| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Thoracic Imaging |
1 From the Dept of Radiology, Hammersmith Hosp, London, England (S.J.C.); Dept of Radiology, King’s College Hosp, London, England (S.R.D.); Depts of Radiology (S.L.S.M., D.M.H., M.B.R.) and Pathology (A.G.N.), and Interstitial Lung Disease Unit (A.U.W., R.M.d.B.), Royal Brompton Hosp, Sydney St, London SW3 6NP, England; Depts of Radiology (R.I.T.) and Respiratory Medicine (A.W.M.), Sir Charles Gairdner Hosp, Perth, West Australia, Australia; Dept of Respiratory Medicine, Middlemore Hosp and Univ of Auckland, New Zealand (P.S.); and Dept of Laboratory Medicine/Pathology, Mayo Clinic, Scottsdale, Ariz (T.V.C.). Received Jun 4, 2002; revision requested Aug 8; final revision received Apr 30, 2003; accepted Jun 16. Address correspondence to D.M.H. (e-mail: d.hansell@rbh.nthames.nhs.uk).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Consecutive patients with a diagnosis of IPF (n = 212) or asbestosis (n = 74) were included. The relationships derived from the initial comparison were tested in a separate group of biopsy-proved UIP (n = 30) and NSIP (n = 23) cases. Two observers independently scored thin-section CT images for extent, distribution, and coarseness of fibrosis; proportion of ground-glass opacification; severity of traction bronchiectasis; and extent of emphysema.
RESULTS: After controlling for extent of fibrosis, patients with asbestosis had coarser fibrosis than those with IPF (odds ratio, 1.52; 95% CI: 1.25, 1.84; P < .001). Compared with the biopsy-proved cases, the asbestosis cases involved coarser fibrosis (after controlling for disease extent) than the NSIP cases (odds ratio, 2.48; 95% CI: 1.49, 4.11; P < .001) but fibrosis similar to that in the UIP cases. A basal and subpleural distribution of disease was usual in all subgroups but significantly more prevalent (P, <.01 to .001) with asbestosis than with UIP or NSIP.
CONCLUSION: The thin-section CT pattern of asbestosis closely resembles that of biopsy-proved UIP and differs markedly from that of biopsy-proved NSIP.
© RSNA, 2003
Index terms: Asbestos • Lung, CT, 60.12111, 60.12115, 60.12118 • Lung, diseases, 60.213, 60.6113, 60.773, 60.792, 60.795 • Lung, fibrosis, 60.6113, 60.792 • Pneumonia, interstitial with fibrosis, 60.213, 60.795
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
In general, emphasis has been placed on the presence or absence of pleural involvement (ie, pleural plaques or diffuse pleural thickening) in discriminating between asbestos-induced pulmonary fibrosis and IPF (2). However, relying on pleural disease alone may be an oversimplified way to differentiate these two entities: An individual previously exposed to asbestos is as susceptible to the same non–asbestos-induced fibrotic lung diseases (including IPF) as the general population (2), and diffuse pleural thickening may be due to other causes.
Histopathologically, asbestosis and IPF have similar features. Both individuals with usual interstitial pneumonia (UIP) and those with nonspecific interstitial pneumonia (NSIP) may have clinical findings of IPF. With asbestosis, the fibrosis is typically patchy and closely resembles that with UIP (3). However, despite histopathologic similarities, the etiologies and biologic behaviors of the two conditions are different (2). Important issues, such as differences in prognosis and eligibility for legal or state compensation for asbestosis, highlight the need for accurate differentiation whenever possible (2).
The aim of the present study was to identify differences, if any, between the thin-section CT morphologic features of asbestosis and those of IPF and to test the findings in a subset of histopathologically proved cases of UIP and NSIP.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
All cases (n = 74) were diagnosed as asbestosis by an experienced occupational chest physician on the basis of clinical and radiologic criteria, an unequivocal history of substantial asbestos exposure, and an appropriate time between exposure and onset of symptoms, according to American Thoracic Society criteria for the diagnosis of asbestosis (4).
Consecutive patients who were referred to the Interstitial Lung Disease Unit of Royal Brompton Hospital between 1991 and 1997 were identified from the hospital patient information system. Two hundred twelve patients with a diagnosis of IPF were included in this study. Patients with a diagnosis of IPF met the following clinical, physiologic, and radiographic criteria: (a) widespread persistent inspiratory crackles at auscultation, (b) a restrictive ventilatory defect and/or an isolated depression of gas transfer, (c) the presence of widespread persistent bilateral radiographic abnormalities consistent with IPF, and (d) no history of exposure to fibrogenic dusts or drugs (5). Patients with pulmonary fibrosis secondary to a connective tissue disease were excluded.
On the basis of findings in a cohort examined in a previous study (6), consecutive patients who received a lung biopsy–based (open or video-assisted thoracoscopic) diagnosis of UIP or NSIP between January 1990 and March 2000 were identified and examined separately (7,8). Of these patients, 53 who underwent thin-section CT 12 months before or after biopsy (median interval, 2 months) and met the described clinical criteria for the diagnosis of IPF were included in the study. Thirty of these patients had UIP and 23 NSIP. Patients with a connective tissue disease were again excluded.
One of several authors (S.J.C., S.R.D., Y.C.G.L., S.L.S.M.) obtained demographic data (ie, age and sex) and smoking histories by independently reviewing patient records. Institutional review board approval and informed patient consent for retrospective review of patient records and images were not required at either center.
Histopathologic Diagnosis
All biopsy results were independently reviewed by two pathologists (A.G.N., T.V.C.). For the histopathologic diagnosis of UIP in 30 patients, the following criteria had to be met: evidence of temporal heterogeneity with nonuniform and variable interstitial changes, including intermingled zones of established interstitial fibrosis, inflammation, fibroblastic activity, honeycomb change, and normal lung parenchyma coexisting in variable proportions (9). NSIP was diagnosed when the fibrosis was either patchy or diffuse (primarily the latter), but the pattern of lung injury remained temporally uniform (10). When there was disagreement, the final diagnosis was reached by consensus.
The 23 patients with a histopathologic diagnosis of NSIP were further separated into either predominantly cellular NSIP (n = 5) or predominantly fibrotic NSIP (n = 18) subgroups. Cellular NSIP was characterized by the presence of a mild to moderate chronic inflammatory cellular infiltrate within the interstitium, with either minimal or no interstitial fibrosis (10). Fibrotic NSIP was characterized by the presence of temporally homogeneous interstitial fibrosis, with or without derangement of the lung architecture, and an accompanying chronic inflammatory cellular infiltrate of varying intensity (10). In cases of disagreement, the final histopathologic diagnosis and grade of NSIP were determined by consensus.
CT Scanning Protocol
CT scanning was performed by using an electron-beam scanner (Imatron, San Francisco, Calif) or helical CT scanners (Somatom Plus, Siemens, Erlangen, Germany; or Toshiba Xpress, Toshiba, Tokyo, Japan), depending on the referral center. Thin (1–3-mm) sections were obtained at 10–30-mm intervals in all patients, who were scanned in a supine position, prone position, or both. In the 212 patients with IPF, 1.5-mm sections were obtained in 121 patients and 3.0-mm sections were obtained in 91. In a proportion of the IPF cases, prone CT images were obtained; however, only the supine images (n = 212) were evaluated by the observers. In all 53 patients with biopsy-proved UIP or NSIP, 1.5-mm sections were obtained. In the 74 patients with asbestosis, 1.0-mm sections were obtained in 38 patients; 1.5-mm sections, in 30; and 3.0-mm sections, in six. In 45 of these patients, prone images were obtained.
All CT images were obtained at full suspended inspiration. The images were reconstructed with a high-spatial-frequency (ie, bone) algorithm and photographed at wide window settings (window level, -300 to -655 HU; window width, 1,500–1,850 HU), and all images were evaluated on film hard copies.
Image Evaluation
CT images were evaluated independently by two thoracic radiologists (D.M.H., M.B.R.), who were blinded to the clinical findings and pulmonary function data. Disagreements about the presence or absence of fibrosis (n = 0) and emphysema (nine regarding asbestosis, two regarding IPF) were resolved by consensus. The images were scored at five levels: the great vessels, the aortic arch, the carina, the right inferior pulmonary vein, and halfway between the right inferior pulmonary vein and the caudal limit of the left costophrenic angle. The images were scored as follows:
1. At each level, the overall extent of pulmonary fibrosis, including the reticular pattern and ground-glass opacification with traction bronchiectasis, was visually estimated to the nearest 5%. The mean value (range, 0.5% to 100%) was considered the extent of fibrosis, irrespective of the predominant pattern.
2. At each level, the relative percentages of ground-glass opacification and the reticular pattern were recorded, and the extent of each was calculated as the proportion of each pattern multiplied by the overall disease extent. The proportion of ground-glass opacification was computed. Ground-glass opacification was defined as a hazy increase in attenuation, with preservation of the bronchial and vascular margins. A reticular pattern was defined as that of innumerable interlacing line shadows—either fine, intermediate, or coarse—with associated distortion of the lung architecture (11).
3. At each level, the coarseness of fibrosis was assigned a grade of 0, indicating ground-glass opacification with no reticular element or cysts; 1, indicating predominantly fine intralobular fibrosis without cysts; 2, indicating a predominantly microcystic reticular pattern (ie, definable airspaces smaller than 3 mm in diameter); or 3, indicating a predominantly macrocystic reticular pattern (ie, airspaces larger than 3 mm in diameter). The scores for the five regions were summed to provide an overall coarseness score (range, 0–15). In cases in which no disease was seen on one or more CT sections, the coarseness score was adjusted proportionately to a five-level score to prevent spurious reductions in the coarseness score due to a localized distribution of disease. For example, in a patient with disease in only three of five sections, a coarseness score of 5 would be adjusted by 5/3 to 8.3.
4. At each level, the extent of emphysema was visually estimated to the nearest 5%. The mean value (range, 1%–100%) was considered the extent of emphysema. Emphysema was defined as an area of decreased attenuation, usually without discrete walls and with a nonuniform distribution causing permeative destruction of the lung parenchyma (11).
5. At each level, the severity of bronchial dilatation within a reticular pattern or ground-glass opacification (traction bronchiectasis or bronchiolectasis) was graded by comparing the diameter of the airway with the diameter of the adjacent pulmonary artery: 0 meant no; 1, mild; and 2, severe dilatation. The five scores were summed to yield an overall traction bronchiectasis score (range, 0–10).
6. In the cases of asbestosis and biopsy-proved UIP or NSIP, all CT sections were reviewed and the distribution of fibrosis was determined by consensus to be random, subpleural, bronchocentric, and/or basal.
7. In cases of asbestosis and IPF, the presence of pleural plaques and diffuse pleural thickening was recorded. Diffuse pleural thickening was defined as a continuous sheet of pleural thickening, whereas pleural plaques were defined as circumscribed areas of pleural thickening with well-demarcated edges (12).
Statistical Analysis
Results are expressed as means with SDs or medians with ranges (for non–normally distributed variables). Group comparisons of continuous variables were performed with data analysis software (Stata; Computing Resource Centre, Santa Monica, Calif) by using the Student t or Wilcoxon rank sum test, depending on the data distribution. Differences in proportions were tested by using 
2 statistics. Univariate correlations were examined by using Spearman correlation coefficients (ie, rs values). Observer variation was evaluated by using the weighted 
coefficient for semicategorical variables (13) or the single-determination SD for quantitative data (14).
