Abstract
The characteristics that predict progression to overt chronic obstructive pulmonary disease (COPD) in smokers without spirometric airflow obstruction are not clearly defined.
We conducted a post hoc analysis of 849 current and former smokers (≥20 pack–years) with preserved spirometry from the Subpopulations and Intermediate Outcome Measures in COPD Study (SPIROMICS) cohort who had baseline computed tomography (CT) scans of lungs and serial spirometry. We examined whether CT-derived lung volumes representing air trapping could predict adverse respiratory outcomes and more rapid decline in spirometry to overt COPD using mixed-effect linear modelling.
Among these subjects with normal forced expiratory volume in 1 s (FEV1) to forced vital capacity (FVC) ratio, CT-measured residual volume (RVCT) to total lung capacity (TLCCT) ratio varied widely, from 21% to 59%. Over 2.5±0.7 years of follow-up, subjects with higher RVCT/TLCCT had a greater differential rate of decline in FEV1/FVC; those in the upper RVCT/TLCCT tertile had a 0.66% (95% CI 0.06%–1.27%) faster rate of decline per year compared with those in the lower tertile (p=0.015) regardless of demographics, baseline spirometry, respiratory symptoms score, smoking status (former versus current) or smoking burden (pack–years). Accordingly, subjects with higher RVCT/TLCCT were more likely to develop spirometric COPD (OR 5.7 (95% CI 2.4–13.2) in upper versus lower RVCT/TLCCT tertile; p<0.001). Other CT indices of air trapping showed similar patterns of association with lung function decline; however, when all CT indices of air trapping, emphysema, and airway disease were included in the same model, only RVCT/TLCCT retained its significance.
Increased air trapping based on radiographic lung volumes predicts accelerated spirometry decline and progression to COPD in smokers without obstruction.
Abstract
Radiographic lung volumes and related computed tomography measures that represent air trapping are associated with an accelerated decline in lung function and can identify susceptible smokers at increased risk of progressing to overt COPD http://bit.ly/32QqiKQ
Introduction
Chronic obstructive pulmonary disease (COPD) is a heterogeneous disease that affects only a fraction of those who smoke tobacco [1–4]. Although nearly all smokers have evidence of chronic airway inflammation [5, 6], only about 20% of them develop chronic airflow obstruction that meets the definition of COPD [1]. The origin of this widely variable susceptibility to develop COPD has not been well elucidated, and the ability to identify which smokers without airflow obstruction are at the highest risk for development of respiratory problems and lung function decline is of great interest for prognostication and intervention purposes.
Air trapping, defined as abnormally increased volume of air remaining in the lungs at the end of exhalation, is a manifestation of obstructive lung disease and is associated with increased dyspnoea, reduced functional capacity and higher mortality [7, 8]. However, its consequence in those at risk for COPD but with preserved spirometry (normal forced expiratory volume in 1 s (FEV1) to forced vital capacity (FVC) ratio) demands further examination. A recent retrospective study of the United States Veterans Administration electronic health records showed abnormal lung volumes and air trapping, as measured by plethysmography, to be present in >30% of smokers with preserved spirometry and to be associated with adverse respiratory outcomes and progression to spirometric COPD [9]. However, there has been no prospective validation of the utility of lung volumes as predictor of future lung function decline and progression to overt COPD.
In this study, we hypothesised that in individuals at risk for COPD due to smoking but with preserved spirometry, those with increased ratio of residual volume (RV) to total lung capacity (TLC), an index that represents air trapping, would have faster rates of lung function decline and progression to develop spirometric COPD. To examine this hypothesis, we used computed tomography (CT)-derived measures of lung volumes and clinical data prospectively collected on current and former smokers without COPD from the National Heart, Lung, and Blood Institute-funded Subpopulations and Intermediate Outcome Measures in COPD Study (SPIROMICS) [10], and investigated whether CT measures of lung volumes representing air trapping could predict subsequent development of spirometric COPD and increased morbidity. We also examined whether other CT indices of air trapping, emphysema and airway disease had additional contributions towards the above outcomes beyond that from lung volumes.
Methods
Study design
The SPIROMICS multicentre observational study enrolled 2981 participants from 2010 through 2015 [10]. The study included persons 40–80 years of age who were either never-smoking healthy persons or current and former smokers who had a smoking history ≥20 pack–years, with or without a clinical diagnosis of obstructive lung disease. Participants were categorised using the Global Initiative on Obstructive Lung Disease (GOLD) staging system according to the results on spirometry performed before and after four inhalations each of albuterol at a dose of 90 μg per inhalation and ipratropium at a dose of 18 μg per inhalation [11]. Current and former smokers who had a concomitant diagnosis of asthma were not excluded. Respiratory symptoms, exacerbation history, exercise capacity by 6-minute walk distance (6-MWD) testing, and CT scans of the lung were obtained, and subjects were followed for a target follow-up time of 3 years with planned annual serial spirometry and symptoms questionnaires, as previously described [10, 12]. Lung volumes representing air trapping were measured from full inspiratory (TLC) and full expiratory (RV) CT imaging of lungs. Other CT indices of air trapping, emphysema, and airway disease were also measured as described below.
