Historical perspective

After more than a decade of model development (Jarabek et al., 1990; Miller et al., 1983, 1988; USEPA, 1988), USEPA encouraged the application of regional deposition models such as the Regional Deposited Dose Ratio model for conducting exposure assessments of inhaled particulate aerosols (USEPA, 1994). Despite this early commitment to regional deposition modeling, risk assessment guidance provided by USEPA through 2022 makes use of integrated risk information system (IRIS)-derived reference concentrations (RfCs) for assessment of noncancer impacts that might result from inhalation exposure to a chemical (USEPA, 1994). Risks from exposure to carcinogens is similarly indexed to unit risk estimates for cancer after inhalation—a static estimate of the lifetime cancer risk associated with 1 μg/m³ of airborne chemical exposure (USEPA, 2005). Established in 2005, unit risk estimates do not accommodate variations in deposition and exposure created by changes in the aerodynamic characteristics of particulate aerosols.

Although USEPA recognized the advantages of adjusting for species-specific differences in the respiratory tract and acknowledged that estimates of regional deposition have utility, body weight or estimated respiratory tract surface area have remained the recommended normalizing factors for extrapolating risk of systemic impacts from animals to humans after inhalation (Foureman & Kenyon, 2006; USEPA, 1994) unless physiologically based pharmacokinetic (PBPK) models were available to predict systemic uptake and distribution (USEPA,1994, 2007a, 2009). Physiologically based pharmacokinetic models are available for extensively studied MMEs such as lead or cadmium (Nordberg & Fowler, 2018) but have not been developed for most MMEs. Indeed, even for MMEs as well studied as lead, successful exposure assessment models that incorporate regional pulmonary deposition patterns into estimates of systemic exposure have yet to be developed. For example, the recently released All-Ages Lead Model (USEPA, 2019a), developed over several decades as a refined successor to the integrated exposure uptake biokinetic (IEUBK) model (USEPA, 2007b), lacks the ability to use particle size information to predict local and systemic impacts after inhalation exposure.

Equations for calculating the regional deposition of particulate aerosols in the lung were compiled into the provisional Regional Deposited Dose Ratio computer model by USEPA almost three decades ago (USEPA, 1994). Despite this early commitment to regional deposition modeling, exposure assessment guidance currently provided by USEPA in online resources (e.g., the inhalation module of the USEPA's EXPOsure toolBOX or Expobox) does not include estimates of regional deposition patterns or rates of clearance (USEPA, 2022). Citations and links are provided by USEPA to approximately three dozen tools for inhalation exposure assessment, but there are no models cited that explicitly incorporate regional deposition rates into estimates of inhalation exposure. Instead, USEPA retained the use of inhalation reference concentrations or inhalation unit risk estimates that are normalized via body weight to “account for inhalation rates in the development of dose–response relationships” and in accordance with allometric scaling equations adopted by USEPA (1988). Normalization to body weight is further employed by the agency in interspecies extrapolation (USEPA, 1994) and the calculation of human equivalent concentration (HECs) that extrapolate from doses producing effects in experimental animals to the exposure that would be experienced by humans under similar exposure conditions (USEPA, 2022). Implicit in this normalization is the assumption that body weight is an acceptable surrogate to compensate for species-specific differences in breathing rates or the surface area of target tissues (Foureman & Kenyon, 2006).

Although the calculation of regional deposition patterns was important for assessing local and systemic exposure risk in 1994, subsequent guidance provided by USEPA through 2022 provides scant mention of the benefits that could be realized with computer simulations of respiratory function.

