Pulmonary Arterial Pruning and Longitudinal Change in Percent Emphysema and Lung Function

Abstract

Background

Pulmonary endothelial damage has been shown to precede the development of emphysema in animals, and vascular changes in humans have been observed in COPD and emphysema.

Research Question

Is intraparenchymal vascular pruning associated with longitudinal progression of emphysema on CT imaging or decline in lung function over 5 years?

Study Design and Methods

The Genetic Epidemiology of COPD Study enrolled ever smokers with and without COPD from 2008 through 2011. The percentage of emphysema-like lung, or “percent emphysema,” was assessed at baseline and after 5 years on noncontrast CT imaging as the percentage of lung voxels < –950 Hounsfield units. An automated CT imaging-based tool assessed and classified intrapulmonary arteries and veins. Spirometry measures are postbronchodilator. Pulmonary arterial pruning was defined as a lower ratio of small artery volume (< 5 mm² cross-sectional area) to total lung artery volume. Mixed linear models included demographics, anthropomorphics, smoking, and COPD, with emphysema models also adjusting for CT imaging scanner and lung function models adjusting for clinical center and baseline percent emphysema.

Results

At baseline, the 4,227 participants were 60 ± 9 years of age, 50% were women, 28% were Black, 47% were current smokers, and 41% had COPD. Median percent emphysema was 2.1 (interquartile range, 0.6-6.3) and progressed 0.24 percentage points/y (95% CI, 0.22-0.26 percentage points/y) over 5.6 years. Mean FEV₁ to FVC ratio was 68.5 ± 14.2% and declined 0.26%/y (95% CI, –0.30 to –0.23%/y). Greater pulmonary arterial pruning was associated with more rapid progression of percent emphysema (0.11 percentage points/y per 1-SD increase in arterial pruning; 95% CI, 0.09-0.16 percentage points/y), including after adjusting for baseline percent emphysema and FEV₁. Arterial pruning also was associated with a faster decline in FEV₁ to FVC ratio (–0.04%/y per 1-SD increase in arterial pruning; 95% CI, –0.008 to –0.001%/y).

Interpretation

Pulmonary arterial pruning was associated with faster progression of percent emphysema and more rapid decline in FEV₁ to FVC ratio over 5 years in ever smokers, suggesting that pulmonary vascular differences may be relevant in disease progression.

Trial Registry

ClinicalTrials.gov; No.: NCT00608764; URL: www.clinicaltrials.gov

Graphical abstract

COPD is a globally important disease and a leading cause of death.^1,2 Significant heterogeneity exists in the characteristics of COPD, including emphysema,³ which is defined on pathologic examination as permanent airspace dilation and also can be measured on CT scan as the percentage of emphysema-like lung, or “percent emphysema.”⁴ Percent emphysema has been associated with greater lung function decline, hospitalizations, and mortality in those with COPD.^{5, 6, 7}

Pulmonary vascular endothelial damage has been implicated as an early step in the pathogenesis of COPD and emphysema.^{8, 9, 10} The pulmonary vasculature regulates perfusion to optimize gas exchange and transports essential cellular mediators to the lung parenchyma (eg, neutrophils, cytokines, and lung progenitor cells).¹¹ In animal models, vascular endothelial growth factor receptor blockade causes endothelial apoptosis, emphysema, and pruning of pulmonary arteries.¹² In humans, a loss of pulmonary capillaries is evident on pathologic sections of emphysematous lung,^13,14 and a lower pulmonary microvascular perfusion on contrast-enhanced MRI was associated with greater percent emphysema in the Multi-Ethnic Study of Atherosclerosis COPD Study.¹⁵ In the Genetic Epidemiology of COPD (COPDGene Study), a higher degree of pulmonary vascular pruning (a lower ratio of volume of vessels < 5 mm² in cross-sectional area to total lung vessel volume) on noncontrast CT scan was found in those with more advanced COPD and with more emphysema.¹⁶ However, it is unknown whether differences in the parenchymal pulmonary vasculature contribute to the progression of emphysema and COPD.

