University of Miami Sylvester Comprehensive Cancer Center Miami, FL
K. Cullison1, K. Samimi1, J. B. Bell1, D. Maziero2, A. Valderrama1, A. L. Breto1, K. Jones3, M. De La Fuente4, G. J. Kubicek1, J. J. Meshman1, G. Azzam1, J. Ford1, R. Stoyanova1, and E. A. Mellon1; 1Department of Radiation Oncology, Sylvester Comprehensive Cancer Center, Miller School of Medicine, University of Miami, Miami, FL, 2UCSD Health Radiation Medicine and Health Sciences, La Jolla, CA, 3West Physics, Atlanta, GA, 4Department of Neurology, Sylvester Comprehensive Cancer Center, Miller School of Medicine, University of Miami, Miami, FL
Purpose/Objective(s): Contrast MRI after chemoradiotherapy (chemoRT) shows areas of possible tumor growth in ~50% of glioblastoma patients compared to pre-RT, but changes during chemoRT are rarely investigated due to logistics of frequent standalone MRI. Daily MRI is feasible with MRI-linac, but the frequency of gadolinium contrast imaging for adaptive RT remains undefined. To identify the need for contrast during chemoRT, we analyzed findings on MRI-linac T2-weighted (T2) MRI and compared those to standalone T1-post contrast (T1+C) and T2 MRI. Materials/
Methods: Using an IRB-approved prospective cohort of glioblastoma patients undergoing 30 fractions of chemoRT to 60 Gy, standalone MRIs (1.5 or 3T) at 3 timepoints: 1) 1 week pre-RT, 2) mid-RT (end of week 4/beginning of week 5 of treatment) when available, and 3) 1-month post-RT, were compared with T2 treatment set-up scans on 0.35T MRI-linac. Two regions of interest (ROI) were contoured: 1) lesion and 2) cavity (post-surgical resection cavity). Lesion was defined as tumor and edema on T2 MRI, and as enhancing disease on T1+C MRI. ROI volumes from MRI-linac and standalone MRI acquired on the same day were compared using linear fits. Percent change from pre- to post-treatment was calculated for lesion. Results: Thirty-five patients that underwent RT on MRI-linac were analyzed.Defined on the MRI-linac pre-RT T2 planning scan, 8 patients had visible cavity only, 15 had a mix of both cavity and lesion, and 12 had lesion only (n=23 for cavity, n=27 for lesion). R2 values for linear fits between standalone MRI vs. MRI-linac can be seen in the table (n=number of comparisons included in model). There was a moderate correlation between T1+C and MRI-linac lesion, despite MRI-linac T2’s inability to separate contrast enhancement from surrounding non-enhancing tumor and edema. From pre- to post-treatment, standalone T1+C and MRI-linac T2 lesions changed together – shrank (n=7) or grew (n=11) – in 18 (51%) patients. Another 9 patients (26%) had growth on MRI-linac T2 while the T1+C component shrank. In no patient did T1+C lesion grow while the MRI-linac T2 shrank. No patients with cavity only on MRI-linac (n=8, 23%) had any lesion growth on T1+C from pre- to post-treatment.
Conclusion: Daily low-field MRI-linac non-contrast imaging demonstrates changes in resection cavity and tumor/edema during chemoRT. We found that these changes can be used as a signal for when to consider IV contrast for adaptive MRI-RT in glioblastoma. If T1+C volumes are desired for adaptation, a selective approach can be employed where MRI-linac T2/non-contrast lesion growth (n=20, 57%) can be used to trigger contrast injection at selected timepoints. In patients with no lesion growth on non-contrast imaging/MRI-linac (n=15, 43%), contrast administration may not be needed.
