An Individualised Medicine Approach to Improve Survival for Patients Diagnosed with Glioblastoma

Home , Talampanel

Toni Rose Jue, DVM

A thesis in fulfilment of the requirements for the degree of

Doctor of Philosophy

Cure Brain Cancer Foundation

Biomarkers and Translational Research Group

Prince of Wales Clinical School

Faculty of Medicine

University of New South Wales

May 2018

PLEASE TYPE THE UNIVERSITY OF NEW SOUTH WALES Thesis/Dissertation Sheet

Surname or Family name: Jue

First name: Toni Rose Other name/s: not applicable

Abbreviation for degree as given in the University calendar: PhD

School: Prince of Wales Clinical School Faculty: Faculty of Medicine

Title: An individualised medicine approach to improve survival for patients diagnosed with glioblastoma

Abstract 350 words maximum: (PLEASE TYPE)

Glioblastoma (GBM) is a highly aggressive primary brain tumour occurring in adults and children. The current standard treatment is maximal tumour resection followed by radiotherapy with concurrent and adjuvant temozolomide (TMZ) chemotherapy. Currently, median survival remains at approximately 15 months and less than 5% of patients survive longer than 5 years. Patients harbouring methylation within the MGMT promoter hold a survival advantage. Unfortunately, the majority of GBM patients (~60%) do not harbour a methylated MGMT promoter and treatment options available for these patients are very limited. Hence, there is an urgent need to develop novel therapeutic approaches focused on this subset of patients.

The overarching aim of this thesis is to undertake a personalised therapeutic approach through precision medicine by identifying targetable molecular markers within a patient’s whole genomic profile. A patient-derived model (G89), for both in vitro and in vivo investigations, was developed and characterised. Three molecular targets were identified using a commercial biotargeting system and in-house whole genome sequencing. These targets were topoisomerase I, mammalian target of rapamycin (mTOR) and poly ADP ribose polymerase (PARP). Inhibition of these targets were investigated in vitro and in vivo on G89 along with other GBM patient-derived models. In addition, mTOR and PARP inhibition was investigated in vitro and in vivo in combination with an organo-arsenic mitochondrial inhibitor (PENAO) and radiotherapy, respectively. Variable responses were observed. Treatment with topoisomerase inhibitors elicited poor response. Promising responses with the combination of a mTOR inhibitor with the mitochondrial inhibitor, PENAO, were observed in vitro (p < 0.05) but did not show appreciable effects in vivo. PARP inhibition in combination with radiotherapy, on the other hand, showed remarkable response both in vitro (p < 0.001) and in vivo (Log-rank p value = 0.042).

Overall, the attempt to carry out a personalised treatment approach led us to identify and develop novel therapeutic combinations. The findings of this thesis have led to a successful novel therapeutic combination for GBM patients with unmethylated MGMT promoter that is currently being tested in a clinical trial.

Declaration relating to disposition of project thesis/dissertation

I hereby grant to the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or in part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all property rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation.

I also authorise University Microfilms to use the 350-word abstract of my thesis in Dissertation Abstracts International (this is applicable to doctoral theses only).

…………………………………………………………… ……………………………………..……………… ……….……………………...…….… Signature Witness Signature Date

The University recognises that there may be exceptional circumstances requiring restrictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be made in writing. Requests for a longer period of restriction may be considered in exceptional circumstances and require the approval of the Dean of Graduate Research.

FOR OFFICE USE ONLY Date of completion of requirements for Award:

ORIGINALITY STATEMENT

‘I hereby declare that this submission is my own work and to the best of my knowledge it contains no materials previously published or written by another person, or substantial proportions of material which have been accepted for the award of any other degree or diploma at UNSW or any other educational institution, except where due acknowledgement is made in the thesis. Any contribution made to the research by others, with whom I have worked at UNSW or elsewhere, is explicitly acknowledged in the thesis. I also declare that the intellectual content of this thesis is the product of my own work, except to the extent that assistance from others in the project's design and conception or in style, presentation and linguistic expression is acknowledged.’

Signed ……………………………………………......

Date ……………………………………………......

‘I hereby grant the University of New South Wales or its agents the right to archive and to make available my thesis or dissertation in whole or part in the University libraries in all forms of media, now or here after known, subject to the provisions of the Copyright Act 1968. I retain all proprietary rights, such as patent rights. I also retain the right to use in future works (such as articles or books) all or part of this thesis or dissertation. I also authorise University Microfilms to use the 350-word abstract of my thesis in Dissertation Abstract International (this is applicable to doctoral theses only). I have either used no substantial portions of copyright material in my thesis or I have obtained permission to use copyright material; where permission has not been granted I have applied/will apply for a partial restriction of the digital copy of my thesis or dissertation.'

Signed ……………………………………………......

Date ……………………………………………......

AUTHENTICITY STATEMENT

‘I certify that the Library deposit digital copy is a direct equivalent of the final officially approved version of my thesis. No emendation of content has occurred and if there are any minor variations in formatting, they are the result of the conversion to digital format.’

Signed ……………………………………………......

Date ……………………………………………......

Abstract

Currently, median survival remains at approximately 15 months and less than 5% of patients survive longer than 5 years. Patients harbouring methylation within the MGMT promoter hold a survival advantage. Unfortunately, the majority of GBM patients (~60%) do not harbour a methylated MGMT promoter and treatment options available for these patients are very limited. Hence, there is an urgent need to develop novel therapeutic approaches focused on this subset of patients.

The overarching aim of this thesis is to undertake a personalised therapeutic approach through precision medicine by identifying targetable molecular markers within a patient’s whole genomic profile. A patient-derived model (G89), for both in vitro and in vivo investigations, was developed and characterised. Three molecular targets were identified using a commercial biotargeting system and in-house whole genome sequencing. These targets were topoisomerase I, mammalian target of rapamycin (mTOR) and poly ADP ribose polymerase (PARP). Inhibition of these targets were investigated in vitro and in vivo on G89 along with other GBM patient-derived models. In addition, mTOR and PARP inhibition were investigated in vitro and in vivo in combination with an organo-arsenic mitochondrial inhibitor (PENAO) and radiotherapy, respectively. Variable responses were observed. Treatment with topoisomerase inhibitors elicited poor response.

Promising responses with the combination of an mTOR inhibitor with the mitochondrial

iii

inhibitor, PENAO, were observed in vitro (p < 0.05) but did not show appreciable effects in vivo. PARP inhibition in combination with radiotherapy, on the other hand, showed remarkable response both in vitro (p < 0.001) and in vivo (Log-rank p value = 0.042).

Publications

1. Jue, T. R., Hovey, E., Davis, S., Carleton, O., & McDonald, K. L. Incorporation of biomarkers in phase II studies of recurrent glioblastoma. Tumor Biology. 2015; 36(1), 153-162.

2. Jue, T. R., & McDonald, K. L. The challenges associated with molecular targeted therapies for glioblastoma. Journal of neuro-oncology. 2016; 127(3), 427-434.

3. Jue, T.R., Nozue, K., Lester, A., Joshi, S., Schroeder, L., Whittaker, S., Nixdorf, S., Rapkins, R., Khasraw, M. and McDonald, K.L. Veliparib in combination with radiotherapy for the treatment of MGMT unmethylated glioblastoma. Journal of translational medicine. 2017; 15(1):61.

4. Jue, T.R., Sena, E., Macleod, M., McDonald, K.L, and Hirst, T.H. Effect of topoisomerase inhibitors in glioblastoma animal models: A systematic review and meta-analysis. Oncotarget. 2018; 9(13), 11387-11401.

5. Jue, T. R., Chung, S., Nixdorf, S., Hoehn, K., Rapkins, R., Dilda, P., Hogg, P., and McDonald, K.L. PENAO and Temsirolimus on patient-derived glioblastoma cell lines: In vitro and in vivo study (under review).

Conference Abstracts

1. Jue, T.R., Chung, S.A., Rapkins, R.W., Dilda, P., Hogg, P. & McDonald, K.L. Targeting glioblastoma mitochondrial metabolism to inhibit cell proliferation & tumor growth. Neuro-Oncology, Volume 19, Issue suppl_3, 1 May 2017, Pages iii59. Presented at the 5th Quadrennial Meeting of the World Federation of Neuro- Oncology Societies (WFNOS); May 4 – 7, 2017; Zurich, Switzerland.

2. Jue, T.R., Hirst, T.C., Sena. E.S., Macleod, M.R., & McDonald, K.L. A systematic review and meta-analysis of topoisomerase inhibition in pre-clinical glioma models. Neuro-Oncology, Volume 19, Issue suppl_3, 1 May 2017, Pages iii59. Presented at the 5th Quadrennial Meeting of the World Federation of Neuro- Oncology Societies (WFNOS); May 4 – 7, 2017; Zurich, Switzerland.

3. Jue, T.R., Chung, S.A., Rapkins, R.W., Dilda, P., Hogg, P., & McDonald, K.L. Targeting glioblastoma metabolism through mTOR and mitochondrial inhibition. Neuro-Oncology, 18 (suppl 6), vi63. Presented at the 21st Annual Scientific Meeting and Education Day of the Society for Neuro-Oncology; November 17- 20, 2016; Scottsdale, Arizona.

4. Jue, T.R., Chung, S., Joshi, S., Rapkins, R., Yin, J., Dilda, P., Hogg, P., and McDonald, K.L. Targeting the Glycolytic Pathway through mTOR and Mitochondrial Inhibition in Glioblastoma. Presented at the EMBL Australia PhD Symposium 2015; November 25-27, 2015; Melbourne, Australia.

5. McDonald, K. L., Khasraw, M., Jue, T. R., Joshi, S., & Yin, J. Abstract A12: Genomically unstable glioblastoma are sensitive to Parp inhibition. Cancer Research, 75(23 Supplement), A12-A12. Presented at the AACR Special Conference: Advances in Brain Cancer Research; May 27-30, 2015; Washington, DC.

6. McDonald, K., Joshi, S., Jue, T. R., Yin, J., & Khasraw, M. ATPS-54 Genomically Unstable Glioblastoma (U-GBM) Show Exquisite Sensitivity to Parp Inhibition. Neuro-Oncology, 17(suppl 5), v30. Presented at the 20th Annual Scientific Meeting of the Society for Neuro-Oncology; November 19-22, 2015; San Antonio, Texas.

Acknowledgements

To begin with, I would like to express my sincerest gratitude to my supervisor Associate Professor Kerrie L. McDonald for her unending support throughout my PhD, for all her advice, motivation, patience, and for sharing her immeasurable knowledge in the field of Neuro-oncology.

Aside from my supervisor, I would also like to thank the members of my panel review, especially Dr Jonathan Erlich, for their encouragement, and for their insightful comments and sometimes difficult questions which broadened my research perspective.

I would like to express my appreciation to all the past and present members of the Biomarkers and Translational Research group: Swapna, Wendy, Julia, Sheri, Irene, Pukar, Kyoko, Han, Shane, Anna, Lauren, Ashleigh, Humaira, Ashraf, Victor, and especially Rob and Sylvia, for all the discussions, stories, advice and technical support that they gave me. All of which has been helpful with how I coped with my projects, PhD and personal life. I would also like to thank all the past and present members of the Adult Cancer Program. It was always interesting and enjoyable chatting with you during lunch, and over the coffee machine.

I would like to extend my gratitude to all my collaborators: Dr Theodore Hirst, Prof Malcolm Macleod and Dr. Emily Sena for sharing their knowledge and teaching me the complexities of a meta-analysis. Associate Professor Kyle Hoehn for his guidance and support in learning and understanding the Seahorse Mitochondrial Assay, and to his assistant Ms. Ellen Ozlomer for teaching me how to use the XF96 analyzer. Dr Mark Rybchyn and Dr Andrea Nuñez for helping me with Seahorse data normalisation. Dr Pierre Dilda for his advice during the beginning of my project tackling mTOR and the mitochondria. Prof Phil Hogg for generously providing me the drug PENAO.

I would also like to thank and acknowledge the Prince of Wales Clinical School, Cure Brain Cancer Foundation, TCRN, and UNSW for providing me with scholarships and travel grants.

I would like to thank my friends, Juan, Angie, Len, Jewel, Siva and Chandini for bearing with me for the last three and a half years. You guys gave me the break I needed from all the stresses I get from work. To Elma, Drea and Carmel, although we’re thousands of miles apart, you girls never fail to make me laugh.

Finally, I would like to thank my Mama Laureen and Tatay Douglas, my sisters, Sierra and Jen for their unconditional love and endless support in all the things I do. To my nieces, Kai and Lily, you girls make my day brighter with your silly antics. To my most patient and loving partner, Jay. I thank you for all your support and encouragement.

I wouldn’t have survived my PhD without all of you. Thank you!

vii

Abbreviations

4E-BP1 eukaryotic translation initiation factor 4E (eIF4E)-binding protein 1

5-ALA 5-aminolevulinic acid

Acetyl-CoA acetyl coenzyme A

ADP adenosine diphosphate

AKT protein kinase B

AMPA α-amino-3-hydroxy-5 methyl-4-isoxazolepropionate

AMPK AMP-activated protein kinase pathway

APE1 apurinic/apyrimidinic endonuclease-1

ATM ataxia telangiectasia mutated kinase

ATP adenosine triphosphate

ATR ataxia telangiectasia and rad3-related kinase

ATRX alpha-thalassemia/mental retardation syndrome X-linked

AYLL average years of life lost

BER base excision repair

BEV bevacizumab

CAMARADES Collaborative Approach to Meta-Analysis and Review of Animal Data from Experimental Studies

CDK4/6 cyclin dependent kinase 4/6

CDKN2A/B/C cyclin dependent kinase inhibitor 2A/2B/2C

CHI3L1/ YKL40 chitinase 3 Like 1

CI combination index

CISH chromogenic in situ hybridization

CNS central nervous system

CNV copy number variation

viii

CoA coenzyme A

CST Cell Signaling Technology

CTD comparative toxicogenomic database

CTMP carboxyl-terminal modulator protein

DCC deleted in colorectal cancer

DDR DNA damage response

DMSO dimethyl sulfoxide

DNA-PK DNA-dependent protein kinase

DRI dose-reduction index

DSBs double-strand breaks

DTNB 5,5′-Dithiobis (2-nitrobenzoic acid)

EGFR epidermal growth factor receptor

ERBB2 erb-B2 receptor tyrosine kinase-2

ERCC1 excision repair cross-complementing group 1

ETC electron transport chain

FA fraction affected

FAD flavin adenine dinucleotide

FADH2 reduced FAD

FDA Food and Drug Administration

FDR false discovery rates

FFPE formalin-fixed paraffin embedded

FI functional interaction

FITC Fluorescein isothiocyante

GABRA1 gamma-aminobutyric acid type A receptor alpha 1 subunit

GBM glioblastoma

GDA GEM-derived allograft

GEM genetically engineered mouse

GICs GBM-initiating cells

GLUT glucose transporter

GSAO 4-(N-(S-glutathionylacetyl)amino) phenylarsonous acid

H&E hematoxylin and eosin

HDAC histone deacetylase

HGF/SF hepatocyte growth factor/scatter factor

HIF1α hypoxia-inducible factor 1 alpha

HR hazard ratio

HRR homologous recombination repair

IC50 half maximal inhibitory concentration

IDH isocitrate dehydrogenase

IGR intergenic region

IHC immunohistochemistry

Indels insertion/deletions

IR irradiation

KEGG Kyoto Encyclopedia of Genes and Genomes

LDHA lactate dehydrogenase A lincRNA long intergenic noncoding RNA

LOH loss of heterozygosity

MERTK MER proto-oncogene, tyrosine kinase

MET MET proto-oncogene, receptor tyrosine kinase

MGMT o6-methylguanin-DNA-methyltransferase

MLH mutL homolog

MMR mismatch repair

MRI magnetic resonance imaging

MSH mutS homolog

MTH1 mutT homolog-1 mTOR mammalian target of rapamycin mTORC1/2 mammalian target of rapamycin complex ½

MTS 3-(4,5-dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4- sulfophenyl)-2H-tetrazolium, inner salt (MTS)

NA not applicable

NAD+ nicotinamide adenine dinucleotide

NADH reduced NAD+

NCI National Cancer Institute;

NCIC National Cancer Institute of Canada

NEFL neurofilament light

NER nucleotide excision repair

NES Nestin

NF1 neurofibromin-1

NF-κβ nuclear factor kappa B

NGS next generation sequencing

NHEJ non-homologous end-joining

NKX2-2 NK2 homeobox 2

NOG NOD/Shi-scid/IL-2Rγnull mouse strain

NOS not otherwise specified

NRC-US National Research Council US

OATP organic anion transporting polypeptide

OCR oxygen consumption rate

OLIG2 oligodendrocyte transcription factor 2

ORR overall response rate

OS overall survival

OXPHOS oxidative phosphorylation p14ARF tumour suppressor protein p14ARF p53 tumour suppressor p53

PARP poly ADP ribose polymerase

PBS Dulbecco’s phosphate-buffered saline

PDCL patient-derived cell line

PDGF platelet derived growth factor

PDGFRA platelet derived growth factor receptor alpha

PDH pyruvate dehydrogenase

PDK pyruvate dehydrogenase kinases

PDK1 pyruvate dehydrogenase kinase

PDX patient-derived xenograft

PENAO 4-(N-(S-penicillaminylacetyl) amino) phenylarsonous acid

PES phenazine ethosulfate

PFS progression free survival

PGC1α PPAR-γ coactivator 1 alpha

PI propidium iodide

PI3K phosphoinositide 3-kinase pathway

PIK3C1 phosphatidylinositol-4,5-bisphosphate 3-kinase, catalytic subunit beta

PIK3CA phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha

PIK3R1 phosphoinositide-3-kinase regulatory subunit 1

PIP2 phosphatidylinositol (4,5)-biphosphate

PIP3 phosphatidylinositol (3,4,5)-triphosphate

PKCα protein kinase C alpha

xii

PMSF phenylmethylsulfonyl fluoride

PTEN phosphatase and tensin homolog

RB1 retinoblastoma-1

Rheb Ras homolog enriched in brain

RT radiotherapy

RTK receptor tyrosine kinase

S6K1 S6 kinase 1

SD standard deviation

SGK1 serum glucocorticoid-induced protein kinase 1

SLC12A5 solute carrier family 12 member 5

SNPs single nucleotide polymorphisms

SNVs simple nucleotide variations

SSBs single-strand breaks

SYT1 synaptotagmin 1

TBE Tris-borate-EDTA

TBST Tris-buffered saline with Tween20

TCA tricarbocylic acid

TCGA The Cancer Genome Atlas

TERT telomerase reverse transcriptase

TMZ temozolomide

TNF Tumour Necrosis Factor

TOR target of rapamycin

TP53 tumour suppressor protein p53

TSC1 tuberous sclerosis 1/hamartin

TSC2 tuberous sclerosis 1/tuberin

TTF tumour-treating fields

xiii

TUNEL terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick end labelling

ULK1 UNC-51 like kinase 1

VEGF vascular endothelial growth factor

WGS whole genome sequencing

WHO World Health Organisation

XP xeroderma pigmentosa

YY1 Ying-Yang 1

xiv

Table of Contents

ORIGINALITY STATEMENT ...... i

AUTHENTICITY STATEMENT ...... ii

Abstract ...... iii

Publications ...... v

Conference abstracts ...... vi

Acknowledgements ...... vii

Abbreviations ...... viii

Table of contents ...... xv

List of figures ...... xx

List of tables ...... xxiv

Literature Review...... 1

Glioblastoma ...... 2

WHO classification of high grade tumours ...... 2

Molecular subtypes ...... 5

Current therapeutic approaches ...... 8

Standard therapy...... 8

Other therapies ...... 9

Personalised medicine to precision medicine ...... 11

History and definition ...... 11

Current approaches...... 14

Molecular-targeted treatments ...... 14

Significance of this study ...... 29

Aims ...... 29

Materials and Methods ...... 30

Ex vivo ...... 31

In vitro ...... 31

Cell culture ...... 31

Seeding density optimisation ...... 34

Cell proliferation assay ...... 34

Treatments ...... 35

DNA ...... 39

Protein ...... 42

Apoptosis assay ...... 51

Bioenergetic profiling ...... 51

In vivo ...... 55

Treatment ...... 56

Immunohistochemistry ...... 58

Meta-analysis ...... 59

Sources ...... 59

Inclusion criteria...... 59

Data extraction ...... 59

Study quality scoring ...... 60

Network analysis ...... 61

Data analysis ...... 61

In vitro and in vivo ...... 61

Meta-analysis and metaregression ...... 62

xvi

Characterisation of a Patient Tumour Sample ...... 63

Introduction ...... 64

Pre-clinical GBM models ...... 64

The Cancer Genome Atlas (TCGA) ...... 66

Aims ...... 67

Results ...... 68

Patient tumour sample ...... 68

Patient-derived cell line...... 71

Patient-derived xenograft model ...... 74

Genomic profile of G89 and G244 ...... 77

Discussion ...... 92

Targeted Therapy Based on Molecular Profiling ...... 96

Introduction ...... 97

Tumour heterogeneity ...... 97

Targeted therapy...... 97

Molecular profiling ...... 99

Aims ...... 99

Results ...... 100

Molecular profiling ...... 100

Cross-validation of CARIS® gene list against Comparative Toxicogenomic Database...... 102

Literature search to identify possible drug treatments ...... 103

Drug target validation by immunoblotting ...... 107

Testing drugs of interest ...... 108

Discussion ...... 110

xvii

mTOR and Mitochondrial Inhibition ...... 112

Introduction ...... 113

Cellular metabolism...... 113

Cellular metabolism regulation ...... 115

mTOR inhibitors ...... 119

Mitochondria and PENAO ...... 120

Aims ...... 122

Results ...... 123

In vitro ...... 123

In vivo ...... 135

Discussion ...... 138

Inhibiting PARP as a Therapeutic Strategy for GBM ...... 141

Introduction ...... 142

GBM and DNA repair mechanisms ...... 142

PARP inhibition ...... 143

Aims ...... 146

Results ...... 147

In vitro ...... 147

In vivo ...... 161

Discussion ...... 166

A Systematic Review and Meta-analysis of GBM in vivo Investigations for Topoisomerase Inhibition...... 169

Introduction ...... 170

Systematic reviews and meta-analyses ...... 170

Topoisomerases ...... 170

xviii

Aims ...... 171

Results ...... 173

Study characteristics ...... 175

Study quality ...... 179

Meta-analysis ...... 180

Meta-regression ...... 180

Publication bias ...... 192

Discussion ...... 194

Summary and Conclusions ...... 200

Appendices ...... 208

References ...... 255

xix

List of Figures

Figure 1.1. Evolution of CNS tumour classification from 1962 to 2016...... 3

Figure 1.2. A summary of progressive genetic alterations of diffuse gliomas in adults. .. 4

Figure 1.3. A summary of the classification of diffuse gliomas based on histological and genetic features...... 5

Figure 1.4. Trends in precision medicine...... 13

Figure 2.1. Drug treatment in 96-well format...... 38

Figure 2.2. Gel electrophoresis of PCR products ...... 41

Figure 2.3. BCA assay template...... 45

Figure 2.4. Template for used for seahorse assay...... 55

Figure 3.1. Major developments in pre-clinical cancer models...... 65

Figure 3.2. T1-weighted MRI images post gadolinium infusion...... 69

Figure 3.3. Flow chart of tissue sample processing...... 71

Figure 3.4. Light microscopy image of G89 and G244...... 72

Figure 3.5. Kaplan Meier survival curve of G89 in comparison to other xenograft models...... 75

Figure 3.6. MRI of G89 PDX model...... 76

Figure 3.7. Tumour mutation summary...... 78

Figure 3.8. Copy number alterations of G89 tumour sample...... 79

Figure 3.9. Copy number alterations of G244 tumour sample...... 80

Figure 3.10. Mutational signature for G89 and G244 ...... 82

Figure 3.11. Sanger sequencing result of POLE point mutation in G89 and G244...... 83

Figure 3.12. Oncoprint of POLE and MMR genes showing mutations and RNA expression from TCGA (Cell, 2013) dataset...... 86

Figure 3.13. FI Network map of all protein changing mutations...... 88

Figure 3.14. Network analysis of G89 driver genes...... 89

Figure 4.1. Drug target...... 98

Figure 4.2. Venn diagram showing a comparison of genes screened by CARIS® and of genes with inferred association to GBM from the CTD...... 102

Figure 4.3. Western blot validation of drug target...... 108

Figure 4.4. Dose response curves of G89 treated with topoisomerase inhibitors...... 109

Figure 5.1. Cellular metabolic pathway...... 114

Figure 5.2. Molecular regulation of cell metabolism...... 116

Figure 5.3. Chemical structure of rapamycin and its analogues...... 119

Figure 5.4. Mechanism of action of PENAO...... 121

Figure 5.5. Dose response curves...... 124

Figure 5.6. Cell viability assay for PENAO and temsirolimus combination treatment...... 126

Figure 5.7. Combination index values. calculated with CompuSyn™...... 126

Figure 5.8. Calculated dose reduction of each drug when used in combination...... 127

Figure 5.9. Apoptosis assay...... 129

xxi

Figure 5.10. Effect of PENAO and temsirolimus on mitochondrial respiration of RN1 cells...... 131

Figure 5.11. Bioenergetic profiling of RN1 cell treated with PENAO and temsirolimus combination...... 133

Figure 5.12. Assessing mTOR and S6K1 phosphorylation by western blot in response to treatment...... 134

Figure 5.13. Kaplan-meier curve comparison show no significance between control and all treatment groups...... 137

Figure 6.1. Pathways involved in DNA damage repair and their targets...... 142

Figure 6.2. PARP DNA Repair mechanism...... 144

Figure 6.3. Summary results of GBM PDCLs treated with different concentrations of ABT-888...... 147

Figure 6.4. Summary results of GBM PDCLs treated with different concentrations of TMZ...... 148

Figure 6.5. Summary results of GBM PDCLs treated with different doses of radiation...... 149

Figure 6.6. Overall summary of cell viability assay performed to assess the efficacy of ABT-888 and TMZ combination...... 150

Figure 6.7. Overall summary of cell viability assay performed to assess the efficacy of the combination of ABT-888 and radiation...... 152

Figure 6.8. Summary of clonogenic assay performed to assess efficacy of the combination of ABT-888 and radiation...... 153

Figure 6.9. Dose-effect curves (A) and median-effect plots (B) for RN1 and G89...... 155

Figure 6.10. Combination Index of ABT-888 and radiation shows synergy in reducing cell viability in GBM PDCLs...... 156

xxii

Figure 6.11. Apoptotic cell death was induced after ABT-888 and radiation treatment combination...... 159

Figure 6.12. Expression of proteins involved in the MRN complex...... 160

Figure 6.13. Treatment plan for in vivo investigation of ABT-888 and radiation combination treatment...... 162

Figure 6.14. Kaplan-Meier curve comparing treated groups with the control group. .. 162

Figure 6.15. Ki-67 and TUNEL staining of brain samples from RN1 PDX treated with ABT-888 and radiation, alone and in combination...... 164

Figure 7.1. Study selection summary ...... 174

Figure 7.2. Study Characteristics...... 178

Figure 7.3. Quality score frequency distribution...... 179

Figure 7.4. Meta-analysis...... 181

Figure 7.5. Multivariable meta-regression based on survival outcome...... 185

Figure 7.6. Univariable meta-regression based on survival outcome...... 186

Figure 7.7. Univariable meta-regression based on tumour volume reduction...... 189

Figure 7.8. Post-hoc univariable meta-regression analysis of choice of carriers...... 191

Figure 7.9. Publication bias...... 193

xxiii

List of Tables

Table 1.1. Common molecular gene alteration within each GBM subtype...... 7

Table 1.2. Summary of drugs investigated in clinical trials for GBM treatment and their molecular targets...... 19

Table 1.3. Clinical trials looking at targeted treatment in GBM from 2010 to the most recent...... 20

Table 2.1. A summary of pharmaceutical agents used in this thesis...... 35

Table 2.2. Titration buffer preparation*...... 36

Table 2.3. Dilutions of reagents used for titration...... 37

Table 2.4. Dilution of 20x diluted PENAO in a 96-well plate format...... 37

Table 2.5. Primers used to detect POLE mutation in Exon 30...... 40

Table 2.6. PCR reaction mixture ...... 41

Table 2.7. Cell lysis buffer preparation...... 42

Table 2.8. Dilution of BSA stock solution for standard curve determination...... 44

Table 2.9. Laemmli buffer...... 46

Table 2.10. Running buffer...... 46

Table 2.11. Transfer buffer...... 47

Table 2.12. Tris-buffered saline with Tween20 (TBST)...... 48

Table 2.13. Probing antibodies used within experiments...... 49

Table 2.14. Dilution of compounds used for the mitochondrial stress test...... 54

xxiv

Table 2.15. 12-item checklist for assessing methodological quality and publication bias ...... 60

Table 3.1. Immunohistochemistry and MGMT methylation results of primary and recurrent tumour...... 70

Table 3.2. STR profile of G89 and G244 PDCLs...... 73

Table 3.3. Alterations observed in MMR genes in patient tumour samples...... 84

Table 3.4. Significant co-occurring alterations between POLE and MMR genes...... 85

Table 3.5. Cancer driver genes within the patient genome ...... 89

Table 3.6. Affected KEGG pathways within the patient genome determined from using the STRING database ...... 90

Table 4.1. Summary results of genes/biomarkers tested in the MI (Molecular Intelligence) Profile™ by CARIS® Life Sciences ...... 100

Table 4.2. Drug agents predicted to have a benefit based on molecular profile according to CARIS® Life Sciences...... 103

Table 4.3. Summary of rationalising compound inclusions for in vitro testing...... 105

Table 4.4. Concentration range used for treatment...... 109

Table 5.1. IC50 concentrations for PENAO and Temsirolimus...... 123

Table 5.2. Effective doses (ED) inhibiting 10, 50 and 90% of cell population in GBM PDCLs...... 124

Table 5.3. Summary of median survival in all treatment groups...... 136

Table 6.1. Summary of pre-clinical investigations of PARP inhibition with IR and/or TMZ...... 145

Table 6.2. Combination index (CI) values for the combined treatment of ABT-888 and radiation...... 156

xxv

Table 7.1. Summary of TP53 mutation status for glioma models included in this meta- analysis...... 176

Table 7.2. Multivariable meta-regression of survival data...... 183

Table 7.3. Comparison of univariable meta-regression for survival and tumour volume experimental comparisons...... 187

xxvi

Literature Review

Parts of this chapter have been previously published in the Journal of Neuro-oncology

and Tumour Biology (1, 2).

Glioblastoma

In 2013, the Australian Institute of Health and Welfare recorded 1,636 new incidences of brain cancer with a predicted estimate of 1,935 new cases in 2018 (3). In the US, 32% of occurring brain tumours are malignant with almost half of this being glioblastoma

(GBM). GBM is a diffuse astrocytic tumour that may arise de novo or as a secondary to a lower grade astrocytic tumour. The incidence rate of GBM has been reported to be 3.2 per 100,000 population and occurs in 2 males for every female with the disease (4). Out of all the malignant cancers, GBM has the highest population burden in terms of the average years of life lost (AYLL) (1, 5, 6). Young people at the peak of their working and child-rearing obligations are the most frequently affected with GBM (1, 7). However, the elderly at the age of 65 years and above has been increasingly reported (1, 8-12).

WHO classification of high grade tumours

Brain tumour classification originated from the works of Bailey and Cushing in 1926 where they classified brain tumours based on histological features (13, 14). This has greatly evolved into a more complex scheme of classifying tumours of the central nervous system (CNS) not just based on histological features but also based on molecular parameters (Figure 1.1). Histological grading of tumours used to be the main contributing factor for selecting what treatment regimen a patient receives (15). Most common CNS tumours in adults were histologically classified as astrocytic, oligoastrocytic and oligodendroglial. The 2007 World Health Organisation (WHO) classification system further graded these tumours into grade II, III and IV. WHO grade II tumours are diffusely infiltrative tumours with cytological atypia alone, grade III tumours additionally show anaplasia and mitotic activity, while grade IV tumours display microvascular proliferation and/or necrosis in addition to those characteristics already mentioned (15).

Figure 1.1. Evolution of CNS tumour classification from 1962 to 2016.

In 1979, Zulch published the first edition of the histological typing of CNS tumours. Kleihues et al. edited the first edition in 1993 to reveal the progress of diagnosing CNS tumours as a result of adding immunohistochemistry into pathology (16). In 2000, Kleihues and Cavenee published the third edition which included genetic profiles to aid the description of CNS tumours (17). In 2007, the WHO working group updated the histological grading scheme and genetic profile of CNS tumours (15). The 2016 edition added molecular parameters to histology in the classification scheme.

With recent advances in technology, access to a patient’s molecular profile has become more available. Over the past years, genetic alterations that cause tumour progression in different types of malignant glioma have been discovered (Figure 1.2) (18).

In 2016, the WHO working group comprised of prominent pathologists and geneticists updated the molecular classification of CNS tumours. The new classification now includes molecular parameters in addition to histology (Figure 1.3) (19).

Figure 1.2. A summary of progressive genetic alterations of diffuse gliomas in adults.

Primary glioblastoma acquire complex genetic and chromosomal alterations involving chromosome 10, 7, 19 and 20; and important cellular pathways, such as the RTK/MAPK/PI3K, p53 and pRB1 pathway. IDH1/2 mutations are common in WHO grade II and III astrocytic, oligoastrocytic and oligodendroglial gliomas, and secondary glioblastoma. TP53 mutations and 1p/19q codeletion are characteristic of diffuse astrocytic gliomas and oligodendroglial tumours, respectively; while oligoastrocytic gliomas have either one. Progression of WHO grade II gliomas to grade III is often associated with 9p losses and CDKN2A/B and p14ARF deletion, while progression to secondary glioblastoma is often associated with loss of 10q and of the DCC gene. Adapted from previous publications (18, 20).

Abbreviations: TERT – telomerase reverse transcriptase; MGMT – o6-methylguanin- DNA-methyltransferase; EGFR – epidermal growth factor receptor; MET – MET Proto- Oncogene, Receptor Tyrosine Kinase; PDGF – platelet derived growth factor; PDGFRA – platelet derived growth factor receptor alpha; NF1 – neurofibromin-1; ERBB2 – erb- B2 receptor tyrosine kinase-2; PTEN – phosphatase and tensin homolog; PIK3C1 – phosphatidylinositol-4,5-bisphosphate 3-kinase, catalytic subunit beta; PIK3CA – phosphatidylinositol-4,5-bisphosphate 3-kinase catalytic subunit alpha; CTMP – carboxyl-terminal modulator protein; TP53 – tumour suppressor protein p53; p14ARF – CDKN2A p14 alternate reading frame; RB1 – retinoblastoma-1; CDKN2A/B/C, cyclin dependent kinase inhibitor 2A/2B/2C; CDK4/6 – cyclin dependent kinase 4/6; ATRX – alpha-thalassemia/mental retardation syndrome X-linked; DCC – DCC netrin 1 receptor.

Figure 1.3. A summary of the classification of diffuse gliomas based on histological and genetic features.

Adapted from ‘The 2016 World Health Organization Classification of Tumors of the Central Nervous System: a summary’ (19).

Abbreviation: IDH – isocitrate dehydrogenase; ATRX – alpha-thalassemia/mental retardation syndrome X-linked; TP53 – tumour suppressor protein p53; 1p – short arm of chromosome 1; 19q – long arm of chromosome 19; NOS – not otherwise specified.

Molecular subtypes

The profiling of the human genome has given scientists and researchers an insight into the possibilities of the DNA being the root source of a person’s susceptibility to a certain disease or condition. This has also allowed the identification of possible genetic targets which therapeutic agents can be directed to. GBM was the first cancer studied by The

Cancer Genome Atlas (TCGA) Research network (2, 21, 22). The TCGA aims to improve cancer prevention, detection and treatment by identifying genomic changes in each cancer type and understanding how the disease is driven by these changes. The TCGA GBM project screened over 500 tumours to provide a comprehensive genome-wide map of the genetic, epigenetic and transcriptomic changes, as well as proteomic changes (2, 23, 24).

This led to the identification of four distinct molecular subtypes of GBM on the basis of

signatory gene expression profiles, namely classical, mesenchymal, proneural, and neural. These have also been associated with particular genetic alterations (Table 1.1)

(25). The classical subtype exhibits an astrocytic expression profile, with frequent chromosome 7 amplification associated with EGFR expression, loss of chromosome 10 associated with PTEN expression, and focal deletions of 9p21.3 affecting CDKN2A expression. TP53 mutations, although common in GBM, are not found in the classical subtype. The mesenchymal subtype is characterised by mesenchymal marker expression, with frequent NF1 and PTEN deletions or mutations. The proneural subtype presents an oligodendrocytic expression signature exhibiting focal amplifications of the chromosome

4q12 region containing PDGFRA, or mutations of the isocitrate dehydrogenase 1 gene,

IDH1. Lastly, the neural subtype exhibits expression of neuronal markers and displays various mutations and copy number alterations including amplification of EGFR and deletion of PTEN (2, 25). No prognostic difference was observed between the classical, mesenchymal and neural subtypes. The proneural subtype was associated with a secondary GBM diagnosis, a younger age group and a longer survival time, all of which has been attributed to the presence of an IDH1 mutation (25-27). However, GBM patients with a wild-type IDH1 gene classified under the proneural subtype did not reflect this survival advantage (28).

Table 1.1. Common molecular gene alteration within each GBM subtype (25).

Proneural Neural Mesenchymal Classical

IDH1 mutation Neuron marker NF1 mutation EGFR expression (NEFL, amplification PDGFRA GABRA1, SYT1 TP53 loss amplification and SLC12A5) TP53 wildtype PTEN loss TP53 mutations Ras activation and LOH CHI3L1/YKL40 activation PTEN deletion PTEN loss MET activation CDKN2A deletion NKX2-2 amplification CD44 NES activation amplification OLIG2 Notch signalling amplification MERTK pathway activation amplification EGFR Sonic hedgehog amplification TNF pathway pathway activation activation CDKN2A loss NF-κβ pathway PIK3CA/PIK3R1 activation mutation

Abbreviations: IDH1 – isocitrate dehydrogenase 1; PDGFRA – platelet-derived growth factor receptor; TP53 – tumour suppressor protein p53; LOH – loss of heterozygosity; PTEN – phosphatase and tensin homolog; NKX2-2 – NK2 homeobox 2; OLIG2 - oligodendrocyte transcription factor 2; EGFR – epidermal growth factor receptor; CDKN2A – cyclin dependent kinase inhibitor 2A; PIK3CA – phosphatidylinositol-4,5- bisphosphate 3-kinase catalytic subunit alpha; PIK3R1 – phosphoinositide-3-kinase regulatory subunit 1; NEFL – neurofilament light; GABRA1 - gamma-aminobutyric acid type a receptor alpha1 subunit; SYT1 – synaptotagmin 1; SLC12A5 – solute carrier family 12 member 5; NF1 – neurofibromin-1; CHI3L1/ YKL40 – chitinase 3 like 1; MET – MET proto-oncogene, receptor tyrosine kinase; MERTK – MER proto-oncogene, tyrosine kinase; TNF – tumour necrosis factor; NF-κβ – nuclear factor kappa b; NES – nestin

Current therapeutic approaches

The survival trends for patients diagnosed with GBM have remained largely stagnant, suggesting a lack of improvement in the therapeutic options for patients. GBM are highly refractory to treatment with local tumour recurrence occurring 2–3 cm from the original resection cavity frequently observed. Relapsed GBMs are difficult to manage with a median survival of only a few months after recurrence (2, 29).

Standard therapy

Patients with a good performance status (KPS ≥ 70%) at the time of medical intervention undergo maximal surgical resection of their tumour followed by the standard treatment.

This standard treatment protocol developed by the European Organisation for Research and Treatment of Cancer (EORTC) and the National Cancer Institute of Canada (NCIC), also known as Stupp protocol, consists of upfront radiotherapy (60Gy/30 fractions) with concomitant and adjuvant temozolomide (TMZ) chemotherapy (1, 30). During the 6- week radiotherapy treatment, concomitant TMZ dosage is at 75 mg/m2 body-surface area for 7 days weekly. This is followed by adjuvant TMZ dosage at 150-200 mg/m2 body- surface area for 5 days out of a 28-day cycle (6 cycles) (30). TMZ is a second-generation imidazotetrazine derivative synthesised in the 1990s (31, 32). It is an atypical oral alkylating agent that induces its cytotoxic effects by alkylating the guanine residues in the DNA. This results in a mispairing with thymine during DNA replication, consequently resulting to cellular arrest in rapidly dividing cells (33, 34). The MGMT gene encodes for an enzyme that protects the DNA from damages by removing the mutagenic alkyl adducts induced by TMZ and other alkylating agents. Loss of MGMT is a common occurrence in different human cancer types (35). Hypermethylation of the CpG islands within the

MGMT promoter and enhancer regions regulate its expression in cells (36-38). In GBMs

with MGMT promoter methylation, MGMT loses its function therefore resulting in enhanced chemotherapeutic effects of TMZ due to the tumour’s inefficiency to repair itself (39).

The standard therapy for GBM results in a median survival of almost 15 months compared to just 12 months with radiotherapy alone. Overall survival was reported to be at almost

10% for 5 years with radiotherapy and TMZ, which is a great improvement from 2% with radiotherapy alone (32, 40). As mentioned above, MGMT promoter methylation is the strongest predictor of a survival advantage from the treatment with standard radiotherapy and TMZ treatment after surgery. Hegi et al. reported the 2-year survival rates of GBM patients with unmethylated MGMT promoter to be 14%, while GBM patients with methylated MGMT promoter had a 2-year survival rate of 46% (39).

Other therapies

In the past 40 years, only a handful of drugs have been US Food and Drug Administration

(FDA) approved for the treatment of GBM in the USA, including TMZ. Other treatment modalities that have been previously approved are nitrosureas (lomustine, carmustine) and a VEGF inhibitor (bevacizumab). Due to lack of survival data showing a benefit, bevacizumab was not approved the Therapeutic Goods Administration (TGA) for treatment of GBM.

Nitrosureas

The earliest report for the use of nitrosureas was in 1970 (41). The most commonly used nitrosureas for GBM are lomustine and carmustine. Nitrosureas are antineoplastic agents that attach to the alkyl group of the DNA subsequently causing DNA damage. In 1970 carmustine was initially administered intravenously while lomustine was given orally

(41). In 1976, Brisman et al. evaluated the use of nitrosureas in a prospective, controlled, randomly allocated clinical trial where they reported no significant improvement in patients receiving nitrosureas as adjuvant to surgery and radiotherapy versus patients who received surgery and radiotherapy alone (41). In 1979, Hochberg et al. reported a retrospective evaluation of 74 patients treated with lomustine as an adjunct to surgery and radiotherapy (42). Here they reported an improved median survival of 11.5 months in comparison to 5.75 months with just surgery and radiotherapy. In 1996, the use of carmustine wafers (Gliadel®) was approved for the treatment of newly diagnosed and recurrent GBM (43). In 1997, a prospective, randomised double-blind study of a group treated with carmustine wafers versus a placebo group was performed (44). In 2003, a bigger phase III clinical trial was performed examining the effect of carmustine wafers in newly diagnosed malignant glioma (45). Both clinical trials showed a survival advantage of 2 to 4 months versus the placebo group. A meta-analysis of 60 publications investigating the use of carmustine wafers reported a 2-year overall survival (OS) of 26% and a median survival of ~17 months for newly diagnosed high grade glioma (HGG); and a 15% 2-year OS and ~10 months median survival for recurrent HGG (46).

Bevacizumab

Bevacizumab (Avastin®) is a recombinant humanised monoclonal immunoglobulin G subclass 1 antibody (47). It inactivates human vascular endothelial growth factor (VEGF) activity by inhibiting it from attaching to its endothelial surface receptors (47). One of

GBM’s characteristics is its high vascularization and high expression of VEGF making bevacizumab a desirable therapeutic agent for treatment (47, 48). In 2007, Vredenburgh et al. investigated the safety and efficacy of combining bevacizumab and irinotecan for the treatment of recurrent grade III and IV glioma. They concluded that the combination

had acceptable toxicity level and recommended that the combination be an active treatment for recurrent high-grade glioma (49). In 2009, the same combination was evaluated in patients diagnosed with recurrent GBM in a larger phase II multicentric clinical trial (50). Here the study resulted in a 6-month progression free survival (PFS) rate of 42.6% and a median OS of 9.2 months in the group treated with bevacizumab alone; and 50.3% and 8.7 months for the group treated with bevacizumab and irinotecan

(50). In 2014, two phase III clinical trials were conducted to evaluate the efficacy of adding bevacizumab to radiotherapy and TMZ treatment. Both clinical trials reported a prolonged PFS but showed no improvement in OS (51, 52).

Personalised medicine to precision medicine

History and definition

The Human Genome Project started in 1990. In 2002, Guttmacher and Collins stated

“With the sequencing of the human genome…, the practice of medicine has now entered an era in which the individual patient's genome will help determine the optimal approach to care,… Genomics, which has quickly emerged as the central basic science of biomedical research, is poised to take center stage in clinical medicine as well.” (53).

Personalised medicine for cancer treatment has been founded on the concept that high- throughput screening of an individual patient’s tumour will lead to analysis of its molecular profile, sequentially driving selection of drugs that could be effective in prolonging the patient’s survival (54, 55). However, an argument that a personalised medicine approach has always been practised by physicians has been pointed out (56), thereby shifting its terminology to precision medicine. Precision medicine has been previously defined as “treatments targeted to the needs of individual patients on the basis of genetic, biomarker, phenotypic, or psychosocial characteristics that distinguish a given

patient from other patients with similar clinical presentations” (56). Both personalised medicine and precision medicine seem to have the same definition and can be used interchangeably. However, a report by the National Research Council US (NRC-US) cautioned that the use of the term personalised medicine may be misinterpreted as developing unique treatments for every individual patient (57). The NRC-US, on the other hand, defined precision medicine as “tailoring of medical treatment to the individual characteristics of each patient” by classifying individual patients “into subpopulations that differ in their susceptibility to a particular disease, in the biology and/or prognosis of those diseases they may develop, or in their response to a specific treatment” (57). This differentiated the latter from the previous definition in a way that precision medicine has a wider application to a bigger population, as opposed to an individual patient. Over the years an increasing trend can be observed in the interest and investigation of personalised and precision medicine (Figure 1.4).

Figure 1.4. Trends in precision medicine.

(A) Number of publications returned in PubMed after searching for the keywords: “precision medicine”, “personalised/personalized medicine” and “targeted therapy”. (B) Shows search interest of each keyword in Google. Interest over time (y-axis) represents the search interest of each keyword in relation to the highest peak in the graph for the given time and region.

Current approaches

In various cancer types, precision medicine is specifically being applied through targeted therapy. Targeted therapy is a form of treatment modality that attacks a specific molecular target in a person’s genomic structure. This arose from Paul Erlich’s “magic bullet” concept, where compounds are developed to selectively target a specific disease-causing organism or cell while avoiding others. MD Anderson defines targeted therapy as treatments designed to interfere with specific molecules or cancer-causing genes that encode these molecules (58). Some targeted therapies work by either forming a drug complex with the specific molecule or the drug blocks the receptor for that specific molecule, therefore inhibiting its function.

The use of the term “targeted therapy” was first recorded in the 1980s (59). The interest with this tailored therapeutic approach for cancer treatment started to rise after the TCGA program started to map out the genomic changes in different cancer types in 2005 (Figure

1.4A).

Molecular-targeted treatments

Increasingly, the development of novel therapies involves defining drug-diagnostic combinations where the presence of a molecular target or marker identifies patients who are most likely to respond to a specific therapy. This model of developing treatment and diagnostic/companion biomarker combinations is the emerging paradigm for novel drug and diagnostic development (2, 60-62). Trials in other cancers such as breast cancer, melanoma, bowel, and non-small cell lung cancer (NSCLC) have demonstrated clear utility in the incorporation of biomarkers to stratify patients to the targeted treatment.

Breast cancer

In breast cancer, studies have shown that 25% of breast cancer tumours overexpress the

HER-2/neu oncoprotein (63). HER-2/neu is related to EGFR, also known as HER1, which is dysregulated in almost 50% of breast cancer tumours (63, 64). HER-2/neu amplification is a known prognostic factor for survival and tumour relapse in patients

(64). Patients with high HER2/neu expression are treated with trastuzumab, a monoclonal antibody against the HER2 oncoprotein (65-68). The addition of trastuzumab to the treatment regimen of breast cancer improved patient survival by 37% (2, 69). HER2- positive patients were also shown to respond to erlotinib, an EGFR inhibitor, having an overall response rate of 67% (70).

Colon cancer

In colorectal cancer, targeted therapy with cetuximab, an EGFR inhibitor, in patients with

KRAS wild-type metastatic colorectal carcinoma has shown enhanced treatment effects as first- and second-line treatment in conjunction with chemotherapy and as a monotherapy for third-line treatment (71-74). Bevacizumab, a VEGF inhibitor, in combination with chemotherapy is also used as first- and second-line treatment (75-77).

However, a recent study has shown that KRAS wild-type colorectal cancer patients who receive bevacizumab after prior treatment with cetuximab show modest if not poorer response (78). Colorectal cancer patients with KRAS mutations, on the other hand, are not given these targeted treatments due to the risk of acquiring treatment resistance. Instead they are given BRAF inhibitors, such as sorafenib, which inhibit ARAF, BRAF and

CRAF found downstream of KRAS (2, 79).

Lung cancer

In patients with non-small cell lung carcinoma, EGFR mutations were found to be prognostic of a sensitive response to EGFR tyrosine kinase inhibitors, such as erlotinib and gefitinib (80, 81). However, non-small cell lung carcinoma patients who harbour a

KRAS mutation as well as EGFR mutation are known to be resistant to such targeted treatments (81). Alternatively, patients who harbour an ALK mutation have a higher chance of responding to crizotinib or ceritinib, which are known ALK tyrosine kinase inhibitors (82).

Melanoma

In melanoma, the most common target is the BRAF V600E activating mutation.

Treatment with a BRAF inhibitor, PLX4032 (also known as RO5185426), resulted to a

60% response rate in patients who are positive for BRAF V600E mutation (83-85).

Glioblastoma

In GBM, despite the different known molecular subtypes, treatment regimens remain limited to those mentioned in Section 1.2. VEGF, VEGFR and EGFR are the most investigated molecular targets. Bevacizumab, which inhibits VEGF, is the most studied molecular-targeted treatment as discussed in Section 1.2.2. However, since it’s approval in 2009 various targeted modalities have been under investigation. A few are mentioned in Table 1.2.

