| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Thoracic Imaging |
1 From the Department of Medical Imaging (E.K., C.P.M., T.B.C., J.C., N.S.P., G.L.W.) and Toronto Lung Transplant Program (C.G., C.C., M.A.H.), Toronto General Hospital, University Health Network, Ontario, Canada. Received April 10, 2003; revision requested June 11; final revision received September 9; accepted September 29. Address correspondence to E.K., Department of Diagnostic Imaging, Chaim Sheba Medical Center, Tel Hashomer 52621, Israel (e-mail: konen@zahav.net.il).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Two hundred ninety-eight thin-section CT scans obtained in 26 lung transplant recipients who did (study group) and 26 lung transplant recipients who did not (control group) develop BOS were reviewed for the presence of mosaic perfusion, bronchiectasis, bronchial wall thickening, and air trapping. BOS was defined by using the recently revised definition of the International Society for Heart and Lung Transplantation. CT scans obtained in the BOS group were divided into three groups: Group A consisted of the last scans obtained before the clinical appearance of BOS; groups B and C consisted of, respectively, the first and last scans obtained after the clinical appearance of BOS. Scans obtained in the control group were acquired during similar posttransplantation periods and matched to scans in each BOS group. Sensitivity, specificity, and positive and negative predictive values were calculated separately for each subgroup. The optimal threshold for each thin-section CT–depicted abnormality was defined by using receiver operating characteristics analysis.
RESULTS: The sensitivities of air trapping for the diagnosis of BOS during the periods in which the scans in groups A, B, and C were obtained were 50%, 44%, and 64%, respectively; specificities were 80%, 100%, and 80% respectively. Sensitivities of mosaic perfusion were 4%, 20%, and 36%, respectively; specificities were 100%, 96%, and 96%, respectively. Sensitivities of bronchiectasis were 25%, 24%, and 32%, respectively; specificities were 80%, 80%, and 96%, respectively. Sensitivities of bronchial wall thickening were 4%, 24%, and 40%, respectively; specificities were 96%, 84%, and 80%, respectively. Air trapping was seen intermittently in nine (43%) of 21 patients with CT scans that depicted this finding at least once.
CONCLUSION: The value of the finding of air trapping before the clinical appearance and during the early stages of BOS is lower than has been previously reported. When using the recently revised criteria for BOS, the role of thin-section CT as a screening test to evaluate patients with lung transplants appears to be limited.
© RSNA, 2004
Index terms: Bronchiolitis obliterans, 60.2191 • Lung, CT, 60.12111, 60.12118 • Lung, diseases, 60.2191 • Lung, function • Lung, transplantation
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
BOS was originally defined as a decrease in forced expiratory volume in 1 second (FEV1) (6). The International Society for Heart and Lung Transplantation, however, recently issued a revised classification of BOS (7) that includes the new subcategory of "potential BOS," or stage BOS 0-p. With this modification, a decrease in forced expiratory flow in the midexpiratory phase and/or a milder decrease in FEV1 is also taken into account, and these criteria are thought to be more sensitive for detection of the early development of BOS.
The results of several studies have suggested that thin-section computed tomography (CT) has an important role in the detection of bronchiolitis obliterans or BOS in lung transplant recipients (8–11). However, the investigators in most such studies have evaluated thin-section CT scans that were obtained in patients with relatively long-standing disease. There is limited information regarding the potential role of thin-section CT in predicting the forthcoming deterioration before the clinical appearance or during the early stages of BOS. This is the most clinically important period because there is no treatment for established BOS and only prompt therapeutic intervention during the early stages may stop the progression of the disease (12).
