Minimalistic Transcriptomic Signatures Permit Accurate Early Prediction of COVID-19 Mortality

Rithwik Narendra ^1,2, Emily C Lydon ¹, Hoang Van Phan ¹

¹University of California San Francisco, San Francisco, CA 94115, USA.
²University of California Los Angeles, Los Angeles CA 90095, USA.
³Precision Vaccines Program, Boston Children's Hospital, Harvard Medical School, Boston, MA 02115, USA.
⁴National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, MD 20814, USA.
⁵Icahn School of Medicine at Mount Sinai, New York, NY 10029, USA.
⁶Research Computing, Department of Information Technology, Boston Children's Hospital, Harvard Medical School, Boston, MA 02115, USA.
⁷Drexel University, Tower Health Hospital, Philadelphia, PA 19104, USA.
⁸Benaroya Research Institute, University of Washington, Seattle, WA 98101, USA.
⁹Emory School of Medicine, Atlanta, GA 30322, USA.
¹⁰Yale School of Public Health, New Haven, CT 06510, USA.
¹¹Yale School of Medicine, New Haven, CT 06510, USA.
¹²Stanford University School of Medicine, Palo Alto, CA 94305, USA.
¹³La Jolla Institute for Immunology, La Jolla, CA 92037, USA.

⁰University of California San Francisco, San Francisco, CA 94115, USA.

Medrxiv : the Preprint Server for Health Sciences +

|

June 6, 2025

View abstract on PubMed

Summary

A novel three-gene blood signature and the OLAH gene accurately predict COVID-19 mortality within 48 hours of hospitalization. These findings offer accessible prognostic tools for early intervention and resource allocation in COVID-19 patients, including vaccinated individuals.

Area Of Science

Genomics and Transcriptomics
Infectious Disease Epidemiology
Clinical Prognostics

Background

Predicting COVID-19 mortality remains a significant clinical challenge.
Accurate prognostic assays are crucial for timely patient management.
This study addresses the unmet need for reliable early mortality prediction in COVID-19.

Purpose Of The Study

To identify and validate parsimonious transcriptomic signatures for predicting COVID-19 mortality.
To develop accessible prognostic tools for early risk stratification.
To assess the performance of these signatures in both unvaccinated and vaccinated patients.

Main Methods

Analysis of host gene expression in 894 hospitalized COVID-19 patients using RNA sequencing.
Development of prognostic classifiers using LASSO regression incorporating gene expression, age, and viral load.
External validation in an independent cohort of 137 COVID-19 patients, including vaccinated individuals.

Main Results

Fatal COVID-19 showed distinct peripheral blood gene expression patterns, differing from sepsis.
A three-gene peripheral blood classifier (CD83, ATP1B2, DAAM2) achieved an AUC of 0.88.
The OLAH gene alone demonstrated strong predictive performance (AUC 0.86), and classifiers were validated in vaccinated patients.

Conclusions

A three-gene peripheral blood signature and OLAH gene expression accurately predict COVID-19 mortality early in hospitalization.
These classifiers show promise as accessible prognostic tools for guiding clinical decisions.
Further research is warranted to implement these signatures for triage and early therapeutic interventions.