Structural Predictors of Lung Function Decline in Young Smokers with Normal Spirometry

Rationale Chronic obstructive pulmonary disease (COPD) due to tobacco smoking commonly presents when extensive lung damage has occurred. Objectives We hypothesized that structural change would be detected early in the natural history of COPD and would relate to loss of lung function with time. Methods We recruited 431 current smokers (median age, 39 yr; 16 pack-years smoked) and recorded symptoms using the COPD Assessment Test (CAT), spirometry, and quantitative thoracic computed tomography (QCT) scans at study entry. These scan results were compared with those from 67 never-smoking control subjects. Three hundred sixty-eight participants were followed every six months with measurement of postbronchodilator spirometry for a median of 32 months. The rate of FEV1 decline, adjusted for current smoking status, age, and sex, was related to the initial QCT appearances and symptoms, measured using the CAT. Measurements and Main Results There were no material differences in demography or subjective CT appearances between the young smokers and control subjects, but 55.7% of the former had CAT scores greater than 10, and 24.2% reported chronic bronchitis. QCT assessments of disease probability–defined functional small airway disease, ground-glass opacification, bronchovascular prominence, and ratio of small blood vessel volume to total pulmonary vessel volume were increased compared with control subjects and were all associated with a faster FEV1 decline, as was a higher CAT score. Conclusions Radiological abnormalities on CT are already established in young smokers with normal lung function and are associated with FEV1 loss independently of the impact of symptoms. Structural abnormalities are present early in the natural history of COPD and are markers of disease progression. Clinical trial registered with www.clinicaltrials.gov (NCT 03480347).

Chronic obstructive pulmonary disease (COPD) is currently characterized by the presence of respiratory symptoms and structural changes in airways and lungs leading to airflow obstruction that is persistent and often progressive (1).Most patients are identified when significant structural damage is present with airspace enlargement, airway thickening, and reduced numbers of small airways (2,3).Existing treatment can ameliorate the resulting symptoms (4) but does not reverse the underlying pathophysiological problem, making earlier disease identification a priority (5).
Our understanding of the natural history of COPD changed with the recognition of early-life loss of lung function in many cases (6).A large proportion of patients with COPD achieve their expected lung growth and experience an accelerated decline in FEV 1 thereafter, usually due to tobacco smoking (6,7).Several life-spanning birth cohorts have shown that smoking causes respiratory symptoms to emerge gradually and is associated with accelerated loss of lung function (8,9).New quantitative computed tomography (CT) imaging techniques allow a more detailed analysis of lung structure in vivo (10,11).To date, such studies have focused on those with wellestablished COPD or older individuals in whom COPD has not developed despite heavy smoking (12,13).Studying such groups is unlikely to identify the earliest stages of COPD development (5).
We hypothesized that even modest tobacco exposure in midlife would lead to structural change and symptoms in susceptible individuals and that these would be associated with an accelerated decline in lung function.To address this, we established a multicenter prospective observational study in the United Kingdom, recruiting smokers (the BEACON [British Early COPD Network] cohort) 30-45 years of age with normal findings on spirometry who underwent quantitative thoracic CT (QCT) imaging, symptom scoring, and detailed physiological measurements over a median of 32 months of follow-up.In addition, we compared our baseline data with those from a group of healthy nonsmokers with similar demographic characteristics.

