Proteomic predictors of mortality in stable COPD

COPD is the 3^rd leading cause of death worldwide. Clinical mortality predictors have been extensively described, including age, symptoms, comorbidities, lung function, and some biomarkers. However, proteomics can provide a more comprehensive blood profile linked to its pathophysiological mechanisms. <bold>We aimed</bold> to identify proteomic predictors of long-term mortality in stable patients. <bold>Methods:</bold> Blind LC-MS/MS and hypothesis-driven multiplex immunoassay approaches were used to analyse plasma samples from 5-year-followed 34 COPD patients (69±9 years old, 29% ♀, FEV₁ 41±15% ref., DL_co 47±18% ref.). Reactome and UniProt were used for protein categorisation and Flame for AI modelling. <bold>Results:</bold> 32% of patients died before the 4^th year, mostly by cardiovascular (45%) or respiratory (45%) events. 402 proteins/peptides were identified: 209 were valid for quantitative comparisons (data present in ≥80% of individuals in both groups), and the remaining were qualitatively analysed. 35 proteins showed differences between survivors and non-survivors, mainly related to the immune response (12), haemostasis (10), and complement cascade (4). However, for the predictor model (5-fold CV: 94% accuracy, 89% MCC, 85% coverage), 10 proteins were selected. Those were linked to hemostasis (4), complement cascade (2), innate immunity (1), and muscle structures (2). <bold>Conclusion:</bold> A blood proteomic signature, composed of elements mainly linked to coagulation and complement cascades, may be a valuable predictor of long-term mortality in stable COPD patients. <bold>Funding</bold>: SAF2014-54371 (Spain-EU); CIBERES, PI21/00785, BA22/00009, FI22/000003 & MV23/00012 (ISCiii-EU); SEPAR 2015, 2016, 2019 & 2021; FUCAP 2014, SOCAP 2020, Menarini Grant 2015; 2021 SGR 00100 (Catalan Gov.)