Población y muestra

7.1.1 Impact of ITH on radiosensitivity

The hypothesis that there are differences in intrinsic radioresponse of glioblastoma cells isolated from spatially distinct regions of human tumours was tested by performing radiosensitivity studies and whole exome sequencing on patient-derived cell lines and corresponding tumour fragments. Comparison of Pearson’s correlation coefficients and superFreq analysis of SNV and CNA data demonstrated that patient-derived cell lines were good models of their tumours of origin. In addition, the J14 cell lines were also found to be good models of their specific corresponding tumour fragments. SuperFreq -generated river plots allowed for the comparison of the presence and relative prevalence of defined clones between samples from a given tumour. These plots revealed that while samples shared clones, they also harboured unique clones and combinations of clones which supported the use of cell lines derived from spatially distinct tumour fragments as models of intratumoral heterogeneity.

Cell lines underwent radiosensitivity testing with γH2AX foci analysis and limiting dilution assays. The J7 cell lines displayed conflicting results in DSB foci induction (1-hour post RT) and repair (24 hours post RT). All other cell lines showed no significant differences in foci analysis. For limiting dilution assays, only the J14 cell lines displayed differences in proliferation capacity after irradiation. Because no cell lines displayed consistent differences across radiosensitivity studies, it seems that cell lines derived from spatially distinct tumour regions exhibit similar levels of radiosensitivity in spite of observed ITH. These results suggest that ITH does not produce differences in intrinsic radiosensitivity and that GB radioresponse may be conserved across heterogeneous clones within a tumour.

7.1.2 Radiation drives evolution of GSC-initiated orthotopic xenografts

Glioblastoma stem-like cell-initiated orthotopic xenografts were utilised to test whether radiation alone can drive the evolution of glioblastoma. The fractionated radiotherapy protocol (3x5Gy) provided a significant survival advantage and caused a tumour growth delay for irradiated mice, which mimics the clinical course of treated GB. Histological analysis revealed that control tumour-bearing brains demonstrated different growth patterns when compared to irradiated tumour-bearing brains. In particular, GB cells in control tumours were more diffusely scattered throughout the right hemisphere both anteriorly and posteriorly from the injection site in the striatum. In contrast, irradiated tumours demonstrated a clearly demarcated tumour border anterior to the right striatum. Such clear differences in growth patterns point towards a fundamental change in recurrent biology after irradiation.

We investigated the potential role of clonal selection in this process by performing Viral integration site analysis. VISA demonstrated a significant reduction in the number of unique integration sites when transitioning from an in vitro to an in vivo environment. Clonal diversity was even further reduced when implanted tumours were irradiated. This reduction in clonal diversity was detected in two different GSC lines, suggesting that radiation has the ability to drive evolution. To test whether radiation alone was sufficient to promote clonal selection, we irradiated NSC11 cells in vitro. However, there was no difference in the number of integration sites between control and irradiated cells in vitro. Furthermore, VISA of U251 xenografts demonstrated that while irradiation of subcutaneous xenografts did lead to a reduction in clonal diversity, the largest reduction in diversity occurred after irradiation of intracerebral xenografts.

This suggests that radiation-induced GB evolution is not specific to GSCs and that the brain microenvironment plays an important role in mediating the evolutionary process.

Whole exome sequencing of morbid control and irradiated tumours further supported the observed impact of radiation on GB evolution. COSMIC mutational signature profiles were found to be different between control and irradiated NSC11 and NSC20 tumours. Additionally, EXPANDS analysis demonstrated extensive subpopulation dynamics as many variants shifted into more or less prevalent subpopulations following irradiation. Similarly, patterns of GB driver genes differed between control and irradiated tumours and GB driver gene variants were also found to shift between defined subpopulations after irradiation. The WES results coupled with VISA, histology, and survival support the radiation -driven evolution of GSC-initiated orthotopic xenografts.

7.1.3 Glioblastoma reirradiation model

To test the hypothesis that radiation-driven evolution leads to the emergence of resistant clones, clonogenic assays of xenograft-derived cell lines as well as reimplantation studies were performed. Cells from control and irradiated NSC11 morbid tumours were put back into culture to establish xenograft-derived cell lines grown as neurospheres. These lines were subjected to clonogenic survival analysis which demonstrated no difference in radiosensitivity between the lines. Next, control and irradiated NSC11 xenograft-derived cell lines were reimplanted into the right striatum of nude mice and treated with fractionated irradiation. No differences in survival after irradiation were noted regardless of the prior treatment received by the xenograft-derived cell lines.

An alternative approach for testing the effects of radiation-driven evolution was the addition of a reirradiation protocol to the previously described GSC-initiated orthotopic xenograft model. Tumour bearing mice that received fractionated irradiation (3x5Gy) were followed by serial BLI imaging to monitor for tumour regrowth. Once an increase in BLI suggested regrowth (recurrence), an additional fractionated irradiation protocol (3x5Gy) was applied to half of the mice. Similar to a single course of irradiation, the second course of irradiation led to an observed tumour growth delay by BLI and mice that were reirradiated experienced a significant increase in median survival compared to mice that only received one course of irradiation. Because the GSC xenograft reirradiation protocol better mirrors the

clinical course of GB by retreating recurrent tumours, this model represents a useful method for studying recurrent GB biology and retreatment strategies. Furthermore, these reirradiation results coupled with clonogenic survival analysis and reimplantation results indicate that radiation-induced evolution does not lead to the emergence of radioresistant clones in GSC- initiated orthotopic xenografts.