We recently developed techniques to classify pulmonary arteries and veins on noncontrast CT imaging.¹⁷ Therefore, we sought to test the hypothesis that pulmonary arterial pruning, measured as a decrease in the ratio of small artery volume to total pulmonary arterial volume on noncontrast CT imaging, would be associated with a faster progression of percent emphysema and decline in lung function measured over 5 years in the COPDGene Study.

Methods

Study Design and Participants

The COPDGene Study is a longitudinal observational study designed to evaluate genetic contributors to COPD (ClinicalTrials.gov Identifier: NCT000608764). Current and former smokers with at least a 10-pack-year history with and without COPD were enrolled at 21 clinical centers in the United States from 2008 through 2011.¹⁸ Participants self-identified as non-Hispanic White or Black and were between 45 and 80 years of age at enrollment. The major exclusion criteria were significant lung disease other than COPD or asthma, major lung surgery (lobar resection, lung volume reduction surgery, or transplantation), and an exacerbation in the prior month treated with antibiotics or steroids. Those enrolled at baseline were invited to return for the 5-year follow-up visit (from 2012 through 2016), and the 10-year follow-up visit is ongoing. All participants provided written informed consent, and institutional review board approval was obtained at all sites. The full protocol is available at www.copdgene.org.

CT Imaging Measurements

Noncontrast CT scan of the chest was performed at full inspiration (total lung capacity) at baseline and the 5-year follow-up visit using the COPDGene Study protocol, with a smooth reconstruction kernel (Bf31 for Siemens scans, Bone for GE scans) and 0.45- to 0.9-mm slice thickness, depending on the scanner manufacturer.¹⁸ Quantitative assessment of lung parenchyma (CT scan densitometry) was performed at a single center blinded to other participant information using automated software that includes lobar segmentation (Thirona). The percent emphysema was calculated as the percentage of lung volume with attenuation < –950 Hounsfield units (percent emphysema_–950). Percent emphysema_–950 also was calculated for each lobe. Secondary analyses used lung density at the lower 15th percentile (PD15) adjusted for percent predicted total lung volume (ie, sponge model to account for different degree of inhalation on CT scans),¹⁹ based on healthy participants in a population-based study undergoing CT scan.²⁰ Per convention, PD15 values are reported as the Hounsfield units plus 1,000 to reflect lung density in grams per liter.

Assessment of the pulmonary vasculature was performed using the Chest Imaging Platform (www.chestimagingplatform.org) after lung and lobe segmentation,²¹ with 3-dimensional vascular reconstruction using a scale-space particle method.^22,23 Arterial and venous vessels then were separated by a deep learning technique (Fig 1), with mean accuracy of 93.6% when compared with manual artery-vein segmentation.¹⁷ The total arterial volume of interparenchymal vessels (TBVa) and total venous volume of interparenchymal vessels in the lung were calculated for each participant, as were volume of pulmonary arteries less than 5 mm² in cross-sectional area (BV5a) and the volume of pulmonary veins less than 5 mm² in cross-sectional area. BV5a was correlated with histologic assessment of vessel volume (r = 0.50, P = .04)²⁴ and overall BV5 was found to correlate with pulmonary perfusion by scintigraphy, including on a regional basis within the lung (r = 0.87, P < .001).²⁵ The main exposure of interest is a decrease in the relative volume of small pulmonary vessels, which we term “pruning,” with pulmonary arterial pruning being a decrease in BV5a to TBVa ratio and pulmonary venous pruning being a decrease in volume of pulmonary veins less than 5 mm² in cross-sectional area per total venous volume of interparenchymal vessels.

Figure 1.

A, B, Pulmonary arteries and veins defined by a deep learning technique in selected participants with Global Initiative for Chronic Obstructive Lung Disease 2 COPD without pruning of small arteries (volume of pulmonary arteries less than 5 mm² in cross-sectional area [BV5a] per total arterial volume of interparenchymal vessels [TBVa] ratio > 50th percentile) (A) and pruning of small arteries (BV5a per TBVa < 50th percentile) (B). Vessels are a 3-dimensional reconstruction with 2-dimensional CT image for anatomic landmarks. Blue = artery; green overlay = emphysematous lung areas; red = vein.

