T2 Cluster analysis of transcriptomic datasets to identify endotypes of idiopathic pulmonary fibrosis

RationaleConsiderable clinical heterogeneity in Idiopathic Pulmonary Fibrosis (IPF) suggests the existence of multiple disease endotypes. Identifying these endotypes could allow for the development of a biomarker-driven personalised medicine approach in IPF.ObjectivesTo improve our understanding of the pathogenesis of IPF by identifying clinically distinct groups of patients with IPF that could represent distinct disease endotypes.MethodsWe systematically selected three publicly available datasets containing gene expression data measured from whole blood (220 IPF cases total). These datasets were co-normalised, pooled and clustered. We then compared clinical and demographic traits across clusters and used gene enrichment analysis to identify biological pathways and processes that were over-represented among the genes that were differentially expressed across clusters. A classifier was developed to assign additional individuals with IPF to a cluster using expression data from a minimal number of genes. We validated the classifier using three additional independent datasets (194 IPF cases total) and compared its performance at predicting survival in IPF to that of a previous transcriptomic prognostic biomarker for IPF.ResultsWe identified three clusters of IPF patients with distinct transcriptomic signatures. These clusters demonstrated statistically significant differences in lung function (P=0.009) and mortality (P=0.009) between groups. One cluster appeared to consist of patients with favourable lung function and survival over time (low risk cluster), whilst the other two clusters contained patients with worse lung function and reduced survival (high risk clusters). Gene enrichment analysis implicated dysregulation of mitochondrial homeostasis, apoptosis, cell cycle and innate and adaptive immunity in the pathogenesis of these groups. We developed and validated a 13-gene cluster classifier that predicted mortality in IPF (figure 1).Abstract T2 Figure 1Survival over time for the IPF subjects in the validation datasets, stratified by risk group according to our 13-gene classifier. The P-value on the plot is from a log-rank test testing the two curves for equality. The dashed line on the plot indicates the median survival time for the risk group if this could be calculatedConclusionsThere are at least two groups of IPF patients with significant differences in survival and lung function that are discernible by blood gene expression signatures. These groups co