure.

 
In the rat, lining the surface along the dorsal air stream is an anterior region of respiratory epithelium. Immediately posterior to this small region of respiratory epithelium is a protruding finger of olfactory epithelium that lines the dorsal meatus. Posterior to this finger is the entire ethmoid turbinate region lined with olfactory epithelium. The ventral side consists of the remaining portions of the rat nasal passage lined with respiratory epithelium. For the nasal cavity description in the model, the large olfactory region in rodents (approximately 50% of the total nasal surface area) was divided into a small anterior portion to accommodate the protruding finger of olfactory epithelium lining the dorsal meatus and a larger posterior compartment. The ventrally located respiratory mucosa was also divided into an anterior and a posterior compartment of almost equal dimensions. The human nasal cavity does not have an anatomical equivalent for the rat ethmoid turbinate; therefore, the relatively small region of olfactory mucosa in humans (approximately 4% of the total nasal surface area) was described by a single compartment.

AA vapor in the air dissociates into the acrylate anion and a proton upon dissolving in the nasal mucus. The pKa for acrylic acid is 4.25 (Weast and Astle, 1978Go), with approximately 99% dissociation expected at physiological pH. Once absorbed in the nasal cavity at the air-mucus interface, where it ionizes, acrylic acid diffuses into the tissue and is eliminated either by metabolism or by venous efflux. At steady state, the net flux of AA from the lumen into the tissue will equal the total elimination rate of AA due to metabolism in the epithelial layer and blood perfusion in the submucosal layer. To characterize this movement of AA across the lumen, mucus, and tissue layers of the nasal cavity, a model substructure similar to the one proposed by Morris et al. (1993) was incorporated by Frederick et al. (1998) into each nasal compartment of the CFD-PBPK hybrid model (Fig. 2Go). In this model substructure, each tissue stack is divided into multiple subcompartments to represent the mucus, epithelial tissue, or submucosal tissue.



View larger version (80K):
[in this window]
[in a new window]
 
FIG. 2. Schematic of a typical unit compartment in the nasal cavity showing the lumen, mucus, epithelial, and submucosal layer.

 
Evaluation of model behavior.
Although the hybrid CFD-PBPK model developed by Frederick et al. (1998) has been previously described, the model has not been fully evaluated with regard to its behavior. Three separate analyses were conducted to evaluate the behavior of the model, as will be discussed below. Before these analyses could be conducted, the toxicity and mode of action information for AA in the available literature was reviewed to identify the appropriate dose metric to be used during the evaluation of model behavior, as well as for the development of the RfC. This literature review also served to identify the critical end point for development of the RfC.

The CFD-PBPK model simulates both unidirectional and cyclic breathing conditions. One exercise was conducted to evaluate the impact of using unidirectional versus cyclic breathing on the dose metrics estimated by the model. A periodic function is used in the model to describe the flow rate during cyclic breathing. This periodic function imposes on the cyclic model a steady and uniform inward flow during inhalation and an equal and opposite outward flow during exhalation. For rats the average breathing time is about 0.4 s and it takes around 500 s for the tissue concentration to reach steady state. These two factors combined make running the cyclic model highly time consuming. Furthermore, to conserve the total mass of inhaled AA in simulations of physiological minute volumes, the inhalation flow rate in the cyclic model is set to twice the flow rate in the unidirectional model. Simulations were run, using either cyclic breathing or unidirectional breathing functions, to evaluate the impact on the estimation of tissue doses by the model using either breathing pattern.

The dose to the target site is a result of a number of competing mass transport processes. Sensitivity analysis is a tool to explore the model behavior under various exposure conditions. For a given exposure, a sensitivity analysis is conducted by measuring the fractional change in the relevant dose metric resulting from a specified change in a particular model input parameter while all the other parameters are held fixed. The ratio of the fractional change in the tissue dose to the fractional change in the input is called a sensitivity coefficient (SC). Once the appropriate tissue dose metric was determined based on the available toxicity and mode of action information, a sensitivity analysis was conducted to identify those parameters that were most critical to the model estimates.

The various model parameters are not known with absolute certainty. The uncertainty in knowledge of these parameters leads to uncertainty in the estimates of the dose metrics obtained from exercising the model. An evaluation of this uncertainty in the model-derived dose estimates can be obtained using a Monte Carlo analysis. Such an analysis is useful for a human health risk assessment, as it can help support reasonable UFs. An uncertainty analysis was conducted by evaluating the distribution in the model-derived dose metric using a Monte Carlo procedure to sample from prespecified distributions for the input model parameters. The uncertainty in the input parameters was described using normal or lognormal distributions. The Monte Carlo algorithm was then used to randomly select a value for each parameter from its distribution, and the model was run to predict an instance of the dose metric. This random selection of parameter values and running of the model was repeated a large number of times (1000 in this case) until the distribution of the dose metric was characterized.


    RESULTS
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Determination of the Relevant Dose Metric for RfC Calculations
In order to apply the hybrid model in a risk assessment application, the existing toxicity literature was reviewed to identify the end point that would form the basis for the RfC. In addition, data on the mode of action of AA was reviewed in order to identify the most relevant dose metric(s) to serve as the basis for an RfC, as well as for evaluating the behavior of the model as it relates to the estimation of an RfC.

