Abstract
Rationale: Airway remodeling in chronic obstructive pulmonary disease (COPD) is due to luminal narrowing and/or loss of airways. Existing computed tomographic metrics of airway disease reflect only components of these processes. With progressive airway narrowing, the ratio of the airway luminal surface area to volume (SA/V) should increase, and with predominant airway loss, SA/V should decrease.
Objectives: To phenotype airway remodeling in COPD.
Methods: We analyzed the airway trees of 4,325 subjects with COPD Global Initiative for Chronic Obstructive Lung Disease stages 0 to 4 and 73 nonsmokers enrolled in the multicenter COPDGene (Genetic Epidemiology of COPD) cohort. Surface area and volume measurements were estimated for the subtracheal airway tree to derive SA/V. We performed multivariable regression analyses to test associations between SA/V and lung function, 6-minute-walk distance, St. George’s Respiratory Questionnaire, change in FEV1, and mortality, adjusting for demographics, total airway count, airway wall thickness, and emphysema. On the basis of the change in SA/V over 5 years, we categorized subjects into predominant airway narrowing [positive ∆(SA/V) more than 0] and predominant airway loss [negative ∆(SA/V) less than 0] and compared survival between the two groups.
Measurements and Main Results: Airway SA/V was independently associated with FEV1/FVC (β = 0.12; 95% confidence interval [CI], 0.09–0.14; P < 0.001) and FEV1% predicted (β = 20.10; 95% CI, 15.13–25.08; P < 0.001). Airway SA/V was also independently associated with 6-minute-walk distance, respiratory quality of life, and lung function decline. Compared with subjects with predominant airway narrowing (n = 2,914; 66.3%), those with predominant airway loss (n = 1,484; 33.7%) had worse survival (adjusted hazard ratio for all-cause mortality = 1.58; 95% CI, 1.18–2.13; P = 0.002).
Conclusions: Computed tomography–based airway SA/V is an imaging biomarker of airway remodeling and provides differential information on predominant airway narrowing and loss in COPD. SA/V is associated with respiratory morbidity, lung function decline, and survival.
Keywords: chronic obstructive pulmonary disease, airway remodeling, phenotypes
At a Glance Commentary
Scientific Knowledge on the Subject
Airway remodeling in chronic obstructive pulmonary disease is due to luminal narrowing and/or loss of airways. Existing computed tomographic metrics of airway disease reflect only components of these processes and do not inform phenotyping of airway disease.
What This Study Adds to the Field
In a cohort of current and former smokers, we demonstrated that the surface area–to-volume ratio of the airway lumen is associated with respiratory morbidity, disease progression, and survival in chronic obstructive pulmonary disease. On the basis of the change in surface area–to-volume ratio over time, we found two trajectories of airway luminal loss, that of predominant airway narrowing and that of predominant airway loss. These airway remodeling phenotypes are seen in almost equal proportions in early and late disease stages, suggesting that these paths are specific to individuals and not necessarily sequential.
Airway disease associated with chronic obstructive pulmonary disease (COPD) is characterized by wall thickening, luminal narrowing, and loss of airways. In cross-sectional analyses, higher disease stages were associated with increased wall thickness, greater number of airways occluded by inflammatory mucus exudates, and loss of terminal bronchioles (1, 2). Although in vivo imaging of human lungs does not permit sufficient resolution to visualize terminal airways, a number of studies have reported similar findings in more proximal airways (3). These include increased thickening of segmental and subsegmental airways and a lower number of visualized airway branches (4–6). We recently calculated the airway fractal dimension to quantify the complex branching patterns of the airways, a measure that is affected by changes in airway luminal narrowing and airway loss (7).
Airflow obstruction in airway disease is due to luminal narrowing of the airways, mucus plugging, or airway loss. Although cross-sectional studies suggest that airway narrowing results in eventual loss of the airways (2, 8), this evolution in airway remodeling has not been demonstrated, and, in a given individual, it is likely that both processes occur concurrently. It remains unknown whether the predominance of either remodeling process results in differences in outcomes. Existing computed tomographic (CT) metrics of airway disease reflect only components of these processes and do not provide information on their relative contributions to overall airway remodeling (3).
