The rationale for performing follow-up of the lungs in patients with cystic fibrosis relies on the high incidence of pulmonary insufficiency, which is the primary cause of death in 90-95% of patients [1, 2]. Patient status is assessed with clinical examination through the Shwachman-Kulczycki or other scoring systems [3-5]. Although these clinical scales are useful, they are not completely and directly correlated with pathologic changes of the lung [6, 7]. This is also true for laboratory tests, which are useful for the diagnosis of lung disease and the detection of complications. Pulmonary function tests are additional tools for lung evaluation that do not supply pathologic information.

Radiologic techniques provide morphologic information. Since the early application of chest radiography in patients with cystic fibrosis, a scoring system was developed to improve the accuracy and reproducibility of the examination at follow-up [8-10].

CT was also proposed for the follow-up of lung impairment in patients with cystic fibrosis [11]. As with chest radiography, a correlation between CT using a scoring system and functional status was found [7, 11, 12].

The CT examination rate that has been proposed in the follow-up of patients with cystic fibrosis is every 4-5 years. However, the data that are available refer to small populations [6, 11], acute settings [3], or short-term or one-shot evaluation and correlation.

The aims of this longitudinal retrospective study were to determine the correlation between CT using the Bhalla scoring system and pulmonary function test results and to assess the role of CT in predicting disease progression.

From the database of the Centre of Respiratory Pathophysiology at the University of Parma, 300 CT examinations of patients with cystic fibrosis performed between 1991 and 2001 were reviewed. Our institution approved this study, and informed consent was not required because of the retrospective nature of this investigation.

All scans were obtained in the follow-up set ting; patients with acute cystic fibrosis were excluded. Inclusion criteria were, first, CT examination with lung and mediastinal windows readable; and, second, pulmonary function tests performed the same day as radiologic assessment. The pulmonary function test results had to provide at least forced vital capacity (VC) and forced expiratory volume in the first second (FEV₁). CT examinations with im ages that were not of adequate quality and incomplete examinations were excluded, and cases with missing pulmonary function test results were also excluded.

Of this group of 300 examinations, 145 CT examinations were included in this retrospective study. These 145 examinations referred to 87 patients with a mean age of 15.6 years (range, 9 months-38 years). All patients had a confirmed dia gnosis of cystic fibrosis obtained with a sweat electrolyte test. Thirty patients underwent one CT examination, 56 underwent two examinations, and one patient underwent three examinations.

The mean interval between two examinations was 36.5 ± 8 months (range, 16.1-55.6 months). The CT examinations were performed on different scanners because of the constantly updating technology of the radiology department. We included in our population patients who underwent a sequential high-resolution scanning protocol of the thorax with standard acquisition parameters (slice thickness, 1 or 1.5 mm; scan interval, 10 or 20 mm) and re construction parameters (high-frequency kernel [B70]; lung window settings [width, -550 H; level, 1,500 H]). All examinations were performed at full inspiration, a respiratory status that is less affected by variability in the dimensions of the airways.

The 145 examinations were independently reviewed by three board-certified radiologists with at least 3 years' experience in thoracic imaging. The radiologists used the Bhalla scoring system [11] to evaluate lung impairment (Table 1) and were blinded to clinical and pulmonary function test results. The examinations were assessed using the scoring system shown in Table 1. Bronchiectasis and emphysema were identified according to criteria described elsewhere [13, 14]. Each of the radiologists provided a CT score using the Bhalla scoring system, and the overall mean CT score was obtained.

Statistical analysis was performed using a software package (SPSS version 11.0, SPSS) on a per-examination basis and on a per-patient basis. The per-examination analysis was performed by assessing the 145 CT examinations as independent statistical units on the basis of the long interscan interval. The per-patient analysis was performed to estimate progression of disease. Association indexes were obtained between patient characteristics, functional pulmonary test results, and CT scores. Differences between groups' means were assessed using the Student's t test. The two-way intraclass correlation coefficient was used to assess interobserver agreement. Linear regression was used for per-patient analysis. Statistical significance was set at a p value of < 0.05.

The rationale for performing chest CT in patients with cystic fibrosis is to obtain additional information with prognostic significance that should be used to modify clinical approach and treatment. For this purpose, standardization of the radiologic evaluation through a scoring system is crucial to obtain correlation with clinical outcome and for reproducibility.

