Published ahead of print on August 11, 2005, doi:10.1164/rccm.200503-401PP
© 2005 American Thoracic Society doi: 10.1164/rccm.200503-401PP
Computed Tomography in the Evaluation of Cystic Fibrosis Lung DiseaseDepartment of Radiology, Cincinnati Children's Hospital Medical Center and the University of Cincinnati College of Medicine, Cincinnati; Columbus Children's Hospital, The Children's Radiological Institute, Columbus, Ohio; Department of Pediatric Pulmonology, Erasmus MC–Sophia, Rotterdam, The Netherlands; Department of Radiology, Vancouver General Hospital, Vancouver, British Columbia, Canada; Department of Radiology, David Geffen School of Medicine at the University of California–Los Angeles, Los Angeles; Division of Pediatric Pulmonology, Packard Children's Hospital at Stanford, Palo Alto, California; State University of New York Upstate Medical University, Syracuse, New York; University of Wisconsin, Madison, Wisconsin; and Children's Hospital of Denver, Denver, Colorado Correspondence and requests for reprints should be addressed to Alan S. Brody, M.D., Professor of Radiology and Pediatrics, Department of Radiology, MLC-5031, Cincinnati Children's Hospital, 3333 Burnet Avenue, Cincinnati, OH 45229-3039. E-mail: alan.brody{at}cchmc.org ABSTRACT The first report of computed tomography (CT) scanning to monitor cystic fibrosis (CF)–related lung disease was published in 1986. Further publications followed, but in general there was little interest in this technique until recently. Two factors in particular have led to this increased interest. First is an increasing realization that pulmonary function tests, long the mainstay of CF evaluation, often underestimate the presence and severity of mild and moderate lung disease. Second is the need for more sensitive outcome measures to assess new therapies. This had led to new interest and a series of important publications. The goal of this article is to present the current status of CT scanning in CF.
Key Words: computed tomography X-ray • cystic fibrosis • radiation effects • research design The first report of computed tomography (CT) scanning to monitor cystic fibrosis (CF)–related lung disease was published in 1986 (1). Further publications followed, but in general there was little interest in this technique until recently. Two factors in particular have led to this increased interest. First is an increasing realization that the morphologic changes seen with CT scanning often precede the functional changes identified with pulmonary function tests (PFTs). Second is the need for more sensitive outcome measures to assess new therapies. In addition, new imag-ing techniques, including controlled ventilation and spirometric control of lung volume, have advanced the capabilities of CT scanning. These factors have led to new interest and a series of important publications. One result of these efforts has been a realization that CT scanning offers information that may be complementary to the information provided by PFTs, both in understanding the pathophysiology of CF lung disease and in evaluating patients. The goal of this article is to present the current status of CT scanning in CF. CT TECHNIQUE
The two CT techniques that have been used for the evaluation of CF lung disease are high-resolution CT (HRCT) and volumetric, or spiral, CT. Incremental HRCT uses thin sections (< 2 mm) obtained at intervals through the lungs. This technique was developed to provide thin sections necessary for detailed parenchymal evaluation at a time when an extensively long breath-hold ( The chief benefit of incremental HRCT scanning is an approximately eightfold decrease in radiation dose for the study compared with volumetric scanning, due primarily to the interval rather than complete exposure of the lung to the CT X-ray beam. Such surveys, if used as outcome measures, are suitable for qualitative evaluation using scoring systems but pose problems for (semi-)automated quantitative analysis and for longitudinal comparison. For example, in one study, each subject had only five to seven airway/vessel pairs large enough and sectioned near perpendicular to the long axis to allow accurate measurement (3). The average lumen diameter of the segmental airways in children between the ages of 0 and 5 yr measured by CT was 1 mm, which is near the limits of resolution (3). Using volumetric CT, the lungs can be rotated three-dimensionally, allowing more airways to be measured accurately anywhere along their course. Repeat volumetric scanning allows one to easily measure the same airway at the same location for longitudinal assessment. To obtain diagnostically useful images, the scan should be taken during