- Dr Lucy Stead: Group Leader
- Ms Marilena Elpidorou: Postdoctoral fellow
- Ms Georgette Tanner: PhD student
- Ms Rhiannon Barrow: PhD student
- Ms Muna Al-Jabri: PhD student
- Mr James Manning: MSc research student
- Mr Stelios Theophanous: Leeds Institute of Data Analytics research intern
- Ms Judith Valluru: PhD student (Co-supervisor with Professor David Beech)
We are investigating intra-tumour heterogeneity in glioblastoma and its role in tumour recurrence. To do this we have active research projects under three key areas, which are complementary and mutually beneficial to one another (Fig.1):
Glioblastoma multiforme (GBM) is the most common, most malignant adult brain cancer. Almost 50% of patients die within a year. GBM tumours are surgically removed and patients then receive radio- and chemotherapy but tumours inevitably regrow – on average just 7 months later. To understand this regrowth, we have collected pairs of patient-matched pre-treatment GBM tumour and recurrence following treatment. This is a hugely valuable dataset because recurrent GBM samples are rare: clinicians seldom repeat brain tumour surgeries as there is often no patient benefit. We plan to apply high-throughput sequencing and array approaches to these paid samples to ascertain which cells survived treatment and why. To do this we are developing integrated approaches at the bulk-tissue and single cell level to yield genotypic and phenotypic information.
In silico Modeling
Many methods exist to attempt to deconvolute the cancer cell populations within a tumour but different methods often give different results and there is no gold standard for determining which is giving the ‘ground truth’ i.e. which of these methods work best. Often times, simulated data is used to benchmark computational methods and, in fact, just this year a tumour genome simulation method has been published, aimed at addressing this problem (http://www.ncbi.nlm.nih.gov/pmc/articles/PMC4349766/). However, there is a huge scope for expanding and improving upon such a method to not just provide benchmarking datasets but to model the effects on tumour evolution of:
- Different mutational mechanisms
- Mutations within known driver genes with known phenotypes
- Variations in clonal fitness under different selective pressures
We must also begin to understand ITH at the transcriptional level: there is increasing evidence that cells within the same genomic subclone behave differently, likely owing to heterogeneity at the epigenetic, transcriptional and/or protein level. As yet, no method exists to model transcriptional profiles that alter within and between the genomic subclones of a tumour. This is invaluable for developing methods to delineate such profiles and to compare simulation results against real world data to begin to understand these phenomena.
The aim of this project is to develop an in silico tumour genome and transcriptome simulation tool and use this to a) model the effects of tumour evolution on emergent ITH, and b) benchmark current methods for delineation of ITH.
This model will be used iteratively with the data we get from patient and model-derived sample, first to inform its design and then so that inferences made from the model will be tested within the lab using in vitro and, where possible, in vivo models.
We have developed several models of GBM ITH. Patient derived models are being used to recreate the patient treatment process in vitro and ascertain whether the cells that survive therapy in this context can further inform us about treatment-resistance mechanisms. However, we are also using the models to control confounding aspects of ITH that commonly co-exist within patient samples so we can focus on, and understand, each in turn. An example of this is the use of gliomaspheroids of admixed subclonal cell populations, that fluorescing different colours, to inform upon subclonal interactions.