BAL cell transcriptome predicts survival in IPF and can be used to gauge and model treatment effects interfering with the TGF-beta pathway

Introduction: Idiopathic pulmonary fibrosis (IPF) is a fatal disease with a variable course. Current prediction models of mortality which are based on clinical and physiological parameters have modest value in predicting survival. In addition there is little information whether changes in molecular pathways in the alveolar compartment are indicative of disease progression. Methods: BAL cells were harvested from 212 IPF patients at three different European tertiary referral centers: Freiburg (Germany), Siena (Italy) and Leuven (Belgium) between 2006-2014. Furthermore, BAL cells from 20 IPF patients were cultured in vitro w/wo a TGF-beta inhibitor, pirfenidone, nintedanib, CXCR4 inhibition. BAL transcriptome was analyzed by Agilent gene expression arrays. Survival analysis was performed by Cox models and component-wise boosting. Results: 1582 genes were predictive of mortality in the IPF derivation cohort in univariate analyses adjusted for age and gender at FDR<0.05. A nine-gene signature, derived from the discovery cohort (Freiburg), performed well in both replication cohorts, Siena (p<0.0032) and Leuven (p=0.0033). In vitro experiments demonstrated strong and significant changes in BAL transcriptome with TGF-beta inhibition and CXCR4 inhibition. More than 1500 genes were differentially expressed after treatment (FDR< 0.05). Combining both datasets we modelled the treatment effect of both compounds on survival in IPF in silico. Conclusion: BAL cell transcriptome can be used to gauge treatment effects interfering with the TGF-beta pathway and indicates TGF-beta driven disease processes.