Development of Combined PET-MR Imaging Biomarkers in Healthy Volunteers and Glioma Patients

Abstract

Gliomas are primary brain tumours that can lead to brain changes beyond the tumour region as detected by conventional magnetic resonance imaging (MRI). To study these alterations non-invasively requires advanced imaging techniques, with their combination offering additional benefits. Magnetic resonance spectroscopy imaging (MRSI) allows the detection of variations in metabolite concentrations such as N-acetylaspartate (NAA), enabling the assessment of neuronal density. High-affinity SV2A positron emission tomography (PET) ligands such as [11C]UCB-J have been recently developed to measure a proxy of synaptic density. Optimisation of MRSI and PET methodology, independently and jointly, is required to validate the combination of these modalities to introduce a novel metric of synaptic-to-neuronal density ratio. The first dynamic PET study presented in this thesis used radiolabelled water ([15O]H2O) with arterial blood sampling to measure absolute cerebral blood flow (CBF) and assess CBF heterogeneity within the tumour and with distance from the tumour edge. Compared to healthy controls, CBF in white matter was significantly reduced in the ipsilateral and contralateral hemispheres in glioblastoma patients and CBF asymmetry was highest closer to the tumour edge. Prompted by challenges posed by the short early frames and short half-life of 15O yielding noisy dynamic images and by the longer positron range of 15O compromising the HRRT high spatial resolution, we investigated the benefits of complementary frame reconstruction and resolution modelling with isotope-specific point-spread function, which were found to improve the detectability of abnormal CBF in glioblastoma patients. The second dynamic PET study pioneers the development and validation of an image-derived input function (IDIF) for [11C]UCB-J PET in humans to avoid the invasive procedure of arterial blood sampling, leveraging the advantages of a PET-MR scanner through the simultaneous acquisition of MR angiography images to segment the internal carotid artery. Spill-out correction of the carotid time-activity curve up to the first pass extraction was required and a significant correlation between the [11C]UCB-J volume of distribution values obtained from the IDIF and the gold-standard arterial plasma input function was observed. In two separate MRS studies, mixed-effect statistical modelling was used to independently assess regional and tissue-specific differences in metabolite concentration. Metabolite concentrations differed between grey matter and white matter in different ways for each metabolite and varied across the brain. The insights gained from the aforementioned studies were used to derive the normal range and spatial distribution of the synaptic-to-neuronal density ratio using [11C]UCB-J volume of distribution and NAA concentration values. These results serve as a reference for future studies in neurological diseases that cause widespread changes on the brain, such as glioma.

Date of Award	30 May 2024
Original language	English
Awarding Institution	The University of Manchester
Supervisor	Marie-Claude Asselin (Main Supervisor), Laura Parkes (Co Supervisor), David Coope (Co Supervisor) & Julian Matthews (Co Supervisor)