The majority of studies that have been conducted in which laboratory animals were exposed to AA via inhalation were for duration of 2 weeks or less (Cascieri and Clary, 1993Go; Gage, 1970Go; Miller et al., 1979Go; Nachreiner and Dodd, 1989Go). In the only subchronic study that has been conducted, mice and rats were exposed to AA via inhalation for 13 weeks (Miller et al., 1981aGo). Statistically significant increases in the incidence of slight focal degeneration of the olfactory epithelium were observed in male and female rats exposed to 75 ppm AA, compared to controls. A NOAEL of 25 ppm was established for rats (Table 2Go). Olfactory lesions were also reported in female mice at 5, 25, and 75 ppm, as well as in male mice at 25 and 75 ppm. These lesions were categorized as very slight to slight focal degeneration of the olfactory epithelium, with no evidence of an inflammatory response in mice exposed to 5 ppm or in most of the animals exposed to 25 ppm. No other treatment-related histopathological changes were reported. A NOAEL for mice was not obtained in this study. This subchronic inhalation study conducted by Miller et al. (1981a) is the critical study for the derivation of an RfC for AA, with the nasal lesions being the critical end point.


View this table:
[in this window]
[in a new window]
 
TABLE 2 Acrylic Acid 13-Week Vapor Inhalation Study. Histopathologic Observations in the Nasal Mucosa of Rats and Mice
 
In order to define and provide support for a dose metric for the risk assessment application of the CFD-PBPK hybrid model, the literature was reviewed to determine the mode of action of AA-induced nasal toxicity. The specific mechanism of action for the nasal toxicity of AA is not known; however, research has been conducted that provides information relevant to determining a mode of action for the nasal lesions observed following inhalation exposure to AA. Several modes of action for the cytotoxicity of AA are possible, including interference with the normal function of mitochondria (Custodio et al., 1998Go), metabolism of AA in the mitochondria to a more reactive or toxic substance (Brass, 1994Go), and direct irritant potential of the accumulated acid (Keenan et al., 1990Go; Trela and Bogdanffy, 1991Go). The results from the majority of studies with a wide range of acids and esters indicate that the most likely mode of action for the degenerative lesions observed in the olfactory region of laboratory animals following inhalation exposure to AA is a response to the irritant property of the acid itself. Toxicity following exposure to AA by oral, inhalation, or dermal exposure is limited to effects related to its irritant properties, until exposure reaches high concentrations. The relatively low systemic toxicity of AA is probably due to its rapid systemic metabolism and elimination. However, at high exposure concentrations, normal detoxification or clearance mechanisms (locally, systemically, or both) may become saturated, resulting in a build up of the acid. This is supported by the results observed in the available toxicity studies. In the acute and subacute studies, degenerative lesions of the liver and kidney, the major metabolic tissues for AA, were observed only following exposure to high inhalation concentrations (Gage, 1970Go). Similarly, focal olfactory degeneration observed in the nasal mucosa was slight to moderate at lower concentrations and was reversible following cessation of exposure (Lomax et al., 1994Go). However, at high concentrations, the irritant-related effects observed were significantly more severe, with the replacement of olfactory epithelium by respiratory epithelium. It is possible that all of the proposed mechanisms play a role in the cytotoxicity observed in the nasal mucosa following inhalation of AA. Regardless of the specifics of the mechanism, however, the currently available literature does provide general support for the use of a dose metric for the nasal toxicity that is based on the concentration of AA in the tissue.

Based on the information provided by the toxicity studies, inhalation exposure to AA in rats and mice resulted in lesions confined to the olfactory mucosa (Miller et al., 1981aGo). Hence the olfactory epithelium in the anterior dorsal compartment in the model was the primary target tissue, and the total AA concentration in the epithelial tissue in this compartment was the relevant dose metric. To further evaluate the selection of this dose metric (concentration of the acid in the olfactory epithelium) for the development of an RfC, the model-derived concentrations of AA in the epithelial layer of the olfactory 1 compartment for in vivo exposures in the rat were compared with concentrations of AA associated with cytotoxicity in the in vitro rat nasal explant studies (Frederick et al., 1998Go). Short-term organ culture of nasal explants from rats with media containing AA resulted in loss of sustentacular and neuronal cells at an AA concentration of 6 mM in the culture medium. For an inhaled AA concentration of 75 ppm, the LOAEL in rats, the hybrid CFD-PBPK model predicts a steady-state average epithelial tissue concentration of 5.49 mM in the dorsal olfactory 1 region and a concentration of 6.47 mM in the top epithelial layer, under cyclic flow conditions. These model-predicted –tissue concentrations of AA compared well with concentrations at which cytotoxicity was observed in the in vitro nasal explant studies. This agreement provides support for using the AA concentration in the epithelial layer of the anterior olfactory compartment as the appropriate dose metric for the development of an RfC and for evaluation of the model behavior.

