INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Postmortem studies have shown that the airway walls of patients dying of asthmatic attacks are thickened (1). Such airway wall thickening involves all wall layers (epithelium, submucosa, muscle, and adventitia) and results from inflammation with edema and inflammatory cell infiltration and from structural changes, such as subepithelial fibrosis, mucous gland and goblet cell hyperplasia, and smooth muscle hypertrophy and hyperplasia (1). The latter is considered a feature of airway wall remodeling, which is attributed to chronic inflammation (4). Analysis using a computer model indicates that a modest degree of airway wall thickening is associated with airway narrowing caused by smooth muscle shortening (7, 8), suggesting that airway wall thickening has an important pathophysiologic role in asthma.

Bronchial biopsy with a fiberoptic bronchoscope has been used to study structural, as well as inflammatory changes of the asthmatic airway (4, 5). However, thickening of the total bronchial wall cannot be evaluated with this procedure. Assessment of airway wall thickness in asthmatic patients has therefore been difficult.

Computed tomography (CT) has recently been used to study airway dimensions in experimental models (9, 10), and CT findings, including airway wall thickening, have been evaluated in asthmatic patients (11). However, most interpretations were subjective (11, 14, 17, 18), and only a few studies have used CT to quantitatively analyze airway wall thickness in patients with asthma (13, 16, 19). Okazawa and colleagues (16) and Awadh and associates (19) have shown that the airways of asthmatic patients are thicker than those of normal controls, whereas Boulet and colleagues (13) have failed to show such a difference. The presence and significance of airway wall thickening as assessed by CT in asthmatic patients thus remains to be clarified.

We quantified airway wall thickness and luminal area in a large number of asthmatic patients and healthy control subjects. First, the reproducibility of the method used for CT measurement was confirmed. Airway wall dimensions were then compared according to the severity of asthma. In addition, we investigated the relation of airway wall thickness to the clinical characteristics of the asthmatic patients, such as age, duration and severity of disease, and degree of airflow obstruction and airway inflammation.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

A total of 81 patients with asthma and 28 healthy volunteers, including members of our hospital staff, were studied in a cross-sectional fashion.

Asthma was defined according to the American Thoracic Society criteria (22). Some patients were enrolled when they first presented at our outpatient clinic, a part of which had not previously received treatment for asthma. Others were already receiving medication and were being followed at our clinic. The severity of asthma was classified according to the Global Initiative for Asthma (23), based on clinical features before treatment and daily medication required to maintain control (24). Asthma was classified as intermittent in seven patients, mild persistent in 13, moderate persistent in 39, and severe persistent in 22. For patients who were receiving treatment at entry, "baseline" clinical features were assessed in a retrospective manner. For patients who had not previously received medication for asthma or who had received treatment but had inadequate disease control, the severity of asthma was evaluated after the minimal medication required to maintain control had been determined. Medication usage at study entry is shown in Table 1.

The healthy control subjects had no respiratory symptoms and no history of asthma or other respiratory diseases.

All subjects were lifetime nonsmokers. Lung function tests and CT scans were performed on the same day, before using bronchodilators. These examinations were done on presentation to our hospital for newly recruited asthmatic patients and at regular follow-up examinations for other patients. The study protocol was approved by the ethics committee of our institution, and written informed consent was obtained from all subjects.

CT

A Toshiba X-Vigor CT scanner (Toshiba, Tokyo, Japan) was used. All scans were obtained at suspended end-inspiratory volume because an increased number of artifacts has been reported in scans obtained at functional residual capacity (12, 13). Although a window level of 450 Hounsfield units (HU) has been recommended for measurement of airway dimensions on CT scans (10, 25), we performed a validation study using a phantom to determine the most appropriate window settings for our CT scanner. The phantom was made of a polystyrene foam block and five plastic right circular cylinders, which represented the lung parenchyma and airways, respectively. The actual size of the cylinders was measured with an optical micrometer caliper to the closest 0.01 mm. The cylinders ranged from 1.09 to 1.65 mm in wall thickness and from 12.4 to 50.2 mm² in luminal area. The actual wall areas of the phantoms were calculated from their actual luminal area and wall thickness, which ranged from 17.3 to 43.8 mm². Figure 1 shows the calculated degree of error for nine different CT window levels at a width of 700 HU (A) and eight different CT window widths at a level of 450 HU (B). As reported previously (10, 25), the most appropriate window level was 450 HU, whereas window width had no effect on the accuracy of these measurements. The optimal window settings for measurement of luminal area were similar to those for measurement of wall area: a window level of 450 to 500 HU, and no effect of window width (data not shown). There was no obvious systematic error in the determination of actual versus calculated luminal area (data not shown). A window level of 450 HU was therefore chosen for the present study. A window width of 1,500 HU was used because narrower widths cause less than optimal visualization of anatomic landmarks at the 450 HU level (10).