In this context, we developed a new metric that may offer information on whether airway narrowing or airway loss is the predominant process over time. The human lung displays a recurrent branching pattern even in the last generations, such that a large surface area is possible in a small amount of space. A large alveolar surface area facilitates efficient gas exchange, and although airways do not participate in gas exchange, they need to necessarily divide in parallel to supply the large air space component without an increase in airway resistance. This results in progressively larger cross-sectional and luminal surface area for a given volume of airways with progression in airway generation. When narrowed, tubular structures such as the small airways lose more volume than surface area; when truncated, more surface area is lost than volume. Hence, we hypothesized that the ratio of surface area to the volume (SA/V) of the airway tree would provide information on the predominant process occurring in a given lung and that these airway phenotypes would be associated with differences in lung function, respiratory quality of life, disease progression, and survival.
Methods
Study Population
We included subjects who completed a 5-year follow-up visit as part of the COPDGene (Genetic Epidemiology of COPD) study. Subject inclusion is summarized in the Consolidated Standards of Reporting Trials diagram (see Figure E1 in the online supplement) included in the supplement. The protocol of the COPDGene study has been previously published (9). Briefly, COPDGene is a large multicenter study that enrolled current and former smokers between the ages of 45 years and 80 years across 21 clinical centers across the United States. Only individuals with at least a 10 pack-year smoking history were enrolled. Smoking status was classified as current if participants smoked within 30 days of study visit. Participants returned for follow-up at approximately 5 years after the initial visit, with continued follow-up phone calls every 6 months.
At each visit, participants underwent prebronchodilator and post-bronchodilator spirometry to assess lung function. Their 6-minute-walk distance (6MWD) was measured to quantify functional capacity (10). The St. George’s Respiratory Questionnaire (SGRQ) score was used to estimate respiratory quality of life (11). SGRQ scores range from 0 to 100, and a higher score indicates a worse quality of life. Post-bronchodilator FEV1/FVC <0.70 was used to define COPD, and severity stages were determined according to the Global Initiative for Chronic Obstructive Lung Disease (GOLD) report (12). We excluded participants with preserved ratio impaired spirometry (FEV1/FVC ≥0.70 and FEV1% predicted <80) (13). The change in lung function was assessed by calculating the difference between FEV1 at baseline and FEV1 at the 5-year follow-up visit. Vital status was assessed at 6-month follow-up phone calls, and deaths were confirmed via a search of the Social Security Death Index. All participants provided written informed consent, and the study protocol was approved by the institutional review board at each of the 21 participating centers.
CT Image Analysis
We analyzed high-resolution volumetric CT images acquired at full inspiration (TLC) (9, 14). CT scans were acquired post-bronchodilator administration. Lungs and airway trees were extracted using Thirona lung quantification software (Thirona) (15). Emphysema was quantified as the percentage of voxels less than −950 Hounsfield units at end inspiration. Segmental airway wall thickness was measured from the inspiratory CT scans using Thirona airway quantification software (15). The total airway count (TAC) was estimated by identifying branch points on the airway tree and counting the number of branches using MATLAB software (MathWorks).
Airway SA/V ratio
Airway trees segmented from the inspiratory CT scans were analyzed to extract quantitative data characterizing airway branching structure using the MATLAB software. The three-dimensional reconstructions of airway lumen were analyzed using built-in functions to provide an estimate of surface area and volume. Surface area was estimated using an algorithm based on run-length encoding of binary objects using the Crofton formula for n-dimensional objects (16). Surface area measurements were adjusted for voxel size. Volume was estimated as the count of the number of voxels in the airway tree, multiplied by the voxel size in millimeters. SA/V ratios were estimated for each airway tree in cm2/ml.
Simulation of airway loss
To evaluate the relative contributions of airway narrowing and airway loss to SA/V, we simulated these processes in a representative airway (7). Airway loss was simulated by iteratively cutting distal branches with specific lumen diameter per iteration from the segmented airway tree. Airway narrowing was achieved by applying a distance map on the airway mask and by implementing an iterative binary morphological thinning algorithm to simulate airway narrowing.
SA/V Ratio–based Subject Categorization
The airway SA/V ratio was estimated for each subject at both baseline and 5-year follow-up visit. We used the difference between baseline and follow-up SA/V ratio (SA/V at visit 2 minus SA/V at visit 1) to categorize subjects into the following two distinct categories: predominant airway narrowing [positive ∆(SA/V) more than 0] and predominant airway loss [negative ∆(SA/V) less than 0].
