Multimodal neuroimaging methods for response assessment to treatment, grade characterization and detection of tumor infiltration in gliomas

Adult-type diffuse gliomas are a group of primary brain tumors that can be classified in different subtypes and grades according to molecular and histological characteristics. This classification reflects tumor aggressiveness and prognosis of the patients. To support clinical decision-making and treatment planning, neuroimaging techniques such as Positron Emission Tomography (PET) and Magnetic Resonance Imaging (MRI) play a fundamental role. In fact, from initial diagnosis to follow-up and diagnosis of disease progression or recurrence, several imaging modalities are routinely acquired. Glioblastoma is the most malignant among high-grade gliomas and patient prognosis is very poor despite multimodal treatment strategies including surgical resection and concomitant radiotherapy and chemotherapy. At recurrence the prognosis is even more dismal, and treatment options are very limited. The recently approved regorafenib treatment provides a tool to improve patients’ survival with respect standard treatment with lomustine treatment, but response to the treatment poses a challenge to neuroimaging techniques due to the variety of pathophysiological processes affected at once. Therefore, the present PhD Thesis investigated the ability of a set of PET- and MRI-derived and clinical features to predict the survival of recurrent glioblastoma patients treated with regorafenib. The resulting variables could be used as independent markers to assess the response to regorafenib and to stratify patients that could benefit more from the treatment. Besides clinically acquired sequences, advanced MRI modalities provide additional information that could be useful in many clinical settings. In particular, Quantitative Susceptibility Mapping is a recent MRI post-processing technique that has been poorly investigated in gliomas. It provides information on magnetic susceptibility of the biological tissues that can be composed of paramagnetic (e.g., iron) and diamagnetic (e.g., myelin, lipids and calcium) compounds. The current PhD Thesis compared different algorithmic combinations for the quantification of magnetic susceptibility, suggesting methods that performed better than others when applied to glioma patients. Further, it explored the differences in magnetic susceptibility between low- and high-grade gliomas, also examining radiomics features that are able to capture the complex texture of the images. Given their specific tissue composition and intra-tumoral heterogeneity different glioma grades were successfully differentiated by a set of radiomics features. Finally, glioblastoma is a highly invasive and infiltrative tumor, reason for which both surgical resection and radiation therapy are not able to remove all tumor cells. By decomposing tissue magnetic susceptibility in paramagnetic and diamagnetic sources based on MRI data, this PhD Thesis proposes a novel imaging marker that non-invasively and in-vivo identifies for the first time areas of altered iron metabolism possibly associated to tumor activities in the edema tissue of glioblastoma. The Ratio between Paramagnetic and Diamagnetic susceptibility components (PDR) uniquely delineates the hyperintense susceptibility area corresponding to a hypothesized Tumor and Immune cells Infiltration Zone (TIZ). The defined patient-specific metrics derived from the PDR in the TIZ were significantly associated to prognostic factors. Further, pathological data revealed the presence of tumor cells and macrophages within the TIZ, demonstrating the biological nature of the PDR. This provides first evidence of the potential of PDR, for example to aid and improve surgical and treatment planning as a marker of tumor-associated cells.