Glioblastoma target volumes change substantially during MR-guided radiotherapy, often within the first week

KEY POINTS

• MARGA-I was a prospective single-center study of 25 patients with glioblastoma treated on a 0.35-T MR-linac using either 15 or 30 fractions. After exclusions, extended resection cavity analysis was available in 22 patients and FLAIR analysis in 19, with imaging at simulation and every fifth fraction.
• Low-field MR-linac imaging showed strong agreement with diagnostic MRI at baseline. Extended resection cavity volumes had a Pearson r of 0.968, while FLAIR volumes had an r of 0.976, both p<0.001.
• Extended resection cavity/tumor volume was highly variable. 59% of patients had a major increase >10%, with a median total increase of 52.9%; 27% had a major decrease, with a median change of -26.5%; only 14% remained within ±10%.
• FLAIR changes were predominantly expansive: 15 of 19 patients (79%) had an increase >10%, with a median total increase of 56.7%. No patient had a major FLAIR volume decrease.
• The extended resection cavity exceeded the original CTV by >1 cc in 6 of 22 patients (27%). FLAIR exceeded the CTV by >1 cc in 14 of 19 (74%) and by >5 cc in 8 of 19 (42%). At the final fraction, 4 of 22 patients (18%) had >1 cc of cavity/tumor extending beyond the original 95% isodose.
• The median interval between simulation and treatment start was 10 days. The authors also observed that all six patients initially classified as RANO 2 after surgery had visible residual tumor on the later simulation imaging, highlighting how anatomy may evolve even before fraction 1.
• No evaluated clinical or molecular factor, including MGMT methylation, resection status, tumor focality, temozolomide, steroids, or baseline target volume, reliably predicted the volumetric trajectory. Early versus later changes were also not statistically different, and volumetric imaging alone could not distinguish tumor progression from edema or treatment-related effects.

CLINICAL TAKEAWAY

SOURCE

Radiotherapy and Oncology