Pulmonary artery and aorta diameters were measured at the bifurcation of the pulmonary artery on baseline inspiratory CT scans by two reviewers using the OsiriX MD DICOM Viewer version 11.0 (Pixmeo SARL), as described previously.^{26, 27, 28}

Lung Function and Covariates

Participant race, sex, educational attainment, current smoking status, smoking history, and diagnosis of diabetes were self-reported and categorized as in Table 1. Height and weight were measured according to standard techniques. Spirometry was performed using the EasyOne Spirometer (ndd) following a standardized protocol and quality review; postbronchodilator values are used in all analyses.^18,29 COPD classification followed the Global Initiative for Chronic Obstructive Lung Disease categories based on FEV₁ percent predicted for those with COPD with a postbronchodilator FEV₁ to FVC ratio of < 0.7.³⁰

Table 1.

Baseline Characteristics by Quintile of Pulmonary Arterial Pruning^a

Characteristic	Quintile 1 (n = 845)	Quintile 2 (n = 846)	Quintile 3 (n = 845)	Quintile 4 (n = 846)	Quintile 5 (n = 845)
	Less Pruning → More Pruning
Age, y	59.8 ± 7.8	60.1 ± 8.4	60.0 ± 8.7	59.8 ± 9.1	59.3 ± 9.4
Male sex	29.1	45.7	53.3	59.1	62.4
Race	…	…	…	…	…
Non-Hispanic White	87.5	77.9	74.6	66.8	54.7
Black	12.5	22.1	25.4	33.2	45.3
Height, cm	166.9 ± 8.5	169.2 ± 9.4	171.5 ± 9.5	170.8 ± 10.0	171.8 ± 9.5
BMI, kg/m2	26.5 ± 4.6	28.4 ± 5.0	29.5 ± 6.1	29.6 ± 6.2	31.7 ± 7.0
Education	…	…	…	…	…
High school or less	22.8	31.2	32.0	36.4	45.5
Some college	29.0	26.1	29.6	28.3	26.3
College or more	48.2	42.7	38.4	35.3	28.2
Smoking status	…	…	…	…	…
Former smoker	54.6	56.4	52.5	52.4	47.3
Current smoker	45.4	43.6	47.5	47.6	52.7
Pack-years	38.7 ± 20.2	41.3 ± 22.8	42.9 ± 23.5	45.6 ± 26.1	44.9 ± 24.9
Lung function	…	…	…	…	…
GOLD stage	…	…	…	…	…
1, FEV1 > 80%	15.3	10.6	8.1	6.2	4.3
2, 50% < FEV1 < 80%	14.2	18.9	20.8	22.7	22.8
3-4, FEV1 < 50%	1.9	3.9	11.8	17.2	25.0
Normal (FEV1 to FVC ratio ≥ 0.7, FVC ≥ LLN)	62.7	57.9	49.6	41.5	28.5
Abnormal, nonobstructive (FVC < LLN, FEV1 to FVC ratio ≥ 0.7)	5.9	8.7	9.7	12.4	19.4
FEV1, L	2.6 ± 0.7	2.5 ± 0.8	2.4 ± 0.9	2.2 ± 0.9	2.0 ± 0.8
FEV1, percent predicted	91.7 ± 17.2	86.7 ± 18.6	80.7 ± 21.7	75.0 ± 23.5	67.6 ± 23.0
FEV1 to FVC ratio	0.72 ± 0.09	0.72 ± 0.11	0.69 ± 0.14	0.66 ± 0.16	0.65 ± 0.16
Percent emphysema–950	2.0 (0.7-5.3)	2.2 (0.7-5.5)	2.1 (0.6-6.2)	2.2 (0.6-7.4)	1.8 (0.4-8.3)
Lung density at the lower 15th percentile adjusted for lung volume, g/L	88.4 ± 20.8	87.5 ± 21.0	86.9 ± 24.9	85.2 ± 28.0	87.0 ± 32.2
Total lung volume on CT scan, L	5.60 ± 1.20	5.57 ± 1.30	5.63 ± 1.41	5.54 ± 1.42	5.19 ± 1.45
Small artery volume (BV5a), mL	110.7 ± 25.5	107.0 ± 25.1	102.3 ± 24.5	95.5 ± 23.1	82.0 ± 21.4
Small venous volume (BV5v), mL	66.8 ± 13.8	66.4 ± 12.9	64.7 ± 12.9	61.2 ± 13.2	54.4 ± 14.5
Total lung arterial volume (TBVa), mL	158.9 ± 37.1	167.7 ± 39.3	171.1 ± 41.1	173.3 ± 41.7	177.2 ± 41.8
Total venous volume of interparenchymal vessels, mL	106.4 ± 21.2	108.5 ± 21.7	108.2 ± 22.4	106.0 ± 23.3	102.1 ± 25.4
BV5a per TBVa (lower values reflect arterial pruning)	0.70 ± 0.03	0.64 ± 0.01	0.60 ± 0.01	0.55 ± 0.02	0.46 ± 0.05
BV5v per TBVv (lower values reflect venous pruning)	0.63 ± 0.04	0.61 ± 0.04	0.60 ± 0.05	0.58 ± 0.05	0.53 ± 0.07
Pulmonary artery diameter to aorta diameter ratio	0.79 ± 0.11	0.82 ± 0.11	0.83 ± 0.12	0.85 ± 0.12	0.88 ± 0.12

