1*1 - Library and Archives Canada · Stereology is a group of statistical and geometrical procedures which yield information about objects in three dimensions from two dimensional

National Library 1*1 of Canada Bibliothèque nationale du Canada

Acquisitions and Acquisitions et Bibliographic Services services bibliographiques

395 Wellington Street 395. rue Wellington OttawaON K1A ON4 OttawaON KlAON4 Canada Canada

Yaur iib V m reterenœ

Our nle Norre n ? f . c e

The author has granted a non- exclusive licence aüowing the National Library of Canada to reproduce, loan, distribute or sel1 copies of this thesis in microfom, paper or electronic formats.

The author retains ownership of the copyright in this thesis. Neither the thesis nor substantial extracts from it may be printed or otherwise reproduced without the author's permission.

L'auteur a accordé une licence non exclusive permettant à la Bibliothèque nationale du Canada de reproduire, prêter, distribuer ou vendre des copies de cette thèse sous la forme de microfiche/fYm, de reproduction sur papier ou sur format électronique.

L'auteur conserve la propriété du droit d'auteur qui protège cette thèse. Ni la thèse ni des extraits substantiels de celle-ci ne doivent être imprimés ou autrement reproduits sans son autorisation.

ABSTRACT

ldiopathic pulmonary fibrosis (IPF) and pulmonary emphysema are chronic lung

diseases exhibiting progressive deteiioration in pulmonary function as the lung architecture is

remodeled. This thesis quantifies these tissue changes using a novel combination of computed

tomography (CT) and quantitative histology. Pre-operative CT scans were obtained from

patients with IPF, patients receiving lung volume reduction surgery for diffuse emphysema and

from patients with minimal to mild emphysema undergoing lobectomy for a small peripheral

tumour. Total lung volume was calculated using the pixel dimensions on the CT scan while

airspace and tissue volume as well as the regional lung expansion were esümated using the X-

ray attenuation values. Tissue samples were obtained at either open lung biopsy (IPF) or

surgical resection (control and emphysema) and prepared for quantitative histology. A method

for correcting the histology specimens to an in vivo level of inflation was developed so that the

tissue composition and surface area could be estimated using stereologic techniques. The

data shows that there is a reorganization of lung parenchyma in IPF with a disproportionate loss

of airspace and surface area without increasing the total amount of tissue. The patients with

ernphysema show evidence of a progressive proteolytic destruction of tissue volume and

surface area. There is a negative correlation between regional lung expansion and surface

area in emphysema and a positive correlation between surface area and the diffusing capacity

of the lung in both diseases. This technique should prove useful in the longitudinal assessment

of chronic lung diseases and the monitoring of response to treatment.

TABLE OF CONTENTS

ABSTRACT

iii TABLE OF CONTENTS

LIST OF TABLES

vii LIST OF FIGURES

viii ACKNOWLEDGEMENTS

PREFACE

CHAPTER 1: INTRODUCTiON TO QUANTlTATlVE ANALYSIS OF THE LUNG

1.1 Quantitative Histology Of The Lung 1.1.1 Sampling 1.1.2 Bias 1.1.3 Variance

1.2 Stereological Methods 1.2.1 Stereological Probes 1.2.2 Cavalieri's VoIume Estimator 1.2.3 Volume Fraction 1.2.4 Surface Area 1 .Z.5 Multi-Level Sampling Design

1.3 Quantitative Gross Analysis Using Computed Tomography

CHAPTER 2: WORKING HYPOTHESIS. SPECIFIC AlMS AND STRATEGY

2.1 Working Hypothesis

2.2 Specific Aims

2.3 Strategy

CHAPTER 3: THE NORMAL HUMAN LUNG

3.1 Descriptions of the Lung 3.1.1 Gross Lung Structure 3.1.2 Cellular Lung Structure 3.1.3 Extra-cellular Matrix

3.2 The Clinical Measurement of Lung Function

3.3 The Pleural Pressure Gradient

3.4 Experiment #1

3.5 Material and Methods 3.5.1 Pulrnonary Function Studies 3.5.2 CT Studies 3.5.3 Quantitative Histology 3.5.4 Statistical Analysis

3.6 Results

3.7 Discussion

CHAPTER 4: INTERSTITIAL LUNG DISEASE

4.1 Introduction to Interstitial Pulmonary Fibrosis (IPF) 4.1.1 Clinical Description of IPF 4.1.2 Radiological Description of I PF 4.1.3 Histological Description of IPF

4.2 Fibrotic Mechanisms 4.2.1 Cellular Mechanisms of IPF 4.2.2 Molecular Mechanisms of IPF

4.3 Quantitative Studies of IPF

4.4 Experiment #2

4.5 Material and Methods 4.5.1 Pulmonary Function Studies 4.5.2 CT Studies 4.5.3 Quantitative Histology 4.5.4 Statistical Analysis

4.6 Results

4.7 Discussion

5.1 Introduction to Pulmonary Einphysema 5.1 .1 Functional Description of Emphysema 5.1.2 Radiological Description of Emphysema 5.1.3 Histological Description of Emphysema

5.2 Pathogenesis of Emphysema 5.2.1 ProteasdAntiprotease Theory 5.2.2 Inflammatory-Repair Mechanisrn

5.3 Quantitative Studies in Emphysema 5.3.1 Gross Analysis 5.3.2 Histologie Analysis 5.3.3 Radiological Analysis

5.4 Experiment #3

5.5 Materials and Methods 5.5.1 Pulmonary Function Studies 5.5.2 CT Studies 5.5.3 Quantitative Histology 5.5.4 Statistical Analysis

5.6 Results

5.7 Discussion

CHAPTER 6: SUMMARY AND DlSCUSSlON

6.1 Summary

6.2 Future Directions

6.3 Conclusion

REFERENCES

LIST OF TABLES

Table 1. Stereologic rules for the selection of a sampling probe.

Table 2. Cellular composition of the lung.

Table 3. Pulmonary Function Data.

Table 4. lndividual lobar volumes measured by CT in 9 patients.

Table 5. Lobar weight and volume.

Table 6. Lung volume and Gas per Gram of Tissue.

Table 7. Stereology.

Table 8. Patient Demographics.

Table 9. Lung Volumes and Weights.

Table 10. CT Estimated Regional Lung Inflation.

Table 1 1. Light Microscopy Volume Fractions (%).

Table 12. Patient Demographics.

Table 13. Lung Volumes and Weights.

Table 14. CT Estimated Regional Lung Inflation.

Table 15. Quantitative Histology.

Table 16. Percent Emphysema of Resected Lobe.

Figure 1.

Figure 2

Figure 3.

Figure 4.

Figure S.

Figure 6.

Figure 7.

Figure 8.

Figure 9.

Figure 10.

Figure 11.

Figure 12.

Figure 13.

Figure 14.

Figure 15.

Figure 16.

Figure 17.

Figure 18.

Figure 19.

Figure 20.

Figure 21.

Figure 22.

Representation of variance and bias in a sample.

Sample point counting grid on an object.

Sample point counting grid on light microscopic section of human l~ng .

Sample intercept counting grid on light microscopic section of humao lung.

Multi-level sampling design.

CT scan of human lung showing segmentation of the different lobes.

A representative CT analysis slice.

Graph of the CT density of the lung.

Graph of the pressure volume curves.

Graph of the pleural pressure gradient.

Classification of interstitial lung diseases based on pathogenesis.

Representative electron micrograph from IPF patient biopsy.

CT density of the lung.

Volume fraction of the tissue in the biopsied regions of lung.

Weight of the interstitial components.

CT scan of human lung with emphysema using the density mask.

Gross lung slice and C f scan.

CT density of the lung.

Mixed effects regression line for surface area per volume and lung inflation.

Mixed effects regression line for surface area and diffusing capacity of the lung for carbon rnonoxide.

Mixed effects regression line for surface area and diffusing capacity of the lung for carbon monoxide for al1 patients.

Three dimensional reconstruction of a human lung with emphysema.

vii

ACKNOWLEGMENTS

No scientific work can be completed without the assistance and support of many people.

As such, I wish to thank my supervisor and mentor Dr. James C. Hogg for getting me started in

this field and forhis guidance and support through al1 of the aspects of my career in science. 1

also wish to thank my supervisory committee, Drs. Peter D. Paré, Clive R. Roberts, and John R.

Mayo for their constructive input and instruction. This project would not have been possible

without collaborations from the University of Iowa under the direction of Gary W. Hunninghake

and Dr. Robert R. Rogers at the University of Pittsburgh. I also wish to thank Ms. Hayedeh

Bezad and Dr. Benard Meshi for their technical assistance with the stereology; the Histology

Laboratory St. Paul's Hospital for processing the histological material; the computed

tomographyfmagnetic resonance imaging staff at St. Paul's Hospital for gathering and

transferring the CT images; Dr. Kenneth P. Whittall and Mr. Don Kirkby for their wizardry with

the computers; Ms. Barbara Moore for collecting the pulmonary function data; Messrs Joe

Comeau and Stuart Greene for their computer and photographic expertise; and Ms. Lorri

Verburgt and Ms. Yulia D'Yachkova for their statistical advice. Special appreciation is also

extended to my good friends (Paul and Mary Lacey and Gary and Heidi Rae) who introduced

me to fly fishing and provided me with so much moral support through this time. Thank you

also to my parents and family who always believed in me. Finally, I wish to thank rny wife

Maureen for her patience, faith and unwavering love. Thank you al1 and God bless you.

viii

PREFACE

Chapters 3 and 4 are modifications of published papers. The introduction has been re- written to match thesis requirements, and the methods have been modified to reduce redundancies. The results, and discussion are as published. The complete publication record is listed below.

Chapter 3: Coxson, H.O., J.R. Mayo. H. Behzad, B.J. Moore, L.M. Verburgt, C.A. Staples, P.D. Paré and J.C. Hogg. The rneasurement of lung expansion with computed tomography and cornparison with quantitative histology. J. Appl. Physid. 79;1525-1530. 1 995.

Chapter 4: Coxson, H.O., J-C. Hogg, J.R. Mayo, H. Behzad, K.P. Whittall, D.A. Schwartz, P.G. Haitley, J.R. Galvin, J.S. Wilson and G.W. Hunninghake. Quantification of idiopathic pulrnonary fibrosis using computed tomography and histology. Am. J. Respir. Crit. Care Med. l55:1649- 1 656. 1 997.

CHAPTER 1: INTRODUClïON TO QUANTiTATlVE ANALYSE OF THE LUNG

1.1 Quantitative histolom of the lunq

Stereology is a group of statistical and geometrical procedures which yield information

about objects in three dimensions from two dimensional sections (48) and the history, theory

and methods of this technique have been reviewed in a number of excellent books and review

articles (1 6,36,4O,48,49,74,l44,233,2~l,253,254). These techniques are very powerful, not

only because they allow the quantification in three-dimensions, but because they are unbiased

and extremely efficient (74.1 44). Stereology began with investigations of geometrical probability

theory in the 17001s, which was then applied to quantitative problems in geology and metallurgy

in the mid 1800s and finally to histopathology starting in the rnid 1900s (4). The popularization

of these methods to investigations of the lung is attributed to Weibel and his classic book,

Morphometfy of the Human Lung (249) and was the point from which stereology began to be

applied to questions of structure in health and disease. This has led to an explosion in the

mathematical theory on which the procedures are based as welf as in refinements of techniques

for sampling and quantification.

This chapter will cover the basic principles for a stereologic analysis of the human lung

including: the sampling protocols, the basic test probes and their associated formulas. The

subsequent chapters will apply these techniques to the normal lung and two disease states, a

fibroproliferative disorder, idiopathic pulmonary fibrosis (IPF) and a desinidive disorder,

emphysema.

1.1.1 Sam~linq

In virtually al1 studies, it is impractical to quantify the whole population. Therefore, the

population is sampled and estimates about the population are made from the measurements

made on this sample. Sampling is extremely important for the stereological analysis of very

small structures because the high level of magnification that is needed to observe small

structure greatly reduces the area that can be examined in one field of view. The reliability of

any estimate is very dependent on the bias of the sample. and the variance inherent in the

technique used to make the estimate (35).

1.1.2 Bias

There is no method for calculating the bias in a sample (35). For this reason

stereological analysis has develop into what is know as 'design-baseda stereology (1 6) where

the procedures used to quantify the lung rely on the sampling design of the study. The power

of this approach is that it does not require any assumptions about size, shape or distribution of

the structures under investigation, as is the case with the 'model-based" study (233).

The !wo main forms for reducing bias within the sample are a random sample of the

lung structure, and a uniform sample of the whole organ. Random sampling gives al1 parts of

the specimen equal opportunity to be selected (144) while uniformity allows the sampling of

structures which may not be randomly distributed throughout the specimen (1 6). 60th of these

criteria are satisfied by choosing the first area, or slice, randomly. and then using a

predeterrnined interval to choose the subsequenl samples. For example, an object's volume

can be estimated by cutting the object into multiple slices of a uniform thickness, randomly

choosing a slice. (i.e. slice 2), and then systematically choosing every third slice until the object

is completely sampled (1 6,144). Variations on this procedure can be used for selecting tissue

biopsies, and microscopie fields of view (1 6,35,74,144) which yields a systematic-random

sample. It is important to note here that these sections are truly random and must be

differentiated from samples that are chosen arbitrarily or because they 'appear interesting."

These later sampling protocols introduce a bias into the sample that can dramatically effect the

reliability of the outcome.

There are many other forms of bias within the sample mostly due to technical limitations.

These include, but are not limited to: tissue processing, resolving power, section thickness,

automatic edge detection algorithms, and image recognition (144). Al1 of these biases must be

taken into account when designing stereological studies and their influence on the results will

be discussed Iater.

1.1.3 Variance

The variance associated with the estimation of a structure is determined by its biological

properties and can be calculated for the sample. The first factor to consider when calculating

the variance of a structure is the biological variation, which takes the form of how the structures

are distributed within the organ and any between subject variability. The second area where

variance can occur is due to the measuring technique used (214). The variance for the sample

can be calculated by adding the variance at each sampling level to obtain the overall standard

error of the mean (SEM) for the sarnple according to equation 1:

where ( s ~ ( ~ ) is the variance between subjects, histologic samples ( ~ ~ ~ h ~ ) , microscopie field of

view ( s ' ~ , ~ ~ ) ) , and the measurernents ( s ~ ~ , , ) , while n is the nurnber of subjects, histologic

samples, fields of view and rneasurernents respectively. It can be seen from this equation that

the greatest effect on SEM will corne from the number of subjects examiiied, even if the ?(,,,

is greater than the s2(abl (21 4). Therefore, a design-based stereological study attempts to

minimize the variance of the eçtimator by applying the most work to the level that has the

greatest impact on the overall SEM.

Figure 1 shows how variance and bias affect the estimate. If the dots represent the

measured values, and the target represents sampling bias, the ideal estimation would have low

bias and low variance (fig 1A). However, it is possible that this estimate can have too large of a

variance (fig 1 B) or too large a bias (fig 1 C) or. at worst, a combination of both (fig 1 D).

Therefore, care

must be taken in

the experimentat

design to reduce

the bias to a

minimum and

concentrate the

quantitative effort

to the

measurement

which has the

greatest variance

because it has

the greatest

impact on the

Low Bias

High Bias

Low Variance

@ High Variance

Figure 1. Representation of variance and bias in a sample. See text for full explanation.

estimation of the population.

1.2 Stereolociicat Methods

1.2.1 Stereoloaical Probes

Classical stereology began in geology in 1842 with the work of Delesse, (4) and it took

almost 100 years for the techniques to become developed enough to apply to histologie

specimens (4). During the last thirty years the techniques have becorne understood by

researchers other than mathematicians and statisticians and have been refined for general use

in estimating the size of three-dimensional structures (1 6,74,144). Stereology obtains three-

dimensional information by applying a geometric probe to a two-dimensional section. The

probe can take the form of a point, a line, a plane, or a volume (1 6), and the intersections

between the probe and the sampling plane provide information about the three-dimensional

relationship of the structure. The rule for the use of probes is that the dimensional quantity of

the object of interest plus the dimension of the probe must equal three. Table 1 shows how a

probe is chosen. In the first Iine, the object of interest is volume which has three dimensions,

therefore, the probe rnust have zero dimensions so the sum equals three. It c m be seen that

for surface area (dimensions = 2) the probe must have one dimension and so on for length and

number.

Table 1. Stereologic rules for the selection of a sampling probe. The object dimension plus the

test grid dimension must total three. See text for full explanation.

Object Dimension 1 Test grid dimensions l Total

Volume (3)

Surface Area (2)

Length (1)

Number (O)

Point (O)

Line (1)

3

3

Plane (2)

Volume (3)

3

3

1.2.2 Cavalieri's Volume Estimator

The most simple application of stereology is in the estimation of volume. The 17th

century mathematician Cavalieri proved that the volume of any object c m be estimated by

cutting the object into parallel slices with a constant thickness. summing the cross-sectional

area for al1 the slices and multiplying by the slice thickness. This relationship holds tnie for

objects of any shape or size as long as the first slice is randomly positioned within the volume

(1 6.74). The cross-sectional area can be measured in any way. but the most simple is to apply

a test probe of points to the surface

and count the number of points falling

on the object. Since the points on the

probe represent an area of the probe

(figure 2), which is simply the distance

between points squared, the area of

the slice can be estimated by summing

the points and rnultiplying by the area

associated with each point. Therefore,

Cavalieri's volume can be estimated

from equation 2:

~ o l u r n e = ~ ~ x d ' x h 121

where d is the distance between points

on the probe, h is the thickness of the

and ZP is the Of the points Figure 2. Sample point counting grid on an object.

falling on the object. Area (A) = distance between points squared. See

text for full explanation.

1.2.3 Volume Fraction

By using the same probe of points as in Cavalieri's volume estimation, the number of

points falling on a structure of interest within the section divided by the total number of points on

the section (figure 3) is an estimate of the areal fraction (An), or fraction of the total area

occupied by the structure (74). It has been shown that for multiple systematic-random sections

through the object, the sample now represents a volume so that the areal fraction is a reliable

estimate of the fraction cll the volume fraction (Vv), occupied by the structure (4,16,35,74,144).

Volume fraction is calculated using equation 3:

where CP is the sum of points falling on the structure and the total object respectively. Since

volume fraction is a volume ratio, the volume of individual structures can be estimated by

multiplying the volume fraction of a structure by the total volume of the organ.

1.2.4 Surface Area

The surface area of the structure is estimated using a using a probe of lines and

counting the intersections between the Iines and the structure of interest, dong with the number

of line end points that fall on the structure (figure 4) (4,74,249). The surface density (Sv), or

surface to volume ratio, is calculated from equation 4:

where U is the sum of the intersects between the structure and the line probe, P is the

sum of the line end points that fall on the structure, and Z is the length of the line. As with the

volume fraction, the surface area is calculated by multiplying the Sv by the total volume. Finally,

Figure 3. Sample point counting grid on light microscopic section of human lung. Volume fraction = number of points on lung structure divided by total number of points.

Figure 4. Sample intercept counting grid on light rnicroscopic section of human lung. Surface density = number of line intersects with lung tissue divided by number of end points on tissue multiplied by 4 divided by the length of the line. See text for full explanation.

Figure 5. Multi-level sampling design. A: Level 1. CT scan representing gross lung structure. B: Level 2, low level light rn icroscopy representing lung parench yma and airspace. C: Level 3, High level light microscopy representing alveolar wall and capillary lumen. D: Level 4, electron microscopy representing cell and extracellular matrix corn positon. See text for full explanation.

since Sv is an estimate of the surface area per volume, the inverse of Sv is an estimate of the

volume per surface area, which is another expression for thickness.

1.2.5 Multi-Level Samplina Desian

To estimate the volume fraction of small structures, the lung must be sampled using a

very high level of magnification. However, since the sample size needed to reliably estimate

the volume fraction, goes up with the magnification a method of optirnizing the sarnple size for

small structures is needed. This sampling protocol is known as the 'Multi-level", or 'cascade

design' (35) and makes use of fact that small structures are located within a larger structure

which can be quantified at a lower level of rnagnification with a smaller sample size. For

example. collagen fibrils (col) are contained within the alveolar wall (awl), which is contained

within parenchyrnal tissue (tis), which is contained within parenchyma of the lung (par). To

quantify the collagen you start at the lowest level of magnification where the structure can be

visualized, and then using increasing levels of magnification the object phase at one level

becomes the reference phase of the next level as shown in figure 5. Finally, the volume

fraction is calculated by multiplying the object Vv by the Vv of the reference space at the

previous level in a cascading manner as shown in equation 5:

Vv,,,=Vv,,,(level#)~Vv,,,(leve~3)~V~,,,(I~e~2)~V~,,,(lew~~) Pl

In summary, the lung can be reliably quantified by applying a simple design-based

sampling protocol to obtain systematic-randorn microscopic fields of view of increasing levels of

magnification. A probe of lines with end points or a probe of points is then applied to these

fields of view as shown in figure 4 and the number of intersects between the sample and the

probe is counted. Equations 3 and 4 are used to obtain the volume fraction and surface

density. Volume fractions of small objects are obtained using the cascade equation 5 which is

then multiplied by the lung volume to estimate the structural component's volume. Surface area

of lung parenchyma is estimated by rnultiplying the surface density by the lung volume, and the

mean parenchymal thickness is calculated by taking the inverse of Sv.

1.3 Quantitative Gross Analvsis Usinq Comwted Tornoaraoh~

Computed tomography (CT) sans are obtained by projecting a beam of X-rays through

the body to a detector on the opposite site which records the absorbante, or attenuation, of the

X-rays by structures within the body (38). This is performed in a complete circle around the

body so that the attenuation values are also given spatial information and the images are

reconstructed on a matrix of 51 2 X 51 2 picture elements, or pixels, which have the dimensions

of the field of view divided by the number of pixels. However, the X-ray beam also has a

thickness, known as collimation, so the pixels of a CT scan image are more appropriately

referred to as voxels, because they are actually volume elements. Therefore, the CT scan can

be used to reliably estimate the volume of the lung by surnrning the voxel dimensions, which is

analogous to the Cavalieri principle described above.

Another important aspect of CT scans is that the voxels contain information about the

linear attenuation of X-rays (p), where p is dependent on the density of the structure, the atornic

number of the structure, and the energy of the electron beam of the scanner. The attenuation

value is then converted to a Hounsfield Unit (HU) scale which is based on the attenuation of

water, according to the foliowing equation (43):

where a is equal to 1000 for the HU scale (43). From this scale, water would have a HU of

zero, air of -1000, and bone of +1000 HU (43). It has been shown that for objects with atomic

numbers in the biological range such as polyethylene foam (1 16), epoxy (41,242), bread (41 ),

wood (242), cork (1 23). and body tissue (41,l 23,156,242), the HU can be converted to

gravimetric density by assuming a Iinear relationship between X-ray values and density (41 ).

Density is then calculated by the following equation (43):

CqHUI + IO00 Density (g / ml) =

IO00 [Tl

For this reason, many studies have focused on the correlation of CT densitometry

measurements to physiologic properties, such as lung inflation and density gradients, and how

to these properties change in response to disease processes (2,8,11,32,41,51,63-

66,78,85,l23,l~~,l55,l~6,l67,l93,l9~,2~,242,Z7O)~ There are, however, artifacts in the CT

values that have effects on CT densitometry due to scanner properties such as: absorbance of

low energy electrons by dense structures (beam hardening) (38,148,276), non-linear partial

volume artifact near lung boundaries and transversing structures with markedly different

densities such as blood vessels (69,l l5,l48,2l6,242,276), and even differences associated

with different manufactures and generations of CT scanners (1 16,1 17,148). Most of these

concerns were raised in the early 1980's (69,l 38,148) when CT densitornetry was in its infancy

and investigators found that there were differences in CT values associated with scanner

manufacture (1 38), slice thickness, and reconstruction filter (69,138,148). Kemerink and

associates have recently studied modern scanners and have reported that a properly calibrated

scanner yields a reliable estimate of densrty (1 14,116,117) and that differences between

scanners can be minimized by daily calibration using water and air phantoms (1 16,117). They

have shown that results obtained by different scanners are within the reproducibility of clinical

practice associated with lung inflation (1 17). They also found that for modem scanners, there is

no difference in mean density due to magnification, slice thickness or reconstruction filter

(1 16.1 17). However. there are differences in density resolution, the ability to discriminate

materials of a different density within a histogram, that are dependent on slice thickness and

reconstruction filter (1 14). Even though high resolution CT scanning, which makes use of thin

sections and sharp reconstruction algorithms, produce an image that is qualibtively easier to

assess with the un-aided eye due to better edge discrimination and the removal of overlapping

structures, the density resolution is poor. This is due to the increase in the signal to noise ratio

of this technique which produces quantitative information with too large a variance for reliable

density resolution. Therefore, they recommend the use of sections thicker than one millimeter

without a high resolution reconstruction algorithm to differentiate different densities with the

lung (1 14).

There are also artifacts associated with patient charactenstics, most notably, the size of

breath that the patient takes during the scan. The density of the lung is very dependent on the

level of inflation (1 55,156,198,248) and differences in density can be seen between expiratory

and inspiratory scans, as well as along the pleural pressure gradient

(32,41,1 55,156,198,242,248). There have been attempts to standardize the inflation level

during the scan, using a spirometn'cally gated scan which assures that al1 images are acquired

at the same level of inspiration (1 O11O8,I95). However, this is a technically demanding

procedure and is not routinely used on clinical scans. In the clinical setting, the scans are

reported at either full inspiration (78,l~5,l~6,l98,248), or at a tidal volume above FRC in which

the patient has been asked to take a normal breath and hold it during the scan. In the 1960's

Hogg and Nepzy measured the volume of gas per gram of lung tissue in frozen exsanguinated

dogs using the following equation:

where specific volume is the inverse of density, the density of blood free tissue was measured

to be 1 .O65 g/ml and the density of lung was measured frorn its weight and volume. Since CT

scans yield an estimate of lung density, we can calculate the volume of gas per gram of tissue

in these iungs during the CT scan by applying the above equation (32). The volume of gas per

gram of tissue at total lung capacity (TLC) c m be estirnated by dividing the patients' measured

TLC by the CT estimated lung weight. The volume of gas per gram of tissue estimated by CT

can then be expressed as a percentage of TLC, which will allow the comparison of CT scans

from different patients because we know at what level of inflation the scans were performed

(32).