Methods Study Design and Participants
The BEACON cohort was recruited from seven academic centers (see Tables E1 and E2 in the online supplement) between February 2018 and February 2020.It enrolled 30-to 45-year-old current tobacco smokers (defined as having regularly consumed tobacco within the preceding 7 d) with .10pack-year smoking histories, postbronchodilator FEV 1 of at least 80% predicted, and body mass index , 35 kg/m 2 .Any history of chronic lung, cardiovascular, diabetic, or autoimmune disease, a current cancer diagnosis, or regular cannabis use led to exclusion (see Table E3; Figure 1).All participants were offered smoking cessation support.At enrollment, demographics, symptoms, and physiological data were recorded (see the online supplement and Table E4).Six-monthly follow-up visits including lung function assessment were scheduled (see the online supplement and Table E4) but were delayed if within 6 weeks of a respiratory infection.Participants underwent quantitative inspiratory-and expiratory-phase CT scans (coached TLC and coached residual volume) within 180 days of recruitment.
Healthy never-smokers (lifetime smoking history , 100 cigarettes) aged 30-50 years without histories of chronic lung disease and with preserved spirometry (FEV 1 .80% predicted and FEV 1 :FVC ratio .0.7) were recruited by the University of Iowa between November 2017 and September 2018 (Figure 1).
All participants gave written informed consent, and the study protocols were

At a Glance Commentary
Scientific Knowledge on the Subject: Understanding the root causes of chronic obstructive pulmonary disease (COPD) and developing effective treatments are crucial for COPD research.Most patients are identified when significant structural damage is present, with airspace enlargement, airway thickening, and reduced numbers of small airways.Existing treatment can ameliorate the resulting symptoms but does not reverse the underlying pathophysiological problem, making earlier disease identification a priority.

What This Study Adds to the
Field: This large prospective cohort study showed that radiological abnormalities measured on quantitative computed tomography relate to smoking history and reflect an already established degree of subclinical small airway dysfunction.Radiological abnormality and, separately, symptom scores related to an increased decline in FEV 1 .Greater decline was seen in those who self-reported chronic bronchitis.This finding was confirmed by several sensitivity analyses.Our data have identified abnormalities early in the natural history of the disease, which can be monitored to assess physiological and pathological impact.More important, we have shown that even a relatively short period of regular tobacco smoking results in identifiable lung abnormalities before smokers typically receive diagnoses of COPD.approved by the relevant research ethics committees (see the online supplement) and registered at ClinicalTrials.gov(NCT 03480347).

QCT Scanning
The imaging protocol for both cohorts was standardized as detailed by Sieren and colleagues (see the online supplement) (14) and analyzed centrally (VIDA Diagnostics, Inc.).The disease probability measure (DPM) approach (15,16) was used to provide quantitative classification of wholelung nonemphysematous gas trapping (DPM-defined air trapping [DPM AirTrap ], indicating functional small airway disease [fSAD]), emphysema (DPM-defined emphysema [DPM Emph ]), and lung volume meeting healthy criteria for each participant (see Table E5).DPM quantifies the voxel-tovoxel difference in Hounsfield units between matched inspiratory and expiratory images to estimate the probability of air trapping such that the probability is inversely proportional to the relative differences in Hounsfield units.When quantified using DPM analysis, air trapping reflects what has been referred to previously as fSAD (17).The upper limit of normal for emphysema has been previously defined (18).
On the inspiratory scans, we also estimated an index of airway wall thickness (Pi10) as the square root of the wall area of an airway with a 10-mm inner perimeter.Larger Pi10 values indicate thicker airway walls relative to luminal size (19).Interstitial components of the lung were characterized using three-dimensional texture analysis (adaptive multiple-feature method [AMFM]) (20)(21)(22).The AMFM is a machine-learning approach that classifies regional parenchymal textures such as ground-glass opacities (GGOs) and bronchovascular bundles (see the online supplement).Briefly, the AMFM-based texture analysis quantifies GGOs, or other lung tissue characteristics, as a percentage of total lung volume at TLC by using grayscale patterns within CT images.
Intrapulmonary vasculature was investigated by assessing the ratio of small vessel volume (SVV 0.75 ; vessels with radii ,0.75 mm) to total pulmonary vessel volume (TPVV).Total lung and TPVV segmentations were generated using convolutional neural networks (23).The size of each vessel was calculated by first identifying the vessel center line and then calculating the distance from each center line voxel to the nearest boundary voxel.