Evaluation of Model Behavior
As a result of the description of mucus buffering and pH used in the hybrid model, at low inhaled AA concentrations (below 0.1 ppm), estimated mucus pH remains close to physiological values, leading to a high ratio of ionized to nonionized AA, which results in a very large effective partitioning from air to mucus (see Discussion). After simulation of inhalation of higher AA concentrations, the mucus pH was reduced, and thus the effective partitioning between mucus and air was lower. In regions where the mucus pH was relatively constant as a function of inhaled AA concentration, the model behaved linearly (Fig. 3Go). Thus, linear changes in the relevant tissue dose (olfactory epithelium concentration) with changes in inhaled AA concentration were observed with the model, both following high inhalation exposures, where the mucus pH is determined by the pKa of AA, and after low exposures, where the buffering capacity of the mucus layer was not exceeded. For the intermediate exposures, mucus pH varied significantly, and the model behaved in a nonlinear fashion. The transition between these two extremes was predicted by the model to occur in the range of concentrations for the animal bioassays. Following these intermediate exposures, when the mucus pH was predicted to change rapidly, the olfactory epithelium tissue dose was a nonlinear function of the inhaled AA concentration. This transition between the two linear regions is demonstrated by the slight curvature in the log-log plot for the olfactory epithelium tissue dose versus inhaled concentration at intermediate exposures in the rat (Fig. 4Go, solid line). The dotted line in Figure 4Go displays the model-predicted dose response in the human at a ventilation rate representative of light exercise (20 l/min). Note that, in contrast to the above discussion of the rat results, the human dose response is essentially linear under these conditions.



View larger version (10K):
[in this window]
[in a new window]
 
FIG. 3. Plot showing steady-state mucus pH in the dorsal respiratory and olfactory 1 compartment as a function of inhaled acrylic acid concentration in a rat model under unidirectional flow conditions.

 


View larger version (11K):
[in this window]
[in a new window]
 
FIG. 4. Plot comparing the tissue dose in the olfactory 1 epithelial layer in rats (solid line) and humans (dotted line) for a range of inhaled acrylic acid concentrations.

 
Cyclic versus Unidirectional Breathing Patterns
Figure 5Go compares the model-derived target tissue dose in rats during cyclic breathing and unidirectional breathing. The cyclic model predicted tissue doses that were about 10% lower than the unidirectional model predictions at higher AA exposures. At lower concentrations, the cyclic model-derived tissue doses were slightly greater than the unidirectional model predictions. At intermediate concentrations, the two different breathing patterns resulted in similar tissue doses.



View larger version (11K):
[in this window]
[in a new window]
 
FIG. 5. Plot showing steady-state acrylic acid concentration in the epithelial layer of the olfactory 1 compartment as a function of inhaled acrylic acid concentration for unidirectional and cyclic breathing conditions in the rat model.

 
Sensitivity Analysis
The sensitivity analysis was conducted for rats by measuring the percent change in the total AA concentration in the dorsal anterior olfactory epithelial tissue compartment for a 1% change in each model input parameter (Table 3Go). Table 3Go does not report SCs that are less than or equal to 0.01 in absolute value. A positive value for the SC indicates a direct correlation between the tissue dose and the corresponding model parameter. Similarly, a negative value for the SC indicates an inverse correlation between dose and model parameter.


View this table:
[in this window]
[in a new window]
 
TABLE 3 Sensitivity Coefficients for the Rat Model at Various Inhaled Acrylic Acid Concentrations
 
The SCs were calculated for six different exposures, two each at the extremes of behavior and two concentrations in the transition region. In the lower exposure range, a 10-fold decrease in the inhaled concentration (0.2 ppm to 0.02 ppm) changes the SCs only marginally, implying that the system behaves linearly in this exposure range. Similarly, at the higher exposure range, a 10-fold increase in the inhaled concentration (35 ppm to 350 ppm) changes the SCs only marginally, once again signifying a linear system behavior. The SCs vary significantly with inhaled concentration at intermediate exposure (2 to 35 ppm).

As mentioned previously, at low AA exposure (i.e., AA concentration < 0.2 ppm) mucus pH was relatively unaffected, and AA extracted from the airstream was predicted to be essentially completely ionized in the mucus. At 0.2 and 0.02 ppm inhaled AA concentration, the SCs were comparable. At these concentrations, the target tissue dose was primarily sensitive to airflow and blood flow parameters, as well as to diffusivity and anatomical parameters (surface area and thickness), primarily those associated with the olfactory compartment. Following simulations of high AA exposure (i.e., AA exposure > 35 ppm), the mucus pH asymptotically approached its lower limit and a relatively larger fraction of the extracted AA in the mucus was estimated to be nonionized. At these concentrations, the most sensitive parameters were the initial mucus pH and the pKa of AA. All the other model parameters showed much less sensitivity following simulations of higher AA exposure. In the intermediate exposure range, parameters showed variable sensitivity, as listed in Table 3Go.

Uncertainty Analysis
The output chosen for the uncertainty analysis was the steady-state concentration of AA in the epithelial tissue of the anterior olfactory region for an intermediate exposure concentration (35 ppm). The mean, range, and coefficient of variation (standard deviation divided by mean) for each parameter distribution used in the Monte Carlo analysis are listed in Table 4Go. Normal distributions were used for all parameters except the diffusivity constants and mass transport coefficients, for which lognormal distributions were used. All distributions were truncated at the limits of the ranges shown in Table 4Go.


