ABSTRACT
Background: Mid-regional pro-adrenomedullin (MR-proADM), an endothelium-related peptide, is a predictor of death and multi-organ failure in respiratory infections and sepsis and seems to be effective in identifying COVID-19 severe forms. The study aims to evaluate the effectiveness of MR-proADM in comparison to routine inflammatory biomarkers, lymphocyte subpopulations, and immunoglobulin (Ig) at an intensive care unit (ICU) admission and over time in predicting mortality in patients with severe COVID-19. Methods: All adult patients with COVID-19 pneumonia admitted between March 2020 and June 2021 in the ICUs of a university hospital in Italy were enrolled. MR-proADM, lymphocyte subpopulations, Ig, and routine laboratory tests were measured within 48 h and on days 3 and 7. The log-rank test was used to compare survival curves with MR-proADM cutoff value of >1.5 nmol/L. Predictive ability was compared using the area under the curve (AUC) and 95% confidence interval (CI) of different receiver-operating characteristic curves. Results: A total of 209 patients, with high clinical severity [SOFA 7, IQR 4-9; SAPS II 52, IQR 41-59; median viral pneumonia mortality score (MuLBSTA)-11, IQR 9-13] were enrolled. ICU and overall mortality were 55.5 and 60.8%, respectively. Procalcitonin, lactate dehydrogenase, D-dimer, the N-terminal prohormone of brain natriuretic peptide, myoglobin, troponin, neutrophil count, lymphocyte count, and natural killer lymphocyte count were significantly different between survivors and non-survivors, while lymphocyte subpopulations and Ig were not different in the two groups. MR-proADM was significantly higher in non-survivors (1.17 ± 0.73 vs. 2.31 ± 2.63, p < 0.0001). A value of >1.5 nmol/L was an independent risk factor for mortality at day 28 [odds ratio of 1.9 (95% CI: 1.220-3.060)] after adjusting for age, lactate at admission, SOFA, MuLBSTA, superinfections, cardiovascular disease, and respiratory disease. On days 3 and 7 of the ICU stay, the MR-proADM trend evaluated within 48 h of admission maintained a correlation with mortality (p < 0.0001). Compared to all other biomarkers considered, the MR-proADM value within 48 h had the best accuracy in predicting mortality at day 28 [AUC = 0.695 (95% CI: 0.624-0.759)]. Conclusion: MR-proADM seems to be the best biomarker for the stratification of mortality risk in critically ill patients with COVID-19. The Ig levels and lymphocyte subpopulations (except for natural killers) seem not to be correlated with mortality. Larger, multicentric studies are needed to confirm these findings.