| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
Thoracic Imaging |
1 From the Department of Radiology, Charité, Campus Virchow-Klinikum, Humboldt-University, Augustenburger Platz 1, 13353 Berlin, Germany (F.D.K., T.W., R.F.); and Department of Cardiothoracic and Vascular Surgery, German Heart Institute, Berlin, Germany (R.E., R.H.). From the 2001 RSNA scientific assembly. Received August 14, 2001; revision requested October 10; revision received January 9, 2002; accepted February 20. Address correspondence to F.D.K. (e-mail: friedrich.knollmann@charite.de).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Forty-nine lung transplant recipients were examined at least 8 months after surgery with spirometrically gated thin-section CT of the lung. In addition to visual signs of small-airway disease at CT, mean lung attenuation and the SD were numerically determined and compared with the results of lung function testing at the time of the CT examination and 1 year later by using factorial analysis of variance.
RESULTS: Mean lung attenuation was significantly lower in patients who developed bronchiolitis obliterans syndrome within 1 year after the CT study (-837 HU ± 3) than in patients with persistent normal lung function (-812 HU ± 3, P < .001). With an optimal threshold, sensitivity was 69%, specificity was 71%, and accuracy was 84%. Visual analysis did not significantly contribute to the prognostic power of CT.
CONCLUSION: Spirometrically gated CT measurements of lung attenuation can be used to predict the onset of bronchiolitis obliterans syndrome after lung transplantation.
© RSNA, 2002
Index terms: Bronchiolitis obliterans, 60.2191 • Computed tomography (CT), quantitative, 60.1211 • Computed tomography (CT), thin-section, 60.12118 • Lung, transplantation, 60.458
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
The purpose of our study, therefore, was to assess the potential use of spirometrically gated lung CT findings in the diagnosis of BOS after lung transplantation.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
All patients were examined with a spirometrically gated CT scanner (Somatom Plus 4; Siemens, Forchheim, Germany) at 20%, 50%, and 80% of their vital capacity (VC). No correction was used for obtaining VC in patients in the supine position, because only relative VC data were used as a guideline for spirometric gating. All lung function tests used for the diagnosis of bronchiolitis obliterans were performed with a standard spirometer (Master Screen; Jäger, Würzburg, Germany) and with the patient in the upright position. Scans were acquired at the level of the carina, 5 cm above it, and 5 cm below it. All thin-section CT scans were obtained with 1-mm section thickness, 1-second scanning time, 140-kV voltage, and 146-mA tube current and were reconstructed with a high-resolution adult body kernel. The reconstruction algorithm was AB 82.
Standard postprocessing software (Pulmo-CT; Siemens) was used to determine the mean attenuation of the lung transplant and the SD within each scan. With this software, lung parenchyma was semiautomatically traced, and the outline was manually corrected where necessary. Manual correction was performed by one of the authors (T.W.). Central airways were automatically excluded, and also were excluded with manual correction where indicated. Numeric analysis of lung attenuation included a histogram of attenuation frequency distribution, and determinations of mean lung attenuation and the SD included attenuation values between -1,000 and -500 HU. For bilateral lung transplants, attenuation data from both lungs were pooled.
In double lung transplants, both grafts were from the same donor and were harvested and grafted simultaneously. Since bronchiolitis obliterans is an immunologically mediated process, the concept of chronic rejection implies no differences between the two grafts. Also, lung function test results serve as the parameters that are used to define BOS, and lung function test results cannot be obtained in the two lungs separately. For all statistical tests used, as a condition before the test, each numeric value had to be independent. Because of the systemic nature of bronchiolitis obliterans, this precondition would not be fulfilled if both lungs in one recipient were introduced into the analysis separately.
For visual analysis, all scans were reviewed by a thoracic radiologist (F.D.K.) to determine the presence of bronchiectasis, air trapping, mosaic attenuation, and other abnormalities. Air trapping was defined as the presence of areas of less than normal increases in parenchymal attenuation on expiratory scans (5). Mosaic attenuation was defined as the presence of parenchymal inhomogeneity on inspiratory scans (5). For visual evaluation, a window center of -600 HU and a window width of 1,600 HU were selected in all instances. In one patient, the expiratory scans were excluded from the analysis because of respiratory artifacts.
