- Email: [email protected]
Quantitative Analysis of Fibroblastic Foci in Usual Interstitial Pneumonia
CHEST
Original Research INTERSTITIAL LUNG DISEASE
Quantitative Analysis of Fibroblastic Foci in Usual Interstitial Pneumonia* Noriyuki Enomoto, MD; Takafumi Suda, MD, PhD; Masato Kato, MD; Yusuke Kaida, MD; Yutaro Nakamura, MD, PhD; Shiro Imokawa, MD, PhD; Masaaki Ida, MD; and Kingo Chida, MD, PhD
Study objectives: Using semiquantitative scoring methods, several studies have shown that the amount of fibroblastic foci (FF), which are one of the pathologic characteristics in usual interstitial pneumonia (UIP), is a significant prognostic factor of UIP. In those studies, the degree of FF was evaluated semiquantitatively on several scales by a panel of pulmonary pathologists. However, the evaluation was somewhat subjective because interobserver variation was not small. Additionally, these methods are not entirely practical because two or more pathologists are required. In this study, we tried to develop a more quantitative scoring method of FF. Patients and methods: With a charge-coupled device camera, we made images of lung sections obtained from 15 patients with UIP associated with collagen vascular disease (CVD) [CVD-UIP] and 16 patients with idiopathic pulmonary fibrosis (IPF) [IPF/UIP], and calculated the proportion of FF areas in the target image areas with an image analytic software. Measurements and results: Our quantitative scoring method enabled us to readily and objectively evaluate the extent of FF as a quantitative percentage of FF area (%FF) score. Interobserver and intraobserver correlations were high in our method (r ⴝ 0.877 and r ⴝ 0.898, respectively). The quantitative %FF score (ⴞ SD) of IPF/UIP patients was 1.67 ⴞ 0.90%, which was significantly higher than that of CVD-UIP patients (0.39 ⴞ 0.24%, p < 0.0001). A Cox proportional hazards model showed that the quantitative %FF score was a significant predictor of survival in UIP patients. The quantitative %FF score had a correlation with scores assessed by the semiquantitative scoring methods previously reported, but patients with the same score assessed by the semiquantitative methods had widely varying scores assessed by our method. Conclusions: These results suggest that our quantitative scoring method for FF is more objective than the semiquantitative scoring methods previously reported, providing accurate information about the prognosis of patients with UIP. (CHEST 2006; 130:22–29) Key words: collagen vascular disease; fibroblastic foci; idiopathic interstitial pneumonia; quantitative analysis; usual interstitial pneumonia Abbreviations: CCD ⫽ charge-coupled device; CVD ⫽ collagen vascular disease; %FF ⫽ percentage of fibroblastic foci area; FF ⫽ fibroblastic foci; IPF ⫽ idiopathic pulmonary fibrosis; UIP ⫽ usual interstitial pneumonia
foci (FF), distinct aggregates of fibroF ibroblastic blasts, are one of the pathologic characteristics of usual interstitial pneumonia (UIP). Because a variety of profibrotic agents have been found in FF and most cells in FF are myofibroblasts, FF are thought
to represent sites of ongoing lung injury that may progress to dense fibrosis.1– 6 Recently, there has been increasing interest in the prognostic significance of FF in UIP. Indeed, several studies7,8 have demonstrated the clinical importance of FF as a
*From the Second Division, Department of Internal Medicine, Hamamatsu University School of Medicine, Hamamatsu, Japan. The authors have reported to the ACCP that no significant conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. Manuscript received November 3, 2005; revision accepted February 24, 2006.
Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (www.chestjournal. org/misc/reprints.shtml). Correspondence to: Takafumi Suda, MD, PhD, 1–20-1 Handayama, Hamamatsu, 431-3192, Japan; e-mail: [email protected] DOI: 10.1378/chest.130.1.22
22
Original Research
prognostic parameter in UIP. In idiopathic pulmonary fibrosis (IPF), poor prognosis was shown to be related to greater degrees of FF.7,8 In those studies, to evaluate the degree of FF, several pathologic scoring systems, such as the Brompton,8 the Denver,7,9 and the Michigan10 FF scoring methods, were employed. Briefly, two or more specialists in pulmonary pathology reviewed specimens obtained by surgical lung biopsy and semiquantitatively scored the degrees of FF on several scales. These scoring methods have at least two disadvantages. First, the For editorial comment see page 3 assessment is somewhat subjective, and not actually quantitative, because interobserver variation is not small when evaluated with the use of unweighted coefficient of agreement.8,9 Second, in the practical setting, it is not easy to consult two or more specialists in pulmonary pathology just for quantifying the degree of FF. Indeed, although Flaherty et al10 showed a positive association among these three scoring methods, the association ratios were relatively low (⬍ 0.56) and some discrepancies were noted in certain clinical parameters, such as predicting survival. Thus, the evaluation of FF degrees may not be identical when different scoring systems are employed. Accordingly, Flaherty et al10 emphasized the need for further studies to define the optimal way to score the degree of FF. Even in the practical setting, there is a need for a more objective and simple scoring method to assess the degree of FF, which would enable us to easily score FF even without consulting a panel of specialists in pulmonary pathology, and to directly compare our scores with published data and/or the data of others. The present study was conducted to develop a more objective and quantitative scoring method to accurately assess the degree of FF in UIP. With a charge-coupled device (CCD) camera, we captured the images of specimens of patients with IPF/UIP and collagen vascular disease (CVD)-associated UIP (CVD-UIP), and measured the percentage of FF area (%FF) in the target image areas using image analysis software. Using our quantitative scoring method, we were able to readily and accurately assess the degree of FF. Further, we found that our quantitative FF score was an independent prognostic factor of the survival of patients with UIP. These results suggest that our quantitative FF scoring method, which is easily performed, can provide a more objective and quantitative assessment of the degree of FF. www.chestjournal.org
Materials and Methods Study Population We experienced consecutive 39 patients with interstitial pneumonia associated with CVD and 37 patients with idiopathic interstitial pneumonia who underwent open or thoracoscopic lung biopsy in our hospital between 1990 and 2003. Surgical lung biopsy slides were independently reviewed by two lung pathologists (S.I., N.E.) who were unaware of clinical or physiologic findings. When the classification differed between the pathologists, a consensus opinion on the overall histopathologic pattern was reached. Histologic features of UIP were based on a previously published report11 and the criteria in the American Thoracic Society/European Respiratory Society consensus classification.12 Of the patients, 17 with CVD and 20 with idiopathic interstitial pneumonia were histologically classified into UIP. Fifteen patients with CVD-UIP and 16 with IPF/UIP for whom clinical findings and lung specimens were available participated in the present study. The study protocol was approved by the Ethical Committee of the Hamamatsu University School of Medicine, and informed consent was obtained from all patients. No patients in an accelerated phase of interstitial pneumonia were included. The patients with CVD-UIP who all fulfilled the diagnostic criteria for their respective CVD included five patients with primary Sjogren syndrome; four patients with rheumatoid arthritis; two patients with systemic sclerosis; two patients with polymyositis/dermatomyositis; one patient with systemic lupus erythematosus and Sjogren syndrome; and one patient with rheumatoid arthritis and polymyositis/dermatomyositis. Quantification of the Area of FF Lung specimens were obtained from at least two lobes, and all available specimens were reviewed. Images of sections stained with hematoxylin-eosin were made using a microscope (Axiophoto; Carl Zeiss Corporation; Oberkochem, Germany) with a CCD camera (SPOT; Diagnostic Instruments; Sterling Heights, MI). At least 10 fields of imaged lesions at 100-fold magnification were randomly selected except for the areas of honeycombing. Briefly, we blindly moved the target fields in a CCD camera within the section. If honeycomb area was present in the target field, we moved the fields again until the field has no honeycombing area. In each selected field, the area of FF was measured using analytic software (Image-Pro Plus, version 4.5.0.24; Media Cybernetics; Atlanta, GA), and %FF was calculated by dividing the area of FF by that of the target field, as in Figure 1. Overall %FF, a quantitative %FF score, in each patient was defined as the average %FF in ⬎ 10 selected fields. To evaluate an intraobserver correlation of our quantitative scoring method, the same pathologist reviewed the same specimen on different days. Additionally, to evaluate an interobserver correlation of our method, the two pathologists (S.I., N.E.) independently reviewed the same specimen. The intraobserver and interobserver correlation were assessed using the Pearson correlation coefficient. To determine the optimal number of fields counted, we measured the area of FF in 5, 10, and 20 fields in the same specimens two times. The %FF scores calculated were almost similar by 10- and 20-field measurement even when measuring at different times but were varied in 5-field measurement. Correlation between the scores at different times was significant in 10and 20-field measurements (r ⫽ 0.970, p ⫽ 0.0365; r ⫽ 0.979, p ⫽ 0.0232, respectively) but not significant in 5-field measurement (r ⫽ 0.797, p ⫽ 0.2761). Thus, we measured at least 10 fields in each specimen. CHEST / 130 / 1 / JULY, 2006
23
Data Collection Clinical data, including sex, age, smoking history, symptoms, treatment, and outcome, were obtained from patient medical records. Laboratory findings, pulmonary function tests, and BAL data at the time of surgical lung biopsy were also recorded. Statistical Analysis
Figure 1. Images of sections made using a microscope with a CCD camera and image analysis software. Lung specimens were stained with hematoxylin-eosin, and at least 10 fields of imaged lesions at 100-fold magnification in each lobe were randomly selected. In each selected field, the %FF was calculated by dividing the area of FF (b) by that of the target field (a). The overall %FF and the quantitative %FF score in each patient are defined as the average of the %FF in ⬎ 10 selected fields.