Cediranib, a VEGFR inhibitor was investigated in a phase III clinical trial including 325 patients. This study failed to demonstrate an improvement in progression free survival when cediranib was used as a monotherapy (hazard ratio [HR] = 1.05; 95% CI, 0.74 to

1.50; two-sided P = .90) or in combination with lomustine (HR = 0.76; 95% CI, 0.53 to

1.08; two-sided P = .16) (86).

Erlotinib, an EGFR inhibitor observed to show effectiveness in lung and breast cancer patients, failed to show effect as a monotherapeutic agent in GBM resulting to just 1.9 months and 6.9 months in PFS and OS, respectively (87). Erlotinib was also investigated in combination with various treatment modalities. One phase II clinical trial investigated erlotinib in combination with bevacizumab and standard chemoradiotherapy showed an improvement in median PFS (13.5 months, 95% CI: 12.2–17.0 months, p=0.03) but failed to show significant improvement in median OS (19.8 months, 95% CI: 17.1–28.6 months, p= 0.33) compared to a historical control (8.6months and 18 months, respectively) (88).

Another phase II clinical trial looked at the same combination in GBM patients with an unmethylated MGMT promoter and showed no improvement in median PFS (9.2 months,

95% CI: 6.4 – 11.3) and median OS for 33 months (13.2 months 95% CI: 10.8 – 19.6)

(89). Another combination that was investigated in a phase II clinical trial was with sorafenib, which again resulted in disappointing results in median PFS (2.5 months, 95%

CI: 1.8–3.7 months) and median OS (5.7 months, 95% CI: 4.5–7.9 months) (90).

Rindopepimut, a vaccine for EGFR variant III, was investigated in a phase II clinical trial where it showed promising results of improving median overall survival to 21.3 months at 29.5 months follow-up in comparison to the Stupp protocol (14.6 months at 28 months follow-up) (91). However, a recent phase III clinical trial investigating Rindopepimut

(Rintega®) in newly diagnosed GBM positive for EGFR variant III (n=745) failed to replicate these positive results. The trial resulted to failed improvement in median overall survival from the treatment arm (20.1 months, 95% CI: 18.5 – 22.1 months) in

comparison to the control arm (20 months, 95% CI: 18.1 – 21.9 months) (HR 1ꞏ01, 95%

CI 0ꞏ79–1ꞏ30; p=0ꞏ93) (92).

Novo-TTF is a portable device that delivers alternating electric fields (Tumour-treating fields; TTF) via topical disposable transducer arrays. The TTF interfere with cell division by preventing the transition of cells to anaphase from metaphase through misalignment of microtubule subunits in the mitotic spindle (93). Additionally, TTF target macromolecules important for chromosomal alignment and separation in proliferating cancer cells in GBM by taking advantage of the heterogenous electric field in these dividing cells (94). The novel therapeutic device showed promise in vivo by inhibiting tumour growth in a malignant melanoma mouse model (94). Novo-TTF, was investigated in a phase III clinical trial involving 237 GBM patients with two prior treatments. The trial failed to show a significant difference in median survival between those treated with

TTF alone (6.6 months) and those treated with chemotherapy alone (6 months) (HR: hazard ratio 0.86, 95% CI: 0.66–1.12, p=0.27) (95).

Other phase II clinical trials investigating molecular-targeted treatments are summarised in Table 1.3.

Table 1.2. Summary of drugs investigated in clinical trials for GBM treatment and their molecular targets.

VEGF HGF/SF

Aflibercept Rilotumumab

Bevacizumab AMPA glutamate receptor

Patupilone Talampanel

VEGFR Signalling pathway

Cedirnaib TLN-4601

EGFR HDAC

Cetuximab Vorinostat

Erlotinib Toposiomerase I & II

Lapatinib Irinotecan

Nimotuzumab Etoposide

Multiple receptor tyrosine kinase mTOR

2 RTK Sirolimus

Imatinib

Sorafenib

Vandetanib

4 or more RTK

Pazopanib

Nindetanib

Sunitinib

Abbreviations: VEGF – vascular endothelial grothfactor; VEGFR – vascular endothelial growth factor receptor; EGFR – endothelial growth factor receptor; RTK – receptor tyrosine kinase; HGF/SF – hepatocyte growth factor/scatter factor; AMPA – α-amino-3-hydroxy-5 methyl-4- isoxazolepropionate; HDAC – histone deacetylase; mTOR – mammalian target of rapamycin

Table 1.3. Clinical trials looking at targeted treatment in GBM from 2010 to the most recent.

Study Drug Phase Patients ORR Median PFS Median OS (n) (%)

de Groot et al. Aflibercept II 39 18 3 months 9.75 months (96) (95% CI: 2 – 4 months)

Galanis et al. (97) Bevacizumab + II 54 18.5 2.9 months 5.6 months Sorafenib (95% CI: 2.3 – 3.6 (95% CI: 4.7 – 8.2 20 months) months)

Hasselback et al. Bevacizumab + II 37 37.9 4.25 months 9.6 months (98) Cetuximab + Irinotecan (range: 1.75 – 31.25 (range: 0.5 – 35.25 months) months)

Hasselback et al. Bevacizumab + II 24 38.1 7.25 months 11.23 months (98) Irinotecan (range: 1.5 – 51.5 (range: 0.25 – 56.25 months) months)

Reardon et al. (99) Bevacizumab + II 40 33 PFS-6: 46.5% 8.3 months Carboplatin + Irinotecan (95% CI: 30.4 – 61%) (95% CI: 5.9 – 10.7 months)

Reardon et al. Bevacizumab + II 13 0 2.03 months 4.75 months (100) Etoposide (95% CI: 1.03 – 3 (95% CI: 2.75 – 6.43 months) months)

Reardon et al. Bevacizumab + II 10 0 1.03 months 3.15 months (100) TMZ (95% CI: 0.75 – 1.98 (95% CI: 1.15 – 5.83

21 months) months)

Raizer et al. (101) Bevacizumab II 50 24.5 PFS-6: 25% 6.4 months

Sathornsumetee et Bevacizumab + II 25 48 4.5 months 11.15 months al. (102) Erlotinib (95% CI: 3 – 5.98 (95% CI: 7.1 – 17.18 months) months)

Soffietti et al. Bevacizumab + II 54 48 5.29 months 9.13 months (103) Fotemustine

Batchelor et al. Cediranib II 31 56.7 3.9 months 7.57 months (104) (95% CI: 2.73 – 4.83 (95% CI: 5.9 – 9.77 months) months)

Lassen et al. (105) Bevacizumab II 13 0 8 weeks OS: 15 weeks +TMZ

22 Hofland et al. Bevacizumab + II 32 32 7.7 months 11.8 months (106) TMZ (95% CI: 5.1 – 10.2 (95% CI: 8.2 – 15.3 months) months)

Hofland et al. Bevacizumab + II 31 23 7.3 months 15.1 months (106) Irinotecan (95% CI: 5.0 – 9.3 (95% CI: 9.6 – 20.6 months) months)

Møller et al. (107) Bevacizumab + II 32 25 5.2 months 7.9 months Irinotecan (95% CI: 3.1 – 7.2 (95% CI, 6.3 – 9.6 months); months);

van Linde et al. Bevacizumab + II 19 19 9.6 months 16 months (108) RT + TMZ (95% CI: 4.3 – 14.4 (95% CI: 8.1 – months) 26.3months)

Peereboom et al. Erlotinib + II 56 5 2.5 months 5.7 months (109) Sorafenib (95% CI: 1.8 – 3.7 (95% CI: 4.5 – 7.9 months) months)

Raizer et al. (110) Erlotinib II 43 0 2 months 14 months

(95% CI: 9 – 17 months) 23

Reardon et al. Erlotinib + II 32 0 PFS-6: 3.1% 8.45 months (111) Sirolimus (95% CI: 5.48 – 13.4 months)

Aoki et al. (112) Carboplatin + II 42 25 PFS-6: 35% 10.7 months Etoposide + Ifosfamide (95% CI: 22 – 43%); (95% CI: 9.7 – 15.2 months);

Schäfer et al. Carboplatin + II 10 0 2 months 5 months (113) Etoposide + Ifosfamide (95% CI: 0.5 – 3.6 (95% CI: 3.6 – 6.4 months) months)

Dresemann et al. Imatinib + II 120 25 1.5 months OS-6: 40% (114) hydroxyurea (95% CI: 1.5 – 1.75 months);

PFS-6: 5%

(95% CI: 1 – 10%) 24 Jaeckle et al. (115) Irinotecan + RT + II 56 3 3.9 months 10.8 months BCNU (95% CI: 3.2 – 5.1 (95% CI: 7.7 – 14.9 months); months)

PFS-6: 28.6%

(95% CI: 18.9 – 43.2%)

Reardon et al. Lapatinib + II 41 1 PFS-6: 15% NA (116) Pazopanib

Muhic et al. (117) Nintedanib II 25 0 1 month 6 months

(95 % CI 0.7 – 1.3 (95 % CI, 3.6 – 8.4 months); months)

PFS-6: 4%

(95 % CI 0.1–20.4%)

Solomon et al. Nimotuzumab + II/III & 19 na na 11.8 months (118) RT IV (95% CI: 8.2 – 15.3 months) 25 Wang et al. (119) Nimotuzumab + ?? 26 na 10 months 15.9 months TMZ + RT (95% CI: 7.8 – 12.4 (95% CI: 13 – 19.2 months); months);

PFS-6: 69.2% OS-6: 88.5%

Oehler et al. (120) Patupilone II 9 0 1.5 months 21.5 months

(95% CI, 1.3 – 1.7 (95% CI, 17 – 25 months); months);

PFS-6: 22% OS-12: 45%

(95% CI, 0 – 46) (95% CI, 14 – 76)

Iwamoto et al. Pazopanib II 35 5.9 12 weeks 35 weeks (121) (95% CI: 8 – 14 weeks) (95% CI: 24 – 47 weeks)

Wen et al. (122) Rilotumumab II 61 0 4.1 weeks 6.5 weeks

(95% CI: 4.0 – 4.1 (95% CI: 4.1 – 9.8 weeks)*; weeks)*;

4.3 weeks 5.4 weeks

(95% CI: 4.1–8.1 (95% CI: 3.4 – 11.4 26 weeks)** weeks)**

Zustovich et al. Sorafenib + TMZ II 43 12 3.2 months 7.4 months (123) (95% CI: 1.8 – 4.8 (95% CI: 5.6 – 9.0 months); months)

PFS-6: 26%

Kreisl et al. (124) Sunitinib II 63 10 PFS-6: 10.4% 9.4 months

(95% CI: 3.2 – 33.8)+; (95 % CI: 6.15 – 21.90)+;

PFS-6: 0%++ 4.37 months

(95 % CI: 3.02 – 6.21)++

Pan et al. (125) Sunitinib II 16 0 PFS-6: 16.7% 12.6 months

Iwamoto et al. Talampanel II 22 5 5.9 weeks 13 weeks (126) (95% CI: 4.6 – 6 weeks); (95% CI: 9.5 – 25.6 weeks) PFS-6: 4.6%

27 (95% CI: 0.1 – 23%)

Mason et al. (127) TLN-4601 II 20 0 PFS-6: 0% 21.43 weeks

Kreisl et al. (128) Vandetanib II 32 12.5 1.3 months 6.3 months

(95% CI: 0.9 – 1.9 (95% CI: 3.8 – 8.5 months); months)

PFS-6: 6.5%

(95% CI: 1.7 – 24.7%)

Friday et al. (129) Vorinostat + II 34 2.9 PFS-6: 0% 3.2 months Bortezomib (range: 0.7 – 24.8 months)

Legends: (*) – Group A: 10mg/kg; (**) – Group B: 20mg/kg; (+) – Group A: BEV-Naïve; (++) – Group B: BEV-resistant

Abbreviations: RT – radiotherapy; ORR – overall response rate; na – not applicable; BEV – bevacizumab 28

Significance of this study

GBM’s heterogeneous characteristic is a significant barrier to treating the disease (130,

131). A one bullet approach to treating all patients with GBM seem impractical and unrealistic. Hence, the beseeching demand to further investigate other therapeutic approaches and modalities.

Aims

A patient newly diagnosed with GBM (2014) was treated at Prince of Wales Hospital.

With her consent, I developed a patient-derived cell culture and patient-derived xenograft model, denoted as G89. During the course of my research, the patient’s tumour recurred and is denoted G244.

The overarching aim of this thesis was to characterise the tumour(s) molecularly and test novel therapies for impending treatment. Specifically, the aims of this thesis are:

AIM 1: To develop and characterise a patient-derived in vitro and in vivo GBM model.

AIM 2: To identify targetable molecular markers using a commercial biotargeting system and whole genome sequencing and validate the identified markers.

AIM 3: To inhibit the identified markers from Aim 2 in patient-derived in vitro and in vivo GBM models developed from Aim 1.

AIM 4: To systematically identify factors that may predict response in pre-clinical animal

GBM models.

Materials and Methods

This chapter describes general procedures and materials used for this thesis. Variations are mentioned within each section with reference to the specific chapter it was used in.

Ex vivo

Glioblastoma tumour samples are taken from the Prince of Wales hospital. Tumour samples were promptly collected after surgery. The tumour tissue was manually dissected to remove any visible blood vessels and necrotic tissues. The tumour tissue was portioned for snap-freezing, formalin-fixed paraffin embedding (FFPE) and RNA extraction. Snap- frozen tissue were used for the development of patient-derived cell lines (PDCL) and xenografts (PDX). This was also used for DNA extraction, which was used whole genome sequencing (Garvan Institute). FFPE was used for immunohistochemistry to detect for specific protein. Sections of the FFPE block were sent to CARIS® Life Sciences for molecular profiling. All details are discussed below.

All experiments conducted on human clinical samples had the required ethics approval,

HREC reference number 10/121, granted by the South Eastern Sydney Illawarra Area

Health Service Human Research Ethics Committee.

In vitro

Cell culture

Cell line development

A portion of the tumour tissue was placed in a gentleMACS™ tube with Accutase® solution (Sigma-Aldrich, USA; Cat. No. A6964). The tissue was gently dissociated using the gentleMACS™ Dissociator (Miltenyi Biotec, Australia). Prior to dissociation of the tissue, necrotic and haemorrhagic tissue were removed as much as possible. After dissociation, trypsin inhibitor (Sigma-Aldrich, USA; Cat. No. T9128) was added to neutralise the proteolytic and collagenolytic effect of the Accutase® solution. Tissue bits were passed through 100µm FalconTM cell strainer (Corning Inc., USA) to ensure a single cell suspension is collected for culture. The cell suspension was centrifuged at 1000 RPM

for 5 minutes at 20˚C. Cell pellets were resuspended in pre-warmed RHB-A medium

(Clontech Laboratories Inc., Sweden) supplemented with 20 ng/µl human Epidermal growth factor (Sigma-Aldrich, USA; Cat. No. E9644) and 20 ng/µl human Fibroblast growth factor – Basic (Sigma-Aldrich, USA; Cat. No. F0291), from here on now referred to as culture media. The cell suspension was seeded into Nunc™ cell culture flasks (T25,

T75, or T175) with filter caps (Thermo Fisher Scientific, USA). The cell culture flasks were previously coated with BD Matrigel™ Basement Membrane Matrix (BD

Biosciences, USA; Cat. No. 354234) diluted at 1:100 in ice cold 1x Dulbecco’s

Phosphate-Buffered Saline (1x D-PBS) (Thermo Fisher Scientific, USA). Containers were incubated at 37˚C for at least 2.5 hours to allow the Matrigel™ coating to set. The

o seeded cells were left undisturbed in a 37 C, 5% CO2 incubator (Thermo Fisher Scientific,

USA) for 5 days before the media was changed to allow single cells to attach as a monolayer.

Cell maintenance

Patient-derived glioblastoma cell lines (G54, G57 and G89) were developed in-house following the procedure described in Section 2.2.1.1. RN1 is a patient-derived cell line that was kindly donated by Dr Bryan Day from the QIMR Berghofer Medical Research

Institute. All GBM PDCLs were maintained in culture media. Culture media was replaced

2 to 3 times per week. GBM PDCLs were subcultured once 80-85% confluence was reached. Supernatant media was discarded, and the cells briefly washed twice with pre- warmed (37˚C) 1x D-PBS. After washing, the cell monolayer was then trypsinised by adding two millilitres of Accutase® solution (Sigma-Aldrich, USA; Cat. No. A6964) followed by incubation for 5 minutes at 37˚C. Trypsinisation was neutralised by adding trypsin inhibitor at a 1:2 dilution. The cells were then centrifuged at 1000 RPM for 5

minutes at 20˚C. The cell pellet was gently resuspended in pre-warmed (37 ˚C) culture media making sure to disperse the cell pellet into a single cell suspension. The amount of culture media used to resuspend the cell pellet depended on the needed seeding concentration.

Cryopreservation techniques

Cell pellets were collected following the same subculturing procedure as described in

Section 2.2.1.2. and resuspended in freezing media instead of culture media. The freezing media was prepared by adding 10% dimethyl sulfoxide (DMSO) (Sigma-Aldrich, USA;

Cat. No. D8418) to the culture media and allowing it to cool to 4°C before use. After resuspension, the media was aliquoted into 2ml cryovials and stored in -80°C freezer overnight. The vials were transferred to the vapour-phase liquid nitrogen tank for long- term storage.

Cryopreserved cells were revived by briefly thawing out the vial in a 37˚C water bath.

The thawed cell suspension was resuspended in 10 ml pre-warmed culture media. The cell suspension was centrifuged at 1000 RPM for 5 minutes at 20°C. The cell pellet was resuspended in pre-warmed culture media and reseeded into freshly Matrigel™-coated

Nunc™ cell culture flasks. The cells were allowed to attach for 24 hours before changing the culture media.

Trypan blue cell counting

Cell counting was performed to determine appropriate seeding densities for each experiment. After the cells were resuspended following the procedure described in

Section 2.2.1.2, 15µl of the cell suspension was mixed with 15µl of 0.4% trypan blue. 10

µl of the mixture was then loaded into each of the chamber ports (2x) of the disposable haemocytometer. Cell numbers and viability were determined using the Countess™ Cell

Counter (Invitrogen, USA). Seeding densities were calculated using the live cell count.

The cell counting procedure was performed for every sample.

Seeding density optimisation

To optimise seeding density, cell counts were obtained from each cell line following the procedure described in Section 2.2.1.4. The cells were seeded in flat-bottom 96-well plates at a range 500 to 20,000 cells in 100µl in triplicates. The plate was then allowed to incubate following standard conditions. Cell viability was assessed by MTS assay as described below in Section 2.2.3.

Cell proliferation assay

Cell viability was assessed with CellTiter 96® Aqueous One Solution Cell Proliferation

Assay (Promega, USA). This assay makes use of a reagent which contains 3-(4,5- dimethylthiazol-2-yl)-5-(3-carboxymethoxyphenyl)-2-(4-sulfophenyl)-2H-tetrazolium, inner salt (MTS) and phenazine ethosulfate (PES). Viable cells reduce the MTS tetrazolium compound into formazan which can be measured by absorbance at 490 nm.

For a 96-well plate assay, each well contained a total volume of 200µl of media. At the time of the assay 90µl of media was removed before adding 20µl of the MTS reagent.

The plate was incubated at 37˚C with 5% CO2 for at least 30 minutes. SpectraMax® M2

Multi-Mode Microplate Reader (Molecular Devices, USA) was used to measure absorbance at 490 nm and 650 nm. Absorbance was measured at 650 nm to account for the background. Data was analysed using SoftMax® Pro 6 Data Acquisition & Analysis

Software (Molecular Devices, USA).

For analysis, background reading was subtracted from the absorbance reading from each well. All treatment groups were normalised to the control and was reported as a percentage of the control.

Treatments

Pharmaceuticals

Pharmaceutical compounds listed in the Table 2.1 below were purchased and resuspended according to manufacturer’s condition where it was applicable. Topotecan and irinotecan were used in Chapter 4. Temsirolimus and PENAO was used in Chapter 5. TMZ and

ABT-888 were used in Chapter 6.

Table 2.1. A summary of pharmaceutical agents used in this thesis.

Compound Manufacturer Catalogue Stock solution Name Number

Topotecan Sigma- T2705 47.46mM in 100% DMSO Aldrich, USA

Irinotecan Sigma- I1406 80.23mM in 100% DMSO Aldrich, USA

Temsirolimus Abcam, UK Ab141999 25mM in 100% DMSO (CC1779)

TMZ Sigma- T2577 160.96mM in 100% DMSO Aldrich, USA

PENAO Dr Phil Hogg NA NA & Dr Pierre Dilda

ABT-888 Abbvie, USA NA 10.23mM in 100% DMSO

4-(N-(S-penicillaminylacetyl)amino) phenylarsonous acid (PENAO) was kindly provide by Dr Phil Hogg (University of Sydney) and Dr Pierre Dilda (UNSW Sydney). To make the stock solution, PENAO was dissolved in titration buffer (Table 2.2) and warmed in a

37˚C water bath for 2 hours. The solution was filter sterilised using a 0.22 µm filter.

Resuspended PENAO was stored at 4˚C for a maximum of 2 weeks. PENAO concentration was determined by titration each time a new stock solution was prepared.

Table 2.2. Titration buffer preparation*.

Chemical Concentration Weight

NaCL 0.14M 4.09g

Hepes 20mM 2.38g

Glycine 20mM 0.75g

EDTA 1mM 0.15g

*Made up in 500 ml MilliQ H2O (pH 7).

PENAO titration

Titration is performed to determine the concentration of the PENAO stock solution prepared following the procedure discussed in Section 2.2.4.1. Titration solution was prepared as shown in Table 2.3. PENAO 20x diluted solution was added into a 96-well plate as shown in

Table 2.4 for further dilution in titration buffer. 10µl of solution B (Table 2.3) was added into all the wells except in the blanks. The plate was placed in the plate shaker for 20 minutes. After which, 19.8mg of 5,5′-Dithiobis (2-nitrobenzoic acid) (DTNB) (Sigma-

Aldrich Pty. Ltd., USA; Cat. No. D8130) (19.8mg) was dissolved in 1ml of DMSO. 5µl of the DTNB-DMSO solution was added in all wells. The plate was further incubated in room temperature for another 20 minutes on the plate shaker before reading the absorbance at 412 nm using the SpectraMax® M2 Multi-Mode Microplate Reader

(Molecular Devices, USA). Data was analysed using SoftMax®Pro Data Acquisition &

Analysis Software (Molecular Devices, USA). Exact concentration was calculated using

Microsoft Excel.

Table 2.3. Dilutions of reagents used for titration.

Chemical/Reagent Dilution Factor Diluent

PENAO 1:20 Titration buffer

DMP (solution A) 1:100 DMSO

DMP-DMSO (Solution B) 1:200 Titration buffer

Table 2.4. Dilution of 20x diluted PENAO in a 96-well plate format.

Well Blanks 1 2 3 4 5 6 7 8

Titration Buffer (µl) 195 185 184 183 182 180 177 175 170

PENAO 20x (µl) 0 0 1 2 3 5 8 10 15

Radiation

Radiation was administered at a dose ranging from 1 to 8Gy using the X-Rad 320

(Precision X-ray, USA). Endpoint experiments were performed according to specific

experiments. Radiation was used in Chapter 6 to irradiate plated cells and mice for treatment.

IC50 Determination

Cells were seeded in 96-well plates coated with Matrigel™ according to optimised seeding densities. The cells were then allowed to attach for 24 hours before treatment is initiated. A wide range of drug concentrations was used to determine the IC50 for each

GBM PDCL for each drug treatment alone or in combination. IC50 is the half-maximal inhibitory concentration of each drug. (Figure 2.1).

Figure 2.1. Drug treatment in 96-well format.

Drug treatments were done using these templates. (A) represents the template used for assessing individual IC50 of two drugs. (B) represents the template used for 2-drug combination treatment. White circles represent evaporating wells filled with 200µl PBS. Pink circles represent blank wells without seeded cells filled with media for background reading. Circles filled with colour gradients represent dose ranges from 0 to highest concentration. Green gradient – drug A. Purple gradient – drug B. Blue gradient – combination of drug A and B.

DNA

DNA was extracted from whole tumour tissue and matching cell pellets of GBM PDCLs.

DNA extraction was performed using the QIAmp® DNA Mini Kit (Qiagen, Germany) following the manufacturer’s instructions. For whole tumour tissue, the tissue was homogenised using a 23G syringe needle while suspended in the lysis buffer provided in the kit. DNA concentration was quantified using the NanoDrop® ND-1000 spectrophotometer (Thermo Fisher Scientific, USA). The sample was stored in -80˚C for later use.

Agarose gel electrophoresis

Amplified DNA products were separated in a 2% agarose gel to evaluate the quality of the designed primers. The gel was prepared in Tris-borate-EDTA (TBE) buffer. The mixture was microwave-heated for 2 minutes. Once all agarose gel powder was homogenously dissolved, GelRed™ nucleic acid gel stain (Biotium, USA; Cat. No.

41003) was added to the solution at a dilution of 1:10000. GelRed™ is a non-toxic and non-mutagenic stain used to visualise the nucleic acid in DNA. The solution was poured into a horizontal gel electrophoresis mould with a holding tray, carefully removing any existing bubbles before inserting the plastic comb. The solution was allowed to solidify for at least 30 minutes. The plastic comb was then removed, and the holding tray with the solidified agarose gel was fully submerged in 0.5x TBE in an electrophoresis chamber.

Five microliters of amplified DNA sample were placed into each well. Two microliters of GeneRuler™ 100bp DNA Ladder (Thermo Fisher Scientific, USA; Cat. No. SM0243) was placed in the first well of the agarose gel. Electrophoresis was run for 20 minutes at

100 V. DNA in the agarose gel was visualised using the Gel Doc™ XR Documentation

System (Bio-Rad, USA) and the QuantityOne Software (Bio-Rad, USA).

Primer design

The nucleotide sequences of POLE were selected for nucleotide alignments. Exons that included the single nucleotide variant from the whole genome sequence was selected for primer design. Details are shown in Table 2.5.

Table 2.5. Primers used to detect POLE mutation in Exon 30.

ID Forward Tm (˚C)

301 GCAGACCCCTCAGAGACAGA 57.00

302 GTGCCTTACCTGGCTGGTT 57.36

ID Reverse Tm (˚C)

301 AGGACTTCGGCCTCGTAAAG 55.00

302 GCTTCACTGGGTCTTTCCAG 57.02

Gradient PCR and PCR amplification

Twenty-five microlitres of PCR reaction mixture was prepared consisting of 100ng DNA sample, 10µM forward primers, 10µM reverse primers and KAPA Taq Ready Mix with dye (Sigma-Aldrich, USA; Cat. No. KK1006) (Table 2.6). The annealing temperature

(Tm) was optimised for the reaction. A 10˚C temperature range was used, starting at 3˚C lower than the lowest Tm of primers of interest. DNA samples were amplified using the optimised Tm (58.4˚C) with the primers in Table 2.5. DNA products were electrophoresed to check for quality before they were sent out for Sanger Sequencing to the Australian Genome Research Facility (Figure 2.2).

Table 2.6. PCR reaction mixture

Reagents Concentration/dilution Amount (µl/reaction)

KAPA Taq 1:2 12.5

Sterile H20 (top-up) ‐ 8.5

Forward Primer 10µM 1

Reverse Primer 10µM 1

DNA sample 100ng 2

Total 25

Figure 2.2. Gel electrophoresis of PCR products

Image shows amplified DNA products of PDCLs G89 and G244. Legends: Ld – ladder, Neg. – Negative, c – cell culture, t – tumour tissue.

Protein

Protein extraction

The cell lysis buffer was prepared in MilliQ water by mixing the ingredients listed in

Table 2.7. 10µl (1:10) of Protease inhibitor (Roche, Switzerland) and 200µl (1:5) of phenylmethylsulfonyl fluoride (PMSF) (Roche, Switzerland) was added in 1000µl of lysis buffer. Prior to lysis, cells were washed with ice cold PBS (2x) before adding the cell lysis buffer. The media suspension and PBS were collected in a labelled falcon tube and centrifuged at 1000 RPM for 5 minutes at 4˚C. The supernatant liquid was then discarded leaving a cell pellet. Concurrently, cell lysis buffer was added (150µl for a fully confluent T75 flask, 25 – 30µl for a one well in a 6-well plate) onto the cell monolayer, the cells were then scraped with a cell scraper to remove the cells from the flask surface and to aid in disrupting the cells. The resulting mixture of cells and lysis buffer were added to the cell pellet previously collected. This were transferred into a 1.5ml Eppendorf tube and left to incubate on ice for 1 hour. The cells were subjected to sonication using the Bioruptor® Pico (Diagenode, Belgium) sonication device.

Table 2.7. Cell lysis buffer preparation.

Components Reagent Final Concentration

Cell lysis buffer Tris-Cl (pH 7.4) 10 mM

NaCl 100 mM

EDTA 1 mM

NaF 1 mM

Na4P2O7 20 mM

SDS 0.1%

Sodium deoxycholate 0.5%

Triton X-100 1%

Glycerol 10%

Protease inhibitor cOmplete™, Mini, 1 tablet in 1.5ml MilliQ EDTA-free protease water inhibitor

PMSF phenylmethylsulfonyl 10 mM fluoride

Protein quantification

Protein quantification was performed using the Pierce™ BCA Protein Assay Kit (Thermo

Fisher Scientific, USA; Cat. No. 23227). Briefly, protein standards were initially prepared by serially diluting (1:2) the BSA stock solution (2 mg/ml) in the lysis buffer (Table 2.7) until a final concentration of 0.0625 mg/ml was reached (Table 2.8). Samples were diluted

1:10 before assay. A working reagent was made by diluting and mixing copper sulphate in BCA reagent A (1:50). 200µl of the working reagent was then added into each working well of a 96-well plate. 10µl of the standard solution and samples were then added into the wells (Figure 2.3).

Table 2.8. Dilution of BSA stock solution for standard curve determination.

Dilution Final concentration (mg/ml)

A Add 50ul of BSA stock solution to 50ul of cell 1.00

lysis buffer (e.g. RIPA buffer) and mix.

B Add 50ul of A to 50ul of cell lysis buffer (e.g. 0.50

RIPA buffer) and mix.

C Add 50ul of B to 50ul of cell lysis buffer (e.g. 0.25

RIPA buffer) and mix.

D Add 50ul of C to 50ul of cell lysis buffer (e.g. 0.1250

RIPA buffer) and mix.

E Add 50ul of D to 50ul of cell lysis buffer (e.g. 0.0625

RIPA buffer) and mix.

Figure 2.3. BCA assay template.

Blank wells and BSA standards are placed in each plate when multiple samples are exceeding the 96-well format. Green gradient represents highest BSA concentration (dark) to lowest (light).

Protein loading

Protein lysates were defrosted in ice, vortexed and centrifuged at 14,000 RPM for 10 minutes. Laemmli buffer (Bio-Rad, USA; Cat. No. 1610737) was prepared as detailed in Table 2.9 in a fume hood. An equivalent of 5µg or 20µg protein lysate, where applicable, was transferred into a fresh Eppendorf tube with 1:4 dilution of Laemmli buffer and topped up with sterile water to reach a total volume of 20 – 40µl. The final solution of protein lysates with other reagents were mixed in an Eppendorf MixMate®

(Eppendorf, Germany) at 1500 RPM for 1 minute. After mixing, the solution was heated for 5 minutes at 95˚C using an Eppendorf Thermostat™ (Eppendorf, Germany), and cooled in ice thereafter. Running buffer (1x) was prepared prior to setting up for gel electrophoresis (Table 2.10).

Table 2.9. Laemmli buffer.

Reagent Amount (µl)

4 x Laemmli Buffer 190.00

Mercapthoethanol 10.00

TOTAL VOLUME 200.00

Table 2.10. Running buffer.

Reagent Amount (ml)

MilliQ water 900.00

tris/glycine buffer/SDS running buffer 100.00

TOTAL VOLUME 1000.00

Bio-Rad 4-15% Mini-PROTEAN® TGX™ precast gels (Bio-Rad, USA; Cat. Nos.

4561085, 4561084) were placed in gel holders and placed in a Mini-PROTEAN® Tetra

Vertical Electrophoresis Cell (Bio-Rad, USA; Cat. No. 1658004) buffer dam. The buffer dam was filled with 1x running buffer at designated levels (2 or 4 gels). Each well in the pre-cast gel was washed with 1x running buffer prior to protein loading. Processed protein lysates were centrifuged at 14000 RPM for 30 seconds. Supernatant was extracted and loaded into the wells, along with 5 to 10μl of Precision Plus Protein™ Kaleidoscope™

Prestained Protein Standards (Bio-Rad, USA; Cat. No. 1610375). Electophoresis was run at 100V for 1 hour.

Wet transfer

Transfer buffer (1x) was prepared as detailed in Table 2.11. Foam pad, filter paper and

Amersham™ Hybond™ ECL Nitrocellulose Membrane (GE Healthcare, UK) was soaked in cold transfer buffer prior to use. The gel was removed from its plastic mould and sandwiched in foam pads, filter paper, and membrane. The sandwiched gel and membrane were placed in a gel holder cassette and transferred in an electrode assembly.

Making sure that the gel is positioned in the correct orientation. The transfer process can sometimes create too much heat that it denatures the sample proteins, hence the buffer tank was placed in an esky filled with ice to cool down. The transfer was done at 100V for 2 hours and 30 minutes.

Table 2.11. Transfer buffer.

Reagent Amount (ml)

MilliQ water 700.00

10x tris/glycine transfer buffer 100.00

Methanol 200.00

TOTAL VOLUME 1000.00

Antibody staining

Protein transfer was validated by staining the membrane with Ponceau S Staining

Solution (Sigma-Aldrich, USA; Cat No. P7170). Staining solution was left on the membrane until pink bands become visible. The stain was removed by washing with cold

1x tris-buffered saline with Tween® 20 (Sigma-Aldrich, USA; Cat No. P9416) prepared as detailed in Table 2.12.

Table 2.12. Tris-buffered saline with Tween20 (TBST).

Reagent Amount (ml)

MilliQ water 900.00

10x Tris-buffered saline 100.00

Tween 20 2.00

TOTAL VOLUME 1000.00

Once completely clean, the membrane was completely submerged in blocking solution at room temperature for 3 hours. Blocking solution was made up with 5% skim milk in 1x

TBST. After blocking, the membrane was incubated with the primary antibody in blocking solution overnight at 4˚C. The membrane was washed five times for five minutes each prior to incubating with secondary antibody. Incubation with secondary antibody was done at room temperature for 2 hours. Antibody dilution differed depending on probing antibodies used (Table 2.13). For the housekeeping gene, EIF4E was used was loading control.

Table 2.13. Probing antibodies used within experiments.

Antibody Type Dilution Expected Mol Wt. Company Cat. No. Chapter

Primary antibodies

TOP2A Ms mAb 1:1000 170 kDa Abcam, UK ab180393 4

TOPO1 Ms mAb 1:500 91 kDa Abcam, UK ab58313 4

49 PTEN Rb mAb 1:500 47 kDa Abcam, UK ab154812 4

p-MTOR (S2448) Rb mAb 1:1000 250 kDa Abcam, UK Ab109268 6

p-S6K1 Rb mAb 1:1000 75 kDa Abcam, UK Ab126818 6

Mre11 [12D7] Ms mAb 1:500 79 kDa Abcam, UK Ab214 7

Rad50 [13B3/2C6] Ms mAb 1:500 150 kDa (146 kDa) Abcam, UK Ab89 7

eIF4E (C46H6) Rb mAb 1:1000 25 kDa Cell Signaling 2067 4 - 6 Technology (CST)

Secondary antibodies

Anti-Rabbit (Polyclonal HRP- 1:2000 - DAKO P0449 4 - 6 Goat) conjugated antibody

Anti-Mouse (Polyclonal Goat) HRP- 1:1000 - DAKO P0447 4 - 6 conjugated 50 antibody

Legends: Ms – mouse; Rb – rabbit; HRP – horseradish peroxidase

Membranes were washed with TBST 5x for 5 minutes each after secondary antibody incubation. Image development was done with Amersham ECL Plus Western Blotting

Detection System (GE Healthcare, UK) according to the manufacturer’s instructions.

LAS-4000 (Fuji Photo Film Co.Ltd., Japan) was used to acquire the images of the blots.

ImageQuant™ TL version 8.1 (GE Healthcare, UK) was used for densitometry analysis.

Apoptosis assay

RN1, G54, G57 and G89 GBM PDCLs were seeded in Matrigel™-coated 6-well plates and was incubated for 24 hours. The following day the cells were treated with specific drug concentrations and incubation period according to specific experiments. Fluorescein isothiocyante (FITC)-conjugated Annexin V and propidium iodide (PI) staining kit

(Roche, Switzerland) was used to measure apoptotic cell death. In brief, the cells were harvested by collecting the supernatant, washing the cell monolayer with warmed PBS and trypsinization as described in Section 2.2.1.2. Trypsinization was inhibited by adding media 4x the amount of trypsin. The supernatant, PBS wash and trypsinization solution are all collected and centrifuged at 1000 RPM for 5 minutes at 20˚C. The supernatant was discarded after centrifugation. The cell pellet was stained with Annexin V-FITC and/or

PI diluted in incubation buffer (1:50) by following the manufacturer’s protocol. Cellular apoptosis was quantitively measured by flow cytometry using the BD FACScanto™ II

(BD Biosciences, USA).

Colony formation assay

PDCLs were plated in 6-well plates coated with BD Matrigel™ Basement Membrane

Matrix (BD Biosciences, USA; Cat. No. 354234) as described in Section 2.2.1.1. Cells were allowed to attach overnight. The cells were treated with DMSO (control) or ABT-

888 (10 μM) in the culture media. Radiation was delivered using a self-contained X-ray

system (X-RAD 320). Plates were incubated at 37˚C with 5% CO2 for 14 days undisturbed. Supernatant was gently removed by suction. The cells were gently washed with PBS followed by staining and fixation with 0.5% crystal violet dissolved in distilled water and further diluted in methanol (1:1) for 15 minutes. Stained colonies consisting of

>50 cells were counted and the number was recorded. Plating efficiency was calculated as the number of colonies counted divided by the number of cells seeded and normalized to the average plating efficiency of untreated samples. The average of these values were reported as “percentage of cells survived compared to the control.”

Bioenergetic profiling

Seahorse Assay media

Unbuffered phenol red-free media was prepared prior to the start of the assay by dissolving 8.3g of DMEM powder (Sigma-Aldrich, USA; Cat. No. D5030) in 1L of distilled water. The pH of the solution was balanced to 7.4 at 37˚C using small amounts of NaOH and HCL. The solution was then supplemented with 1mM Na pyruvate (Sigma-

Aldrich, USA; Cat. No., P2256), 0.07 mM L-glutamine (Sigma-Aldrich, USA; Cat. No.,

G7513) and 35mM Glucose. The seahorse media was made up so that the concentration of the components previously mentioned matched that of the culture media. The solution was then filter sterilised using a 500ml Nalgene™ Rapid-Flow™ Sterile Disposable Filter

Units with PES Membrane (Thermo Fisher Scientific, USA; Cat. No. 166-0045). The solution was stored at 4˚C for later use.

Cell seeding of XF96 cell culture plate

A 96-well plate format was used for this assay using a XF96 cell culture plate. The working surface of a XF96 cell culture plate is 33-40% small than a typical 96-well plate format. This was taken into consideration when the cell culture plate was coated with

Matrigel™ following the procedure described in Section 2.2.1.1. A cell seeding density experiment was performed to determine the appropriate seeding density needed for the assay. A single-step process was followed to seed the cells in the cell culture plate. The cells were trypsinised and collected from a confluent flask as described in Section 2.2.1.1.

The cells were resuspended at the desired number in culture media at a total volume of

80µl. Each well was seeded with 80 µl of cell suspension leaving all four corner wells unseeded for background correction. All background correction wells were filled with

80µl of Seahorse assay media. Seeded cells were then allowed to settle and attach to the bottom of the plate for 1 hour in an incubator at room temperature. The cells were then allowed to grow overnight at 37˚C with 5% CO2.

Sensor cartridge hydration for the XF96 Assay

200 µl of the XF96 calibrant pH 7.4 (Seahorse Bioscience, USA) was placed in all the wells of a 96-well utility plate (Seahorse Bioscience, USA). The sensor cartridge was then placed on top of the utility plate and incubated at 37˚C without CO2.

Mitochondrial Stress test

The mitochondrial stress test enables the assessment of the mitochondrial function of cells and their ability to respond to stress. The metabolism of the cells is agitated by sequentially adding Oligomycin, BAM15, Antimycin A and Rotenone. A stock solution of each compound was made up in 100% DMSO. Each compound was diluted in seahorse assay media and were made up to a total content of 1% DMSO. All compounds were made up to 10x more concentrated than the desired final concentration for loading. The compounds were kept on ice until further use. Compounds were made fresh for each assay in a total volume of 2ml.

Table 2.14. Dilution of compounds used for the mitochondrial stress test.

Compound [Stock] (mM) [Final] (µM) Dilution Factor

Oligomycin 1 1 100

BAM15 (uncoupler) 10 5 200

Antimycin A 100 10 1000

Rotenone 2 1 200

XF96 assay preparation

Cellular bioenergetic profile was assessed using the XF96 Extracellular Flux Analyzer

(Seahorse Bioscience, USA). Cells were seeded in a XF96 cell culture plate (Seahorse

Bioscience, USA) following a single-step procedure. The XF96 cell culture plate is similar to a typical 96-well plate format except for the seeding surface area being 33-40% smaller. The cell culture plate was initially coated with Matrigel to allow the cells to adhere and grow in an even monolayer. RN1 cells were seeded at 10,000 cells in a final volume of 80µL. Cells were allowed to settle and adhere to the bottom of the plate for 1 hour at room temperature. The plate was then incubated at 37˚C with 5% CO2 for 24 hours. Following the 24-hour incubation period, the cells were further incubated for 24 hours with specified doses of PENAO and temsirolimus alone or in combination (Figure

2.4). On the day of the assay, the cells were washed with pre-warmed PBS twice before replacing the cell culture media with an in-house Seahorse assay media resulting to a total volume of 180µL. The cell culture plate was then incubated for 1 hour at 37˚C without

CO2. The XF96 Extracellular Flux Analyzer measures oxygen consumption rate and extracellular acidification rate. A baseline metabolic rate was first measured followed by

the subsequent addition of 1mM Oligomycin (ATP coupler), 5mM BAM15 (uncoupler), and a combination of 10mM Antimycin A (Complex III inhibitor) and 1mM Rotenone

(Complex I inhibitor). The addition of these compounds allows the bioenergetic profile of the cells to shift and allow assessment of cellular metabolism.

The cell number was then normalised by measuring the protein concentration within each well. Protein concentration was determined using the BCA assay (Section 2.2.6.2). Data analysis was performed using the recommended Wave software v2.1.3 (Seahorse

Bioscience, USA).

Figure 2.4. Template for used for seahorse assay.

Treatment wells are specifically assigned in this orientation to account for the temperature gradient observed within the Seahorse XF Analyzer chamber.

In vivo

These studies were conducted in accordance to the Animal Care and Ethics committee of

NSW Australia (15/102A and 14/117B). 6-7-week-old female Balb/c nude mice were purchased form the Animal Resources Centre (Perth, Australia). The mice were all

housed in a controlled environment in groups of 4 and were given ad libitum food and water. The animals were allowed to reach an average weight of 18g before the intracranial injection was performed. After acclimatisation, the animals were subjected to the intracranial injection of the tumour cells. Anaethesia was induced with 4% isoflurane and was further maintained at 2.5%, with an oxygen flow rate of 1L/min and 0.25L/min, respectively. A midline incision was made on the dorsal side of the head to expose the bregma of the skull. A burr hole was made 1.5mm lateral and 1mm anterior to the bregma.

The syringe needle was then injected into the caudate putamen at a depth of 3mm from the skull to deliver a total of 2 x105 cells (RN1) in a total volume of 2ul.

Tumour growth was confirmed by sacrificing animals for H&E. Once tumour growth was confirmed the mice was randomly assigned into different groups according to the experiment.

Mice were euthanised when they exhibited symptoms indicative of significant compromise to neurologic function and/or a greater than 20% body weight loss.

Treatment

PENAO and temsirolimus

Once tumour growth was confirmed, the animals were randomised into different treatment groups and treated at 62 days post-tumour induction. In Chapter 5, treatment groups for randomisation were: i) untreated control (n=6), ii) PENAO 3 mg/kg once a day through a subcutaneous pump (n=7), iii) low dose Temsirolimus 1 mg/kg once a day intraperitoneally (n=7), iv) high dose Temsirolimus 5 mg/kg once a day intraperitoneally

(n=7), v) combination of PENAO and low dose temsirolimus (n=8), and vi) combination of PENAO and high dose temsirolimus (n=8).

Drugs used for treatment were made fresh each day of administration. Temsirolimus drug preparation was as previously described (132). Briefly, temsirolimus was resuspended in

100% EtOH at a concentration of 20 mg/ml and was diluted in 5% polyethylene glycol

400 (Sigma-Aldrich, USA; Cat. No. 202398) and 5% Tween® 80 (Sigma-Aldrich, USA;

Cat. No. P6474) in PBS. The drug was administered intraperitoneally at a bolus concentration of 0.1 ml/g of body weight per mice. PENAO was prepared as previously described (133). The drug was resuspended in titration buffer (0.14mM NaCl, 20mM

HEPES, 20mM Glycine and 1mM EDTA in MilliQ water) and further diluted in sterile

0.9% NaCl for administration. PENAO was continuously administered subcutaneously via a mini-osmotic pump at a rate of 0.48 microlitres per hour (Alzet® 2002, USA). The pump was replaced every 2 weeks with a new pump and fresh PENAO for the duration of the treatment cycle (28 days).

ABT-888 and radiation

In chapter 7, similar to Section 2.3.1.1, randomization was performed once tumour growth was confirmed. Mice were randomly assigned to 4 groups; i) untreated control (n=5); ii) veliparib only (12.5mg/kg, twice daily gavage for 5 days in a 28-day treatment cycle)

(n=5); iii) radiation treatment (total of 4 Gy over 2 days) (n=5) and iv) veliparib combined with radiation (n=7).

ABT-888 (Veliparib) was given by gavage at 12.5 mg/kg twice daily for 5 days in a 28- day treatment cycle. Whole brain radiation was performed on the second day of ABT-888 administration. This was delivered using a self-contained X-ray system (X-RAD 320)

(Precision X-Ray, USA). Prior to irradiation, mice were anesthetised with isoflurane for

5 minutes before placing them in a customised lead box to shield the body and allow radiation to be delivered directly to the entire brain. The total radiation dose administered

was 4 Gy at a clinically relevant 2 Gy/fraction schedules on 2 consecutive days. Two cycles of veliparib were administered to the mice, before endpoint was reached.

Immunohistochemistry

Brain samples were collected at endpoint, formalin fixed and embedded in paraffin. 4m sections were cut and mounted on ultrafrost slides. Hematoxylin and eosin (H&E) staining was done to confirm tumour growth. FFPE mounted sections were then deparaffinised in xylene and rehydrated in PBS using an ethanol gradient. Target

Retrieval Solution (Dako, Denmark; Cat. No. S1700) was used to perform a heat- mediated antigen retrieval step at 95C for 20 minutes. Ki67 antibody (Monoclonal mouse anti-human Ki67 [MIB-1 clone] 1:100) (Dako, Denmark; Cat. No. M7240) was used for staining to detect cellular proliferation within the specimen. The secondary antibody used was a polyclonal goat anti-mouse immunoglobulins/biotinylated (1:300)

(Dako, Denmark; Cat. No. E0433).

Ki67-positive staining tumours cells were counted in 5 random fields at 20x magnification and was represented as percentage positive staining.

The hallmark of apoptosis is DNA fragmentation which can be detected using the TUNEL

Assay. TUNEL stands for terminal deoxynucleotidyl transferase (TdT)-mediated dUTP nick end labelling. TdT catalyzes the reaction of dUTP labelled nucleotides to free 3’-OH termini of single- and double-stranded DNA breaks. This process was performed using the In situ Cell Death Detection Kit, POD (Roche, Switzerland). The sections were then examined using light microscopy.

TUNEL-positive staining tumours cells were counted in 5 random fields at 20x magnification and was represented as percentage positive staining.

Meta-analysis

The meta-analysis was conducted as described in a protocol, published online on 15

October 2015. Available at http://www.dcn.ed.ac.uk/camarades/research.html#protocols.

(Appendix I).

Sources

Prior to the study performes in Chapter 5, a literature search of phase II clinical trials revealed that doxorubicin, epirubicin, etoposide, irinotecan and topotecan are the most commonly used FDA-approved topoisomerase inhibitors for the treatment of GBM. In

August 2014, we searched PubMed, Medline and Embase for the following keywords:

(glioblastoma or glioblastoma multiforme or GBM or high-grade glioma) AND

(Doxorubicin OR Epirubicin OR Etoposide OR Irinotecan OR Topotecan). The search was limited to in vivo studies with predeveloped filters (134, 135), with no language or publication date restrictions. The search was updated in July 2016.

Inclusion criteria

During the screening process, criteria for inclusion required the following information from the studies: i) a topoisomerase inhibitor used as monotherapy, ii) use of an adult high-grade glioma model, iii) intracranial or subcutaneous tumour implantation and iv) tumour volume or median survival reported as the outcome measure.

Data extraction

All data extracted were entered into the CAMARADES data manager application. Data were extracted regarding the publication (author names, year of publication, and title), intervention (drug used, dose, dose frequency, route of administration, drug delivery method, delay to treatment, and time of assessment based on day 0 of treatment), animal population (species, strain, sex, age, and number of animals per group), tumour

implantation (cell type used, site of implantation, number or volume of cells inoculated and inoculation method used) and outcome measures (tumour volume and median survival data).

The median survival and tumour volume data were estimated from graphs using a desktop ruler when these were not provided in the text.

Study quality scoring

A modified 12-item checklist adapted from previously published studies was used to assess the quality of the studies and determine possible publication bias (Table 2.15) (136-

138).

Table 2.15. 12-item checklist for assessing methodological quality and publication bias

1. Peer-reviewed publication

2. Sample size calculation

3. Randomised allocation of drug (or control) treatment

4. Blinded assessment of outcome

5. Compliance with animal welfare regulations

6. Statement of conflict of interests

7. Uniform number of cells implanted

8. Site of implantation is consistent in all animals

9. “Take rates” of implanted tumour cells is mentioned in the publication

10. Number of excluded animals must be stated with reasons for exclusion mentioned

11. Drug action justified

12. Drug-carrier justified

Network analysis

Somatic mutations observed from the patient sample, G89, was used as input for a network analysis. Network analyses was performed as described by Wu et al. using the

Reactome knowledgebase (139, 140). Reactome FIViz application in Cytoscape version

3.5.1 was used for visualization.