Statistical Analysis
Pearson’s correlations were used to estimate correlations between baseline airway SA/V ratio and lung function measures. Multivariable regression analyses were performed to test associations between baseline airway SA/V ratio and FEV1/FVC and FEV1% predicted after adjusting for age, race, sex, smoking status, pack-years of smoking, body mass index (BMI), CT emphysema, segmental airway wall thickness, baseline TLC, TAC, and CT scanner type. Associations between baseline airway SA/V ratio and 6MWD, SGRQ, and change in lung function were also adjusted for baseline FEV1 and number of exacerbations in the year before enrollment. Cox proportional hazards models were created to test the association of baseline SA/V ratio with all-cause mortality. Covariates included age, race, sex, smoking status, smoking pack-years, BMI, baseline TLC, and CT scanner type. We used Kaplan-Meier survival analysis with a log-rank test to compare survival between the predominant airway remodeling phenotypes and tested Cox proportional hazards models to calculate adjusted hazard ratio (HR) for all-cause mortality in the predominant airway loss category compared with the predominant airway narrowing category with adjustments for age, race, sex, smoking status, smoking pack-years, BMI, CT scanner type, CT emphysema, segmental airway wall thickness, TAC, and TLC (all measured at visit 2) and change in TLC between baseline and follow-up. The follow-up time for the survival analyses for associations with baseline SA/V ratio was initiated at the first visit and for comparisons between the two airway phenotypes was initiated at the second visit. A two-sided α <0.05 was considered statistically significant. All analyses were performed using R statistical software (version 3.6).
Results
Participant Characteristics
We analyzed data from 4,398 participants who completed a baseline and 5-year follow-up visit (Table 1), including 73 never-smokers, and 2,345, 449, 952, 472, and 107 participants with GOLD disease severity stages 0 through 4, respectively. The mean age of the cohort was 60.1 (8.7) years and the cohort included 49% women and 28.4% African American participants. Current smokers comprised 46.6% of the cohort. The mean (SD) airway SA/V ratio at baseline was 90.02 (11.93) cm2/ml for nonsmokers and was 84.61 (12.89) cm2/ml, 80.00 (12.92) cm2/ml, 73.80 (11.54) cm2/ml, 69.18 (10.81) cm2/ml, and 67.93 (9.87) cm2/ml for participants with GOLD stages 0 through 4, respectively (Jonckheere-Terpstra trend test; P < 0.001). Similarly, the mean (SD) airway SA/V ratio at the 5-year follow-up visit was 93.27 (11.89) cm2/ml for nonsmokers and was 87.68 (14.07) cm2/ml, 82.39 (13.91) cm2/ml, 74.41 (12.89) cm2/ml, 69.38 (11.18) cm2/ml, and 67.96 (10.35) cm2/ml for subjects with GOLD stages 0 through 4, respectively (trend test; P < 0.001). Table E1 in the online supplement shows a comparison of change in CT parameters between baseline and follow-up visits, stratified by GOLD stage.
Table 1.
Baseline Demographics of Participants (n = 4,398)
| Parameters | Results |
|---|---|
| Age, yr | 60.1 (8.7) |
| Sex, F, n (%) | 2,156 (49.0) |
| African American, n (%) | 1,251 (28.4) |
| Body mass index, kg/m2 | 28.5 (5.7) |
| Smoking, pack-years | 42.0 (24.1) |
| Current smokers, n (%) | 2,050 (46.6) |
| FEV1, L | 2.40 (0.9) |
| FEV1% predicted | 82.2 (23.7) |
| FEV1/FVC | 0.7 (0.1) |
| GOLD severity, n (%) | |
| 0 | 2,345 (53.3) |
| 1 | 449 (10.2) |
| 2 | 952 (21.6) |
| 3 | 472 (10.7) |
| 4 | 107 (2.4) |
| Nonsmokers, n (%) | 73 (1.7) |
| CT emphysema (percentage of voxels < −950 HU), % | 5.7 (8.4) |
| CT air trapping (percentage of voxels < −856 HU), % | 20.2 (17.8) |
| Airway wall thickness (segmental), mm | 1.1 (0.2) |
| Total airway count | 151.8 (69.3) |
| Airway SA/V ratio (baseline), cm2/ml | 79.8 (13.7) |
Definition of abbreviations: CT = computed tomographic; GOLD = Global Initiative for Chronic Obstructive Lung Disease; HU = Hounsfield units; SA/V = surface area–to-volume ratio of subtracheal airways.
All values are expressed as the mean (SD) unless specified otherwise.