Another piece of useful data for the CT density is the ability to estimate volume

fractions. As was mentioned previously, the volume fraction of tissue can be estimated

histologically by applying a grid of points to a two-dimensional sample and dividing the number

of points falling on tissue by the total number of points within the lung. As has been shown, a

volume fraction, is simply a volume divided by a volume, so a volume fraction can be calculated

using volumes that are derived using any method. Therefore, since we have an estimate of the

specific volume of tissue from the literature, and an estimate of the specific volume of the whole

lung calculated from equation 8 using the CT scan, a volume fraction of tissue can be estimated

from the CT according to the following equation:

Speci/ic VolumetmC, Volume FractionIrimrC, =

Specijc volurne,, PI

In summary, there is clear evidence that the CT scans obtained from a properly

calibrated, modern scanner can be used to estimate lung volume, density, volume of gas per

gram of tissue and the volume fraction of tissue and airspace of the lung. The purpose of this

thesis is to use these parameters to assess the normal lung and the changes that occur in

diseases such as fibrosis and emphysema.

Chapter 1 provides background for two techniques useful for the quantitative

assessment of lung structure. Stereology allows three-dimensional information to be obtained

from systematic-random histologie samples of the lung. It is unbiased, and efficient, but very

invasive in that it relies on lung tissue obtained at autopsy. Although resected lung lobes can

be evaluated by this technique it is not suitable for analysis of lung biopsies because of the

collapse of the specimen which makes re-inflation to an in vivo level very difficult. Quantitative

analysis of the X-ray attenuation data obtained by computed tomography is minimally invasive

and provides estimates of gross lung characteristics such as volume. density and weight, as

well as estimates of the volume fraction of tissue and airspace. The goal of this thesis is to

combine these two techniques with a view to validate the less invasive computed tomography in

terms of the more invasive stereology.

2.1 Workina Hv~othesis

The working hypothesis of this thesis developed from a need for a non-invasive method

for quantifying the structural changes in the lung in chronic disease:

The combination of c o m ~ ~ t e d tomoara~hv scans and stereoIoaic auantification of

histolooical soecimens allows the assessment of luna tissue chanoes in the chronic lunq

diseases idio~athic ~ulmonarv fibrosis and em~hvsema with minimal destructive im~act

on the ~atient.

The goal of this study is to move analysis of Cf scans beyond the qualitative observations

currently made by diagnostic radiologists and make them more useful to clinical physiologists

and physicians in quantifying structural defects in a way that will be useful to establishing the

natural history of the disease and masure the effect of treatment.

2.2 Specific Aims

1. To use lung volume and density rneasurements obtained by CT to measure differences in

regional lung expansion and calculate the pleural pressure gradient.

2. To validate measurements of lung structure observed on CT with the structural studies

based on quantitative histology.

3. To quantify the structural defects in chronic interstitial lung disease and emphysema using

the cornbined CT and quantitative histological approach

2.3 Strate~l

Specific Aim 1 and 2 will be accomplished on patients undergoing lung resection for

bronchogenic carcinoma. Pre-operative study of these patients established that their lung

function was within normal limits. Specific aim # 3 will be accomplished using open lung biopsy

specirnens from patients with idiopathic pulmonary fibrosis (IPF), chapter 4, and surgically

resected specimens from patients with emphysema (Chapter 5).

2.4 Summay

The analysis of X-ray attenuation values from a CT scan is combined with the estimates

of lung structure obtained from a histologic quantification of the lung specirnen. This procedure

allows the correlation of lung structure and function using techniques that are minimally invasive

to the patient. It also adds information about the process of lung re-modeling caused by

chronic lung disease.

3.1 ûescriptions of the Lunq

A quantitative analysis of the lung is an extremely complex problem because of the

sizes of the tissue components which range from several centirneters to only a few micrometers

and their intricate three dimensional arrangement. The early studies of the lung were

descriptive and as imaging techniques improved with the refinement of light microscopes and

the invention of the electron microscope, the structure of the lung and the cellular and extra-

cellular composition have been described in great detail (1 59,256). One of the first quantitative

histologie studies of the lung was published in 1731 by the Reverend Stephen Hales (256) who

reported calf lung alveoli to be cuboid boxes about 1/100 part of an inch in diameter (254 pm).

From these measurements he was able to estimate the surface area of the lung to be

approximately 27 m2 which led him to conclude that this enormous surface area made it very

probable that oxygen entered the blood through the lung rather than through food (256). This

was a major shift in how people viewed the function of the lung and led to many more

quantitative studies which attempted to relate lung structure to function. However, the biggest

problem with any quantitative study is how to maintain the three dimensional structure of the

lung microanatomy while obtaining accurate and reliable measurements. This was first

undertaken by cutting serial sections of the lung and then making three dimensional

reconstructions of the images, as described by William Snow-Miller in his book The Lung, which

is one of the first complete quantitative studies of the human lung (159). Another technique

was to dry the lung so the walls would become opaque and then use a microscope to look

through the pleural surface at the intemal structures (159). Still another approach for the study

of the ainnrays was to create casts of them using a material such as Woods Metal which could

be poured into the airways of the lungs and then polymerized and the surrounding material

removed. The Swiss anatomist Christoph Theodor Aeby used this approach in the 1870s to

make painstaking measurernents of the bronchi and their branching pattern (256), and even

though his conclusions about the monopodial branching pattern ran counter to Kolliker, who

was the first to describe the complete epithelium in the alveoli, Aeby's study is considered the

first quantitative analysis of the airway structure (256).

The last century has seen great improvements in the quality of microscopes used to

image the lung and nowhere is this more evident than with the advent of the electron

microscope which has allowed investigators to visualize structures down to the macro-

molecular level. However, the intrinsic problem of quantification with the microscope was still

that they were very time consuming, labour intensive and reduced the three dimensional

structure of the lung to two dimensional sections. Perhaps the greatest advance in the

quantification of the lung came with the application of stereological methods, as outlined in

Chapter 1, which began in 1961 when Hans Elias assembled the International Society of

Stereology. This society brought together investigators from biology. geology, metallurgy and

mathematics to develop and refine techniques for the reliable and efficient quantification of

three dimensional structures (256). Ewald Weibel in his book Morphometry of the Human Lung

(249) applied these new methods to a quantification of the hurnan lung and from this point on

quantitative morphology of the lung has produced unbiased estimates of virtually al1 aspects of

lung structure.

The physiological description of the lung has been under intense scrutiny for hundreds

of years but many of the advances have corne since the Second World War (151,262). During

this time investigators have developed techniques to quantify the functional aspects of the lung

and described the important principles of airfiow, the pleural pressure gradient and gas

exchange.

This chapter provides a bnef overview of lung structure and function and then uses

modern techniques to quantify the human lung.

3.1.1 Gross Lunci Structure

The hurnan lung consists of two separate lungs located (anatomically) on the left and

right side of the thoracic cavity connected to the trachea by the main stem bronchus. The right

lung has three lobes: the upper, rniddle and lower, while the left side has two lobes with the left

upper lobe containing the lingula which is analogous, but smaller in volume, to the right middle

lobe. The lobes are separated from each other by a wrapping of visceral pleura which forms a

major fissure on the left and a major and minor fissure on the right but this separation is often

incomplete allowing collateral ventilation between lobes. The lung is further partitioned into

smaller units towards the periphery using the branchial anatomy as a basis for division and

nomenclature. A lobar segment is the next major division of the lung and consists of the lung

that is supplied by the second division of the main stem bronchus. The secondary lobule

measures 1-2.5 cm in diameter and can be visualized either on the cut surface of the lung, or

by CT, and consists of a bronchiole and artery in the lobular core bounded by tissue septae that

are continuous with the visceral pleura and contain the pulmonary veins and lymphatics (247).

Each secondary lobule contains three to five acini which are described as the complex of al1

airways distal to the terminal bronchiole (75). mis structural definition is compatible with a

functional definition of the largest lung unit in which al1 airways participate in gas exchange (75).

The final region of the lung is the alveoli which are blind air sacs containing the greatest surface

area to volume ratio in the lung and are surrounded by thin walled capillaries which is the

primary site of gas exchange.

3.1.2 Cellular Luna Structure

There are 24 different cell types in the lung (table 2) (255) which are arranged into three

layers: the epithelial layer, which is in contact with the airspace of the external environment, the

endothelial layer, which is in contact with the blood, and the interstitial cornpartment which both

separates and binds these layers together.

In addition, a fourth group of cells, the blood cells, move through the lung and transport

Me oxygen to the tissues, remove the carbon dioxide from the body, and combat infection. The

red blood cells are responsible for transporting the oxygen to the tissues. The red colour is

produced by the hemoglobin molecules which bind or release oxygen depending on the

gradient surrounding the cells. The cells responsible for the elimination of foreign substances

are the white blood cells which include the polymorphonuclear cells (PMN), monocytes.

lymphocytes, eosinophils, and

basophils. Al1 of these cells

develop from a cornmon

progenitor cell in the bone

marrow and then differentiate

into specialized cells. For

example, the lymphocytes are

the primary immune cells

which produce cytokines to

direct the host response and

antibodies to combat foreign

particles. Eosinophils are

particularly effective against

parasitic infections and

Table 2. Cellular composition of the lung (250)

Structure

Parenchyma

Nonparenchyma Airways

BIood Vessels

Total Alveolar Type I Alveolar Type Il Endothelium Mesenchymal Total

Ciliated Glandular

% of Total Lung cells

86 4 6

33 43 14 5

2-3 <1 9

Connective Tissue Structure Component % of Total Lung

Connective Tissue Parenchyma 1 Total 1 62

Nonparenchyma

Collagen Elastin Total Collagen Elastin

46 16 38 28 10

basophils are capable of generating leukotrienes and supplement rnast cells in immediate

hypersensitivity reactions (1,106). However, the first line of defense are the PMN which are the

most abundant white blood cells in the blood and migrate quickly from the blood in response to

a chemotactic stimulus to phagocytose the foreign particles, and release proteolytic enzymes

and chemotactic substances which direct the rest of the immune response. The monocytes

follow the PMN into the tissue from the blood but then they differentiate into the phagocytic

alveolar macrophages which are long Iived in the lung and release important cytokines that

direct the host response to both living and inert material entering the airspaces.

The first of the two types of epithelium cells are the lining cells, which are ciliated in the

central airways for the movement of the mucus. ln the periphery of the lung, the alveoli are

lined by the alveolar type I cells which cover the rnajority of the alveolar surface area even

though they only account for a small percentage of the total number of cells in the lung (table 2)

(253,256). The second cell type are the glandular (secretory) cells which include the goblet

cells in the trachea and the bronchi, and the Clara cells in the bronchioles (76,105,266) whose

function is to secrete mucus which lines the airways and traps inhaled foreign particles

(76,105,266). In the alveoli, the type II cells secrete surfactant which lowers the surface tension

thereby preventing the collapse of the alveoli and reducing the work required for lung inflation

(76,105,253,266). The endothelium is a simple squamous layer of ceils which is extremely thin

in the alveolar capillaries to allow optimal exchange of gases (76,105,253,266).

The interstitium is the space between the basement membranes of the epithelium and

the endothelium and consists of fibroblasts and pericytes as well as cells of the immune system

such as mast cells and plasma cells and the extra-cellular matrix. The fibroblasts in the lung

are responsible for the synthesis of the extra-cellular matrix which gives the lung its structural

properties (1 25.1 39.1 49.1 78,2Ti3. During a fibroproliferative response, the fibroblasts have

been shown to stain positive for smooth muscle actin and have contractile properties

(3.1 12.1 25.1 36) so that they are postulated to be important for wound contraction (1 25,136).

Their exact role within the normal lung has been postulated as regulators for capillary blood

flow, compliance of the interstitial space, and tissue elasticity of the lung parenchyma as well as

turnover and maintenance of the extra-cellular matrix (1 10,112). Pericytes are cells which are

associated with the alveolar capillaries and may be a special differentiation of the smooth

muscle cells, or of a completely different cell line but these cells have been shown to have

contractile propeicies which rnay be responsible for regulating capillary blood flow (1 10,215).

The major cell line of the a i m y and blood vessel interstitium is the smooth muscle cell and

their contractile ability is responsible for the conducting properties of the airways and vessels by

reacting to nerve and chernical stimuli to dilate or constrict the caliber of the aiway or vessel

they surround.

3.1.3 Extra-cellular Matrix

Table 2 summarizes the state of knowledge up into the 1970's when elastin and

collagen were considered the major components of the extracellular matrix. The last twenty

years have seen an explosion in the identification and description of other important molecules

within the interstitium, most notably the proteoglycans (PG) and laminin

(1 3,18,19,50,77,163,200,234,235,268).

At least 19 different collagens have been described to date, of which three are important

in the interstitium of the lung, the fibrillar collagens type I and III and the sheet forrning collagen

type IV of the basement membrane (1 39,191,236). The collagens are a trimer assembled in an

a-helix with each molecule of the helix consisting of repeating chains of glycine-X-Y where X is

predominately proline and Y is often hydroxyproline (1 39,236). The individual collagen

molecules are assembled within the Golgi of the fibroblasts, and secreted in a pro-collagen

f o n into the extratellular matrix (1 39,143,277) where they are cleaved and assembled into

collagen fibers (139,143). For example, the type III collagen is a cylindrical fiber 40-200 nrn

thick which shows a characteristic banding pattern of 64 nm periodicity. by electron microscopy.

This pattern is due to the fiber consisting of the collagen fibrils arranged in a quarter stagger

pattern resulting in the negative charges of the fibrils overiapping to produce a high affinity

binding of the positively charged heavy metal stains used in electron microscopy (139,236).

The collagen fibers have a very high elastic modulus that gives the lung its tensile strength

(1 32,139).

Elastin fibers contain a core of polymeric insoluble elastin molecules with a mantle of

microfibrils (1 52,211 ). The two identified microfibrils are the glycoproteins: microfibrillar

associated glycoprotein and fibrillin (152). The protein structure of elastin contains a high

concentration of hydrophobic amino acids like valine with a low content of acidic and basic

amino acids and large numbers of lysine derivatives that provide cross linkage (1 52,211). This

composition makes elastin extremely insoluble and the extensive cross linking renders it

resistant to degradation (1 52). Elastin is usually found as amorphous bundles of dense staining

material due to its large negative charge and hydrophobic characteristics (1 52,211). It is very

Iikely that elastin and associated rnicrofibrils are synthesized early in the developrnent of the -

organ and then remain relatively constant throughout the life of the organism, although the

number of rnicrofibrils does appear to decrease with age (128,152,189,211). There are reports

of new elastin synthesized in disease conditions such as atherosclerosis (220) and drug

induced pulmonary fibrosis (22). In contrast to collagen. elastin has high extensibility and low

tensile strength (1 52).

The proteoglycans (PG) are a large group of molecules defined by a protein core

with attached side chains of glycosaminoglycans (GAG) which are repeating disaccharides

(77,268). These molecules have been shown to be important for tissue hydration

(77,200,220,268), cell migration (1 3,29,7ï,268.274), and the binding. and possible regulation,

of growth factors within the tissue matrix (18,50,268).

The early descriptions of how lung structure was related to function focused mostly on

the collagens and the elastin since these are the largest of the extraçellular molecules and the

easiest to characterize. However, it has becorne apparent that these molecules are not the

whole story for the lung and it is now clear that proteoglycans play a large role in lung structure.

especially in the formative and the repair phases.

3.2 The Clinical Measurement of Luna Function

Lung function is measured in ternis of static lung volumes, recorded at predeterrnined

points in the respiratory maneuver, and dynamic lung volumes, recorded during the forced

expiration. Total lung capacity (TLC) is defined as the volume of gas in the lung at full inflation

and the residual volume (RV) is the point where no more air can be forced from the lung by the

action of the expiratory muscles (261). Functional residual capacity (FRC) on the other hand is

the equilibrium point reached at the end of a quiet expiration where the inward force of the lung

recoil is balanced by the outward force of the chest wall (261). These volumes are most

accurately measured by seating the subject in an airtight body plethysrnograph and measuring

the patient's FRC using Boyle's law (44, which states that at a constant temperature the

volume of any gas varies inversely as the pressure to which the gas is subjected. For this

study, the subject inhales maximally and then fully exhales to RV. The measured volume of

gas that was inspired is referred to as the inspired capacity (IC) and the exhaled volume the

vital capacity (VC). TLC is then calculated by adding the IC to the FRC, while RV is the TLC

minus the VC (261).

The dynamic lung volumes are recorded during a forced expiratory maneuver where the

subject inhales to TLC and then forcibly expires as quickly as possible to RV. The expired

volume is referred to as the forced vital capacity (WC) and the volume expelled during the first

one second of the rnaneuver is referred to as the forced expiratory volume in one second

(FEV,). In normal subjects, the ratio of the FEU1 to FVC is approxirnately 80%, and this ratio

can be decreased by airway obstruction that decreases the F N l , and increased by restncted

lung volume that reduces the W C . The characteristics of obstructive iung disease are:

decreased dynamic lung volumes and FEVl/FVC ratio due to airflow obstruction or reduced

driving pressure, and increased static volumes due to gas trapping because of the reduction of

expiration. Another form of lung dysfunction is referred to as restrictive lung disease in which,

as the name suggests, the lung is restncted by a stiffening of either the chest wall or lung

parenchyma so that the static lung volumes and the W C are decreased which causes an

increase in the FEVl/FVC ratio even if the FEV, is minimally reduced (261).

The ability of gas to diffuse from the alveoIar air into the biood is measured by the

diffusing capacity (DLCO) which determines the movernent of a trace amount of carbon

monoxide (CO) from air to blood. In this test, the subject inhales a mixture of CO and He (0.3%

and 10% respectively) and breath holds for 10 seconds (1 57,261). The subject then exhales

and after discarding the volume of dead space gas in the central airways, the concentrations of

CO and He are measured. Since He is not absorbed by the blood, the difference between the

inspired and expired concentration of He is used to calculate the initial alveolar concentration of

CO by gas dilution. Carbon rnonoxide. on the other hand, is absorbed into the blood at

approxirnately the same rate as oxygen so that the difference between the initial alveolar and

expired CO concentration indicates the amount of CO absorbed by the blood. The diffusing

capacity can be influenced by many factors including a change in the lung surface area, the

volume of the blood in the alveolar capillaries and the tirne spent by the erythrocytes at the air-

blood interface (257,261 ).

3.3 The Pleural Pressure Gradient

Investigation of the lung with radioactive gases established that ventilation of the lung is

uneven (109,154,261). Milic-Emili and coworkers showed that when lung volume was

increased the apical regions of the lung showed an initial greater change in volume cornpared

to the basal regions, which then reversed at higher lung volumes (1 09,154,261). These and

other studies established that regional ventilation was influenced by a gravity dependent pleural

pressure gradient which causes upper regions of the lung to be expanded more fully than lower

Iung regions due to a higher trans-pulmonary pressure. It has been hypothesized that this

pressure gradient is caused by the weight of the lung below the given region which must be

supported by the lung parenchyma, sornewhat Iike a spring which is suspended at the top

(68,91,109). The distance between the coils of the spnng is largest at the top and decreases

towards the bottom because to the weight of the spring on each coi1 which is greatest at the top

and minimal at the bottom. Since the upper regions are exposed to this higher pressure, they

are more inflated at lower lung volumes and inflate first upon inspiration (1 56). This hypothesis

is supported by Glazier and colleagues who undertook a morphometric analysis of alveolar size

to show that alveoli in the upper regions of the lung are larger than alveoli in the iower regions

at FRC (68) and Hogg and Nepszy who obtained sirnilar results using densRy measurements of

frozen dog lungs to show that the non-gravity dependent regions have a greater volume of gas

per weight of tissue than the dependent regions (91). Gravity also influences pulmonary

perfusion (109,263,265) and led West to desctibe lung perfusion in terms of specific zones. In

Zone 1 the alveolar pressure is greater than the arterial pressure and the venous pressure,

resulting in compressed capillaries and reduced blood flow. In Zone 2 the arterial pressures are

greater than the alveolar pressure which is greater than the venous pressure and blood flow is

now determined by the arterial and the alveolar pressure with the venous pressure being

irrelevant. In Zone 3, the arterial pressure is greater than the venous pressure which in turn is

greater than the alveolar pressure and results in maximal dilatation of the vessels and

maximum blood flow. These factors al1 contribute to a difference in lung density which al1

investigators state can not be appreciated unless the lung is fixed in vivo (closed chest) so that

the pleural pressure gradient is intact.

The introduction of CT in the 1980's allowed differences in regional lung volume to be

appreciated from measurements of lung density (82.107.1 55,156). These investigators report

that the non-dependent portions of the lung are less dense than the dependent regions, and

that this relationship is dependent on body position during the scan.

3.4 Experiment #1

In this study we have used CT densitometry to measure the pleural pressure gradient in

human subjects undergoing surgery for bronchogenic carcinoma. We have combined the CT

measurements of total and regional lung volume with quantitative histology to measure the

structure of the human lung and to develop a technique to correct biopsy specimens to an

appropriate level of inflation within the thorax to alleviate the problems associated with collapse

of histologie specimens.

3.5 Material and Methods

Studies were performed on 19 subjects who are part of an ongoing study of lung

structure and function at the University of British Columbia in which patients requiring lung

resection for a srnall peripheral tumor are studied just prior to surgery. To be included in the

study the patient's forced expiratory volume in one second (FEV,), forced vital capacity (FVC).

FEVJWC ratio. diffusing capacity for carbon monoxide (DLCO) and total lung capacity (TLC)

had to be within the nomial range. In a preliminary study, the images from 10 patients were

used to evaluate the point counting technique by comparing the size of the tumor on the CT

image to its size in the resected specimen (Group 1). In the nine remaining patients, in whom

the CT X-ray attenuation data were available, regional lung volumes were calculated.

3.5.1 Pulmonarv Function Studies

Spirometry, lung volumes and lung pressure-volume curves were measured in the

patients from both group 1 and 2 seated in a volume displacement body plethysmograph.

Functional residual capacity (FRC) was measured using the Boyle's Law technique. TLC was

calculated by adding inspiratory capacity (IC) to FRC. Residual volume (RV) was calculated by

subtracting vital capacity (VC) from TLC. In six of the nine patients in group 2, trans-pulmonary

pressure was measured using a differential pressure transducer (45 mp I 100 cmH20; Validyne;

NoRhridge, CA) to compare mouth pressure to intra-thoracic pressure, measured with an

esophageal balloon and PV curves were constructed by comparing these values to the

simultaneously rneasured lung volume (146). In the remaining three patients in Group 2, the

subdivisions of lung volume were determined using a dry rolling seal spirometer and the

multiple breath Helium (He) dilution technique. Spirometry and He dilution FRC were

performed on a P.K. Morgan computerized pulrnonary function testing system (P.K. Morgan,

Boston, Mass.). DLCo for al1 9 subjects was measured by the single breath method of Miller and

associates (1 57) on the P.K. Morgan automated diffusing capacity analyzer. The results are

corrected for alveolar volume (VA) and reported as NA.

3.5.2 CT Studies

All the patients had a conventional C f scan (10 mm thick contiguous slices) without IV

contrast on a GE 9800 Highlight Advantage C f scanner (General Electric Medical Systems,

Milwaukee, WI) approximately one week prior to resection. All scans were perfonned with the

subject supine during breath holding following inspiration. Conventional scanning parameters

in use at our institution were used (1 20 kVe, 100 mA, 2 sec, reconstructed on standard

algorithm). The images for al1 19 patients were printed onto standard radiological film using a

window of 1200 and a level of -700 HU. For the nine patients studied completely, the CT data

was transferred to a Sparc2 Workstation (Sun Microsystems, Mountain View CA) and the lungs

were segmented out of the chest using the program Medical Image Viewer (Arkansas

Children's Hospital Little Rock, AR, GE Medical Systems) using X-ray attenuation values of - 1000 to -500 HU. The volume of the whole Iung and the individual lobes was calculated using

the Cavalieri principle (1 53.1 83). This was accomplished by summing the segrnented pixel

area in each slice and multiplying by the slice thickness to get total lung volume. The horizontal

and oblique fissures were noted on each slice and the lung volume was apportioned to upper.

lower and rniddle lobes (figure 6). The segmented images were then passed to the numencal

analysis package, PV- Wave (Visual Numerics, Boulder CO), where a pixel wide stnp around the

lung was subtracted away to eliminate partial volume artifact, due to the cuwature of the lung

and the chest wall, and each lung slice was divided into 16 ml sections (40 X 40 X 10 mm;

figure 7). The mean X-ray attenuation of each of these sections was calculated, as well as the

vertical distance from the middle of each section to the most gravity dependent (posterior)

portion of the lung. The CT values were converted to true lung density (@ml) by adding 1000 to

the CT value and dividing by 1 0 (equation 7) (82). The weight of the lung or lobe to be

resected was calculated by multiplying the mean density of the lung (or lobe), by its volume

calculated by the Cavalieri principle. The mean total lung gas volumes per gram of tissue at

TLC, FRC and RV were detemined by dividing the physiologically measured values of lung gas

volume at TLC, FRC and RV by the total lung weight calculated from the CT scan. The X-ray

attenuation for each 16 ml cubes was then used to calculate the volume of gas per gram of

tissue for that cubic region of the lung according to equation 8.

Figure 6. CT scan of human lung showing segmentation of the different lobes. A: unsegmented CT scan, B: segmented CT scan. RUL: right upper lobe, RML: right middle lobe, RLL: right lower lobe, LUL: left upper lobe (contains the lingula), LLL: left lower lobe.

31

Figure 7. A representative CT slice showing the 40 X 40 X 10 (slice thickness) mm sampling regions drawn on the segmented lung. The mean X-ray attenuation for each of these regions, as well as the distance to the middle of the region f rom the most posterior (gravity dependent) reg ion was calculated.

3.5.3 Quantitative Histoloqy

The lobectorny specimens were obtained directly from the operating room and taken to

the laboratory where they were weighed. inflated with Optimal Cutting Temperature (OCT)

cornpound (Miles Laboratones, Elkhart, IN) diluted 1 :1 with normal saline, re-weighed, and

frozen in liquid nitrogen. Once frozen, the specimen was cut into 2 cm thick slices in the

transverse plane on a band-saw and cores of lung 2 cm in diarneter were sampled with a power

driven hole-saw. These frozen cores of lung tissue were stored at -70 O C for other purposes.