From the 849 current and former smokers with preserved spirometry (FEV1/FVC ≥0.70 after bronchodilator use and FVC equal to or above the predicted lower limit of normal [13], complete data for this analysis were available for 814 subjects (figure 1). CT-measured lung volumes with high confidence in their accuracy were available from 618 of the 814 subjects as described in the supplementary appendix and shown in supplementary figures S1 and S2. Using this cohort (described in supplementary table S1), we conducted a post hoc analysis to determine whether baseline radiographically measured lung volumes representing air trapping (ratio of CT-measured RV to TLC (RVCT/TLCCT)) could predict more rapid decline in spirometry to overt COPD and worse respiratory symptoms.
Subject flow. CT: computed tomography; SVC: slow vital capacity; VCCT: CT-measured vital capacity; RVCT/TLCCT: CT-measured residual volume to total lung capacity ratio.
CT Indices of lung volumes, air trapping, emphysema and small airways
The detailed protocol and quality assessment of SPIROMICS CT scans have been described previously [14]. Briefly, SPIROMICS has an established quantitative CT lung assessment system (QCT-LAS), which includes scanner-specific imaging protocols for lung assessment at TLC and RV. Written breath-holding instructions were supplied to the CT technologists, who were instructed to coach the subject, as in a pulmonary function laboratory, to achieve both TLC and RV with a series of preceding deep inspirations. To provide imaging speeds that allow proper breath-holds from subjects, only 64-detector rows or higher scanners were used.
In addition to RVCT/TLCCT, other CT indices of air trapping, including the percentage of the lung voxels with attenuation <−856 HU on the expiratory CT images (Exp−856) [15, 16] and parametric response mapping of functional small airway disease (PRMfSAD) [17, 18], were also used in the analysis. Moreover, measures of emphysema including the percentage of the lung voxels on inspiratory CT images with attenuation <−950 HU (Insp−950) and parametric response mapping of emphysema (PRMEMPH) [17, 18], and measures of airway disease including the average and thickest values for the square root of wall area of a hypothetical airway with 10 mm internal perimeter (Pi10) [19] were also examined as additional predictors in the analysis.
Statistical and data analysis
The distribution of RVCT/TLCCT was computed and its correlations with airflow obstruction indices (FEV1/FVC and FEV1) were examined using the Pearson correlation test. To control for age, sex, height and weight covariates when examining the raw RVCT/TLCCT values, partial correlations corrected for covariates were derived and examined [20]. To examine these distributions in more detail, airflow indices were partitioned in 5% increments and summary statistics were calculated across each partition.
Outcome variables, including spirometric indices, symptoms (modified Medical Research Council dyspnoea scale (mMRC), COPD assessment test (CAT), Saint George's Respiratory Questionnaire (SGRQ) and Short Form 12-item Survey (SF12)), body–mass index, airflow obstruction, dyspnoea and exercise capacity (BODE) index, exercise capacity (6-MWD test) and respiratory exacerbations (frequency and time to event) were examined longitudinally. Changes in the outcomes were calculated by subtracting the subsequent visits (V2, V3 or V4) outcome values from those of baseline visit (V1), and then analysed using mixed effect modelling as described below.
Because there are no validated reference values for CT-measured lung volumes, we divided the subjects into three equal groups based on their RVCT/TLCCT to form distinct categories of low, intermediate, and high RVCT/TLCCT, with the assumption that low and high RVCT/TLCCT tertile groups would likely represent those subjects with normal and abnormal lung volumes, respectively. We used these tertile groups in the analysis as a categorical variable that would represent risk of progression to spirometric COPD.
The effects of RVCT/TLCCT (both as a continuous and as a categorical variable) on changes in outcomes were examined using mixed-effect linear regression, with a nested random subject and study site effect, and fixed effect variables, including age, sex, height, weight, smoking status (current versus former), smoking burden (pack–years of smoking), corresponding baseline lung function or symptom or activity measurements (e.g. baseline FEV1/FVC when evaluating change in FEV1/FVC as an outcome or baseline mMRC when evaluating change in mMRC as an outcome), and follow-up time to repeat outcome measurement as described in the supplementary appendix. Interaction models were fit with the inclusion of the main effect for 1) follow-up time and 2) smoking status (current versus former smoker), and their interactions with RVCT/TLCCT strata. To demonstrate statistical significance, p-values from mixed-effect linear regression modelling, as well as the 95% confidence intervals for comparisons of each RVCT/TLCCT category effect estimate to that of the reference value, were calculated.
The analysis of association between progression to spirometric COPD and RVCT/TLCCT was performed using mixed-effect logistic regression modelling with a nested random time and site effect, and fixed effect variables including age, sex, height, weight, smoking status (former versus current smoking), and smoking burden (pack–years of smoking).