The aim of this study was to determine whether there are thin-section CT features that predict BOS in lung transplant recipients before the clinical appearance and during the early stages of the disease.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
|
The thin-section CT examinations were performed by using one of the following scanner models: HiSpeed CTi, LightSpeed QXi, or LightSpeed (GE Medical Systems, Milwaukee, Wis). The scans were obtained in all patients during inspiration and expiration, with the patient lying in the supine position, by using 1.0-mm collimation, a 1-second scanning time, 120 kVp, and 280–340 mAs. The images were reconstructed by using a high-spatial- frequency (bone) algorithm. The inspiratory scans were obtained from the thoracic inlet to the lung bases at 10-mm intervals. Three expiratory scans were obtained at the following levels: the aortic arch, the tracheal carina, and above the diaphragm.
All thin-section CT scans were evaluated independently by a pair of chest radiologists, with each pair consisting of one senior fellow in thoracic imaging (E.K. or C.P.M.) and one staff member (T.B.C. or J.C., with 3 and 4 years experience, respectively). The reviewers were blinded to each other’s interpretation as well as to the patients’ clinical and functional information. To standardize the assessments, all observers reviewed from previous studies (9–11,13) the thin-section CT findings in patients with BOS that developed after lung transplantation; the highest quality digital versions of the images were reviewed. The scans were assessed in several sessions by using a picture archiving and communication system, or PACS (eFilm Workstation 1.7.1; Merge eFilm, Toronto, Ontario, Canada). In each session, different images from the study and control groups were evaluated in a random order. For each patient, the inspiratory scans were reviewed first, with the expiratory scans of the same patient reviewed immediately afterward. The reviewers were allowed to change window and magnification settings freely and to compare the inspiratory and expiratory scans side by side on the screen, with no time limitation.
For each inspiratory CT examination, five sections at the following levels were selected: the great arteries, the aortic arch, the carina, the inferior pulmonary vein, and 1 cm above the diaphragm. The sections were selected by one of the two radiologists who reviewed the case and were recorded on a separate form, which allowed the second reviewer to assess the same sections while remaining blinded to the scoring of the first reviewer. The inspiratory thin-section CT scans were evaluated for bronchial dilatation, bronchial wall thickening, and mosaic perfusion, and the expiratory thin-section CT scans were evaluated for air trapping. A finding was considered to be positive for bronchial dilatation when the bronchoarterial ratio was greater than 1 at subjective evaluation or the bronchial lumen was seen within 1 cm of the costal pleura. Findings positive for bronchial wall thickening were based on subjective assessment. Findings of bronchial dilatation and bronchial wall thickening were scored by using a three-point scale, on which a score of 0 meant no abnormality was visible; a score of 1, the finding was borderline abnormal; and a score of 2, the finding was definitely abnormal. A single score was assigned for both lungs.
For mosaic perfusion, the extent of low-attenuating regions, expressed as a percentage of the cross-sectional lung surface area, was assessed for each section and for each lung by using a five-point scale that was used in a previous study (11): A score of 0 meant no low-attenuating areas were visible; a score of 1, 1%–25% of the cross-sectional area of the lung was affected; a score of 2, 26%–50% of the cross-sectional area of the lung was affected; a score of 3, 51%–75% of the cross-sectional area of the lung was affected; and a score of 4, 76%–100% of the cross-sectional area of the lung was affected. For each lung, the potential maximum score could reach 20; thus, for both lungs, the potential maximum score could reach 40.
The extent of air trapping was similarly assessed on thin-section expiratory CT scans at the levels of the aortic arch, of the tracheal carina, and above the diaphragm for each section and for each lung by using a five-point scale that was used in previous studies (11,13–15): A score of 0 meant no air trapping was visible; a score of 1, 1%–25% of the cross-sectional area of the lung was affected; a score of 2, 26%–50% of the cross-sectional area of the lung was affected; a score of 3, 51%–75% of the cross-sectional area of the lung was affected; and a score of 4, 76%–100% of the cross-sectional area of the lung was affected. For each lung, the potential maximum score could reach 12; thus, for both lungs, the potential maximum score could reach 24.
