Developing an integrated temporal neuro-oncology dataset
Research type
Research Study
Full title
Integrating clinical, genomic, imaging and pathological data in patients with confirmed or suspected primary and secondary brain tumours over time
IRAS ID
265404
Contact name
Matt Williams
Contact email
Sponsor organisation
Imperial College Healthcare NHS Trust
Duration of Study in the UK
11 years, 4 months, 31 days
Research summary
Brain tumours are a leading cause of cancer death in the under-40’s, and are a CRUK cancer of unmet need and an NIHR focus. Although relatively uncommon, they are associated with disproportionate morbidity and mortality. For WHO Grade IV glioblastoma multiforme, the most common type of primary brain tumour, only one third of patients receive maximal treatment despite which their median survival is 12-15 months. The management of brain tumours is highly dependent on MRI-based imaging, but involves integrated consideration of patient factors, imaging, pathology, tumour genomics and clinical data.
Up to one third of adult cancer patients develop brain metastases from an extracranial primary. As systemic therapy for metastatic disease improves it is increasingly important to focus on local treatment options for brain metastases. Although the development of brain metastases has historically portended a short prognosis, in the era of modern treatment including stereotactic radiosurgery an increasing subset of patients are achieving longer survival. Understanding the factors which lead to more favourable outcomes is therefore of significant interest.
Previous research has drawn upon routine clinical nation-wide datasets to understand national incidence, outcomes and treatment in brain tumours. These national datasets contain coded data on treatment, hospital admission, radiotherapy and chemotherapy. The actual images from these datasets and their correlation with clinicopathological data, however, have not yet been fully exploited due to the inherent difficulty in presenting an image as data in itself. New developments in computational analysis of medical imaging could change this, with deep learning having already been applied to brain tumour MRI images. However, these deep learning methods require large datasets to become more effective, as well as linked clinical and pathological information.
This work therefore aims to bridge that gap: to conduct work on a large, unbiased dataset that allows us to link pathological, clinical, genomic and imaging data, and provide a route to developing and deploying computational tools to improve the care of patients with brain tumours.
REC name
London - Bloomsbury Research Ethics Committee
REC reference
19/LO/1763
Date of REC Opinion
29 Oct 2019
REC opinion
Favourable Opinion