Mass spectrometry (MS)-based proteomics and mortality analysis in idiopathic pulmonary fibrosis (IPF)

Rationale: Circulating proteins may be associated with mortality in patients with IPF. Methods: The cohort included 299 patients from the IPF-PRO Registry. Proteins were measured in plasma collected at enrollment using an Evosep One coupled to an Orbitrap Exploris. Data were analyzed with Spectronaut 14. Protein data were log2 transformed. Missing values were imputed; pyCombat was used for batch correction. The cohort was split 75/25 into train and test sets. Univariable Cox regression assessed associations between each protein and a composite of respiratory death or lung transplant. The false discovery rate was controlled at 5%. A multivariable Cox model with elastic net penalty identified predictors of the outcome in the train set. All proteins and select clinical factors (sex, age, FVC and DLco % predicted, oxygen use) were considered. Model performance was evaluated in the train and test sets using the C-index. Results: A total of 761 protein groups were detected. Forty-seven proteins were associated (p<0.05) with the outcome. Multivariable analyses identified a set of 7 predictors (4 proteins, 3 clinical) with a C-index of 0.78 in the train set and 0.72 in the test set. The variable importance of these predictors is shown in the Figure. Conclusions: MS-based proteomic analyses confirmed known and identified new candidate biomarkers of mortality risk in patients with IPF. erj;64/suppl_68/OA4555/F1F1F1