Temporal assessment of airway remodeling in severe asthma using quantitative computed tomography.

To the Editor: Heterogeneity in asthma is evident in every aspect of the disease process (1–3). Quantitative computed tomography (QCT) has emerged as a reliable, noninvasive tool for assessment of proximal airway remodeling and air trapping in asthma (4). We have identified three asthma clusters based on QCT indices, using factor and cluster analysis (3). Subjects in clusters 1 and 3, with more severe asthma, had distinct patterns of proximal airway remodeling: cluster 1 showing a dilated right upper lobe apical segmental bronchus (RB1) lumen with wall thickening and cluster 3 had no wall thickening and markedly narrowed lumen. Subjects in cluster 2 had milder asthma, and there was a lack of proximal airway remodeling. It remains elusive whether airway structural changes reflect cause or effect; namely, are they a consequence of asthma and represent different stages of disease progression or the distinct remodeling changes that are fundamental to the pathogenesis of asthma, representing distinct asthma endotypes (5)? Our aim was to assess the temporal pattern of proximal airway remodeling in QCT-derived asthma clusters. Some of the results of this study have been previously reported in the form of an abstract (6). Twenty-two patients with severe asthma of mean (SEM) disease duration 28.6 (4) years, who were in the placebo arm of a previous study (7), were included in the analysis. All 22 patients had undergone two inspiratory thoracic CT scans to image RB1 and further inspiratory and expiratory full thoracic CT scans as part of research studies at our institute (3, 7). All CT scans were performed after administration of long-acting β2-agonist. The mean (range) duration between the first (baseline) and second CT scan was 1.6 (0.9–2.7) years and between the second and third was 2.6 (1.9–3.7) years. QCT-derived asthma clusters were determined on the basis of full thoracic paired inspiratory and expiratory CT scans obtained at time point 3 (3). Only inspiratory scans were used for the current analysis. Informed consent was obtained from all subjects and the studies were approved by the Leicestershire, Northamptonshire, and Rutland Research Ethics Committees. Fully automated software (VIDA Pulmonary Workstation, version 2.0; VIDA Diagnostics, Coralville, IA) was used for quantitative airway morphometry as described previously (3). RB1 wall area (WA)/body surface area (BSA) demonstrated a significant increase over time (mean [SEM]: first CT, 14.3 [0.9]; second CT, 14.7 [0.9]; third CT, 16.5 [1.3] mm2/m2; repeated measure analysis of variance [ANOVA], P = 0.008). No significant change was seen in RB1 lumen area (LA)/BSA (mean [SEM]: first CT, 9.1 [1.0]; second CT, 9.6 [1.0]; third CT, 9.9 [0.9]; repeated measure ANOVA, P = 0.4). There was an increase in RB1 length at the time of the third CT (mean [SEM]: first CT, 11.3 [0.8]; second CT, 11.0 [0.7]; third CT, 13.1 [0.6] mm; repeated measure ANOVA, P < 0.01). The change in RB1 WA/BSA (ΔRB1 WA/BSA = RB1 WA/BSA third CT − RB1 WA/BSA first CT) negatively correlated with change in RB1 length (Pearson correlation, −0.5; P = 0.03). When the subjects with severe asthma were split into previously described QCT-derived clusters (3), the mean (SEM) change in interval normalized RB1 WA/BSA and LA/BSA, respectively, was as follows: cluster 1 (n = 3), 3.6 (0.8) mm2/m2/year, 1.7 (1.1) mm2/m2/year; cluster 2 (n = 9), 1.0 (0.5) mm2/m2/year, −0.02 (0.4) mm2/m2/year; cluster 3 (n = 10), −0.1 (0.3) mm2/m2/year, 0.1 (0.4) mm2/m2/year (Figure 1). A one-way between-groups analysis of covariance (ANCOVA) was performed to compare the differences between clusters (independent variable), of airway mophometry at the time of the second and third CT (dependent variables) after controlling for airway morphometry at the time of first CT (covariate). After adjusting for airway morphometry (first CT), there were significant differences between the three clusters for RB1 WA/BSA (third CT) [F(2, 18) = 21, P < 0.001, partial η2 = 0.70] and for RB1 LA/BSA (third CT) [F(2, 18) = 32, P < 0.001, partial η2 = 0.78]. No significant difference was seen between the three clusters for RB1 WA/BSA (second CT) and RB1 LA/BSA (second CT) (data not shown). A comparison of airway morphometry in healthy control subjects at time point 3 with airway morphometry in severe asthma clusters at time points 1, 2, and 3 is presented in Table 1. Figure 1. Temporal assessment of airway remodeling in asthma clusters. Asthma clusters were determined on the basis of data from the third computed tomography (CT). Retrospective scans were available for temporal assessment of RB1 (right