Academic research community (GECIP)

Glioblastoma is a devastating brain cancer which affects over 2,000 people per year in England. Despite aggressive treatments with surgery, radiotherapy and chemotherapy most patients do not survive more than 15 months after diagnosis. There is an urgent need to find more effective treatments. The Sheffield Translational Brain Tumour Research Group is developing a number of new ways to tackle glioblastoma. The research group’s work includes: 1. Looking for new treatments for patients through tumour testing. We can monitor how tumour samples from surgery react to treatments using microscopes in the laboratory. This will help us to find the medications that work best. 2. Strengthening our understanding of how potential treatments will work. We can match our results with the genetic information of patients with the disease to reveal when a specific treatment may work. 3. Unveiling what keeps glioblastoma cancer cells growing and how they repair. This can help us to find treatments to stop tumours coming back again after chemotherapy and surgery. This work will improve our understanding of glioblastoma biology, while access to data from Genomics England’s 100,000 Genomes Project will allow us to be more confident about our results and their potential to improve treatment for patients. This research could help to uncover low-cost medications that have not previously been tested for this cancer. Sadly, many glioblastoma patients are not able to take part in clinical trials for new treatments that could possibly help them or future patients. This project can help to increase the availability of clinical trials looking for treatments for patients with glioblastoma and help to find more effective future treatments.

Glioma is a devastating disease. Rarely some families have a genetic predisposition to developing glioma. We are looking to identify new genes that can cause familial glioma by analyzing large datasets.

Among the different types of cancer, Glioblastoma Multiforme (GBM) is the most common primary malignant brain tumor, and despite the availability of chemotherapy and radiotherapy to combat the disease, overall survival remains low with a high incidence of tumor recurrence. GBM accounts for approximately 65% of all primary brain tumors and is characterized by low survival, with only 10% of patients surviving 5?years. (Ellis et al., 2015) Although GBM is a rare tumor with global incidence of less than 10 per 100,000 people, its poor prognosis makes it a crucial public health issue. (Hanif et al., 2017) Unfortunately, being able to identify relevant etiological factors or potential drug targets is hindered by its low prevalence. (Amirian et al., 2015) Common methodological challenges among glioma studies include small sample sizes, heterogeneity of tumor subtypes, and retrospective exposure assessment. Taking into account the limitations of GBM research and the advantages of multi-modal network analyses for sub-phenotyping cancer patients, in this study we will use similarity network fusion (SNF) to create the patient similarity networks of the available GBM patients inside the Genomics England (GeL) dataset. To be able to describe and evaluate the heterogeneity and similarity (Chiu et al., 2018) among these GBM patients we are going to pre-process appropriately, analyse and fuse: i) their germline and somatic variants data; ii) their cancer treatment related data; iii) the relevant HES data. The constructed fused networks, by the aforementioned layers of data, will be further analysed with clustering to identify subtypes among existing samples and predict labels for future, new samples. Our motivation behind this analysis is to recognise specific GBM subtypes that could be effective for a better prediction of glioblastoma patients’ survival. An additional goal will be to demonstrate significant ML analytical processes (e.g., handling of missing and imbalanced data, creation of synthetic data), during the pre-processing and validation stage, that will be able to support the extrapolation of our findings to the general population of GBM patients.

This study focuses on glioblastoma (GBM) which is a type of cancer affecting the brain. It is one of the most aggressive cancers with devastating effects on quality of life and very poor survival despite maximal treatment. Researchers aim to find new ways of detecting GBM and monitoring the response to treatment whilst avoiding invasive and risky procedures such as surgical biopsies. Living cells, including cancer cells, release products that enable them to interact and ‘communicate’ with each other. These products are contained in microscopic spherical structures called extracellular vesicles (EVs) and can be found in blood and other fluids. Within the EVs there is a variety of molecules such as genetic material (DNA, RNA), proteins, lipids and products of cellular metabolism. There is a potential for some of these molecules to be used as biomarkers (evidence of disease) in GBM. We will study the relationship between the characteristics of the tumour cells and the content of the secreted vesicles that were isolated from blood.

Cancers result from an evolutionary process, akin to the Darwinian process driving the evolution of species. In this project we aim to leverage on mathematical techniques we have recently developed to measure this process in human cancers. This will allow us to make sense to the data in light of tumour evolution, but also to potentially identify new prognostic biomarkers that will be useful for patient stratification.

By looking at how many variants there are in different regions of the genome in cancer cells it is possible to estimate how the DNA is packed inside the nucleus of cancer cells. This is very informative for tumour biology because unpacked DNA indicate that nearby genes may be active, whereas packed DNA correlates with inactive genes. We will use this to estimate what regions of the cancer genome are active and which ones are most likely inactive.

So-called ‘negative’ or ‘stabilising’ selection removes cancer cells that are unfit either because they acquire deleterious variants in genes that are necessary for the cancer to grow, or because they have acquired variants that make them visible by the immune system. We can measure the extent at which these cancer cells are removed, allowing for the identification of potentially new therapeutic targets as well as predictive and prognostic biomarkers in gliomas.

