Chronic obstructive pulmonary disease (COPD) and emphysema are, jointly, the fourth leading cause of death in the United States and third leading cause globally. COPD prevalence and mortality have doubled in the US in the last several decades, particularly among women and minorities, despite large reductions in smoking and in part due to population aging. The Multi-Ethnic Study of Atherosclerosis (MESA) Lung Study found that emphysema measured quantitatively on computed tomography (CT) is common among older adults in the general population, usually occurs in the absence of COPD on spirometry, is a strong correlate of cardiac function, and independently predicts all-cause mortality. However, standard quantitative measures of emphysema simplify 20-30 Mb of data per lung CT scan to one number. We therefore applied unsupervised machine learning to emphysema-like voxels in a second study and found six highly reproducible CT emphysema subtypes. In preliminary work we have found that genetic and environmental risk and prognosis varies substantially by these subtypes. For this renewal of the MESA Lung Study, we therefore propose to perform non-contrast CT, spirometry and oxygen saturation among 1,750 participants in MESA Exam 7 to ascertain CT emphysema subtypes over 11 years and to collect nasal brushings and hair follicles for gene expression from 500 to test if the diffuse emphysema subtype is progressive and independently predicts lung- related hospitalizations and mortality; the obstructive CPFE subtype is progressive and independently predicts lung-related mortality; the senile subtype is benign, and these distinct subtypes do not progress from one to another, findings which are consistent across race/ethnic groups, among men and women, and among smokers and nonsmokers. Second, we will examine if different environmental risk factors including ambient air pollution and occupational exposures are associated with distinct progression of the diffuse and obstructive CPFE subtypes. Third, we hypothesize that gene expression profiles differ between CT emphysema subtypes. Innovative aspects of this proposal include longitudinal evaluation of CT emphysema subtypes and epithelial cell gene expression in a large, highly genomically characterized multiethnic general-population sample. Confirmation of the aims would help solve the conundrum of chronic lower respiratory disease, provide precision phenotypes for ‘omic analyses, and suggest biologically based preventative and therapeutic strategies for emphysema in the elderly.

Chronic obstructive pulmonary disease (COPD) and emphysema are, jointly, the fourth leading cause of death in the United States and third leading cause globally. The Multi-Ethnic Study of Atherosclerosis (MESA) Lung Study found that emphysema measured quantitatively on computed tomography (CT) is common and morbid among older adults in the general population; however, standard measures underestimate the complexity of the disease. In this renewal, we hypothesize that machine-learned CT emphysema subtypes are distinct, have different causes and prognoses, and that distinguishing pathological emphysema from that resulting from normal aging with help the prevention and treatment of COPD and emphysema.