Abstract
Formaldehyde is a nasal carcinogen in rodents at high doses and is an endogenous compound that is present in all living cells. Due to its high solubility and reactivity, quantitative risk estimates for inhaled formaldehyde have relied on internal dose estimates in the upper respiratory tract. Dosimetry calculations are complicated by the presence of endogenous formaldehyde concentrations in the respiratory mucosa. Anatomically accurate computational fluid dynamics (CFD) models of the rat, monkey, and human nasal passages were used to simulate uptake of inhaled formaldehyde. An epithelial structure was implemented in the nasal CFD models to estimate formaldehyde absorption from air:tissue partitioning, species-specific metabolism, first-order clearance, DNA binding, and endogenous formaldehyde production. At an exposure concentration of 1 ppm, predicted formaldehyde nasal uptake was 99.4, 86.5, and 85.3% in the rat, monkey, and human, respectively. Endogenous formaldehyde in nasal tissues did not significantly affect wall mass flux or nasal uptake predictions at exposure concentrations > 500 ppb; however, reduced nasal uptake was predicted at lower exposure concentrations. At an exposure concentration of 1 ppb, predicted nasal uptake was 17.5 and 42.8% in the rat and monkey; net desorption of formaldehyde was predicted in the human model. The nonlinear behavior of formaldehyde nasal absorption will affect the dose-response analysis and subsequent risk estimates at low exposure concentrations. Updated surface area partitioning of nonsquamous epithelium and average flux values in regions where DNA-protein cross-links and cell proliferation rates were measured in rats and monkeys are reported for use in formaldehyde risk models of carcinogenesis.
Formaldehyde is used in many manufacturing processes, including the production of fertilizer, wood and paper products, and in the textiles, rubber, and cement industries. Human exposure to formaldehyde is common from both occupational and environmental sources. Ambient air concentrations vary widely depending on proximity to nearby emission sources that include automobile exhaust, power plants, petroleum refineries, and tobacco smoke. Ambient air monitoring data in the United States yielded overall averages of formaldehyde concentrations in the 2.5–3 ppb range (ATSDR, 1999). Indoor formaldehyde concentrations are typically higher and have been estimated to range from 16 to 32 ppb (Salthammer et al., 2010).
Formaldehyde is a highly water soluble and reactive gas that is readily absorbed by the upper respiratory tract upon inhalation and rapidly metabolized by enzymes in the respiratory mucosa. Formaldehyde causes nasal squamous cell carcinomas in rats chronically exposed to concentrations of 6 ppm and higher (Kerns et al., 1983). Formaldehyde inhalation leads to the formation of DNA-protein cross-links (DPX) and increased cell proliferation rates in the nasal passages of rats and monkeys (Casanova et al., 1991, 1994; Monticello et al., 1991). High levels of formaldehyde-induced DPX and cell proliferation rates have been correlated with the locations of nasal lesions and inspiratory airflow patterns (Casanova et al., 1994; Monticello et al., 1996; Morgan et al., 1991). DPX likely play a role in formaldehyde toxicity and carcinogenicity and were used as biomarkers for delivered dose in earlier cancer risk assessments of formaldehyde (Hernandez et al., 1994).
Three-dimensional, anatomically accurate computational fluid dynamics (CFD) models of the rat, monkey, and human nasal passages have been developed to predict inhaled airflow patterns and formaldehyde uptake to elucidate the roles of nasal anatomy and airflow on formaldehyde toxicity (Kepler et al., 1998; Kimbell et al., 1997, 2001a,b; Subramaniam et al., 1998). The results from the CFD models provided further support that interspecies differences in formaldehyde-induced lesion locations were due to differences in airflow and gas uptake patterns and that localized formaldehyde dose plays a major role in the distribution of nasal lesions in formaldehyde-exposed animals. CFD simulation results were also used to estimate flux at sites within the nasal passages of rats and monkeys where DPX and cell proliferation rates were measured (Conolly et al., 2000; Kimbell et al., 2001b). Predictions of localized DPX showed good agreement with experimental data in rats and monkeys using a model developed to simulate formaldehyde disposition and DPX kinetics in the nasal mucosa (Conolly et al., 2000). CFD-derived flux data were subsequently used in the development of biologically motivated computational models for formaldehyde carcinogenesis in rats and humans (Conolly et al., 2003, 2004).
Formaldehyde is also produced endogenously from normal cellular metabolism and is present in all tissues of the body, including the respiratory mucosa. Formaldehyde has been measured in exhaled breath at median levels of a few parts per billion, but there is a large degree of variability in measurements due to the sensitivity of the analytical methods used for gas analysis (Moser et al., 2005; Riess et al., 2010; Wang et al., 2008). Recent studies have used mass spectrometry to differentiate between DNA adducts arising from endogenous formaldehyde and those arising from inhaled exogenous formaldehyde in the nasal epithelium of rats and monkeys (Edrissi et al., 2013; Lu et al., 2011; Moeller et al., 2011). It is currently unknown how the presence of endogenous formaldehyde concentrations in respiratory tissues affects absorption of inhaled exogenous formaldehyde. To further understand this process, tissue dosimetry models are needed that consider the contribution from endogenous sources, formaldehyde clearance due to metabolism and reactivity, and the delivered dose to the respiratory mucosa from exogenous formaldehyde inhalation (Andersen et al., 2010).
In this study, CFD models of the rat, monkey, and human nasal passages were used to simulate nasal absorption of inhaled formaldehyde due to effects of nasal airflow and the underlying pharmacokinetics, including the presence of endogenous formaldehyde, in nasal tissues. The effects of endogenous formaldehyde on nasal uptake and wall mass flux estimates were examined and updated flux values are provided at sites within the rat and monkey nasal passages where DPX and cell proliferation rates were measured for subsequent use in computational models of formaldehyde carcinogenesis.
MATERIALS AND METHODS
CFD models of the nasal passages of a F344 rat, a rhesus monkey, and an adult human were used to simulate inhaled airflow and the transport and absorption of inhaled formaldehyde vapor. Boundary conditions for formaldehyde absorption at nasal airway walls were determined from formaldehyde physicochemical characteristics and clearance properties in the nasal mucosa and included the effects of endogenous formaldehyde production in nasal tissues. Endogenous formaldehyde production rates were estimated from an independent formaldehyde tissue disposition model.
Nasal CFD models
Anatomically accurate, three-dimensional reconstructions of the rat and monkey nasal passages were developed in previous studies and were constructed from histological section tracings of the right nasal passages from the nostril to the nasopharynx (Kepler et al., 1998; Kimbell et al., 1997). Details of the CFD model development for the rat and monkey can be found in earlier formaldehyde dosimetry modeling efforts (Kimbell et al., 2001a,b). For this study, the original numerical meshes from the rat and monkey nasal CFD models were removed and the surfaces were imported into the medical imaging software Mimics (Materialise, Inc.) so that the surface contours of the original models could be smoothed and refined (Fig. 1). The human nasal CFD model used in this study was reconstructed from de-identified computed tomography (CT) scan images (0.7 mm resolution) of the nasal passages of a 37-year-old female (56.7 kg) with no radiological evidence of nasal abnormalities using Mimics (Fig. 1). This model differs from the human CFD model used by Kimbell et al. (2001b) and was developed from higher-resolution scan images, which leads to a smoother, higher-fidelity surface. The approximate locations of nonmucus-coated and mucus-coated squamous epithelia were mapped onto the surfaces of the rat and monkey nasal CFD models according to the descriptions in Kimbell et al. (2001b). The extent of nonmucus-coated squamous epithelium in the human model was defined according to the description by Lang (1989).