Other Variables
Spirometry and body plethysmography lung volume measurements were obtained within the BEACON cohort at each visit according to the American Thoracic Society and European Respiratory Society guidelines (24,25), and predicted values were estimated using the Global Lung Function Initiative reference equations (26).Respiratory symptoms were assessed using the COPD Assessment Test (CAT) questionnaire.Chronic bronchitis was self-reported as a productive cough for at least three months in two consecutive years.Smoking burden was measured in pack-years.

Statistical Analysis
Categorical data are summarized as frequencies and percentages.For continuous data, normally distributed variables are summarized as mean (SD) and nevernormally distributed variables as median (interquartile range [IQR]).We compared the baseline demographics and spirometric values of BEACON participants and neversmoking control subjects using two-sample t tests for continuous variables and chisquare statistics for categorical variables.The imaging data were highly skewed, so we tested for between-group differences using Mann-Whitney tests.
For the longitudinal analysis of the follow-up measurements made every six months during follow-up (median 32 mo) in the BEACON cohort, we used randomeffects linear regression to investigate the association between radiological abnormalities and change in FEV 1 over time.To examine the correlations between CT abnormalities and lung function, we chose lung function parameters that were expressed as percentage predicted or ratios to minimize issues with sex and body size (see Figure E4).We adjusted for age and sex at initial recruitment and smoking status at each follow-up visit (model 1).We also investigated the association between baseline symptoms (measured using the CAT score) and change in FEV 1 over follow-up, using random-effects linear regression (model 2).As chronic bronchitis is known to be associated with accelerated FEV 1 decline, the presence of self-reported chronic bronchitis (see the online supplement) was explored in two sensitivity analyses.The first sensitivity analysis used random-effects linear regression to assess the interaction with FEV 1 change when adjusted for age, sex, and smoking status.The second sensitivity analysis added chronic bronchitis as a covariate to statistical models 1 and 2.
Follow-up lung function measurements were not available in the never-smoking control group.P values were two sided, with Bonferroni adjustment for multiple comparisons (with adjustment for the 12 comparisons across the two groups), and values less than 0.05 were considered to indicate statistical significance for all analyses.All statistical analyses were conducted using Stata version 17 (StataCorp) or Prism version 9.2 (GraphPad Software).

Study Sample
Of 550 individuals in the BEACON cohort, 431 (78%) (all current smokers, 41% women), with a median age of 38 (IQR, 35-42) years and a mean FEV 1 of 101% predicted (SD, 11.6%), underwent CT scanning (Figure 1).The CT scans of five participants were technically unacceptable and excluded from analysis.One hundred fourteen cohort members either declined consent for scanning or withdrew.Of the remaining 431 participants (median smoking history, 16 2).This AMFM honeycomb texture can alternatively be related to an enhanced "CT density gradient" at the borders of vascular bundles, possibly associated with inflammatory processes (27).Figure 3 summarizes visual differences between these two groups.
QCT abnormalities increased with age and tobacco consumption independently of age, but there was substantial collinearity between tobacco pack-years and age at recruitment.(see Figures E1-E3).There was no significant gender difference in the occurrence of QCT abnormality.

Associations among CT Measures, Lung Function, and Symptoms in Smokers
Although some of the associations between FEV 1 and FEV 1 :FVC ratio achieved nominal statistical significance, generally the relationships were weak.There were significant associations between DPM AirTrap, Pi10, and ground-glass opacification and the ratio of residual volume to TLC (see Figure E4).The median CAT score in the BEACON cohort participants was 10.5 (IQR, 6-15).There were no significant differences in QCT measurements between those reporting high (.10) and low (<10) CAT scores (see Figure E5).