In this study, no significant correlations were found between patient age and pulmonary function (VC), whereas VC was shown in a previous report to decline with patient age [15]. These results are probably due to the fact that the rate of decline varies among individuals and even in the same patient. In addition, pulmonary function test results are not linearly correlated with clinical status. VC and FEV₁ and derived ratios are reproducible but are not particularly sensitive to the early stages of disease, which, on the contrary, are better detected using small-airways function tests [16, 17]. Recently, investigators have shown that the lack of correlation between CT data and pulmonary function test results may be due to the higher sensitivity of CT in showing changes in the small airways before impairment can be detected on functional tests [18, 19].

The per-examination analysis showed significant but rather fair correlations between the mean CT score and VC (r = 0.39) and between the mean CT score and FEV₁ (r = 0.52). Helbich et al. [7] reported in 117 patients a high correlation between CT score and VC (r = 0.62) and CT score and FEV₁ (r = 0.70). Santamaria et al. [6] also reported a moderate correlation between CT score and VC (r = 0.60) and CT score and FEV₁ (r = 0.50) in a limited population of 30 patients. Conversely, Bhalla et al. [11] did not find significant correlations between CT score and VC (r = 0.48) or CT score and FEV₁ (r = 0.42), whereas they described a significant correlation between CT score and the ratio of FEV₁ to forced VC (r = 0.69). However, that study [11] was performed retrospectively on a fairly small population of patients (n = 14).

These different results and the fair correlation between CT scores and pulmonary function parameters can be explained by four factors.

The first factor is that CT does not allow one to distinguish between acute and chronic injury or to predict the reversibility of findings [20]. Therefore, acute and chronic radiologic findings are evaluated as similar [21].

The second factor is related to the fact that the Bhalla scoring system uses an ordinal scale. The same score may describe two different degrees of injury, and the intervals between scores are not necessarily equal or proportional. In addition, the characteristics that are considered in the scoring system are not weighted in accordance with their clinical importance.

The third factor is that there is large intraindividual variability and interindividual variability in the commonly used outcome measurements [17].

The fourth factor is the inclusion criteria, which may have allowed an excessive number of patients with disease at early stages to be included in the study cohorts, and the low number of patients in these studies.

It is not clear how to explain the optimal performance of the Bhalla scoring system for chest radiography providing better results than CT that has been reported in earlier investigations [9, 10, 22]. Conversely, other investigations have shown high-resolution CT to be superior to chest radiography [23, 24]. This could probably be due to the capability of CT to depict bronchiectasis, which is associated with pulmonary function tests [6, 7], whereas bronchiectasis was missed on chest radiography in 44.5% of bronchopulmonary segments and mucous plugs were missed in 91.5% of bronchopulmonary segments in one study [11]. High correlation between the chest radiography score and high-resolution CT score was found, and the authors concluded that in many cases CT results could be predicted by combining chest radiography findings with pulmonary function test results. This observation suggests that CT could be indicated in patients with mild respiratory symptoms for identifying early changes, whereas in those with stable disease, chest radiography could be the first imaging technique [6].

The interobserver agreement in our study shows that the Bhalla scoring system is highly reproducible, supporting the results of previous studies [7, 11].

The per-patient analysis showed a significant correlation and significant difference in mean CT scores between the first and second examinations. These results may suggest that the CT score is more sensitive to mild changes than functional pulmonary tests. The lack of significant differences between the functional test scores obtained at the first and second examinations explains why the CT score cannot be used to predict changes in pulmonary function test results between two examinations. The lack of significant correlation between the variations in functional parameters (VC and FEV₁) and mean CT score, the time elapsed between the first and second CT scans, and the lack of significant correlation between the variations in functional parameters (VC and FEV₁) and mean CT score show that these parameters cannot be used to predict the progression of disease. This finding could be indirectly confirmed by the insufficient correlation between the age-based severity curve of the Brasfield chest radiograph score and the decline of pulmonary function in a population of 230 patients [25]. In addition, this finding may be explained by the interval of 36.5 months between examinations in our study, which may not be sufficient to find significant correlation. However, the clinical role of variations in CT score, which is able to predict changes only when scans are obtained several years apart, is not clear.

Address correspondence to F. Cademartiri ([email protected]).

FOR YOUR INFORMATION

The AJR has made getting the articles you really want really easy with an online tool, Really Simple Syndication, available at www.ajronline.org. It's simple. Click the yellow RSS button located in the menu on the left of the page. You'll be on your way to syndicating your AJR content in no time.