Design-Based Sampling and Quantitation of the Respiratory Airways

Abstract

Design-based quantitation of the nasal cavity, larynx and tracheobronchial conducting airways after exposure to inhaled toxicants requires complete measurement of all respiratory airways or appropriate sampling followed by morphometric measurements. In vivo imaging (MRI or CT) of the nasal cavity, larynx and conducting airways provides anatomical detail of all the airways down to the distal airways. Since inhaled toxicants show predictable deposition patterns in the airways, identification and sampling of conducting airways becomes essential in a precise toxicological evaluation. Lengths, diameters and luminal surface areas can be directly measured on fixed specimens using a steromicroscope. Estimates of cell numbers, extracellular matrix volumes and vessel/nerve lengths per airway or epithelial basal laminar surface are estimated stereologically. Selected airways are cut into smaller pieces using a “fractionator” for uniform sampling of the airways. Cell numbers are estimated using a “disector.” Volumes are estimated using point probes, while length and surface areas are estimated by isotropically oriented sections with plane and line probes; an approach free of assumptions of shape, size or spatial orientation. True biological variance and the average sampling variance of the stereological measurement define the minimal sampling required to achieve precise estimates of the nasal cavity, larynx and conducting airways.

Introduction

Measurement of nasal cavity, larynx and conducting airways yields quantitative data for parameters like volume, surface area, length, cell number and cell size. While measurement of structure in general is known as morphometry, the methods to obtain these data in microscopy are referred to as stereological methods, because of the requirement of tissue sections to observe structure by microscopy. Stereology can be defined as the science of sampling structures with geometric probes. Over the past few decades, there has been an evolution of rigorously uniform sampling designs of stereology that ensure unbiased estimates of number, length, surface area and volume. This approach has been termed ‘design-based’ stereology because of the reliance on sampling design rather than geometric model-based stereology that makes assumptions (Gundersen et al., 1988a, 1988b). The aim of this paper is to establish new design-based stereological approaches for the direct estimation of anatomical structures and epithelial, interstitial and endothelial cells of specific regions of the nasal turbinates, larynx and conducting airways normalized to their epithelial basal laminar surface.

The extrapolation to humans of studies of toxic agents injurious to the respiratory system using animal models assumes comparability in the structure and function of these model species and humans. An underlying assumption is that data, especially data that is quantitative in nature related to respiratory system structure in model species, can be extrapolated to humans. The rationale for using nonhuman primates as animal models to study lung disease has been their proximity to man on the phylogenetic scale, the similarities of postnatal development of the inflammatory/immune systems and pulmonary structure and function. We will use the examples of nonhuman primates extensively throughout this paper. The key issues in quantitative analysis of the nonhuman primate respiratory system are that injury is site-specific, target cell populations vary in distribution, metabolic functions of cell populations vary by microenvironment, and the pattern of injury varies by: species, age, injurant class, exposure route and gender and trophic interactions between all cellular and acellular components at site dictate injury and repair.

The nose is the primary site of entry for inhaled air in the respiratory system of mammals. Because of its critical location it is ideally suited to provide a sense of smell and to filter, warm and humidify the inhaled air before it enters the conducting airways of the lung. Even though it serves a critical role as the entry portal to the respiratory system, it is also vulnerable to the highest doses of inhaled toxicants or infectious agents in the respiratory system. After the transitional airway, the larynx, which is distal to the nasopharynx and pharynx, we enter the remainder of the respiratory system. In mammals it is organized into two compartments based on function: the conducting airways and the gas exchange area. The three dimensional organization of the conducting airways and the cell populations that comprise the walls of the airways, serve a protective function for the more fragile gas exchange area. This protective function includes further warming and humidification of the inspired air, filtering and detoxifying the air, clearing cellular and foreign debris that accumulates on the airway surfaces. The conducting airways are comprised of the extrapulmonary bronchi which branch from the trachea in the mediastinum and carry air to the left and right lungs; and the intrapulmonary bronchi and bronchioles, which form a branching tree within the lungs to carry air to the gas exchange area.

Methods

We will use the following guiding principles for quantitation: (1) use design-based methods to quantify structure, (2) use structural hierarchies to link and interpret experimental data, and (3) collect data with the most statistical efficiency to detect structural changes.