Rate of FEV 1 Decline and Baseline Radiological Abnormalities and Symptoms in Younger Adult Smokers
The annualized FEV 1 decline rate in the BEACON cohort was 236.4 ml/yr (95% confidence interval [CI], 244.4 to 228.4 ml/yr; P , 0.001).When adjusting for age, sex, and smoking, we found that increases in six pathological QCT measurements were associated with accelerated annualized FEV 1 decline (Table 2).Increased DPM AirTrap was associated with an additional 21.5 ml/yr per 1% increase in the CT measurement (95% CI, 22.1 to 20.1 ml/yr; P , 0.001), while DPM Emph was associated with an additional 219.5 ml/yr decline per 1% increase (95% CI, 233.5 to 25.6 ml/yr; P , 0.001).For a 0.1-mm increase in the Pi10 measurement, there is an additional 29.1 ml/yr (95% CI, 233.5 to 27.1 ml/yr; P , 0.001) loss of  and FEV 1 decline, we added the FEV 1 value to our models as a covariate.This made little difference to the magnitude or direction of the relationships between CT and FEV 1 decline (see Table E7).Self-reported chronic bronchitis was associated with an additional 232.6 ml/yr (95% CI, 248.3 to 217.0 ml/yr; P , 0.001) loss of FEV 1 .When including chronic bronchitis as a covariate in the model assessing the association of radiological abnormality and FEV 1 change, DPM Emph was the only parameter for which chronic bronchitis augmented FEV 1 decline (228.6 ml/yr [95% CI, 245.3 to 212.0 ml/yr]; P = 0.006) (see Table E7).
Not all subjects enrolled and scanned then attended follow-up to provide AMFM-defined bronchovascular bundles (pink) and 0.6% ground-glass opacities (green) and a participant in the smoking cohort with 15% bronchovascular bundles and 18% ground-glass opacities.Middle panels demonstrate two members of the baseline smoking cohort.On the left is a participant with disease probability measure (DPM)-defined 25% air trapping (yellow) but no emphysema, and on the right is a participant with DPM-defined 29% air trapping (yellow) and 5.7% emphysema (red).The DPM results are shown within a midcoronal section of a topographic multiplanar reformatted (VIDA Diagnostics) (37) view in which the airways are flattened into a single plane with the parenchyma similarly warped.In the bottom panels, smoking participants at the extremes of the SVV .75/TVV distribution are displayed, with ratios of 0.085% (left) and 0.404% (right).Vessels with diameters of 0.75 mm or less are displayed in red, and larger vessels are displayed in dark purple.AMFM = adaptive multiple-feature method; SVV .75= small vessel volume (vessels with radii <0.75 mm); TVV = total vessel volume.
longitudinal lung function, but this group did significantly differ from those who remained (see Table E9).

Smoking Status Change during Follow-Up
All participants were active smokers at enrollment.Change in smoking status was included in the longitudinal modeling because 58 participants (15.8%) intermittently quit smoking at one or more follow-up visits.A total of six (1.6%) participants quit tobacco throughout the study observation period.