To examine the relevance of RVCT/TLCCT in the risk-prediction model for COPD development, we performed receiver operating characteristic (ROC) analysis and calculated its incremental contribution to the model beyond that of other covariates.
Cox proportional hazards regression modelling was used to analyse the association of RVCT/TLCCT and CT indices of air trapping with time to the first hospitalisation. In addition, the association of those indices with number of severe respiratory exacerbations, as defined by number of emergency room and hospital admissions, were analysed using mixed-effect Poisson regression modelling to determine the incident rate ratios (IRR) of such events with consideration of follow-up time and study site.
We performed sensitivity analyses by simultaneous inclusion of variables that could act as confounders as additional terms in the regression models including hip-to-waist ratio and bronchodilator responsiveness (≥12% and ≥200 mL increase in FEV1 after bronchodilator administration). Separate sensitivity analyses were performed to evaluate the effect of presence or absence of respiratory symptoms (as measured by CAT questionnaire score of < versus≥10) on associations of RVCT/TLCCT and other CT air trapping indices with lung function outcomes. Additional sensitivity analyses were also performed by excluding patients with specific characteristics that could act as confounders, including smoking status (current versus former smoker), obesity or asthma separately.
Results
Correlation between baseline lung volumes and airflow indices
Among the 618 subjects with high concordance between CT-measured vital capacity (VCCT) and slow vital capacity (SVC), baseline RVCT/TLCCT had weak-to-moderate inverse correlations with baseline FEV1/FVC and FEV1 (covariate-corrected correlations of 0.21 and 0.28, respectively; p<0.001) (figure 2) (supplementary table S2). Nevertheless, RVCT/TLCCT had a wide distribution across normal ranges of these airflow indices spanning from 21% to 59%. This distribution corresponded to maximum coefficient of variations (standard deviation to mean ratio) of 19.3% and 20.5% across 5% (percent predicted) increments of FEV1/FVC and FEV1, respectively (figure 2).
Correlation between computed tomography-measured residual volume to total lung capacity ratio (RVCT/TLCCT) and forced expiratory volume in 1 s (FEV1)/forced vital capacity (FVC) or FEV1 in smokers with preserved spirometry. Boxplots show the distribution of RVCT/TLCCT (raw value) by 5% increments in FEV1/FVC % predicted (Panel a), and 5% increments in FEV1 % predicted (Panel b). Subjects were stratified into tertiles of RVCT/TLCCT represented by green, blue and magenta for low, intermediate, and high RVCT/TLCCT strata, respectively. The black line represents the regression line for all the points. r: correlation coefficient; rp: partial correlation, which is the correlation coefficient between the dependent variable and the targeted independent variable with the effect of other controlling random variables removed.
Association of lung volumes with progression to spirometric COPD
Follow-up spirometry was available in 496 out of 618 subjects with high VCCT and SVC concordance. The median follow-up time from baseline spirometry (V1) to the last spirometry available was 2.7 years (interquartile range from 2.0 to 3.0 years and total range from 0.5 to 4.2 years; average follow-up time was 2.5±0.7 years). Among the 496 subjects with at least one follow-up spirometry, 295 had two and 157 had three follow-up spirometries (table 1). The average raw value of RVCT/TLCCT was 40±7% for the entire cohort, and 33±3%, 40±2% and 48±4% for the low, intermediate and high tertiles, respectively. Overall, 16.7% of the 496 subjects progressed to meet the spirometric definition of COPD during the median 2.7 years of follow up (unadjusted proportions of 6.2%, 16.1% and 27.7% for low, intermediate and high RVCT/TLCCT groups, respectively; table 1).
Characteristics of smoker subjects with preserved spirometry who had follow-up spirometry
During this follow-up period, and after adjustment for covariates (age, sex, height, weight, smoking status, smoking burden and baseline FEV1/FVC), FEV1/FVC ratio declined in an RVCT/TLCCT-dependent manner such that a 10% higher baseline RVCT/TLCCT was associated with a 1.1% higher absolute decline in FEV1/FVC over the follow-up period (p<0.001) (tables 2 and 3 and figure 3a). Accordingly, subjects with higher baseline RVCT/TLCCT were more likely to develop spirometric COPD (table 4). For example, a 1% higher absolute RVCT/TLCCT value on baseline CT resulted in 10.8% higher likelihood of developing spirometric COPD during the follow-up period (OR (95%CI) 1.108 (1.056–1.162); p<0.001), which translates to nearly tripling the likelihood of developing COPD for every 10% higher RVCT/TLCCT (OR (95%CI) 2.779 (1.721–4.486); p<0.001).
