Grant: DRA : Canada Graduate Research Scholarship - Doctoral Program (CGRS D)

Glioblastoma (GB) is the most aggressive adult brain tumour. Despite maximal therapy, the tumour typically recurs near its original site within a year, limiting survival to about 12-18 months. Surgeons strive to remove as much tumour as possible but taking extra tissue risks damaging brain regions essential for speech, movement, and memory. A key unmet need is the ability to pinpoint, at the time of surgery, which areas around the tumour are most likely to cause regrowth. This project develops and applies a new strategy to meet that need. An MRI-registered surgical tool links each tissue sample collected to its exact 3D location on the patient's scans and allows tracking of those locations on follow-up MRIs. In parallel, rare whole-lobe specimens enable comprehensive sampling from the tumour core outward into the surrounding brain. These samples are profiled with single-cell and spatial transcriptomic methods to reveal which genes are active in individual cells and how different cell types interact across space. The resulting maps capture how GB invades healthy brain, and which cellular neighbourhoods are associated with early return of disease. Machine learning models then analyze this data to identify patterns such as specific cell types, gene programs, and microenvironments, that best predict recurrence risk at each sampled location. The expected outcome is a set of practical markers and risk maps that can guide care. Surgeons could extend removal or perform focused adjunct treatments only where risk is highest, while radiation plans could be tailored to target high-risk margins and spare low-risk brain. By making surgery and radiotherapy more precise, this approach aims to delay recurrence, reduce side effects, and improve outcomes for people facing this otherwise uniformly fatal cancer.