Discussion
Our data show for the first time the extent to which smokers younger than 45 years with normal findings on spirometry have changes in their small airways, lung parenchyma, and pulmonary vasculature.These abnormalities are unrelated to their symptomatology but increase with greater tobacco exposure.In addition, these changes predict a faster rate of lung function loss but are not the only mechanism for this process, as an increased symptom score also independently predicts functional decline.
Prior studies quantitatively analyzing QCT scan-derived metrics suggested that fSAD and emphysema can develop before measurable change in larger airways (12,13).However, those studies focused on older individuals with already established COPD (12,16).Our study provides evidence that smokers commonly begin to develop small airway dysfunction and emphysema at a much younger age than previously reported (15).Although this observational study was not designed to test causality, the relationship found between the presence of radiological abnormalities and smoking, in particular pack-years accrued, would fit with smoking's being the primary cause of these radiological abnormalities.
FEV 1 decline has remained the main marker of COPD progression, and it is central to current concepts of COPD development during adulthood (28).Our FEV 1 decline rate of 36.4 ml/yr compares with similar data (8,28), as do the effect of chronic bronchitis and the association we find between baseline CAT score and accelerated FEV 1 decline (8,29).We also describe for the first time a relationship between a number of QCT measurements and subsequent rate of FEV 1 decline.
Increased fSAD, emphysema, and bronchial wall thickening (Pi10) are associated with an increase in the decline of subsequently measured FEV 1 .The overall burden of these radiological abnormalities found in the BEACON cohort conforms to expectations compared with other studies of COPD in older patients (15).Overall, the quantity of emphysema measured in our participants was low, but we found that a 1% increase in this measurement was related to a 10-fold increase in FEV 1 loss compared with fSAD and Pi10.Relatively novel measurements of interest, such as the increase in ratio of SVV 0.75 to TPPV, GGOs, and prominence of bronchovascular bundles were associated with a subsequently accelerated rate of FEV 1 decline.The higher proportion of small vessels in the BEACON cohort is potentially explained by downstream increased resistance in inflamed lung regions (30).An increased regional pulmonary parenchymal perfusion heterogeneity has been shown to be reduced by a single dose of oral sildenafil, supporting the hypothesis that the increased resistance is associated with regional hypoxic vasoconstriction in the inflamed regions (31,32).An alternative explanation is that the greater enlargement of the SVV 0.75 relative to the TPVV is because the increased size of the small vessels brings them within the resolution of the scanner, and thus more small vessels enter into the calculation as opposed to enlargement of existing vessels.
Not all smokers showed each of these changes.The was a modest association between DPM AirTrap and DPM Emph , with a similar association between DPM Emph and pulmonary vessel volume.The parenchymal changes were strongly related to each other, but the airway wall thickness was not related to other changes, emphasizing the heterogeneity of lung damage present.The lack of a clear association between the changes and baseline physiology is unsurprising given the study entry criteria but shows that these changes are yet to significantly affect lung mechanics and patient symptomatology, thereby representing a truly early phase of the process.Tobacco burden expressed as packyears was related the extent of small airways damage as seen in older smokers (33).Age per se was not responsible for the absence of this significant abnormality in our control group.Although our subjects reported a range of CAT scores, and 25% had symptoms of chronic bronchitis, there were no differences in QCT appearance between symptomatic and asymptomatic individuals.
A study strength is the comparison between our younger smokers and nonsmokers who followed an identical scanning protocol.The prebronchodilator lung function of BEACON subjects was lower than that of control subjects, a finding noted among older "healthy" smokers (34).Although the geographical differences between these two groups are a possible limitation, it is uncertain whether this would lead to a material difference in results.Multiple lung function trajectories can lead to symptomatic COPD, and we chose to study one clearly defined closely associated with tobacco smoking and classically described by Fletcher and Peto (35).It is possible that a different pattern of radiological damage would be seen in individuals with suboptimal lung growth.Understanding the root causes of COPD and developing effective interventions are a priority for COPD research (5).Not all subjects enrolled then provided follow-up, but these participants did not significantly differ from those who remained.Additional analyses, such as the degree of airway mucus plugging (36) and total airway count (3), would shed further light on the lung structure in our cohort, while interval CT scanning would help understand damage progression and the occurrence of new abnormalities.Nevertheless, these limitations do not detract from our finding that radiological abnormalities, classically associated with established COPD, are prevalent among young adult smokers.

Conclusions
Our study shows that these radiological abnormalities relate to prior smoking history, reflect an already established degree of subclinical small airway dysfunction, and provide novel QCT data that relates to subsequent FEV 1 decline.Understanding the root causes of COPD and developing effective treatments are priorities for COPD research (5,28).Our data have identified abnormalities early in the natural history of the disease, which can be monitored to assess physiological and pathological impact.More important, we have shown that even a relatively short period of regular tobacco smoking results in identifiable lung abnormalities before smokers typically receive diagnoses of COPD.

Figure 1 .
Figure 1.Flow diagram outlining selection of the study cohort and healthy control subjects.COPD = chronic obstructive pulmonary disease; CT = computed tomography; qCT = quantitative computed tomography.