Associations of changes in lung function or symptoms with computed tomograpy-measured residual volume/total lung capacity (RVCT/TLCCT)
Associations of changes in lung function or symptoms with computed tomography-measured residual volume/total lung capacity (RVCT/TLCCT) strata
Association of spirometric COPD development with computed tomography-measured residual volume/total lung capacity (RVCT/TLCCT)
Comparison of change in spirometry from different SPIROMICS visits across CT-measured RV/TLC (RVCT/TLCCT) strata. Line graphs of FEV1/FVC values predicted from mixed-effect regression modelling (“fitted values”) through time across RVCT/TLCCT strata. Subjects were stratified into tertiles of RVCT/TLCCT represented by green, blue and magenta for low, intermediate and high RVCT/TLCCT tertiles, respectively. The tick marks on the x-axes represent the time that each spirometry was performed during the course of the study. Panel a shows the change in FEV1/FVC (predicted from the main model) and panel b shows the difference in rate of FEV1/FVC change per year (predicted from the spirometry follow-up time interaction model). FEV1: forced expiratory volume in 1 s; FVC: forced vital capacity.
To better understand the risk associated with developing COPD in smokers with preserved spirometry, we examined the likelihood of lung function decline to spirometric COPD for subjects in low, intermediate and high tertiles of RVCT/TLCCT. We found that subjects with high RVCT/TLCCT had greater decline compared with those with intermediate RVCT/TLCCT, and those with intermediate RVCT/TLCCT had greater decline compared with those with low RVCT/TLCCT (p=0.005) (table 3 and figure 4). Overall, subjects with high RVCT/TLCCT (≥42.7% absolute value) were nearly 6-times more likely to develop COPD compared with those with low RVCT/TLCCT (≤36.6% absolute value) over the follow-up period (OR (95%CI) 5.689 (2.446–13.228); p<0.001) (table 4). Furthermore, ROC analyses showed that inclusion of RVCT/TLCCT in the models improved the area under the curve (AUC) beyond the contribution of other covariates (including age and sex) (supplementary table S3).
Comparison of change in airflow indices on follow-up spirometry across CT-measured RV/TLC (RVCT/TLCCT) strata. Graphs represent means and 95% confidence intervals for change in airflow indices across RVCT/TLCCT strata relative to the reference group (subjects in the lowest tertile of RVCT/TLCCT) from mixed-effect linear regression modelling with adjustment for age, sex, height, weight, smoking status (former versus current), baseline lung function and time to follow-up spirometry. Subjects were stratified into tertiles of RVCT/TLCCT represented by green, blue and magenta for low, intermediate and high RVCT/TLCCT tertiles, respectively. Ref: reference value; FEV1: forced expiratory volume in 1 s; FVC: forced vital capacity; FEF25–75%: forced expiratory flow at 25–75% of FVC; FEF75%: forced expiratory flow at 75% of FVC.
The RVCT/TLCCT and follow-up time interaction analysis showed the rate of decline in FEV1/FVC to be greater in those with high RVCT/TLCCT compared with those with low (or intermediate) RVCT/TLCCT (a differential 0.66% (95%CI 0.06%–1.27%) increase in the rate of decline per year; p=0.015) (figure 3b and supplementary table S4).
The changes in FEV1 and FVC (or forced expiratory flow at 25–75% of FVC (FEF25–75%) and forced expiratory flow at 75% of FVC (FEF75%)) were not statistically significantly different between the RVCT/TLCCT strata during the follow-up period. However, the decline in FEV1/FVC was, at least in part, due to a relative increase in FVC among those with high RVCT/TLCCT compared to those with low RVCT/TLCCT (tables 2 and 3 and figure 4).
Smoking status (current versus former smoker) or burden (pack–years of smoking) did not significantly contribute to any of the observed associations, nor did the interaction term between smoking status and RVCT/TLCCT (supplementary table S5).
Association of other CT measures of air trapping, emphysema and airway disease with progression to spirometric COPD
Similar to RVCT/TLCCT, PRMfSAD and Exp−856 showed wide but less varied distributions across FEV1/FVC and FEV1 (supplementary figures S3 and S4). Among the 496 subjects with high VCCT and SVC concordance and follow-up data, FEV1/FVC declined in PRMfSAD- and Exp−856-dependent manners such that those in high and intermediate tertiles for either PRMfSAD or Exp−856 had greater decline compared to those in the low tertile (p-values of 0.038 and 0.035 for PRMfSAD and Exp−856, respectively, for FEV1/FVC decline) (supplementary tables S6, S7 and S8). Accordingly, those in the groups with higher PRMfSAD or Exp−856 were more likely to progress to develop spirometric COPD (supplementary table S9). Similar to what was observed with lung volumes, smoking status (current versus former smoker) or burden (pack–years of smoking) did not significantly contribute to any of the observed associations, nor did the interaction terms between smoking status and PRMfSAD or Exp−856 (supplementary table S10).
Other CT indices of emphysema and airway disease examined, including PRMEMPH, Insp−950 and Pi10, were not associated with progression to spirometric COPD (supplementary table S11).
To better understand the importance of RVCT/TLCCT compared with other CT parameters, we included all measured CT indices of air trapping, emphysema and airway wall thickness in the same model along with RVCT/TLCCT. In this all-inclusive and fully adjusted model, RVCT/TLCCT (included as either a continuous or a categorical variable) was the only significant predictor for FEV1/FVC decline and COPD development in smokers with preserved spirometry (supplementary table S12).