Design-Based Methods

It has been calculated that if all the material that had ever been in focus in any of the transmission electron microscopes in the world were gathered together it would total less than 1 cubic centimeter in volume (Howard and Reed, 2005). Given that this statement is reasonably true, then sampling is everything in quantitation. We are aware of sampling bias when boys pick the tallest players for basketball in drafting a playground team. Here the height of the first 5 boys selected is not representative of the 25 boys that are eligible to play. We see similar bias at work in the sampling of airways and cells within airways by sections. The probability of a cell being selected in a section is directly proportional to its 3-dimensional size (height along the sectioning direction to be exact). Thus a sample is considered unbiased when all the compartments of the structure have an equal chance of being sampled. The most reliable method of satisfying this need for unbiased sampling is to introduce randomness in the sampling process. For example, if a structure such as the lung does not have a uniform distribution of all of its components, then the best way to avoid sampling bias is to collect samples with a design-based approach (Gundersen and Jensen, 1987; Ogbuihi and Cruz-Orive, 1990). Design-based sampling ensures unbiased sampling at all magnifications and even when the components exhibit striking anisotropy. Another obvious benefit is that all of the data derived from design-based sampling allows experimental questions, data and interpretations to be related to the 3-dimensional space of real world biology. These design-based methods are assumption-free, mathematically rigorous and cannot be validated with data.

As we all know, statistical and stereological advice is always most effective when it is given before the experiment is begun. The primary objective of statistical analysis is to infer characteristics of a group of data by analyzing the characteristics of a small sampling of the group. Basic to statistical analysis is the desire to draw conclusions about a group of measurements of a variable being studied, if we use design-based methods we have confidence in the conclusions we draw from our data. Systematic sampling is a more efficient method for taking replicated spatial samples than random sampling (Howard and Reed, 2005). The only problem inherent in systematic sampling is if there is a natural periodicity in the structure being sampled which happens to coincide with the period of the systematic sample. This approach uses a random start in the first period of sampling. Systematic, Uniformly Random Sampling (SURS) is a sampling strategy that almost always has been used in stereological sampling and estimation procedures. SURS finds its greatest refinement in the smooth fractionator that is much more efficient than random sampling (Gundersen, 2002). Like biochemical fractionation, the smooth fractionator can be used to estimate tissue composition using a small sample of the original feature. The fractionator can be used to sample the entire lung (Ogbuihi and Cruz-Orive, 1990) or subsets of the respiratory system, like the nasal cavity (Figures 1 and 2) or a specific airway generation (Figure 3). The sampling design illustrates that the fractionator may effortlessly be combined with other sampling requirements, like a specific orientation distribution of the sections and estimates of global volume shrinkage, necessary for robust estimates of surface area and length of lung structures (Hyde et al., 2004). Critical characteristics of quantitative estimates, bias (not reporting true data) and error (individual estimates scattered around the true data), are direct consequences of sampling and thus are dependent on design-based methods.

The interface of probes and structural features that enable investigators to estimate quantitative values requires isotropic orientation of either the tissue section or the test probe (Figure 4). This probe-structural feature interaction causes dimensional reduction when estimates are performed on sections: 3D volumes in the 3D world become 2D areas in sections; 2D surfaces in the real world become 1D lines in sections; 1D lengths in the 3D world become 0D points in sections. It also becomes clear that it requires a 3D probe to estimate number of objects. This follows from the sampling discussion above that a single section contains size bias relative to individual features.

In this probe-structural feature interaction the most efficient stereological estimators are point hits for volume, line intersections for surface, plane transects for length and volume sampling of number. While volume and number are not orientation-dependent, surface and length are influenced by the orientation of the probe relative to the structural feature to be estimated. Methods have been developed to address this orientation problem by either isotropically orientating the tissue before it is sectioned (orientator (Mattfeldt et al., 1990) and isector (Nyengaard and Gundersen, 1992) or orienting the tissue with a known (correctable bias) relative to the probe (vertical sections). There are only a few rules for coherent test systems of probes. Keep them simple for easy counting of probe/feature interactions (see section on Statistical Efficiency later) and use the counting rule of Gundersen when counting profiles (Gundersen, 1977).