Airway SA/V Ratio, Lung Function, and Respiratory Morbidity
There was a positive correlation between airway SA/V ratio and FEV1/FVC (r = 0.43; P < 0.001) as well as FEV1% predicted (r = 0.43; P < 0.001). After adjusting for age, race, sex, current smoking status, pack-years of smoking, BMI, CT emphysema, airway wall thickness, TLC, TAC, and CT scanner type, airway SA/V ratio (per 100 cm2/ml) was significantly associated with FEV1/FVC (adjusted β = 0.12; 95% confidence interval [CI], 0.09–0.14; P < 0.001) and FEV1% predicted (adjusted β = 20.10; 95% CI, 15.13–25.08; P < 0.001). Airway SA/V ratio (per 100 cm2/ml) was also associated with 6MWD (adjusted β = 148.86; 95% CI, 53.77–243.95; P = 0.002) and SGRQ score (adjusted β = −5.99; 95% CI, −10.86 to −1.11; P = 0.016) after also adjusting for baseline FEV1 and number of exacerbations in the year before enrollment.
Airway SA/V Ratio, Lung Function Decline, and Mortality
The median duration between the baseline and follow-up visits was 100 months (25th to 75th percentile = 91–110). Airway SA/V ratio at baseline (per 100 cm2/ml) was inversely associated with change in FEV1 (adjusted β = −21.77; 95% CI, −36.94 to −6.60; P = 0.004) after adjusting for age, race, sex, current smoking status, pack-years of smoking, BMI, CT scanner type, CT emphysema, airway wall thickness, TAC, number of exacerbations in the year before enrollment, TLC, and baseline FEV1. Baseline SA/V was also inversely associated with all-cause mortality (adjusted HR, 0.12; 95% CI, 0.04–0.35; P < 0.001), after adjusting for age, race, sex, BMI, CT scanner type, current smoking status, TLC, and pack-years of smoking.
Simulation of Airway Narrowing and Loss
Figure 1 shows the change in airway SA/V as a function of remaining airway volume percentage by iteratively narrowing airway branches and by randomly cutting branches from an airway tree of a representative participant. The SA/V ratio increases with progressive narrowing of the airway branches, whereas the SA/V ratio decreases with the iterative loss of airways. Although the SA/V ratio is affected in opposite directions by airway narrowing and loss, the change for a given decrease in airway volume is greater for airway loss, suggesting that airway loss is the main driver of changes in the SA/V ratio. The separation in curves can be noted with as little as 5% volume loss (Figure 1).
Figure 1.
Change in airway surface area–to-volume ratio as a function of airway volume by simulating airway narrowing and airway loss in an airway tree of a representative subject with Global Initiative for Chronic Obstructive Lung Disease stage 0 disease. The airway surface area–to-volume ratio increases with predominant airway narrowing and decreases with predominant airway loss. The x-axis shows remaining airway volume compared with baseline airway volume. The separation in airway volume between the two airway phenotypes can be seen with approximately 5% loss of airway volume.
Predominant Airway Narrowing versus Predominant Airway Loss
Of the 4,398 participants, 2,914 (66.3%) had a positive ∆SA/V between the baseline and follow-up visits, placing them in the predominant airway narrowing category, whereas 1,484 (33.7%) had a negative ∆SA/V, placing them in the predominant airway loss category. The mean (SE) of ∆SA/V was −5.60 (0.16) cm2/ml in the predominant loss category and 6.0 (0.09) cm2/ml in the predominant narrowing category (P < 0.001). Figure 2 shows the density distribution of ∆SA/V in the two airway phenotypes, stratified by baseline GOLD stage. The mean change in FEV1 between visits 1 and 2 over 5 years was −38.0 (48.6) ml/yr in the predominant airway narrowing group, compared with −45.7 (51.6) in the predominant airway loss group. Characteristics of participants in these airway remodeling categories is shown in Table 2. Of note, there was a significant decrease in TAC in the airway loss category but not in the airway narrowing category. The change in FEV1 was greater in those with predominant airway loss than in those with predominant airway narrowing (mean change, 45.7 ± 51.6 vs. 38.0 ± 48.6 ml/yr; P < 0.0001). Participants with predominant airway loss category had significantly greater CT gas trapping than those with predominant narrowing (26.7% ± 21.5% vs. 19.1% ± 17.7%; P < 0.001) at the 5-year visit. Compared with individuals with predominant airway narrowing, those with predominant airway loss also had greater degree of emphysema, thicker segmental airway walls, and lower TAC at both baseline and follow-up visits (Table 2). At visit 1, the proportion of individuals in each GOLD stage with predominant narrowing was 72.7%, 65.4%, 58.6%, 52.9%, and 50.4% for GOLD stages 0 through 4, respectively. At visit 2, the proportion of individuals in each GOLD stage with predominant narrowing was 75.1%, 68.3%, 57.8%, 54.5%, and 46.0% for GOLD stages 0 through 4, respectively. The comparable distribution of the two airway phenotypes in each GOLD stage (Table 2) suggests that these may be two separate pathways. Of those with mild disease (GOLD stages 0 and 1 combined), a higher proportion with predominant airway narrowing remained in the milder GOLD stages at visit 2 than did those with predominant airway loss (87% vs. 77%).