Association of lung volumes with exercise capacity, symptoms, and respiratory exacerbations
Among the 496 subjects with high VCCT and SVC concordance and follow-up data, the subjects with higher RVCT/TLCCT (intermediate and high tertiles) walked a shorter distance on their subsequent 6-MWD testing compared with those with low RVCT/TLCCT (p=0.041) (tables 2 and 3). For example, subjects with high and intermediate RVCT/TLCCT had a differential 6-MWD decline of 15 m and 19 m, respectively, compared with those with low RVCT/TLCCT, reflecting an absolute decline in 6-MWD distance of 22 m to 30 m in those with higher RVCT/TLCCT (supplementary table S13).
Similarly, subjective assessments of physical activity as measured by SF12 and SGRQ questionnaires showed a higher decline in the self-described level of subject activity in those with high and intermediate RVCT/TLCCT compared with those with low RVCT/TLCCT although this effect was not statistically significant when measured by SGRQ activity score (p=0.138) (tables 2 and 3). Other CT indices of air trapping (PRMfSAD or Exp−856) were not associated with changes in exercise capacity (supplementary tables S6 and S7).
Higher RVCT/TLCCT was statistically significantly associated with worsening respiratory symptom scores measured only by mMRC (p=0.031). Changes in symptoms measured by SGRQ and CAT, although in the hypothesised direction, were not statistically significant (tables 2 and 3). The BODE index also showed a trend towards an RVCT/TLCCT-dependent worsening with higher RVCT/TLCCT, but this did not reach statistical significance (p=0.074). Other CT indices (PRMfSAD or Exp−856) were not associated with changes in respiratory symptoms (supplementary tables S6 and S7).
The time and RVCT/TLCCT interaction analyses were not statistically significant with any of the measured exercise or symptoms score outcomes (supplementary table S4).
Among the 618 subjects with high VCCT and SVC concordance, a total of 36 subjects had severe respiratory exacerbation events (including emergency department and hospital admissions). Neither the number of those events nor the time to the first hospitalisation was significantly different between the tertiles of RVCT/TLCCT, PRMfSAD, or Exp−856 (supplementary table S14).
Sensitivity analyses
Among the 496 subjects with high VCCT and SVC concordance and follow-up data, 37.1% of patients were obese (BMI >30), 16.7% had asthma diagnoses and 12.1% had bronchodilator responsiveness. Sensitivity analyses with exclusion of patients with obesity, asthma or bronchodilator responsiveness did not change any of the observed associations, with the exception of one of the outcomes (the continuous RVCT/TLCCT model of progression to COPD measured at V3) becoming non-significant (p=0.097) when obese subjects were excluded. Sensitivity analyses with simultaneous inclusion of bronchodilator responsiveness (in terms of FEV1) and hip-to-waist ratio in the models did not affect any of the observed associations except for change in 6-MWD test, which showed similar but a non-significant association (p=0.072). Inclusion of symptom score (CAT < or ≥10) in the models along with RVCT/TLCCT or other CT indices of air trapping (PRMfSAD or Exp−856) did not affect the observed associations.
Furthermore, use of lower limit of normal criteria for diagnosis of COPD, instead of fixed ratio per GOLD criteria, produced similar associations between RVCT/TLCCT and outcomes (supplementary tables S15, S16, and S17).
Finally, sensitivity analysis with inclusion of all 814 subjects regardless of their VCCT and SVC concordance did not affect the overall associations of RV/TLC with lung function outcomes.
Discussion
In this longitudinal study of a prospective cohort of smokers at risk for COPD but with preserved spirometry, we found radiographically-measured RV to TLC ratio (RVCT/TLCCT) to vary widely across the normal ranges of spirometric indices used for COPD definition (FEV1/FVC and FEV1). We then explored this wide variance and found patients with higher RVCT/TLCCT to have greater decline in lung function at a faster rate, greater likelihood of developing spirometric COPD, and greater reduction in exercise capacity compared with those with lower RVCT/TLCCT. The relationship between higher RVCT/TLCCT and worse respiratory symptoms as measured by respiratory questionnaires reached statistical significance only across one of the three survey tools used. These findings were robust, as adjustment of analyses for several possible confounders, including smoking status (current versus former smoker), smoking burden (pack–years of smoking), obesity (including hip-to-waist ratio), concomitant asthma, respiratory symptoms score (score of < versus ≥10 on CAT questionnaire) or bronchodilator responsiveness, did not change the observed associations. Furthermore, CT indices of air trapping including PRMfSAD and Exp−856 also showed similar patterns of association with the FEV1/FVC decline and progression to COPD. Remarkably, when all CT parameters of air trapping, emphysema, and airway disease were analysed together in the same model, the RVCT/TLCCT was the only significant predictor for lung function decline and progression to spirometric COPD in smokers with preserved spirometry.
