The correspondence of cell state changes in diseased organs to peripheral protein signatures is currently unknown. Here, we generated and integrated single‐cell transcriptomic and proteomic data from multiple large pulmonary fibrosis patient cohorts. Integration of 233,638 single‐cell transcriptomes (n = 61) across three independent cohorts enabled us to derive shifts in cell type proportions and a robust core set of genes altered in lung fibrosis for 45 cell types. Mass spectrometry analysis of lung lavage fluid (n = 124) and plasma (n = 141) proteomes identified distinct protein signatures correlated with diagnosis, lung function, and injury status. A novel SSTR2+ pericyte state correlated with disease severity and was reflected in lavage fluid by increased levels of the complement regulatory factor CFHR1. We further discovered CRTAC1 as a biomarker of alveolar type‐2 epithelial cell health status in lavage fluid and plasma. Using cross‐modal analysis and machine learning, we identified the cellular source of biomarkers and demonstrated that information transfer between modalities correctly predicts disease status, suggesting feasibility of clinical cell state monitoring through longitudinal sampling of body fluid proteomes.

Rationale Probe-based confocal endomicroscopy provides real time videos of autoflourescent elastin structures within the alveoli. With it, multiple changes in the elastin structure due to different diffuse parenchymal lung diseases have previously been described. However, these evaluations have mainly relied on qualitative evaluation by the examiner and manually selected parts post-examination. Objectives To develop a fully automatic method for quantifying structural properties of the imaged alveoli elastin and to perform a preliminary assessment of their diagnostic potential. Methods 46 patients underwent probe-based confocal endomicroscopy, of which 38 were divided into 4 groups categorizing different diffuse parenchymal lung diseases. 8 patients were imaged in representative healthy lung areas and used as control group. Alveolar elastin structures were automatically segmented with a trained machine learning algorithm and subsequently evaluated with two methods developed for quantifying the local thickness and structural connectivity. Measurements and main results The automatic segmentation algorithm performed generally well and all 4 patient groups showed statistically significant differences with median elastin thickness, standard deviation of thickness and connectivity compared to the control group.

No abstract