Abstract Background The association between immune-cell-specific transcriptomic profiles and mortality in IPF is unknown. Methods We profiled peripheral blood mononuclear cells (PBMC) by single-cell RNA sequencing (scRNA-seq), identified distinct immune-cell subpopulations and generated immune-cell gene signatures using the differentially expressed genes in each immune-cell subpopulation. Subsequently, we investigated which immune-cell-specific gene signature predicted IPF outcomes consistently. Prognostic accuracy was investigated in PBMC, Bronchoalveolar Lavage (BAL) and lung tissue. Findings were validated by cellular deconvolution, flow cytometry and analysis of independent scRNA-seq datasets. We investigated the function of this transcriptomic profile and its cellular source in lung tissue (overall sample size:1054, IPF:555, other:499). Connectivity map and LASSO regression were used to identify drug candidates and a subset of genes with prognostic potential, respectively. Results A 230-gene-up-score from CD14+CD163-HLA-DRlow monocytes predicted mortality in PBMC cohorts (Pittsburgh-HR: 3.00, 95% CI: 1.37-6.54, p = 0.006, Chicago-HR: 6.58, 95%CI:2.15-20.13, p = 0.001), in BAL pooled analysis (HR: 2.20, 95%CI: 1.44-3.37, p = 0.0003) and negatively correlated with Forced Vital Capacity (FVC) in lung tissues (ρ=-0.2, p = 0.02). Results were replicated through cellular deconvolution (p 0.05). Flow cytometry analysis in PBMC demonstrated that the proportion of CD14+CD163⁻HLA-DRlow monocytes was negatively correlated with FVC% predicted (ρ = -0.499, p = 0.0297). ScRNA-seq analysis showed that CD14+CD163-HLA-DRlow monocytes were higher in progressive versus stable IPF (12.59%,95%CI:9.66-16.23, vs 7.61%,95%CI:6.68-10.21, p = 0.014). High risk IPF patients had decreased expression of T-cell co-stimulatory genes (Pittsburgh and Chicago p 0.01). CD14+HLA-DRlow monocytes had higher expression of profibrotic, proangiogenic and chemotactic factors including Transforming Growth Factor-beta (TFB-b) one, Vascular Endothelial Growth Factor A, C-C chemokine receptor 1, Toll-like Receptor 2, and C-X-C motif chemokine ligand 2 compared to CD14+HLA-DRhi monocytes (p 0.05). ScRNA-seq analysis demonstrated that the 230-gene-up-score correlated with the SPP1+fibrosis-associated macrophages-gene-score in lung tissues (ρ = 0.19, p 2.2e-16). Connectivity map identified compounds such as modulators of phosphatidylinositol 3-kinase (PI3K) signaling, p38 mitogen-activated protein kinase inhibitors, nicotinamide phosphoribosyltransferase inhibitors, and TGFb inhibitors with potential to decrease the 230-gene-signature. Notably, connectivity map also showed that corticosteroids were among the compounds predicted to upregulate these genes. Finally, a 6-gene subset of the 230-gene signature retained predictive performance in PBMC, BAL and lung tissues (pooled PBMC cohorts - HR: 4.79,95%CI:2.58-8.92, p 0.0001, pooled BAL - HR: 2.03, 95% CI: 1.31-3.16, p = 0.0016, lung tissues - ρ = -0.37, p 0.0001). Conclusions The transcriptome of CD14+CD163-HLA-DRlow monocytes is associated with increased mortality in patients with IPF. Its reversal should be investigated as a precision-based-therapy in IPF. This abstract is funded by: Ubben Family Fund
Karampitsakos et al. (Fri,) studied this question.