Figure 2.
Density distribution of the change (∆) in surface area–to-volume ratio (SA/V) in the predominant airway loss versus airway narrowing categories stratified by baseline GOLD stage. The difference between baseline and follow-up SA/V (SA/V at visit 2 minus SA/V at visit 1) was used to categorize subjects into the following two distinct categories: predominant airway narrowing [positive ∆(SA/V) > 0] and predominant airway loss [negative ∆(SA/V) < 0]. The mean ∆SA/V was −5.60 cm2/ml, with an SE of 0.16 cm2/ml in the predominant loss category, whereas the mean ∆SA/V was 6.0 cm2/ml with an SE of 0.09 cm2/ml in the predominant narrowing category (P < 0.001). GOLD = Global Initiative for Chronic Obstructive Lung Disease.
Table 2.
Comparison of Predominant Airway Remodeling Categories Defined by Airway SA/V Ratio
| Parameters | Airway Narrowing (n = 2,914) |
Airway Loss (n = 1,484) |
||
|---|---|---|---|---|
| Baseline | Follow-up | Baseline | Follow-up | |
| Age, yr | 59.7 (8.6) | 65.2 (8.5) | 60.8 (9.0) | 66.4 (8.9) |
| Sex, F, n (%) | 1,427 (49.0) |
729 (49.1) |
||
| African American, n (%) | 721 (24.7) |
530 (35.7) |
||
| Body mass index, kg/m2 | 28.5 (5.6) | 28.6 (5.9) | 28.5 (5.8) | 28.4 (6.3) |
| Smoking pack-years | 40.4 (23.1) | 41.8 (23.6) | 45.3 (25.8) | 46.8 (26.1) |
| Current smokers, n (%) | 1,284 (44.1) | 1,062 (36.5) | 766 (51.6) | 575 (38.8) |
| FEV1, L | 2.5 (0.8) | 2.3 (0.8) | 2.2 (0.9) | 1.9 (0.9) |
| FEV1% predicted | 84.9 (22.5) | 83.1 (24.4) | 76.9 (25.0) | 72.7 (26.5) |
| FEV1/FVC | 0.7 (0.1) | 0.7 (0.1) | 0.6 (0.2) | 0.6 (0.2) |
| GOLD severity, n (%) | ||||
| 0 | 1,705 (58.5) | 1,495 (51.9) | 640 (43.1) | 498 (34.4) |
| 1 | 294 (10.1) | 315 (10.9) | 155 (10.4) | 146 (10.1) |
| 2 | 558 (19.1) | 491 (17.0) | 394 (26.6) | 358 (24.7) |
| 3 | 250 (8.6) | 262 (9.1) | 222 (15.0) | 218 (15.0) |
| 4 | 54 (1.8) | 93 (3.2) | 53 (3.6) | 109 (7.5) |
| Nonsmokers, n (%) | 53 (1.8) | 49 (1.7) | 20 (1.4) | 17 (1.2) |
| CT emphysema (percentage of voxels < −950 HU), % | 5 (7.3) | 4.9 (8.1) | 7.3 (10.1) | 8.1 (11.5) |
| CT air trapping (percentage of voxels < −856 HU), % | 18.3 (16.4) | 19.1 (17.7) | 23.8 (19.8) | 26.7 (21.5) |
| Airway wall thickness (segmental), mm | 1.00 (0.21) | 0.99 (0.21) | 1.04 (0.22) | 1.06 (0.21) |
| Total airway count | 153.4 (69.3) | 165.1 (72.3) | 148.7 (69.2) | 120.6 (61) |
| Airway SA/V ratio (baseline), cm2/ml | 79.7 (13.8) | 85.7 (14.2) | 80.0 (13.6) | 74.5 (14.3) |
For definition of abbreviations, see Table 1.
All values are expressed as the mean (SD) unless specified otherwise.
Calculating survival from the 5-year follow-up visit onward, we had follow-up data for a median of 34 months (25th to 75th percentile = 23–45). A total of 246 (5.5%) participants died on follow-up. Figure 3 shows the survival curves for the two airway remodeling categories. After adjusting for age, race, sex, BMI, current smoking status, pack-years of smoking, FEV1, CT emphysema (%), airway wall thickness, TAC, CT scanner type, TLC, and change in TLC between baseline and follow-up (all at visit 2), individuals with predominant airway loss had worse survival compared with individuals with predominant airway narrowing (adjusted HR, 1.58; 95% CI, 1.18–2.13; P = 0.002).
