BioProject

Idiopathic pulmonary fibrosis (IPF) is a chronic fibrosing lung disease that is difficult to diagnose and follows an unpredictable clinical course. The object of this study was to develop a predictive gene signature model of IPF from whole lung tissue. We collected whole lung samples from 11 IPF patients undergoing diagnostic surgical biopsy or transplantation. Whenever possible, samples were obtained from different lobes. Normals consisted of healthy organs donated for transplantation. We measured gene expression on microarrays. Data were analyzed by hierarchical clustering and Principal Component Analysis. By this approach, we found that gene expression was similar in the upper and lower lobes of individuals with IPF. We also found that biopsied and explanted specimens contained different patterns of gene expression; therefore, we analyzed biopsies and explants separately. Signatures were derived by fitting top genes to a Bayesian probit regression model. We developed a 153-gene signature that discriminates IPF biopsies from normal. We also developed a 70-gene signature that discriminates IPF explants from normal. Both signatures were validated on an independent cohort. The IPF Biopsy signature correctly diagnosed 76% of the validation cases (p < 0.01), while IPF Explant correctly diagnosed 78% (p < 0.001). Examination of differentially expressed genes revealed partial overlap between IPF Biopsy and IPF Explant and almost no overlap with previously reported IPF gene lists. However, several overlapping genes may provide a basis for developing therapeutic targets. Overall design: 17 samples from 11 patients with IPF (6 patients provided a pair of samples from upper and lower lobes; 5 patients contributed singleton samples); 6 control specimens were obtained from routine lung volume reduction of healthy donor lungs at the time of lung transplantation.