Pulmonary artery enlargement and mortality risk in moderate to severe COPD: results from COPDGene

Summary:

Pulmonary artery enlargement on chest CT imaging is independently associated with all-cause mortality in moderate-severe COPD, after adjustment for other known risk factors for COPD mortality and cardiovascular disease.

To the Editor:

Chronic obstructive pulmonary disease (COPD) is one of the leading causes of mortality worldwide (1). Pulmonary hypertension (PH) has been associated with reduced survival among individuals with COPD (2) and is an independent risk factor for mortality following acute exacerbations of COPD (AECOPD) (3). Measurement of the pulmonary artery to aorta (PA:A) ratio by computed tomography (CT) and assessment of PA enlargement (PA:A>1) outperforms echocardiography in identifying PH in severe COPD (4), and PA enlargement has been independently associated with risk for total and severe AECOPD in two large prospective COPD cohorts (5). Studies of population-based (6) and non-COPD (7) cohorts have indicated an association between PA enlargement on CT and mortality. However, these studies did not include adjustment for other factors associated with COPD severity, known risk factors for COPD mortality, and risk factors for atherosclerotic cardiovascular disease (ASCVD), which is a major cause of mortality in COPD.

We analysed CT imaging and clinical data from the COPDGene study and hypothesised that PA enlargement is associated with all-cause mortality, even after adjustment for other known predictors of mortality in COPD and risk factors for ASCVD. We analysed data from 3464 COPDGene participants with GOLD spirometry grades 2-4 COPD (5) who had CT imaging, and vital status assessment approximately six years later. This cohort underwent detailed baseline assessment (8), volumetric non-contrast CT scans of the chest with measurement of PA and aorta at the level of the PA bifurcation (5), and long-term longitudinal clinical follow-up (9) as previously described. All-cause mortality was determined via longitudinal follow-up, contact with next-of-kin, clinical records, death certificates, and information from the Social Security Death Index (10).

Participants had a median follow-up period of 6.3 years (IQR = 1.7 years, range = 8.4 years). Mean age was 63±9 years, 44% were female, and 78% were white. The mean FEV1 was 50±18% predicted, and 28% used supplemental oxygen. Mean PA:A was 0.90±0.14 and 810 (23.4%) participants had PA:A ratio >1. During the follow-up period, 755 (21.8%) participants died. In Kaplan-Meier analysis (SPSS), participants with PA:A >1 had shorter mean estimated survival (6.7 (95% CI [6.50, 6.89]) vs 7.4 (95% CI [7.35, 7.53]) years, P<0.001 (log-rank), Figure).

Figure: Association between PA enlargement and all-cause mortality in COPD.

Kaplan-Meier survival curves stratified by the presence or absence of PA enlargement (P<0.001, log-rank test).

Cox proportional hazards (CPH) models were used to investigate for potential association between PA enlargement (PA:A >1) and mortality. The multivariable CPH model included covariates that adjusted for other known factors associated with mortality and disease severity in COPD, as well as ASCVD. These included demographics (age, sex, race), body mass index (BMI), use of supplemental oxygen, six-minute walk distance (6MWD), Modified Medical Research Council (mMRC) dyspnoea score, FEV1 (% predicted), history of severe AECOPD within the preceding 12 months (self-reported using questions modified from standardised questionnaires), percent emphysema on CT (Slicer software), current smoking, and self-reported histories of coronary artery disease, diabetes mellitus, hypertension, and hyperlipidaemia. In the unadjusted CPH model, PA enlargement was associated with HR for mortality of 1.75 (95% CI [1.50, 2.04], P<0.001). In the multivariable CPH model, PA enlargement was associated with a 42% increase in mortality risk (HR 1.42; 95% CI [1.17, 1.72], P<0.001). Older age, male sex, lower BMI, lower FEV1 (% predicted), shorter 6MWD, supplemental oxygen use, higher mMRC dyspnoea scores, and current smoking were also associated with increased mortality risk in the adjusted CPH model. ASCVD-specific factors were not associated with risk for mortality in this cohort.