Data analysis

In vitro and in vivo

GraphPad Prism software was used to perform statistical analyses for both in vitro and in vivo studies. For in vitro data, statistical analysis was performed utilizing a one-way or two-way ANOVA with a Bonferroni’s post hoc test, Students t-test or log-rank (Mantel-

Cox) test, where appropriate (GraphPad Software Inc., San Diego, CA, USA). Data are represented as the mean  standard deviation (S.D.). Kaplan-Meier survival curve were made using. For in vivo data, animal survival was defined as the time taken from tumour injection until euthanasia and survival curves were established using the Kaplan–Meier estimator.

Data from in vitro drug combinations were further evaluated by calculating for the combination index based on the median effect principle. Calculation was performed using the CompuSyn© version 1.

Meta-analysis and metaregression

A summary statistic was initially performed for each individual study. For survival data, we used median survival ratio as described previously (141) and for tumour volume data we used response ratio (142). Data were then normalised via log transformation and pooled using DerSimonian and Laird random-effects meta-analysis. We weighted survival studies by the number of animals in the study as a surrogate marker for inverse variance (141).

We assessed for the presence of heterogeneity using the I2 statistic and investigated sources of heterogeneity using meta-regression. Meta-regression was performed using

‘metareg’ in STATA 12.0. Previous meta-analyses in this field have suggested a substantial degree of covariance between study design parameters, particularly the choice of tumour model (137, 138). To limit this covariance, we excluded those studies using glioma models reported in fewer than five experiments. We then performed a univariable meta-regression followed by a multivariable meta-regression where sufficient data were present; further details are described in the results section. Where multiple comparisons were made from the same dataset, we adjusted the critical p-value using Bonferroni correction. In univariate meta-regression, critical p-value was adjusted to p=0.0042 for survival data (12 comparisons) and p=0.0056 for volume data (nine comparisons); in multivariate meta-regression, critical p was set at 0.05 (five comparisons). Publication bias was assessed using contour-enhanced funnel plots, Egger’s regression test and ‘Trim and Fill’ analysis using ‘forestplot’ and ‘metafor’ R packages (143-146). Contour enhancement of funnel plots was performed to help identify cause of funnel plot asymmetry (146). All tests reported are two-sided.

Characterisation of a Patient Tumour Sample

Introduction

Due to the high heterogeneity observed in GBM, patient-derived models and molecular characterisation of patient tumour samples have been invaluable in applying targeted therapeutic approaches for treating GBM. Additionally, the use of patient-derived cell lines has faithfully recapitulated the biology of the human tumour in vitro (147, 148).

Pre-clinical GBM models

In 1968, Pontén and Macintyre from Uppsala, Sweden were the first to establish long- term tissue culture of benign and malignant glial tumours (149). Their investigation would later lead to the establishment of immortal glial cell lines – 87MG, 105MG, 118MG and

138MG. 87MG would later be known as U87MG and become one of the most commonly used cell line for human glioma research. Pre-clinical in vivo tumour models were concurrently being developed (Figure 3.1). This was motivated by the need to recapitulate the complex physiology of the human body and to navigate through the different pathophysiological mechanisms of cancer. From 1960 to 1970s, initial tumour models were developed as isografts of mouse tumours grown in immune-competent mouse strains for use in drug screening studies (150).

Figure 3.1. Major developments in pre-clinical cancer models.

This figure has been adapted from “Preclinical Mouse Cancer Models: A Maze of Opportunities and Challenges” (150).

Legends: GEM – genetically engineered mouse; NCI – National Cancer Institute; NSG – NOD scid gamma (NOD.Cg-Prkdcscid Il2rgtm1Wjl/SzJ) mouse strain; NOG - NOD/Shi- scid/IL-2Rγnull mouse strain; PDX – patient derived xenograft; GDA – GEM-derived allograft

Isograft mouse models evolved into cell line-derived xenografts and patient-derived xenografts. The first evidence of GBM patient-derived xenografts was in 1981, where

Horten et al. performed a histopathological study of malignant glial tumours in orthotopic xenografts (151). Currently, the U87 immortalised cell line is most commonly used for developing xenograft models. Immortalised cell lines are much more accessible and reliable in vitro and in vivo. However, recent studies have shown that these cell lines may have diverged from the original tumours from which they were grown (152).

At the present time, processes for culturing patient tumours have improved to retain the tumour specific phenotype and tumour initiating capacity of patient-derived cell lines

(152). Moreover, a recent study by Lan et al. demonstrated the evolution of patient

derived cell line in vivo models is independent of a changing mutational signature (153).

This study supports the use of patient-derived models for drug discoveries and possible selection of targeted therapies for clinical use.

The Cancer Genome Atlas (TCGA)

Characterisation of the molecular landscape of GBM was started, being one of the first tumours studied by The Cancer Genome Atlas (TCGA) in 2005. The TCGA GBM project currently contains 617 cases with different data types including raw sequencing data, gene expression data, data for simple nucleotide variations (SNVs) including single nucleotide polymorphisms (SNPs) and small insertion/deletions (indels), copy number variation

(CNV) data, and DNA methylation data. Studies of the TCGA GBM dataset resulted to the identification of different molecular subtypes, markers and phenotypes that are important for clinical diagnoses and treatment (25, 154, 155). In 2008, genomic characterisation of the TCGA GBM dataset resulted in the identification of tumour suppressor p53 (p53), retinoblastoma (Rb), and receptor tyrosine kinase

(RTK)/Ras/phosphoinositide 3-kinase (PI3K) signalling as the core pathways involved in the GBM oncogenic process (154). In 2010, Verhaak et al. classified GBM into four distinct molecular subtypes, namely proneural, neural, classical and mesenchymal subtypes, discussed in detail in Section 1.1.2 (25). Following this Brennan et al. (2013) defined the somatic genomic landscape of GBM through analysing the available data for the whole genome, exome, RNA sequencing, copy number, transcriptome, epigenome and proteome in TCGA.

Characterisation of the human genome has gained significant importance in clinical medicine, especially in the treatment of heterogenous cancers. The drug discovery process

in cancer therapeutics has become fundamentally reliant to the identification of molecular targets within a patient’s genome as broadly discussed in Chapter 1 (Section 1.3.3).

Aims

The overall aim of this chapter is to establish a patient-derived cell line and matching animal model. Furthermore, to perform whole genome sequencing on the tumour and its derivatives to identify viable targets for treatment.

Specifically, this chapter aims to:

1. Collect fresh tumour samples to develop a patient-derived cell line (in vitro) and

xenograft (in vivo) model.

2. Characterise the patient genome using whole genome sequencing.

3. Identify viable targets for treatments with the aid of pathway-based network

analysis.

Results

Patient tumour sample

Clinical history

A 51-year old female patient was admitted to the Prince of Wales Hospital in April 2014.

Her presenting clinical signs included headaches, nausea, and alexia. Upon magnetic resonance imaging (MRI), a mass with hyperintense signal was observed on the left temporoparietal lobe of the brain (Figure 3.2A). The patient had no previous history of brain tumour malignancy. Maximal surgical resection was performed to remove the tumour. A fluorescent dye, 5-aminolevulinic acid (5-ALA), was infused to guide the tumour removal.

The patient was treated with the standard “Stupp” protocol consisting of concurrent radiotherapy and temozolomide (TMZ) followed by 6 weeks of adjuvant TMZ. The patient’s disease remained stable for two and a half years. New lesions were identified on a MRI scan, prompting a new treatment with pembrolizumab (Keytruda®). The patient was treated with Keytruda® over a course of 4 weeks, before a MRI scan revealed a new lesion (Figure 3.2B). In September 2016, the patient again underwent surgery for removal of the recurrent lesion, which was subsequently identified as recurrent GBM. The patient was re-irradiated. Shortly after, the patient was treated with bevacizumab (Avastin®).

Several lesions distal to the original tumour bed were noted on MRI scanning. The patient was then treated with a combination of Avastin® and ABT-414, a novel EGFR inhibitor.

The patient passed away on 13th of October 2017.

Figure 3.2. T1-weighted MRI images post gadolinium infusion.

T1-weighted MRI images of patient at (A) primary tumour occurrence and (B) recurrence. Images are taken approximately 2 years apart.

Histopathology

Histopathology examination and diagnosis were performed by SEALS (South Eastern

Area Laboratory Services) Pathology Laboratory. Macroscopic examination of the brain tissue samples described the appearance of both primary and recurrent tumours to be pale grey in colour, with brown haemorrhagic areas noted in the recurrent tumour.

Microscopically, frequent mitosis was observed in both the primary (30 per 10HPF) and recurrent tumour (38 per 10 HPF). Palisading tumour necrosis and microvascular proliferation was also observed. The primary tumour cells consisted of large pleomorphic atypical nuclei with amophilic or eosinophilic cytoplasm. The recurrent tumour showed

hypercellularity, consisting mostly of small cells with cytologic atypia, closely resembling a small cell glioblastoma pattern. The microscopic appearance of the tumour cells was a typical characteristic of glioblastoma (WHO grade IV), thus confirming diagnosis. Both tumours were tested for GFAP, IDH1, p53, and MGMT methylation. The recurrent tumour was additionally tested for synaptophysin.

Immunohistochemistry analysis was performed on paraffin-embedded tissue sections of the primary and recurrent tumour. Both tested positive for GFAP and p53 protein expression. Both tumours had a 50% proliferation index after staining for the proliferative marker, Ki-67. We extracted DNA from the frozen tissue specimen of both tumours and performed MGMT methylation analysis using pyrosequencing. The primary tumour tissue sample was MGMT unmethylated, this status was retained in the recurrent tumour sample

(Table 3.1).

Table 3.1. Immunohistochemistry and MGMT methylation results of primary and recurrent tumour.

Primary tumour Recurrent tumour

GFAP Positive Positive

IDH-1 Negative Negative

p53 Positive Positive

Ki-67 proliferation index 50% 50%

Synaptophysin NA Negative

MGMT methylation 3% 1.7%

Patient-derived cell line

Tumour processing

At the time of both surgeries, fresh viable pieces of tumour were collected by a laboratory staff member. The tissue was brought back to the laboratory. A section of the tumour was preserved in formalin (PaxGene) and a paraffin block was made for future studies.

Another section was flash-frozen (cryo-preserved) in liquid nitrogen and stored at -80˚C.

Lastly, a section of the tumour was dissociated into single cells and seeded in matrigel- coated plates and grown in RBA-H media for cell growth (Figure 3.3). The full process is described in Section 2.2.1.1. Figure 3.4 shows the morphology of the patient-derived tumour cell lines at 70-80% confluency. The primary and recurrent tumour cell line will hereon be called G89 and G244, respectively.

Figure 3.3. Flow chart of tissue sample processing.

Tumour tissue sample (A) from the patient was portioned for pathology and molecular profiling (B), in vitro cell line development (C) and in vivo patient-derived xenograft (PDX) model development (D). Drug treatments are then identified based on the tumour’s molecular profile (E). Drug response are then determined in vitro (F) and in vivo (G).

Figure 3.4. Light microscopy image of G89 and G244.

The cells were dissociated and seeded at 2 x 106 cells into Matrigel™-coated cell culture flasks with 75cm2 surface area. The PDCLs were cultured in SFM, as described above, and grown as an attached monolayer culture in a 37˚C incubator humidified with 5% CO2. The images were obtained after 7 and 14 days of plating for (A) G89 and (B) G244, respectively. Magnification at 40x.

Cell line validation

Cell and tissue samples were sent to CellBank Australia for short tandem repeat (STR) profiling. STR profiling is necessary for the identification and validation of patient- derived cell lines. Original tissue samples from G89 and G244 were sent out as reference for the established cell lines. Most cell lines have intrinsic genetic instability that may cause some variation to its STR profile. Cell lines with more than 80% identical alleles to the parent tumour samples are validated as a true match (156). In this case, out of the

16 loci that was profiled, 100% of the alleles of the G89 and G244 cell lines matched the reference tissue samples (Table 3.2).

Table 3.2. STR profile of G89 and G244 PDCLs.

Sample Name G89 tissue G89 cell G244 tissue G244 cell

D3S1358 16,17 16,17 16,17 16,17

TH01 9 9 9 9

D21S11 29,33.2 29,33.2 29,33.2 29,33.2

D18S51 13,17 13,17 13,17 13,17

Penta E 9,10 9,10 9,10 9,10

D5S818 12 12 12 12

D13S317 8,13 8,13 8,13 8,13

D7S820 8,12 8,12 8,12 8,12

D16S539 9,14 9,14 9,14 9,14

CSF1PO 10,11 10,11 10,11 10,11

Penta D 10 10 10 10

Amel X X X X

vWA 14,16 14,16 14,16 14,16

D8S1179 14 14 14 14

TPOX 8,11 8,11 8,11 8,11

FGA 19,21 19,21 19,21 19,21

Patient-derived xenograft model

Dissociated cells were intracranially injected into NSG mice. Two different cell populations were used, one was taken from the tumour margin that was highly fluorescent with 5-ALA, while the second was from the core of the tumour. The mice injected with the core tumour sample (G89C) had a median survival of 184.5 days while the mice injected with the high-fluorescent margin (G89HFM) had a median survival of 204 days.

There was no significant difference between G89C (n=4) and G89HFM (n=4) (Log-rank

(Mantel-Cox) test: p=0.3557).

Figure 3.5 shows the survival curves of both G89C and G89HFM PDX models in comparison to RN1, a patient-derived model from the QIMR Berghofer Medical Research

Institute, and U87, a commercially available immortalised cell line. G89C showed significantly longer survival than RN1 (Log-rank (Mantel-Cox) test: p=0.0035) and U87

(Log-rank (Mantel-Cox) test: p=0.0044). MRI confirmed tumour growth in 50-75% of injected animals (Figure 3.6).

Figure 3.5. Kaplan Meier survival curve of G89 in comparison to other xenograft models.

Kaplan Meier survival curve comparing G89 PDX models from other in-house xenograft models, namely RN1, a patient-derived cell line, and U87, an immortalised commercial cell line. Cell lines were intracranially injected into 8 – 9-week-old female NOD SCID gamma mice. The G89 PDX model was developed in conjunction with Ms Wendy Ha and Dr Sylvia Chung’s experiment. The data for the RN1 and U87 xenograft models were taken from unpublished in-house historical data. G89 and RN1 were injected at a density of 2 x 105 in a total volume of 2µl, while U87 was injected at a density of 5 x 104 cells in 2µl.

Figure 3.6. MRI of G89 PDX model.

Representative coronal MR-images of PDX model injected with G89 PDCLs. MRI was taken at 130 days post-tumour injection. Mice were euthanised the next day. Image acquired by Dr Sheri Nixdorf.

Genomic profile of G89 and G244

Paraffin embedded tissue was sent to CARIS® Life Sciences for molecular profiling.

Details of this test will be discussed in the following chapter (Chapter 4). The patient submitted a saliva sample post-operatively and DNA was extracted. This DNA served as the control to differentiate germline from somatic mutations. Control DNA and tumour

DNA for G89 and G244 were sent to the Kinghorn Centre for Clinical Genomics for whole genome sequencing.

SNVs and small genomic indels were identified for both G89 and G244. These results were compared to the control DNA. SNVs were called using Strelka v1.1.7 and functionally annotated using Oncotator v1.8 (157, 158). Indels were annotated using the

Genome Analysis ToolKit (159). Sequence processing and annotations was performed by

Ms. Julia Yin from the Garvan Institute.

The G89 primary tumour sample had 1,346,303 somatic SNVs and 120,634 indels, while

G244 had 1,346,302 somatic SNVs and 118,658 indels. Both tumour samples had a prevalence of 421 substitutions per megabase. Figure 3.7 shows a summary of mutations found in G89 and G244 based on variant classifications. Both tumour samples could be categorised as “hyper-mutated” since both strikingly surpass the average mutation frequency in GBM tumours of less than one substitution per megabase (160).

Copy number alterations (CNAs) were quantified using Sequenza (161). The CNA profile of G89 (Figure 3.8) had low concordance with that of G244 (Figure 3.9). The copy number gains were observed in chromosome 7 for G89. Copy number gains were also observed in chromosome 7 for G244, as well as copy number losses in chromosomes 6,

10, and 13. These alterations are comparable to profiles reported in literature (21, 155,

162). Similar results were observed in an experiment performed by Ms. Sylvia Chung, comparing the mutational landscape of G89 tumour sample and its PDX (133).

Figure 3.7. Tumour mutation summary.

Summary of SNVs and indels found in G89 and G244 categorised based on variant classifications annotated using Oncotator. Top figure shows the summary for all mutations including those within the intergenic region (IGR), introns, RNA, long intergenic noncoding RNA (lincRNA), 3’ UTR, 5’UTR and 5’ flank. The bottom figure only shows a summary of mutations that are likely to be pathogenic.

Figure 3.8. Copy number alterations of G89 tumour sample.

Figure shows CNA in G89 tumour sample. Each plot represents chromosomes 1 to 22, and X (left to right, top to bottom). The x-axis represents the genomic position of the copy number, while the y-axis shows the copy number profile of each chromosome. The green dots represent normal copy numbers of 2, red dots represent gains, and blue dots represent losses. Copy number gains can be noted on chromosomes 7.

Figure 3.9. Copy number alterations of G244 tumour sample.

Figure shows CNA in G244 tumour sample. Each plot represents chromosomes 1 to 22, and X (left to right, top to bottom). The x-axis represents the genomic position of the copy number, while the y-axis shows the copy number profile of each chromosome. The green dots represent normal copy numbers of 2, red dots represent gains, and blue dots represent losses. Copy number gains can be noted on chromosomes 7, while copy number losses can be observed on chromosomes 6, 10 and 13.

Mutational signatures of G89 and G244

Most common gene mutations occurring in GBM are in PTEN, TP53, RB1, CDKN2A,

CDKN2B, EGFR, IDH1, IDH2, NF1, PIK3CA, and PIK3R1 (163, 164). Whole genome sequencing revealed 587,109 point mutations occurring in the patient genome (G89 and

G244). The majority of the somatic variants were located in the intragenic, UTR and flank regions. Four thousand one hundred sixty-two mutations (0.7%) were identified to be potentially damaging. Of all 10 commonly mutated genes in GBM, only PTEN was identified to be potentially damaging in the G89 patient tumour sample, while G244 had mutations in RB1, EGFR and PTEN.

The mutational landscape of both tumours was determined by identifying the six classes of base pair substitutions with 96 sub-classifications based on the same base pair substitution within different contexts in the DNA sequence (Figure 3.10) (165). C↔T transitions were most frequently observed than transversions (166). The analysis revealed three distinct signatures from both G89 and G244, specifically signature 1A, 5, and 16

(Figure 3.10). Signature 1 was previously reported to have high correlation with age and was related to the deamination of cytosine resulting to C>T transitions, while signature 5 and 16 were reported to be associated with high transcriptional bias for T>C substitutions

(167).

Figure 3.10. Mutational signature for G89 and G244

Figure shows 96 sub-classifications of base pair substitution (x-axis) based on their context within the DNA sequence. C>T base pair substitution was the most common variations observed. Three distinct mutational signatures were observed in both G89 and G244.

POLE and MMR genes

Unlike other cancers such as endometrial and colorectal cancers where hypermutators are more commonly studied, hypermutation in GBM is a rare occurrence. Mutations in POLE and mismatch repair (MMR) genes (MLH1, MSH2, MSH6, and PMS2) have been well reported and associated with hypermutation in both endometrial and colorectal cancers

(168, 169). For GBM, the association of hypermutation to POLE or MMR gene mutations is limited and significant results are yet to be determined. Hence, we found it important to investigate POLE and MMR genes in both G89 and G244. Somatic alterations in

POLE, MLH1 and PMS2 were observed in both G89 and G244. WGS detected a C > T transition in POLE causing a E1240K protein change in both G89 and G244. Sanger sequencing was used to validate this in cultured PDCLs to determine if the mutation is retained in culture (Figure 3.11). Alterations were also observed in the MLH1 and PMS2 genes (Table 3.3), however these were not further investigated since both genes have previously been studied (170).

Figure 3.11. Sanger sequencing result of POLE point mutation in G89 and G244.

G89 and G244 DNA were extracted from cultured PDCLs. Sanger sequencing results shown are in reverse strand. Red box shows position of POLE mutation.

Table 3.3. Alterations observed in MMR genes in patient tumour samples.

Sample Gene Chr. Position Variant Classification Variant Reference Alteration Protein type Allele change

G89 MLH1 3 37053568 Missense_Mutation SNP A G p.I219V

G244 MLH1 3 37053568 Missense_Mutation SNP A G p.I219V

G89 PMS2 7 6026708 Missense_Mutation SNP C A p.R563L 84

G244 PMS2 7 6026708 Missense_Mutation SNP C A p.R563L

To assess the association of POLE and MMR genes, we queried the TCGA GBM (Cell,

2013) dataset for this gene set. Alterations in this gene set was observed in 63 of 580 patients with sequencing data (Figure 3.12). Four gene pairs returned with significant co- occurring alterations (Table 3.4). Log odds ratio was used to determine how strong the association is between the presence or absence of each gene in the co-occurring gene pair.

Significance was calculated using Fisher’s Exact test (p = 0.05). All data was extracted from the cBio Cancer Genomics Portal (cBioPortal) (Appendix II), an open-access resource for exploring TCGA datasets (171).

Table 3.4. Significant co-occurring alterations between POLE and MMR genes.

Gene A Gene B p-value Log odds ratio Association

PMS2 MLH1 0.0242 1.65 Tendency towards co-occurrence

MSH2 MSH6 0.0001 4.66 Tendency towards co-occurrence

MSH2 POLE 0.0101 3.14 Tendency towards co-occurrence

MSH6 POLE 0.0010 3.39 Tendency towards co-occurrence

86 Figure 3.12. Oncoprint of POLE and MMR genes showing mutations and RNA expression from TCGA (Cell, 2013) dataset.

The TCGA (Cell, 2013) dataset was queried for alterations in the POLE and MMR genes. Alterations were observed in 10.9% of 580 sequenced patients (black arrowhead).

Pathway analysis of protein changing mutations

Of the 4,162 potentially damaging mutations, we found 2,801 unique genes with associated somatic mutations causing protein changes identified from the G89 genome.

A network analysis was performed using the Reactome Functional Interaction (FI) app in

Cytoscape v3.5.1 (172). A network map of all 2,801 genes are shown in Figure 3.13. Due to the complexity of the network map, we decided to screen for driver mutations within the patient genome and use this for further cluster analysis. Parson et al. previously identified 42 GBM candidate cancer genes (173). Another study by Vogelstein et al. identified 43 cancer predisposing genes and 138 cancer driver genes (174). For the purpose of this study, we combined these two sets to form a reference list of cancer driver genes (Appendix III). This was used to differentiate between the driver and passenger genes from the patient genome. From this we identified 29 possible cancer driver genes within the patient genome (Table 3.5). Of these 29 genes, SCNA9A, LRP2, COL3A1,

PTEN and ZNF497 were matched to the GBM candidate genes identified by Parson’s et al. No interactions were observed between these five genes (result not shown).

Using the list of 29 identified driver genes (Table 3.5), we performed a cluster analysis using the Reactome FI app. The cluster analysis identified five clusters, with a modularity score of 0.3254. In this context, modularity is a measure of the fraction of interactions

(edges) in the network that connects the genes (nodes) within clusters (modules) minus the expected value of the same number of interactions in a network with the same cluster divisions but random connections between genes (175, 176). The genes within the top three cluster was used to further investigate the affected Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways using the STRING database (Figure 3.14). We opted to use the STRING database as this allows the user to filter based on organism, in this case

homo sapiens (177). The affected pathways of the top three largest clusters according to the KEGG pathway are listed in Table 3.6. False discovery rates (FDR) were used to adjust the p-values to define affected pathways that were statistically significant. FDR of less than 0.05 was deemed significant.

Figure 3.13. FI Network map of all protein changing mutations.

Network map for all 2,801 unique genes with mutations causing protein changes found in G89. Network map was created using the Reactome FI app using the Cytoscape software. Cluster analysis identified 70 different clusters with at least 2 to a maximum of 148 genes within a cluster.

Table 3.5. Cancer driver genes within the patient genome

ATRX DNMT1 FGFR3 PDGFRA TET2

AXIN1 EP300 HR PMS2 TSC2

BRCA2 ERBB2 KIT PTCH1 WAS

CACNA1H ERCC2 LRP2 PTEN WRN

CCND1 ERCC5 MLH1 RNF43 ZNF497

COL3A1 FANCA NOTCH1 SCN9A

Figure 3.14. Network analysis of G89 driver genes.

A list of 29 driver genes found mutated in the G89 genome was used for the FI network analysis. The cluster analysis of the network revealed five different clusters (Cluster 1 – pink, cluster 2 – green, cluster 3 – light blue, cluster 4 – blue, cluster 5 – yellow). Each node represents a gene. Circular nodes represent genes that are found within the G89 genome. Square nodes represent linker genes that are found to be statistically connected to the altered genes in the G89 genome. Connecting lines represent the interactions between each gene. Solid lines represent directly curated interactions based on multiple pathway databases. Broken lines represent predicted interactions.

Table 3.6. Affected KEGG pathways within the patient genome determined from using the STRING database

Pathway description Observed gene FDR adjusted Matching altered genes in the protein network count p-values

Cluster 1

Pathways in cancer 5 0.00056 AXIN1, ERBB2, FGFR3, PTCH1, PTEN

MicroRNAs in cancer 4 0.00056 ERBB2, FGFR3, NOTCH1, PTEN 90 Endometrial cancer 3 0.00072 AXIN1, ERBB2, PTEN

Bladder cancer 2 0.027 ERBB2, FGFR3

Nucleotide excision repair 2 0.0329 ERCC2, ERCC5

Hedgehog signalling pathway 2 0.0329 LRP2, PTCH1

Basal cell carcinoma 2 0.0341 AXIN1, PTCH1

Adherens junction 2 0.0482 ERBB2, WAS

Cluster 2

Thyroid hormone signalling pathway 5 1.83E-07 CCND1, EP300, ESR1, HDAC2, TSC2

Cell cycle 3 0.00237 CCND1, EP300, HDAC2

MicroRNAs in cancer 3 0.00261 CCND1, DNMT1, EP300

91 Viral carcinogenesis 3 0.00395 CCND1, EP300, HDAC2

Notch signalling pathway 2 0.0106 EP300, HDAC2

Cluster 3

Fanconi anemia pathway 3 8.33E-05 BRCA2, MLH1, PMS2

Mismatch repair 2 0.0024 MLH1, PMS2

Homologous recombination 2 0.0024 BRCA2, NBN

Discussion

As a pre-requisite to the following chapters of this thesis, we aimed to develop a patient- derived model for a patient diagnosed with GBM. We also wanted to characterise the whole genome of the patient. The patient was a 51-year old female who was diagnosed with GBM in April 2014 with a recurrence after two and a half years from primary diagnosis. Both tumour samples were collected after surgical resection. PDCLs were developed for both primary (G89) and recurrent (G244) tumour. PDXs were developed and established for G89 but not for G244 due to the slow growth observed in vitro.

PDCLs from both tumours were successfully grown as a monolayer culture and maintained in serum free media. STR marker profiling of the patient tumour samples and

PDCL derivatives confirmed a 100% match. PDX for G89 (Figure 3.6) was successfully grown and showed similar growth patterns to the patient tumour (Figure 3.2) as observed by MRI. Both patient tumour and PDX model showed hyperintense tumour margins with areas of necrosis in the tumour core. Although no direct correlation can be assessed between the patient and the PDX, the recapitulation of the patient tumour in the PDX can be deduced from the similarity in its growth pattern, as well as in the survival of PDXs which was much longer than most commonly used GBM models.

The patient genome was characterised using whole genome sequencing. We chose whole genome sequencing over whole exome sequencing because we were interested in not only mutations in the exomic regions of the gene, but also copy number variations and translocations. A hyper-mutated phenotype was observed in both G89 and G244 (421 substitutions/Mb). A hypermutated phenotype has been previously shown to be associated with mutations in the exonuclease domain of the DNA polymerase epsilon

(POLE) gene in GBM and other cancer types (168, 169, 178). In the previous study by

Erson et al., four out of the 55 adult high-grade gliomas were reported to have a hypermutated phenotype and POLE mutations (178). The progression-free survival of our patient (29 months) was comparable to what has been previously reported (26.93 months) in patients with hyper- and POLE mutations (170) However, the average age of the four patients with hyper- and POLE mutation was 35.5 which is significantly younger to our patient at time of diagnosis (51-years). POLE mutational variants have been previously reported in codons 286, 411, 297 and 459 causing protein changes in high grade glioma

(168, 169, 178). POLE mutational variants observed in P286R, V411L and S297Y were found in adult high-grade glioma, while S297F and S459F were observed in paediatric high grade-glioma samples (178). P286R and V411L POLE mutant variants has been previously reported to be frequently observed in colorectal and endometrial cancers (168,

169, 179, 180). S297F has been reported to be frequent in ovarian cancers (180). Herein we observed a variant (E1240K) that has not been previously reported in GBM. The significance of POLE mutation in hyper-mutated GBM is yet to be established. G89 and

G244 were also found to have mutations on MLH1 and PMS2. Alterations in MMR genes have been previously shown to be associated with hyper-mutation (170). However, in this case, a statistically significant conclusion could not be drawn out since we are only looking at samples taken from one patient.

We attempted to perform a network analysis to identify affected pathways within the patient’s genome. Recent studies have used pathway-based analyses to identify clusters of affected pathways based on the frequency of somatic mutations within the TCGA GBM datasets (172, 176). Two independent studies by Wu et al. and Cerami et al. identified genes and pathways involved in the core GBM functional network. Wu et al. identified

five frequently altered driver genes (CDKN2A, CDKN2B, EGFR, PTEN, and TP53) shared between two GBM data sets concluding that these genes contribute to the core network of GBM (172, 176, 181). Similarly, Cerami et al. identified two major pathways,

RB1 and PIK3R1, which included the genes previously mentioned, thus iterating their involvement in the GBM functional network (176). Our analysis looked at an individual patient with a hypermutated phenotype in contrast to previous studies that used published

TCGA datasets from GBM cohorts (176, 182). A challenge we encountered was the high mutational load found within the patient’s tumour genome. Hence, it should be iterated that the network analysis only included mutations that were functionally damaging based on the observed protein changes. Additionally, we only included previously identified somatic driver mutations found in the patient’s genome that corresponded to a published

GBM candidate cancer gene list and a pan-cancer driver gene list (173, 174).

The network analysis identified three clusters, two of which were pathways of interest for targeted treatment. These pathways include genes in clusters 1 and 3 (Table 3.5). The first cluster are pathways involving PTEN, and the second cluster are pathways involving genes associated with mismatch repair and homologous recombination. Targeted therapeutic combinations were investigated based on these pathways and will be further discussed in Chapters 5 and 6.

In conclusion, we successfully developed a matched primary and recurrent PDCL that will be used for further experimentation. In addition, the PDX model of G89 was also established, allowing us to determine the survival times. Interestingly, the tumour spread observed in the PDX model was very similar to what occurred clinically. The patient’s tumour was hypermutated, with mutations in POLE-1 and MMR. Unfortunately, the high

mutational load affecting key oncogenes and tumour suppressor genes within the patient tumour made targeted therapy problematic.

Targeted Therapy Based on Molecular Profiling

Introduction

Tumour heterogeneity

Tumour heterogeneity is a hallmark of GBM, with intratumoural heterogeneity as a major contributing factor to varied treatment response and resistance to available treatments

(183). A number of studies have established the presence of intratumoural heterogeneity in GBM (183-185). Sottoriva et al. (2013) investigated the presence of intratumoural heterogeneity in 10 GBM patients with each tumour having four to six biopsy samples.

Their study showed that 60% of their samples had at least two different GBM subgroups between different biopsy samples. They also demonstrated intratumoural heterogeneity at the DNA level by showing varying levels of copy number alterations between different biopsy samples of the same tumour (184). Patel et al. (2014) demonstrated intratumoural heterogeneity through RNA-seq at the single cell level showing the presence of cell to cell variability within one patient’s tumour (185). Parker et al. (2016) corroborated these results in 14 GBM cases, with each case having two to six biopsy samples for investigation. Their study confirmed intratumoural heterogeneity by showing an observed variation in MGMT promoter methylation in 40% of the cases (epigenetic level), by showing 14 to 50% heterogeneity in gene expression of at least one gene from the base- excision repair (BER) and mismatch repair (MMR) pathway (DNA level), and by demonstrating that 43% of the cases had at least two different hierarchal cluster groups between different biopsy samples (183).

Targeted therapy

The human genome is made up of approximately 30,000 genes. Of these, 10% are known to transcribe functional targetable proteins and another 10% of the genes are known to be

disease-related. Between this two groups, there is a 2 to 5% overlap of possible drug targets (Figure 4.1) (186).

Figure 4.1. Drug target.

Out of the ~30,000 genes in the human body, only 2-5% can be potential drug targets. These genes are those that transcribe proteins that have the ability to bind to small molecules, as well as those considered disease-modifying. Adapted from “The druggable genome” by Hopkins et al (186).

Targeted therapy has been well established in other cancer types. In breast cancer, patients with high HER2/neu expression are treated with the HER2 inhibitor, trastuzumab, or with an EGFR inhibitor, erlotinib (65-68, 70). In colon cancer, patients are stratified based on

KRAS mutation which determines if they are given cetuximab (KRAS-wild type) or sorafenib (KRAS-mutant) (2, 71-74, 79). In lung cancer, a couple of molecular targets have been identified with known inhibitors such as EGFR (erlotinib and gefitinib), and

ALK (crizotinib or ceritinib) (80-82). In melanoma, BRAF V600E mutations are targeted with BRAF inhibitors such as PLX4032 (83-85). Targeted therapy in GBM has been limited to VEGF, VEGFR and EGFR inhibition with very little to poor advantages over standard therapy.

Molecular profiling

Schilsky said in a Nature perspective that “knowledge of the molecular profile of the tumour is necessary to guide selection of therapy for the patient” (61). Molecular profiling has been made possible by recent technologies such as whole genome sequencing and superimposing this information with an exhaustive annotation of databases (60).

A clinical trial was conducted in 2010 to investigate the use of molecular profiling to identify potential targets for treatment in refractory cancers. Of the 86 patients that were molecularly profiled, 77% (n=66) were treated based on their molecular profile.

Treatment was deemed beneficial if the PFS ratio (PFS of period B/PFS of period A) was

≥ 1.3. PFS of period A was defined as time to progression (TTP) after last therapy received. PFS of period B was defined as PFS while on the treatment selected through molecular profiling. Eighteen (27%) patients had a PFS ratio ≥ 1.3. Patients with PFS ratio ≥ 1.3 had an advantage of 4.7 months on overall survival in comparison to all 66 patients (Mantel-Cox log-rank, p=0.026) (187). This study concluded that molecular profiling of a patient’s tumour to identity potential targets for treatment will increase the clinical outcome for an individual patient.

Aims

The specific aims of this chapter are:

1. To acquire a molecular profile of the model developed from Chapter 3 and identify

targetable molecular markers for treatment.

2. To determine valid targeted therapies based on Aim 1.

3. To validate the molecular markers identified from Aim 2.

4. To perform a drug screening assay based on the results of Aims 2 and 3.

Results

Molecular profiling

To address Aim 1, molecular profiling was performed on the G89 paraffin sample by

CARIS® Life Sciences, which is a commercial laboratory offering molecular profiling services. The laboratory makes use of different technologies to profile the tumour sample, such as immunohistochemistry (IHC), chromogenic in situ hybridization (CISH), sanger sequencing, and pyrosequencing for MGMT methylation. The complete details of Table

4.1 is shown in the molecular intelligence tumour report (Appendix IV).

Table 4.1. Summary results of genes/biomarkers tested in the MI (Molecular Intelligence) Profile™ by CARIS® Life Sciences

Gene / Biomarker Test Result

EGFRvIII Fragment Analysis Indeterminate Sequencing

IDH2 Sanger Sequencing Wild type

EGFR NGS Without alterations

KRAS NGS Without alterations

BRAF NGS Without alterations

c-KIT NGS Variant of Unknown Significance

NRAS NGS Without alterations

PIK3CA NGS Wild type

MGMT Pyrosequencing Unmethylated

TUBB3 IHC Positive

100

RRM1 IHC Positive

PTEN IHC Positive

TOPO1 IHC Positive

TOP2A IHC Positive

TLE3 IHC Positive

PD-L1 IHC Negative

SPARC Polyclonal IHC Negative

SPARC Monoclonal IHC Negative

TS IHC Positive

PR IHC Positive

Androgen Receptor IHC Negative

ER IHC Negative

HER2/Neu IHC Negative

HER2/Neu CISH Not amplified

PGP IHC Negative

cMET IHC Negative

BRCA1 NGS Mutation undetected

BRCA2 NGS Mutation undetected

Abbreviations: NGS – next generation sequencing, IHC – immunohistochemistry.

101

Cross-validation of CARIS® gene list against Comparative Toxicogenomic Database

The CARIS® molecular profile tested 56 genes in total (Appendix V). This list was cross- validated with the publicly available Comparative Toxicogenomic Database (CTD) to confirm the validity of the gene panel for GBM. The CTD uses manually curated interactions between genes and disease to make inferred associations (188). CTD returned

19,765 genes with inferred associations to GBM which included all the genes listed on the CARIS® gene panel (Figure 4.2).

Figure 4.2. Venn diagram showing a comparison of genes screened by CARIS® and of genes with inferred association to GBM from the CTD (188).

102

Literature search to identify possible drug treatments

Based on the molecular profile discussed in Section 4.3.1, a list of potentially beneficial drug agents was shortlisted based on the CARIS® reference database (Table 4.2).

Table 4.2. Drug agents predicted to have a benefit based on molecular profile according to CARIS® Life Sciences.

Gene/Biomarker Test Results Drug agent

PGP IHC Negative Docetaxel, paclitaxel

TLE IHC Positive

TUBB3 IHC Positive

Her2/Neu CISH Not Doxorubicin, liposomal- amplified doxorubicin, epirubicin

PGP IHC Negative

TOP2A IHC Positive

TOPO1 IHC Positive Irinotecan, topotecan

ER IHC Negative Tamoxifen, toremifene, fulvestrant, letrozole, PR IHC Positive anastrozole, exemestane, megestrol acetate, leuprolide, goserelin

103

Androgen IHC Negative Goserelin, leuprolide, receptor triptorelin, abarelix, degarelix ER IHC Negative

PR IHC Positive

To address Aim 2, the drugs listed in Table 4.2 were assessed through the “rule-of-five” by Lipinski et al. The rule associates to poor absorption/permeability if: (i) the compound has five hydrogen-bond donors, (ii) the compound has more than 10 nitrogen and oxygen atoms, (iii) the molecular weight is greater than 500 and (iv) the calculated Log P (CLogP) is greater than 5 (Table 4.3) (189). In addition to the “rule-of-five”, information about the ability of the compounds to permeate the blood-brain barrier was also acquired. All information for this assessment was taken from the DrugBank database (Appendix VI)

(190). A scoping review was also performed in PubMed database to determine if the drugs are being investigated in GBM or other cancers.

Of the 18 compounds identified by CARIS® to have a beneficial effect for G89, only five compounds passed the assessment based on the “rule-of-five”, namely topotecan, letrozole, anastrozole, exemestane, and megestrol acetate. In addition, the scoping review revealed three additional compounds that have been investigated in GBM, either in pre- clinical or clinical studies, namely doxorubicin, irinotecan and tamoxifen (Table 4.3).

Topotecan, although resulting to a high probability (0.9659) of not permeating the blood brain barrier, was still included in the compounds to be tested in vitro since enough literature was gathered that suggests its efficacy in GBM (191-196). Letrozole, anastrozole, examestane and megestrol acetate were excluded due to lack of evidence of

104

its efficacy for treatment of GBM. Doxorubicin was excluded because it did not pass the rule-of five assessment, had high probability of not permeating the blood-brain barrier and the scoping review revealed poor results from phase II clinical trials (197-199).

Irinotecan, even though it did not pass the rule-of-five assessment, was added in the compounds to be tested in vitro since it had moderate permeability (0.6284) of the blood brain barrier, as well as enough literature to support its use for GBM treatment (115, 200-

209). Lastly, tamoxifen, even though had the probability (0.5838) of passing through the blood brain barrier was excluded for in vitro testing because it failed the “rule-of-five” assessment and the scoping review revealed poor results from phase II clinical trials (210-

212).

Table 4.3. Summary of rationalising compound inclusions for in vitro testing.

BBB Scoping Pass permeability Drug 1 2 3 4 review RO5? (value, (cancer types) probability)

Docetaxel N Y N Y No Neg, 0.9659 Gastric (213), lung (214), prostate (215)

Paclitaxel N Y N Y No Neg, 0.9748 Breast (216), bladder (217), cervical (218), lung (219)

Doxorubicin Y Y N Y No Neg, 0.9951 GBM (197- 199)

Epirubicin Y Y N Y No Neg, 0.9951 Breast (220, 221), lung (222)

105

Irinotecan N Y N N No Pos, 0.6284 Colorectal (223, 224), GBM (115, 200-209), lung (225)

Topotecan N N N N Yes Neg, 0.9659 Lung (226), GBM (191- 196)

Tamoxifen N N Y N No Pos, 0.5838 GBM (210- 212, 227-231)

Toremifene N N Y N No Pos, 0.7488 Breast (232)

Fulvestrant N Y Y N No Pos, 0.9217 Breast (233)

Letrozole N N N N Yes Pos, 0.9737 Breast (234- 236)

Anastrozole N N N N Yes Pos, 0.9382 Breast (237- 239)

Exemestane N N N N Yes Pos, 0.9778 Breast (240- 242)

Megestrol N N N N Yes Pos, 0.9617 Breast (243), acetate lung (244)

Leuprolide NA Y NA Y No NA Prostate (245)

Goserelin Y Y N Y No Neg, 0.8816 Breast (246, 247), prostate (248)

Triptorelin Y Y N Y No NA Breast (249), prostate (250- 252)

106

Abarelix NA Y NA Y No NA Prostate (253- 256)

Degarelix Y Y N Y No Neg, 0.953 Prostate (254, 257)

Abbreviations: Y – yes, N – no, Pos – positive; Neg – negative; NA – not available

Drug target validation by immunoblotting

Two compounds were identified for validation based on the “rule-of-five” assessment, blood-brain barrier permeability and scoping review results, namely irinotecan and topotecan. Both drugs according to the DrugBank database (190) target topoisomerase I, a protein found in both the nucleus and mitochondria of cells. To determine if this target of interest is present, whole cell protein extraction was performed from GBM PDCLs

G89 (the patient cell line developed in Chapter 3), BAH1 and WK1 (additional GBM

PDCLs included for comparison). Western blotting was performed to validate the presence of the drug targets. Expression of the target protein, topoisomerase I, in G89 was low relative to the expression observed in WK1 and BAH1, and the house-keeping protein (α-tubulin). However, presence of the target protein in G89 warranted further testing of inhibiting topoisomerase I in vitro.

107

Figure 4.3. Western blot validation of drug target.

Protein separation was performed in 4 – 15% SDS-PAGE gel (BioRad) by loading 20µg of protein lysates into each well. TOP1 mouse monoclonal antibody at 1:500 dilution (Abcam; ab58313).

Testing drugs of interest

All GBM PDCLs were seeded in 96-well plates coated with Matrigel™ according to optimised seeding densities. The cells were allowed to attach for 24 hours prior to treatment. WK1, BAH1 and G89 were all treated with irinotecan and topotecan for 72 hours (Table 4.4). Cell viability was determined by performing a MTS assay at 72 hours.

Treatment response varied between cell lines. The 50% inhibitory concentration (IC50) for WK1 was 12.98 ± 1.09µM when treated with irinotecan and 2.59 ± 1.6768 µM for topotecan treatment. BAH1 had an IC50 of 1.18 ± 1.06µM for irinotecan and 0.02

±1.20µM for topotecan. G89 showed the least sensitivity to both compounds with an IC50 of 102.30 ± 1.07µM for irinotecan and 17.5 ± 1.11µM for topotecan. Figure 4.4 shows the dose response curves for all cell lines treated with irinotecan and topotecan. For the purpose of this thesis, we assigned a cut-off of 5µM to determine drug sensitivity.

108

Table 4.4. Concentration range used for treatment.

GBM PDCL Irinotecan Topotecan

WK1 0.10 – 50µm 0.01 – 1µm

BAH1 0.05 – 10 µm 0.005 – 1µm

G89 0.10 – 500µm 1 – 50µm

Figure 4.4. Dose response curves of G89 treated with topoisomerase inhibitors.

Cell viability was determined by performing a MTS assay. Cell viability (y-axis) was plotted as percent of the control against dose concentrations (x-axis). Dose concentrations were plotted on the x-axis in log scale. Each data point represents the mean ± SE of six replicates. Dose response curves was created using GraphPad Prism v7.02.

109

Discussion

This chapter aimed to test recommended drugs based on the patient’s molecular profile as provided by a commercial biotargeting system (CARIS® Life Sciences). Based on the targets identified from the molecular profile of G89, a list of beneficial compounds was suggested from the system’s reference database (Table 4.2, Appendix IV). Most of these references were related to ovarian, breast, prostate, pancreatic, lung, colorectal, and other solid cancers. Thus, I performed a scoping review focusing on the literature based on

GBM and other high-grade glioma studies. The list of drugs was also screened based of

Lipinski’s rule-of-five criteria to assess the compound’s absorbability. In addition, the ability of the compound to permeate the BBB was also included in the criteria for inclusion for in vitro testing. Based on these three criteria, 16 of the 18 suggested compounds were excluded. Irinotecan and topotecan were identified as the most ideal treatment for G89. Aside from the positive IHC results that the CARIS® molecular profile reported, we also validated the target of interest through immunoblotting and confirmed the presence of the target in G89. Based on this result we proceeded to investigate the response of G89, as well as other GBM PDCLs (WK1 and BAH1), to treatment with irinotecan and topotecan. Drug sensitivity was assigned at a cut-off of IC50

< 5µm. G89 showed high resistance to treatment with both irinotecan (IC50: 102.30 ±

1.07µM) and topotecan (IC50: 17.5 ± 1.11µM). A possible explanation for this resistance is the low expression of the target protein identified in the western blot. The treatment was not carried out clinically. Standard therapy was instead given to the patient.

In the previous clinical trial, the use of a patient’s tumour molecular profile to identify targeted therapies specific for individual patients has shown to have benefit in improving patients’ clinical outcome (187). However, in this chapter, we have shown that this

110

approach was not applicable for G89. A few limitations were encountered. The reference database of CARIS at the time this study was being conducted lacked material specific for GBM. The panel of genes that were used for G89 were also mostly associated to other more popular cancer types (i.e. breast, ovarian, prostate, pancreatic, etc), although these were cross-validated and confirmed to be also associated to GBM (Figure 4.2). It must also be taken into consideration that although the clinical trial reported a positive result, benefit was only observed in 27% of the patients which means that most patients like G89 would have a higher probability of falling into the majority of the population that does not respond or might not even qualify for treatment based on their molecular profile.

In conclusion, commercial companies offering to perform molecular profiling for a fee produce an enormity of data, however caution must be taken. Based upon the profiling, topoisomerase I inhibitors were recommended for G89, however we showed in vitro that

G89 was in fact resistant to this treatment. More research is needed before profiling/treatment planning can be conducted.

111

mTOR and Mitochondrial Inhibition

Parts of this chapter will be submitted for manuscript publication.

112

Introduction

It has been observed that GBM cells, similar to other cancers, undergo a phenomena known as the Warburg effect, to produce energy through aerobic glycolysis (258). Taking advantage of this modification in energy production by targeting the mitochondria has been observed to be effective in previous studies (259-262).

Cellular metabolism

Non-proliferative, proliferative and cancer cells induce different pathways of cellular metabolism to produce adenosine triphosphate (ATP) (Figure 5.1). Non-proliferative cells found in differentiated tissues can efficiently produce ATP through oxidative phosphorylation, and less efficiently through anaerobic glycolysis. As a prerequisite to this, glucose is metabolised into pyruvate through glycolysis in the presence of oxygen.

This allows pyruvate to undergo further oxidation in the mitochondria via the tricarboxylic acid (TCA) cycle. This results in the production of NADH (reduced nicotinamide adenine dinucleotide (NAD+)), which is important for efficient ATP production through the electron transport chain and oxidative phosphorylation occurring in the inner mitochondrial membrane. In instances when there is low oxygen supply, pyruvate is converted into lactate and produces ATP using anaerobic glycolysis (263).

On the other hand, actively proliferating and tumour cells go through the phenomenon known as the Warburg effect, also known as aerobic glycolysis, to produce ATP. In this case, the majority of the glucose carbon is rapidly metabolised into pyruvate and converted into lactate, which will be later on excreted from the cell (263, 264). Much of this process produces ATP in the cytoplasm of the cell and bypasses pyruvate oxidation in the mitochondria for oxidative phosphorylation (265). GBM cell cultures have been previously shown to convert 90% of accessible glucose into lactate (263, 266).

113

Figure 5.1. Cellular metabolic pathway.

Normal non-proliferating differentiated cells oxidise glucose into CO2 through the TCA cycle that occurs in the mitochondrial matrix. Non-proliferating cells efficiently produce ATP through the ETC and OXPHOS in the presence of ample oxygen supply or may convert glucose to lactate when there is low oxygen supply through anaerobic glycolysis. However, proliferating and tumour cells produce ATP via glycolysis even in the presence of sufficient oxygen supply. Aerobic glycolysis causes a significant surge in glycolytic flux which precipitously produces ATP in the cytoplasm. The majority of the pyruvate produced is converted into lactate and excreted out of the cell. This process increases the regeneration of NADH to NAD+, which consequently causes the inhibition of pyruvate to be converted into acetyl-CoA for the TCA cycle (263, 265, 267).

Abbreviations: Acetyl-CoA – acetyl coenzyme A; ADP – adenosine diphosphate; ATP – adenosine triphosphate; CO2 – carbon dioxide; ETC – electron transport chain; FAD – + flavin adenine dinucleotide; FADH2 – reduced FAD; NAD – nicotinamide adenine dinucleotide; NADH – reduced NAD+; OXPHOS – oxidative phosphorylation; TCA – tricarbocylic acid

114

Cellular metabolism regulation

At the molecular level, cellular metabolism is controlled by different regulators including tumour suppressor p53 gene, phosphoinositide 3-kinase (PI3K) pathway and AMP- activated protein kinase (AMPK) pathway (264, 268) (Figure 5.2).