In a previous retrospective study of electronic health records from the Veterans Health Administration [9], we found plethysmographically-measured RV/TLC, as well as other lung volume indices that represent air trapping (such as the ratio of functional residual capacity (FRC) to TLC), to predict morbidity and progression to COPD in smokers with preserved spirometry. The current study increases our confidence in those conclusions by providing prospective validation of the findings of that study and expands our understanding of multi-dimensionality of susceptibility to develop COPD. Overall, these findings indicate the predictive usefulness of lung volume measurements, regardless of whether determined radiographically or physiologically, and argue for use of air trapping parameters for prognostication in those at risk for COPD.
Given the baseline differences in spirometric indices between those with higher and lower RVCT/TLCCT in this cohort, a possible explanation for the faster rate of lung function decline may simply be that those with higher RVCT/TLCCT had longer duration or higher amount of smoking, causing them to have more prolonged or more severe lung damage. However, the association of RVCT/TLCCT with progression to COPD was unaffected by adjustment for baseline lung function, smoking status (being former versus current smoker) or the amount of smoking (pack–years) in these unobstructed subjects with a minimum of 20 pack–years of smoking. Indeed, these data suggest that once at least 20 pack–years of smoking has been achieved, even smoking cessation does not affect the subsequent lung function decline in those unobstructed smokers who have developed air trapping. Overall, these findings suggest that distinct underlying biological mechanisms may be involved in determining susceptibility of smokers to develop COPD as has been previously suggested in “the Dutch hypothesis” [21], and that lung volumes representing air trapping may provide early evidence for identifying the “susceptible” smokers.
An interesting finding in this study is the manner by which the FEV1/FVC ratio declined in those with high RVCT/TLCCT to reach the threshold to be considered COPD (i.e. FEV1/FVC <0.70). Although the differential changes in FEV1 or FVC across RVCT/TLCCT strata did not reach statistical significance, there appeared to be contributions from decline in FEV1 and increase in FVC as seen in figure 4. Our previous retrospective study of electronic health records from the Veterans Health Administration [9], showed statistically significant change in only FVC (and no change in FEV1) to contribute to the decline in FEV1/FVC seen with higher RV/TLC and air trapping. Together, these findings may implicate increase in FVC as the important mechanism responsible for progression to spirometric COPD in the early stages of the disease. It is remarkable, however, that despite the trend of increases in FVC in those with higher RVCT/TLCCT, there was not an increasing trend in the expiratory time on spirometry and, in fact, the expiratory time was shorter in those with higher baseline RVCT/TLCCT and higher FVC on follow-up spirometry. These findings may indicate that regional loss of lung elastance with subsequent expansion of chest wall and increased TLC may contribute to the higher FVC and development of spirometric COPD [9].
Our study has several limitations. First, although in some subjects, assessing lung volumes using CT imaging was challenging, the majority of subjects had accurate RV and TLC measurements as we demonstrated by the high concordance between radiographically and physiologically measured VC. The cases of lung volume measurement inaccuracy by CT may be due to the challenges with breath-hold manoeuvres during the full inspiratory and expiratory CT imaging. Remarkably, the strength of the associations of lung volumes representing air trapping with lung function decline and COPD development was so robust that inclusion of all subjects versus stringent inclusion of only those with high VCCT and SVC concordance did not affect the study findings. Second, only limited number of repeat spirometries and relatively short duration of follow-up were available from the SPIROMICS cohort. Nevertheless, it is indeed remarkable that statistically significant changes in lung function albeit small were detected within the available follow-up duration. Lastly, while we found higher RVCT/TLCCT to be associated with faster decline in lung function and exercise capacity over the follow-up period, we found statistically significant worsening of symptoms via only one of the three respiratory questionnaires used in SPIROMICS. However, other questionnaires did show a similar albeit non-significant trend towards worsening symptoms with higher RVCT/TLCCT. This may be due to the relatively short duration of follow-up available and/or the limited sensitivity of the survey tools that were used to assess the symptoms. A longer follow-up period may then provide further convincing evidence regarding symptom progression.
In conclusion, in smokers with preserved spirometry, radiographic lung volumes representing air trapping prospectively predict higher rate of spirometry decline and COPD development, and may be predictive of more rapid decline in exercise capacity and respiratory symptoms associated with COPD. Further investigation of underlying biological mechanisms involved in development of air trapping should be useful in understanding the susceptibility to develop COPD at its early stages.
Supplementary material
Supplementary Material
Please note: supplementary material is not edited by the Editorial Office, and is uploaded as it has been supplied by the author.
Supplementary material ERJ-02214-2018.SUPPLEMENT
Shareable PDF
Supplementary Material
This one-page PDF can be shared freely online.