The precision and efficiency of probe-structural feature interaction on sections is not intuitively obvious as many investigators still believe that computer digitization is better than test probes. However, experimental comparisons of these approaches favors the use of test probes even in computerized systems (Mathieu et al., 1981; Heidsiek et al.,1987; Howard and Reed, 2005). It should be noted that even when image analysis can be applied to lung tissue, such as quantitating the volume of stored mucosubstance per surface area of epithelial basal lamina, it is only 12-fold more efficient than manual methods (Heidsiek et al., 1987). This discussion of sections does not carry over to 3-dimensional data sets like MRI and CT where the density of voxels and the sophistication of algorithms provide robust estimates of morphometric parameters.

Structural Hierarchies (Multicascade Sampling Schemes)

The respiratory system is a complex organ composed of numerous compartments ranging in size from molecules to tissues. Hierarchies allow us to organize data according to the size of the structures and the equations define relationships and link data within and across hierarchical levels. Traditionally we call this approach a multicascade sampling scheme that has been used to estimate the number and composition of nonciliated bronchiolar cells in respiratory bronchioles in response to inhaled oxidants (Fujinaka et al., 1985; Moffatt et al., 1987). In hierarchical organization two general guidelines emerge: (1) use the lowest reasonable magnification (acceptable resolution) to increase sample size for measurements; and (2), if major compartments and their subcompartments cannot be measured at the same magnification, then the magnification should be increased to optimize resolution in the subcompartment. This approach fits perfectly with fractionator sampling as estimates of number of alveoli is directly obtained from fractionator sampling, alveolar volume and diameter estimates require use of a modest multicascade sampling scheme for determination of the volume of the lung in which alveoli reside (Hyde et al., 2004).

Statistical Efficiency

The phrase “Do more less well,” coined by Professor Ewald Weibel at the Fifth International Congress for Stereology in Salzburg in 1979 in the discussion of Professor Hans Gundersen’s paper entitled “Sampling Efficiency and Biological Variation in Stereology,” is in essence the key to statistical efficiency in quantitation (Gundersen and Osterby 1981; Gundersen et al., 1999). Knowing the components that comprise biological variation in a system like the respiratory system is the key to the design of an experiment (number of animals, tissue samples, fields per slide, frame size, points and line length per test system to estimate a particular structural feature). The key equation is simply

$${\text{OCV}}^2(\widehat{\text{X}})={\text{CV}}^2(\text{X})+\text{mean}[{\text{CE}}^2(\widehat{\text{X}})]$$

where OCV²(X̂) is the observed coefficient of variation, CV²(X) is the true biological variation (unknown) and mean[CE²(X̂)] is the estimate of the stereological variation (Gundersen and Osterby, 1981). Estimation of the contributions to stereological variation to direct volume estimates using the Cavalieri method has been provided by Cruz-Orive (1999) and in a nomogram by Gundersen and Jensen (1987). Contributions to stereological variation for ratio estimators like volume, number, surface and length densities have also been derived (Cruz-Orive, 1980). However, some simple guidelines will usually suffice for stereological sample size within an animal (primary sampling unit) as follows: 100–200 probe interactions (e.g., point hits), 50 fields and 10 blocks. This approximation is based on the t-test as follows:

$$\begin{array}{l}{(t\hspace{0.17em}\text{test})}^2=\left(\frac{{\text{diff}}^2}{{\text{SEM}}^2}\right)>4\\ {\text{n}}_{\text{blocks}}>2\frac{{\text{OCV}}_{\text{blocks}}^2}{{\text{CV}}_{\text{animals}}^2}\mathrm{\hspace{0.17em} }\mathrm{\hspace{0.17em} }\mathrm{\hspace{0.17em} }\text{then\hspace{0.28em}}\mathrm{\hspace{0.17em} }\mathrm{\hspace{0.17em} }\text{n\hspace{0.28em}}\mathrm{\hspace{0.17em} }\mathrm{\hspace{0.17em} }\text{blocks}\sim 10\\ {\text{n}}_{\text{fields}}>2\frac{{\text{OCV}}_{\text{fields}}^2}{{\text{CV}}_{\text{blocks}}^2}\mathrm{\hspace{0.17em} }\mathrm{\hspace{0.17em} }\mathrm{\hspace{0.17em} }\text{then}\mathrm{\hspace{0.17em} }\mathrm{\hspace{0.17em} }\mathrm{\hspace{0.17em} }\text{n\hspace{0.28em}}\mathrm{\hspace{0.17em} }\mathrm{\hspace{0.17em} }\text{fields}\sim 50\\ {\text{n}}_{\text{probes}}>2\frac{{\text{OCV}}_{\text{probes}}^2}{{\text{CV}}_{\text{fields}}^2}\mathrm{\hspace{0.17em} }\mathrm{\hspace{0.17em} }\mathrm{\hspace{0.17em} }\text{then\hspace{0.28em}}\mathrm{\hspace{0.17em} }\mathrm{\hspace{0.17em} }\text{n\hspace{0.28em}}\mathrm{\hspace{0.17em} }\mathrm{\hspace{0.17em} }\text{probes}\sim 100-200\\ {\text{OCV}}_{\text{probes}}^2\sim \frac{1}{\sum \text{hits}}\end{array}$$