View this table:
[in this window]
[in a new window]
 
TABLE 4 List of Rat Model Parameters Used in the Variability Analysis Showing the Preferred Values, Range, and Coefficient of Variation (CV)
 
The percentiles of the resulting distribution of target tissue concentrations are shown in Table 5Go. The coefficient of variation for the distribution was 0.37 and the 5th and 95th percentiles of the distribution were within roughly a factor of two of the median. Thus, in spite of the many parameters in the model that must be estimated, and the uncertainty regarding the values of those parameters, the predictions of the model are relatively stable. The narrow distribution of tissue concentration predictions provides increased confidence in the use of the model to support quantitative dose-response calculations.


View this table:
[in this window]
[in a new window]
 
TABLE 5 Distribution of Target Tissue Concentrations (CDOE11) from Monte Carlo Analysis
 
Development of a Model-Based Reference Concentration (RfC)
Inhalation toxicity studies in rats and mice have clearly identified the olfactory epithelium as the critical effect for establishing an exposure standard (Miller et al., 1981aGo). Although the mouse appeared to be more sensitive to these olfactory effects, CFD dosimetry modeling to date has focused on airflow and uptake characteristics in nasal casts from rats and humans. The strategy used for the development of the hybrid model-based RfC uses the rat as the basis for the animal:human dosimetry comparisons, due to the lack of dosimetric information for the mouse. Once the appropriate mouse data become available, a similar analysis could be conducted for the mouse.

Duration Adjustment and Dose Adjustment
The mode of action of AA toxicity appears to be associated with the accumulation of the acid in target tissues. The nasal lesions observed following inhalation exposure to AA are observed in the anterior-most region of the olfactory epithelium. The CFD-PBPK model for AA was used to calculate two dose metrics for the risk assessment: the concentration of AA in the uppermost layer of the olfactory epithelium in the anterior-most section (CDOE-11), and an average AA concentration in all of the olfactory epithelial layers in this section (CDOE-AVG). The rat NOAEL was used as the point of departure, i.e., 25 ppm for a subchronic 90-day study with 6 h/day, 5 days/week exposures (Miller et al., 1981aGo). This exposure concentration was converted to a tissue concentration of AA in the rat using a cyclic airflow pattern in the CFD-PBPK model for the two dose metrics (Table 6Go). At an exposure of 25 ppm, the rat model predicts a steady-state tissue concentration (CDOE-11) of 3.0 mM. In our tissue dosimetry-based approach for estimating the RfC, this tissue dose was then directly adjusted to continuous, 24-h/day exposures, the exposure scenario of concern in the human population. This adjustment was performed by multiplying the tissue dose by the two ratios 6/24 and 5/7, which relate the toxicity test exposures to a chronic exposure in humans. Using this adjustment, the duration-adjusted tissue concentration would be 0.5 mM.


View this table:
[in this window]
[in a new window]
 
TABLE 6 Target Tissue Dose (µgm/ml) in the Mouse, Rat, and Human Model Computed at the NOAEL (25 ppm) and Duration-Adjusted NOAEL (4.464 ppm) under Cyclic Flow Conditions
 
Uncertainty Factors
The duration-adjusted dose metric obtained from the toxicity studies was next adjusted by appropriate uncertainty factors (UF) before calculating the human equivalent concentration. The UFs considered included interspecies variability, intraspecies variability, subchronic/chronic extrapolation, and modifying factors. The usual interspecies factor of 10 was reduced to {surd}10, as the CFD-PBPK model accounts for dosimetry differences between species (U.S. EPA, 1994Go). The default UF of 10 for intraspecies variability, which comprises a factor of {surd}10 for intraspecies differences in pharmacokinetics and a factor of {surd}10 for intraspecies differences in pharmacodynamics (Jarabek, 1994Go), was used for this analysis.

A subchronic to chronic factor was considered because the duration of the study on which the assessment is based was only 90 days (Miller et al., 1981aGo; Miller et al., 1985Go; Reininghaus et al., 1991Go). However, chronic studies with acrylate esters can be used to examine the necessity for this factor. These esters also exert effects on olfactory epithelium due to metabolism to AA (Frederick et al., 1994Go; Morris and Frederick, 1995Go; Stott and McKenna, 1985Go). In chronic studies conducted with ethyl, methyl, or butyl acrylate, nasal lesions virtually identical to those observed following inhalation exposure to AA were observed following initial interim sacrifices (after 3–12 months of exposure). No appreciable change in the extent or severity of the lesions was observed as the studies progressed. The results from these studies indicate that a UF for subchronic to chronic duration of exposure is unnecessary.

The combination of these UFs results in a total uncertainty adjustment of 30, in comparison to the UF of 300 used by the U.S. EPA. The composite for the PBPK-based approach consists of a factor of {surd}10 for interspecies pharmacodynamics and a factor of 10 for human variability (intraspecies). The factor of 300 used by the U.S. EPA was applied to a LOAEL for nasal lesions in the mouse and is based on a factor of 10 for intraspecies variability, a factor of 3 for extrapolation from subchronic to chronic duration due to the limited progression between short-term and subchronic exposures and due to rapid metabolism, and a factor of 10 to account for both interspecies extrapolation, because dosimetric adjustments were applied, and use of a LOAEL (in the mouse) because the effect was considered mild.