Lateral views of the rat (top), monkey (middle), and human (bottom) nasal CFD models. The approximate location of nonmucus-coated squamous epithelium is highlighted in dark gray. Model sizes are not to scale.
The locations of areas in which DPX and cell proliferation rates were measured were mapped onto the surface meshes of the CFD models of the rat and monkey so that average formaldehyde wall mass flux values could be calculated in these regions. In the rat, areas in which cell proliferation rates were measured included the anterior lateral meatus (ALM), posterior lateral meatus (PLM), anterior midseptum (AMS), posterior midseptum (PMS), anterior dorsal septum (ADS), and the medial maxilloturbinate (MMT) (Monticello et al., 1996). Cell proliferation was measured in the monkey at histological cross-sections corresponding to Levels B–E, as described by Monticello et al. (1989). The locations of areas where DPX were measured in the rat included the high tumor region (which consisted of the lateral meatus) and the low tumor region consisting of the medial aspects of the naso- and maxilloturbinates, the posterior lateral wall, and the posterior dorsal septum, as described by Casanova et al. (1994). In the monkey, DPX areas consisted of the anterior lateral walls and septum (ALWS), the nasopharynx (NP), and the middle turbinates (MT) (Casanova et al., 1991). Surface areas of nasal epithelia, DPX regions, and cell proliferation sites are given in Table 1. The locations of cell proliferation sites and DPX regions in the rat and monkey models are shown in Supplementary Data. The descriptions of DPX and cell replication regions in the CFD models were similar to those used in Kimbell et al. (2001a) and Conolly et al. (2000).
Surface Areas in the Rat, Monkey, and Human Nasal CFD Models
| Surface area (mm2) | |||
|---|---|---|---|
| Rat | Monkey | Human | |
| Whole nose | 1704.8 | 7568.6 | 20,159.6 |
| Dry squamous | 67.4 | 896.2 | 1259.8 |
| Wet squamous | 105.4 | 108.6 | |
| Cell replication sites | ALM: 49.4 | Level B: 187.1 | |
| PLM: 42.2 | Level C: 207.7 | ||
| AMS: 7.7 | Level D: 113.6 | ||
| PMS: 9.7 | Level E: 51.4 | ||
| ADS: 3.8 | |||
| MMT: 5.4 | |||
| DPX regions | High tumor: 85.8 | ALWS: 482.6 | |
| Low tumor: 59.5 | MT: 379.8 | ||
| NP: 233.3 | |||
| Surface area (mm2) | |||
|---|---|---|---|
| Rat | Monkey | Human | |
| Whole nose | 1704.8 | 7568.6 | 20,159.6 |
| Dry squamous | 67.4 | 896.2 | 1259.8 |
| Wet squamous | 105.4 | 108.6 | |
| Cell replication sites | ALM: 49.4 | Level B: 187.1 | |
| PLM: 42.2 | Level C: 207.7 | ||
| AMS: 7.7 | Level D: 113.6 | ||
| PMS: 9.7 | Level E: 51.4 | ||
| ADS: 3.8 | |||
| MMT: 5.4 | |||
| DPX regions | High tumor: 85.8 | ALWS: 482.6 | |
| Low tumor: 59.5 | MT: 379.8 | ||
| NP: 233.3 | |||
Surface Areas in the Rat, Monkey, and Human Nasal CFD Models
| Surface area (mm2) | |||
|---|---|---|---|
| Rat | Monkey | Human | |
| Whole nose | 1704.8 | 7568.6 | 20,159.6 |
| Dry squamous | 67.4 | 896.2 | 1259.8 |
| Wet squamous | 105.4 | 108.6 | |
| Cell replication sites | ALM: 49.4 | Level B: 187.1 | |
| PLM: 42.2 | Level C: 207.7 | ||
| AMS: 7.7 | Level D: 113.6 | ||
| PMS: 9.7 | Level E: 51.4 | ||
| ADS: 3.8 | |||
| MMT: 5.4 | |||
| DPX regions | High tumor: 85.8 | ALWS: 482.6 | |
| Low tumor: 59.5 | MT: 379.8 | ||
| NP: 233.3 | |||
| Surface area (mm2) | |||
|---|---|---|---|
| Rat | Monkey | Human | |
| Whole nose | 1704.8 | 7568.6 | 20,159.6 |
| Dry squamous | 67.4 | 896.2 | 1259.8 |
| Wet squamous | 105.4 | 108.6 | |
| Cell replication sites | ALM: 49.4 | Level B: 187.1 | |
| PLM: 42.2 | Level C: 207.7 | ||
| AMS: 7.7 | Level D: 113.6 | ||
| PMS: 9.7 | Level E: 51.4 | ||
| ADS: 3.8 | |||
| MMT: 5.4 | |||
| DPX regions | High tumor: 85.8 | ALWS: 482.6 | |
| Low tumor: 59.5 | MT: 379.8 | ||
| NP: 233.3 | |||
Computational meshes consisting of unstructured tetrahedral elements were generated in each nasal model using the commercial software ICEM-CFD (ANSYS, Inc.). A thin layer of prism elements along airway walls was extruded from the surface mesh to improve mass transfer calculations at the air-wall interface. The rat, monkey, and human nasal CFD models each contained 4–5 million elements.
Airflow simulations
Steady-state, inspiratory airflow simulations were conducted in each model using Fluent version 14.0 (ANSYS, Inc.) to solve the viscous, incompressible Navier-Stokes equations for laminar flow. Airflow simulations were conducted with an air density of 1.204 kg/m3 and a dynamic viscosity of 1.8 × 10−5 kg/(m s). Airflow boundary conditions consisted of a zero-pressure condition at the nostril(s), a negative pressure at the model outlet, and no-slip conditions on all nasal airway walls. Pressure values at the outlet of each model were calibrated to generate inhalation flow rates equal to twice the minute volume (equal to the tidal volume divided by the inhalation time) for resting breathing in each species. Estimated minute volumes for the rat and monkey were 288 ml/min and 2.4 l/min, respectively (Kimbell et al., 2001b). The estimated minute volume for the human model was determined from allometric scaling to be 6.9 l/min (Garcia et al., 2009).
Formaldehyde uptake
The proportionality constant, k, is referred to as a mass transfer coefficient and was calculated based upon the physicochemical and pharmacokinetic properties of formaldehyde in air and nasal tissues. Following the approach taken by Kimbell et al. (2001b), the mass transfer coefficient was determined separately for nonmucus-coated squamous epithelium in the nasal vestibule and for mucus-coated epithelial types throughout the rest of the nose. For dry squamous epithelium, uptake of formaldehyde was assumed to be similar to formaldehyde absorption in human epidermal tissue. A constant mass transfer coefficient of knm = 0.41 cm/s, as calculated by Kimbell et al. (2001b), was implemented in the dry squamous region in all species. Endogenous formaldehyde production was not applied to the dry squamous region.
