IMPROVING THE TREATMENT OF INTERMEDIATE RISK (IR) B-CELL PRECURSOR (BCP) ACUTE LYMPHOBLASTIC LEUKAEMIA

Abdulmohsen M Alruwetei (MSc)

Children’s Cancer Institute and School of Women’s and Children’s Health Faculty of Medicine University of New South Wales (UNSW)

A thesis submitted to the University of New South Wales in fulfilment of the requirements for the degree of Doctorate of Philosophy

April, 2016

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

Surname or Family name: Alruwctei

Pi rsl name: Abdulmohsen Other name/s:

Abbrev.iation for degree as given in the University calendar:

School: School of Women's & Children's 1lealth faculty: Faculty of Medicine

'fitle: Improving the lrcalment of intermediate risk (JR) 13-cell prccw·sor (BCP) acute lymphoblastic leukaemia

Abstract Most childhood ALL patients are stratified at diagnosis as Jnicrmediate Risk ( IR), although an unacceptably high proportion of th ese patients relapse. The lack of specific prognostic featu res makes it dilTicult to predict the response of IR patients to treatment. Recent progress in the development of paticnl•derived xenografts in Immune-deficient mice represents an opportunity to improve relapse prediction in ALL patients.

In this :;tudy, a pilot experiment was carried out to optimise the most appropriate engrallment strategy, which could stratify engrathncnt of samples from IR 8 cell Precursor (BCP) ALL patients according lo clinical outcomes. l-1 igh efficiency and quicker cngraftmenl of JR ALL patient samples were observed in NSO mice and via the intravenous route of inoculation over NOD/SCID mice and the intrafcmoral route. ·rhe response ofxenogra·fts to combination chemotherapy showed evidence of renccting patiem outcome. Validation oflhe capability of the optimised cngraftmcnt strategy to identify relapse in IR ALL patients us ing a larger patient cohort emphasised the i-ole of drug treatment to discriminate between relapsed and non-relapsed IR ALL patients. Certainly, the time required for ALL patient samples to reach 25% ongrallment in t·he peripheral blood oftbe dntg· treated mice was the most significanl -parameter for predicting patient outcomes.

Analysis of the ex vivo drug sensitivity of ALL .xenog1·ans revealed thal tho responses to vincristine and L-asparaginasc reflected tl,e in vivo responses to drug solt.:ction. Moreover, samples from one ALL xonograft selected in vivo will, drug treatmcnLs howed increased resistance to vincristine ex 11ivo compared with the non drug-treated samples of the same xenogran. Microarray analysis of expression identified HSP90 as a potential target for reve1·sing resistance, although lhe HSP90 inhibitor , I 7DMAG failed to enhance the inhibitory effect of vii1cristine ex vivo. rurther analysis of gene cxpl"ession revealed significant up-reguh11ion of involved in dynamics and/or stability of the microtubu.le network in vincristine-rcsistant samples.

In summary, engraf\mont of IR ALL samples under selective chemotherapy treatment cou Id provide a clinical approach for upfront prediction of outcome in JR ALL patients, which could allow tailoring the intensity of treatment according t<> lhe risk of rclause within this patient subgroup.

Declaration relating to disposition of project thesis/dissertation

I hereby grant to Lhc 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 ot media, now or here after known, subject lo the provisions of the Copyright Act 1968, 1 rctnin all property ri ghts, such as potent rights. l also retain the right lo use in future works (such as articles or books) nl l or part of this thesis or dissertation.

t also authorise U11ivcrsity Microlilms 10 use the 350 word abstract ofmy thesis in Dissertation Abstracts International (Lbis is applicable to doctoral theses only). I • .. .'1. ..I..[ ..

The University recognises that there may be exceptional circumstances requiring res trictions on copying or conditions on use. Requests for restriction for a period of up to 2 years must be madt! in writing. Requests for a longor period of restriction may be considered in cxccntionril circ11ms1n11ces .and require the approvnl of the Dean of Graduate Resci,rch.

F'OR OFFICE USE ONLY Date of completion ofrequirements for Award:

THIS SI IEET IS TO ae GLUl!O TO THB INSIDE f.RONT COVER OF' Tl IE TH8SIS 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 …………………. Dated …………………...

ii COPYRIGHT STATEMENT ‘I hereby grant the University of New South Wales or its agents 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 …………………. Dated …………………...

iii 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 …………………. Dated …………………...

iv ACKNOWLEDGEMENTS Firstly, I would like to thank my supervisor Prof. Richard Lock for his continued support and great guidance throughout the last four years. Thank you for giving me the opportunity to be a part of your group, for helping me to participate in national and international meetings and for reviewing and proofreading my thesis. I also would like to thank my co-supervisors Prof. Glenn Marshall and Dr. Rosemary Sutton for their support and giving feedback and advice about my research work. I am also grateful to the Qassim University for sponsoring me during my PhD study and the Steven

Walter Children's Cancer Foundation for funding this project.

I would like to express my sincere gratitude to both past and present members of

Leukaemia Biology Program who have help me during PhD. Particular thanks to

Kathryn Evans, Jennifer Richmond, Dr. Santi Syuriani for training and assistance in many practical issues. Thank you to Dr. Hernan Carol for his help to inoculate patient samples for the Pilot Study. Thanks you Dr. Babasaheb Yadav for your contribution to the mice work for the Main Study.

Last but not the least, I would like to thank my family: Dad and Mum (Menwer and

Mutairah), for their love, care, support and encouragement in many different ways over the years and until today. Thanks to my wife, Abeer for her love, patience, support throughout the years and taking care of me and my lovely kids Ghala, Hala and Omar who make my life more interesting and meaningful. I really look forward to celebrate with you.

v CONFERENCES, PUBLICATIONS AND AWARDS Conferences Alruwetei, A.M., Carol, H., Sutton, R., Marshall, G.M., and Lock, R.B. (2015). Evaluation of Patient-Derived Xenografts for Modeling Outcome of Pediatric B-Cell Precursor Acute Lymphoblastic Leukemia (poster presentation). In: Proceedings of American Society of Hematology Annual Meeting Exposition 57th, 2015 December 5- 8; Orlando, Florida: Abstract 3759.

Alruwetei, A.M., Carol, H., Sutton, R., Marshall, G.M., and Lock, R.B. (2015). Utilisation of Patient-Derived Xenograft Models to Improve Relapse Prediction in Intermediate Risk Paediatric Acute Lymphoblastic Leukaemia (poster presentation). ASMR Scientific Meeting 2015, Sydney, NSW, Australia.

Alruwetei, A.M., Carol, H., Sutton, R., Marshall, G.M., and Lock, R.B. (2014).. Utilisation of Patient-Derived Xenograft Models to Improve Relapse Prediction in Intermediate Risk Paediatric Acute Lymphoblastic Leukaemia (Poster presentation). New Directions in Leukaemia Research Meeting 2014, Noosa, QLD, Australia

vi TABLE OF CONTENTS AUTHENTICITY STATEMENT ...... iv ACKNOWLEDGEMENTS...... v CONFERENCES, PUBLICATIONS AND AWARDS ...... vi LIST OF FIGURES ...... xiii LIST OF TABLES ...... xvii ABBREVIATIONS AND ACRONYMS ...... xix ABSTRACT ...... xxii Chapter 1 INTRODUCTION ...... 1 Lymphopoiesis: A general approach to the nature of lymphoid leukaemia ...... 2 1.1.1 Normal development of blood cells ...... 2 1.1.2 B cell development ...... 3 1.1.3 Lymphocyte malignant transformation ...... 4 Childhood acute lymphoblastic leukaemia: An overview ...... 7 Demographics and aetiology of the disease ...... 7 Diagnosis of childhood acute lymphoblastic leukaemia ...... 8 Classification of childhood acute lymphoblastic leukaemia ...... 10 Basic principles and evolution of chemotherapy-based treatment of childhood acute lymphoblastic leukaemia ...... 11 1.6.1 Treatment protocols used for acute lymphoblastic leukaemia patients ...... 12 1.6.1.1 Induction of Remission ...... 13 1.6.1.2 Post-induction therapy (Intensification) ...... 14 1.6.1.3 Maintenance therapy ...... 15 1.6.2 Common chemotherapeutic agents used in acute lymphoblastic leukaemia treatment protocols ...... 15 1.6.2.1 Glucocorticoids ...... 15 1.6.2.2 Vincristine ...... 18 1.6.2.3 L-Asparaginase ...... 18 1.6.3 Treatment of recurrent acute lymphoblastic leukaemia ...... 19 Prognosis of childhood acute lymphoblastic leukaemia ...... 20 1.7.1 Clinical factors ...... 21 1.7.2 Immunophenotype ...... 22 1.7.3 Cytogenetic features and genetic alterations ...... 24 1.7.4 Response to initial treatment: Early studies ...... 27 vii 1.7.5 Response to initial treatment: Monitoring of minimal residual disease ...... 28 Risk-adapted therapy protocols ...... 29 1.8.1 Intermediate-risk BCP-acute lymphoblastic leukaemia group as a heterogeneous class of the disease ...... 30 Understanding the biology of acute lymphoblastic leukaemia relapse ...... 32 1.9.1 Leukaemia heterogeneity: Evidence of concepts ...... 36 1.9.2 Current concepts about the biology of relapse in acute lymphoblastic leukaemia disease ...... 37 Personalised therapy: Towards “Omics”-oriented treatment of childhood acute lymphoblastic leukaemia ...... 42 Disease modelling strategies ...... 47 1.11.1 Primary patient cells ...... 48 1.11.2 Immortalised cell lines ...... 48 1.11.3 Genetically engineered mouse models ...... 49 patient-derived xenografts ...... 52 1.12.1 Progress in acute lymphoblastic leukaemia patient-derived xenografts ...... 54 Aims and summary of the topic ...... 59 Chapter 2 MATERIALS AND METHODS ...... 60 2.1 Xenograft mouse model ...... 61 2.1.1 Reagents and equipment ...... 61 2.1.2 Preparation of patient samples for inoculation into mice ...... 62 2.1.3 Intravenous (IV) transplantation of leukaemia cells ...... 62 2.1.4 Intra-femoral (IF) transplantation of leukaemia cells ...... 63 2.1.5 In vivo treatment with VXL chemotherapy ...... 63 2.1.6 Monitoring of engraftment ...... 64 2.1.7 Harvesting of leukaemia cells from engrafted mice ...... 65 2.1.8 Analysis of mouse EFS at different levels of engraftment ...... 66 2.2 Cytotoxicity assay ...... 67 2.2.1 Reagents and equipment ...... 67 2.2.2 Preparation of cells for ex vivo cytotoxicity analysis ...... 67 2.2.3 Alamar Blue cytotoxicity assay ...... 68 2.2.4 Trypan blue exclusion assay...... 70 2.3 Preparation of RNA for analysis ...... 70 2.3.1 Reagents and equipment ...... 70 viii 2.3.2 RNA extraction ...... 71 2.3.3 RNA amplification ...... 73 2.4 Illumina gene expression ...... 76 2.4.1 Normalisation and transformation ...... 76 2.4.2 Hierarchical clustering...... 76 2.4.3 Differential gene expression ...... 76 2.4.4 Gene Set Enrichment Analysis (GSEA) ...... 77 2.4.5 Connectivity Map (CMap) analysis ...... 77 2.5 Quantitative real-time reverse transcriptase polymerase chain reaction (qRT-PCR) ...... 78 2.5.1.1 Reagents and equipment ...... 78 2.5.1.2 cDNA synthesis ...... 79 2.5.1.3 Real-time qRT-PCR ...... 79 2.6 Statistical analysis ...... 80 2.7 Ethics ...... 80 Chapter 3 PILOT STUDY: OPTIMISATION OF XENOGRAFT MODELS TO PREDICT OUTCOME IN B CELL PRECURSOR (BCP) INTERMEDIATE RISK (IR) ALL PATIENTS ...... 81 3.1 Introduction ...... 82 3.2 Selection of IR ALL Patients for the Pilot study ...... 86 3.3 Experimental design of the Pilot Study ...... 89 3.4 Establishment and characterisation of IR ALL PDXs ...... 92 3.4.1 ALL-64 ...... 92 3.4.2 ALL-65 ...... 95 3.4.3 ALL-66 ...... 98 3.4.4 ALL-67 ...... 103 3.5 Comparison of engraftment efficiency between IR ALL xenografts established using various engraftment conditions ...... 108 3.6 Comparison of rate of engraftment between IR ALL xenografts established using various engraftment conditions ...... 110 3.6.1 NSG versus NOD/SCID ...... 110 3.6.2 IV versus IF inoculation ...... 114 3.7 Assessing the capability of different transplantation conditions to improve prediction of outcomes in IR ALL patients ...... 117 ix 3.7.1 Stratification of IR ALL xenografts based on engraftment in NSG mice at time to 1% human CD45+ cells ...... 118 3.7.2 Stratification of IR ALL xenografts based on engraftment in NOD/SCID mice at time to 1% human CD45+ cells ...... 123 3.8 Assessing the engraftment features of IR ALL xenografts based on time to 25% human CD45+ cells in mice PB ...... 127 3.8.1 NSG versus NOD/SCID ...... 127 3.8.2 IV versus IF inoculation ...... 127 3.8.3 Stratification of IR ALL xenografts based on EFS of NSG mice at time to 25% human CD45+ cells ...... 133 3.8.4 Stratification of IR ALL xenografts based on EFS of NOD/SCID mice at time to 25% human CD45+ cells ...... 137 3.9 Assessing the engraftment features of IR ALL xenografts based on time to leukaemia ...... 140 3.10 Summary and Discussion ...... 149 Chapter 4 MAIN STUDY: ASSESSING THE CAPABILITY OF XENOGRAFTS TO PREDICT RELAPSE IN IR BCP-ALL PATIENTS ...... 156 4.1 Introduction ...... 157 4.2 Patient selection for the Main Study ...... 160 4.3 Establishing and characterising engraftment of the IR BCP-ALL patient primary samples ...... 164 4.3.1 Xenografts established from the first panel of primary ALL patient samples…...... 167 4.3.1.1 Summary of xenograft engraftment kinetics ...... 173 4.3.2 Xenografts established from the second panel of primary ALL patient samples...... 175 4.3.2.1 Summary of xenograft engraftment kinetics ...... 181 4.3.3 Xenografts established from the third panel of primary ALL patient samples...... 183 4.3.3.1 Summary of xenograft engraftment kinetics ...... 189 4.4 Assessment of the reliability of combining the engraftment data from the entire cohort for analysis of relapse prediction ...... 191 4.4.1 Patients’ clinical characteristics ...... 191 4.4.2 Engraftment characteristics ...... 197 x 4.5 Systematic analysis of modelling IR ALL patient outcomes in patient derived xenografts ...... 199 4.5.1 Impact of engraftment in VXL-treated and non-treated mouse models ...... 203 4.5.2 The impact of time to event on stratification of IR ALL xenografts ...... 209 4.5.3 Median versus first mouse EFS ...... 215 4.6 Characterising the engraftment at TT25% human cells in the VXL-treated mice as a predictor for relapse in IR ALL patients ...... 221 4.6.1 Stratification of IR ALL patients based on median TT25% ...... 222 4.6.2 Stratification of IR ALL patients according to the LGD values at TT25% . 229 4.7 Characterising the TTL in the VXL-treated mice as a predictor for identifying early from late relapsed IR ALL patients ...... 233 4.8 Summary and Discussion ...... 237 Chapter 5 CHARACTERISATION OF THE HETEROGENEITY BETWEEN IR ALL XENOGRAFTS AND THEIR RESPONSE TO CHEMOTHERAPEUTIC DRUGS………...... 246 5.1 Introduction ...... 247 5.2 Diversity in transcriptional profiles of an IR ALL xenograft and its corresponding patient sample ...... 250 5.3 Ex vivo sensitivity of ALL xenografts to induction chemotherapy ...... 262 5.3.1 Response of xenografts derived from ALL patients with different outcome (relapse vs. non-relapse) to induction chemotherapy ...... 263 5.3.2 Assessment of the development of chemotherapy induced resistance ...... 269 5.4 Identification of drug leads that could overcome vincristine resistance in VXL- treated ALL-67 cells ...... 275 5.5 Identification of genes associated with vincristine resistance in the ALL-67 xenograft ...... 284 5.6 Summary and Discussion ...... 288 5.6.1 Diversity in transcriptional profiles of an IR ALL xenograft and its corresponding patient sample ...... 288 5.6.2 Response of xenografts derived from ALL patients with different outcome (relapse vs. non-relapse) to induction chemotherapy ...... 290 5.6.3 Assessment of the development of chemotherapy induced resistance ...... 293 5.6.4 Identification of drug leads that could overcome vincristine resistance in the VXL-treated ALL-67 cells ...... 294 xi 5.6.5 Identification of genes associated with vincristine resistance in the ALL-67 xenograft ...... 297 Chapter 6 FINAL DISCUSSION AND FUTURE DIRECTIONS ...... 299 REFERENCES ...... 308

xii LIST OF FIGURES Figure 1.1. Stages of B cell development showing molecules that regulate normal development...... 6 Figure 1.2. The risk of isolated central nervous system (CNS) relapse over a period of seven years in ALL patients treated with steroids ...... 17 Figure 1.3. Frequency of genetic abnormalities in childhood ALL...... 26 Figure 1.4. Schematic diagram illustrates the mechanisms associated with antimicrotubule drug resistance ...... 34 Figure 1.5. Models of clonal progression in acute leukaemia ...... 39 Figure 3.1. Schematic diagram of the experimental plan used to establish xenografts from the IR ALL cohort for the Pilot Study ...... 91 Figure 3.2. Patient sample A5072 engrafted in NSG mice to establish the ALL-64 xenograft ...... 93 Figure 3.3. Infiltration of ALL-64 into organs of NSG mice at day of harvest ...... 94 Figure 3.4. Engraftment of patient sample A1795 inoculated into VXL treated and non- treated mice to establish ALL-65 xenograft ...... 96 Figure 3.5. Infiltration of ALL-65 into organs of non-drug treated mice at day of harvest ...... 97 Figure 3.6. Engraftment of patient sample A1839 inoculated into mice to establish ALL-66 in the presence or absence of VXL selection ...... 99 Figure 3.7. Infiltration of NSG mouse organs with ALL-66 at the time of harvest ..... 100 Figure 3.8. Infiltration of NOD/SCID mice organs with ALL-66 at day of harvest .... 101 Figure 3.9. Image of ALL-66 established in NSG mice showing splenomegaly and lymphadenopathy ...... 102 Figure 3.10. Engraftment of patient sample A4334 inoculated into VXL-treated and non-treated mice to establish ALL-67 xenograft ...... 105 Figure 3.11. Infiltration of NSG mice organs with ALL-67 at day of harvest ...... 106 Figure 3.12. Infiltration of NOD/SCID mice organs with ALL-67 at harvest ...... 107 Figure 3.13. Graphs of individual engraftment conditions showing how each patient sample engrafted at 1% human CD45+ in NSG mice compared to NOD/SCID mice . 113 Figure 3.14. Graphs of individual engraftment conditions showing mouse EFS when mice were inoculated via IV and IF routes ...... 116

xiii Figure 3.15. Comparison of the EFS of NSG mice inoculated with IR ALL patient samples using different transplantation conditions ...... 121 Figure 3.16. Comparison of the EFS of NOD/SCID mice inoculated with IR ALL patient samples using different transplantation conditions ...... 125 Figure 3.17. Graphs of individual engraftment conditions showing how each patient sample engrafted at 25% human CD45+ in NSG mice compared to NOD/SCID mice ...... 128 Figure 3.18. Graphs of individual engraftment conditions comparing the mouse EFS at time to 25% human CD45+ cells when mice were inoculated via IV and IF routes .... 130 Figure 3.19. Comparison of the EFS of NSG mice at time to 25% human CD45+ cells inoculated with IR ALL patient samples using different transplantation conditions ... 135 Figure 3.20. Comparison of the EFS at time to 25% human CD45+ cells of NOD/SCID mice inoculated with IR ALL patient samples using different transplantation conditions ...... 139 Figure 3.21. Graphs of individual engraftment conditions showing how each patient sample engrafted in NSG mice compared to NOD/SCID mice at time to leukaemia (TTL) ...... 143 Figure 3.22. Graphs of individual engraftment conditions comparing the mouse EFS at TTL when mice were inoculated via IV and IF routes ...... 145 Figure 3.23. Comparison of the EFS of NSG mice at time to leukaemia when IR ALL patient samples were inoculated using different transplantation conditions ...... 146 Figure 3.24. Comparison of the EFS at time to leukaemia of NOD/SCID mice inoculated with IR ALL patient samples using different transplantation conditions ... 147 Figure 4.1. Experimental scheme for the Main Study ...... 166 Figure 4.2. Engraftment of samples derived from the relapsed patients in the first panel ...... 170 Figure 4.3. Engraftment of samples derived from patients who maintain CR1 in the first panel ...... 172 Figure 4.4. Engraftment of samples derived from the relapsed patients in the second panel ...... 177 Figure 4.5. Engraftment of samples derived from the CR patients in the second panel ...... 180 Figure 4.6. Engraftment of samples derived from the relapsed patients in the third panel ...... 186 xiv Figure 4.7. Engraftment of samples derived from patients who maintain complete remission in the third panel ...... 188 Figure 4.8. Distribution of patients based on duration of complete remission ...... 192 Figure 4.9. Distribution of patients between the three panels based on patient clinical features ...... 195 Figure 4.10. Assessment of the difference in the engraftment profiles of patient samples between the three panels at various time points ...... 198 Figure 4.11. Compiled engraftment data of the three panels of patient samples ...... 202 Figure 4.12. Plots comparing the number of censored and uncensored mice at TT1% 204 Figure 4.13. Plots comparing the number of censored and uncensored mice at TT25% ...... 206 Figure 4.14. Plots comparing the number of censored and uncensored mice at TTL .. 208 Figure 4.15. Comparison of the EFS at TT1% between mice used to establish xenografts from patients with different outcomes ...... 210 Figure 4.16. Comparison of the EFS at TT25% between mice used to establish xenografts from patients with different outcomes ...... 211 Figure 4.17. Comparison of the EFS at TTL between mice used to establish xenografts from patients with different outcomes ...... 212 Figure 4.18. Comparison of the LGD values between mice used to establish xenografts from patients with different outcomes ...... 214 Figure 4.19. Comparison of the engraftment of IR ALL patients based on median and first mouse EFS at TT25% ...... 218 Figure 4.20. Comparison of the engraftment of IR ALL patients based on median and first mouse EFS at TTL ...... 220 Figure 4.21. Evaluation of IR ALL patient outcomes in relation to the median TT25% ...... 226 Figure 4.22. Evaluation of the relapse probability in IR ALL patients based on the LGD values at TT25% ...... 232 Figure 4.23. Evaluation of IR ALL patient outcomes based on the EFS of mice at TTL in the VXL-treated mice ...... 236 Figure 5.1. Hierarchical clustering dendrogram represents broad differences in gene expression profiles between samples of ALL-67 and primary samples of the corresponding patient ...... 252

