Radiation Effects on Normal Brain in Subjects With Recurrent Glioblastoma by Spectroscopic MRI

Glioblastoma (GBM) is the most common and aggressive primary malignant brain tumor in adults and almost always recurs. In recurrent GBM, conventional MRI has limited ability to detect microscopic tumor infiltration and distinguish progression from treatment-related change. Spectroscopic MRI (sMRI) provides quantitative metabolic maps across much of the brain, including choline (Cho) and N-acetylaspartate (NAA), a marker of healthy tissue. Because Cho/NAA is commonly used to define metabolically abnormal tissue, and prior radiation therapy (RT) can alter metabolite levels, we evaluated whether prior RT shifts baseline Cho/NAA in recurrent disease. We retrospectively studied 20 patients with recurrent GBM previously treated with maximal safe resection, RT, and temozolomide who underwent whole-brain sMRI at recurrence. The median interval from RT completion to sMRI was 8.99 months. T1-weighted images were co-registered to sMRI in MIDAS and aligned with planning CT in MIM to generate radiation dose maps. MIDAS generated tissue-water-referenced Cho, NAA, and creatine (Cr) maps, and Cho/NAA was normalized to contralateral normal-appearing white matter. Voxel-wise linear regression was performed between prior radiation dose and metabolite values at recurrence. Higher prior dose was associated with reduced NAA (-0.26%/Gy, r² = 0.011, p < 0.001) and Cr (-0.18%/Gy, r² = 0.004, p < 0.001), while Cho changed minimally (-0.023%/Gy, r² = 0.001, p < 0.001). Accordingly, normalized Cho/NAA increased with dose, with a mean slope of 0.0018/Gy (p < 0.001). A standard Cho/NAA threshold of 2 corresponded to a median of 2.39 (range, 1.91-2.72) in tissue previously receiving 60 Gy. These findings suggest prior RT modestly elevates baseline Cho/NAA, primarily through NAA reduction, and that dose-corrected Cho/NAA maps may improve tumor delineation in recurrent GBM.