This is not to say regional deposition modeling lacked merit in USEPA—individual USEPA efforts supported research focused on aerodynamic determinants of particle deposition (USEPA, 2002), assessment of exposure to susceptible subgroups in the general population (USEPA, 2007a, 2007b), and the development of modeling software for regional deposition (Kim & Hu, 2017; USEPA, 2021a, 2021b). Unfortunately, the detailed reports developed by most of these efforts are not among the online resources made available by the agency. The utility of deposition modeling had been evident in USEPA's evaluation of some air pollutants, provided that one had advanced knowledge of the agency's procedures for the conduct of technical reviews. For example, USEPA efforts to develop health protective standards limiting population exposure to particulate matter (USEPA, 2019b) paid attention to regional patterns of particle deposition with tabular summaries being provided in the supporting criteria document of more than a dozen approaches to assess inhalation exposure. However, computer simulation tools for assessing regional deposition patterns are summarized only in passing. This surprisingly limited profile for regional deposition modeling is justified on the basis that methods for assessing regional deposition had been reviewed in a previous criteria document (USEPA, 2019b).

In contrast, multiple agencies and private sector interests have promoted the development of models for deposition after inhalation exposure. Historical precedents were set more than 100 years ago in evaluations of occupational exposure to coal dust (Schins & Borm, 1999) and asbestos (Lippmann, 1988). Table 1 summarizes some of the tools developed for prediction of inhalation exposure. Existing models range from those developed to enhance the precision of risk assessment such as the multipath particle deposition model (MPPD; Asgharian & Price, 2009) to models that support the design and manufacture of delivery systems for aerosolized pharmaceuticals (Ahookhosh et al., 2020).

Other scientific bodies have placed a strong public focus on regional deposition models. With the development of nuclear technology in the 1940s, standardized values for parameters descriptive of the deposition of insoluble radionuclide aerosols were needed to set exposure limits in the occupational setting (International Commission on Radiological Protection [ICRP], 1979). The work of Weibel and colleagues is particularly noteworthy for the detailed casting studies that permitted development of the first realistic models of the human respiratory tract (Weibel, 1963). Over the next several decades, deposition modeling increased in sophistication, defining three different anatomical compartments of physiological and morphological significance (alveolar regions of the deep lung, bronchial and tracheal tissues of the thoracic region, and extrathoracic regions containing the nasal and pharyngeal tissues). Estimates of clearance rates specific to each compartment were also developed, increasing the precision of exposure estimates and the prediction of health impacts subsequently associated with radionuclide exposure (ICRP, 1979, 1993, 1995, 2002, 2017). The ICRP models make extensive use of observational data for pulmonary deposition in humans and have served as the basis for other models for pulmonary deposition.

The definition of distinct regions of the respiratory tract facilitated efforts to align toxicologically relevant exposures with the target cell populations for adverse impacts. The initial inhalation risk assessments of ICRP were indexed to the total blood-filled mass of the lungs and assumed homogeneous exposure throughout pulmonary tissues (ICRP, 1979). The definition of different regions of the airways permitted exposure assessments indexed to regional deposition estimates. Thus, estimates of inhalation exposure could be developed that were specific to the airway regions and cell types at risk. Moreover, particle deposition was determined to be relatively independent of chemical composition except to the extent that particle composition influenced aerosol aerodynamic properties. For example, particle density can vary with chemical speciation and thereby influence the mass-based mean aerodynamic diameter (MMAD) and the most likely regions of particle penetration after inhalation. Similarly, hygroscopic particles can accumulate water vapor, or particles can agglomerate, to alter MMADs and particle deposition patterns. However, it emerges that specification of a relatively limited set of aerosol properties such as density, size, shape, and statistical descriptors of the particle size distribution are adequate to make species-specific predictions of the regional deposition pattern that will be associated with an aerosol (Asgharian & Price, 2009; ICRP, 1979; USEPA, 1994, 2021a).

The aerodynamic properties of particulate aerosols combine with species-specific airway morphology and breathing patterns to determine initial particle deposition (Asgharian & Price, 2009; ICRP, 1979; USEPA, 1994, 2021a; Vincent, 2007). Respiratory tract morphology determines the specific conditions of air pressure, flow rate, flow direction, flow volume, temperature, and humidity that govern the likelihood of sedimentation and impaction events in different respiratory tract regions. Predictive deposition modeling can be performed for animal species that have adequate characterization of the structural and functional determinants of breathing. Accommodations can also be made for species-specific differences in breathing patterns such that impacts in obligate nose-breathers such as the rat can be compared and contrasted with predicted impacts associated with human oronasal breathing patterns. Finally, multiple aspects of pulmonary function are altered by factors such as age, sex, and levels of physical activity (Miller et al., 1988; USEPA, 2007a, 2007b). Accommodations can be made for such factors although doing so increases the complexity of modeling.