The PI3K pathway has been found to be a key regulator for cell survival, proliferation and differentiation; and is found to be frequently altered in human cancers, specifically

GBM (268-270). Transactivation of the phosphatase and tensin homolog deleted on chromosome 10 by p53 negatively regulates PI3K action (271). Mutations in PTEN increase the activation of PI3K. PI3K is responsible for the activation of protein kinase

B, also known as AKT, which is a strong regulator of mTOR. The activation of mTOR directly affects cell growth through protein and lipid biosynthesis by stimulating mRNA translation and ribosome biosynthesis (268). Indirectly, changes in the cell’s metabolism can be caused by mTOR through the activation of transcription factors such as hypoxia- inducible factor 1 (HIF1) (268). HIF1 activation is responsible for the transcription of genes needed to increase glycolysis in the cell (268, 272), as well as the activation of pyruvate dehydrogenase kinases (PDK) and lactate dehydrogenase A (LDHA) which consequently reduces the movement of glucose-derived pyruvate into the TCA cycle

(264, 268, 273-275). AMPK on the other hand antagonises the effects of AKT by directly inhibiting mTOR. Loss of AMPK signalling results in increased glycolytic phenotype in cancer cells (268).

115

Figure 5.2. Molecular regulation of cell metabolism.

Key regulators within the mTOR signalling pathway are PI3K and p53. Both of which control the signalling of AKT, a strong regulator of mTOR. The cascade of upstream and downstream signalling events of mTOR ultimately leads to the regulation of cellular metabolism (264, 268, 270, 276-278).

Abbreviations: 4E-BP1 – eukaryotic translation initiation factor 4E (eIF4E)-binding protein 1; AKT – protein kinase B; CoA – Coenzyme A; GLUT – glucose transporter; HIF1α - hypoxia-inducible factor 1 alpha; mTORC1/2 – mammalian target of rapamycin complex 1/2; P53 – tumour suppressor p53; PDH – pyruvate dehydrogenase; PDK1 – pyruvate dehydrogenase kinase; PGC1α – PPAR-γ coactivator 1 alpha; PIP2 – phosphatidylinositol (4,5)-biphosphate ; PIP3 – phosphatidylinositol (3,4,5)- triphosphate; PKCα – protein kinase C alpha ; PTEN – phosphatase and tensin homolog deleted on chromosome 10, Rheb – Ras homolog enriched in brain; RTK – receptor tyrosine kinase; S6K1 – S6 kinase 1; SGK1 – serum glucocorticoid-induced protein kinase 1; TSC1 - tuberous sclerosis 1/hamartin; TSC2 – tuberous sclerosis 1/tuberin ; ULK1 – UNC-51 like kinase 1; YY1 – Ying-Yang 1

116

The study of mTOR as a therapeutic target has been of interest since its discovery. The target of rapamycin (TOR) was first discovered in the 1990s from yeast and was later on purified from mammals (mTOR). mTOR forms two complexes, mTORC1 and mTORC2. Each complex has distinct cellular functions. mTOR binds to PRAS40, regulatory-associated protein of mTOR (RAPTOR), DEP domain-containing mTOR- interacting protein (DEPTOR) and mLST8/GβL to form mTORC1; and to rapamycin- insensitive companion of mTOR (RICTOR), protein observed with RICTOR

(PROTOR), DEPTOR, mSin1 and mLST8 to form mTORC2 (270, 276, 279, 280). The mTORC1 controls protein translation, cell growth, proliferation, and survival, while mTORC2 acts as an upstream activator of Akt (270). mTORC2 also activates other substrates, such as serum glucocorticoid-induced protein kinase (SGK) and protein kinase C alpha (PKCα), which are also important for cell proliferation and growth (270,

280).

As mentioned in the above section, regulation of mTOR in mTORC1 is controlled by

PDK, AKT, and AMPK. PDK activates AKT through phosphorylation leading to a cascade of signalling events downstream (269). One of which is the regulation of mTOR in mTORC1 through phosphorylation of tuberous sclerosis complex 2 (TSC2) (281).

Phosphorylation of TSC2 causes its dissociation from its stable complex with tuberous sclerosis complex 1 (TSC1) (279). TSC2 is a GTPase-activating protein of Rheb, guanosine triphosphate (GTP) and binding to Rheb consequently activates mTORC1

(281-284). Conversely, the stable complex of TSC1 and TSC2 inhibits mTORC1 by converting Rheb-bound GTP to guanosine diphosphate (GDP) (281). AMPK, additionally, directly inhibits mTOR function (268). All these processes subsequently inhibit cell growth and proliferation.

117

mTORC1 controls different cellular functions through downstream signalling of eukaryotic translation initiation factor 4E-BP1, S6K1, YY1, HIF1α and ULK1. The most described function of mTOR is in relation to the downstream effect of mTORC1 in protein synthesis by directly phosphorylating eukaryotic translation initiation factor 4E-

BP1 and S6K1 (276). In addition to this, mTORC1 is also responsible for the generation of the lipid membrane required for cell proliferation (276). mTORC1, through S6K1, controls lipogenic gene expression through the sterol regulatory element-binding protein

1/2 (SREBP1/2) transcription factors. mTOR directly regulates mitochondrial oxidation through the YY1 protein which is coactivated by PGC1α. Inhibition of mTOR leads to the inactivation of YY1 which is responsible for mitochondrial gene expression and respiration (285). mTORC1 also controls cellular metabolism and energy production by increasing glycolysis through the activation of HIF1α transcription and translation (276). mTORC1, when in abundant nutritional conditions, inhibits autophagy by phosphorylating the ULK1 – autophagy-related protein 13 (Atg13) complex. Autophagy is induced in conditions of starvation or when mTORC1 is inhibited (286, 287).

118

mTOR inhibitors

ClinicalTrials.gov have over a thousand clinical trials registered examining at the effects of mTOR inhibition across different cancer types. The most commonly investigated mTOR inhibitors in GBM are rapamycin (sirolimus), temsirolimus (CCI-779; Torisel™) and everolimus (RAD-001; Afinitor™) (Figure 5.3).

Figure 5.3. Chemical structure of rapamycin and its analogues.

Chemical structure of rapamycin and its analogues taken from DrugBank (288). Rapalogs maintain the basic chemical structure of rapamycin (sirolimus) except for the changed lactonic macrocycle. Rapalogs also have the same binding site for FKBP12 (green bracket) and mTOR (blue bracket) (289).

Sirolimus, also known as rapamycin, was first discovered in 1975. It was extracted from

Streptomyces hygroscopicus and was initially used as an antifungal antibiotic (290). It’s currently approved for prophylactic use in patients receiving renal transplant due to its immunosuppressive properties (291-293); and for treating patients with lymphangioleiomyomatosis (294). Its use as an anticancer agent was investigated in the

1980s but was discontinued due to its poor pharmacokinetic activity (289, 295). The use of mTOR inhibitors as anticancer treatments was later rejuvenated when analogues of rapamycin (rapalogs) were discovered. Rapalogs have better pharmacokinetic activity

119

and do not induce immunosuppression (289). Rapalogs, like rapamycin, form a stable complex with FK506-binding protein 12 (FKBP12) and mTOR (296). This inhibits mTOR from forming its complexes, consequently inhibiting its function. The difference between the rapalogs is in its formulation and how they are metabolised in the body. The change in temsirolimus and everolimus’ chemical structure allows these drugs to be formulated for intravenous and oral administration, respectively (Figure 5.3).

Temsirolimus is a prodrug that needs to be metabolised and converted into rapamycin to be active. Everolimus, on the other hand, does not need this bioconversion to elicit an effect (289).

Mitochondria and PENAO

The most important organelle in the cell responsible for ATP production is the mitochondria.

4-(N-(S-penicillaminylacetyl) amino) phenylarsonous acid (PENAO) is a second generation organo-arsenical compound that targets and inactivates the adenine nucleotide translocase (ANT) molecule located in the inner-mitochondrial membrane (Figure 5.4).

The inactivation of the ANT molecule is said to lead to proliferation arrest and apoptosis mediated by the mitochondria by directly affecting oxidative phosphorylation (259, 261,

297). PENAO is a more potent derivative of its first generation counterpart, GSAO (4-

(N-(S-glutathionylacetyl)amino) phenylarsonous acid), causing a 44-fold increase in anti- proliferative activity and 20-fold increase in anti-tumour activity in vivo (298). PENAO is currently in a phase I dose escalation trial for patients with solid tumours recalcitrant to standard therapy.

120

Figure 5.4. Mechanism of action of PENAO.

PENAO enters the cell through the organic anion transporting polypeptide (OATP). This consequently inhibits ANT from allowing the exchange of ATP and ADP, which is essential for oxidative phosphorylation. Multidrug resistance associated protein isoforms 1 and 2 regulate PENAO cytosolic levels by exporting it out of the cell. (261, 262, 299). Figure adapted from different publications and manuscripts (261, 262, 299).

121

Aims

The aim of this study is to combine both temsirolimus and PENAO to inhibit both glycolysis and oxidative phosphorylation. This mechanism is hypothesised to synergistically impede GBM cell proliferation and tumour growth.

1. To determine the response of a panel of GBM PDCLs to temsirolimus and

PENAO treatment.

2. To assess if the combination of temsirolimus and PENAO can elicit a synergistic

effect on GBM PDCLs viability.

3. To determine the effect of temsirolimus, PENAO and the combination treatment

on mitochondrial respiration and acid production.

4. To determine the efficacy of temsirolimus in combination with PENAO in vivo.

122

Results

In vitro

Cell viability of GBM PDCLs affected by PENAO and temsirolimus in vitro

To determine the response of GBM PDCLs to PENAO and temsirolimus, a concentration curve was set up to determine the half maximal inhibitory concentration (IC50) for each drug (Table 5.1). Three PDCLs (G54, G89 and RN1) were treated with increasing doses of PENAO (range: 0.18 to 6.6 µM) and temsirolimus (range: 6.5 to 49.2 µM) for 72 hours prior to performing a MTS assay (Figure 5.5). Each cell line responded differently to the

PENAO and temsirolimus treatment. G54 showed little response to PENAO with an IC50 of (2.33µM) compared to G89 (0.85µM) and RN1 (1.70µM). For Temsirolimus treatment, G54 and G89 showed better sensitivity (IC50: 17.42 µM and 17.43µM, respectively) than RN1 (22.07µM).

Table 5.2 summarises the extrapolated effective doses inhibiting 10, 50 and 90% cell viability based on the dose response curves shown in Figure 5.5. CompuSyn™ software was used to perform this calculation.

Table 5.1. IC50 concentrations for PENAO and Temsirolimus.

PDCL ID PENAO (µM) Temsirolimus (µM)

G54 2.33 ± 0.95 17.42 ± 3.54

G89 0.85 ± 0.30 17.43 ± 3.23

RN1 1.70 ± 0.54 22.07 ± 4.68

123

Figure 5.5. Dose response curves.

GBM PDCLs were individually treated with increasing concentrations of PENAO and temsirolimus for 72 hours prior to assessing cell viability by MTS assay. The figure shows a summary of fitted dose-response curves with 95% confidence interval (dotted lines with shaded area). Results represent five independent experiments, with each experiment having duplicate measurements.

Table 5.2. Effective doses (ED) inhibiting 10, 50 and 90% of cell population in GBM PDCLs.

ED10 (µM) ED50 (µM) ED90 (µM)

PENAO

G54 2.24 ± 0.88 3.21 ± 0.85 4.72 ± 0.83

G89 0.21 ± 0.11 1.02 ± 0.06 3.79 ± 1.63

RN1 0.95 ± 0.53 1.94 ± 0.34 4.4 ± 0.99

Temsirolimus

G54 9.94 ± 2.39 15.11 ± 1.46 23.32 ± 1.42

G89 12.33 ± 2.39 18.94 ± 3.08 29.14 ± 3.9

RN1 14.99 ± 6.12 22.30 ± 3.9 34.39 ± 3.94

124

Combination of PENAO and Temsirolimus shows variable synergism in vitro.

To determine the synergistic effect of PENAO and temsirolimus, all three GBM PDCLs were treated with both drugs for 72 hours at multiples of IC50 before performing a cell viability assay (Figure 5.6). The interaction between PENAO and temsirolimus was analysed by calculating the combination index (CI) using the CompuSyn™ software

(Figure 5.7). Dose-reduction index (DRI) was also analysed by the software. DRI was used to calculate for how much the dose of each drug was reduced when used in combination as opposed to each drug alone (Figure 5.8). Synergism of both PENAO and temsirolimus can only be noted at higher FA (fraction affected) levels for both G89 and

RN1, with RN1 showing synergism at 70% inhibitory dose (Figure 5.7). On the other hand, antagonistic effects can be observed with G54. The combination of PENAO and temsirolimus significantly reduced the doses of either drug when used together in comparison to either drug alone (Figure 5.8).

125

Figure 5.6. Cell viability assay for PENAO and temsirolimus combination treatment.

Summary of fitted dose-response curves with 95% confidence interval (dotted lines with shaded area) for GBM PDCLs treated with PENAO and/or temsirolimus. Each cell line was treated with PENAO and temsirolimus, alone and in combination, for 72 hours. MTS assay was performed to determine cell viability. Results represent five independent experiments, with each experiment having duplicate measurements.

Figure 5.7. Combination index values. calculated with CompuSyn™.

Each cell line was treated with PENAO and temsirolimus combination at multiples of IC50 for 72 hours. MTS assay was performed at endpoint to determine cell viability. Actual experimental data was used to calculate CI values using the CompuSyn™ software. CI values were plotted against affected fractions of the whole cell population. Results presented as mean ± SD of three individual experiments.

126

Figure 5.8. Calculated dose reduction of each drug when used in combination.

Drug dose reduction was calculated based on the DRI estimated from the actual experimental data using the CompuSyn™ software. Actual reduced drug concentrations were calculated by taking the inverse of the DRI and multiplying this to the actual drug concentrations (µM) that was based on actual experimental data points. Black bars represent concentration of each drug alone, while grey bars represent the concentration of each drug in combination with the other drug. The brackets underneath the x-axis shows statistically significant (Multiple t-test: p < 0.05) difference between one drug alone compared to when it is used in combination with the other drug.

127

PENAO and temsirolimus induce apoptosis in GBM patient-derived cell lines.

G54, G89 and RN1 were treated with IC50 concentrations of PENAO and temsirolimus.

Apoptosis was quantitively assessed with Annexin-PI staining and flow cytometry after

24, 48 and 72 hours of treatment. A time-dependent increase of apoptotic cells was observed in all cell lines (Figure 5.9). Combination treated cell lines had 4 to 6-fold more apoptotic cells than the control group. At 72 hours the rate of apoptosis in G54, G89 and

RN1 was 23%, 50% and 25% for PENAO only treated group; 25%, 41%, and 17% for temsirolimus only treated group; and 45%, 60% and 38% for the group treated with the combination, respectively. Although dramatic increase of apoptotic cells can be observed in the combination-treated groups in comparison to the control, statistically significant difference of individual PENAO and temsirolimus treated groups compared to the control group was not observed. Statistical significance was determined by performing a two- way ANOVA with Tukey’s multiple comparison test using GraphPad Prism version 7.02.

128

Figure 5.9. Apoptosis assay.

All three cell lines were treated with IC50 of PENAO and temsirolimus, alone and in combination for 24, 48 and 72 hours. Cells were stained with Annexin-PI and apoptosis was quantitatively analysed by flow cytometry at the end of each timepoint. Data represent mean ± SD of replicated experiments.

129

Individual effect of PENAO and temsirolimus on mitochondrial respiration in RN1 cells

A mitochondrial stress test was performed on RN1 cells treated with PENAO or temsirolimus for 2, 4, 8, and 12 hours. Oligomycin, BAM15, rotenone and antimycin were sequentially injected at the end of treatment. Basal respiration, proton leak, maximal respiration, spare respiratory capacity, non-mitochondrial consumption rate and ATP production were measured. No significant difference was observed between the different timepoints of cells treated with PENAO in comparison to the control (Figure 5.10A and

Figure 5.10C). RN1 cells treated with temsirolimus, on the other hand, showed significant decrease in oxygen consumption rate after 12 hours of treatment (p <0.0001) (Figure

5.10B). This too was reflected in a significant decrease in the spare respiratory capacity

(p=0.01), maximal respiration (p=0.01) and ATP production (p=0.03) of temsirolimus- treated cells (Figure 5.10D).

130

A B

Figure 5.10. Effect of PENAO and temsirolimus on mitochondrial respiration of RN1 cells.

Oxygen consumption rate (OCR) of RN1 cells treated with (A) 1.13 µM PENAO and (B) 20 µM temsirolimus at different timepoints (2, 4, 8 and 12 hours). Individual parameters for basal respiration, proton leak, maximal respiration, spare respiratory capacity, non- mitochondrial respiration, and ATP production for (C) PENAO and (D) temsirolimus treated cells. OCR values in (A and B) were plotted against time. The area under the curve (AUC) of (A and B) were calculated using graphpad. Using AUC values, ordinary one- way anova was used to determine if difference between timepoints were statistically significant (p-value = 0.05). For the individual parameters (C and D), multiple t-test was performed to determine significance. Results represent mean ± SD of three to four replicates.

131

Combination effect of PENAO and temsirolimus on mitochondrial respiration in RN1 cells

RN1 cells were subjected to a mitochondrial stress test after treatment with 0.5x and 1x

IC50 of PENAO and temsirolimus combination for 12 hours. As shown in Section 5.4.1.4,

PENAO did not significantly affect the OCR of RN1 cells. However, when combined with temsirolimus, the OCR of RN1 cells significantly decreased in comparison to the control (p <0.0001). Nevertheless, it should be noted that the change in OCR of the cells treated with the combination of PENAO and temsirolimus did not significantly differ from cells treated with temsirolimus only (Figure 5.11).

Drug combination affects targeted protein expression

The protein expression of phosphorylated mTOR and S6K1 post-treatment with PENAO,

Temsirolimus or combination were assessed using western blotting. PENAO treatment of

G89 and RN1 PDCLs resulted in significantly impeded phosphorylation of mTOR with a reduction of 25 and 1.7-fold respectively compared to the untreated control.

Interestingly, for G54, the expression of phosphorylated mTOR was unchanged compared to the untreated control (Figure 5.12). When cells were treated with temsirolimus, G89 and RN1 PDCLs demonstrated reduced phosphorylated mTOR expression by 2.85 and

4-fold respectively compared to the untreated control. When PENAO and Temsirolimus was combined, strong inhibition of phosphorylated mTOR was observed in all 3 cell lines.

Phosphorylated S6K1 expression was also affected by PENAO, temsirolimus and combination treatment. Expression of phosphorylated S6K1 followed the same trend as the expression of phosphorylated mTOR in all cell lines except for G54, where PENAO treatment increased phosphorylated S6K1 expression by 25% in comparison to the control group.

132

Figure 5.11. Bioenergetic profiling of RN1 cell treated with PENAO and temsirolimus combination.

(A) OCR of RN1 cells treated with PENAO and temsirolimus alone and in combination for 12 hours. (B) Area under the curve analysis of OCR and extracellular acidification rate (ECAR) measurements. (C) A comparison of different treatment groups was performed by calculating for the individual parameters based on oxygen consumption rate. Results represent mean ± SD of three to four replicates. (A and C) uses the same color legends. Legends: (ns) – not significant; (*) – p < 0.01; (****) – p <0.0001

133

Figure 5.12. Assessing mTOR and S6K1 phosphorylation by western blot in response to treatment.

Protein separation was performed by loading 5µg of protein in each lane of a 4 – 15% Mini-PROTEAN® TGX™ precast gel. Proteins were transferred to a nitrocellulose membrane by wet blotting at 100V for 2 hours and 30 minutes. (A) Membranes were probed with a rabbit monoclonal anti-mTOR (phospho S2448) antibody [EPR426(2)] (ab109268) and rabbit polyclonal anti-S6K1 (phospho T389) antibody (ab126818), both at 1:1000 dilution in 5% skim milk in TBST. Rabbit monoclonal EIF4E antibody diluted at 1:2000 in 5% skim milk was used as a loading control. (B) Densitometry analysis of Western blot was performed using ImageQuant™ TL software. Bar graphs represent the band density relative to the control group, which was adjusted to the loading control (EIF4E) (B). Representative images of duplicated experiments.

134

In vivo

Combination effect of PENAO and temsirolimus on survival in vivo.

The combinatorial effects of PENAO and temsirolimus was investigated in an orthotopic patient-derived animal model. The PDCL, RN1 was intracranially injected into the brain.

Tumour growth was confirmed by H&E sections at 62 days post tumour induction and treatment with PENAO, Temsirolimus and the combination commenced. The drugs on their own or in combination were well-tolerated by the mice. Overall survival was calculated, and a Kaplan Meier curve was constructed to compare survival times (Figure

5.13). We observed no survival advantages with any of the treatment groups when compared to the untreated control group. For control mice, the median survival was 75.5 days. Treatment with PENAO alone (3 mg/kg) conferred a 2.5-day advantage (78 days) however this was not significant (p=0.306). We tried two doses of temsirolimus (1 and

5mg/kg). The 1mg/kg dose did not result in any survival advantage (75 days; p=0.863) however the median survival for mice treated with the 5mg/kg dose was 87 days.

Although there was a survival benefit of 11.5 days, this did not reach significance

(p=0.081) when compared to the control mice as the benefit was seen in individual mice, not as a group. Two combination schedules with temsirolimus (1mg/kg and 5mg/kg) and

PENAO (3mg/kg) were tested in our mouse model. Using the low dose of temsirolimus with PENAO did not result in any survival advantages for the mice compared to the untreated control (79 days; p=0.144). The combination of PENAO and high dose temsirolimus, however, led to a statistically significant benefit on overall survival (80.5 days or a 5-day advantage) compared to the control group (log-rank test p = 0.0112)

(Figure 5.13). The combination was not statistically significant to when mice were treated with high dose temsirolimus alone. Table 5.3 summarises the median survival of each treatment group.

135

Table 5.3. Summary of median survival in all treatment groups.

Treatment group Median Survival (days) Range (days)

Control 75.5 71 – 80

P3 78 64 – 83

T1 75 67 – 79

T5 87 67 – 97

P3T1 79 66 – 85

P3T5 80.5 75 – 85

Legends: P3 – 3mg/kg PENAO; T1 – 1mg/kg temsirolimus; T5 – 5mg/kg temsirolimus

136

137

Figure 5.13. Kaplan-meier curve comparison show no significance between control and all treatment groups.

RN1 cells were orthotopically inoculated into BALB/c nude mice. Once tumour growth was confirmed by H&E staining, mice were randomly allocated into different treatment groups: control, 3 mg/kg PENAO via subcutaneous pump (P3), 1 mg/kg temsirolimus via IP injection (T1), 5 mg/kg temsirolimus via IP injection (T5), combination of P3T1 and combination of P3T5. Survival was defined as the day of euthanasia and was put in GraphPad Prism to establish a Kaplan – Meier survival curve. Statistical significance was determined by performing a log-rank (Mantel – Cox) test.

Discussion

The use of temsirolimus as a single agent therapy has been investigated in both a Phase I and II clinical trial for GBM (300, 301). PTEN loss, which is typically observed in 40 to

60% of GBM patients, has been shown to sensitise tumours to temsirolimus treatment. In the phase I clinical trial, the investigators were able to demonstrate the ability of temsirolimus to cross the blood brain barrier and inhibit mTOR in tumours (301). In the phase II clinical trial, although results were not strikingly positive, they were able to demonstrate that the drug was well tolerated and did not cause severe toxicities. The conclusion of this trial was that temsirolimus was warranted for further investigation, however it was recommended to be used in combination with other therapeutic agents

(300). In this chapter, we aimed to combine temsirolimus with PENAO, a mitochondrial inhibitor. PENAO interferes with the mitochondria by inhibiting the ANT molecule found in the mitochondrial permeability transition pore of the inner mitochondrial membrane

(302). The hypothesis behind this study was that the inhibition of mTOR would lead to a cascade of downstream events that would lead to diminished cell proliferation, or even apoptosis; and the addition of inhibiting the ANT molecule from the mitochondrial permeability transition pore would amplify this effect by decreasing oxygen consumption for energy production while increasing acid production.

In this study, we used three MGMT unmethylated GBM PDCLs (G54, G89 and RN1) with known PTEN protein expression. G54 and G89 were PTEN deficient cell lines

(Appendix VII). We treated all cell lines with temsirolimus, PENAO, and the combination of both, and assessed for cell viability and apoptosis in vitro. For temsirolimus treatment, we observed RN1 cell line that had high PTEN expression to have the highest ED50, while cell lines that had low to no expression of PTEN needed relatively lower

138

temsirolimus concentrations to elicit response (Table 5.2), thus keeping consistent with the current literature that PTEN loss evokes sensitisation to temsirolimus. All 3 cell lines responded to very low concentrations of PENAO (1.02 to 3.21µM). When combined together, temsirolimus and PENAO displayed striking synergistic effects (validated by

CompuSyn), resulting in loss of cell viability and increased apoptosis in RN1 and G89.

Interestingly, an antagonistic effect with the two drugs was observed with G54, despite being PTEN deficient. The effects of temsirolimus and PENAO on downstream phosphorylation of mTOR and S6K1 were confirmed in RN1 and G89, with expression levels significantly suppressed with monotherapy and combination therapy. No appreciable effects were observed with G54, where the combination treatment was found to be antagonistic.

Mitochondrial respiration was assessed in RN1 cells. In comparison to the control, oxygen consumption rate was not substantially affected by PENAO treatment in RN1 cells after 12 hours. Treatment with temsirolimus significantly (p<0.0001) reduced OCR.

The combination of PENAO and temsirolimus treatment, although showed significant

(p<0.0001) OCR reduction in comparison to the control, did not show any appreciable difference to the temsirolimus only treated group. ECAR measurements significantly increased in temsirolimus treated cells (p = 0.0133) and was not affected by either

PENAO treatment or combination treatment. This is in contrast to a previous published study where PENAO significantly reduced OCR and increased ECAR measurements in an immortalised GBM cell line, U87, and a patient-derived cell line, BAH1 (259).

However, differences in this study and the previous study should be noted. Previously, a

24-well format was used for the Seahorse assay, treatment regimens also differed in a

139

way that the previous study used a much higher dose of PENAO (5µM) and a much longer treatment time (24 hours) (259).

I used a well-established orthotopic mouse model where RN1 cells were implanted intracranially into the brains of Balb/c nude mice to determine if the combination of

PENAO and temsirolimus led to a survival advantage. The mouse model utilises established tumours prior to treatment rather than giving the mice treatment early at the onset of tumour injection as I was looking for the ability of the two drugs to lead to tumour shrinkage and regression. All treatments: PENAO (3mg/kg), Temsirolimus (1mg/kg and

5mg/kg) and the combinations (PENAO 3mg/kg + Temsirolimus (1mg/kg) or (5mg/kg) were well tolerated by the mice. We observed a positive signal, measured by increased survival times, in the mice treated with temsirolimus (5mg/kg) and in combination with

PENAO, however these increases did not equate to significant changes in survival compared to the control. Unfortunately, the failure of the treatment combination lies in the fact that PENAO concentrations may not have been high enough in the brain to synergise with temsirolimus. Because of its extremely short half-life, PENAO was continuously administered through an osmotic pump. Because of these pumps, the dose of PENAO could not be escalated. Whilst, brain concentrations of PENAO were not measured in this experiment, other members of the lab have shown that the levels of

PENAO reaching its tumour target have been negligible.

In summary, although exciting synergistic effects were measured when temsirolimus was combined with PENAO in vitro, this effect could not be translated to the in vivo setting.

140

Inhibiting PARP as a Therapeutic Strategy for GBM

Parts of this chapter have been published in the Journal of Translational Medicine (303).

141

Introduction

GBM and DNA repair mechanisms

DNA damage regularly occurs as a normal physiologic response of the body and when left unrepaired may lead to genomic mutations and instability (304). However, there are multiple mechanisms of DNA repair, collectively known as DNA Damage Response

(DDR) mechanisms that can restore damaged DNA. Some of the major pathways involved include BER, homologous recombination repair (HRR), non-homologous end- joining (NHEJ), nucleotide excision repair (NER), and mismatch repair (MMR). Each mechanism is induced depending on the type of the DNA damage present. Figure 6.1 summarises the different molecular targets within each repair pathway involved in specific types of DNA damage (305). Recent reviews have discussed in detail the different mechanisms and how they play in the DNA damage response mechanism (304-

306).

Figure 6.1. Pathways involved in DNA damage repair and their targets (305, 307).

Abbreviations: APE1 – apurinic/apyrimidinic endonuclease-1; PARP – poly ADP ribose polymerase; ATR – ataxia telangiectasia and rad3-related kinase; ATM – ataxia telangiectasia mutated kinase; DNA-PK – DNA-dependent protein kinase; ERCC1 – excision repair cross-complementing group 1; XP – xeroderma pigmentosa; MLH – mutL homolog; MSH – mutS homolog; MTH1 – mutT homolog-1; SSBs – single strand breaks; DSBs – double strand breaks

142

GBM has high DNA repair capabilities which is attributed to its resistance to radiotherapy and chemotherapy (307, 308). This makes the inhibition of the DNA damage response pathway an attractive target for GBM therapeutics (304). This approach has been reported to be more effective for tumours carrying a deficiency in their DNA repair mechanism.

This allows the inhibition of one DNA repair pathway which drives the tumour cells to be dependent on another, less efficient, pathway. This concept is known as synthetic lethality. This approach has been reported to be effective in killing cancer cells and leaving normal cells unharmed. Inhibiting one gene product in a DNA repair pathway would push the cancer cells to be dependent on a deficient pathway, while normal cells compensate with a full complement of DNA repair pathways (305).

PARP inhibition

PARP is a family of nuclear proteins involved in DNA damage repair. In GBM treatment, the standard therapy involves radiation therapy and/or the use of an alkylating agent, such as TMZ. As shown in Figure 6.1, radiation therapy and alkylating agents induce SSBs in the DNA. PARP activation then occurs and binds to the SSBs. PARP recruits other proteins to repair the damage DNA. Inhibiting PARP ultimately results in a stall in the repair process, secondarily causing the formation of DSBs, eventually leading to cell death (Figure 6.2).

Inhibition of the DNA repair mechanism with the use of PARP inhibitors has undergone pre-clinical and clinical investigation in other solid cancers such as breast, ovarian, rectal, prostate and lung cancer (309-315). Sensitivity to PARP inhibition has been well described in BRCA1/2 deficient breast cancer and ovarian cancer (316-319), which consequently have defective homologous recombination pathway function. Currently, five phase 3 clinical trials are on-going looking at the response of breast cancer

143

(NCT02163694, NCT02032277), lung cancer (NCT02264990, NCT02106546) and ovarian cancer (NCT02470585) to PARP inhibition in combination with other chemotherapeutic agents. Although sensitivity to PARP inhibition is known to BRCA 1/2 deficient tumours, response have also been observed in tumours with no BRCA 1/2 deficiency. Mutations in other genes cause recombination defects, hence increasing sensitivity to PARP inhibition. Some of the genes described above are PTEN (320, 321),

ATM (322, 323), PALB2 (324), CHEK2 (323, 325, 326), FANCA (325) and HDAC2

(326).

Figure 6.2. PARP DNA Repair mechanism.

Radiation or some chemotherapeutic drugs induce DNA damage through the formation of single strand breaks. PARP is then activated and initiates repair by recruiting other proteins within the BER pathway. Inhibition of PARP goes through two mechanisms of action. First is that it interferes with the repair of single strand break in the DNA. Second is that it forms a stable PARP-DNA complex thus causing double strand breaks which eventually lead to cell death. However, resistance can be observed when cells go through homologous recombination, or other less efficient repair mechanisms such as Fanconi anaemia pathway (FA), template switching (TS), ATM (ataxia-telangiectasia mutated), FEN1(replicative flap endonuclease) and polymerase β.

PARP inhibitors are used as radio- or chemo-sensitisers in the treatment of GBM (Table

6.1). A previous study has shown that the use of PARP inhibitors increase the sensitivity of, even resistant, GBM cell lines to TMZ treatment, however, this effect was not

144

observed in vivo (327). Other studies have pre-clinically investigated the radio-sensitizing effect of PARP inhibition using in vitro and subcutaneous in vivo models of GBM (328-

330). In vitro, GBM can be susceptible to the radio-sensitizing effect of inhibiting PARP due to its high proliferative characteristic and high PARP1 expression (328, 330). In vivo, the combination treatment of a PARP inhibitor and irradiation delayed tumour growth in subcutaneous GBM models and inhibited the tumour initiating function of GBM- initiating cells (GICs) in an orthotopic model (329, 330). A more recent study in 2016 investigated the effect of combining a PARP inhibitor, veliparib (ABT-888), with irradiation (IR) and TMZ in a genetically engineered intracranial GBM mouse model

(331). This showed a significant improvement in the overall survival of the mice treated with the triple combination compared to other treatment groups. However, the clinical relevance of this study may be questionable due to the compounding cytotoxic effects of the concurrent administration of ABT-888 with irradiation and TMZ.

Table 6.1. Summary of pre-clinical investigations of PARP inhibition with IR and/or TMZ.

Author Year Pre-clinical model Treatment

F. A. Dungey, et al. 2008 In vitro (U373-MG, T9G, U87- KU-0059436 + IR (328) MG, UVW)

A.L. Russo, et al. 2009 In vitro (U251) E7016 + IR (329) In vivo – s.c. model (U251)

M. Venere, et al. 2014 In vitro (primary GBM 3691, Olaparib + IR (330) 08-387, 3359)

In vivo – s.c. model (primary GBM 3691)

145

S.K. Gupta, et al. 2014 In vitro (GBM12, U251, T98G, Veliparib + TMZ (327) and GBM12)

In vivo – s.c. model (U251 and GBM12)

B. Lemasson, et al. 2016 In vivo – genetically engineered Veliparib + IR + (331) GBM mouse model (i.c.) TMZ

Legends: s.c. – subcutaneous, i.c. – intracranial

The investigation of PARP inhibition in combination with irradiation in clinically relevant GBM models is lacking. This paucity adds to the need of investigating the said combination for MGMT unmethylated GBM.

Aims

1. To determine the efficacy of ABT-888 as a monotherapy for GBM patient-derived

cell lines (PDCLs).

2. To determine the suitability of ABT-888 as a chemosensitiser for TMZ in MGMT

unmethylated GBM PDCLs.

3. To determine if PARP inhibition can be used as a radiosensitizing agent for GBM

PDCLs with unmethylated MGMT.

4. To determine if aim 3 can be effective in vivo using a GBM patient-derived

xenograft (PDX) with an unmethylated MGMT phenotype.

5. To explore possible biomarkers predictive of response to PARP inhibition.

146

Results

In vitro

Effects of ABT-888, TMZ and radiation as monotherapy on cell viability of GBM PDCLs

GBM PDCLs (RN1, G54, G57 and G89) were processed and prepared for seeding into

96-well plates as described in (Section 2.2.2). The monotherapeutic effect of ABT-888,

TMZ and radiation was assessed by treating the cells with a range of drug concentrations and radiation dose. MTS assay (Section 2.2.3) was performed after 8 days of incubation at 37˚C with 5% CO2. ABT-888 alone did not show any significant effects on G54 and

G57. RN1 and G89 only showed reduced cell viability that was significantly different from the control at a high dose of 10 µM for ABT-888. A decrease of 27% and 24% in viable cells were observed in RN1 and G89 cells treated with 10 µM of ABT-888 (Figure

6.3).

Figure 6.3. Summary results of GBM PDCLs treated with different concentrations of ABT-888.

GBM PDCLs treated with 1, 5 and 10µM of ABT-888 for 8 days. Cell viability was assessed by MTS assay. Treatment with 10µM of ABT-888 showed the most significant response for RN1 and G89 compared to control. Figure shows mean ± SD from three independent experiments. Significance was determined by 2-way ANOVA. Legends: p < 0.0332 (*), 0.0021 (**).

147

TMZ as a monotherapy was also assessed on G54, G57, G89 and RN1. The cells were treated with 25 to 600µM of TMZ and was incubated at 37˚C with 5% CO2 for 8 days.

All four GBM PDCLs did not show sensitivity to TMZ treatment at lower concentrations

(25, 50 and 100µM). G54 showed sensitivity at 200µM of TMZ, while G89 and RN1 only showed sensitivity at the highest concentration of TMZ (600µM) (Figure 6.4).

Sensitivity, for the purpose of this thesis, is defined as a reduction of 25% of cell viability compared to the untreated control group.

Figure 6.4. Summary results of GBM PDCLs treated with different concentrations of TMZ.

Dose response curves showing effect of TMZ (25 to 600µM) after 8 days of treatment on GBM PDCLs with unmethylated MGMT. Cell viability was assessed by MTS assay. No sensitivity was observed at lower TMZ concentrations in all GBM PDCLs. Figure shows mean ± SD of three independent experiments.

The effect of radiation as an individual treatment was also assessed. RN1, G54, G57 and

G89 were irradiated with 2, 4, 6 and 8Gy (Section 2.2.4.3). All cell lines showed significant response to radiation as a monotherapy. G54 and G57 showed sensitivity at 2

Gy while RN1 and G89 showed sensitivity at 6Gy (Figure 6.5).

148

Figure 6.5. Summary results of GBM PDCLs treated with different doses of radiation.

Response of MGMT unmethylated GBM PDCLs after irradiation (2 to 8Gy). MTS assay was performed to assess cell viability after 8 days of incubation. Figure represent mean ± SD of three independent experiments. Significance was determined by 2-way ANOVA. Legends: p < 0.0332 (*), 0.0001 (****).

Cytotoxic effect of TMZ in combination with ABT-888

To assess the sensitizing effects of ABT-888, RN1, G54, G57 and G89 were treated with the combination of TMZ (25 to 600µM) and ABT-888 (5µM). The cells were incubated for 8 days at 37˚C with 5% CO2 before MTS assay was performed. Response to TMZ and

ABT-888 combination was not observed at lower concentrations of TMZ (25 and

100µM). Treatment with higher doses of TMZ (300 and 600µM) in combination with

ABT-888 showed more significant effects when compared to the control in all cell lines, except G57. However, no statistically significant difference was observed between the high dose TMZ only treated group versus the groups treated with the combination of both drugs (Figure 6.6).

149

Figure 6.6. Overall summary of cell viability assay performed to assess the efficacy of ABT-888 and TMZ combination.

MGMT unmethylated GBM PDCLS treated with 5µM ABT-888 in combination with a dose range of TMZ (25 to 600µM) for 8 days. Cell viability was determined with an MTS assay. Figure shows mean ± SD of three independent experiments. Significance was determined by 2-way ANOVA. Comparisons were made to the control. Legends: p < 0.0332 (*), 0.0021 (**), 0.0001 (****).

150

Radio-sensitizing effects of ABT-888

Consequently, we looked at the response of the GBM PDCLs to the treatment combination of irradiation and ABT-888. For this experiment, G54, G57, G89 and RN1 were treated with different concentrations of ABT-888 (1, 5, and 10µM) 2 hours prior to irradiation. Irradiation was given at a dose range of 1 to 8 Gy following the procedure described in Section 2.2.4.3. After treatment, the cells were incubated for 8 days at 37˚C with 5% CO2. Figure 6.7 shows a summary of the MTS assay (Section 2.2.3) performed to assess the efficacy of the combination of irradiation and ABT-888 treatment.

G54 and G57 PDCLs significantly showed good response with radiation alone compared to control. However, looking at the groups treated with the combination of ABT-888 and radiation, the combination groups did not show any significant difference from those treated with radiation alone. G89 and RN1 showed sensitivity at the lowest ABT-888 concentration of 5µM in combination with 2Gy radiation. Increased response can be observed with increasing dose of ABT-888 and radiation.

151

Figure 6.7. Overall summary of cell viability assay performed to assess the efficacy of the combination of ABT-888 and radiation.

GBM PDCLs treated with the combination of irradiation (2 to 8Gy) with 1, 5 and 10µM of ABT-888 for 8 days. Cell viability was assessed by MTS assay. Figure shows mean ± SD from three independent experiments. Significance was determined by 2-way ANOVA. Legends: p < 0.0332 (*), 0.0021 (**), 0.0002 (***), 0.0001 (****).

Cell viability was also measured by the ability of the PDCLs to form colonies. Clonogenic assays were performed on RN1, G54 and G57 (Section 2.2.8). Cells were pre-treated with

10µM of ABT-888 two hours before irradiation (1, 2, or 4Gy) and were incubated for 10 days to allow colony formation. Colonies were counted and are presented as a percentage of the control. RN1 showed the most sensitivity to the combination treatment of ABT-

888 (10µM) and radiation (1, 2 and 4Gy), with more than 50% of colonies being inhibited.

152

On the other hand, the combination treatment did not significantly affect G54 and G57 when compared to radiation treatment alone (Figure 6.8).

Figure 6.8. Summary of clonogenic assay performed to assess efficacy of the combination of ABT-888 and radiation.

Clonogenic assays performed by Ms. Kyoko Nozue. Statistical significance was determined by Multiple t-test.

153

Combination of ABT-888 and radiation have a synergistic effect

To confirm the presence of a synergistic effect between the combination of ABT-888 and radiation, the CI was calculated based on the median-effect principle (332). The CI theorem by Chou and Talalay gives a quantitative definition to concluding if the combination of two drugs does offer a synergistic effect. By principle, a CI value greater than one denotes an antagonistic effect of the two drugs combined, a CI value equal to one is additive and a CI value less than one is synergistic. CI values were calculated for the two cell lines, RN1 And G89, that showed statistically significant results in Section

6.3.1.3. The inhibitory effect of the individual and combined treatments was entered into the CompuSyn© software along with the corresponding dosages. A dose effect curve and a median-effect plot are initially mapped out as a prerequisite to the calculation of the CI values (Figure 6.9). The median-effect plot enables the calculation of the slope (m) and the median effect dose (Dm) which are used for the calculation of the CI (333).

Dose reduction indices (DRI) were also calculated using the CompuSyn© software (Table

6.2). DRI is a measure of how many folds the dose of both drugs in a synergistic combination can be reduced to have the same effect as each individual drug (334). DRI equal to one has no dose-reduction, more than one gives a favourable dose-reduction, and less than one gives a non-favourable dose-reduction.

154

Figure 6.9. Dose-effect curves (A) and median-effect plots (B) for RN1 and G89.

The combination of ABT-888 (1, 5 and 10µM) and radiation (2, 4, 6, and 8Gy) conferred synergy in both RN1 and G89 at different dose combinations of both drugs. It can also be observed that synergy increased as the doses of both ABT-888 and radiation increased. However, it should be noted that the low dose of radiation (2Gy) and ABT-888 (1µM) produced antagonistic effects in RN1 (Figure 6.10).

155

Figure 6.10. Combination Index of ABT-888 and radiation shows synergy in reducing cell viability in GBM PDCLs.

The CI values were calculated based on the pooled average of three independent experiments. Actual CI values and degrees of synergism and antagonism are shown in Table 6.2.

Table 6.2. Combination index (CI) values for the combined treatment of ABT-888 and radiation.

Cell IR ABT- fa CI Degree of DRI RT DRI line (Gy) 888 synergism/antagonism ABT-888 (µM)

G89 2 1 0.118 0.73145 ++ 1.52771 13.0089

2 5 0.165 0.72435 ++ 1.99152 4.50007

2 10 0.238 0.60146 +++ 2.71858 4.28052

4 1 0.242 0.74745 ++ 1.37970 44.1444

4 5 0.280 0.71968 ++ 1.57772 11.6484

4 10 0.348 0.62114 +++ 1.95661 9.08667

6 1 0.337 0.80431 ++ 1.26180 84.8402

6 5 0.382 0.73861 ++ 1.44126 22.3347

156

6 10 0.424 0.68620 +++ 1.62300 1432735

8 1 0.410 0.86192 + 1.17051 131.642

8 5 0.428 0.84667 + 1.23086 29.2106

8 10 0.475 0.76647 ++ 1.40049 19.0716

RN1 2 1 0.121 1.19765 ̶ 1.33751 2.2229

2 5 0.240 0.82044 ++ 2.01252 3.09074

2 10 0.313 0.68977 +++ 2.41032 3.63783

4 1 0.356 0.77195 ++ 1.32546 57.1454

4 5 0.460 0.64171 +++ 1.63976 31.3793

4 10 0.553 0.53421 +++ 1.97026 27.5070

6 1 0.507 0.83790 ++ 1.19933 243.591

6 5 0.577 0.73621 ++ 1.37817 94.2355

6 10 0.639 0.64983 +++ 1.56670 86.5926

8 1 0.653 0.82679 ++ 1.21097 999.046

8 5 0.680 0.78162 ++ 1.28559 265.387

8 10 0.707 0.73630 ++ 1.36855 178.550

157

Combination treatment of ABT-888 and radiation induces apoptosis in GBM PDCLs

The PDCLs RN1, G54, G57 and G89 were plated in 6-well plates as described in Section

2.2.7. The cells were then treated with 10µM of ABT-888 2 hours before irradiation was performed at a dose of 4 Gy. Apoptotic cell death was quantified at 8 days post treatment.

The media containing ABT-888 (10µM) was replenished every day. Apoptotic cell death was measured by harvesting the cells and staining them with Annexin-V-FITC and propidium iodide (Section 2.2.7). Labelling the cells with Annexin-V-FITC and counterstaining with propidium iodide allowed single-cell quantification of apoptotic cells after treatment. Figure 6.11(A) shows representative data for the 8-day timepoint.

The bar graphs in Figure 6.11 summarises the percentage apoptotic cells in the whole cell population.

Out of the 4 GBM PDCLs treated with ABT-888 and radiation combination treatment,

RN1 showed the most significant increase of apoptotic cells compared to control, ABT-

888 only and RT only. After 8 days of treatment, percentage of apoptotic cells reached

33.8% for cells treated with ABT-888 and radiation combination, while apoptotic cells for control, ABT-888 only and radiation only groups were at 6.6%, 13% and 14% respectively (p<0.001). No statistically significant changes were observed for G54, G57 and G89 (Figure 6.11). Significance was determined using Two-way ANOVA and

Dunnett's multiple comparisons test.

158

Figure 6.11. Apoptotic cell death was induced after ABT-888 and radiation treatment combination.

(A) Representative image of flow cytometric analysis of GBM PDCLs stained with Annexin-V-FITC/PI after 8 days of treatment with 10µM ABT-888 and 4Gy irradiation. (B) shows the mean ± SD of two independent experiments.

159

Loss of the MRN complex confers sensitivity to veliparib and radiotherapy

To explore predictive biomarkers for PARP inhibition, Mre11 and Rad50 proteins were examined using Western blot analysis. Both proteins form a complex with the Nbs protein, which is essential in detecting, processing and repairing double strand breaks. A notable observation was the low expression of Mre11 and Rad50 in RN1, one of the sensitive cell lines treated with ABT-888 and radiation (Figure 6.12A and B).

A B

Figure 6.12. Expression of proteins involved in the MRN complex.

Protein was extracted from untreated PDCLs (Rn1, G54, G57 and G89). Protein separation was performed by loading 5µg of protein in each lane of a 4 – 15% Mini- PROTEAN® TGX™ precast gel. Proteins were transferred to a nitrocellulose membrane by wet blotting at 100V for 2 hours and 30 minutes. (A) Membranes were probed with a rabbit monoclonal anti-Mre11 antibody [12D7] (ab214) and mouse monoclonal anti- Rad50 antibody [13B3/2C6] (ab89), both at 1:500 dilution in 5% skim milk in TBST. Rabbit monoclonal EIF4E antibody (CST 2067) diluted at 1:2000 in 5% skim milk was used as a loading control. (B) Densitometry analysis of Western blot was performed using ImageQuant™ TL software. Bar graphs represent the band density relative to the control group, which was adjusted to the loading control (EIF4E) (B).

160

In vivo

Combination of ABT-888 and radiation improves survival in vivo

As a result of the significant results observed in Section 6.3.1, the efficacy of ABT-888 and radiation was investigated in vivo using patient-derived GBM xenograft animal models. Orthotopic GBM xenograft models were developed by injecting 2 x 105 RN1 cells intracranially into BALB/c nude mice following the procedures described in Section

2.3. Treatment was commenced after tumour growth was confirmed via histological staining with H&E (approximately 50 days after implantation). Mice were randomly assigned to different treatment groups: control (n=5), ABT-888 (n=5), radiation (n=5) and combination (n=7). ABT-888 was given at 12.5mg/kg twice a day for 5 days. ABT-

888 was given per orem via gavage. A second cycle was given to the animals that still had not reached endpoint by the 28th day after initiation of treatment. Radiation was administered at 2Gy increments for two consecutive days (4Gy total dose) after 24 hours of treatment with ABT-888 (Figure 6.13). Endpoint was based on the neurological and physiological status of the animal. The animals were humanely euthanised once significant changes were observed in their neurological function and/or the animals lost

20% of their body weight. Animal survival was defined as the time of euthanasia. Median survival was estimated by plotting the animal survival into a Kaplan-Meier Curve (Figure

6.14). Statistically significant improvement was observed in the median survival of those treated with the combination (83 days) compared to control (66 days), ABT-888 only treated group (64 days) and radiation only treated group (73 days) (Log-rank (Mantel-

Cox) test: p=0.0419).

161

Figure 6.13. Treatment plan for in vivo investigation of ABT-888 and radiation combination treatment.

ABT-888 was administered twice a day for 5 days from day 0. Day 0 was assigned as 50 days post-tumour injection. Radiation treatment was given 24 hours after first dose of ABT-888. A second cycle of ABT-888 was given 28 days after the first course of treatment. Red cross represents the end of the experiment.

Figure 6.14. Kaplan-Meier curve comparing treated groups with the control group.

Once tumour growth was confirmed, mice were rendomly assigned to control (n=5), ABT-888 (n=5), radiation (n=5) and combination (n=7) treatment groups. Statistical significance was calculated using the Log-rank (Mantel-Cox) test (p-value=0.0419).

162

ABT-888 and radiation combination treatment cause reduced proliferation and increased apoptosis in vivo

Immunohistochemistry was performed to validate the combinatorial effect of ABT-888 and radiation treatment following the procedures described in Section 2.3.2. Briefly, 4µM sections of formalin-fixed paraffin embedded brain tissues were mounted onto ultrafrost slides. Tissue sections were processed by deparaffinization with xylene and rehydration in PBS using an ethanol gradient. A heat-mediated antigen retrieval step was performed to break methylene bridges that are formed during tissue fixation. During the antigen retrieval step, the cross-linked proteins are released, and antigenic sites are exposed for antibody binding.

Initially, tissue sections were stained with H&E to confirm the presence of tumour in the brain tissue of sacrificed mice. H&E stained sections confirmed GBM tumour growth with the characteristic increase in mitotic bodies, necrosis and microvascular proliferation.

The combinatory effect of ABT-888 and radiation on proliferation was examined by staining the tissues sections with Ki-67 (details in Section 2.3.2). The expression of the protein Ki-67 is generally correlated as a marker for proliferation since the protein is only present in the active phases of the cell cycle, namely G1, S, G2 and mitosis phase (335).

The rate of apoptosis was assessed by performing a TUNEL assay (Section 2.3.2). The

TUNEL assay detects DNA fragmentation which is a hallmark of apoptosis. Both Ki-67 and TUNEL staining were visualised under a light microscope. 5 random fields at 20x magnification were counted and represented as a percentage of positive staining cells

(Figure 6.15). Statistical significance was determined by One-way ANOVA using

GraphPad Prism version 7.02.

163

B C

Figure 6.15. Ki-67 and TUNEL staining of brain samples from RN1 PDX treated with ABT-888 and radiation, alone and in combination.

(A) Immunohistochemistry staining of control (i, v, ix), ABT-888 (ii, vi, x), radiation (iii, vii, xi) and combination (iv, viii, xii) treated mice brain samples. (B) Ki-67 index (%) counted from 5 random fields at 20x magnification. (C) TUNEL positive staining cells (%) counted from 5 random fields at 20x magnification.

Ki-67 positive staining cells were markedly reduced in the brain tissue sections from animals treated with the combination of ABT-888 and radiation (Figure 6.15A). A statistically significant reduction of Ki-67 positive cells in the combination-treated group versus the control group (p=0.0013) was observed. Ki-67 staining was measured in an average of less than 5% of tumour cells from mice treated with the combination of ABT-

888 and radiation, in comparison to 12% and 14% of tumour cells treated with radiation alone and ABT-888 alone. TUNEL staining was strikingly evident in the combination

164

treated group, exhibiting statistically significant results compared to the control group

(p<0.0001). Approximately 92.5% of the tumour cells from mice treated with the combination of ABT-888 and radiation demonstrated apoptosis. Apoptosis was also observed in 41% of tumour cells from mice treated with radiation only, and fewer cells from those treated with ABT-888 only (3.75%) and control group (3.25%) (Figure

6.15B).

165

Discussion

In this chapter, the combination of ABT-888 with radiation is an effective therapeutic strategy for GBM patient-derived cell lines with unmethylated MGMT. Existing literature has shown that the use of PARP inhibition to sensitise GBM cells to TMZ treatment has been efficacious, however this favorable result is only observed in MGMT methylated

GBM in vitro (327).

Four GBM PDCLs were utilised to address the aims of this chapter. All GBM PDCLs were initially tested for MGMT promoter methylation and G54, G89 and RN1 were found to have unmethylated MGMT. All four cell lines were treated with ABT-888 alone or in combination of either TMZ or radiation.

Monotherapeutic response to TMZ and ABT-888 alone was first examined in all four

GBM PDCLs. The treatment of all four PDCLs with ABT-888 alone did not show any significant effects at lower concentrations of 1µM and 5µM. However, the MTS assay revealed a slight significance at 10µM for both G89 (p=0.0215) and RN1 (p=0.0048)

(Figure 6.3). This is concordant to previous studies where they have shown that treatment with ABT-888 alone does not have any appreciable effects (336). All four GBM PDCLs were also tested for response to TMZ alone at a range concentration of 25µM to 600µM.

Three of the four PDCLs were MGMT unmethylated (G54, G89 and RN1), while G57 had a methylated MGMT promoter. A significant response was detected on G54, G89 and RN1 only at higher TMZ concentrations of 200 to 600µM. Surprisingly, G57 having a methylated MGMT promoter showed no response to TMZ treatment (Figure 6.4). Next, we looked at the response of the four GBM PDCLs to the combination of ABT-888 and

TMZ. The results showed that although a significant decrease of viable cells can be observed on G54, G89 and RN1 with the combination of ABT-888 and high dose TMZ

166

(300 and 600µM), this significance was not perceived in comparison to cells treated with

TMZ alone. This is in concordance with published literature where the combination of

ABT-888 and TMZ exhibited significant synergism only at high concentrations and could not be translated in vivo due to intolerably high dosing regimens (327).

This therefore lead us to hypothesize that ABT-888 can sensitise GBM PDCLs to radiation. Out of the four GBM PDCLs, we observed that the combination of ABT-888 and radiation elicited the most significant response to RN1 as shown in the results of the

MTS, clonogenic and apoptosis assay. G89 also showed some response but results were variable across all different assays. G54 and G57 showed little to no response to the combination treatment. Synergism was also determined and confirmed as shown in Figure

6.10 and Table 6.2. Synergism was scaled following Chou’s scaling system (334). ABT-

888 and radiation showed moderate (++) to synergistic (+++) combination effects in all dose combinations for both G89 and RN1. However, it must be noted that a slight antagonism (CI=1.19765) was observed for RN1 when treated with 1µM ABT-888 and

2Gy radiation dose. The synergistic effects of ABT-888 and radiation is similar to results in a previous study conducted on immortalised commercial cell lines (337).

It should be noted that not all MGMT unmethylated GBM PDCLs responded to the combination treatment. We have hypothesized that response to PARP inhibition works as a synthetic lethality in conjunction with other DNA repair mechanisms. We therefore looked at the protein expression levels of Mre11 and Rad50. Both proteins together with

NSB1 form the MRN complex which has a vital role in detecting and repairing DSBs thru acting as an effector in the HRR and NHEJ repair pathways (338). As shown in Figure

6.12, RN1 exhibited the most significant reduction in MRE11 protein expression compared to the other GBM PDCLs. It has been previously shown in endometrial and

167

colorectal cancers that reduced MRE11 expressions is associated in arbitrating sensitivity to PARP inhibition (339, 340). Biallelic mutations in MRE11 rendered colorectal cancer cells more sensitive to a PARP inhibitor, LT-626. Furthermore, overexpression and knockdown of MRE11 in the colorectal cancer cell lines showed correlation with sensitivity to PARP inhibition (340).

Following significant in vitro results, we examined the effect of the combination of ABT-

888 and radiation in vivo. RN1 cells were intracranially injected in female BALB/s nude mice. This is to our knowledge, the first study looking at the response of a GBM orthotopic model with an unmethylated MGMT promoter. Similar to a previous publication ABT-888 alone did not have an effect in vivo (341). However, the combination treatment of ABT-888 and radiation significantly extended survival by 10 days when compared to radiation as a monotherapy. Tumour sections from the combination-treated group demonstrated high levels of apoptosis and a lower proliferation index when compared to the other treatment groups (Figure 6.15).

The results of this chapter have contributed to the foundations for a randomised Phase II clinical trial in Australia (ANZCTR: U1111-1167-6365), looking at the combination of

ABT-888 (Veliparib) and radiation therapy with adjuvant TMZ and veliparib versus standard therapy in patients with newly diagnosed GBM with unmethylated MGMT promoter.

168

A Systematic Review and Meta-analysis of GBM in vivo Investigations for Topoisomerase Inhibition.

Parts of this chapter have been published in the Oncotarget journal (342).

169

Introduction

Systematic reviews and meta-analyses

Pre-clinical animal studies form an important stage in the development and planning of new treatment designs for human clinical trials. Systematic review and meta-analysis is a powerful tool for the assessment of any treatment and is considered to have the highest level of clinical evidence accepted by many regulatory bodies. Although systematic reviews and meta-analyses of clinical trials are widely reported, until recently there has been a relative paucity concerning pre-clinical literature. Groups such as the

Collaborative Approach to Meta-Analysis and Review of Animal Data from

Experimental Studies (CAMARADES) have undertaken systematic review and meta- analysis across a range of pre-clinical disease groups which have illustrated a number of important themes relevant to the design and interpretation of animal experiments (343,

344).

There are a number of previous systematic reviews of glioma model treatments, assessing

BCNU/CCNU (136), TMZ (137) and gene therapies (138). In these reviews, particularly in relation to chemotherapies, results were overall concordant with human trials, although the underlying data were limited in quality and design. Several factors were found to be associated with treatment efficacy, including factors relating to study quality such as randomization and blinding, and those relating to study design, in particular the selection of tumour model.

Topoisomerases

Mutations in the gene TP53 are associated with changes in the regulation of topoisomerase I and II activity (345-347). Topoisomerase I and II enzymes are essential in the uncoiling of supercoiled DNA to promote DNA metabolic processes.

170

Topoisomerase I catalyse single-strand breaks that allows DNA transcription needed for protein synthesis. Topoisomerase II, on the other hand, catalyses double strand breaks that are important for chromosome condensation. Inhibition of topoisomerases leads to the increased formation of stable topoisomerase-cleaved DNA breaks which eventually lead to cell death (348, 349). The earliest evidence of topoisomerase classification dates from 1972 (350) and has been a target of interest in a range of different cancer types including glioma. In the last five years, Phase I and II clinical trials evaluating the use of topoisomerase inhibitors in combination with other chemotherapeutic drugs to treat GBM have shown variable results (99, 106, 107, 351-355).

Topoisomerase inhibition in humans

In 2013, a meta-analysis performed by Leonard and Wolff of phase I-III clinical trials from 1976 to 2011 including 44,850 patients from 624 publications examined the treatment efficacy of four different topoisomerase inhibitors (topotecan, irinotecan, etoposide and teniposde) (356). Out of these four drugs they were able to establish that etoposide had the most statistically significant effect on overall survival while others showed no significant improvement or even worsened survival outcomes.

This variability in effects observed in clinical trials suggests the need to identify conditions that optimise conditions for treatment efficacy (i.e. patient selection, timing of treatment, etc.). This contrasts with the preclinical literature where the subjective impression is one of a consistent treatment efficacy.

Aims

1. To summarise the overall efficacy of topoisomerase inhibition in pre-clinical

glioma models.

171

2. To appraise study quality and search for evidence of publication bias.

3. To describe the variation in and impact of different study design parameters on

outcome measures.

4. To appraise the construct validity of pertinent literature.

172

Results

Three databases were used for the literature search: PubMed, Medline, and EMBASE.

Five hundred forty-seven publications were identified and screened for eligibility, and 76 publications satisfied the inclusion criteria (Figure 7.1). One study was excluded for using bioluminescent imaging to assess change in tumour volume, since this does not have any direct measurement of the tumour volume (357). Other reasons for exclusion were missing or unreported data (n=5), no true control used for the study (n=11), no measure of spread reported for the volume data (n=2) or the tumour volumes not reported (n=2), no quantitative data reported for the outcome measure (n=1), no individual drug treatment results reported (n=2) and the drug being assessed in the study is not included in the list of drugs specified in the protocol (n=1).

Following these exclusions, 52 publications remained reporting 61 experiments assessing animal survival and 29 assessing tumour volume data, which we have included in the systematic review. Of these, we proceeded to meta-analyse studies reporting glioma models used in five or more studies reporting the corresponding outcome measure.

Previous meta-analyses in this field have suggested a substantial degree of covariance between study design parameters, particularly the choice of tumour model (137, 138).

Hence the limitation to only include studies that use glioma models reported in five or more experiments. This involved 47 survival and 16 tumour volume experimental comparisons utilizing 1,192 animals.

173

Figure 7.1. Study selection summary

174

Study characteristics

Of the 90 comparisons, mice were the most commonly used species for assessing both survival (n=35) and tumour volume reduction (n=20). No other species of animal was used besides mice and rats. Comorbid animals were commonly used. Athymic animal models were most commonly observed in studies assessing survival (n=26) and in studies assessing tumour volume (n=20). Other comorbidities observed in the survival studies were RAG2-M (n=2) and SCID (n=2), which are both immunocompromised models.

SCID animals were reported in two out of 29 tumour volume reduction experimental comparisons. Surprisingly, 24 experimental comparisons assessing survival did not use animals with any comorbidity.

The most frequently used cell line was U87 in both survival (n=14) and volume studies

(n=16). Other glioma models more commonly observed were U251 (n=7), 9L (n=8),

BT4Ca (n=5), GBM (n=7), and 101/8 (n=6). Aside from one study which used a patient- derived GBM xenograft model (358), none of the studies screened for TP53 mutation

(Table 7.1). Majority of the survival studies used xenograft models (n=57), and six of the seven experiments that used ‘GBM’ models were serially passaged subcutaneous patient- derived xenografts. Most tumour models developed were orthotopic (n=67).

Experiments most commonly used doxorubicin (n=49), followed by irinotecan (n=25), etoposide (n=7), topotecan (n=6), and epirubicin (n=3). The most frequently tested route of drug delivery was intravenous (n=48), intracranial (n=21) and intraperitoneal (n=15) administration. Figure 7.2 illustrates a graphical representation of the summaries of the different study characteristics observed in all the studies included in the review.

175

Table 7.1. Summary of TP53 mutation status for glioma models included in this meta-analysis.

Cell line group TP53 mutation status

U87 Wildtype (359)

U251 Mutant (360)

GBM SJ – GBM2 – mutant (358)

BT4Ca Unknown

9L Mutant (361)

101/8 Unknown

176

A B C D

177 E F G H

Figure 7.2. Study Characteristics.

Frequency distribution of all studies included in the systematic review (n=90) based on (A) outcome measure, (B) animal species, (C) site of tumour implantation, (D) glioma model, (E) comorbidity of animals, (F) drug, (G) route of drug administration and (H) frequency of administration. (A) is presented in percentage, while (B to H) are presented in absolute counts. Blue bars represent survival data, while green bars represent tumour volume data.

Abbreviations: IV – intravenous; IC – intracranial; IP – intraperitoneal; SC – subcutaneous; PO – per orem; IA – intra-arterial.

178

Study quality

Study quality was scored based on a 12-item checklist (Table 2.15). Study quality was modest, the median number of checklist items scored 6 out of 12 (IQR – 5-7, range 3-9)

(Figure 7.3). Ninety-six percent of the comparisons included in the systematic review were published in a peer-reviewed journal, 56% reported random allocation to treatment groups, 10% blinded the assessment of outcome measures, none reported sample size calculation, 88% reported compliance with animal welfare regulations, 34% stated a potential conflict of interest, 26% published “take rates”, 13% reported reasons for exclusions, 98% had a consistent site of tumour implantation, 87% reported a standard number of implanted cells in all the animals, 23% justified drug action and 62% justified the use of carriers. Full details of study quality score are in Appendix VIII.

Figure 7.3. Quality score frequency distribution.

Frequency distribution of all included studies in the review based on stratified quality scores. Frequency is reported as percent of total number of studies included in the review (n=90).

179

Meta-analysis

In the 90 comparisons included in the systematic review, 18 different glioma models were reported (Appendix IX). However, previous meta-analyses in this field have suggested a substantial degree of covariance between study design parameters, particularly the choice of tumour model (137, 138). To limit this covariance, we proceeded to meta-analyse studies reporting experiments using glioma models which had been used in five or more studies reporting the corresponding outcome measure. This involved 47 survival and 16 tumour volume experimental comparisons using a total of 1,192 animals (Figure 7.1).

Overall, animals bearing gliomas treated with a topoisomerase inhibitor survived 1.32 times (95% CI 1.23 – 1.43) longer than control animals, a significant between-study heterogeneity was observed (F= 7.1, p<0.001, I2= 92.2%; Figure 7.4A). Topoisomerase inhibition, on the other hand, reduced tumour growth by 68.75% in comparison to control

(TVR response ratio: 3.2, 95% CI 1.99 – 5.18), significant between-study heterogeneity was also observed (F=4.78, p<0.001, I2=95.5%, Figure 7.4B).

Meta-regression

A high level of heterogeneity was observed in both datasets, warranting further analysis to identify study quality or design items associated with treatment efficacy. As specified in the protocol, the strategy of interest was a multivariable meta-regression, designed to identify and account for covariance between study characteristics which is common in heterogeneous datasets. In this instance, glioma model selection was suspected to be a frequent confounder in univariable analyses based on previous publications (136, 137,

362).

180

Figure 7.4. Meta-analysis.

Forest plots showing meta-analysis results for both (A) survival and (B) tumour volume data. Effect sizes for each experimental comparison is plotted with the 95% confidence interval (CI). Each comparison was weighted by the corresponding sample size. Sample size is the total number of control animals and total number of animals in the treatment group served by one control. The solid grey line represents the level of neutral effect. Value greater than one favours treatment, while values less than one favours the control.

181

Study characteristics relating to both survival and tumour volume data was initially assessed in a univariable meta-regression. Variables tested included the glioma model, animal species, site of tumour cell implantation, drug used, dosage, type of control used, route of drug delivery, frequency of drug administration, randomisation, blinding of outcome assessment and quality score. The data was deemed sufficient to proceed to a multivariable meta-regression with the survival dataset (47 comparisons) but not the tumour volume dataset (18 comparisons).

For the multivariable meta-regression model, based on previous experience, inclusion of one variable for every ten comparisons was allowed, thereby permitting input of five predictors. These five variables were selected a priori to be glioma model, site of implantation, drug used, route of delivery, quality score and type of control used for the experiment. However, because the majority of the tumours were orthotopic (n=44/47), site of implantation was not included; instead, the variable which had returned the largest

F-value in the univariable analysis was included, that being the nature of the control used.

Following the meta-regression, each predictor was tested post hoc with a Wald test – with a significant result implying an independent predictive value of the variable of interest.

Survival data

The multivariable meta-regression was significantly associated with treatment outcome, suggesting a predictive value of at least one variable offered (F=6.08, p<0.0001), and residual I2 of 74.24%. When tested with a Wald test, 4 of 5 variables offered to the meta- regression model (glioma model, drug, type of control, and route of drug delivery) were independently associated with heterogeneity in the survival data (Table 7.2).

182

Table 7.2. Multivariable meta-regression of survival data.

Variables F P

Model 6.08 0.0005*

Type of control 3.85 0.0185*

Route of drug delivery 8.61 0.001*

Drug 3.74 0.0207*

Quality score 0.91 0.3482

Legend: (*) - p-value <0.05

All 4 drugs used appeared to be associated with treatment efficacy (Figure 7.5A), with the most commonly used drugs (doxorubicin and irinotecan) associated with similar outcomes (MSR: 1.29, 95% CI 1.17-1.43 and 1.38, 1.16-1.65, respectively). Epirubicin, used in only 2 experiments, was associated with greater efficacy (1.78, 95% CI 1.19-

2.68). The choice of drug was independently predictive of treatment outcome in the multivariable model (F=3.74, p<0.05), although not in the univariable model (F=1.14, p>0.0042). The route of drug delivery was associated with survival outcome in the multivariable model (F=8.61, p<0.05, Figure 7.5B), with intracranial treatment appearing to be associated with greater efficacy than systemic routes. There was no association seen on univariable analysis (F=3.92, p>0.0042). The type of control, included into the multivariable analysis because of a large F-value on univariable analysis (F=9.94, p<0.0042), was also predictive of outcome on multivariable meta-regression (F=3.85, p<0.05). Studies where the control was untreated or given a carrier were associated with greater efficacy than those controlled with vehicle or saline (Figure 7.5C).

183

All 6 glioma models used in survival experiments were associated with treatment efficacy

(Figure 7.5D). The choice of glioma model was associated with outcome heterogeneity on multivariable meta-regression (F=6.08, p<0.05), with human GBM (1.58, 1.22 – 2.07) and rat 101/8 (1.63, 1.34 – 1.97) cell lines associated with greater efficacy than the most frequently reported U87 cell line (1.14, 0.90-1.43). Tumour model selection was not associated with heterogeneity on univariable meta-regression (F=3.08, p>0.0042). There were no associations between total study quality score and survival outcome on either multivariable (F=0.91, p>0.05) or univariable (F=0.18, p>0.0042) analysis.

The remaining variables were tested on univariable meta-regression only (Table 7.3,

Figure 7.6). There were no associations between survival outcome and frequency of drug administration (F=1.68, p>0.0042), site of tumour implantation (F=2.32, p>0.0042), species (F=1.68, p>0.0042), comorbidity (1.14, p>0.0042), the reporting of randomised group allocation (F=0.79, p>0.0042) or blinded assessment of outcome (F=0.79, p>0.0042).

184

A C

B D

Figure 7.5. Multivariable meta-regression based on survival outcome.

Forest plots of (A) choice of drug, (B) route of drug delivery, (C) type of control and, (D) choice of glioma model for the study. Median survival ratio was calculated by dividing the mean outcome of the treatment groups by the mean outcome of the control group. The solid grey line represents neutral effect. The x-axis is shown in log-scale.

185

A D

B E

C F

Figure 7.6. Univariable meta-regression based on survival outcome.

Forest plots of (A) frequency of drug administration, (B) tumour site implantation, (C) animal species, (D) comorbidity, (E) randomisation, and (F) blinding of outcome assessment. The solid grey line represents neutral effect. The x-axis is shown in log- scale.

186

Table 7.3. Comparison of univariable meta-regression for survival and tumour volume experimental comparisons.

Survival Tumour volume

Variables F p F p

Animal species 1.68 0.2011 - -

Model 3.05 0.0198 na na

Comorbidity 1.14 0.3526 - -

Site of tumour 2.32 0.1349 2.47 0.1386

implantation

Drug 1.14 0.3448 1.05 0.405

Route of drug delivery 3.92 0.0271 1.17 0.3601

Type of control 9.94 <0.0001* 0.29 0.8779

Frequency of 1.07 0.3518 0.21 0.8168

administration

Randomisation 0.79 0.3801 0.06 0.8146

Blinded outcome 0.79 0.3785 na na

assessment

Quality score 0.18 0.671 0.81 0.691

Legends: (*) - p-value <0.05; na – not applicable

187

Tumour volume dataset

Multivariate meta-regression model was not applied to tumour volume data as there were only 16 experimental comparisons included in this meta-analysis. All 16 experiments used a U87 glioma model, which was the reason why this glioma model was not included in the univariable meta-regression. Similarly, the majority of studies used mice (13/16) and athymic animals (14/16), and all were not blinded. Consequently, species, comorbidity or blinded assessment of outcome variables were not included in the univariate metaregression of the tumour volume data. None of the remaining variables, when applied to univariable meta-regression, were found to be predictive of volume outcome (Table 7.3, Figure 7.7), including drug (F=1.05, p>0.0056), route of delivery

(F=1.17, p>0.0056), type of control (F=0.29, p>0.0056), frequency of drug administration (F=0.21, p>0.0056), site of tumour implantation (F=2.47, p>0.0056), randomised group allocation (F=0.06, p>0.0056) and quality score (F=0.81, p>0.0056).

188

A D

B E

C F

Figure 7.7. Univariable meta-regression based on tumour volume reduction.

Forest plots of (A) drug of choice, (B) route of administration, (C) type of carrier, (D) frequency of administration, (E) site of tumour implantation, and (F) randomisation. The solid grey line represents neutral effect. The x-axis is shown in log-scale.

189

Post hoc assessment of drug carriers

Two of the significant results from the two analyses pertained to drug delivery by way of route of delivery directly and indirectly in the way of the type of control used. A variance in the use of the drug carriers has been observed across the different studies, and it is possible that the use and choice of carriers could be affecting both of these significant outcome results. As such, the descripiton of the frequency of carrier use was considered important (Appendix X) and have subsequently included this stratification post hoc in the univariable metaregression.

Of the 52 publications included in the systematic review, 58% (n=30) used a carrier for drug delivery. The most frequently reported carriers were nanoparticles (n=8) and liposomes (n=7) as drug carriers, all of which were administered intravenously except for one study using nanoparticles that administered via intraperitoneal route. Other carriers observed were polymers (n=6), drug eluting beads (n=3), nanoliposomes (n=2), albumin

(n=1), micelles (n=1), microbubbles (n=1), and microspheres (n=1). Of the 38 publications included in the meta-analysis, 25 used a carrier intended to enhance drug delivery into the brain. The choice of carrier was not associated with heterogeneity in the survival (F=2.04, p=0.0824) or volume (F=0.95, p=0.472) datasets on univariable meta- regression (Figure 7.8).

190

Figure 7.8. Post-hoc univariable meta-regression analysis of choice of carriers.

Post-hoc univariable analysis of carriers used in studies assessing (A) survival and (B) tumour volume reduction.

191

Publication bias

Assessing all studies included in this review, evidence of publication bias has been found in both datasets. For the survival dataset, publication bias was observed by means of an asymmetric funnel plot (Figure 7.9A), and a significantly positive intercept in the Egger’s regression test (B=11.24, t=8.67, p<0.001, Figure 7.9B). A dramatic funnel plot asymmetry was observed in the volume dataset (Figure 7.9C), although Egger’s regression test did not show a significantly positive intercept (B=0.377, t=0.151, p>0.05,

Figure 7.9D). Trim and Fill analysis suggested the presence of 14 ‘missing’ studies in the tumour volume dataset (Figure 7.9C). Addition of the ‘missing’ studies dramatically reduced overall efficacy from 2.35 to 1.35. A contour overlay was applied to the funnel plot (Figure 7.9C) to help identify the cause of funnel plot asymmetry (146). Contour enhancement in funnel plots simply add on contour lines representing “milestones” of statistical significance (146). Asymmetry can be attributed to publication bias if

“missing” studies are clustered only in areas of high or non-statistical significance. In this case, the imputed ‘missing’ tumour volume studies are equally distributed between the areas of high, conventional and non-statistical significance (Figure 7.9C). This can be interpreted as asymmetry being not because of publication bias but because of other confounding factors such as variable study quality or heterogeneous study design.

192

A C

B D

Figure 7.9. Publication bias.

Funnel plots and Egger’s regression publication bias plot for all studies included in the systematic review for both (A and B) survival and (C and D) tumour volume datasets. Survival studies show significant publication bias (p < 0.001) as shown by the asymmetry in (A) and by means of a positive intercept (11.24) in (B). For the tumour volume studies, publication bias was observed by means of funnel plot asymmetry observed in (C). Trim and fill imputed 14 “missing” studies (red dots) thus reducing overall efficacy estimate in (C). Egger’s regression analysis of the tumour volume dataset did not return a significantly positive intercept (D). Funnel plots show the effect size (x-axis) against study precision (y-axis). Egger’s regression publication bias plot shows the standardised effects size (x-axis) against study precision (y-axis). Solid black lines in (A) and (B) represent the level of neutral effect. Funnel plot in (C) is contour-enhanced. The light grey shading represents area of high statistical significance (p > 0.01), while the white area represents the area of non-statistical significance (p > 1). The darker shaded contour line represents the area of associated statistical significance (p = 0.05).

193

Discussion

In this study, 52 publications (90 comparisons) satisfied the inclusion criteria. Of these,

38 publications with 63 experimental comparisons consisting of 1,192 animals for treatment efficacy was used for the meta-analysis. Five different drugs were assessed, namely doxorubicin, epirubicin, etoposide, irinotecan and topotecan.

Overall, animals harbouring glioma tumours treated with a topoisomerase inhibitor both survived longer (1.33x) and showed reduced tumour growth (3.21x) compared to controls. Efficacy estimates favoured treatment for doxorubicin, epirubicin and irinotecan in survival experiments, with topotecan subjectively appearing to confer a lesser benefit.

Similarly, efficacy estimates favoured treatment in all 4 drugs used in tumour volume studies (doxorubicin, epirubicin, etoposide, irinotecan) although confidence intervals were broad. Performing the meta-regression showed that the choice of drug did not account for heterogeneity in the univariable phase, and this was therefore analysed as a single pooled dataset.

A major limitation of the meta-regression model is that, it only describes the observational association between studies, and therefore lacks the benefit of specifying a causal relationship that can be interpreted in randomised studies (363). On the other hand, the advantage of performing a multivariable model is that it can offer an estimate of the degree of collinearity between variables and attempt to correct for this. In previous reviews of the glioma literature (136-138), a large number of significant results on stratified meta-analysis and a particularly large range of efficacies observed across a large number of glioma models has generated suspicion of collinearity and, as such, rendered results difficult to interpret.

194

Therefore, two strategies have been devised to counter this: firstly, to exclude infrequently reported glioma models (those reported in fewer than five experiments), and secondly to apply multivariable meta-regression to the remaining dataset following a standard (univariable) meta-regression. The univariable meta-regression returned fewer significant results than previous stratified meta-analyses in glioma studies, and this likely is a manifestation of the fact that meta-regression is a more conservative tool than stratified meta-analysis (364, 365), and of a smaller number of included studies. Based on the multivariable meta-regression analysis, four variables were independently associated with survival heterogeneity. Three of these, namely glioma model, drug and route of drug delivery were non-significant on univariable meta-regression. The revelation of these associations on multivariable modelling confirms the presence of collinearity in this dataset as suspected, and thus validates multivariable modelling in this context.

The selection of carrier has been suspected as a confounding factor of the analyses of the route of delivery and type of control used, both of which returned significant results from the meta-regression model. On this basis, the selection of carrier was included in the meta- regression post hoc. While this result was non-significant, this variable was not included in a multivariable model and so it remains possible that a confounding effect of drug carriers may exist. No significant associations were observed on the univariable analysis of the tumour volume dataset, and this is likely a manifestation of a large degree of heterogeneity (I2=95.5%) and small sample size.

On further inspection of the included studies, a large range of experimental design was observed relating to both design parameters and measures to reduce the risk of bias.

Overall, study quality was modest. Less than 50% of the studies reported randomisation,

195

while 90% of the included studies did not report blinding of outcome assessment, and no study reported sample size calculation. These factors are important measures of overall study quality and inflated treatment efficacy in the context of a lack of randomisation or blinding is a phenomenon well-described in previous glioma studies (137, 138).

Furthermore, evidence of publication bias was evident in both datasets by way of asymmetrical Funnel plots, a positive Egger intercept from the survival data and “missing studies” imputed from the tumour volume data by Trim and Fill analysis – with a dramatic reduction in global efficacy estimate after inclusion of these ‘missing studies’ into the meta-analysis. These suggest that there is a relative lack of small, inefficacious studies that are not reported for a number of reasons (366) and resulting in a consequent inflation of efficacy perceived both in subjective impressions of the literature and in this meta- analysis. It should also be noted that although publication bias was concluded from the funnel plot asymmetry, other confounding factors should also be considered as a cause of the funnel plot asymmetry.

Taking these together, we conclude that this literature is of low quality and therefore at high risk of bias with an overstatement of perceived treatment efficacy. For this reason, as well as those highlighted below, the results from this review should be interpreted with caution. While study quality appears to be improving with time (367), more should be done to improve quality in preclinical research and consequently to improve transferability between pre-clinical and clinical research, efficiency of treatment translation to clinical trial and reduce a potential overuse of animal laboratory resources.

Animal experiments may often be viewed as a stepping stone from bench to bedside, given the innate similarities between both hosts and their diseases – in physiological and

196

genetic terms. Unfortunately, animal models have a number of important features that, although not preclusive of their value in research, render them imperfect.

One difference between the animal study and the high-quality randomised control trial is the population homogeneity: animal experiments are designed to minimise variables and thus tend to report the use of genetically similar individuals seeded with tumours derived from a single cell line. This contrasts with the clinical setting, defined by genetically diverse patients suffering from tumours that are phylogenetically distinct. Indeed, an important feature of the successful clinical trial is a treatment’s efficacy across this range of heterogeneity, after stratification or selection to account for known predictors. The animal study often lacks this; and pooling of studies in meta-analysis provides a way to simulate this heterogeneity and retest a treatment in this context. However, its value is dependent on the underpinning data – in terms of quality as described above – and in terms of the study’s ability to replicate the biology and heterogeneity of human disease.

A prime example of this are animal glioma models. Immortalised cell lines used to produce xenograft models are well established and easy to cultivate, although have a number of key limitations. While each model displays different features advantageous for the study of tumour cell biology, such as angiogenesis or immunogenicity, their overall behaviour appears to be divergent from their parent tumours and from de novo human

GBM disease (138). The most commonly used glioma xenograft – U87 – has been observed to grow in a pattern not typical of a human GBM (368, 369), and the DNA profile of commercially available U87 cells are distinct from the original ‘U87’ tumour

(152). They are genetically homogenous and behave predictably – thus lacking aforementioned heterogeneity.

197

Herein we observed the use of 18 different tumour models, 16 of which were xenograft- based, and only two studies reported the intention of testing a topoisomerase inhibitor across a number of different tumour models. While one reported the use of virus-induced tumours, there were only two instances of patient-derived xenografts – a model known to behave more closely to human disease in terms of progression and response to treatment

(147, 370, 371). We have shown that glioma model selection is one of the most important predictors of treatment efficacy: more so than features such as drug of choice and several quality parameters. Median survival ratio estimates ranged widely, from 1.14 for U87 to

1.63 for 101/8 tumours. Consequently, glioma model selection should be considered very carefully in the design phase of future studies. We have reservations about the use of a single xenograft model to test hypotheses wider than the scope of cell biology because we feel that a single xenograft model cannot be extrapolated to the breadth of human disease.

An example of note in this context is the p53 status of tumour model used in the study.

Of the six glioma models included in the meta-analysis, four have known p53 mutations

(Table 7.1). We can observe a trend for p53 mutant cell lines to be associated with better response than p53 wildtype cell lines (Figure 7.5).

Topoisomerase inhibition appears to be effective in human clinical trials, in certain circumstances, as evidenced in a systematic review and meta-analysis performed by

Leonard and Wolff. The meta-analysis suggests that etoposide has a favourable effect on median survival in GBM patients while the use of irinotecan worsens outcome (356). An immediate discrepancy between this review of clinical studies and our review is an absence of doxorubicin in human trials, contrasting with animal studies in which it was the most commonly used drug. Secondly, while the use of carriers was unacknowledged

198

in the review of Leonard and Wolff, we found that carriers were used in 58% of preclinical studies. Finally, we have shown that the route of delivery of treatment is an important predictor of outcome, with experiments using intracranial drug delivery associated with greater efficacy. While convection-enhanced delivery and Gliadel® wafers are examples of intracranial chemotherapy tested in humans, trials have unfortunately yet to reveal any dramatic advantages over systemic chemotherapy (372, 373), thus representing a further discord between the animal and clinical literature. This discrepancy has been a consistent finding across the pre-clinical glioma literature (137, 362).

In conclusion, based on this study, we believe that topoisomerase inhibition remains a viable treatment option for patients diagnosed with GBM. However, a number of concerns arise relating to the internal and construct validity of the animal literature: the existing literature is at high risk of bias, with evidence of augmented perceived treatment efficacy, and animal research has limitations in the recapitulation of human disease.

Nonetheless, factors such as glioma model, route of administration and drug delivery methods appear to predict outcome and must be taken into consideration when planning future studies. We believe that further high-quality in vivo studies accommodating these conclusions would be invaluable in helping to further define the role for topoisomerase inhibition in clinical GBM.

199

Summary and Conclusions

200

Glioblastoma (GBM) is one of the most devastating cancers arising in adults. In Australia, the current statistics show that six patients are being diagnosed with brain cancer every day, with a possibility of 55% of them being diagnosed with GBM (4, 374).

Improvements in the 5-year survival rate has remained at 1% for the last three decades

(374). The most prominent concern for the treatment of GBM is its high molecular heterogeneity. This molecular heterogeneity is also reflected with the lack of appropriate pre-clinical in vitro and in vivo models that faithfully recapitulate the patient’s tumour.

Another challenge for treatment is the presence of the blood brain barrier, where drug agents observed to be effective in other cancer types cannot be used for the treatment of

GBM.

Targeted therapeutics focused on the perturbation of genetic or protein targets has been of great interest in the past years, however investigations in clinical trials have yet to demonstrate a significant result. A targeted treatment approach has been largely applied in cohorts of patients, with a positive result being determined based on a collated statistical outcome within that group of patients. A previous study by Von Hoff et al. demonstrated a positive outcome in 27% of patients with refractive cancers who received treatment based on their molecular profile (187). The overarching aim of this thesis was to use molecular profiling to perform an individualised treatment approach for a patient based on the tumour’s genomic profile.

Specifically, three major aims were pursued. Aim 1 addressed in chapter 3, was to develop a patient-derived in vitro and in vivo model for a specific patient’s tumour. Aim 2 sought to characterise the patient’s tumour molecular profile to enable the identification of possible molecular targets. Characterisation of the patient tumour’s molecular profile was conducted in two parts. First was the whole genome sequencing performed in chapter 3,

201

and second was through a commercial molecular profiling service as shown in chapter 4.

For aim 3, based on results achieved in chapters 3 and 4, identified molecular targets were further investigated in chapters 4 to 6. Topoisomerases were directly identified as targets of interest based on the report from the commercial molecular profiling service.

Additionally, the mammalian target of rapamycin and the poly (ADP-ribose) polymerase family of proteins were deemed targets of interest based on the perturbations observed on

PTEN and DNA repair pathways, respectively. Additionally, I also performed a systematic review and meta-analysis of pre-clinical glioma literature with the aim to validate the use of patient-derived tumour models and identify underlying variables that can influence treatment response in animal models. The systematic review and meta- analysis included studies investigating the efficacy of topoisomerase inhibitors in vivo.

In Chapter 3, I performed the characterisation and molecular profiling of the patient primary (G89) and its recurrent (G244) tumour as a pre-requisite to the identification of logical therapeutic regimens based on the molecular make-up of the individual patient.

The profiling performed on G89 and G244 was conducted to compare if differences can be observed between the primary and recurrent tumour. The mutational load of G89 and

G244 had very high similarity (97.6%). However, differences were observed in copy number alterations where G244 acquired copy number losses in chromosomes 6, 10 and

13, while retaining the copy number gain in chromosome 7 as observed in G89. For the purpose of target identification, GBM cancer driver genes in the mutational profile of

G89 were classified based on previous works of Parsons et al. and Vogelstein et al. (173,

174). A pathway-based network analysis identified clusters that were functionally affected based on these driver gene mutations within the patient tumour. Specifically, major pathways included gene families involved in DNA repair pathway and PTEN.

202

Perturbations of mTOR and PARP, targets found to have major contributions with pathways previously mentioned, were further investigated in Chapters 5 and 6.

In addition, the POLE mutation was investigated due to the tumour sample’s high mutational load. Other cancer types have previously reported a significant correlation between POLE mutations and tumour mutational load (168, 180). My investigation revealed an unreported mutational variant (E1240K) of the POLE gene. This adds to the current knowledgebase of occurring somatic POLE mutational variants in GBM – V411L,

P286R, S297F/Y and S459F (178). These mutational variants have been previously observed in GBM tumour samples, and other cancer types with high mutational load.

Erson et al. concluded that POLE mutations were significantly associated with hyper mutations in GBM and prolonged survival in a comparison of 4 POLE mutant and 51

POLE wild type tumour samples (178). However, a more recent investigation by Hodges et al. in early 2017 reported no significant association to high tumour mutational load and

POLE mutations. Hence, future studies to provide stronger evidence is still needed to validate the importance of POLE mutations in the occurrence of hyper mutations in GBM due to this incoherent result observed in literature.

In Chapter 4, CARIS molecular profiling was performed on the G89 tumour sample.

CARIS molecular intelligence tumour profiling is a commercial service offered by the

CARIS Life Sciences, USA. A CARIS molecular intelligence profile report was provided.

The report suggested possible beneficial drugs based on the targets identified in the patient tumour. I screened the suggested list of drugs for the patient based on glioma literature, Lipinski’s rule of 5 and its ability to pass through the blood brain barrier.

Lipinski’s rule of 5 was used for screening with the prospect to select a drug that had high oral bioavailability and increase the confidence that it can be recommended for use in the

203

clinic. Based on these screening strategies I identified topotecan and irinotecan as relevant treatment options. Prior to screening the G89 PDCL, I validated the presence of the target of interest by performing immunoblotting techniques. Treatment of G89 along with other

PDCLs from QIMR with topoisomerase inhibitors such as irinotecan and topotecan revealed varying treatment responses, with G89 showing low sensitivity to the treatment.

Based on the results of the network analysis from Chapter 3, treatment responses to the inhibition of mTOR and PARP were investigated in Chapters 5 and 6, respectively. mTOR targeting with temsirolimus in combination with PENAO was investigated, while

PARP inhibition was examined in combination with radiotherapy. These drug combinations were also investigated on a panel of GBM PDCLs, including G89.

In Chapter 5, the temsirolimus and PENAO combination showed promising results in vitro as evidenced by decreased cell proliferation, decreased clonogenicity and increased apoptosis. The lowest synergistic dose combination was noted at IC70, with moderate to high reduction in temsirolimus and PENAO doses, respectively, when used in combination. Despite promising results observed in vitro, further investigation of the combination in a PDX model injected with RN1 did not result in a survival advantage.

This was attributed to PENAO not being able to reach the brain tumour at therapeutic concentrations. Drug administration was restricted to continuous subcutaneous release in small amounts via an osmotic pump. This was due to a previous in-house study where subcutaneous administration of higher amounts of the organo-arsenical compound resulted in tissue necrosis at the site of administration (Chung et al. - unpublished). Future studies will be focused on improving drug delivery of PENAO to the brain to replicate the results we observed in vitro.

204

In Chapter 6, I investigated the efficacy of PARP inhibition with ABT-888 in combination with radiotherapy. Here we focused at the response of unmethylated GBM PDCLs, such as G89, G54 and RN1, to the treatment combination. I was able to demonstrate concordant in vitro and in vivo experimental results. In vitro data exhibited synergistic treatment combination effects demonstrated by cell viability, clonogenic and apoptosis assays supported by the Chou-Talalay combination index calculation. The drug combination significantly extended survival in PDX models by 10 days (p=0.0419). Low expression of MRE11 in GBM PDCLs was identified as a possible prognostic marker for a favourable treatment response from PARP inhibition. The results of this chapter contributed to the establishment of a randomised clinical trial in Australia investigating the combination of ABT-888 and radiotherapy with adjuvant TMZ and ABT-888, versus standard therapy in GBM patients with unmethylated MGMT promoter (VERTU).

A major limitation I have observed in performing an individualised treatment approach is the lack of a comprehensive pathway database containing information regarding gene- protein interactions in the context of GBM. Most available resources map gene-protein interactions to normal physiological and molecular functions. Improvements in databases, such as CTD (188) and KEGG (375), that contain these associations, specifically genotypic, proteomic, and drug phenotypic response interactions are still needed. The development of such databases will unquestionably improve the efficiency in how we identify appropriate targeted treatments that can be given to individual GBM patients, and in turn improve survival.

In the last few years, calls for modifying traditional clinical trials to accommodate an individualised treatment approach. N-of-1, basket and umbrella trials have been proposed, all of which focus on the treatment effect of drugs based on a subset of patients

205

who all share the same genetic abnormality (376). An example of this is the US NCI

Molecular Analysis for Therapy Choice (MATCH) trial (2015), a basket trial, where patients who shared a specific genetic marker were assigned to specific treatment brackets regardless of their cancer type or stage. Patients were switched to different treatment arms if they don’t have a positive response to their first treatments (377). Currently, a clinical trial known as GBM AGILE is in development, which would utilise a new continuous adaptive clinical trial system. Similar to the MATCH trial, such trial system would allow treatment iterations based on the responses, be it physiological or molecular, observed in patients during treatment.

Lastly, in Chapter 7, I examined the validity of use of pre-clinical animal models by performing a systematic review and meta-analysis of in vivo studies investigating topoisomerase inhibition. Overall, a favoured treatment efficacy was noted in all topoisomerase inhibitors included in the study. Further analysis with a multivariable meta-regression revealed four variables associated with survival heterogeneity, namely glioma model, type of control, route of drug delivery, and type of drug used. Limitations observed in this study were the presence of publication bias, large degree of heterogeneity between the studies and the small sample size. Therefore, interpretation of this chapter should be done with admonition. Nevertheless, this chapter has identified the variables that should be considered in planning future pre-clinical trials to increase translation of in vivo investigations as well as prevent the over use of animals in these types of studies.

In summary, the personalised approach of identifying targeted therapeutics for an individual patient was challenging, in this case, due to the high mutational load observed in the patient tumour. Individualising treatment for GBM patients based on their tumour’s genomic profile may seem unattainable at the moment but it is not impossible. Further

206

studies should be focused on the integration of different molecular types to improve the accuracy and efficiency of identifying treatment and predicting treatment responses from available pathways, that could be used for analysing molecular profiles and network interactions. Overall, it is in my opinion that the findings of this thesis have contributed to the existing knowledgebase of GBM therapy at different levels of translational research.

207

Appendices

Appendix I. Meta-analysis protocol (Chapter 2) ...... 209

Appendix II. All co-occurring and mutually exclusive mutations observed between POLE and MMR genes extracted from cBioPortal (Chapter 3)...... 213

Appendix III. List of GBM driver gene mutations (Chapter 3) ...... 215

Appendix IV. CARIS Molecular Intelligence Report (Chapter 4) ...... 218

Appendix V. CARIS® gene panel (Chapter 4) ...... 230

Appendix VI. Information extracted from DrugBank for rule-of-five assessment. (Chapter 4) ...... 234

Appendix VII. Immunoblotting of a panel of GBM PDCLs for PTEN expression (Chapter 5)...... 237

Appendix VIII. Study quality scores (Chapter 7)...... 237

Appendix IX. Study characteristics (Chapter 7)...... 241

Appendix X. Carrier types (Chapter 7)...... 252

208

Appendix I. Meta-analysis protocol (Chapter 2)

209

210

211

212

Appendix II. All co-occurring and mutually exclusive mutations observed between POLE and MMR genes extracted from cBioPortal (Chapter 3).

Gene A Gene B p-Value Log Odds Ratio Association

PMS2 MSH2 0.4858 0.5838105165224796 Tendency towards co-occurrence

PMS2 MSH6 0.6073 0.17034536574723894 Tendency towards co-occurrence

213 PMS2 POLE 0.1132 1.1696233243112983 Tendency towards co-occurrence

MLH1 POLE 0.2990 1.2113529116235584 Tendency towards co-occurrence

PMS2 MLH1 0.0242 1.6462518855568167 Tendency towards co-occurrence(Significant)

MSH2 MSH6 0.0001 4.661077817926388 Tendency towards co-occurrence(Significant)

MSH2 POLE 0.0101 3.142714463902637 Tendency towards co-occurrence(Significant)

MSH6 POLE 0.0010 3.3904252846802443 Tendency towards co-occurrence(Significant)

MLH1 MSH2 0.8386 -Infinity Tendency towards mutual exclusivity

MLH1 MSH6 0.7808 -Infinity Tendency towards mutual exclusivity 214

Appendix III. List of GBM driver gene mutations (Chapter 3)

Parsons et al. GBM candidate cancer genes (CAN-genes) (173)

TP53 COL3A1 PIK3R1 DSG4

CDKN2A ARNT2 RBM27 KIAA0774

PTEN KIAA1804 SERPINA12 NGEF

EGFR Q8NDH2_HUMAN PKHD1 PHIP

IDH1 C6orf170 C21orf29 ASTN

CDK4 IMP4 LMX1A FRMPD4

GML IRX6 ZNF497 SCN9A

RB1 KIAA1441 LRRC7 GRM3

ENST00000355324 LRP2 KIAA0133 CACNA1H

NF1 OR2L13 MYO1B

SKP2 PIK3CA TRPV5

Vogelstein et al. Predisposition cancer genes (174)

FLCN ERCC4 GPC3 SDH5

BLM ERCC5 CDC73 SDHB

BMPR1A EXT1 MUTYH SDHC

BRIP1 EXT2 NBS1 SDHD

215

BUB1B FANCA PALB2 SUFU

CDH1 FANCC PHOX2B TSC2

CDK4 FANCD2 PMS1 WAS

CHEK2 FANCE PMS2 WRN

DICER1 FANCF PRKAR1A XPA

ERCC2 FANCG RECQL4 XPC

ERCC3 FH SBDS

Vogelstein et al. Mutation driver genes (174)

CCND1 ATM DAXX HRAS NCOR1 SETD2

CDKN2C ATRX DNMT1 IDH1 NF1 SETBP1

IKZF1 AXIN1 DNMT3A IDH2 NF2 SF3B1

LMO1 B2M EGFR JAK1 NFE2L2 SMAD2

MAP2K4 BAP1 EP300 JAK2 NOTCH1 SMAD4

MDM2 BCL2 ERBB2 JAK3 NOTCH2 SMARCB1

MDM4 BCOR EZH2 KDM5C NPM1 SMO

MYC BRAF FAM123B KDM6A NRAS SOCS1

MYCL1 BRCA1 FBXW7 KIT PAX5 SOX9

MYCN BRCA2 FGFR2 KLF4 PBRM1 SPOP

216

NCOA3 CARD11 FGFR3 KRAS PDGFRA SRSF2

NKX2-1 CASP8 FLT3 MAP2K1 PHF6 STAG2

SKP2 CBL FOXL2 MAP3K1 PIK3CA SMARCA4

ABL1 CDC73 FUBP1 MED12 PIK3R1 STK11

ACVR1B CDH1 GATA1 MEN1 PPP2R1A TET2

AKT1 CDKN2A GATA2 MET PRDM1 TNFAIP3

ALK CEBPA GATA3 MLH1 PTCH1 TRAF7

APC CIC GNA11 MLL2 PTEN TP53

AR CREBBP GNAQ MLL3 PTPN11 TSC1

ARID1A CRLF2 GNAS MPL RB1 TSHR

ARID1B CSF1R H3F3A MSH2 RET U2AF1

ARID2 CTNNB1 HIST1H3B MSH6 RNF43 VHL

ASXL1 CYLD HNF1A MYD88 RUNX1 WT1

217

Appendix IV. CARIS Molecular Intelligence Report (Chapter 4)

218

219

220

221

222

223

224

225

226

227

228

229

Appendix V. CARIS® gene panel (Chapter 4)

Gene symbol NCBI Gene ID

ABL1 25

AKT1 207

ALK 238

APC 324

AR 367

ATM 472

BRAF 673

BRCA1 672

BRCA2 675

CDH1 999

CMET 53561

CSF1R 1436

CTNNB1 1499

EGFR 1956

ER 246870

ERBB2 2064

ERBB4 2066

230

FBXW7 55294

FGFR1 2260

FGFR2 2263

FLT3 2322

GNA11 2767

GNAQ 2776

GNAS 2778

HER2 30300

HNF1A 6927

HRAS 3265

IDH1 3417

JAK2 3717

JAK3 3718

KDR 3791

KIT 3815

KRAS 3845

MPL 4352

NOTCH1 4851

231

NPM1 4869

NRAS 4893

PDGFRA 5156

PGP 283871

PGR 5241

PIK3CA 5290

PTEN 5728

PTPN11 5781

RB1 5925

RET 5979

SMAD4 4089

SMARCB1 6598

SMO 6608

STK11 6794

TLE3 7090

TOP1 7150

TOP2A 7153

TP53 7157

232

TUBB3 10381

TYMS 7298

VHL 7428

233

Appendix VI. Information extracted from DrugBank for rule-of-five assessment. (Chapter 4)

Drug Target UniProt DrugBank Type H+ H+ Molecular cLogP Oxygen (O) and Blood brain Rule ID Accession No. Donor Acceptor Mass nitrogen (N) barrier of Count Count (Ave.± atom count (value, Five SD) probability)

Docetaxel TBA1A Q71U36 DB01248 Small 5 10 807.35 2.64 ± O=14 Neg, 0.9659 No (APRD00932) molecule 0.26 N=1

Paclitaxel TBA1A Q71U36 DB01229 Small 4 10 853.33 3.24 ± O=14 Neg, 0.9748 No

234 (APRD00259, molecule 0.27 N=1 DB05261, DB05927, DB05526)

Doxorubicin TOP2A P11388 DB00997 Small 6 12 543.17 1.2 ± O=11 Neg, 0.9951 No (APRD00185, molecule 0.25 N=1 DB05331, DB05847)

Epirubicin TOP2A P11388 DB00445 Small 6 12 543.17 0.61 ± O=11 Neg, 0.9951 No (APRD00361) molecule 0.99 N=1

Irinotecan TOP1 P11387 DB00762 Small 1 6 586.28 3.3 ± O=6 Pos, 0.6284 No (APRD00579) molecule 0.59 N=4

Topotecan TOP1 P11387 DB01030 Small 2 6 421.16 0.76 ± O=5 Neg, 0.9659 Yes (APRD00687) molecule 1.1 N=3

Tamoxifen ESR1 P03372 DB00675 Small 0 2 371.51 6.46 ± O=1 Pos, 0.5838 No (APRD00123) molecule 0.59 N=1

Toremifene ESR1 P03372 DB00539 Small 0 2 405.96 6.24 ± O=1 Pos, 0.7488 No SHBG P04278 (APRD00391) molecule 0.58 N=1 235 Fulvestrant ESR1 P03372 DB00947 Small 2 3 606.77 7.67 ± O=3 Pos, 0.9217 No (APRD00654) molecule 1.18 N=0

Letrozole CYP19A1 P11511 DB01006 Small 0 4 285.30 2.43 ± O=0 Pos, 0.9737 Yes (APRD01066) molecule 0.54 N=5

Anastrozole CYP19A1 P11511 DB01217 Small 0 4 293.37 2.58 ± O=0 Pos, 0.9382 Yes (APRD00016) molecule 0.39 N=5

Exemestane CYP19A1 P11511 DB00990 Small 0 2 296.40 3.41 ± O=2 Pos, 0.9778 Yes (APRD00144) molecule 0.65 N=0

Megestrol PGR P06401 DB00351 Small 0 3 384.52 3.54 ± O=4 Pos, 0.9617 Yes acetate NR3C1 P04150 (APRD01092) molecule 0.29 N=0

Leuprolide GNRHR P30968 DB00007 Biotech NA NA 1209.40 NA O=12 NA NA (BTD00009, N=16 BIOD00009)

Goserelin LHCGR P22888 DB00014 Small 17 18 1269.41 -2.3 ± O=14 Neg, 0.8816 No

236 GNRHR P30968 (BTD00113, molecule 2.76 N=18 BIOD00113)

Triptorelin GNRHR P30968 DB06825 Small 18 17 1311.47 -1.265 O=13 NA No molecule ± 3.30 N=18

Abarelix GNRHR P30968 DB00106 Biotech NA NA 1416.06 NA O=14 NA NA (BTD00051, N=14 BIOD00051)

Degarelix GNRHR P30968 DB06699 Small 17 18 1632.29 1.42 ± O=16 Neg, 0.953 No molecule 1.75 N=18

Appendix VII. Immunoblotting of a panel of GBM PDCLs for PTEN expression (Chapter 5).

Appendix VIII. Study quality scores (Chapter 7).

Each of the publication was scored according to a 12-item checklist, as listed below, to determine publication bias.

1. Peer-reviewed publication

2. Sample size calculation

3. Randomised allocation of drug (or control) treatment

4. Blinded assessment of outcome

5. Compliance with animal welfare regulations

6. Statement of conflict of interests

7. Uniform number of cells implanted

8. Site of implantation is consistent in all animals

9. “Take rates” of implanted tumour cells is mentioned in the publication

10. Number of excluded animals must be stated with reasons for exclusion mentioned

11. Drug action justified

12. Drug-carrier justified

Name Year 1 2 3 4 5 6 7 8 9 10 11 12 Quality Score

Wang, W. 2015 + + + + + + 6

237

Marrero, L. 2014 + + + + + + + + + 9

Zhong, Y. 2014 + + + + + 5

Lin, J. 2014 + + + + + + + 7

Kovac, Z. 2014 + + + + + 5

Sonabend, A. 2014 + + + + + + 6

Tarasenko, N. 2014 + + + + + + 6

Jiang, P. 2014 + + + + + + 6

Yang, Y. 2013 + + + + + 5

Escoffre, J.M. 2013 + + + + + 5

Jaszberenyi, 2013 + + + + + + + 7 M.

Alhenn, D. 2013 + + + + + + 6

Morfouace, M. 2012 + + + 3

Munson, J. 2012 + + + + 4

Cheema, T. 2011 + + + + + + + 7

Serwer, L. 2011 + + + + + + + 7

Guo, L. 2011 + + + + + 5

Lopez, K. 2011 + + + + + 5

Vinchon-Petit, 2010 + + + + + + + 7 S.

238

Panigrahy, D. 2010 + + + + + + 6

Pozsgai, E. 2000 + + + + + 5

Arai, T. 2010 + + + + + + 6

Kuroda, J. 2010 + + + + + + + 7

Lu, J. 2009 + + + + + + + + 8

Hekmatara, T. 2009 + + + + + + + + 8

Kuroda, J. 2009 + + + + + 5

Kreuter, J. 2008 + + + 3

Petri, B. 2007 + + + + + + + 7

Ambruosi, A. 2006 + + + + + + 6

Gomez- 2006 + + + + + 5 Manzano, C.

Mamot, C. 2005 + + + + + 5

Lesniak, M. 2005 + + + + + + + 7

Steiniger, S. 2004 + + + + + + 6

Prasad, G. 2002 + + + + 4

Houghton, P. 2000 + + + + 4

Sharma, U. 1997 + + + + + 5

Pechman, K. 2012 + + + + 4

239

Glage, S. 2011 + + + + + + + + + 9

Verreault, M. 2012 + + + + + 5

Baltes, S. 2010 + + + + + + + 7

Recinos, V.R. 2010 + + + + + + + 7

Manome, Y. 2006 + + + + + 5

Hsu, W. 2005 + + + + + + + 7

Morita, K. 2003 + + + 3

Chen, P.Y. 2013 + + + + + + + 7

Hosokawa, Y. 2015 + + + + 4

Li, J. 2015 + + + + + + 6

Zhang, C.X. 2015 + + + + + + + 7

Verreault, M. 2015 + + + + + 5

Zhao, Y. 2016 + + + + + + + 7

Byeon, H.J. 2016 + + + + + + 6

Ramachandran 2016 + + + + 4 , C.

240

Appendix IX. Study characteristics (Chapter 7).

This table lists down all the study characteristics extracted from all the references included in the study. Notice some of the references are

supplicated which mean that the study included multiple experiments with different study characteristics (i.e. glioma model, type of control,

outcome measure, etc.)

Name Year Quality Total Animal Comorbidity Broad Drug Broad Single/ Type of Type of Outcome score N glioma Route of Multiple Control Carrier used Measure model Delivery

Wang, W. 2015 6 6.00 Mouse Athymic U251 Irinotecan IP Multiple Vehicle NA Median Survival 241 Marrero, L. 2014 9 12.00 Mouse Athymic U87 Doxorubicin IV Multiple Vehicle Albumin Median Survival

Zhong, Y. 2014 5 9.00 Mouse Athymic U87 Doxorubicin IV Multiple Saline Nanoparticles Median Survival

Zhong, Y. 2014 5 9.00 Mouse Athymic U87 Doxorubicin IV Multiple Saline Nanoparticles Volume

Lin, L. 2014 7 16.00 Mouse Athymic U-118MG Doxorubicin IV Multiple Saline Nanoliposome Median Survival

Lin, L. 2014 7 10.00 Mouse Athymic U-118MG Doxorubicin IV Multiple Carrier Nanoliposome Volume

Kovac, Z. 2014 5 5.33 Mouse None GL261 Doxorubicin IV Single Untreated Microbubbles Median Survival

Kovac, Z. 2014 5 5.67 Mouse None SMA-560 Doxorubicin IV Single Untreated Microbubbles Median Survival

Sonabend, A.M. 2014 6 21.00 Mouse Unknown Murine Etoposide Intracranial Multiple Saline NA Median 242 glioma survival

Sonabend, A.M. 2014 6 20.00 Mouse Unknown Murine Etoposide Intracranial Continuous Saline NA Median glioma Survival

Sonabend, A.M. 2014 6 28.00 Mouse Unknown Murine Etoposide Intracranial Continuous Saline NA Median glioma Survival

Tarasenko, N. 2014 6 8.40 Mouse Athymic U251 Doxorubicin IP Multiple Saline NA Median Survival

Tarasenko, N. 2014 6 8.40 Mouse Athymic U251 Doxorubicin IP Multiple Saline NA Volume

Jiang, P. 2014 6 7.50 Mouse Athymic U87 Irinotecan IP Multiple Vehicle NA Volume

Yang, Y. 2013 5 9.33 Mouse Athymic U87 Doxorubicin IV Multiple Saline Lipoosme Median Survival

Escoffre, J.M. 2013 5 7.50 Mouse Athymic U87 Irinotecan IV Multiple Untreated NA Volume

Jaszberenyi, M. 2013 7 12.50 Mouse Athymic U87 Doxorubicin IV Continuous Unknown NA Volume 243

Alhenn, D. 2013 6 7.50 Rat None F98 Etoposide Carrier Microspheres Median Survival

Morfouace, M. 2012 3 13.33 Mouse Athymic P7CSC Etoposide IP Continuous Unknown NA Volume

Munson, J. 2012 4 9.00 Rat None eGFP-RT2 Doxorubicin IV Single Saline NA Median Survival

Munson, J. 2012 4 9.00 Rat None eGFP-RT2 Doxorubicin IV Single Saline NA Volume

Cheema, T. 2011 7 13.33 Mouse Athymic BT74 Etoposide IP Continuous Saline NA Median Survival

Serwer, L. 2011 7 9.50 Mouse Athymic U87 Topotecan IV Multiple Saline Nanoliposome Median Survival

Serwer, L. 2011 7 15.00 Mouse Athymic GBM Topotecan IV Multiple Saline Nanoliposome Median Survival

Serwer, L. 2011 7 12.00 Mouse Athymic GBM Topotecan IV Multiple Saline Nanoliposome Median Survival

Guo, L. 2011 5 7.00 Mouse Athymic U87 Doxorubicin IV Multiple Saline Lipoosme Median

244 Survival

Lopez, K. 2011 5 4.00 Rat Unknown Retrovirus Topotecan Intracranial Single Saline NA Median Survival

Lopez, K. 2011 5 4.00 Rat Unknown Retrovirus Topotecan Intracranial Multiple Saline NA Median Survival

Lopez, K. 2011 5 4.00 Rat Unknown Retrovirus Topotecan Intracranial Continuous Saline NA Median Survival

Vinchon-Petit, 2010 7 11.75 Rat None 9L Doxorubicin IV Single Carrier Drug-eluting Median S. beads survival

Panigrahy, D. 2010 6 20.00 Mouse SCID U87 Etoposide Oral Continuous Unknown NA Volume

Pozsgai, E. 2000 5 11.25 Mouse Athymic DBTRG-05 Doxorubicin IV Multiple Vehicle NA Volume

Arai, T. 2010 6 6.67 Mouse Athymic U87 Doxorubicin SubCut Single Carrier Polymer Volume

Kuroda, J. 2010 7 5.71 Mouse Athymic U87 Irinotecan IV Multiple Saline Micelles Median Survival

Kuroda, J. 2010 7 5.71 Mouse Athymic U87 Irinotecan IV Multiple Saline Micelles Median 245 Survival

Lu, J. 2009 8 6.67 Mouse SCID U87 Doxorubicin IP Multiple Saline NA Volume

Hekmatara, T. 2009 8 28.00 Rat None 101/8 Doxorubicin IV Multiple Untreated Nanoparticles Volume glioblastoma

Kuroda, J. 2009 5 9.00 Mouse Athymic U87 Irinotecan Intracranial Multiple Saline Micelles Median Survival

Kuroda, J. 2009 5 4.50 Mouse Athymic U87 Irinotecan SubCut Multiple Saline Micelles Volume

Kreuter, J. 2008 3 7.50 Rat Unknown 101/8 Doxorubicin IV Multiple Untreated Nanoparticles Median Survival

Petri, B. 2007 7 23.33 Rat None 101/8 Doxorubicin IV Multiple Untreated Nanoparticles Median Survival

Ambruosi, A. 2006 6 11.67 Rat None 101/8 Doxorubicin IV Multiple Untreated Nanoparticles Median Survival

Gomez- 2006 5 36.00 Mouse Athymic U87 Irinotecan IP Multiple Vehicle Nanoparticles Median Manzano, C. survival

Gomez- 2006 5 58.00 Mouse Athymic U87 Irinotecan IP Multiple Vehicle Nanoparticles Median

246 Manzano, C. survival

Mamot, C. 2005 5 12.50 Mouse Athymic U87 Doxorubicin IV Multiple Saline Lipoosme Volume

Mamot, C. 2005 5 12.50 Mouse Athymic U87 Epirubicin IV Multiple Saline Lipoosme Volume

Lesniak, M. 2005 7 15.00 Rat None 9L Doxorubicin Intracranial Single Carrier Polymer Median survival

Steiniger, S. 2004 6 27.20 Rat None 101/8 Doxorubicin IV Multiple Untreated Nanoparticles Median Survival

Steiniger, S. 2004 6 25.60 Rat None 101/8 Doxorubicin IV Multiple Carrier Nanoparticles Median Survival

Steiniger, S. 2004 6 10.50 Rat None 101/8 Doxorubicin IV Multiple Untreated Nanoparticles Median Survival

Prasad, G. 2002 4 6.00 Mouse Athymic U87 Irinotecan IP Multiple Saline NA Volume

Houghton, P. 2000 4 11.63 Mouse None GBM Irinotecan IV Multiple Untreated NA Volume 247 Sharma, U. 1997 5 11.00 Rat None 9L Doxorubicin IV Multiple Saline Lipoosme Median Survival

Sharma, U. 1997 5 9.00 Rat None 9L Doxorubicin IV Multiple Saline Lipoosme Median Survival

Sharma, U. 1997 5 9.50 Rat None 9L Doxorubicin IV Multiple Saline Lipoosme Median Survival

Pechman, K. 2012 4 8.60 Rat Athymic U87 Irinotecan IV Single Untreated NA Volume

Glage, S. 2011 9 16.00 Rat None BT4Ca Doxorubicin Intracranial Single Carrier Drug-eluting Median beads survival

Glage, S. 2011 9 17.00 Rat None BT4Ca Doxorubicin Intracranial Single Carrier Drug-eluting Median beads survival

Glage, S. 2011 9 18.00 Rat None BT4Ca Irinotecan Intracranial Single Carrier Drug-eluting Median beads survival 248 Glage, S. 2011 9 15.00 Rat None BT4Ca Doxorubicin Intracranial Single Carrier Drug-eluting Volume beads

Glage, S. 2011 9 13.00 Rat None BT4Ca Doxorubicin Intracranial Single Carrier Drug-eluting Volume beads

Glage, S. 2011 9 14.00 Rat None BT4Ca Irinotecan Intracranial Single Carrier Drug-eluting Volume beads

Verreault, M. 2012 5 7.20 Mouse RAG2-M U251 Doxorubicin IV Multiple Untreated Lipoosme Median Survival

Baltes, S. 2010 7 34.00 Rat None BT4Ca Irinotecan Intracranial Single Carrier Drug-eluting Median beads Survival

Baltes, S. 2010 7 25.00 Rat None BT4Ca Doxorubicin Intracranial Single Carrier Drug-eluting Median beads Survival

Recinos, V.R. 2010 7 8.00 Rat None 9L Epirubicin IP Multiple Untreated Polymer Median survival

Manome, Y. 2006 5 7.50 Rat None RT2 Doxorubicin SubCut Single Polymer Volume 249

Hsu, W. 2005 7 20.33 Rat None 9L Doxorubicin Intracranial Single Carrier Polymer Median survival

Morita, K. 2003 3 9.60 Rat None C6 Doxorubicin IV Single Untreated NA Median Survival

Morita, K. 2003 3 9.60 Rat None C6 Doxorubicin IArterial Single Untreated NA Median Survival

Chen, P.Y. 2013 7 10.80 Mouse Athymic GBM Irinotecan IV Multiple Untreated NA Median Survival

Chen, P.Y. 2013 7 10.80 Mouse Athymic GBM Irinotecan Intracranial Single Untreated NA Median Survival

Chen, P.Y. 2013 7 10.80 Mouse Athymic GBM Irinotecan Intracranial Multiple Untreated NA Median Survival

Chen, P.Y. 2013 7 8.80 Mouse Athymic GBM Irinotecan Intracranial Multiple Untreated NA Median Survival

Hosokawa, Y. 2015 4 6.67 Mouse Athymic U251 Doxorubicin IP Continuous Saline NA Volume 250

Li, J. 2015 6 16.00 Mouse Athymic U87 Doxorubicin IV Multiple Saline Polymer Median Survival

Zhang, C.X. 2015 7 7.20 Mouse Athymic U251 Epirubicin IV Multiple Saline Lipoosme Median Survival

Verreault, M. 2015 5 7.20 Mouse SCID U251 Doxorubicin IV Multiple Untreated NA Median Survival

Verreault, M. 2015 5 7.50 Mouse SCID U251 Doxorubicin IV Multiple Untreated NA Median Survival

Zhao, Y. 2016 7 8.00 Mouse Athymic U87 Doxorubicin IV Multiple Saline Lipoosme Median survival

Byeon, H.J. 2016 6 10.67 Mouse Athymic U87 Doxorubicin IV Multiple Saline Nanoparticles Median Survival

Ramachandran, 2016 4 26.67 Mouse Athymic U87 Irinotecan IP Continuous Vehicle NA Median C. Survival

Ramachandran, 2016 4 26.67 Mouse Athymic U87 Irinotecan IP Continuous Vehicle NA Volume C. 251

Appendix X. Carrier types (Chapter 7).

This table contains a list of the types of carriers observed in the studies included in this study.

First Author Year Outcome Route of Type of carrier Measure administration

Marrero, L. 2014 Survival IV albumin

Zhong, Y. 2014 Survival IV nanoparticles

Zhong, Y. 2014 Volume IV nanoparticles

Yang, Y. 2013 Survival IV liposome

Serwer, L. 2011 Survival IV nanoliposome

Guo, L. 2011 Survival IV liposome

Vinchon-Petit, S. 2011 Survival IV drug eluting beads

Arai, T. 2010 Volume SC polymer

Kuroda, J. 2009 Survival IV micelles

Kuroda, J. 2009 Volume IV micelles

Kreuter, J. 2009 Survival IV nanoparticles

Petri, B. 2007 Survival IV nanoparticles

Ambruosi, A. 2006 Survival IV nanoparticles

252

Gomez-Manzano, C. 2006 Survival IP nanoparticles

Mamot, C. 2005 Volume IV liposome

Lesniak, M. 2005 Survival IC polymer

Steiniger, S. 2004 Survival IV nanoparticles

Sharma, U. 1997 Survival IV liposome

Glage, S. 2011 Survival IC drug eluting beads

Verreault, M. 2012 Survival IV liposome

Baltes, S. 2010 Survival IC drug eluting beads

253

Recinos, V. 2010 Survival IP polymer

Hsu, W. 2005 Survival IC polymer

Li, J. 2015 Survival IV polymer

Zhang, C. 2015 Survival IV liposome

Zhao, Y. 2016 Survival IV liposome

Byeon, H. 2016 Survival IV nanoparticles

254

References

1. T. R. Jue, E. Hovey, S. Davis, O. Carleton, K. L. McDonald, Incorporation of biomarkers in phase II studies of recurrent glioblastoma. Tumour Biol 36, 153- 162 (2015).

2. T. R. Jue, K. L. McDonald, The challenges associated with molecular targeted therapies for glioblastoma. Journal of Neuro-Oncology 127, 427-434 (2016).

3. A. I. o. H. a. Welfare, " Cancer in Australia 2017," Cancer series no.101. Cat. no. CAN 100. (AIHW, Canberra, 2017).

4. Q. T. Ostrom, H. Gittleman, J. Xu, C. Kromer, Y. Wolinsky, C. Kruchko, J. S. Barnholtz-Sloan, CBTRUS Statistical Report: Primary Brain and Other Central Nervous System Tumors Diagnosed in the United States in 2009–2013. Neuro- Oncology 18, v1-v75 (2016).

5. T.-M. Pham, Y. Fujino, S. Matsuda, T. Yoshimura, Premature mortality due to cancer in Japan, 1995 and 2005. International Journal of Cancer 127, 190-194 (2010).

6. N. G. Burnet, S. J. Jefferies, R. J. Benson, D. P. Hunt, F. P. Treasure, Years of life lost (YLL) from cancer is an important measure of population burden [mdash] and should be considered when allocating research funds. Br J Cancer 92, 241- 245 (2005).

7. Q. T. Ostrom, H. Gittleman, P. M. de Blank, J. L. Finlay, J. G. Gurney, R. McKean-Cowdin, D. S. Stearns, J. E. Wolff, M. Liu, Y. Wolinsky, C. Kruchko, J. S. Barnholtz-Sloan, American Brain Tumor Association Adolescent and Young Adult Primary Brain and Central Nervous System Tumors Diagnosed in the United States in 2008-2012. Neuro Oncol 18 Suppl 1, i1-i50 (2016).

8. I. Chakrabarti, M. Cockburn, W. Cozen, Y.-P. Wang, S. Preston-Martin, A population-based description of glioblastoma multiforme in Los Angeles County, 1974–1999. Cancer 104, 2798-2806 (2005).

9. N. Laperriere, M. Weller, R. Stupp, J. R. Perry, A. A. Brandes, W. Wick, M. J. van den Bent, Optimal management of elderly patients with glioblastoma. Cancer Treatment Reviews 39, 350-357 (2013).

10. L. Paszat, N. Laperriere, P. Groome, K. Schulze, W. Mackillop, E. Holowaty, A population-based study of glioblastoma multiforme. Int J Radiat Oncol Biol Phys 51, 100-107 (2001).

11. F. M. Iwamoto, A. S. Reiner, K. S. Panageas, E. B. Elkin, L. E. Abrey, Patterns of care in elderly glioblastoma patients. Ann Neurol 64, 628-634 (2008).

255

12. M. Wrensch, Y. Minn, T. Chew, M. Bondy, M. S. Berger, Epidemiology of primary brain tumors: current concepts and review of the literature. Neuro Oncol 4, 278-299 (2002).

13. V. Collins, BRAIN TUMOURS: CLASSIFICATION AND GENES. Journal of Neurology, Neurosurgery, and Psychiatry 75, ii2-ii11 (2004).

14. P. Bailey, H. Cushing, A classification of the tumours of the glioma group on a histogenetic basis, with a correlated study of prognosis. (J. B. Lippincott Company, Philadelphia, 1926), pp. 175.

15. D. N. Louis, H. Ohgaki, O. D. Wiestler, W. K. Cavenee, P. C. Burger, A. Jouvet, B. W. Scheithauer, P. Kleihues, The 2007 WHO Classification of Tumours of the Central Nervous System. Acta Neuropathologica 114, 97-109 (2007).

16. Histological typing of tumours of the central nervous system. P. Kleihues, P. C. Burger, B. W. Scheithauer, Eds., World Health Organization international histological classification of tumours (Springer, Heidelberg, 1993).

17. Pathology and genetics of tumours of the nervous system. P. Kleihues, W. K. Cavenee, Eds., World Health Organization Classification of Tumours (IARC Press, Lyon, 2000).

18. M. J. Riemenschneider, J. W. M. Jeuken, P. Wesseling, G. Reifenberger, Molecular diagnostics of gliomas: state of the art. Acta Neuropathologica 120, 567-584 (2010).

19. D. N. Louis, A. Perry, G. Reifenberger, A. von Deimling, D. Figarella-Branger, W. K. Cavenee, H. Ohgaki, O. D. Wiestler, P. Kleihues, D. W. Ellison, The 2016 World Health Organization Classification of Tumors of the Central Nervous System: a summary. Acta Neuropathologica 131, 803-820 (2016).

20. M. E. Hegi, R. Stupp, Neuro-oncology: In search of molecular markers of glioma in elderly patients. Nat Rev Neurol 9, 424-425 (2013).

21. F. B. Furnari, T. Fenton, R. M. Bachoo, A. Mukasa, J. M. Stommel, A. Stegh, W. C. Hahn, K. L. Ligon, D. N. Louis, C. Brennan, L. Chin, R. A. DePinho, W. K. Cavenee, Malignant astrocytic glioma: genetics, biology, and paths to treatment. Genes Dev 21, 2683-2710 (2007).

22. Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature 455, 1061-1068 (2008).

23. T. C. G. A. R. Network, Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature 455, 1061-1068 (2008).

24. Cameron W. Brennan, Roel G. W. Verhaak, A. McKenna, B. Campos, H. Noushmehr, Sofie R. Salama, S. Zheng, D. Chakravarty, J. Z. Sanborn, Samuel H. Berman, R. Beroukhim, B. Bernard, C.-J. Wu, G. Genovese, I. Shmulevich, J. Barnholtz-Sloan, L. Zou, R. Vegesna, Sachet A. Shukla, G. Ciriello, W. K. Yung, W. Zhang, C. Sougnez, T. Mikkelsen, K. Aldape, Darell D. Bigner, Erwin G.

256

Van Meir, M. Prados, A. Sloan, Keith L. Black, J. Eschbacher, G. Finocchiaro, W. Friedman, David W. Andrews, A. Guha, M. Iacocca, Brian P. O’Neill, G. Foltz, J. Myers, Daniel J. Weisenberger, R. Penny, R. Kucherlapati, Charles M. Perou, D. N. Hayes, R. Gibbs, M. Marra, Gordon B. Mills, E. Lander, P. Spellman, R. Wilson, C. Sander, J. Weinstein, M. Meyerson, S. Gabriel, Peter W. Laird, D. Haussler, G. Getz, L. Chin, C. Benz, J. Barnholtz-Sloan, W. Barrett, Q. Ostrom, Y. Wolinsky, Keith L. Black, B. Bose, Paul T. Boulos, M. Boulos, J. Brown, C. Czerinski, M. Eppley, M. Iacocca, T. Kempista, T. Kitko, Y. Koyfman, B. Rabeno, P. Rastogi, M. Sugarman, P. Swanson, K. Yalamanchii, Ilana P. Otey, Yingchun S. Liu, Y. Xiao, J. T. Auman, P.-C. Chen, A. Hadjipanayis, E. Lee, S. Lee, Peter J. Park, J. Seidman, L. Yang, R. Kucherlapati, S. Kalkanis, T. Mikkelsen, Laila M. Poisson, A. Raghunathan, L. Scarpace, B. Bernard, R. Bressler, A. Eakin, L. Iype, Richard B. Kreisberg, K. Leinonen, S. Reynolds, H. Rovira, V. Thorsson, I. Shmulevich, Matti J. Annala, R. Penny, J. Paulauskis, E. Curley, M. Hatfield, D. Mallery, S. Morris, T. Shelton, C. Shelton, M. Sherman, P. Yena, L. Cuppini, F. DiMeco, M. Eoli, G. Finocchiaro, E. Maderna, B. Pollo, M. Saini, S. Balu, Katherine A. Hoadley, L. Li, C. R. Miller, Y. Shi, Michael D. Topal, J. Wu, G. Dunn, C. Giannini, Brian P. O'Neill, B. A. Aksoy, Y. Antipin, L. Borsu, Samuel H. Berman, Cameron W. Brennan, E. Cerami, D. Chakravarty, G. Ciriello, J. Gao, B. Gross, A. Jacobsen, M. Ladanyi, A. Lash, Y. Liang, B. Reva, C. Sander, N. Schultz, R. Shen, Nicholas D. Socci, A. Viale, Martin L. Ferguson, Q.-R. Chen, John A. Demchok, Laura A. L. Dillon, Kenna R. M. Shaw, M. Sheth, R. Tarnuzzer, Z. Wang, L. Yang, T. Davidsen, Mark S. Guyer, Bradley A. Ozenberger, Heidi J. Sofia, J. Bergsten, J. Eckman, J. Harr, J. Myers, C. Smith, K. Tucker, C. Winemiller, Leigh A. Zach, Julia Y. Ljubimova, G. Eley, B. Ayala, Mark A. Jensen, A. Kahn, Todd D. Pihl, David A. Pot, Y. Wan, J. Eschbacher, G. Foltz, N. Hansen, P. Hothi, B. Lin, N. Shah, J.-g. Yoon, C. Lau, M. Berens, K. Ardlie, R. Beroukhim, Scott L. Carter, Andrew D. Cherniack, M. Noble, J. Cho, K. Cibulskis, D. DiCara, S. Frazer, Stacey B. Gabriel, N. Gehlenborg, J. Gentry, D. Heiman, J. Kim, R. Jing, Eric S. Lander, M. Lawrence, P. Lin, W. Mallard, M. Meyerson, Robert C. Onofrio, G. Saksena, S. Schumacher, C. Sougnez, P. Stojanov, B. Tabak, D. Voet, H. Zhang, L. Zou, G. Getz, Nathan N. Dees, L. Ding, Lucinda L. Fulton, Robert S. Fulton, K.-L. Kanchi, Elaine R. Mardis, Richard K. Wilson, Stephen B. Baylin, David W. Andrews, L. Harshyne, Mark L. Cohen, K. Devine, Andrew E. Sloan, Scott R. VandenBerg, Mitchel S. Berger, M. Prados, D. Carlin, B. Craft, K. Ellrott, M. Goldman, T. Goldstein, M. Grifford, D. Haussler, S. Ma, S. Ng, Sofie R. Salama, J. Z. Sanborn, J. Stuart, T. Swatloski, P. Waltman, J. Zhu, R. Foss, B. Frentzen, W. Friedman, R. McTiernan, A. Yachnis, D. N. Hayes, Charles M. Perou, S. Zheng, R. Vegesna, Y. Mao, R. Akbani, K. Aldape, O. Bogler, Gregory N. Fuller, W. Liu, Y. Liu, Y. Lu, G. Mills, A. Protopopov, X. Ren, Y. Sun, C.-J. Wu, W. K. A. Yung, W. Zhang, J. Zhang, K. Chen, John N. Weinstein, L. Chin, Roel G. W. Verhaak, H. Noushmehr, Daniel J. Weisenberger, Moiz S. Bootwalla, Phillip H. Lai, Timothy J. Triche Jr, David J. Van Den Berg, Peter W. Laird, David H. Gutmann, Norman L. Lehman, Erwin G. VanMeir, D. Brat, Jeffrey J. Olson, Gena M. Mastrogianakis, Narra S. Devi, Z. Zhang, D. Bigner, E. Lipp, R. McLendon, The Somatic Genomic Landscape of Glioblastoma. Cell 155, 462-477 (2013).

257

25. R. G. Verhaak, K. A. Hoadley, E. Purdom, V. Wang, Y. Qi, M. D. Wilkerson, C. R. Miller, L. Ding, T. Golub, J. P. Mesirov, G. Alexe, M. Lawrence, M. O'Kelly, P. Tamayo, B. A. Weir, S. Gabriel, W. Winckler, S. Gupta, L. Jakkula, H. S. Feiler, J. G. Hodgson, C. D. James, J. N. Sarkaria, C. Brennan, A. Kahn, P. T. Spellman, R. K. Wilson, T. P. Speed, J. W. Gray, M. Meyerson, G. Getz, C. M. Perou, D. N. Hayes, Integrated genomic analysis identifies clinically relevant subtypes of glioblastoma characterized by abnormalities in PDGFRA, IDH1, EGFR, and NF1. Cancer Cell 17, 98-110 (2010).

26. H.-B. Cheng, W. Yue, C. Xie, R.-Y. Zhang, S.-S. Hu, Z. Wang, IDH1 mutation is associated with improved overall survival in patients with glioblastoma: a meta- analysis. Tumor Biology 34, 3555-3559 (2013).

27. N. K. Kloosterhof, L. B. C. Bralten, H. J. Dubbink, P. J. French, M. J. van den Bent, Isocitrate dehydrogenase-1 mutations: a fundamentally new understanding of diffuse glioma? The Lancet Oncology 12, 83-91 (2011).

28. H. Noushmehr, D. J. Weisenberger, K. Diefes, H. S. Phillips, K. Pujara, B. P. Berman, F. Pan, C. E. Pelloski, E. P. Sulman, K. P. Bhat, R. G. W. Verhaak, K. A. Hoadley, D. N. Hayes, C. M. Perou, H. K. Schmidt, L. Ding, R. K. Wilson, D. Van Den Berg, H. Shen, H. Bengtsson, P. Neuvial, L. M. Cope, J. Buckley, J. G. Herman, S. B. Baylin, P. W. Laird, K. Aldape, Identification of a CpG Island Methylator Phenotype that Defines a Distinct Subgroup of Glioma. Cancer Cell 17, 510-522 (2010).

29. M. R. Gilbert, Recurrent Glioblastoma: A Fresh Look at Current Therapies and Emerging Novel Approaches. Seminars in Oncology 38, S21-S33 (2011).

30. R. Stupp , W. P. Mason , M. J. van den Bent , M. Weller , B. Fisher , M. J. B. Taphoorn , K. Belanger , A. A. Brandes , C. Marosi , U. Bogdahn , J. Curschmann , R. C. Janzer , S. K. Ludwin , T. Gorlia , A. Allgeier , D. Lacombe , J. G. Cairncross , E. Eisenhauer , R. O. Mirimanoff Radiotherapy plus Concomitant and Adjuvant Temozolomide for Glioblastoma. New England Journal of Medicine 352, 987-996 (2005).

31. E. S. Newlands, M. F. G. Stevens, S. R. Wedge, R. T. Wheelhouse, C. Brock, Temozolomide: A review of its discovery, chemical properties, pre-clinical development and clinical trials. Cancer Treatment Reviews 23, 35-61 (1997).

32. G. Minniti, R. Muni, G. Lanzetta, P. Marchetti, R. M. Enrici, Chemotherapy for glioblastoma: current treatment and future perspectives for cytotoxic and targeted agents. Anticancer Res 29, 5171-5184 (2009).

33. H. S. Friedman, T. Kerby, H. Calvert, Temozolomide and treatment of malignant glioma. Clin Cancer Res 6, 2585-2597 (2000).

34. W. P. Roos, B. Kaina, DNA damage-induced cell death: From specific DNA lesions to the DNA damage response and apoptosis. Cancer Letters 332, 237-248 (2013).

258

35. M. Esteller, S. R. Hamilton, P. C. Burger, S. B. Baylin, J. G. Herman, Inactivation of the DNA repair gene O6-methylguanine-DNA methyltransferase by promoter hypermethylation is a common event in primary human neoplasia. Cancer Res 59, 793-797 (1999).

36. M. Esteller, CpG island hypermethylation and tumor suppressor genes: a booming present, a brighter future. Oncogene 21, 5427-5440 (2002).

37. M. Esteller, Epigenetic lesions causing genetic lesions in human cancer. European Journal of Cancer 36, 2294-2300 (2000).

38. J. Felsberg, N. Thon, S. Eigenbrod, B. Hentschel, M. C. Sabel, M. Westphal, G. Schackert, F. W. Kreth, T. Pietsch, M. Löffler, M. Weller, G. Reifenberger, J. C. Tonn, N. for the German Glioma, Promoter methylation and expression of MGMT and the DNA mismatch repair genes MLH1, MSH2, MSH6 and PMS2 in paired primary and recurrent glioblastomas. International Journal of Cancer 129, 659- 670 (2011).

39. M. E. Hegi , A.-C. Diserens , T. Gorlia , M.-F. Hamou , N. de Tribolet , M. Weller , J. M. Kros , J. A. Hainfellner , W. Mason , L. Mariani , J. E. C. Bromberg , P. Hau , R. O. Mirimanoff , J. G. Cairncross , R. C. Janzer , R. Stupp MGMT Gene Silencing and Benefit from Temozolomide in Glioblastoma. New England Journal of Medicine 352, 997-1003 (2005).

40. R. Stupp, M. E. Hegi, W. P. Mason, M. J. van den Bent, M. J. Taphoorn, R. C. Janzer, S. K. Ludwin, A. Allgeier, B. Fisher, K. Belanger, P. Hau, A. A. Brandes, J. Gijtenbeek, C. Marosi, C. J. Vecht, K. Mokhtari, P. Wesseling, S. Villa, E. Eisenhauer, T. Gorlia, M. Weller, D. Lacombe, J. G. Cairncross, R. O. Mirimanoff, Effects of radiotherapy with concomitant and adjuvant temozolomide versus radiotherapy alone on survival in glioblastoma in a randomised phase III study: 5-year analysis of the EORTC-NCIC trial. Lancet Oncol 10, 459-466 (2009).

41. R. Brisman, E. M. Housepian, C. Chang, P. Duffy, E. Balis, Adjuvant nitrosourea therapy for glioblastoma. Arch Neurol 33, 745-750 (1976).

42. F. H. Hochberg, R. Linggood, L. Wolfson, W. H. Baker, P. Kornblith, Quality and duration of survival in glioblastoma multiforme: combined surgical, radiation, and lomustine therapy. Jama 241, 1016-1018 (1979).

43. H. C. Bock, M. J. A. Puchner, F. Lohmann, M. Schütze, S. Koll, R. Ketter, R. Buchalla, N. Rainov, S. R. Kantelhardt, V. Rohde, A. Giese, First-line treatment of malignant glioma with carmustine implants followed by concomitant radiochemotherapy: a multicenter experience. Neurosurgical Review 33, 441-449 (2010).

44. S. Valtonen, U. Timonen, P. Toivanen, H. Kalimo, L. Kivipelto, O. Heiskanen, G. Unsgaard, T. Kuurne, Interstitial chemotherapy with carmustine-loaded polymers for high-grade gliomas: a randomized double-blind study. Neurosurgery 41, 44-48; discussion 48-49 (1997).

259

45. M. Westphal, D. C. Hilt, E. Bortey, P. Delavault, R. Olivares, P. C. Warnke, I. R. Whittle, J. Jaaskelainen, Z. Ram, A phase 3 trial of local chemotherapy with biodegradable carmustine (BCNU) wafers (Gliadel wafers) in patients with primary malignant glioma. Neuro Oncol 5, 79-88 (2003).

46. S. A. Chowdhary, T. Ryken, H. B. Newton, Survival outcomes and safety of carmustine wafers in the treatment of high-grade gliomas: a meta-analysis. Journal of Neuro-Oncology 122, 367-382 (2015).

47. M. H. Cohen, Y. L. Shen, P. Keegan, R. Pazdur, FDA drug approval summary: bevacizumab (Avastin) as treatment of recurrent glioblastoma multiforme. Oncologist 14, 1131-1138 (2009).

48. S. P. Weathers, J. de Groot, VEGF Manipulation in Glioblastoma. Oncology (Williston Park) 29, 720-727 (2015).

49. J. J. Vredenburgh, A. Desjardins, J. E. Herndon, 2nd, J. M. Dowell, D. A. Reardon, J. A. Quinn, J. N. Rich, S. Sathornsumetee, S. Gururangan, M. Wagner, D. D. Bigner, A. H. Friedman, H. S. Friedman, Phase II trial of bevacizumab and irinotecan in recurrent malignant glioma. Clin Cancer Res 13, 1253-1259 (2007).

50. H. S. Friedman, M. D. Prados, P. Y. Wen, T. Mikkelsen, D. Schiff, L. E. Abrey, W. K. Yung, N. Paleologos, M. K. Nicholas, R. Jensen, J. Vredenburgh, J. Huang, M. Zheng, T. Cloughesy, Bevacizumab alone and in combination with irinotecan in recurrent glioblastoma. J Clin Oncol 27, 4733-4740 (2009).

51. O. L. Chinot, W. Wick, W. Mason, R. Henriksson, F. Saran, R. Nishikawa, A. F. Carpentier, K. Hoang-Xuan, P. Kavan, D. Cernea, A. A. Brandes, M. Hilton, L. Abrey, T. Cloughesy, Bevacizumab plus radiotherapy-temozolomide for newly diagnosed glioblastoma. N Engl J Med 370, 709-722 (2014).

52. M. R. Gilbert, J. J. Dignam, T. S. Armstrong, J. S. Wefel, D. T. Blumenthal, M. A. Vogelbaum, H. Colman, A. Chakravarti, S. Pugh, M. Won, R. Jeraj, P. D. Brown, K. A. Jaeckle, D. Schiff, V. W. Stieber, D. G. Brachman, M. Werner- Wasik, I. W. Tremont-Lukats, E. P. Sulman, K. D. Aldape, W. J. Curran, Jr., M. P. Mehta, A randomized trial of bevacizumab for newly diagnosed glioblastoma. N Engl J Med 370, 699-708 (2014).

53. A. E. Guttmacher , F. S. Collins Genomic Medicine — A Primer. New England Journal of Medicine 347, 1512-1520 (2002).

54. I. F. Tannock, J. A. Hickman, Limits to Personalized Cancer Medicine. New England Journal of Medicine 375, 1289-1294 (2016).

55. K. Pavelić, T. Martinović, S. Kraljević Pavelić, Do we understand the personalized medicine paradigm?: Personalized medicine marks the beginning of a new attitude in medicine. EMBO Reports 16, 133-136 (2015).

56. J. L. Jameson, D. L. Longo, Precision Medicine — Personalized, Problematic, and Promising. New England Journal of Medicine 372, 2229-2234 (2015).

260

57. National Research Council Committee on a Framework for Developing a New Taxonomy of Disease in Toward Precision Medicine: Building a Knowledge Network for Biomedical Research and a New Taxonomy of Disease. (National Academies Press (US) National Academy of Sciences., Washington (DC), 2011).

58. The University of Texas MD Anderson Cancer Center. Targeted Therapy (Precision Medicine). (2017).

59. W. R. Vogler, Therapeutic apheresis: where we've been and where we are going. Prog Clin Biol Res 106, 1-13 (1982).

60. A. V. Biankin, S. J. Chanock, The road ahead: less travelled and more arduous than initially envisioned. Human Genetics 130, 1-2 (2011).

61. R. L. Schilsky, Personalized medicine in oncology: the future is now. Nat Rev Drug Discov 9, 363-366 (2010).

62. T. Harris, Gene and drug matrix for personalized cancer therapy. Nat Rev Drug Discov 9, 660-660 (2010).

63. D. S. Salomon, R. Brandt, F. Ciardiello, N. Normanno, Epidermal growth factor- related peptides and their receptors in human malignancies. Crit Rev Oncol Hematol 19, 183-232 (1995).

64. D. J. Slamon, G. M. Clark, S. G. Wong, W. J. Levin, A. Ullrich, W. L. McGuire, Human breast cancer: correlation of relapse and survival with amplification of the HER-2/neu oncogene. Science 235, 177-182 (1987).

65. G. Gasparini, R. Longo, F. Torino, A. Morabito, Therapy of breast cancer with molecular targeting agents. Ann Oncol 16 Suppl 4, iv28-36 (2005).

66. K. McKeage, C. M. Perry, Trastuzumab: a review of its use in the treatment of metastatic breast cancer overexpressing HER2. Drugs 62, 209-243 (2002).

67. M. A. Cobleigh, C. L. Vogel, D. Tripathy, N. J. Robert, S. Scholl, L. Fehrenbacher, J. M. Wolter, V. Paton, S. Shak, G. Lieberman, D. J. Slamon, Multinational study of the efficacy and safety of humanized anti-HER2 monoclonal antibody in women who have HER2-overexpressing metastatic breast cancer that has progressed after chemotherapy for metastatic disease. J Clin Oncol 17, 2639-2648 (1999).

68. J. Baselga, D. Tripathy, J. Mendelsohn, S. Baughman, C. C. Benz, L. Dantis, N. T. Sklarin, A. D. Seidman, C. A. Hudis, J. Moore, P. P. Rosen, T. Twaddell, I. C. Henderson, L. Norton, Phase II study of weekly intravenous recombinant humanized anti-p185HER2 monoclonal antibody in patients with HER2/neu- overexpressing metastatic breast cancer. J Clin Oncol 14, 737-744 (1996).

69. E. A. Perez, E. H. Romond, V. J. Suman, J. H. Jeong, G. Sledge, C. E. Geyer, Jr., S. Martino, P. Rastogi, J. Gralow, S. M. Swain, E. P. Winer, G. Colon-Otero, N. E. Davidson, E. Mamounas, J. A. Zujewski, N. Wolmark, Trastuzumab Plus Adjuvant Chemotherapy for Human Epidermal Growth Factor Receptor 2-

261

Positive Breast Cancer: Planned Joint Analysis of Overall Survival From NSABP B-31 and NCCTG N9831. J Clin Oncol 32, 3744-3752 (2014).

70. R. H. Alvarez, V. Valero, G. N. Hortobagyi, Emerging Targeted Therapies for Breast Cancer. Journal of Clinical Oncology 28, 3366-3379 (2010).

71. D. Cunningham, Y. Humblet, S. Siena, D. Khayat, H. Bleiberg, A. Santoro, D. Bets, M. Mueser, A. Harstrick, C. Verslype, I. Chau, E. Van Cutsem, Cetuximab monotherapy and cetuximab plus irinotecan in irinotecan-refractory metastatic colorectal cancer. N Engl J Med 351, 337-345 (2004).

72. C. Bokemeyer, I. Bondarenko, J. T. Hartmann, F. de Braud, G. Schuch, A. Zubel, I. Celik, M. Schlichting, P. Koralewski, Efficacy according to biomarker status of cetuximab plus FOLFOX-4 as first-line treatment for metastatic colorectal cancer: the OPUS study. Ann Oncol 22, 1535-1546 (2011).

73. D. J. Jonker, C. J. O'Callaghan, C. S. Karapetis, J. R. Zalcberg, D. Tu, H. J. Au, S. R. Berry, M. Krahn, T. Price, R. J. Simes, N. C. Tebbutt, G. van Hazel, R. Wierzbicki, C. Langer, M. J. Moore, Cetuximab for the treatment of colorectal cancer. N Engl J Med 357, 2040-2048 (2007).

74. E. Van Cutsem, C. H. Kohne, E. Hitre, J. Zaluski, C. R. Chang Chien, A. Makhson, G. D'Haens, T. Pinter, R. Lim, G. Bodoky, J. K. Roh, G. Folprecht, P. Ruff, C. Stroh, S. Tejpar, M. Schlichting, J. Nippgen, P. Rougier, Cetuximab and chemotherapy as initial treatment for metastatic colorectal cancer. N Engl J Med 360, 1408-1417 (2009).

75. B. J. Giantonio, P. J. Catalano, N. J. Meropol, P. J. O'Dwyer, E. P. Mitchell, S. R. Alberts, M. A. Schwartz, A. B. Benson, 3rd, Bevacizumab in combination with oxaliplatin, fluorouracil, and leucovorin (FOLFOX4) for previously treated metastatic colorectal cancer: results from the Eastern Cooperative Oncology Group Study E3200. J Clin Oncol 25, 1539-1544 (2007).

76. L. B. Saltz, S. Clarke, E. Diaz-Rubio, W. Scheithauer, A. Figer, R. Wong, S. Koski, M. Lichinitser, T. S. Yang, F. Rivera, F. Couture, F. Sirzen, J. Cassidy, Bevacizumab in combination with oxaliplatin-based chemotherapy as first-line therapy in metastatic colorectal cancer: a randomized phase III study. J Clin Oncol 26, 2013-2019 (2008).

77. H. Hurwitz, L. Fehrenbacher, W. Novotny, T. Cartwright, J. Hainsworth, W. Heim, J. Berlin, A. Baron, S. Griffing, E. Holmgren, N. Ferrara, G. Fyfe, B. Rogers, R. Ross, F. Kabbinavar, Bevacizumab plus irinotecan, fluorouracil, and leucovorin for metastatic colorectal cancer. N Engl J Med 350, 2335-2342 (2004).

78. K. O. Lam, V. H. F. Lee, R. K. Y. Liu, T. W. Leung, D. L. W. Kwong, Bevacizumab-containing regimens after cetuximab failure in Kras wild-type metastatic colorectal carcinoma. Oncology Letters 5, 637-640 (2013).

262

79. H. Prenen, S. Tejpar, E. V. Cutsem, New Strategies for Treatment of KRAS Mutant Metastatic Colorectal Cancer. Clinical Cancer Research 16, 2921-2926 (2010).

80. M. Reck, N. van Zandwijk, C. Gridelli, Z. Baliko, D. Rischin, S. Allan, M. Krzakowski, D. Heigener, Erlotinib in advanced non-small cell lung cancer: efficacy and safety findings of the global phase IV Tarceva Lung Cancer Survival Treatment study. J Thorac Oncol 5, 1616-1622 (2010).

81. P. A. Janne, J. A. Engelman, B. E. Johnson, Epidermal growth factor receptor mutations in non-small-cell lung cancer: implications for treatment and tumor biology. J Clin Oncol 23, 3227-3234 (2005).

82. A. T. Shaw, D.-W. Kim, R. Mehra, D. S. W. Tan, E. Felip, L. Q. M. Chow, D. R. Camidge, J. Vansteenkiste, S. Sharma, T. De Pas, G. J. Riely, B. J. Solomon, J. Wolf, M. Thomas, M. Schuler, G. Liu, A. Santoro, Y. Y. Lau, M. Goldwasser, A. L. Boral, J. A. Engelman, Ceritinib in ALK-Rearranged Non–Small-Cell Lung Cancer. New England Journal of Medicine 370, 1189-1197 (2014).

83. K. L. Nathanson, Using genetics and genomics strategies to personalize therapy for cancer: focus on melanoma. Biochemical pharmacology 80, 755-761 (2010).

84. K. Flaherty, I. Puzanov, J. Sosman, K. Kim, A. Ribas, G. McArthur, R. J. Lee, J. F. Grippo, K. Nolop, P. Chapman, Phase I study of PLX4032: Proof of concept for V600E BRAF mutation as a therapeutic target in human cancer. Journal of Clinical Oncology 27, 9000-9000 (2009).

85. J. Tsai, J. T. Lee, W. Wang, J. Zhang, H. Cho, S. Mamo, R. Bremer, S. Gillette, J. Kong, N. K. Haass, K. Sproesser, L. Li, K. S. Smalley, D. Fong, Y. L. Zhu, A. Marimuthu, H. Nguyen, B. Lam, J. Liu, I. Cheung, J. Rice, Y. Suzuki, C. Luu, C. Settachatgul, R. Shellooe, J. Cantwell, S. H. Kim, J. Schlessinger, K. Y. Zhang, B. L. West, B. Powell, G. Habets, C. Zhang, P. N. Ibrahim, P. Hirth, D. R. Artis, M. Herlyn, G. Bollag, Discovery of a selective inhibitor of oncogenic B-Raf kinase with potent antimelanoma activity. Proc Natl Acad Sci U S A 105, 3041- 3046 (2008).