The remainder of the specimen was transferred to 10% fomalin and fixed at room temperature

for at least 24 hours. Samples were taken from these specimens and processed into paraffin,

sectioned at 5 V r n and stained with hematoxyiin and eosin for quantitative histologic analysis.

The histologic slides of the lobectorny specirnens from the nine Group 2 patients were

quantified using a cascade-design technique to estimate the volume fraction (Vv) of airspace,

parenchymal tissue and blood vessek, as well as the surface density (Sv) and thickness of the

parenchyma as described in Chapter 1. In summary, this technique allows the volume fraction

of very small components to be estimated by using increasing levels of rnagnification so that

larger objects are subdivided into their components at each successive level (35). Volume

fractions are estimated by casting a grid of regular points on the field of view and counting the

number of points that fall on each lung component and the total number of points on the lung

(1 6). Equation 3 is then used to calculate the volume fractions of each lung component.

In the classical description of the technique, the first level was determined using the

unmagnified gross specirnen (35). Therefore, to determine if the C f image would substitute for

the gross specimen, the volume fractions of parenchyrna, bronchovascular bundle and the

tumour in the lobes of Group 1 patients was compared with the resected gross specimen. A

grid of points was cast on each slice of the frozen fixed lobes and the CT images of that lobe of

the 10 Group 1 patients. Once this technique was verified, the full cascade design was

implemented on the Group 2 patients using the CT scan image for level 1.

Levels 2 and 3 were perfonned at the light microscopy Ievel using the point counting

program Gridder (Wilrich Tech. Vancouver. B.C.) which generated random fields of view,

projected a grid on to the field of view via a camera-lucida attachment on a Nikon Labophot

Iight microscope and tabulated the counts. Level2 used lOOx magnification with a grid of 80

points (d = 0.1 1 mm ) and 40 lines (1 = 0.1 1 mm). The nurnber of points falling on airspace,

tissue (lung parenchyma), and medium sized blood vessels (50-1 000 pm) as well as the

number of intersects behveen the grid lines and the parenchyrnal-airspace interface were

tabulated. The surface density of the parenchyma was estimated using equation 4. Since

surface density is the surface area in a given volume, the surface area of the parenchyma is

calculated by multiplying the surface density by the volume of the lung (1 6). The mean

parenchyma thickness is calculated by taking the inverse of the surface density. Level 3 was

perfomed on 10 random fields of view per slide at 4UOx magnification and the number of points

falling on airspace components (Alveolar macrophages, alveolar PMN, other objects in the

airspace, and empty space) as well as the tissue components (alveolar wall, capillary lumen,

and small blood vessels (20-50 pm)) were counted using a 100 point grid. The volume

fractions for al1 the lung components at level2 and 3 were calculated using equation 3, and the

overall VV for the individual lung cornponents was calculated by multiplying the Vv of that lung

component by the VV of the component that contained it in the previous levels using equation 5

as described eariier.

The volume fraction of the lung that is gas and tissue was also estimated by dividing the

specific volume of tissue by the specific volume of the lung obtained from the CT scans

according to equation 9. The volume fraction of tissue and gas estimated using CT (Vvo)

compared to the Vv of tissue estimated a i Level2 using the morphometric technique to

determine the validity of this technique.

3.5.4 Statistical Analvsis

All data were analyzed using independent t-tests or the one way analysis of variance.

Transformations were made on certain variables to norrnalize distributions and to make

variances homogeneous. A Bonferroni sequential rejective procedure was used to correct for

multiple comparisons (94). A conected p-value of less than 0.05 was considered significant.

3.6 Results

Table 3 shows anthropometric and lvng function data for al1 19 patients where group 1

(n=10) represents the patients in the preliminary study done to evaluate the point counting

method and group 2 are the 9 patients in which the lungs were cornpletely analyzed for volume

and density. The data for these 9 patients (table 4) shows that on average the right lung

accounted for 55 i 2 O/O and the left lung 45 I 2 5% of the total lung volume with the right upper

lobe (RUL) accounting for 21 + 5 %, right middle lobe (RML) 9 I 3 %, right lower lobe (RLL) 25

I 3 %, left upper lobe (LUL) including the lingula 24 I 4 % and left lower lobe (LLL) 22 + 5 Oh.

Table 5 compares the calculated weight and volume of the lobe to be resected with the fresh

weight of the resected specirnen and its volume after inflation with cryo-protectant rnaterial.

This shows that the fresh weight of the resected specirnen was 76 I: 19 % of CT calculated lobe

weight and that the inflated volume of the specimen was 94 î 36 % of the volume determined in

vivo.

The volume of gas per gram of tissue was calculated from the physiologically

detemined TLC, FRC, and RV gas volume divided by the lung weight (1 069 I 327 g) and is

shown in table 6. The lung weight was calculated by dividing the CT estimated lung density by

the CT estimated lung volume. The average volume was 6.0 * 1.1 mVgrarn ai TLC, 3.6 * 0.9 at

FRC. and 2.4 I 0.8 at RV and the lung volume at which the CT was obtained was 3.8 I 0.8 mVg

or 65.5 I 11.9% of TLC. Regional lung expansion calculated in ml of gas per gram of tissue

from the CT density was expressed as a percent of TLC. When this measure of lung

expansion is plotted against lung height (figure 8) the regions of the lung lower in the

gravitational field (posterior in a supine scan) were less well inflated than the upper (anterior)

regions with a mean slope of 1.8 * 0.09 O/oTLC/cm. Figure 9a shows the individual pressure

volume curves for the six patients in Group 2 where the volume component is expressed as the

volume of gas per gram of tissue and CT derived values of volume of gas per gram of tissue

catculated from the 16 ml cubes of the CT scan are shown as individual points. Figure 9b

shows the same data expressed as a percent of each patient's TLC. Figure 10 shows the

mean pleural pressure gradient calculated using the PV curve in figure 9b and the lung volume

data in figure 8 and has a slope of 0.24 ~t 0.08 cmH20/cm.

Cornpanson of volume fraction of lung parenchyma, large blood vessels and tumor

obtained by point counting on the CT scan and directly on the resected specimen

(table 7a) showed an over estimation of blood vesse1 volume fraction on the CT scan compared

to the lobectomy specimen, with a conesponding decrease in the tissue and tumor fractions.

The full cascade-design stereologic analysis of the whole lungs (table 7b) estimated a mean

surface area of lung parenchyma to be 101.8 t 35.1 m2, which gives a mean thickness of 7.2 r

1.0 Pm. At this level of inflation, the lung is cornposed of 60 I 4 % airspace, 26 I 4 % blood

vessels (including capillaries), and 14 I 3 % parenchymal tissue. The fraction of the lung

occupied by parenchyma that includes tissue, capillaries, small and medium sized blood

vessels estimated by the CT was slightly higher at 22 I 5 % than the 17 I: 5 % estimated from

the level2 quantitative histology.

TABLE 5:

Lobar weight and volume

Resected specimen vs. CT

CASE LOBE

LUL

LLL

LUL

LL

LLL

LUL

RML

RLL

RUL

WEIGHT (g)

SPECIMEN"

* calculated from inflated weight

** prior to inflation with OCT

VOLUME (ml)

SPECIMEN'

Left upper lobe: LUL, left lower lobe:LLL, right upper lobe: RUL, right middle lobe:RML, nght

lower lobe:RLL, and left lung:LL.

$ Significantly greater than the lobectomy weights, P < 0.05.

Total Lung Weight = 1069 I 327 g

The CT scans were obtained at 65.5 r 1 1.9 % TLC. i.e. just above FRC where the lung

contained an average of 3.8 i 0.8 ml of gas per gram of tissue. Total lung capacity: TLC,

functional residual capacity: FRC, and residual volume: RV.

TABLE 6:

Lung volume and Gas per Gram of Tissue:

TLC

FRC

RV

Liters

6.1 I 1.4

3.7 I 1 .O

2.5 I 1 .O

mug

6.0 I 1.1

3.6 * 0.9 2.4 I 0.8

% TLC

1 O0

59.6 * 7.2

40.2 I 10.9

10 15

Lung Height (cm)

Figure 8. A graph of the CT density of the lung expressed as ml of gadgram of tissue and expressed as a percent of TLC against lung height in cm measured from the most posterior portion of the lung, measured according to Figure 7, and calculated according to equation 8. As the data for the nine patients was not statistically different, the data was pooted, and the random effects mean linear regression is ptotted ( r d line), with the 95% confidence intervals of the line (blue lines).

Pressure (cm&O)

Figure 9. A graph of the pressure volume curves of the 6 patients with trans- pulmonary pressure measured, plotting the CT derived volume measurement of ml of gaslgram of tissue against the transpulmonary pressure (cm of H,O). The individual curves are shown in A and the same curves are shown in B when corrected for percent of individual TLC.

10 15

Lung Height (cm)

Figure 10. A graph of the pleural pressure gradient in the 6 patients from Figure 9. The pleural pressure is estimated from Figure 9 and the lung height is estimated from Figure 8. The red line shows the random effects mean linear regression with the 95% confidence intervals of the line as the blue lines.

3.7 Discussion

The values obtained for lung and lobe volume (table 4 and 5) using this technique are

similar to reported values from autopsy specimens (1 59,188,249), radiographic, bronchoscopic,

scintigrams and inert gas dilution (1 88). The technique also provides the advantage that it is

minimally invasive to the patient and no assumptions are made about lung shape. The fact that

the weight of the lobe calculated from the Cl density was significantly greater than that of the

resected specimen, can be partially attributed to the loss of blood from the specimen during

and after resection. However, as the volume of the OCT inflated specimens was 94 I 36 O' of

the volume calculated from the CT, the specimen was inflated close to it's volume within the

intact thorax.

The technique described here is less technicaily demanding than those that seek to

track lung volume during CT using spirometry and the results show that on average the entire

lung was inflated to 65.5 I 11 -9% of TLC during the scan. The regional lung data also show a

linear relationship between lung volume expressed as a percent of TLC and height within the

thorax (figure 8) with a slope of 1.8 I 0.1 O/oTLC/cm calculated using the restricted maximum

likelihood analysis (53). This confirms the classic findings of Milic-Emili et al (1 54) and Kanako

et al (109) using eno on'? This difference in lung volume is due to a gradient in pleural

pressure where the dependent portions of the lung are at a lower transmural pressures than the

upper regions (figure 9). Our data shows that this pressure gradient was 0.24 I 0.08

cmH20/cm (figure 10) which compares very favorably to the 0.2 cmH,O/cm that Milic-Emili

reported on healthy volunteers (1 54). This rneans that at any level of inflation, the upper regions

of the lung are more inflated than the lower regions. As Millar and CO-workers (1 56) found that

tissue volume including vessels under 5 mm in diameter was unifonn from the top of the Iung to

the bottom, the obsemed change in lung attenuation must be due to regional differences in gas

inflation. These observations confirm others in the Iiterature (1 55,l56,198,242,248) and

extends them by providing a simple quantitative method for estimating the regional differences

in lung expansion (91). The individual PV curves can also be used to determine the local

transmural pressure at each lung height and calculate the pleural pressure gradient within the

thorax. Furtherrnore comparisons of the regional lung volumes at a given lung height allows a

determination of whether the lung volume of each region is appropriate for that lung height or

not. This approach should be useful in defining emphysematous destruction where the regional

volume will be shifted up and chronic interstitial disease where the regional volume should be

shifted down.

The volume fraction of blood vessels obtained by point counting the CT images was

overestirnated compared to point counting directly from the lung surface (table 7a). This could

be related to two factors. Firstly, the volume fraction estimates rely on a slice with zero

thickness, such as the cut surface of a physical slice, but the CT image is an average of al1 the

components within the slice. Therefore, the orientation of the vesse1 in the slice. may cause its

volume fraction to be overestimated. This is known as the Holrnes effect in the stereologic

literature (4), and volume averaging in the radiological literature (165). Secondly, part of the

difference between the CT and specimen results could also be due to the inevitable loss of

blood from the specimen as a result of surgery. The large size and relatively spherical shape of

the tumor in relation to the slice thickness reduces the Holmes effect and accounts for the good

agreement between the estimate of tumor volume from both the CT and the resected

specimen. Therefore, we attribute the overestimate of volume fraction of blood vessels to the

problem of point counting of objects that are significantly smaller than the thickness of the CT

slice. This problem can be rninimized with high-resolution CT which was not available for this

study.

The surface area of the entire lung of the nine patients in Group 2 (table 7b) was

estimated at Level 2 of the histologic analysis by counting the grid line intersects with lung

parenchyma and multiplying by the volume of the lung obtained from CT. This shows a mean

surface area of 101.8 * 35.1 m2 and a mean parenchyma tissue thickness is 7.2 I 1 .O Pm (table

7b). These values are consistent with published data on pst-mortem lungs (249) and previous

studies from ourlaboratory (89). At the level of lung inflation achieved during the CT scan

(65.5% of TLC) 60.1% of the specimen consisted of air. The parenchymal volume fraction,

which is the sum of the capillary, medium sized blood vessels (c l mm) and actual lung tissue

volume fraction accounted for 17% of the total volume which also compares favorably with

published data (249). The fraction of the lung that is occupied by air and parenchyma can also

be deterrnined from the CT densrty by dividing the specific volume of tissue (assuming a density

of 1 .O65 g/ml (91)) by the calculated specific volume of the lung (tissue and air). This produces

a slightly higher parenchymal volume fraction (22% than the histology (1 7%) which suggests

that the CT estirnate may include larger blood vessels than were observed histologically.

However, these CT estirnates of the fraction of tissue and air can be used to correct the

histology to the appropriate air and tissue fraction in the intact thorax. Although this correction

is small in the present study because the specimen was inflated, it should be particularly useful

in quantitative studies of lung biopsies where Iung inflation is an uncontrolled variable.

The results of this study confi rm that the CT can be used to obtain accurate estimates of

the total lung volume and weight as well as regional lung expansion

(82,91,146,155,156,183,198,242,248,270). They also extend those observations by providing a

simple method of converting lung density to the volume of gas per gram of tissue. This

approach allows the CT data to be integrated with physiologic measurements of the pressure

volume characteristics of the sarne lung; to calculate the volume fractions, surface area and

tissue thickness appropriate for the entire lung; and to correct these vaiues to the appropriate

intra-thoracic lung volume.

CHAPTER 4: INTERSTiTIAL LUNG DISEASE

4.1 Introduction to Pulmonarv Fibrosis

ldiopathic pulmonary fibrosis (IPF) forms what is probably a heterogeneous group of

disorders within the broad category of interstitial lung disease. The original description of IPF

was based upon a pathologic description of the fibroproliferative expansion of the interstitium

seen on autopsy or at open lung biopsy (25,46,113,217). However, it has becorne evident that

the disease process involves an inflammatory infiltrate into the airspaces with subsequent

reorganization of both the infiltrate and the interçtitium and not the interstitiurn alone

(88.1 13,124,126). Therefore. the t e n infiltrative lung disease has been proposed for the group

rather than interstitial lung disease because it is more descriptive of the process (1 65). Hogg

suggested that the entire group of diseases should be reclassified in terms of their pathogenic

origin instead of

descriptive histology

(88) and his

classification divides

the diseases into

two main groups, or

pathways,

depending on

whether the process

is based on the

inflammatory or

neoplastic process

(figure 11). This is

Figure 1 1. Classification of interstitial lung diseases based on

pathogenesis (88)

an important distinction because it elevates the classification beyond qualitative descriptions

into a grouping of mechanisms that are responsible for the tissue changes observed. It is

important to note that the end result of al1 of these conditions is the end stage lung which, on

both CT and gross examination, has the appearance of dense fibrotic lung filled with

'honeycomb" cysts at which point the origin of the process is lost within the massive fibrotic

changes seen in the lung tissue.

The most common form of IPF is the result of the nongranulomatous inflarnmatory

processes referred to as the usual form of interstitial pneumonia (UIP) in North American

(88,113), or cryptogenic fibrosing alveolitis in Europe (46,88). These ternis are based on a

histologic description of the tissue changes within the lung, but when this description is

combined with the radiological and clinical features it is known as idiopathic pulrnonary fibrosis

(96,l 13). The etiology responsible for IPF is unknown but auto immune disorders

(46,99,ll3,li2,232), occupational exposure (34.1 02,113). and genetic abnormalities (34.1 13)

have al1 been suggested.

4.1.1 Clinical Oescri~tion of IPF

Most patients with IPF die within five years of diagnosis and less than 20% of patients

respond to any sort of treatment (52,l O4,l84,206,208,2O9). It usually begins with an insidious

onset of breathlessness with lung function tests showing a reduction in static lung volumes

(TLC, FRC, RV) with a normal. or even elevated, FEVl/FVC ratio due to a greater decrease in

the FVC than FEVi because the conducting ainnrays are relatively normal (184,261). The

diffusing capacity of the lung is also reduced and this was first thought to be due to the fibrosis

creating an anatornic barrier for the diffusion of oxygen. However, subsequent studies showed

that the major problem is ventilation-perfusion mismatching and that the barrier to diffusion only

becomes important during exercise (37,54,62,84,98,118,240). The pressure volume cuwe of

patients with IPF is shifted downward and to the right signifying that the lungs in l PF are less

cornpliant, or stiffer, than normal lungs (37,62,261,275).

4.1.2 RadioIoaical Oescri~tion of IPF

Radiologically IPF shows reticular, reticulonodular, and "ground glass" pattern on chest

X-ray and CT that usually involves both lungs, but is predominantly located in the lung bases

(1,165,231) The subpleural distribution of the disease is best appreciated with the use of

high-resolution computed tomography (HRCT) which has greatly increased visualization of the

lung. HRCT uses narrow beam coltirnation to rninimize volume averaging and a high spatial

frequency reconstruction algorithm to show the lung parenchyma cleariy. As a result, srnaIl

changes in lung density can be appreciated. As the disease advances, the cystic, 'h~neycomb'~

pattern becomes evident with most of the normal lung being replaced by dense connective

tissue (1 64,165).

4.1.3 Histoloaic Descriation of IPF

The hallmark of UIP is the variegated appearance of the lung on biopsy, with regions of

marked inflammatory and fibrotic changes right next to normal appearing regions

(34,37,46,113). The affected regions show several stages of the disease starting from signs of

lung injury and repair that include epithelial necrosis and alveolar collapse with exudation of

fluid and inflammatory cells (34,37,113). The subsequent interstitial and intra-alveolar fibrosis

is characterized by an increase in fibroblast numbers and a large increase in the ECM

(34,37,113). The interstitium is thickened in these regions and there is often hyperplasia of

alveolar type II cells consistent with the repair process (34,37,113). Associated with these

interstitial changes are proliferation and thickening of smooth muscle surrounding both the

arteries and the bronchi.

Histologie examination of the peripheral lung shows that the inflammatory exudate

contains predorninately lymphocytes, plasma cells and alveolar macrophages, with some PMNs

(37,46), while Bronchoalveolar lavage (BAL) shows a large proportion of ?MN, macrophages

and eosinophils (37,46,205), al1 of which are increased in patients who smoke (205).

4.2 Fibrotic Mechanisms

4.2.1 Cellular Mechanism of IPF

Lung tissue injury results in a proteinaceous exudate into the air space associated with a

recniitment of PMNs followed by monocytes from the microvasculature (126). The neutrophil

response is acute and their pnmary purpose is to destroy infectious agents. After migration,

the monocytes differentiate into alveolar macrophages that phagocytose and kill pathogenic

organisms. scavenge tissue debris and release cytokines. These cytokines include interleukin-

1 (IL-1). platelet derived growth factor (PDGF). epidermal growth factor, tumour necrosis factor

(TNF), and transforming growth factors -a and -p (TGF-a. TGF-p). The replacement of

surfactant by this exudate increases the surface tension causing alveoli to collapse ont0 the

alveolar ducts (34.88.126). Fibroblast proliferation, differentiation into myofibroblasts and

migration into the region organises the exudate into a fibrotic scar

(1 5,29,34,88,113,125,126,160,187.191). The fibroblasts initially synthesize type III collagen.

which in later stages becomes predominantly type 1, fibronectin, and proteoglycans

(9.15,29,125,149,191). The new ECM is then re-epithelialized by migrating type II epithelial

cells and the end result is a cystic, fibrotic lung, with a reduction in alveolar surface area and

total lung volume (34,37.88,126). The most obvious histological and biochemical change in

areas of fibrosis is the increased amount of collagen. There are several mechanisms for the

increased collagen aside from the obvious increase due to more fibroblasts

(9,42,126,149,190,199) and the increased synthetic ability of the fibroblasts (9,126,149).

Fibroblasts within the gingiva have been shown to have a reduced ability to phagocytose

collagen (1 47) although analogous mechanisms have not been shown in the lung. Additionally

there is evidence for decreases in the synthesis of the collagen degradation enzymes ( i.e.

collagenase and stromelysin) which are known as metalloproteinases, in conjunction with an

increase in the production of molecules that function as tissue inhibitors of metalloproteinases

(TIMP) (20.34.190). Therefore. the increase in ce11 nurnber. synthetic activity, and decreased

degradation leads to an increase in the ECM components.

4.2.2 Molecular Mechanisrns of IPF

TGF-$ is the most well studied of the cytokines released by the process, and appears to

play the central role in controlling the response of the mesenchymal cells in terms of ce11

differentiation, proliferation, synthesis of matrix components, and secretion of the other

cytokines responsible for the propagation of the response (1 8.2O,ll9.l47,l9O, 1 96). TGF-p is

released initially by macrophages and platelets and is a very strong chemoattractant for

phagocytes and fibroblasts (1 19.1 47.1 90,196.268). Once stimulated, fibroblasts synthesize

TGF-b which has an autocnne effect on itself so that the process can continue without the aid

of inflammatory cells (20). TGF-p has been shown to increase the production of collagen.

proteoglycanç. fibronectin, TI MPs, and decrease the synthesis of collagenase (20.1 90). It also

inhibits endothelial cell division. and the proliferation of IL-1 dependant inflammatory cells (1 90).

The proteoglycan, decorin. has been shown to bind to TGF-B and inactivate it (20) and the

increase in decorin in the eady stages of fibrosis led to the postulate that decorin binds TGF-B

and directs it to the mesenchymal cells to potentiate the migratory and synthetic effects (1 96).

Another growth factor, PDGF, is released from damaged endothelium. platelets and

fibroblasts and has also been shown to cause proliferation of fibroblast cell lines (20,122,190).

IL-1 is a cytokine that stimulates the immune system as well as acting as a mitogen for

endothelial cells causing them to proliferate and migrate over the new connective tissue (1 90).

Studies of the kidneys have shown that collagen breakdown products stimulate fibroblasts to

proliferate and produce collagen in the absence of any inflarnrnatory response (212) suggesting

that a prolonged inflarnrnatory response is not necessary for fibrosis (20,122).

In summary, the remodelling of the lung matnx to produce the changes in IPF requires

input from many sources, ail of which stimulate fibroblasts through the production of TGF-P.

This response is an expression of the normal response to injury that is necessary for the

elimination of harmful agents and the repair of the damaged tissue (1 5,34,258). However, the

fibrotic response in IPF is maladaptive in that it does not abate and the tissue is never returned

to a normal state (1 5,34).

4.3 Quantitative Studies of IPF

There are several problems with quantitative studies of fibrosis in human lungs. The

first problem is obtaining suitable tissue for analysis because autopsy specimens have often

reached the end-stage with only large masses of connective tissue remaining. Open lung

biopsy specimens are better than autopsy specimens, but because of the variable distribution of

the disease it may be diffÏcult to obtain enough specimens to get a tmly representative sample

of the lung. Also, the biopsy collapses during the surgical removal and must be reinflated to an

appropriate level to perform a quantitative analysis. However, since the biopsy is usually from

penpheral lung the airway structure is no longer available for instillation of fixative and the

pleural surface is disrupted so that it does not contain the fixative at an appropriate pressure

within the lung parenchyma to inflate the airspaces.

BAL is a very popular rnethod for assaying the extent and the progression of IPF. Many

attempts have been made to correlate the cells and molecules retrieved with BAL to pulmonary

function (37,39,62,78,l76,i 86,207) or disease activity

(6OI68,86,l 41,145,170,l 82,186,194,203,205,221,223,243,245,278). However, BAL and other

biochemical studies have the sarne intrinsic problem in that local acting cytokines and

inflammatory cells can al1 get mixed up in the same test tube with no regard to their location

within the tissue so that functional or pathogenic conclusions are difficult at best. Also, BAL

washes cells predominately from the larger airways and may not be truly representative of the

alveolar situation.

To date there are very few quantitative studies on IPF lungs. The histologic studies are

either qualitative descriptions of the disease process (1 13,124) or the extracellular matrix

changes

(9,21,23,30,55,61,86,95,100,111,125,137,143,161,169,171,177,185,191,201,203,221,269) or

semi-quantitative anatysis involving the use of complicated scoring systems. Several scoring

systems have been derived but these are very complex and for best results involve the use of a

panel of trained observers (25-27,101,244). Hyde has reported a cornparison between a

stereologic quantification of biopsy specimens and a panel grading system and reports that

there is good correlation between the two arguing that the time consuming quantitative

approach is not justified for an analysis of the tissue changes in IPF (26,101 ).

CT analysis has fallen into this same arena as histology. Semiquantitative scoring

systems are reported to be faster and less expensive than the computer and time intensive

quantitative studies

(5,6,8,10,51,63,67,79,80,85,133,140,167,175,180,193,204,225,238,259,260,267). Also,

quantitative studies of the CT scans in IPF are relatively new and the results are still open to

interpretation as to what aspect of the CT scan should be quantified, or what aspect gives a

reliable estimate of the disease process. lnvestigators have reported an increase in overall

lung density and changes in the frequency distribution curves of the voxel attenuation values of

the CT scan (31,78,155,156). Hartley studied the changes in the frequency distribution curve in

interstitial lung disease and showed a correlation between the moments (mean, median, mode,

kurtosis. skewness) of the frequency curves and other markers of disease activity such as lung

function and BAL results (78).