xv Figure 5.2. Heat map showing the top 100 most differentially expressed genes between ALL-67 and the two patient samples ...... 254 Figure 5.3. Characterisation of Whiteford pediatric cancer marker gene set identified in ALL-67 xenograft ...... 259 Figure 5.4. Hierarchical clustering of samples of ALL-67 xenograft with a panel of ALL xenografts representing different ALL subtypes ...... 261 Figure 5.5. Ex vivo assessment of ALL xenografts in response to dexamethasone. .... 266 Figure 5.6. Ex vivo assessment of ALL xenografts in response to vincristine ...... 267 Figure 5.7. Ex vivo assessment of ALL xenografts response to L-asparaginase ...... 268 Figure 5.8. Comparison of ex vivo sensitivity of ALL-66 and ALL-67 cells from VXL- treated mice and cells from the non-VXL treated mice to dexamethasone...... 272 Figure 5.9. Comparison of ex vivo sensitivity of ALL-66 and ALL-67 cells from VXL- treated mice and cells from the non-VXL treated mice to vincristine ...... 273 Figure 5.10. Comparison of ex vivo sensitivity of ALL-66 and ALL-67 cells from VXL- treated mice and cells from the non-VXL treated mice to L-asparaginse ...... 274 Figure 5.11. Heat map showing the top 100 most differentially expressed genes between the VXL-treated and non-treated ALL-67 cells ...... 278 Figure 5.12. mRNA expression of HSP90 gene family in the VXL-treated versus non- treated cells of ALL-67 ...... 280 Figure 5.13. Ex-vivo assessment of the cytotoxic effect of 17 DMAG ± 10 µM of vincristine on ALL-67 VXL-treated samples ...... 283 Figure 5.14. Representation of microtubule related genes in the ALL-67 xenograft. .. 287

xvi LIST OF TABLES Table 2.1. Optimised cell densities for Alamar Blue assay ...... 69 Table 2.2. First Strand Master Mix ...... 75 Table 2.3. Second Strand Master Mix ...... 75 Table 2.4. In vitro (IVT) Master Mix ...... 75 Table 3.1. Disease characteristic features of the two pairs of IR ALL patient ...... 88 Table 3.2. Summary of the engraftment efficiency for the IR ALL xenograft panel ... 109 Table 3.3. Summary of engraftment kinetics of IR ALL panel in NSG mice at time to 1% human CD45+ cells ...... 122 Table 3.4. Summary of EFS of NOD/SCID mice inoculated with IR ALL patient samples and assessed at time to 1% human CD45+ cells...... 126 Table 3.5. Summary of EFS at TT25% of IR ALL panel in NOD/SCID mice and LGD values in response to VXL treatment ...... 136 Table 3.6. Summary of EFS at TTL of IR ALL panel in NOD/SCID mice and LGD values in response to VXL treatment ...... 148 Table 4.1. Clinical characteristics of IR ALL patient samples used for the Main Study ...... 162 Table 4.2. Summary of engraftment kinetics of the first panel of patient samples ...... 174 Table 4.3. Summary of engraftment kinetics of the second panel of patient samples .. 182 Table 4.4 Summary of engraftment kinetics of the third panel of patient samples ...... 190 Table 4.5. Analysis the impact of different thresholds points of engraftment data at the TT25% on identifying patients according to their outcomes ...... 225 Table 4.6. Analysis of the association of clinical features of the relapsed IR ALL patients with the median TT25% in the VXL-treated mice...... 228 Table 4.7. Analysis the impact of different thresholds of LGD values at the TT25% on identifying patients according to their outcomes ...... 231 Table 5.1. terms down-regulated in the ALL-67 xenograft compared to the patient sample ...... 256 Table 5.2. Gene Ontology terms up-regulated in the ALL-67 xenograft compared to the patient sample ...... 258 Table 5.3. CMap results of drug leads associated with differentially expressed genes between the VXL-treated and non-treated ALL-67 cells ...... 279

xvii Table 5.4. Gene Ontology terms up-regulated in the VXL-treated ALL-67 cells compared to non-treated cells ...... 286

xviii ABBREVIATIONS AND ACRONYMS

°C Degrees Celsius µg Microgram µL Microlitre µM Micromolar AIEOP Associazione Italiana Ematologia Oncologia Pediatrica ALL Acute lymphoblastic leukaemia AML Acute myeloid leukaemia ANZCHOG Australia and New Zealand Children’s Haematology and Oncology Group APC Allophycocyanin ASO-PCR Allele-specific oligonucleotide- Polymerase chain reaction ASP Asparaginase B-ALL B cell Acute lymphoblastic leukaemia BCL-2 B-cell lymphoma 2 BCP-ALL B-cell precursor ALL BCR Breakpoint cluster region BIM bcl-2 interacting mediator of cell death BLI Bioluminescence imaging BM Bone marrow CCG Children's Cancer Group CCI Children Cancer Institute CD Cluster of differentiation CD19 B-lymphocyte surface antigen, B4, Leu-12 CD45 Leukocyte common antigen, Ly-5 cDNA Complementary DNA CLL Chronic lymphocytic leukaemia cm2 Centimetres-squared CMap Connectivity map CNA Copy number alteration CNS Central nervous system COG Children’s Oncology Group CLPs Common lymphoid progenitors CO2 Carbon dioxide CR Complete response cRNA Complementary RNA CR1 First complete remission CR2 Second complete remission CRLF2 Cytokine receptor-like factor 2 D Diversity segment DMSO Dimethyl sulfoxide DNA Deoxyribonucleic acid DOD Died of disease E2A Enhancer binding factor E12/E47 EBF1 Early B-Cell Factor 1 EDTA Ethylenediaminetetraacetic acid EF1a Elongation factor1 alpha EFS Event-free survival

xix ES Enrichment score FAB French American British FACS Fluorescence Activated Cell Sorting FBS Foetal bovine serum FDA Food and Drug Administration FDR False discovery rate FLT-3 FMS-like tyrosine kinase-3 g G-force GCs Glucocorticoids GEMM Genetically engineered mouse model GFP Green fluorescent GRs Glucocorticoid receptors GSEA Gene Set Enrichment Analysis HLA-DR Human leukocyte antigen-D related HPC Haematopoietic progeniter cells HSCs Haematopoietic stem cells HR High risk hr Hour/s HSP90 Heat shock protein 90 IF Intrafemoral IG Immunoglobulin IKZF1 IKAROS Family Zinc Finger 1 IP Intraperitoneal IR Intermediate risk IU International unit IV Intravenous J Junctional segment JAK2 Janus kinase 2 kg Kilogram L-ASP L-asparaginase LGD Leukaemia growth delay LICs Leukaemia initiating cells Limma Linear model for microarray analysis LMPPs Lymphoid-primed multipotent progenitors LSC Leukaemia stem cell M Molar MAPs Microtubule-associated MDR Multidrug resistance mg Milligram min Minute/s mL Millilitre MLL Mixed-lineage-leukaemia mm Millimetre mM Millimolar MMLV Moloney murine leukemia virus MRD Minimal residual disease MRI Magnetic resonance imaging MRC Medical Research Council mRNA Messenger RNA NGS Next generation sequencing xx NK Natural Killer NCI National Cancer Institute NES Normalised enrichment score NOD Non-obese diabetic NSG NOD/SCID/interleukin-2 receptor gamma knockout OD Optical density PB Peripheral blood PBS Phosphate-buffered saline PCR Polymerase chain reaction PDX Patient-derived xenograft p-DFS Probability of disease-free survival PE Phycoerythrin PEG Polyethylene glycol PER Poor early responder PET Positron emission tomography Ph+ Philadelphia -positive PMC Passage-matched control POG Pediatric Oncology Group POMP Prednisone, vincristine, and 6-MP PPTP Pediatric Preclinical Testing Program PSG Penicillin / Streptomycin / L-glutamine QBSF-60 Quality Biologicals Serum-Free-60 RIN RNA integrity number RNA Ribonucleic acid RPMI Roswell Park Memorial Institute RT Room temperature SCID Severe combined immune-deficiency Sec Second SER Slow early response SNP Single nucleotide polymorphism SR Standard risk STAT Signal transducer and activator of transcription SVT Simvastatin T-ALL T-cell acute lymphoblastic leukaemia TCF3 3 TCR T-cell receptor TdT Terminal deoxynucleotidyl transferase TT1% Time to 1% human cells TT25% Time to 25% human cells TTL Time to leukaemia TFs Transcription factors TKIs Tyrosine kinase inhibitors V Variable segment VXL Vincristine / dexamethasone / L-asparaginase WBC White blood cell WHO World health organization

xxi ABSTRACT

Most childhood ALL patients are stratified at diagnosis as Intermediate Risk (IR), although an unacceptably high proportion of these patients relapse. The lack of specific prognostic features makes it difficult to predict the response of IR patients to treatment.

Recent progress in the development of patient-derived xenografts in immune-deficient mice represents an opportunity to improve relapse prediction in ALL patients.

In this study, a pilot experiment was carried out to optimise the most appropriate engraftment strategy, which could stratify engraftment of samples from IR B cell

Precursor (BCP) ALL patients according to clinical outcomes. High efficiency and quicker engraftment of IR ALL patient samples were observed in NSG mice and via the intravenous route of inoculation over NOD/SCID mice and the intrafemoral route. The response of xenografts to combination chemotherapy showed evidence of reflecting patient outcome. Validation of the capability of the optimised engraftment strategy to identify relapse in IR ALL patients using a larger patient cohort emphasised the role of drug treatment to discriminate between relapsed and non-relapsed IR ALL patients.

Certainly, the time required for ALL patient samples to reach 25% engraftment in the peripheral blood of the drug-treated mice was the most significant parameter for predicting patient outcomes.

Analysis of the ex vivo drug sensitivity of ALL xenografts revealed that the responses to vincristine and L-asparaginase reflected the in vivo responses to drug selection.

Moreover, samples from one ALL xenograft selected in vivo with drug treatment showed increased resistance to vincristine ex vivo compared with the non drug-treated samples of the same xenograft. Microarray analysis of gene expression identified

xxii HSP90 as a potential target for reversing resistance, although the HSP90 inhibitor

17DMAG failed to enhance the inhibitory effect of vincristine ex vivo. Further analysis of gene expression revealed significant up-regulation of genes involved in dynamics and/or stability of the microtubule network in vincristine-resistant samples.

In summary, engraftment of IR ALL samples under selective chemotherapy treatment could provide a clinical approach for upfront prediction of outcome in IR ALL patients, which could allow tailoring the intensity of treatment according to the risk of relapse within this patient subgroup.

xxiii

CHAPTER 1 INTRODUCTION

1 Lymphopoiesis: A general approach to the nature of lymphoid leukaemia

1.1.1 Normal development of blood cells

Blood is highly specialised tissue and is characterised by a high repopulative capacity in its ability to generate a large number of cells daily that maintain normal homeostasis.

This remarkable function requires a mechanistic process to generate all blood cell lineages with a tight dynamic balance of different blood cell types, referred to as . Blood synthesis takes place in the bone marrow (BM) of adult where mature blood cells (red and white blood cells and platelets) are derived from common ancestors called haematopoietic stem cells (HSCs) (Doulatov et al., 2012;

Orkin and Zon, 2008).

Distinctively, an HSC retains its potency for multilineage differentiation and has the capacity to maintain its reserve. During the process of blood cell formation, differentiation of stem cells gives rise to functionally distinct subsets of progenitor cells.

HSCs develop into myeloid and lymphoid progenitors that produce series of erythroid, myeloid and lymphoid cell lineages with no or restricted capability of haematopoietic reconstitution (Doulatov et al., 2012; Orkin and Zon, 2008). Lymphoid specification is believed to begin in lymphoid-primed multipotent progenitors (LMPPs) and further differentiation stimulus results in the production of common lymphoid progenitors

(CLPs), which give rise to B and T lymphocytes and natural killer cells. Although this model of haematopoiesis is used as a standerd paradigm, recent progress in the field of haematopoiesis presents some controversy about the relationship between different proginitors and their differentiation behaviour. This process primarily occurs within a

2 unique microenvironment in the BM that provides an interaction of haematopoietic cells

with blood vessels, stromal cells, growth factors and certain cytokines (Dorantes-Acosta

and Pelayo, 2012; LeBien, 2000). The B lymphocyte maturation process is described

below along with aberrations that lead to leukaemia.

1.1.2 B cell development

Based on characterisation of the B cell development process, the maturation of

lymphoid progeny is represented by complex alterations in morphological and

immunological phenotypes. The early phase of B cell development takes place in the

BM where a number of cytokines, growth factors and stromal cells regulate the

differentiation steps. A series of immature B cell precursors mediate mature B cell

formation. These precursors are identified by an overlap in changes in cell surface

markers and heavy and light chain immunoglobulin (IG) rearrangement status (Perez-

Vera et al., 2011). The early precursor that results from the commitment of the HSC

progenitors towards the B cell lineage is recognised as the pre-pro-B cell (Figure 1.1).

In the next stage, a progenitor B cell (pro-B cell) is characterised by a process of IG

heavy chain (IGH) rearrangement. The next developmental stage is represented by

the pre-B cells, marked by expression of pre-B cell receptors (Nagasawa, 2006; Perez-

Vera et al., 2011). Migration of immature B cells to peripheral blood (PB) permits

antigen-dependent maturation. After that, they undergo a conformational change and

selection process for non-auto-reactive B cells to complete their maturation in lymphoid

tissues. Thereafter, B cell activation occurs when encountering an antigenic stimulus or

specialised signals from T cells that permit differentiation of these cells into memory B

cells or plasma cells (LeBien and Tedder, 2008).

3 Large sets of proteins govern the development of non-committed progenitor cells toward B cells and may modulate B cell lymphopoiesis. The stepwise programing process of B cells is initiated by extracellular signals of lymphoid-specific transcription factors (TFs) to induce the expression of downstream genes involved in B cell maintenance. These TFs can exert a negative or positive mode of regulation on the receptors to control the expression of developmental or inhibitory genes. A major failure to maintain normal B cell development results in maturation arrest and programmed cell death (apoptosis) to prevent them from entering the . An understanding of these regulatory elements and their mode of action is crucial for distinguishing molecules that produce impairment of B cell maturation (Matthias and

Rolink, 2005).

1.1.3 Lymphocyte malignant transformation

The integrity of lymphopoiesis regulatory components is a critical determinant for the production of normal B cells. Many genetic and/or epigenetic aberrations to B cells are known to be the founders for B cell malignant transformation. Leukaemia and lymphoma are the two major classes of blood cancers that affect B lymphoid cell maturation. Leukaemia occurs due to accumulation of genetic aberrations in the primitive haematopoietic stages or early lymphoid precursor phases of maturation

(Somasundaram et al., 2015). Several can disrupt normal cell differentiation and increase blast cell proliferative capability. Many TFs are targeted by deleterious mutations that could condition B cell ontogeny. For instance, aberrations of molecules such as PU.1, E2A, PAX5, EBF1, TCF3 and IKAROS are frequently involved in the mechanism of lymphoid disorders. Disruptions in the regulated functions of these molecules can cause developmental arrest of B lineages (Mullighan et al., 2007; Perez-

Vera et al., 2011). In some cases, impairment of cell death signalling molecules [a group 4 of anti-apoptotic and pro-apoptotic molecules, such as the B cell lymphoma (BCL2)

family] and loss of cell death activators such as BIM (which alters the response) are

also involved in leukaemogenesis (Tzifi et al., 2012).

The transformed cells acquire molecular features that make them able to escape the

regulatory processes of normal lymphocyte development and to expand, carrying the

genetic and/or epigenetic alterations of the transformed cells’ particular developmental

stage (Somasundaram et al., 2015). These cells are also able to interact with the BM

microenvironment and cause the stromal niches to contribute to the pathology of the

disease by disrupting the normal haematopoietic progenitor cell (HPC) BM niches and

create abnormal microenvironments (Colmone et al., 2008).

5

SPL1 (PU.1) BCL1A PAX5 IKZF1 (IKAROS) TCF3 (E2A) FOXP1 EBF1 LEF1

Figure 1.1. Stages of B cell development showing molecules that regulate normal development. During B cell maturation, different proteins are maintained at specific levels to enhance the controlled maturation of B cells. For instance, IKZF1 (IKAROS) is expressed in undifferentiated stem cells and drives cells toward lymphoid development. Inactivation of IKZF1 leads to a differentiation block at the early stage of B cell development and complete absence of B lineage differentiation. Activation of the B cell lineage from common lymphoid progenitors (CLPs) is determined by molecules such as EBF1, and loss of EBF1 leads to a differentiation block at the pro-B stage. PAX5 is expressed in the pro-B stage and maintains lineage commitment; loss of PAX5 causes a differentiation block at the pro-B cell stage. Figure adapted with modifications from Bernt and Hunger, 2014.

6 Childhood acute lymphoblastic leukaemia: An overview

Childhood acute lymphoblastic leukaemia (ALL) is a group of clonal disorders

characterised by aggressive accumulation of transformed B cell precursors that

demonstrate diversity in morphology and genetic features with pathological

consequences. The propagation of altered cells leads to decreased production and

function of normal haematopoietic cells, as the redundant abnormal lymphoblasts

outcompete healthy cells for the normal BM environment. These cells are then able to

mobilise to the PB and infiltrate reticuloendothelial tissues – particularly in the liver,

lymph nodes and the spleen and may invade the central nervous system (CNS). Owing

to very high numbers of leukaemic cells in the BM, normal haematopoiesis is affected

which results in symptoms such as anaemia (a reduction of blood haemoglobin

concentration), leukopaenia (a reduction in leukocyte count) and thrombocytopaenia (a

reduction in platelet count). As a result, patients may suffer from progressive weakness

and fatigue, infection and bleeding (Kebriaei et al., 2002).

Demographics and aetiology of the disease

Leukaemia occurs with varying frequencies at different ages, with a much higher

incidence in children than adults. Childhood cancer is the most common cause of

disease-related deaths in children in Australia, and leukaemias account for around one

third (33%) of all cases (Cancer Council QLD, 2014) and 30% of paediatric

malignancies worldwide. It was estimated that 3114 American children and adolescents

less than 15 years old are at risk of ALL in 2015 (Cancer Facts & Figures 2015. Atlanta,

GA: American Cancer Society; 2015). Based on patient age, the pattern of incidence for

leukaemia is characterised by a peak between 2 and 5 years then the rate drops for the

remaining childhood period (Robison, 2011). Sex can change the incidence ratio for

7 leukaemia, as it is more predominant in males than females with a minor exception during infancy (Holmes et al., 2012).

Worldwide epidemiological studies of leukaemia occurrence show large geographical variation, with higher prevalence in Northern and Western Europe, North America,

Oceania and in Chinese populations in Hong Kong and Singapore, and lower incidence in Africa, Eastern Europe, Japan, much of Latin America and the Middle East. These geographical disease clusters suggest environmental-related causes of leukaemia

(Stiller, 2004). Exposure of mothers during pregnancy to physical and chemical environmental factors has been reported to be directly or indirectly involved in the cause of childhood leukaemia, with risk factors including large doses of ionising radiation, benzene intoxication, tobacco smoke, certain recreational drugs, and pesticides. Secondary leukaemia has also been seen in patients treated with chemotherapeutic agents; for example, alkylating agents and topoisomerase II binding agents such as etoposide were found to produce ALL in children characterised by MLL gene fusions (Greaves, 1997).

Diagnosis of childhood acute lymphoblastic leukaemia

The goal of leukaemia diagnosis is to ensure a precise characterisation of clinical and biological features associated with the condition to allow better treatment strategies.

Establishing a diagnosis of ALL is not a standardised process and requires a complete analysis of clinical findings and cellular features of the blood and BM. Childhood leukaemia is commonly of lymphoid origin and occurs in an acute form distinct from adult leukaemia, which is usually chronic lymphocytic leukaemia and myeloid leukaemia. Common symptoms may not specifically represent the disease and may occur either insidiously or acutely. Non-specific clinical presentation may include bone 8 pain, general fatigue, unexplained fever and bleeding or bruising. Physically, the

dissemination of leukaemia blasts is seen in the form of enlargement to different body

organs, including the lymph nodes, liver, spleen and the involvement of the CNS or

testicular organs (Kebriaei et al., 2002; Stanulla and Schrappe, 2009).