Statistical analysis was performed using statistical software (StatView J-4.5; SAS Institute; Cary, NC). Categorical data were compared between CVD-UIP and IPF/UIP using the 2 test for independence, and continuous data were compared using MannWhitney U test. The intraobserver and interobserver correlation, and the relationship between the quantitative %FF score and clinical continuous data were analyzed using the Pearson correlation coefficient. The relationship between the quantitative %FF score and the semiquantitative scores was analyzed using the Spearman rank correlation coefficient. The overall survival experience for each group was estimated using Kaplan-Meier curves. The log-rank test was used to compare survival between two groups. The effect of the quantitative %FF score on the risk of death was modeled using the Cox proportional hazards regression. All tests were two sided and performed at the 0.05 significance level.
Results Semiquantification of the Degrees of FF To compare our quantitative scoring method using the analytic software with semiquantitative scoring systems previously reported, each pathologist scored the profusion of FF semiquantitatively in three different ways, according to the Brompton,8 the Denver,7,9 and the Michigan10 FF scoring methods. When the score differed between the pathologists, a consensus score was reached. Interobserver variation was quantified using an unweighted coefficient of agreement to take into account the degree of disagreement on the semiquantitative categorical scales.
Clinical Characteristics, Laboratory Findings, Pulmonary Function Testing, and BAL Findings The clinical characteristics and laboratory findings of the patients with CVD-UIP and IPF/UIP are summarized in Table 1. There were no significant differences in clinical characteristics between CVDUIP and IPF/UIP, although the proportion of female patients with CVD-UIP tended to be higher than that with IPF/UIP, and pack-years of smoking in
Table 1—Clinical Characteristics, and Laboratory, Physiologic and BAL Fluid Cell Findings of the Study Populations* Variables Clinical characteristics Male/female gender, No. Age at biopsy, yr Pack-yr of smoking Symptom onset, mo Laboratory findings Serum lactate dehydrogenase, U/L Serum KL-6, U/L Serum surfactant protein D, U/L Physiologic FVC, % predicted FEV1, % predicted Resting Pao2, mm Hg BAL fluid cell analysis, % Lymphocytes Neutrophils Eosinophils
CVD-UIP
IPF/UIP
p Value, 2 Test
8/7 58.6 ⫾ 8.6 (42–70) 23.5 ⫾ 23.4 (0–50) 7.9 ⫾ 13.1 (1–43)
13/3 62.0 ⫾ 7.7 (50–79) 47.7 ⫾ 48.7 (0–177) 7.6 ⫾ 12.1 (3–48)
0.20 0.43 0.17 0.86
234 ⫾ 64 (174–377) 785 ⫾ 865 (254–2,320) 123 ⫾ 47 (89–207)
264 ⫾ 100 (180–526) 782 ⫾ 562 (303–1,400) 311 ⫾ 185 (100–447)
0.50 0.65 0.30
82.1 ⫾ 15.2 (66–111) 83.4 ⫾ 8.4 (73–100) 89.6 ⫾ 9.2 (76–103)
81.1 ⫾ 25.2 (45–128) 85.4 ⫾ 8.7 (72–100) 79.6 ⫾ 11.2 (64–102)
0.98 0.58 0.08
7.0 ⫾ 5.8 (0.5–20) 1.2 ⫾ 1.7 (0–39) 4.4 ⫾ 12.2 (0–7)
4.4 ⫾ 2.0 (0–26) 0.9 ⫾ 2.2 (0–8.1) 1.1 ⫾ 2.0 (0–8.6)
0.18 0.42 0.15
*Data are presented as mean ⫾ SD (range) unless otherwise indicated. 24
Original Research
IPF/UIP tended to be higher than those in CVDUIP. In laboratory findings, serum surfactant protein-D levels of IPF/UIP were higher than that of CVD-UIP, although the difference was not significant. In pulmonary function tests or BAL analysis, no significant difference was found between CVD-UIP and IPF/UIP. Quantitative Analysis of the Degrees of FF: the Quantitative %FF Score Using a CCD camera and the analytic software, we readily calculated the %FF in the sections of surgical lung biopsy (Fig 1). Even in IPF/UIP, the FF areas occupied only ⬍ 4% of the target fields, but our method enabled us to accurately quantify the extent of FF. The calculated proportion of the FF areas, the quantitative %FF score, was 1.67 ⫾ 0.90% (⫾ SD) in IPF/UIP patients. Interestingly, patients with CVD-UIP had much lower %FF score (0.39 ⫾ 0.24%, p ⬍ 0.0001, compared with patients with IPF/UIP) [Fig 2]. Intraobserver and Interobserver Correlation in the Quantitative Scoring Method Using the Pearson correlation coefficient, we examined the intraobserver and interobserver correlation of the quantitative %FF scores. When our quantitative scoring was repeated on different days by the same observer, a significantly high correlation was found between the first time scores and second time scores (r ⫽ 0.898, p ⬍ 0.0001; Fig 3, top, A). Additionally, when the scoring was done by different observers, a significantly high correlation was also found between them (r ⫽ 0.877, p ⬍ 0.0001; Fig 3, bottom, B).
Figure 3. Intraobserver and interobsever correlations of %FF score. Significantly high intraobserver and interobserver correlations were noted (top, A: r ⫽ 0.898, p ⬍ 0.0001; bottom, B: r ⫽ 0.877, p ⬍ 0.0001, respectively).