a breath-hold at near full-inflation and end-exhalation volumes to eliminate motion artifacts, improve resolution, increase reproducibility, and optimize the detection of abnormalities. In young children, this can be done using the controlled ventilation technique for both inspiratory and expiratory images (4). In older children, inspiratory and expiratory images can be obtained by voluntary breath-holds at full inflation and deflation, or by spirometer-controlled scan acquisitions at defined full-inspiratory and -expiratory lung volumes (5). When determining the number of images required for the evaluation of a specific lung disease, one should take the extent and distribution of the lung disease into account. On the basis of our preliminary experience, early CF lung disease may affect any lobe and is patchy in distribution, rather than evenly distributed through the lungs (6). A recent study has shown that incremental HRCT images at intervals greater than 10 mm will underestimate the severity of CF lung disease (7). In summary, conventional HRCT techniques allow lower dose CT scanning and may be useful for qualitative evaluation. Volumetric CT is necessary for optimal longitudinal evaluation and for quantitative analysis. USE IN CLINICAL CARE For clinical management of CF-related lung disease to be effective, onset and progression of the lung disease should be closely monitored. This can be done either indirectly by monitoring lung function or more directly by monitoring lung structure. PFTs, such as spirometry and body plethysmography, are considered the most important tools for measuring lung function from the age of 5 or 6 yr and onwards. More recently, multiple-breath washout techniques were described as a promising alternative to detect early pulmonary changes (8, 9). To monitor lung structure, chest radiographs have been used. Although quantitative chest radiograph scores have been shown to be higher at 2 yr of age than in infancy, the technique is described by the investigators as "complicated to use in non-study settings" (10). In adults, chest radiographs have been shown to be insensitive to acute changes, such as pulmonary exacerbations (11). To monitor structural lung changes, it is current practice at the Columbus Children's Hospital to do a controlled-ventilation HRCT at diagnosis. In contrast to chest radiographs, CT scans often show abnormalities in young children with CF. In a study of 32 asymptomatic infants and toddlers with CF, 10 to 20% of airways measured were bronchiectatic and 20% had bronchial wall thickening (12). In addition, the airway lumen/vessel diameter ratio was increased relative to control subjects and was progressive with increasing age. To monitor lung structure at the Erasmus MC–Sophia, all patients with CF from the age of 4 yr have a thin-slice CT repeated every other year. To monitor lung function, dynamic flows and dynamic and static volumes (FEV1, FVC, maximal expiratory flow at 25% of FVC [MEF25], total lung capacity [TLC], residual volume [RV]) are done routinely at the annual check-up. Figure 1 shows the dissociation between changes in lung function and lung structure in two patients who were part of a cohort study of 48 patients who were monitored for 2 yr (13). In the first patient, there were moderate structural abnormalities and mild lung function abnormalities initially. Two years later, structural abnormalities had markedly increased, although PFTs improved. Despite the improvement in pulmonary function, this patient has progressive damage to the lungs with increasing bronchiectasis and mucus plugging. On the basis of the CT findings, changes in treatment strategy, such as introduction of antiinflammatory treatment, more aggressive antibiotic treatment, and/or early introduction of recombinant human (rh)DNase, will be made. The second patient had marked structural abnormalities initially, indicated by the high CT score in the presence of only mildly abnormal PFTs. Two years later, the CT score was unchanged and PFTs had improved and were within the normal range. The CT findings indicate that there is definite lung damage despite normal PFTs. However, structural changes did not progress and lung function improved. It is likely that the treatment strategy for this patient will remain unchanged. This dissociation between structure and function was observed in many patients of the cohort study (Figure 2) (13). Other studies have also shown this dissociation. In children aged 6 to 10 yr, bronchiectasis was present in 30% of 37 children with normal PFTs (14). In a study of patients 12 yr to adult with an FEV1 of greater than 90%, bronchiectasis was found in 73% (15).