View larger version (16K): [in this window] [in a new window]   Figure 1.   Effect of computed tomography (CT) window level (A) and window width (B) on measurement of wall area (WA) of the phantom. The absolute difference between "actual" WA and WA measured from CT images is plotted. Values are means ± SD for all five cylinders, each measured 3 times. The accuracy of measurement was greatest at the window level of 450 Hounsfield units (HU) (A). Window width had no effect on the accuracy of measurement (B).

Thin-section helical scans, rather than conventional scans, were used for measurement (21), to more easily obtain the optimal slice to be measured and to allow the possibility of conducting a future longitudinal study to examine the effect of treatment on airway wall thickness. Such studies require comparison of scans at identical airway levels preintervention and postintervention in each patient (10, 12, 16, 21), and an unacceptably high radiation dose would have been necessary if conventional CT scanning was used. In a preliminary investigation, scans were obtained in 32 subjects (29 asthmatic patients and three healthy control subjects) at two selected levels, the trunk of the apical bronchus of the right upper lobe and the trunk of the posterior basal bronchus of the right lower lobe. Two parameters of airway wall thickness obtained at these two sites showed good correlation (r = 0.655, p = 0.0003 for wall area and r = 0.555, p = 0.002 for wall area%) by Spearman's rank correlation test. We therefore decided to measure airway wall dimensions at only one of two sites in this study, the origin of the apical bronchus of the right upper lobe. This site was chosen because of its more convenient orientation for obtaining a tangential view of the airway, as well as an outer perimeter view of the airway not abutted by vessels or other bronchi.

Helical scanning was performed at 120 kVp, 50 mA, 3 mm collimation, and pitch 1.0. To obtain one section through the apical segmental bronchus, 3.8 cm of lung were scanned in the helical section. The starting point of the scan was determined on scout film according to this value, assuming that scanning would be terminated at the origin of the right upper lobe bronchus. No contrast media were used. Images were reconstructed using the FC10 algorithm at 1-mm spacings. A targeted reconstruction of the right lung was performed using a subject-specific field of view (FOV) (153 to 214 mm). Each image was composed of a 512 × 512 matrix of numeric data (CT numbers) in HU. The slice to be used for measurement was selected at the origin of the bronchus, avoiding oblique orientation, branching from the upper lobe bronchus, branching into the subsegments of the apical bronchus, and artifacts.

Airway wall dimensions were quantified according to a validated method described by McNamara, Okazawa, and colleagues (10, 16). A slight modification to their method was made in that the CT images were enlarged 5 times and were analyzed on a workstation (X-tension; Toshiba, Tokyo, Japan). Regions of interest (ROI) were traced manually, using a mouse along the internal perimeter (the luminal border of the airway) and the external perimeter (the parenchymal border). In airways where a vessel abutted the external perimeter, an extrapolated line was traced on the assumption that the airway wall thickness was constant throughout the areas of vascular contact (10, 16). Luminal area (Ai) and total airway area (Ao) in mm² were determined automatically. Five measurements of both Ai and Ao in each airway were performed and averaged. Airway wall area (WA) was calculated as Ao  Ai. The percentage wall area (WA%) was calculated as (WA/Ao) × 100. Because Ai and WA may be affected by body size (21), Ai and WA corrected by body surface area (Ai/BSA and WA/BSA, mm²/m²) were also calculated as indices of airway wall dimensions. The largest luminal diameter (DL) and the largest luminal diameter perpendicular to DL (DS) were measured. Airways with a DL/DS ratio of  1.5 were not included because they were considered obliquely oriented (10).

