Updated project metadata. Single cell genomics enables characterization of disease specific cell states, while improvements in mass spectrometry workflows bring the clinical use of body fluid proteomics within reach. However, the correspondence of peripheral protein signatures to changes in cell state in diseased organs is currently unknown. Here, we leverage single cell RNA-seq and proteomics from large patient cohorts of pulmonary fibrosis to establish that predictive protein signatures in body fluids correspond to specific cellular changes in the lung. We determined transcriptional changes in 45 cell types across three patient cohorts and quantified bronchoalveolar lavage fluid and plasma proteins to discover protein signatures and associated cell state changes that were linked to diagnosis, lung function, smoking and injury status. Altered expression of the novel marker of lung health CRTAC1 in alveolar epithelium is robustly reported in patient plasma. With further improvements of this concept and deeper coverage of plasma proteomes, we envision future longitudinal profiling of body fluid signatures coupled to machine learning for non-invasive prediction and monitoring of pathological cell state changes in patient organs.