Figure 2 .
Figure 2. Comparison of quantitative computed tomography parameters by disease probability measure and three-dimensional texture analysis methods between the BEACON cohort and healthy control subjects.(A-D) Differences in DPM AirTrap (A), DPM Emph (B), DPM Healthy (C), and Pi10 (D).(E-G) Difference in prominent bronchovascular bundles (expressed as a percentage of lung field) (E), ground-glass opacity (F), and honeycombing (G), using the three-dimensional texture analysis method.(H) Difference in the ratio of SVV 0.75 to TPVV.Comparisons were made using the Mann-Whitney test.BEACON = British Early COPD Network; DPM AirTrap = disease probability measure-defined air trapping; DPM Emph = disease probability measure-defined emphysema; DPM Healthy = disease probability measure-defined healthy lung; Pi10 = square root of the wall area of a hypothetical airway with a 10-mm inner perimeter (larger values indicate thicker airway walls); SVV 0.75 = small vessel volume (vessels with radii <0.75 mm); TPVV = total pulmonary vessel volume.

Figure 3 .
Figure 3. Visual examples of the image-based metrics.Top panels demonstrate midcoronal sections from a nonsmoking subject (right) with 9%AMFM-defined bronchovascular bundles (pink) and 0.6% ground-glass opacities (green) and a participant in the smoking cohort with 15% bronchovascular bundles and 18% ground-glass opacities.Middle panels demonstrate two members of the baseline smoking cohort.On the left is a participant with disease probability measure (DPM)-defined 25% air trapping (yellow) but no emphysema, and on the right is a participant with DPM-defined 29% air trapping (yellow) and 5.7% emphysema (red).The DPM results are shown within a midcoronal section of a topographic multiplanar reformatted (VIDA Diagnostics)(37) view in which the airways are flattened into a single plane with the parenchyma similarly warped.In the bottom panels, smoking participants at the extremes of the SVV .75/TVV distribution are displayed, with ratios of 0.085% (left) and 0.404% (right).Vessels with diameters of 0.75 mm or less are displayed in red, and larger vessels are displayed in dark purple.AMFM = adaptive multiple-feature method; SVV .75= small vessel volume (vessels with radii <0.75 mm); TVV = total vessel volume.

Table 1 .
Demographics and Clinical Characteristics of Participants Definition of abbreviations: BEACON = British Early COPD Network; BMI = body mass index; CAP = community-acquired pneumonia; CAT = COPD Assessment Tool; IQR = interquartile range; LAA = low-attenuation area; LRTI = lower respiratory tract infection; mMRC = Modified Medical Research Council Dyspnea Scale; RV = residual volume.*Chi-squaretest.†Chronic bronchitis was defined as a productive cough for at least three months in two consecutive years.‡Fisher exact test.FEV 1 .An increase in the of small blood vessels (as assessed by the ratio of SVV 0.75 to TPVV), was significantly associated with faster FEV 1 decline (an additional 21.1 ml/yr per 0.01 increment in the ratio [95% CI, 21.3 to 20.8 ml/yr]; P , 0.001).Increased ground-glass opacification (23.4 ml/yr per 1% increase [95% CI, 25.5 to 21.3 ml/yr]; P , 0.001) and percentage of prominent bronchovascular bundles (22.6 ml/yr per 1% increase [95% CI, 23.2 to 22.0 ml/yr]; P , 0.001) also related to accelerated FEV 1 decline.To explore the effect of baseline lung function on the relationship between QCT measurements

Table 2 .
Examining the Association between Computed Tomography Measurements and Symptom Burden on Baseline FEV 1 and FEV 1 Longitudinal Change in the Study CohortDefinition of abbreviations: CAT = COPD Assessment Test; CI = confidence interval; CT = computed tomography; DPM AirTrap = disease probability measure-defined air trapping; DPM Emph = disease probability measure-defined emphysema; NS = nonsignificant; Pi10 = measurement of bronchial wall thickness; SVV 0.75 = small vessel volume (vessels with radii <0.75 mm); TPVV = total pulmonary vessel volume.Separate generalized least squares random-effects models are presented for each CT parameter.Models were adjusted for FEV 1 decline interaction with time, smoking status (e.g., quit), gender, and age.All P values are adjusted for multiple comparisons.