Reproducibility of Airway Dimensions

Airway dimensions were measured in a blind fashion. All airway dimensions were measured by one observer (A.N.). Intraobserver error was tested by having this observer measure WA, WA%, and Ai in 20 randomly selected asthmatic or control subjects 2 times, separated by an interval of 2 wk. Interobserver error was determined by having two observers (A.N. and H.M.) measure WA, WA%, and Ai in 22 randomly selected asthmatic patients.

Intraobserver and interobserver reproducibility were assessed by plotting the difference between the two measurements against the average value of the two (26).

Serum ECP Measurement

To assess the intensity of airway inflammation, serum eosinophil cationic protein (ECP) levels were measured in the asthmatic patients as described previously (27). Briefly, blood was collected, stored for 60 ± 5 min at 25° C, and subsequently centrifuged at 1,300 g at 4° C for 10 min. The sera were stored at 20° C until measurement of ECP concentration by radioimmunosorbent assay (Pharmacia Diagnostics, Uppsala, Sweden).

IgE Measurement

Total and specific serum IgE antibody titers were measured by radioimmunosorbent testing using commercially available kits (Pharmacia & Upjohn, Tokyo, Japan). Patients were considered atopic when one or more specific IgE antibodies against house dust mite, cat dander, dog dander, weed pollen, grass pollen, and mold were positive.

Pulmonary Function

Indices of airflow obstruction, FEV₁, FEV₁/FVC, and mean forced expiratory flow during the middle half of the FVC (FEF_25-75%) were measured using a Chestac-65V (Chest, Tokyo, Japan).

Statistical Analysis

Data are expressed as means ± SD unless stated otherwise. Comparison between two groups was performed using unpaired t tests. Comparison of multiple groups was performed using analysis of variance (ANOVA) and Fisher's protected least significant difference tests or chi-square tests. For analysis of correlations, Pearson's correlation coefficient was calculated for data with a normal distribution, and Spearman's rank correlation test was used otherwise. A p value of < 0.05 was considered to indicate statistical significance.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Reproducibility

Plots of the average of, and difference between, the two measurements of WA or WA%, obtained to assess intraobserver or interobserver reproducibility, are shown in Figure 2. For each plot, the mean difference did not appreciably deviate from zero, and the limits of agreement were small. In addition, there was no obvious relation between measurement error and airway size.

View larger version (20K): [in this window] [in a new window]   Figure 2.   Intraobserver (A, B) and interobserver error (C, D) for the measurement of wall area (WA) and WA%. The average of and difference between two measurements are plotted. The dotted and dashed lines represent the mean and the mean ± 2 SD of the difference, respectively.

Intraobserver and interobserver reproducibility of Ai measurement was also good (data not shown).

Comparison Among Control Subjects and Patients with Various Severities

The clinical characteristics of the control subjects and the asthmatic patients classified according to severity are shown in Table 1. Among the five groups, age, but not sex distribution or BSA, differed significantly (p < 0.05 by ANOVA). Patients with intermittent asthma were significantly younger than control subjects and patients with moderate persistent or severe persistent asthma. The duration of asthma, total IgE level, prevalence of atopy, and serum ECP level did not differ among the four asthmatic subgroups.

Satisfactory CT images were obtained in all subjects. Among the five groups, FEV₁ (percentage of predicted), FEV₁/FVC, FEF_25-75% (percentage of predicted), WA, WA/BSA, and WA% differed significantly (p < 0.001 by ANOVA). Among patients with mild, moderate, and severe persistent asthma, WA, WA/BSA, and WA% were significantly higher in all patient groups and FEV₁, FEV₁/FVC, and FEF_25-75% were significantly lower in almost all patient groups than control groups. There was no significant difference between patients with intermittent asthma and control subjects. Patients with severe persistent asthma differed significantly from those with intermittent, mild, or moderate persistent asthma with respect to all these indices, except for WA%, which was similar in patients with severe persistent asthma and those with mild persistent asthma. Patients with moderate persistent asthma differed significantly from those with intermittent asthma with respect to FEV₁/FVC, FEF_25-75% (percentage of predicted), and WA/BSA and from those with mild persistent asthma with respect to FEV₁ (percentage of predicted), FEV₁/FVC, and FEF_25-75% (percentage of predicted). There was no difference between patients with intermittent asthma and those with mild persistent asthma in indices of airflow obstruction and airway wall thickness (Table 1, Figures 3 and 4). Neither Ai nor Ai/BSA differed among the five groups on ANOVA. These variables showed no decrease with increasing severity in asthmatic patients (Table 1).