RfC Calculation
The duration-adjusted tissue dose from the animal model was adjusted by the UF of 30 to result in the target tissue dose. The human model was then run at various external exposure concentrations to determine the inhaled AA concentration that resulted in the same target tissue dose in humans at steady state. Using the rat model, the internal dose metric (i.e., olfactory 1 epithelial AA tissue concentration) for our RfC calculation is 3.0 mM at the NOAEL of 25 ppm under cyclic flow condition. Adjusting this internal dose for duration (i.e., multiplying by 6/24 and 5/7) and dividing by the UF of 30 results in a target tissue dose of 0.018 mM. Using the human PBPK model, we determined that this target tissue dose would result from an inhaled concentration of 0.079 ppm. Thus, using the pharmacokinetic model in the risk assessment resulted in an RfC of 79 ppb for AA. This estimate assumed a human ventilation rate of 13.9 l/min, based on the U.S. EPA's standard breathing rate of 20 m3/day (U.S. EPA, 1998Go).

The U.S. EPA's default approach (U.S. EPA, 1994Go) applies the duration adjustment to the external exposure concentration level that is the NOAEL. Next, a regional gas dose ratio (RGDR) is calculated to derive a human equivalent concentration (HEC); UFs are then applied to obtain the RfC. The RGDR applied as an interspecies adjustment is technically the ratio of rat to human target tissue dose at the duration-adjusted NOAEL value. Using this approach, the ratio of rat to human doses for the duration-adjusted NOAEL (CDOE-AVG concentration at 4.464 ppm from Table 6Go) results in a RGDR of 1.38 for cyclic flow conditions. Applying this RGDR value and an UF of 30 on the duration-adjusted NOAEL of 4.464 ppm results in an alternative RfC of 205 ppb.

In a third possible approach for incorporation of the PK model into the standard U.S. EPA (1994) RfC methodology, the animal model is run at the duration-adjusted NOAEL (4.464 ppm) to estimate the target tissue dose associated with the NOAEL. The human model is run to determine an external exposure concentration associated with the NOAEL tissue dose metric. Once the external exposure or human equivalent concentration is determined, the UFs are applied to the human equivalent concentration. This method results in an RfC of 233 ppb, which is again less conservative than the consistent tissue-dose approach.

The differences between the three approaches for calculating the RfC that incorporate the PK model are primarily where the duration adjustment is applied, and where the UFs are applied. The default U.S. EPA approach assumes that tissue dose and inhaled concentrations are linearly related. The exposure-tissue dose relationship in the human is practically linear (Fig. 4Go); therefore, differences in where the UFs are applied are not the reason for the differences in the RfCs estimated by the different approaches. However, nonlinearity in the exposure-tissue dose relationship in the experimental animal (Fig. 4Go) results in differences in the estimated human equivalent concentration, depending on whether the duration adjustment is applied to the external concentration or the internal tissue-dose metric.


    DISCUSSION
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Estimation of an RfC Using the Hybrid Model
The U.S. EPA's RfC methodology (U.S. EPA, 1994Go) provides default procedures for estimating RfCs. The first step in developing an RfC is analyzing the results from available subchronic/chronic inhalation toxicity studies conducted in various mammalian species to arrive at a highest exposure level that does not result in adverse effects (NOAEL) below the lowest exposure level that is associated with adverse effects (LOAEL). Once this exposure level is determined, it becomes the basis for deriving an RfC. Because inhalation toxicity studies on laboratory animals are conducted with discontinuous exposure regimes (typically 6–8 h/day, 5 days/week) and an RfC, by definition, is derived for a continuous exposure, the experimentally assessed NOAEL/LOAEL must be adjusted for the duration of exposure. The basis of this correction is Haber's Law, which relates the product of exposure concentration and time to a toxic effect (Lomax et al., 1994Go).

The critical end point observed in the available toxicity studies for AA is toxicity in the olfactory epithelium of the nasal cavity in both rats and mice (Miller et al. 1981aGo). In the derivation of an RfC, it must be considered that various species used in inhalation toxicology studies do not receive identical doses in comparable respiratory tract regions under similar exposure conditions. To adjust for differences in dose, dosimetry defaults are applied by the U.S. EPA (1994) based on the region of the respiratory tract affected—the upper respiratory tract, the tracheobronchial region, or the alveolar region. The standard factors included in these dosimetry adjustments are breathing rate, surface areas of specific regions of the respiratory tract, and the blood:air partition coefficients of the inhaled gas or vapor. For soluble vapors affecting the upper respiratory tract, the default used is the ratio of the regional flux to the corresponding surface area at risk. The comparison of this ratio between species is the RGDR. This correction is applied to the duration-adjusted LOAEL or NOAEL in the animal study to arrive at an HEC as a starting point for considering other UFs in deriving the RfC.