Data are presented as percentage, mean SD, or median (interquartile range). BV5a = volume of pulmonary arteries less than 5 mm² in cross-sectional area; BV5v = volume of pulmonary veins less than 5 mm² in cross-sectional area; GOLD = Global Initiative for Chronic Obstructive Lung Disease; LLN = lower limit of normal; percent emphysema_–950 = percentage of lung volume with attenuation < –950 Hounsfield units; TBVa = total arterial volume of interparenchymal vessels; TBVv = total venous volume of interparenchymal vessels

^aHighest quintile has the most pruning, that is, the lowest BV5a per TBVa.

Statistical Analysis

Participant characteristics are shown stratified by quintile of vascular pruning for descriptive purposes (highest quintile has the greatest pruning, ie, lowest small arterial to total lung vessel volume [BV5a/TBVa]). Analyses used a mixed linear regression model with a random intercept to assess the interaction between time and baseline vascular variables (all treated continuously). For emphysema, results are shown unadjusted, adjusted for CT imaging scanner, and the full model is adjusted for baseline age, sex, race, education, time-varying height, weight, and smoking status. In lung function analyses, the base model adjusted for baseline age, sex, race, education, and time-varying height, weight, smoking status, and pack-years of smoking, and a full model also adjusted for percent emphysema_–950 and clinical center. For analyses of percent change in FEV₁, the outcome was log-transformed. All analyses were adjusted for COPD status given the case-control study design. Nonlinearity was assessed using quintiles of BV5 per total blood volume of intraparenchymal vessels in the linear models.

Additive interactions were assessed between arterial pruning and age, sex, race, BMI, smoking status, COPD, baseline emphysema, CT imaging scanner manufacturer, and pulmonary artery to aorta ratio for change in percent emphysema_–950 (all categorized as shown in Fig 2). For lung function analyses, interactions were assessed for age, sex, race, smoking status, and COPD category, as well as baseline FEV₁ for the FEV₁ analyses. Further analyses for percent emphysema used lobar measures of vascular pruning and emphysema, and sensitivity analyses were adjusted for potential confounders: baseline percent emphysema_–950, percent predicted FEV₁, time-varying pack years, cigarettes smoked within 24 h, self-reported diabetes, oxygen saturation, and pulmonary artery to aorta ratio. Additional analyses excluded those scanned on a different CT imaging scanner at the two visits (45%), with more than 20% change in CT scan lung volume between scans (13%), with a CT scan lung volume of < 80% or > 120% predicted (23%), and where smoking status changed during follow-up (14%). For percent emphysema, alternate measures of vascular pruning also were assessed: the ratio of small vessel volume to lung volume on CT scan and small vessel volume (BV5a or volume of pulmonary veins less than 5 mm² in cross-sectional area) and total lung vessel (arterial or venous) volume as separate terms in the model. We also evaluated lobe-specific measures and used PD15 corrected for CT scan lung volume as an alternate lung density outcome.¹⁹ For lung function, sensitivity analyses excluded outliers in the change in FEV₁: those with the greatest gain (> 90 mL/y, n = 50) and greatest loss (> 180 mL/y, n = 45), and adjusting for baseline lung function parameters. Those missing covariates for the main analyses were not included. All analyses were performed in SAS version 9.4 software (SAS Institute) and R software (R Foundation for Statistical Computing). P values were two-sided, and a value of less than .05 was considered significant.

