Rapid Etiological Classification of Meningitis by NMR Spectroscopy Based on Metabolite Profiles and Host Response

Bacterial meningitis is an acute disease with high mortality that is reduced by early treatment. Identification of the causative microorganism by culture is sensitive but slow. Large volumes of cerebrospinal fluid (CSF) are required to maximise sensitivity and establish a provisional diagnosis. We have utilised nuclear magnetic resonance (NMR) spectroscopy to rapidly characterise the biochemical profile of CSF from normal rats and animals with pneumococcal or cryptococcal meningitis. Use of a miniaturised capillary NMR system overcame limitations caused by small CSF volumes and low metabolite concentrations. The analysis of the complex NMR spectroscopic data by a supervised statistical classification strategy included major, minor and unidentified metabolites. Reproducible spectral profiles were generated within less than three minutes, and revealed differences in the relative amounts of glucose, lactate, citrate, amino acid residues, acetate and polyols in the three groups. Contributions from microbial metabolism and inflammatory cells were evident. The computerised statistical classification strategy is based on both major metabolites and minor, partially unidentified metabolites. This data analysis proved highly specific for diagnosis (100% specificity in the final validation set), provided those with visible blood contamination were excluded from analysis; 6-8% of samples were classified as indeterminate. This proof of principle study suggests that a rapid etiologic diagnosis of meningitis is possible without prior culture. The method can be fully automated and avoids delays due to processing and selective identification of specific pathogens that are inherent in DNA-based techniques.