Correlation Between the Quantitative %FF Score and Clinical Data
Figure 2. Quantitative %FF scores in patients with CVD-UIP and in IPF/UIP. The quantitative %FF scores were significantly higher in IPF/UIP than in CVD-UIP (p ⬍ 0.0001). www.chestjournal.org
No significant correlation was found between the quantitative %FF score and FVC or resting Pao2 at surgical lung biopsy in CVD-UIP or IPF/UIP (data not shown). There was no correlation between the quantitative %FF score and BAL findings in CVDUIP (data not shown), while a significant correlation was found between the quantitative %FF score and the percentage of neutrophils (p ⫽ 0.026, r ⫽ 0.55) and eosinophils (p ⫽ 0.046, r ⫽ 0.50) in IPF/UIP. CHEST / 130 / 1 / JULY, 2006
25
Table 3—Multivariate Cox Proportional Hazards Models of Survival Adjusting for Age, FVC at Surgical Lung Biopsy, and IPF/UIP Diagnosis in All UIP Patients
Figure 4. Comparison of survival between CVD-UIP and IPF/UIP. Kaplan-Meier analysis shows that patients with CVDUIP had significantly better survival than those with IPF/UIP (log-rank test, p ⫽ 0.045).
Correlation Between the Quantitative %FF Score and Survival Comparison of survival between IPF/UIP and CVD-UIP is shown in Figure 4. Consistent with the study of Flaherty et al10 and our previous data,13 patients with CVD-UIP had significantly better survival than those with IPF/UIP (log-rank test, p ⫽ 0.045). Univariate Cox proportional hazards regression models showed that the quantitative %FF score (hazard ratio, 3.03; p ⫽ 0.004) and period from symptom onset (hazard ratio, 1.07; p ⫽ 0.019) were significant predictors of survival (Table 2). Multivariate Cox proportional hazards regression models, which were adjusted for age, percentage of predicted FVC at surgical lung biopsy, and IPF/UIP diagnosis, showed that the quantitative %FF score was a significant predictor of survival in all UIP patients (hazard ratio, 7.24, p ⫽ 0.010; Table 3). Period from symptom onset (hazard ratio, 1.07; p ⫽ 0.026) and resting Pao2 (hazard ratio, 0.91; p ⫽ 0.046) were also significant predictors (Table 3). Even in IPF patients
Table 2—Univariate Cox Proportional Hazards Models of Survival in All UIP Patients
Variables
Hazard Ratio
Age at biopsy, yr Male gender Symptom onset, mo IPF/UIP diagnosis FVC, % predicted Resting Pao2, mm Hg Quantitative FF score, %
1.04 0.45 1.07 4.72 0.98 0.93 3.03
26
95% Confidence Interval Lower
Upper
p Value
0.94 0.11 1.01 0.89 0.95 0.86 1.42
1.16 1.93 1.13 24.9 1.02 1.01 6.48
0.468 0.286 0.019 0.068 0.394 0.071 0.004
Variables
Hazard Ratio
Male gender Symptom onset, mo Resting Pao2, mm Hg Quantitative FF score, %
0.12 1.07 0.91 7.24
95% Confidence Interval Lower
Upper
p Value
0.02 1.01 0.83 1.59
0.79 1.14 1.00 32.9
0.028 0.026 0.046 0.010
alone, the quantitative %FF score was an independent predictor of survival (hazard ratio, 5.53; p ⫽ 0.032; Table 4). Comparison Between the Quantitative Scoring Method and Semiquantitative Scoring Methods We compared our quantitative scoring method with three previously reported semiquantitative scoring methods.7–10 With the use of coefficient, interobserver agreement was found to be moderate with the Michigan FF scoring method10 (unweighted coefficient, 0.53; p ⬍ 0.001), whereas the agreement was relatively low with the Denver FF scoring method7,9 (unweighted coefficient, 0.33; p ⬍ 0.001) and with the Brompton FF method8 (unweighted coefficient, 0.27; p ⬍ 0.001). The quantitative %FF score had a significant correlation with the Michigan FF score ( ⫽ 0.794, p ⬍ 0.001; Fig; 5, top, A), Denver FF score ( ⫽ 0.601, p ⫽ 0.001; Fig 5, center, B), and Brompton FF score ( ⫽ 0.608, p ⬍ 0.001; Fig 5, bottom, C), but patients with the same score assessed by the semiquantitative methods had widely varying scores assessed by our method. Discussion The main aim of this study was to develop a more quantitative procedure for scoring the degrees of FF Table 4 —Multivariate Cox Proportional Hazards Models of Survival Adjusting for FVC at Surgical Lung Biopsy in IPF/UIP Patients
Variables
Hazard Ratio
Age at biopsy, yr Male gender Symptom onset, mo Resting Pao2, mm Hg Quantitative FF score, %
1.04 0.28 1.06 0.94 5.53
95% Confidence Interval Lower
Upper
p Value
0.90 0.44 1.00 0.85 1.16
1.19 1.77 1.13 1.03 26.3
0.598 0.175 0.057 0.178 0.032
Original Research
Figure 5. Correlation between the quantitative scoring method and semiquantitative scoring methods. The quantitative %FF score had a significant correlation with Michigan FF score (top, A; ⫽ 0.794, p ⬍ 0.001), Denver FF score (center, B; ⫽ 0.601, p ⫽ 0.001), and Brompton FF score (bottom, C; ⫽ 0.608, p ⬍ 0.001), whereas patients with the same score assessed by the semiquantitative methods had widely varying scores as assessed by the quantitative method. www.chestjournal.org
than the semiquantitative methods previously reported. Our scoring method using a CCD camera and analytic software enabled us to accurately measure the %FF, and evaluate the degree of FF as the quantitative %FF scores. Additionally, we found that our quantitative %FF score was an independent prognostic factor in patients with UIP in the Cox proportional hazards regression model. Interestingly, patients with CVD-UIP had a significantly lower %FF score than those with IPF/UIP, which may be related to the better prognosis of CVD-UIP than IPF/UIP. In the present study, we found the interobserver variations were not small in the semiquantitative scoring methods, especially in the methods that had more categorical scales, such as the Denver (six scales) and the Brompton (seven scales) methods. In contrast, our quantitative scoring method had high interobserver correlation (r ⫽ 0.877). Moreover, the intraobserver correlation was also shown to be high (r ⫽ 0.898). In addition, our quantitative scoring method basically needs only one pathologist, who does not necessarily need to specialize in pulmonary pathology, and the pathologist could easily calculate the proportion of FF using a CCD camera and the analytic software. Taken together, these data suggest that our quantitative scoring method, which does not need two or more of pulmonary pathologists, is more objective and reproducible than the semiquantitative scoring methods. We examined whether the quantitative %FF score defined by our scoring method had clinical significance in UIP. Most notably, we found that this score was an independent predictor of survival in all UIP patients as well as IPF patients in the Cox proportional hazards regression model; higher FF score was associated with poor prognosis. In IPF, conflicting reports using semiquantitative scoring systems exist concerning the significance of FF score for survival in the proportional hazards model. Flaherty et al10 showed that the Michigan FF score was not a significant predictor of survival, while Nicholson et al8 demonstrated that the Brompton FF score was independently linked to survival. This may reflect the difference in categorization of FF among the scoring systems. In addition, we found no significant correlation between the quantitative %FF score and FVC or resting Pao2 at surgical biopsy in CVD-UIP or IPF/UIP, indicating that this FF score was not associated with baseline pulmonary function. Interestingly, Nicholson et al8 demonstrated that increasing semiquantitative FF score was independently associated with greater declines at both 6 months and 12 months in FVC and carbon monoxide diffusing capacity in IPF.8 Unfortunately, we did not examine the trends of pulmonary function tests at CHEST / 130 / 1 / JULY, 2006
27
such intervals. Thus, further studies will be needed to clarify whether our quantitative %FF score is related to functional decline. There was no correlation between the quantitative %FF score and BAL findings in CVD-UIP, while the quantitative %FF score was significantly correlated with the percentage of neutrophils (p ⫽ 0.026, r ⫽ 0.55), and eosinophils (p ⫽ 0.046, r ⫽ 0.50) in IPF. Interestingly, previous studies14 –20 demonstrated that higher percentages of neutrophils or eosinophils were associated with more extensive disease and poor prognosis in IPF. Although the correlation rates were not high, there might be some association between extent of FF and the number of BAL neutrophils and eosinophils. Our previous study13 showed CVD-UIP had more favorable prognosis than IPF. In the present study, we found that patients with CVD-UIP had significantly lower %FF scores and a better prognosis than those with IPF. Consistent with our results, Flaherty et al10 showed that the degree of FF was smaller in CVD-UIP than in IPF/UIP, as assessed by their semiquantitative scoring method, which might be associated with the favorable prognosis of CVD-UIP. The reason for the variation in the degree of FF between IPF and CVD-UIP is unclear. The etiology is thought to differ between IPF and CVD-UIP: the formation of FF might be induced by the injury of the alveolar epithelium in IPF,1,21 while endothelial injury probably causes FF formation in CVDUIP.22–24 This etiologic difference might be responsible for the variation in the degree of FF. In addition, the present data demonstrated that the quantitative %FF score was an independent predictor of prognosis in the multivariate Cox proportional hazards regression model adjusted by IPF/UIP diagnosis, suggesting that FF play a pivotal role in determining the prognosis of patients having UIP, regardless of etiology. The main weakness of the present study is its retrospective design and the relatively small number of patients. In addition, the survival rate of our IPF patients seemed to be relatively higher than that of typical IPF patients. Approximately 38% of our IPF patients came to the hospital because of interstitial abnormalities on the chest radiograph of annual medical checkups without respiratory symptoms, suggesting that our study population included early stage IPF patients. This may account for the relatively better prognosis of our IPF patients. To confirm our findings, larger prospective studies will be required. In conclusion, our quantitative FF scoring method using a CCD camera and analytic software, which is easily performed, is more objective and quantitative than previously reported semiquantitative scoring 28
methods, and provides accurate information about prognosis in patients with UIP.
References 1 Selman M, Pardo A. The epithelial/fibroblastic pathway in the pathogenesis of idiopathic pulmonary fibrosis. Am J Respir Cell Mol Biol 2003; 29:S93–S97 2 Phan SH. Fibroblast phenotypes in pulmonary fibrosis. Am J Respir Cell Mol Biol 2003; 29:S87–S92 3 Phan SH. The myofibroblast in pulmonary fibrosis. Chest 2002; 122:286S–289S 4 Kuhn C III, Boldt J, King TE Jr, et al. An immunohistochemical study of architectural remodeling and connective tissue synthesis in pulmonary fibrosis. Am Rev Respir Dis 1989; 140:1693–1703 5 Kuhn C, McDonald JA. The roles of the myofibroblast in idiopathic pulmonary fibrosis: ultrastructural and immunohistochemical features of sites of active extracellular matrix synthesis. Am J Pathol 1991; 138:1257–1265 6 Pardo A, Selman M. Molecular mechanisms of pulmonary fibrosis. Front Biosci 2002; 7:d1743– d1761 7 King TE Jr, Schwarz MI, Brown K, et al. Idiopathic pulmonary fibrosis: relationship between histopathologic features and mortality. Am J Respir Crit Care Med 2001; 164:1025– 1032 8 Nicholson AG, Fulford LG, Colby TV, et al. The relationship between individual histologic features and disease progression in idiopathic pulmonary fibrosis. Am J Respir Crit Care Med 2002; 166:173–177 9 Cherniack RM, Colby TV, Flint A, et al. Quantitative assessment of lung pathology in idiopathic pulmonary fibrosis: the BAL Cooperative Group Steering Committee. Am Rev Respir Dis 1991; 144:892–900 10 Flaherty KR, Colby TV, Travis WD, et al. Fibroblastic foci in usual interstitial pneumonia: idiopathic versus collagen vascular disease. Am J Respir Crit Care Med 2003; 167:1410 – 1415 11 Katzenstein AL, Myers JL. Idiopathic pulmonary fibrosis: clinical relevance of pathologic classification. Am J Respir Crit Care Med 1998; 157:1301–1315 12 American Thoracic Society/European Respiratory Society International Multidisciplinary Consensus Classification of the Idiopathic Interstitial Pneumonias. This joint statement of the American Thoracic Society (ATS), and the European Respiratory Society (ERS) was adopted by the ATS board of directors, June 2001 and by the ERS Executive Committee, June 2001. Am J Respir Crit Care Med 2002; 165:277–304 13 Nakamura Y, Chida K, Suda T, et al. Nonspecific interstitial pneumonia in collagen vascular diseases: comparison of the clinical characteristics and prognostic significance with usual interstitial pneumonia. Sarcoidosis Vasc Diffuse Lung Dis 2003; 20:235–241 14 The diagnosis, assessment and treatment of diffuse parenchymal lung disease in adults. Introduction. Thorax 1999; 54(Suppl)1:S1–S14 15 Haslam PL, Turton CW, Lukoszek A, et al. Bronchoalveolar lavage fluid cell counts in cryptogenic fibrosing alveolitis and their relation to therapy. Thorax 1980; 35:328 –339 16 Rudd RM, Haslam PL, Turner-Warwick M. Cryptogenic fibrosing alveolitis: relationships of pulmonary physiology and bronchoalveolar lavage to response to treatment and prognosis. Am Rev Respir Dis 1981; 124:1– 8 17 Turner-Warwick M, Haslam PL. The value of serial bronchoalveolar lavages in assessing the clinical progress of paOriginal Research
18
19 20
21
tients with cryptogenic fibrosing alveolitis. Am Rev Respir Dis 1987; 135:26 –34 Watters LC, Schwarz MI, Cherniack RM, et al. Idiopathic pulmonary fibrosis: pretreatment bronchoalveolar lavage cellular constituents and their relationships with lung histopathology and clinical response to therapy. Am Rev Respir Dis 1987; 135:696 –704 Peterson MW, Monick M, Hunninghake GW. Prognostic role of eosinophils in pulmonary fibrosis. Chest 1987; 92:51–56 Hallgren R, Bjermer L, Lundgren R, et al. The eosinophil component of the alveolitis in idiopathic pulmonary fibrosis: signs of eosinophil activation in the lung are related to impaired lung function. Am Rev Respir Dis 1989; 139:373– 377 Selman M, King TE, Pardo A. Idiopathic pulmonary fibrosis:
www.chestjournal.org
prevailing and evolving hypotheses about its pathogenesis and implications for therapy. Ann Intern Med 2001; 134:136 –151 22 Hammar SP, Winterbauer RH, Bockus D, et al. Endothelial cell damage and tubuloreticular structures in interstitial lung disease associated with collagen vascular disease and viral pneumonia. Am Rev Respir Dis 1983; 127:77– 84 23 Wusirika R, Ferri C, Marin M, et al. The assessment of anti-endothelial cell antibodies in scleroderma-associated pulmonary fibrosis: a study of indirect immunofluorescent and western blot analysis in 49 patients with scleroderma. Am J Clin Pathol 2003; 120:596 – 606 24 Ihn H, Sato S, Fujimoto M, et al. Characterization of autoantibodies to endothelial cells in systemic sclerosis (SSc): association with pulmonary fibrosis. Clin Exp Immunol 2000; 119:203–209
CHEST / 130 / 1 / JULY, 2006
29
Original Research INTERSTITIAL LUNG DISEASE
Quantitative Analysis of Fibroblastic Foci in Usual Interstitial Pneumonia* Noriyuki Enomoto, MD; Takafumi Suda, MD, PhD; Masato Kato, MD; Yusuke Kaida, MD; Yutaro Nakamura, MD, PhD; Shiro Imokawa, MD, PhD; Masaaki Ida, MD; and Kingo Chida, MD, PhD
Study objectives: Using semiquantitative scoring methods, several studies have shown that the amount of fibroblastic foci (FF), which are one of the pathologic characteristics in usual interstitial pneumonia (UIP), is a significant prognostic factor of UIP. In those studies, the degree of FF was evaluated semiquantitatively on several scales by a panel of pulmonary pathologists. However, the evaluation was somewhat subjective because interobserver variation was not small. Additionally, these methods are not entirely practical because two or more pathologists are required. In this study, we tried to develop a more quantitative scoring method of FF. Patients and methods: With a charge-coupled device camera, we made images of lung sections obtained from 15 patients with UIP associated with collagen vascular disease (CVD) [CVD-UIP] and 16 patients with idiopathic pulmonary fibrosis (IPF) [IPF/UIP], and calculated the proportion of FF areas in the target image areas with an image analytic software. Measurements and results: Our quantitative scoring method enabled us to readily and objectively evaluate the extent of FF as a quantitative percentage of FF area (%FF) score. Interobserver and intraobserver correlations were high in our method (r ⴝ 0.877 and r ⴝ 0.898, respectively). The quantitative %FF score (ⴞ SD) of IPF/UIP patients was 1.67 ⴞ 0.90%, which was significantly higher than that of CVD-UIP patients (0.39 ⴞ 0.24%, p < 0.0001). A Cox proportional hazards model showed that the quantitative %FF score was a significant predictor of survival in UIP patients. The quantitative %FF score had a correlation with scores assessed by the semiquantitative scoring methods previously reported, but patients with the same score assessed by the semiquantitative methods had widely varying scores assessed by our method. Conclusions: These results suggest that our quantitative scoring method for FF is more objective than the semiquantitative scoring methods previously reported, providing accurate information about the prognosis of patients with UIP. (CHEST 2006; 130:22–29) Key words: collagen vascular disease; fibroblastic foci; idiopathic interstitial pneumonia; quantitative analysis; usual interstitial pneumonia Abbreviations: CCD ⫽ charge-coupled device; CVD ⫽ collagen vascular disease; %FF ⫽ percentage of fibroblastic foci area; FF ⫽ fibroblastic foci; IPF ⫽ idiopathic pulmonary fibrosis; UIP ⫽ usual interstitial pneumonia
foci (FF), distinct aggregates of fibroF ibroblastic blasts, are one of the pathologic characteristics of usual interstitial pneumonia (UIP). Because a variety of profibrotic agents have been found in FF and most cells in FF are myofibroblasts, FF are thought