Formaldehyde absorption on mucus-coated epithelia is much more rapid than on dry squamous epithelia due to the high water solubility of formaldehyde. In the study by Kimbell et al. (2001b), a constant value of km = 4.7 cm/s was used for the mass transfer coefficient on mucus-coated epithelia in all species. This value was calibrated by fitting nasal uptake in the rat CFD model to the experimental uptake data of Patterson et al. (1986), who measured formaldehyde uptake of 97% in the rat nose at concentrations from 2 to 15 ppm. Use of a constant mass transfer coefficient governing the absorption of formaldehyde does not permit inclusion of endogenous production in the nasal mucosa. Therefore, in this study, the mass transfer coefficient for mucus-coated epithelia was calculated using an approach derived by Schroeter et al. (2008) for another reactive aldehyde, acrolein. In this approach, vapor uptake was modeled as a nonlinear process according to air:tissue partitioning, air and tissue phase diffusivity, and metabolic clearance in the nasal mucosa.
Schematic of the epithelial structure used to define the boundary condition for formaldehyde absorption on mucus-coated nasal airway walls in the CFD models.
Formaldehyde Parameter Values Used in the Tissue Clearance Model
| Parameter | Symbol | Units | Value | Source | ||
|---|---|---|---|---|---|---|
| Rat | Monkey | Human | ||||
| Partition coefficient | Pt:a | Unitless | 72,464 | 72,464 | 72,464 | Asgharian et al. (2012) |
| Tissue diffusivity | Dt | cm2/s | 1.74 × 10−5 | 1.74 × 10−5 | 1.74 × 10−5 | Risk Assessment Information Systema |
| Maximum saturable clearance rate | Vmax | pmol/(mm3 min) | 1008.0 | 91.0 | 15.7 | Conolly et al. (2000) |
| Half-saturation concentration | Km | pmol/mm3 | 70.8 | 6.7 | 6.7 | Conolly et al. (2000) |
| First-order clearance rate constant | kf | min−1 | 1.08 | 1.08 | 1.08 | Conolly et al. (2000) |
| Pseudo first-order rate constant for DNA binding | kb | min−1 | 6.406 × 10−6 | 6.406 × 10−6 | 6.406 × 10−6 | Conolly et al. (2000) |
| Endogenous production rate constant | k0/V | pmol/(mm3 min) | 134.1 | 21.6 | 12.0 | Estimated |
| Mucus layer thickness | Lm | μm | 10 | 20b | 20 | Morris et al. (1993) |
| Epithelial layer thicknessc | Lt | μm | 35.5 | 177.5 | 177.5 | Conolly et al. (2000) |
| Submucosal layer thicknessc | Lb | μm | 35.5 | 177.5 | 177.5 | Conolly et al. (2000) |
| Cardiac output | CO | ml/min | 99 | 1510 | 5200 | Brown et al. (1997) |
| Parameter | Symbol | Units | Value | Source | ||
|---|---|---|---|---|---|---|
| Rat | Monkey | Human | ||||
| Partition coefficient | Pt:a | Unitless | 72,464 | 72,464 | 72,464 | Asgharian et al. (2012) |
| Tissue diffusivity | Dt | cm2/s | 1.74 × 10−5 | 1.74 × 10−5 | 1.74 × 10−5 | Risk Assessment Information Systema |
| Maximum saturable clearance rate | Vmax | pmol/(mm3 min) | 1008.0 | 91.0 | 15.7 | Conolly et al. (2000) |
| Half-saturation concentration | Km | pmol/mm3 | 70.8 | 6.7 | 6.7 | Conolly et al. (2000) |
| First-order clearance rate constant | kf | min−1 | 1.08 | 1.08 | 1.08 | Conolly et al. (2000) |
| Pseudo first-order rate constant for DNA binding | kb | min−1 | 6.406 × 10−6 | 6.406 × 10−6 | 6.406 × 10−6 | Conolly et al. (2000) |
| Endogenous production rate constant | k0/V | pmol/(mm3 min) | 134.1 | 21.6 | 12.0 | Estimated |
| Mucus layer thickness | Lm | μm | 10 | 20b | 20 | Morris et al. (1993) |
| Epithelial layer thicknessc | Lt | μm | 35.5 | 177.5 | 177.5 | Conolly et al. (2000) |
| Submucosal layer thicknessc | Lb | μm | 35.5 | 177.5 | 177.5 | Conolly et al. (2000) |
| Cardiac output | CO | ml/min | 99 | 1510 | 5200 | Brown et al. (1997) |
bThe mucus thickness in the monkey was assumed to be equal to that in humans.
cThe epithelial and submucosal layers were assumed to be the same thickness and were determined from the mucosal thickness from Conolly et al. (2000) after subtracting the mucus thickness.
Formaldehyde Parameter Values Used in the Tissue Clearance Model
| Parameter | Symbol | Units | Value | Source | ||
|---|---|---|---|---|---|---|
| Rat | Monkey | Human | ||||
| Partition coefficient | Pt:a | Unitless | 72,464 | 72,464 | 72,464 | Asgharian et al. (2012) |
| Tissue diffusivity | Dt | cm2/s | 1.74 × 10−5 | 1.74 × 10−5 | 1.74 × 10−5 | Risk Assessment Information Systema |
| Maximum saturable clearance rate | Vmax | pmol/(mm3 min) | 1008.0 | 91.0 | 15.7 | Conolly et al. (2000) |
| Half-saturation concentration | Km | pmol/mm3 | 70.8 | 6.7 | 6.7 | Conolly et al. (2000) |
| First-order clearance rate constant | kf | min−1 | 1.08 | 1.08 | 1.08 | Conolly et al. (2000) |
| Pseudo first-order rate constant for DNA binding | kb | min−1 | 6.406 × 10−6 | 6.406 × 10−6 | 6.406 × 10−6 | Conolly et al. (2000) |
| Endogenous production rate constant | k0/V | pmol/(mm3 min) | 134.1 | 21.6 | 12.0 | Estimated |
| Mucus layer thickness | Lm | μm | 10 | 20b | 20 | Morris et al. (1993) |
| Epithelial layer thicknessc | Lt | μm | 35.5 | 177.5 | 177.5 | Conolly et al. (2000) |
| Submucosal layer thicknessc | Lb | μm | 35.5 | 177.5 | 177.5 | Conolly et al. (2000) |
| Cardiac output | CO | ml/min | 99 | 1510 | 5200 | Brown et al. (1997) |
| Parameter | Symbol | Units | Value | Source | ||
|---|---|---|---|---|---|---|
| Rat | Monkey | Human | ||||
| Partition coefficient | Pt:a | Unitless | 72,464 | 72,464 | 72,464 | Asgharian et al. (2012) |
| Tissue diffusivity | Dt | cm2/s | 1.74 × 10−5 | 1.74 × 10−5 | 1.74 × 10−5 | Risk Assessment Information Systema |
| Maximum saturable clearance rate | Vmax | pmol/(mm3 min) | 1008.0 | 91.0 | 15.7 | Conolly et al. (2000) |
| Half-saturation concentration | Km | pmol/mm3 | 70.8 | 6.7 | 6.7 | Conolly et al. (2000) |
| First-order clearance rate constant | kf | min−1 | 1.08 | 1.08 | 1.08 | Conolly et al. (2000) |
| Pseudo first-order rate constant for DNA binding | kb | min−1 | 6.406 × 10−6 | 6.406 × 10−6 | 6.406 × 10−6 | Conolly et al. (2000) |
| Endogenous production rate constant | k0/V | pmol/(mm3 min) | 134.1 | 21.6 | 12.0 | Estimated |
| Mucus layer thickness | Lm | μm | 10 | 20b | 20 | Morris et al. (1993) |
| Epithelial layer thicknessc | Lt | μm | 35.5 | 177.5 | 177.5 | Conolly et al. (2000) |
| Submucosal layer thicknessc | Lb | μm | 35.5 | 177.5 | 177.5 | Conolly et al. (2000) |
| Cardiac output | CO | ml/min | 99 | 1510 | 5200 | Brown et al. (1997) |
bThe mucus thickness in the monkey was assumed to be equal to that in humans.
cThe epithelial and submucosal layers were assumed to be the same thickness and were determined from the mucosal thickness from Conolly et al. (2000) after subtracting the mucus thickness.