Studying of blood indices and other haematological parameters can be useful but are not

conclusive in the diagnosis of leukaemia. Initial blood counts show a variety of

abnormal findings in the leukocyte count and morphology; leukocyte numbers may

decrease or increase, with a large number of lymphoblasts and some cases may not

show any circulating blasts in the peripheral blood. Therefore, most oncology centres

correlate the findings of the PB with more than 25% of leukaemic cells in a BM aspirate

as a standard requirement to confirm a diagnosis. Examination of the aspirate also

includes assessing the complete morphological picture, immunological markers, genetic

and cytogenetic features of the blast cells (Lanzkowsky, 2011; Roganovic, 2013;

Stanulla and Schrappe, 2009). Details of some of these features will be described when

discussing the classification and prognosis of the disease below, to highlight their

importance in disease management.

The degree of leukaemia cell invasion leads to pathological abnormalities that disturb

metabolic activity. For instance, some patients may produce high levels of serum uric

acid due to the catabolic reaction of leukocyte purine, or leukaemic infiltration in the

bone may lead to hypercalcaemia and elevation of serum lactate dehydrogenase (LDH)

and phosphorus levels, which are associated with tumour cell lysis (Bain, 2007).

9 Classification of childhood acute lymphoblastic leukaemia

At one time leukaemia was only recognised as a disease with an abnormal high level of white blood cells. Since then, remarkable progress has been made in describing and managing the disease. The earliest diagnostic indicator of the disease classes was given based on the organ of disease origin, initially distinguished either as splenic or lymphatic (Freireich et al., 2014). In the second half of the 19th century, the duration of the disease was introduced in defining leukaemia cases and thereafter leukaemia could be classified as acute or chronic (Piller, 2001). The French American British (FAB) cooperative working group incorporated morphological features to formally classify

ALL cases into three categories: ALL-L1, ALL-L2 and ALL-L3 (Bennett et al., 1976).

The World Health Organization (WHO) grouping system subdivided ALL cases into categories with diagnostic relevance (McGregor et al., 2012). Traditionally, ALL stratification incorporated morphological, biological and genetic features of the disease to classify ALL patients into different classes (McGregor et al., 2012). Acute lymphoblastic leukaemias are now divided into two major categories: precursor B cell lymphoblastic leukaemia (B-ALL) and precursor T cell lymphoblastic leukaemia (T-

ALL). Acute lymphoblastic leukaemia of B cell origin is further subdivided into distinct entities defined largely by specific recurring chromosomal and genetic abnormalities

(Campo et al., 2011; Harris et al., 2000; Harrison, 2013). Although these classification approaches are used to distinguish between ALL patients, the utility of these classifications is mainly limited to the study of disease biology and remain to be integrated into clinical practice. The most efficient categorisation strategy should facilitate using appropriate treatment protocols for each disease subtype and to improve patient outcomes. Hence, factors that designate cases into subtypes based on their 10 response to treatment and risk of relapse have been established and used by many

treatment protocols.

Basic principles and evolution of chemotherapy-based treatment of

childhood acute lymphoblastic leukaemia

Treatment of ALL is considered one of the major achievements in the history of cancer

management. The first worthy attempts to treat cancer patients identified nitrogen

mustard as an effective agent against lymphoma, and the principle that cancer cells

might be more susceptible to toxins than normal tissues. This principle ultimately led to

the development of chemotherapy agents used to treat cancer in the clinic (Chabner and

Roberts, 2005). In 1948, the folic acid antagonist aminopterin was described to produce

temporary remissions in childhood ALL (Farber et al., 1948), which was replaced by a

safer folate antagonist methotrexate in the 1950's.

A number of chemotherapeutic agents–including drugs that interfere with DNA

synthesis [6-mercaptopurine (6-MP)], cyclophosphamide, daunorubicin and cytarabine;

glucocorticoids (GCs) such as cortisone; antimicrotubule agents such as vincristine, and

antimetabolites eg. L-asparaginase – were then introduced in the treatment of childhood

ALL as single agents. Initial testing of these drugs resulted in temporary remission

periods for ALL patients (Burchenal et al., 1953; Cameron et al., 1951; Hudson et al.,

2014). In 1965, Holland, Frei and Freireich revolutionised the concept of treating cancer

patients by introducing a new rationale. They hypothesised that chemotherapies should

be administrated in combination; each with a different mechanism of action, to target

cancer cells in different cell cycle phases and that this could prevent the emergence of

resistance (Freireich et al., 2014). Accordingly, they tested their hypothesis by

11 designing a treatment protocol for ALL patients, referred to as POMP (prednisone, vincristine, methotrexate and 6-MP). The study showed that the POMP regimen could induce long-term remissions in children with ALL (Frei et al., 1965). Further evidence came from St Jude Children’s Research Hospital in the U.S., which showed complete remission (CR) in 50% of ALL patients treated with various multi-agents given in combination (Pinkel et al., 1972).

The golden age in leukaemia treatment was started by these research efforts and clinical trials, which dramatically improved disease outcome, and their results are used to guide contemporary clinical practice for the management of ALL patients (Pui and Evans,

2013). It has become abundantly clear that successful treatment protocols require consideration of pharmacological and biochemical features of the drug used and the cell cycle, and the kinetics of the tumours to be treated. Therefore, greater chemotherapeutic benefits can be achieved by using a combination of multiple agents that act with additive or synergistic anti-leukaemic effects and that are given in a prolonged treatment

.

1.6.1 Treatment protocols used for acute lymphoblastic leukaemia patients

Successful treatment of most ALL patients has been established by standardising the treatment of all newly diagnosed cases with a set of backbone chemotherapies given in an efficient manner. Guidelines for treating children with ALL have been developed for different subtypes of leukaemia patients based on the results from many clinical trials that incorporated varied clinical and biological features of the disease (Alvarnas et al.,

2012). These guidelines aim to achieve CR via a strategic treatment plan that comprises multiple chemotherapeutic agents with different mechanisms of action, for a total duration of 2–3 years. Acute lymphoblastic leukaemia treatment protocols include the 12 induction of remission, consolidation and maintenance therapy (Pui and Evans, 2006).

Because, at the time of treatment, not all leukaemic cells are in the same phase of the

cell cycle, chemotherapy protocols are given in cycles. The cycle of chemotherapy is

designed to introduce an eradicative effect on proliferating cancer cells by using

multiple chemotherapy treatments over different time intervals and then allowing the

body to recover from the side effects associated with these agents (Berger, 1986).

Patients are assigned to different treatment approaches dependent on the risk of relapse

and remission period of relapsed cases. Time to relapse is an indicator used to monitor

the response of patients to applied therapy and determine the survival period with and

without clinical events (Nguyen et al., 2008). Event-free survival (EFS) is defined as the

time elapsed from starting treatment to occurrence of important events, such as relapse,

second malignancy or death (Tubergen et al., 1993b). With experience gained from

systematic consecutive trials, the Berlin-Frankfurt-Münster (BFM) study group defined

time to relapse as ‘very early’ relapse if a patient shows evidence of disease recurrence

within 18 months after initial diagnosis, ‘early relapse’ if a patient maintains CR for

about 18 months after initial diagnosis and for up to 6 months after the cessation of

frontline treatment. Patients who relapse beyond 6 months after the cessation of

frontline treatment are considered to have ‘late relapse’ (Borgmann et al., 2003; Parker

et al., 2010).

1.6.1.1 Induction of Remission

The aim of applying an induction schedule in the protocol of ALL treatment is to

eradicate leukaemic clones, therefore leading to CR of disease clones and to recover

normal BM cellularity. An induction course consists of several chemotherapeutic drugs

13 given over a period of 4–6 weeks. These include daily administration of GC drugs

(prednisone, prednisolone or dexamethasone), and weekly vincristine and L- asparaginase (Pui and Evans, 2006; Seibel, 2008). In the case of BFM trials, anthracycline such as daunorubicin is also given weekly (Conter et al., 2010).

Intensifying the induction protocols with the same drugs used in that phase has been suggested to prevent resistance to chemotherapy, due to rapid and complete clearance of leukaemia cell burden (Reiter et al., 1994; Tubergen et al., 1993a). Although an aggressive induction regimen has been credited with improving the EFS rate for patients with a high risk of relapse, it may increase morbidity and mortality in some cases due to excessive toxicity (Pui and Evans, 2006). ALL cases stratified into the standard risk group do not necessarily need intensive induction therapy, considering that those patients receive adequate post-remission intensification therapy (Pui et al., 2012) .

1.6.1.2 Post-induction therapy (Intensification)

Intensification therapy follows the induction course. This phase of therapy is essential for all patients with ALL to eradicate any residual leukaemic cells under the detection limit. There are some variations between clinical trials in actual drugs and duration of different treatment protocols. Consolidation involves administering similar drugs to those initially used during the induction regimen, and adding other potentially effective therapy to avoid drug resistance. In a study that measured outcome improvements of

ALL patients who received intensive methotrexate treatment, the methotrexate- beneficial effect was limited to intermediate- or high-risk ALL patients, but not low-risk patients (Pui et al., 2012; Stanulla and Schrappe, 2009).

14 1.6.1.3 Maintenance therapy

The concept of maintenance therapy was introduced to prevent disease recurrence by

targeting slowly dividing leukaemic cells with an extended treatment at least two years

thus preventing re-emergence of disease (Stanulla and Schrappe, 2009; Toyoda et al.,

2000). The usual regimen involves daily administration of 6-MP and weekly doses of

methotrexate. High-dose methotrexate and 6-MP are important components of

maintenance therapy reported by many study groups and their benefit is well known

(Schmiegelow et al., 2014; Toyoda et al., 2000). Intermittent pulses of a GC and

vincristine are also administered in some protocols (Arico et al., 2005; Toyoda et al.,

2000).

1.6.2 Common chemotherapeutic agents used in acute lymphoblastic

leukaemia treatment protocols

The United States food and Drug Administration (FDA) has approved a number of

therapeutic agents for treating ALL patients. Most treatment protocols use combinations

of several drugs such as vincristine, L-asparaginase, glucocorticoids, daunorubicin,

etoposide, methotrexate, 6-mercaptopurine, 6-thioguanine, cyclophosphamide and

cytarabine at different times during the treatment plan. The following describes three of

the most common chemotherapeutic agents used in the induction treatment of ALL

patients.

1.6.2.1 Glucocorticoids

As GC hormones are known to induce programmed cell death – apoptosis – acting

through the GC receptors (GRs) in lymphocytes, a number of synthetic derivatives to

the naturally occurring steroids were developed to interact with GRs and mediate cell

cycle arrest and apoptosis of sensitive leukaemic blast cells. Dexamethasone and

15 prednisone have been extensively used in the treatment of many lymphoid malignancies, including leukaemia, for more than 30 years (Tissing et al., 2003; Walsh and Avashia, 1992). These agents have shown a good efficacy and prognostic relevance in the treatment of paediatric ALL. For instance, in several early clinical trials conducted by the BFM group, ALL patients were treated in the first week with 60 mg/m2/day of prednisone and a single intrathecal dose of methotrexate, and their initial response to prednisone was prognostic of patient outcome (Moricke et al., 2009).

Indeed, measuring of GC activity showed that dexamethasone is superior to prednisone and has better penetration into the CNS, resulting in lower CNS relapse rate, and an advantage in EFS was observed in children receiving dexamethasone (Bostrom et al.,

2003) (Figure 1.2).

The Medical Research Council (MRC) in the United Kingdom carried out a trial on children with ALL by comparing the remission rate, EFS and overall survival of ALL patients upon receiving either prednisone or dexamethasone during induction and continuing phases of treatment. The EFS was significantly higher in patients who received dexamethasone and the risk of relapse was much lower in all patients treated with dexamethasone (Mitchell et al., 2010).

There is no single optimal type and dosage of GC drugs to treat ALL disease; the choice between these drugs and doses should be based on the risk of relapse, the treatment phase, and the concomitant chemotherapeutic drugs (Inaba and Pui, 2010; McNeer and

Nachman, 2010). Most ALL regimens give about 40–60 mg/m2/day of prednisone or 6–

10 mg/m2/day of dexamethasone to newly diagnosed ALL cases (Bostrom et al., 2003).

Some paediatric patients may tolerate the use of these doses but others may suffer from 16 various side effects, including suppression of the immune system that leads to increased

infection recurrence, cushingoid appearance, weight gain, adrenal insufficiency, obesity,

hypertension, gastritis, diabetes mellitus or osteoporosis (Liu et al., 2013).

Figure 1.2. The risk of isolated central nervous system (CNS) relapse over a period of seven years in ALL patients treated with steroids. The risk of isolated CNS relapse in ALL patients randomised to receive dexamethasone is lower than that for those randomised to receive prednisone (Bostrom et al., 2003).

17 1.6.2.2 Vincristine

Vincristine is an alkaloid extracted from the leaves of the periwinkle plant

Catharanthus roseus. The drug has been used to treat many type of cancers, including lymphomas, Wilms tumour, rhabdomyosarcoma, Ewings sarcoma, neuroblastoma and leukaemia (Moudi et al., 2013). Vincristine exerts cytotoxic effects by binding to tubulin dimers with high affinity at specific recognition sites of tubulin. Binding of the drug to tubulin subunits interferes with microtubule polymerisation, which is involved in the formation of mitotic spindles required for chromosomal segregation during cell division. Thus, the ability of eukaryotic cells to replicate is halted because of poorly formed mitotic spindles and subsequently damaged cells then die (Drukman and

Kavallaris, 2002; Dumontet and Sikic, 1999; Groninger et al., 2002). The standard dose given to children less than 18 years is 1.5 mg/m2 or 2 mg/m2 once per week (Estlin et al.,

2000). There are several side effects associated with the use of vincristine for paediatric patients. Vincristine is neurotoxic and can impair the function of motor neurons, resulting in hyporeflexia, constipation, paresthesias, muscle weakness and impaired motor function (Lavoie Smith et al., 2013).

1.6.2.3 L-Asparaginase

Normal human cells produce an amino acid called L-asparagine, which is a critical amino acid involved in protein synthesis. This molecule is hydrolysed by an enzyme called L-asparaginase into aspartic acid and ammonia, reducing the activity of asparagine. This forms the basis of using L-asparaginase for treatment of some cancer types including leukaemia as their cells are unable to produce enough asparagine synthetase to maintain levels of L-asparagine. This eventually leads to cessation of cellular protein synthesis and therefore inhibition of cancer cell proliferation and consequent cell death (Avramis, 2012; McCredie et al., 1973). 18 Three forms of L-asparaginase are available for cancer patients: it can be isolated from

natural sources such as E.coli or Erwina chrysanthemi or a modified form of E.coli,

which conjugates with polyethylene glycol (PEG) polymer chain (PEG-ASP) to

enhance the pharmacokinetics of the drug and avoid its uptake by the reticuloendothelial

system. A study that tested the difference between the native and modified form of L-

asparaginase as a part of induction and delayed intensification of treatment protocols for

children newly-diagnosed with standard risk ALL, showed better efficacy and rapid

decrease in leukaemia burden for patients treated with the PEG form (Avramis et al.,

2002).

The recommended dose for ALL patients less than 18 years of age is to receive 2500

IU/m2 of PEG-ASP, and 6000 IU/m2 of the natural form given every third day for eight

doses (Shinnick et al., 2013). Administration of L-asparaginase is associated with side

effects in some patients, which include anaphylaxis, pancreatitis and coagulopathy due

to inhibition of protein synthesis. A large number of patients may develop

hypersensitivity to E coli L-asparaginase and produce a high level of antibodies against

the drug, especially if they receive repeated courses during the consolidation phase

(Appel et al., 2007; Shinnick et al., 2013).

1.6.3 Treatment of recurrent acute lymphoblastic leukaemia

Treatment failure remains a major problem in the management of ALL patients.

Regression of leukaemia to undetectable level of residual disease often implies a long-

term cure. However, patients may fail to achieve remission of leukaemia cells during

treatment or may relapse within two years of CR. Unfortunately, relapse occurs in about

15 to 20% of ALL patients and the cure rate is only 30-60% for relapsed cases, since

19 they may develop resistance to chemotherapies and other treatment approaches

(Einsiedel et al., 2005; Locatelli et al., 2012; Parker et al., 2010; Rivera et al., 2005).

The treatment options for relapsed cases include chemotherapy alone if patients have a late relapse, or very good minimal residual disease (MRD) response to induction therapy, which is one of the main prognostic indicators of overall outcome. ALL cases with very early relapses or poor MRD response that are incurable by chemotherapy are prioritised for chemotherapy and allogeneic HSC transplantation. A few ALL cases may encounter relapse in the CNS or testes with undetectable leukaemia cells in BM; such cases are treated with systemic therapy and radiotherapy directed to the site of relapse

(Locatelli et al., 2012; Parker et al., 2010).

Prognosis of childhood acute lymphoblastic leukaemia

The current picture of childhood ALL outcomes emphasises the need to increase efforts to improve outcomes of other cancer patients. This picture is based on optimum administration of chemotherapy, accurate risk categorisation and improved supportive care of ALL patients being responsible for successful treatment in more than 90% of children with ALL, from the first induction time and though maintaining long-term remission in 85% of these children (Curran and Stock, 2015; Pui and Evans, 2013).

There are certain prognostic factors commonly used to stratify ALL patients into groups according to their risk of relapse; some factors vary among different treatment protocols and others may only identify a few patients with poor prognosis. These criteria are regularly reviewed and updated to improve the outcome of ALL patients. Alongside the importance of these factors in disease prognosis, they can also guide research work that

20 improves our understanding of the disease heterogeneity. The great challenge for

clinical researchers is to determine how to best incorporate the prognostic parameters

into clinically-applicable measures (Friedmann and Weinstein, 2000; Schultz et al.,

2007). Therefore, modern risk stratification schemes use concise parameter strategies to

stratify patients for different treatment protocols. The most common prognostic factors

and their roles in disease prognosis are detailed below.

1.7.1 Clinical factors

White blood cell count (WCC), age at the time of initial diagnosis and sex were

identified in early clinical trials to have adverse impacts on the outcome of ALL patients

with B-lineage ALL regardless of the treatment regimen used. High WCC counts

(hyperleukocytosis) in ALL samples can reduce the uptake and the apoptotic effect of

chemotherapy and influence disease prognosis (Eguiguren et al., 1992; Kobayashi et al.,

1998). A 50 x 109 cell/L is the usual cut-off that determines the outcome favourability

for ALL patients. Children who have WCC >50 x 109 cell/L are stratified as ‘high risk

of relapse’ in most treatment protocols. Presenting with a lower WCC usually stratifies

patients into ‘low risk of relapse’ (Assumpção et al., 2013; Teachey and Hunger, 2013).

Age is considered a major adverse factor in the clinical outcome of ALL cases. For

instance, infants <12 months of age are relatively resistant to GCs and different

treatment protocols. Patients who are 1.5–10 years of age tend to have a good prognosis

while, compared with younger children, ALL patients older than 10 years of age have a

higher risk of relapse (Pieters et al., 1998; Teachey and Hunger, 2013). The correlation

between age and prognosis could be explained by the predominance of common genetic

abnormalities that associate with the outcome in each age group. The effect of age on

prognosis of ALL children is decreased by intensive treatment regimens used in some 21 treatment protocols (Harrison, 2013). Sex was also been shown to influence the risk of relapse. For instance, early UKALL clinical trials conducted by the Medical Research

Council on 4,000 children with ALL (1972-1990) showed that the prognosis for girls is significantly better than boys (Chessells et al., 1995).

The National Cancer Institute (NCI) published uniform criteria to stratify ALL children aged >10 years or with a diagnostic WCC >50 x 109/L as patients at high risk of relapse

(Smith et al., 1996). Many treatment protocols use this set of well-recognised prognostic features, although it is not used by the BFM or Australian and New Zealand Children’s

Haematology/Oncology Group (ANZCHOG) protocols (Conter et al., 2010; Marshall et al., 2003). However, additional criteria are now considered by modern treatment protocols (Pui and Evans, 2013). The Pediatric Oncology Group (POG) and Children's

Cancer Group (CCG) introduced additional prognostic factors to assign ALL patients based on strong prognostic indicators that have been shown to influence patient outcomes (e.g., ploidy, blast karyotype, and early morphologic response) (Schultz et al.,

2007; Seibel, 2008).

1.7.2 Immunophenotype

Immunophenotype of ALL cells is an important investigation measure in the initial workup of the disease. Multiparametric flow cytometry allows the identification of leukaemia based on a combination of monoclonal antibodies to cell membrane or cytoplasmic proteins. Leukaemia cells display a pattern of antigen markers that represent the cell type and lineage at which the cell became malignant. Therefore, these antigens can be used to determine ALL subtypes as B-ALL or T-ALL. Leukaemias of B cell origin can be subdivided into B cell precursor (BCP) ALL and mature B cell ALL.

22 The BCP-ALL subtype represents 80–85% of leukaemias and includes pro-B-ALL,

common B-ALL and pre-B-ALL (Sędek et al., 2012). The pro-B-ALL blast is

characterised by expression of CD19, cytoplasmic CD79a, CD22, CD10, nuclear TdT

and HLA-DR antigens but lack of surface immunoglobulin expression. Pre-B-ALL is a

more differentiated form of leukaemia that occurs in 20% of patients and is

characterised by expression of cytoplasmic µ chains and absence of surface antigens.

The expression of CD10 antigen identifies ALL blasts known as “common ALL” (Pui

et al., 1993; Szczepański et al., 2006). Heterogeneity in antigen marker expression is

also well known in B-ALL disease. Some blasts also co-express myeloid CD33 and

CD13 as aberrant markers (Szczepański et al., 2006). Many reports have shown

instability of markers such as CD10, CD34, CD19 and CD20 under the early phases of

treatment (Dworzak et al., 2008; Gaipa et al., 2008).

Stratification of ALL patients according to the immunological subtypes was initially

found to have a good correlation with the response to treatment. For instance, early

studies indicated that the pre-B ALL is a risk factor for poor prognosis in ALL patients.