Shareable PDF ERJ-02214-2018.Shareable
Acknowledgements
The authors thank the SPIROMICS participants and participating physicians, investigators and staff for making this research possible. More information about the study and how to access SPIROMICS data are at www.spiromics.org. We would like to acknowledge the following current and former investigators of the SPIROMICS sites and reading centres: Neil E. Alexis, Wayne H. Anderson, Mehrdad Arjomandi, Igor Barjaktarevic, R. Graham Barr, Lori A. Bateman, Surya P. Bhatt, Eugene R. Bleecker, Richard C. Boucher, Russell P. Bowler, Stephanie A. Christenson, Alejandro P. Comellas, Christopher B. Cooper, David J. Couper, Gerard J. Criner, Ronald G. Crystal, Jeffrey L. Curtis, Claire M. Doerschuk, Mark T. Dransfield, Brad Drummond, Christine M. Freeman, Craig Galban, MeiLan K. Han, Nadia N. Hansel, Annette T. Hastie, Eric A. Hoffman, Yvonne Huang, Robert J. Kaner, Richard E. Kanner, Eric C. Kleerup, Jerry A. Krishnan, Lisa M. LaVange, Stephen C. Lazarus, Fernando J. Martinez, Deborah A. Meyers, Wendy C. Moore, John D. Newell Jr, Robert Paine, III, Laura Paulin, Stephen P. Peters, Cheryl Pirozzi, Nirupama Putcha, Elizabeth C. Oelsner, Wanda K. O'Neal, Victor E. Ortega, Sanjeev Raman, Stephen I. Rennard, Donald P. Tashkin, J. Michael Wells, Robert A. Wise and Prescott G. Woodruff. The project officers from the Lung Division of the National Heart, Lung, and Blood Institute were Lisa Postow and Lisa Viviano.
Footnotes
This article has supplementary material available from erj.ersjournals.com
Author contributions: Conceived and designed the current study: M. Arjomandi. Developed study protocols: M. Arjomandi, S. Zeng, I. Barjaktarevic, R.G. Barr, E.R. Bleecker, R.P. Bowler, R.G. Buhr, G.J. Criner, A.P. Comellas, C.B. Cooper, D.J. Couper, J.L. Curtis, M.T. Dransfield, M.K. Han, N.N. Hansel, E.A. Hoffman, R.J. Kaner, R.E. Kanner, J.A. Krishnan, R. Paine III, S.P. Peters, S.I. Rennard and P.G. Woodruff. Collected data: I. Barjaktarevic, R.G. Barr, E.R. Bleecker, R.P. Bowler, R.G. Buhr, G.J. Criner, A.P. Comellas, C.B. Cooper, D.J. Couper, J J.L. Curtis, M.T. Dransfield, M.K. Han, N.N. Hansel, E.A. Hoffman, R.J. Kaner, R.E. Kanner, J.A. Krishnan, R. Paine III, S.P. Peters, S.I. Rennard and P.G. Woodruff. Analysed and interpreted data: M. Arjomandi, S. Zeng, I. Barjaktarevic, R.G. Barr, R.P. Bowler, R.G. Buhr, A.P. Comellas, C.B. Cooper, D.J. Couper, J.L. Curtis, M.T. Dransfield, M.K. Han, N.N. Hansel, E.A. Hoffman, R.J. Kaner, R.E. Kanner, J.A. Krishnan, R. Paine III, S.P. Peters, S.I. Rennard and P.G. Woodruff. Prepared and edited the manuscript: M. Arjomandi, S. Zeng, I. Barjaktarevic, R.G. Barr, R.P. Bowler, R.G. Buhr, A.P. Comellas, C.B. Cooper, D.J. Couper, J.L. Curtis, M.T. Dransfield, M.K. Han, N.N. Hansel, E.A. Hoffman, R.J. Kaner, R.E. Kanner, J.A. Krishnan, R. Paine III, S.P. Peters, S.I. Rennard and P.G. Woodruff. Obtained funding: M. Arjomandi, I. Barjaktarevic, R.G. Barr, E.R. Bleecker, R.P. Bowler, R.G. Buhr, G.J. Criner, A.P. Comellas, C.B. Cooper, D.J. Couper, J.L. Curtis, M.T. Dransfield, M.K. Han, N.N. Hansel, E.A. Hoffman, R.J. Kaner, R.E. Kanner, J.A. Krishnan, R. Paine III, S.P. Peters, S.I. Rennard and P.G. Woodruff.
Support statement: SPIROMICS was supported by contracts from the NIH/NHLBI (HHSN268200900013C, HHSN268200900014C, HHSN268200900015C, HHSN268200900016C, HHSN268200900017C, HHSN268200900018C, HHSN268200900019C, HHSN268200900020C), and supplemented by contributions made through the Foundation for the NIH and the COPD Foundation from AstraZeneca/MedImmune, Bayer, Bellerophon Therapeutics; Boehringer Ingelheim Pharmaceuticals, Inc., Chiesi Farmaceutici S.p.A., Forest Research Institute, Inc., GlaxoSmithKline, Grifols Therapeutics, Inc., Ikaria, Inc., Novartis Pharmaceuticals Corporation, Nycomed GmbH, ProterixBio, Regeneron Pharmaceuticals, Inc., Sanofi, Sunovion, Takeda Pharmaceutical Company and Theravance Biopharma. Funding for the work on this manuscript was also provided by the Flight Attendant Medical Research Institute (M. Arjomandi). Funding information for this article has been deposited with the Crossref Funder Registry.