The reviewers ignored areas of decreased attenuation in the bare region—that is, in the region of the minor fissure—and in the apical segments of the lower lobes. They also ignored air trapping in single secondary lobules, which is seen in some healthy individuals (14). For statistical calculations, in cases of single-lung transplantation, the score for the transplanted lung was multiplied by two.
Pulmonary Function Tests
At our institution, the routine follow-up of all lung transplant recipients includes PFTs, which are performed every month for life, and the acquisition of inspiratory and expiratory thin-section CT scans at 3, 6, 9, 12, 18, and 24 months after transplantation and then every 12 months thereafter. All patients in our study underwent thin-section CT and PFTs during the same week; most of them underwent these examinations on the same day. Spirometry was performed with a dry rolling seal spirometer (PK Morgan, Kent, United Kingdom). The PFTs included measurements of the FEV1, forced vital capacity, FEV1/forced vital capacity ratio, and forced expiratory flow in the midexpiratory phase. The diagnosis and staging of BOS were based on the results of periodically performed PFTs, with use of the recently revised classification system for BOS established by the International Society for Heart and Lung Transplantation (Table 2) (7).
|
First, we selected scans from the study group that were obtained during three clinically important periods: The group A scans consisted of the last thin-section CT scans obtained in each patient before the clinical appearance of BOS; the group B scans consisted of the first thin-section CT scans obtained in each patient after the clinical appearance of BOS; and the group C scans consisted of the last thin-section CT scans obtained in each patient after the clinical appearance of BOS. We defined the respective periods during which these scans were obtained as periods A, B, and C. The median days on which and the ranges of times during which the scans in each group were obtained in relation to the date of the clinical appearance of BOS are shown in Table 3. We matched each thin-section CT scan obtained during the above periods to a scan from the control group that was obtained during a similar posttransplantation period.
|
The correlation between either air trapping or mosaic perfusion score and clinical stage of BOS was assessed by using Spearman correlation coefficients. Kruskal-Wallis analysis of variance of the ranks was used to compare the radiologists’ scores according to severity of BOS. Interobserver agreement on the assessment of each thin-section CT abnormality was analyzed by using weighted 
statistics (17). All analyses were performed by using statistical software (SAS, version 8.02 for Windows; SAS Institute, Cary, NC).
The value of each thin-section CT finding in the patients in whom several repeat scans were obtained was further evaluated by generating graphic timelines that included all of the thin-section CT results obtained for each of the study group patients. Direct analysis of these timelines enabled us to calculate the frequencies of the appearance of each thin-section CT finding during the different periods in the patients who underwent multiple examinations.
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
|
Interobserver agreement was good in the assessment of bronchial dilatation ( = 0.76) and air trapping ( = 0.61) and moderate in the assessment of mosaic perfusion ( = 0.57) and bronchial wall thickening ( = 0.54).
Ten (38%) of the 26 patients in the study group were judged at PFT to have BOS 0-p, which remained at this stage until the last day of the study. An additional seven patients (27%) were initially judged to have BOS 0-p, which subsequently progressed to higher stages. The remaining nine patients (35%) were initially judged to have BOS 1 or more severe disease at their first PFT. As of this writing, none of the patients who were judged to have BOS 0-p during the period of the study had recovered—that is, their lung function results had not returned to normal values. No significant correlation between the air trapping score at expiratory thin-section CT and the clinical stage of BOS was observed.
When the scans that were positive for air trapping were combined with the scans that were positive for mosaic perfusion, the combined sensitivity of thin-section CT for the detection of BOS remained unchanged (50%, 12 of 24 scans) during period A, increased from 44% (11 of 25 scans) to 48% (12 of 25 scans) during period B, and increased from 64% (16 of 25 scans) to 72% (18 of 25 scans) during period C.