These studies do not address the use of CT scanning for acute changes in patient status. Although CT scanning has been shown to reflect improvement in patients treated for respiratory tract exacerbations (16–18), no clinical advantage has been shown for this evaluation. The use of CT scanning has been suggested in the follow-up of patients with nontuberculous mycobacteria infection (19). In summary, functional lung changes can be dissociated from structural changes. Thin-slice CT scanning is a more sensitive method than PFTs to detect structural changes and disease progression. Whether using CT in patient management improves the health of patients is not yet known. USE IN CLINICAL RESEARCH To monitor CF lung disease in clinical trials as a primary endpoint, CT must be shown to provide a valid outcome surrogate. An outcome surrogate is a laboratory measurement or physical sign used as a substitute for a clinically meaningful endpoint. In CF, the only outcome surrogate currently accepted by the U.S. Food and Drug Administration is FEV1. Outcome surrogates must meet strict criteria to ensure accurate results from clinical trials. One set of criteria was proposed by Robert Temple at the U.S. Food and Drug Administration (20). He wrote that a surrogate must be biologically plausible, reflect clinical severity, improve rapidly with effective therapy, and correlate with true outcomes. A recent review of imaging outcomes concluded that the presence of the surrogate should be closely linked to the presence of disease; detection of the surrogate must be accurate, reproducible, and feasible over time; and the changes must be closely linked to changes in the true endpoint (21). Looking at CT scanning in terms of these criteria, we know that CT images are similar to pathologic specimens, and the criteria for bronchiectasis, for example, were developed from pathologic correlation studies (22). HRCT is a sensitive indicator of early CF lung disease and has provided evidence of bronchiectasis, peribronchial thickening, mosaic perfusion, and mucus plugging in infants and children with CF (23–25). HRCT differentiates CF lung disease from other conditions as shown by very rare identification of abnormalities typical of CF in normal subjects (3, 26). HRCT scores have been shown to correlate with other evaluations, including clinical status and PFTs, in multiple cross-sectional studies of persons with CF (16, 17, 23–25, 27–32), indicating that higher CT scores are more common in subjects with worse PFTs despite the frequent dissociation of CT scores and PFTs in individual subjects and in longitudinal studies. A recent study by de Jong and coworkers (13) has shown that CT scores are reproducible. Studies have used HRCT to evaluate pulmonary disease progression in CF (27, 33). As part of a multicenter study to determine the effect of rhDNase on pulmonary function in subjects with CF aged 6 to 10 yr, serial HRCT scans were performed on 11 subjects over a 2-yr period. Visible peripheral bronchi were counted using a methodology used in children with difficult-to-treat asthma (34). Even though lung function remained stable over this 2-yr study, the number of visible airways on HRCT increased, suggesting that bronchiectasis worsened over this 2-yr period (unpublished data). Furthermore, findings from a 3-mo pilot study revealed improvements in HRCT subscores in young children with CF treated with rhDNase (18). Multiple studies have shown that CT scores improved after treatment of an exacerbation (16, 17, 32), indicating that CT scans improve after effective treatment. Recently, a 2-yr study was completed that showed that the change in CT score correlated with other outcome measures—in this case, respiratory tract exacerbations. In this study, there was no correlation between change in FEV1 and the number of exacerbations (35). CT scanning has the potential to detect disease sooner than FEV1, allowing evaluation of therapies directed toward very early disease. In more advanced disease, CT scanning appears to be more sensitive to changes in lung disease than FEV1 (27, 32), suggesting that if CT scanning becomes an accepted outcome surrogate, sample size and duration may be decreased when using CT as an outcome surrogate. Many of the recent CT data have come from long-duration trials without specific interventions. It will be necessary to demonstrate that CT reflects the effect of an effective intervention over a time course compatible with a therapeutic intervention trial to regard CT scanning as a valid outcome surrogate. QUALITATIVE AND QUANTITATIVE CT ANALYSIS The data provided by a CT scan are a series of images. These graphic data must be converted to numeric data to be used in clinical trials, and to be most helpful in clinical care. This conversion is an essential component in the use of CT scanning in CF. Both expert reader systems and computer analysis programs have been used for this conversion. Bhalla and colleagues (28) published the first HRCT scoring system for CF in 1991. Since then, several modifications have been published (17, 27, 30). In all systems, the reader identifies and assesses the severity of features associated with CF. Bronchiectasis, mucus plugging, airway wall thickening, and parenchymal opacities are included in all systems. Small nodules, mosaic attenuation, sacculations, and air trapping on expiratory images are included in some of the systems. The systems differ in how the severity of the features is assessed and, most strikingly, in how features are localized anatomically. Despite these differences, a comparison of five scoring systems—Castile and colleagues (12), Brody and colleagues (17), Helbich and coworkers (27) Santamaria and collleages (30), and Bhalla and coworkers (28)—showed similar results for each system (13). Twenty-five CT scans from children with CF aged 5 to 18 yr were scored and rescored after an interval of 1 to 2 wk and again after 1 to 2 mo by three observers. Between- and within-observer agreement was good with intraclass correlation coefficients generally greater than 0.8 (13). After this validation, the five systems were used in a 2-yr longitudinal study with 48 children with CF. There was no difference in the ability to track disease progression between the five scoring systems (32), likely reflecting the contribution of the features common to these different systems.
