Geavanceerde Magnetische Resonantie Beeldvorming in de Diagnose en Opvolging van Gliomas

Gliomas are the most common primary brain tumours in adults. They cover a broad range of lesions with distinct differences in malignancy. Glioblastoma are the most aggressive and the most frequent of gliomas. With the current standard of care consisting of surgery and combined radiochemotherapy, the average life expectancy for patients with glioblastoma is 14 months. As such, there is an obvious need for more effective therapies. Among several innovative treatment strategies, dendritic cell immune therapy has emerged as a promising, safe and feasible treatment option for inducing long-term survival in at least a subpopulation of patients with high grade gliomas. The diagnosis and follow-up of patients with gliomas relies mainly on medical imaging and in specific magnetic resonance imaging (MRI). However, routinely used MRI techniques still present many difficulties in clinical practice. Current conventional MRI techniques are insufficient to accurately define glioma grade. However, adequate determination of the aggressiveness of gliomas is of capital importance because treatment regimens and prognosis depend on malignancy grade. Moreover, in the follow-up of patients with glioblastoma, the differentiation between therapy-induced inflammatory reactions and early tumour relapse remains challenging, as the radiological characteristics of both entities are similar. This is a frequently encountered problem in patients treated with the standard of care as well as in patients treated with more innovative therapies such as immune therapy.Currently used conventional MRI techniques do not enable adequate documentation of the nature of the tumoural pathophysiology of gliomas. Advanced MR methods, however, address parameters like diffusion, tissue organisation, blood flow or metabolism, induced by pathological changes. Therefore, advanced MR techniques potentially provide more insights in the problems encountered with the currently used conventional MR techniques as various aspects of tumoural pathophysiology can be demonstrated. Diffusion MRI allows non-invasive mapping of the diffusion process of water molecules in a biological environment and is therefore able to provide information on the organization of biological tissue. In this thesis work, we use diffusion weighted imaging (DWI) and diffusion kurtosis imaging (DKI). DKI is the most recent technique in diffusion MRI and provides information on diffusional heterogeneity and microstructural complexity. Furthermore, we used dynamic susceptibility-weighted imaging, a perfusion technique that measures brain perfusion and provides markers of blood flow and blood volume in healthy and diseased tissue. Finally, chemical shift imaging (CSI) was used to study the metabolism of gliomas. For example, CSI provides markers of neuronal density, cell turn over and energy metabolism. In this thesis work, we explore the potential of these advanced MRI techniques to improve the workup of patients with gliomas in five experiments. In a first experiment, we validated the reproducibility of a 2D - CSI acquisition protocol accelerated with parallel imaging, using automated repositioning software. CSI provides measures relating to the metabolism of the tissue under investigation. The reproducibility was tested on ten healthy volunteers, each imaged three times on two different 3T MRI scanners, receiving anatomical imaging and two identical CSI measurements. The automated repositioning software was tested for consistent repositioning of the CSI grid. This proved to be very robust which is of utmost importance in conducting longitudinal studies. Comparison of CSI data acquisition on the same volunteers using different MR scanners indicated acceptable reproducibility, However, great care should be taken for the interpretation of the results of metabolite signals with an inherent low SNR as the error estimates were still substantial. Moreover, the effect of the MRI scanner on the variability of the results was negligible, which implies that scanners could be interchanged, a pre-requisite for inter-institutional studies. As such, in this first experiment, we proposed a CSI protocol with good reproducibility, applicable in clinical routine. The proposed protocol was afterwards implemented in two clinical studies in this thesis work. In a second experiment of this thesis, we assessed the diagnostic accuracy of a novel diffusion imaging technique, diffusion kurtosis imaging (DKI), in determining glioma grade. We found that kurtosis parameters better separated between low and high grade gliomas, compared to conventional diffusion metrics. Although the outcome of this study was promising, results were demonstrated on a group level. Nonetheless, to get acceptance in clinical setting, diagnosis should be made for the individual patient. Therefore, in a third study, in order to increase diagnostic performance, we combined DKI with dynamic susceptibility-weighted MRI (DSC-MRI) and CSI, as established in the first experiment, in a multiparametric approach to assess glioma grade. By integrating these three advanced MRI techniques in a decision tree algorithm, diagnostic accuracy could be increased up to individual patient level. The best performing modality, when considering the techniques separately, was DSC-MRI. These results are considerably beneficial for patients with gliomas because using this algorithm, an accurate diagnosis can be made before any invasive procedure is performed. An early diagnosis allows optimizing the therapeutic strategy in view of the patient's individual needs. The last part of this thesis work focused on the follow-up of glioblastoma patients treated with dendritic cell immune therapy. The two experiments presented in this thesis work are, to our knowledge, the first studies to explore immune therapy-induced inflammatory responses with MRI. In a first small, retrospective study, we reviewed the MR imaging follow-up in patients with recurrent GBM, treated with immune therapy as single treatment approach. We focused on the available DSC-MRIand DWI exams obtained in routine clinical setting, and found that mainly perfusion measures and to a lesser extent diffusion measures, were potential markers to differentiate between immune therapy inflammatory responses and tumour relapse.These findings were further explored in a more elaborate prospective, longitudinal imaging follow-up study. Patients with relapsed glioblastoma treated with immune therapy as the single approach as well as patients with primary diagnosis of GBM receiving the standard treatment strategy combined with immune therapy were imaged with an extensive imaging protocol, consisting of DSC-MRI, DKI and CSI. The aims of this study were threefold: (i) documentation of the time course of immune therapy-induced inflammatory changes versus tumour relapse on conventional imaging, (ii) characterization of the vaccine-induced inflammatory immune responses and tumour relapse using advanced MRI techniques and (iii) defining prognostic biomarkers. The first results of this ongoing study indicate that patients with relapsed glioblastoma display distinct imaging patterns compared to patients with the primary diagnosis of glioblastoma receiving immune therapy. Furthermore, we were able to confirm the results of the retrospective study. Perfusion related measures were able to differentiate inflammatory reactions from tumour relapse and possibly provide clues for the patient s prognosis. Also kurtosis related parameters seemed promising. As such, these results bring further insights in the working mechanism of immune therapy and can help guide clinical trials. Furthermore, treatment strategy can be improved and the patient can be informed more accurately on his or her disease status. Although the results of this thesis contribute to establish the role of advanced MRI techniques in the characterization of the pathophysiology of gliomas, future research and further standardization is warranted to gain acceptance for these advanced MRI techniques as clinically valid and applicable techniques.status: publishe