Independent differences in CT features between diseases were identified by using stepwise logistic regression. The extent of fibrosis, extent of emphysema, coarseness of fibrosis, proportion of ground-glass opacification, severity of traction bronchiectasis, and distribution of fibrosis were included as covariates. Variables were retained in logistic regression models if they contributed to the explanatory power of the regression equation (P < .10).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
coefficient, 0.57).
|
coefficient for interobserver agreement of coarseness of fibrosis was 0.78.
|
As shown in Table 2, fibrosis was more extensive in the IPF group (55%; range, 5%–96%) than in the asbestosis group (26%; range, 0.5%–90.5%) (P < .001) (Table 2). The prevalence of emphysema did not differ significantly between the asbestosis and IPF groups, but the extent of emphysema was significantly greater in the asbestosis group (P < .001). Traction bronchiectasis scores were higher in the IPF group (before adjustment for extent of fibrosis in multivariate analysis).
Relationships between the extent of fibrosis and other CT features are summarized in Table 3. Strong correlations were seen between traction bronchiectasis scores and the extent of fibrosis in the asbestosis (rs = 0.81, P < .001) and IPF (rs = 0.72, P < .001) groups. Increasing coarseness scores and decreasing proportions of ground-glass opacification were more closely linked to increasing disease extent in the asbestosis group than in the IPF group.
|
Asbestosis versus Biopsy-proved UIP
At univariate analysis, fibrosis was more extensive in the UIP group than in the asbestosis group (46.6% ± 22.3 vs 28.0% ± 20.6, P < .001). No single CT feature consistently enabled the differentiation between UIP and asbestosis. Coarseness scores and proportions of ground-glass opacification did not differ significantly between the UIP and asbestosis groups (before adjustment for disease extent in multivariate analysis). The distribution of fibrosis was basal and subpleural in the majority of patients, but it was more often basal in the asbestosis group—in 72 (97%) of 74 patients versus 25 (83%) of 30 in the UIP group (P < .01)—and more often subpleural in the asbestosis group—in 67 (91%) patients versus 20 (67%) in the UIP group (P < .005). A bronchocentric distribution was seen in only two cases of UIP and was not seen with asbestosis.
Examination of the logistic regression model revealed that after adjustment for age, sex, and extent of fibrosis, none of the CT features (ie, coarseness of fibrosis, proportion of ground glass opacification, extent of emphysema, traction bronchiectasis scores, and presence of a subpleural or basal distribution) differed significantly between the patients with asbestosis and those with biopsy-proved UIP.
Asbestosis versus Biopsy-proved NSIP
At univariate analysis, fibrosis was marginally more extensive in the NSIP group than in the asbestosis group (37.9% ± 23.4 vs 28.0% ± 20.6, P = .05). There were significant differences in most CT features between the asbestosis and NSIP groups. The asbestosis cases, as compared with the NSIP cases, were characterized by higher coarseness scores (P < .001) (Figure), a lower proportion of ground-glass opacification (P < .001), and higher likelihoods of basal (in 72 [7%] of 74 vs 16 [70%] of 23 patients, P < .001) and subpleural (in 67 [91%] of 74 vs 14 [61%] of 23 patients, P < .001) distributions. A bronchocentric distribution was not seen with NSIP.
|
|
Univariate and multivariate logistic analyses were repeated in the comparison of the asbestosis and fibrotic NSIP groups, with five cases of cellular NSIP excluded. All findings at both univariate analysis and multivariate analysis remained statistically significant.
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Despite the convention that patients with IPF will have UIP at histopathologic examination of lung biopsy specimens (15), the clinical diagnosis of IPF (ie, that based on clinical criteria and compatible radiographic features) necessarily includes NSIP as well as UIP. Therefore, a difference in the coarseness of fibrosis between asbestosis and NSIP groups will also extend to the combined IPF group, but not when the analysis is confined to biopsy-proved UIP.
