Soluble TREM-1 plasma concentration predicts poor outcome in COVID-19 patients

Abstract

Background: The immuno-receptor Triggering Expressed on Myeloid cells-1 (TREM-1) is activated during bacterial infectious diseases, where it amplifies the inflammatory response. Small studies suggest that TREM-1 could be involved in viral infections, including COVID-19. We here aim to decipher whether plasma concentration of the soluble form of TREM-1 (sTREM-1) could predict the outcome of hospitalized COVID-19 patients.

Methods: We conducted a multicentre prospective observational study in 3 university hospitals in France. Consecutive hospitalized patients with confirmed infection with SARS-CoV-2 were enrolled. Plasma concentration of sTREM-1 was measured on admission and then at days 4, 6, 8, 14, 21, and 28 in patients admitted into an ICU (ICU cohort: ICUC) or 3 times a week for patients hospitalized in a medical ward (Conventional Cohort: ConvC). Clinical and biological data were prospectively recorded and patients were followed-up for 90 days. For medical ward patients, the outcome was deemed complicated in case of requirement of increased oxygen supply > 5 L/min, transfer to an ICU, or death. For Intensive Care Unit (ICU) patients, complicated outcome was defined by death in the ICU.

Results: Plasma concentration of sTREM-1 at inclusion was higher in ICU patients (n = 269) than in medical ward patients (n = 562) (224 pg/mL (IQR 144-320) vs 147 pg/mL (76-249), p < 0.0001), and higher in patients with a complicated outcome in both cohorts: 178 (94-300) vs 135 pg/mL (70-220), p < 0.0001 in the ward patients, and 342 (288-532) vs 206 pg/mL (134-291), p < 0.0001 in the ICU patients. Elevated sTREM-1 baseline concentration was an independent predictor of complicated outcomes (Hazard Ratio (HR) = 1.5 (1.1-2.1), p = 0.02 in ward patients; HR = 3.8 (1.8-8.0), p = 0.0003 in ICU patients). An sTREM-1 plasma concentration of 224 pg/mL had a sensitivity of 42%, and a specificity of 76% in the ConvC for complicated outcome. In the ICUC, a 287 pg/mL cutoff had a sensitivity of 78%, and a specificity of 74% for death. The sTREM-1 concentrations increased over time in the ConvC patients with a complicated outcome (p = 0.017), but not in the ICUC patients.

Conclusions: In COVID-19 patients, plasma concentration of sTREM-1 is an independent predictor of the outcome, although its positive and negative likelihood ratio are not good enough to guide clinical decision as a standalone marker.

Keywords: ARDS; COVID-19; Prognostication; TREM-1.

Fig. 1

Flow chart of the study

Fig. 2

sTREM-1 concentration and its discriminatory power for complicated outcome. a sTREM-1 plasma concentrations at admission in the conventional cohort (medical ward) and the ICU cohort of patients according to their outcome. P values were calculated with the Mann–Whitney U test. *p < 0.0001. b Receiver-operating curves for sTREM-1 based on the outcome of the conventional cohort, and c the ICU cohort of patients

Fig. 3

Survival curves according to baseline sTREM-1 plasma concentration. Kaplan Meier curves (a) in the conventional cohort for complicated outcome with sTREM-1 cutoff set at 224 pg/mL, and (b) in the ICU cohort for death with sTREM-1 cutoff set at 287 pg/mL. LogRank tests with respective p values at 0.0013 and < 0.0001

Fig. 4

Mortality in ICU according to sTREM-1 kinetics. ICU patients were segregated into four different groups according to the sTREM-1 baseline level and kinetics. Group A: patients with low baseline sTREM-1 (< median: 224 pg/mL) that remained low (n = 59); Group B: low sTREM-1 that increased by more than 20% (n = 49); Group C: high sTREM-1 (> 224 pg/mL) that decreased of more than 20% (n = 39); and Group D: high sTREM-1 that stayed high (n = 85). The associated mortality rates were, respectively, 5.1, 10.2, 23.1, and 33% in these different groups. p values were obtained with Chi-2 test. *p < 0.0001