Formaldehyde wall mass flux (i.e., the rate at which formaldehyde vapor is absorbed) was computed at every nodal point on the surface meshes of all the models. Maximum and average flux values were computed on mucus-coated nonsquamous epithelium from over 200,000 surface elements in the rat, monkey, and human models. “Maximum” flux values reported in this study represent the 99th percentile flux, defined as the flux value which is greater than 99% of all fluxes. The 99th percentile flux was believed to be a more accurate dose metric than a maximum flux value that is based on a single element, which may be spurious due to calculations on irregular surfaces found in nasal geometries (Garcia et al., 2009). Average flux values were also computed in the DPX and cell replication regions in the rat and monkey models. All reported flux values were computed at air flow rates equal to twice the minute volume for resting breathing. Simulated exposure concentrations ranged from < 0.001 ppm up to 10 ppm in all species. Additionally, exposure concentrations were conducted at 0.7, 2, 6, 10, and 15 ppm in the rat and at 0.7, 2, and 6 ppm in the monkey because these exposure concentrations were used in experimental studies measuring DPX and cell proliferation. In each species, predicted net nasal uptake was calculated as (Cnostril(s) – Coutlet)/Cnostril(s), where Cnostril(s) is the exposure concentration and Coutlet is the predicted concentration at the model outlet. Nasal uptake calculations measured net overall absorption and potential desorption of formaldehyde on nasal airway walls.
To examine the distribution of formaldehyde flux across the nasal surface, the nonsquamous epithelia of each species was partitioned into 20 evenly spaced bins between the minimum and maximum flux values following the approach of Kimbell et al. (2001a). Average fluxes were computed in each bin. User-defined functions were implemented in Fluent for all flux binning calculations. Flux binning calculations were conducted at an exposure concentration of 1 ppm in each species and also at exposure concentrations of 0.001, 0.01, and 0.1 ppm in the human model.
Formaldehyde tissue disposition model
A nasal epithelial model was developed in Matlab (Mathworks, Inc.) to simulate the tissue disposition of absorbed formaldehyde. The model was designed to be run independently of the CFD simulations and was used to calibrate the endogenous formaldehyde production rate constant in each species. The structure of the model was identical to the epithelial structure used in the CFD models (described above and shown in Fig. 2) and used the same model parameters except that the air-phase formaldehyde concentration (Cair) was set to 0 to simulate inhalation of clean air. Endogenous formaldehyde concentrations in the nasal mucosa of unexposed F344 rats were measured to be about 0.4 μmol/g (Heck et al., 1982). This concentration was also used for monkeys and humans. The zero-order endogenous production rate constant, k0, was calibrated in each species to calculate average steady-state tissue concentrations of 0.4 μmol/g, assuming inhalation of clean air and a blood concentration of 2.5 mg/l using the model parameters from Table 2 to simulate formaldehyde tissue disposition. The endogenous production rates that were derived using the nasal epithelial model were used in all formaldehyde uptake calculations with the CFD models.
RESULTS
Steady-state, inspiratory airflow simulations were conducted at twice the resting minute volume in each species. The pressure drops required to generate these airflow rates were 88, 55, and 12 Pa in the rat, monkey, and human models, respectively. Actual flow rates used for the formaldehyde uptake simulations were 287.7 ml/min in the rat, 2.4 l/min in the monkey (flow rates reflect half of the total volumetric flow because the rat and monkey models only contained one side of the nose), and 13.8 l/min in the human. Nasal airflow patterns in each species were similar to the results from Kimbell et al. (2001b). Mass balance errors for airflow simulations were < 0.01% in the rat, monkey, and human CFD models.
Endogenous formaldehyde production rate constants estimated from the nasal tissue disposition model were 8.9 × 103, 2.7 × 104, and 8.6 × 104 pmol/min in the rat, monkey, and human nasal mucosa, respectively. When normalized by tissue volume (as implemented in the CFD models), endogenous rate constants were 134.1, 21.6, and 12.0 pmol/(mm3 min) (Table 2). The endogenous formaldehyde production rates derived using the nasal epithelial model were used in the epithelial structure of the CFD models of the rat, monkey, and human for all subsequent CFD simulations of formaldehyde exposure scenarios.
The CFD models were used to simulate nasal uptake of inhaled formaldehyde in each species at flow rates equal to twice the estimated minute volume. The three-tiered epithelial structure was incorporated into the CFD models to include formaldehyde physicochemical characteristics that govern absorption at nasal airway walls and endogenous production of formaldehyde in the nasal mucosa. At an exposure concentration of 1 ppm, predicted net nasal uptake was 99.4, 86.5, and 85.3% in the rat, monkey, and human, respectively. Nasal uptake predictions were unchanged in each species at exposure concentrations > 1 ppm. The presence of endogenous formaldehyde in nasal tissues did not significantly affect nasal uptake predictions at exposure concentrations > 500 ppb. However, for exposure concentrations < 500 ppb, estimated net nasal uptake decreased as exposure concentration decreased (Fig. 3). In the rat, nasal uptake was > 90% for exposure concentrations > 10 ppb. As exposure concentrations decreased below 10 ppb, formaldehyde nasal uptake decreased rapidly to a value of 17.5% at a concentration of 1 ppb. Similar behavior was predicted in the monkey and human, with slightly decreased nasal uptake for exposure concentrations in the 10–100 ppb range and significantly reduced uptake for exposure levels below 10 ppb. In the human model, at exposure concentrations of 1 ppb and below, the concentration at the model outlet was estimated to be slightly greater than 1 ppb, indicating net desorption of formaldehyde in the human nose due to the presence of endogenous formaldehyde in nasal tissues. Net desorption was predicted in the rat and monkey nasal models at exposure concentrations < 800 and 500 ppt, respectively.
Formaldehyde nasal uptake predictions in the rat, monkey, and human nasal CFD models.
Decreased nasal uptake at low exposure concentrations is primarily due to the difference in tissue kinetics in the nasal mucosa with endogenous formaldehyde production. At a zero exposure concentration, predicted tissue levels were at or below the 2.5 mg/l level in the blood, which acts as a boundary condition at the tissue:blood interface (Fig. 4). As air concentrations increase, this leads to higher formaldehyde concentrations at the air:tissue interface, thereby producing a gradient in tissue concentrations near the air:tissue interface that drives absorption from the air phase. This effect was most pronounced at exposure concentrations > 0.1 ppm (Fig. 4).
Predicted formaldehyde nasal tissue concentrations in the human model at exposure concentrations of 0, 0.001, 0.01, 0.1, and 1 ppm. Air concentrations were determined from the CFD model in the human nose near the region of highest flux on nonsquamous epithelium. A tissue depth of 0 corresponds to the air:tissue interface; a tissue depth of 375 μm corresponds to the tissue:blood interface.