4.4 Experirnent # 2

In this study, we report results obtained using a new method for quantifying structural

changes in the lungs of living patients with IPF using data obtained from CT scans and

quantitative stereology (32). Total and regional lung volumes are estimated from X-ray

attenuation values on CT. The surface area of the lung parenchyma and the volume fraction of

each of the components of the lung is quantified by line intercept and point counting techniques

using both the light and electron microscope. This study shows that a combination of the CT

and histologie data provides accurate information about the changes present in the lung and

could provide a basis for measuring the progression of disease in subsequent CT studies.

4.5 Material and Methods

The procedures used in this study were approved by the ethical review boards of the

University of lowa Hospitals. St. Paul's Hospital, and the Universities of lowa and British

Columbia. All the patients in this study signed infonned consent foms that allowed the use of

physiologie data. CT scans, and the surgical tissue. The patients with IPF were enrolled by the

National Institutes of Heakh Specialized Center of Research in Interstitial Lung Disease at the

University of lowa. To be included in tbis group patients had to have a clinical history, chest X-

ray, and pulmonary function data suggestive of IPF, and a pathological diagnosis of usual

interstiti al pneumonia on open lung biopsy. The control subjects were enrolled in the lung

registry of the University of British Columbia Pulmonary Research Laboratory and were part of

an ongoing study of lung structure and function. These patients required either a lobectomy or

pneumonectorny for a small. non-obstructing, peripheral bronchogenic carcinoma whose lung

function was measured a few days prior to surgery. The patients selected had normal lung

function and were matched for age, sex and smoking history with the IPF patients.

4.5.1 Pulmonarv Function Studies

Control Patients:

Spirometry, lung volumes and lung compliance were measured with subjects seated in a

volume displacement body plethysmograph at the Pulrnonary Research Laboratory. in

Vancouver. Functional residual capacity (FRC) was measured using the Boyle's Law technique

(441 46.1 8 1 ). Total lung capacity (TLC) was calculated by adding inspiratory capacity (IC) to

FRC. Residual volume (RV) was calculated by subtracting vital capacity (VC) from TLC.

Measurernents of lung volume and its subdivisions were obtained on a P.K. Morgan

computerized pulmonary function testing systern (P.K. Morgan. Boston, Mass.) using a dry

rolling seal spirometer and multiple breath heliurn dilution techniques on subjects who were

unwilling to enter the plethysmograph. Diffusing Capacity (DLcO) by the single breath method

as described by Miller and associates (157) on the P.K. Morgan automated diffusing capacity

analyzer. The results are conected for alveolar volume (VA) and reported as both Dtco and

DL&*. The predicted nonal values for FEV,, FVC, and DLCo were those of Crapo et al (33)

and TLC was predicted using Goldman's values (70).

1 PF Patients:

The lung function for this patient group was studied at the University of Iowa.

Spirometry was obtained using a Medical Graphics 1070 systern (St. Paul. MN) and the lung

volumes were obtained with the patients seated in a body plethysmograph (Medical Graphics

1085 systern). The Dm was measured using the single breath technique on the Medical

Graphics 1070 system. The predicted normal values were those of Morris et al (162) for FM,,

FVC, Goldman et a/ (70) for TLC , and Van Ganse et al (237) for DLm (78).

4.5.2 CT Studies

Control Patients:

The control subjects in the study received a conventional CT scan (1 0 mm thick

contiguous slices) on a GE 9800 Highlight Advantage CT scanner (General Electric Medical

Systems, Milwaukee, WI) approximately one week prior to resection. All scans were performed

with the subject supine during breath holding at full inspiration. Conventional clinical scanning

parameters in use at St. Paul's Hospital were used (120 kVe, 100 mA, 2 sec scan time). The

images were reconstructed using a standard algorithm. and printed ont0 radiologic viewing film

at a window of 1200 and a level of -700 HU. The image data was transferred to a Silicon

Graphics lndy Workstation (Mountainview, CA) for analysis of the X-ray attenuation values.

I PF Patients:

The CT scans of the IPF patients were obtained at the University of Iowa on an lmatron

C-100 ultrafast scanner during inspiration with the patient prone. These high resolution CT

(HRCT) scans were obtained using 3 mm sections spaced by 20 mm gaps from the lung apex

to the diaphragm (130 kVe, 640 mA, 0.6 s scan time) and were reconstructed using a high

spatial frequency algorithm. The images were printed ont0 standard radiologic film at a window

of 2000 and a level of -500 HU, and sent, dong with the image data, to the Pulmonary

Research Labofatory in Vancouver. The image data was transferred to a Silicon Graphics lndy

Workstation (Mountainview. CA) for attenuation analysis.

A program to segment the lungs from the chest and the large central blood vessels, and

analyze the X-ray attenuation was written for the numerical analysis package PV- Wave (Visual

Numerics, Boulder CO). The lungs were segmented using threshold settings of -1000 to -500

HU. The volume of the whole lung (tissue and airspace) was calculated by summing the pixel

dimensions in each slice and multiplying by the slice thickness. The mean CT attenuations in

HU of each pixel was calculated and converted to gravimetric density @/ml) by adding 1000 to

the HU value and dividing by 1000 (equation 7) (82). The weight of the lung was calculated by

multiplying the mean density of the lung by its volume. Regional volumes of gas per gram of

tissue were calculated according to equation 8. To eliminate the affect of body position during

the scan (supine versus prone) on the distribution curves of lung inflation, the volume of gas per

gram of tissue was expressed as percent of the patient's TLC which was obtained by dividing

the TLC measured in the upnght position (were ail the alveoli are evenly inflated) by the

measured lung weight.

Because of the inflation artifact associated with the biopsy specimens in the IPF

patients, the VV(C~) of the biopsy was estimated by identifying the segment on CT that was

biopsied using the surgical report, and the position of the surgical clips on the post-operative

chest X-rays (PA and lateral) and calculating the Vv(rn from equation 9 (32). This value was

then used as the level 2 Vv in the stereological analysis below.

4.5.3 Quantitative Histoloav

Control Patients:

The resected lung specimens were obtained directly from the operating room and taken

to the laboratory where they were weighed, infiated with Optimal Cutüng Temperature (OCT)

compound (Miles Laboratones, Elkhart, IN) diluted 1 :1 with n o n a l saline, re-weighed, and

frozen in liquid nitrogen without clamping the airways (32). Once frozen, the specimen was cut

into 2 cm thick slices on a band-saw and aie tissue sampled with a power driven hole-saw.

These sarnples were stored at -70 O C for other purposes. The remainder of the specimen was

transferred to 10% buffered formalin and fixed at rom temperature for at least 24 hours.

Random samples were taken for light microscopy from the fixed slices, embedded in paraffin

and stained with hematoxylin and eosin. Small samples were taken for electron microscopy .

(EM) from the peripheral lung of five of the patients prior to OCT inflation and freezing and fixed

by inflating with 2.5% glutaraldeyhyde in 0.1 M sodium cacodylate buffer using an 18 gauge

needle. These specimens were post fixed in 1 % 0sO4 in 0.1 M sodium cacodylate, dehydrated

through graded ethanol and infiltrated with LR-White. The tissue blocks were sectioned with a

diamond knife on an ultra-microtome ( Reichert Ultra-Cut or RMC MT-6000 XL) at 60-90 nm,

picked up on formvar coated 200 mesh copper gnds and stained with Uranyl Acetate, and

Sato's Lead solutions.

IPF Patients:

At thoracotomy the lung was biopsied from at least 2 regions, one that appeared normal,

and one that appeared diseased on HRCT. The site of the biopsy was confirmed by post-

operative chest X-rays in the PA and the lateral position. The biopsy specimens were split and

a small portion was prepared for EM as above, and the remainder was processed for light

microscopy.

To optimize the sampling for the stereologic analysis, a cascade design technique was

used as descnbed in Chapter 1 and represented in figure 5. In this design the lung is quantified

in a series of steps that increase in rnagnification so that what is quantified at one level can be

further sub-divided into it's components at the successive level (35).

Level 1 was performed on the C f scans of both groups, using al1 of the available image

slices by counting the number of points falling on normal lung, densely fibrotic lung, 'ground-

glass opacification', bronchovascular bundle, and turnor (figure 5A). Levels 2 and 3 were

performed at the light microscopy level using the point counting program Gndder (Wilrich Tech,

Vancouver, B.C.) which generated random fields of view, projected a grid on to the field of view

via a camera-lucida attachment on a Nikon Labophot light microscope and tabulated the -

counts- Level 2 used 100x magnification with a grid of 80 points (d = 0.1 1 mm ) and 40 lines (1 =

0.1 1 mm). The number of points falling on airspace, tissue (lung parenchyma), and medium

sized blood vessels (50-1000 vm) as well as the number of intersects between the grid lines

and the parenchymal-airspace interface were tabulated (figure 5B-C). Level3 was performed

on 10 random fields of view per slide at 400x rnagnification and the number of points falling on

airspace components (Alveolar macrophages, alveolar PMN, other objects in the airspace, and

ernpty space) as well as the tissue cornponents (alveolar wall, capillary lumen. and small blood

vessels (20-50 pm)) were counted using a 100 point grid.

Level 4 was performed using TEM images. 10 systematic area weighted fields per grid

(35) were photographed ont0 35 mm slide film at a rnagnification of 1080x using a Phillips 300

transmission electron microscope. These slides were projected ont0 a grid of 120 points with a

magnification of 10x to give a final magnification 10800x and the number of points falling on

electron-lucent space, collagen fibers, elastin fibers, interstitial cells, inflamrnatory cells and

unidentifiable substances were tabulated using Gridder (figure 5D and figure 12).

The volume fraction (Vv) of each of the lung components, (VI,(~,), were estimated

at each level according to equation 3, and the surface density (Sv(wd) was estimated using

equation 4. Since surface density is the surface area in a given volume, the surface area of the

parenchyma is calculated by multiplying the surface density by the volume of the lung. The

inverse of Sv is an estimate of parenchymal thickness. The overall Vv is calculated by

multiplying the Vv of the lung component at the highest level by the VV that contained it in the

previous levels according to equation 5.

Figure 1 2. Representative electron micrograph from 1 PF patient biopsy. IC: interstitial cell, Col: collagen, El: elastin, ES: electron-lucent space.

4.5.4 Statistical Analvsis

All data were analyzed using independent t-tests or the one way analysis of variance.

Transformations were made on certain variables to normalize distributions and to make

variances homogeneous. A Bonferroni sequential rejective procedure was used to correct for

multiple comparisons (94). A corrected p-value of less than 0.05 was considered significant.

4.6 Results

Table 8 shows anthropometric and lung function data for the patients studied. The

control patients have normal lung function are of similar sex distribution but are slightly younger

and Iighter than the IPF group. The IPF group show the pattern characteristic of restrictive lung

disease with a reduction in FR/,, W C , DLCo and in the subdivisions of lung volume (RV, FRC,

and TLC), and an increased FEVJFVC ratio.

The CT estimates of lung volume and weight (table 9) show a reduction in total lung

volume in IPF compared to controls due to a reduction in the volume of the airspace. The

tissue volume and therefore the lung weight was not different between the two groups. The

frequency distribution in ml of gas per gram of tissue present in each voxel (figure 13) was

different between the two groups (p < 0.001). The control lungs showed a normal distribution

(table 10) with mean, median and mode values that are closely similar (4.6 î 0.3, 4.5 I 0.2. and

4.3 i 0.2 mVg) and a relatively small variance (1 8.0 8.2 mllg) and skew (1.6 0.6 mVg), while

the IPF lungs show a left shifted distribution with a positive skew (6.9 I 3.4 mUg) and a very

large variance (1 16.8 30.8 mVg). The mode (1.4 * 0.3 mllg) for the IPF lungs was reduced

from control levels. whereas the mean (5.7 I 0.7 mVg) and the median (4.1 I 0.4 ml&) were not

different frorn the controls due to the large variance. When the median values were expressed

as percentage TLC, the values were 72.1 * 2.5 in control patients. and 80.6 i 5.2 in IPF

patients (p > 0.05) indicating that there was little effect of body position on the degree of lung

inflation.

Figure 14 compares the volume fraction of tissue estimated from CT to that obtained

with histology. This shows that in the control cases the resected lung lobes were inflated to

approximately the same level during both CT and histologic analysis (figure 14a). However, the

biopsies obtained from the IPF lungs were under-inflated during histologic preparation

compared to the CT studies (figure 14b). The ability to measure this difference and use the CT

data to correct the biopsies to the correct levei of inflation on an individual basis is critical to the

subsequent analysis of these biopsy specimens.

The light microscopic data (table 11) show that after the correction to the level of lung

inflation obtained during the CT scan, the volume fraction of tissue is increased and the

airspace is decreased in the regions of the IPF lungs where there was radiologie evidence of

disease, compared to the regions of the IPF lungs that appeared normal and the control lungs.

The regions of the IPF lungs thought to be normal on CT showed an increase in the volume

fraction of tissue compared to controls and a decrease in the capillary volume fraction from

controis, while the medium sized blood vessels were decreased in both the normal and

diseased regions of the IPF lungs. A most striking finding was the difference in surface area

which was 113 I 12 m2 in the control lungs and 30 I 7 m2 in the diseased lungs. This reduction

in surface area was associated with an increase in the mean parenchymal thickness from the

control level of 8 * 1 pm to 40 * 11 prn in the regions of the IPF lungs judged to be normal on

CT, and 97 I 22 prn in the diseased regions.

The airspace of both regions of the IPF group contained a signifiant amount of

inflammatory exudate consisting of both proteinaceous fluid and cells. The percentage of the

alveolar airspace occupied by fluid was significantly increased in the diseased regions of the

IPF lungs when compared with the control lungs. The fraction of the airspace occupied by

macrophages was also increased from control levels in both regions of the IPF lungs and when

these are expressed as number of cells per square rneter of surface area, the macrophage

number increased from 12 r 3 X 10' cells/rn2 to 30 r 22 X 10' ce1ls/m2 in 'normal" regions of

the IPF patients and was significantly higher. 150 I 100 X 10' ce11s/m2 (p < 0.001). in the lung

regions judged to be diseased on CT.

Figure 15 summarizes the results of the stereological analysis of the tissue

cornpartment of the lung based on electron microscopy and are normalized to the lung weight

obtained by CT and expressed per 100 grams of lung. The data show that the electron-lucent

space tended to be lower in the normal regions and higher in the diseased regions of the IPF

lungs compared to control lungs. The collagen content showed a progressive increase from

control to diseased IPF regions, whereas the elastin content of the tissue was decreased in the

"normaln regions, and there was a trend for an increase in the inflammatory cells which did not

reach statistical significance owing to the large variance in number in the different biopsies

(p=0.08).

5 I O 15

ml gasl g tissue

Figure 13. Shows the CT density of the lung expressed as a percent of the voxels on the CT scan aaainst oas volume per gram of tissue. The red line represenk the control group, and the blue line the IPF group. p < 0.001

Volume Fraction Difference (CT-Histology) Volume Fnctlon Difference (CT-Histology)

V, = a a a p a + a a o o o g g g

O O O O O o O O O O O O ~ ~ s s a , ~ ~ s s o o o s , , , a . . . . . ,

( D i I

Figure 15.

Collagen Elastin Electron- Lucent Space

Interstitial Cells

lnflamm atory Cells

Tissue Component

ows the weight of the interstitial components estimated from histologic analysis at the EM level and ~ressed per 100 g of lung tissue. Red bars are the control group, blue bars are IPF normal group, and Iow bars are IPF diseased group. Values are rneans I S.E.M. a < 0.05.

4.7 Discussion

This study provides new quantitative information on the composition of human lungs

with IPF obtained using a combination of CT and histologie analysis. The CT scans show a

reduction in total Iung and airspace volume in IPF (table 9), which correlates with the reduction

of static and dynamic lung volumes obtained by pulmonary function tests (table 8). The

distribution of volume of gas per gram of lung in IPF lungs, (figure 13) shows that the larger

proportion of the lung has lower volume of gas per gram of tissue than controls, and that there

are also over-inflated regions indicating larger airspaces. The lung regions with low gas

volumes per weight are diseased with reduced airspace, surface area and a reorganization of

the tissue components. The regions with a high gas volume per gram are probably the result of

the formation of cystic spaces in the diseased areas that are thought to result from a collapse of

alveoli ont0 alveolar ducts with distortion of the ducts by the repair process (88J 26). However,

it is also possible that some of the overinflation is due to a concomitant emphysematous

process (207).

Differences in total and regional Iung volume will be influenced by the degree of lung

inflation during the CT scan, however, the estimates of lung weight should be reliable because

it is based on density of the lung which reflects the degree of lung expansion (volume). The

effect of body position on lung inflation was accounted for by expressing the volume of gas per

gram of tissue of each of the voxels as a percentage of the total lung weight divided into the

TLC measured in the upnght position where the alveoli are know to be evenly inflated

throughout the lung (1 54). As there was no statistically significant difference in the median

values of percentage TLC between the two groups, we conclude that the lung parameters

estimated from CT can be compared between the two groups even though some CT scans

were conducted prone and others in the supine position. There may have also been

differences introduced due to the different scanners that were used for the study, and the slice

thickness and reconstruction algorithms used to acquire the images. However. Kemerink and

colleagues have shown that the reconstruction algorithms and slice thickness have a negligible

effect on densitometry measurements (1 16). Further, the same group tested numerous

scanners for densitometry measurements and concluded that cornparisons between properly

calibrated scanners is possible (1 17). The largest difference in densitornetry measurements will

corne from inhomogeneous lungs and since normal human lungs contain a more homogeneous

parenchyma than the diseased condition, thicker slices will be appropriate but thinner slices

should be used for the changes associated with fibrotic lung disease. Therefore, considering al1

of the variances associated with different human lungs, which includes the varying degrees of

disease, we believe that the differences in scanners and scan acquisition are of minimal

consequence.

As we have previously shown. there was a good correlation between the air to tissue

fraction calculated by both CT and histology in the control cases where the resected specimens

were fixed in inflation. However, in the IPF lungs the histologic fraction of air to tissue was

markedly reduced because the lung biopsies collapse (figure 14). By locating the segments of

the lung that were biopsied on the pre-operative CT scan, the volume fraction of tissue and

airspace can be estimated in the intact chest using equation 9. These values can then be used

to correct the histologic estimates to the appropriate inflation at the time of the CT scan and is a

very important advance in quantitative histology. Surprisingly the data show that lung weight

and tissue volume was the same in IPF and control lungs which establishes that the CT

appearance refiects a reduction in airspace more than an increase in tissue. The histology

suggests that the reduction in airspace is due to a collapse of alveoli on ducts and that this

accounts for the marked reduction in surface area. Hogg (88) has previously postulated that

the exudate of fluid through the tissue and into the airspaces increases surface tension and

causes the alveoli to collapse onto alveolar ducts. Kuhn et al (1 26) showed that the

subsequent organization of this process with epithelial growth over the exudate incorporates the

material present in the airspace into tissue. These events change the relative fraction of tissue

to air (table 11) but does not significantly increase in the volume of tissue or lung weight. The

major consequence is a decrease in the alveolar surface area frorn 1 13 I 12 m2 in control

patients to 30 I 7 m2 in IPF and a ten fold increase in mean parenchymal thickness frorn 8 I 1

to 97 I 22 Pm because the alveoli account for the majority of the lung surface area (249). The

large variance in these values is the result of regions of dense fibrosis being Iocated

geographically close to normal regions (1 13). This reduction in surface area contributes to the

reduction of the diffusing capacity of the lung (table 8). but because the alveolar volume is also

reduced, the DL&* remained within noma1 limits in these cases.

There was an increase in the volume of alveolar airspace occupied by proteinaceous

and cellular components in the IPF biopsy specimens (table 11). Although we cannot discount

the effect of surgical trauma on the efflux of edema into the airspace, it is unlikely that this is the

sole cause of the cellular infiltrate as the tirne period between surgical removal of the tissue and

fixation is relatively short. Therefore, the increase in volume of alveolar exudate with increased

numbers of PMN in the fibrotic regions of the lung is consistent with the hypothesis described

previously, and the literature conceming the histologic appearance of IPF (88). The increase in

macrophage number in both regions of the IPF lungs is also of interest in that these cells are

considered to be key players in directing the fibrotic response (1 91,213). They are capable of

synthesizing large amountç of growth factors including transforming growth factor+ (1 91), as

well as platelet derived growth factor (21 3) and intedeukin-1 (121). Our data demonstrate the

variability of the disease process where some regions contain numerous cells. while others are

similar to controls. This provides a more accurate picture of the peripheral lung condition than

that represented in bronchoalveolar lavage.

The degree of tissue reorganization is best appreciated on EM examination (figure15)

where the diseased regions of the lung showed an increase in electron-lucent space, collagen

and interstitial cells per 100 grams of tissue, compared to the control lungs and regions of the

IPF lungs not grossly involved by disease. The normal appearing lung regions in the patients

with IPF were not significantly different from the controls for any of the variables but did have

less infiammatory cells than the diseased regions. lnterestingly the amount of elastin present

was decreased in the normal regions of the IPF lungs and tended to be intermediate in the

diseased regions. The obsewed reduction in elastin content in the IPF lungs is consistent with

reports of minimal synthesis of elastin in mature human lungs (21 l), and we postulate that

proteolytic mechanisms associated with migration inflamrnatory cells may aIso result in the

reduction in elastin content. The increase in the estimated collagen content within the

interstitium has been welt documented in IPF (46,126,191), and Our obsewations suggest that

the fibrotic process is well advanced in these areas.

Bensadoun and CO-workers have recentiy shown that in there is an increase in the

proteoglycan content of the interstitiurn in regions of IPF lungs where there is active disease

(13). Proteoglycans play an important rote in the fibrotic response by acting as receptors for

growth factors (18), directing effects on the synthesis and degradation of elastin (1 50) and

collagen (1 3). Proteoglycans are highly negatively charged molecules that require hydration to

maintain their shape and hence are important for the modulation of tissue hydration (1 3). As

the EM fixation used in Our study does not preserve these molecules (234), we speculate that

they account for the increase in electron-lucent space present in the diseased areas of the lung.

We conclude that the CT scan can be combined with histology to provide a quantitative

estimate of the tissue changes occuning within the lungs of patients with IPF. The CT densrty

provides the ratio of tissue and air which can be used to correct open lung biopsies to the level

of inflation at which the CT was perforrned. This approach could provide a method of following

the progress of disease using the CT analysis to esti-mate the histologie changes.

CHAPTER 5: PULMONARY EMPHYSEMA

5.1 Introduction to Pulmonarv Emphvsema

Chronic obstructive pulmonary disease refers a group of disorders presenting with

chronic airflow limitation and are usually placed in one of Wo basic classes. Type A patients

have hyperinflation of their lungs without cough and wheezing and are considered to have

predominantly emphysema (14,197,261). The type B patients are considered to have chronic

bronchitis and they present with cough and sputum and are often more hypoxic and

hypercapnic than the type A patients (1 4,197,261)- However, this classification actually

represents the extremes of these groups and there is a great degree of overlap between these

extreames. The pathologic studies of COPD lungs show even more overiap between the

groups with al1 patients having a degree of emphysema and some airway disease (1 4,197,229).

Emphysema is a pathologic diagnosis, and while there are several established criteria

that must be met, it is still a subjective diagnosis. An eariy definition of emphysema published

by the Ciba Guest symposium in 1959 defined emphysema as 'enlargement of the acinus that

might or might not be accompanied by destruction of the respiratory tissue' (28). In the eady

1960's, the Wodd Health Organization (271) and the American Thoracic Society (7) adopted a

similar but different definition that limited ernphysema to 'enlargement of any part of the acinus

with destruction of the respiratory tissue'. The current definition: "the abnomiai permanent

enlargement of airspaces distal to the terminal bronchioles. accompanied by destruction of their

walls, without obvious fibrosis,' was published in 1985 in an attempt to define the destruction

term of the ATS definition and to exclude diseases where there is airspace enlargement but the

predominant feature is fibrosis (21 9). Each of these terms was carefully chosen to differentiate

between the enlargement of airspaces which is a natural process in the aging (senile) lung and

the cornpensatory overinflation of the remaining lung following a lobectomy or pneumonectomy,

or the generalized airspace enlargement tbat occurs in Down's syndrome. Therefore, in order

to be defined as emphysema there must be destruction of the normal tissues, defined as "the

reduction of that tissue to a useless form or nothingnessn (219). The qualification "without

obvious fibrosis" was added to the definition to exclude the airspace enlargement seen in

interstitial lung diseases such as sarcoidosis, eosinophilic granuloma and in the end stage

fibrotic lung where the predominant feature is the fibroproliferative response which is due to a

completely different pathogenic process than the emphysematous destruction. It should be

noted here that for a diagnosis by this definition the lung tissue must be directly observed and

any other method of obtaining this diagnosis is presumptive. However, there are many features

of the disease that when combined enable the physician to make a diagnosis of the disease

without having to obtain lung tissue.

5.1.1 Furictional Description of Emphvsema

The most obvious functional characteristic of COPD is the classic airflow limitation

shown by a reduced FEV1, W C and FEV,/FVC ratio as well as increases in al1 of the

subdivisions of lung volume (197,261). The pressure volume cuive of these patients is shifted

up and to the left and there is a reduction in the elastic recoil of the lung (93,181,197,261)

signifying a more cornpliant lung than the normal condition so that at any trans-pulmonary

pressure the lung will have a higher volume. The airfiow limitation in these patients. shown by a

reduction in the FEV,, can be due to obstructions within the airways, or to dynamic limitations of

the airways as the equal pressure point rnigrates towards the periphery of the lung and

becomes located within the small airways (1 97,261). The FEV,/FVC ratio is decreased due to

the reduction in the F EV,. The subdivisions of lung volume are elevated in this condition

partially due to the airflow limitation and partially due to the increased cornpliance of the lung.

The greatest change is seen in the RV resulting in a greatly increased RVKLC ratio and an RV

that approaches the VC volume (1 97). There is often a decrease in Da which is attributed to

the loss of alveolar surface area for gas exchange and the reduction in the blood volume of the

lung (261).