Figure 3.
Kaplan-Meier survival curves for the two airway phenotypes based on the change in airway surface area–to-volume ratio between baseline and at 5-year visit. Follow-up time was initiated at the 5-year visit.
Discussion
We demonstrated that the SA/V ratio of the airway lumen is associated with respiratory morbidity, disease progression, and survival in COPD. Our study has two important findings. First, there are potentially two trajectories of airway luminal loss, including that of predominant airway narrowing and that of predominant airway loss. The direction of change in the SA/V ratio can be used to phenotype predominant airway remodeling processes. Second, these airway remodeling phenotypes are seen in almost equal proportions in early and late disease stages, suggesting that these paths are specific to individuals and not necessarily sequential.
In contrast to quantification of emphysema, the assessment of airway disease has been challenging given the complex recurring branching patterns and the change in airway size by generation, precluding the quantification of airway disease in a way that captures all aspects of airway remodeling. Wall thickness of the segmental airways is affected by the size of the airway measured, which confounds comparisons between individuals. Wall area percentage, the percentage of the airway cross-sectional area occupied by the wall, adjusts for airway size but appears to be more affected by wall thickness than actual changes in the luminal diameter (5). A summary measure to account for changes in the wall thickness of airways of various sizes that is not confined to the larger airways is the Pi10, the square root of the wall area of a hypothetical airway with a 10-mm internal perimeter (4). Although these measures of airway wall thickness are associated with respiratory outcomes, they do not account for eccentric wall thickening or luminal changes. In any case, the main determinant of airway resistance and, hence, airflow obstruction is not the wall thickness but the luminal size.
Metrics of lumen size are reliant on the airway generation that is measured. Airway volume estimates have included both the luminal volume of the segmental airways (17) and the volume of the entire airway tree as a summary measure of airway luminal size (18). The latter, however, is disproportionately affected by large airway size. We recently used fractal analyses to quantify airway remodeling and showed significant associations between airway fractal dimension and important patient-reported outcomes in COPD (7). Although both airway volume and airway fractal dimension account for airway narrowing and loss, they do not provide information on the relative importance of each process. We extend the literature by showing that temporal trends in the SA/V ratio can distinguish the predominant process involved in airway remodeling. Compared with individuals with predominant airway narrowing, those with predominant airway loss had a greater degree of emphysema and gas trapping, thicker segmental airway walls, and a lower TAC at both baseline and follow-up visits. These data suggest that predominant airway loss is associated with worse downstream parenchymal and airway measures than airway narrowing. It is pertinent to note that an assessment of TAC can provide temporal information about airway loss, but this measure does not account for narrowing. Importantly, although SA/V change over 5 years was used to determine predominant airway groups, it is apparent from the simulation data that the curves diverge when only 5% of the airway volume is lost.
Although it is likely that luminal narrowing leads to airway loss and that these occur as a continuum, these pathways do not seem to converge over a period of at least 5 years, which is a considerable amount of time. A few mechanisms may be at play to cause this divergence. These may be parallel processes whereby some individuals may have nonuniform narrowing and disproportionate airway loss in some lung segments, whereas in some individuals, airway narrowing occurs rather uniformly without losing many airways. This is supported by the fact that there was an equal distribution of predominant narrowing and loss even at baseline in individuals who had accumulated considerable amount of disease and had stages 3 and 4 airflow obstruction. It is also possible that in some airways, there is damage to the radial fibers that predominantly results in a narrowing of the airways, whereas others have more damage to the axial fibers linking the parent and daughter airways with resultant axial recoil of the airways that leads to loss (19).
We acknowledge that in many participants, narrowing and loss of airways occur in parallel, but detecting the dominant trajectory is important. The SA/V ratio is influenced by both these processes, and hence, the classification based on predominant change adds value to airway phenotyping. The efficacy of bronchodilator and other inhaled therapies in COPD remains modest, and the response is not uniform in all individuals. It is likely that airway loss is irreversible, but airway narrowing, by virtue of a multitude of possible mechanisms, including untethering of airways due to adjacent emphysema, mucus hypertrophy, mucus, and bronchospasm, may be reversible. Although mucus plugging can also result in an appearance of loss of airways, these plugs are accounted for in the SA/V calculation. These mucus plugs can be persistent over many years and are not necessarily a reversible cause of airway loss. In individuals with milder disease, those with predominant narrowing were more likely to remain in the milder GOLD stages than those with predominant airway loss. The identification of these airway phenotypes is therefore of importance.