However, integration of the immunophenotype features with cytogenetic markers

revealed that the less favourable response to therapy was due to the presence of t(1;19)

chromosomal translocation in some cases of pre-B-ALL. A more aggressive treatment

for those patients led to improve their prognosis (Raimondi et al., 1990). Similarly, the

prognostic significance of CD10 expression was diminished by improvement in

chemotherapy treatments (Behm, 2003). A subsequent study conducted by Kaspers et al

(2005) tested the in vitro sensitivity of 207 relapsed ALL samples presented with

different lineages of B cells (pro-B, common and pre-B-ALL) to 19 chemotherapeutic

drugs widely used in treating leukaemia patients. The results of that study showed no

23 significant difference among patients with diverse B lineages to the 19 drugs (Kaspers et al., 2005).

1.7.3 Cytogenetic features and genetic alterations

Several common cytogenetic abnormalities of leukaemic blasts have been linked to development of ALL disease and therefore may have a direct or indirect association with disease prognosis (Kempski et al., 1999; Moorman et al., 2014). Structural and numerical alterations of that associate with a relatively favourable prognosis in B-ALL include hyperdiploidy state, especially trisomies 4, 10 or 17 and t(12;21)/[TEL-AML1](ETV6-RUNX1) translocation (Heerema et al., 2000; McLean et al., 1996; Sutcliffe et al., 2005). In contrast, an inferior outcome is seen with the presence of MLL rearrangements, t(9;22)/[BCR-ABL1] and hypodiploidy (<45 chromosomes). Pre-B-ALL with the t(1;19)(q23;p13) translocation and expression of the TCF3-PBX1 fusion was identified as a high-risk entity in trials conducted by St Jude

Children’s Research Hospital. (Jeha et al., 2009), although these cytogenetic factors are not considered by some study groups. This could be attributed to the ability of their contemporary treatment regimens to improve the outcome for patients with such cytogenetic abnormalities (Pui et al., 2012).

A number of genetic alterations that accompany, and may cooperate with, many chromosomal translocations have been shown to correlate with inferior prognosis of

ALL patients (Moorman et al., 2014; Schwab et al., 2013; Virely et al., 2010). Genetic aberrations of high risk of relapse patients include IKZF1 deletions, CDKN2A/B deletions, Janus kinases 2 (JAK2) mutations, and cytokine receptor-like factor 2

(CRLF2) rearrangements (Mullighan et al., 2009; Schwab et al., 2013). Patients who harbour such genetic alterations are categorised into different genetic subtypes and 24 treated using risk-adapted therapy in most contemporary treatment protocols. The

frequencies of cytogenetic and genetic alterations are different among ALL patients and

some aberrations may only be found in small populations of ALL patients. Figure 1.3

shows the distribution of genetic abnormalities among paediatric ALL patients.

25

Figure 1.3. Frequency of genetic abnormalities in childhood ALL. The majority of ALL cases are of B cell origin (shown in the blue and yellow segments of the chart). The most common cytogenetic abnormalities present in those patients are the ETV6- RUNX1 and hyperdiploid status. About 10% of B-ALL cases present with unclassified B-ALL and 6% of cases have MLL translocations. Hypodiploid chromosomes, TCF3- PBX1 and some submicroscopic genetic alterations were each found in less than 5% of B-ALL cases. The yellow segments indicate the few cases of B-ALL targeted by newly- discovered genetic alterations known to associate with poor prognosis. The genetic lesions that exclusively associated with T-ALL are shown in red. Figure adapted from Mullighan, 2012.

26 1.7.4 Response to initial treatment: Early studies

Early systematic clinical trials identified that the rate of leukaemia blasts clearance from

BM and PB in the early stages of treatment is highly prognostic and reflects

chemosensitivity of the disease (Lilleyman, 1998; Miller et al., 1989). Persistence of

leukaemia in PB and/or BM after 7 and/or 14 days of chemotherapy and after 30–35

days at the end of induction studies were confirmed to be highly predictive for ALL

patient outcome (Gaynon et al., 1997).

The BFM group considered early response of ALL patients to prednisone as a good

prognostic criterion, especially if the reduction in peripheral blast count is below 1000

blasts/µL after seven days with prednisone treatment, whereas a higher blast number is

associated with poor outcome (Moricke et al., 2009; Schrappe et al., 2000). In addition,

in a study that investigated the percentage of leukaemia blasts before and after one week

of induction with multichemotherapy agents in ALL patients enrolled in the St Jude

Children’s Research Hospital Total Therapy Study XI, persistence of leukaemia cells

after one week of induction with multiagent chemotherapy proved to be the most

significant adverse feature in patient outcomes (Gajjar et al., 1995).

The in vitro testing of leukaemic cells obtained at the time of diagnosis can also predict

the sensitivity of leukaemic cells to chemotherapies. Various studies, which used human

patient biopsy, have ascertained the prognostic significance of the in vitro response of

patients who had diverse treatment outcomes to these agents (Den Boer et al., 2003;

Frost et al., 2003; Styczynski et al., 2012). In a retrospective study, Kaspers et al. (1997)

reported the in vitro response of 152 samples from newly-diagnosed ALL patients to 12

27 chemotherapies including prednisone, vincristine and L-asparaginase. In the same year,

Hongo et al. (1997) published similar data on samples from 209 children with ALL, using a panel of 16 drugs. Both studies found that a combination of data for prednisolone, L-asparaginase and vincristine provide a drug-resistance profile with prognostic significance for probability of disease-free survival (p-DFS) (Hongo et al.,

1997; Kaspers et al., 1997).

1.7.5 Response to initial treatment: Monitoring of minimal residual disease

Traditionally, assessment of treatment response has relied on morphological detection of 5% blasts in the BM as a cut-off for defining remission status (Bjorklund et al.,

2003). However, those techniques do not allow sensitive counting of leukaemic cells and estimation of leukaemic burden in the body. Without minimal residual disease measurements, patients with high leukaemic burden at end of induction are treated on the same regimen as those with very low-level leukaemia or, perhaps, with no leukaemia at all (Campana and Pui, 1995).

Monitoring of minimal residual disease (MRD) levels after chemotherapy induction and during the course of the disease in BM samples has been shown to be the major prognostic factor in prediction of relapse in ALL patients (van Dongen et al., 1998).

Different time points and detection thresholds for monitoring MRD levels are now used by contemporary treatment protocols to distinguish between patients with a high or low risk of relapse (Borowitz et al., 2008; Bruggemann et al., 2010). Various methods used to monitor MRD levels involve techniques that distinguish disease-specific immunological or molecular markers from their normal counterparts. Although the immunophenotypes of most leukaemic cells are similar to those of their normal

28 counterparts, some leukaemia may under- or over-express molecules during the

leukaemogenesis process and others may present with a translocation that make genes

encoding chimeric proteins specific for leukaemia (Campana and Coustan-Smith, 1999).

The gold standard technique for monitoring of leukaemia regression relies on

quantifying residual disease using quantitative PCR based on an allele-specific

oligonucleotide approach (ASO-PCR). This technique detects the unique junctional

sequences of the immunoglobulin (IG) and T cell receptor genes (TCR), created through

rearrangement and joining of the multiple variable (V), diversity (D) and joining (J)

segments of these genes. These represent a highly specific sequences for PCR targeting,

which reflects the amount and homogeneity of disease clones present during the course

of the disease (Langerak et al., 2012). A recent study suggests that sensitive of MRD

detection can also be achieved using high-throughput deep sequencing technology,

which would enable monitoring of ALL patient response to their treatments (Faham et

al., 2012).

Risk-adapted therapy protocols

The importance of determining prognostic criteria for ALL patients is mainly to stratify

patients into different risk categories, to predict their outcome and determine a range of

treatment approaches for each subgroup accordingly. The aim is to intensify the

treatment protocols for high-risk patients and prevent lower-risk relapse cases from

receiving intensive doses of chemotherapy, while maintaining long-term survival (Jeha

and Pui, 2009). Many risk-adapted therapy protocols stratify ALL patients according to

their risk of relapse into three or four major categories: standard risk (SR), intermediate

risk (IR), high risk (HR) and very high risk (VHR) based on monitoring of molecular,

29 genetic features and clinical response of ALL patients to treatment (Bassan and Hoelzer,

2011; Schultz et al., 2007). This stratification system defined patients according to the risk of relapse and therefore most of B-ALL patients with good prognosis are stratified into the standard and intermediate categories whereas fewer patients with poor prognosis are stratified into the higher risk categories.

1.8.1 Intermediate-risk BCP-acute lymphoblastic leukaemia group as a heterogeneous class of the disease

The intermediate-risk ALL (IR-ALL) subset represents the largest group, and is clinically and biologically the most heterogeneous risk group of children with ALL

(Conter et al., 2010). Intermediate-risk ALL cases comprise more than 50% of ALL patients and the majority of patients present with intermediate NCI risk features (WCC and age), favourable cytogenetic markers, chromosomal diploid status and MRD values that suggest a remission period greater than that for high risk ALL cases (Biondi and

Cazzaniga, 2013; Chiaretti et al., 2014). However, 20–25% of patients may relapse at any stage after treatment (Biondi et al., 2000). A major clinical challenge is presented by the current inability to readily identify indolent from aggressive IR ALL patients who present with favourable prognostic factors. The absence of information to guide treatment options can lead to a significant “overtreatment” of patients who would otherwise require only conventional management (Biondi et al., 2000; Conter et al.,

2007). Thus, the impact of toxicity from overtreatment with chemotherapies can reduce the benefits of survival for those who require minimal treatment intervention.

Based on earlier trials, several MRD clinical trails including the Associazione Italiana

Ematologia Oncologia Pediatrica SG and BFM SG (AIEOP-BFM-ALL-2000) and

30 ANZCHOG ALL Study VIII– have identified high risk patients by measuring MRD

levels at day 79 after treatment initiation. However, MRD at day 79 in these studies was

not able to predict relapse in 20% of IR ALL patients (Conter et al., 2010). Based on

MRD levels at day 33, the AIEOP-BFM-ALL 2000 study group was able to identify a

subset of IR ALL patients with slow early response (SER) to therapy and treat them

with a more intensive treatment protocol than the other IR ALL patients in the new

AIEOP-BFM-2009 trial. Retrospective analysis of the prognostic values of measuring

MRD at day 15 and day 33 together in BCP-ALL patients who enrolled in the

ANZCHOG ALL Study VIII trial has also identified a subgroup of IR ALL patients

who presented with higher levels of MRD at day 15 and at day 33 and lower five-year,

relapse-free survival compared with other IR-ALL patients. This subset of patients was

called the poor early responder (PER) as those patients maintained the high level of

MRD at both time points (day 15 and 33) (Karsa et al., 2013).

Few reports have identified biomarkers that predict poor prognosis in IR-ALL patients.

For instance, Krentz et al. (2013) showed that certain genetic alterations (IKZF1

deletion and TP53 alteration) have a very adverse impact on survival of ALL patients

who initially presented with low MRD levels and were stratified into the IR-ALL

subtype (Krentz et al., 2013). In addition, the poor prognostic impact of CRLF2 over-

expression and P2RY8-CRLF2 has been identified in a group of BCP-ALL patients

stratified into the IR category (Harrison, 2013). The prognostic benefits of these features

were limited to a small group of IR ALL patients who presented with such genetic

alterations (Palmi et al., 2012). These data highlight the clinical heterogeneity among IR

ALL patients and suggest expanding research efforts beyond the HR group to improve

outcomes of patients who may not show signs of aggressive disease, yet their clinical

31 response is dismal. This challenge could be addressed by better understanding the molecular basis of clinical heterogeneity, which should eventually lead to identifying patients at high risk of relapse for direct treatment interventions.

Understanding the biology of acute lymphoblastic leukaemia relapse

Functional characterisation of leukaemia progression has been described over twenty years. In spite of scientific efforts devoted to improve treatment protocols, current treatment strategies apparently do not protect child patients who show signs of highly chemoresistant leukaemic cells (Locatelli et al., 2012). Improved survival of ALL patients demands earlier identification of patients at high risk of relapse, which then could help to tailor therapies to the unique biology of these ALL cases (Raetz and

Bhatla, 2012).

Treatment failure in cancer – including leukaemia – is thought to be inherited or acquired due to creation of resistance to chemotherapies during an induction regimen

(Gorlick et al., 1996; Kerbel et al., 1994). Most relapses occur due to the failure of chemotherapies to completely eliminate leukaemic clones during treatment courses

(Locatelli et al., 2012). Resistance to chemotherapeutic agents, which influences prognosis of ALL patients, has been characterised in the hope of understanding the pattern of resistance and preventing its development.

Resistance to GC in ALL is one of the major challenges observed in the clinical setting.

Mechanisms associated with this resistance include inability of the GC to bind its receptor, improper translocation of the subsequent signals to the nucleus and defects in

32 the apoptosis pathways, especially in the BCL2 family members such as BIM

(Bachmann et al., 2007; Kofler et al., 2003). Another mechanism by which leukaemia

cells survive GC and other agent effects is via the multidrug resistance (MDR) pathway,

which acts on the kinetics of transmembrane transporter proteins to alter the influx or

efflux of proteins and drugs. Many MDR proteins are currently linked to cancer

treatment outcomes, including in paediatric ALL. For instance, the cellular drug efflux

pump known as P-glycoprotein (ABCD1; MDR1) modulates the efflux for drugs such

as GC, vincristine and mitoxantrone that reduce intracellular drug accumulation

(Hodges et al., 2011; Tissing et al., 2003; Wu et al., 2008).

Treatment failure due to primary or secondary development of resistance to vincristine

is also a major clinical challenge to the treatment of ALL. Insensitivity of cancer cells to

vincristine may depend on the ability of the drug to reach its target, mechanisms that

reduce the ability of the drug to interact with its cellular targets or affect the response of

cancer cells to the executioner signals induced by vincristine (Verrills and Kavallaris,

2005). For instance, the alteration of the tubulin/microtubule system due to mutations in

tubulin, drug-binding regions or change in expression levels of genes encoding alpha

and beta tubulin isotypes and microtubule-associated proteins (MAPs) are also

responsible for decreasing the sensitivity to vincristine in a range of cancer types.

Modifications in events downstream of the drug target have also been reported to

decrease sensitivity to vincristine. As vincristine and other drugs mediate cell death via

apoptosis, defective apoptotic signals can interfere with the signal mediating cell death

(Dumontet and Jordan, 2010; Kavallaris, 2010). Figure 1.4 explains mechanisms

responsible for resistance to antimicrotubule drugs.

33

Figure 1.4. Schematic diagram illustrates the mechanisms associated with antimicrotubule drug resistance. Figure adapted from Verrills and Kavallaris, 2005. The proposed mechanisms of antimicrotubule drug resistance include inability of microtubule targeting agents (MTAs) to reach their cellular target due efflux of the drug out of the cell by drug transport proteins, inability of MTAs to inhibit microtubule dynamics due to structural and/or functional changes in the tubulin/microtubule system (tubulin alter drug binding, change in expression of tubulin isotypes and MAPs alter drug stability). Another possible mechanism by which MTAs fail to inhibit survival of cells is due to alterations in cell death pathways, which cause lack of response to death signals. Mt, microtubules; MAPs, microtubule associated proteins; MF, microfilaments.

34 The source of resistance to L-asparaginase is not fully understood and different

mechanisms are proposed in diminishing the activity of this drug. For instance, L-

asparaginase is known to produce its anti-leukaemic effect through inhibiting protein

synthesis via consumption of L- Asparagine. Resistance to L-asparaginase is also linked

to accumulation of asparagine and increased uptake of glutamine/aspartic acid due to

changes in expression of genes that encode transmembrane transporter proteins

(Aslanian and Kilberg, 2001). Deprivation of asparagine may change the intracellular

amino acid balance due to interference with the activity of proteins regulating the

translation process such as protein translation repressors, which reduce the activity of

ribosomal proteins and translation initiation factors. Imbalanced protein synthesis may

also protect cells against cell death (Pui, 2006). Genetic variations in genes that affect

the pharmacokinetics and pharmacodynamics of drugs have also been reported to have a

plausible influence on the effectiveness of treatment. For instance, polymorphisms in

the NR3C1 gene are reported to be associate with a reduction in survival probability in

children with ALL (Labuda et al., 2010; Fleury et al., 2004).

The main limitation with many approaches used to understand the mode of resistance to

chemotherapies that mediate recurrence of the disease is that they investigate

mechanisms that maintain the relapse without taking into account that relapse in

leukaemia could not share the same mechanisms identified at time of diagnosis, and

may involve more than one mechanism for maintaining the disease. In addition, it has

been also known for a long period that leukaemia is a heterogeneous disease. The multi-

clonal compositions may form the basis for clonal selection, evolution and resistance to

treatments (Brown et al., 2009; Jan and Majeti, 2013). However, less attention had been

paid to the clinical consequences of this phenomenon in leukaemia research conducted

35 before the last two decades. Currently, with the advent in high-throughput “omics” technologies that enable functional characterisation of cells at clonal level, interest has been renewed in studying the role of heterogeneity in the mechanisms underlying cancer cell relapse (Dick, 2008).

1.9.1 Leukaemia heterogeneity: Evidences of concepts

Based on the data obtained from primary patient samples and animal models, leukaemia cell populations exhibit marked variability on their cell of origin, biological and phenotypic features and cellular proliferative capability. Leukaemia heterogeneity is even seen within leukaemic clones originated from the same initiating cells (Dick,

2008).

Studies investigating the nature and frequency of leukaemia initiating cells (LICs) represent ongoing controversy about the functional importance of certain leukaemia populations in development or progression of ALL disease. Various ALL subgroups were thought to initiate at unique haematopoietic maturation stages, with each exhibiting different proliferation potencies. Earlier reports have described that ALL cells preserve the LICs in the immature CD34+/CD38- or CD34+/CD19- stage

(Cobaleda et al., 2000; Cox et al., 2004). The LICs are also found to reside in the more mature CD19+/CD34+/CD38dim populations in patients with good risk ALL (Castor et al., 2005; Hong et al., 2008). Subsequently, a study produced by Cox et al. (2009) showed activity of LICs at the unique immature compartment CD133+/CD19− in ALL cells (Cox et al., 2009). It is notable that studies indicating an immature phase of development as the origin of cell expansion used samples from patients with HR ALL whereas the good risk patient samples showed LIC activity in the mature phase.

36 However, further studies have implicated both mature and immature stages of

maturation in the initiation of the disease (le Viseur et al., 2008). This suggests that the

site of LICs could depend on the cellular stage at which the mutation arrest occurred

and LIC activity could rely on conditions that confer selective advantage between

multiple competing leukaemia clones.

Single-cell analysis has shown the existence of variation in genetic abnormalities within

leukemic cell populations. Tracing of clonal composition in a defined group of patients

based on DNA copy number variation profiling revealed a multi-clonal landscape, with

different clones possessing large or small copy number alterations (CNA) that

frequently involve tumour suppressor genes, oncogenes and lymphocyte regulatory

genes (Anderson et al., 2011; Mullighan et al., 2008). Further evidence underlying the

clonal heterogeneity was presented from identical twins; molecular profiling of

leukaemia cell population from twins who harboured an identical translocation

abnormality showed similar in utero pre-leukemic clones that subsequently developed

overt leukaemia in both twins with different CNA profiles (Bateman et al., 2010).

1.9.2 Current concepts about the biology of relapse in acute lymphoblastic

leukaemia disease

The observations that the clonal compositions of ALL disease influence the response of

ALL patients to chemotherapies highlighted the critical importance of clonal

heterogeneity in the biology of ALL disease. Therefore, studying the clonal populations

can be especially important in distinguishing treatment-resistant cell populations from

those that are sensitive (Choi et al., 2007).

37 The role of clonal heterogeneity in ALL disease progression has been explained by the following concepts. Leukaemia stem cell (LSC) theory proposes that only a subpopulation of leukaemia blasts have unique features including long-term stemness, immortality, ability to differentiate and insensitivity to chemotherapies that resemble normal stem cells, and that the other dominant non- cancer cells have heterogeneous genotypes and phenotypes (Lane and Gilliland, 2010; Lobo et al., 2007). This theory emphasises the hierarchical order of events that occurs during disease progression, as it proposes that premalignant cells acquire an oncogenic change followed by genetic or epigenetic aberrations, which allow transformation to give rise to complete disease

(Bomken et al., 2010; Shackleton et al., 2009). Treatments that follow this concept impose targeting LSCs as a source of relapse and therefore disease recurrence can only occur through escaped LSCs (Huntly and Gilliland, 2005).

Alternatively, a non-deterministic model proposes that most or all leukaemia cells are able to self-renew to initiate the disease. Therefore relapse in ALL in this model could arise due to selective pressure derived from chemotherapies on diagnosis clones, which leads to selection of a minor treatment-resistant sub-clone or leads to genetic evolution in predominant cell populations. Relapse could also arise due to the persistence of the dominant cell populations at initial presentation or acquisition of mutations in the preleukaemic cells that leads to re-emergence of the disease (Jan and Majeti, 2013).

More details about the proposed models to explain relapse in ALL patients are depicted in Figure 1.5.

38

Figure 1.5. Models of clonal progression in acute leukaemia. Clonal heterogeneity could present in the pre-leukemic stage, which could then contribute to development of an overt leukaemia disease with multiple clones. Relapse could arise due to: (A) persistence of the dominant cell populations at initial presentation after treatment; (B) selection of a minor treatment-resistant subclone through the course of therapy; (C) genetic evolution in predominant cell populations induced by treatment with DNA- damaging therapeutic agents; or (D) acquisition of additional mutations due to damage induced by DNA-damaging therapeutic agents on pre-leukemic cells, which results in a novel relapsed clone. Figure adapted from Jan and Majeti, 2013.