View larger version (73K): [in this window] [in a new window]   View larger version (77K): [in this window] [in a new window]   View larger version (75K): [in this window] [in a new window]   Figure 3.   Representative CT images of (A) a control subject, (B) a patient with mild persistent asthma, and (C ) a patient with severe persistent asthma. The views of the apical bronchus of the right upper lobe are indicated by arrows.

View larger version (15K): [in this window] [in a new window]   Figure 4.   The distribution of WA/BSA in the control and four asthmatic groups. Bars represent means. The difference among the groups was significant by ANOVA (p < 0.001). See Table 1 for p values for comparison of each pair of groups. See Table 2 for the correlation coefficient between the asthma severity score and WA/BSA.

Relation between Airway Wall Thickness and Clinical Indices in All Asthmatic Patients

Comparison of all 81 asthmatic subjects with the 28 control subjects revealed significantly higher values for WA (26.8 ± 8.2 mm² versus 17.6 ± 4.3 mm², p < 0.001), WA/BSA (16.7 ± 4.9 mm²/m² versus 11.2 ± 3.0 mm²/m², p < 0.001), and WA% (63.7 ± 7.5% versus 55.1 ± 6.7%, p < 0.001).

When the 81 asthmatic subjects were analyzed together (Table 2), WA and WA/BSA showed a weak but significant correlation with the duration of asthma. Both WA and WA/ BSA also correlated with the severity of asthma when it was scored as 1 = intermittent, 2 = mild persistent, 3 = moderate persistent, and 4 = severe persistent. In addition, WA and WA/BSA negatively correlated with FEV₁, FEV₁/FVC, and FEF_25-75%. The only correlation found for WA% was with FEV₁.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our study shows that airway wall dimensions in asthmatic and healthy subjects can be reproducibly quantified by helical CT scanning, and that airway wall thickening is present in asthmatic patients. Such thickening may reflect the duration and severity of disease and the degree of airflow obstruction.

Quantitative analysis of airway wall thickness was first performed by Boulet and colleagues (13), who measured the thickness of the intermediate bronchus with the use of an electronic caliper and calculated the ratio of wall thickness to outer diameter (T/D ratio). Asthmatic patients with (n = 13) or without (n = 11) fixed airflow obstruction and 10 healthy control subjects were evaluated. The T/D ratio correlated with bronchial reactivity in asthmatic patients with fixed airflow obstruction, but it did not differ significantly among the three groups. They attributed the absence of a difference in airway wall thickness to limitations in measurement technique or to the fact that the measurement site (intermediate bronchus) was a central airway. Okazawa and colleagues (16) measured WA and WA% in six asthmatic patients and four normal control subjects with airways of various size, and showed that smaller airways (luminal diameter < 6 mm) are thicker in asthmatic patients. The method they used to measure airway dimensions, which was also used in our study, was developed on the basis of studies using excised canine lungs and was shown to be valid and reproducible (10). Recently, Awadh and colleagues (19) compared airway wall thickness in 40 patients who had various severities of asthma (15 with a history of a near-fatal attack, 13 with moderate asthma, and 12 with mild asthma) with that in 14 control subjects. They measured airway wall thickness and short axis luminal diameter in all segmental and subsegmental bronchi that had a luminal diameter of 1 mm or more. Calculated thickness/diameter ratio and WA% (as defined previously) were used as parameters of airway wall thickness. They found that all patient groups had greater airway wall thickness than the normal subjects and that patients with more severe asthma had greater airway wall thickening than those with milder asthma. They did not, however, assess the interobserver or intraobserver reproducibility of their method for CT measurement or investigate the relation of airway wall thickness to clincal indices other than severity.