upper lobe apical segmental ... Table 1. RB1 Dimensions of Subjects with Severe Asthma and Healthy Subjects The subjects did not show any significant change in postbronchodilator FEV1% predicted (mean [SEM] change from baseline, −1.8 [2.7]; paired sample t test, P = 0.5), postbronchodilator FEV1/FVC (%) (mean [SEM] change from baseline, −0.7 [1.3]; paired sample t test, P = 0.6), asthma quality of life questionnaire (AQLQ) score (mean [SEM] change from baseline, 0.07 [1.3]; paired sample t test, P = 0.7), and sputum neutrophils (mean [SEM] change from baseline, 5.4 [7.1]; paired sample t test, P = 0.5) at the time of third CT scan compared with baseline. There was a statistically significant increase in the asthma control questionnaire (ACQ) (mean [SEM] change from baseline, 0.4 [0.2]; paired sample t test, P = 0.03). The change in RB1 QCT indices (LA/BSA, WA/BSA, and length) between third and first CT did not show any significant correlation with change in postbronchodilator FEV1% predicted, postbronchodilator FEV1/FVC%, ACQ, and AQLQ. Previous longitudinal studies have demonstrated a significant decrease in proximal airway wall dimensions after use of inhaled corticosteroids (ICS) (8, 9), an ICS/long-acting β2 agonist (LABA) combination (10), and anti-IgE treatment (11). In contrast, Brillet and colleagues found no change in CT-assessed airway dimensions in subjects with poorly controlled asthma treated for 12 weeks with inhaled LABA and ICS despite improvement in physiological measures of airway obstruction and air trapping (12). A follow-up of subjects with asthma on ICS from a previous study (8) for a mean duration of 4.2 years did not show any significant change in airway dimensions, with a reported mean (SEM) change in interval-normalized RB1 WA/BSA of −0.27 (0.59) mm2/m2/year (13). We have previously shown a decrease in RB1 WA/BSA in subjects with severe asthma after 1 year of treatment with anti–IL-5 compared with placebo, with an approximately 10% between-group change (7). In the current analysis subjects with severe asthma demonstrate a small, albeit significant temporal increase in RB1 WA/BSA but no change in RB1 LA/BSA. These varied patterns of airway remodeling exhibited by subjects with asthma may be explained by the heterogeneous nature of the disease, and differences in patient selection and duration of treatment and/or follow-up. A longitudinal study in subjects with severe asthma has demonstrated that in a multivariate regression model baseline %WA was a predictor of subsequent airway remodeling (14). In our analysis after adjusting for the RB1dimensions at time of first CT, significant differences were found in RB1 dimensions between severe asthma QCT-derived clusters at the time of third CT but not at the time of second CT. Patients with severe asthma, when grouped on the basis of QCT-derived clusters, show a differential temporal pattern of airway remodeling, particularly patients in cluster 3, where no significant change in airway wall or lumen dimensions was demonstrated over a period of 2.6 years. This suggests that the mechanism of lumen narrowing in this asthma phenotype may be due to decreased compliance of the airway wall or alteration between intrinsic and extrinsic airway wall properties (15), rather than thickened airway wall encroaching on the lumen. Mathematical modeling studies (16, 17) have also shown that thickening of the adventitia can uncouple the airway smooth muscle (ASM) from the lung’s elastic recoil forces, abating the airway–parenchymal interdependence. QCT based phenotyping could thus help us unravel novel asthma subtypes which may have distinct pathophysiological mechanisms. The inverse correlation between the change in RB1 WA/BSA and RB1 length suggests that despite bronchodilation, ASM shortening resulting in shortening of airway length may contribute to QCT-assessed airway wall thickening. We acknowledge that QCT-derived clusters were determined on the basis of full thoracic paired inspiratory (third CT in the current analysis) and expiratory CT scans as part of a recent study (3) and that temporal CT (first and second CT in the current analysis) data were obtained from retrospective scans. We therefore are unable to assess the stability of CT-derived phenotypes. Moreover, data are lacking in the current literature on the temporal stability of airway morphometry in healthy subjects. Temporal assessment was possible only in a small number of subjects in each cluster, with only three subjects in cluster 1, and therefore further verification of these findings is required by large longitudinal studies. Despite this limitation, temporal analysis may provide useful insight into the natural history of airway remodeling.