Magnetic resonance spectroscopy in the diagnosis and management of brain tumours

Introduction Clinical Magnetic Resonance Spectroscopy (MRS) is developing as a non-invasive adjunct to conventional Magnetic Resonance Imaging (MRI). It measures concentrations of metabolites within a selected area (voxel). The investigator chooses the region for examination at the time of scanning by placing a voxel C3-dimensional cube-shaped area of interest) over it. Brain tumour biopsy carries a risk of post-operative haematoma, stroke and even death. It is therefore advisable that all possible non-invasive diagnostic investigations be exhausted before resorting to a biopsy especially in cases where surgery may otherwise be avoided. Spectroscopy may also be clinically useful post-operatively in the context of a patient being monitored for disease recurrence when the conventional imaging (MRI and CT) is equivocal. Methods and Materials Two hundred and eighty four patients, in all, at St George's University, London (SGUL) were studied. (One hundred and twenty five of these had low grade tumours and were studied from the perspectives of outcome of aggressive surgical management and post-operative surveillance 2D-CSI, Chapters 6 and 7). One hundred and fifty nine patients with intracranial space-occupying lesions had brain spectroscopy. Spectra of benign entities are included in the dataset but the majority of spectra are from tumours. The work undertaken pursued 3 main avenues ... 1. Spectral characterisation of common intracranial space-occupying lesions, using single voxel proton spectroscopy (lH-MRS), attempting to improve non-invasive tumour diagnosis (work conducted as part of INTERPRET, a European Union (EU)-funded, multicentre study). The "added value" of SV IH-:MRS in the context of diagnostic magnetic resonance studies is addressed in Chapter 4. 2. Delineation of difficulties of managing low grade intrinsic brain tumours and an attempt to assess the effect of maximal surgical resection on survival and clinical outcome (Chapters 5 and 6). 3. Surveillance imaging of individuals with low grade intrinsic intracranial lesions using 2D-CSI, in an effort to pre-emptively detect a change in tumour grade prior to a perceptible change on conventional imaging sequences (Chapter 6). While the main thrust of this work concerns IH-MRS of brain tumours, J have also included work detailing complications of different management strategies for low grade tumours and survival of patients with low grade tumours for whom the primary treatments were different (Chapters 6 and 7). One hundred and fifty four patients underwent early, radical tumour resection in SOUL. Their outcome is compared to that for a group whose primary treatment modality was radiotherapy. A niche for two dimensional chemical shift imaging (2D-CSJ) surveillance in the management of these tumours is also described. Results The current "gold-standard" for diagnosis of intracranial mass lesions is neurohistopathology. Chapter 3 details a study of the accuracy of neuropathological diagnosis. Neuropathologists at different European sites were asked to make a diagnosis based on some clinical information and histopathological evaluation. They failed to agree the exact histopathological diagnosis in 22.9% of 175 cases. The discrepancy would have had implications for treatment in 15% of cases. This illustrates that neurohistopathology has a subjective component and although it is the current gold standard it may not be as accurate as we assume. This therefore defines a potential niche for a non-invasive technique such as lH-MRS, complimenting conventional, clinical MRI in diagnosis. To enumerate this potential clinical role for single voxel 1H-MRS of intracranial mass lesions as an adjunct to conventional MRl sequences one hundred consecutive patients evaluated with MR1 and lH-MRS were studied. In 26 cases the pre-operative radiological diagnosis was inaccurate. The inaccuracy related to lesion grade in 17 of these and to lesion type in 9 cases. Taking the spectral data into consideration would have improved the pre-operative diagnosis in 6 (6%) cases and would have had an "added value" of 5.6%. It is likely that the number of potential beneficiaries will be greater in the future when we have more spectral data from unusual pathologies. Some cases illustrate that diagnosis of benign pathologies may also he improved with lH-MRS used as an adjuctively with MRI. Low grade intrinsic brain tumours (LGIBTs) can be the most challenging intracranial tumours to manage. Optimising intra-operative conditions to protect the neurological welfare of the patient (using the technique of "awake" craniotomy with multiple marginal biopsies) survival at 5 and ]0 years is good (85.9% and 65.1% respectively). Survival is improved in cases where a "complete" resection (defined by negative marginal biopsies) is achieved. This patient group benefits from surveillance imaging to monitor disease status. Conventionally, FLAIR and TIWC sequences are used. The former is too sensitive and the later too insensitive to predict presence or otherwise of residual/recurrent tumour. Two-dimensional CSI may be used as an adjunct to improve prediction of resection cavity tumour status with maximum Choline:creatine ratios of less than 1.5 suggesting the absence of tumour and a ratio of greater than 1.8 predicting the presence of tumour. Conclusions The work performed shows that lH-MRS can play a valuable diagnostic and follow-up role in management of brain tumours. A potential clinical role for spectroscopy is well demonstrated in the context of low grade tumours, where long-term surveillance is important. Some demographic studies suggest that low grade lesions are best managed by primary resection, by "awake" craniotomy (if the tumour is located adjacent to eloquent cortex) using the technique of multiple marginal biopsies. Follow-up of these lesions with regular 2D-CSJ to characterise the tumour resection bed is yielding promising results. The hurdles to clinical uptake of spectroscopy can only be overcome by improving the technique accuracy and usability. Optimisation of pulse sequences to derive the maximum, pertinent, uncontaminated data for any given subject, including improvement of acquisition times is important to improving the clinical applicability of MRS. Additionally, an appreciation of the limitations of MRS and the use of various spectroscopy modalities to various pathologies and interpretation of the data is central to understanding the potential clinical role of MRS.EThOS - Electronic Theses Online ServiceGBUnited Kingdo