Our results are the first to demonstrate an association between PA enlargement on CT and mortality in moderate to severe COPD, even when adjusted for factors including AECOPD, components of the BODE index, and ASCVD. Several studies have previously reported associations between CT-based PA:A ratio and mortality. Using coronary CT angiography from patients with suspected coronary artery disease, Nakanishi et al. demonstrated an association between PA enlargement (defined as PA:A >0.9 in that study) and mortality after adjustment for cardiovascular risk factors and left ventricular function. However, this study neither reported the prevalence of, nor performed adjustment for factors associated with chronic lung disease (7). ASCVD is a major cause of mortality in COPD, and several factors such as age, sex, and smoking are related to both ASCVD risk and COPD severity/mortality. However, the results of our adjusted model suggest that the increased risk is unlikely attributable to ASCVD.

Shin et al. found an association between PA:A ≥1 and mortality over a shorter follow-up interval, in a smaller cohort of COPD patients referred for lung transplant/lung volume reduction evaluation, after adjustment for age, BMI, and FEV1 (% predicted) (11). Terzikhan et al. also reported an independent association between PA:A ratio and mortality among participants with COPD in a large, predominantly white population in the Netherlands. However, only 5% of the study population had moderate-severe COPD, and 17 of 2197 total participants had PA:A ≥1 (6). In contrast to prior studies, COPDGene is a multicentre, disease-specific study that included larger numbers of African-American participants (approximately 1/3 of total enrolment) (8) as well large numbers of subjects with moderate to severe disease. Our results also indicate a high prevalence of PA enlargement in the study cohort, with PA enlargement in nearly 1/4 of participants. Our findings add to the mounting body of evidence to indicate an association between PA:A ratio and adverse outcomes in COPD including severe acute exacerbations (5), reduced exercise tolerance (12), risk for development of resting hypoxaemia (13), as well as cardiac injury and more complicated hospital course during severe acute exacerbations (14).

Our results demonstrate increased mortality risk after adjustment for supplemental oxygen use, FEV1, and percent emphysema. This suggests that COPD severity cannot completely account for the observed associations, and supports the importance of mechanisms other than chronic hypoxia and capillary dropout due to emphysema in the pathophysiology of pulmonary vascular disease in COPD. Direct pulmonary vascular injury as a result of cigarette smoke exposure could also contribute to the development of PH in early or mild disease, in the absence of emphysema or hypoxaemia (15). Genetic variants in IREB2 and GALC that were identified through a genome-wide association study of COPDGene and ECLIPSE participants with PA:A>1 (16) might also account for earlier onset of PH in certain individuals.

The strengths of our study include use of a well-defined cohort, length of follow-up, and adjustment for other known predictors of mortality. PA:A measurement represents a reproducible, practical method for evaluation of patients with COPD for pulmonary vascular disease. Our study has several limitations. First, our results should not be used to imply a direct causal relationship between PA enlargement and mortality. Since we evaluated all-cause mortality, there is insufficient evidence to reach conclusions regarding specific causes of death among individuals with concurrent COPD and PA enlargement. Measurement of disease-specific outcomes and study of PA:A ratio in common comorbid conditions such as obesity, obstructive sleep apnoea, and congestive heart failure would be valuable in elucidating the nature of the associations observed in this study. Our conclusions are also based on a single cohort with moderate to severe disease, using noncontrast, non-gated CT imaging, though others have reached similar conclusions using different study populations and CT imaging protocols (6, 7, 17).

These findings extend the known utility of PA enlargement on CT scan as an imaging-based biomarker for morbidity and mortality in COPD. The increased use of CT imaging for clinical indications (i.e. lung cancer screening) coupled with the relative ease in ascertaining PA enlargement make this a practical metric for aiding in COPD prognosis.