Proton magnetic resonance spectroscopic imaging in newly diagnosed glioblastoma: predictive value for the site of postradiotherapy relapse in a prospective longitudinal study.

PURPOSE To investigate the association between magnetic resonance spectroscopic imaging (MRSI)-defined, metabolically abnormal tumor regions and subsequent sites of relapse in data from patients treated with radiotherapy (RT) in a prospective clinical trial. METHODS AND MATERIALS Twenty-three examinations were performed prospectively for 9 patients with newly diagnosed glioblastoma multiforme studied in a Phase I trial combining Tipifarnib and RT. The patients underwent magnetic resonance imaging (MRI) and MRSI before treatment and every 2 months until relapse. The MRSI data were categorized by the choline (Cho)/N-acetyl-aspartate (NAA) ratio (CNR) as a measure of spectroscopic abnormality. CNRs corresponding to T1 and T2 MRI for 1,207 voxels were evaluated before RT and at recurrence. RESULTS Before treatment, areas of CNR2 (CNR > or =2) represented 25% of the contrast-enhancing (T1CE) regions and 10% of abnormal T2 regions outside T1CE (HyperT2). The presence of CNR2 was often an early indicator of the site of relapse after therapy. In fact, 75% of the voxels within the T1CE+CNR2 before therapy continued to exhibit CNR2 at relapse, compared with 22% of the voxels within the T1CE with normal CNR (p < 0.05). The location of new contrast enhancement with CNR2 corresponded in 80% of the initial HyperT2+CNR2 vs. 20.7% of the HyperT2 voxels with normal CNR (p < 0.05). CONCLUSION Metabolically active regions represented a small percentage of pretreatment MRI abnormalities and were predictive for the site of post-RT relapse. The incorporation of MRSI data in the definition of RT target volumes for selective boosting may be a promising avenue leading to increased local control of glioblastomas.