Phedias Diamandis – 2020 Feature Grant Recipient
Generously Funded by ‘DUNN with Cancer‘, A Movement of Hope in Memory of Allison
Phedias Diamandis – University Health Network/Princess Margaret Cancer Centre
Project Title: “Development of deep learning approaches for deciphering and targeting intra-tumoural heterogeneity in Glioblastoma”
Description of Project:
Glioblastoma is the most common and aggressive form of brain cancer with an expected survival of only ~12-15 months from diagnosis despite spirited surgical/medical therapy. This poor outlook has remained fairly unchanged over the past half century despite many breakthroughs in our understanding of glioblastoma biology.
One recent explanation for these previous failures is that while we often think of a patient’s tumour being a homogenous mass of identical cancer cells, there are in fact a number of tumour sub-clones within each glioblastoma that respond differently to therapy. This means that existing treatments may not equally target all tumour cells and allow resistant subclones to survive and drive disease recurrence/progression.
Approaches that can help the routine characterization and personalized management of these tumour subpopulations, could provide an effective means of better controlling glioblastoma. To address this, Dr. Diamandis’ Lab plans to harness expertise in artificial intelligence to explore if this technology can help automate detection of biologically distinct tumour subregions.
Routine detection and characterization of tumour subclones, within each individual patients’ tumour, could help propose personalized and effective drug combinations that together target a larger fraction of the overall tumour biology. This could ultimately provide more durable responses for patients.
What receiving this award means:
I am humbled and excited that our work was selected for a 2020 Feature Research Grant by Brain Tumour Foundation of Canada. In this project, we couple some of our most exciting technologies we’ve worked on including mass spectrometry-based proteomics and artificial intelligence to develop a platform that could allow doctors and scientists to more routinely assess biological variation found in patients’ glioblastomas.
We expect this tool to help better decipher the regional molecular variation in individual patient tumours and guide the design of advanced combination therapies that aim to target a larger fraction of the tumour’s true biology. We hope this will improve outcomes for this deadly disease and other cancer in general.
This award is also important in helping us grow the brain tumour research community. Each opportunity where we can recruit and train additional young and promising scientists in the areas of brain tumours research, has exponentially positive and lasting effects. They help contribute to making our brain tumour community larger and sustainable in the long term. By creating valuable new insights and qualified personnel in brain tumours, we improve our visibility and competitiveness in other more general research opportunities.
Progress Update: March 2022
While we now understand that each person’s brain tumor is slightly different than anyone else’s, it has recently become clear that individual tumors can also show profound region to region variation within themselves. This “intra-tumoral heterogeneity” can wreak havoc on current diagnosis and treatment strategies due to challenges in sufficiently sampling the cancer’s biodiversity geographically distributed across large tumor samples. To address this, the Diamandis Lab has developed a tool that leverages artificial intelligence to generate “Histomic Atlases of Variation Of Cancers” (HAVOC). Importantly, they demonstrated that this tool can capture cancer variation across large tumor samples. They are now validating this tool on a larger cohort of glioblastomas to highlight its generalizability. Routine generation of these histomic cancer atlases offers an objective means to guide regional deployment of limited molecular resources to the most relevant and biodiverse tumor niches.