We would like to generate additional data, such as gene expression, from glioma tumours sequenced by Genomics England. In combination with the sequence data from these tumours this will enable us to understand more about how gliomas progress, and identify potential markers for prognosis and response to treatment.

The incidence of glioblastoma (the most malignant form of brain tumour) is rising in the elderly population. However, these patients generally have a very poor prognosis due to reasons that are not well understood and are commonly underrepresented in clinical trials. There is not much information on the genetic profile of glioblastoma in the elderly (over the age of 70 years). We would aim to identify variants in the glioblastoma sample that are specific/new to this particular population. Furthermore, there are currently two clinical trials ongoing (HCQ and PARADIGM) that could potentially co-register patients and scientific insight could be combined.

The mainstay of treatment for high grade glioma (an aggressive form of brain tumour) patients is surgical resection followed by a combination of chemotherapy and radiotherapy treatment. It is known that patients with a glioma may respond very differently to these treatments, varying from good response to a very poor response depending on the molecular profile of these patients' tumours. It is also known that the tumour profile is altered during these treatments. The variation in response to post-surgical chemotherapy and radiotherapy treatments strongly suggests that a better understanding of how these therapies work has the potential to improve patient selection and ultimately the therapy itself. We also want to look at the evolution of these tumours and the exact alterations to the genome so as to help with developing rescue therapies/drugs.

We will use state-of-the-art approaches to identify new groupings and pathways in glioma tumours on the basis of shared gene variants as well as copy number gains and losses. These findings will hopefully lead to improvements in classification and treatment of gliomas.

Brain cancers are sometimes driven by specific tumour genes that, instead of being present in only two copies (mother and father), they are present in hundreds of copies inside cancer cells. Peculiarly, these fragments of DNA containing cancer genes, are not embedded inside chromosomes, as it would be normal, but rather they are floating in the nucleus as extrachromosomal DNA. As this represents a possible therapeutic target in glioma, we will study this phenomenon using a combination of bioinformatics and mathematical modelling methods.

Due to recent advances in treatment strategies for patients with a glioblastoma (malignant brain tumour) several new challenges have arisen. One of these is that during post-operative follow-up imaging with regular interval magnetic resonance imaging the radiologist is faced with a new dilemma: due to the new treatment with chemotherapy/radiotherapy it can be very hard to distinguish between patients whose tumours are growing back/growing further and patients whose tumours have undergone radiotherapy-induced cell-death. These changes on magnetic resonance imaging may look very much the same and the difference can be difficult to ascertain even for the very experienced clinician. With this project we would look to identify genetic variants at the time of patient presentation that may predict subsequent change on magnetic resonance imaging that would seem like progression but is not. This would inform the clinical team on whether a change in treatment regime would be required at an early stage or not.

In some patients with a low grade (benign) brain tumour, the tumour can transform over time to a more aggressive high grade brain tumour. These changes can be detected on normal magnetic resonance imaging of the brain. With more advanced magnetic resonance imaging techniques using (the blood flow in the brain and detecting the flow of water molecules) these changes may be captured at an earlier time point. The aim of this project is to determine whether a genetic marker can predict and be correlated with a time point of transformation from low grade to high grade tumours on normal magnetic resonance imaging of the brain. And furthermore, whether a genetic marker can predict an earlier time point if transformation on advanced magnetic resonance imaging.

Gliomas are a heterogenous entity whose classification has changed with the knowledge of a new molecular classification. Data generated through radiological imaging will likely reveal certain imaging characteristics that can be correlated with the data generated through genetic profiling. This may help with the understanding of the underlying biology of those tumours, and help in the future with gaining a diagnosis, prognosis and planning of treatment.

We plan to further investigate initial findings from analysis of glioma tumours in GeL by comparing with additional sources of data as well as by carrying out functional experiments in tumour/brain cells. This will hopefully increase our understanding into the effects of these findings on glioma development, and potentially suggest new targets for treatment.

There is an inherited component to glioma risk, as shown by relatives of glioma patients having approximately two-fold increased risk compared with the general population. We would like to analyse both normal and tumour DNA sequence from ~300 glioma patients in order to identify new genetic risk factors for glioma development as well as better understand how both inherited and cancer variants contribute to disease progression. This will hopefully increase our understanding of glioma causation and potentially suggest new therapeutic opportunities.

The cohort of patients with glioma can be heterogenous. We would like to identify outliers in terms of survival after initial diagnosis, age at diagnosis (especially young patients), and try to correlate these findings with a particular genetic profile.

Gliomas with a certain genetic mutational profile may respond better or worse to different treatments used in clinical trials. For patients that are participating in Genomics England and are currently enrolled in a clinical trial the results of the genetic analysis as part of Genomics England may alter their presumed treatment response significantly. Therefore we will integrate the obtained data as part of Genomics England into clinical trials data.

We plan to search glioma genome sequence for presence of viral genomes that are not present in genomes from normal tissue. This will hopefully improve our understanding into whether some gliomas can be caused by viral infection, which could suggest potential therapeutic strategies.