MSACL 2025 Abstract

Urinary Metabolites as Noninvasive Biomarkers for Neonatal Sepsis

Dace Sniedze (1), Zanda Daneberga (1), Theresa Schiemer (2), Jingzhi Fan (2), Julia Kapzems (3), Annija Vaska (2), Diāna Dakiča (4), Dace Gardovska (1), Kristaps Klavins (2)
(1) Rīga Stradiņš University, Riga, Latvia (2) Riga Technical University, Riga, Latvia (3) Johanniter Krankenhaus Bonn, Bonn, Germany (4) Children’s Clinical University Hospital, Riga, Latvia

Kristaps Klavins, PhD (Presenter) Riga Technical University

Relevant Financial Disclosures (within past 24 months, reported on Jul 17, 2025) No relevant financial relationship(s) to disclose. Conflict mitigation NOT REQUIRED.

INTRODUCTION:
Neonatal bacterial sepsis is a major contributor to morbidity and mortality worldwide. Early diagnosis is critical but remains challenging due to nonspecific clinical signs and limitations of current biomarkers. Blood sampling in neonates, especially those with very low birth weight, carries risks such as iatrogenic blood loss and pain. As a result, there is a strong clinical need for accurate, noninvasive diagnostic tools.

OBJECTIVE(S):
To identify urinary metabolites that serve as potential noninvasive biomarkers for neonatal sepsis and to develop a predictive model based on targeted metabolomics.

METHODS:
A prospective study was conducted involving 93 neonates, including 23 diagnosed with sepsis and 70 age-matched controls. Urine samples were collected noninvasively using standard neonatal care procedures. A targeted LC-MS-based workflow was employed to quantify 57 urinary metabolites. In the control group, confounding factors such as antibiotic exposure, chronological age, postmenstrual age, and feeding type were assessed to exclude metabolites influenced by these variables. Multivariate analysis and logistic regression were applied to identify the most discriminatory metabolites and construct a predictive model.

RESULTS:
Septic neonates showed significant decreases in normalized concentrations of adenosine and hydroxylysine. These changes were independent of potential confounding variables. A logistic regression model based on the absolute concentrations of three key metabolites yielded an area under the curve (AUC) of 0.9115, with 91% sensitivity and 90% specificity for sepsis diagnosis.

CONCLUSION:
Targeted urinary metabolomics offers a promising noninvasive approach for early diagnosis of neonatal sepsis. The identified metabolite-based model demonstrated high diagnostic accuracy and independence from common confounding factors. These findings support further validation and potential clinical integration of urinary metabolomics in neonatal care settings.