1133 - Defining the Occult Margin in Recurrent Glioblastoma Using Spectroscopic MRI: Implications for Radiation Therapy Volumes

J. B. Bell¹, S. Sheriff², M. Goryawala², K. Cullison¹, G. Azzam¹, J. J. Meshman¹, M. C. Abramowitz³, M. Ivan⁴, M. De La Fuente⁵, and E. A. Mellon¹; ¹Department of Radiation Oncology, Sylvester Comprehensive Cancer Center, Miller School of Medicine, University of Miami, Miami, FL, ²Department of Radiology, Miller School of Medicine, University of Miami, Miami, FL, ³Department of Radiation Oncology, University of Miami/Sylvester Comprehensive Cancer Center, Miami, FL, ⁴Department of Neurosurgery, University of Miami Medical Center, Miami, FL, ⁵Department of Neurology, Sylvester Comprehensive Cancer Center, Miller School of Medicine, University of Miami, Miami, FL

Purpose/Objective(s): Overall survival is improved in glioblastoma (GBM) by multimodality therapy including surgical resection and adjuvant radiation (RT). In primary GBM (pGBM), a classic problem is the delineation of the infiltrating microscopic tumor margin for focal therapies. In recurrent GBM (rGBM), the imaging is further confounded by changes from the pGBM treatments. As a result, rGBM is often not given RT due to confusing imaging or delivered with a tiny clinical target volume (CTV) that is likely missing high-risk non-enhancing tumor due to fears about re-RT radiation necrosis. Whole-brain spectroscopic MRI (sMRI) is an emerging MRI technique with similar resolution to brain PET that can detect native metabolites and has been applied to define CTV in pGBM. We wished to determine whether it may also have usefulness in defining the CTV for rGBM. Materials/

Methods: Patients with pGBM (n=21) and rGBM (n=21) were prospectively consented for sMRI. T1-post contrast (T1PC) and T2/FLAIR MRI volumes were contoured. sMRIs were used to generate relative choline:N-acetylaspartate>2 (rCN>2) maps normalized to contralateral normal appearing white matter, which have been validated to correlate strongly with the presence of occult tumor invasion. sMRI-defined volumes were contoured using the rCN>2 maps. Hausdorff distances were calculated to define the margin necessary to cover rCN>2 in both pGBM and rGBM. In rGBM, mock CTV expansions from the T1PC volume were created to determine non-selective CTV expansions needed to cover occult tumor invasion.
Results: For pGBM, the mean T1PC volume was 49.9 cc, mean T2/FLAIR volume was 108.5 cc, and mean rCN>2 was 72.5 cc. For rGBM, the mean T1PC volume was 28.6 cc, mean T2/FLAIR volume was 140.7 cc, and mean rCN>2 was 51.8 cc. The ratio of T1PC to T2/FLAIR volumes was 0.53 in pGBM and 0.25 in rGBM (p<0.001). The mean Hausdorff distance indicating the maximum additional distance between T1PC and rCN>2 was 22.9 mm in pGBM and 28.3 mm in rGBM. Our findings support ~2 cm expansions in pGBM as per clinical standards. But, in rGBM, no CTV expansion from the T1PC volume (i.e., per RTOG 1205 trial protocol) resulted in only 54% of total disease coverage. With 10-, 15- and 20-mm CTV expansions from the T1PC volume, 94%, 98%, and 99% of total disease was covered, respectively.
Conclusion: sMRIs rCN>2 maps define occult disease in both pGBM and rGBM. These volumes extend approximately 2-3 cm beyond the disease seen on T1PC MRI. In rGBM, CTV expansions based on rCN>2 maps could allow for increased coverage of occult disease typically missed with standard radiation therapy volumes.