Abstract P073: Co-morbidities and Cardiac Associated Mortality in Smokers

Abstract

Background: Chronic Obstructive Pulmonary Disease (COPD) is a complex syndrome involving all aspects of the lungs which is strongly associated with cigarette smoking. Parenchymal destruction and remodeling are disease processes involving inflammatory pathways likely to have systemic vascular effects outside of the lungs. Indeed cardiovascular disease (CVD) is a leading cause of mortality in people affected by COPD and many co-morbid conditions are also associated with the disease. We explored CVD mortality in smokers considering aspects of COPD as well as co-morbidities in the longitudinal follow-up of the COPDGene study. Methods: The COPDGene study includes baseline and longitudinal assessment of mortality for 8,157 participants with (3,604) and without COPD (4,553) all of whom reported >10 pack-years smoking exposure. Aspects of COPD including CT phenotyping and pulmonary function were combined using Principal Components Analysis (PCA), co-morbidities were combined using Latent Class Analysis (LCA) and cause specific mortality was assessed using study center reports, SSRI searches and single clinician adjudication. Cox Proportional Hazards models accounting for the effects of competing risks were used to assess the association between PCA factors and CVD mortality and PCA factors plus LCA classes and CVD mortality. Results: PCA analysis resulted in 5 factors describing emphysema, airway disease, gas trapping, BMI and its effect on CT measurement and hyperinflation and LCA analysis identified 7 classes of co-morbidities. CVD associated mortality occurred in 128 participants and competing causes of mortality occurred in 605. The PCA factor describing airway disease predicted CVD mortality in the PCA only model (HR 1.8, 95%C.I. 1.4-2.3,p<0.0001), as well as in the LCA model (HR 1.7, 95% C.I. 1.3-2.2, p<0.0001). LCA classes associated with CVD mortality include a class describing diabetes, high BP and high Cholesterol (HR 3.5, 95% C.I. 1.8-6.8, p=0.0003) and a class describing known CVD (HR 2.9, 95% C.I. 1.3-6.7, p=0.01). Conclusions: Co-morbidities of COPD represent independent predictors of CVD associated mortality in smokers accounting for pulmonary disease and competing mortality risks. Clustering of comorbidities using LCA is an approach that may be informative in complex diseases.