[Oral Communication] Correlation Between MRI-based Hyper-perfused Areas and Tumor Recurrence in High-Grade Gliomas

I. Chabert1,2, I. Belladjou2, F. Poisson2, F. Dhermain2, V. Martin2, S. Ammari2, S. Vauclin3, P. Pineau3, I. Buvat4, E. Deutsch1,2, C. Robert1,2
1 INSERM, U1030 Radiothérapie Moléculaire, Villejuif, France

2 Gustave Roussy, Radiothérapie, Villejuif, France
3 DOSIsoft, Recherche et Développement, Cachan, France
4 CEA, Service Hospitalier Frédéric Joliot, Orsay, France

Presented at ESTRO 2016

Purpose and Objective: Patients suffering from high-grade gliomas currently have a median survival time of 14 months despite treatment. Our purpose was to investigate whether MR perfusion and relative Cerebral Blood Volume (rCBV) maps could predict tumor recurrence areas and improve treatment planning.

Material and Methods: This retrospective study included 19 patients suffering from high grade gliomas (3 and 4) who received standard radiotherapy [60 Gy, 2 Gy/fraction] and
Temodal chemotherapy. Subjects underwent pre-treatment CT, gadolinium-enhanced T1-weighted, T2 FLAIR acquisitions and a DSC-MR scan. rCBV maps were calculated using READE View Advantage Workstation (GE) and normalized to the normal white matter perfusion value. The PLANET software (DOSIsoft) was used to register all MR images to the planning CT. A senior radiologist and a senior radiotherapist delineated Gross Tumor Volumes (GTV) on anatomical MR images. The Planning Target Volumes (PTV) were defined by a physicist. Threshold of 1.7 was applied to the rCBV maps to define hyper-perfused volumes (Vperf). Follow-up anatomical MR images were used to localize recurrence areas (GTV’). Correlations between all volumes were analyzed using several indexes. I1 is the percentage of Vperf not included in the GTV. I2, I3, and I5 are respectively the percentage of GTV’ included in Vperf, GTV, and PTV. I4 is the percentage of Vperf’ not included in the GTV which was predictive of tumor recurrence outside GTV. This index is meaningful only if GTV’ and GVT are different.

Results: Indexes obtained for each patient are presented in Table 1. For two patients, a threshold of 2 was applied to the rCBV maps at the physician request to facilitate the hyperperfused area visualization. I1 values are in a range of 4 to 82% (mean = 43%) and are greater than 20% for almost 90% of the patients, indicating that hyper-perfused areas and GTV can be different. Hence, rCBV maps provide supplementary information. At least 40% of GTV’ is included in Vperf for 16 patients (I2 index). For 10 patients, GTV’ is not completely included in the GTV (I3 < 85%). In all these cases except one, the I4 index is greater than 20%, suggesting that a part of Vperf is predictive of the recurrence localization (Figure 1). I5 being almost always equal to 1 points out that all recurrence areas received the same dose as the GTV.