Cerebrospinal Fluid Metabolome in Central Nervous System Infections: A Study of Diagnostic Accuracy

Authors:
  • IGI‐PACE Study Group
    I‐PACE Study Group
Open Access
Publisher website
Google Scholar
Abstract
Objective: To assess the diagnostic accuracy of metabolites in cerebrospinal fluid (CSF) for central nervous system (CNS) infections. Methods: Patients were derived from three prospective cohort studies in the Netherlands. All studies included adults suspected of a CNS infection who underwent a diagnostic lumbar puncture. Metabolomics was performed on CSF using ultra‐high‐performance liquid chromatography with tandem mass spectrometry on a discovery and validation cohort. Metabolite quantification was the index test; a microbiologically confirmed diagnosis was the reference standard. Results: In total, 343 episodes were included, of whom 170 (50%) had a CNS infections and 173 (50%) episodes had other diagnoses. CNS infections included bacterial meningitis in 88 (26%), viral meningoencephalitis in 50 (15%), and other CNS infections in 32 (9%) episodes. Other diagnoses consisted of CNS autoimmune disorders in 21 (6%), other neurological diseases in 84 (24%), and systemic infections in 68 (20%) episodes. A distinct metabolomic profile was observed in CSF of CNS infections, particularly bacterial meningitis. Glucose, glycerate, 1.3‐diphosphoglyceric acid, pyruvate, lactate, taurine, and alpha‐ketoglutarate had the highest diagnostic accuracy (area under the curve 0.87 to 0.95). Combinations further improved diagnostic accuracy, resulting in models that outperformed both individual metabolites and CSF leukocytes. Episodes with CSF leukocytes between 5 and 1,000 cells per mm 3 showed similar results. Interpretation: CSF metabolites demonstrate high diagnostic accuracy for CNS infections, particularly bacterial meningitis. Combinations further improve the diagnostic performance, exceeding that of CSF leukocytes alone. These findings highlight the potential of cerebrospinal fluid metabolites to improve diagnostic accuracy in clinical practice. ANN NEUROL 2025;98:851–863
Funding Information
  • ZonMw (917.17.308)
  • H2020 European Research Council (101001237)

This publication has 25 references indexed in Scilit: