Flash Talk and Poster Presentation 36th Lorne Cancer Conference 2024

Deciphering the response to temozolomide in heterogeneous glioblastoma populations (#202)

Zachery Moore 1 2 , Montana Spiteri 1 2 , Daniel Brown 1 2 , Piper O'Keeffe 1 2 , Adam Valkovic 1 2 , Alana Fakhri 3 , Kate Drummond 2 4 , Sarah A. Best 1 2 , Saskia Freytag 1 2 , Jim R. Whittle 1 2 4 5
  1. Walter and Eliza Hall Institute of Medical Research, Parkville, Victoria, Australia
  2. Department of Medical Biology, University of Melbourne, Melbourne, Victoria, Australia
  3. Department of Pathology, The Royal Melbourne Hospital, Melbourne, Victoria, Australia
  4. Department of Neurosurgery, The Royal Melbourne Hospital, Melbourne, Victoria, Australia
  5. Department of Medical Oncology, Peter MacCallum Cancer Centre, Melbourne, Victoria, Australia

Glioblastoma (GBM) remains both the most common and deadliest form of brain cancer with increased tumour heterogeneity associated with poorer survival. Current research views this heterogeneity as a spectrum of tumour states that parallel early neurodevelopmental lineages. Despite these advances, the clinical management of GBM remains unchanged with a multimodal approach of surgical resection, radiotherapy, and use of the sole chemotherapeutic temozolomide (TMZ). Widely-used in vitro 2D cell line models of GBM are often homogeneous, lacking the extensive heterogeneity seen in patients. These limitations often mask true biological effects of interest, which includes drug responses. Neurospheres are self-aggregating 3D cultures of brain cancer cells generated from surgical resections, and in contrast to traditional 2D models, recapitulate heterogeneity seen in patient tumours. Here, we utilise TMZ resistant and sensitive neurosphere models to decipher how heterogenous populations are altered in response to TMZ. Sensitivities to TMZ were first determined via multiplexed live-cell imaging with a custom machine learning segmentation model across an 8-point dose-response curve. For our sensitive neurosphere line, we integrated matched bulk RNAseq to imaging data across all concentrations at 3 timepoints. Using these data, we investigated the temporal relationship between alterations in transcription and decreases in cellular growth. Using deconvolution methods to infer cellular composition across samples, we demonstrate a decrease in the number of tumour subtypes with increases in TMZ concentration. In contrast, we interrogated our resistant line via single-cell RNAseq at a single high dose. Though classified as resistant via imaging analysis, we observe changes in cellular heterogeneity, with increases in astrocytic-like and mesenchymal-like states in treated neurospheres when compared to vehicle controls. This indicates that TMZ is still able to elicit effects on sensitive cells within a larger heterogenous and overall resistant population. Our data demonstrate that neurospheres provide adequate heterogeneity to investigate altered drug sensitivities. Crucially, we have uncovered that even in classically resistant cells, different tumour subtypes have altered sensitivity, unveiling new insights into the breadth of TMZ responses.