Nasal uptake predictions were sensitive to the endogenous production rate in all species, most notably at low exposure concentrations (Table 3). Higher endogenous production rates yielded higher formaldehyde tissue concentrations, which in turn decreased the concentration gradient between air and tissue at lower exposure concentrations, leading to reduced absorption. At higher exposure concentrations, these effects were not as magnified due to higher formaldehyde air concentrations. The sensitivity of the predicted formaldehyde concentration at the model outlet was also examined in the human model by computing normalized sensitivity coefficients as a function of a ± 10% change in model parameter (Table 4). At an exposure concentration of 1 ppm, model outlet concentrations were insensitive to all parameters due to the strong influence of the partition coefficient on uptake (i.e., formaldehyde uptake is primarily driven by its high solubility). Sensitivity coefficients for most model parameters were higher at an exposure concentration of 1 ppb due to the decreased uptake levels. Model predictions were most sensitive to the partition coefficient, tissue diffusivity, and rate constants.
Sensitivity of Nasal Uptake Predictions to the Endogenous Production Rate (k0)
| Percent change in endogenous production rate | |||
|---|---|---|---|
| Concentration (ppm) | −10% | 0% | +10% |
| Nasal uptake (%) | |||
| Rat | |||
| 1 | 99.4 | 99.4 | 99.4 |
| 0.1 | 98.7 | 98.6 | 98.6 |
| 0.01 | 92.1 | 91.3 | 90.5 |
| 0.001 | 25.5 | 17.5 | 9.4 |
| Monkey | |||
| 1 | 86.5 | 86.5 | 86.5 |
| 0.1 | 86.5 | 86.5 | 86.5 |
| 0.01 | 84.2 | 84.1 | 83.0 |
| 0.001 | 47.5 | 42.8 | 37.3 |
| Human | |||
| 1 | 85.3 | 85.3 | 85.3 |
| 0.1 | 84.8 | 84.7 | 84.6 |
| 0.01 | 78.0 | 77.1 | 76.1 |
| 0.001 | 8.1 | n/aa | n/aa |
| Percent change in endogenous production rate | |||
|---|---|---|---|
| Concentration (ppm) | −10% | 0% | +10% |
| Nasal uptake (%) | |||
| Rat | |||
| 1 | 99.4 | 99.4 | 99.4 |
| 0.1 | 98.7 | 98.6 | 98.6 |
| 0.01 | 92.1 | 91.3 | 90.5 |
| 0.001 | 25.5 | 17.5 | 9.4 |
| Monkey | |||
| 1 | 86.5 | 86.5 | 86.5 |
| 0.1 | 86.5 | 86.5 | 86.5 |
| 0.01 | 84.2 | 84.1 | 83.0 |
| 0.001 | 47.5 | 42.8 | 37.3 |
| Human | |||
| 1 | 85.3 | 85.3 | 85.3 |
| 0.1 | 84.8 | 84.7 | 84.6 |
| 0.01 | 78.0 | 77.1 | 76.1 |
| 0.001 | 8.1 | n/aa | n/aa |
aThe predicted formaldehyde concentration exiting the nasal model was greater than the exposure concentration.
| Percent change in endogenous production rate | |||
|---|---|---|---|
| Concentration (ppm) | −10% | 0% | +10% |
| Nasal uptake (%) | |||
| Rat | |||
| 1 | 99.4 | 99.4 | 99.4 |
| 0.1 | 98.7 | 98.6 | 98.6 |
| 0.01 | 92.1 | 91.3 | 90.5 |
| 0.001 | 25.5 | 17.5 | 9.4 |
| Monkey | |||
| 1 | 86.5 | 86.5 | 86.5 |
| 0.1 | 86.5 | 86.5 | 86.5 |
| 0.01 | 84.2 | 84.1 | 83.0 |
| 0.001 | 47.5 | 42.8 | 37.3 |
| Human | |||
| 1 | 85.3 | 85.3 | 85.3 |
| 0.1 | 84.8 | 84.7 | 84.6 |
| 0.01 | 78.0 | 77.1 | 76.1 |
| 0.001 | 8.1 | n/aa | n/aa |
| Percent change in endogenous production rate | |||
|---|---|---|---|
| Concentration (ppm) | −10% | 0% | +10% |
| Nasal uptake (%) | |||
| Rat | |||
| 1 | 99.4 | 99.4 | 99.4 |
| 0.1 | 98.7 | 98.6 | 98.6 |
| 0.01 | 92.1 | 91.3 | 90.5 |
| 0.001 | 25.5 | 17.5 | 9.4 |
| Monkey | |||
| 1 | 86.5 | 86.5 | 86.5 |
| 0.1 | 86.5 | 86.5 | 86.5 |
| 0.01 | 84.2 | 84.1 | 83.0 |
| 0.001 | 47.5 | 42.8 | 37.3 |
| Human | |||
| 1 | 85.3 | 85.3 | 85.3 |
| 0.1 | 84.8 | 84.7 | 84.6 |
| 0.01 | 78.0 | 77.1 | 76.1 |
| 0.001 | 8.1 | n/aa | n/aa |
aThe predicted formaldehyde concentration exiting the nasal model was greater than the exposure concentration.