Our study has several strengths. First, we included well-characterized participants who had a wide range of disease severity. Second, the COPDGene study had stringent quality-control measures for both spirometry and CT. Third, we showed the utility of this novel measure independently of TAC and airway wall thickness. Our study also has a few limitations. The phenotyping into two main categories was done using data collected 5 years apart. We do, however, note that even a 5% change in airway volume is sufficient to identify predominant airway remodeling categories, a change that is likely achieved within 1 year of follow-up in most individuals with COPD. We did not include the trachea in the calculation of the SA/V ratio because this dominates the signal and reduces the sensitivity to detect change in the smaller airways. Although the tracheal size may determine the size of the downstream airways, we believe our methodology is robust, as we did include the large main stem airways. Excluding the trachea may also enable the detection of changes in the smaller airways where disease accrues. We adjusted all analyses for CT scanner type, but some influence of scanner variation is expected. The CT scans were not volume gated, but subjects were coached to full inspiration. Deep inhalation may result in bronchodilation, which theoretically can increase surface area more than the volume and cause a rise in SA/V. We attenuated some of this influence by adjusting for lung volume in all analyses.
Conclusions
CT-based airway SA/V ratio is an imaging biomarker of airway remodeling in COPD, which is significantly associated with respiratory morbidity, lung function decline, and survival, and provides differential information on airway remodeling trajectories. Predominant airway narrowing and airway loss are both associated with respiratory morbidity, but mortality is higher in those with predominant airway loss.
Supplementary Material
Footnotes
Supported by NHLBI grants R01 HL151421 (S.P.B. and A.N.) and K23HL133438 (S.P.B.) and National Institute of Biomedical Imaging and Bioengineering grant R21EB027891 (S.P.B.). The COPDGene (Genetic Epidemiology of Chronic Obstructive Pulmonary Disease) study is supported by NIH grants U01 HL089897 and U01 HL089856. COPDGene is also supported by the COPD Foundation through contributions made to an Industry Advisory Board consisting of AstraZeneca, Boehringer Ingelheim, Genentech, GlaxoSmithKline, Novartis, Pfizer, Siemens, and Sunovion.
Author Contributions: Study concept and design: S.B. and S.P.B. Acquisition, analysis, or interpretation of data: all authors. Drafting of the manuscript: S.B. and S.P.B. Critical revision of the manuscript for important intellectual content: all authors. Statistical analysis: S.B. and S.P.B. S.P.B. had full access to all of the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.
This article has an online supplement, which is accessible from this issue’s table of contents at www.atsjournals.org.
Originally Published in Press as DOI: 10.1164/rccm.202004-0951OC on August 5, 2020
Author disclosures are available with the text of this article at www.atsjournals.org.
References
- 1.Hogg JC, Macklem PT, Thurlbeck WM. Site and nature of airway obstruction in chronic obstructive lung disease. N Engl J Med. 1968;278:1355–1360. doi: 10.1056/NEJM196806202782501. [DOI] [PubMed] [Google Scholar]
- 2.Hogg JC, Chu F, Utokaparch S, Woods R, Elliott WM, Buzatu L, et al. The nature of small-airway obstruction in chronic obstructive pulmonary disease. N Engl J Med. 2004;350:2645–2653. doi: 10.1056/NEJMoa032158. [DOI] [PubMed] [Google Scholar]
- 3.Bodduluri S, Reinhardt JM, Hoffman EA, Newell JD, Jr, Bhatt SP. Recent advances in computed tomography imaging in chronic obstructive pulmonary disease. Ann Am Thorac Soc. 2018;15:281–289. doi: 10.1513/AnnalsATS.201705-377FR. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Grydeland TB, Dirksen A, Coxson HO, Eagan TM, Thorsen E, Pillai SG, et al. Quantitative computed tomography measures of emphysema and airway wall thickness are related to respiratory symptoms. Am J Respir Crit Care Med. 2010;181:353–359. doi: 10.1164/rccm.200907-1008OC. [DOI] [PubMed] [Google Scholar]
