Soluble triggering receptor expressed on myeloid cells 1 (sTREM-1) predicts mortality in patients with febrile illness in southern Mozambique

Abstract

Background

Fever is a leading reason for seeking healthcare globally. Early in the course of febrile illness, it is challenging to identify patients at risk of severe and fatal infections. Quantifying biomarkers of immune and endothelial activation may facilitate patient triage.

Methods

We prospectively enrolled children ≥2 months and adults with fever visiting two Mozambican hospitals from December 2018 to February 2021. Standard clinical and laboratory parameters, including lactate levels, were assessed at presentation. Plasma levels of Angpt-2, CHI3L1, CRP, IL-6, IL-8, PCT, sFlt-1, sTNFR1, sTREM-1, and suPAR at presentation were retrospectively quantified. Clinical outcomes were evaluated up to 28 days. We assessed the prognostic performance of biomarkers for 28-day mortality and explored their association with other adverse outcomes.

Results

This study includes 1955 participants, with 93 deaths occurring within 28 days. We show that all biomarker levels are elevated in inpatients compared to outpatients and are associated with 28-day mortality (all p < 0.001). sTREM-1 is the best biomarker predicting 28-day mortality with an AUROC of 0.82 (95% CI: 0.78-0.86), superior to that of PCT (p < 0.001), CRP (p < 0.001), and lactate (p = 0.0033). Its prognostic performance is consistent across age and sex, but is reduced in HIV-positive individuals (AUROC = 0.73, 95% CI: 0.66-0.80). Adding sTREM-1 improves the discrimination of clinical severity scores for 28-day mortality. Among discharged inpatients, sTREM-1 is positively correlated with duration of hospitalisation (p < 0.001). Among outpatients, sTREM-1 levels are higher in those seeking further care (p = 0.0022) or subsequently hospitalised (p = 0.012).

Conclusions

sTREM-1 is a promising biomarker for risk stratification of all-age, all-cause febrile illnesses in resource-limited settings.