Abstract
One of the major challenges in successfully treating glioblastoma (GBM) is the significant heterogeneity in cellular composition observed within and between patients. Recent single cell transcriptomics suggests there can be as many as eighteen distinct cell types in a single tumour (Al-Dalahmah et al. 2021). Furthermore, advances in cellular deconvolution techniques, such as CIBERSORTx, allow us to accurately determine the cellular composition of imaged localised biopsies from bulk RNA-Seq (Steen et al. 2020). Understanding this heterogeneity and how the complex interactions between cellular populations impacts the progression of GBM may lead to novel treatments which exploit the unique cellular composition within individual tumours. We group these eighteen cell types into sub-populations, e.g., glioma, immune, astrocyte, then attempt to learn the dynamics of these sub-populations by considering various interacting ODE/PDE models. Typically, a GBM patient will have biopsies taken at most twice, as well as only a handful of MRI scans. Therefore, the number of temporal data points to fit any model to are very limited. Thus, we apply trajectory inference methods, such as Monocle, to biopsy data, which allows us to order samples via pseudotime, an arbitrary unit of progress akin to real time (Trapnell et al. 2014). We illustrate our modelling approach with a simplified two species Lotka-Volterra style competition model.
Poster
References
Citation
@misc{harbour2023,
author = {Harbour, Nicholas and Curtin, Lee and Velez, Sebastian and
Chappell, Michael and Hubbard, Matthew and Al-Dalahmah, Osama and
Canoll, Peter and Swanson, Kristin and Owen, Markus},
title = {Mathematical Modelling of Cell Cycle Dynamics in Glioblastoma
Subpopulations},
date = {2023-07-16},
url = {https://2023.smb.org/ONCO/PS01-ONCO-12.html},
langid = {en}
}