lapdMouse: a data archive for advancing computational models of inhaled aerosol dosimetry

Risk assessment of inhaled toxic particles is often based on the extrapolation to humans of a critical dose observed to cause adverse effects in laboratory animals (5, 10). Extrapolation of animal doses to humans requires sophisticated computational models that account for interspecies differences in breathing rates, airway anatomy, particle deposition patterns, and clearance rates (1, 4, 6). Accurate prediction of inhaled aerosol dosimetry is challenging due to the multiscale nature of the respiratory tract anatomy (from the nose to the alveoli), complex flow patterns, multiple physical mechanisms of particle capture (inertial impaction, diffusion, and sedimentation), and respiratory physiology factors (e.g., lung compliance, mucociliary clearance, etc.) (9). Progress in computational toxicology is hindered by the paucity of high-resolution, quantitative experimental data to test model predictions. For example, although mice are commonly used in toxicology studies, there is a paucity of quantitative experimental data to validate theoretical models of inhaled particle dosimetry in mice (2, 8). Another factor hindering progress in computational toxicology is the scarcity of publicly-available data repositories that would allow researchers from different centers to validate their computational methods against experimental benchmarks.

In this issue of the Journal of Applied Physiology, Bauer and coauthors (3) report the creation of an open-access data archive of three-dimensional mouse lung models, breathing parameters, and high-resolution regional doses of inhaled 0.5, 1, and 2 μm particles. Complete data sets are available for 34 mice encompassing four strains (B6C3F1, BALB/C, C57BL/6, and CD-1) and both sexes (16 male, 18 female). The lung models include the anatomy from the trachea to terminal bronchioles (diameters ∼100 μm). The data are made publicly available through the Lung Anatomy + Particle Deposition (lapd) Mouse Archive (https://doi.org/10.25820/9arg-9w56). Compared with previous studies, the lapdMouse Archive represents a giant leap in terms of the number of specimens investigated and resolution of particle deposition locations. This achievement was made possible by significant innovations in image-processing and airway segmentation algorithms (3).

A major contribution of the lapdMouse Archive is that it provides the opportunity to investigate how variability in lung morphology (due to gender, mouse strain, or individual differences) affects the dose of inhaled particles deposited in different lung regions. For example, the authors reported that the trachea and main bronchi were narrower in BALB/C mice compared to the other mouse strains. Given that inertial impaction is the dominant capture mechanism of micrometer-sized particles, greater deposition is expected in the proximal airways of BALB/C mice compared to the other strains. This prediction is supported by Fig. 18 in Bauer et al. (3), which shows a colormap of relative deposition in the four mouse strains. Unfortunately, the authors did not report a quantitative comparison of regional deposition in the different mouse strains, but this illustrates how researchers can utilize the lapdMouse Archive to explore the interplay between lung anatomy and regional doses of inhaled particles. Consistent with expectations for flows where particle inertia dominates particle motion, Bauer and coauthors (3) reported hot spots of particle deposition at airway bifurcations and higher deposition of 2-μm particles than 1-μm particles in the proximal airways. Importantly, the authors provide a detailed comparison of breathing parameters (breathing rate, tidal volume, and minute ventilation) and lung morphology (lobe volumes, branching angles, etc.) that can be used to develop strain-specific dosimetry models.

One limitation of the lapdMouse Archive is the small particle size range studied (0.5 to 2 μm), which was limited by the fact that particles had to be small enough to reach the lungs and large enough to provide sufficient fluorescent signal for imaging. There is a growing need for risk assessment of inhaled engineered nanomaterials (10); thus future studies should explore technological innovations to expand this database to include nanoparticles. Another limitation is that the lung models start at the distal trachea and thus do not include the nasal cavity, pharynx, larynx, and proximal trachea. The upper respiratory tract has been predicted to capture a significant percentage of inhaled micrometer-sized particles (almost 40% of 1-μm particles) in mice (2). Since the lapdMouse Archive reports only relative doses, experimental validation of absolute doses predicted by nose-to-lung models will require expanding this or other data repositories to include three-dimensional models and associated particle deposition data for the upper respiratory tract. Finally, another challenge to validate dosimetry models with in vivo measurements is the confounding factor of mucociliary clearance. Bauer and coauthors (3) reported that mice were exposed to the monodisperse aerosol for 10–15 min before being euthanized. Thus, optimal correlation with regional doses in the lapdMouse Archive may require dosimetry models to incorporate an estimate of mucociliary clearance that occurred during the exposure and before tissue preparation.

In summary, computational models of inhaled aerosol dosimetry provide human-equivalent concentrations of critical doses observed to cause adverse effects in animal models, and thus they play a major role in establishing occupational exposure limits of inhaled chemicals. The ability of these models to predict site-specific doses has not been fully validated yet due to the lack of experimental data to test model predictions. The lapdMouse Archive provides the first high-resolution data set of regional doses in the mouse lung. By sharing their data via a public archive, Bauer and colleagues (3) have made an impactful contribution that will certainly accelerate the development of computational fluid dynamics (CFD) models and one-dimensional “typical path” dosimetry models. Advances in dosimetry models will benefit not only the risk assessment of occupational exposures, air pollution, and tobacco smoke, but it will also advance respiratory drug delivery, where CFD models are being used to improve the performance of drug delivery devices, including maximizing drug delivery to specific target sites (7).

GRANTS

The author acknowledges funding from the National Heart, Lung, and Blood Institute via Grant HL-122154.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the author.

AUTHOR CONTRIBUTIONS

G.J.M.G. drafted, edited, revised, and approved the final version of the manuscript.