Our study findings confirm that asbestosis resembles UIP but is strikingly different from NSIP. Our observations are in agreement with histopathologic data in that asbestosis and IPF have been reported to have similar appearances, with the exception that asbestos or ferruginuous bodies are present with asbestosis (3,16). Histopathologically, the fibrosis with asbestosis is typically patchy and has been described as closely resembling the UIP subtype that characterizes IPF (3).
Virtually all of the patients with asbestosis in our study had a subpleural and basal distribution of fibrosis. This finding is in agreement with the observations made by Akira et al regarding patients with early asbestosis; they found that the small subpleural dots that became more confluent on subsequent CT images correlated histopathologically with subpleural fibrosis (17). The distribution of abnormalities on CT images is probably due to the fact that asbestos fibers are deposited in the small airways in the lung periphery, where they first elicit an inflammatory response.
In the current study, the patients with asbestosis also had less extensive fibrosis than did those with IPF. One explanation for this discrepancy is some form of selection bias. The asbestosis cases were selected on the basis of CT reports, whereas the IPF cases were selected by using a combination of clinical, radiographic, and physiologic criteria, with which the presence of more extensive fibrosis might be required. The discrepancy also may be due to the earlier detection of disease in individuals who are exposed to asbestos because of closer radiographic and clinical surveillance. Increased patient awareness of symptoms, due to potential legal or state compensation for asbestosis, may be another factor. However, the observed differences in the quality of fibrosis (ie, fine vs coarse) are robust despite these potential selection biases according to our results, which were controlled for the extent of fibrosis.
Gaensler and Carrington first studied the correlation between pathologic, radiographic, and physiologic features in subjects with interstitial lung disease, including IPF and asbestosis, but a direct comparison of structure and function between the two disease entities was not made (18). Cookson et al compared the radiographic features and functional indexes of asbestosis and IPF and found a greater correlation between the transfer factor of the lungs and the radiographic profusion of small parenchymal opacities in subjects with asbestosis, as compared with this correlation in subjects with IPF (after exclusion of subjects with pleural thickening in the asbestosis cases) (19).
Similar to us in the present study, Cookson et al found that subjects with IPF had more extensive disease than did those with asbestosis, as measured according to the radiographic extent of small round opacities. However, comparison of neither the quality of fibrosis nor the coexistence of emphysema could be performed in detail with chest radiographs.
al-Jarad et al compared the distribution and configuration of lung opacities in asbestosis and IPF groups at thin-section CT (1). Ground-glass opacification was more frequent with IPF than with asbestosis (although traction bronchiectasis was not recorded); this finding is similar to our finding of more extensive ground-glass opacification with NSIP. al-Jarad et al also reported that there was a qualitative impression of greater distortion of the lung architecture owing to fibrosis and emphysema in subjects with IPF as compared with the impression of lung architecture distortion in those with asbestosis, but no quantitative data were given to support this subjective observation.
The al-Jarad study involved a small number of subjects (18 with IPF and 24 with asbestosis), and only the frequencies of CT features with each condition were analyzed (1). Another important difference is that al-Jarad et al concentrated on the frequencies of specific CT features that might allow differentiation between the two conditions instead of quantifying the extent of individual CT patterns as we did in the current study. The quality of fibrosis was more specifically categorized in our study: al Jarad et al grouped subjects into those with a crescentic fine, reticular pattern with small cysts and those with ground-glass opacification. No categories for coarser pattern and traction bronchiectasis existed, and, thus, direct comparisons between the two studies are impossible.
Thin-section CT has advantages over other methods of comparing populations with different interstitial lung diseases (16). Thin-section CT enables a global assessment of the lung, unlike histopathologic analysis of open lung biopsy specimens, in which a relatively small amount of lung parenchyma is sampled. Furthermore, the semiquantitative scoring of disease patterns with CT involves considerably less observer variation than does grading of histopathologic features (20,21). The disadvantage of using pulmonary function indexes to control for the disease extent between these two conditions is that coexisting diseases (eg, pleural thickening with asbestosis) cannot be taken into account.
