Magnetic Resonance Imaging Relaxometry for Glioblastoma Response Assessment During Radiation Therapy on a 0.35 T Magnetic Resonance Imaging Linear Accelerator.

Abstract

The integration of magnetic resonance imaging (MRI) and linear accelerator (MRI-Linac) enables daily imaging during radiation therapy (RT). This study implements MRI-Linac relaxometry to evaluate quantitative imaging changes in patients with glioblastoma during RT and identify associations with disease progression and survival outcomes. Thirty-eight patients with glioblastoma were treated on a 0.35 T MRI-Linac with Strategically Acquired Gradient Echo and T2 multiecho acquisitions every other day. Per voxel changes in tumor T2, T2*, and T1 values were assessed by parametric response mapping comparing each treatment fraction with pre-RT baselines. Statistical analyses included the Wilcoxon test for group comparisons and Cox proportional hazards models for survival associations. Progressors had higher proportions of voxels with increased T2 values at week 2 (49% vs 40%, P = .008) and week 6 (58% vs 43%, P = .012) and higher T2* values at week 1 (47% vs 43%, P = .016), week 2 (48% vs 43%, P = .016), week 3 (50% vs 44%, P = .012), and the final week (53% vs 43%, P = .021). Cox modeling linked increased T2 values at week 4 with overall survival (hazard ratio [HR], 4.72; 95% CI, 1.24-12.9) and progression-free survival (HR, 9.26; 95% CI, 1.88-24.5). Increased T2* values at weeks 2 and 3 correlated with progression-free survival (HR, 5.02; 95% CI, 1.44-17.6; HR, 6.04; 95% CI, 1.59-22.9) and overall survival at week 3 (HR, 3.09; 95% CI, 0.94-10.1). Quantitative changes in T2 and T2* values during RT, particularly in weeks 3 to 4, were associated with progression and survival outcomes. Early detection of poor responders may enable therapy adaptation, improving glioblastoma treatment outcomes.