Figure 2.

Stratified analyses of the mean difference in yearly progression of percentage of lung volume with attenuation < –950 Hounsfield units (percent emphysema_–950) for a 1-SD increase in arterial pruning (decrease in volume of pulmonary arteries less than 5 mm² in cross-sectional area per total arterial volume of interparenchymal vessels). Model adjusts for baseline age, sex, race, education, and COPD status and time-varying height, weight, and smoking status, except when stratified by the variables. Three-way interaction P values for age, race, BMI category, smoking status, and spirometric category and baseline percent emphysema were < .001; P = .015 for sex; and P = .145 pulmonary artery diameter to aorta diameter ratio (P = 0.064 for CT imaging scanner manufacturer; not shown). GOLD = Global Initiative for Chronic Obstructive Lung Disease.

Results

Of the 10,263 ever smoking participants enrolled at baseline, a total of 6,409 participants returned for the 5-year follow-up visit (62%, or 65% of those living). At baseline, 9,541 participants had percent emphysema measures and 7,970 (84%) had valid pulmonary vascular measures; of these, 4,227 had 5-year percent emphysema measures and 4,449 had baseline and 5-year lung function measures. Compared with those in the current analyses, baseline participants not included were more likely to be Black, to be men, to be current smokers with COPD, to show more percent emphysema, and to show more arterial pruning (e-Table 1).

At baseline, the 4,227 participants included in the emphysema analyses were a mean of 60 ± 9 years of age, 50% were women, 72% were non-Hispanic White, and 28% were Black; 47% were current smokers and 41% had COPD. Those with the most pulmonary arterial pruning (quintile 5 in Table 1) were more likely to be men, Black, and current smokers with greater height and weight, lower educational status, lower lung function, and greater mean percent emphysema. The median baseline percent emphysema_–950 was 2.1 (interquartile range, 0.6-6.3), and mean progression was 0.24 percentage points/y (95% CI, 0.22-0.26 percentage points/y) over an average of 5.6 years. Mean baseline FEV₁ was 2330 ± 830 mL, with an average decline of 2.0%/y (95% CI, –1.9 to –2.1 %/y) and 37.6 mL/y (95% CI, –39.1 to –36.1 mL/y). Mean FEV₁ to FVC ratio was 68.5 ± 14.2%, with an average decline of 0.26%/y (95% CI, –0.30 to –0.23 %/y).

Percent Emphysema

Pulmonary arterial pruning was associated with a faster progression of percent emphysema_–950 in the main model (0.11 percentage points/y per 1-SD increase in arterial pruning; 95% CI, 0.09-0.13 percentage points/y) (Table 2). Pulmonary venous pruning also predicted faster progression of percent emphysema_–950, although results were of lower magnitude (0.03 percentage points/y per 1-SD increase in venous pruning; 95% CI, 0.01-0.05 percentage points/y) (Table 2). Results were similar when restricted to the right upper lobe and right lower lobe (e-Table 2).

Table 2.

Predicted Yearly Change in Percent Emphysema_–950 by Quintile of Vascular Pruning^a

Variable	Quintile of Vascular Pruning Less Pruning → More Pruning					Mean Difference per Year per 1-SD Greater Pruning (95% CI)b	P Valueb
	1	2	3	4	5
Arterial vessels (BV5a/TBVa)
Unadjusted	Reference	0.10	0.23	0.37	0.46	0.19 (0.17-0.21)	< .001
Scanner adjusted	Reference	0.05	0.15	0.23	0.29	0.12 (0.10-0.14)	< .001
Main model	Reference	0.05	0.13	0.22	0.25	0.11 (0.09-0.13)	< .001
Venous vessels (BV5v/TBVv)
Unadjusted	Reference	0.02	0.06	0.12	0.23	0.08 (0.06-0.11)	< .001
Scanner adjusted	Reference	0.003	0.03	0.06	0.12	0.04 (0.02-0.06)	< .001
Main model	Reference	0.01	0.03	0.06	0.10	0.03 (0.01-0.05)	< .001

