Development of an MR spectroscopic index to differentiate tumor from treatment induced gliosis

Introduction The goal of this study was to identify MR spectroscopic (MRS) markers that are likely to be able to differentiate active tumor from treatment induced gliosis. This is an important problem because while tumor regions should be included in follow-up treatment plans, regions represented by gliosis should be left untouched since they represent areas of normal brain that are being influenced by treatment. Typically high MRS choline levels are used to indicate tumor presence. The challenge in differentiating tumor from gliosis using this metric is that both of them could result in elevated levels of choline. While the components that contribute to the choline peak may be different in tumor and gliosis, this is not evident from an in-vivo spectrum, which suffers from low spectral resolution. Hence proliferation and hypertrophy of astrocytes in response to surgical trama and various therapeutic modalities may be indistinguishable from tumor growth. To evaluate strategies for overcoming this ambiguity we use the following approach: 1) ex-vivo samples from the UCSF tissue bank that are confirmed as being tumor or gliosis with histopathology were studied using High Resolution Magic Angle Spectroscopy (HRMAS) to identify specfic markers that differ between the two cases (non-image guided study), 2) the behavior of the same markers were investigated with ex-vivo HRMAS of biopsies obtained from within tumor regions from patients with newly diagnosed GBM with image guidance and 3) these markers were compared against known measures of tumor presence, choline-NAA index (CNI) [1] derived in-vivo from the 3T MRS voxel that contained the biopsy site. The high spectral resolution of the HR-MAS implies that it can reveal several differentiating markers between tumor and gliosis. Because the long-term goal of this study is to identify in-vivo markers we focused on metabolites that can be investigated with the 3T whole body scanner.