Predictive gene expression signature diagnoses neonatal sepsis before clinical presentation

Area of Science:

• Genomics
• Neonatal Medicine
• Biomarker Discovery

Background:

• Neonatal sepsis presents with non-specific signs, delaying critical diagnosis and treatment.
• Early biomarkers are crucial for timely intervention in neonatal sepsis.
• This study aimed to identify gene expression biomarkers for predicting sepsis at birth.

Purpose of the Study:

• To identify neonatal sepsis gene expression biomarkers predictive of sepsis at birth.
• To develop a predictive signature for early-onset sepsis (EOS) prior to clinical presentation.

Main Methods:

• RNA sequencing (RNA-seq) on peripheral blood from 720 neonates at birth and within the first week.
• Identified differentially expressed genes (DEGs) in neonates who later developed sepsis (EOS/LOS) versus controls.
• Applied machine learning (sPLS-DA, LASSO) to identify predictive gene signatures at birth.

Main Results:

• Neonates who later developed EOS showed ~1000 DEGs at birth compared to controls.
• A 4-gene signature (HSPH1, BORA, NCAPG2, PRIM1) accurately predicted EOS at birth (AUC=0.94).
• This signature was validated in an external cohort (AUC=0.72) and showed high sensitivity and specificity.

Conclusions:

• Distinct whole blood gene expression changes at birth predict later development of EOS.
• A 4-gene predictive signature enables early recognition of neonatal sepsis.
• Early identification and treatment of EOS can potentially mitigate long-term sequelae.