Among the UFs considered, there is a correction for interspecies differences. A factor of 10 is usually used for animal to human differences. This factor is now regarded to consist of both pharmacokinetic and pharmacodynamic components (Jarabek, 1994Go) in a manner similar to that suggested by Renwick (1993). To account for intraspecies variability and to accommodate sensitive subpopulations, a UF of 10 is typically used. A UF of 10 is also typically used when a NOAEL is derived using a less-than-chronic animal study or when using a LOAEL instead of a NOAEL to derive the RfC. These defaults are applied in the absence of compound- or species-specific information on dose delivered to target tissues of the respiratory tract or the mode of action by which the compounds cause toxicity. The RfC documentation indicates that these defaults should be replaced by compound- and species-specific information when they are available (U.S. EPA, 1994Go).

The U.S. EPA (1998) has derived an RfC for AA based on the incidence of olfactory degeneration in female mice reported by Miller et al. (1981a). A LOAEL of 5 ppm AA was used, with no NOAEL identified. The LOAEL was adjusted by the duration of exposure (6 h/24 h x 5 days/7 days) resulting in a duration-adjusted LOAEL(ADJ) of 0.89 ppm AA. An HEC was derived from the duration-adjusted animal LOAEL by applying an RGDR. The current guidelines for deriving the RGDR are based on a gas categorization scheme (U.S. EPA, 1994Go). This scheme uses water solubility and reactivity of the chemical to categorize gases into three categories. The U.S. EPA (1998) has evaluated AA as a Category 1 gas with an RGDR of 0.122. Category 1 gases are highly water soluble and/or rapidly and irreversibly reactive in the surface liquid/tissue of the respiratory tract. These chemicals are anticipated to be extracted mainly in the extrathoracic region.

The duration-adjusted LOAEL of 0.89 ppm is then multiplied by the RGDRET of 0.122 to give HEC based on a LOAEL (LOAEL(HEC)) of 0.11 ppm. The LOAEL(HEC) is then multiplied by uncertainty and modifying factors to derive the RfC. For AA, the U.S. EPA (1998) has used a UF of 300. This composite UF consists of a factor of 10 for human variability (intraspecies), a factor of 10 for interspecies extrapolation, and a factor of 3 for extrapolation from subchronic to chronic duration. For the subchronic to chronic duration extrapolation, a factor of 3 rather than the default of 10 was used due to both the limited progression of lesions between short-term and subchronic exposures and the rapid metabolism of AA. The resulting RfC for AA using the U.S. EPA approach was 0.37 ppb.

The application of the CFD-PBPK hybrid model leads to a significant change from the RfC of 0.37 ppb calculated by U.S. EPA to 79 ppb using the CFD-PBPK modeling approach. This 213-fold difference arises due to the following reasons:

When incorporating a PK model into the standard risk assessment approach, it is generally recommended that the duration adjustments and UFs be applied to the target tissue dose metric. The rationale for applying the UFs to the internal tissue dose is the same as for using the dose metrics in dose-response modeling. It is expected that the observed effects of a chemical will be more directly related to a measure of target tissue dose than to a measure of administered dose. The dose metrics are specifically chosen to provide more useful measures of the biologically effective dose. Therefore, it is the dose metrics that should be adjusted to assure that the biologically effective dose is reduced to the extent desired. As a counter-example, consider the case where toxicity has been observed for exposure to a chemical at a concentration well above the point where saturation of the metabolism of the chemical occurs, and where the metabolism of the chemical is responsible for the toxicity. In this case, applying the UF to a dose metric based on total metabolism would assure that the extent of metabolism would be reduced by the same factor. However, if the concentration at which the effect was observed was sufficiently higher than the concentration at which metabolism is saturated, applying the UF to the exposure concentration might not actually reduce the extent of metabolism to any appreciable extent.

A reduced interspecies UF is justified due to an increased confidence in tissue dosimetry achieved with a more optimal compound-specific and species-specific model for regional accumulation of AA in target tissues within the nasal cavity. Other reductions reflect the developing characterization of the mode of action of these acids and esters. The toxicity appears to be related to acid accumulation regardless of the source, i.e., either parent acid or ester hydrolysis. The results from 2-year studies with other compounds, specifically the acrylate esters, indicate that an uncertainty adjustment for subchronic to chronic situations for this end point is unnecessary.

Evaluation of Model Behavior
As mentioned in the Results section describing the evaluation of model behavior, the pH of the mucus layer used in the model determines the effective mucus:air partition coefficient of AA. Mucus has a limited buffering capacity for the extracted AA, and the mucus pH decreases with increasing mucus AA concentration. Hence, the characteristics of the mucus layer influence the concentration of AA in the target tissue.

The net flux of AA from the airstream into the mucus is described as a function of the airstream AA concentration, the mass transport coefficient across the lumen-mucus interface and the effective air:mucus partition coefficient. The assumed pH of the mucus layer determines the effective partition coefficient of AA between air and mucus. The thermodynamic partition coefficient (Pm:a) is a ratio of the concentration of the unionized acid in the mucus to the concentration in air. The effective partition coefficient (Peffm:a) is assumed to be the ratio of the total concentration of the acid and acrylate in the mucus to the air concentration, as described by the following equation.