39 The observations that some leukaemic cells are more resistant to chemotherapeutic drugs at time of relapse compared with cells at time of diagnosis in some patients urged defining the relationship between the diagnosis and relapsed genetic profiles (Kawamata et al., 2009; Klumper et al., 1995). Indeed, most findings of clonal heterogeneity studies in ALL disease support the notion that dominant cells at time of relapse mostly descend from minor cell populations at diagnosis and various biological features confer leukaemia selection at a clonal level (Jan and Majeti, 2013). A detailed analysis of clonal markers of ALL patients sampled at time of diagnosis and relapse based on MRD markers has identified selection of pre-existing drug-resistant sub-clones. Choi et al.

(2007) have studied the clonal markers of 20 patient samples using IG gene rearrangements. The initial observation drawn from their work was that 13 ALL patient samples tested in their study developed new clonal populations at time of relapse.

Backtracking of the relapse clones in diagnostic samples using highly sensitive RQ-

PCR assay then identified that the relapse clones were actually present at diagnosis in 8 patient samples. Interestingly, the quantity and relapse clonal markers predicted relapse in these patients (Choi et al., 2007).

Researchers who employ high-resolution genome-wide analysis of DNA copy number variation based on single nucleotide polymorphisms (SNPs) and next generation sequencing (NGS) have frequently reported changes in genetic composition at relapse, and these changes can be tracked in samples collected at time of diagnosis. Mullighan et al. (2008), for example, analysed CNA of paired samples from paediatric ALL at time of diagnosis and relapse to determine how relapse clones are genetically altered compared with the disease at diagnosis. This elegant study showed that the clonal constitution of some relapsed cases was either similar to the diagnosis clones or

40 represented cell populations evolved from diagnostic clones. Interestingly, the majority

of relapsed cases showed different patterns of CNAs and shared only a few genetic

markers with the predominant diagnostic clones. The analyses also identified that

relapse cases have a significant increase in the mean number of CNAs compared with

diagnosis cases (10.8 at diagnosis versus 14.0 at relapse). In addition, most of CNA

target genes were involved in cell cycle regulation and B cell development, such

CDKN2A/B, IKZF1, RAG, PAX5 and EBF1 (Mullighan et al., 2008). In agreement with

this study, Ma et al. (2015) have also shown that deep whole-exome sequencing of

samples from B-ALL patients at time of relapse can identify clones with mutations that

were already present in diagnostic samples, albeit as minor populations. Their study

also found that some samples have novel clones established at time of relapse with

some mutations frequently observed in relapsed patients (Ma et al., 2015).

Assessment of clonal markers based on immunoglobulin (IG) and T-cell receptor (TCR)

gene rearrangement and genetic composition of leukaemia cells at different time points

of disease progression indicate that clonal heterogeneity has profound clinical

implications on prognosis of ALL. Evidence presented from previous studies indicates

that relapse-driving leukemic cells could arise due to expansion of genetically evolved

cell populations in a considerable number of ALL patients. Mutations that were

frequently enriched at the time of relapse were involved in cell cycle progression, anti-

apoptosis and DNA repair pathways, suggesting a role in selective advantage of these

genes. This indicates that these molecules may mediate the resistance by increasing the

proliferative capacity and leukaemia blast survival that render cells to be resistant to

common chemotherapies. Overall, translating these data into the clinical realm requires

41 an early detection of mutations driving relapse and applying an alternative treatment approach to avoid mutation selection (Ma et al., 2015; Mullighan et al., 2008).

Personalised therapy: Towards “Omics”-oriented treatment of childhood acute lymphoblastic leukaemia

Oncologists hope to improve current treatment protocols and integrate prognostic factors that impose clinical variations between patients to enhance the precision of treatments and utilise more efficient methods to stratify ALL patients into different risk categories (Nersting et al., 2011). The basic approach behind personalised cancer treatment is that achieving good efficacy and safety dictates using the right treatment plan for the right patients. Currently, patients stratified into the non-HR ALL categories receive similar treatment despite the heterogeneity in biological features and clinical outcomes of their disease. Indeed, not all cases need harsh treatment protocols to achieve maintained complete response of the disease; on the other hand, undertreating of aggressive disease is associated with very poor prognosis. Therefore, patients should receive tailored intensity of chemotherapy approaches according to their risk of disease recurrence (Pui et al., 2012; Raetz and Bhatla, 2012; Teachey and Hunger, 2013).

Another strategy to improve the survival rate in ALL patients is to incorporate novel agents into the front-line treatments. A good candidate drug should be highly selective and interfere with molecular aberrations involved in cancer growth and progression.

Unlike chemotherapeutic agents, which inhibit all rapidly dividing cells, good targeted drug candidates should have less damage to normal cells and organ function. It has to accommodate that cancer cells may not present with a constant mutation rate and genetically stable cancers might acquire a mutator phenotype during tumour

42 progression. A further characteristic feature of a good target molecule is that the

aberration to be targeted is preferred to occur as an early event during disease

progression, which implies that this aberrant feature presents in all cancer cells. Target

molecules should not be highly overexpressed in normal cellular compartments, nor

interfere with the effect of conventional chemotherapies and researchers should

accurately measure the response using a non-biased strategy (Lee-Sherick et al., 2010;

Turner and Reis-Filho, 2012).

The advent of the genome, transcriptome and proteome era is promising to help in

stratifying ALL cases into clinically-related subgroups, tailoring the therapy based on

the characteristic features of the individual patient, directing use of current antitumour

strategies or developing of new treatment strategies. Many research groups – including

ours – have classified ALL cases into disease-relevant classes based on the genetic

profiles of ALL patients. This strategy has helped to classify ALL disease into broadly

B cell and T cell ALL and mixed lineage leukaemia (MLL) subtypes. Armstrong and

his colleagues proved that MLL has genetic phenotypes distinct from ALL and acute

myeloid leukaemia (AML) (Armstrong et al., 2003). A novel subgroup called

Philadelphia-like ALL occurs in approximately 10–12% of BCP-ALL cases with a

genetic signature similar to that of Ph-positive ALL but this group does not harbour a

translocation between the chromosomes that carry the BCR and ABL1 genes (Den Boer

et al., 2009; Mullighan et al., 2009). A group of ALL patients has been identified to

have CRLF2 gene rearrangements. Leukaemia cells of those patients are characterised

by high expression of the CRLF2 gene, which enhances signalling through oncogenic

pathways and by the presence of mutations in genes encoding Ikaros (IKZF1) and JAK

proteins that leads to constitutive activation of the JAK-STAT pathway (Tasian et al.,

43 2012). The many emerging ALL subtypes supports the concept that not all cases should be treated similarly and perhaps targeting the unique disease biology could offer an alternative approach toward disease treatment and prognosis

A number of therapies are under development to target molecular pathways involved in leukaemia progression with less toxicity to normal cells. Many tyrosine kinase inhibitors (TKIs) have been developed to target leukaemia blasts with mutations in their protein kinase signalling cascades. For instance, Imatinb is the first kinase inhibitor approved to be used for treating ALL patients with a Ph-positive chromosome. Several studies have shown the activity of JAK inhibitors against leukaemia cells overexpressing CRLF2 genes (Loh et al., 2015; Tasian et al., 2012). Novel therapies – such as molecules that inhibit BCL2 protein – are promising to enhance the cell death process in leukaemia cells that exhibit upregulation of the antiapoptotic BCL2 protein family. ABT-263 is an enhanced version of the BH3 mimetic BCL2 inhibitors that demonstrated the ability to improve the treatment of relapsed chronic lymphocytic leukaemia and is effective in xenografted ALL (Roberts et al., 2011; Suryani et al.,

2014).

These new therapies are being tested for use as single effective agents or agents to reverse resistance of leukaemia cells to chemotherapies. Although the clinical use of targeted therapies is mostly unproven to date, with the exceptions of TKIs in BCR-ABL1 positive leukaemia, some agents seem promising to improve the quality of treatment approaches for ALL patients. Perhaps using targeted therapies will expand treatment options for relapsed ALL cases or may help to induce remission in ALL cases refractory to frontline chemotherapies, so that these patients can proceed to stem cell

44 transplantation (Locatelli et al., 2012; Raetz and Bhatla, 2012). Indeed, current

viewpoints suggest that cure of many ALL patients is still dependent on chemotherapies

and therefore the workup for improving outcomes of ALL patients should also include

optimising the most suitable treatment approaches of available chemotherapies and

selecting patients who will benefit from risk-adjusted therapies based on direct

measurement of the biological features of the leukaemic cells (Krishnan et al., 2011; Pui

et al., 2012).

One of the major challenges related to personalised treatment is identifying biomarkers

in cancer that can guide therapeutic decisions or predict who will benefit from designed

therapy protocols. A biomarker is a biological indicator, which can be used to make

objective estimates and monitor changes in physiological and pathological conditions or

response to a therapeutic medication. Biomarkers can be divided into three categories:

diagnostic, prognostic or predictive. A diagnostic biomarker is usually a characteristic

feature used to identify a particular type of disease in an individual. A prognostic

marker estimates the likely course of a disease in patients so a patient with a poor

outcome can be distinguished from one with favourable outcomes and treated

accordingly. Predictive biomarkers are used to evaluate the efficiency of a particular

therapy by identifying patients as responders or non-responders. These biomarkers are

based on disease clinical features or are DNA, RNA or protein molecules altered in

cancer cells and have a crucial role in mechanisms of the disease progression or

mechanism of drug action and response (Biomarkers Definitions Working, 2001).

Evidence from studying mechanisms underlying resistance of ALL cells to

chemotherapies based on gene expression techniques showed that some genes are 45 expressed at different levels between drug-sensitive and drug-resistant ALL cells. This could provide a promise to identify key genes altered after exposure to chemotherapies and therefore reflect the genetic signature of ALL clones resistant to drugs or the acquired phenotypes to escape the effect of chemotherapies. In two studies conducted to elucidate mechanisms of resistance to conventional chemotherapy agents (prednisolone, vincristine, L-asparaginase and daunorubicin) used to treat ALL patients in the clinic, the differentially-expressed genes between drug-sensitive and drug-resistant ALL cells was used to determine response of ALL patients to these agents. Holleman et al. (2004) showed that analysis of gene expression based on microarray profiling of ALL-sensitive and -resistant cells can identify a signature of genes that associate with drug resistance for each of the four drugs. Interestingly, the validity of this signature was proved using an independent cohort of paediatric ALL patients who were treated according to different treatment protocols using the same chemotherapy agents (Holleman et al.,

2004). A subsequent study supported the concept of gene signatures related to drug resistance where distinct gene sets were identified to correlate with the multi-drug resistance developed to the four chemotherapy agents and similarly the gene expression linked to the drug resistance profiles was able to identify a subgroup of patients with

ALL who had a markedly inferior treatment outcome (Lugthart et al., 2005).

Hulleman et al. (2009) also studied the pattern of gene expression associated with the sensitivity to prednisolone and identified a genetic signature that was differentially expressed between prednisolone-sensitive and prednisolone-resistant precursor B- lineage leukemic patients. Based on a defined signature, targeting of the glycolysis pathway using the 2-deoxy-D-glucose (2-DG) resulted in an increase in the in vitro cytotoxicity of Jurkat and Molt4 cell lines to GCs, while it did not influence the drug

46 cytotoxicity of the prednisolone-sensitive Tom-1 and RS4 cell lines. Interestingly, the

reversing effects of the 2-DG have been also observed in primary cells isolated from

ALL patients (Hulleman et al., 2009). Consistently, Eberhart et al. (2009) also showed

that the combination of 2-DG and dexamethasone can potentiate cell death in GC-

resistant cells (Eberhart et al., 2009).

Despite the great promise from many treatment approaches used to treat ALL patients,

resistance develops in some patients to both conventional chemotherapies and targeted

therapies. A major obstacle that exacerbates challenges for personalised treatment for

leukaemia patients is linking the discrepancy of patient responses to drugs applied in

clinical settings with available sets of prognostic features to guide the type and intensity

of drugs for different cancer patients (Nersting et al., 2011). One of the frequently

expressed reasons for treatment failure is the lack of preclinical models that recapitulate

the heterogeneity of ALL patient disease. The application of personalised leukaemia

treatment, targeted therapy and patient stratification based on their outcomes requires

integration of clinical information with a powerful preclinical model to reflect the

characteristics of ALL patient disease and therefore provide predictive strategies to

identify ALL patients with a high risk of relapse.

Disease modelling strategies

Over the years, various models have been utilised by cancer biologists to gain an

improved understanding of mechanisms involved in cancer and to test the efficiency of

drugs in targeting of cancer cells. The vast majority of research questions have been

investigated in in vitro systems using immortalised human cell lines, mouse cell lines

and patient primary samples. However in vivo models also provide a means to study

47 many features of the disease. These models can be more valuable to study the disease depending on the question asked and resources available to conduct research.

1.11.1 Primary patient cells

Isolation of patient cells from BM is an optimal tool to reflect the biology of the disease and provide the standard for genomic/genetic features of leukaemic blasts. Primary patient samples also provide direct testing of the drug resistance profiles that correlate with clinical outcome (Holleman et al., 2004; Pieters et al., 1998). However, utility of this source is limited by the fact that patient samples are limited in size and do not survive in culture long-term, which could hinder testing large numbers of available treatment approaches (Nijmeijer et al., 2009).

1.11.2 Immortalised cell lines

Cell lines are the basis for the in vitro analysis of the biological and molecular features of acute leukaemia and other cancer types. Leukaemia cell lines originate from a population of cells with high potency to proliferate in culture and that have been exposed to genetic modification in genes responsible for cell immortality, enabling them to divide for an indefinite time. Cell lines can be easy to grow continuously if supplied with suitable culture conditions and therefore provide an unlimited source of human cells (Drexler and MacLeod, 2003; Drexler et al., 2000). Development of cancer cell lines has helped to establish many chemotherapies and targeted molecules used to treat cancer patients in the clinic and therefore this model provides an affordable tool to evaluate the cytotoxic effect of drugs against leukaemia cells. It also has merit to artificially produce resistance to drugs and therefore help to investigate the mechanism by which leukaemia could acquire resistance to applied therapy (McDermott et al.,

2014).

48

Extensive work has been done to expand the utility of cell lines to represent leukaemias

harbouring different translocations and oncogenes implicated in disease biology. For

instance, establishment of the Kasumi-1 cell line, which carries an AML1-ETO

translocation, has helped to understand the role of this fusion in blocking the

differentiation process of B cells via inactivation of PU.1 transcription factor (Vangala

et al., 2003). In addition, this model tends to be more reliable for lentiviral and retroviral

transductions, which facilitate the study of biological functions of individual genes

(Christoph et al., 2013; Fazzina et al., 2012).

Despite the significant contribution of cell lines in our understanding of cancer biology,

cell lines are limited in their ability to retain features of the original patient disease. For

instance, the functional significance of many genes has failed to be validated in vivo

despite strong evidence for their importance using the in vitro approach. Indeed, in vitro

culture condition and expanding of cell lines long-term are known to change gene

expression profiles, which may under- or overestimate the importance of phenotypic

changes in cancer cell lines. Cell lines can be easily contaminated with mycoplasma,

which affects the growth pattern of infected cells, or contaminated with other cell lines

(Drexler et al., 2000; Wilding and Bodmer, 2014).

1.11.3 Genetically engineered mouse models

Understanding the role of genetic alterations that affect tumour suppressor genes and

oncogenes in leukaemogenesis has led to generation of mouse models that resemble

leukaemia development in humans (Jacoby et al., 2014). As mice share many

physiological, anatomical and genomic features with humans, the mouse model has

49 been confirmed as a good candidate for studying the early events involved in leukaemogenesis and the consequences of many hits responsible for full leukemic transformation (Kennedy and Barabe, 2008). Mouse models can be genetically manipulated by various techniques, including ethylnitrosourea, Cre-lox-mediated inversions and deletions of chromosomes and gene trap, as well as pro-viral insertion mutagenesis to mutate one or both copies of tumour suppressor genes in embryonic stem cells (van der Weyden et al., 2002). As a result, an engineered mouse carries on its genome genetic lesions which express phenotypes similar to leukaemia disease in humans.

Investigations of the effect of knocking the expression of many oncogenes on and off by altering a gene with the oncogenic fusion product have led to the study of the influence of many oncogenes on the behaviour of cancer cells. Genetically engineered mouse models (GEMMs) have provided an insight into the mechanism by which many oncogenes mediate leukaemia development and dependence of cancer cells on oncogenic signals to survive. For example, GEMMs that express the cancer inducing oncogene in hematopoietic cells. Mice engineered to maintain sustained expression of the MYC mice transgene develop stable phenotypes of malignant T cell lymphomas and acute myeloid leukaemias. Inactivation of MYC transgene causes rapid proliferative arrest and apoptosis of malignant cells and restoration of normal mouse haematopoiesis (Felsher and Bishop, 1999).

GEMMs provide interaction of tumour tissue with the stromal and haematopoietic components, which makes targeting of tumours occurring in an environment resembling the disease in patients. This feature is important for assessing cytotoxicity in vivo.

50 Another feature highlighting the importance of using GEMMs in preclinical testing of

therapy against malignancy is the addiction of these cells to oncogene expression (Politi

and Pao, 2011).

Although there is widespread use of cell lines and GEMMs in cancer biology research

to screen thousands of anticancer molecules, these models have made a minimal

contribution in defining targets that impact patients survival in the clinic (Williams et

al., 2013). The inability to prove the functional significance of some targets despite

strong evidence from the in vitro results may be attributable to the nature of tumour

models that were used to determine druggable targets. In addition, cancer phenotypes

have been explored without considering the homogeneous nature of the cell lines and

effect of host microenvironment in drug resistance. For instance, along with the genetic

and epigenetic alterations that take place within cancer cells, the tumour

microenvironment in patients may influence the gene expression of patient cells to

avoid drug actions (Iwamoto et al., 2007; Kennedy and Barabe, 2008; Wilding and

Bodmer, 2014). Tumour cells may maintain their survival due to more than one

oncogenic driver mutation and /or mutated tumour suppressor genes, which may

increase the chance of developing resistance to a target therapy by a different

mechanism. Because the range of driver mutations varies between leukaemia cell

populations, the general utility of these models is uncertain. However, these models

could work as essential complementary approaches for exploring the molecular and

cellular biology behind certain genetic defects (Williams et al., 2013).

51 Human patient-derived xenografts

The standard method in which to expand and study leukaemia progression is through establishing human patient-derived xenografts (PDXs) using immunodeficient mice.

Thorough understanding of the mammalian immunological processes and their aberrations has led to development of a number of immunodeficient murine models.

Historically, mutations producing severe combined immunodeficiency (SCID) in mice has been demonstrated to permit transplantation of haematopoietic cells, leukaemia cell lines and patient primary cells, into mice although with lower capacity to engraft.

Xenotransplantation of human cells in SCID murine recipients was limited by the residual immune response of these mice (Greiner et al., 1998; McCune et al., 1988).

Crossing of SCID mice onto a non-obese diabetic (NOD) mice strain established a new strain called NOD/SCID. The resultant strain retains a partial immune system, which makes this mouse more amenable to engrafting human cells than SCID mice (Shultz et al., 1995). The NOD/SCID mouse strain was the gold standard model for establishing many disease models, including leukaemia (Pearson et al., 2008). However, some patient samples may never engraft in NOD/SCID mice even with a long period of monitoring. To overcome these challenges, a number of NOD/SCID-derived mice strains have been developed, including the NOD/SCID/IL-2Rγnull (NSG) strain. NSG mice have an additional impairment in the immune system due to lack of the IL-2R γ- chain, the absence of which disrupts natural killer (NK)-cell activity that restrains efficient engraftment of primary human cells (Shultz et al., 2005).

Patient-derived xenografts are proven to be a valid preclinical model due to the ability to recapitulate the biology and clinical features of cancer in humans. Unlike transgenic

52 models or cell lines that help to investigate the biology of a monoclonal population of

cells and specific genetic aberrations in leukaemia disease, establishing a xenograft

system in immunodeficient mice retains the vital disease features with excellent

relevance to the biological phenotypes among different subgroups and phases of the

disease. The patient-derived xenograft model has the advantage of developing systemic

organ infiltration that is highly relevant to the patient setting. The extent of the disease

progression in mice can be indirectly monitored in real time using disease-specific

markers and testing of drug toxicity against animal health. Therefore response to

therapy is evaluated in real time with an introduction to the pharmacokinetic and

dynamic activities of the drug used to treat xenografted cells. Genetic profiling of PDXs

and patient primary samples showed that PDX samples cluster with their corresponding

primary ALL patient samples and also showed the expected close clustering of genetic

profiles between xenografts derived from unique leukaemia subtype (MLL-ALL, BCP-

ALL or T-ALL) (Neale et al., 2008; Suryani et al., 2014).

ALL PDXs of different disease subtypes show a wide range of sensitivity to established

chemotherapies and other drugs, which provides an insight into the drug response

profiles among leukaemia subgroups with different disease biology. Response of

leukaemia cells to various drug strategies can be studied in an environment close to the

patient setting with an approach to investigate the consequences of intrinsic

(microenvironment) and extrinsic (treatment) selective pressure applied on leukaemia

clones. These factors make many scientists and pharmaceutical companies interested in

using the PDX model for studying cancer biology and testing anti-cancer agents in the

pre-clinical phase. The Pediatric Preclinical Testing Program (PPTP) sponsored by the

NCI is an example of an innovative research consortium that tests drugs prioritised by

53 experts in this field against several types and subgroups of cancers including leukaemia.

Stringent criteria for reporting the drug response are applied to precisely prioritise novel agents that exhibit good efficacy and low toxicity for clinical trials (Houghton et al.,

2007).

