Quantitative Assessment of Lung Pathology in Idiopathic Pulmonary Flbrosts"?
REUBEN M. CHERNIACK, THOMAS V. COLBY, ANDREW FLINT, WILLIAM M. THURLBECK, JAMES WALDRON, LYNN ACKERSON, TALMADGE E. KING, JR., and the BAL COOPERATIVE GROUP STEERING COMMITTEE'
Introduction
The diagnosis and classification of most interstitial lung diseases requires histologic evaluation of biopsy material. In the case of idiopathic pulmonary fibrosis (IPF), an open lung biopsy is necessary to confirm the diagnosis. The disease activity (i.e., the stage of the disease) is thought to be represented by the degree of cellularity present, and it is generally accepted that the prognosis and response to therapy is related to the relative amounts of cellular and fibrotic involvement of the lung parenchyma (1-3). In routine practice, a qualitative assessment of the relative extent and severity of these pathologic changes is usually provided by pathologists, so that correlation between pathologic changes and the clinical and physiologic alterations is at best crude. It is possible that a more rigorous quantitative assessment of pathologic parameters would provide a better correlation with clinical parameters and physiologic abnormalities, as well as the response to therapeutic intervention. As part of a multicenter study in which the role of bronchoalveolar lavage constituents in patients with IPF wereevaluated (4), a panel of four pathologists was established to quantify the pathologic changes present in IPF. The purpose of this report is to describe the pathologic scoring system that was developed, and the variables that wereassessed. In addition, the reproducibility (i,e., intraobserver variability) and the comparability (i.e., interobserver variability) of the readings of the four pathologists are presented. Methods All suspected cases of IPF from the multicenter study were reviewed and scored prospectively by the panel of four pathologists (T.V.C., A.F., W.M.T., and JYV.) who were unaware of the clinical or physiologic findings. Cases in which there were differing diagnoses werereevaluated by the panel at peri892
SUMMARY The diagnosis and classification of most Interstitial lung diseases requires histologic evaluation of lung tissue, obtained by an open lung biopsy to confirm the diagnosis. In addition, It Is generally accepted that response to therapy In Idiopathic pulmonary fibrosis (IPF) Is related to the relative degree of cellularity and fibrosis present. Because only a qualitative assessment of the relative extent and severity of these changes Is generally provided, correlation with clinical and physiologic alterations Is difficult. This report describes resulta of a semiquantitative assessment by four pathologists of Inflammatory/exudative changes, fibrotic/reparative changes, and airway alterations, In addition to an overall assessment of cellularity and fibrosis In 50 patients with IPF. In 10 randomly selected biopsies examined twice In a blinded fashion, absolute agreement between assessments for a given pathologist varied between 54 and 64% (mean = 57.5%) and In the majority of Instances the agreement was greater than would have occurred by chance. There was good agreement for most verlables across the four raters on the 101samples. The mean score for some of the parameters reported by a given rater deviated occasionally from those of the other raters, but no single rater was consistently different from the other raters. A principal component factor analysis revealed that the pathologic features fell Into four general groupings: alveolar wall mataplssla, fibrosis, honeycombing, smooth muscle, and Vlscular changes fell Into one group; severity and extent of cellularity In the alveolar wall Into a second group; severity and extent of cellularity In the alveolar space Into a third group; and Interstitial young connective tissue along with granulation tissue In the airways formed the fourth group. The correlations for the factors were significant (p < 0.0001) as follows: 0.7588 for fibrosis, 0.5542 for alveolar well cellularity, 0.5112 for alveolar space cellularity, and 0.4858 for connective/granulation tissue; the SE of each of these coefficients was 0.0406. Weconclude that the scoring system Is simple and provides a semiquantitative assessment of pathologic features that msy be useful In the aveluatlon of IPF. The observed verlatlons In scoring point out the Importance of tissue evaluation by a panel of pathologists In clinical reAM REV RESPIR DIS 1991; 144:892-900 search and In reporting on the disease activity In IPF.