Certainly 200 point hits per animal seems like very few, but 50 fields and 10 blocks seems like a lot of sampling per animal. However, if we apply the principle of doing more less well, then we sample widely within the organ, embed 10 small blocks (selected by a smooth fractionator) in one block so we only have to cut one section per animal and sample the 50 fields and 200 point hits throughout the 10 blocks (Hyde et al., 2004). This sampling approach is not only precise but highly efficient (Gundersen et al., 1999).

Next we will consider the critical data required to detect and interpret quantitative data in the respiratory system. These include the measures of volume (V), surface (S), length (L) and number (N) and their ratios or densities. It should be mentioned that ratio densities to a reference volume (Vv, Sv, Lv, and Nv) can be misleading independent of knowledge of changes in the reference compartment and has been coined the “reference trap” by Gundersen (Braendgaard and Gundersen, 1986). In regards to the reference trap he states “never ever not measure the reference space.” Densities are further defined by symbols (i, ref) that defines the ratio of the two compartments for the densities (e.g., Vv (i, ref)). The compartment of interest, the small i, is related to the reference compartment, “ref.” For example, the volume of mucous cells (mu) in conducting airways (aw) is represented as Vv (mu, aw). We usually use the first 2 letters of the compartment to abbreviate the names, such as lung = lu, capillary = ca, trachea = tr and so on.

Volume

Volume is an obvious starting point for quantitation of the respiratory system. One of the most direct methods is to systematically cut the parts of the respiratory system into slabs of equal thickness and determine their cumulative area by point counting, and multiply by the average slab thickness (Cavalieri method) (Michel and Cruz-Orive, 1988). (The Cavalieri method is named for Italian mathematician, Bonaventura Cavalieri (1598–1647), who first proposed the method for estimating volume.) Of course, MRI derived images (Figures 1 and 2) provide in vivo images that can be used to estimate volume, but there are potential errors as illustrated in subcompartment MRI volume estimation of the brain (Garcia-Finana et al., 2003). The good news is these errors can be compensated for in the brain (error prediction formulae), but remain to be defined for the respiratory system. The other direct method that works well for lung is the buoyant weight in saline (Scherle, 1970). The only drawback to this approach is that leaks in the lung influence the estimate and the nasal cavity cannot be measured by this method. Hence, of the 2 methods available, the Cavalieri estimator is the more robust.

Surface

Surface is usually estimated by surface density, Sv, and is influenced by both the sectioning angle and the shape of anisotropic structures. For isotropic structures, surface density can be defined as follows:

$${\text{Sv}}_{\text{i},\text{r}}=2{\text{I}}_{\text{i}}/{\text{L}}_{\text{r}}$$

where I_i is the number of intersections of the object surface by a linear probe and L_r is the total probe length in the reference component (Smith and Guttman, 1953). This equation is valid for test lines that are isotropic uniform random in 3D space. To meet this requirement using a lattice grid, the structures of interest must be distributed uniformly and randomly and their orientation must be isotropic. Both the orientator (Mattfeldt et al., 1990) and the isector (Nyengaard and Gundersen, 1992) are approaches that ensure tissue isotropy and mixed with a smooth fractionator uniform randomness is achieved. The fractionator approach provides total surface of the feature of interest for the component of the respiratory system. Other approaches require multiplication of the appropriate volumes and subcomponent volumes (multicascade design) to arrive at the total surface.

For tissues with evident anisotropy, the use of vertical sections, defined along the plane of preferred orientation for anisotropic microstructures, and a cycloid test system, gives surface density estimates that correct for anisotropic orientation directly using the Sv equation here (Baddeley et al., 1986; Figure 5). This approach can be used to sample tubular organs like the conducting airways (Figure 3).

