Plasma multi-omics profiling unravels molecular alterations and biomarker panels during recovery from severe coronavirus disease 2019: a pilot study.

BACKGROUND

The potential multidimensional molecular alterations during recovery of severe patients with coronavirus infectious disease (COVID-19) remain to be elucidated. Early assessment of the prognosis of severe COVID-19 may facilitate appropriate medical interventions.

METHODS

In this small-cohort exploratory study, plasma proteomic and widely targeted metabolomic profiling were conducted on 24 severe COVID-19 patients: 12 patients who underwent severe-to-mild transformation with plasma samples available at three consecutive time points (T1: severe stage, T2: moderate stage, T3: mild stage) for longitudinal analysis, and 12 patients who remained persistently severe or progressed to death with only T1 samples. Temporal analyses were performed to identify altered molecules with consistent trends during remission in severe COVID-19. Subsequently, we compared the differential traits at T1 between severe COVID-19 with two opposite outcomes: those with amelioration from severe to mild illness and those with persistent-severe illness. We also applied a machine learning model to explore biomarker panels predictive of severe COVID-19 prognosis.

RESULTS

During the remission phase of severe COVID-19, a distinct dynamic balance in the regulation of inflammation-associated molecules was observed. Notably, acute phase proteins, including SAA1, SAA2, and CRP, were all remarkably downregulated. Pathway analysis emphasized the essential role of lipid metabolism in the dynamic improvement of severe COVID-19. Furthermore, molecules implicated in lipid metabolism demonstrated significant consistency of alteration as the condition ameliorated, such as the consistent upregulation of APOC1 and various phospholipids, contrasted with the sustained downregulation of certain acylcarnitines. Through LASSO logistic regression, a biomarker panel comprising the proteins RPLP0, CKB and the metabolite Leu-Asp was identified as a promising predictive model. It significantly differentiated the prognosis of severe COVID-19 patients, with superior predictive accuracy (AUC: 0.922 in the training set; AUC: 0.875 in the validation set).

CONCLUSIONS

In the small cohort, multi-omics investigation provides novel exploratory insights into the intricate and dynamic regulation of the inflammatory response and lipid metabolism during the recovery phase of severe COVID-19. Furthermore, we have established a highly valuable biomarker panel that facilitates the early identification of severe COVID-19 prognoses.

CLINICAL TRIAL NUMBER: Not applicable.