The utility of leukaemia PDX models as surrogate cell lines is limited for short-term experiments. Xenograft cells do not proliferate in vitro, although a recent study showed that co-incubation of PDX cells with human mesenchymal stem cells allowed long term proliferation of PDX cells (Pal et al., 2016). Furthermore, PDX cells are difficult to transduce with lentiviral and retroviral methods. Thus, in vitro testing the importance of manipulation of genes in PDX cells is compromised by these practical challenges.

Furthermore, the NOD/SCID mouse strain has a high frequency to develop spontaneous thymic lymphomas by the age of 40 weeks, resulting in early death (Prochazka et al.,

1992).

1.12.1 Progress in acute lymphoblastic leukaemia patient-derived xenografts

Recent research on xenograft-based leukaemia models has been an essential stage for advances in the utilisation of PDXs to greatly inform our understanding about the diversity in biology of ALL disease and reflect the clinical course of ALL patient disease. The clinical course of the disease in mice was characterised based on different recipient mouse strains, different engraftment setup and using variable definition of leukaemia manifestation to improve the predictive value of this preclinical model.

The utility of PDXs has been extended to investigate the functional importance of unique cell biology that drives the disease in mouse models and addresses the impact of

54 clonal heterogeneity in outcomes of leukaemia patients. PDXs were recently used to

assess the repopulation capacity of mixed or unique leukaemia populations to initiate

leukaemia disease. Research studies that used the NOD/SCID strain identified that rare

populations of leukaemia cells have leukaemia-initiating activity and this feature is

mainly enriched in cells with immature phenotypes. In contrast, leukaemia-initiating

activity was found to be common and reside in cells with a more mature phenotype,

based on engrafting patient samples into NSG and other mouse strains (Heidenreich and

Vormoor, 2009). However, further research by le Viseur et al. (2008) showed that

leukaemia cells at different stages of maturation are able to transfer in both NK cell-

depleted NOD/SCID and NSG mice. This indicates that different leukaemia sub-

populations could have stem cell-like properties.

ALL PDXs provide a good approach to study ALL clonal heterogeneity and expand

leukaemia populations with more relevance to patient disease. Comparing leukaemia

subpopulations between PDXs established from ALL patients at time of diagnosis and

their corresponding samples at relapse, revealed that certain leukaemia clones which

could cause disease progression were represented in ALL PDXs. Based on analysis of B

and T cell gene rearrangement for some xenografts derived from diagnostic samples of

HR ALL patients which expanded up to the tertiary passage in NOD/SCID mice, the

pattern of clones in xenografts was similar to the clonal rearrangement in the relapse

samples (Liem et al., 2004). Concordantly, leukaemia sub-clones with various genetic

alterations have been shown to be selected frequently in xenografts derived from

diagnosis samples and the genetic constituents of PDX cells are closer to relapse clones

(Clappier et al., 2011).

55 Engraftment of leukaemia cells in mouse models is being assessed to see whether engrafted cells show evidence toward association between the engraftment patterns of

PDXs and outcomes of patients from whom the samples were derived. Correlation between poor outcome and engraftment is very significant and it has been suggested as being predictive of relapse in leukaemia patients. Several studies have shown correlations between outcomes of leukaemia patients and in vivo proliferation of their leukaemia cells transplanted into immunodeficient mice, although with some inconsistency between reports. In a large cohort of BCP-ALL samples, including all risk groups, this correlation did not hold when the SCID mouse strain was used for xenotransplantation (Uckun et al., 1998). However, a small study from our lab showed a good correlation between the length of patient CR and the time required for ALL patient samples collected at time of relapse, but not at diagnosis, to engraft in irradiated

NOD/SCID mice (Lock et al., 2002).

Perhaps the most compelling evidence for using xenograft models for upfront prediction of outcome has been highlighted by Meyer et al. (2011). This study showed that time to leukaemia (TTL) can reproducibly identify the high risk group of paediatric ALL cases according to the short time required for development of overt leukaemia manifestation in non-conditioned NOD/SCID mice. Interestingly, leukaemia cells extracted from early

TTL xenografts were characterised by genetic signatures that reflected the high-risk phenotypes of some patients who relapsed after treatment. This signature includes activated pathways that involve cell growth and apoptosis. Validating the importance of

TTL on xenografts established from another group of ALL patients showed independency of the TTL factor as a strong predictor for patient outcome (Meyer et al.,

2011).

56 Further to these is a study conducted by Schmitz et al. (2011) that demonstrated the

higher engraftment of de novo resistant VHR-ALL samples in NSG mice compared

with standard risk ALL cells. Interestingly, the VHR-ALL cells were able to maintain

high levels of leukaemia cells in the BM and spleen of xenografted mice upon serial

transplantation between mice recipients but not the standard risk-ALL cases. This

finding indicates a high capacity of disease clones in VHR samples to establish the

disease (Schmitz et al., 2011).

Several reports have also measured the influence of establishing PDXs using different

engraftment conditions and applying various selection strategies to improve the

relevance of mouse models to patient clinical information and improve the efficiency of

the engraftment process. For instance, the role of the immune system in influencing

leukaemia engraftment was investigated by comparing leukaemia engraftment between

mouse strains with different levels of immunodeficiency using unsorted leukaemia cells.

The efficiency and speed of leukaemia cell engraftment were shown to be frequently

higher in the NSG compared with NOD/SCID mouse strain (Agliano et al., 2008;

Diamanti et al., 2012).

Although the majority of ALL patient samples tend to engraft using the intravenous

(IV) route of inoculation, cell homing in mice may confine the growth of some clones

that might be representative for the patient disease. Therefore, injecting ALL cells

directly into the BM via the femur (IF) could circumvent this problem. Recent research

has reported superior transplantation of leukaemia patient samples injected via the intra-

femoral route using the NSG mouse strain. Interestingly, one hundred leukaemic cells

transplanted into NSG mice via the IF route repopulated the same patient clones with

57 slight changes in CNA (Schmitz et al., 2011). The ability of leukaemia xenografts to retain similar patient clones after IF injection of a few cells suggests that this method has high efficiency in displaying the disease features at least for ALL disease.

Since ALL diseases constitutes a mixture of clones with a different potency to engraft and progress, administration of a combined chemotherapy regimen could play a major role in improving xenotransplantation strategies. It may act as a selection method for variable clones and therefore exert a distinct pressure for selection of a sub-clone to dominate the disease. Interestingly, recently published data from our group showed that in vivo treatment of xenografts with multiple cycles of combined chemotherapy in a protocol [vincristine, dexamethasone, L-asparaginase and daunorubicin (VXLD)] mimics the induction phase in paediatric T-ALL patients, inducing resistance in two highly aggressive leukaemia samples (Samuels et al., 2014). Relapse xenografts were characterised by a decrease in leukaemia growth delay (LGD) and mouse survival rate when compared with their non-treated counterparts. These resistant lines also exhibited resistance to dexamethasone and alteration in expression of genes for lipid and cholesterol metabolism. Analysis of the connection between the genetic signature associated with resistance to induction chemotherapy drugs based on Connectivity map analysis identified the cholesterol pathway inhibitor simvastatin (SVT) as a potential therapy to overcome resistance. Ex vivo testing of the combination of dexamethasone and SVT showed that the combined therapy synergistically inhibits the percentage of viable cells compared with the effect of each drug alone. However, administration of

SVT with the VXLD in vivo did not delay leukaemia progression due to the limit of the maximum tolerated dose of SVT in mice (Samuels et al., 2014).

58 Aims and summary of the topic

Current clinical criteria are reasonably predictive of high-risk and standard risk disease

in ALL patients but are insufficient to infer the prognosis of IR ALL patients who

represent the largest group with relapses. A complete and highly effective cure for IR

ALL patients could be assured with proper adjustment of treatment plans based on

patient clinical factors, the intrinsic biological properties of leukaemia cells and

interaction between the disease and host. Several common themes emerged from the

literature, suggesting that PDXs in many ways represent a valuable tool for testing the

hypothesis that ultimately drives personalised strategies for leukaemia patients.

In this project I hypothesised that ALL PDXs may serve as an integral approach toward

precise identification of previously unrecognised subgroups of IR ALL patients with

high risk of relapse and therefore my specific objectives were:

1- To conduct a Pilot Study experiment to optimise the most appropriate PDX

model that can predict an outcome of IR BCP-ALL patients according to the

engraftment of their patient cells in murine models.

2- To test the feasibility of using an optimised engraftment design from the Pilot

Study to improve relapse prediction in a larger cohort of IR BCP-ALL patients.

3- To characterise the heterogeneity between IR BCP-ALL xenografts and study

their response to chemotherapeutic drugs

59 CHAPTER 2 MATERIALS AND METHODS

60 2.1 Xenograft mouse model

2.1.1 Reagents and equipment

Cryopreserved vials of mononuclear cells, which were derived from ALL patients, were

obtained from the Tumour Bank at Children’s Cancer Institute (CCI, UNSW,

Kensington, NSW, AUS) and transferred into our internal liquid nitrogen tank in the

Leukaemia Biology Program. Five to nine-week-old Nonobese Diabetic/Severe

Combined Immunodeficiency (NOD/SCID) and NOD/SCID common cytokine receptor

gamma chain–/– (NSG) mice were purchased from the Australian BioResources facility

(Moss Vale, NSW, AUS). Upon arrival at the Animal Facility in CCI, mice were placed

in micro-isolator cages (6 mice/cage). Xenotransplantation of human cells into non-

conditioned mice was performed after one week of mice arrival and acclimatisation.

Roswell Park Memorial Institute 1640 medium (RPMI1640), Penicillin (10,000 U/mL),

Streptomycin (10 mg/mL), L-glutamine (29.2 mg/mL) liquid (PSG) were supplied by

Invitrogen (Carlsbad, CA, USA). Trypan blue (0.4% solution) and dimethyl sulfoxide

(DMSO) were purchased from Sigma-Aldrich (St Louis, MO, USA). Neubauer

haemocytometer chambers were purchased from Dutec Diagnostics (Sydney, NSW,

AUS). Needles (23 gauge x 1 1/4”), FACS lysing solution, cell strainers (40 µm),

Fluorescein isothiocyanate (FITC)-conjugated anti-murine CD45, allophycocyanin

(APC)-conjugated anti-human CD45 and APC-conjugated anti-human HLA-DR and

FITC- and APC-conjugated anti-murine IgG isotype control antibodies were purchased

from BD bioscience (San Jose, CA, USA). Insulin syringes (1 mL and 0.5 mL, 29 and

27 gauge x 1/2”) were purchased from Livingstone International (Rosebery, NSW,

AUS). Mini-Collect® EDTA blood collection tubes were purchased from Interpath

Services (Heidelberg West, VIC, USA). LymphoPrep was purchased from Axis-Shield

(Oslo, Norway). Fluorescence Activated Cell Sorting (FACS) analysis was performed

61 using a FACSCanto multiparametric flow cytometer purchased from BD and acquired samples were analysed using BD FACSDiva software version 6.1.2.

2.1.2 Preparation of patient samples for inoculation into mice

On the day of inoculation, vials of frozen ALL patient samples were thawed by placing them in a 37°C water bath. The cell suspension from each vial was then transferred into a 15 mL centrifuge tube and 10 mL of pre-warmed RPMI1640 medium plus 10% FCS

(heat inactivated for 30 min at 56°C) and PSG (100 U/mL penicillin, 100 µg/ml streptomycin, and 2 mM L-glutamine) was added to the cell suspension. The first 3 ml of medium was added dropwise with constant mixing of cell suspension to avoid the osmotic shock of ALL cells during the process of thawing. Each sample was then centrifuged at 490 x g for 5 min and pelleted cells were resuspend in 5 mL of the

RPMI1640 medium. Cells were counted and assessed for viability using the trypan blue exclusion method as described in Section 2.2.4. One million cells of each sample were resuspended in an appropriate volume of PBS for inoculation into mice (100 µL for intravenous, and 25 µL for intra-femoral injections). Samples of ALL patient cells were then transferred into 1.5 mL microcentrifuge tubes, placed on ice and transferred to the

Animal Facility at CCI for injection into mice.

2.1.3 Intravenous (IV) transplantation of leukaemia cells

To prepare mice for inoculation of leukaemia cells via the intravenous route, each box containing mice to be inoculated with ALL patient samples was placed in front of an infrared heat lamp until the veins of the mouse tail appeared dilated. Mice in each cage were then moved into a biological safety cabinet and each mouse positioned in a perspex restrainer with the lateral vein exposed for injection. Using a 23 gauge needle attached to a 0.5 mL syringe, the leukaemia cell suspension (100 µL) was inoculated via the lateral tail vein. Compression tissues were applied on the injection site until 62 bleeding had ceased and the mouse was returned to its cage and monitored for general

health and activity.

2.1.4 Intra-femoral (IF) transplantation of leukaemia cells

To inoculate patient samples directly into the bone marrow via intra-femoral transplant,

mice were anaesthetised with a mixture of isoflurane /oxygen gas. Mice were placed on

an electrical heating pad during anaesthesia and recovery to maintain normal body

temperature. To determine if an appropriate depth of general anaesthesia was achieved,

a pain stimulus was applied to one foot of a mouse and in case of foot withdrawal, the

level of anaesthesia was adjusted prior to carrying out any procedure. A small puncture

was created in the head of the right femur with a 29 gauge needle by an expert

researcher who optimised this injection method in our lab. Twenty-five µL of leukaemia

cell suspension was then injected via the IF route, mice were allowed to recover from

anaesthesia and activity was recorded for 2 hrs after the procedure.

2.1.5 In vivo treatment with VXL chemotherapy

Before initiating treatment with VXL chemotherapy, mice were randomised to avoid

any bias that could be introduced from the process of inoculation. Mouse body weights

were measured after randomisation to calculate VXL chemotherapy doses. Drug

treatment was started after two weeks of cell injection. Half of the group of mice

injected with each patient sample received a combination of vincristine (VCR) (0.15

mg/kg once every 7 days for 2 weeks), dexamethasone (DEX) (5 mg/kg Mon-Fri for 2

weeks) and L-asparaginase (ASP) (1000 IU/kg Mon-Fri for 2 weeks) via the

intraperitoneal route (IP) and the rest of mice received an equivalent volume of 0.9

normal saline using the same dosing route. Drugs were prepared weekly, diluted using

0.9% normal saline and stored at 4°C.

63 To administer the IP injection, each mouse was held by scruffing the loose skin above the mouse shoulders using two fingers and the other three fingers of the same hand were used to restrain the mouse body and the tail. The mouse head was tilted down at 35-40° angle so that the body organs fall away from the injection sites. A 27 gauge needle was used to inject drugs into the lower part of the abdomen inside the peritoneal cavity.

2.1.6 Monitoring of engraftment

Leukaemia engraftment was monitored weekly by analysing the percentage of human cells in the murine PB. Monitoring commenced 2-3 weeks after inoculation of patient samples and continued throughout the whole study period. The non-VXL treated mice were bled every week for monitoring of engraftment. Mice that received VXL chemotherapy were bled on alternate weeks until engraftment was evident in the PB of the control mice and then the VXL treated mice were bled weekly until euthanasia. To prepare for bleeding, mice were exposed to the infrared heat lamp, until the tail veins became dilated and then mice were placed in a perspex restrainer with the lateral tail vein exposed for bleeding. The veins on the lower part of the tail were punctured using a 23 gauge x 1/4” needle and 2-3 drops (50–75 µL) of blood were collected in Mini- collect EDTA blood tubes and blood samples were mixed with the anticoagulants thoroughly. Compression tissue was applied on the puncture site to stop bleeding.

Approximately 50 µL of each sample of the PB was transferred into a 5 mL flow cytometery tube and 100 µL of an antibody mixture containing 90 µL PBS, 4 µL mouse serum, 1 µL FITC-conjugated anti-murine CD45 and 5 µL APC-conjugated anti-human

CD45 or HLA-DR was added to blood samples. At least one sample was stained with

IgG isotype antibody (APC-conjugated anti-mouse and FITC-conjugated antimouse) to use for accurate setting of the gates during flow cytometer analysis. The tubes were then

64 mixed using a vortex mixer and incubated for 30 min in the dark at RT. Erythrocytes

were lysed with 800 µL FACS lysing solution for 10 min at 37°C in the dark and 3 mL

of flow cytometry buffer (0.2% BSA, 0.1% sodium azide in PBS) was added and

samples centrifuged at 490 x g for 5 min at RT. The supernatants were discarded and

cell pellets were resuspended in 200 µL of PBS. Cells were analysed by dual-color

multiparametric flow cytometry on a FACSCanto cytometer with the FACS-Diva v6.1.2

software using appropriate compensation settings.

Mouse cells which stained positive for FITC-conjugated anti-mouse CD45 were

detected in the FL1 fluorescence channel at an emission wavelength of 530 nm and

human cells which stained positive for APC-conjugated anti-human CD45 or HLA-DR

were detected in the FL4 channel at an emission wavelength of 670 nm. A minimum of

10000 events were recorded per sample. The percentage of leukaemia engraftment in

murine PB was determined based on the proportion of human CD45 or HLA-DR

positive cells versus total human and mouse cells, as it has been shown that this

parameter accurately reflects the overall leukaemic burden (Rice et al. 2000). Previous

studies have shown that this method reliably detects as low as 0.1% human CD45 cells

in murine PB (Nijmeijer et al. 2001).

2.1.7 Harvesting of leukaemia cells from engrafted mice

Mice were euthanised by carbon dioxide asphyxiation if they showed signs of suffering

and distress (hunched, lethargy, ruffled fur) or the total body weight loss was equal to or

> 20% in accordance with the UNSW Animal Ethics guidelines (See Appendix A).

Each mouse was dissected immediately after death and a blood sample was collected

via cardiac puncture into a MiniCollect® EDTA tube. The mouse spleen was collected

into a 5 mL tube containing RPMI1640 medium for harvest of leukaemia cells for

further use. Right and left femur bones were flushed with 1 mL of RPMI1640 medium 65 using a 27 gauge needle to harvest human leukaemia cells in the BM. Small pieces were collected from liver, kidney, lung, lymph nodes and the brain to determine the extent of leukaemia infiltration into body organs.

In order to harvest human leukaemia cells from the spleens of engrafted mice, spleen cell suspensions were prepared by placing individual organs into a sterile tea strainer and homogenising with RPMI1640 medium using the plunger of a 10 mL syringe to mince spleen tissues. The suspension of spleen contents was filtered by passing the homogenate through a 40 µm cell strainer into a 50 mL falcon tube containing

RPMI1640 to a final volume of 35 mL. Fifteen mL of LymphoPrep was underlaid carefully, by using a 25 mL syringe fitted with a sterile cannula, to minimise mixing of the cell suspension with LymphoPrep. Mononuclear cells from spleens were separated by density gradient centrifugation (800 x g for 30 min at RT). The interphase

(mononuclear cell) layer was then transferred into a clean tube and diluted to 40 mL with RPMI1640 medium and viable cells were counted using the trypan blue exclusion method. Xenograft cells were cryopreserved at 15-60 million cells/ml in FCS containing

10% DMSO. Cells harvested from the BM were also filtered by passing the cell suspension through a 40 µm cell strainer into a 10 mL tube and cells were then washed with RPMI1640 medium. After counting viable cells using the trypan blue exclusion method, xenograft cells harvested from the BM were also cryopreserved in FCS containing 10% DMSO.

2.1.8 Analysis of mouse EFS at different levels of engraftment

To analyse mouse EFS values at different levels of engraftment, the percentage of engraftment of each mouse inoculated with patient sample was determined weekly.

Values were entered into GraphPad Prism software and EFS was calculated and

66 visualised by Kaplan-Meier analysis, and significant differences in EFS were calculated

based on the log-rank test.

2.2 Cytotoxicity assays

2.2.1 Reagents and equipment

Quality Biologicals Serum Free-60 (QBSF-60) media was supplied by Quality

Biologicals (Gaithersburg, MD, USA), and FMS-like tyrosine kinase-3 ligand (FLT-3-

L) was purchased from Amgen (Thousand Oaks, CA, USA). Roswell Park Memorial

Institute 1640 medium (RPMI1640), Penicillin (10,000 U/mL), Streptomycin (10

mg/mL), L-glutamine (29.2 mg/mL) liquid (PSG), and phosphate-buffered saline (PBS)

were supplied by Invitrogen (Carlsbad, CA, USA). Dexamethasone, 17-

dimethylaminoethylamino-17-demethoxygeldanamycin (17 DMAG), trypan blue (0.4%

solution), dimethyl sulfoxide (DMSO) and Alamar Blue were purchased from Sigma-

Aldrich (St Louis, MO, USA). Falcon® 96 Well Clear Round Bottom TC-Treated Cell

Culture Microplates were purchased from BD Bioscience (San Jose, CA, USA).

Neubauer haemocytometer chambers were purchased from Dutec Diagnostics (Sydney,

NSW, AUS). A Victor X3 Multilabel Plate Reader was purchased from PerkinElmer

(Massachusetts, USA). Vincristine and L-asparaginase (Leunase) were purchased from

the Prince of Wales Hospital pharmacy (POWH, Randwick, Sydney, AUS). All cell

culture procedures were performed in a Biological Safety Cabinet Class II (AES

Environmental Pty Ltd, Sydney, AUS). All ex vivo experiments were carried out at

37°C/5% CO2 in humidified incubators.

2.2.2 Preparation of cells for ex vivo cytotoxicity analysis

The ex vivo sensitivity of ALL xenografts to chemotherapy drugs (vincristine,

dexamethasone or L-asparaginase) was tested using the Alamar Blue assay. To

67 undertake these experiments, ALL xenograft cells harvested from mouse spleens were retrieved from cryostorage and thawed rapidly by placing them in a 37°C water bath.