Normalized Sensitivity Coefficients Corresponding to Changes in the Formaldehyde Concentration at the Model Outlet as a Function of a 10% Change in Model Parameter
| Normalized sensitivity coefficient | |||
|---|---|---|---|
| Parameter | Symbol | 1 ppm | 1 ppb |
| Partition coefficient | Pt:a | 0.13 | 0.86 |
| Tissue diffusivity | Dt | 0.10 | 0.37 |
| Endogenous production rate constant | k0 | 0.01 | 0.50 |
| First-order rate constant | kf | 0.03 | 0.23 |
| Maximum saturable clearance rate | Vmax | 0.01 | 0.35 |
| Half-saturation concentration | Km | 0.02 | 0.25 |
| Blood flow rate | Qb | 0.02 | 0.01 |
| Blood concentration | Cblood | 0.01 | 0.01 |
| Tissue thickness | Lm+Lt+Lb | 0.06 | 0.20 |
| Normalized sensitivity coefficient | |||
|---|---|---|---|
| Parameter | Symbol | 1 ppm | 1 ppb |
| Partition coefficient | Pt:a | 0.13 | 0.86 |
| Tissue diffusivity | Dt | 0.10 | 0.37 |
| Endogenous production rate constant | k0 | 0.01 | 0.50 |
| First-order rate constant | kf | 0.03 | 0.23 |
| Maximum saturable clearance rate | Vmax | 0.01 | 0.35 |
| Half-saturation concentration | Km | 0.02 | 0.25 |
| Blood flow rate | Qb | 0.02 | 0.01 |
| Blood concentration | Cblood | 0.01 | 0.01 |
| Tissue thickness | Lm+Lt+Lb | 0.06 | 0.20 |
Normalized Sensitivity Coefficients Corresponding to Changes in the Formaldehyde Concentration at the Model Outlet as a Function of a 10% Change in Model Parameter
| Normalized sensitivity coefficient | |||
|---|---|---|---|
| Parameter | Symbol | 1 ppm | 1 ppb |
| Partition coefficient | Pt:a | 0.13 | 0.86 |
| Tissue diffusivity | Dt | 0.10 | 0.37 |
| Endogenous production rate constant | k0 | 0.01 | 0.50 |
| First-order rate constant | kf | 0.03 | 0.23 |
| Maximum saturable clearance rate | Vmax | 0.01 | 0.35 |
| Half-saturation concentration | Km | 0.02 | 0.25 |
| Blood flow rate | Qb | 0.02 | 0.01 |
| Blood concentration | Cblood | 0.01 | 0.01 |
| Tissue thickness | Lm+Lt+Lb | 0.06 | 0.20 |
| Normalized sensitivity coefficient | |||
|---|---|---|---|
| Parameter | Symbol | 1 ppm | 1 ppb |
| Partition coefficient | Pt:a | 0.13 | 0.86 |
| Tissue diffusivity | Dt | 0.10 | 0.37 |
| Endogenous production rate constant | k0 | 0.01 | 0.50 |
| First-order rate constant | kf | 0.03 | 0.23 |
| Maximum saturable clearance rate | Vmax | 0.01 | 0.35 |
| Half-saturation concentration | Km | 0.02 | 0.25 |
| Blood flow rate | Qb | 0.02 | 0.01 |
| Blood concentration | Cblood | 0.01 | 0.01 |
| Tissue thickness | Lm+Lt+Lb | 0.06 | 0.20 |
Estimates of maximum and average formaldehyde wall mass flux on nonsquamous epithelium were reported in each species at exposure concentrations of 0.001, 0.01, 0.1, and 1 ppm (Table 5). The maximum predicted formaldehyde fluxes at an exposure concentration of 1 ppm were 9068.9, 8570.2, and 10,183.8 pmol/(mm2 h) in the rat, monkey, and human, respectively; average fluxes were 503.0, 1680.7, and 1551.2 pmol/(mm2 h). For exposure concentrations > 1 ppm, maximum and average flux values were within 1% of fluxes that were linearly scaled from the 1 ppm exposure case. As exposure concentration decreased, estimated maximum and average flux decreased nonlinearly due to the effects of endogenous formaldehyde with a sharp decrease in flux at exposure concentrations < 2 ppb (Fig. 5). Similar behavior was observed in all species. At an exposure concentration of 1 ppb, the estimated maximum flux in the human model was positive (0.01 pmol/(mm2 h)) yet the average flux was negative (−0.04 pmol/(mm2 h)), indicating overall net desorption of formaldehyde. Flux contours in the rat, monkey, and human models at formaldehyde exposure concentrations of 1 ppm and 1 ppb are shown in Supplementary Data. Even though total nasal uptake was reduced at the 1 ppb exposure concentration, flux contours in the rat and monkey were still highly nonuniform with anterior-to-posterior gradients. Formaldehyde wall mass flux was negative (indicating desorption) throughout most of the human model at a 1 ppb exposure concentration.
Wall mass flux predictions of inhaled formaldehyde on nonsquamous epithelium in the rat, monkey, and human nasal CFD models: (A) maximum flux; (B) average flux.
Formaldehyde Wall Mass Flux Predictions on Nonsquamous Epithelium in the Rat, Monkey, and Human Models
| Exposure concentration (ppm) | Rat | Monkey | Human |
|---|---|---|---|
| Maximum flux (pmol/(mm2 h)) | |||
| 1 | 9068.9 | 8570.2 | 10,183.8 |
| 0.1 | 898.6 | 870.3 | 1017.1 |
| 0.01 | 82.9 | 83.4 | 93.1 |
| 0.001 | 1.4 | 4.1 | 1.0 × 10−2 |
| Average flux (pmol/(mm2 h)) | |||
| 1 | 503.0 | 1680.7 | 1551.2 |
| 0.1 | 49.8 | 169.4 | 148.8 |
| 0.01 | 4.6 | 15.7 | 13.5 |
| 0.001 | 0.1 | 0.8 | −4.0 × 10−2 |
| Exposure concentration (ppm) | Rat | Monkey | Human |
|---|---|---|---|
| Maximum flux (pmol/(mm2 h)) | |||
| 1 | 9068.9 | 8570.2 | 10,183.8 |
| 0.1 | 898.6 | 870.3 | 1017.1 |
| 0.01 | 82.9 | 83.4 | 93.1 |
| 0.001 | 1.4 | 4.1 | 1.0 × 10−2 |
| Average flux (pmol/(mm2 h)) | |||
| 1 | 503.0 | 1680.7 | 1551.2 |
| 0.1 | 49.8 | 169.4 | 148.8 |
| 0.01 | 4.6 | 15.7 | 13.5 |
| 0.001 | 0.1 | 0.8 | −4.0 × 10−2 |
| Exposure concentration (ppm) | Rat | Monkey | Human |
|---|---|---|---|
| Maximum flux (pmol/(mm2 h)) | |||
| 1 | 9068.9 | 8570.2 | 10,183.8 |
| 0.1 | 898.6 | 870.3 | 1017.1 |
| 0.01 | 82.9 | 83.4 | 93.1 |
| 0.001 | 1.4 | 4.1 | 1.0 × 10−2 |
| Average flux (pmol/(mm2 h)) | |||
| 1 | 503.0 | 1680.7 | 1551.2 |
| 0.1 | 49.8 | 169.4 | 148.8 |
| 0.01 | 4.6 | 15.7 | 13.5 |
| 0.001 | 0.1 | 0.8 | −4.0 × 10−2 |
| Exposure concentration (ppm) | Rat | Monkey | Human |
|---|---|---|---|
| Maximum flux (pmol/(mm2 h)) | |||
| 1 | 9068.9 | 8570.2 | 10,183.8 |
| 0.1 | 898.6 | 870.3 | 1017.1 |
| 0.01 | 82.9 | 83.4 | 93.1 |
| 0.001 | 1.4 | 4.1 | 1.0 × 10−2 |
| Average flux (pmol/(mm2 h)) | |||
| 1 | 503.0 | 1680.7 | 1551.2 |
| 0.1 | 49.8 | 169.4 | 148.8 |
| 0.01 | 4.6 | 15.7 | 13.5 |
| 0.001 | 0.1 | 0.8 | −4.0 × 10−2 |
Average formaldehyde wall mass flux estimates were computed in regions where cell replication and DPX were measured in rats and monkeys (Table 6). Predicted flux values displayed an anterior-to-posterior effect, with higher fluxes predicted in regions located in the more anterior sections of the nasal passages. This anterior-posterior flux gradient is consistent with cell proliferation measurements in rats and monkeys where higher rates were observed in the more anterior regions of the nose (Monticello et al., 1989, 1996). The reported flux results were not averaged over a breathing cycle as in Kimbell et al. (2001b) due to uncertainties in formaldehyde lung uptake due to endogenous formaldehyde. Mass balance errors for formaldehyde uptake simulations, computed as ([mass entering nostril(s)] – [mass exiting outlet] – [mass absorbed by airway walls])/[mass entering nostrils], were < 2% in all cases. Due to the low mass balance errors, the remaining mass was not redistributed across the nasal surfaces, as was done in Kimbell et al. (2001b). Average flux values in DPX and cell replication regions for inhalation exposure concentrations of 0.7, 2, 6, 10, and 15 ppm can be linearly scaled from the 1 ppm exposure case because the presence of endogenous formaldehyde did not have any significant effect on tissue dose at these concentration levels.