to represent sites of ongoing lung injury that may progress to dense fibrosis.1– 6 Recently, there has been increasing interest in the prognostic significance of FF in UIP. Indeed, several studies7,8 have demonstrated the clinical importance of FF as a
*From the Second Division, Department of Internal Medicine, Hamamatsu University School of Medicine, Hamamatsu, Japan. The authors have reported to the ACCP that no significant conflicts of interest exist with any companies/organizations whose products or services may be discussed in this article. Manuscript received November 3, 2005; revision accepted February 24, 2006.
Reproduction of this article is prohibited without written permission from the American College of Chest Physicians (www.chestjournal. org/misc/reprints.shtml). Correspondence to: Takafumi Suda, MD, PhD, 1–20-1 Handayama, Hamamatsu, 431-3192, Japan; e-mail: [email protected] DOI: 10.1378/chest.130.1.22
22
Original Research
prognostic parameter in UIP. In idiopathic pulmonary fibrosis (IPF), poor prognosis was shown to be related to greater degrees of FF.7,8 In those studies, to evaluate the degree of FF, several pathologic scoring systems, such as the Brompton,8 the Denver,7,9 and the Michigan10 FF scoring methods, were employed. Briefly, two or more specialists in pulmonary pathology reviewed specimens obtained by surgical lung biopsy and semiquantitatively scored the degrees of FF on several scales. These scoring methods have at least two disadvantages. First, the For editorial comment see page 3 assessment is somewhat subjective, and not actually quantitative, because interobserver variation is not small when evaluated with the use of unweighted coefficient of agreement.8,9 Second, in the practical setting, it is not easy to consult two or more specialists in pulmonary pathology just for quantifying the degree of FF. Indeed, although Flaherty et al10 showed a positive association among these three scoring methods, the association ratios were relatively low (⬍ 0.56) and some discrepancies were noted in certain clinical parameters, such as predicting survival. Thus, the evaluation of FF degrees may not be identical when different scoring systems are employed. Accordingly, Flaherty et al10 emphasized the need for further studies to define the optimal way to score the degree of FF. Even in the practical setting, there is a need for a more objective and simple scoring method to assess the degree of FF, which would enable us to easily score FF even without consulting a panel of specialists in pulmonary pathology, and to directly compare our scores with published data and/or the data of others. The present study was conducted to develop a more objective and quantitative scoring method to accurately assess the degree of FF in UIP. With a charge-coupled device (CCD) camera, we captured the images of specimens of patients with IPF/UIP and collagen vascular disease (CVD)-associated UIP (CVD-UIP), and measured the percentage of FF area (%FF) in the target image areas using image analysis software. Using our quantitative scoring method, we were able to readily and accurately assess the degree of FF. Further, we found that our quantitative FF score was an independent prognostic factor of the survival of patients with UIP. These results suggest that our quantitative FF scoring method, which is easily performed, can provide a more objective and quantitative assessment of the degree of FF. www.chestjournal.org
Materials and Methods Study Population We experienced consecutive 39 patients with interstitial pneumonia associated with CVD and 37 patients with idiopathic interstitial pneumonia who underwent open or thoracoscopic lung biopsy in our hospital between 1990 and 2003. Surgical lung biopsy slides were independently reviewed by two lung pathologists (S.I., N.E.) who were unaware of clinical or physiologic findings. When the classification differed between the pathologists, a consensus opinion on the overall histopathologic pattern was reached. Histologic features of UIP were based on a previously published report11 and the criteria in the American Thoracic Society/European Respiratory Society consensus classification.12 Of the patients, 17 with CVD and 20 with idiopathic interstitial pneumonia were histologically classified into UIP. Fifteen patients with CVD-UIP and 16 with IPF/UIP for whom clinical findings and lung specimens were available participated in the present study. The study protocol was approved by the Ethical Committee of the Hamamatsu University School of Medicine, and informed consent was obtained from all patients. No patients in an accelerated phase of interstitial pneumonia were included. The patients with CVD-UIP who all fulfilled the diagnostic criteria for their respective CVD included five patients with primary Sjogren syndrome; four patients with rheumatoid arthritis; two patients with systemic sclerosis; two patients with polymyositis/dermatomyositis; one patient with systemic lupus erythematosus and Sjogren syndrome; and one patient with rheumatoid arthritis and polymyositis/dermatomyositis. Quantification of the Area of FF Lung specimens were obtained from at least two lobes, and all available specimens were reviewed. Images of sections stained with hematoxylin-eosin were made using a microscope (Axiophoto; Carl Zeiss Corporation; Oberkochem, Germany) with a CCD camera (SPOT; Diagnostic Instruments; Sterling Heights, MI). At least 10 fields of imaged lesions at 100-fold magnification were randomly selected except for the areas of honeycombing. Briefly, we blindly moved the target fields in a CCD camera within the section. If honeycomb area was present in the target field, we moved the fields again until the field has no honeycombing area. In each selected field, the area of FF was measured using analytic software (Image-Pro Plus, version 4.5.0.24; Media Cybernetics; Atlanta, GA), and %FF was calculated by dividing the area of FF by that of the target field, as in Figure 1. Overall %FF, a quantitative %FF score, in each patient was defined as the average %FF in ⬎ 10 selected fields. To evaluate an intraobserver correlation of our quantitative scoring method, the same pathologist reviewed the same specimen on different days. Additionally, to evaluate an interobserver correlation of our method, the two pathologists (S.I., N.E.) independently reviewed the same specimen. The intraobserver and interobserver correlation were assessed using the Pearson correlation coefficient. To determine the optimal number of fields counted, we measured the area of FF in 5, 10, and 20 fields in the same specimens two times. The %FF scores calculated were almost similar by 10- and 20-field measurement even when measuring at different times but were varied in 5-field measurement. Correlation between the scores at different times was significant in 10and 20-field measurements (r ⫽ 0.970, p ⫽ 0.0365; r ⫽ 0.979, p ⫽ 0.0232, respectively) but not significant in 5-field measurement (r ⫽ 0.797, p ⫽ 0.2761). Thus, we measured at least 10 fields in each specimen. CHEST / 130 / 1 / JULY, 2006
23
Data Collection Clinical data, including sex, age, smoking history, symptoms, treatment, and outcome, were obtained from patient medical records. Laboratory findings, pulmonary function tests, and BAL data at the time of surgical lung biopsy were also recorded. Statistical Analysis
Figure 1. Images of sections made using a microscope with a CCD camera and image analysis software. Lung specimens were stained with hematoxylin-eosin, and at least 10 fields of imaged lesions at 100-fold magnification in each lobe were randomly selected. In each selected field, the %FF was calculated by dividing the area of FF (b) by that of the target field (a). The overall %FF and the quantitative %FF score in each patient are defined as the average of the %FF in ⬎ 10 selected fields.