odic meetings and a consensus diagnosis (or lack thereof) was agreed upon. Fifty patients in whom there was a consensus for the diagnosis of IPF formed the basis of this study. In all, lung biopsy tissue was obtained by open biopsy from two different sites, generally the upper and lower lobes of the same lung (when technically feasible). A total of 101 specimens were obtained from the 50 patients. Immediately following the biopsy, all specimens were gently inflated with formalin, using a syringe and needle and injecting through the pleura. The specimens were then processed in routine fashion, and multiplestep sections showing representative pathology, as selected by one observer (J.W.), were cut and distributed among the four pathologists. Each pathologist received five slides from each block recut: one stained with hematoxylin and eosin, one with pentachrome stain (which provides differential staining of elastic tissue, collagenized connective tissue, and mucopolysaccharide-rich stroma), one
(Received in originalform August 24, 1990and in revised form May 6, 1991) 1 From the Departments of Medicine and Pathology, National Jewish Center for Immunology and Respiratory Medicine, Division of Respiratory Science and Critical Care, University of Colorado, Denver; Mayo Clinic, Minnesota; University of Michigan, Michigan; University of British Columbia, Canada. 2 Supported by SCOR Grants No. HL-27353, HL-142112, and HL-15092 from the National Heart, Lung, and Blood Institute, National Institutes of Health, and by the GCRC Program, Bethesda, Maryland. 3 Correspondence and requests for reprints should be addressed to ReubenM. Cherniack, M.D., Department of Medicine, National Jewish Center for Immunology and RespiratoryMedicine, University of Colorado, 1400Jackson Street, Denver, CO 80206. 4 Members of the Committee were Daniel E. Banks, 'Iulane University, Dianne Y. Bell, Duke University,Gerald S. Davis, University of Vermont, and Janet M. Hughes, Tu'ane University.
QUANTITATIVE ASSESSMENT OF LUNG PATHOLOGY IN IPF
893 TABLE 1 SCORING OF OPEN LUNG BIOPSIES IN IPF Variable Score Grade (% tissue involvement)'
o 1 2 3 4 5 (absent) toccesionet; « 25%) (25-49%) (50-75%) (> 75%) Overall assessment Cellularity (inflammatory cells) Fibrosis Inflammatory/exudative changes Alveolar wall Cell infiltrate, extent Cell infiltrate, severity Cell metaplasia Lymphoid Aggregates, total no. Alveolar space, including alveolar duct, respiratory bronchioles Cellularity, extent Cellularity, severity Granulation tissue, degree; Absent, present, or marked Fibrotic/reparative changes Alveolar wall Interstitial young connective tissue Interstitial fibrosis, % alveolar walls including honeycombing Honeycombing alone, % Smooth muscle, % low power Vessel myointimal changes Airways, including respiratory and terminal bronchioles; Absent, present, or marked Mural inflammation, degree Luminal granulation tissue, degree Mural fibrosis • Whichever is appropriate for the individual parameter.
with Prussian blue (for iron), one with toluidine blue stain, and one unstained slide. The pentachrome stain was used for evaluation of connective tissue and vessels,the toluidine blue stain to assess metachromatic properties, and the iron stain to look for ferruginous bodies and other hemosiderin-containing structures. For each case, a composite grading was developed by each pathologist from all of the slides, and individual slides were not scored separately. As can be seen in table I, each pathologist first provided a qualitative assessment, on a scale of zero to 5, of the overall degree of cellularity and fibrosis present in the biopsy specimens. Figure 1presents the frequency distributions of the mean of the four scores provided by the pathologists for the overall assessments of fibrosis and cellularity in the 50 cases of IPF. Table 1 also indicates that, in addition to the overallassessments of fibrosis and cellularity, 15 specific histopathologic features that were divided into three sections - inflammatory/exudative changes, fibrotic/reparative changes, and airway alterations - were scored in a semiquantitative fashion. The parameters that were evaluated, and the scoring system that was used, were agreed upon in a series of preliminary discussions among the pathologists, and after informal review and quantification of a small number of cases. As shown in table I, a grade of zero to 5, roughly correlating with percentages (absent, occasional, < 25070, 25
to 49,50 to 75, and> 75%) was used for most of the parameters. By concensus it was thought that for the inflammatory/exudative changes, both the extent and the severity of cellularity in the alveolar wall and in the alveolar space (often referred to as desquamation), should be assessed. This was necessitated by the lack of uniformity often seen from one site to the next in IPF. For example, in a given biopsy cell infiltrates might be quite dense (high severity) but only in a small focus (low extent). Metaplastic cells lining the airspaces were assessed for an entire tissue specimen, even in foci of honeycombing. A simpler grading system (absent, present, marked) was used for granulation tissue in the alveolar space, as this was considered to be a more subjective observation and difficult to distinguish from alveolar wallchanges; lymphoid aggregateswere simply expressed as the number actually seen. The fibrotic/reparative changes in the interstitium and alveolar walls were assessed according to the percentage of tissue involved using a zero to 5 grading system as well. "Young connective tissue," rich in fibroblasts and metachromatic stroma and relativelypoor in mature collagen, was considered distinct from fibrosis (with well-developed"collagenization") and honeycombing. Fibrotic changes in the alveolar wall werescored by assessment of the total extent of interstitial fibrosis (including honeycombing and foci with young connective tissue). The percents of tissue
showing honeycombing alone and young connective tissue alone were evaluated separately. The extent of metaplastic smooth muscle in the stroma, as distinct from normal smooth muscle associated with airways and vasculature, and of myointimal mural thickening in the vessel walls, a common feature in areas of scarring also was assessed. During the initial discussions of the panel, evaluation of the airways was recognized to CELLULARITY