- 5.Washko GR, Diaz AA, Kim V, Barr RG, Dransfield MT, Schroeder J, et al. Computed tomographic measures of airway morphology in smokers and never-smoking normals. J Appl Physiol (1985) 2014;116:668–673. doi: 10.1152/japplphysiol.00004.2013. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Kirby M, Tanabe N, Tan WC, Zhou G, Obeidat M, Hague CJ, et al. CanCOLD Collaborative Research Group; Canadian Respiratory Research Network; CanCOLD Collaborative Research Group, the Canadian Respiratory Research Network. Total airway count on computed tomography and the risk of chronic obstructive pulmonary disease progression: findings from a population-based study. Am J Respir Crit Care Med. 2018;197:56–65. doi: 10.1164/rccm.201704-0692OC. [DOI] [PubMed] [Google Scholar]
- 7.Bodduluri S, Puliyakote ASK, Gerard SE, Reinhardt JM, Hoffman EA, Newell JD, Jr, et al. COPDGene Investigators. Airway fractal dimension predicts respiratory morbidity and mortality in COPD. J Clin Invest. 2018;128:5374–5382. doi: 10.1172/JCI120693. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.McDonough JE, Yuan R, Suzuki M, Seyednejad N, Elliott WM, Sanchez PG, et al. Small-airway obstruction and emphysema in chronic obstructive pulmonary disease. N Engl J Med. 2011;365:1567–1575. doi: 10.1056/NEJMoa1106955. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Regan EA, Hokanson JE, Murphy JR, Make B, Lynch DA, Beaty TH, et al. Genetic epidemiology of COPD (COPDGene) study design. COPD. 2010;7:32–43. doi: 10.3109/15412550903499522. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.ATS Committee on Proficiency Standards for Clinical Pulmonary Function Laboratories. ATS statement: guidelines for the six-minute walk test. Am J Respir Crit Care Med. 2002;166:111–117. doi: 10.1164/ajrccm.166.1.at1102. [DOI] [PubMed] [Google Scholar]
- 11.Jones PW, Quirk FH, Baveystock CM, Littlejohns P. A self-complete measure of health status for chronic airflow limitation: the St. George’s Respiratory Questionnaire. Am Rev Respir Dis. 1992;145:1321–1327. doi: 10.1164/ajrccm/145.6.1321. [DOI] [PubMed] [Google Scholar]
- 12.Vogelmeier CF, Criner GJ, Martinez FJ, Anzueto A, Barnes PJ, Bourbeau J, et al. Global strategy for the diagnosis, management, and prevention of chronic obstructive lung disease 2017 report: GOLD executive summary. Am J Respir Crit Care Med. 2017;195:557–582. doi: 10.1164/rccm.201701-0218PP. [DOI] [PubMed] [Google Scholar]
- 13.Wan ES, Castaldi PJ, Cho MH, Hokanson JE, Regan EA, Make BJ, et al. COPDGene Investigators. Epidemiology, genetics, and subtyping of preserved ratio impaired spirometry (PRISm) in COPDGene. Respir Res. 2014;15:89. doi: 10.1186/s12931-014-0089-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Bhatt SP, Washko GR, Hoffman EA, Newell JD, Jr, Bodduluri S, Diaz AA, et al. Imaging advances in chronic obstructive pulmonary disease: insights from COPDGene. Am J Respir Crit Care Med. 2019;199:286–301. doi: 10.1164/rccm.201807-1351SO. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15.Charbonnier JP, Pompe E, Moore C, Humphries S, van Ginneken B, Make B, et al. COPDGene investigators. Airway wall thickening on CT: relation to smoking status and severity of COPD. Respir Med. 2019;146:36–41. doi: 10.1016/j.rmed.2018.11.014. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Lehmann G, Legland GL. Efficient N-dimensional surface estimation using Crofton formula and run-length encoding. Insight J. 2012 [Google Scholar]
- 17.Diaz AA, Rahaghi FN, Ross JC, Harmouche R, Tschirren J, San José Estépar R, et al. COPD Gene investigators. Understanding the contribution of native tracheobronchial structure to lung function: CT assessment of airway morphology in never smokers. Respir Res. 2015;16:23. doi: 10.1186/s12931-015-0181-y. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 18.De Backer W, De Backer J, Vos W, Verlinden I, Van Holsbeke C, Clukers J, et al. A randomized study using functional respiratory imaging to characterize bronchodilator effects of glycopyrrolate/formoterol fumarate delivered by a metered dose inhaler using co-suspension delivery technology in patients with COPD. Int J Chron Obstruct Pulmon Dis. 2018;13:2673–2684. doi: 10.2147/COPD.S171707. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Mitzner W. Emphysema: a disease of small airways or lung parenchyma? N Engl J Med. 2011;365:1637–1639. doi: 10.1056/NEJMe1110635. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.