Statistical analysis was performed using statistical software (StatView J-4.5; SAS Institute; Cary, NC). Categorical data were compared between CVD-UIP and IPF/UIP using the 2 test for independence, and continuous data were compared using MannWhitney U test. The intraobserver and interobserver correlation, and the relationship between the quantitative %FF score and clinical continuous data were analyzed using the Pearson correlation coefficient. The relationship between the quantitative %FF score and the semiquantitative scores was analyzed using the Spearman rank correlation coefficient. The overall survival experience for each group was estimated using Kaplan-Meier curves. The log-rank test was used to compare survival between two groups. The effect of the quantitative %FF score on the risk of death was modeled using the Cox proportional hazards regression. All tests were two sided and performed at the 0.05 significance level.
Results Semiquantification of the Degrees of FF To compare our quantitative scoring method using the analytic software with semiquantitative scoring systems previously reported, each pathologist scored the profusion of FF semiquantitatively in three different ways, according to the Brompton,8 the Denver,7,9 and the Michigan10 FF scoring methods. When the score differed between the pathologists, a consensus score was reached. Interobserver variation was quantified using an unweighted coefficient of agreement to take into account the degree of disagreement on the semiquantitative categorical scales.
Clinical Characteristics, Laboratory Findings, Pulmonary Function Testing, and BAL Findings The clinical characteristics and laboratory findings of the patients with CVD-UIP and IPF/UIP are summarized in Table 1. There were no significant differences in clinical characteristics between CVDUIP and IPF/UIP, although the proportion of female patients with CVD-UIP tended to be higher than that with IPF/UIP, and pack-years of smoking in
Table 1—Clinical Characteristics, and Laboratory, Physiologic and BAL Fluid Cell Findings of the Study Populations* Variables Clinical characteristics Male/female gender, No. Age at biopsy, yr Pack-yr of smoking Symptom onset, mo Laboratory findings Serum lactate dehydrogenase, U/L Serum KL-6, U/L Serum surfactant protein D, U/L Physiologic FVC, % predicted FEV1, % predicted Resting Pao2, mm Hg BAL fluid cell analysis, % Lymphocytes Neutrophils Eosinophils
CVD-UIP
IPF/UIP
p Value, 2 Test
8/7 58.6 ⫾ 8.6 (42–70) 23.5 ⫾ 23.4 (0–50) 7.9 ⫾ 13.1 (1–43)
13/3 62.0 ⫾ 7.7 (50–79) 47.7 ⫾ 48.7 (0–177) 7.6 ⫾ 12.1 (3–48)
0.20 0.43 0.17 0.86
234 ⫾ 64 (174–377) 785 ⫾ 865 (254–2,320) 123 ⫾ 47 (89–207)
264 ⫾ 100 (180–526) 782 ⫾ 562 (303–1,400) 311 ⫾ 185 (100–447)
0.50 0.65 0.30
82.1 ⫾ 15.2 (66–111) 83.4 ⫾ 8.4 (73–100) 89.6 ⫾ 9.2 (76–103)
81.1 ⫾ 25.2 (45–128) 85.4 ⫾ 8.7 (72–100) 79.6 ⫾ 11.2 (64–102)
0.98 0.58 0.08
7.0 ⫾ 5.8 (0.5–20) 1.2 ⫾ 1.7 (0–39) 4.4 ⫾ 12.2 (0–7)
4.4 ⫾ 2.0 (0–26) 0.9 ⫾ 2.2 (0–8.1) 1.1 ⫾ 2.0 (0–8.6)
0.18 0.42 0.15
*Data are presented as mean ⫾ SD (range) unless otherwise indicated. 24
Original Research
IPF/UIP tended to be higher than those in CVDUIP. In laboratory findings, serum surfactant protein-D levels of IPF/UIP were higher than that of CVD-UIP, although the difference was not significant. In pulmonary function tests or BAL analysis, no significant difference was found between CVD-UIP and IPF/UIP. Quantitative Analysis of the Degrees of FF: the Quantitative %FF Score Using a CCD camera and the analytic software, we readily calculated the %FF in the sections of surgical lung biopsy (Fig 1). Even in IPF/UIP, the FF areas occupied only ⬍ 4% of the target fields, but our method enabled us to accurately quantify the extent of FF. The calculated proportion of the FF areas, the quantitative %FF score, was 1.67 ⫾ 0.90% (⫾ SD) in IPF/UIP patients. Interestingly, patients with CVD-UIP had much lower %FF score (0.39 ⫾ 0.24%, p ⬍ 0.0001, compared with patients with IPF/UIP) [Fig 2]. Intraobserver and Interobserver Correlation in the Quantitative Scoring Method Using the Pearson correlation coefficient, we examined the intraobserver and interobserver correlation of the quantitative %FF scores. When our quantitative scoring was repeated on different days by the same observer, a significantly high correlation was found between the first time scores and second time scores (r ⫽ 0.898, p ⬍ 0.0001; Fig 3, top, A). Additionally, when the scoring was done by different observers, a significantly high correlation was also found between them (r ⫽ 0.877, p ⬍ 0.0001; Fig 3, bottom, B).
Figure 3. Intraobserver and interobsever correlations of %FF score. Significantly high intraobserver and interobserver correlations were noted (top, A: r ⫽ 0.898, p ⬍ 0.0001; bottom, B: r ⫽ 0.877, p ⬍ 0.0001, respectively).