12
en en
LLI
«
0
10
8 6 4
2 II.
0
0
II: 1 0 ID :E 8
FIBROSIS
LLI
:> Z
6 4
2 0
4
Fig. 1. Frequency distribution of the average of the scores for overall assessment of cellularity and fibrosis by the four pathologists in 50 cases of IPF.
CHERNIACK, COLBY, FLINT, ET AL.
894
Fig. 2. (A and B) Idiopathic pulmonary fibrosis with foci of old scarr ing (straight arrows) and zones where more edematous loose-appear lnq fibroblasts are present (curved arrows) . The looser appearing foci show more intense staining with stains for acid mucopolysaccharide, and contrast with the dense collagenized foci. (C) In some fields, it is difficult to separate fibroblastic proliferatio n with in airspaces from fibroblastic proliferation incorporated into the interstitium (arrows), although in other foci, airspace organization is definitely present (0) (straight arrows) . A tiny remnant of fibrinous exudate, undergoing organization is present (0) (curved arrow). These fields would be scored (see table 1) as follows: alveolar wall cell infiltrate (extent = 4, severity = 2; from A); alveolar space granulation tissue = present; from 0; young connective tissue = 2; from A to C; fibrosis = 3; from A; and honeycombing = O.
be very subjective, and therefore it was considered that this might not lend itself to good intraobserver agreement. A relati vely simple scoring system (absent, present , marked) was therefore used to evaluate airway mural inflammation,lumenai granulation tissue (bronchiolitis obliterans), and mural fibrosis. Examples of some of the pathologic features and their assigned scores are shown in figure 2, and illustrative cases demonstrating the use of the scoring system are presented in figures 3 through 9.
Statistical Analysis The reproducibility of the readings of the pathologists was assessed by resubmission of a separate set of slides cut from the same blocks of 10randomly selected, unlabeled biopsies to each of the pathologists 3 to 4 months later. The intraobserver variability of readings for each pathology variable was as-
sessed in two ways: (1) the percentage of times each rater assigned the same score for each pathology variable, and (2 ) by the K statistic, which is used for nominal or ordinal data (5) and which indicates the proportion of agreement between rating s after accounting for chance agreement. The range of K values range from -1 to I, with I being perfect agreement, a score of zero indicating pure chance agreement, a positive value (zero to I) indicating greater agreement than expected by chance, and a negative value ( -I to zero) agreement that is even less than expected by chance. The interobserver variability (i.e., the degree to which the four rater s agreed with each other in their evaluat ion of each variable) in the 101 biopsies also was assessed by determination of K statistics (5). A principal component factor analysis (6, 7) with varimax rotation was performed for each observation in order to determine wheth-
er there were logical groupings of related pathologic variables. The factor anal ysis was performed on each rater's data separately, and then on the mean ratings acros s all four pathologis ts, using the procedure provided by SAS version 6.04 (SAS Institute, Cary, NC). The number of factors to be fit was deter-mined by the number of eigenvalues greater than one ·(6). Results
The reproducibility of the assessment of the pathologic features is indicated by the percentage of times there was absolute agreement between the two ratings on different slides of the 10 cases for each rater, as well as the associated K statistic and its standard error (table 2). The K values that are omitted are undefined and occurred when the same rating level was
Fig. 3. Scoring of IPF in open lung biopsy material (based on A except as noted). Scoring was performed using the following scale: 0 - absent, 1 ~ occasional, 2 = < 25% , 3 - 25 to 49% , 4 = 50 to 75%, and 5 - > 75%. Overall assessment scores: cellularity = 3; fibrosis = 2. Alveolar wall scores : cell infiltrate (extent) - 3; cell infiltrate (severity) = 3; type II cell metaplasia = 2 (based on B); lymphoid aggregates (total) = O. Alveolar space scores (includes alveolar duel and respiratory bronch ioles): cellula rity (extent) = 2; cellularity (severity) = 3; granulation tissue (absent , present , or marked) = O. Alveolar wall scores : interstitial young connective tissue = t (arrows); fibrosis (percentage of all alveo iar walls inciuding honeycomb ing) = 3; honeycombing = 0; smooth muscie (percentage involved at low power) = 0; myointimal vascular changes = O. Airways scores (terminal and respiratory bronchioles; grade absent , present, or marked): mural Inflammation = not evaluable on illustration (NE); lumina l granulation tissue = NE; mural fibrosis = NE. (B) Detail of alveolar wall young connective tissue (arrows in A and B). (e) Detail of interstitial fibros is with focal desquamation. (D) Myoint imal change from field (not shown in A).