Formaldehyde Wall Mass Flux Predictions in DPX Regions and Cell Proliferation Sites at a 1 ppm Exposure Concentration
| Flux (pmol/(mm2 h)) | ||||
|---|---|---|---|---|
| Rat | Monkey | |||
| DPX regions | ||||
| High tumor | 1728.7 | ALWS | 2136.3 | |
| Low tumor | 493.4 | MT | 2946.7 | |
| NP | 483.0 | |||
| Cell proliferation sites | ||||
| ALM | 2213.0 | Level B | 2832.8 | |
| PLM | 653.5 | Level C | 1271.9 | |
| ADS | 374.7 | Level D | 451.8 | |
| MMT | 846.7 | Level E | 351.5 | |
| AMS | 958.1 | |||
| PMS | 719.4 | |||
| Flux (pmol/(mm2 h)) | ||||
|---|---|---|---|---|
| Rat | Monkey | |||
| DPX regions | ||||
| High tumor | 1728.7 | ALWS | 2136.3 | |
| Low tumor | 493.4 | MT | 2946.7 | |
| NP | 483.0 | |||
| Cell proliferation sites | ||||
| ALM | 2213.0 | Level B | 2832.8 | |
| PLM | 653.5 | Level C | 1271.9 | |
| ADS | 374.7 | Level D | 451.8 | |
| MMT | 846.7 | Level E | 351.5 | |
| AMS | 958.1 | |||
| PMS | 719.4 | |||
Formaldehyde Wall Mass Flux Predictions in DPX Regions and Cell Proliferation Sites at a 1 ppm Exposure Concentration
| Flux (pmol/(mm2 h)) | ||||
|---|---|---|---|---|
| Rat | Monkey | |||
| DPX regions | ||||
| High tumor | 1728.7 | ALWS | 2136.3 | |
| Low tumor | 493.4 | MT | 2946.7 | |
| NP | 483.0 | |||
| Cell proliferation sites | ||||
| ALM | 2213.0 | Level B | 2832.8 | |
| PLM | 653.5 | Level C | 1271.9 | |
| ADS | 374.7 | Level D | 451.8 | |
| MMT | 846.7 | Level E | 351.5 | |
| AMS | 958.1 | |||
| PMS | 719.4 | |||
| Flux (pmol/(mm2 h)) | ||||
|---|---|---|---|---|
| Rat | Monkey | |||
| DPX regions | ||||
| High tumor | 1728.7 | ALWS | 2136.3 | |
| Low tumor | 493.4 | MT | 2946.7 | |
| NP | 483.0 | |||
| Cell proliferation sites | ||||
| ALM | 2213.0 | Level B | 2832.8 | |
| PLM | 653.5 | Level C | 1271.9 | |
| ADS | 374.7 | Level D | 451.8 | |
| MMT | 846.7 | Level E | 351.5 | |
| AMS | 958.1 | |||
| PMS | 719.4 | |||
The ranges of formaldehyde wall mass flux estimates in the rat, monkey, and human nasal models were partitioned into 20 evenly spaced flux bins, where bin 1 corresponds to the lowest fluxes and bin 20 corresponds to the highest fluxes. The percentage surface area of nonsquamous epithelium was calculated for each flux bin at an exposure concentration of 1 ppm in each species (Fig. 6). Average flux values in each bin and the corresponding surface areas are provided in Supplementary Data. Flux bins allow for a comparison of the distribution of formaldehyde absorption patterns among rats, monkeys, and humans. The bin corresponding to the highest flux values (bin 20) contained < 0.2% of the nonsquamous surface area of each species. At an exposure concentration of 1 ppm, the rapid absorption of inhaled formaldehyde in the anterior nose quickly reduced concentration levels in the nasal airspace, leading to a sharp gradient in flux distribution. This led to a large portion of the nasal surfaces receiving low flux, which was most evident in the rat with 80% of the surface area located in the lowest flux bin (bin 1).
Partitioning of nonsquamous nasal surface area into 20 flux bins based on predicted formaldehyde flux in the rat, monkey, and human models at an exposure concentration of 1 ppm.
At lower exposure concentrations, the distribution of flux throughout the human nasal passages changed dramatically (Fig. 7). As exposure concentration decreased below 1 ppm, the highest surface area flux bins shifted to the higher bins (those with larger fluxes). At an exposure concentration of 0.001 ppm, over 99% of the surface area was located in bin 20. As exposure concentration decreased, the average flux in each bin decreased and also contained more negative flux values, indicating larger regions in the human nose where formaldehyde was predicted to desorb from nasal tissues due to endogenous tissue concentrations (Supplementary Data).
Partitioning of nonsquamous nasal surface area in the human model into 20 flux bins based on predicted formaldehyde flux at exposure concentrations of 0.001, 0.01, and 0.1 ppm.
DISCUSSION
Formaldehyde is a soluble and reactive gas that, at high exposure concentrations, is rapidly absorbed by the upper respiratory tract, causes nasal tumors in laboratory animals and may cause nasopharyngeal cancer in humans. Although human exposure is widespread, exposure levels are typically in the low parts per billion. Risk estimates of formaldehyde inhalation in humans must account for the dose-response behavior at concentrations well below where effects were observed in animals. Formaldehyde is also an endogenous chemical that is present as a metabolic intermediate in all living cells. Risk assessments for endogenous compounds such as formaldehyde that rely on estimates of target tissue dose are complicated by the presence of background concentrations in these tissues. Toxic effects may be exerted at high doses yet low-level exposures may not appreciably contribute to tissue concentrations, potentially leading to dose-dependent transitions and nonlinear dose-response behavior. Accounting for endogenous formaldehyde in respiratory tissues while assessing the contribution from inhaled exogenous formaldehyde is necessary for an accurate extrapolation of high-dose effects in animals to low-dose exposures in humans.
Andersen et al. (2010) developed a compartmental pharmacokinetic model for formaldehyde that accounted for formaldehyde production, interaction of formaldehyde with glutathione (GSH), and saturable metabolism of the formaldehyde thioacetal to formic acid. Nonlinear effects in nasal tissues occur due to saturable metabolism and the finite concentration of GSH to complex tissue formaldehyde. Earlier models examining DNA binding of inhaled mass-labeled formaldehyde used saturable metabolism, but did not explicitly include GSH (Casanova et al., 1989). In the present work, we included saturable oxidation of formaldehyde in the tissue pharmacokinetic model without explicitly accounting for GSH. This pharmacokinetic structure was present throughout the mucus-coated epithelial surface of the nasal cavity and allowed predictions of net flux from tissue to air or from air to tissue in the presence or absence of inhaled formaldehyde in the airstream. This approach also allowed for prediction of tissue formaldehyde concentrations, although inclusion of additional formaldehyde kinetics such as those described above may be needed to obtain more accurate estimates of formaldehyde tissue levels and amounts entering the blood.
Earlier formaldehyde dosimetry modeling efforts used anatomically accurate CFD models of the nasal passages of a rat, monkey, and human to predict localized formaldehyde uptake patterns and analyze interspecies differences in the spatial distribution of formaldehyde flux (Kimbell et al., 2001a,b). Results from these studies showed significant interspecies differences in formaldehyde absorption patterns that were primarily due to differences in nasal anatomy and ventilation between species. Correlation of the site specificity of formaldehyde-induced nasal lesions with regions of high flux predicted from the CFD models was instrumental in assessing risk from exposure to inhaled formaldehyde using biologically based computational models (Conolly et al., 2003, 2004). Although these earlier CFD efforts provided significant findings on the heterogeneity of formaldehyde absorption, they were unable to account for physiological background concentrations of formaldehyde due to the use of a constant mass transfer coefficient as a boundary condition that was calibrated to nasal uptake measurements in rats conducted at high exposure concentrations.