The cell suspension from each vial was then transferred into a 15 mL centrifuge tube and 10 mL of pre-warmed RPMI1640 medium was carefully added to the cell suspension. Cells were then centrifuged at 490 x g for 5 min and washed with complete

RPMI medium. Cell viability and number were determined by trypan blue exclusion assay as describe in Section 2.2.4. Xenograft cells to be incubated with the chemotherapy drugs were resuspended in QBSF-60 media supplemented with 20 ng/mL

FLT3 and PSG at an appropriate density and 100 uL seeded into 96-well cell culture plates at a concentration of 2-5×105cells/well (Table 2.1). Each plate was then incubated at 37°C/5% CO2 for 5-6 hrs before adding drugs.

Cytotoxic drugs were diluted in an appropriate vehicle and made up to 1 mL with

QBSF-60 medium to prepare a concentration 6 times higher than the final concentration. Twenty µL of the drug was added to the 100 µL of cells in each plate

-12 -5 such that the final concentration of the drug ranged from 10 to 10 M. Each plate included 3 wells of untreated xenograft cells consisting of cells with media plus vehicle equivalent to the highest concentration of drug and 3 wells of ALL xenograft cells incubated with the serial drug concentrations. Plates were then incubated for 48 hrs in a

37°C/5% CO2 incubator.

2.2.3 Alamar Blue cytotoxicity assay

The ex-vivo sensitivity to drugs was assessed based on using the Alamar Blue cytotoxicity assay. The Alamar Blue is a blue non-fluorescent dye, which is reduced to the pink-coloured, highly fluorescent resorufin. The Alamar Blue assay was developed by Ahmed et al. (1994) to monitor and determine the viability and proliferation of cells

68 based on reduction of the Alamar Blue stain by mitochondrial dehydrogenase enzymes.

This assay measures the degree of change in fluorescence after the enzymatic reduction

measured by a fluorescence microplate reader to reflect the mitochondrial activity of

cells in the plate as a surrogate measure of cell viability.

To measure the cytotoxic effect of drugs on ALL xenografts, the Alamar Blue reagent

was prepared by diluting the Resazurin solution in sterile pre-warmed PBS (1:10) and

the solution was then filtered using Whatman filter paper. Alamar blue reagent (14 µL)

was added into each well and the fluorescence in each well was estimated based on 560

nm excitation wavelength and 590 nm emission wavelength at 0 hr and after 24 hrs of

reagent addition using the Victor X3 Multilabel Plate Reader (560 nm excitation

wavelength, 590 nm emission wavelength). The difference in the level of fluorescence

was calculated by subtracting the 0 hr background from the 24 hrs reading and the cell

viability in the drug-treated wells was calculated relative to untreated vehicle control

cells as follows: (mean fluorescence drug-treated cells/mean fluorescence control

cells)*100.

Table 2.1. Optimised cell densities for Alamar Blue assay

Xenograft Cell densityX105/100 µL

ALL-64 5

ALL-65 5

ALL-66 2

ALL-67 4.5-5

69 2.2.4 Trypan blue exclusion assay

Determination of cell number and percentage of viability in primary ALL patient samples and ALL xenografts were performed using trypan blue exclusion assay. This assay allows differentiation between viable cells and dead cells based on the integrity of their cell membranes. It is based on the principle that dead cells permit passing of the trypan blue dye through their cell membranes whereas viable cells prevent the dye from passing through their intact membrane. The difference in cell viability is visualised under a light microscope whereby dead cells appear as dark blue round cells but viable cells appear as clear white round cells.

To perform this assay, an aliquot of cell suspension being tested was mixed with trypan blue stain (0.4% in PBS) at ratios of 1:1 to 1:10. Cells in the mixture were then transferred into the Neubauer haemocytometer and visualised using a light microscope.

Viable cells (unstained) and dead cells (blue stain) were counted from at least 4 individual 0.1 mm2 fields of the chambers, and then averaged to give the number of cells per 0.1 mm2 counting chamber, which is equivalent in volume to 0.1 µL. The concentration of cells/mL was calculated by multiplying the number of cells per 0.1 mm2 field by the trypan blue dilution factor and by a factor of 104. Cell viability was calculated as: % viable cells = (total no. viable cells/total no. of cells)*100.

2.3 Preparation of RNA for gene expression analysis

2.3.1 Reagents and equipment

RNase/DNase free water and Trizol reagent were purchased from Invitrogen.

Chloroform and nucleic acid grade absolute ethanol were purchased from Sigma-

Aldrich. The RNeasy RNA isolation kit was purchased from QIAGEN (Valencia, CA,

USA). Total RNA was amplified using Illumina Total Prep RNA Amplification Kit

70 (Ambion, Foster City, CA, USA). The Nanodrop ND-1000 spectrophotometer was used

to quantify RNA, provided by Nanodrop Technologies (Wilmington, DE, US). RNA

quality and integrity were measured on an Agilent Bioanalyzer 2100, purchased from

Agilent Biotechnologies (Santa Clara, CA, USA). Data were analysed and interpreted

using Agilent 2100 Expert Software by the Ramaciotti Centre (UNSW, Kensington,

NSW, AUS).

2.3.2 RNA extraction

To extract RNA samples, cryopreserved xenograft cells were thawed rapidly by

incubating vials in a 37°C water bath. Then, cells were transferred to a 15 mL tube and

pre-warmed RPMI1640 medium was added drop wise to the suspension of thawed cells.

ALL cells were counted and viability was assessed as described in Section 2.2.4. Cells

were then washed by centrifugation at 490 x g for 5 min and the supernatant was

discarded in order to remove DMSO traces. Cells were transferred into an eppendorf

tube and centrifuged again at 490 x g for 5 min at RT. Following aspiration of the

medium, cell pellets were released by flicking the tubes and 1 mL of Trizol solution was

added to lyse cells (5-10x106). Each sample was then incubated at RT for 5 min before

the addition of 200 µL of chloroform. Each sample was vigorously shaken for 15 sec

and incubated at RT for 2-3 min. The sample was then centrifuged at 12000 x g for 15

min at 4°C. The aqueous phase resulted after the centrifugation was transferred into a

new eppendorf tube and an equal volume of 70% ethanol was added to the same tube.

The RNeasy RNA isolation kit was used to isolate and purify total RNA and the

manufacturer’s instructions were followed for every step in the protocol. Firstly, 700 µL

of the aqueous phase and ethanol mixture was transferred into Qiagen RNeasy spin

columns. The RNeasy columns were then centrifuged at 8000 x g for 30 sec at RT and

71 the flow-through was discarded. Secondly, 700 µL of RWI buffer was pipetted onto the column and the column was centrifuged at 8000 x g for 30 sec to allow for the washing buffer to filter through. The flow-through and collection tubes were discarded. Thirdly,

RNeasy columns were placed in a new 2 mL collection tube. RPE buffer (500 µL) was pipetted onto the column, and columns centrifuged at 8000 x g for 30 sec. Another 500

µL of the same buffer was added again and the column was centrifuged at 8000 x g for

30 sec. The RNeasy columns were placed in a 2 mL collection tube and centrifuged at

16000 x g for 2 min to remove any ethanol carried over from the previous step. The spin columns were then transferred to a new 1.5 mL collection tube and RNA was then eluted from the column by adding 30 µL of pre-warmed nuclease-free water into the middle of the column and the samples were incubated at RT for 2-3 min. The tubes were centrifuged at 8000 x g for 1 min. The eluate containing RNA was collected in the collection tube and tubes were transferred onto ice during analysis of RNA quantity.

RNA quality was evaluated through two approaches based on using the Nanodrop

Spectrophotometer and the Agilent Bioanalyzer in the Ramaciotti Centre for Gene

Function Analysis at the University of New South Wales. The Optical Density (OD),

260/280 and 260/230 ratios, were measured on the Nanodrop to assess the purity and concentration of extracted RNA samples. Extracted RNA samples with good purity showed OD 260/280 and 260/230 ratios around 2.0. An aliquot of each sample was sent for assessing the integrity of RNA based on calculating of RNA integrity number (RIN) using an algorithm, which estimates the level of RNA degradation in samples. The RIN factor was determined based on using RNA 6000 LabChip kits and an Agilent 2100

Bioanalyzer. Samples with an OD 260/280 and 260/230 ratio of more than 1.80, and an

72 RNA integrity number (RIN) of greater than 8.0 were selected for processing of gene

expression analysis using Illumina microarray chips.

2.3.3 RNA amplification

To amplify RNA samples for running on Illumina Arrays, RNA was first converted to

complementary RNA (cRNA), labelled with biotin and amplified for hybridisation with

Illumina microarray chips using an Illumina® TotalPrep™ RNA Amplification kit. The

experimental procedure was carried out in accordance with the Ambion Illumina® Total

Prep kit manual. The protocol includes synthesis of first and second strand cDNA based

on reverse transcription of RNA, purification of synthesised cDNA, amplification and

labelling of cRNA based on in vitro transcription followed by purification of

synthesised cRNA.

The recommended amount of RNA for Illumina RNA Amplification experiments is 50-

500 ng of total RNA. In these experiments, 500 ng of total RNA was brought up to 11

µL with RNase-free water in a non-stick, RNase-free 0.5 mL microcentrifuge tube. The

reverse transcription master mix was prepared at RT as in Table 2.2. The master mix

was mixed well by gently vortexing, centrifuged briefly and then placed on ice. Nine

microlitres of the reverse transcription master mix was added to each RNA sample,

mixed thoroughly by pipetting up and down 2-3 times, flicking the tube 3-4 times and

then centrifuged briefly. The samples were then placed in a thermal cycler at 42°C for 2

hrs. After incubation, the tubes were centrifuged briefly and placed on ice.

A Second Strand Master Mix was prepared on ice as described in Table 2.3 and 80 µL

was added to the 20 uL from the previous step. Tubes were then incubated for 2 hrs at

16°C (no lid) in a thermal cycler. During the incubation time, nuclease-free water was

preheated to 55°C for 10 min and the cDNA elution columns were placed in the wash 73 tubes. After the 2 hrs of incubation, tubes were taken out of the thermal cycler and placed on ice. To purify cDNA samples, 250 µL of the cDNA Binding Buffer was added to each sample and the cDNA sample/cDNA Binding Buffer solution was mixed thoroughly then pipetted onto the centre of a cDNA Filter Cartridge. The solution was centrifuged at 10000 x g for 1 min at RT. Flow through was discarded and the cDNA

Filter Cartridge was placed in a tube for sample washing. The washing buffer (500 uL) was applied to each cDNA Filter Cartridge followed by centrifugation at 10000 x g for

1 min at RT. After 2 min incubation at RT, the cDNA sample was eluted with 20 µL preheated nuclease free water (50-55°C). The resulting double-stranded cDNA was then transferred to a fresh PCR tube.

The final step of RNA amplification includes labelling of cRNA based on an in vitro transcription (IVT) method. An IVT Master Mix was prepared at RT as described in

Table 2.4 and 7.5 µl was added to each cDNA sample. Samples were placed in a thermal cycler at 37oC for 14 hrs to maximise cRNA yield, followed by the addition of

75 µL nuclease-free water to stop the reaction. The samples were then transferred to a sterile 1.5 ml Eppendorf tube and 350 µl cRNA binding buffer and 250 µl 100% ethanol were added to each tube. Each sample was then transferred to cRNA filter cartridges, and catridges were centrifuged at 10,000 x g for 1 min at RT. The flow-through discarded, 650 µL Wash Buffer was added, and cartridges were centrifuged at 10000 x g for 1 min. One hundred µL of Pre-warmed nuclease-free water (55oC) was added to the centre of each cartridge, and cRNA eluted by centrifugation at 10,000 x g for 90 sec at RT.

74 Table 2.2. First Strand Master Mix

Amount per reaction Component

1 µL T7 (oligo dt) primer

2 µL 10X first strand buffer

4 µL dNTP mix

1 µL RNase inhibitor

1 µL Array Script

Table 2.3. Second Strand Master Mix

Amount per reaction Component

63 µL Nuclease-free water

10 µL 10X Second strand buffer

4 µL dNTP mix

2 µL DNA polymerase

1 µL RNase H

Table 2.4. In vitro transcription (IVT) Master Mix

Amount per reaction Component

2.5 µL T7 10X reaction buffer

2.5 µL T7 Enzyme mix

2.5 µL Biotin-NTP mix

75 2.4 Illumina gene expression

Microarray studies were performed by hybridisation of cDNA samples to HumanHT-12 v4 BeadChip arrays. The Illumina HT-12 Array was run at the Ramaciotti Centre for

Genomics and the raw data were obtained in the form of an IDAT file. To perform analysis of gene expression, the raw data were processed using GenePattern software suite (provided by the Garvan Institute, http://pwbc.garvan.unsw.edu.au).

2.4.1 Normalisation and transformation

Each IDAT file was converted into the GenePattern GCT file format using the Illumina

Expression File Creator module within GenePattern with the collapse mode set at

‘max’. The GCT file was then subjected to log transformation (log2) to stabilise variance among the range of expression, using the Log Transform module in

GenePattern. The log transformed GCT file was then normalised using the scale normalisation module to remove non-biological differences introduced during the process of hybridising RNA to microarray chips.

2.4.2 Hierarchical clustering.

The pattern of sample clustering based on RNA expression was analysed using the

Heirarchical Clustering module within GenePattern. The log transformed normalised

GCT file was subjected to unsupervised hierarchical clustering using the Pairwise complete-linkage feature to define the clustering pattern of samples and the generated dendogram was visualised using the Heirarchical Clustering viewer.

2.4.3 Differential gene expression

Comparison of gene expression between samples was performed using the limma

(linear model for microarray analysis) module function. This module is a moderated t- test, which has the ability to reduce variance or noise estimate for each gene and thus

76 provide more statistical power than other normal t-tests for identifying truly

differentially expressed genes. Analysis of gene expression data using the limma

module allows comparing the entire gene set (34,694 genes) between samples and

determining significant differences in gene expression based on Benjamini and

Hochberg false discovery rate (FDR) (Benjamini and Hochberg 1995) and unadjusted P

value.

2.4.4 Gene Set Enrichment Analysis (GSEA)

Gene Set Enrichment Analysis (GSEA) was used to delineate the change in gene

expression between samples in relation to publically available gene networks and

reported signatures uploaded into the Molecular Signatures Database (Subramanian et

al., 2005). To specify which pathways or gene signatures were highly enriched in the

differentially expressed genes defined by the limma module; the rank file created by the

limma GenePattern module was submitted to the GSEA preranked module within

GenePattern and evaluated against the c5 all v2 (gene ontology) and also c2-all v4

curated collections of gene sets from the Molecular Signatures Database with 1000

permutations.

2.4.5 Connectivity Map (CMap) analysis

The CMap database is a collection set of gene expression signatures, linked with the

action of bioactive compounds against a variety of cancer cell lines, available through

the Broad Institute (http://www.broadinstitute.org/cmap). It allows discovery of

connections between drugs and genes through querying the match between an uploaded

gene list and thousands of gene signatures available in the database (Lamb, 2007). The

CMap analysis was performed using the most significantly differentially expressed

genes (FDR < 0.05, P value < 0.01) between the VXL-treated and non-treated samples

of ALL-67 for prediction of a potential reversing agent of resistance to vincristine 77 observed in ALL-67 VXL-treated samples. As resources in the CMap database were generated using Affymetrix_U113A gene chips, it was required first to convert the differentially expressed genes from Unigene Code IDs to Affymetrix U113_A IDs.

The Unigene Code IDs were converted to their corresponding Affymetrix U113_A IDs using Biomart portal ID Converter tool which is publically available through

(http://www.biomart.org/converter/#!/ID_converter/gene_ensembl_config_2). The list of recovered Affymetrix U133_A IDs was then uploaded into the CMap database. The program compares the query signature against the reference library and assigns a connectivity score between +1 to -1 based on the strength of enrichment to the reference gene-expression profile. A highly positive connectivity score indicates that the corresponding perturbagen induces the expression of the query signature and a highly negative connectivity score indicates that the corresponding perturbagen reversed the expression of the query signature. A zero connectivity score is assigned when no positive or negative enrichment is identified (Lamb et al, 2006).

2.5 Quantitative real-time reverse transcriptase polymerase chain reaction (qRT-PCR)

2.5.1.1 Reagents and equipment MMLV Reverse Transcriptase, 5X buffer [250 mM Tris-HCl (pH 8.3), 375 mM KCl,

15 mM MgCl2], 0.1M DTT and RNase-free water were purchased from Invitrogen.

Random primers (diluted to 0.1 mg/ml) and dNTPs (diluted to 2.5 mM) were purchased from Amersham Biosciences. RNasin Plus Ribonuclease Inhibitor was purchased from

Promega (Madison, WI, USA). Taqman Master Mix and MicroAmp Fast Optical 96-

Well Reaction Plates were purchased from Applied Biosystems (Mulgrave, VIC, AUS). mRNA expression was measured using TaqMan Gene Expression Assays (Applied

78 Biosystems). Real-time quantitative PCR (RQ-PCR) was performed using an ABI

7900HT PCR machine purchased from Applied Biosystems.

2.5.1.2 cDNA synthesis mRNA samples were reverse transcribed to cDNA using MMLV reverse transcriptase,

which initiates the reverse transcription to cDNA of all mRNA molecules present in a

sample using the random primers method. mRNA samples were thawed on ice and 2 µg

of mRNA from each sample was prepared using DNase/RNase-free water. mRNA

samples were added to 500 ng of random primers and 0.25 µL of RNasin plus in a total

reaction volume of 5 µL made up with water in a 0.2 mL thin walled PCR tube. The

mixture was then incubated at 70ºC for 10 min to denature the secondary structure of

RNA then rapidly cooled on ice. Five µL of reverse transcriptase master mix (2 µL of

5X buffer, 1 µL of 0.1 M DTT, 1 µL of 2.5 mM dNTPs and 1 µL of 200 U/µL MMLV

reverse transcriptase) was added to the mRNA and incubated in a thermal cycler at 37ºC

for 1 hr and then 70ºC for 15 min to inactivate the enzyme. The mixture was made up to

50 ul with DNase/RNase-free water and the resultant cDNA concentration was assumed

to be 40 ng/µL.

2.5.1.3 Real-time qRT-PCR

PCR reactions were prepared in MicroAmp Fast Optical 96-Well Reaction Plates to

which 1 µL of synthesised cDNA was added along with 5 µL of TaqMan Master Mix,

consisting of 0.3 µL of 10 µM Elongation factor-1 α (EF-1α) forward primer, 0.3 µL of

10 µM EF-1α reverse primer, 0.4 µL of 10 µM EF-1α probe, 1 µL of 20X TaqMan gene

expression assay and 3 µL of DNAse/RNAse-free water. The reaction was then made to

10 ul with DNAse/RNAse-free water. Each plate was sealed and placed in ABI 7500HT

PCR machine and thermal cycling conditions set as 50°C for 2 min and 95°C for 10

min, followed by 40 cycles of 95°C for 15 sec and 60°C for 1 min. Ct values generated 79 for each gene of interest were then normalised to the internal loading control gene (EF-

1α), so the delta Ct can be calculated. The average delta Ct of triplicate repeats for each gene was calculated and used for measuring difference in gene expression between samples.

2.6 Statistical analysis

All data were analysed using GraphPad Prism software (version 6.03 and version 6.0g).

The non-parametric Mann-Whitney U test was used to determine whether relative differences between samples were significant for non-normally distributed data and results were expressed as the median (range). The Student t-test was used for data with normal distribution and results were expressed as the mean ± standard error of the mean

(SEM). The Fisher's exact test was used to test the differences in the proportions of censored/uncensored mice. The Kaplan-Meier data plots of EFS were compared by log- rank test. For all statistical tests, the level of significance was set at 0.05. Analysis of the sensitivity, specificity, positive and negative predictive values and 95% confidence intervals were performed using MedCalc Software available online at www.medcalc.org/calc/diagnostic_test.php.

2.7 Ethics

All experimental studies were approved by the Human Research Ethics Committees of the South Eastern Sydney Area Health Service and the University of New South Wales

(SESIAHS-HREC #10/114; UNSW HREC 10442) and the Animal Care and Ethics

Committee of the University of New South Wales (ACEC: 12/39A, 15/15B).

80 CHAPTER 3 PILOT STUDY: OPTIMISATION OF XENOGRAFT MODELS TO PREDICT OUTCOME IN B CELL PRECURSOR (BCP) INTERMEDIATE RISK (IR) ALL PATIENTS

81 3.1 Introduction

Children with acute lymphoblastic leukaemia (ALL) are stratified at diagnosis based on molecular/cytogenetic characteristics and their response to initial treatment to receive risk-adapted multi-agent chemotherapy. The majority of ALL patients are stratified as

IR and present with moderate levels of minimal residual disease (MRD) upon receiving induction therapy, although an unacceptably high proportion of these patients relapse

(Conter et al., 2010; Karsa et al., 2013). The lack of specific prognostic features makes it difficult to predict the response of IR ALL patients to treatment

Recent progress in xenograft research has showed the reliability of xenograft models to represent the clinical heterogeneity of ALL disease and to test hypotheses for translation of leukaemia research into the clinical setting. Research indicates that the development of xenograft models allows studying the effectiveness of many drug interventions on leukaemia cell populations to prioritise potential therapies for clinical trials. Xenografts derived from ALL patients also provide a good model to track changes in molecular pathways associated with drug responses, to understand the molecular determinants of drug resistance and to explore the difference in leukaemia cells among various ALL subtypes or even within different clonal populations (Jacoby et al., 2014; Meyer and

Debatin, 2011).