The main model adjusts for baseline age, sex, race, education, and time-varying height, weight, smoking status, and scanner terms. All models adjust for COPD status. Boldface values indicate that the quintile was significantly different from the reference group (P < .05). BV5a = volume of pulmonary arteries less than 5 mm² in cross-sectional area; BV5v = volume of pulmonary veins less than 5 mm² in cross-sectional area; percent emphysema_–950 = percentage of lung volume with attenuation < –950 Hounsfield units; TBVa = total arterial volume of interparenchymal vessels; TBVv = total venous volume of interparenchymal vessels.

^aHighest quintile has the most pruning, that is, the lowest BV5a per TBVa.

^bEffect estimate and P value are from a model with a continuous independent variable.

Additional adjustment for baseline percent emphysema_–950 increased the magnitude of the results, whereas adjustment for baseline percent predicted FEV₁ attenuated them; with inclusion of both of these measures, the results were of similar magnitude and significance as the main model (Fig 3). With adjustment for pack years, cigarettes smoked within 24 h, diabetes, oxygen saturation, and pulmonary artery to aorta ratio, the results essentially were unchanged (Fig 3). Also, little variation was found when limiting the analysis to those with a repeat CT scan on the same scanner, with < 20% change in lung volume between the two CT scans, with CT scan lung volume between 80% and 120% predicted, and those without change in smoking status during follow-up (Fig 3). In stratified analyses, the association was of lower magnitude in those younger than 60 years, women, Black people, obese individuals, current smokers, those with a reduced FEV₁ and preserved FEV₁ to FVC ratio, and those with less than 5% emphysema at baseline (P < .01 for each interaction); however, in each subcategory, pulmonary arterial pruning was associated significantly with faster emphysema progression (Fig 2).

Figure 3.

Sensitivity analyses of the mean difference in yearly progression of percentage of lung volume with attenuation < –950 Hounsfield units for a 1-SD increase in arterial pruning (decrease in volume of pulmonary arteries less than 5 mm² in cross-sectional area per total arterial volume of interparenchymal vessels). Model adjusts for baseline age, sex, race, education, and COPD status and time-varying height, weight, smoking status, and scanner terms. PA/A = pulmonary artery diameter to aorta diameter ratio.

Results when using the ratio of small vessel volume to total CT scan lung volume were of lower magnitude, but remained statistically significant (e-Table 3). When adjusting for small vessel volume and total lung vessel volume separately, results were similar to those using their ratio as the primary exposure (e-Table 4). In addition, using PD15 adjusted for CT scan lung volume as the outcome instead of percent emphysema_–950 showed very similar results (e-Table 5).

Lung Function

Pulmonary arterial pruning was associated with a more rapid decline in FEV₁ to FVC ratio (–0.04%/y per 1-SD increase in arterial pruning; 95% CI, –0.08 to –0.001%/y; P = .042) (e-Table 6). After excluding outliers and adjusting for baseline FEV₁ to FVC ratio, results were of greater magnitude (e-Fig 1). No significant interactions were identified in the association between pulmonary arterial pruning and decline in FEV₁ to FVC ratio.

Pulmonary arterial pruning was associated with a more rapid percent decline in FEV₁ (0.18%/y per 1-SD increase in arterial pruning; 95% CI, –0.27 to –0.08%/y; P < .001) (e-Table 7). However, results were in the opposite direction when looking at absolute change in FEV₁, with a slower decline in FEV₁ by 2.4 mL/y per 1-SD increase in arterial pruning (95% CI, 0.7-4.1 mL/y; P = .005) (e-Table 8). After excluding outliers, the association for percent FEV₁ was of greater magnitude and absolute FEV₁ was attenuated, and after adjusting for baseline FEV₁, both results were nonsignificant (e-Fig 1). Interactions by spirometric category were present for both percentage change and absolute change in FEV₁, and the most divergent findings were in those with COPD (e-Table 9). In addition, results for absolute change in FEV₁ differed by baseline FEV₁ (P = .079 for interaction), whereas those for percentage FEV₁ did not (e-Table 9). No significant associations were found between pulmonary venous pruning and change in any measure of lung function.

