Abstract

Introduction: The molecular mechanisms linked to the pathology of severe COVID-19 and its outcomes are poorly described.

Aim: To analyze the proteomic profile of bronchial aspirates (BAS) samples from critically ill COVID-19 patients in order to identify factors associated with the disease and its prognosis.

Methods: Multicenter study including 74 critically ill non-COVID-19 and COVID-19 patients. BAS was obtained by bronchoaspiration after invasive mechanical ventilation (IMV) initiation. Proximity extension assay (PEA) technology was used for proteomic profiling. Random forest (RF) statistical models were used to predict the variable importance.

Results: After adjusting for confounding factors, CST5, NADK, SRPK2 and TGF-a showed differences between COVID-19 and non-COVID-19 patients. Reduced levels of ENTPD2 and PTN were observed in non-survivors, even after adjustment. AGR2, NQO2, IL-1a, OSM and TRAIL, were the top five strongest predictors for ICU mortality and were used to build a prediction model. PTN (HR=4.00) ENTPD2 (HR=2.14) and the prediction model (HR=6.25) were associated with higher risk of death. In survivors, FCRL1, NTF4 and THOP1 correlated with lung function (DLCO levels) 3-months after hospital discharge. Similar findings were observed for Flt3L and THOP1 and radiological features (TSS). The proteins identified are expressed in immune and non-immune lung cells. A poor control of viral infectivity and an inappropriate reparative response seems to be linked to the disease and fatal outcomes, respectively.

Conclusion: In critically ill COVID-19 patients, specific proteomic profiles are associated with the pathology, mortality and lung sequelae.