86. T. T. Batchelor, P. Mulholland, B. Neyns, L. B. Nabors, M. Campone, A. Wick, W. Mason, T. Mikkelsen, S. Phuphanich, L. S. Ashby, J. Degroot, R. Gattamaneni, L. Cher, M. Rosenthal, F. Payer, J. M. Jurgensmeier, R. K. Jain, A. G. Sorensen, J. Xu, Q. Liu, M. van den Bent, Phase III randomized trial comparing the efficacy of cediranib as monotherapy, and in combination with lomustine, versus lomustine alone in patients with recurrent glioblastoma. J Clin Oncol 31, 3212-3218 (2013).

87. K. Kesavabhotla, C. D. Schlaff, B. Shin, L. Mubita, R. Kaplan, A. J. Tsiouris, S. C. Pannullo, P. Christos, E. Lavi, R. Scheff, J. A. Boockvar, Phase I/II study of oral erlotinib for treatment of relapsed/refractory glioblastoma multiforme and anaplastic astrocytoma. J Exp Ther Oncol 10, 71-81 (2012).

263

88. J. L. Clarke, A. M. Molinaro, J. J. Phillips, N. A. Butowski, S. M. Chang, A. Perry, J. F. Costello, A. A. DeSilva, J. E. Rabbitt, M. D. Prados, A single-institution phase II trial of radiation, temozolomide, erlotinib, and bevacizumab for initial treatment of glioblastoma. Neuro-Oncology 16, 984-990 (2014).

89. J. J. Raizer, P. Giglio, J. Hu, M. Groves, R. Merrell, C. Conrad, S. Phuphanich, V. K. Puduvalli, M. Loghin, N. Paleologos, Y. Yuan, D. Liu, A. Rademaker, W. K. Yung, B. Vaillant, J. Rudnick, M. Chamberlain, N. Vick, S. Grimm, I. W. Tremont-Lukats, J. De Groot, K. Aldape, M. R. Gilbert, A phase II study of bevacizumab and erlotinib after radiation and temozolomide in MGMT unmethylated GBM patients. J Neurooncol 126, 185-192 (2016).

90. D. M. Peereboom, M. S. Ahluwalia, X. Ye, J. G. Supko, S. L. Hilderbrand, S. Phuphanich, L. B. Nabors, M. R. Rosenfeld, T. Mikkelsen, S. A. Grossman, C. for the New Approaches to Brain Tumor Therapy, NABTT 0502: a phase II and pharmacokinetic study of erlotinib and sorafenib for patients with progressive or recurrent glioblastoma multiforme. Neuro-Oncology 15, 490-496 (2013).

91. IMMUNOTHERAPY. Neuro-Oncology 13, iii34-iii40 (2011).

92. M. Weller, N. Butowski, D. D. Tran, L. D. Recht, M. Lim, H. Hirte, L. Ashby, L. Mechtler, S. A. Goldlust, F. Iwamoto, J. Drappatz, D. M. O'Rourke, M. Wong, M. G. Hamilton, G. Finocchiaro, J. Perry, W. Wick, J. Green, Y. He, C. D. Turner, M. J. Yellin, T. Keler, T. A. Davis, R. Stupp, J. H. Sampson, Rindopepimut with temozolomide for patients with newly diagnosed, EGFRvIII-expressing glioblastoma (ACT IV): a randomised, double-blind, international phase 3 trial. Lancet Oncol, (2017).

93. S. X. Lee, A. Tunkyi, E. Wong, K. D. Swanson, Mitosis interference of cancer cells during anaphase by electric field from NovoTTF-100A: An update. Journal of Clinical Oncology 30, e21078-e21078 (2012).

94. E. D. Kirson, Z. Gurvich, R. Schneiderman, E. Dekel, A. Itzhaki, Y. Wasserman, R. Schatzberger, Y. Palti, Disruption of cancer cell replication by alternating electric fields. Cancer Res 64, 3288-3295 (2004).

95. R. Stupp, E. T. Wong, A. A. Kanner, D. Steinberg, H. Engelhard, V. Heidecke, E. D. Kirson, S. Taillibert, F. Liebermann, V. Dbaly, Z. Ram, J. L. Villano, N. Rainov, U. Weinberg, D. Schiff, L. Kunschner, J. Raizer, J. Honnorat, A. Sloan, M. Malkin, J. C. Landolfi, F. Payer, M. Mehdorn, R. J. Weil, S. C. Pannullo, M. Westphal, M. Smrcka, L. Chin, H. Kostron, S. Hofer, J. Bruce, R. Cosgrove, N. Paleologous, Y. Palti, P. H. Gutin, NovoTTF-100A versus physician's choice chemotherapy in recurrent glioblastoma: a randomised phase III trial of a novel treatment modality. Eur J Cancer 48, 2192-2202 (2012).

96. J. F. de Groot, K. R. Lamborn, S. M. Chang, M. R. Gilbert, T. F. Cloughesy, K. Aldape, J. Yao, E. F. Jackson, F. Lieberman, H. I. Robins, Phase II study of aflibercept in recurrent malignant glioma: a North American Brain Tumor Consortium study. Journal of Clinical Oncology 29, 2689-2695 (2011).

264

97. E. Galanis, S. K. Anderson, J. M. Lafky, J. H. Uhm, C. Giannini, S. K. Kumar, T. K. Kimlinger, D. W. Northfelt, P. J. Flynn, K. A. Jaeckle, Phase II study of bevacizumab in combination with sorafenib in recurrent glioblastoma (N0776): A North Central Cancer Treatment Group Trial. Clinical Cancer Research 19, 4816- 4823 (2013).

98. B. Hasselbalch, J. G. Eriksen, H. Broholm, I. J. Christensen, K. Grunnet, M. R. Horsman, H. S. Poulsen, M. T. Stockhausen, U. Lassen, Prospective evaluation of angiogenic, hypoxic and EGFR-related biomarkers in recurrent glioblastoma multiforme treated with cetuximab, bevacizumab and irinotecan. APMIS 118, 585-594 (2010).

99. D. A. Reardon, A. Desjardins, K. B. Peters, S. Gururangan, J. H. Sampson, R. E. McLendon, J. E. Herndon, 2nd, A. Bulusu, S. Threatt, A. H. Friedman, J. J. Vredenburgh, H. S. Friedman, Phase II study of carboplatin, irinotecan, and bevacizumab for bevacizumab naive, recurrent glioblastoma. J Neurooncol 107, 155-164 (2012).

100. D. A. Reardon, A. Desjardins, K. Peters, S. Gururangan, J. Sampson, J. N. Rich, R. McLendon, J. E. Herndon, 2nd, J. Marcello, S. Threatt, A. H. Friedman, J. J. Vredenburgh, H. S. Friedman, Phase II study of metronomic chemotherapy with bevacizumab for recurrent glioblastoma after progression on bevacizumab therapy. J Neurooncol 103, 371-379 (2011).

101. J. J. Raizer, S. Grimm, M. C. Chamberlain, M. K. Nicholas, J. P. Chandler, K. Muro, S. Dubner, A. W. Rademaker, J. Renfrow, M. Bredel, A phase 2 trial of single-agent bevacizumab given in an every-3-week schedule for patients with recurrent high-grade gliomas. Cancer 116, 5297-5305 (2010).

102. S. Sathornsumetee, A. Desjardins, J. J. Vredenburgh, R. E. McLendon, J. Marcello, J. E. Herndon, A. Mathe, M. Hamilton, J. N. Rich, J. A. Norfleet, S. Gururangan, H. S. Friedman, D. A. Reardon, Phase II trial of bevacizumab and erlotinib in patients with recurrent malignant glioma. Neuro Oncol 12, 1300-1310 (2010).

103. R. Soffietti, E. Trevisan, R. Ruda, L. Bertero, C. Bosa, M. G. Fabrini, I. Lolli, Phase II trial of bevacizumab with fotemustine in recurrent glioblastoma: Final results of a multicenter study of AINO (Italian Association of Neuro-oncology). J Clin Oncol 29, 2027 (2011).

104. T. T. Batchelor, D. G. Duda, E. di Tomaso, M. Ancukiewicz, S. R. Plotkin, E. Gerstner, A. F. Eichler, J. Drappatz, F. H. Hochberg, T. Benner, D. N. Louis, K. S. Cohen, H. Chea, A. Exarhopoulos, J. S. Loeffler, M. A. Moses, P. Ivy, A. G. Sorensen, P. Y. Wen, R. K. Jain, Phase II study of cediranib, an oral pan-vascular endothelial growth factor receptor tyrosine kinase inhibitor, in patients with recurrent glioblastoma. J Clin Oncol 28, 2817-2823 (2010).

105. U. Lassen, M. Sorensen, T. B. Gaziel, B. Hasselbalch, H. S. Poulsen, Phase II study of bevacizumab and temsirolimus combination therapy for recurrent glioblastoma multiforme. Anticancer Res 33, 1657-1660 (2013).

265

106. K. F. Hofland, S. Hansen, M. Sorensen, S. Engelholm, H. P. Schultz, A. Muhic, K. Grunnet, A. Ask, J. C. Costa, C. Kristiansen, C. Thomsen, H. S. Poulsen, U. Lassen, Neoadjuvant bevacizumab and irinotecan versus bevacizumab and temozolomide followed by concomitant chemoradiotherapy in newly diagnosed glioblastoma multiforme: A randomized phase II study. Acta Oncol 53, 939-944 (2014).

107. S. Moller, K. Grunnet, S. Hansen, H. Schultz, M. Holmberg, M. Sorensen, H. S. Poulsen, U. Lassen, A phase II trial with bevacizumab and irinotecan for patients with primary brain tumors and progression after standard therapy. Acta Oncol 51, 797-804 (2012).

108. M. E. van Linde, J. J. Verhoeff, D. J. Richel, W. R. van Furth, J. C. Reijneveld, H. M. Verheul, L. J. Stalpers, Bevacizumab in combination with radiotherapy and temozolomide for patients with newly diagnosed glioblastoma multiforme. Oncologist 20, 107-108 (2015).

109. D. M. Peereboom, M. S. Ahluwalia, X. Ye, J. G. Supko, S. L. Hilderbrand, S. Phuphanich, L. B. Nabors, M. R. Rosenfeld, T. Mikkelsen, S. A. Grossman, NABTT 0502: a phase II and pharmacokinetic study of erlotinib and sorafenib for patients with progressive or recurrent glioblastoma multiforme. Neuro Oncol 15, 490-496 (2013).

110. J. J. Raizer, L. E. Abrey, A. B. Lassman, S. M. Chang, K. R. Lamborn, J. G. Kuhn, W. K. Yung, M. R. Gilbert, K. A. Aldape, P. Y. Wen, H. A. Fine, M. Mehta, L. M. Deangelis, F. Lieberman, T. F. Cloughesy, H. I. Robins, J. Dancey, M. D. Prados, A phase II trial of erlotinib in patients with recurrent malignant gliomas and nonprogressive glioblastoma multiforme postradiation therapy. Neuro Oncol 12, 95-103 (2010).

111. D. A. Reardon, A. Desjardins, J. J. Vredenburgh, S. Gururangan, A. H. Friedman, J. E. Herndon, 2nd, J. Marcello, J. A. Norfleet, R. E. McLendon, J. H. Sampson, H. S. Friedman, Phase 2 trial of erlotinib plus sirolimus in adults with recurrent glioblastoma. J Neurooncol 96, 219-230 (2010).

112. T. Aoki, T. Mizutani, K. Nojima, T. Takagi, R. Okumura, Y. Yuba, T. Ueba, J. A. Takahashi, S. Miyatake, K. Nozaki, W. Taki, M. Matsutani, Phase II study of ifosfamide, carboplatin, and etoposide in patients with a first recurrence of glioblastoma multiforme. J Neurosurg 112, 50-56 (2010).

113. N. Schafer, J. Tichy, S. Thanendrarajan, Y. Kim, M. Stuplich, F. Mack, J. Rieger, M. Simon, B. Scheffler, J. Bostrom, J. P. Steinbach, U. Herrlinger, M. Glas, Ifosfamide, carboplatin and etoposide in recurrent malignant glioma. Oncology 80, 330-332 (2011).

114. G. Dresemann, M. Weller, M. A. Rosenthal, U. Wedding, W. Wagner, E. Engel, B. Heinrich, R. Mayer-Steinacker, A. Karup-Hansen, O. Fluge, A. Nowak, M. Mehdorn, E. Schleyer, D. Krex, I. N. Olver, J. P. Steinbach, C. Hosius, C. Sieder, G. Sorenson, R. Parker, Z. Nikolova, Imatinib in combination with hydroxyurea versus hydroxyurea alone as oral therapy in patients with progressive pretreated

266

glioblastoma resistant to standard dose temozolomide. J Neurooncol 96, 393-402 (2010).

115. K. A. Jaeckle, K. V. Ballman, C. Giannini, P. J. Schomberg, M. M. Ames, J. M. Reid, R. M. McGovern, S. L. Safgren, E. Galanis, J. H. Uhm, P. D. Brown, J. E. Hammack, R. Arusell, D. A. Nikcevich, R. F. Morton, D. B. Wender, J. C. Buckner, Phase II NCCTG trial of RT + irinotecan and adjuvant BCNU plus irinotecan for newly diagnosed GBM. J Neurooncol 99, 73-80 (2010).

116. D. A. Reardon, M. D. Groves, P. Y. Wen, L. Nabors, T. Mikkelsen, S. Rosenfeld, J. Raizer, J. Barriuso, R. E. McLendon, A. B. Suttle, A phase I/II trial of pazopanib in combination with lapatinib in adult patients with relapsed malignant glioma. Clinical Cancer Research 19, 900-908 (2013).

117. A. Muhic, H. S. Poulsen, M. Sorensen, K. Grunnet, U. Lassen, Phase II open-label study of nintedanib in patients with recurrent glioblastoma multiforme. Journal of Neuro-Oncology 111, 205-212 (2013).

118. M. T. Solomon, N. Miranda, E. Jorrin, I. Chon, J. J. Marinello, J. Alert, P. Lorenzo-Luaces, T. Crombet, Nimotuzumab in combination with radiotherapy in high grade glioma patients: a single institution experience. Cancer Biol Ther 15, 504-509 (2014).

119. Y. Wang, L. Pan, X. F. Sheng, S. Chen, J. Z. Dai, Nimotuzumab, a humanized monoclonal antibody specific for the EGFR, in combination with temozolomide and radiation therapy for newly diagnosed glioblastoma multiforme: First results in Chinese patients. Asia Pac J Clin Oncol 12, e23-29 (2016).

120. C. Oehler, K. Frei, E. J. Rushing, P. M. McSheehy, D. Weber, P. R. Allegrini, D. Weniger, U. M. Lutolf, A. Knuth, Y. Yonekawa, K. Barath, A. Broggini-Tenzer, M. Pruschy, S. Hofer, Patupilone (epothilone B) for recurrent glioblastoma: clinical outcome and translational analysis of a single-institution phase I/II trial. Oncology 83, 1-9 (2012).

121. F. M. Iwamoto, K. R. Lamborn, H. I. Robins, M. P. Mehta, S. M. Chang, N. A. Butowski, L. M. Deangelis, L. E. Abrey, W. T. Zhang, M. D. Prados, H. A. Fine, Phase II trial of pazopanib (GW786034), an oral multi-targeted angiogenesis inhibitor, for adults with recurrent glioblastoma (North American Brain Tumor Consortium Study 06-02). Neuro Oncol 12, 855-861 (2010).

122. P. Y. Wen, D. Schiff, T. F. Cloughesy, J. J. Raizer, J. Laterra, M. Smitt, M. Wolf, K. S. Oliner, A. Anderson, M. Zhu, E. Loh, D. A. Reardon, A phase II study evaluating the efficacy and safety of AMG 102 (rilotumumab) in patients with recurrent glioblastoma. Neuro-Oncology 13, 437-446 (2011).

123. F. Zustovich, L. Landi, G. Lombardi, C. Porta, L. Galli, A. Fontana, D. Amoroso, C. Galli, M. Andreuccetti, A. Falcone, V. Zagonel, Sorafenib plus daily low-dose temozolomide for relapsed glioblastoma: A phase ii study. Anticancer Research 33, 3487-3494 (2013).

267

124. T. N. Kreisl, P. Smith, J. Sul, C. Salgado, F. M. Iwamoto, J. H. Shih, H. A. Fine, Continuous daily sunitinib for recurrent glioblastoma. J Neurooncol 111, 41-48 (2013).

125. E. Pan, D. Yu, B. Yue, L. Potthast, S. Chowdhary, P. Smith, M. Chamberlain, A prospective phase II single-institution trial of sunitinib for recurrent malignant glioma. Journal of neuro-oncology 110, 111-118 (2012).

126. F. M. Iwamoto, T. N. Kreisl, L. Kim, J. P. Duic, J. A. Butman, P. S. Albert, H. A. Fine, Phase 2 trial of talampanel, a glutamate receptor inhibitor, for adults with recurrent malignant gliomas. Cancer 116, 1776-1782 (2010).

127. W. P. Mason, K. Belanger, G. Nicholas, I. Vallieres, D. Mathieu, P. Kavan, A. Desjardins, A. Omuro, D. Reymond, A phase II study of the Ras-MAPK signaling pathway inhibitor TLN-4601 in patients with glioblastoma at first progression. J Neurooncol 107, 343-349 (2012).

128. T. N. Kreisl, K. A. McNeill, J. Sul, F. M. Iwamoto, J. Shih, H. A. Fine, A phase I/II trial of vandetanib for patients with recurrent malignant glioma. Neuro Oncol 14, 1519-1526 (2012).

129. B. B. Friday, S. K. Anderson, J. Buckner, C. Yu, C. Giannini, F. Geoffroy, J. Schwerkoske, M. Mazurczak, H. Gross, E. Pajon, K. Jaeckle, E. Galanis, Phase II trial of vorinostat in combination with bortezomib in recurrent glioblastoma: a north central cancer treatment group study. Neuro Oncol 14, 215-221 (2012).

130. A. Olar, K. D. Aldape, Using the molecular classification of glioblastoma to inform personalized treatment. J Pathol 232, 165-177 (2014).

131. T. F. Cloughesy, W. K. Cavenee, P. S. Mischel, Glioblastoma: from molecular pathology to targeted treatment. Annu Rev Pathol 9, 1-25 (2014).

132. H. Zhao, K. Cui, F. Nie, L. Wang, M. B. Brandl, G. Jin, F. Li, Y. Mao, Z. Xue, A. Rodriguez, J. Chang, S. T. Wong, The effect of mTOR inhibition alone or combined with MEK inhibitors on brain metastasis: an in vivo analysis in triple- negative breast cancer models. Breast Cancer Res Treat 131, 425-436 (2012).

133. S. A. Chung, University of New south Wales, Sydney, Australia (2017).

134. R. B. M. de Vries, C. R. Hooijmans, A. Tillema, M. Leenaars, M. Ritskes- Hoitinga, A search filter for increasing the retrieval of animal studies in Embase. Laboratory Animals 45, 268-270 (2011).

135. C. R. Hooijmans, A. Tillema, M. Leenaars, M. Ritskes-Hoitinga, Enhancing search efficiency by means of a search filter for finding all studies on animal experimentation in PubMed. Laboratory Animals 44, 170-175 (2010).

136. S. Amarasingh, M. R. Macleod, I. R. Whittle, What is the translational efficacy of chemotherapeutic drug research in neuro-oncology? A systematic review and meta-analysis of the efficacy of BCNU and CCNU in animal models of glioma. J Neurooncol 91, 117-125 (2009).

268

137. T. C. Hirst, H. M. Vesterinen, E. S. Sena, K. J. Egan, M. R. Macleod, I. R. Whittle, Systematic review and meta-analysis of temozolomide in animal models of glioma: was clinical efficacy predicted? Br J Cancer 108, 64-71 (2013).

138. T. Hirst, H. Vesterinen, S. Conlin, K. Egan, A. Antonic, A. Lawson McLean, M. Macleod, R. Grant, P. Brennan, E. Sena, A systematic review and meta‐analysis of gene therapy in animal models of cerebral glioma: why did promise not translate to human therapy? Evidence-based Preclinical Medicine 1, 21-33 (2014).

139. A. Fabregat, K. Sidiropoulos, P. Garapati, M. Gillespie, K. Hausmann, R. Haw, B. Jassal, S. Jupe, F. Korninger, S. McKay, L. Matthews, B. May, M. Milacic, K. Rothfels, V. Shamovsky, M. Webber, J. Weiser, M. Williams, G. Wu, L. Stein, H. Hermjakob, P. D'Eustachio, The Reactome pathway Knowledgebase. Nucleic Acids Res 44, D481-487 (2016).

140. M. Milacic, R. Haw, K. Rothfels, G. Wu, D. Croft, H. Hermjakob, P. D'Eustachio, L. Stein, Annotating cancer variants and anti-cancer therapeutics in reactome. Cancers 4, 1180-1211 (2012).

141. H. M. Vesterinen, E. S. Sena, K. J. Egan, T. C. Hirst, L. Churolov, G. L. Currie, A. Antonic, D. W. Howells, M. R. Macleod, Meta-analysis of data from animal studies: a practical guide. J Neurosci Methods 221, 92-102 (2014).

142. L. V. Hedges, J. Gurevitch, P. S. Curtis, The Meta-Analysis of Response Ratios in Experimental Ecology. Ecology 80, 1150-1156 (1999).

143. S. Duval, R. Tweedie, Trim and fill: A simple funnel-plot-based method of testing and adjusting for publication bias in meta-analysis. Biometrics 56, 455-463 (2000).

144. M. Egger, G. Davey Smith, M. Schneider, C. Minder, Bias in meta-analysis detected by a simple, graphical test. Bmj 315, 629-634 (1997).

145. M. Gordon, T. Lumley. (2017).

146. J. L. Peters, A. J. Sutton, D. R. Jones, K. R. Abrams, L. Rushton, Contour- enhanced meta-analysis funnel plots help distinguish publication bias from other causes of asymmetry. Journal of clinical epidemiology 61, 991-996 (2008).

147. P. C. Huszthy, I. Daphu, S. P. Niclou, D. Stieber, J. M. Nigro, P. Ø. Sakariassen, H. Miletic, F. Thorsen, R. Bjerkvig, In vivo models of primary brain tumors: pitfalls and perspectives. Neuro-oncology 14, 979-993 (2012).

148. K. Lenting, R. Verhaak, M. ter Laan, P. Wesseling, W. Leenders, Glioma: experimental models and reality. Acta Neuropathologica 133, 263-282 (2017).

149. J. A. N. PontÉN, E. H. Macintyre, LONG TERM CULTURE OF NORMAL AND NEOPLASTIC HUMAN GLIA. Acta Pathologica Microbiologica Scandinavica 74, 465-486 (1968).

269

150. C.-P. Day, G. Merlino, T. Van Dyke, Preclinical Mouse Cancer Models: A Maze of Opportunities and Challenges. Cell 163, 39-53 (2015).

151. B. C. Horten, G. A. Basler, W. R. Shapiro, Xenograft of Human Malignant Glial Tumors into Brains of Nude MiceA Histopathological Study. Journal of Neuropathology & Experimental Neurology 40, 493-511 (1981).

152. M. Allen, M. Bjerke, H. Edlund, S. Nelander, B. Westermark, Origin of the U87MG glioma cell line: Good news and bad news. Science Translational Medicine 8, 354re353-354re353 (2016).

153. X. Lan, D. J. Jörg, F. M. G. Cavalli, L. M. Richards, L. V. Nguyen, R. J. Vanner, P. Guilhamon, L. Lee, M. M. Kushida, D. Pellacani, N. I. Park, F. J. Coutinho, H. Whetstone, H. J. Selvadurai, C. Che, B. Luu, A. Carles, M. Moksa, N. Rastegar, R. Head, S. Dolma, P. Prinos, M. D. Cusimano, S. Das, M. Bernstein, C. H. Arrowsmith, A. J. Mungall, R. A. Moore, Y. Ma, M. Gallo, M. Lupien, T. J. Pugh, M. D. Taylor, M. Hirst, C. J. Eaves, B. D. Simons, P. B. Dirks, Fate mapping of human glioblastoma reveals an invariant stem cell hierarchy. Nature 549, 227-232 (2017).

154. N. The Cancer Genome Atlas Research, Comprehensive genomic characterization defines human glioblastoma genes and core pathways. Nature 455, 1061-1068 (2008).

155. Cameron W. Brennan, Roel G. W. Verhaak, A. McKenna, B. Campos, H. Noushmehr, Sofie R. Salama, S. Zheng, D. Chakravarty, J. Z. Sanborn, Samuel H. Berman, R. Beroukhim, B. Bernard, C.-J. Wu, G. Genovese, I. Shmulevich, J. Barnholtz-Sloan, L. Zou, R. Vegesna, Sachet A. Shukla, G. Ciriello, W. K. Yung, W. Zhang, C. Sougnez, T. Mikkelsen, K. Aldape, Darell D. Bigner, Erwin G. Van Meir, M. Prados, A. Sloan, Keith L. Black, J. Eschbacher, G. Finocchiaro, W. Friedman, David W. Andrews, A. Guha, M. Iacocca, Brian P. O’Neill, G. Foltz, J. Myers, Daniel J. Weisenberger, R. Penny, R. Kucherlapati, Charles M. Perou, D. N. Hayes, R. Gibbs, M. Marra, Gordon B. Mills, E. Lander, P. Spellman, R. Wilson, C. Sander, J. Weinstein, M. Meyerson, S. Gabriel, Peter W. Laird, D. Haussler, G. Getz, L. Chin, C. Benz, J. Barnholtz-Sloan, W. Barrett, Q. Ostrom, Y. Wolinsky, Keith L. Black, B. Bose, Paul T. Boulos, M. Boulos, J. Brown, C. Czerinski, M. Eppley, M. Iacocca, T. Kempista, T. Kitko, Y. Koyfman, B. Rabeno, P. Rastogi, M. Sugarman, P. Swanson, K. Yalamanchii, Ilana P. Otey, Yingchun S. Liu, Y. Xiao, J. T. Auman, P.-C. Chen, A. Hadjipanayis, E. Lee, S. Lee, Peter J. Park, J. Seidman, L. Yang, R. Kucherlapati, S. Kalkanis, T. Mikkelsen, Laila M. Poisson, A. Raghunathan, L. Scarpace, B. Bernard, R. Bressler, A. Eakin, L. Iype, others, The Somatic Genomic Landscape of Glioblastoma. Cell 155, 462-477 (2013).

156. Y. Reid, D. Storts, T. Riss, L. Minor, in Assay Guidance Manual [Internet]. G. S. Sittampalam, N. P. Coussens, K. Brimacombe, A. Grossman, M. Arkin, D. Auld, C. Austin, J. Baell, B. Bejcek, T. D. Y. Chung, J. L. Dahlin, V. Devanaryan, T. L. Foley, M. Glicksman, M. D. Hall, J. V. Hass, J. Inglese, P. W. Iversen, S. D. Kahl, S. C. Kales, M. Lal-Nag, Z. Li, J. McGee, O. McManus, T. Riss, O. J. Trask, Jr.,

270

J. R. Weidner, M. Xia, X. Xu, Eds. (Eli Lilly & Company and the National Center for Advancing Translational Sciences, Bethesda (MD), 2013).

157. K. Cibulskis, M. S. Lawrence, S. L. Carter, A. Sivachenko, D. Jaffe, C. Sougnez, S. Gabriel, M. Meyerson, E. S. Lander, G. Getz, Sensitive detection of somatic point mutations in impure and heterogeneous cancer samples. Nat Biotech 31, 213-219 (2013).

158. A. H. Ramos, L. Lichtenstein, M. Gupta, M. S. Lawrence, T. J. Pugh, G. Saksena, M. Meyerson, G. Getz, Oncotator: Cancer Variant Annotation Tool. Human Mutation 36, E2423-E2429 (2015).

159. A. McKenna, M. Hanna, E. Banks, A. Sivachenko, K. Cibulskis, A. Kernytsky, K. Garimella, D. Altshuler, S. Gabriel, M. Daly, M. A. DePristo, The Genome Analysis Toolkit: A MapReduce framework for analyzing next-generation DNA sequencing data. Genome Research 20, 1297-1303 (2010).

160. A. Shlien, B. B. Campbell, R. de Borja, L. B. Alexandrov, D. Merico, D. Wedge, P. Van Loo, P. S. Tarpey, P. Coupland, S. Behjati, A. Pollett, T. Lipman, A. Heidari, S. Deshmukh, N. a. Avitzur, B. Meier, M. Gerstung, Y. Hong, D. M. Merino, M. Ramakrishna, M. Remke, R. Arnold, G. B. Panigrahi, N. P. Thakkar, K. P. Hodel, E. E. Henninger, A. Y. Goksenin, D. Bakry, G. S. Charames, H. Druker, J. Lerner-Ellis, M. Mistry, R. Dvir, R. Grant, R. Elhasid, R. Farah, G. P. Taylor, P. C. Nathan, S. Alexander, S. Ben-Shachar, S. C. Ling, S. Gallinger, S. Constantini, P. Dirks, A. Huang, S. W. Scherer, R. G. Grundy, C. Durno, M. Aronson, A. Gartner, M. S. Meyn, M. D. Taylor, Z. F. Pursell, C. E. Pearson, D. Malkin, P. A. Futreal, M. R. Stratton, E. Bouffet, C. Hawkins, P. J. Campbell, U. Tabori, C. for the Biallelic Mismatch Repair Deficiency, Combined hereditary and somatic mutations of replication error repair genes result in rapid onset of ultra-hypermutated cancers. Nat Genet 47, 257-262 (2015).

161. F. Favero, T. Joshi, A. M. Marquard, N. J. Birkbak, M. Krzystanek, Q. Li, Z. Szallasi, A. C. Eklund, Sequenza: allele-specific copy number and mutation profiles from tumor sequencing data. Annals of Oncology 26, 64-70 (2015).

162. I. Crespo, A. L. Vital, A. B. Nieto, O. Rebelo, H. Tão, M. C. Lopes, C. R. Oliveira, P. J. French, A. Orfao, M. D. Tabernero, Detailed Characterization of Alterations of Chromosomes 7, 9, and 10 in Glioblastomas as Assessed by Single-Nucleotide Polymorphism Arrays. The Journal of Molecular Diagnostics : JMD 13, 634-647 (2011).

163. C. W. Brennan, R. G. W. Verhaak, A. McKenna, B. Campos, H. Noushmehr, S. R. Salama, S. Zheng, D. Chakravarty, J. Z. Sanborn, S. H. Berman, R. Beroukhim, B. Bernard, C.-J. Wu, G. Genovese, I. Shmulevich, J. Barnholtz-Sloan, L. Zou, R. Vegesna, S. A. Shukla, G. Ciriello, W. K. Yung, W. Zhang, C. Sougnez, T. Mikkelsen, K. Aldape, D. D. Bigner, E. G. Van Meir, M. Prados, A. Sloan, K. L. Black, J. Eschbacher, G. Finocchiaro, W. Friedman, D. W. Andrews, A. Guha, M. Iacocca, B. P. O’Neill, G. Foltz, J. Myers, D. J. Weisenberger, R. Penny, R. Kucherlapati, C. M. Perou, D. N. Hayes, R. Gibbs, M. Marra, G. B. Mills, E. Lander, P. Spellman, R. Wilson, C. Sander, J. Weinstein, M. Meyerson, S.

271

Gabriel, P. W. Laird, D. Haussler, G. Getz, L. Chin, The Somatic Genomic Landscape of Glioblastoma. Cell 155, 462-477 (2013).

164. A. Olar, K. D. Aldape, Using the Molecular Classification of Glioblastoma to Inform Personalized Treatment. The Journal of pathology 232, 165-177 (2014).

165. L. B. Alexandrov, S. Nik-Zainal, D. C. Wedge, S. A. J. R. Aparicio, S. Behjati, A. V. Biankin, G. R. Bignell, N. Bolli, A. Borg, A.-L. Børresen-Dale, S. Boyault, B. Burkhardt, A. P. Butler, C. Caldas, H. R. Davies, C. Desmedt, R. Eils, J. E. Eyfjörd, J. A. Foekens, M. Greaves, F. Hosoda, B. Hutter, T. Ilicic, S. Imbeaud, M. Imielinsk, N. Jäger, D. T. W. Jones, D. Jones, S. Knappskog, M. Kool, S. R. Lakhani, C. López-Otín, S. Martin, N. C. Munshi, H. Nakamura, P. A. Northcott, M. Pajic, E. Papaemmanuil, A. Paradiso, J. V. Pearson, X. S. Puente, K. Raine, M. Ramakrishna, A. L. Richardson, J. Richter, P. Rosenstiel, M. Schlesner, T. N. Schumacher, P. N. Span, J. W. Teague, Y. Totoki, A. N. J. Tutt, R. Valdés-Mas, M. M. van Buuren, L. van ’t Veer, A. Vincent-Salomon, N. Waddell, L. R. Yates, I. Australian Pancreatic Cancer Genome, I. B. C. Consortium, I. M.-S. Consortium, I. PedBrain, J. Zucman-Rossi, P. A. Futreal, U. McDermott, P. Lichter, M. Meyerson, S. M. Grimmond, R. Siebert, E. Campo, T. Shibata, S. M. Pfister, P. J. Campbell, M. R. Stratton, Signatures of mutational processes in human cancer. Nature 500, 415-421 (2013).

166. C. Kandoth, M. D. McLellan, F. Vandin, K. Ye, B. Niu, C. Lu, M. Xie, Q. Zhang, J. F. McMichael, M. A. Wyczalkowski, M. D. M. Leiserson, C. A. Miller, J. S. Welch, M. J. Walter, M. C. Wendl, T. J. Ley, R. K. Wilson, B. J. Raphael, L. Ding, Mutational landscape and significance across 12 major cancer types. Nature 502, 333-339 (2013).

167. L. B. Alexandrov, S. Nik-Zainal, D. C. Wedge, S. A. J. R. Aparicio, S. Behjati, A. V. Biankin, G. R. Bignell, N. Bolli, A. Borg, A.-L. Børresen-Dale, S. Boyault, B. Burkhardt, A. P. Butler, C. Caldas, H. R. Davies, C. Desmedt, R. Eils, J. E. Eyfjörd, J. A. Foekens, M. Greaves, F. Hosoda, B. Hutter, T. Ilicic, S. Imbeaud, M. Imielinski, N. Jäger, D. T. W. Jones, D. Jones, S. Knappskog, M. Kool, S. R. Lakhani, C. López-Otín, S. Martin, N. C. Munshi, H. Nakamura, P. A. Northcott, M. Pajic, E. Papaemmanuil, A. Paradiso, J. V. Pearson, X. S. Puente, K. Raine, M. Ramakrishna, A. L. Richardson, J. Richter, P. Rosenstiel, M. Schlesner, T. N. Schumacher, P. N. Span, J. W. Teague, Y. Totoki, A. N. J. Tutt, R. Valdés-Mas, M. M. van Buuren, L. van ’t Veer, A. Vincent-Salomon, N. Waddell, L. R. Yates, I. Australian Pancreatic Cancer Genome, I. B. C. Consortium, I. M.-S. Consortium, I. PedBrain, J. Zucman-Rossi, P. Andrew Futreal, U. McDermott, P. Lichter, M. Meyerson, S. M. Grimmond, R. Siebert, E. Campo, T. Shibata, S. M. Pfister, P. J. Campbell, M. R. Stratton, Signatures of mutational processes in human cancer. Nature 500, 415 (2013).

168. S. M. Ahn, A. A. Ansari, J. Kim, D. Kim, S. M. Chun, J. Kim, T. W. Kim, I. Park, C. S. Yu, S. J. Jang, The somatic POLE P286R mutation defines a unique subclass of colorectal cancer featuring hypermutation, representing a potential genomic biomarker for immunotherapy. Oncotarget 7, 68638-68649 (2016).

272

169. B. E. Howitt, S. A. Shukla, L. M. Sholl, et al., Association of polymerase e– mutated and microsatellite-instable endometrial cancers with neoantigen load, number of tumor-infiltrating lymphocytes, and expression of pd-1 and pd-l1. JAMA Oncology 1, 1319-1323 (2015).

170. T. R. Hodges, M. Ott, J. Xiu, Z. Gatalica, J. Swensen, S. Zhou, J. T. Huse, J. de Groot, S. Li, W. W. Overwijk, D. Spetzler, A. B. Heimberger, Mutational burden, immune checkpoint expression, and mismatch repair in glioma: implications for immune checkpoint immunotherapy. Neuro-Oncology 19, 1047-1057 (2017).

171. E. Cerami, J. Gao, U. Dogrusoz, B. E. Gross, S. O. Sumer, B. A. Aksoy, A. Jacobsen, C. J. Byrne, M. L. Heuer, E. Larsson, Y. Antipin, B. Reva, A. P. Goldberg, C. Sander, N. Schultz, The cBio Cancer Genomics Portal: An Open Platform for Exploring Multidimensional Cancer Genomics Data. Cancer discovery 2, 401-404 (2012).

172. G. Wu, X. Feng, L. Stein, A human functional protein interaction network and its application to cancer data analysis. Genome biology 11, R53 (2010).

173. D. W. Parsons, S. Jones, X. Zhang, J. C.-H. Lin, R. J. Leary, P. Angenendt, P. Mankoo, H. Carter, I.-M. Siu, G. L. Gallia, A. Olivi, R. McLendon, B. A. Rasheed, S. Keir, T. Nikolskaya, Y. Nikolsky, D. A. Busam, H. Tekleab, L. A. Diaz, J. Hartigan, D. R. Smith, R. L. Strausberg, S. K. N. Marie, S. M. O. Shinjo, H. Yan, G. J. Riggins, D. D. Bigner, R. Karchin, N. Papadopoulos, G. Parmigiani, B. Vogelstein, V. E. Velculescu, K. W. Kinzler, An Integrated Genomic Analysis of Human Glioblastoma Multiforme. Science 321, 1807-1812 (2008).

174. B. Vogelstein, N. Papadopoulos, V. E. Velculescu, S. Zhou, L. A. Diaz, K. W. Kinzler, Cancer Genome Landscapes. Science 339, 1546-1558 (2013).

175. M. E. J. Newman, M. Girvan, Finding and evaluating community structure in networks. Physical Review E 69, 026113 (2004).

176. E. Cerami, E. Demir, N. Schultz, B. S. Taylor, C. Sander, Automated Network Analysis Identifies Core Pathways in Glioblastoma. PLOS ONE 5, e8918 (2010).

177. L. J. Jensen, M. Kuhn, M. Stark, S. Chaffron, C. Creevey, J. Muller, T. Doerks, P. Julien, A. Roth, M. Simonovic, P. Bork, C. von Mering, STRING 8--a global view on proteins and their functional interactions in 630 organisms. Nucleic Acids Res 37, D412-416 (2009).

178. E. Z. Erson-Omay, A. O. Çağlayan, N. Schultz, N. Weinhold, S. B. Omay, K. Özduman, Y. Köksal, J. Li, A. Serin Harmancı, V. Clark, G. Carrión-Grant, J. Baranoski, C. Çağlar, T. Barak, S. Coşkun, B. Baran, D. Köse, J. Sun, M. Bakırcıoğlu, J. Moliterno Günel, M. N. Pamir, K. Mishra-Gorur, K. Bilguvar, K. Yasuno, A. Vortmeyer, A. J. Huttner, C. Sander, M. Günel, Somatic POLE mutations cause an ultramutated giant cell high-grade glioma subtype with better prognosis. Neuro-Oncology 17, 1356-1364 (2015).

273

179. S. Briggs, I. Tomlinson, Germline and somatic polymerase ϵ and δ mutations define a new class of hypermutated colorectal and endometrial cancers. The Journal of Pathology 230, 148-153 (2013).

180. Y. Zou, F. Y. Liu, H. Liu, F. Wang, W. Li, M. Z. Huang, Y. Huang, X. Q. Yuan, X. Y. Xu, O. P. Huang, M. He, Frequent POLE1 p.S297F mutation in Chinese patients with ovarian endometrioid carcinoma. Mutat Res 761, 49-52 (2014).

181. S. K. Merid, D. Goranskaya, A. Alexeyenko, Distinguishing between driver and passenger mutations in individual cancer genomes by network enrichment analysis. BMC Bioinformatics 15, 308 (2014).

182. Y. Gu, H. Wang, Y. Qin, Y. Zhang, W. Zhao, L. Qi, Y. Zhang, C. Wang, Z. Guo, Network analysis of genomic alteration profiles reveals co-altered functional modules and driver genes for glioblastoma. Molecular BioSystems 9, 467-477 (2013).

183. N. R. Parker, A. L. Hudson, P. Khong, J. F. Parkinson, T. Dwight, R. J. Ikin, Y. Zhu, Z. J. Cheng, F. Vafaee, J. Chen, H. R. Wheeler, V. M. Howell, Intratumoral heterogeneity identified at the epigenetic, genetic and transcriptional level in glioblastoma. Sci Rep 6, 22477 (2016).

184. A. Sottoriva, I. Spiteri, S. G. Piccirillo, A. Touloumis, V. P. Collins, J. C. Marioni, C. Curtis, C. Watts, S. Tavare, Intratumor heterogeneity in human glioblastoma reflects cancer evolutionary dynamics. Proc Natl Acad Sci U S A 110, 4009-4014 (2013).

185. A. P. Patel, I. Tirosh, J. J. Trombetta, A. K. Shalek, S. M. Gillespie, H. Wakimoto, D. P. Cahill, B. V. Nahed, W. T. Curry, R. L. Martuza, D. N. Louis, O. Rozenblatt- Rosen, M. L. Suvà, A. Regev, B. E. Bernstein, Single-cell RNA-seq highlights intratumoral heterogeneity in primary glioblastoma. Science 344, 1396-1401 (2014).

186. A. L. Hopkins, C. R. Groom, The druggable genome. Nat Rev Drug Discov 1, 727-730 (2002).

187. D. D. Von Hoff, J. J. Stephenson, Jr., P. Rosen, D. M. Loesch, M. J. Borad, S. Anthony, G. Jameson, S. Brown, N. Cantafio, D. A. Richards, T. R. Fitch, E. Wasserman, C. Fernandez, S. Green, W. Sutherland, M. Bittner, A. Alarcon, D. Mallery, R. Penny, Pilot study using molecular profiling of patients' tumors to find potential targets and select treatments for their refractory cancers. J Clin Oncol 28, 4877-4883 (2010).

188. A. P. Davis, T. C. Wiegers, B. L. King, J. Wiegers, C. J. Grondin, D. Sciaky, R. J. Johnson, C. J. Mattingly, Generating Gene Ontology-Disease Inferences to Explore Mechanisms of Human Disease at the Comparative Toxicogenomics Database. PLoS One 11, e0155530 (2016).

274

189. C. A. Lipinski, F. Lombardo, B. W. Dominy, P. J. Feeney, Experimental and computational approaches to estimate solubility and permeability in drug discovery and development settings. Adv Drug Deliv Rev 46, 3-26 (2001).

190. X. Liu, H. Qu, W. Cao, Y. Wang, Z. Ma, F. Li, H. Wang, Efficacy of combined therapy of goserelin and letrozole on very young women with advanced breast cancer as first-line endocrine therapy. Endocr J 60, 819-828 (2013).

191. G. Klautke, M. Schutze, I. Bombor, R. Benecke, J. Piek, R. Fietkau, Concurrent chemoradiotherapy and adjuvant chemotherapy with Topotecan for patients with glioblastoma multiforme. J Neurooncol 77, 199-205 (2006).

192. T. Lesimple, L. Riffaud, D. Frappaz, M. Ben Hassel, D. Gedouin, J. O. Bay, C. Linassier, A. Hamlat, G. Piot, M. Fabbro, S. Saikali, B. Carsin, Y. Guegan, Topotecan in combination with radiotherapy in unresectable glioblastoma: a phase 2 study. J Neurooncol 93, 253-260 (2009).

193. J. M. Pipas, L. P. Meyer, C. H. Rhodes, L. D. Cromwell, C. E. McDonnell, L. S. Kingman, J. R. Rigas, C. E. Fadul, A Phase II trial of paclitaxel and topotecan with filgrastim in patients with recurrent or refractory glioblastoma multiforme or anaplastic astrocytoma. J Neurooncol 71, 301-305 (2005).

194. G. G. Grabenbauer, K. D. Gerber, O. Ganslandt, A. Richter, G. Klautke, J. Birkmann, M. Meyer, Effects of concurrent topotecan and radiation on 6-month progression-free survival in the primary treatment of glioblastoma multiforme. Int J Radiat Oncol Biol Phys 75, 164-169 (2009).

195. D. Macdonald, G. Cairncross, D. Stewart, P. Forsyth, C. Sawka, N. Wainman, E. Eisenhauer, Phase II study of topotecan in patients with recurrent malignant glioma. National Clinical Institute of Canada Clinical Trials Group. Ann Oncol 7, 205-207 (1996).

196. B. Fisher, M. Won, D. Macdonald, D. W. Johnson, W. Roa, Phase II study of topotecan plus cranial radiation for glioblastoma multiforme: results of Radiation Therapy Oncology Group 9513. Int J Radiat Oncol Biol Phys 53, 980-986 (2002).

197. S. Ananda, A. K. Nowak, L. Cher, A. Dowling, C. Brown, J. Simes, M. A. Rosenthal, Phase 2 trial of temozolomide and pegylated liposomal doxorubicin in the treatment of patients with glioblastoma multiforme following concurrent radiotherapy and chemotherapy. J Clin Neurosci 18, 1444-1448 (2011).

198. C. P. Beier, C. Schmid, T. Gorlia, C. Kleinletzenberger, D. Beier, O. Grauer, A. Steinbrecher, B. Hirschmann, A. Brawanski, C. Dietmaier, T. Jauch-Worley, O. Kolbl, T. Pietsch, M. Proescholdt, P. Rummele, A. Muigg, G. Stockhammer, M. Hegi, U. Bogdahn, P. Hau, RNOP-09: pegylated liposomal doxorubicine and prolonged temozolomide in addition to radiotherapy in newly diagnosed glioblastoma--a phase II study. BMC Cancer 9, 308 (2009).

275

199. S. L. Chua, M. A. Rosenthal, S. S. Wong, D. M. Ashley, A.-m. Woods, A. Dowling, L. M. Cher, Phase 2 study of temozolomide and Caelyx in patients with recurrent glioblastoma multiforme. Neuro-Oncology 6, 38-43 (2004).

200. C. E. Fadul, L. S. Kingman, L. P. Meyer, B. F. Cole, C. J. Eskey, C. H. Rhodes, D. W. Roberts, H. B. Newton, J. M. Pipas, A phase II study of thalidomide and irinotecan for treatment of glioblastoma multiforme. Journal of neuro-oncology 90, 229-235 (2008).

201. D. Schiff, B. Purow, Bevacizumab in combination with irinotecan for patients with recurrent glioblastoma multiforme. Nature clinical practice. Oncology 5, 186-187 (2008).

202. V. K. Puduvalli, P. Giglio, M. D. Groves, K. R. Hess, M. R. Gilbert, S. Mahankali, E. F. Jackson, V. A. Levin, C. A. Conrad, S. H. Hsu, H. Colman, J. F. de Groot, M. G. Ritterhouse, S. E. Ictech, W. K. Yung, Phase II trial of irinotecan and thalidomide in adults with recurrent glioblastoma multiforme. Neuro Oncol 10, 216-222 (2008).

203. S. Chowdhary, E. T. Wong, Bevacizumab combined with irinotecan for recurrent glioblastoma multiforme--improvement over available therapy? Nature clinical practice. Neurology 4, 242-243 (2008).

204. J. A. Quinn, S. X. Jiang, D. A. Reardon, A. Desjardins, J. J. Vredenburgh, A. H. Friedman, J. H. Sampson, R. E. McLendon, J. E. Herndon, 2nd, H. S. Friedman, Phase II trial of temozolomide (TMZ) plus irinotecan (CPT-11) in adults with newly diagnosed glioblastoma multiforme before radiotherapy. Journal of neuro- oncology 95, 393-400 (2009).

205. M. D. Prados, K. Lamborn, W. K. Yung, K. Jaeckle, H. I. Robins, M. Mehta, H. A. Fine, P. Y. Wen, T. Cloughesy, S. Chang, M. K. Nicholas, D. Schiff, H. Greenberg, L. Junck, K. Fink, K. Hess, J. Kuhn, A phase 2 trial of irinotecan (CPT-11) in patients with recurrent malignant glioma: a North American Brain Tumor Consortium study. Neuro Oncol 8, 189-193 (2006).

206. A. A. Brandes, A. Tosoni, U. Basso, M. Reni, F. Valduga, S. Monfardini, P. Amista, L. Nicolardi, G. Sotti, M. Ermani, Second-line chemotherapy with irinotecan plus carmustine in glioblastoma recurrent or progressive after first-line temozolomide chemotherapy: a phase II study of the Gruppo Italiano Cooperativo di Neuro-Oncologia (GICNO). J Clin Oncol 22, 4779-4786 (2004).

207. D. A. Reardon, J. A. Quinn, J. N. Rich, S. Gururangan, J. Vredenburgh, J. H. Sampson, J. M. Provenzale, A. Walker, M. Badruddoja, S. Tourt-Uhlig, J. E. Herndon, 2nd, J. M. Dowell, M. L. Affronti, S. Jackson, D. Allen, K. Ziegler, S. Silverman, C. Bohlin, A. H. Friedman, D. D. Bigner, H. S. Friedman, Phase 2 trial of BCNU plus irinotecan in adults with malignant glioma. Neuro Oncol 6, 134- 144 (2004).

208. M. L. Gruber, W. P. Buster, Temozolomide in combination with irinotecan for treatment of recurrent malignant glioma. Am J Clin Oncol 27, 33-38 (2004).

276

209. M. C. Chamberlain, Salvage chemotherapy with CPT-11 for recurrent glioblastoma multiforme. J Neurooncol 56, 183-188 (2002).

210. M. Napolitano, F. Keime-Guibert, A. Monjour, C. Lafitte, A. Ameri, P. Cornu, P. Broet, J. Y. Delattre, Treatment of supratentorial glioblastoma multiforme with radiotherapy and a combination of BCNU and tamoxifen: a phase II study. J Neurooncol 45, 229-235 (1999).