Fig. 4. Scoring of IPF in open lung biopsy material. Overall assessment scores : cellularity = 3; fibrosis = 2. Alveolar wall scores : cell infiltrate (extent) = 2; cell infiltrate (severity) = 4; type II cell metaplasia = 2; lymphoid aggregates (total) = 3 (small arrows). Alveolar space scores (includes alveolar duct and respiratory bronch ioles) : cellularity (extent) = 1; cellularity (severity) = 1; granulation tissue (absent , present , or marked) = O. Alveolar wall scores: interstitial young connective tissue = 0; fibrosis (percentage of all alveolar walls including honeycombing) = 3; honeycombing = 0 (large arrows, detail in B); smooth muscle (percentage involved at low power) = 1; myointimal vascular changes = O. Airways scores (terminal and respiratory bronchioles; absent , present, or marked): mural inilammation - (NE) ; lum inal granulation tissue = NE ; mural fibrosis = NE. (For abbrev iations and explanation of scoring, see figure 3.)
co Ul
II)
~ Z ...:;;
:i!o
~
l:l
z
E
...
~ o
m
3:
IJ)
rn
IJ)
~
s
~
:!
z
o ~
Fig . 5. Scoring of IPF in open lung biopsy material. Overall assessment scores: cellularity = 2; fibrosis = 4. Alveolar wall scores : cell infiltrate (extent) = 4; cell infiltrate (severity) = 2; type II cell metaplasia = 3; lymphoid aggregates (total) ~ 1. Alveolar space scores (includes alveolar duct and respiratory bronchioles): cellularity (extent) = 2; cellularity (severity) = 3; granulation tissue (absent, present, or marked) = O. Alveola r wall scores : interstitial young connective tissue R 0; fibrosis (percentage of all alveolar walls inctudlnq honeycombing) R 5; honeycombing • 4; smooth muscle (percentage involved at low power) - 0; myointimal vascular changes = 0 (curved arrows). Airways scores (terminal and respiratory bronchioles ; absent, present, or marked): mural inflammation = 0; luminal granulation tissue = 0; mural fibrosis = O. (8) Detail of honeycombing . (For explanation of scoring. see figure 3.)
R
Fig. 6. Scoring of IPF in open lung biopsy material. Overall assessment scores : cellularity = 3; fibrosis = 5. Alveolar wall scores : cell infiltrate (extent) = 5; cell infiltrate (severity) = 3; type II cell metaplas ia = 3 (seen in B); lymphoid aggregates (total) = 1. Alveolar space scores (includes alveolar duct and respiratory bronchioles): cellularity (extent) R 1; cellular ity (severity) = 1; granulation tissue (absent , present, or marked) = O. Alveolar wall scores: interstitial young connective tissue = 0; fibrosis (percentage of all alveolar walls including honeycombing) = 5; honeycombing = 2; smooth muscle (percentage involved at low power) 0; myointimal vascular changes = zero. Airways scores (terminal and respiratory bronchioles ; absent , present, or marked): mural inflammation = NE; luminal granulation tissue = NE; mural fibrosis = NE. (8) Detail of smooth muscle metaplasia . (For explanation of abbreviat ions and scoring, see figure 3.)
i!'i
r-
»,
ITl
:1
.
z
[!