5.1.2 Radiolociical Description of Emphvsema

The chest X-ray is an imperfect tool for visualizing emphysema because the overlapping

structures hide the emphysematous changes in the lung parenchyma (56,202). Hyperinflation

of the chest is shown as the flattening of the diaphragm, best seen on the lateral view, and an

increased retrosternal airspace (192,202,230). There is also a decrease in the caliber or

presence of vessels in the outside third of the lung (1 92,202,230). The detection of

emphysema on chest X-rays does not correlate very well with pathologie scores and only has a

60-80% diagnostic accuracy with significant false positive rates (202).

The introduction of CT scans advanced in the visualization of ernphysema because it

removes the overlapping structures allowing the lung parenchyma to be visualized in cross

section. On CT emphysema shows hypodense regions and is usually associated with pruning

or obliteration of the pulmonary vessels (202). CT has shown high sensitivity and accuracy in

the diagnosis of emphysema with better than 80% sensitivity and only 2-30h false positives

(202,229). A further advantage of CT over chest X-rays is that CT scans contain the

information on the X-ray attenuation values of the lung. These values provide the means for

quantification of lung disease in a sensitive and reproducible fashion that visual grading

systems can never achieve.

5.1.3 Histoloaic Description of Emahvsema

There are three main forms of emphysema that can be identified on the gross

specimen: centrilobular, panlobular and paraseptaf. The most common fom is centrilobular

(centriacinar) emphysema which is predominately located in the upper lobes but becomes more

diffuse with increasing severity (226,272). Centrilobular emphysema affects the center of the

pulmonary lobule and is surrounded by normal lung parenchyma so that on gross examination

the affected lobule collapses in on itself below the level of the bronchi and vessels in the center

of the lobule. Histologically, airspace enlargement is observed in the center of the lobule and is

often associated with a distorted respiratory bronchiole (272).

Panlobular (panacinar) emphysema is primarily located in the lower lobes and grossly

affects the whole lobule so that the lobular septae, the airways and the vessels are elevated

above the parenchymal surface (226,272). Microscopically, there is enlargement of airspaces

throughout the entire lobule. As the severity of the emphysema increases. it becomes more

difficult to differentiate between panlobular and centrilobular emphysema, especially

histologically, and the obsewer must try and find a relatively normal region on the lung to

determine the type of emphysema.

The third type of emphysema is paraseptal (distal acinar) which is located subpleurally

and is characterized by bullae in the upper lobes (272). the walls of which may be fibrotic, while

the surrounding airspace is appears normal (272).

Emphysema is differentiated from simple airspace enlargement by the anatomical

location of the tissue changes, which are focally located in emphysema, and more uniformly

distributed in the senile lung. Histologically, the senile lung is usually associated with

enlargement of the alveolar ducts and saccules rather than any changes in aie alveoli while

Down's syndrome shows widened alveolar ducts and enlarged alveoli in suggesting an

incomplete alveolar development rather than a destructive process (272).

5.2 Pathoaenesis of Emphvsema

5.2.1 ProteaseJAntiwotease Theorv

Laurell and Ericksson were the first to show that widespread emphysema was present in

younger patients with a,-antitrypsin (al-PI) deficiency (1 31). At about the same time Gross et

al (73) showed that emphysema could be produced by instilling the enzyme papain into the

lung. These observations led to the theory that an imbalance between the release of proteases

from stimulated PMN and macrophages and the inactivation by serum components could result

in emphysema (1 7,87,103,272). It has been hypothesized that this balance can be upset

through a global inactivation of inhibitors, an overwhelming influx of proteases, or through the

creation of a local environment where the proteases can be released and the antiproteases

would be excluded, such as - a "pocket" formed between the PMN and the endotheliai cell

(57,87.131,272). Originally this theory was developed for the human neutrophil elastase which

degrades collagen, fibronectin and proteoglycans as well as elastin but has since been

extended to take into account other proteases such as the metalloproteases of the alveolar

macrophages and cathepsin G from the PMN (1 ï157,87,lO3,2l8,2i2). As well as proteases,

activated PMN and macrophages release reactive oxygen products through a transfer of

electrons from NADPH to oxygen to form the superoxide ion Of (57.87)- Superoxide reacts

with hydrogen ions to form hydrogen peroxide (H202) that in presence of transition metal ions

can generate the highly reactive hydroxyl radical OH- (57,87). PMN granules also contain

myloperoxidase which combines with H202 in the presence of chloride ions to form the

hypochlorous acid OHCI- (57,87). These free radicals have very darnaging consequences for

living tissue by oxidizing membrane lipids, oxidizing, cleaving and cross-linking proteins,

cleaving DNA, and changing cell pemeability. While they are usually targeted to foreign

substances (bacteria), they can also have their effects on the host tissue (57).

Cigarette smoke is considered the major cause of centrilobular emphysema because it

is known to contain a multitude of chemicals, some of which are powerful oxidants with long half

lives. and others inactive al-PI by oxidizing the active site (57.272). Cigarette smoke has also

been shown to cause the sequestration of PMN and monocytes within the lung (142). and

promote the release of their granules which contain the proteases and oxidants (87).

5.2.2 Inflammatory-Re~air Mechanism

The second theory for the degradation of tissue in emphysema is the inflammatory

repair mechanisrn (57,87,l74,272) which operates simiiarly to the fibrotic response described in

the previous chapter. In this mechanism, the inflamrnatory cells are activated and recruited to

the site of an insult where they release their proteases and oxidants which have an initial local

degradative effect on the tissue (272). This step is followed by activation of the repair process

which increases the extracellular matrix components in the damaged area (58,59). There are

numerous reports on the ultra-structural changes in emphyserna which show thickened alveolar

walls (58,59,168), rearrangement of collagen and elastin fibers (1 2). initial collagen destruction

followed by collagen synthesis (273) and increases in the amount of collagen per unit area of

alveolar wall (24, 128,129). These data al1 support the theory that emphysema is not a simple

destructive mechanism, but has an initial degradation response that is followed by a

fibroproliferative repair phase. It is important to note that these fibrotic changes are al1 at the

rnicroscopic level and the predominant feature of emphysema is destruction and not fibrosis as

seen in interstitial lung disease. Therefore, while the distinction of "without obvious fibrosis" is

debated by some investiqators (272), because this fibrosis is mild it is argued that the

qualification should rernain as part of the definition to exclude the end stage Jung of interstitial

lung disease (229).

Clearly there is a potential for overlap between the two aieofles as both mechanisms

involve elernents of the inflammatory process which has an exudative and proliferative (i.e.

repair) phase. Proteolytic destruction occurring during the period of exudation of fluid and cells

followed by a proliferative phase with partially destroyed lung architechure could account for the

emphysematous lesions.

5.3 Quantitative Studies in Em~hvsema

Quantitative anatomical studies on lungs with emphysema range from those using semi-

quantitative scoring systems to detailed stereologic and other morphometric analysis of the

lung. The introduction of CT resulted in renewed interest in these areas where several

investigators developed semi-quantitative scoring systems for the C f scans and detailed

analysis of the X-ray attenuation values to quantify the extent of the emphysematous changes

within the lung. Most of these studies attempt to correlate the observed morphologie changes

with the changes in the physiology of the lung.

5.3.1 Grass Analvsis:

The original analysis of the emphysematous change was initiated by Gough and

Wentworth (71) who pioneered the use of paper mounted thin sections and led to the original

description of the extent and severity of the tissue changes in the disease process

(1 35,229,272). The preparation of the lung for this method is extremely tedious and time

consuming so Thurlbeck and colleagues developed a modification of this technique which uses

a picture grading systern where lung slices are compared to a panel of photographs and

assigned a score, or grade, between O and 100. Scores of 5-25 indicate mild emphysema, 30-

50 moderate emphysema and 60 or greater are defined as severe (227,229). It has been

shown that the semiquantitative panel grading system provides a fast, efficient and reliable

estimate of the extent and severity of ernphysema (227,229), but was never designed to be

used with lobectomy specimens (272) and does not provide quantitative data in ternis of

absolute volumes or volume fractions of the lung involved by disease (227). Therefore, a

quantitative analysis of the lung must use techniques that provide three-dimensional information

on the lung specirnen and not cornparisons to a picture. Quantitative analysis can be achieved

using the standard stereologic point counting grid (227) or a grid of squares (158,224)

superimposed over the gross specimen and the number of points. or squares, over emphysema

is divided by the number over normal lung.

5.3.2 Histolocric Analvsis:

A gross analysis gives data on the extent and the severity of the changes in the lung

tissue while a histologic analysis attempts to quantify the changes in the tissue at the cellular

and molecular level. Since the definition of emphysema centers on enlargement of airspaces

with destruction of lung tissue, the most common analysis is quantification of the airspace size

performed using the rnean linear intercept (Lm) technique (4,45,70,93,249). This procedure is

performed by counting intercepts between a test grid of Iines and lung parenchyma on

systematic random sections of the lung and is a derivation of the surface area method,

described in the opening chapter (45,249,252,253,257). A variation by Lang and coworkers

(128,129) uses an automated image analysis system to rneasure the length of the alveolar

walls and then calculates the surface area per unit volume (AWUV). Point counting techniques

have also been used at this level to estirnate the tissue and airspace volume changes as a

result of the destruction (83,227).

These data al1 show that in emphysema t h e is a reduction in the tissue volume fraction

(83), with a corresponding decrease in the surface area (Sv or AWUV) (56.72.128,129), and an

increase in the size of the airspaces as measured by Lm (93,181,229). However, in human

subjects these results are not sîraight forward as shown by the wide range of alveolar sires and

surface areas with some estimates of severe emphysematous lungs within the normal range

(229). Undoubtedly some of this variation is due to the sample used, which tend to be patients

in their sixth decade with significant smoking histories who would have significant age related

changes as well as varying degrees of emphyserna (229,239).

Another analysis is the destructive index (1 27,229), which records breaks in the

parenchymal tissue, and bronchial attachments to peripheral airways. This technique does not

provide three-dimensional lung measurements and does not seem to be very sensitive to mild

changes in the lung parenchyrna (229).

There have also been atternpts to quantify the number of inflarnmatory cells within the

vasculature and airspaces of emphysematous lungs (83,89,241). These studies have shown

that there is an increase in the number of macrophages within the airspaces of patients with

ernphysema, which is further increased in patients classified as current smokers (241). They

also show that the number of PMN in the microvasculature increases in subjects that are

actively smoking (142). Studies have also shown that here is an increase in both the

degradation and synthesis of collagen and elastin producing changes in the parenchymal tissue

thickness indicative of a fibroproliferative response (1 2,24,128-130,179).

5.3.3 RadioIoaical Analvsis

The CT scan has proven to demonstrate the presence, extent and severity of the

emphysema and to correlate better with pathology and pulmonary function tests

(72,120,222,229) than the chest film (1 73,192,2O2,Z8). CT scans also yield an image of the

lung slice that is similar in appearance to the gross slice and has led investigators to modify the

picture grading systern to be used on gross sections cut in the transverse plane so that

cornparisons can be made to the CT image (66). While these correlations have been good,

they consistently underestimate the extent of the mild emphysema (66,158,166) which is

attributed to volume averaging within the slice because the thinner slices used for HRCT yield

better correlation with pathology (64,66,246).

A major source of artifact with CT is due to the variable inflation volumes of the lung

during the scan which can differ from slice to slice and from scan to scan. This has led

investigators to design apparatus that enable scans to be obtained at a known and reproducible

spirometric level (1 0.1 08.1 95) or to develop correction criteria which express the lung inflation

as a percent of the individual's TLC (31.32). This latter technique enables the CT scan to be

compared to the gross specirnen and to use stereologic techniques on biopsy material.

Investigators have also shown that expiratory CT scans are more reproducible intra and inter

scan. and very efficient at dernonstrating the hyperinflated regions of the lung due to gas

trapping (47,65).

As previously mentioned, the CT scan data contains infornation on the attenuation of X-

rays within the lung parenchyma, and Hayhurst showed that patients with emphysema had

more voxels in the -900 to -1000 HU range (81). In another study Gould used a computer to

assess the frequency distribution of the X-ray attenuation values and found the lowest srn

percentile of the voxels correlated with the extent of the emphysema and had a negative

correlation with measurements of AWUV (72). Müiler et a l used a similar technique to compare

the extent of emphyserna quantified by a 'density mask" at different HU cutoff values with the

extent of emphysema quantified on the gross lung specimen (120.166). He showed that the

best correlation for conventional CT scans used a density mask of -91 0 HU and Gevenois

followed this study by demonstrating that -950 HU was the best cut off for HRCT (64).

5.4 Exneriment #3

Chapters 3 and 4 of this thesis have described a technique which combines CT

measurements with quantitative histology to provide infornation about the structure of normal

lungs and lung with idiopathic pulmonary fibrosis (31). This technique uses X-ray attenuation

values to estimate lung density and calculate lung weight, tissue and gas volumes. The CT

estimates of the proportion of tissue and air in the total volume is then used to correct the

histologie measurernents made on resected lung tissue to the level of lung inflation present

during the CT scan.

This chapter compares the lung structure of heavy smokers who had maintained nearly

normal lung function with minimal emphysema. as defined by a 'density mask" technique, to

that present in patients with similar smoking histories and advanced emphysematous lung

destruction. This procedure provides quantitative data on lung structure that wiil be useful in

foltowing the natural history of the devetoping emphysematous process, correlating these

structural changes with function and assessing the benefit of lung volume reduction surgery.

5.5 Materials and Methods

The procedures used in this study were approved by the ethical review boards of St.

Paul's Hospital, the University of Pittsburgh Hospital. and the Universities of British Columbia

and Pittsburgh. Al1 of the patients signed informed consent forms that allowed the use of

physiologic data, CT scans and the surgically resected tissue. They were divided into control,

mild-ernphyserna, and severe-emphysema groups according to the severity of emphysematous

destruction of lung tissue. The patients in the control and mild-emphysema groups required

either a lobectomy or pneumonectomy for a small, non-obstructing, peripheral bronchogenic

carcinoma and were part of an ongoing study of lung structure and function at the University of

British Columbia Pulmonary Research Laboratory in Vancouver. The severe-emphysema

group were selected for lung volume reduction surgery at the University of Pittsburgh. These

patients were separated into their groups based on the percent of the lung defined as

emphysernatous using the "density mask" technique (166).

5.5.1 Pulmonarv Function Studies

Spirometry and lung volumes were measured pre-operatively with the subjects seated in

a volume displacement body plethysmograph using previously described techniques

(31,32,93,210). Functional residual capacity (FRC) was measured using the Boyle's Law

technique (44,146,181). Total lung capacity (TLC) was calculated by adding inspiratory

capacity (IC) to FRC, and residual volume (RV) was calculated by subtracting vital capacity

(VC) from TLC. Diffusing Capacity (ilLCO) was measured by the single breath method as

described by Miller and associates (1 57). The predicted normal values for FEV, and W C were

those of Crapo et al (33), for Drco; Miller et al (157) and TLf: was predicted using Goldman's

values (70).

5.5.2 CT Studies

The subjects in the study received a conventional, non-contrast CT scan (1 0 mm thick

contiguous slices) on a GE 9800 Highlight Advantage CT scanner (General Electric Medical

Systems, Milwaukes, WI) approxirnately one week prior to surgery. All scans were performed

with the subject supine during breath holding after an inspiration. The image data was

transferred to a Silicon Graphics Indy Workstation (Mountainview. CA) for analysis of the X-ray

attenuation values.

The CT scan analysis used to evaluate the lung has been described in detail elsewhere

(31,32). Briefly it was performed using a prograrn wntten for the numerical analysis package

PV-Wave (Visual Numerics, Boulder CO). The lung parenchyma was segmented from the

chest and the large central blood vessels, and the volume of the whole lung (tissue and

airspace) was calculated by summing the voxel dimensions in each slice. The density of the

lung (glml) was estimated using equation 7 (31,82). Lung weight was estimated by multiplying

the mean lung density by the volume. The volume of gas per gram of tissue for each voxel was

calculated according to the equation 8 (31,32). The frequency distribution of the individual

voxels was plotted and the moments of the cuwe were obtained.

Volume fraction (V,) of tissue and airspace in the resected portions of lung for the

severe-emphyserna cases and specific regions identified on the lobectorny specimen was

calculated according to equation 9, which has been fully discussed above (31,32).

This CT estimation of volume fraction is used to correct the histologie estimates of the tissue

and airspace to the level of inflation the patient achieved during the CT scan (31).

The extent of the emphysema in each patient was estimated by applying a "density

maskn to the CT scans using the settings of Müller et a l (1 66) (figure 16). This procedure

identifies al1 of the voxels within the lung that have a HU value of -91 0 or less and expresses

them as a percent of the total. An HU value of -91 0 represents a lung inflation value of 10.2 ml

gas per gram of tissue (equation 2). This is three standard deviations above the 6.0 ml/g

established for TLC in patients of similar age and smoking history which have normal lung

function (32). The patients with mild disease had greater than five percent but tess than 20

percent of their lung volume inflated to volumes above 10.2 mlfg, while patients with severe-

emphysema al1 had greater than 20 percent of their lung inflated above this volume. This

technique for measuring emphysema was validated by comparing the CT measurement of the

percent emphysema in a lobe to the quantitative histological estimates of the actual amount of

emphysema present in that lobe after it was resected.

5.5.3 Quantitative Histoloav

Resected Lobes:

The specimens were prepared for quantitative histology as previously described (31,32).

Briefly, this was perforrned by inflating the fresh surgical specimen with Optimal Cutting

Temperature (OCT) compound (Miles Laboratories, Elkhart, IN) diiuted 1 :1 with normal saline,

Figure 16. A: CT scan of human lung. 8. CT scan with density mask (-91 0 HU = 10.2 mVg) applied, attenuation values less than -91 0 is shown in red and values greater than -910 is in blue.

and freezing in liquid nitrogen. The frozen specimens were cut into 2 cm thick slices in the

transverse plane on a band-saw and fixed in 10% buffered formalin at room temperature for at

least 24 hours. The volume fraction of normal and emphysematous lung was estimated from

the gross lung slices by floating them in water and overiaying a grid of points. The number of

points falling on emphysematous holes (severe, >5 mm diameter, moderate, 2-5 mm, mild <2

mm) and normal lung parenchyma was counted using a magnifying lens to estirnate the volume

fraction of emphysema and normal lung (figure 17). Hematoxylin and eosin stained sections

were prepared from random samples of the lobectomy specimens for the stereologic analysis.

A second set of slides were prepared from a subset of the patients where the site of the biopsy

could be identified on the CT scan.

Luna Volume Reduction Suraerv S~ecimens:

The tissue from the patients with severe-emphysema was received fresh from the

operating room following the lung volume reduction surgery procedure and fixed, as received, in

10% formalin. Representative samples were embedded in paraffin and stained with

hematoxyiin and eosin.

Stereolow:

To optimize the sampling for the stereologic analysis, a cascade design technique was used as

has been previously described in chapter 1 (31,32,35).

Level 1 was performed on the fixed slices of the lobectomy specimens from the subset

of patients where the histology sample site could be identified on the CT scan. Levels 2 and 3

were performed on al1 available sections at the light microscopy level using the point counting

program Gridder (Wilrich Tech, Vancouver, B.C.) which generated random fields of view,

projected a grid on to the field of view via a camera-lucida attachment on a Nikon Labophot

Iight microscope and tabulated the counts. Level2 used 100x magnification with a grid of 80

points and 40 lines. The number of points falling on airspace, tissue (lung parenchyma), and

medium sized blood vessels (50-1 000 pm) as well as the number of intersects between the grid

Iines and the parenchymal-airspace interface were tabulated. Level3 was performed on 10

random fields of view per slide at 4 0 x magnification and the number of points falling on

airspace components (Alveolar macrophages, alveolar PMN, other objects in the airspace, and

empty space) as welI as the tissue components (alveolar wall, capillary lumen, and small blood

vessels (20-50 pm)) were counted using a 100 point grid.

The volume fraction (VY) of each of the lung cornponents, (VV,,), were estimated at

each level according to equation 3, and the surface density Sm, was estimated using

equation 4. Since surface density is the surface area in a given volume, the surface area of the

parenchyma is calculated by multiplying the surface density by the volume of the lung. This

analysis was performed on the random sections for an estimate of the total lung surface area

as well as the biopsies frorn the specific sites to estimate the surface density in specific regions

of the lung that could be identified on CT. The surface density measurements for al1 patients

were pooled and tested for correlation with the CT measurements of ml gas per gram of tissue

from the same region using a mixed effects regression analysis. The overall VV is calculated by

multiplying the Vv of the lung component at the highest level by the Vv that contained it in the

previous levels according to equation 5.

5.5.4 Statistical Analvsis

AH data were analysed using the one way analysis of variance and the multivariate

analysis of variance. Transformations were made on certain variables to normalise distributions

and to make variances homogeneous. The correlation between CT measurements of lung

expansion and surface area per volume as well as the diffusing capacity of the lung and surface

area were analysed using a mixed effects regression analysis. A p-value of less than 0.05 was

considered significant.

5.6 Results

The control group (N=23) has less than five percent of their lung volume in the

emphysematous category ( greater than 10.2 ml gas per gram of tissue). Those with mild

disease (N=7) have a mean of 13 percent (range 5-20%) in this categoiy and those with severe

emphyserna (N=14) have a mean of 46 percent (range 24-6096) in this category.

The patient demographics (table 12) show that the control group is slightly younger but

have a similar sex distribution, and body size to the emphysema groups. The smoking history

is not statistically different between the three groups but the patients with mild emphysema

tended to srnoke less. Those with mild-emphysema have FEV1, W C , TLC and RV values that

are similar to the controls but the FEVi/FVC ratio and DLm afe reduced while the FRC is

elevated. Those with severe emphyserna have grossly abnormal lung function with al1 values

showing the classic obstructive pattern of reduced FEV,, FVC, FEVJFVC and DLm and

elevated TLC, FRC, and RV.

The CT estimates of lung volume (table 13) show an increase in total lung and airspace

volume in the severe-emphysema group compared to the mild-emphysema group which in tum

is greater than the controls. The tissue volume and therefore the lung weight is decreased in

severe-emphysema, but there is no difference in lung weight between the mild-emphysema

group and the control subjects.

The cumulative frequency distribution curves (figure 18) of ml gas per gram of tissue

present in each voxel are different between the three groups. The vertical arrow indicates the

cut off at -91 0 HU in the density mask technique which is equivalent to 10.2 ml gas per gram of

tissue. In the control group 99 I O percent of the voxels are below this cut off, compared to 87

1 percent of the mild-emphysema and 54 I 3 percent of the severe cases of emphysema

(table 13). Those with severe-emphyserna have 20% of their lung inflated beyond 20 mUg

which is more than three tirnes the amount of air contained in the normal hurnan lung at TLC

(32). Further information about the frequency distribution of the mVg values is shown in table

14 where the control lungs show a normal distribution with mean, median and mode values that

are closely sirnilar (4.5 I 0.2, 4.4 I 0.1, and 4.4 I 0.2 mVg) and a relatively srnall variance (2.9 * 2.3 mlfg). The mild-emphysema group show a distribution which is slightly shifted to the right

around a rnean of 7.1 0.3 ml/g, median of 5.8 I 0.3 rnVg, and mode of 5.3 * 0.6 rnlfg with a

very large variance (207.8 * 1 11.9 ml/g). The severe-emphysema group has a flattened

distribution which is shifted to the right with a mean of 14.0 r 1.2 mVg, a median of 9.8 I 0.4

ml/g, mode of 8.1 * 0.5 ml/g and the largest variance (578.5 199.6 ml/g). When the median

values are expressed as percentage of rneasured TLC, the control and rnild-emphyserna

patients are not different (66.0 I 2.2% versus 74.6 I 3.2%) however, the median vaiue for the

severe-emphysema patients was significantly greater (99.7 I 4.9%) than the other two groups

indicating that these patients lungs are hyper-inflated.

Table 15 shows the stereology data for al! three groups corrected to the level of lung

inflation present during the CT scan. This shows a decrease in the percentage of the lung

occupied by tissue, and an increase in the percentage occupied by airspace in the group with

severe-emphysema. There is a small but significant decrease in the surface area per volume

of lung in rnild-ernphysema and a very marked reduction in this value in the resected lung

regions of the group with severe-emphysema.

The airspace of these lungs contains an inflarnmatory exudate. The cellular component

of the exudate varies with PMN increasing from 1 * O X 10' cells/m2 in the control group to 9 I 6

X 1 o8 cellslm2 in the mild disease and 197 I 34 X 10' cells/m2 in the lungs with severe

ernphysema. The macrophages, on the other hand, were not increased in the mild emphysema

and showed a large but variable increase in the group with severe emphysema. The fiuid

volume present in the exudate was increased only in the mild-emphysematous group.

Figure 19 shows the mixed effects regression analysis between lung volume (ml gadg

tissue) and surface area per volume which has a negative slope (-3.7) and an intercept of 124

cm2/ml both of which are significantly different from zero (p=û.001). The mixed effects

regression analysis of surface area versus measured DLCo (figure 20) shows a positive

correlation with a dope of 0.1 rnl/min/rnm~grn~ and an intercept of 7.1 mVminfmmHg.

The amount of emphysema detected in the same lobe by either CT or histological

analysis is compared in table 16- This shows that the volume fraction of the lesions greater

than 5 mm in diameter was similar using both techniques. However, the histologic analysis

shows that a large fraction of the lobe in both groups of specimens contains lesions smaller

than 5 mm in diameter which were not detected by the CT scan.

Table 13:

Lung Volumes and Weights

Values are rneans * S.E.M.

