Pre-irradiation tumour volumes defined by MRI and dual time-point FET-PET for the prediction of glioblastoma multiforme recurrence: A prospective study.

Abstract

The diagnostic accuracy of magnetic resonance imaging (MRI) for glioblastoma multiforme (GBM) is suboptimal. We analysed pre-treatment MRI- and dual time-point 18F-fluoroethylthyrosine-PET (FET-PET)-based target volumes and GBM recurrence patterns following radiotherapy with temozolomide. Thirty-four patients with primary GBM were treated according to MRI-based treatment volumes (GTVRM). Patients underwent dual time-point FET-PET scans prior to treatment, and biological tumour volumes (GTVPET) were contoured but not used for target definition. Progressions were classified based on location of primary GTVs. Volume and uniformity of MRI- vs. FET-PET/CT-derived GTVs and progression patterns assessed by MRI were analysed. FET-based GTVs measured 10min after radionuclide injection (a.r.i.; median 37.3cm(3)) were larger than GTVs measured 60min a.r.i. (median 27.7cm(3)). GTVPET volumes were significantly larger than corresponding MRI-based GTVs. MRI and PET concordance for the identification of glioblastoma GTVs was poor (mean uniformity index 0.4). 74% of failures were inside primary GTVPET volumes, with no solitary progressions inside the MRI-defined margin +20mm but outside the GTVPET detected. The size and geometry of GTVs differed in the majority of patients. The GTVPET volume depends on time after radionuclide injection. FET-PET better defined failure site than MRI alone.