:
REUBEN M. CHERNIACK, THOMAS V. COLBY, ANDREW FLINT, WILLIAM M. THURLBECK, JAMES WALDRON, LYNN ACKERSON, TALMADGE E. KING, JR., and the BAL COOPERATIVE GROUP STEERING COMMITTEE'
Introduction
The diagnosis and classification of most interstitial lung diseases requires histologic evaluation of biopsy material. In the case of idiopathic pulmonary fibrosis (IPF), an open lung biopsy is necessary to confirm the diagnosis. The disease activity (i.e., the stage of the disease) is thought to be represented by the degree of cellularity present, and it is generally accepted that the prognosis and response to therapy is related to the relative amounts of cellular and fibrotic involvement of the lung parenchyma (1-3). In routine practice, a qualitative assessment of the relative extent and severity of these pathologic changes is usually provided by pathologists, so that correlation between pathologic changes and the clinical and physiologic alterations is at best crude. It is possible that a more rigorous quantitative assessment of pathologic parameters would provide a better correlation with clinical parameters and physiologic abnormalities, as well as the response to therapeutic intervention. As part of a multicenter study in which the role of bronchoalveolar lavage constituents in patients with IPF wereevaluated (4), a panel of four pathologists was established to quantify the pathologic changes present in IPF. The purpose of this report is to describe the pathologic scoring system that was developed, and the variables that wereassessed. In addition, the reproducibility (i,e., intraobserver variability) and the comparability (i.e., interobserver variability) of the readings of the four pathologists are presented. Methods All suspected cases of IPF from the multicenter study were reviewed and scored prospectively by the panel of four pathologists (T.V.C., A.F., W.M.T., and JYV.) who were unaware of the clinical or physiologic findings. Cases in which there were differing diagnoses werereevaluated by the panel at peri892
SUMMARY The diagnosis and classification of most Interstitial lung diseases requires histologic evaluation of lung tissue, obtained by an open lung biopsy to confirm the diagnosis. In addition, It Is generally accepted that response to therapy In Idiopathic pulmonary fibrosis (IPF) Is related to the relative degree of cellularity and fibrosis present. Because only a qualitative assessment of the relative extent and severity of these changes Is generally provided, correlation with clinical and physiologic alterations Is difficult. This report describes resulta of a semiquantitative assessment by four pathologists of Inflammatory/exudative changes, fibrotic/reparative changes, and airway alterations, In addition to an overall assessment of cellularity and fibrosis In 50 patients with IPF. In 10 randomly selected biopsies examined twice In a blinded fashion, absolute agreement between assessments for a given pathologist varied between 54 and 64% (mean = 57.5%) and In the majority of Instances the agreement was greater than would have occurred by chance. There was good agreement for most verlables across the four raters on the 101samples. The mean score for some of the parameters reported by a given rater deviated occasionally from those of the other raters, but no single rater was consistently different from the other raters. A principal component factor analysis revealed that the pathologic features fell Into four general groupings: alveolar wall mataplssla, fibrosis, honeycombing, smooth muscle, and Vlscular changes fell Into one group; severity and extent of cellularity In the alveolar wall Into a second group; severity and extent of cellularity In the alveolar space Into a third group; and Interstitial young connective tissue along with granulation tissue In the airways formed the fourth group. The correlations for the factors were significant (p < 0.0001) as follows: 0.7588 for fibrosis, 0.5542 for alveolar well cellularity, 0.5112 for alveolar space cellularity, and 0.4858 for connective/granulation tissue; the SE of each of these coefficients was 0.0406. Weconclude that the scoring system Is simple and provides a semiquantitative assessment of pathologic features that msy be useful In the aveluatlon of IPF. The observed verlatlons In scoring point out the Importance of tissue evaluation by a panel of pathologists In clinical reAM REV RESPIR DIS 1991; 144:892-900 search and In reporting on the disease activity In IPF.