Group

Control

Mild-Emphysema

Severe-Emphysema

Denslty

Mask

(%<-910 HU)

1 * O

13 & 1 "

46 I 3**

Lung

Weight

(9)

1019 I 37

1104 î 35

81 O I 37'"

Total

Volume

(ml)

47721223

6232 A 41 O"

6725 k 384"

different from Control, p < 0.005

" different frorn Control and Mild-Emphysema, p < 0.008

Air

Volume

(ml)

38151194

51 95 388"

5964 I 353'

Tissue

Volume

(ml)

11201167

1 037 * 33

760 k 35"'

Volume Fraction

Tissue

("w 17.8 î 0.5

14.3 î 0.8"

8.9 * 0.4'"

Table 15:

Quantitative Histology

Control

Mild-Emphysema

Severe-Em physema

Airspace Tissue Capillary Vessel Vessel

Lumen Lumen Lumen

(20-50 pm) (50-1000 (im)

[%l [%l ["hl vôl [XI

81.4* 1.4 15.41 1,2 1.8k 0.2 0.4 k0.1 1 .O î 0.3

85.6i2.6' 11.812.1 1.410.3 0.1 10.0' 1.2 I 0.3

93.7 î 0.6'* 5.7 î 0.6" 0.5 * 0.1"" 0.0 I 0.0' 0.6 I 0.0"

Alveolar #Mac~hq #PMN - Exudate Area Area

Surface Surface

Area Areai

Volume

m21 [m2M

118111 2513

10718 1812'

2915" 4*1**

Values are means î SEM.. alveolar macrophage: Macphg, polymorphonuclear cells: PMN

* different from Control, p < 0.05

" different from Control and Mild-Emphysema, p < 0.005

10 ml gaslg tissue

Figure 18. Shows the CT density of the lung expressed as a cumulative distribution of the voxels on the CT scan against gas volume per gram of tissue. The red line represents the control group, and the blue line the mild-emphysema group, and the purple line the severe-emphysema group. p < 0.001

100

Surface Area (m2)

Figure 20. Shows the mixed affects regression line (red line) and the 95% confidence limits (Mue lines) of the histologie measured surface area of the lung and the measured diffusing capacity of the lung for carbon monoxide.

5.7 Discussion

The results of this study show (table 12) the three groups of patients have a similar sex

distribution, body size and smoking history but the patients in the two emphysematous groups .

are slightly older than those in the control group. Those with mild emphysema have a lower

FEV,/FVC ratio and DLCo with a slightly higher FRC than the control group, whereas those with

severe disease have reduced dynamic lung volumes combined with elevation in al1 of the

subdivisions of lung volume, indicating that they have a marked reduction in the ability to empty

their lungs. These patients also have a reduction in their diffusing capacity which is greatest in

the severe group.

Table 13 shows that severe-emphysema is associated with an increase in gas volume

and a reduction in lung tissue volume and weight. Some of this decrease in tissue volume

could be due to a reduced blood volume associated with the shift of the lung into zone 1 and 2

caused by airway closure and high alveolar gas pressure relative to puimonary venous and

arterial pressure (197,264,265). However, further analysis of the data (table 15) shows that

there is a very large decfease in the tissue volume fraction and the surface area per volume

consistent with lung destruction. The mild-emphysema cases show an increased total and

airspace volume without a significant decrease in the tissue volume, lung weight (table 13) or

lung surface area (table 15). These data suggest that early emphysema is associated with

minimal destruction of surface area associated with an increase in Iung volume to produce a

significant reduction in surface to volume raüo whereas severe emphysema is dominated by a

destruction of the parenchymal surface. This interpretation is consistent with early lesions in

the peripheral aitways that would prolong the time constants of peripheral lung units and

increase lung volume before the onset of the destruction of the lung surface area (90).

There is a marked inctease in the volume of exudate ont0 the alveolar surface in the

cases with mild emphysema and this fluid contains excess numbers of PMN. In the more

severe cases the fluid volume returns to control levels but the number of PMN and alveolar

macrophages present on the surface are both increased. A change in the nature of the

exudate in severe disease could be important in the pathogenesis of lung destruction. The

relative importance of the proteolytic enzymes produced by PMN and those produced by

alveolar macrophages and other cells in the pathogenesis of emphysema is controversial

(1 7,89). The cigarette smoking habit increases the number of PMN and macrophages in lung

tissue and airspaces (205,241) and active smoking increases PMN concentration in the lung

microvessels (1 42). Our data show that mild lung destruction is only associated with an

increase in PMN whereas advanced lung destruction is associate with large numbers of both

PMN and macrophages on the alveolar surface. As the controls in this study had a significant

smoking history, their macrophage level is probably already elevated from that in non-smokers.

The alveolar macrophage and the PMN are linked through a network of cytokine and growth

factors (57) and the increased macrophages may keep the recniitment of PMN high enough

that the PMN degradative enzymes can destroy the lung tissue to produce the severe disease.

Other investigators have estimated the severity of emphysema using either a 'density

rnask" with a single cut off value (1 58,166) or the lowest fifth percentile (72) of the x-ray

attenuation values expressed in Hounsfield Units (HU). The simple calculation we use to

convert HU obtained frorn conventional CT scans to a more physiologically meaningful unit, ml

of gas per gram of tissue (32), allows us to relate regional to total lung volume and determine

the overall lung volume at which the CT scan was obtained. This data show (table 14) that

when the lung volume at the time of the CT scan is expressed as a percent of measured TLC

the control group was at 66.0 I 2.2% of TLC the mild emphysematous group was at 74.6 I

3.2% and the severely emphysematous group was at 99.7 I 4.9% of TLC. This tendency for

those with severe disease to reach a higher overall lung volume is consistent with the fact that

emphysematous lesions reach full inflation at very low transpulmonary pressures (92).

Further analysis of frequency distributions of regional lung volumes (ml/g) show that the

control patients have a normal distribution of Iung inflation with a similar mean, median and

mode and 99% of the lung being below the density mask cutoff value (table 14, figure 18).

Emphysema shifts this curve away from normality with increasing proportion of the lung being

above the cut off value in the severely affected group (table 14, figure 18). This value (-91 0 HU

= 10.2 ml/g) is three standard deviations above that obtained by dividing measured TLC by

measured lung weight in the control group (6.0 mVg) (32). Therefore, some of the lung

between 6.1 and 10 mVg should also be abnomal. This is confirmed by the anatomic studies

of the resected lobes which show a large percentage of emphysematous holes less than five

mm in diameter are not detected by CT (table 16). This confins previous reports (1 58.1 66)

showing that the CT technique accurately identifies holes larger than 5 mm in diameter but fails

to identify the smaller lesions. These smaller lesions are probabiy represented by values

between normal TLC (6.0 mVg) and the cut off (1 0.2 mVg) in the density mask technique. The

inability to detect these small lesions with CT is the result of volume averaging on 10 mm thick

slices which affects the segmentation of objects less than 5 mm in diameter (38,276). Thin

slices of a high resolution scans allow better visualization of these smaller lesions (64,97) but

Kemennk and associates have shown that the signal to noise ratio is so high that 1 dirninishes

the ability to discriminate different lung densities within the same slice (1 14). This means that

the thinner slice provides better visualization of srnall structures but the thicker slice and a lower

spatial frequency reconstruction algorithm provides better lung density discrimination and more

reliable estimates of the extent of the ernphysema in the lung.

In earlier studies, Gould et al also showed a negative correlation between the EMI

measurement of X-ray attenuation and a histologic rneasurement of surface area of alveolar

wall per unit lung volume (AWUV) (72). The relationship between surface area and volume of a

sphere is one of radius squared divided by radius cubed which means that the surface to

volume ratio will decrease as the sphere enlarges. Our data (figure 19) shows the predicted

negative relationship between lung surface area per volume (cm2/ml) and lung volume (mVg)

with lung expansion. The regression line for the mean value has quite tight 95% confidence

Iimits in the range of the normal values for RV, FRC and TLC and this relationship persists up

to the cut off for the detection of holes greater than 5 mm in diameter (10.2 Mg) . The

confidence tirnits widen at higher lung volume presumably because of a variable destruction of

the lung surface. This relationship allows lung surface to volume rations and lung surface area

to be predicted form the CT and these measurements could be used to track the progression of

lung destruction in individuaI patients.

There was a good correlation between the surface area calculated from lung histology

and the diffusing capacity of the lung for carbon monoxide (figure 20) that is in agreement with

published data (72). The ability to relate lung surface area to CT measurements of lung volume

per gram and to a functional assessment of gas exchange impairment provides a powerful tool

for the analysis of the structural and functional changes in chronic lung diseases. The

algorithms currently used to assess the CT could be easily modified to rnake these calculation

available to clinicians who might use them to assess the progress of lung destruction in

emphysema and to evaluate the impact of lung volume reduction surgery and possibly other

therapies on Iung structure.

In summary, our data show values obtained for the CT scan can provide an accurate

assessment of the tissue and airspace changes in emphysema that should be useful in the

longitudinal assessment of emphysematous lung destruction.

CHAPTER 6: Summarv and Discussion

6.1 Sumrnary

The data'presented in this thesis has established that the CT scan can be combined

with quantitative histology to quantify the tissue changes in chronic lung diseases. The CT

scan provides a powerful tool for Me estimation of total and regional lung volumes and weight,

and provides a method of correcting open lung biopsy matenal to the level of lung inflation in

the intact thorax.

The data from the IPF studies presented in chapter 4 show that the decrease in total

lung volume which is due to a loss of the airspace volume with minimal changes in the tissue

volume. In emphysema, there is an increase in the total lung volume which is due. first of all, to

an increase in the airspace volume, with minor changes in the tissue volume, which is then

followed by a decrease in the tissue component through proteolytic destruction. The frequency

distribution cvrves of these studies show that the control lungs have a normal distribution

around a rnean and median of 4.5 ml g a d g tissue which is 66% of the patient's TLC divided by

lung weight (6.0 mllg). In IPF the curve is grossly shifted to the left (figure 13) indicating that

there is an increase in the proportion of the lung occupied by tissue, and the long tail suggests

that there are regions of hyperinflation probably due to the cystic changes of the end stage

lung, concomitant emphysematous changes. or a combination of both. In emphysema (figure

18), the cuive is shifted toward more gas volume per gram of tissue and this is especially true

in the severe cases where over half of the lung has a density greater than 10.2 mVg, which has

been shown to be a good estimate of the emphysema present on conventional CT scans

(120,158,166). Also, the CT density provides data on the tissue and airspace volume fraction

which is essential for the correction of lung biopsy specimens to an appropriate level of inflation

for cornpansons between individuals and even different biopsy from the same lung.

The frequency distribution data in the CT analysis provides an important tool for the

longitudinal studies of lung disease because it is sensitive to changes in lung composition. This

present study confirms the work of Hayhurst (81) in emphysema and Hartley (78) in interstitial

lung disease, who showed that the properties of the attenuation curve correlate with changes in

lung structure. We have modified their technique to express the x-ray attenuation values in

terms of ml of gas per gram of tissue which we propose is a more physiologically useful

expression because it allows the analysis of the lung in terms of the pleural pressure gradient

and regional lung inflation. The greatest problem with this analysis is that it does not separate

airspace infiltration from fibrotic remodeling and more investigations need to be done in this

area with careful correlations between the frequency distribution curves of the radiological

categories 'ground-glassn attenuation and ' honey-combing' with the histological estirnates of

the lung structure.

A quantitative histological analysis of the lung assumes that the specirnen has been

inflated to an appropriate level. We have shown in chapter 3 that this is a valid assumption if

the gross specirnen is able to be inflated through the airways. However, figure 14 shows that

when the inflation is not controlled. the histologie information is not comparable between

individuals, or even different biopsies. This thesis describes a technique where CT

rneasurements of lung density a n be used to calculate the volume fraction of tissue (equation

9) in specific ragions of the lung. These values can then be used to correct open lung biopsies.

where the inflation cannot be controlled, to an appropriate level of inflation for quantitative

anatomic analysis (chapter 4).

The stereology data shows that there is an increase in the tissue volume fractions of

lungs with IPF and a decrease in emphysema. There is, however, a loss of surface area in

both condÏtions which is due to a reorganization of the lung parenchyma in I PF and a

destruction of alveolar walls in emphysema. Figure 20 shows that the surface area of the lung

is related to the diffusing capacity in the emphysematous patients. When the surface area and

DLco measurements from the patients with IPF are added to the analysis, a statistical difference

is not detected to the emphysematous cases so the groups can be combined. The rnixed

effectç regression line for al1 of the patients (figure 21) has a slope of 0.09 rnVmin/rnrn~g/rn~

and an intercept of 9.4 ml/min/mmHg, both of which are highly significantly different from zero

(p < 0.001). This is an important finding because CT measurements of lung inflation have been

shown to have a negative correlation with the surface area (figure 19. (72)) and this provides a

tool for the clinician to quantitatively follow the patient with chronic lung disease and to assess

the changes in lung structure and function over the course of the disease.

The inflammatory response in the lung is very important in the pathogenesis of both IPF

and emphysema and this study has shown that there are increased inflammatory cells in the

airspaces of both diseases. These cells consist mostly of aiveolar macrophages because of

the chronic condition of the process, and the central role that macrophages play in terms of

modulating the inflammatory and the fibroproliferative response. It is very tempting to

characterite these changes as different outcornes to the same process because it has been

shown that there is a fibroproliferative response in the early stages of emphysema that results

in modifications to the elastin and collagen content of the parenchyma which is characterized as

fibrosis. However, this fibrosis is mild and the predominant feature of emphysema is the

destruction of the lung tissue. The exact molecular events of the fibrosis still need more

clarification, but it is now obvious that the larger molecules of collagen and elasün are not the

only players in the game and that proteoglycans are a very dynamic constituent of the fibrotic

response. Proteoglycans appear very early in the fibroproliferative response and have been

linked to key roles in fibrosis such as tissue hydration. cytokine binding and cellular adhesion

(77,200,268). Their exact role in emphysema is still to be investigated, but the increased

alveolar wall thickness seen early on in the emphysematous lesions is an attractive site for PG

synthesis. Why one process proceeds to the proliferative stage while the other one elevates

the degradative response still needs to be delineated.

6.2 Future Directions

Quantitative three dimensional analysis of the lung in chronic lung diseases is an

important step. Correlations with semi-quantitative sconng systems and functionaI

characteristics has proven to be weak and of Iimited value aside from a quick process for

assessing the severity of the disease. It is time to move beyond these semi-quantitative

analysis and describe the lung in terrns of the three-dimensional structural parameters that it

possesses. The CT scan avails itsetf for quantitative longitudinal studies of the lung because it

is minimally invasive and easy to perform. A few simple calculations of the data may provide

important information that allows the assessment of the lung in terms of airspace enlargement,

Ioss of surface area and tissue destruction or reorganization. This may become very important

in the long terni follow up of patients with IPF to assess the time course of the disease and to

assess treatment protocols and in emphysema for the assessment of the lung structure in

ternis of selecting patients for the palliative surgical treatment of Iung volume reduction surgery.

Lung volume reduction surgery is a controversial and very expensive procedure and the ability

to choose patients pre-operatively that will benefit from the intervention has great ramifications

for both the patient's health and the resources of the health cafe provider. A quantitative CT

analysis of the lung may be able to delineate the structural factors in the lung that are

responsible for the improvement or non-improvement of these patients. Also, by use of a

density mask and three-dimensional reconstruction, a map can be provided to the surgical team

clearly demonstrating the extent and location of aie emphysema to help decide where the best

site for reduction surgery (figure 22).

6.3 Conclusion

In conclusion, it has been shown that the combination of quantitative CT and stereology

provides reliable quantitative data on the lung structure in chronic lung disease. These studies

have detailed the procedure that enables the CT to rneasure the total and regional volume

changes within the lung and to quantify the ultra-structural changes responsible for the patho-

physiological changes seen by the clinician.

Figure 22. Three dimensional reconstruction of a human lung with emphysema using CT scan images. Emphysema is shown in red. and normal lung parenchyma in blue, while dense tissue is green.

REFERENCES

Abbas, A.K., A.H. Lichtman, and J.S. Pober. Cells and tissues of the immune system. In Cellular and Molecular Immunology, 1 st ed. W.B. Saunders Co. Philadelphia. pp.13-34. 1991.

Adams, H., M.S. Bernard and K. McConnochie. An appraisal of CT pulmonary density mapping in normal subjects. Ciinical Radioiogy- 43:238-242. 1991 .

Adler, K.B., RB. Low, K.O. Leslie, J. Mitchell and J.N. Evans. Contractile cells in normal and fibrotic lung . Laboratory investigation. 60:473-485. 1 989.

Aherene W.A. and M.S. Dunnill. Morphometry, Edward Arnold, London. 1982.

Akira. M., H. Hamada, M. Sakatani, C. Kobayashi, M. Nishioka and S. Yamamoto. CT findings during phase of accelerated deterioration in patients with idiopathic pulmonary fibrosis. American Journal of Roentgenology. 168:79-83. 1 997.

Akira. M.. M. Sakatani and E. Ueda. ldiopathic pulmonary fibrosis: progression of honeycombing at thin-section CT. Radiology. 189:687-691. 1 993.

American Thoracic Society. Chronic bronchitis, asthma and pulmonary emphysema. A statement by the Cornmittee on Diagnostic Standards for NonTuberculous Respiratory Diseases. Arnetfcan Review of Respiratory Diseases. 85762-768- 1 962.

Balazs, G., S. Noma, A. Khan, T. Eacobacci and P.G. Herman. Bleomycin-induced fibrosis in pigs: evaluation with CT. Radioiogy. 191 :269-272. 1994.

Bateman, E.D., M. Turner-Warwick and B.C. Adelmann-Grill. lmmunohistochemical study of collagen types in human foetal lung and fibrotic lung disease. Thorax. 36545-653. 1981.

10. Beinert, T., J. Behr, F. Mehnert, P. Kohz, M. Seemann, R. Rienmulier and ReiserM. Spirometrically controlled quantitative CT for assessing diffuse parenchymal lung disease. Journal of Cornputer Assisted Tomography. 19:924-931. 1 995.

11. Bellamy, E.A., U. Nicholas and J.E. Husband. Quantitative assessrnent of lung damage due to bleornycin using computed tomography. The Bna'sh Journal of Radiology. 60:1205- 1209. 1987.

12. Belton, L.C., N. Crise, R.F. Mclaughlin and E.E. Tueller. Ultrastructural alterations in collagen associated with rnicroscopic foci of human emphysema. Human Pathology. 8:6691977.

13. Bensadoun, ES., A.K. Burke, J.C. Hogg and C.R. Roberts. Proteoglycan deposition in pulrnonary fibrosis. Amerkan Journal of Respiratory & Critical Care Medicine. lW:l819- 1828. 1996.

14. Biemacki, W., G.A. Gould, K.F. Whyte and D.C. Fienley. Pulmonary hemodynarnics, gas exchange, and the severity of emphysema as assessed by quantitative CT scan in chronic bronchitis and ernphysema. Amencan Review of Respiratory Diseases. l39:l~O9-lSl5. 1989.

15. Bitterman, P.B. and C.A. Henke. Fibroproliferative disorders. Chest. W:8l S-845. 1991.

16. Bolender, R.P., D.M. Hyde and R.T. Dehoff. Lung morphometry: a new generation of tools and experiments for organ, tissue, ceII, and molecular biology. American Journal of Physiology (Lung Cellular and Molecular Physiolosy). 265:L521 -L548. 1 993.

1 7. Bowden, D. H. The alveolar macrophage. Environmental Health Perspectives. 55327-341. 1984.

18. Boyd, F.T., S. Cheifetz, J. Andres, M. Laiho and J. Massagué. Transfoning growth factor- B receptors and binding proteoglycans. Joumal of Ce11 Science. I3:13l-l38. 1 990.

19. Bratt, P., M.M. Anderson, B. Mansson-Rahemtulla, J.W. Stevens, C. Zhou and F. Raherntulla. Isolation and characterization of bovine gingival proteoglycans versican and decorin. The International Joumal of Biochemistty. 24: 1 573- 1 583. 1 992.

20. Burt, A.D. Cellular and molecular aspects of hepatic fibrosis. Joumal of Pathology. l?O:IOS-l14. 1993.

21. Cantin, A.M.. R. Boileau and R. Begin. lncreased procollagen III aminoterminal peptide- related antigens and fibroblast growth signals in the lungs of patients with idiopathic pulmonary fibrosis. American Review of Respiratory Diseases. lW:S72-578. 1 988.

22. Cantor, J.O., S. Keller, 1. Mandl and G.M. Turino. Increased synthesis of elastin in amiodarone-induced pulmonary fibrosis . Joumal of Labofatory & Clinical Medicine. 1 O9 1480-485. 1 987.

23. Capelli, A., M. Lusuardi, C.G. Cenitti and C.F. Donner. Lung alkaline phosphatase as a marker of fibrosis in chronic interstitial disorders. Amencan Joumal of Respiratory & Critical Care Medicine. 1553249-253. 1 997.

24. Cardoso, W.V., H.S. Sekhon, D.M. Hyde and W.M. Thurlbeck. Collagen and elasün in human pulmonary emphysema. Amencan Review of Respiratoiy Diseases. 147:975-981. 1993.

25. CarrÏngton, C.B., E.A. Gaensler, RE. Coutu, M.X. fitzgerald and R.G. ~ub ta . Natural history and treated course of usual and desquamative intersUetial pneumonia. The New England Journal of Medicine. 298:801-809. 1 978.

26. Chemiack, R.M., T.V. Colby, A. Flint, W.M. Thurlbeck, J. Waldron, L Ackerson and T.E. King, Jr. Quantitative assessrnent of lung pathology in idiopathic pulmonary fibrosis. The BAL Cwperative Group Steering Cornmittee. American Review of Respimfory Diseases. l44:892-9OO. 1 991.

27. Chemiack, R.M., T.V. Colby, A. Flint, W.M. Thurlbeck, J.A. Waldron, Jr., M.I. Schwarz and

T.E. King, Jr. Correlation of structure and function in idiopathic pulmonary fibrosis. Amekan Joumal of Respiratory & Critical Care Medicine. 151 :Il804 1 88. 1 995.

28. Ciba Foundation Guest Symposium. Terminology, definitions, and classification of chronic pulmonary emphysema and related conditions. Thorax. 14:286-299. 1959.

29. Clark, R. The commonality of cutaneous wound repair and lung injury. Chest. 99~575- 60s. 1991.

30. Corrin, B., D. Butcher, B.J. McAnulty, R.M. Dubois, C.M. Black, G.J. Laurent and N.K. Harrison. lmmunohistochemical localization of transforrning growth factor43 in the lungs of patients with systemic sclerosis, cryptogenic fibrosing alveolitis and other lung disorders. Histopathology. 24: 1 45-1 50. 1 994.

31. Coxson, H.O., J.C. Hogg, J.R. Mayo, H. Behzad, K.P. Whittall, D.A. Schwartz, P.G. Hartley, J.R- Galvin, J.S. Wilson and G.W. Hunninghake. Quantification of idiopathic pulmonary fibrosis using computed tomography and histology. American Joumal of Respiratory & Critical Care Medicine. 155:1649-16Ei6. 1 997.

32. Coxson, H.O., J.R. Mayo, H. Behzad, B.J. Moore, L.M. Verburgt, C.A. Staples, P.D. Paré and J-C. Hogg. The rneasurement of lung expansion with computed tomography and cornpanson with quantitative histology. Joumal of Applied fhysiology. 79:1525-1530. 1 995.

33. Crapo, R. Spirometric values using techniques and equiprnent that meet ATS recommendations. American Review of Respiratov Diseases. 123:659-664. 1 981.

34. Crouch, E. Pathobiology of pulmonary fibrosis. Amencan Joumal of Physiology (Lung Cellular and Molecular Physiologyl. 259:L159-LI84 1 990.

35. Cruz-Orive, LM. and E.R. Weibel. Sam pling designs for stereology. Joumal of Microscopy. 1 22:235-257. 1 981 .

36. Cruz-Orive, L.M. and E.R. Weibel. Recent stereological methods for cell biology: a brief survey. American Joumal of Physiology (Lung Cellular and Moiecular Physiology). 258:Ll48-LlS6. 1 990.

37. Crystal, R.G., J.D. Fulmer, W.C. Roberts, M.L Moss, B.R. Line and H.Y. Reynolds. ldiopathic pulmonary fibrosis. Clinical. histologie, radiographie, physiologic, scintigraphie, cytologie, and biochemical aspects. Annals of Interna1 Medicine. 85769-788. 1 976.

38. Curry T.S.. J.E. Dowdey, and R.C. Murry. Christensen's Introduction to the Physics of Diagnostic Radiology, Lea & Febiger, Philadelphia. 1984.

39. de Cremoux, H., J.F. Bemaudin, P. Laurent, P. Brochard and J. Bignon. Interactions between cigarette smoking and the natural history of idiopathic pulmonary fibrosis. Chest 98:71-76. 1990.

40. Dehoff R.T. and F.N. Rhines. Quantitative Microscopy, McGraw-Hill, New York. 1968.

41. Denison, D.M.. M.D.L. Morgan and A.B. Millar. Estimation of regional gas and tissue volumes of the lung in supine man using computed tomography. Thorax. 41 :620-628. 1986.

42. Diamond, J.R. Analogous pathobiologic mechanisms in glomerulosclerosis and atherosclerosis. Kidney lntemational. 39:S29-S34. 1 991.

43. Dohring, W., G. Linke, and H.-S. Stender. CT densitometry of the lung. In Radiology Today 1 ., Springer-Verlag, Berlin. pp.99-112. 1981 .

44. DuBois, A.B., S.Y. Botelho. G.W. Bedell, R. Marshall and J.H. Comroe. A rapid plethysmographic method for measuring thoracic gas volume: a companson with nitrogen washout method for measuring functional residual capacity in nonal subjects. Journal of Clinical Investigations. 35:322-326. 1 956.

45. Dunnill, M.S. Quantitative methods in the study of pulmonary pathology. Thorax. 17:320- 328. 1962.

46. Dunnill, M.S. Pulmonary fibrosis. HistopathoI~gy~ 16:321-329. 1990.

47. Eda, S., K. Kubo, K. Fujimoto, Y. Matsuzawa, M. Sekiguchi and F. Sakai. The relations beniveen expiratory chest CT using helical CT and pulmonary function tests in emphysema. Amencan Journal of Respira tory & Critical Care Medicine. 155: 1 290-1 294. 1 997.

48. Elias H. Stereology, Spnnger-Verlag, New York. 1967.

49. Elias H. and D. Hyde. A Guide to Practical Stereology, S. Karger, New York. 1983.

50. Engel, J. Domains in proteins and proteoglycans of the extracellular matrix with functions in assembly and cellular activities. lntemational Journal of Biological Macromolecules. 1311 47-1 51. 1991.

51. Engeler, C.E., J.H. Tashjian, C.M. Engeler, R.A. Geise, J.C. Holrn and E.R. Ritenour. Volumetric high-resolution CT in the diagnosis of interstitial lung disease and bronchiectasis: diagnostic accuracy and radiation dose. Amencan Journal of Roentgenology. 163:31-35. 1 993.

