The purpose of this study was to evaluate the metabolite profiles obtained with short echo time (TE) magnetic resonance spectroscopic imaging (MRSI) of patients with non-enhancing gliomas. The major metabolites, such as Cho, Cr and NAA, were compared to those acquired from conventional long TE MRSI in order to assess the differences in properties that were observed within the lesions. Forty patients with suspected non-enhancing glioma were studied before surgical resection. Tumors were graded by histological examination of tissue samples from the surgeries. There were 11 patients with a diagnosis of grade 2 oligodendroglioma (OD), 8 with grade 2 astrocytoma (AS), 6 with grade 2 oligoastrocytoma (OA), 1 with grade 3 OD and 14 with grade 3 AS. The pre-surgical 3T MR exam included pre- and post-contrast T1-weighted spoiled gradient echo, T2-weighted fluid attenuated inversion recovery, short TE (N=40) and lactate-edited long TE (N=39) 3D H-1 MRSI data [1]. The MRSI data were obtained using CHESS water suppression, VSS outer volume suppression and PRESS volume selection (TE=35/144ms, spectral array =16x16x16, spatial resolution =1cm3). Flyback trajectories were applied in the S/I dimension to speed up the total acquisition time [2]. Regions of interest include the area with T2 hyperintensity (T2L) and normal appearing white matter (NAWM). The 3D spectral data were combined and processed as described previously [3,4]. All the spectra were quantified by LCModel [5] with simulated basis-sets. LCModel results were then converted to DICOM and visualized using SIVIC [6]. Metabolite levels were expressed relative to Cr, or normalized to the median concentrations in NAWM. Metabolites included in the analysis were those with Cramer-Rao lower bounds lower than 20%. The differences in metabolite levels from short TE MRSI between NAWM and T2L, or between grades were examined using signed-rank or rank-sum tests, with a cut-off of P=0.05 being used to define significance. An example of both short and long TE 3D MRSI data is illustrated in Figure 1. Voxels overlapping by at least 50% with the T2L were considering as being abnormal, and at least 75% for NAWM. The differences in median metabolite ratios from short TE MRSI between the NAWM and T2L were statistically significant for Cho/Cr (p<0.001), NAA/Cr (p<0.001), mIG/Cr (p<0.001), MM+Lip at 1.3ppm (p<0.001) and MM+Lip at 0.9ppm (p=0.007) (Figure 2). The levels of NAA/tCr and Glu/tCr were significantly higher for the grade 2 lesions relative to the grade 3 lesions (0.85±0.34 vs. 0.64±0.25, p=0.032; 1.55±0.32 vs. 1.32±0.27, p=0.019). Within the AS subtype, the difference in NAA/Cr between grade 2 and grade 3 lesions were sustained (0.85±0.29 vs. 0.59±0.19, p=0.038). For these patients who also had long TE MRSI, the Cho/Cr, NAA/Cr ratios and normalized levels between two acquisitions were plotted voxel by voxel in Figure 3. This suggests that there are differences in T2 between Cho and Cr/NAA within the T2L. This study has evaluated the metabolite profiles obtained using short TE MRSI in patients with non-enhancing gliomas at the time of pre-surgical resections. Differences in metabolite ratios were detected between the T2L and NAWM. Low grade OD has been reported to have increased normalized Glx compared to astrocytoma in the contrast-enhanced lesion [7], and mI/Cr is associated with tumor grade in AS based upon single voxel acquisitions [8]. However, these metabolites were not found to be statistically significant within the T2L in our study. Whether this is due to variations in the populations considered or due to the methods used is unclear. Combining the MRSI with other imaging modalities such as diffusion could be valuable to localize and interpret spectral data for such non-enhancing glioma. The observed differences in Cho/Cr, NAA/Cr, and normalized metabolite levels between long/short TE acquisitions reflects the variations in relaxation times that were consistent with previous findings [9] and indicated that the levels of tumor marker, Cho/NAA, would be higher at long TE. Future studies will perform a more detailed analysis that combines DWI and clinical information, and evaluate metabolites from long TE MRSI [10].