Timelines demonstrating the temporal relationship between the appearance of air trapping on thin-section CT scans and the date of the clinical appearance of BOS for each patient in the study group are shown in the Figure and summarized in Table 5. The results of similar analyses of timelines demonstrating this relationship with mosaic perfusion, bronchiectasis, and bronchial wall thickening also are presented in Table 5. The percentages of patients who had findings positive for air trapping on at least one thin-section CT scan during the period ranging from 6 months before to up to 6 months after the clinical appearance of BOS ranged from 42% to 50%. These values increased gradually when CT scans were acquired more than 6 months after the appearance of BOS (Table 5). Nine (43%) of 21 patients who showed air trapping at thin-section CT had at least one sequential normal scan thereafter (Figure).
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
Although our study findings are in agreement with those in previous series in which air trapping was shown to be the most reliable indicator of BOS, we found the sensitivity of this finding to be significantly lower (44%–50%) than previously reported (74%–91%) when thin-section CT scans were obtained during periods A and B—respectively, 41–136 days before and 30–135 days after the clinical appearance of BOS. The sensitivities of mosaic perfusion, bronchiectasis, and bronchial wall thickening were even lower during these periods, not exceeding 25%. However, the sensitivities of air trapping, mosaic perfusion, and bronchiectasis for the diagnosis of BOS increased when CT scans were obtained later during the course of the disease—that is, during period C (median day scan obtained after BOS appearance, day 409). A similar trend was noted when we analyzed the timelines: The repeat scans obtained in the patients who had had BOS for longer than 12 months showed abnormal air trapping at least once in 79% of the patients, mosaic perfusion at least once in 57%, and bronchiectasis at least once in 43%. These findings confirm the assumption of Lee et al (13) that there is a gradual increase in abnormal thin-section CT findings as BOS progresses.
In most previous reports, the duration of BOS at the time of CT scanning either was not specified (8,11) or was averaged (9). Lee et al (13) evaluated thin-section CT scans that were obtained during a relatively early stage of the disease (mean, 19 weeks), and the air trapping (74%) and mosaic perfusion (17%) seen on these images also had relatively low sensitivity in the diagnosis of BOS (Table 6).
|
Both bronchiectasis and bronchial wall thickening had relatively low sensitivity in the diagnosis of BOS during all periods (24%–36% and 4%–40%, respectively) but higher specificity (80%–96% for both findings). These results are in agreement with those in previous series, which similarly showed bronchiectasis (9,11,13) and bronchial wall thickening (8,11,13) to have low sensitivity and bronchiectasis to have higher specificity (8,9) (Table 6).
We intentionally adopted the scoring systems for air trapping and mosaic perfusion used in previous studies (11,13–15) so that our results would be comparable. Because an air trapping score of 2 may be seen in subjects with normal airways (14), some investigators (11,13,15) have adapted the threshold for findings positive for air trapping and considered scores of 3 or higher to indicate air trapping. The relatively large number of patients in the present study enabled us to base our statistical analysis on separate threshold calculations for each posttransplantation period and for each thin-section CT finding by using receiver operating characteristics analysis. However, adopting a uniform threshold score of 3 to indicate air trapping would result in an even lower sensitivity of air trapping during period A (33%), with a gradual increase in sensitivity during periods B and C (48% and 68%, respectively). These data emphasize the limited value of thin-section CT for the diagnosis of BOS during the last 3 months before the clinical appearance of the disease.
Our use of the revised criteria for BOS (7) for the first time also might have contributed to the lower sensitivity of air trapping in this study compared with the sensitivities of this finding in previous series. The revised classification includes the new stage of potential BOS, or BOS 0-p, which is defined as a 10%–19% decrease in FEV1 and/or a 25% decrease in the forced expiratory flow in the midexpiratory phase from the baseline value. This revised classification is believed to be more sensitive for the detection of early-stage BOS. Seventeen (65%) of the 26 patients in our study group were initially judged to have BOS 0-p. With the previous classifications, these patients would have been judged to be disease free. Bankier et al (10) referred to this issue in their discussion and reported performing a complementary analysis of their data by using criteria similar to those of the revised BOS staging system. They reported observing lower values of sensitivity and specificity without specifying the exact change in values. Since most transplantation centers are already using the new criteria and initiating treatment in patients with BOS 0-p, we believe that any future study in which thin-section CT findings are compared with PFT results should refer to the new criteria.