Correlation Between the Quantitative %FF Score and Clinical Data
Figure 2. Quantitative %FF scores in patients with CVD-UIP and in IPF/UIP. The quantitative %FF scores were significantly higher in IPF/UIP than in CVD-UIP (p ⬍ 0.0001). www.chestjournal.org
No significant correlation was found between the quantitative %FF score and FVC or resting Pao2 at surgical lung biopsy in CVD-UIP or IPF/UIP (data not shown). There was no correlation between the quantitative %FF score and BAL findings in CVDUIP (data not shown), while a significant correlation was found between the quantitative %FF score and the percentage of neutrophils (p ⫽ 0.026, r ⫽ 0.55) and eosinophils (p ⫽ 0.046, r ⫽ 0.50) in IPF/UIP. CHEST / 130 / 1 / JULY, 2006
25
Table 3—Multivariate Cox Proportional Hazards Models of Survival Adjusting for Age, FVC at Surgical Lung Biopsy, and IPF/UIP Diagnosis in All UIP Patients
Figure 4. Comparison of survival between CVD-UIP and IPF/UIP. Kaplan-Meier analysis shows that patients with CVDUIP had significantly better survival than those with IPF/UIP (log-rank test, p ⫽ 0.045).
Correlation Between the Quantitative %FF Score and Survival Comparison of survival between IPF/UIP and CVD-UIP is shown in Figure 4. Consistent with the study of Flaherty et al10 and our previous data,13 patients with CVD-UIP had significantly better survival than those with IPF/UIP (log-rank test, p ⫽ 0.045). Univariate Cox proportional hazards regression models showed that the quantitative %FF score (hazard ratio, 3.03; p ⫽ 0.004) and period from symptom onset (hazard ratio, 1.07; p ⫽ 0.019) were significant predictors of survival (Table 2). Multivariate Cox proportional hazards regression models, which were adjusted for age, percentage of predicted FVC at surgical lung biopsy, and IPF/UIP diagnosis, showed that the quantitative %FF score was a significant predictor of survival in all UIP patients (hazard ratio, 7.24, p ⫽ 0.010; Table 3). Period from symptom onset (hazard ratio, 1.07; p ⫽ 0.026) and resting Pao2 (hazard ratio, 0.91; p ⫽ 0.046) were also significant predictors (Table 3). Even in IPF patients
Table 2—Univariate Cox Proportional Hazards Models of Survival in All UIP Patients
Variables
Hazard Ratio
Age at biopsy, yr Male gender Symptom onset, mo IPF/UIP diagnosis FVC, % predicted Resting Pao2, mm Hg Quantitative FF score, %
1.04 0.45 1.07 4.72 0.98 0.93 3.03
26
95% Confidence Interval Lower
Upper
p Value
0.94 0.11 1.01 0.89 0.95 0.86 1.42
1.16 1.93 1.13 24.9 1.02 1.01 6.48
0.468 0.286 0.019 0.068 0.394 0.071 0.004
Variables
Hazard Ratio
Male gender Symptom onset, mo Resting Pao2, mm Hg Quantitative FF score, %
0.12 1.07 0.91 7.24
95% Confidence Interval Lower
Upper
p Value
0.02 1.01 0.83 1.59
0.79 1.14 1.00 32.9
0.028 0.026 0.046 0.010
alone, the quantitative %FF score was an independent predictor of survival (hazard ratio, 5.53; p ⫽ 0.032; Table 4). Comparison Between the Quantitative Scoring Method and Semiquantitative Scoring Methods We compared our quantitative scoring method with three previously reported semiquantitative scoring methods.7–10 With the use of coefficient, interobserver agreement was found to be moderate with the Michigan FF scoring method10 (unweighted coefficient, 0.53; p ⬍ 0.001), whereas the agreement was relatively low with the Denver FF scoring method7,9 (unweighted coefficient, 0.33; p ⬍ 0.001) and with the Brompton FF method8 (unweighted coefficient, 0.27; p ⬍ 0.001). The quantitative %FF score had a significant correlation with the Michigan FF score ( ⫽ 0.794, p ⬍ 0.001; Fig; 5, top, A), Denver FF score ( ⫽ 0.601, p ⫽ 0.001; Fig 5, center, B), and Brompton FF score ( ⫽ 0.608, p ⬍ 0.001; Fig 5, bottom, C), but patients with the same score assessed by the semiquantitative methods had widely varying scores assessed by our method. Discussion The main aim of this study was to develop a more quantitative procedure for scoring the degrees of FF Table 4 —Multivariate Cox Proportional Hazards Models of Survival Adjusting for FVC at Surgical Lung Biopsy in IPF/UIP Patients
Variables
Hazard Ratio
Age at biopsy, yr Male gender Symptom onset, mo Resting Pao2, mm Hg Quantitative FF score, %
1.04 0.28 1.06 0.94 5.53
95% Confidence Interval Lower
Upper
p Value
0.90 0.44 1.00 0.85 1.16
1.19 1.77 1.13 1.03 26.3
0.598 0.175 0.057 0.178 0.032
Original Research
Figure 5. Correlation between the quantitative scoring method and semiquantitative scoring methods. The quantitative %FF score had a significant correlation with Michigan FF score (top, A; ⫽ 0.794, p ⬍ 0.001), Denver FF score (center, B; ⫽ 0.601, p ⫽ 0.001), and Brompton FF score (bottom, C; ⫽ 0.608, p ⬍ 0.001), whereas patients with the same score assessed by the semiquantitative methods had widely varying scores as assessed by the quantitative method. www.chestjournal.org
than the semiquantitative methods previously reported. Our scoring method using a CCD camera and analytic software enabled us to accurately measure the %FF, and evaluate the degree of FF as the quantitative %FF scores. Additionally, we found that our quantitative %FF score was an independent prognostic factor in patients with UIP in the Cox proportional hazards regression model. Interestingly, patients with CVD-UIP had a significantly lower %FF score than those with IPF/UIP, which may be related to the better prognosis of CVD-UIP than IPF/UIP. In the present study, we found the interobserver variations were not small in the semiquantitative scoring methods, especially in the methods that had more categorical scales, such as the Denver (six scales) and the Brompton (seven scales) methods. In contrast, our quantitative scoring method had high interobserver correlation (r ⫽ 0.877). Moreover, the intraobserver correlation was also shown to be high (r ⫽ 0.898). In addition, our quantitative scoring method basically needs only one pathologist, who does not necessarily need to specialize in pulmonary pathology, and the pathologist could easily calculate the proportion of FF using a CCD camera and the analytic software. Taken together, these data suggest that our quantitative scoring method, which does not need two or more of pulmonary pathologists, is more objective and reproducible than the semiquantitative scoring methods. We examined whether the quantitative %FF score defined by our scoring method had clinical significance in UIP. Most notably, we found that this score was an independent predictor of survival in all UIP patients as well as IPF patients in the Cox proportional hazards regression model; higher FF score was associated with poor prognosis. In IPF, conflicting reports using semiquantitative scoring systems exist concerning the significance of FF score for survival in the proportional hazards model. Flaherty et al10 showed that the Michigan FF score was not a significant predictor of survival, while Nicholson et al8 demonstrated that the Brompton FF score was independently linked to survival. This may reflect the difference in categorization of FF among the scoring systems. In addition, we found no significant correlation between the quantitative %FF score and FVC or resting Pao2 at surgical biopsy in CVD-UIP or IPF/UIP, indicating that this FF score was not associated with baseline pulmonary function. Interestingly, Nicholson et al8 demonstrated that increasing semiquantitative FF score was independently associated with greater declines at both 6 months and 12 months in FVC and carbon monoxide diffusing capacity in IPF.8 Unfortunately, we did not examine the trends of pulmonary function tests at CHEST / 130 / 1 / JULY, 2006
27
such intervals. Thus, further studies will be needed to clarify whether our quantitative %FF score is related to functional decline. There was no correlation between the quantitative %FF score and BAL findings in CVD-UIP, while the quantitative %FF score was significantly correlated with the percentage of neutrophils (p ⫽ 0.026, r ⫽ 0.55), and eosinophils (p ⫽ 0.046, r ⫽ 0.50) in IPF. Interestingly, previous studies14 –20 demonstrated that higher percentages of neutrophils or eosinophils were associated with more extensive disease and poor prognosis in IPF. Although the correlation rates were not high, there might be some association between extent of FF and the number of BAL neutrophils and eosinophils. Our previous study13 showed CVD-UIP had more favorable prognosis than IPF. In the present study, we found that patients with CVD-UIP had significantly lower %FF scores and a better prognosis than those with IPF. Consistent with our results, Flaherty et al10 showed that the degree of FF was smaller in CVD-UIP than in IPF/UIP, as assessed by their semiquantitative scoring method, which might be associated with the favorable prognosis of CVD-UIP. The reason for the variation in the degree of FF between IPF and CVD-UIP is unclear. The etiology is thought to differ between IPF and CVD-UIP: the formation of FF might be induced by the injury of the alveolar epithelium in IPF,1,21 while endothelial injury probably causes FF formation in CVDUIP.22–24 This etiologic difference might be responsible for the variation in the degree of FF. In addition, the present data demonstrated that the quantitative %FF score was an independent predictor of prognosis in the multivariate Cox proportional hazards regression model adjusted by IPF/UIP diagnosis, suggesting that FF play a pivotal role in determining the prognosis of patients having UIP, regardless of etiology. The main weakness of the present study is its retrospective design and the relatively small number of patients. In addition, the survival rate of our IPF patients seemed to be relatively higher than that of typical IPF patients. Approximately 38% of our IPF patients came to the hospital because of interstitial abnormalities on the chest radiograph of annual medical checkups without respiratory symptoms, suggesting that our study population included early stage IPF patients. This may account for the relatively better prognosis of our IPF patients. To confirm our findings, larger prospective studies will be required. In conclusion, our quantitative FF scoring method using a CCD camera and analytic software, which is easily performed, is more objective and quantitative than previously reported semiquantitative scoring 28
methods, and provides accurate information about prognosis in patients with UIP.