odic meetings and a consensus diagnosis (or lack thereof) was agreed upon. Fifty patients in whom there was a consensus for the diagnosis of IPF formed the basis of this study. In all, lung biopsy tissue was obtained by open biopsy from two different sites, generally the upper and lower lobes of the same lung (when technically feasible). A total of 101 specimens were obtained from the 50 patients. Immediately following the biopsy, all specimens were gently inflated with formalin, using a syringe and needle and injecting through the pleura. The specimens were then processed in routine fashion, and multiplestep sections showing representative pathology, as selected by one observer (J.W.), were cut and distributed among the four pathologists. Each pathologist received five slides from each block recut: one stained with hematoxylin and eosin, one with pentachrome stain (which provides differential staining of elastic tissue, collagenized connective tissue, and mucopolysaccharide-rich stroma), one
(Received in originalform August 24, 1990and in revised form May 6, 1991) 1 From the Departments of Medicine and Pathology, National Jewish Center for Immunology and Respiratory Medicine, Division of Respiratory Science and Critical Care, University of Colorado, Denver; Mayo Clinic, Minnesota; University of Michigan, Michigan; University of British Columbia, Canada. 2 Supported by SCOR Grants No. HL-27353, HL-142112, and HL-15092 from the National Heart, Lung, and Blood Institute, National Institutes of Health, and by the GCRC Program, Bethesda, Maryland. 3 Correspondence and requests for reprints should be addressed to ReubenM. Cherniack, M.D., Department of Medicine, National Jewish Center for Immunology and RespiratoryMedicine, University of Colorado, 1400Jackson Street, Denver, CO 80206. 4 Members of the Committee were Daniel E. Banks, 'Iulane University, Dianne Y. Bell, Duke University,Gerald S. Davis, University of Vermont, and Janet M. Hughes, Tu'ane University.
QUANTITATIVE ASSESSMENT OF LUNG PATHOLOGY IN IPF
893 TABLE 1 SCORING OF OPEN LUNG BIOPSIES IN IPF Variable Score Grade (% tissue involvement)'
o 1 2 3 4 5 (absent) toccesionet; « 25%) (25-49%) (50-75%) (> 75%) Overall assessment Cellularity (inflammatory cells) Fibrosis Inflammatory/exudative changes Alveolar wall Cell infiltrate, extent Cell infiltrate, severity Cell metaplasia Lymphoid Aggregates, total no. Alveolar space, including alveolar duct, respiratory bronchioles Cellularity, extent Cellularity, severity Granulation tissue, degree; Absent, present, or marked Fibrotic/reparative changes Alveolar wall Interstitial young connective tissue Interstitial fibrosis, % alveolar walls including honeycombing Honeycombing alone, % Smooth muscle, % low power Vessel myointimal changes Airways, including respiratory and terminal bronchioles; Absent, present, or marked Mural inflammation, degree Luminal granulation tissue, degree Mural fibrosis • Whichever is appropriate for the individual parameter.
with Prussian blue (for iron), one with toluidine blue stain, and one unstained slide. The pentachrome stain was used for evaluation of connective tissue and vessels,the toluidine blue stain to assess metachromatic properties, and the iron stain to look for ferruginous bodies and other hemosiderin-containing structures. For each case, a composite grading was developed by each pathologist from all of the slides, and individual slides were not scored separately. As can be seen in table I, each pathologist first provided a qualitative assessment, on a scale of zero to 5, of the overall degree of cellularity and fibrosis present in the biopsy specimens. Figure 1presents the frequency distributions of the mean of the four scores provided by the pathologists for the overall assessments of fibrosis and cellularity in the 50 cases of IPF. Table 1 also indicates that, in addition to the overallassessments of fibrosis and cellularity, 15 specific histopathologic features that were divided into three sections - inflammatory/exudative changes, fibrotic/reparative changes, and airway alterations - were scored in a semiquantitative fashion. The parameters that were evaluated, and the scoring system that was used, were agreed upon in a series of preliminary discussions among the pathologists, and after informal review and quantification of a small number of cases. As shown in table I, a grade of zero to 5, roughly correlating with percentages (absent, occasional, < 25070, 25
to 49,50 to 75, and> 75%) was used for most of the parameters. By concensus it was thought that for the inflammatory/exudative changes, both the extent and the severity of cellularity in the alveolar wall and in the alveolar space (often referred to as desquamation), should be assessed. This was necessitated by the lack of uniformity often seen from one site to the next in IPF. For example, in a given biopsy cell infiltrates might be quite dense (high severity) but only in a small focus (low extent). Metaplastic cells lining the airspaces were assessed for an entire tissue specimen, even in foci of honeycombing. A simpler grading system (absent, present, marked) was used for granulation tissue in the alveolar space, as this was considered to be a more subjective observation and difficult to distinguish from alveolar wallchanges; lymphoid aggregateswere simply expressed as the number actually seen. The fibrotic/reparative changes in the interstitium and alveolar walls were assessed according to the percentage of tissue involved using a zero to 5 grading system as well. "Young connective tissue," rich in fibroblasts and metachromatic stroma and relativelypoor in mature collagen, was considered distinct from fibrosis (with well-developed"collagenization") and honeycombing. Fibrotic changes in the alveolar wall werescored by assessment of the total extent of interstitial fibrosis (including honeycombing and foci with young connective tissue). The percents of tissue
showing honeycombing alone and young connective tissue alone were evaluated separately. The extent of metaplastic smooth muscle in the stroma, as distinct from normal smooth muscle associated with airways and vasculature, and of myointimal mural thickening in the vessel walls, a common feature in areas of scarring also was assessed. During the initial discussions of the panel, evaluation of the airways was recognized to CELLULARITY