In this study, the boundary conditions governing formaldehyde absorption in the nasal CFD models of the rat, monkey, and human were modified by including tissue partitioning, species-specific saturable metabolism, first-order clearance, and endogenous formaldehyde production. Using this approach, formaldehyde nasal absorption was determined from its physicochemical characteristics and pharmacokinetics in the nasal mucosa. This model structure allowed for localized formaldehyde absorption from the nasal airspace into tissues or desorption from the tissues into the airway lumen depending on predicted air concentrations from the CFD model and tissue formaldehyde concentrations from the epithelial model. Using this methodology, the parameters affecting formaldehyde absorption (e.g., partitioning and endogenous tissue levels) are directly implemented into the CFD models allowing for the study of the effect of pharmacokinetic processes in nasal tissues on formaldehyde nasal uptake. Similar approaches using a virtual epithelial structure linked to CFD models (without endogenous production) have been used to study uptake of other reactive gases in the respiratory tract (Corley et al., 2012; Schroeter et al., 2006, 2008). These epithelial structures have also been used in physiologically based modeling efforts to estimate nasal tissue concentrations of inhaled gases (Frederick et al., 1998; Teeguarden et al., 2008).
The use of formaldehyde tissue partitioning in place of a constant mass transfer boundary condition led to significant differences in nasal uptake and regional flux predictions compared with the results from Kimbell et al. (2001b). At an exposure concentration of 1 ppm, formaldehyde nasal uptake predictions were higher than those predicted by Kimbell et al. (2001b), which were 90% in the rat, 67% in the monkey, and 76% in the human. Formaldehyde partitioning led to higher absorption rates in the anterior nasal passages when compared to the mass transfer coefficient approach and therefore increased overall nasal uptake. Maximum flux values on nonsquamous epithelium predicted in the rat and human were higher in this study than those predicted by Kimbell et al. (2001b). Maximum flux values in the monkey model were comparable to those in the Kimbell study. Predicted flux values at cell proliferation sites in the rat were lower in this study, except for the ALM region which is located in an area of high formaldehyde flux. For all of the other regions, the highest formaldehyde flux values were located in more anterior regions of the nose, so the higher flux in the anterior nose predicted using the formaldehyde partitioning approach resulted in lower average flux values in these regions. In the cell replication sites in the monkey, higher flux values were predicted in this study in the more anterior levels and lower flux values in the more posterior levels due to the lower air concentration levels.
Other differences between the models used in this study and the Kimbell et al. (2001b) models include smoother surface contours, a different human model based on high-resolution CT data, higher-density numerical meshes, and nostril airflow profiles derived from the transnasal pressure drop. None of these model differences led to significant differences in predicted formaldehyde uptake. Garcia and colleagues (2009) showed that variations in model geometry among seven subjects led to a < 1.6-fold difference in predicted wall mass flux of soluble and reactive vapors. “Adult 5” in the Garcia study was derived from the same CT data as the human model in this study. Additionally, the smoother surfaces and higher mesh density led to smaller mass balance errors, but did not significantly affect predictions of overall uptake (Kimbell et al., 2010).
At exposure concentrations > 0.5 ppm, formaldehyde nasal uptake was controlled by tissue partitioning, even at concentrations as high as 10 ppm, as evidenced by the linear relationship between exposure concentration and wall mass flux. The presence of endogenous formaldehyde did not have any quantitatively significant effect on nasal uptake or flux values for exposure concentrations > 0.5 ppm. However, endogenous formaldehyde in nasal tissues did affect nasal uptake at lower exposure concentrations, most notably at air concentrations < 10 ppb. At a concentration of 1 ppb, predicted formaldehyde uptake was greatly reduced, to the point that formaldehyde desorption from nasal tissues was predicted in humans. Exposure concentrations < 1 ppb also yielded net desorption of inhaled formaldehyde from nasal tissues in the rat and monkey models.
Respiratory tract absorption of inhaled formaldehyde occurs from the diffusive transfer of vapor molecules from regions of high to low concentrations. Gases with high water solubility rapidly partition into respiratory tissues, leading to high uptake. Likewise, gases that are reactive or metabolized in nasal tissues are rapidly absorbed because the deposited vapor molecules are cleared from the air:tissue interface, thereby retaining a large concentration gradient with the nasal airspace. With endogenous chemicals such as formaldehyde that are naturally present in respiratory tissues, the background tissue concentrations lower the concentration gradient at the air:tissue interface, potentially leading to reduced net absorption. At exposure concentrations > 0.5 ppm, formaldehyde air concentrations were large enough that the presence of endogenous formaldehyde had little effect. But at low exposure concentrations, the presence of formaldehyde in the tissues due to endogenous production created a smaller concentration gradient, thereby reducing the rate of absorption.
Formaldehyde is a ubiquitous air pollutant present in indoor and outdoor environments, but is also an essential metabolite in all living cells. Both endogenous and exogenous formaldehyde need to be considered in risk assessment. The National Academies of Science review of the U.S. EPA draft formaldehyde risk assessment recently concluded that the presence of endogenous formaldehyde complicates the risk assessment due to formaldehyde inhalation (NRC, 2011). By simulating formaldehyde absorption or desorption due to material properties and kinetics in the nasal mucosa, we were able to quantify the reduced net flux of inhaled exogenous formaldehyde due to the presence of endogenous formaldehyde. These results can help inform human cancer risk estimates from formaldehyde exposure, most notably at low exposure levels. The nasal uptake predictions suggest that risk estimates based on formaldehyde dosimetry in the upper respiratory tract may be overly conservative due to net desorption of formaldehyde at ambient air concentrations < 1 ppb. The CFD simulation results at high exposure concentrations are consistent with experimental data showing high upper respiratory tract uptake in rats at exposure concentrations > 1 ppm. Although further studies measuring formaldehyde nasal uptake would have to be conducted to verify the CFD results at lower exposure concentrations, the CFD predictions are based on mass transfer theory of vapors and are consistent with studies showing exhaled breath concentrations of several parts per billion, which would indicate off-gassing of formaldehyde from respiratory tissues.
It should be noted that the CFD simulations presented in this study used steady-state airflow and formaldehyde transport assumptions, which assumed equilibrium conditions in the air and tissue phases. Transient simulations over multiple breaths would have to be conducted in a complete respiratory tract model to analyze the potential for formaldehyde buildup in tissues and to assess the effects of the steady-state assumption on endogenous formaldehyde and exhaled breath levels.
FUNDING
Research Foundation for Health and Environmental Effects.
The authors would like to thank Dr Todd Yokley for providing de-identified CT scans of radiologically normal human nasal cavities.
REFERENCES
Author notes
Disclaimer: The information in this document has been subjected to review by the National Health and Environmental Effects Research Laboratory, Office of Research and Development, U.S. Environmental Protection Agency and approved for publication. Approval does not signify that the contents reflect the views of the Agency, nor does mention of trade names or commercial products constitute endorsement or recommendation for use.
Comments