Multimodal approach for diagnosis of bacterial etiology in brain abscess

AbstractBackground and purpose magnetic resonance spectroscopy (PMRS) has high sensitivity and specificity for the detection of pyogenic brain abscess and the categorization of bacteria. But the metabolite patterns failed to evaluate the etiology of disease when the culture results are sterile. The aim of the present study is to compare the multimodality techniques viz., conventional culture, MR spectroscopy and 16S rRNA PCR and sequencing for rapid diagnosis of etiology in brain abscess and evaluate the PMRS in culture sterile samples and also demonstrate the sensitivity and specificity of these techniques. s five patients underwent MRI on a 3 T MRI and in-vivo PMRS for the diagnosis and evaluation of various resonances of metabolites such as lipid (LIP), lactate (LAC), acetate (AC), amino acid (AC), succinate (SUC). Pus was collected for identification of etiologic agents by culture and molecular method. s samples, metabolite patterns were as follows: LIP/LAC/AA, n = 17, LIP/LAC/AA/SUC with or without AC, n = 17 and LIP/LAC/AA/AC, n = 1. Culture showed bacterial growth in 22 samples (18 aerobic/facultative anaerobic, 9 anaerobic) whereas molecular method was detected 26 aerobic/facultative anaerobic, 13 anaerobic, 4 microaerophilic bacteria. Among the 13 sterile samples, molecular method detected 16 microorganisms along with 3 mixed infections and PMRS recognized metabolite patterns as LIP/LAC/AA, n = 5 and LIP/LAC/AA/SUC with or without AC, n = 8. The sensitivity of in-vivo PMRS in sterile samples was 100% and 75%, and specificity was 75% and 100% for aerobic and anaerobic organisms respectively. sion on metabolite resonances, PMRS can detect slow growing and fastidious organisms and classify them into aerobic and anaerobic bacteria which are difficult to culture by conventional method. It can categorize microorganisms even in culture sterile samples with rational sensitivity and specificity which may allow early choice of targeted therapy.