Length

Length is usually estimated by length density, Lv, and is influenced by both the sectioning angle and the shape of anisotropic structures. For isotropic structures, length density can be defined as follows:

$${\text{L}}_{\text{V}(\text{i},\text{r})}=2*\frac{{\text{Q}}_{\text{i}}}{\frac{\text{a}}{\text{p}}*{\text{P}}_{\text{r}}}$$

where Q_i is the number of profiles (feature transects) per unit area of the test probe estimated by the product of the area per point (a/p) and the points that hit the reference space (Pr) (Smith and Guttman, 1953). The estimation of length density then is simply making design-based estimates of the number of profiles per unit area, which is accomplished by an IUR section and the use of the unbiased counting rule of Gundersen for areal probe profile counting (Gundersen, 1977). In thick vertical sections a ‘virtual’ IUR surface can be generated by projecting a cycloid line through the section if the major axis of the cycloid is parallel with the vertical direction (Gokhale, 1990). However, there is still the requirement for uniform random sampling and isotropic rotation about the vertical axis for the thick vertical section. For thick vertical sections we use the following formula:

$${\widehat{\text{L}}}_{\text{V}(\text{i},\text{r})}=\frac{2}{\text{t}}*\frac{\text{p}}{\text{l}}*\frac{\sum \text{Ii}}{\sum Pr}$$

where t is section thickness, p/l is the length per point on the test system, Ii is the number of intersections observed through the thick vertical section and Pr is the number of point that hit the reference space (Figure 6).

Number

Count the number of structural features in the respiratory system, whether they be cells, organelles, airways, vessels or nerves should be done in 3-dimensional space. As mentioned previously profile counts are not equivalent to number counts in volume and should be avoided as an endpoint because of their inherent bias. Direct counting in volume is possible with in vivo imaging (MRI and CT), but can be challenging if the sample size is large if a robust sampling scheme is not employed. In microscopy, a “disector” is used as a volume probe (Sterio, 1984). The disector approach can be used with thick sections and viewed with a brightfield or laser confocal microscopes that optically focus through the section (Figure 7) (Postlethwait et al., 2000). The usual section thickness is 30 to 50 μm and a short depth of focus (usually about 1 μm) is essential to optically section the tissue. A length gauge is required to record the distance moved in the Z direction. This unbiased counting method, called an optical disector, is direct and efficient, provided we use a 2-dimensional unbiased counting frame (Gundersen, 1977) and extend the counting frame concept by excluding structures counted on either the top or bottom of the counting cube. We estimate the reference volume by point counting an optical section in the middle of the cube that provides us with a reference area that is multiplied by the distance traveled in the Z direction for counting structures. We use the following formula:

$${\text{N}}_{\text{V}(\text{i},\text{r})}=2*\frac{{\text{Q}}_{\text{i}}}{\frac{\text{a}}{\text{p}}*{\text{P}}_{\text{r}}*\text{h}}$$

where Qi is the feature count, a/p is the area per point of the test probe, Pr is the number of test points that hit the reference component and h is the height of the disector. Usually an optical disector requires feature identification because of lack of resolution, hence the use of immunohistochemistry (Figure 7). For straight morphological identification a physical disector is used where the disector height can be used to control sample size and we aim for 200 counts over all sections and fields (do more less well) (Figure 8). In general, disector height should not be greater than 0.75 h of the smallest feature to be counted. We use the same formula as for the optical disector above. Usually the sections of a physical di-sector are as thin as possible and the top section is use as the counting section and the bottom section as the lookup section. Only those features intersecting only the counting section are counted, thus the height is the thickness of the counting section and the distance to the lookup section. Naturally, we use a 2-dimensional unbiased counting frame on the counting section (Gundersen, 1977). Any dissector count can be combined with a fractionator to provide an unbiased estimate of the total number in the respiratory system as follows:

$$\text{N}=\frac{1}{\text{bsf}}*\frac{1}{\text{ssf}}*\frac{1}{\text{asf}}*\frac{1}{\text{hsf}}*\sum {\text{Q}}^{-}$$

where the fractions are bsf = block sampling fraction, ssf = section sampling fraction, asf = area sampling fraction, hsf = height sampling fraction and Q⁻ is the feature count within the sampled fraction. Figure 2b provides an example of a physical disector count using fractionator sampling in the nasal cavity of a rat.

Results

These design-based methods can be used to evaluate the distribution of airway epithelial cells along the conducting airways of the right middle lobe of a rhesus monkey (Figure 9). We can use a stereologic measure of the abundance of epithelial cell types (volume of the cell per epithelial basal lamina surface area, μm³/μm²) per selected airway generation within the lung lobe. This normalization of epithelial cells to a local surface area generally shows less variance than the total number of cells at a particular airway generation and allows direct comparisons of the epithelium among airways of different generations. It is also useful in airway epithelial comparisons between species. As emphasized in Figures 10 and 11, there is a substantial difference in the amount of epithelium that lines the luminal surface in the trachea. The thickness of the epithelium in the trachea of rhesus monkeys is approximately twice that of mice and rats and one-half to one-third that in humans. Other major differences between species are the composition of the epithelium, the density of cells lining the surface, and the proportion of cell phenotypes in the epithelium. A substantial percentage of the airway in primates is occupied by mucus cells, but they are not generally found to the same extent in the trachea of healthy, pathogen-free mice and rats. The proportion of ciliated cells in the epithelium is relatively similar in all species, but the proportion of basal cells varies with species.

The largest, most proximal intrapulmonary bronchi exhibit some of the most organizational variation between species. While smooth muscle is present in the walls of all mammalian species, there is a substantial difference in the amount of cartilage found in the lobar bronchus in laboratory mammals and in the distribution of sub-mucosal glands. Epithelium is reduced in thickness in more distal airways compared to the trachea. There also are major differences in the organization of the surface epithelial population, with mucus cell and basal cells predominating in primates and Clara cells being the principal nonciliated cell population in other laboratory animals. These differences are even more marked in more distal conducting airways such as the mid-level bronchi. In the most distal conducting airways, the bronchioles, the major differences between species are related to the epithelial surface lining. In laboratory mammals, the Clara cell is the primary secretory cell phenotype, there are no mucous cells, and the numbers of basal cells in the epithelium are related to the extent of alveolarization (Figure 11). The bronchioles of rhesus monkeys have an extensive smooth muscle portion which is arranged in large bundles and is interspersed with extensive connective tissue not generally observed in laboratory mammals.

It still remains for us to compare statistical efficiencies among common cell quantitation approaches in the conducting airways. We used 6 rats and sampled 200 cells with disectors, both optical and physical, in 4 blocks of 4 randomly selected terminal bronchioles. We used immunohistochemistry of Cytochrome P450 for cellular identification of non-ciliated bronchiolar cells in thick 30 μm sections for the optical disector and half-micron and 60 nm thick sections for morphologic identification for physical disectors by light and transmission electron microscopy (TEM) (Figure 8), respectively. The efficiency of data collection (time to prepare the specimen, section, collect images on the microscope and use a test system to estimate the cell number) was the optical disector > light microscopic physical disector > TEM physical disector (Figure 12). In this instance the stereological precision of the estimate was very similar between the three approaches as evidenced by similar variability within blocks. When the variability between blocks was compared, the optical disector showed about half the variability of the 2 physical disector methods (Figure 13). The reason for this result is not immediately apparent as we would expect all 3 methods to have the same variability. However, it has little impact on the final estimate because the animal variability is the same for all 3 methods, as one would expect because the same 6 animals were used for all methods (Figure 14). Since the variability between blocks is about the same as the animal variability, our sampling appears to be very efficient (doing more less well). Our conclusion would be that if we want to improve our estimate, then we should use more animals. If we want the most efficient estimator of nonciliated bronchiolar cell number, then clearly the optical disector is the best.

Discussion

It is clear from the methods and examples provided that design-based quantitation of the nasal cavity, larynx and tracheobronchial conducting airways after exposure to inhaled toxicants provides a robust, unbiased assessment of the structural components of the respiratory airways. These design-based methods interface well with in vivo imaging (MRI or CT) of the nasal cavity, larynx and conducting airways that provide useful anatomical detail of all the airways down to the distal airways. Careful sampling using a smooth fractionator for the components of the respiratory system allows us to efficiently and precisely evaluate the effects of inhaled toxicants on the epithelium and submucosal cells and extracellular matrix. Using normalization of the epithelial basal lamina, we can compare changes in these compartments among animals at the same airway generation or among different airway generations. Thus, these methods provide a morphologic assay that is efficient, precise and statistically rigorous.