Introducing advanced MRI to individualize radiotherapy planning of patients with glioblastoma

Abstract

Patients diagnosed with glioblastoma face poor prognosis, as the median survival from initial diagnosis is less than 15 months. Radiotherapy is considered to be one of the fundamentals of glioblastoma therapy and is given after maximum safe tumour resection or biopsy. Delineation of the gross tumour volume (GTV) and the clinical target volume (CTV) required for radiotherapy planning is performed on a combination of computed tomography and conventional,
structural magnetic resonance (MR) imaging, which only visualize macroscopic features of the tumour. Due to the infiltrative nature of glioblastomas, the conventional CTV is typically defined as the GTV with an expansion of 1.5 cm. In this project, the aim was to introduce advanced MR imaging in order to visualize microscopic tumour infiltration and construct a personalized approach for CTV delineation to integrate the information provided by the advanced MR imaging. With this biological CTV, the 1.5 cm margin could be omitted and potentially lead to reduced radiation toxicity or improved local tumour control. The advanced MR biomarkers APT, VSI and rCBV correlate to cell proliferation, vasodilatation and microvascular density, respectively. Brain maps of these biomarkers from four treated glioblastoma patients were introduced in MIM Maestro®, the software package used at the Department of Radiotherapy of the Erasmus MC Cancer Institute for target delineation. In addition, imaging used for radiotherapy, the corresponding delineations, i.e. GTV, CTV and organs at risk, and follow-up imaging that showed first progression according to the response assessment in neuro-oncology criteria were uploaded to MIM Maestro®. To generate a biological CTV, a semi-automatic workflow was created that incorporates a region growing algorithm to delineate high-risk regions on the biomarker maps. Comparison of
biological CTVs with the conventional CTVs showed a major reduction (ranging from 56.7% to 87.6%) in target size. Pattern of failure analysis revealed adequate coverage of the recurrence volume by the biological CTV. In three of the four patients, the biological CTV covered more than 95% of the recurrence volume. In the other patient, the coverage by the biological CTV was poor. This was partially due to the large spread of the recurrence volume. In this master’s thesis, a semi-automatic workflow is presented to take an initial step towards personalized radiotherapy for patients with glioblastoma. The introduction of advanced MR
techniques was shown highly promising; future research should focus on
optimization of the biological CTV generation and validation of the biomarkers in a larger cohort.