Mechanisms for the clearance of particles from the respiratory tract vary with the region of deposition. Upper airway deposition can be expelled by mechanical processes (e.g., sneezing and/or coughing) or ingested (by swallowing) after deposition has occurred. Gastrointestinal uptake kinetics determine the rate at which swallowed material enters systemic circulation. Dissolution in, and uptake from, physiological fluids present in the upper airways can also occur. Material released from particles by dissolution will be available for uptake via passive diffusion or through active transport systems (Nordberg & Fowler, 2018). Removal from the upper airways is relatively rapid with a removal half-time on the order of minutes (ICRP, 1994; USEPA, 2007c). Material deposited in the thoracic region can be subject to removal by mucociliary clearance (with subsequent translocation to the GI tract), dissolution in pulmonary fluids followed by systemic uptake, or uptake by cells lining the respiratory tract. Removal half-times on the order of minutes to hours would be expected (ICRP, 1979, 2017). Alveolar deposition will be slowly removed by dissolution in interstitial fluids or cellular uptake of dissolved or phagocytized material (by macrophages or alveolar epithelium) over a time frame that can extend from days or weeks for soluble compounds to months or longer for sparingly soluble substances.

Regional deposition models from 1995 to 2023

There have been significant advances in model development over the past several decades. Table 1 summarizes some of these modeling advances, many of which had been stimulated by the development of aerosol delivery inhalers for administering pharmaceuticals. The types and utility of different inhalation models have been recently reviewed in depth (Ahookhosh et al., 2020; Longest & Holbrook, 2012) and only a summary overview is provided here.

Initial deposition models were semi-empirical in nature, aligning observational data on regional deposition with aerosol properties that govern depositional processes such as impaction. Initial modeling efforts of ICRP (1979) employed such approaches. An increase in model sophistication and utility was realized with the development of single path models such as the ICRP or MPPD models that predict the average deposition at each branch of the respiratory tract in accordance with preexisting estimates of the correlation between deposition and processes such as impaction, sedimentation, and diffusion. Outputs for such models included estimates of total deposition as well as regional deposition estimates for the nasopharynx, the tracheobronchial region, and the alveolar region of the lung. Models such as MPPD can further employ a Monte Carlo lung geometry simulation approach to predict deposition from the nose and/or mouth to the deep lung. Adaptations of such models further permitted the prediction of deposition in diseased lungs and in pediatric populations (Longest & Holbrook, 2012). The MPPD model is continually refined, most recently to incorporate estimates of alveolar mixing and multiple breaths on predicted rates of deposition in humans (Asgharian et al., 2022). Outputs from MPPD now include estimates of particle clearance time from different respiratory tract regions although the origin of these estimates is often unclear. Over the past several decades, MPPD has also been updated to permit regional deposition modeling for multiple species, including humans, rats, mice, hamsters, and rabbits (Miller et al., 2016). This versatility has combined with the computational ease of the MPPD software interface to promote the use of this model in a variety of risk assessment applications.

