Myo-inositol concentration in MR spectroscopy for differentiating high grade glioma from primary central nervous system lymphoma.

ABSTRACT

It is sometimes difficult to distinguish gliomas from other tumors on routine imaging. In this study, we assessed whether 3-T magnetic resonance spectroscopy (MRS) with LCModel software might be useful for discriminating glioma from other brain tumors, such as primary central nervous system lymphomas (PCNSLs) and metastatic tumors. A total of 104 cases of brain tumor (66 gliomas, 20 PCNSLs, 6 metastatic tumors, 12 other tumors) were preoperatively investigated with short echo time (35 ms) single-voxel 3-T MRS. LCModel software was used to evaluate differences in the absolute concentrations of choline, N-acetylaspartate, N-acetylaspartylglutamate, glutamate + glutamine, myo-inositol (mIns), and lipid. mIns levels were significantly increased in high-grade glioma (HGG) compared with PCNSL (p < 0.001). In multivariate logistic regression analysis, mIns was the best marker for differentiating HGG from PCNSL (p < 0.0001, odds ratio 1.9927, 95% confidence interval 1.3628-3.2637). Conventional MRS detection of mIns resulted in a high diagnostic accuracy (sensitivity, 64%; specificity, 90%; area under the receiver operator curve, 0.80) for HGG. The expression of inositol 3-phosphate synthase (ISYNA1) was significantly higher in gliomas than in PCNSLs (p < 0.05), suggesting that the increased level of mIns in glioma is due to high expression of ISYNA1, the rate-limiting enzyme in the mIns-producing pathway. In conclusion, noninvasive analysis of mIns using single-voxel MRS may be useful in distinguishing gliomas from other brain tumors, particularly PCNSLs.