Pulmonary function test results that were obtained within 2 weeks of the CT examination were used as a reference for diagnosing BOS. Lung function data were obtained retrospectively from the closest possible time either before or after CT, with a maximum of 2 weeks either before or after the CT examination. We also obtained the results of pulmonary function testing 1 year after CT to determine if CT findings could have been used to predict the course of lung function. Pulmonary function was classified according to values for the forced expiratory volume in 1 second (FEV1), as defined by the International Society for Heart and Lung Transplantation (6). According to that definition, BOS is diagnosed if the FEV1 decreases below 80% of its baseline value. The baseline value is defined as the average of the two highest consecutive measurements obtained 3–6 weeks apart. Since this value may increase over time in some patients, baseline lung function was determined separately for the time of the CT examination and for the follow-up visit 1 year later. Details of this staging system have been published by the International Society of Heart and Lung Transplantation (6) and were applied throughout our investigation.
For statistical analysis, factorial analysis of variance was used to compare mean lung attenuation and the SD at the three image levels and the three levels of inspiration for patients with and without BOS, at both the time of the examination and 1 year later. Data from the different lung levels and inspiratory levels were compared by means of analysis of variance by using anatomic level and level of inspiration as factors. Thus, the attenuation measurements from different levels were not pooled.
Results were recorded as the mean and standard error of the mean. Lung attenuation and the SD (homogeneity) were determined at three levels of inspiration and at three anatomic levels, which accounts for nine images per patient and 18 numeric results for each patient. Pairwise comparisons were determined with the Scheffé correction.
The results of visual analysis were also compared with lung function (Fisher exact test) and lung attenuation (factorial analysis of variance). The level of statistical significance was set at 5% in all instances. Reported results were from all three examined inspiratory levels, and the level with the highest discriminatory power was used to illustrate the potential clinical use.
To determine the diagnostic power of lung attenuation measurements for predicting BOS, an analysis of the receiver operating characteristics (ROCs) was performed. ROCs were computed with a computer program (CLABROC; MacIntosh version, Charles E. Metz, MD, University of Chicago, Illnois, 1991). The Student t test was used for comparisons of ROC curves.
Logistic regression analysis was used to compare the diagnostic powers of lung attenuation measurements and visual signs of small-airway disease at CT.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
Homogeneity of lung attenuation was measured as the SD of lung attenuation at each level. Patients who developed BOS within 1 year after the CT examination had a more homogeneous distribution of lung attenuation (84 HU ± 1) than did patients whose lung function remained normal (88 HU ± 1, P = .016, Fig 3). The homogeneity of lung attenuation was not different between patients with BOS and those with normal lung function test results at CT (88 HU ± 1 and 86 HU ± 1, respectively, P = .3).
|
At CT, 15 patients displayed a decline of lung function, which was consistent with BOS (Table 1, Fig 4). After 1 year, features of this study were as follows: Five additional patients were classified as having BOS. Three patients died during follow-up. In six patients, no follow-up lung function test results were available. In one patient, lung function improved from initially poor to normal.
|
|
|
At CT, visual signs considered indicative of small-airway disease were not predictive of BOS at 1 year after the CT study (Fig 5, Table 3). At CT, only mosaic attenuation was significantly associated with BOS, and the sensitivity of this sign was 27% (4 of 15 [Table 3]).
|
|
|
|
|
At CT, only mosaic attenuation was associated with a diagnosis of BOS (P = .02; relative hazard, 2.3; no CI available because of the small number of events).
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
While the decrease of lung attenuation in patients with future BOS confirms a concept of small-airway disease with pulmonary hyperinflation as the pathophysiologic basis in chronic graft rejection after lung transplantation, visual signs of small-airway disease, such as bronchiectasis, air trapping, and mosaic attenuation, at CT were less useful for predicting the course of pulmonary function. Mosaic attenuation was the most powerful diagnostic sign of BOS at CT examination, but it could not be used to predict the course of pulmonary function thereafter. It remains to be seen whether mosaic attenuation indicates an already advanced degree of bronchiolitis obliterans (11) and is, therefore, not a good predictor of later pulmonary disease or if mosaic attenuation may reverse to normal later (12).
Our finding that the usefulness of signs of small-airway disease at CT for diagnosing BOS is limited is in good agreement with findings in other investigations (3). According to an analysis of signs of small-airway disease at thin-section CT in volunteers, air trapping appears as a nonspecific sign, with increased incidence in smokers but with no particular correlation with lung function test results (13). These findings are in contrast to those of another investigation in which advanced stages of air trapping were suggestive of bronchiolitis obliterans later after transplantation (14). There is one important difference between our data and information in earlier reports: The time between surgery and CT examination was shorter in our study (mean, 28 months) than it was in the study reported by Bankier et al (14) (29 months until the first and 51 months until the last CT examination, on average).