Conflict of interest: S. Zeng reports salary support from United States Department of Veterans Affairs, during the conduct of the study.
Conflict of interest: I. Barjaktarevic reports grants from AMGEN, grants and personal fees from GE Healthcare, personal fees from Grifols, AstraZeneca, CSL Behring, Boehringer Ingelheim, Verona Pharma and Fisher and Pykel Healthcare, outside the submitted work.
Conflict of interest: R.G. Barr reports grants from NIH, Foundation for the NIH and COPD Foundation, during the conduct of the study; grants from Alpha1 Foundation, personal fees (royalties) from UpToDate, outside the submitted work.
Conflict of interest: E.R. Bleecker reports grants from SARP, AsthmaNET, SPIROMICS, Pharmacogenetics and Foundation NIH, involvement in clinical trials administered through Wake Forest School of Medicine for Amgen, AstraZeneca/MedImmune, Boehringer Ingelheim, Genentech/Roche, GlaxoSmithKline, Janssen/Johnson & Johnson, Novartis, Pfizer, Sanofi-Regeneron and Teva, personal fees for consultancy from Amgen, AstraZeneca/MedImmune, Boehringer Ingelheim, Genentech/Roche, GlaxoSmithKline, Knopp, Novartis and Sanofi/Regeneron, outside the submitted work.
Conflict of interest: R.P. Bowler reports having served on advistory boards for Boehringer-Ingelheim and Abbott Nutrition, outside the submitted work.
Conflict of interest: R.G. Buhr reports personal fees from GlaxoSmithKline, outside the submitted work.
Conflict of interest: G.J. Criner has nothing to disclose.
Conflict of interest: A.P. Comellas reports grants from NIH, during the conduct of the study; non-financial support for consultancy from VIDA Diagnostics, outside the submitted work.
Conflict of interest: C.B. Cooper has nothing to disclose.
Conflict of interest: D.J. Couper reports grants from NHLBI (NIH) and COPD Foundation, during the conduct of the study; grants from NHLBI (NIH), outside the submitted work.
Conflict of interest: J.L. Curtis reports grants from NIH/NHLBI (U01HL137880), during the conduct of the study; and grants from Department of Veterans Affairs (I01 CX000911), NIH/NIAID (R01 AI120526, R21 AI 117371), Department of Defense (W81XWH-15-1-0705, PR150432) and MedImmune, Corp. Ltd, outside the submitted work.
Conflict of interest: M.T. Dransfield reports grants from NIH, during the conduct of the study; grants from Department of Defense, NIH and the American Lung Association, personal fees for consultancy and involvement with contracted clinical trials for Boehringer Ingelheim, GlaxoSmithKline, AstraZeneca, PneumRx/BTG and Boston Scientific, involvement with contracted clinical trials for Novartis, Yungjin and Pulmonx, personal fees for consultancy from Genentech, Quark Pharmaceuticals and Mereo, outside the submitted work.
Conflict of interest: M.K. Han reports personal fees from GSK, BI and AZ, non-financial support from Novartis and Sunovion, outside the submitted work.
Conflict of interest: N.N. Hansel reports grants and personal fees for consultancy from AstraZeneca and GSK, grants from Boehringer Ingelheim, NIH and COPD Foundation, personal fees for consultancy from Mylan, outside the submitted work.
Conflict of interest: E.A. Hoffman reports grants from NIH, during the conduct of the study; and is a founder and shareholder of VIDA Diagnostics, a company commercialising lung image analysis software developed, in part, at the University of Iowa.
Conflict of interest: R.J. Kaner reports personal fees from Boehringer Ingelheim, Roche/Genentech, Medimmune/AstraZeneca, Gilead, Celgene and Janssen, outside the submitted work.
Conflict of interest: R.E. Kanner has nothing to disclose.
Conflict of interest: J.A. Krishnan has nothing to disclose.
Conflict of interest: R. Paine III reports grants from NHLBI and COPD Foundation, during the conduct of the study; grants from Department of Veterans Affairs, outside the submitted work.
Conflict of interest: S.P. Peters, reports grants from NIH, NHLBI as the PI of the Wake Forest Clinical Site for the SPIROMICS COPD Program, during the conduct of the study.
Conflict of interest: S.I. Rennard is employed by AstraZeneca, Cambridge, UK and also retains Professorship and a part-time appointment at the the University of Nebraska Medical Center, Omaha, NE, USA.
Conflict of interest: P.G. Woodruff reports personal fees for consultancy from Theravance, AstraZeneca, Regeneron, Sanofi, Genentech, Roche and Janssen, outside the submitted work.
Conflict of interest: M. Arjomandi reports salary support from United States Department of Veterans Affairs, grants and salary support from United States National Institute of Health, during the conduct of the study.
- Received November 21, 2018.
- Accepted July 17, 2019.
- Copyright ©ERS 2019
References