(1A)

Given the dissociation constant for AA (Ka), the concentration of the ionized and the unionized form are related as follows:

(1B)
By substituting for [A-] from Equation 1bGo into Equation 1aGo, the effective partition coefficient can be expressed in terms of the thermodynamic partition coefficient, mucus pH, and pKa of AA, as follows:

(2A)

or


(2B)

The net AA flux from the lumen to the mucus is dependent on the overall mass transport coefficient (K), which in turn is a function of the gas phase mass transport (kg) and the effective mucus phase mass transport coefficient (kmPAeffm:a), 1/K = 1/kg + 1/(kmPAeffm:a). Hence, acid extraction from the air stream will depend on the mucus pH, as this alters the partition coefficient.

The sensitivity analysis indicates that characteristics of the mucus in terms of buffering capacity, the initial pH, and the form of the relationship accounting for acidification are important parameters for the accumulation of AA in target tissues within the olfactory epithelium at higher exposure concentrations. The AA risk assessment presented in this paper was conducted using the CFD-PBPK model developed by Frederick et al. (1998). Hence, the empirical formulation for pH changes in the mucus developed by Frederick et al. (1998) underlies this analysis. Alternate formulations for acidification, expressing pH changes in terms of rates of change of intracellular proton concentration, are presented in the literature (Plowchalk et al., 1997Go).

Additional Considerations
There are data available in the literature that can be compared to the model estimates to determine whether the dose metrics adequately reflect the toxicity expected. In a study conducted by Lomax et al. (1994), application of concentration times time was evaluated relative to nasal irritation for AA. The results of this study indicated that after administration of AA at concentrations and durations (25 ppm x 4.4 or 6 h, 5 ppm for 22 h) similar to those considered in the current analysis (25 ppm x 6 h, 4.464 ppm x 24 h), similar incidences of nasal irritation were observed. These data demonstrate an equivalent response for nasal irritation with equivalent C x T exposure. However, the dose metrics predicted by the model are nonlinear in this region of C x T. This nonlinearity appears to be related to the formulation for acidification used in the hybrid model. Indeed, it is this nonlinearity that leads to the difference between the RfCs obtained with the various approaches described. The dose response for acidification is thus a key area for further investigation.

In conclusion, the dosimetry model integrates physiological, anatomical, biochemical, and biophysical characteristics of nasal airflow and nasal extraction to improve on previous default dosimetry corrections. After the application of appropriate UFs, the CFD-PBPK model-derived RfC for ambient exposure is 79 ppb.


    NOTES
 
1 To whom correspondence should be addressed. Fax: (318) 255-4960. E-mail: robinangentry{at}icfconsulting.com. Back


    REFERENCES
 TOP
 ABSTRACT
 INTRODUCTION
 MATERIALS AND METHODS
 RESULTS
 DISCUSSION
 REFERENCES
 
Bogdanffy, M. S., Gladnick, N. L., Kegelman, T., and Frame, S. R. (1997). Cell proliferation responses in rat nasal epithelium following repeated exposures to vinyl acetate vapor. Inhal. Toxicol. 9, 331–350.[ISI]

Brass, E. P. (1994). Overview of coenzyme A metabolism and its role in cellular toxicity. Chem. Biol. Interact. 90, 203–214.[ISI][Medline]

Cascieri, T., and Clary, J. (1993). Acrylic acid health effects overview. In Health Effect Assessments of the Basic Acrylates (T. Tyler, S. Murphy, and H. Ek, Eds.), pp. 9–31. CRC Press, Boca Raton, FL.

Cruzan, G., Cushman, J. R., Andrews, L. S., Granville, G. C., Miller, R. R., Hardy, C. J., Coombs, D. W., and Mullins, P. A. (1997). Subchronic inhalation studies of styrene in CD rats and CD-1 mice. Fundam. Appl. Toxicol. 35, 152–165.[ISI][Medline]

Custodio, J. B., Palmeira, C. M., Moreno, A. J., and Wallace, K. B. (1998). Acrylic acid induces the glutathione-independent mitochondrial permeability transition in vitro. Toxicol. Sci. 43, 19–27.[Abstract]

Frederick, C., Udinsky, J., and Finch, L. (1994). The regional hydrolysis of ethyl acrylate to acrylic acid in the rat nasal cavity. Toxicol. Lett. 70, 49–56.[ISI][Medline]

Frederick, C., Bush, M. L., Lomax, L. G., Black, K. A., Finch, L., Kimbell, J. S., Morgan, K. T., Subramaniam, R. P., Morris, J. B., and Ultman, J. S. (1998). Application of a hybrid computational fluid dynamics and physiologically based inhalation model for interspecies dosimetry extrapolation of acidic vapors in the upper airways. Toxicol. Appl. Pharmacol. 152, 211–231.[ISI][Medline]

Frederick, C., Bush, M. L., Lomax, L. G., Black, K. A., Finch, L., Kimbell, J. S., Morgan, K. T., Subramaniam, R. P., Morris, J. B., and Ultman, J. S. (2000). 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. (submitted).