We believed that displaying the thin-section CT findings by using graphic timelines would enable the reader to have an additional and direct visual impression of the value of each thin-section CT finding in the detection of BOS, especially in the patients who underwent several repeat CT examinations. Analysis of these timelines, compared with the statistical results (ie, sensitivity, specificity, and positive and negative predictive values), revealed similar rates of BOS detection with all thin-section CT findings. The timeline analysis also led to an interesting observation: that air trapping was a nonconstant finding in 43% of the patients. In other words, in nine of 21 patients who demonstrated air trapping at CT, at least one subsequent thin-section CT scan that did not show the same findings was obtained. Transient infections or, less likely, subclinical fluctuations in the activity of the disease might explain such intermittent changes. These findings further emphasize the need for future studies to include more than one scan per patient in assessments of the value of thin-section CT in patients with lung transplants.
A potential limitation of our study was the use of lung transplant recipients as a control group. Findings positive for air trapping were seen at least once in 11 of the 26 patients in this group. Although air trapping is commonly encountered in subjects with normal PFT results (18), if any of these patients will develop BOS in the future, it could be argued that their CT scans that are positive for air trapping should be considered true-positive rather than false-negative. We believe, however, that the effect of such an error is minor since a relatively long clinical follow-up (median, 13 months; range, 5–31 months) of these patients did not reveal any clinical evidence of airway disease. In addition, the disadvantage of using healthy individuals as control subjects might be greater owing to bias of the reviewers, who could easily identify the absence of a lung transplant on the CT scans.
In conclusion, our study results are in agreement with the findings of previous studies, which suggest that air trapping is the most frequent thin-section CT abnormality observed in patients with BOS. However, our study findings suggest that the sensitivity of CT-depicted air trapping before the clinical appearance of BOS and during the early stages of the disease is lower than has been previously reported. In addition, the appearance of air trapping in patients with BOS is frequently intermittent. When using the recently revised criteria for BOS, the role of thin-section CT as a screening test for the evaluation of patients with lung transplants appears to be limited.
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, E.K.; study concepts, E.K., C.G., G.L.W., M.A.H.; study design, E.K., C.C., C.G., G.L.W.; literature research, E.K., C.C.; clinical studies, E.K., C.P.M., T.B.C., J.C.; data acquisition, E.K., C.C., C.P.M., T.B.C., J.C.; data analysis/interpretation, E.K., C.G., M.A.H., N.S.P., G.L.W., C.C.; statistical analysis, C.G.; manuscript preparation, E.K., C.G., C.C.; manuscript definition of intellectual content and editing, E.K., C.G., C.C., G.L.W.; manuscript revision/review, C.P.M., T.B.C., J.C., M.A.H., N.S.P., G.L.W.; manuscript final version approval, M.A.H., N.S.P., G.L.W.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODS RESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  T. B. Reece, J. D. Mitchell, M. R. Zamora, D. A. Fullerton, J. C. Cleveland, M. Pomerantz, D. M. Lyu, F. L. Grover, and M. J. Weyant Native lung volume reduction surgery relieves functional graft compression after single-lung transplantation for chronic obstructive pulmonary disease. J. Thorac. Cardiovasc. Surg., April 1, 2008; 135(4): 931 - 937. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 P A de Jong, J D Dodd, H O Coxson, C Storness-Bliss, P D Pare, J R Mayo, and R D Levy Bronchiolitis obliterans following lung transplantation: early detection using computed tomographic scanning Thorax, September 1, 2006; 61(9): 799 - 804. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