12
en en
LLI
«
0
10
8 6 4
2 II.
0
0
II: 1 0 ID :E 8
FIBROSIS
LLI
:> Z
6 4
2 0
4
Fig. 1. Frequency distribution of the average of the scores for overall assessment of cellularity and fibrosis by the four pathologists in 50 cases of IPF.
CHERNIACK, COLBY, FLINT, ET AL.
894
Fig. 2. (A and B) Idiopathic pulmonary fibrosis with foci of old scarr ing (straight arrows) and zones where more edematous loose-appear lnq fibroblasts are present (curved arrows) . The looser appearing foci show more intense staining with stains for acid mucopolysaccharide, and contrast with the dense collagenized foci. (C) In some fields, it is difficult to separate fibroblastic proliferatio n with in airspaces from fibroblastic proliferation incorporated into the interstitium (arrows), although in other foci, airspace organization is definitely present (0) (straight arrows) . A tiny remnant of fibrinous exudate, undergoing organization is present (0) (curved arrow). These fields would be scored (see table 1) as follows: alveolar wall cell infiltrate (extent = 4, severity = 2; from A); alveolar space granulation tissue = present; from 0; young connective tissue = 2; from A to C; fibrosis = 3; from A; and honeycombing = O.
be very subjective, and therefore it was considered that this might not lend itself to good intraobserver agreement. A relati vely simple scoring system (absent, present , marked) was therefore used to evaluate airway mural inflammation,lumenai granulation tissue (bronchiolitis obliterans), and mural fibrosis. Examples of some of the pathologic features and their assigned scores are shown in figure 2, and illustrative cases demonstrating the use of the scoring system are presented in figures 3 through 9.
Statistical Analysis The reproducibility of the readings of the pathologists was assessed by resubmission of a separate set of slides cut from the same blocks of 10randomly selected, unlabeled biopsies to each of the pathologists 3 to 4 months later. The intraobserver variability of readings for each pathology variable was as-
sessed in two ways: (1) the percentage of times each rater assigned the same score for each pathology variable, and (2 ) by the K statistic, which is used for nominal or ordinal data (5) and which indicates the proportion of agreement between rating s after accounting for chance agreement. The range of K values range from -1 to I, with I being perfect agreement, a score of zero indicating pure chance agreement, a positive value (zero to I) indicating greater agreement than expected by chance, and a negative value ( -I to zero) agreement that is even less than expected by chance. The interobserver variability (i.e., the degree to which the four rater s agreed with each other in their evaluat ion of each variable) in the 101 biopsies also was assessed by determination of K statistics (5). A principal component factor analysis (6, 7) with varimax rotation was performed for each observation in order to determine wheth-
er there were logical groupings of related pathologic variables. The factor anal ysis was performed on each rater's data separately, and then on the mean ratings acros s all four pathologis ts, using the procedure provided by SAS version 6.04 (SAS Institute, Cary, NC). The number of factors to be fit was deter-mined by the number of eigenvalues greater than one ·(6). Results
The reproducibility of the assessment of the pathologic features is indicated by the percentage of times there was absolute agreement between the two ratings on different slides of the 10 cases for each rater, as well as the associated K statistic and its standard error (table 2). The K values that are omitted are undefined and occurred when the same rating level was
Fig. 3. Scoring of IPF in open lung biopsy material (based on A except as noted). Scoring was performed using the following scale: 0 - absent, 1 ~ occasional, 2 = < 25% , 3 - 25 to 49% , 4 = 50 to 75%, and 5 - > 75%. Overall assessment scores: cellularity = 3; fibrosis = 2. Alveolar wall scores : cell infiltrate (extent) - 3; cell infiltrate (severity) = 3; type II cell metaplasia = 2 (based on B); lymphoid aggregates (total) = O. Alveolar space scores (includes alveolar duel and respiratory bronch ioles): cellula rity (extent) = 2; cellularity (severity) = 3; granulation tissue (absent , present , or marked) = O. Alveolar wall scores : interstitial young connective tissue = t (arrows); fibrosis (percentage of all alveo iar walls inciuding honeycomb ing) = 3; honeycombing = 0; smooth muscie (percentage involved at low power) = 0; myointimal vascular changes = O. Airways scores (terminal and respiratory bronchioles; grade absent , present, or marked): mural Inflammation = not evaluable on illustration (NE); lumina l granulation tissue = NE; mural fibrosis = NE. (B) Detail of alveolar wall young connective tissue (arrows in A and B). (e) Detail of interstitial fibros is with focal desquamation. (D) Myoint imal change from field (not shown in A).