References 1 Selman M, Pardo A. The epithelial/fibroblastic pathway in the pathogenesis of idiopathic pulmonary fibrosis. Am J Respir Cell Mol Biol 2003; 29:S93–S97 2 Phan SH. Fibroblast phenotypes in pulmonary fibrosis. Am J Respir Cell Mol Biol 2003; 29:S87–S92 3 Phan SH. The myofibroblast in pulmonary fibrosis. Chest 2002; 122:286S–289S 4 Kuhn C III, Boldt J, King TE Jr, et al. An immunohistochemical study of architectural remodeling and connective tissue synthesis in pulmonary fibrosis. Am Rev Respir Dis 1989; 140:1693–1703 5 Kuhn C, McDonald JA. The roles of the myofibroblast in idiopathic pulmonary fibrosis: ultrastructural and immunohistochemical features of sites of active extracellular matrix synthesis. Am J Pathol 1991; 138:1257–1265 6 Pardo A, Selman M. Molecular mechanisms of pulmonary fibrosis. Front Biosci 2002; 7:d1743– d1761 7 King TE Jr, Schwarz MI, Brown K, et al. Idiopathic pulmonary fibrosis: relationship between histopathologic features and mortality. Am J Respir Crit Care Med 2001; 164:1025– 1032 8 Nicholson AG, Fulford LG, Colby TV, et al. The relationship between individual histologic features and disease progression in idiopathic pulmonary fibrosis. Am J Respir Crit Care Med 2002; 166:173–177 9 Cherniack RM, Colby TV, Flint A, et al. Quantitative assessment of lung pathology in idiopathic pulmonary fibrosis: the BAL Cooperative Group Steering Committee. Am Rev Respir Dis 1991; 144:892–900 10 Flaherty KR, Colby TV, Travis WD, et al. Fibroblastic foci in usual interstitial pneumonia: idiopathic versus collagen vascular disease. Am J Respir Crit Care Med 2003; 167:1410 – 1415 11 Katzenstein AL, Myers JL. Idiopathic pulmonary fibrosis: clinical relevance of pathologic classification. Am J Respir Crit Care Med 1998; 157:1301–1315 12 American Thoracic Society/European Respiratory Society International Multidisciplinary Consensus Classification of the Idiopathic Interstitial Pneumonias. This joint statement of the American Thoracic Society (ATS), and the European Respiratory Society (ERS) was adopted by the ATS board of directors, June 2001 and by the ERS Executive Committee, June 2001. Am J Respir Crit Care Med 2002; 165:277–304 13 Nakamura Y, Chida K, Suda T, et al. Nonspecific interstitial pneumonia in collagen vascular diseases: comparison of the clinical characteristics and prognostic significance with usual interstitial pneumonia. Sarcoidosis Vasc Diffuse Lung Dis 2003; 20:235–241 14 The diagnosis, assessment and treatment of diffuse parenchymal lung disease in adults. Introduction. Thorax 1999; 54(Suppl)1:S1–S14 15 Haslam PL, Turton CW, Lukoszek A, et al. Bronchoalveolar lavage fluid cell counts in cryptogenic fibrosing alveolitis and their relation to therapy. Thorax 1980; 35:328 –339 16 Rudd RM, Haslam PL, Turner-Warwick M. Cryptogenic fibrosing alveolitis: relationships of pulmonary physiology and bronchoalveolar lavage to response to treatment and prognosis. Am Rev Respir Dis 1981; 124:1– 8 17 Turner-Warwick M, Haslam PL. The value of serial bronchoalveolar lavages in assessing the clinical progress of paOriginal Research
18
19 20
21
tients with cryptogenic fibrosing alveolitis. Am Rev Respir Dis 1987; 135:26 –34 Watters LC, Schwarz MI, Cherniack RM, et al. Idiopathic pulmonary fibrosis: pretreatment bronchoalveolar lavage cellular constituents and their relationships with lung histopathology and clinical response to therapy. Am Rev Respir Dis 1987; 135:696 –704 Peterson MW, Monick M, Hunninghake GW. Prognostic role of eosinophils in pulmonary fibrosis. Chest 1987; 92:51–56 Hallgren R, Bjermer L, Lundgren R, et al. The eosinophil component of the alveolitis in idiopathic pulmonary fibrosis: signs of eosinophil activation in the lung are related to impaired lung function. Am Rev Respir Dis 1989; 139:373– 377 Selman M, King TE, Pardo A. Idiopathic pulmonary fibrosis:
www.chestjournal.org
prevailing and evolving hypotheses about its pathogenesis and implications for therapy. Ann Intern Med 2001; 134:136 –151 22 Hammar SP, Winterbauer RH, Bockus D, et al. Endothelial cell damage and tubuloreticular structures in interstitial lung disease associated with collagen vascular disease and viral pneumonia. Am Rev Respir Dis 1983; 127:77– 84 23 Wusirika R, Ferri C, Marin M, et al. The assessment of anti-endothelial cell antibodies in scleroderma-associated pulmonary fibrosis: a study of indirect immunofluorescent and western blot analysis in 49 patients with scleroderma. Am J Clin Pathol 2003; 120:596 – 606 24 Ihn H, Sato S, Fujimoto M, et al. Characterization of autoantibodies to endothelial cells in systemic sclerosis (SSc): association with pulmonary fibrosis. Clin Exp Immunol 2000; 119:203–209
CHEST / 130 / 1 / JULY, 2006
29