Abstract
The development of sensitive and specific diagnostic methods for secondary bacterial meningitis remains an urgent challenge in neurosurgical and intensive care units. A combination of various clinical and biochemical parameters, as well as biomarkers and metabolites in cerebrospinal fluid (CSF), can be considered for constructing multivariate diagnostic models. In this study, 96 CSF samples from 53 patients with suspected secondary meningitis were analyzed. The first cohort, consisting of patients with sequelae of severe brain damage, included 7 patients (21 CSF samples) with and 29 patients (56 CSF samples) without secondary bacterial meningitis. The second cohort comprised patients after neurosurgical interventions, including 10 patients (12 CSF samples) with and 7 patients (7 CSF samples) without secondary bacterial meningitis. Combined group 1 with 33 CSF samples from patients with secondary bacterial meningitis and combined group 2 with 63 CSF samples from patients without secondary bacterial meningitis had statistically different cell and biochemical compositions and higher CSF concentrations of biomarkers (interleukin-6 and S100 protein) and lactate-containing aromatic metabolites in group 1. Univariate prognostic models constructed on 4-hydroxyphenyllactic, phenyllactic, and indole-3-lactic acids demonstrated outstanding AUC-ROC of more than 0.91. A multivariate model built on all biomarkers and metabolites resulted in AUC-ROC = 0.94 with a sensitivity of 0.94 and specificity of 0.86, and was found to be the most accurate method for the diagnosis of secondary bacterial meningitis.