52. Erbes, R., T. Schaberg and R. Loddenkemper. Lung function tests in patients with idiopathic pulmonary fibrosis. Are they helpful for predicting outcome?. Chest. 1 11 51 -57. 1997.

53. Feldman, H.A. Families of lines: random effects in linear regression analysis. Joumal of Applied PhysioIogy. 64: 1 721 - 1 732. 1 988.

54. Finley, T.N., E.W. Swenson and J.H.J. Comroe. The cause of arterial hypoxemia at rest in patients withnalveolar-capillary block syndrome". Joumal of Clinical Investigation. 41 :6f 8- 622. 1962.

55. Fischer, B. and K. Morgenroth. Ultrastructural study on hurnan lung in alveolitis versus pulmonary fibrosis . The Clinical lnve~tigator. 71 :452460. 1 993.

56. Flenley, D.C. Diagnosis and follow-up of emphysema. European Respiratory Journal. 3 Suppl 9 5 S-8 S. 1990.

57. Flenley, D.C., 1. Downing and A.P. Greening. The pathogensis of emphysema. Bulletin Europeen de Physiopathologie Respiratoire. 22:245 s-252 S. 1986.

58. Frasca, J.M., O. Auerbach, H.W. Carter and V.R. Parks. Morphologic alterations induced by short-term cigarette smoking. American Journal of Pathoiogy. 1 1 1 :Il-20. 1983.

59. Frasca, J.M., O. Auerbach, V.R. Parks and J.D. Jamieson. Electron microscopic observations on pulmonary fibrosis and emphysema in smoking dogs. Experimental & Mo/ecu/ar Pathology. 1 5: 1 08- 1 25. 1 97 1 .

60. Fujirnoto, K., K. Kubo, S. Yamaguchi, T. Honda and Y. Matsuzawa. Eosinophil activation in patients with pulmonary fibrosis. Chest. 108:48-54. 1 995.

61 . Fulmer, J.D., R.S. Bienkowski, M.J. Cowan, S.D. Breul, K.M. Bradley, V.J. Ferrans and R.G. Crystal. Collagen concentration and rates of synthesis in idiopathic pulmonary fibrosis. American Review of Respiratory Diseases. 122:289-301. 1980.

62. Fulmer, J.D., W.C. Roberts, ER. von Gal and R.G. Crystal. Morphologic-physiologic correlates of the seveflty of fibrosis and degree of cellularity in idiopathic pulmonary fibrosis. Journal of Clinical Investigation. 63:665-676. 1 979.

63. Gamsu, G., C.J. Salmon, M.L. Warnock and P.D. Blanc. CT quantification of interstitial fibrosis in patients with asbestosis: a comparison of two methods. American Joumal of Roentgenology. 1 64:63-68. 1 995.

64. Gevenois, P.A., V. de Maertelaer, P. De Vuyst, J. Zanen and J.C. Yernault. Comparison of computed density and macroscopic morphometry in pulmonary emphysema. Amencan Journal of Respiratory & Cntical Care Medicine. 1 52:653-657. 1 995.

65. Gevenois, P.A., P. De Vuyst, M. Sy, P. Scillia, L. Chaminade, V. de Maertelaer, J. Zanen and J.-C. Yemault. Pulrnonary emphysema: quantitative CT during expiration. Radiology. 1 99~825-829. 1 996.

66. Gevenois, PA., M.-C. Koob, 0. Jacobovitz, P. De Vuyst, J.-C. Yemault and J. Stniyven. Whole lung sections for computed tomographic-pathologic correlations. Modified Gough- Wentworth technique. lnvestigative Radiology. 28:242-246. 1993.

67. Gevenois, P.A., E. Pichot, S. Dedeire, R. Vande Weyer and P. De Vuyst. Low grade coal workefs pneumoconiosis. Comparison of CT and chest radiography. Acta Radiologica. 3S:35l -356. 1 994.

68. Glazier, J.B., J.M.B. Hughes, J.E. Maloney and J.B. West. Vertical gradient of alveolar size in lungs of dogs f rozen intact. Journal of Applied Physiology. 23:694-705. 1 967.

69. Glover, G.H. and N.J. Pelc. Nonlinear partial volume arMacts in x-ray computed tomograph y. Medical Physics. 7:238-248. 1 980.

70. Goldman, H.I. and M.R. Becklake. Respiratory function test: normal values at median altitudes and the prediction of normal results. American Review of Tuberculosis. 79:457- 467. 1959.

71. Gough, J. and J.E. Wentworth. The use of thin sections of entire organs in morbid anatomical studies. Journal of the Royal Microscopy Society. 69:231-235. 1 949.

72. Gould, G.A., W. MacNee, A. McLean, P.M. Warren, A. Redpath, J.J. Best and D. Lamb. CT measurements of lung density in life can quantitate distal airspace enlargement--an essential defining feature of human emphysema. American Review of Respiratw Diseases. 1 37:38&392. 1 988.

73. Gross, P.M., A. Babyak, E. Tolker and M. Kaschak. Enzymatically produced pulmonary emphysema. A preliminary report. Journal of Occupational Medicine. 6:48 1 484. 1 964.

74. Gundersen, H.J.G., T.F. Bendtsen, L. Korbo, N. Marcussen, A. Moller, K. Nielsen, J.R. Nyengaard, B. Pakkenberg, F.B. Sorensen, A. Vesterby and M.J. West. Some new, simple and efficient stereological methods and their use in pathological research and diagnosis. Acta Pathologica, Microbiologica et lmmunologica Scandinavica. 96:379-394. 1 988.

75. Haefeli-Bleuer, B. and E.R. Weibel. Morphometry of the human pulmonary acinus. Anatomical Record. 220:401-414. 1 988.

76. Ham, A.W. and D.H. Cormack. The respiratory system. ln Histology, 8th ed. J.B. Lippincott Co, Philadelphia. pp.725-754. 1979.

7ï . Hardingham, T.E. and A.J. Fosang. Proteoglycans: many forms and may functions. FASEB Journal. 6:861-870. 1992.

78. Hartley, P.G., J.R. Galvin, G.W. Hunninghake, J.A. Merchant, S.J. Yagla, S.B. Speakman and D.A. Schwartz. High-resolution CT-derived measures of lung density are valid indexes of interstitial lung disease. Journal of Applied Physiology. 76:271-277. 1994.

79. Hartman, T.E., S.L. Primack. E.-Y. Kang, S.J. Swensen. D.M. Hansell, G. McGuinness and N.L Miiller. Disease progression in usual interstitial pneumonia compared with desquamative interstitial pneumonia. Chest. 11 0:378-382. 1 996.

80. Hartman, TE., S.L. Primack, S.J. Swensen, D. Hansell, O. McGuinness and N. Miiller. Desquamative intersütial pneumonia: thin-section CT findings in 22 patients. Radiology. 187:787-79O. 1 993.

81. Hayhurst, M.D., D.C. Flenley, A. Mclean. A.J.A. Wightman, W. MacNee, D. Wright, D. Lamb and J. Best. diagnosis of pulmonary emphysema by computenzed tomography. Lancer. 2:320-322. 1 984.

82. Hedlund, L.W., P. Vock and E.L. Effmann. Evaluating lung density by computed tomography. Seminars in Respiratory Medicine. 5~76-87. 1 983.

83. Heemskerk-Gerritsen, B.A.M., J.H. Dijkrnan and A.A.W. Ten Have-Oproek. Stereological

methods; a new approach in the assessment of pulmonary emphysema. Microscopy Research and Technique. 34:556-562. 1 996.

84. Hempleman, S.C. and J.M. Hughes. Estimating exercise DL02 and diffusion limitation in patients with interstitial fibrosis. Respiration Physiology. 83:167-178. 1991.

85. Hirose, N., D.A. Lynch, R.M. Chemiack and D.E. Doherty. Correlation between high resolution computed tomography and tissue morphometry of the lung in bleomycin-induced pulmonary fibrosis in the rabbit. Amencan Review of Respiratory Diseases. 147:730-738. 1993.

86. Hiwatari, N., S. Shimura, K. Yamauchi, M. Nara, W. Hida and K. Shirato. Significance of elevated procollagen-Ill-peptide and transforming growth factor-beta levels of bronchoalveolar lavage fIuids from idiopathic pulmonary fibrosis patients. Tohoku Journal of Experimental Medicine. 1 81 :285-295. 1 997.

87. Hogg, J-C. Neutrophil kinetics and lung injury. Physiological Reviews. 67:1249-1295. 1987.

88. Hogg, J.C. Chronic interstitial lung disease of unknown cause: a new classification based on pathogenesis. American Journal of Roentgenology. 156:225-233. 1 991.

89. Hogg, J-C., H.O. Coxson, M.-L. Brumwell. N. Beyers, C.M. Doerschuk and W. MacNee. Erythrocyte and polymorphonuclear cell transit time and concentration in human pulmonary capillaries. Journal of Applied Physiobgy. 77: 1 795-1 800. 1 994.

90. Hogg, J.C., P.T. Macklem and W.M. Thurlbeck. Site and nature of airway obstruction in chronic obstructive lung disease. The New England Journal of Medicine. 2783 355-1 360. 1968.

91. Hogg, J.C. and S. Nepszy. Regional lung volume and pleural pressure gradient estimated from lung density in dogs. Journal of Applied Physiology. 27:198-203. 1969.

92. Hogg, J.C., S.J. Nepszy, P.T. Macklem and W.M. Thuribeck. Elastic properties of the centrîlobular ernphysematous space. Joumal of Clinical Investigation. 48:1306-1312. 1 969.

93. Hogg, J.C., J.L. Wright, B.R. Wiggs, H.O. Coxson, A.O. Saez and P.D. Par& Lung structure and function in cigarette srnokers. 'Ihorax. 49:473-478. 1 994.

94. Holland, B.S. and M.D. Copenhauer. An improved sequentially rejective Bonferroni test procedure. Biometncs. 43~417-423. 1 987.

95. Homma, S., 1. Nagaoka, H. Abe, K. Takahashi, K. Seyarna, T. Nukiwa and S. Kira. Localkation of platelet-derived growth factor and insulin-like growth factor I in the fibrotic lung. Arnerican Journal of Respiratory & Critical Care Medicine. 152:2084-2089. 1995.

96. Homolka, J. Idiopathic pulmonary fibrosis: a historical review. Canadian Medical Association Joumal. 1 37: 1 003- 1 005. 1 987.

97. Hruban, RH., M.A. Meziane, E.A. Zerhouni, N.F. Khouri, E.K. Fishman, P.S. Wheeler, S.

Dumler and G.M. Hutchins. High resolub'on computed tomography of inflation-fixed lungs: pathologie-radiologic correlation of centrilobular emphysema. American Review of Respiratoryûiseases. 136:935-940. 1 987.

98. Hughes, J.M., D.N. Lockwood, H.A. Jones and R.J. Clark. DLCO/Q and diffusion limitation at rest and on exercise in patients with interstitial fibrosis [published erratum appears in Respir Physiol 1 991 Ju1;85(1):137]. Respiration Physiollogy. 83~155-166. 1 991 .

99. Hunninghake, G.W., J.E. Gadek. T.J. Lawley and R.G. Crystal. Mechanisms of neutrophil accumulation in the lungs of patients with idiopathic pulmonary fibrosis. Journal of Clinical Investigation. 68:259-269. 1 981.

100. Hunt, L.W., T.V. Colby, D.A. Weiler, S. Sur and J.H. Butterfield. lmmunofluorescent staining for mast cells in idiopathic pulmonary fibrosis: quantification and evidence for extracellular release of mast cell tryptase. Mayo Clinic Proceedings. 67:941-948. 1 992.

101. Hyde, D.M., T.E. King, Jr., T. McDermott, J.A. Waldron, Jr., T.V. Colby, W.M. Thurlbeck, A. flint, L. Ackerson and R.M. Chemiack. ldiopathic Pulmonary Fibrosis. Quantitative assessment of lung pathology. Cornparison of a semiquantitative and a morphornetric histopathologic sco ring system . . American Revie w of Respiratory Diseases. 146: 1 042- 1047. 1992.

102. Iwai, K., T. Mon, N. Yamada, M. Yamaguchi and Y. Hosoda. ldiopathic pulmonary fibrosis. Epiderniologic approaches to occupational exposure. American Journal of Respiratory & Crifical Care Medicine. 150:670-675. 1994.

103. Janoff, A. Elastases and emphysema. Current assessment of the protease- antiprotease hypothesis. American Review of Respiratory Diseases. l32:417433. 1 985.

104. Javaheri, S. and L. Sicilian. Lung function, breathing pattern, and gas exchange in interstitial lung disease. Thorax. 47:93-97. 1992.

105. Junqueira, L.C., J. Carneiro, and R.O. Kelley. Respiratory System. In Basic Histology, 6th ed. Appleton & Lange, Norwalk pp.334-353. 1 989.

106. Junqueira, L.C., J. Cameiro, and R.O. Kelley. Blood Cells. In Basic Histology, 6th ed. Appleton & Lange, Nowalk. pp.229-244. 1989.

107. Kalender, W.A., H. Fichte, W. Bauh and M. Skalej. Semiautomatic evaluation procedures for quantitative CT of the lung. Journal of Cornputer Assisted Tomography. l5:248-255. 1991.

108. Kalender, W.A., R. Rienmuller, W. Seissler, J. Behr, M, Welke and H. Fichte. Measurement of pulmonary parenchymal attenuation use of spirometric gating with quantitative CT. Radiology. t 75:265-268. 1 990.

109. Kaneko, K., J. Milic-Emili, M.B. Dolovich, A. Dawson and D.V. Bates. Regional distribution and perfusion as a function of body position. Journal of Applied Physiology. 21 :767-m. 1966.

110. Kapanci, Y., A. Assimacopoulos, C. Ide, A. Zwahlen and G. Gabbiani. 'Contractile interstitial cellsn in pulmonary alveolar septa: a possible regulator of ventilationlperfusion ratio? . Joumal of Cell Biology. 60:375-392. 1 974.

11 1. Kapanci, Y., A. Desrnouliere, J-C. Pache, M. Redard and G. Gabbiani. Cytoskeletal protein modulation in pulmonary alveolar myofibroblasts during idiopathic pulmonary fibrosis. Possible role of transfoming growth factor beta and tumor necrosis factor alpha. American Joumal of Respira tory & Cntical Care Medicine. 1 52:2 1 63-2 1 69. 1 995.

1 1 2. Kapanci, Y., C. Ribaux, C. Chaponnier and G. Gabbiani. Cytoskeletal features of alveolar myofibroblasts and pericytes in nomal human and rat lung. Joumal of Histochemistry and Cytochemistry. 40:1955-1963. 1992.

1 13. Katzenstein, A.-L.A. and F.B. Askin. Idiopathic Interstitial Pneumonia/ldiopathic Pulmonary Fibrosis. In Surgical Pathology of Non-Naoplastic Lung Disease, 2nd ed. Saunders Co., pp.58-96.1990.

1 14. Kemerink, G.J., H.H. Kruize, R.J.S. Lamers and J.M.A. van Engelshoven. Density resolution in quantitative cornputed tomography of foam and lung. Mediml Physics. 23: 1 697- 1 708. 1 996.

115. Kemerink, G.J., R.J. Lamers, G.R. Thelissen and J.M. van Engelshoven. The nonlinear partial volume effect and computed tomography densitometry of foam and lung. Medical P ~ Y S ~ C S . 22: 1 445- 1 450. 1 995.

11 6. Kemerink, G.J., R.J. Larners, G.R. Thelissen and J.M. van Engelshoven. CT densitometry of the lungs: scanner performance. Joumal of Cornputer Assisted Tomography. 20:24-33. 1996.

1 17. Kemerhk, G.J., R.J.S. Lamers, G.R.P. Thelissen and J.M.A. van Engelshoven. Scanner conformity in CT densitornetry of the lungs. Radioiogy. 197:749-752. 1995.

1 18. Keogh, B.A., E. Lakatos, D. Price and R.G. Crystal. Importance of the lower respiratory tract in oxygen transfer. Exercise testing in patients with interstitial and destructive lung disease. American Revie w of Respira tory Diseases. 1 29: S76-S80. 1 984.

1 19. Khalil, N., O. Bereznay, M. Sporn and A.H. Greenberg. Macrophage production of transforming growth factor43 and fibroblast collagen synthesis in chronic pulmonary inflammation. The Joumal of Expenmen taMedicine. 1 70:7Z7-737. 1 989.

120. Kinsella, M., N.L. Müller, R.T. Abboud, N.J. Morrison and A. DyBuncio. Quantitation of emphysema by computed tomography using a 'density mask' program and correlation with pulmonary function tests. Chest. 97:315-321. 1990.

121. Kline, J.N., D.A. Schwartz, M.M. Monick, C.S. Floerchinger and G.W. Hunninghake. Relative release of interleukin-1 beta and interîeukin-1 receptor antagonist by alveolar macrophages. A study in asbestos-induced lung disease, sarcoidosis, and idiopathic pulmonary fibrosis. Chest. 1 O4:47-S3. 1993.

122. Knecht, A., L.G. Fine, K.S. Kleinman, H.P. Rodemann, G.A. Muller, D.D.L. Woo and J.T.

Norman. Fibroblasts of rabbit kidney in culture. II. American Journal of Physiology. 261:F292-299. 1991.

123. Kohda, E. and N. Shigematsu. Measurement of lung density by computed tomography: implication for radiotherapy. Keio Journal of Medicine. 38:454-463. 1 989.

124. Kuhn, C. Patterns of lung Repair. A morphologist's view. Chest. 99:11 S-14s. 1991.

125. Kuhn, C. and J.A. McDonald. The rotes of the myofibroblast in idiopathic pulmonary fibrosis. Amencan Joumal of Pathology. 138: 1257-1 265. 1 991 .

126. Kuhn, C., III, J. Boldt, T.E. King, Jr., E. Crouch, T. Vartio and J.A. McDonald. An imrnunohistochemical study of architectural remodeling and connective tissue synthesis in pulmonary fibrosis. American Review of Respimtory Diseases. l40:1693-1703. 1 989.

127. Kuwano, K., K. Matsuba, T. Ikeda, J. Murakami, A. Araki, H. Nishitani, T. Ishida, K. Yasumoto and N. Shigematsu. The diagnosis of mild emphysema. Correlation of computed tomography and pathology scores. Amen'can Review of Respiratory Diseases. 141 11 69-1 78- 1997.

128. Lang, M.R., G.W. Fiaux, M. Gillooly, J.A. Stewart, D.J. Hulmes and D. Lamb. Collagen content of alveolar wall tissue in ernphysematous and non-emphysematous lungs. Thorax. 49:3 1 9-326. 1 994.

129. Lang, M.R., G.W. Fiaux, D.J. Hulmes, D. Lamb and A. Miller. Quantitative studies of human lung airspace wall in relation to collagen and elastin content. Matrix. 13:471-480. 1993.

130. Last, J.A. Biochemical alterations of lung structure as predictors of chronic lung disease. Environmental Health Perspectives. 52:159-163. 1 983.

131. Laurell, C.B. and S. Ericksson. The electrophoretic alpha1 -globulin patterns of serum in alpha1 -antitrypçin deficiency . Scandinavian Joumal of Clinical hhratoty Investigations. 15:132-140. 1963.

132. Laurent, G.J. Lung collagen: more than scaffolding. Thorax. 41 :418-428. 1 986.

133. Lee, J.S., J.G. lm, J.M. Ahn, Y.M. Kim and M.C. Han. Fibrosing alveolitis: prognostic implication of ground-glass attenuation at high-resolution CT. Radiology. 184:451454. 1992.

134. Lehnert, S. and E. El-Khatib. The use of CT densitometry in the assessrnent of radiation- induced damage to the rat lung: a comparison with other endpoints. International Journal of Radiation Omlogy Biology Physics. 1 6: 1 1 7- 1 24. 1 989.

135. Leopdd, J.G. and J. Gough. The centrilobular fom of hypertrophie emphysema and its relation to chronic bronchitis. Thorax. 12:219-235. 1 957.

136. Leslie, K.O., J. Mitchell and R. Low. Lung myofibroblasts. Ce11 Motility & the

Cytoskeleton. 22:92-98. 1 992.

137. Lesur, O., A.M. Bernard and R.O. Begin. Clara cell protein (CC-1 6) and surfactant- associated protein A (SP-A) in asbestos-exposed workers. Chest. 109:467-474. 1996.

138. Levi, C., J.E. Gray, E.C. McCullough and R.R. Hattery. The unreliability of CT numbers as absolute values. American Joumal of Roentgenology. 139:443-447. 1 982.

139. Linsenmayer, T.F. Collagen. ln Cell Biology of Extracellular Matrix, 2nd ed. Plenum Press, New York. pp.7-43. 1991.

140. Lynch, D.A., J.D. Newell, P.M. Logan, T.E. King, Jr. and N.L. Müller. Can CT distinguish hypersensitivity pneumonitis from idiopathic pulmonary fibrosis?. A m e m n Journal of Roentgenology. l65:8O7-8ll. 1 995.

141. Lynch, J.P., T.J. Standiford, M.W. Rolfe, S.L. Kunkel and R.M. Strieter. Neutrophilic alveolitis in idiopathic pulmonary fibrosis. The role of interleukin-8. American Review of Respiratory Diseases. 145:1433-1439. 1 992.

142. MacNee, W., B. W~ggs, A. Belzberg and J.C. Hogg. The effect of cigarette smoking on neurophil kinetics in human lungs. New England Journal of Medicine. 321 :324-328. 1989.

143. Madri, J.A. and H. Furthmayr. Collagen polymorphism in the lung. Human Pathology. 1 1 ~353-366. 1980.

144. Mayhew, T.M. The new stereological methods for i~terpreting functional morphology from sliceç of cells and organs. Experimental Physiology. 76:639-665. 1991.

145. McCormack, F.X.. T.E. King, Jr., D.R. Voelker, P.C. Robinson and R.J. Mason. ldiopathic pulmonary fibrosis. Abnomalities in the bronchoalveolar lavage content of surfactant protein A. Amencan Review of Respiratory Diseases. 1443 6û-166. 1991.

146. McCuaig, K.E., S. Vessal, C. Coppin, BR. Wiggs, R. Dahlby and P.D. Par& Variability in measurernents of pressure volume curves in normal subjects. American Review of Respimtory Diseases. 131 :656-658. 1 985.

147. McCulloch, C.A.G. and G.C. Knowles. Deficiencies in collagen phagocytosis by human fibroblasts in vitro: a mechanism for fibrosis?. Journal of Cellular Physiolog~t. 155:461-471. 1993.

148. McCullough, E.C. and R.L. Morin. CT-nurnber variability in thoracic geometry. Amencan Journal of Roentgenology. 141 :135- 140. 1 983.

149. McDonald, J.A., T.J. Broekelmann, M.-L. Matheke, E. Crouch, M. K m and C. Kuhn, III. A monoclonal antibody to the carboxyterminal domain of procollagen type I visualizes callagen-synthesizing fibroblasts. Journal of Clhical Investigations. 78:1237-1244. 1 986.

150. McGowan. S.E.. R. Liu and C.S. Harvey. Effects of heparin and other glycosaminoglycans on elastin production by cultured neonatal rat lung fibroblasts. Archives of Biochemistry & Biophysics. 302322-331. 1 993.

151. Mead, J. Mechanics of Lung and Chest Wall. In Respiratory Physiology People and Ideas, Oxford University Press, New York. pp.173-207. 1 996.

152. Mecham, R.P. and J.E. Heuser. The elastic fiber. In Cell Biology of Extracellular Matrix, 2nd ed. Plenum Press, New York. pp.79-109. 1991.

153. Michel, R.P. and L.M. Cruz-Orive. Application of the Cavalieri principle and vertical sections method to lung: estimation of volume and pleural surface area. Journal of Microscopy. 150:117-136. 1988.

154. Milic-Emili, J., J.A.M. Henderson, M.B. Dolovich, D. Trop and K. Kaneko. Regional distribution of inspired gas in the lung. Journal of Applied Physiology. 21 :749-759. 1966.

155. Millar, A.B. and D.M. Denison. Vertical gradients of lung density in supine subjects with fibrosing alveolitis or putmonary emphysema. Thorax. 45:602-605. 1990.

156. Millar, AB. , B. Fromson, B.A. Strickfand and D.M. Denison. Computed tomography based estimates of regional gas and tissue volume of the lung in supine subjects with chronic airflow limitation or fibrosing alveolitis. Thorau. 41:932-939. 1 986.

157. Miller, A., J.C. Thornton, R. Warshaw, H. Anderson, A S . Teirstein and I.J. Selikoff. Single breath diffusing capacrty in a representative sample of the population of Michigan, a large industrial state. Predicted values, lower limits of normal, and frequencies of abnormality by smoking history. American Revie w of Respiratory Diseases. l27:27O-S77. 1 983.

158. Miller, R.R., N.L. Müller, S. Vedal, N.J. Morrison and C.A. Staples. Limitations of computed tomography in the assessrnent of emphysema. American Review of Respiratov Diseases. l39:980-983. 1 989.

159. Miller W.S. The Lung, Thomas Books, Springfield. 1937.

160. Mio, T., S. Nagai, M. Kitaichi, A. Kawatani and T. Izumi. Proliferative characteristics of fibroblast lines derived from open lung biopsy specimens of Patients with IPF (UIP). Chest. 1 O2:832-837. 1 992.

161. Montano, M., C. Ramos, G. Gonzalez, F. Vadillo, A. Pardo and M. Selman. Lung collagenase inhibitors and spontaneous and latent collagenase activity in idiopathic pulmonary fibrosis and hypersensitivity pneumonitis. Chest. 96:1115-1119. 1 989.

162. Morris, J.F., A. Koski and L.C. Johnson. Spirometnc standards for healthy nonsmoking adults. AmenCan Review of Respiratory Diseases. 1 O3:5%67. 1 97 1 .

163. Murdoch, A.D. and R.V. lozzo. Periecan: the multidomain heparan sulphate proteoglycan of basement membrane and extracellular matrix. Virchows Archiv A.Pathologica1 Anatomy and Histology. 423:237-242. 1 993.

164. Müller, N.L. Computed tomography in chronic interstitial lung disease. Radiologie Clinies of North Amen'ca. 29:1085-1094. 1991.

165. Müller, N.L. and R.R. Miller. Cornputed tomography of Chronic diffuse infiltrative lung disease. Part 1. Arnerican Review of Respiratory Diseases. f42:12O6-1215. 1 990.

166. Müller, N.L., C.A. Staples, R.R. Miller and R.T. Abboud. 'Density mask'. An objective method to quantitate emphysema using computed tomography. Chest. 94:782-787. 1988.

167. Müller, N.L., C.A. Staples, R.R. Miller, S. Vedal, W.M. Thurlbeck and D.N. Ostrow. Disease activity in idiopathic pulmonary fibrosis: CT and pathologic correlation. Radiology. l65:?31-734. 1 987.

168. Nagai, A. and W.M. Thurlbeck. Scanning electron microscopie observations of emphysema in humans. American Review of Respiratory Diseases. 144:9011991.

169. Nagaoka, I., B.C. Trapnell and R.G. Crystal. Upregulation of platelet-derived growth factor-A and -B gene expression in alveolar macrophages of individuals with idiopathic pulmonary fibrosis. Journal of Clinical Investigation. 85:2023-2027. 1990.

170. Nakamura, H., S. Fujishima, Y. Waki, T. Urano, K. Sayama, F. Sakamaki, K. Soejima, S. Tasaka, A. lshizaka and et al. Priming of alveolar macrophages for interieukin-8 production in patients with idiopathic pulmonary fibrosis. American Journal of Respiratory & Critical Care Medicine. 152: 1 579-1 586. 1 995.

171. Nakao, A., Y. Hasegawa, Y. Tsuchiya and K. Shimokata. Expression of ce11 adhesion molecules in the lungs of patients with idiopathic pulmonary fibrosis. Chest. 108:233-239. 1995.

172. Nakos, G., A. Adams and N. Andnopoulos. Antibodies to collagen in patients with idiopathic pulmonary fibrosis. Chest. 103:1051-1058. 1993.

173. Nicklaus, T.M., D.W. Stowell, W.R. Christiansen and A.D. Renzetti. The accuracy of the roentgenologic diagnosis of chronic pulmonary emphysema. American Review of Respiratory Diseases. 93:889-899. 1 966.

174. Niewoehner, D.E. and J.R. Hoidal. Lung fibrosis and emphysema: divergent responses to a common injury?. Science. 21 7:359-3ôû. 1 982.

175. Nishimura, K., M. Kitaichi, T. Izumi, S. Nagai, M. Kanaoka and H. Itoh. Usual interstitial pneumonia: histologie correlation with high-resolution CT. Radiology. 182:337-342. 1992.

176. Nugent, K.M., M.W. Peterson, H. Jolles, M.M. Monick and G.W. Hunninghake. Correlation of chest roentgenograms with pulmonary function and bronchoalveolar lavage in interstitial lung disease. Chest. 96:1224-1227. 1989.

177. Ohta, K., R.L. Mortenson, R.A. Clark, N. Hirose and T.E. King, Jr. lmrnunohistochemical identification and characterization of smooth muscle-like cells in idiopathic pulmonary fibrosis. American Journal of Respiratory & Criticai Care Medicine. 1 52:1659-1665. 1 995.

178. Ohtani, H., A. Kuroiwa, M. Obinata, A. Ooshima and H. Nagura. Identification of type I collagen-producing cells in human gastrointestinal carcinomas by non-radioactive in situ hybridization and immunoelectron microscopy. The Journal of Histochemistry an

179. Oldrnixon, E.H. and F.G. Hoppin. Jr. Distribution of elastin and collagen in canine lung alveolar parenchyma . Journal of Applied Physiology. 67:1941-1949. 1989.

180. Orens, J.B., €.A. Kazerooni, F.J. Martinez, J.L. Curüs, B.H. Gross, A. Flint and J.P. Lynch. The sensitivity of high-resolution CT in detecting idiopathic pulmonary fibrosis proved by open lung biopsy. A prospective study. Chest. lO8:lO9-ll5. 1995.

181. Osborne, S., J.C. Hogg, J.L. Wright, C. Coppin and P.D. Paré. Exponential analysis of the pressure-volume curve. Correlation with mean Iinear intercept and emphysema in human lungs. Amencan Review of Respiratory Diseases. 137: 1 083-1 088. 1 988.

182. Ozaki, T., H. Hayashi, K. Tani, F. Ogushi, S. Yasuoka and T. Ogura. Neutrophil chemotactic factors in the respiratory tract of patients with chronic airway diseases or idiopathic pulmonary fibrosis. American Review of Respiratory Diseases. l45:85-91, 1 992.

183. Pakkenberg, B., J. Boesen, M. Albeck and F. Gjerris. Unbiased and efficient estimation of total ventricular volume of the brain obtained from CT-scans by a stereological method. Neuroradiology. 31 :4 1 3-41 7. 1 989.

184. Panos, R.J., R.L. Mortenson, S.A. Niccoli and T.E. King, Jr. Clinical deterioration in patients with idiopathic pulmonary fibrosis: causes and assessment. American journal of Medicine. 88:396-404. 1 990.

185. Pardo, A., M. Selman, R. Ramirez, C. Ramos, M. Montano, O. Stricklin and G. Raghu. Production of collagenase and tissue inhibitor of metalloproteinases by fibroblasts derived from normal and fibrctic human lungs. Chest. 102:1085-1089. 1 992.

186. Peterson, M W . , M. Monick and G.W. Hunninghake. Prognostic role of eosinophils in pulmonary fibrosis. Chest. 92:W -56. 1987.

187. Peyrol, S., J.-F. Cordier and J.-A. Grimaud. Intra-alveolar fibrosis of idipathic bronchiolitis obliterans-organizing pneumonia. American Journal of Pathology. 137:155-170. 1 990.

188. Pierce, R.J., D.J. Brown and D.M. Denison. Radiographic, scintigraphie, and gasdilution estimates of individual lung and lobar volumes in man. Thom. 35:773-780. 1980.

189. Quaglino, D., C. Fornieri, 1.8. Nanney and J.M. Davidçon. Extracellular matrix modifications in rat tissues of different ages. Correlations between elastin and collagen type I m RNA expression and lysyl-oxidase activity. Matrix- 13:481-4QO. 1 993.

190. Raghow, R. Role of transfonning growth factor43 in repair and fibrosis. Chest. 99361 S- 64s. 1991.

191. Raghu, G.. L.J. Striker, L.D. Hudson and G.E. Striker. Extracellular matrix in normal and fibrotic human lungs. American Review of Respiratory Diseases. 131 :281-289. 1 985.

192. Reid, L. and F.J.C. Millard. Correlation between radiological diagnosis and structural lung changes in emphysema. Cliniwl Radiology. l5:30%3ll. 1 964.

193. Remy-Jardin, M., F. Giraud, J. Remy, M.C. Copin, B. Gosselin and A. Duhamel. Importance of ground-gIass attenuation in chronic diffuse infiltrative lung disease: pathologic-CT correlation. Radiology. 189:693-698. 1 993.

194. Reynolds, H.Y., J.D. Fulmer, J.A. Kamierowski, W.C. Roberts, M.M. Frank and R.G. Crystal. Analysis of cellular and protein content of broncho-alveolar lavage fluid from patients with idiopathic pulmonary fibrosis and chronic hypersensitivity pneumonitis. Joumal of Clinical Investigation. 59: 1 65-1 75. 1 977.

195. Rienmuller, R.K., J. Behr, W.A. Kalender, M. Schatzl, 1. Altmann, M. Merin and T. Beinert. Standardized quantitative high resolution CT in lung diseases. Journal of Computer Assisted Tomography. 1 5:742-749. 1 99 1 .

196. Roberts, A.B., G.K. McCune and M.B. Sporn. TGF-6: regulation of extracellular matrix. Kidney lntemational. 4 1 :557-559. 1 992,

1 97. Robins, A.G. Pathophysiology of emphysema. Clinics in Chest Medicine. 4:413-420. 1983.

198. Robinson, P.J. and L. Kreel. Pulmonary tissue attenuation with computed tomography: Comparison if Inspiration and expiration scans.. Journal of Computer Assisted Tomography. 3:740-748. 1 979.

199. Ross, R. Atherosclerosis: a defense mechanism gone awry. American Journal of Pathology. 143:98%IOOS. 1 993.

200. Roughley, P.J. and E.R. Lee. Cartilage proteoglycans: structure and potential functions. Microscopy Research and Technique. 28:385-397. 1 994.

201 Saldiva, P.H., V.C. Delmonte, C.R. de Carvalho, R.A. Kairalla and J.O. AuIer Junior. Histochemical evaluation of lung collagen content in acute and chronic interstitial diseases. Chest. 95953-957. 1989.

202. Sanders, C. The radiographie diagnosis of emphysema. Radiologie Clinics of North Amerka. 29:1019-1030. 1991.

203. Schaberg, T., M. Rau, H. Stephan and H. Lode. lncreased nurnber of alveolar macrophages expressing surface molecules of the CD1 IlCD1 8 family in sarcoidosis and idiopathic pulmonary fibrosis is related to the production of superoxide anions by these cells. American Review of Respiratory Diseases. 147: 1 507-1 51 3. 1 993.

204. Schwartz. D.A.. J.R. Galvin. S.J. Yagla, S.B. Speakman, J.A. Merchant and G.W. Hunninghake. Restrictive lung function and asbestos-induced pleural fibrosis. Journal of Clinical Investigations. 91 :26t35-2692. 1 993.

205. Schwartz, D.A., R.A. Helmers, C.S. Dayton. R.K. Merchant and G.W. Hunninghake. Deteminants of bronchoalveolar lavage cellulanty in idiopathic pulmonary fibrosis. Journal of Applied Physiology. 71 : 1 688- 1 693. 1 99 1 .

206. Schwartz, D.A., R.A. Helmers, J.R. Galvin, D.S. Van Fossen, K . 1 Frees, C.S. Dayton and G.W. Hunninghake. Deterrninants of survival in idiopathic pulmonary fibrosis. American Journal of Respiratory & Critical Care Medicine. 1 49:450-454. 1 994.

207. Schwartz, D.A., R.K. Merchant, RA. Helmers, S.R. Gilbert and C.S. Dayton. The influence of cigarette smoking on lung fundion in patients with idiopathic pulmonary fibrosis. American Review of Respiratory Diseases. 1 44:504-506. 1 99 1 .

208. Schwartz, D.A., D.S. Van Fossen. C.S. Davis, R.A. Helmers, C.S. Dayton, BurmeisterLF. and G.W. Hunninghake. Determinants of progression in idiopathic pulmonary fibrosis. American Joumal of Respiratory & Cntical Care Medicine. 149:444-449. 1 994.

209. Schwarz, M.I. Treatment of 1PF. What does the future hold?. Chest. 108:601-602. 1995.

210. Sciurba, F.C., R.M. Rogers, R.J. Keenan, W.A. Slivka, J. Gorcsan III, P.F. Ferson, J.M. Holbert, M.L. Brown and R.J. Landreneau. lmprovement in pulmonary function and elastic recoil after lung-reduction surgery for difuse ernphysema. The New England Journal of Medicine. 334: 1 095-1 099. 1 996.

21 1. Shapiro, S.D., S.K. Endicott, M.A. Province, J.A. Pierce and E.J. Campbell. Marked longevity of human lung parenchyrnal elastic fibers deduced from prevalence of D-aspartate and nuclear weapons-related radiocarbon. Joumal of Clinical Investigations. 87:1828-1834. 1991.

21 2. Shama, K. and EN. Ziyadeh. The transforming growth factor4 systern and the kidney. Seminars in Nephrology. 1 3: 1 1 6- 1 28. 1 993.

21 3. Shaw, R.J., S.H. Benedict, R.A.F. Clark and T.E. King, Jr. Pathogenesis of pulmonary fibrosis in interstitial lung disease. American Review of Respiratory Diseases. 143:167-173. 1991.

21 4. Shay, J. Economy of effort in electron microscope morphometry. American Journal of Pathology. 81 503-51 2. 1 975.

21 5. Sims, D.E. The pericyte--a review. Tissue & CeIl. 18:153-174. 1986.

21 6. Smeulders, A.W.M. and L. Dorst. Measurement issues in morphometry. Analflical and Quantitative Cytoogy and Histology. 7:242-249. 1 985.

21 7. Snider, G.L. Interstitial pulmonary fibrosis. Chest. 89:115S-121 S. 1 986.

21 8. Snider, G.L., D.E. Ciccolella, S.M. Morris, P.J. Stone and E.C. Lucey. Putative role of neutrophil elastase in the pathogenesis of emphysema. Annals of the New York Academy of Sciences. 624:45-59. 1 99 1 .

21 9. Snider, G.L., J. Kleinenan, W.M. Thurlbeck and Z.H. Bengali. The definition of emphysema. Report of a National Heart, Lung and Wood Institute, Division of Lung Diseases Workshop. American Review of Respiratory Diseases. 132: 1 82-1 85. 1 985.

220. Snow, A.D., R.P. Bolender, T.N. Wight and A.W. Clowes. Heparin modulates the

composition of the extracellular matrix domain surrounding arterial smooth muscles cells. American Journal of Pathology. 1 37:313-330. 1 990.

221 . Standiford, T.J., M.W. Rolfe, S.L. Kun kel, J.P. Lynch, M.D. Burdick, A.R.O. Gilbert, R.I. Whyte and R.M. Strieter. Macrophage inflammatory protein- 1 alpha expression in interstitial lu ng disease. Journal of lmmunology. 1 51 :2852-2863. 1 993.

222. Stern, E.J. and M.S. Frank. CT of the lung in patients with pulmonary emphysema: diagnosis, quantification, and correlation wiai pathologic and physiologic findings. Amerkan Journal of Roentgenology. 162:791-798. 1 994.

223. Striz, I ., L. Zheng, Y.M. Wang, H. Pokoma, P.C. Bauer and U. Costabel. Soluble CD1 4 is increased in bronchoalveolar lavage of active sarcoidosis and correlates with alveolar macrophage membrane-bound CD1 4. American Journal of Respiratory & Critical Care Medicine. 1 51 :544-547. 1 995.

224. Sweet, H.C., J.P. Wyatt and P.W. Kinsella. Correlation of lung macrosections with pulmonary function in emphyserna. The Amekan Journal of Medicine. 29:277-281. 1960.

225. Terriff, B.A., S.Y. Kwan, M.M. Chan-Yeung and N.L. Müller. Fibrosing alveolitis: chest radiography and CT as predictors of clinical and functional impairment at follow-up in 26 patients. Radiology. l84:445-449. 1 992.

226. Thurlbeck, W.M. Overview of the pathology of pulmonary emphysema in the human. Clinics in Chest Medicine. 4:337-350. 1 983.

227. Thurlbeck, W.M., M.S. Dunnill, W. Hartung, B.E. Heard, A.G. Heppleston and R.C. Ryder. A comparison of three methods of measuring emphyserna. Human Pathdogy. 1 :16W 78. 1 970.

228. Thurlbeck, W.M., J.A. Henderson, R.G. Fraser and D.V. Bates. Chronic obstructive lung disease. A comparison between clinicat, roentgenologic, functional and morphologie criteria in chronic bronchitis, ernphysema, asthma and bronchiectasis. Medicine. 49:8l-l45. 1970.

229. Thurlbeck, W.M. and N.L. Müller. Emphysema: definition, imaging, and quantification. American Journal of Roentgenology. 1 63: 1 01 7-1 025. 1 994.

230. Thuribeck, W.M. and G. Simon. Radiographie appearance of the chest in emphysema. American Joumal of Roentgenology. 130:429-440. 1 978.

231. Tung, K.T., A.U. Wells, M.B. Rubens, J.M. Kirk, R.M. du Bois and D.M. Hansell. Accuracy of the typical computed tomographie appearances of fibrosing alveolitis [see comments]. Thorax. 48~334-338. 1993.

232. Turner-Waiwick, M. and P. Haslam. Antibodies in some chronic fibrosising lung disease. 1. Non-organ specific autoantibodies. Clinical Allergy. 1 :83-95. 1 971.

233. UnderWood E.E. Quantitative Stereology, Addison-Wesley, Reading, Mass. 1997.

234. Vaccaro, C.A. and J.S. Brody. Ultrastructural localization and characterization of

proteoglycans in the pulmonary alveolus. Ameiican Review of Respiratory Diseases. 12O:9Ol-91 O. 1 979.

235. Vaccaro, C.A. and J.S. Brody. Structural features of alveolar wall basement membrane in the adult rat lung . Journal of Cell Biology. 91 :427-437. 1 981 .

236. van der Rest, M. and R. Gamone. Collagen family of proteins. FASEB Journal. 5:2814- 2823. 1991.

237. Van Ganse, W.F., B.G. Ferris and J.E. Cotes. Cigarette smoking and pulmonary diffusing capacity (transfer factor). American Review of Respira tory Diseases. 1 OS:30-41. 1 972.

238. Vedal, S., E.V. Welsh, R.R. Miller and N.L. Muller. Desquamative interstitial pneumonia. Computed tomographic findings before and after treatment with corticosteroids. Chest. 93:215-217. 1988.

239. Verbeken, E.K., M. Cauberghs, 1. Mertens, J. Clement, J.M. Lauweryns and K.P. Van de Woestijne. The senile lung. Comparison with normal and emphysematous lungs. 1. Structural aspects. Chest. 101 :793-799. 1 992,

240. Wagner, P.D., D.R. Dantzker, R. Dueck and et al. Distribution of ventilation-perfusion ratios in patients with interstitial lung disease. Chest. 69 (suppl):256-257. 1 976.

241. Wallace, W.A.H., M. Gillooly and 0. Lamb. Intra-alveoloar macrophage num bers in current smokers and non-smokers: a morphometric study of tissue sections. Thorax. 47:437-440. 1 992.

242. Wandtke, J.C., R.W. Hyde, P.J. Fahey, M.J. Utell, D.B. Plewes, M.J. Goske and H.W. Fischer. Measurement of lung gas volume and regional density by computed tomography in dogs. Investigative Radiology. 21 :108-117. 1 986.

243. Waters, L.C., T.E. King, R.M. Chemiack, J.A. Waldron, R.E. Stanford, M.L. Willcox and M.I. Schwarz. Bronchoalveolar lavage fluid neutrophils increase after corticosteroid therapy in smokers with idiopathic pulmonary fibrosis. American Review of Respiratory Diseases. l33:lO4-lO9. 1 986.

244. Watters, L.C., T.E. King, M.I. Schwarz, J.A. Waldron, RE. Stanford and R.M. Chemiack. A clinical, radiographic, and physiologic scoring system for the longitudinal assessment of patients with idiopathic pulmonary fibrosis. American Review of Respiratory Diseases. l33:97-103. 1 986.

245. Watters, L.C., M.I. Schwarz, R.M. Chemiack, J.A. Waldron, T.L. Dunn and R.E. Stanford. ldiopathic pulmonary fibrosis. Pretreatrnent bronchoalveolar lavage cellular constituents and their relationships with lung histopathology and clinical response to therapy. Amencan Review of Respira tory Diseases. 1 35:696-704. 1 987.

246. Webb, W. R. High-resolution corn puted tomography of obstructive lung disease. Radiologie Clinics of North America. 32745-757. 1 994.

247. Webb, W.R., N.L. Müller and D.P. Naidich. Standardized terms for high-resolution

cornputed tomography of the lung: A proposed glossary. Journal of Thoracic Imaging. 8~167-175. 1993.

248. Wegener, O.H., P. Koeppe and H. Oeser. Measurement of Lung Density by Computed Tomography. Journal of Computer Assisted Tomography. 2:263-273. 1 978.

249. Weibel E.R. Morphometry of the human lung, Springer-Verlag, Berlin. 1963.

250. Weibet, E.R. Morphometric estimation of pulmonary diffusion capacity. V. Comparative morp homet ry of alveolar lungs. Respiration Physiology. 1 4:26-43. 1 972.

251. Weibel E.R. Stereological Method, Volume 2. Theoretical Foundations, Academic Press, London. 1980.

252. Weibel E R . The Pathway for Oxygen. Stucture and Function in the Mammalian Respiratory System., Harvard University Press, Cambridge. 1984.

253. Weibel, E.R. Functional rnorphology of lung parenchyma. In Handbook of Physiology Section 3:The Respiratory System Volume III, Part 1, American Physiological Society, Bethesda. pp.89-111. 1986.

254. Weibel, E.R. Measuring through the microscope: development and evolution of stereological rnethods. Journal ov Microscopy. 155:393-403. 1 989.

255. Weibel, E.R. Morphometry: Stereological Theory and Practical Methods. In Models of Lung Disease. Microscopy and Stuctural Methods, Marcel Dekker, Inc, New York. pp. 1 99-252. 1 990.

256. Weibel, ER. The structural basis of lung function. In Respiratory Physiology People and Ideas, Oxford University Press, New York. pp.3-46. 1996.

257. Weibel, ER., P. Gehr, L.M. Cruz-Orive, A.E. Müller, D.K. Mwangi and V. Haussener. Design of the mamrnalian respiratory system. IV. Morphometric estimation of pulmonary diffusing capacity; critical evaluation of a new sampling method. Respiration Physiology. 44~39-59. 1981.

258. Weissler, J.C. ldiopathic pulmonary fibrosis: cellular and molecular pathogenesis. Amencan Journal of the Medical Sciences. 297:91-104. 1 989.

259. Wells, A.U., A.D. King, M.B. Rubens, D. Cramer, R.M. du Bois and D.M. Hansell. Lone cryptogenic fibrosing ahotitis: a functional-rnorphologic correlation based on extent of disease on thin-section computed tomography. Arnerican Journal of Respiratory & Critical Care Medicine. 155:1367-1375. 1 997.

260. Wells, AU., M.B. Rubens, R.M. du Bois and D.M. Hansell. Serial CT in fibrosing alveolitis: prognostic significance of the initial pattern. Amencan Journal of RoentgenoIogy. 161:1159- 1165. 1993.

261. West J.B. Pulmonary Pathophysiology -the essentials, 3rd ed. Williams & Wilkens, Baltimore. 1 987.

262. West, J.B. Pulmonary Blood Flow and Gas Exchange. In Respiratory Physiology People and fdeas, Oxford University Press, New York. pp.140-172. 1996.

263. West, J.B. and C.T. Dollery. Distribution of blood flow and ventilation-perfusion ratio in the lung. measured with radioactive carbon dioxide. Journal of Applied Physiology. 15:405-410. 1 960.

264. West, J.B. and C.T. Dollery. Distribution of blood flow in isolated lung; relation to vascular and alveolar pressures. Joumal of Applied Physiology. 19:7l3-724. 1 964.

265. West, J.B. and C.T. Dollery. Distribution of blood flow and the pressure-flow relations of the whole lung. Joumal of Applied Physiology. 20:175-183. 1 965.

266. Wheater, P.R.. H.G. Burkitt, and V.G. Daniels. Respiratory System. In Functional Histology, 2nd ed. Churchill Livingstone, Edinburgh. pp.178-190. 1987.

267. Wiggins, J., B. Strickland and M. Tumer-Warwick. Combined crypotgenic fibrosing alveolitis and emphysema: the value of high resolution computed tomography in assessment. Respiratory Medicine. 84365-369. 1 990.

268. Wight. T.N., D.K. Heioegard. and V.C. Hascall. Proteoglycans. In Cell Biology 01 Extracellular Matrix, 2nd ed. Plenum Press, New York. pp.111-135. 1991.

269. Wilborn, J., M. Bailie, M. Coffey, M. Burdick, R. Strieter and M. Peters-Golden. Constitutive activation of 5-lipoxygenase in the lungs of patients with idiopathic pulmonary fi b rosis. Joumal of Clinical Investigation. 97: 1 827- 1 836. 1 996.

270. Wollmer, P., U. Albrechtsson, K. Brauer, L. Eriksson, B. Jonson and U. Tylèn. Measurement of Pulmonary density by means of X-ray computerized tomography. Chest. 90~387-391. t 986.

271. World Health Organization. Chronic cor pulmonale: report of an expert committee. Wodd Health Organization Technical Reporf Series. 21 3: 1 5 1 961 .

272. Wright, J.L. Emphyserna: concepts under change-a pathologist's perspective. Modem Pathology. 8:873-880. 1 995.

273. Wright, J.L. and A. Churg. Smoke-induced emphysema in guinea pigs is associated with morphometric evidence of collagen breakdown and repair. American Joumal of Physiology. 2681L17-L20. 1 995.

274. Yamada, K.M. Fibronectin and other cell interactive glycoproteins. In Cell Biology of Extracellular Matrix, 2nd ed. Plenum Press, New York. pp.lll-135. 1991.

275. Zapletal, A., J. Houstek, M. Samanek, M. Copova and T. Paul. Lung function in children and adolescents with idiopathic interstitial pulmonary fibrosis. Pediatric Pulmonology. 1:154-166. 1985.

276. Zerhouni, E.A., J.F. Spivey, R.H. Morgan, F.P. Leo, F.P. Stitik and S.S. Siegelman.

Factors influencing quantitative CT measurements of solitary pulmonary nodules. Journal of Cornputer Assisted Tomography. 6: 1 075- 1 087. 1 982.

27ï. Zhang, K., MD. Rekhter. D. Gordon and S.H. Phan. Myofibroblasts and their role in lung collagen gene expression during pulmonary fibrosis. A combined imrnunohistochemical and in situ hybridization study. American Journal of Pathology. 145: 1 14- 1 25. 1 994.

278. Zhang, Y., TC. Lee, B. Guillemin, M.C. Yu and W.N. Rom. Enhanced IL4 beta and tumor necrosis factor-alpha release and messenger RNA expression in macrophages from idiopathic pulmonary fibrosis or after asbestos exposure. Journal of Immunology. 15O:4188-4196. 1 993.

IMAGE NALUATION TEST TARGET (QA-3)

APPLIED - i LIMAGE. lric = 1653 East Main Street - -- - - Rochester. NY 14609 USA -- -- - - Phone: 71 6/482-O3OO -- -- - - Fax: 716/28&5989

O 1993. App(ied Image. lx.. All Rigrifs Resewed