Discussion

A higher degree of pulmonary arterial pruning was associated with a faster longitudinal progression of percent emphysema_–950 over 5 years in ever smokers in the COPDGene Study. These results were consistent using alternate measures of vascular pruning and PD15 instead of percent emphysema, and when limited to those with both CT scans obtained with the same scanner. Over the same period, pulmonary arterial pruning was associated with a faster decline in FEV₁ to FVC ratio. Together, these results suggest that in smokers, pulmonary arterial structure may be relevant in the progression of emphysema and COPD.

Pulmonary vascular differences have been demonstrated in COPD, with a loss of pulmonary capillaries seen in excised emphysematous lung specimens compared with control specimens.^13,14 Imaging studies also have found reduced pulmonary microvascular perfusion on MRI and vascular pruning on CT imaging in COPD and emphysema.^15,16,31,32 In 74 participants with COPD, Saruya et al³³ found a moderate correlation (r = –0.46) between an increase in percent emphysema and decrease in volume of small vessels (< 5 mm²) over 3 years on axial CT scan slices; however, this was not replicated in another study by the same group.³⁴ In the present study, we found that baseline vascular, particularly arterial, differences on CT imaging were associated with a faster longitudinal progression of percent emphysema and loss of lung function that was robust to adjustment for multiple covariates.

Pulmonary vascular pruning may be the result of vascular injury, pulmonary vasoconstriction, or hyperinflation and gas trapping with compression of the vessels. Although this study cannot determine the cause, several potential mechanisms may link vascular pruning to progression of percent emphysema or decline in lung function. First, vascular pruning may impede the augmentation of pulmonary blood flow usually seen in infection, resulting in fewer cellular mediators to fight infection and aid in resolving lung injury.^11,35 Second, pulmonary vascular pruning may reflect endothelial dysfunction or activation, leading to increased recruitment of inflammatory cells.³⁶ Third, arterial pruning may impact capillary blood flow directly, leading to a slower transit of neutrophils or activated platelets through the pulmonary vasculature, leading to more inflammation.^{37, 38, 39, 40} Further evaluation of the vasculature, and these potentially related pathways, may lead to a better understanding of disease progression.

We found interactions in almost every category for the effect of arterial pruning on the progression of percent emphysema, with greater magnitude results in those with more emphysema, older individuals, and former smokers, and lower magnitude results in women, obese individuals, and Black people. It is not unexpected to see faster emphysema progression in those with more baseline emphysema, because it may reflect disease susceptibility and may be a similar relative increase, and this also may explain the greater magnitude results seen in older participants. Stronger associations in former smokers compared with current smokers may be the result of acute changes in lung attenuation with smoking^41,42 or differences in the pulmonary vasculature, because current smokers and those with greater air pollution exposures show an increase in the volume of smaller pulmonary vessels instead of the decrease seen in emphysema and COPD.^43,44 The interactions by sex and obesity may relate to pulmonary vascular disease, which is more common in women and obese individuals,^45,46 and emphysema progression, which was found to be faster in women and slower in obese individuals.⁴⁷ Racial differences also have been described in pulmonary hypertension and in the extent of emphysema.^20,48 At baseline less arterial pruning was seen in women, and more was seen in obese individuals and Black people. However, more work is needed to understand fully how these factors relate to pulmonary arterial pruning and to modify the relationship between pruning and emphysema progression.

Results for FEV₁ differed when assessed by percentage and absolute change, and both results were nonsignificant after adjusting for baseline FEV₁, leading us to interpret an overall null result for FEV₁. The most divergent results were in those with COPD and those with a low baseline FEV₁ and in participants who have more arterial pruning and a smaller absolute change in FEV₁ compared with those with normal lung function. Although it is used less commonly, percentage change in FEV₁ has been reported to be more reliable than absolute change,⁴⁹ and our findings for FEV₁ to FVC ratio, where FEV₁ is normalized to FVC, are more consistent with results for percentage change in FEV₁.