Lee et al (3) also included findings of earlier CT examinations but used lung biopsy findings and additional pulmonary function parameters as indicators of bronchiolitis obliterans. Therefore, their study findings cannot be directly compared with our data, which are based on definitions provided by the International Society of Heart and Lung Transplantation.
The cause of reduced lung attenuation in patients who will develop BOS within the next year as compared with patients who, at presentation, have BOS initially can only be speculated on. One explanation is that the course of chronic rejection results in greatest attenuation reduction early and that features such as mosaic attenuation and air trapping counteract the reduction of parenchymal attenuation as BOS progresses with time.
To compare our results with those of earlier investigations, we need to take into account that there are few studies of CT lung attenuation in lung transplant recipients available to date, and most other studies (2,3,11,15,16) included smaller cohorts and only visual analysis and used an unspecified time between lung transplantation and CT, uncertain inclusion bias, or different end points. One major problem in the assessment of the CT signs of small-airway disease is the fact that the visual CT diagnosis is operator dependent (17), and the level of inspiration often remains uncertain.
It has been found previously that ungated CT of the lung does not allow for reproducible levels of lung attenuation (18). With the use of spirometrically gated CT and numeric analysis of lung attenuation, both factors can be approached. For the semiautomatic analysis of global lung attenuation used in this study, operator-dependent variability has been estimated as less than 1 HU for spirometrically gated scans (9). Overall reproducibility of the lung attenuation measurements has been cited as 10 HU, or 1% (8). This variability still allows for a meaningful prediction of lung function as evidenced by the predictive powers in our group.
Observer variation for visually diagnosing small-airway disease is acknowledged as a study limitation, but other researchers (14) demonstrated the magnitude of the problem. This variation was one rationale for conducting the present investigation of applying numeric estimates of lung attenuation in a spirometrically controlled setting.
The validity of our measurements is confirmed by the observed decrease of parenchymal attenuation in inspiration and near the lung base, which parallels the expected physiologic distribution of lung ventilation. Further evidence for the validity of our measurements is offered by a more homogeneous distribution of parenchymal attenuation during inspiration and at apical section positions.
Surprisingly, visually apparent instances of air trapping did not display any significantly higher degree of lung attenuation inhomogeneity at numeric analysis. Evidence of the fact that visually apparent inhomogeneity of lung attenuation did not correlate with numeric measurements is shown in Table 4. Visually apparent inhomogeneity of lung attenuation manifests as either mosaic attenuation or air trapping, and both features did not correlate with the SD of lung attenuation in the patients in our study group.
This lack of correlation may be attributable to the facts that lung homogeneity decreases at expiration (19), and the visual diagnosis of air trapping depends on expiratory scans. Apparently, the decrease of lung homogeneity on expiratory scans confounds further decreases caused by air trapping to such an extent that the latter cannot be discriminated by means of exploratory numeric analysis. The effect of expiration on lung homogeneity indicates that a constant level of inspiration must be ascertained for a valid diagnosis of air trapping, which necessitates spirometric gating. An alternative explanation for the lack of an association between air trapping and BOS is that the selected level of expiration at 20% VC may not be sufficient for detecting air trapping. This matter awaits further clarification.
However, direct correlation of lung attenuation with lung function test results confirms that attenuation inhomogeneities were not associated with bronchiolitis obliterans in the patients in our study group.
The significant correlation of visually determined mosaic attenuation with BOS at CT, but a lack of such an association with lung attenuation SD, suggests that changes of lung homogeneity in mosaic attenuation may be too subtle to be reliably detected by using numeric measurements in a larger cohort, which most likely is due to significant interindividual variability and the physiologic ventrodorsal attenuation gradient (20).
Differences in lung attenuation among patients with and without air trapping displayed wide overlap and were not statistically different. It is, therefore, assumed that the numeric mean attenuation differences are due to chance and do not have any specific significance.
The limited capability of CT for assisting in the diagnosis of BOS at examination may reflect the time course of BOS, since our patients were examined earlier after surgery than were patients in other published series. Thus, we speculate that early BOS may correlate with numeric estimates of lung attenuation and that as BOS progresses to clinically overt disease, correlation, as defined by lung function tests, is lost. Also, it is important to note that findings of only a few investigations (2,11,14–16) regarding that matter have indicated that CT can be used to diagnose BOS by using visual analysis. In the other investigations (2,3,11,14–16), another confounding factor was that the exact level of inspiration was not controlled for. The detection of signs of BOS at CT may require more complete expiration than was selected in our investigation.