The modeling of simple monodisperse aerosols has demonstrated the complexity of pulmonary function, the significant extent of interindividual variability, and the difficulties inherent in predicting deposition for even simple monodisperse aerosols. Exposure variation with age, disease state, or breathing patterns can be significant—this technical complexity introduces uncertainty that will likely be the focus of future model improvements. Despite these complexities, existing models have demonstrated good utility in predicting regional deposition for simple environmental aerosols. The modeling of complex aerosols, such as those generated by nebulizers and inhalers for the delivery of pharmaceutical preparations, has been more problematic. Modeling the delivery of pharmaceutical aerosols composed of either liquid droplets or fine dry powders is critical to maximizing the efficacy and safety of aerosol drug delivery systems. Initial regional deposition models lacked the ability to account for important determinants of drug delivery, such as changing droplet size, droplet charge, or inhaler spay velocity, and spurred the development of computational fluid dynamics (CFD) technology incorporating three-dimensional models of both the respiratory tract and drug delivery devices (Longest & Holbrook, 2012). Computational fluid dynamics modeling has been facilitated by advances in medical imaging and casting that have permitted the development of realistic models for the respiratory system (Ahookhosh et al., 2020). Computational fluid dynamics models can make more precise predictions of regional drug delivery but are computationally intensive and require an array of detailed input parameters to define the aerosol, the delivery device, and the receiving respiratory tract. Once developed, CFD models can be adapted to account for different airway geometries although their complexity is such that a given modeling exercise will often develop predictions for only a portion of the respiratory tract. A version of MPPD that incorporates CFD modeling has been developed for more accurate prediction of regional deposition in the human lung under a broader array of conditions (Kuprat et al., 2021). This CFD version of MPPD is awaiting validation but may enhance MPPD modeling of human exposure. Extension of CFD–MPPD modeling to other species is expected.

Several CFD models are included in Table 1 to illustrate the variety of models now available (Ahookhosh et al., 2020; Koullaois et al., 2018), but the computational intensity associated with these models has thus far limited their application in hazard and risk assessment. Research efforts within USEPA promoted the development of CFD models of human lungs (Isaacs et al., 2006), and CFD modeling was further employed within USEPA to refine understanding of aerosol deposition in human airways (Fleming et al., 2006). However, applications in the study of environmental agents appear to have been limited. Instead, primary efforts seem to have focused on use of CFD modeling for studying drug delivery by pharmaceutical nebulizers in the treatment of human pulmonary disease (Ali et al., 2008). The pharmaceutical applications of CFD models have maintained a focus on the human lung although the introduction of physiologically based pharmokinetic (PBPK) modeling has begun to promote the use of PBPK–CFD modeling approaches to supporting animal-to-human data extrapolation. The number of computational fluid dynamics models appropriate to modeling exposure and health impacts in experimental animals is limited but should contribute to future refinements in inhalation exposure assessment (Beckett et al., 2019; Gloede et al., 2011).

Inhalation of MMEs

The USEPA's derivation of inhalation reference concentrations (USEPA, 1994) provides an extensive review of the factors that influence the deposition and clearance of aerosols. Factors such as age, sex, activity level, and breathing patterns combine with the aerodynamic properties of an aerosol to determine the probability of deposition in functionally distinct regions of the respiratory tract in accordance with basic processes such as sedimentation and impaction. The deposition patterns of MMEs in the respiratory tract determine the target cells at risk for toxicity and the pathways available for the local or systemic uptake of an MME. As noted earlier, extrathoracic deposition in the nose and throat will either be expelled or swallowed. Mucociliary clearance transports upper airway deposition to the GI tract. The probability of deep lung deposition increases as aerosol particle d_ae,50 become smaller than 10 µm. Systemic exposure after MME inhalation will be the summation of material absorbed from the GI and respiratory tracts. Depending on the MME under evaluation, there can be significant differences in the uptake rates from the GI and respiratory tracts. Risk of local impacts would generally be related to the amount of material deposited in the respiratory tract normalized to the surface area of the receiving tissues.

Although inhalation exposures are not unique to MMEs and can occur with many organic and/or inorganic substances, inhalation is often the principal route of human exposure to MMEs. The occupational setting can provide the most significant airborne exposure intensity to MMEs and provides examples of the type of issues that can be addressed with regional deposition modeling. Although it can be rationalized that USEPA's mandate does not extend to the occupational setting, the basic principles evident in the following MME case studies are applicable to inhalation exposures in the workplace, from consumer products and/or environmental exposures. Indeed, agencies such as the ECHA conduct integrated assessments that evaluate occupational, consumer, and indirect environmental exposures. These integrated assessments provide much of the MME case study material.