However, there are limitations of using thin-section CT to compare different disease entities that are common to our study and others. These include problems in making histopathologic inferences from CT appearances. For example, in the context of IPF, the histopathologic correlate of ground-glass opacification may represent either partial filling of airspaces with macrophages and alveolar septal inflammation (22) or thickening of alveolar walls owing to fibrosis or edema (23). In a carefully constructed study by Hartley et al in which an objective computer-assisted method was used, the attenuation of the lung parenchyma was examined with thin-section CT in 24 subjects with IPF and in 60 who had been heavily exposed to asbestos (24). Gray-scale histograms were obtained in each group, and the subjects with IPF had a flatter gray scale, which was shifted to the right. This indicated increased parenchymal attenuation, which could be analogous to the finer fibrosis seen in the subjects with IPF and the greater proportions of ground-glass opacification seen in those with NSIP in our study. In the study by Hartley et al, textural analysis of the lung parenchyma was not performed, so a direct comparison between their results and ours is difficult (24).
In summary, the patients with asbestosis had a coarser pattern of fibrosis than did the subgroup of patients with histopathologically proved NSIP, but there was no statistically significant difference in coarseness between the asbestosis and biopsy-proved UIP groups. Thus, our study findings indicate that asbestosis closely resembles UIP but is strikingly different from NSIP.
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, D.M.H.; study concepts and design, S.J.C., D.M.H., A.U.W.; literature research, S.J.C.; clinical studies, A.U.W., A.W.M.; data acquisition, S.J.C., S.R.D., S.L.S.M.; data analysis/interpretation, all authors; statistical analysis, A.U.W.; manuscript preparation, editing, and final version approval, S.J.C., D.M.H., A.U.W.; manuscript definition of intellectual content, D.M.H., A.U.W.; manuscript revision/review, all authors
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  H. Arakawa, K. Fujimoto, K. Honma, N. Suganuma, H. Morikubo, Y. Saito, H. Shida, and Y. Kaji Progression from Near-Normal to End-Stage Lungs in Chronic Interstitial Pneumonia Related to Silica Exposure: Long-Term CT Observations Am. J. Roentgenol., October 1, 2008; 191(4): 1040 - 1045. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 H. Arakawa, T. Johkoh, K. Honma, Y. Saito, Y. Fukushima, H. Shida, and N. Suganuma Chronic Interstitial Pneumonia in Silicosis and Mix-Dust Pneumoconiosis: Its Prevalence and Comparison of CT Findings With Idiopathic Pulmonary Fibrosis Chest, June 1, 2007; 131(6): 1870 - 1876. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 J. Xu, A. L. Mora, J. LaVoy, K. L. Brigham, and M. Rojas Increased bleomycin-induced lung injury in mice deficient in the transcription factor T-bet Am J Physiol Lung Cell Mol Physiol, October 1, 2006; 291(4): L658 - L667. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. J. Yun, P. Y. Lee, and A. N. Gerber Integrating systems biology and medical imaging: understanding disease distribution in the lung model. Am. J. Roentgenol., April 1, 2006; 186(4): 925 - 930. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Misumi and D. A. Lynch Idiopathic pulmonary fibrosis/usual interstitial pneumonia: imaging diagnosis, spectrum of abnormalities, and temporal progression. Proceedings of the ATS, January 1, 2006; 3(4): 307 - 314. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Chong, K. S. Lee, M. J. Chung, J. Han, O J. Kwon, and T. S. Kim Pneumoconiosis: Comparison of Imaging and Pathologic Findings RadioGraphics, January 1, 2006; 26(1): 59 - 77. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. A. Souza, N. L. Muller, J. Flint, J. L. Wright, and A. Churg Idiopathic Pulmonary Fibrosis: Spectrum of High-Resolution CT Findings Am. J. Roentgenol., December 1, 2005; 185(6): 1531 - 1539. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