The present investigation differs from earlier CT investigations (2,3,11,14–16) of lung transplant recipients in several regards. One other earlier report (15) on the use of CT for the diagnosis of bronchiolitis obliterans describes better diagnostic powers for the signs of airway disease at CT. This report included patients who were examined at a later period after transplantation, and the diagnosis of bronchiolitis obliterans was established by means of biopsy findings within 6 months from the study index time in only five of 11 patients. Thus, the diagnosis of bronchiolitis obliterans may not have reflected the status at the time of CT. Since the validity of lung biopsy findings for diagnosing bronchiolitis obliterans is notoriously low, use of this method as a standard of reference remains questionable.
If one must search for an alternative method for diagnosing bronchiolitis obliterans, one must consider that a close correlation with lung function test results at the time of the test cannot reveal any diagnostic powers beyond those of the method of reference. Therefore, comparisons of CT findings with lung function test results at CT alone (16) cannot reveal the potential of CT as a clinically useful method. Since early prediction of bronchiolitis obliterans during the later clinical course remains the most important potential application, comparison of CT parameters with follow-up lung function test results seems the only possible approach for establishing a prognostic method for predicting BOS at a time when preventive intervention may still be feasible.
To date, to our knowledge, augmentation of the immunosuppressive regimen, with standard therapy and the use of new, more potent immunosuppressive drugs, is the only possible response to signs of chronic graft rejection (1). However, until now, early prediction of chronic rejection has not been feasible, and the advent of a highly predictive method would greatly improve the chances for a more effective response. Although quantitative CT of the lung could not be used to predict BOS in all instances, a significant correlation was found as soon as 1 year after the CT examination. Future investigations are needed to confirm these findings in a larger cohort and over a longer follow-up period.
To assess the implications of differences in mean lung attenuation for a specific individual, ROC analysis was performed. According to the results of that analysis, the predictive powers of an optimally selected diagnostic threshold were determined. While the predictive power of this test is clearly less than ideal, there is no alternative method that can be used for predicting the future course of lung function available at present. Therefore, we view our study as a starting point for future investigations, specifically on the potential survival benefit of an earlier prediction of lung function and, consequently, intensified immunosuppressive therapy.
Measurements of lung attenuation were not used for planning clinical treatment in this prospective investigation, since the predictive power of these parameters, to our knowledge, was heretofore unknown. It is our hope that measurements of lung attenuation can be used for clinical purposes in the future. Further, the effectiveness of such an approach to selecting candidates for an intensified regimen of immunosuppression needs to be evaluated in another prospective investigation. Since the definition of BOS by the International Society of Heart and Lung Transplantation is based on lung function only, clinical decisions will be based on lung function tests in the near future. However, if our results are corroborated by those of other investigators, and results of prospective clinical investigations can be used to establish a clinical benefit of predicting BOS by using CT measurements of lung attenuation, imaging could add hitherto unavailable intelligence that is vitally important both for the individual lung transplant recipient and for the role of lung transplantation in general.
In conclusion, results of this investigation indicate that spirometrically gated CT measurements of lung attenuation can be used to predict the onset of BOS in lung transplant recipients. Potential clinical applications include the selection of high-risk candidates for whom aggressive medical therapy may be warranted to prevent the predicted course of lung disease.
|
ACKNOWLEDGMENTS |
|---|
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, F.D.K.; study concepts and design, F.D.K., R.E.; literature research, F.D.K., R.E., T.W.; clinical studies, F.D.K., R.E., T.W.; data acquisition and analysis/ interpretation, F.D.K., R.E., T.W.; statistical analysis, F.D.K.; manuscript preparation, F.D.K., R.E.; manuscript definition of intellectual content, F.D.K., R.E., R.H., R.F.; manuscript editing, R.E., R.H., R.F.; manuscript revision/review, F.D.K., R.E., R.H., R.F.; manuscript final version approval, all authors.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  A. A. Bankier, S. Mehrain, D. Kienzl, M. Weber, M. Estenne, and P. A. Gevenois Regional Heterogeneity of Air Trapping at Expiratory Thin-Section CT of Patients with Bronchiolitis: Potential Implications for Dose Reduction and CT Protocol Planning Radiology, June 1, 2008; 247(3): 862 - 870. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. D. Knollmann, S. Kapell, H. Lehmkuhl, B. Schulz, H. Bottcher, R. Hetzer, and R. Felix Dynamic High-Resolution Electron-Beam CT Scanning for the Diagnosis of Bronchiolitis Obliterans Syndrome After Lung Transplantation Chest, August 1, 2004; 126(2): 447 - 456. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
