Distinguishing smoking related lung disease phenotypes via imaging and molecular features

Using unsupervised machine learning applied to computed tomography-based imaging characteristics we have found three distinct phenotypes of lung disease in two large cohorts of ever-smokers and have identified their molecular correlates. Overall design: Using K-means clustering, we clustered participants from the COPDGene study (n=5273) based on computed tomography (CT) imaging characteristics and then evaluated their clinical phenotypes. These clusters were replicated in the Detection of Early Lung Cancer Among Military Personnel (DECAMP) cohort (n=360), and were further characterized using bronchial epithelial gene expression, including differential gene expression analysis, over-representation analysis, gene set enrichment analysis and gene set variation analysis.