Case study: Zinc

Multiple aspects of metal toxicokinetics are modulated by capacity-limited binding to “carrier systems” specific to different MMEs. This is especially true for MMEs that are essential trace elements (ETEs) required for maintenance of optimal health, development, and reproductive function. Blood and tissue concentrations of ETEs are under tight homeostatic control that maintains ETE blood concentrations within the range consistent with good health (Becking, 1998; IPCS, 2002). These homeostatic controls operate on rate-limiting processes that modulate systemic ETE levels by (1) increasing uptake when ETE deficiency occurs and decreasing uptake when systemic levels become excessive, (2) increasing or decreasing excretion to compensate for deficiency or exposure excess, and/or (3) modulating the presence and distribution of systemic carrier proteins that bind and transport ETEs. Such homeostatic controls maintain optimal systemic levels of ETEs by compensating for temporal variations in external ETE levels present from environmental, diet, or occupational sources. Blood concentrations for an essential MME are thus relatively constant over a broad range of exposures and cannot be used to assess levels of external exposure.

The Existing Substances Program of the European Union (EU) undertook a risk assessment of zinc and zinc compounds in the late 1990s, evaluating occupational, professional, consumer, and environmental exposures associated with the production, use, and recycling of zinc (Battersby & Boreiko, 2004; Bodar et al., 2005; Boreiko, 2010; European Commission, 2008). European Union risk assessments are prepared by a multistakeholder process that merges the technical input of public and private sector interests. The Zinc Risk Assessment initially evaluated inhalation risk by applying simplified assumptions of homogeneous exposure throughout the respiratory tract, assuming a “worst case” deposition rate of 50% and a “worst case” uptake of 100%. The resulting exposure assessment concluded that systemic exposure to zinc would exceed health protective limits in many settings (Boreiko, 2010). In addition to systemic health risk, the clinical nature of which could not be defined, the risk of welding fume fever was suggested to be present in most occupational settings. Metal fume fever is a short-term, self-limiting, flu-like illness associated with the inhalation of hot metal fumes capable of alveolar deposition (Drinker et al., 1927). Curiously, welding was one of the few occupational scenarios that was not associated with excess risk of welding fume fever in the initial predictions of the EU Zinc Risk Assessment.

The conclusions of the Zinc Risk Assessment shifted significantly when regional deposition estimates were applied to exposure assessment. Using an early version of MPPD, it was determined that 80% or more of the mass in zinc oxide occupational aerosols exhibited an inhalable size range that would not yield significant deep lung deposition. After correction for the limited extent of alveolar deposition that would occur, metal fume fever risk was no longer a risk in most settings (Battersby & Boreiko, 2004). The initial Zinc Risk Assessment drafts also concluded that the pulmonary deposition of zinc would result in systemic zinc uptake in excess of that compatible with good health. Concerns over possible systemic impacts were also subsequently resolved through MPPD modeling (Boreiko, 2010). The levels of zinc in biological fluids amenable to sampling for purposes of exposure assessment are relatively constant under conditions of deficiency and/or exposure excess until extreme deficiency or extreme exposure excess occurs. Biomonitoring the levels of zinc in blood was not a reliable indicator of exposure status and could not be used to validate the excess zinc exposure risk suggested by the initial deposition and uptake assumptions of the Zinc Risk Assessment. However, subsequent MPPD modeling indicated that deposition from occupational exposure would be predominantly extrathoracic (Boreiko, 2010). Clearance to the GI tract would be rapid and, if excessive, would exhibit sharply limited systemic uptake as homeostatic controls downregulated zinc-active transport systems. As a result, risk of excess systemic exposure to zinc would not be induced by inhalation exposure. Similar strategies should be applicable to any MME that is an essential element or is otherwise under the control of homeostatic mechanisms.

In the preceding example, simplified default assumptions about particle deposition after inhalation resulted in the prediction of systemic risk where no risk was present and predicted lack of risk for acute toxicity (welding fume fever) when risk was present. Indeed, had more advanced inhalation modeling tools not been brought to bear, conclusions could have been reached that would have restricted zinc exposure to low intake levels that would induce severe zinc deficiency that adversely affected human growth, reproduction, immune function, and cognitive development. Although much of the preceding was conducted in an analysis of occupational exposure, regional deposition estimates contribute to analysis of potential exposure risk in a range of consumer and environmental exposure scenarios.

