Glioblastoma Response during Chemoradiation by Daily Quantitative Multiparametric MRI

Abstract

<h3>Purpose/Objective(s)</h3> Quantitative Multiparametric MRI (qMP-MRI) is a fast and robust technique for providing relaxometry information such as T1, R2 and R2* maps. This information is correlated to glioblastoma (GBM) cellular density and may be useful to predict sites of tumor progression. For example, T1 has longer and R2 and R2* have shorter relaxation times for voxels adjacent to gadolinium enhancing tumor. The availability of MRI-Linac systems provides the opportunity to detect qMP-MRI changes daily through the course of radiation therapy (RT) and to adapt plans during RT. We hypothesize that qMP-MRI can detect tumor changes during a course of MRI-Linac treatment and correlates with patient outcomes. <h3>Materials/Methods</h3> We analyzed an IRB-approved prospective cohort of 21 consecutive GBM patients treated on a 0.35T MRI-Linac system with 30 fractions to 60 Gy and temozolomide. Daily qMP-MRI data were obtained with multi-TE and the Strategically Acquired Gradient Echo (STAGE) on alternating treatment days (15 qMP-MRI acquisitions during RT). For STAGE, 36 slices of 1.5 × 1.5 × 5 mm³ resolution were acquired using varying TEs and flip angles. A multi-TE sequence with 18 slices of 1.6 × 1.6 × 6 mm³ were also acquired. Quantitative T1 and R2* (1/T2*) maps were obtained from STAGE acquisitions and R2 (1/T2) maps from the multi-TE acquisition. Image changes related to tumor area and resection cavity were contoured on the T2 weighted images from daily patient setup using a standard contouring software. A ROI (Peri-tumor) was created based on a 10 mm expansion around the tumor or resection cavity. Changes of T1, R2* and R2 from each voxel within the Peri-tumor were analyzed by parametric response mapping comparing the images from each fraction to those obtained during pre-RT simulation. <h3>Results</h3> Of the 21 GBM patients, 10 had no tumor growth on first post-RT diagnostic MRI (48%), 4 (19%) were determined to have pseudoprogression based on regression or long-term stability of findings, and 7 (33%) had true progression due to continued progression of disease past 6 months, rapid patient death from disease, or tissue sampling showing active malignancy. The analysis of qMP-MRI obtained during RT showed that true progressors presented increases of T1 and decreases of R2* and R2 values for at least 10, 20 and 15%, respectively, of the voxels within the Peri-tumor when compared to patients with no change (Fisher's exact test, p<0.05 for each metric). All true progressors showed such changes before the 15^th fraction (3^rd week) of treatment. <h3>Conclusion</h3> Changes of qMP-MRI based parameters obtained during RT, such as increases in T1 and decreases in R2 and R2* within the Peri-tumor, were found to correlate with true progression. These changes were observed in images obtained before the third week of treatment, suggesting that adaptive RT or other therapies may be useful to intervene early for patients with GBM unresponsive to conventional therapy.