We studied approximately twice the number of asthmatic and control subjects studied by Awadh and colleagues (19). Intraobserver and interobserver reproducibility of WA, WA%, and Ai measurements were tested and proven to be good. Increased airway wall thickness was most prominent in patients with more severe asthma. Both WA and WA/BSA significantly correlated with the duration, although weakly, and severity of asthma. Our findings may support the hypothesis that there is a progressive increase in airway wall thickening with increasing duration as well as severity of asthma (28). Significant negative correlations were found between WA or WA/BSA and FEV₁, FEV₁/FVC, or FEF_25-75%, and between WA% and FEV₁. As for the latter relation, WA% may have been affected by bronchoconstriction, which is supposed to decrease Ai and Ao (21). In contrast, this does not seem to apply to the relation between WA or WA/BSA and the indices of airflow obstruction, because previous CT studies have shown that methacholine-induced bronchoconstriction does not affect WA in asthmatic patients (16) or healthy control subjects (29). Our findings therefore indicate that airway wall thickening, as demonstrated by increased WA and WA/BSA, is related to airflow obstruction. The phenomenon that Ai or Ai/BSA did not decrease with increasing severity of asthma is difficult to interpret, but could be explained at least partly by the following speculations. The presence of bronchial dilatation or bronchiectasis has been reported in many descriptive CT studies of airway abnormalities in asthmatic patients (11, 14, 17, 18, 20, 21). The proportion of patients with evidence of bronchiectasis on CT may increase with increased severity of asthma (14, 17). Although the mechanism for the development of bronchiectasis in asthmatic patients is poorly understood and the presence or absence of "classic bronchiectasis" (30) was not evaluated in our study, it might have been present in our patients with more severe asthma and might have caused airway luminal area to increase. Another possible explanation involves the effect of lung volumes on airway dimensions. Lung volumes increase with increasing severity in asthmatic patients (31), and airway diameter, or luminal area, becomes larger as lung volumes increase (32). Increased severity of asthma therefore might have caused airway luminal area to increase by increasing lung volumes. In any case, our findings that airway wall thickness, but not luminal area, measured at a segmental bronchi related to the severity of asthma may suggest that the wall thickening of central airways may be a surrogate for luminal diameter changes at other sites within the tracheobronchial tree. The lack of a corelation between indices of airway wall thickness and the serum ECP level, which reflects ongoing airway inflammation (27, 33), may be related to the use of systemic or inhaled corticosteroids, which were being received by a substantial proportion of patients and were not withheld during the study. The lack of a relation between the serum ECP level and the severity of asthma, which conflicts with the findings of some previous reports (27, 33), may also have been the result of corticosteroid usage.

In this study, the "entire" airway wall thickness was quantified by measuring Ai and Ao and then calculating WA and WA% (10, 16), whereas in two previous studies, a single measurement of wall thickness and outer or luminal diameter itself was performed at a selected point in each bronchus, and the T/D ratio was calculated as described previously (13, 19). We believe that the former method yields less measurement bias (19). In addition, T/D ratio, or WA%, which was used in our study, is obviously sensitive to the degree of airway narrowing or dilatation (21) in contrast to WA, which is not affected by bronchoconstriction (16, 29). In our study, WA and WA/BSA, but not WA%, correlated with the duration and severity of asthma. Because bronchodilators were not used before CT scanning, WA% might have been affected by transient bronchoconstriction in some patients and therefore did not relate to "overall" asthma severity. The fact that Ai did not relate to asthma severity, as discussed earlier, might also have been involved. Regardless of the underlying cause, the stronger correlation of WA and WA/BSA with the clinical indices of asthma may support the hypothesis that WA is a more appropriate measure of airway wall thickness than WA% (21). We measured airway wall thickness at a single site, the apical bronchus of the right upper lobe, because it was less time-consuming and minimized the radiation dose (13). Airway wall thickness measured at this site was considered representative of that of the other bronchi, because a preliminary study showed a rather strong correlation between WA or WA% measured at this site and the values obtained at another segmental bronchus. We did not quantify the dimensions of smaller airways, because of problems in technique and reproducibility, as suggested by several investigators (10, 13, 15).

We conclude that the methodology used in our study is useful for the assessment of airway wall thickness in patients with asthma. We believe that it can be used in cross-sectional as well as longitudinal studies, such as those designed to evaluate the response to anti-inflammatory treatment.