Gene Co-Expression Modules In Peripheral Blood Mononuclear Cells Predict Clinical Outcome Of Idiopathic Pulmonary Fibrosis

Early identification of idiopathic pulmonary fibrosis (IPF) patients with poor prognosis remains a significant challenge. We Background: hypothesized that gene expression profiles in peripheral blood mononuclear cells (PBMC) may serve as predictors for IPF progression. The testing and validation cohorts consisted of 45/16 and 22/10 IPF/healthy individuals, respectively. The average follow-up time Method: was 18 months in testing and 36 months in validation cohort. Total RNA isolated from PBMC were hybridized on Affymetrix Exon 1.0ST array. Gene expression levels were summarized from exon intensity using dChip software. The R package WGCNA was used to parse genes in the array into 7 co-expressed modules designated with different colors. The module property was determined as the first component of principle component analysis (PCA) and used to calculate the association with individual clinical traits. Death after IPF diagnosis Result: was significantly correlated with the property of green-module containing 156 genes and anti-correlated with the property of turquoise-module containing 1252 genes. The turquoise-module property was also simultaneously correlated with the forced vital capacity (FVC) and diffusion capacity of lung (DLCO) and anti-correlated with composite progression index (CPI). Gene Ontology analysis revealed enrichment of green-module genes in defense response to bacterium and xenobiotic metabolic process (adjusted p-value 9.2x10-9 and 2.4x10-4), and enrichment of turquoise-module genes in DNA repair, mitosis and cell division (adjusted p-value 0.005, 0.01 and 0.0016, respectively). In the validation dataset, Receiver-Operating-Characteristic analysis for the prediction of death revealed area-under-curve (AUC) of 0.75 with green-module and 0.82 with turquoise-module. Moreover, turquoise-module demonstrated AUC of 0.81 for the diagnosis of IPF stability prior to PBMC sampling. The sample scatter plot by PCA accurately distinguished all deceased IPF patients from healthy individuals, with living IPF scattered across both groups. Two gene co-expression modules correlated Conclusion: with pulmonary function and prognosis of IPF were identified and validated in two large cohorts. Simultaneous correlation of multiple clinical traits with gene modules can greatly facilitate the elucidation of the relationships between individual clinical parameters along with the underpinning molecular signaling pathways. These findings indicate that genomic strategy may be a powerful tool in the prediction and mechanism exploration of the progression of chronic lung diseases.