Another scoring system has been developed more recently (36). This system has been evaluated as part of the tracking of subjects in the Wisconsin Benefits of Neonatal CF Screening Project. This system was designed to provide lobar localization of findings and to independently assess each finding in each lobe. These scores could then be combined to provide an assessment of the total abnormality in each lobe, the extent of a specific finding in each patient, or the overall severity of CF lung disease. Evaluation of this system has shown A composite CT/PFT score has also been proposed (37). In a 1-yr, double-blind, placebo-controlled study of rhDNase in children with CF, the largest difference in treatment effects measured by the change from baseline was the composite CT/PFT score which showed a 30% change compared with 13% for FEF25–75 and 7% for the CT score alone. For future studies, it will be important to develop a scoring system by consensus on the basis of previous studies and include features to be assessed using specific definitions. This system should be able to show the severity of a specific feature at a specific anatomic location. A set of reference images would also reduce observer variation. This approach would also need to be tested (and validated) in different age groups and over a broad range of severity. The system would ideally be able to track progression of structural lung abnormalities from birth until end-stage lung disease. Computer-based methods for quantifying the structure of the lung fall into two main areas: evaluation of the parenchyma and of the airway walls. Both of these approaches are important in CF where large airway wall abnormalities can be assessed by airway wall measurement, and small airway and parenchymal abnormalities can be assessed by changes in lung parenchymal attenuation or air trapping. The accuracy of CT in the diagnosis and assessment of severity of parenchymal lesions, such as emphysema, has been well documented (38–42). Quantitative studies of lung CT have focused on the absorption of X-rays by the lung, and have related that to the volume of tissue and gas within the lung. It should be noted that lung density measurements on CT can be affected by a number of variables, including patient size, depth of inspiration, the type of CT scanner used, collimation, and the reconstruction algorithm. The best-studied parameter using computerized analysis of CT scans in CF is air trapping. Air trapping is a finding present in early CF lung disease, which reflects small airway obstruction that can be indirectly measured with quantitative CT techniques. In a study comparing 25 subjects with mild CF with 10 normal subjects, only the RV/TLC was statistically different between groups for the PFTs (43). The quantitative air-trapping technique showed a far greater statistical difference between the groups (p = 0.0003). These subjects were then evaluated during a 1-yr intervention with rhDNase. No significant differences between groups were noted for any of the PFT measurements after 3 and 12 mo of treatment. There was a significant difference in quantitative air-trapping measurements between groups at 3 mo and a nearly significant difference between groups at 12 mo despite a smaller number of subjects (44). The measurements of airway wall dimensions also have very important implications to the study of lung disease. Validation studies of airway measurements consistently show that there is a reasonable correlation between CT and pathologic measurements (45–48). The data suggest that accurate measurement requires sections 2 mm or less in thickness and is currently limited to airways 2 mm or more in diameter (45). In summary, both scoring by expert observers and computer analysis has been used successfully to interpret CT scans in CF. Currently, expert observer scoring systems include more features of CF lung disease than can be evaluated by computer analysis. Computer analysis is an area of active investigation and new methods are likely to be developed, increasing the number of features that can be analyzed. Computer analysis does not require specially trained observers and could also be more sensitive to subtle changes, particularly in air trapping. Future studies should also further evaluate the use of an outcome measure that combines CT scanning and PFTs. RADIATION RISK The best estimate of stochastic, or chance event, radiation risk, which includes the risk of developing cancer, is the patient-effective dose (49). Estimates from energy-imparted data (50, 51) or Monte Carlo simulations on mathematical phantoms (52–55) result in similar patient effective dose values of 0.25 to 0.50 millisieverts (mSv) for a 10- to 15-kg patient using a scanning technique expected to be as follows: 100 kV(p), 10 to 20 mA, pitch of 1.35, beam collimation of 8 x 1.25 mm or 16 x 1.25 mm; which reflects recent National Cancer Institute recommendations on reducing radiation dose to pediatric and small adult patients by using reduced kV(p), reduced mAs, and increased pitch values (56). The cumulative effective dose from one scan every other year from birth to age 30 with increasing X-ray techniques to reflect the increase in patient size would be approximately 10 to 15 mSv. The effective dose from CT scanning can be compared with natural background. In the United States, cosmic radiation, natural radioactivity, and domestic radon result in an average exposure of approximately 3 mSv/yr. An average U.S. resident would thus expect to receive 90 mSv in their first 30 yr of life, or six times the radiation burden of the proposed 15 CT scans for patients with CF. Shielding in X-ray departments is designed to limit doses to the public to less than 1 mSv/yr, and to radiation workers to less than 50 mSv/yr. It is noteworthy that the dose limit to the embryo/fetus of a pregnant radiation worker is 5 mSv for the duration of the gestation period, with no more than 0.5 mSv received in any one month.