211. H. I. Robins, M. Won, W. F. Seiferheld, C. J. Schultz, A. K. Choucair, D. G. Brachman, W. F. Demas, M. P. Mehta, Phase 2 trial of radiation plus high-dose tamoxifen for glioblastoma multiforme: RTOG protocol BR-0021. Neuro Oncol 8, 47-52 (2006).

212. D. I. C. A, G. Carrabba, G. Lanfranchi, C. Menghetti, P. Rampini, M. Caroli, Continuous tamoxifen and dose-dense temozolomide in recurrent glioblastoma. Anticancer research 33, 3383-3389 (2013).

213. L. Chen, Y. Shi, J. Yuan, Q. Wu, Y. Han, R. Qin, B. Jia, B. Wei, L. Wei, G. Dai, S. Jiao, Evaluation of docetaxel- and oxaliplatin-based adjuvant chemotherapy in postgastrectomy gastric cancer patients reveals obvious survival benefits in docetaxel-treated mixed signet ring cell carcinoma patients. Med Oncol 31, 159 (2014).

214. V. Kroeber, S. Nagel, W. Schuette, T. Blankenburg, A Prospective Phase I/II Study: Combination Chemotherapy with Docetaxel and Pemetrexed as Second- Line Treatment in Patients with Stage IIIB/IV Non-Small Cell Lung Cancer. Case Rep Oncol 7, 435-443 (2014).

215. L. Derosa, L. Galli, P. Orlandi, A. Fioravanti, T. Di Desidero, A. Fontana, A. Antonuzzo, E. Biasco, A. Farnesi, R. Marconcini, G. Francia, R. Danesi, A. Falcone, G. Bocci, Docetaxel plus oral metronomic cyclophosphamide: A phase II study with pharmacodynamic and pharmacogenetic analyses in castration- resistant prostate cancer patients. Cancer, (2014).

216. J. Salvador, L. Manso, J. de la Haba, A. Jaen, E. Ciruelos, M. C. de Villena, M. Gil, A. Murias, A. Galan, C. Jara, J. Bayo, J. M. Baena, J. Casal, J. R. Mel, I. Blancas, P. Sanchez Rvira, Final results of a phase II study of paclitaxel, bevacizumab, and gemcitabine as first-line therapy for patients with HER2- negative metastatic breast cancer. Clin Transl Oncol, (2014).

217. J. M. McKiernan, D. D. Holder, R. A. Ghandour, L. J. Barlow, J. J. Ahn, M. Kates, G. M. Badalato, A. Roychoudhury, G. J. Decastro, M. C. Benson, Phase II Trial of Intravesical Nanoparticle Albumin-Bound (nab-) Paclitaxel for the Treatment of Non-Muscle Invasive Urothelial Carcinoma of the Bladder after BCG Treatment Failure. J Urol, (2014).

218. S. S. Varghese, T. S. Ram, S. P. Pavamani, E. M. Thomas, V. Jeyaseelan, P. N. Viswanathan, Concurrent chemo-irradiation with weekly cisplatin and paclitaxel in the treatment of locally advanced squamous cell carcinoma of cervix: A phase II study. J Cancer Res Ther 10, 330-336 (2014).

277

219. H. K. Ahn, M. Jung, S. J. Sym, D. B. Shin, S. M. Kang, S. Y. Kyung, J. W. Park, S. H. Jeong, E. K. Cho, A phase II trial of Cremorphor EL-free paclitaxel (Genexol-PM) and gemcitabine in patients with advanced non-small cell lung cancer. Cancer Chemother Pharmacol 74, 277-282 (2014).

220. D. Crivellari, O. Pagani, A. Veronesi, D. Lombardi, F. Nole, B. Thurlimann, D. Hess, M. Borner, J. Bauer, G. Martinelli, R. Graffeo, C. Sessa, A. Goldhirsch, High incidence of central nervous system involvement in patients with metastatic or locally advanced breast cancer treated with epirubicin and docetaxel. Ann Oncol 12, 353-356 (2001).

221. N. Marschner, G. A. Nagel, J. H. Beyer, M. Adler, A. Ammon, High-dose epirubicin in combination with cyclophosphamide (HD-EC) in advanced breast cancer: final results of a dose finding study and phase II trial. Onkologie 13, 272- 278 (1990).

222. X. Quantin, A. Riviere, J. P. Daures, P. Oliver, M. Comte-Bardonnet, F. Khial, I. Marcillac, J. L. Pujol, Phase I-II study of high dose epirubicin plus cisplatin in unresectable non-small-cell lung cancer: searching for the maximal tolerated dose. Am J Clin Oncol 23, 192-196 (2000).

223. M. S. Braun, S. D. Richman, P. Quirke, C. Daly, J. W. Adlard, F. Elliott, J. H. Barrett, P. Selby, A. M. Meade, R. J. Stephens, Predictive biomarkers of chemotherapy efficacy in colorectal cancer: results from the UK MRC FOCUS trial. Journal of Clinical Oncology 26, 2690-2698 (2008).

224. I. Kostopoulos, V. Karavasilis, M. Karina, M. Bobos, N. Xiros, G. Pentheroudakis, G. Kafiri, P. Papakostas, E. Vrettou, G. Fountzilas, Topoisomerase I but not thymidylate synthase is associated with improved outcome in patients with resected colorectal cancer treated with irinotecan containing adjuvant chemotherapy. BMC cancer 9, 339 (2009).

225. M. Morise, S. Niho, S. Umemura, S. Matsumoto, K. Yoh, K. Goto, H. Ohmatsu, Y. Ohe, Low-dose Irinotecan as a Second-line Chemotherapy for Recurrent Small Cell Lung Cancer. Jpn J Clin Oncol, (2014).

226. S. F. Powell, A. Beitinjaneh, M. Tessema, R. L. Bliss, R. A. Kratzke, J. Leach, A. Z. Dudek, Phase II study of topotecan and bevacizumab in advanced, refractory non--small-cell lung cancer. Clin Lung Cancer 14, 495-501 (2013).

227. W. T. Couldwell, D. R. Hinton, A. A. Surnock, C. M. DeGiorgio, L. P. Weiner, M. L. Apuzzo, L. Masri, R. E. Law, M. H. Weiss, Treatment of recurrent malignant gliomas with chronic oral high-dose tamoxifen. Clin Cancer Res 2, 619-622 (1996).

228. P. Tang, G. Roldan, P. M. Brasher, D. Fulton, W. Roa, A. Murtha, J. G. Cairncross, P. A. Forsyth, A phase II study of carboplatin and chronic high-dose tamoxifen in patients with recurrent malignant glioma. J Neurooncol 78, 311-316 (2006).

278

229. A. M. Spence, R. A. Peterson, J. D. Scharnhorst, D. L. Silbergeld, R. C. Rostomily, Phase II study of concurrent continuous Temozolomide (TMZ) and Tamoxifen (TMX) for recurrent malignant astrocytic gliomas. J Neurooncol 70, 91-95 (2004).

230. T. Muanza, G. Shenouda, L. Souhami, R. Leblanc, G. Mohr, R. Corns, A. Langleben, High dose tamoxifen and radiotherapy in patients with glioblastoma multiforme: a phase IB study. Can J Neurol Sci 27, 302-306 (2000).

231. A. A. Brandes, M. Ermani, S. Turazzi, E. Scelzi, F. Berti, P. Amista, A. Rotilio, C. Licata, M. V. Fiorentino, Procarbazine and high-dose tamoxifen as a second- line regimen in recurrent high-grade gliomas: a phase II study. J Clin Oncol 17, 645-650 (1999).

232. M. Sawaki, M. Wada, Y. Sato, Y. Mizuno, H. Kobayashi, K. Yokoi, M. Yoshihara, K. Kamei, M. Ohno, T. Imai, High-dose toremifene as first-line treatment of metastatic breast cancer resistant to adjuvant aromatase inhibitor: A multicenter phase II study. Oncol Lett 3, 61-65 (2012).

233. J. F. Robertson, J. P. Lindemann, A. Llombart-Cussac, J. Rolski, D. Feltl, J. Dewar, L. Emerson, A. Dean, M. J. Ellis, Fulvestrant 500 mg versus anastrozole 1 mg for the first-line treatment of advanced breast cancer: follow-up analysis from the randomized 'FIRST' study. Breast Cancer Res Treat 136, 503-511 (2012).

234. V. Guarneri, D. G. Generali, A. Frassoldati, F. Artioli, C. Boni, L. Cavanna, E. Tagliafico, A. Maiorana, A. Bottini, K. Cagossi, G. Bisagni, F. Piacentini, G. Ficarra, S. Bettelli, E. Roncaglia, S. Nuzzo, R. Swaby, C. Ellis, C. Holford, P. Conte, Double-blind, placebo-controlled, multicenter, randomized, phase IIb neoadjuvant study of letrozole-lapatinib in postmenopausal hormone receptor- positive, human epidermal growth factor receptor 2-negative, operable breast cancer. J Clin Oncol 32, 1050-1057 (2014).

235. P. A. Fasching, S. M. Jud, M. Hauschild, S. Kummel, M. Schutte, M. Warm, V. Hanf, D. Grab, J. Krocker, E. Stickeler, R. Kreienberg, T. Muller, T. Kuhn, C. Wolf, S. Kahlert, S. Paepke, M. Berghorn, M. Muth, M. Baier, B. Wackwitz, R. Schulz-Wendtland, M. W. Beckmann, M. P. Lux, FemZone trial: a randomized phase II trial comparing neoadjuvant letrozole and zoledronic acid with letrozole in primary breast cancer patients. BMC Cancer 14, 66 (2014).

236. C. Villanueva, G. Romieu, J. Salvat, L. Chaigneau, Y. Merrouche, T. N'Guyen, A. T. Vuillemin, M. Demarchi, E. Dobi, X. Pivot, Phase II study assessing lapatinib added to letrozole in patients with progressive disease under aromatase inhibitor in metastatic breast cancer-Study BES 06. Target Oncol 8, 137-143 (2013).

237. S. Massarweh, Y. L. Tham, J. Huang, K. Sexton, H. Weiss, A. Tsimelzon, A. Beyer, M. Rimawi, W. Y. Cai, S. Hilsenbeck, S. Fuqua, R. Elledge, A phase II neoadjuvant trial of anastrozole, fulvestrant, and gefitinib in patients with newly

279

diagnosed estrogen receptor positive breast cancer. Breast Cancer Res Treat 129, 819-827 (2011).

238. R. Nishimura, K. Anan, Y. Yamamoto, K. Higaki, M. Tanaka, K. Shibuta, Y. Sagara, S. Ohno, S. Tsuyuki, T. Mase, S. Teramukai, Efficacy of goserelin plus anastrozole in premenopausal women with advanced or recurrent breast cancer refractory to an LH-RH analogue with tamoxifen: results of the JMTO BC08-01 phase II trial. Oncol Rep 29, 1707-1713 (2013).

239. M. Ishitobi, O. Suzuki, Y. Komoike, S. Ohsumi, S. Nakahara, T. Yagi, T. Yoshinami, Y. Tomita, H. Inaji, Phase II study of neoadjuvant anastrozole and concurrent radiotherapy for postmenopausal breast cancer patients. Breast Cancer, (2012).

240. O. Pagani, M. M. Regan, B. A. Walley, G. F. Fleming, M. Colleoni, I. Lang, H. L. Gomez, C. Tondini, H. J. Burstein, E. A. Perez, E. Ciruelos, V. Stearns, H. R. Bonnefoi, S. Martino, C. E. Geyer, Jr., G. Pinotti, F. Puglisi, D. Crivellari, T. Ruhstaller, E. P. Winer, M. Rabaglio-Poretti, R. Maibach, B. Ruepp, A. Giobbie- Hurder, K. N. Price, J. Bernhard, W. Luo, K. Ribi, G. Viale, A. S. Coates, R. D. Gelber, A. Goldhirsch, P. A. Francis, Adjuvant exemestane with ovarian suppression in premenopausal breast cancer. N Engl J Med 371, 107-118 (2014).

241. D. A. Yardley, S. Noguchi, K. I. Pritchard, H. A. Burris, 3rd, J. Baselga, M. Gnant, G. N. Hortobagyi, M. Campone, B. Pistilli, M. Piccart, B. Melichar, K. Petrakova, F. P. Arena, F. Erdkamp, W. A. Harb, W. Feng, A. Cahana, T. Taran, D. Lebwohl, H. S. Rugo, Everolimus plus exemestane in postmenopausal patients with HR(+) breast cancer: BOLERO-2 final progression-free survival analysis. Adv Ther 30, 870-884 (2013).

242. S. Jones, C. Vogel, A. Arkhipov, L. Fehrenbacher, P. Eisenberg, B. Cooper, S. Honig, A. Polli, F. Whaley, E. di Salle, J. Tiffany, A. Consonni, L. Miller, Multicenter, phase II trial of exemestane as third-line hormonal therapy of postmenopausal women with metastatic breast cancer. Aromasin Study Group. J Clin Oncol 17, 3418-3425 (1999).

243. J. Bines, R. Dienstmann, R. M. Obadia, L. G. Branco, D. C. Quintella, T. M. Castro, P. G. Camacho, F. A. Soares, M. E. Costa, Activity of megestrol acetate in postmenopausal women with advanced breast cancer after nonsteroidal aromatase inhibitor failure: a phase II trial. Ann Oncol 25, 831-836 (2014).

244. L. Wood, M. Palmer, J. Hewitt, R. Urtasun, E. Bruera, E. Rapp, J. F. Thaell, Results of a phase III, double-blind, placebo-controlled trial of megestrol acetate modulation of P-glycoprotein-mediated drug resistance in the first-line management of small-cell lung carcinoma. Br J Cancer 77, 627-631 (1998).

245. J. Braeckman, D. Michielsen, Efficacy and tolerability of 1- and 3-month leuprorelin acetate depot formulations (Eligard((R))/Depo-Eligard((R))) for advanced prostate cancer in daily practice: a Belgian prospective non- interventional study. Arch Med Sci 10, 477-483 (2014).

280

246. H. Yang, X. Zong, Y. Yu, G. Shao, L. Zhang, C. Qian, Y. Bian, X. Xu, W. Sun, X. Meng, X. Ding, D. Chen, D. Zou, S. Xie, Y. Zheng, J. Zhang, X. He, C. Sun, X. Yu, J. Ni, Combined effects of goserelin and tamoxifen on estradiol level, breast density, and endometrial thickness in premenopausal and perimenopausal women with early-stage hormone receptor-positive breast cancer: a randomised controlled clinical trial. Br J Cancer 109, 582-588 (2013).

247. A. Uslu, B. Zengel, G. Akpinar, H. Postaci, H. Yetis, B. Corumlu, E. Kebapci, A. Aykas, The outcome effect of double-hormonal therapy in premenopausal breast cancer patients with high nodal-status: Result of a prospective randomized trial. Indian J Cancer 51, 582-586 (2014).

248. M. Nozawa, T. Inagaki, K. Nagao, T. Nishioka, T. Komura, A. Esa, M. Kitagawa, M. Imanishi, Y. Uekado, T. Ogawa, H. Kajikawa, S. Uejima, H. Matsuyama, I. Hara, H. Uemura, Phase II trial of zoledronic acid combined with androgen- deprivation therapy for treatment-naive prostate cancer with bone metastasis. Int J Clin Oncol, (2013).

249. Z. Blumenfeld, Gonadotropin-releasing hormone agonist for preservation of ovarian function during (neo)adjuvant chemotherapy for breast cancer. J Clin Oncol 30, 3310; author reply 3312-3313 (2012).

250. G. Ploussard, P. Mongiat-Artus, Triptorelin in the management of prostate cancer. Future Oncol 9, 93-102 (2013).

251. M. Hellstrom, M. Haggman, K. Pedersen, C. Busch, S. Brandstedt, A 3-year follow-up of patients with localized prostate cancer operated on with or without pre-treatment with the GnRH-agonist triptorelin. Br J Urol 78, 432-436 (1996).

252. E. A. Lundstrom, R. K. Rencken, J. H. van Wyk, L. J. Coetzee, J. C. Bahlmann, S. Reif, E. A. Strasheim, M. C. Bigalke, A. R. Pontin, L. Goedhals, D. G. Steyn, C. F. Heyns, L. A. Aldera, T. M. Mackenzie, D. Purcea, P. Y. Grosgurin, H. C. Porchet, Triptorelin 6-month formulation in the management of patients with locally advanced and metastatic prostate cancer: an open-label, non-comparative, multicentre, phase III study. Clin Drug Investig 29, 757-765 (2009).

253. M. B. Garnick, N. Mottet, New treatment paradigm for prostate cancer: abarelix initiation therapy for immediate testosterone suppression followed by a luteinizing hormone-releasing hormone agonist. BJU Int 110, 499-504 (2012).

254. R. S. Kirby, J. M. Fitzpatrick, N. Clarke, Abarelix and other gonadotrophin- releasing hormone antagonists in prostate cancer. BJU Int 104, 1580-1584 (2009).

255. T. M. Beer, C. Ryan, G. Bhat, M. Garnick, Dose-escalated abarelix in androgen- independent prostate cancer: a phase I study. Anticancer Drugs 17, 1075-1079 (2006).

256. M. Koch, C. Steidle, S. Brosman, A. Centeno, F. Gaylis, M. Campion, M. B. Garnick, An open-label study of abarelix in men with symptomatic prostate cancer at risk of treatment with LHRH agonists. Urology 62, 877-882 (2003).

281

257. S. Ozono, T. Ueda, S. Hoshi, A. Yamaguchi, H. Maeda, Y. Fukuyama, K. Takeda, Y. Ohashi, T. Tsukamoto, S. Naito, H. Akaza, The efficacy and safety of degarelix, a GnRH antagonist: a 12-month, multicentre, randomized, maintenance dose-finding phase II study in Japanese patients with prostate cancer. Jpn J Clin Oncol 42, 477-484 (2012).

258. O. Warburg, On the origin of cancer cells. Science 123, 309-314 (1956).

259. H. Shen, S. Decollogne, P. J. Dilda, E. Hau, S. A. Chung, P. P. Luk, P. J. Hogg, K. L. McDonald, Dual-targeting of aberrant glucose metabolism in glioblastoma. J Exp Clin Cancer Res 34, 14 (2015).

260. A. Ramanathan, S. L. Schreiber, Direct control of mitochondrial function by mTOR. Proceedings of the National Academy of Sciences 106, 22229-22232 (2009).

261. P. J. Dilda, S. Decollogne, L. Weerakoon, M. D. Norris, M. Haber, J. D. Allen, P. J. Hogg, Optimization of the antitumor efficacy of a synthetic mitochondrial toxin by increasing the residence time in the cytosol. J Med Chem 52, 6209-6216 (2009).

262. A. S. Don, P. J. Hogg, Mitochondria as cancer drug targets. Trends Mol Med 10, 372-378 (2004).

263. M. G. V. Heiden, L. C. Cantley, C. B. Thompson, Understanding the warburg effect: The metabolic requirements of cell proliferation. Science 324, 1029-1033 (2009).

264. A. J. Levine, A. M. Puzio-Kuter, The Control of the Metabolic Switch in Cancers by Oncogenes and Tumor Suppressor Genes. Science 330, 1340-1344 (2010).

265. R. J. DeBerardinis, J. J. Lum, G. Hatzivassiliou, C. B. Thompson, The Biology of Cancer: Metabolic Reprogramming Fuels Cell Growth and Proliferation. Cell Metabolism 7, 11-20 (2008).

266. T. Cheng, J. Sudderth, C. Yang, A. R. Mullen, E. S. Jin, J. M. Matés, R. J. DeBerardinis, Pyruvate carboxylase is required for glutamine-independent growth of tumor cells. Proceedings of the National Academy of Sciences 108, 8674-8679 (2011).

267. Berg JM, Tymoczko JL, S. L, Entry to the Citric Acid Cycle and Metabolism Through It Are Controlled. Biochemistry (W H Freeman, New York, ed. 5th, 2002).

268. R. A. Cairns, I. S. Harris, T. W. Mak, Regulation of cancer cell metabolism. Nature Reviews Cancer 11, 85-95 (2011).

269. P. Liu, H. Cheng, T. M. Roberts, J. J. Zhao, Targeting the phosphoinositide 3- kinase (PI3K) pathway in cancer. Nature reviews. Drug discovery 8, 627-644 (2009).

282

270. D. Akhavan, T. F. Cloughesy, P. S. Mischel, mTOR signaling in glioblastoma: lessons learned from bench to bedside. Neuro-Oncology 12, 882-889 (2010).

271. V. Stambolic, D. MacPherson, D. Sas, Y. Lin, B. Snow, Y. Jang, S. Benchimol, T. W. Mak, Regulation of PTEN Transcription by p53. Molecular Cell 8, 317-325 (2001).

272. G. L. Semenza, HIF-1: upstream and downstream of cancer metabolism. Curr Opin Genet Dev 20, 51-56 (2010).

273. C. W. Lu, S. C. Lin, K. F. Chen, Y. Y. Lai, S. J. Tsai, Induction of pyruvate dehydrogenase kinase-3 by hypoxia-inducible factor-1 promotes metabolic switch and drug resistance. J Biol Chem 283, 28106-28114 (2008).

274. J. W. Kim, I. Tchernyshyov, G. L. Semenza, C. V. Dang, HIF-1-mediated expression of pyruvate dehydrogenase kinase: a metabolic switch required for cellular adaptation to hypoxia. Cell Metab 3, 177-185 (2006).

275. I. Papandreou, R. A. Cairns, L. Fontana, A. L. Lim, N. C. Denko, HIF-1 mediates adaptation to hypoxia by actively downregulating mitochondrial oxygen consumption. Cell Metab 3, 187-197 (2006).

276. M. Laplante, D. M. Sabatini, MTOR signaling in growth control and disease. Cell 149, 274-293 (2012).

277. M. J. Groenewoud, F. J. T. Zwartkruis, Rheb and mammalian target of rapamycin in mitochondrial homoeostasis. Open Biology 3, 130185 (2013).

278. M. G. Vander Heiden, L. C. Cantley, C. B. Thompson, Understanding the Warburg effect: the metabolic requirements of cell proliferation. Science 324, 1029-1033 (2009).

279. Z. Duzgun, Z. Eroglu, C. Biray Avci, Role of mTOR in glioblastoma. Gene 575, 187-190 (2016).

280. Q.-W. Fan, W. A. Weiss, Inhibition of PI3K-Akt-mTOR Signaling in Glioblastoma by mTORC1/2 Inhibitors. Methods in Molecular Biology (Clifton, N.j.) 821, 349-359 (2012).

281. J. Huang, B. D. Manning, A complex interplay between Akt, TSC2 and the two mTOR complexes. Biochem Soc Trans 37, 217-222 (2009).

282. K. Inoki, Y. Li, T. Xu, K. L. Guan, Rheb GTPase is a direct target of TSC2 GAP activity and regulates mTOR signaling. Genes Dev 17, 1829-1834 (2003).

283. H. Stocker, T. Radimerski, B. Schindelholz, F. Wittwer, P. Belawat, P. Daram, S. Breuer, G. Thomas, E. Hafen, Rheb is an essential regulator of S6K in controlling cell growth in Drosophila. Nat Cell Biol 5, 559-565 (2003).

284. A. R. Tee, B. D. Manning, P. P. Roux, L. C. Cantley, J. Blenis, Tuberous sclerosis complex gene products, Tuberin and Hamartin, control mTOR signaling by acting

283

as a GTPase-activating protein complex toward Rheb. Curr Biol 13, 1259-1268 (2003).

285. J. T. Cunningham, J. T. Rodgers, D. H. Arlow, F. Vazquez, V. K. Mootha, P. Puigserver, mTOR controls mitochondrial oxidative function through a YY1- PGC-1[agr] transcriptional complex. Nature 450, 736-740 (2007).

286. J. D. Rabinowitz, E. White, Autophagy and Metabolism. Science (New York, N.Y.) 330, 1344-1348 (2010).

287. C. H. Jung, S.-H. Ro, J. Cao, N. M. Otto, D.-H. Kim, mTOR regulation of autophagy. FEBS letters 584, 1287-1295 (2010).

288. V. Law, C. Knox, Y. Djoumbou, T. Jewison, A. C. Guo, Y. Liu, A. Maciejewski, D. Arndt, M. Wilson, V. Neveu, A. Tang, G. Gabriel, C. Ly, S. Adamjee, Z. T. Dame, B. Han, Y. Zhou, D. S. Wishart, DrugBank 4.0: shedding new light on drug metabolism. Nucleic Acids Res 42, D1091-1097 (2014).

289. S. Faivre, G. Kroemer, E. Raymond, Current development of mTOR inhibitors as anticancer agents. Nat Rev Drug Discov 5, 671-688 (2006).

290. C. Vezina, A. Kudelski, S. N. Sehgal, Rapamycin (AY-22,989), a new antifungal antibiotic. I. Taxonomy of the producing streptomycete and isolation of the active principle. J Antibiot (Tokyo) 28, 721-726 (1975).

291. R. Y. Calne, S. Lim, A. Samaan, D. S. J. Collier, S. G. Pollard, D. J. G. White, S. Thiru, RAPAMYCIN FOR IMMUNOSUPPRESSION IN ORGAN ALLOGRAFTING. The Lancet 334, 227 (1989).

292. S. L. Schreiber, Chemistry and biology of the immunophilins and their immunosuppressive ligands. Science 251, 283-287 (1991).

293. R. N. Saunders, M. S. Metcalfe, M. L. Nicholson, Rapamycin in transplantation: A review of the evidence. Kidney International 59, 3-16 (2001).

294. F. X. McCormack , Y. Inoue , J. Moss , L. G. Singer , C. Strange , K. Nakata , A. F. Barker , J. T. Chapman , M. L. Brantly , J. M. Stocks , K. K. Brown , J. P. I. Lynch , H. J. Goldberg , L. R. Young , B. W. Kinder , G. P. Downey , E. J. Sullivan , T. V. Colby , R. T. McKay , M. M. Cohen , L. Korbee , A. M. Taveira-DaSilva , H.-S. Lee , J. P. Krischer , B. C. Trapnell Efficacy and Safety of Sirolimus in Lymphangioleiomyomatosis. New England Journal of Medicine 364, 1595-1606 (2011).

295. R. Yuan, A. Kay, W. J. Berg, D. Lebwohl, Targeting tumorigenesis: development and use of mTOR inhibitors in cancer therapy. Journal of Hematology & Oncology 2, 45-45 (2009).

296. L. G. Feun, A. Marini, H. Landy, A. Markoe, D. Heros, C. Robles, C. Herrera, N. Savaraj, Clinical trial of CPT-11 and VM-26/VP-16 for patients with recurrent malignant brain tumors. Journal of Neuro-Oncology 82, 177-181 (2007).

284

297. A. S. Don, O. Kisker, P. Dilda, N. Donoghue, X. Zhao, S. Decollogne, B. Creighton, E. Flynn, J. Folkman, P. J. Hogg, A peptide trivalent arsenical inhibits tumor angiogenesis by perturbing mitochondrial function in angiogenic endothelial cells. Cancer Cell 3, 497-509 (2003).

298. E. E. Ramsay, P. J. Hogg, P. J. Dilda, Mitochondrial metabolism inhibitors for cancer therapy. Pharm Res 28, 2731-2744 (2011).

299. H. Shen, University of New South Wales, Sydney, Australia (2014).

300. S. M. Chang, P. Wen, T. Cloughesy, H. Greenberg, D. Schiff, C. Conrad, K. Fink, H. I. Robins, L. De Angelis, J. Raizer, K. Hess, K. Aldape, K. R. Lamborn, J. Kuhn, J. Dancey, M. D. Prados, Phase II study of CCI-779 in patients with recurrent glioblastoma multiforme. Invest New Drugs 23, 357-361 (2005).

301. T. F. Cloughesy, K. Yoshimoto, P. Nghiemphu, K. Brown, J. Dang, S. Zhu, T. Hsueh, Y. Chen, W. Wang, D. Youngkin, L. Liau, N. Martin, D. Becker, M. Bergsneider, A. Lai, R. Green, T. Oglesby, M. Koleto, J. Trent, S. Horvath, P. S. Mischel, I. K. Mellinghoff, C. L. Sawyers, Antitumor activity of rapamycin in a Phase I trial for patients with recurrent PTEN-deficient glioblastoma. PLoS medicine 5, e8 (2008).

302. D. Park, J. Chiu, G. G. Perrone, P. J. Dilda, P. J. Hogg, The tumour metabolism inhibitors GSAO and PENAO react with cysteines 57 and 257 of mitochondrial adenine nucleotide translocase. Cancer Cell Int 12, 11 (2012).

303. T. R. Jue, K. Nozue, A. J. Lester, S. Joshi, L. B. Schroder, S. P. Whittaker, S. Nixdorf, R. W. Rapkins, M. Khasraw, K. L. McDonald, Veliparib in combination with radiotherapy for the treatment of MGMT unmethylated glioblastoma. J Transl Med 15, 61 (2017).

304. S. P. Jackson, J. Bartek, The DNA-damage response in human biology and disease. Nature 461, 1071-1078 (2009).

305. M. J. O'Connor, Targeting the DNA Damage Response in Cancer. Mol Cell 60, 547-560 (2015).

306. H. Erasimus, M. Gobin, S. Niclou, E. Van Dyck, DNA repair mechanisms and their clinical impact in glioblastoma. Mutation Research/Reviews in Mutation Research 769, 19-35 (2016).

307. S. P. Jackson, J. Bartek, The DNA-damage response in human biology and disease. Nature 461, 1071-1078 (2009).

308. S. Bao, Q. Wu, R. E. McLendon, Y. Hao, Q. Shi, A. B. Hjelmeland, M. W. Dewhirst, D. D. Bigner, J. N. Rich, Glioma stem cells promote radioresistance by preferential activation of the DNA damage response. Nature 444, 756-760 (2006).

309. O. Dizdar, C. Arslan, K. Altundag, Advances in PARP inhibitors for the treatment of breast cancer. Expert opinion on pharmacotherapy 16, 2751-2758 (2015).

285

310. A. Lavery, C. Gilson, S. Chowdhury, PARP inhibitors and stratified treatment of prostate cancer. Expert Review of Anticancer Therapy, (2016).

311. T. K. Owonikoko, G. Zhang, X. Deng, M. R. Rossi, J. M. Switchenko, G. H. Doho, Z. Chen, S. Kim, S. Strychor, S. M. Christner, Poly (ADP) ribose polymerase enzyme inhibitor, veliparib, potentiates chemotherapy and radiation in vitro and in vivo in small cell lung cancer. Cancer medicine 3, 1579-1594 (2014).

312. A. Papa, D. Caruso, M. Strudel, S. Tomao, F. Tomao, Update on Poly-ADP-ribose polymerase inhibition for ovarian cancer treatment. Journal of Translational Medicine 14, 267 (2016).

313. T. A. Greenhalgh, C. Dearman, R. A. Sharma, Combination of Novel Agents with Radiotherapy to Treat Rectal Cancer. Clinical Oncology 28, 116-139 (2016).

314. B. Lupo, L. Trusolino, Inhibition of poly(ADP-ribosyl)ation in cancer: Old and new paradigms revisited. Biochimica et Biophysica Acta (BBA) - Reviews on Cancer 1846, 201-215 (2014).

315. M. A. Morgan, L. A. Parsels, J. Maybaum, T. S. Lawrence, Improving the Efficacy of Chemoradiation with Targeted Agents. Cancer Discovery 4, 280-291 (2014).

316. F. A. Dungey, K. W. Caldecott, A. J. Chalmers, Enhanced radiosensitization of human glioma cells by combining inhibition of poly(ADP-ribose) polymerase with inhibition of heat shock protein 90. Mol Cancer Ther 8, 2243-2254 (2009).

317. H. Farmer, N. McCabe, C. J. Lord, A. N. Tutt, D. A. Johnson, T. B. Richardson, M. Santarosa, K. J. Dillon, I. Hickson, C. Knights, N. M. Martin, S. P. Jackson, G. C. Smith, A. Ashworth, Targeting the DNA repair defect in BRCA mutant cells as a therapeutic strategy. Nature 434, 917-921 (2005).

318. H. E. Bryant, N. Schultz, H. D. Thomas, K. M. Parker, D. Flower, E. Lopez, S. Kyle, M. Meuth, N. J. Curtin, T. Helleday, Specific killing of BRCA2-deficient tumours with inhibitors of poly(ADP-ribose) polymerase. Nature 434, 913-917 (2005).

319. P. C. Fong, D. Boss, C. Carden, M. Roelvink, J. De Greve, C. Gourley, J. Carmichael, J. De Bono, J. Schellens, S. Kaye, AZD2281 (KU-0059436), a PARP (poly ADP-ribose polymerase) inhibitor with single agent anticancer activity in patients with BRCA deficient ovarian cancer: results from a phase I study. Journal of Clinical Oncology 26, 5510-5510 (2008).

320. F. Lin, M. C. de Gooijer, E. M. Roig, L. C. Buil, S. M. Christner, J. H. Beumer, T. Wurdinger, J. H. Beijnen, O. van Tellingen, ABCB1, ABCG2, and PTEN determine the response of glioblastoma to temozolomide and ABT-888 therapy. Clin Cancer Res 20, 2703-2713 (2014).

286

321. J. Majuelos-Melguizo, M. I. Rodriguez, L. Lopez-Jimenez, J. M. Rodriguez- Vargas, J. M. Marti Martin-Consuegra, S. Serrano-Saenz, J. Gavard, J. M. de Almodovar, F. J. Oliver, PARP targeting counteracts gliomagenesis through induction of mitotic catastrophe and aggravation of deficiency in homologous recombination in PTEN-mutant glioma. Oncotarget 6, 4790-4803 (2015).

322. M. S. Gilardini Montani, A. Prodosmo, V. Stagni, D. Merli, L. Monteonofrio, V. Gatti, M. P. Gentileschi, D. Barila, S. Soddu, ATM-depletion in breast cancer cells confers sensitivity to PARP inhibition. J Exp Clin Cancer Res 32, 95 (2013).

323. J. Mateo, S. Carreira, S. Sandhu, S. Miranda, H. Mossop, R. Perez-Lopez, D. Nava Rodrigues, D. Robinson, A. Omlin, N. Tunariu, G. Boysen, N. Porta, P. Flohr, A. Gillman, I. Figueiredo, C. Paulding, G. Seed, S. Jain, C. Ralph, A. Protheroe, S. Hussain, R. Jones, T. Elliott, U. McGovern, D. Bianchini, J. Goodall, Z. Zafeiriou, C. T. Williamson, R. Ferraldeschi, R. Riisnaes, B. Ebbs, G. Fowler, D. Roda, W. Yuan, Y.-M. Wu, X. Cao, R. Brough, H. Pemberton, R. A’Hern, A. Swain, L. P. Kunju, R. Eeles, G. Attard, C. J. Lord, A. Ashworth, M. A. Rubin, K. E. Knudsen, F. Y. Feng, A. M. Chinnaiyan, E. Hall, J. S. de Bono, DNA-Repair Defects and Olaparib in Metastatic Prostate Cancer. New England Journal of Medicine 373, 1697-1708 (2015).

324. K. Obermeier, J. Sachsenweger, T. W. Friedl, H. Pospiech, R. Winqvist, L. Wiesmuller, Heterozygous PALB2 c.1592delT mutation channels DNA double- strand break repair into error-prone pathways in breast cancer patients. Oncogene 35, 3796-3806 (2016).

325. N. McCabe, N. C. Turner, C. J. Lord, K. Kluzek, A. Bialkowska, S. Swift, S. Giavara, M. J. O'Connor, A. N. Tutt, M. Z. Zdzienicka, G. C. Smith, A. Ashworth, Deficiency in the repair of DNA damage by homologous recombination and sensitivity to poly(ADP-ribose) polymerase inhibition. Cancer Res 66, 8109-8115 (2006).

326. A. Krumm, C. Barckhausen, P. Kücük, K.-H. Tomaszowski, C. Loquai, J. Fahrer, O. H. Krämer, B. Kaina, W. P. Roos, Enhanced histone deacetylase activity in malignant melanoma provokes RAD51 and FANCD2 triggered drug resistance. Cancer Research, (2016).

327. S. K. Gupta, A. C. Mladek, B. L. Carlson, F. Boakye-Agyeman, K. Bakken, S. Kizilbash, M. A. Schroeder, J. Reid, J. N. Sarkaria, Discordant in vitro and in vivo chemo-potentiating effects of the PARP inhibitor veliparib in temozolomide- sensitive versus -resistant glioblastoma multiforme xenografts. Clinical cancer research : an official journal of the American Association for Cancer Research 20, 3730-3741 (2014).

328. F. A. Dungey, D. A. Loser, A. J. Chalmers, Replication-dependent radiosensitization of human glioma cells by inhibition of poly(ADP-Ribose) polymerase: mechanisms and therapeutic potential. Int J Radiat Oncol Biol Phys 72, 1188-1197 (2008).

287

329. A. L. Russo, H. C. Kwon, W. E. Burgan, D. Carter, K. Beam, X. Weizheng, J. Zhang, B. S. Slusher, A. Chakravarti, P. J. Tofilon, K. Camphausen, In vitro and in vivo radiosensitization of glioblastoma cells by the poly (ADP-ribose) polymerase inhibitor E7016. Clin Cancer Res 15, 607-612 (2009).

330. M. Venere, P. Hamerlik, Q. Wu, R. D. Rasmussen, L. A. Song, A. Vasanji, N. Tenley, W. A. Flavahan, A. B. Hjelmeland, J. Bartek, J. N. Rich, Therapeutic targeting of constitutive PARP activation compromises stem cell phenotype and survival of glioblastoma-initiating cells. Cell Death and Differentiation 21, 258- 269 (2014).

331. B. Lemasson, H. Wang, S. Galbán, Y. Li, Y. Zhu, K. A. Heist, C. Tsein, T. L. Chenevert, A. Rehemtulla, C. J. Galbán, E. C. Holland, B. D. Ross, Evaluation of Concurrent Radiation, Temozolomide and ABT-888 Treatment Followed by Maintenance Therapy with Temozolomide and ABT-888 in a Genetically Engineered Glioblastoma Mouse Model. Neoplasia (New York, N.Y.) 18, 82-89 (2016).

332. T. C. Chou, P. Talalay, Quantitative analysis of dose-effect relationships: the combined effects of multiple drugs or enzyme inhibitors. Adv Enzyme Regul 22, 27-55 (1984).

333. T.-C. Chou, Drug Combination Studies and Their Synergy Quantification Using the Chou-Talalay Method. Cancer Research 70, 440-446 (2010).

334. T.-C. Chou, Theoretical Basis, Experimental Design, and Computerized Simulation of Synergism and Antagonism in Drug Combination Studies. Pharmacological Reviews 58, 621-681 (2006).

335. T. Scholzen, J. Gerdes, The Ki-67 protein: from the known and the unknown. J Cell Physiol 182, 311-322 (2000).

336. Y. Somnay, S. Lubner, H. Gill, J. B. Matsumura, H. Chen, The PARP inhibitor ABT-888 potentiates dacarbazine-induced cell death in carcinoids. Cancer Gene Ther 23, 348-354 (2016).

337. L. Barazzuol, R. Jena, N. G. Burnet, L. B. Meira, J. C. Jeynes, K. J. Kirkby, N. F. Kirkby, Evaluation of poly (ADP-ribose) polymerase inhibitor ABT-888 combined with radiotherapy and temozolomide in glioblastoma. Radiation oncology (London, England) 8, 65 (2013).

338. B. J. Lamarche, N. I. Orazio, M. D. Weitzman, The MRN complex in Double- Strand Break Repair and Telomere Maintenance. FEBS letters 584, 3682-3695 (2010).

339. R. Koppensteiner, E. P. Samartzis, A. Noske, A. von Teichman, I. Dedes, M. Gwerder, P. Imesch, K. Ikenberg, H. Moch, D. Fink, M. Stucki, K. J. Dedes, Effect of MRE11 Loss on PARP-Inhibitor Sensitivity in Endometrial Cancer In Vitro. PLOS ONE 9, e100041 (2014).

288

340. L. A. McPherson, Y. Shen, J. M. Ford, Poly (ADP-ribose) polymerase inhibitor LT-626: Sensitivity correlates with MRE11 mutations and synergizes with platinums and irinotecan in colorectal cancer cells. Cancer Letters 343, 217-223 (2014).

341. M. J. Clarke, E. A. Mulligan, P. T. Grogan, A. C. Mladek, B. L. Carlson, M. A. Schroeder, N. J. Curtin, Z. Lou, P. A. Decker, W. Wu, E. R. Plummer, J. N. Sarkaria, Effective sensitization of temozolomide by ABT-888 is lost with development of TMZ resistance in glioblastoma xenograft lines. Molecular cancer therapeutics 8, 407-414 (2009).

342. T. R. Jue, E. S. Sena, M. R. Macleod, K. L. McDonald, T. C. Hirst, A systematic review and meta-analysis of topoisomerase inhibition in pre-clinical glioma models. Oncotarget 9, 11387-11401 (2018).

343. M. R. Macleod, A. L. McLean, A. Kyriakopoulou, S. Serghiou, A. de Wilde, N. Sherratt, T. Hirst, R. Hemblade, Z. Bahor, C. Nunes-Fonseca, Risk of bias in reports of in vivo research: a focus for improvement. PLoS Biol 13, e1002273 (2015).

344. E. S. Sena, C. L. Briscoe, D. W. Howells, G. A. Donnan, P. A. G. Sandercock, M. R. Macleod, Factors Affecting the Apparent Efficacy and Safety of Tissue Plasminogen Activator in Thrombotic Occlusion Models of Stroke: Systematic Review and Meta-Analysis. Journal of Cerebral Blood Flow & Metabolism 30, 1905-1913 (2010).

345. K. Søe, F. Grosse, p53 stimulates human topoisomerase I activity by modulating its DNA binding. Nucleic Acids Research 31, 6585-6592 (2003).

346. C. Gobert, A. Skladanowski, A. K. Larsen, The interaction between p53 and DNA topoisomerase I is regulated differently in cells with wild-type and mutant p53. Proceedings of the National Academy of Sciences 96, 10355-10360 (1999).

347. M. I. Sandri, R. J. Isaacs, W. M. Ongkeko, A. L. Harris, I. D. Hickson, M. Broggini, F. Vikhanskaya, p53 Regulates the Minimal Promoter of the Human Topoisomerase IIα Gene. Nucleic Acids Research 24, 4464-4470 (1996).

348. M. Ataka, M. Ikeguchi, M. Yamamoto, M. Inoue, T. Tanida, S. OKA, K. Katano, Topoisomerase I protein expression and prognosis of patients with colorectal cancer. Yonago Acta medica 50, 81-87 (2007).

349. D. A. Burden, N. Osheroff, Mechanism of action of eukaryotic topoisomerase II and drugs targeted to the enzyme. Biochimica et Biophysica Acta (BBA) - Gene Structure and Expression 1400, 139-154 (1998).

350. J. J. Champoux, R. Dulbecco, An activity from mammalian cells that untwists superhelical DNA--a possible swivel for DNA replication (polyoma-ethidium bromide-mouse-embryo cells-dye binding assay). Proc Natl Acad Sci U S A 69, 143-146 (1972).

289

351. G. Reynes, V. Martinez-Sales, V. Vila, C. Balana, P. Perez-Segura, M. A. Vaz, M. Benavides, O. Gallego, I. Palomero, M. Gil-Gil, T. Fleitas, E. Reche, Phase II trial of irinotecan and metronomic temozolomide in patients with recurrent glioblastoma. Anticancer Drugs 27, 133-137 (2016).

352. J. R. Whittle, J. D. Lickliter, H. K. Gan, A. M. Scott, J. Simes, B. J. Solomon, J. A. MacDiarmid, H. Brahmbhatt, M. A. Rosenthal, First in human nanotechnology doxorubicin delivery system to target epidermal growth factor receptors in recurrent glioblastoma. J Clin Neurosci 22, 1889-1894 (2015).

353. K. B. Peters, E. Lou, A. Desjardins, D. A. Reardon, E. S. Lipp, E. Miller, J. E. Herndon, 2nd, F. McSherry, H. S. Friedman, J. J. Vredenburgh, Phase II Trial of Upfront Bevacizumab, Irinotecan, and Temozolomide for Unresectable Glioblastoma. Oncologist 20, 727-728 (2015).

354. B. Chauffert, L. Feuvret, F. Bonnetain, L. Taillandier, D. Frappaz, H. Taillia, R. Schott, J. Honnorat, M. Fabbro, I. Tennevet, F. Ghiringhelli, J. S. Guillamo, X. Durando, D. Castera, M. Frenay, C. Campello, C. Dalban, J. Skrzypski, O. Chinot, Randomized phase II trial of irinotecan and bevacizumab as neo-adjuvant and adjuvant to temozolomide-based chemoradiation compared with temozolomide- chemoradiation for unresectable glioblastoma: final results of the TEMAVIR study from ANOCEFdagger. Ann Oncol 25, 1442-1447 (2014).

355. G. Reynes, C. Balana, O. Gallego, L. Iglesias, P. Perez, J. L. Garcia, A phase I study of irinotecan in combination with metronomic temozolomide in patients with recurrent glioblastoma. Anticancer Drugs 25, 717-722 (2014).

356. A. Leonard, J. E. Wolff, Etoposide improves survival in high-grade glioma: a meta-analysis. Anticancer Res 33, 3307-3315 (2013).

357. A. Sato, B. Klaunberg, R. Tolwani, In vivo bioluminescence imaging. Comp Med 54, 631-634 (2004).

358. P. J. Houghton, C. F. Stewart, P. J. Cheshire, L. B. Richmond, M. N. Kirstein, C. A. Poquette, M. Tan, H. S. Friedman, T. P. Brent, Antitumor activity of temozolomide combined with irinotecan is partly independent of O6- methylguanine-DNA methyltransferase and mismatch repair phenotypes in xenograft models. Clinical cancer research : an official journal of the American Association for Cancer Research 6, 4110-4118 (2000).

359. R. Villalonga-Planells, L. Coll-Mulet, F. Martínez-Soler, E. Castaño, J.-J. Acebes, P. Giménez-Bonafé, J. Gil, A. Tortosa, Activation of p53 by Nutlin-3a Induces Apoptosis and Cellular Senescence in Human Glioblastoma Multiforme. PLOS ONE 6, e18588 (2011).

360. B. Leroy, L. Girard, A. Hollestelle, J. D. Minna, A. F. Gazdar, T. Soussi, Analysis of TP53 Mutation Status in Human Cancer Cell Lines: A Reassessment. Human mutation 35, 756-765 (2014).

290

361. R. F. Barth, B. Kaur, Rat brain tumor models in experimental neuro-oncology: the C6, 9L, T9, RG2, F98, BT4C, RT-2 and CNS-1 gliomas. Journal of neuro- oncology 94, 299-312 (2009).

362. T. C. Hirst, H. M. Vesterinen, S. Conlin, K. J. Egan, A. Antonic, A. Lawson McLean, M. R. Macleod, R. Grant, P. M. Brennan, E. S. Sena, I. R. Whittle, A systematic review and meta-analysis of gene therapy in animal models of cerebral glioma: why did promise not translate to human therapy? Evidence-based Preclinical Medicine 1, 21-33 (2014).

363. S. G. Thompson, J. P. Higgins, How should meta-regression analyses be undertaken and interpreted? Stat Med 21, 1559-1573 (2002).

364. E. Walker, A. V. Hernandez, M. W. Kattan, Meta-analysis: Its strengths and limitations. Cleve Clin J Med 75, 431-439 (2008).

365. W. Viechtbauer, Accounting for Heterogeneity via Random-Effects Models and Moderator Analyses in Meta-Analysis. Zeitschrift für Psychologie / Journal of Psychology 215, 104-121 (2007).

366. P. J. Easterbrook, J. A. Berlin, R. Gopalan, D. R. Matthews, Publication bias in clinical research. Lancet 337, 867-872 (1991).

367. M. R. Macleod, A. Lawson McLean, A. Kyriakopoulou, S. Serghiou, A. de Wilde, N. Sherratt, T. Hirst, R. Hemblade, Z. Bahor, C. Nunes-Fonseca, A. Potluru, A. Thomson, J. Baginskaite, K. Egan, H. Vesterinen, G. L. Currie, L. Churilov, D. W. Howells, E. S. Sena, Risk of Bias in Reports of In Vivo Research: A Focus for Improvement. PLoS Biol 13, e1002273 (2015).

368. L. Chen, Y. Zhang, J. Yang, J. P. Hagan, M. Li, Vertebrate animal models of glioma: understanding the mechanisms and developing new therapies. Biochim Biophys Acta 1836, 158-165 (2013).

369. K. Jellinger, Glioblastoma multiforme: Morphology and biology. Acta Neurochirurgica 42, 5-32 (1978).

370. Kyeung M. Joo, J. Kim, J. Jin, M. Kim, Ho J. Seol, J. Muradov, H. Yang, Y.-L. Choi, W.-Y. Park, D.-S. Kong, J.-I. Lee, Y.-H. Ko, Hyun G. Woo, J. Lee, S. Kim, D.-H. Nam, Patient-Specific Orthotopic Glioblastoma Xenograft Models Recapitulate the Histopathology and Biology of Human Glioblastomas In Situ. Cell Reports 3, 260-273 (2013).

371. D. William, C. S. Mullins, B. Schneider, A. Orthmann, N. Lamp, M. Krohn, A. Hoffmann, C.-F. Classen, M. Linnebacher, Optimized creation of glioblastoma patient derived xenografts for use in preclinical studies. Journal of Translational Medicine 15, 27 (2017).

372. S. Kunwar, S. Chang, M. Westphal, M. Vogelbaum, J. Sampson, G. Barnett, M. Shaffrey, Z. Ram, J. Piepmeier, M. Prados, D. Croteau, C. Pedain, P. Leland, S. R. Husain, B. H. Joshi, R. K. Puri, P. S. G. for the, Phase III randomized trial of

291

CED of IL13-PE38QQR vs Gliadel wafers for recurrent glioblastoma(). Neuro- Oncology 12, 871-881 (2010).

373. J. Perry, A. Chambers, K. Spithoff, N. Laperriere, Gliadel wafers in the treatment of malignant glioma: a systematic review. Current Oncology 14, 189-194 (2007).

374. A. I. o. H. Welfare, "Cancer in Australia 2017," (AIHW, 2017).

375. M. Kanehisa, M. Furumichi, M. Tanabe, Y. Sato, K. Morishima, KEGG: new perspectives on genomes, pathways, diseases and drugs. Nucleic Acids Research 45, D353-D361 (2017).

376. N. J. Schork, Personalized medicine: Time for one-person trials. Nature 520, 609- 611 (2015).

377. H. Bando, N. Takebe, Perspectives on research activity in the USA on Cancer Precision Medicine. Japanese Journal of Clinical Oncology 46, 106-110 (2016).

292