Purpose To evaluate the utility of 3D-MR proton spectroscopic imaging for treatment planning and its implications for early response assessment in glioblastoma multiforme. Methods and Materials Eighteen patients with newly diagnosed, histologically confirmed glioblastoma had 3D-MR proton spectroscopic imaging (MRSI) along with T2 and T1 gadolinium enhanced MR images at simulation and at boost treatment planning after 17-20 fractions of radiotherapy. All patients received standard radiotherapy with temozolomide and follow-up with every two month MR scans. Progression free survival was defined using MacDonald criteria. MRSI images obtained at initial simulation were analyzed for choline / N-acetylaspartate ratios (Cho/NAA) on a voxel by voxel basis with abnormal activity defined as Cho/NAA ≥ 2. These images were compared on anatomically matched MRSI data collected after 3 weeks of radiotherapy. Changes in Cho/NAA between pre-therapy and 3rd week RT scans were tested using Wilcoxon matched-pairs signed rank tests and correlated with progression free survival, radiation dose and location of recurrence using Cox proportional hazards regression. Results After 8.6 months (median follow-up), 50% of patients had progressed based on imaging. Patients with a decreased or stable mean or median Cho/NAA values had less risk of progression (p< 0.01). Patients with an increase in mean or median Cho/NAA values at the 3rd week RT scan had a significantly greater chance of early progression (p <0.01). An increased Cho/NAA at the 3rd week MRSI scan carried a hazard ratio of 2.72 (95% confidence interval 1.10-6.71, p= 0.03). Most patients received the prescription dose of RT to the Cho/NAA ≥ 2 volume, which was where recurrence most often occurred. Conclusion Change in mean and median Cho/NAA detected at 3 weeks was a significant predictor of early progression. The potential impact for risk-adaptive therapy based on early spectroscopic findings is suggested.

No abstract