Fig. 4. Scoring of IPF in open lung biopsy material. Overall assessment scores : cellularity = 3; fibrosis = 2. Alveolar wall scores : cell infiltrate (extent) = 2; cell infiltrate (severity) = 4; type II cell metaplasia = 2; lymphoid aggregates (total) = 3 (small arrows). Alveolar space scores (includes alveolar duct and respiratory bronch ioles) : cellularity (extent) = 1; cellularity (severity) = 1; granulation tissue (absent , present , or marked) = O. Alveolar wall scores: interstitial young connective tissue = 0; fibrosis (percentage of all alveolar walls including honeycombing) = 3; honeycombing = 0 (large arrows, detail in B); smooth muscle (percentage involved at low power) = 1; myointimal vascular changes = O. Airways scores (terminal and respiratory bronchioles; absent , present, or marked): mural inilammation - (NE) ; lum inal granulation tissue = NE ; mural fibrosis = NE. (For abbrev iations and explanation of scoring, see figure 3.)
co Ul
II)
~ Z ...:;;
:i!o
~
l:l
z
E
...
~ o
m
3:
IJ)
rn
IJ)
~
s
~
:!
z
o ~
Fig . 5. Scoring of IPF in open lung biopsy material. Overall assessment scores: cellularity = 2; fibrosis = 4. Alveolar wall scores : cell infiltrate (extent) = 4; cell infiltrate (severity) = 2; type II cell metaplasia = 3; lymphoid aggregates (total) ~ 1. Alveolar space scores (includes alveolar duct and respiratory bronchioles): cellularity (extent) = 2; cellularity (severity) = 3; granulation tissue (absent, present, or marked) = O. Alveola r wall scores : interstitial young connective tissue R 0; fibrosis (percentage of all alveolar walls inctudlnq honeycombing) R 5; honeycombing • 4; smooth muscle (percentage involved at low power) - 0; myointimal vascular changes = 0 (curved arrows). Airways scores (terminal and respiratory bronchioles ; absent, present, or marked): mural inflammation = 0; luminal granulation tissue = 0; mural fibrosis = O. (8) Detail of honeycombing . (For explanation of scoring. see figure 3.)
R
Fig. 6. Scoring of IPF in open lung biopsy material. Overall assessment scores : cellularity = 3; fibrosis = 5. Alveolar wall scores : cell infiltrate (extent) = 5; cell infiltrate (severity) = 3; type II cell metaplas ia = 3 (seen in B); lymphoid aggregates (total) = 1. Alveolar space scores (includes alveolar duct and respiratory bronchioles): cellularity (extent) R 1; cellular ity (severity) = 1; granulation tissue (absent , present, or marked) = O. Alveolar wall scores: interstitial young connective tissue = 0; fibrosis (percentage of all alveolar walls including honeycombing) = 5; honeycombing = 2; smooth muscle (percentage involved at low power) 0; myointimal vascular changes = zero. Airways scores (terminal and respiratory bronchioles ; absent , present, or marked): mural inflammation = NE; luminal granulation tissue = NE; mural fibrosis = NE. (8) Detail of smooth muscle metaplasia . (For explanation of abbreviat ions and scoring, see figure 3.)
i!'i
r-
»,
ITl
:1
.
z
[!
:
