Editors' ChoiceCORONAVIRUS

Blood-borne biomarkers may help predict COVID-19 mortality

See allHide authors and affiliations

Science Translational Medicine  27 May 2020:
Vol. 12, Issue 545, eabb7102
DOI: 10.1126/scitranslmed.abb7102

Abstract

A machine learning algorithm identifies blood-borne markers that are predictive of COVID-19 mortality.

The COVID-19 pandemic is a global health crisis. To curtail the pandemic and enable a return to normalcy, rapid and widely accessible COVID-19 testing is urgently needed. Much of the current efforts to develop COVID-19 testing are devoted to tests that determine whether a person is currently infected or whether a person has previously been infected. Now, Yan et al. take a different perspective and report a blood test that can predict COVID-19 mortality.

The authors analyzed 75 clinical features, including the concentrations of markers in the blood of 485 infected individuals in Wuhan, and developed a mortality prediction model using machine learning tools. Their modeling identified three blood-borne markers that together classified the mortality of individual patients more than 10 days in advance with more than 90% accuracy. One of these markers alone, the level of lactate dehydrogenase (LDH) in the blood, was highly indicative of COVID-19 mortality. LDH is an enzyme involved in energy production and is found in almost all cells in the body. Tests that measure the concentration of LDH in the blood are commonly used to monitor tissue damage associated with a wide range of disorders, including liver disease and interstitial lung disease. The results from Yan et al. therefore suggest that tissue damage markers can be leveraged to predict COVID-19 outcomes.

The markers and model developed by the authors are important because they can possibly help to identify patients that require immediate medical attention, which will be vital to guide clinical decision making. The test could also find use as a surrogate marker to be included in clinical trials of candidate COVID-19 treatments, but additional studies will be required to assess this possibility. The results require validation in larger and independent patient cohorts and call for the development of additional host injury markers, including markers that are cell- and tissue-type specific, to help guide patient management.

Highlighted Article

View Abstract

Stay Connected to Science Translational Medicine

Navigate This Article