Clinical and surgical validation of advanced neuroimaging for enhanced assessment of brain tumors.

Advanced neuroimaging techniques have been introduced in the last years to overcome the limits of MR and CT modalities. Preliminary studies showed extremely promising results for modern methods of perfusion (PWI), diffusion (dMRI), blood-oxygen-level-dependent (BOLD), and spectroscopic MR acquisition, as well as for PET with a multitude of innovative radiotracers. However, many uncertainties still abound for most of these techniques, preventing their systematic introduction as a standard for clinical practice, and further validation in the setting of brain tumors is required. The general aim of this thesis was to validate their application in the assessment and surgical management of brain tumors, specifically in terms of: (1) characterization of tumor biological heterogeneity and ability to guide the neurosurgeon in a more accurate bioptic sampling (2) increase of the rate of maximum safe resection, patient’s quality of life, and eventual survival. In the first experiment, we presented an innovative PET/MRI approach for the evaluation of Hypoxia (from 18F-FAZA-PET), vascularization (from PWI), and cellularity (dMRI) in high-grade gliomas to derive combined habitats representing intra-tumor biologic heterogeneity. Our preliminary analyses in 17 patients confirmed that these maps may reflect tumor heterogeneity non-invasively and can consistently divide malignant gliomas into some small number of clusters. We were able to identify the unique corresponding morphological and physiological MR characteristics and histopathologic correlate of each cluster, observing an outstanding reproducibility in pattern distribution among different cases and allowing us to even speculate about the possible evolution over the time of one habitat into each other. In the second experiment, we investigated “VERDICT” MR imaging, a pioneering multi-compartment model developed to extract quantitative maps from advanced dMRI acquisitions in body tumors, representing their vascular, extracellular and restricted component. We optimized this model for the assessment of brain neoplasms, eventually obtaining a four-compartment model without extra free parameters, showing its improved fitting performance, especially for peritumoral areas. The results from our VERDICT model were validated against PWI parameters and histopathology from corresponding stereotactic sampling, noting a good correlation. We observed higher intracellular and vascular fractions and lower extracellular fractions in more malignant histologies and tumor areas, an extremely high free water fraction in radionecrosis, and a trend towards higher free water fraction and extracellular diffusivity in purely vasogenic edemas. In the third experiment, we investigated the clinical application of advanced DTI tractography and BOLD functional MRI (fMRI), as well as the generation of elaborated 3D anatomic-functional-pathological models, in the pre- and intraoperative management of 234 brain gliomas undergoing surgical resection. The role of modern modalities of image-guided surgery, as neuronavigation and ultrasound, was also investigated. All these imaging techniques generally proved advantageous in terms of extent of resection and functional outcome. In particular, DTI and fMRI were found significantly correlated with a longer survival at multivariate analysis; the former was also significantly associated with lower postoperative morbidity and a higher neurological improvement. Conclusively, we presented, for the first time, innovative approaches for enhanced brain tumor characterization and evaluated DTI and fMRI techniques, together with pioneering anatomic-functional-pathological 3D models, in a large surgical series. We demonstrated that advanced neuroimaging techniques could represent a formidable tool to increase maximum safe resection and survival and to assess tumor heterogeneity, histopathologic/molecular characteristics, and expected behavior.