Case study: Antimony

Inhalation exposure to diantimony trioxide produces lung tumors in rats and mice (National Toxicology Program of the United States [NTP], 2017). Initial studies had only been conducted in rats and produced inconsistent lung tumor responses hypothesized to be mediated by “overload” of pulmonary clearance mechanisms (Groth et al., 1986; Newton et al., 1994). The National Toxicology Program of the United States (NTP, 2017) exposed rats to diantimony trioxide for two years at airborne concentrations of 3, 10, and 30 mg/m³. Dose-dependent increases in lung tumors were observed in both male and female rats. In contrast, Newton et al. (1994) had not observed lung tumors in rats exposed to diantimony trioxide at airborne concentrations of 0.055, 0.511, or 4.5 mg/m³ for one year followed by one year of observation. Although the two-year exposure employed by NTP would probably have had greater impact than the one-year exposure employed by Newton et al. (1994), an exposure of 12 months had been sufficient to induce tumors in other rat studies of diantimony trioxide (Watt, 1983).

A comparison of the predicted deposition patterns expected for the aerosols employed by NTP and by Newton et al. suggests that differences in study outcome might be related to aerosol particle size distribution. The NTP studies used diantimony trioxide aerosols with MMAD of 1.1 μm and a geometric standard deviation (GSD) of 1.95, whereas those of Newton et al. employed aerosols with MMAD of 5.06 μm and a GSD of 2.13. With a few simplifying assumptions regarding aerosol characteristics, expected rates of deposition in the two studies can be compared. Table 2 presents the predicted airborne diantimony trioxide regional deposition fractions for the two different studies calculated using MPPD 3.02. The studies of Newton et al. (1994) used a coarser aerosol that would yield lower deposition rates in the target tissues of the pulmonary and tracheobronchial regions. Thus, although the highest airborne concentration used by Newton et al. (1994) may have been 50% greater than the lowest aerosol concentrations producing tumors in the rat lung in the studies of NTP, the pulmonary and tracheobronchial deposition rate in the Newton et al. (1994) studies was only 5%–7% of the deposition rate in the NTP studies. The negative outcome of the Newton et al. (1994) study likely reflected lower exposure of target tissues.

Although not a factor in the outcome of cancer bioassays, the uptake kinetics for antimony highlight yet another issue for MMEs. Although inhalation exposure can be the dominant exposure pathway for antimony compounds, many antimony aerosols are sufficiently coarse so as to result in extrathoracic deposition (Boreiko & Rossman, 2020). Whereas deep lung deposition can yield close to complete uptake over time, most antimony inhaled in the occupational setting will be translocated to the GI tract where GI uptake kinetics of less than 1% are expected. Gastrointestinal uptake will also exhibit nonlinearity due to saturation of the aquaglyceroporin receptors that mediate the uptake of antimony into cells (Boreiko & Rossman, 2020). For most soluble MMEs, high rates of uptake are to be expected after alveolar deposition. Much lower rates of MME uptake are to be expected from the GI tract, with saturation of the transport systems responsible for uptake yielding nonlinear (decreasing) uptake kinetics with increasing administered dose. Accurate prediction of regional deposition in the lung can thus be critical to determining the combined and/or individual contribution of pulmonary and GI uptake to systemic exposure. As noted in the original Metal Framework Directive (USEPA, 2007a), the complex interactions responsible for MME uptake reinforce the contribution that PBPK modeling can make to MME exposure assessment.