Gage, J. (1970). The subacute inhalation toxicity of 109 industrial chemicals. Br. J. Ind. Med. 27, 1–18.[Medline]

Jarabek, A. M. (1994). Inhalation RfC methodology: Dosimetric adjustment and dose-response estimates of non-cancer toxicity in the upper respiratory tract. Inhal. Toxicol. 6, 301–325.[ISI]

Keenan, C. M., Kelly, D. P., and Bogdanffy, M. S. (1990). Degeneration and recovery of rat olfactory epithelium following inhalation of dibasic esters. Fundam. Appl. Toxicol. 15, 381–393.[ISI][Medline]

Kimbell, J. S., Godo, M. N., Gross, E. A., Joyner, D. R., Richardson, R. B., and Morgan, K. T. (1997). Computer simulation of inspiratory airflow in all regions of the F344 rat nasal passages. Toxicol. Appl. Pharmacol. 145, 388–398.[ISI][Medline]

Lomax, L., Brown, D., and Frederick, C. (1994). Regional histopathology of the mouse nasal cavity following two weeks of exposure to acrylic acid for either 6 or 22 hours per day. Inhal. Toxicol. 6, 445–449.

Lomax, L., Krivanek, N., and Frame, S. (1997). Chronic inhalation toxicity and oncogenicity of methyl methacrylate in rats and hamsters. Food Chem. Toxicol. 35, 393–407.[ISI][Medline]

McCarthy, T. J., and Witz, G. (1997). Structure-activity relationships in the hydrolysis of acrylate and methacrylate esters by carboxylesterase in vitro. Toxicology 116, 153–158.[ISI][Medline]

Mery, S., Larson, J., Butterworth, B., Wolf, D., Harden, R., and Morgan, K. (1994). Nasal toxicity of chloroform in male F-344 rats and female B6C3F1 mice following a 1-week inhalation exposure. Toxicol. Appl. Pharmacol. 125, 214–227.[ISI][Medline]

Miller, R., Ayres, J., and Jersey, G. (1979). Acrylic acid 90-day vapor inhalation study with rats and mice. Final report. Dow Chemical Company, Toxicology Research Laboratory, Health and Environmental Science, 79RC-1024, Midland, MI.

Miller, R. R., Ayres, J. A., Jersey, G. C., and McKenna, M. J. (1981a). Inhalation toxicity of acrylic acid. Fundam. Appl. Toxicol. 1, 271–277.[Medline]

Miller, R., Ayres, J., Rampy, L., and McKenna, M. (1981b). Metabolism of acrylate esters in rat tissue homogenates. Fundam. Appl. Toxicol. 1, 410–414.[Medline]

Miller, R. R., Young, J. T., Kociba, R. J., Keyes, D. G., Bodner, K. M., Calhoun, L. L., and Ayres, J. A. (1985). Chronic toxicity and oncogenicity bioassay of inhaled ethyl acrylate in Fischer 344 rats and B6C3F1 mice. Drug Chem. Toxicol. 8, 1–42.[ISI][Medline]

Morris, J. B., and Frederick, C. B. (1995). Upper respiratory tract uptake of acrylate ester and acid vapors. Inhal. Toxicol. 7, 557–574.[ISI]

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.[ISI][Medline]

Nachreiner, D., and Dodd, D. (1989). Acrylic Acid: Acute Vapor Inhalation Toxicology Test in Rats. Bushy Run Research Center, Export, PA. Project Report 51–577.

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.[ISI][Medline]

Reininghaus, W., Koestner, A., and Klimisch, H.-J. (1991). Chronic toxicity and oncogenicity of inhaled methyl acrylate and n-butyl acrylate in Sprague-Dawley rats. Food Chem. Toxicol. 29, 329–339.[ISI][Medline]

Renwick, A. G. (1993). Data-derived safety factors for evaluation of food additives and environmental contaminants. Food Addit. Contam. 10, 275–305.[ISI][Medline]

Sanders, J., Burka, L., and Matthews, H. (1988). Metabolism and disposition of n-butyl acrylate in male Fischer rats. Drug Metab. Dispos. 16, 429–434.[Abstract]

Silver, E., and Murphy, S. (1981). Potentiation of acrylate ester toxicity by prior treatment with the carboxylesterase inhibitor triorthotolyl phosphate (TOTP). Toxicol. Appl. Pharmacol. 57, 208–219.[ISI][Medline]

Stott, W., and McKenna, M. (1985). Hydrolysis of several glycol ether acetates and acrylate esters by nasal mucosal carboxylesterase in vitro. Fundam. Appl. Toxicol. 5, 399–404.[ISI][Medline]

Subramaniam, R., Richardson, R., B., Morgan, K. T., Kepler, G., Kimbell, J. S., and Guilmette, R. A. (1998). Computational fluid dynamics simulations of inspiratory airflow in the human nose and nasopharynx. Inhal. Toxicol. 10, 91–120.[ISI]

Trela, B. A., and Bogdanffy, M. S. (1991). Cytotoxicity of dibasic esters (DBE) metabolites in rat nasal explants. Toxicol. Appl. Pharmacol. 110, 259–267.[ISI][Medline]

U.S. EPA (1994). Methods for Derivation of Inhalation Reference Concentrations and Application of Inhalation Dosimetry. Office of Research and Development, EPA/600/8–90/066F, Washington, DC.

U.S. EPA (1998). IRIS Substance FileAcrylic Acid. U.S. Environmental Protection Agency, Cincinnati, OH.

Weast, R. C., and Astle, M. J. (1978). CRC Handbook of Chemistry and Physics. CRC Press, Inc., West Palm Beach, FL.