Although this was a large prospective observational study of smokers with and without COPD, is has several limitations. First, noncontrast CT scan measures of the pulmonary vasculature do not include the microvasculature, and it is an aggregate measure of the entire lung or lobe. Further, the lower ratio (BV5a to TBVa) that we call “pruning” may reflect loss, narrowing, or decreased filling of smaller vessels, or even proximal vessel dilation. Nonetheless, BV5a has been correlated with small vessels on histologic analysis and BV5 was correlated with pulmonary perfusion by scintigraphy.^24,25 We found very similar results in analyses in which BV5a was normalized to the total lung volume instead of TBVa, and also when evaluating BV5a by itself while adjusting for TBVa. Adjusting for pulmonary artery to aorta ratio showed no significant impact on our results, suggesting that proximal artery dilation was not an explanation for our findings. Although results for emphysema and vascular pruning were similar when restricted to a specific lobe, further studies are needed to understand whether arterial pruning predicts localized progression of emphysema.

Second, the clinical implications of change in percent emphysema_–950 on CT scan is uncertain, and the magnitude of change in FEV₁ to FVC ratio and FEV₁ related to arterial pruning were small. However, the reported average difference in emphysema progression resulting from arterial pruning was 0.55 percentage points over the 5-year study, representing a 40% faster rate of emphysema progression. Because percent emphysema has been linked to increased mortality,^7,50 even this magnitude change may be important. The changes in lung function were more modest, with an approximately 15% faster decline in FEV₁ to FVC ratio.

Third, selection bias is possible because of loss to follow-up that impacted inclusion in the longitudinal study sample. Those not included in the study were more likely to have COPD, emphysema, and pulmonary arterial pruning; however, those with COPD at baseline constituted 41% of the current sample, and results were of greater magnitude among that group. Although we cannot exclude the possibility of bias, it is unlikely to have led to significant bias in our findings.

Fourth, although the longitudinal study design allows us to make some inference as to causality, pulmonary vascular pruning has been associated with more severe emphysema and airflow obstruction in cross-sectional studies,¹⁶ making baseline lung disease a possible confounder. Because we used a random intercept, we did not adjust for baseline values of the dependent variable in the main models, but did explore this in sensitivity analyses. In the emphysema analysis, results were of greater magnitude with adjustment for baseline percent emphysema_–950 and were attenuated, but still significant, with adjustment for baseline FEV₁. Further, the emphysema findings remained significant when considering only those with percent emphysema_–950 of < 5% and those without COPD. After adjustment for baseline values, both FEV₁ results were null, whereas results for decline in FEV₁ to FVC ratio were of greater magnitude.

Finally, measures of emphysema used a threshold-based technique, and differences in protocol among the different sites and scanner changes may have impacted the results. The coefficient of variation for repeat CT scans in the COPDGene Study was 16%, in part because of differences in inhalation. We showed similar results in analyses restricted to those whose CT scans were obtained with the same scanner and without a large change in lung volume. We also showed similar results for measures of lung density at the lower 15th percentile (PD15) with lung volume correction.

In conclusion, greater pulmonary arterial pruning was associated with a faster progression of percent emphysema_–950 and decline in FEV₁ to FVC ratio over 5 years among ever smokers with and without COPD in the COPDGene Study. This study adds to prior data suggesting that the pulmonary vasculature may be important in emphysema and COPD, and further study of small pulmonary arteries and vascular pruning may yield insights into disease progression.

Take-home Point.

STUDY QUESTION: Is vascular pruning at baseline associated with a faster progression of emphysema on chest CT scan or decline in lung function?

RESULTS: In the COPDGene Study, an observational study of ever smokers with and without COPD, we found that a greater degree of arterial pruning at baseline was associated with faster progression of emphysema on CT imaging and a more rapid decline in FEV₁ to FVC ratio over 5 years.

INTERPRETATION: These findings suggest that the pulmonary vasculature may be important in emphysema and COPD, and further study of small pulmonary arteries may provide further insight into disease progression.