BACKGROUND: The purpose of this study was to use ex vivo NMR spectroscopy of image guided tissue samples in order to characterize the metabolic signatures of gliomas and to determine which in vivo parameters are likely to be of interest for directing treatment, evaluating response and predicting outcome. METHODS: A total of 254 image guided tissue samples were obtained from 126 patients with glioma who were undergoing surgical resection for either a new or recurrent lesion. Half of each sample was flash frozen for analysis using ex vivo high resolution magic angle spinning (HRMAS) 1H spectroscopy with a 500MHz NMR scanner. Metabolite levels were estimated from the ex vivo spectra using the HR-QUEST software package. The other half of each sample was fixed in paraffin for subsequent histological analysis. Confirmation of the presence of tumor, the definition of tumor grade and evaluation of histology were made by an experienced pathologist. Associations between samples with different pathologies and/or from different anatomic regions were evaluated using a proportional odds logistic regression model with repeated measures in order to take account of the multiple samples per subject. RESULTS: Significant differences in the ex vivo metabolite levels were observed between samples from groups that were separated based upon (i) primary tumor grade, (ii) IDH status, (iii) having arisen as primary vs secondary GBM, (iv) being classified as corresponding to recurrent GBM vs treatment effect and (v) coming from regions of GBM that were contrast enhancing versus non-enhancing on T1-weighted post-Gadolinium images. Key metabolites that were relevant for making distinctions between the various groupings included levels of 2-hydroxglutarate, myo-inositol, phosphocholine, glycerophosphcholine, glutathione, glycine, glutamate, creatine, N-acetylasparate and mobile lipids. CONCLUSIONS: Many of the metabolites that were highlighted as being associated with different pathologies and regional imaging characteristics can be measured using clinically available 3T whole body MR scanners in conjunction with in vivo 1H MR spectroscopic imaging. Recent advances in the methodology for acquiring and analyzing such data have made it possible to obtain maps of metabolite levels from the brain in a scan time of 5-10 minutes. Applications to patient care include assessment of tumor burden, prediction of outcome and tailoring treatment to individual patient characteristics.Supported by: NIH P50 CA097257, RO1 CA127612 and P01 CA118816. SECONDARY CATEGORY: Neuropathology & Tumor Biomarkers.