Case study: Nickel

The preceding example illustrated the utility of regional deposition modeling for comparing the delivered dose with target tissues in studies with similar protocols but different outcomes. The sophistication of such evaluations can be expanded to include comparisons of the dosimetry associated with qualitatively distinct outcomes generated from studies of humans, animals, and cultured tissues. The Risk Assessment Committee (RAC) of ECHA recently conducted a quantitative assessment of nickel-induced changes associated with carcinogenesis in the development of occupational exposure limit values (RAC, 2018a, 2018b). Nickel compounds are classified as known human carcinogens based on strong epidemiological evidence from occupationally exposed cohorts (Agency for Toxic Substances and Disease Registry [ATSDR], 2005; ICNCM, 1990). Lung tumor induction has also been observed after inhalation exposure of rats (but not mice or hamsters; ATSDR, 2005), and it had been determined that nickel compounds differ in their ability to induce preneoplastic changes presumed to bear a mechanistic relationship with tumor development (Buxton et al., 2019). This permitted a comparative analysis of the relationship between lung cancer risk and nickel-induced changes in animals. This analysis was then evaluated for consistency with cancer mortality in epidemiology studies.

The Risk Assessment Committee noted (RAC, 2018a, 2018b) that nickel compounds are weak mutagens that likely have an indirect genotoxic mode of action with thresholds for carcinogenesis. Nickel compounds produce inconsistent findings in some assays for genotoxicity, with mechanistic studies suggesting that nickel compounds induce genotoxic changes via an indirect mechanism of action that likely includes the induction of reactive oxygen species (Akerlund et al., 2018). Nickel's carcinogenic mode of action has also been associated with histone modifications and altered DNA methylation patterns consistent with nickel-induced changes in gene expression (Ji et al., 2008; Karaczyn et al., 2006). These molecular changes mirrored those seen in studies of rat lung tissue after inhalation exposure to carcinogenic nickel compounds at or above doses capable of causing inflammation (Efremenko et al., 2014, 2017). Other studies had indicated that inflammatory responses were necessary but not sufficient for nickel carcinogenicity (Buxton et al., 2019). Inflammation thus provided the mechanistic basis for a threshold in nickel carcinogenesis.

Suggestions of a threshold for nickel-induced carcinogenesis were consistent with the observational epidemiology studies of workers exposed to nickel and nickel compounds. The exposures producing tumors in rats had been evaluated for consistency with the observational epidemiology data (Oller & Oberdörster, 2010; Oller et al., 2014). Oller and Oberdörster (2010) surveyed the published animal literature on nickel carcinogenicity, extracting information on the dose–response for nickel-induced tumor induction and putative preneoplastic changes from eight cancer bioassays and seven 90-day inhalation studies. Particle size distribution data were available for most studies, permitting the application of MPPD for estimation of the “delivered dose” to different regions of the rat respiratory tract. Retention half-time assumptions were then made, and cumulative deposition estimates were used to refine the delivered dose exposure estimate to yield the retained dose. Finally, deposition and retained dose metrics were normalized to the surface area of target tissues.

Rat inhalation exposure levels of 0.03 and 0.5 mg/m³ mg were identified as the No Observed Adverse Effect Concentration (NOAEC) and Lowest Observed Adverse Effect Concentration (LOAEC) for nickel sulfate and nickel oxide, respectively. The observed retained dose in the whole rat lung (0.26–0.37 μg/m³) was close to the MPPD prediction for nickel retention of 0.2 μg/m³ in the alveolar regions of the lung (Oller et al., 2014), indicating that the MPPD modeling had successfully approximated pulmonary exposure to nickel. A multipath particle deposition model was then applied to estimate the expected deposition of experimental aerosols used in rodent cancer bioassays in the human respiratory tract. Nickel deposition throughout the respiratory tract was predicted to be greater in humans and with greater fractional deposition in the alveoli and the tracheobronchial regions. As a result, HEC derivations based on the rat aerosol particle size distributions yield a predicted deposition dose value that is less than the human deposited dose by a factor of 2 or greater. However, occupational nickel aerosols are coarser than the experimental aerosols used for rat inhalation studies. Multipath particle deposition model calculations using occupational aerosol particle size distributions (PSDs) indicated that lower levels of deep lung deposition will result in modern occupational settings, and humans would require airborne nickel levels 4–20-fold higher than the rat to yield comparable loading of human deep lung tissues in the workplace.

Occupational exposure data, inclusive of personal monitoring data and particle size distributions, were also available for job activities at nickel production facilities evaluated in cancer epidemiology studies (Oller et al., 2014). The rat NOAEC and/or LOAEC of 0.03/0.5 mg/m³ derived above was then compared with the occupational exposures evaluated in human epidemiology studies. The relationship between nickel exposure and lung cancer (as defined in the human epidemiology data) was plotted and a “point of departure” of 0.1 mg/m³ identified for excess cancer risk. This point of departure was similar to the rat NOAEC and/or LOAEC, an observation that provided plausible reinforcement of the dose–response for nickel carcinogenesis inclusive of a threshold.

Several simplifying assumptions were included in the nickel studies, and multiple adjustments and refinements to quantitative exposure estimates were possible. Moreover, there was seeming precision in the MPPD-derived dosimetry. However, perceptions of precision can be misleading because these calculations contained significant sources of uncertainty (Oller et al., 2014). For example, estimates of retained dose depend on assumptions on rates of particle clearance. Unfortunately, data on retention and clearance are limited, and there is significant uncertainty associated with estimates of retained dose. Indeed, many values used in the preceding examples were imputed or were “expert judgment” evaluations based on limited data. Yet additional uncertainty was introduced by normal morphometric variation in the structure and function of the human respiratory tract and breathing patterns, but quantitative estimates of this source of uncertainty were not made.

No single line of evidence provided adequate definition of mechanism(s) or dose–response(s) for nickel-induced changes. Rather, it was the consistency of effect at the molecular, cellular, and whole animal levels that was most telling and which permitted the definition of associations between the results of laboratory and epidemiology studies. Modeling of the regional deposition patterns associated with inhalation exposure in animals and humans with tools such as MPPD provided the critical linkages required for interspecies extrapolation between studies of nickel and predictions of human risk. As with the preceding example of zinc, the occupational setting provided the exposure intensity required for relevant epidemiological studies. However, deposition modeling was instrumental in interpretating the data from animal cancer bioassays and defining probable mechanisms of action. The findings were thus relevant to all levels of exposure—including those associated with environmental exposure.

Case study: Poorly soluble low toxicities

Although modes of action specific to different MMEs have been postulated to mediate carcinogenic, genotoxic, and/or cytotoxic impacts observed after inhalation exposure, there are aspects of pulmonary toxicity that relate to nonspecific effects that can be induced by poorly soluble low toxicity (PSLT) particles. Metal and metalloid elements such as diantimony trioxide are relatively inert and have been suggested to pose the risk of pulmonary damage by this nonspecific mechanism (Boreiko & Rossman, 2020). As the acronym suggests, PSLTs persist in the lung due to low solubility in pulmonary fluids and exhibit low or no toxicity related to their chemical constituents. Proposals have been made for the definition of “poorly soluble,” but no consensus exists as to the “low level” of toxicity that can be exhibited by a PSLT (Driscoll & Borm, 2020). Poorly soluble low toxicities can accumulate in the lungs of exposed animals and, when their accumulation exceeds the clearance capacity of the lung, inflammatory responses are induced. For rats exposed to PSLTs, this inflammation may precede the development of lung neoplasms. This “particle overload” and resulting inflammation can also be induced in humans and mice but has yet to be linked to the development of lung tumors in these species (Warheit et al., 2016). Indeed, compounds such as nickel sulfate induce inflammation in the lungs of rodents but do not induce tumors. Inflammation may thus be necessary for the action of some carcinogens, but inflammation is not a sufficient basis for tumor induction (Buxton et al., 2019).

The frequency with which MME aerosols act as carcinogenic PSLTs is not known, and the significance of PSLTs for humans remains to be determined. Indeed, a rigorous definition of PSLTs and delineation of their mechanism of action is still being sought (Driscoll & Borm, 2020). Poorly soluble low toxicities can be operationally described, but the mode of action (inclusive of the nature and extent of permissible toxicity and tissue persistence) that produces overload impacts remains to be determined. Until the boundary conditions that define a PSLT are established, the relevance of PSLT activity to the action of MMEs in either humans or experimental animals will remain a matter of conjecture.