It is also possible to estimate the risk of fatal cancer to a small infant from a CT scan with an effective dose of 0.5 mSv. The International Commission on Radiological Protection (49) recommends using a nominal risk coefficient of 5% per sievert, which is increased threefold to account for the higher radiosensitivity of infants. The resultant risk is approximately one additional fatal cancer in 13,000. It is important to note that there is considerable uncertainty in this risk estimate, which is obtained by performing a linear extrapolation from known effects at high organ doses (> 0.1 Gy) to the much lower organ doses expected in these CT scans ( Most CT protocols have been applied in a research setting, and it is unclear what benefit would accrue to patients from routine clinical examinations. Radiation risk needs to be weighed against the anticipated benefit to the patient from diagnostic CT scans. CONCLUSIONS Effective techniques are available to allow CT scanning to be performed in patients with CF of all ages. Quantitative and qualitative parameters derived from CT scans are more sensitive than currently used routine pulmonary function parameters to estimate structural lung changes in CF. Current studies suggest that CT scanning may require substantially fewer patients in clinical studies. Efforts to refine observational scoring systems and expand computerized scoring systems are in progress. Further studies showing that CT scanning is an effective outcome measure in intervention studies will be needed before CT scanning can be fully accepted as a valid outcome measure for CF lung disease. Acknowledgments Additional meeting attendees who participated in discussion included: Richard Ahrens, Andrew Bush, Stephanie Davis, Lynn Fordham, Michael Knowles, Ruth Milner, Robert Tepper, and Claire Wainwright. The affiliations of the presenters as listed on the opening page, are: Cincinnati Children's Hospital Medical Center and the University of Cincinnati College of Medicine, Department of Radiology, Cincinnati, OH (A.S.B.); Erasmus MC-Sophia, Rotterdam, Department of Pediatric Pulmonology, Rotterdam, The Netherlands (H.A.W.M.T. and P.A.d.J.); Vancouver General Hospital, Department of Radiology, Vancouver, British Columbia, Canada (H.O.C.); Columbus Children's Hospital, The Children's Radiological Institute, Columbus, OH, (R.G.C. and F.R.L.); David Geffen School of Medicine at University of California Los Angeles, Department of Radiology, Los Angeles, CA (J.G. and M.M.-G.); State University of New York Upstate Medical University, Syracuse, NY (W.H.); University of Wisconsin, Madison, WI (M.R.); Packard Children's Hospital at Stanford, Division of Pediatric Pulmonology, Palo Alto, CA (T.E.R.); Children's Hospital of Denver, Denver, CO (S.D.S.) FOOTNOTES This conference was supported by the Cystic Fibrosis Foundation. Alan Brody and Harm Tiddens jointly organized and chaired the CT scanning in Cystic Fibrosis Special Interest Group meeting held October 17, 2003, in Anaheim, California, sponsored by the North American Cystic Fibrosis Foundation. This report, jointly written by Alan Brody and Harm Tiddens, summarizes the presentations by those listed above. See ACKNOWLEDGMENTS at end of article for additional information. Originally Published in Press as DOI: 10.1164/rccm.200503-401PP on August 11, 2005 Conflict of Interest Statement: A.S.B. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. H.A.W.M.T. has acted as consultant in research and development, in 2001 and 2002, as member of an advisory board on aerosolized antibiotics from Chiron Biopharmaceuticals, and received within the last 3 yr honoraria and travel expenses for lectures and workshops from Chiron, Hoffman-LaRoche, and Genentech, Inc., and holds no stock options. The BV Kindergeneeskunde of the Erasmus MC–Sophia has in the last 3 yr received research grants from Hoffman-LaRoche. R.G.C. will receive $161,836 between 2004 and 2006 from Genentech, Inc., as a research grant. H.O.C. received $2,500 in 2002 and £1,500 in 2003 for serving on an advisory board for GlaxoSmithKline (GSK). In addition, he is the coinvestigator on two multicenter studies sponsored by GSK and has received travel expenses to attend meetings related to the project. A percentage of his salary between 2003 and 2006 ($15,000/yr) derives from contract funds provided to a colleague, Peter D. Pare, by GSK for the development of validated methods to measure emphysema and airway disease using computed tomography. P.A.d.J. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. J.G. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. W.H. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. F.R.L. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. M.M.-G. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. M.R. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. T.E.R. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. S.D.S. does not have a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form March 19, 2005; accepted in final form August 9, 2005 REFERENCES
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
60 s) and higher radiation dose would be required if contiguous thin sections were obtained through the entire chest (
coefficients with values greater than 0.4, indicating good agreement, in 74 of 96 subcomponents. The sensitivity, which was the square root of film-to-film variance, was quite good at 27.9. The reproducibility of the overall score was excellent at 98.1%, although individual lobe scores show a reproducibility that is around 40% or lower. 
0.001 Gy). 