Further to this, it has been shown that xenografts derived from ALL patients who have an increased risk of relapse engraft quickly in immune-deficient mice (Meyer et al.,

2011; Schmitz et al., 2011). This finding indicates that xenografts may provide an accurate representation of ALL disease in humans and thus annotation of xenograft characteristics and patient outcomes could have clinical implications on identifying

82 ALL patients who are destined to relapse at an earlier stage of their diseases. Given the

clinical diversity among IR ALL patients, xenograft models could help to separate

between patients who have different outcomes yet were assigned to the same risk

category. Although several studies highlighted the relevance of engraftment of cells

derived from various risk subtypes of ALL patients with their clinical outcomes, the

reliability of xenografts to identify outcomes of IR ALL patients is uncertain and the

optimal experimental design for such an approach has not been published.

Investigations toward this goal should take into account that engraftment of ALL patient

samples could be affected by certain factors related to the ALL patient, the recipient

mouse and the experimental setup of transplantation procedures. The quality and source

of ALL patient samples are important determinants of ALL cell engraftment into mice

(Meyer and Debatin, 2011). Removal of ALL patient cells from the BM environment

and cryopreservation affect cell viability and therefore the success of engraftment of

freshly isolated ALL samples could be higher than frozen samples (Greystoke et al.,

2013). However, cryopreserved samples could be the only practical option for

retrospective analysis of the ability of xenografts to predict the outcomes of patients.

The level of residual murine immunity affects engraftment of leukaemia cells into

immunodeficent mice model. Transplantation of leukaemia cells into NSG mice, which

display greater levels of immunodeficiency compared to other mice strains, is more

supportive for cell engraftment than NOD/SCID mice (Agliano et al., 2008; Diamanti et

al., 2012). The tendency of ALL cells to engraft in immunodeficent mice is specifically

higher in young and female mice in comparison to old and male mice (Meyer and 83 Debatin, 2011). Generation of conditioned mice using total body irradiation (TBI) before transplantation of ALL cells has also been used by many researchers to improve the engraftment rates although it has a direct negative impact on cell homing to the BM in recipient mice (Spiegel et al., 2004).

Engraftment of ALL patient samples is also associated with routes of administration of leukaemia cells. Injection of patient samples into mice via the IV route is a routine procedure used to develop a systemic ALL disease in mice although cells injected via the tail vein have the potential to spread into various bodily organs which could allow fewer cells to reach into the BM niches. Recent reports credit direct injection of ALL patient cells into the femur bone as this allows for immediate placing of leukaemia cells into the BM niches (Schmitz et al., 2011). Thus, leukaemia cell expansion occurs in the

BM first and is then followed by dissemination of cells into extramedullary organs.

High engraftment efficiency and relevance of xenograft clonal populations to ALL patient disease have been shown when cells were injected via the IF method to establish xenografts (Rehe et al., 2013; Schmitz et al., 2011).

Leukaemia xenograft models have been extensively used in our program for many years and research from our group has showed that selection of ALL xenografts with the combination of vincristine, dexamethasone, and L-asparaginase chemotherapy regimen causes longer delays in progression of xenografts derived from long-term surviving patients compared to xenografts derived from patients who died of their disease

(Szymanska et al., 2012).

The ability of xenograft models to reflect the outcomes of ALL patients might be also influenced by the endpoint of monitoring used to infer the engraftment features of 84 patient samples (Meyer and Debatin, 2011). Currently, there is no agreement about the

ideal time point or level of leukaemia engraftment which could represent the patient

disease. Various levels and durations of engraftment have been used as readouts for

assessing the ability of patient samples to engraft in the immunodeficent mice . Our

group and others had previously defined the time to engraftment based on 1% or 25%

human CD45+ cells in mice PB to ascertain the clinical relevance of ALL xenografts to

the patients from whom they were derived. Meyer et al. (2011) showed that assessment

of the time to clinical manifestation of the disease (termed Time to Leukaemia, TTL) in

recipient mice can identify early relapse in ALL patients who were stratified into

various risk subtypes.

The possibility of establishing a model of systematic analysis of the influence of these

factors on engraftment of IR ALL patient samples would help to use the most efficient

xenografts model available to date to predict relapse in IR paediatric ALL soon after

diagnosis. In this study, I have decided first to design a Pilot Study for optimising the

most appropriate xenograft models which can predict outcome of IR BCP-ALL patients

according the engraftment of their patient cells in murine models. Specifically, the Pilot

Study will test the role of the following engraftment variables: (1) two mouse strains

with different levels of immunodeficiency (NOD/SCID versus NSG); (2) the site of

inoculation (IV versus IF); and, (3) the ability of an induction-type chemotherapy

regimen consisting of vincristine, dexamethasone and L-asparaginase (VXL) to improve

the predictive power of the current model for identifying relapses in IR ALL patients. In

this chapter, the engraftment properties of each patient sample will be characterised

based on the efficiency and speed of engraftment using different engraftment variables

85 and prediction of patient outcome will be discussed, prior to the Main Study in subsequent chapters.

3.2 Selection of IR ALL Patients for the Pilot study

Between 2002 and 2011 the ANZCHOG enrolled children with ALL on ANZCHOG

Study VIII protocol at Sydney Children’s Hospital and 5 other centres across Australia and New Zealand. All patients received chemotherapy treatment according to the

ANZCHOG protocol for a duration of two years (Karsa et al., 2013; Marshall et al.,

2013). Bone marrow biopsy specimens were harvested from ALL patients at time of diagnosis as part of the treatment protocol. Mononuclear cells were purified from these samples using Ficoll-Lympho Prep density gradient technique and cryopreserved in different sites including the tumour bank at CCI.

My associate supervisor, Dr Rosemary Sutton, and I have identified 4 IR BCP-ALL patients who were treated with standard chemotherapies under the ANZCHOG Study

VIII clinical trial and experienced different outcomes. These samples constitute two pairs of IR ALL patients who are matched based on clinical and genetic features, except that one patient from each pair relapsed early (defined as relapsed on therapy ≤ 25 months) while the other remains relapse-free for more than 60 months. The disease characteristics of the patients are shown in Table 3.1. In the first pair, there were two female patients aged 19 and 24 months at diagnosis who presented with favourable cytogenetic features, white blood cell count (WCC) less than 100 x109/L, and low level of MRD values at time point one after treatment initiation. In the second pair, there were two female patients aged 27 and 29 months at diagnosis who presented with favourable cytogenetic features, WCC more than 100 x109/L, and low level of MRD values at time point one. Each pair has patients who relapsed within 19 months of

86 diagnosis and another patient who maintained complete remission for longer than 60

months from diagnosis.

87

Table 3.1. Disease characteristic features of the two pairs of IR ALL patient

CR1 Xenograft Age at Dx WCC MRD level Current Duration ID Sample (Months) (109/L) TP1 (M) Status Patient Sex Cytogenetics (M) ID Pairs

-4 ALL-64 A5072 F 19 Hyperdiploidy 3.1 1x10 19 DOD 1st -4 ALL-65 A1795 F 24 Hyperdiploidy 9.8 1x10 > 91 CR1 Pair

-4 nd ALL-66 A1839 F 29 t(12;21) 135.8 1x10 19 Relapsed 2 -4 ALL-67 A4334 F 27 t(12;21) 166.9 1x10 > 64 CR1 Pair

F, female; Dx, diagnosis: WCC, white cell count: MRD, minimal residual disease; TP1, time point 1 (day 33); M, months;

DOD, dead of disease: CR1, first complete remission .

88 3.3 Experimental design of the Pilot Study

Cryopreserved samples from 4 IR ALL patients collected at the time of diagnosis were obtained from the CCI Tumour Bank. Samples were thawed and washed, and one million cells from each patient sample were inoculated into 8 groups of mice consisting of 4 mice per group. These 8 groups represent a 2 x 2 x 2 matrix assessing the following engraftment variables: (1) Injection of cells via the IV versus IF route; (2) Inoculation of

NSG versus NOD/SCID mice; and, (3) administration of a 2-week treatment of VXL or vehicle control via intraperitoneal (IP) injection commencing 2 weeks post inoculation.

The VXL protocol has been previously optimised to mimic paediatric ALL remission- induction therapy (Szymanska et al., 2012), and consists of vincristine (VCR) (0.15 mg/kg once every 7 days for 2 weeks), dexamethasone (DEX) (5 mg/kg Mon-Fri for 2 weeks) and L-asparaginase (ASP) (1000 IU/kg Mon-Fri for 2 weeks).

All four patient samples were tested for cell surface immunophenotype and showed strong positivity for human CD45. Therefore, leukaemia engraftment was monitored weekly based on the proportion of human CD45-positive cells in the murine PB as previously described (Liem et al., 2004). The minimum criterion for defining successful engraftment was 1% human CD45 cells in the PB. Monitoring commenced 2-3 weeks after inoculation of the patient samples and the progress of engraftment was calculated for each mouse until the endpoint determined by either signs of morbidity and distress, or reaching the maximum holding time approved by the animal ethics committee for this study. The time to engraftment was calculated from the day of inoculation to the time of an event, and expressed as event-free survival (EFS). Since this was a Pilot

Study, events were defined a priori as the date from which cells were inoculated into mice to the date when the proportion of human CD45+ cells in the PB reached either

89 1% or 25% human CD45+, or when animals reached Time To Leukaemia (TTL). TTL was defined as clear manifestation of clinical signs of leukaemic infiltration into mouse body organs (spleen enlargement, weight loss, lethargy and ruffled fur, hunched posture). The median EFS using all of these event definitions were compared between different xenotransplantation strategies and also compared according to patient outcomes. The leukaemia growth delay (LGD) was calculated as the difference between the median EFS of the VXL-treated mice compared with that of the non-treated groups.

Samples were also taken from the BM liver, kidney, lung and brain to assess for leukaemic infiltration. Figure 3.1 shows the experimental design and process used to establish xenografts for the pilot study.

90

NOD/SCID +/- 2 weeks Weekly monitoring or NSG of VXL of engraftment using FACS 4 Samples One million cells of IR ALL

patients injected IV or IF Human CD45 % Mouse CD45 % 8 Model conditions x 4 Mice/Sample Progression of 5 NSG IF VXL 1 NSG IF engraftment 2 NSG IV 6 NSG IV VXL 3 NOD/SCID IF 7 NOD/SCID IF VXL 8 NOD/SCID IV VXL 4 NOD/SCID IV Mice euthanased at TTL Analysis of EFS

In vitro Cryopreservation experiments

Purification of mononuclear cells

Figure 3.1. Schematic diagram of the experimental plan used to establish xenografts from the IR ALL cohort for the Pilot Study . Each patient sample was inoculated into immunodeficient (NOD/SCID or NSG) mice and the engraftment was monitored based on the proportion of human CD45+ cells in the peripheral blood. IR, Intermediate Risk; NOD/SCID, Nonobese Diabetic/Severe Combined Immunodeficiency; NSG, NOD/SCID common cytokine receptor gamma chain–/–; IF, intrafemoral; IV, intravenous; VXL, combination of vincristine, dexamethasone and L- asparaginase; FACS, fluorescence-activated cell sorting machine; EFS, event-free survival.

91 3.4 Establishment and characterisation of IR ALL PDXs

3.4.1 ALL-64

This xenograft was derived from a patient who relapsed early after treatment initiation

(A5072). ALL-64 demonstrated an early appearance of leukaemia in the PB of all NSG mice that were inoculated via both the IF and IV routes (Figure 3.2). However, the engraftment levels fluctuated for several weeks before the proportion of human CD45+ cells increased gradually and reached high levels in the PB of most of the non-VXL- treated mice. Engraftment of ALL-64 in NSG mice that received VXL treatment also showed early appearance and the same pattern of fluctuation of human CD45+ cells in the PB, but levels barely reached 15% and were maintained at a lower level with no evidence of signs related to leukaemia morbidity until mice had reached the maximum holding time (Figure 3.2). In contrast, ALL-64 did not show any evidence of engraftment in the PB of VXL treated and non-treated NOD/SCID mice over the monitoring period.

Assessment of organ infiltration in NSG mice revealed high levels of human CD45+ cells in the PB, BM, spleen and liver, with lower levels of cellular infiltration into the kidneys, lungs and brain at the time of harvest (Figure 3.3). Evaluation of the organ infiltration by ALL-64 cells in VXL-treated mice revealed that with the low PB levels there was moderate engraftment in the BM, and minimal evidence of leukaemia in spleen but none of other body organs (liver, lung, kidney and brain) showed evidence of leukaemia infiltration (Figure 3.3). As expected, no human CD45+ cells were detected in the PB or body organs of VXL treated and non-treated NOD/SCID mice at time of harvest.

92

A

1 0 0 A L L -6 4 N S G IF

A L L -6 4 N S G IF V X L 8 0 B

P

n 6 0 i

+ 5

4

D 4 0 C u h

% 2 0

0 0 2 5 5 0 7 5 1 0 0 1 2 5 1 5 0 1 7 5 2 0 0 2 2 5 2 5 0 2 7 5 D a y s p o st in o c u la tio n

B 1 0 0 A L L -6 4 N S G IV 8 0 A L L -6 4 N S G IV V X L B P

n

i 6 0

+ 5 4

D 4 0 C u h

% 2 0

0

0 2 5 5 0 7 5 1 0 0 1 2 5 1 5 0 1 7 5 2 0 0 2 2 5 2 5 0 2 7 5

D a y s p o st in o c u la tio n

Figure 3.2. Patient sample A5072 engrafted in NSG mice to establish the ALL-64 xenograft. Engraftment of A5072 in VXL-treated and non-treated NSG mice inoculated via the IF (A) or IV (B) routes. Each line indicates the percentage of human CD45+ cells in the PB of a single mouse. Dashed black lines represent ALL-64 engrafted into mice without selection with VXL treatment while the solid red lines represent VXL- treated mice. The solid black squares indicate the VXL treatment period.

93

A B 1 0 0

1 0 0 )

l 8 0 a t ) l 8 0 o a t t

f o t o

6 0 f o

% 6 0 (

% + (

5 +

4 4 0 5 4

D 4 0 D C u C h u 2 0 h 2 0

0 0

r r y r r r g y n B n u r g in B n u u e n e i u e n e P e v n a P e m v u n a e m m i u r e m i r l e L L id l fe L L id B p f fe B p fe S t K t K h ft S h ft g e g e i i R L R L

C

5 0 D 5 0

) 4 0 l

a 4 0 ) t l o a t t

o f t

o f 3 0 o

3 0 %

( %

(

+ + 5 5

4 2 0 4 2 0 D D C

C u u h h 1 0 1 0

0 0 r r B n r g y n r r r g y n u u e n e i B n u u e n e i P e v n a P e v n a e m m i u r e m m i u r l e L L id l e L L id p f fe B p f fe B S t K S t t K h ft h f g e g e i i L R L R

Figure 3.3. Infiltration of ALL-64 into organs of NSG mice at day of harvest. (A) NSG IF, (B) NSG IV, (C) NSG IF VXL treated, and (D) NSG IV VXL treated. PB, peripheral blood. Bars represent the median.

94 3.4.2 ALL-65

This xenograft was derived from a BM sample of an ALL patient who has been in complete remission for more than 7 years after diagnosis (A1795) and is paired with

ALL-64. In NSG mice, ALL-65 demonstrated a prolonged time for appearance of human CD45+ cells in the PB, but levels continued to increase over the monitoring period in all engrafted mice (Figure 3.4). Regardless of the route of inoculation, all non- treated NOD/SCID mice also demonstrated a slow engraftment rate and constant increase in the level of human CD45+ cells over the monitoring period (Figure 3.4).

Most of the NSG mice that received no treatment showed high levels of human CD45+ cells in the PB, spleen, BM and liver (Figure 3.5). However, the levels of human CD45 infiltration into lungs, kidneys and brain were lower than those infiltrating the spleen and BM at the time of harvest. In NOD/SCID mice, comparable high levels of human

CD45+ cells to that observed in NSG mice were detected in all organs of one mouse inoculated via the IF route except the brain, and moderate to high levels in the PB, BM, and spleen of three engrafted NOD/SCID mice which were inoculated via the IV route

(Figure 3.5).

Engraftment was not apparent in the PB of any of the NSG and NOD/SCID mice following treatment with VXL over the monitoring period, nor detected in their body organs at the time of harvest (Figure 3.4).

95

B A

1 0 0 1 0 0 A L L -6 5 N S G IF A L L -6 5 N S G IV A L L -65 N S G IF V X L A L L -65 N S G IV V X L 8 0 8 0 B B P P

n n 6 0 6 0 i i

+ 5

5 4 4 D D 4 0 4 0 C C u u h H

% 2 0 % 2 0

0 0

0 2 5 5 0 7 5 1 0 0 1 2 5 1 5 0 1 7 5 2 0 0 2 2 5 2 5 0 2 7 5 0 2 5 5 0 7 5 1 0 0 1 2 5 1 5 0 1 7 5 2 0 0 2 2 5 2 5 0 2 7 5 3 0 0 D a y s p o st in o c u la tio n D a y s p o st in o c u la tio n

C D 1 0 0 A L L -6 5 N O D /S C ID IF 1 0 0 A L L -6 5 N O D /S C ID IV 8 0 A L L -6 5 N O D /S C ID IF V X L 8 0 A L L -6 5 N O D /S C ID IV V X L B B P

P n

i n 6 0 i

+ 6 0 5 + 4 5 4 D 4 0 D C

4 0 C

u

u H H

% 2 0 % 2 0 0 0 0 2 5 5 0 7 5 1 0 0 1 2 5 1 5 0 1 7 5 2 0 0 2 2 5 2 5 0 2 7 5 3 0 0 0 2 5 5 0 7 5 1 0 0 1 2 5 1 5 0 1 7 5 2 0 0 2 2 5 2 5 0 2 7 5 3 0 0 D a y s p o st in o c u la tio n D a y s p o st in o c u la tio n

Figure 3.4. Engraftment of patient sample A1795 inoculated into VXL treated and non-treated mice to establish ALL-65 xenograft. Engraftment patterns for ALL-65 in NSG (A, B) or NOD/SCID (C, D) mice inoculated via the IF (A, C) or IV (B, D) routes. Dashed black lines represent ALL-65 established in mice without treatment with VXL drugs while the solid red lines represent the VXL-treated mice. The solid black squares indicate the VXL treatment period.

96

A B

1 0 0 1 0 0

)

l 8 0 )

a 8 0 l t a o t t

o t f

o f

6 0 o

6 0 % ( %

( +

+ 5 5 4 4 0

4 4 0 D D C C u u h

2 0 h 2 0 0 r r 0 B n r g y n e u u e n e i P e m iv u n ra r r r g y n l e m d B n u u e n e i p f e L L i B P e v n a f e m m i u r S t t K l e L L id h f p f fe B g e S t t K i L h f R ig e R L

C

D 1 0 0 1 0 0

)

l 8 0 a ) t l o a

t 8 0 t

f o t o

6 0 f o %

( 6 0

% +

(

5 + 4 4 0 5 D 4 4 0 C

D u C h

2 0 u h 2 0

0 0 r r r g y n B n u u e n e i P e v n a r r r g y n le m m i u r B n u u e n e i e e L L id B P e v n a p f f e m m i u r S t K l e L L id h ft p f fe B g e S t t K i L h f R ig e R L

Figure 3.5. Infiltration of ALL-65 into organs of non-drug treated mice at day of harvest. (A) NSG IF (B) NSG IV (C) NOD/SCID IF and (D) NOD/SCID IV. PB, peripheral blood. Bars represent the median.

97 3.4.3 ALL-66

This xenograft was derived from a BM sample of an ALL patient who relapsed early after treatment initiation (A1839). The engraftment of this patient sample was more robust compared to other xenografts I have established for the pilot study. ALL-66 exhibited rapid and uniform appearance of human CD45+ cells in the PB of all mice

(Figure 3.6). VXL treatment of NSG mice did not appreciably delay the appearance of leukaemia in the PB or change the pattern of growth over the monitoring period. ALL-

66 also demonstrated high levels of human CD45+ cells in the PB of all NOD/SCID mice over the monitoring period. NOD/SCID mice that received VXL treatment also maintained comparable patterns of growth to that observed in the non-treated

NOD/SCID mice (Figure 3.6).

Regardless of the transplantation conditions used to establish ALL-66 in NSG mice and drug treatment status, there was extensive infiltration of human CD45+ cells into all mouse organs but slightly lower levels were detected in the lungs (Figure 3.7).

Likewise, assessment of human CD45+ cells in NOD/SCID mouse organs revealed high levels in the PB, BM, liver, kidneys and brain but moderate levels in the lungs (Figure

3.8). The pattern of human CD45+ cell infiltration in NOD/SCID mice that received

VXL treatment was consistent with that observed in non-treated mice except that moderate high levels were detected in the kidneys and brain of mice inoculated via the

IF route and kidneys of mice inoculated via the IV route. Leukaemia infiltration was also observed in the lymph nodes of NSG mice, and a representative sample of an enlarged lymph node revealed a high percentage of human CD45+ cells (Figure 3.9).

98

A B 100 ALL-66 NSG IF 100 ALL-66 NSG IV ALL-66 NSG IF VXL ALL-66 NSG IV VXL 75 75 in PB in + in PB in

50 + 50

25 %huCD45 %huCD45 25

0 0 0 25 50 75 100 125 150 175 200 225 250 275 300 0 25 50 75 100 125 150 175 200 225 250 275 300

Days post inoculation Days post inoculation

C D 100 100 ALL-66 NOD/SCID IF ALL-66 NOD/SCID IV ALL-66 NOD/SCID IF ALL-66 NOD/SCID IV VXL 75 75 VXL in PB in in PB in + 50 + 50

%huCD45 25 25 %huCD45

0 0

0 25 50 75 100 125 150 175 200 225 250 275 300 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Days post inoculation

Figure 3.6. Engraftment of patient sample A1839 inoculated into mice to establish ALL-66 in the presence or absence of VXL selection. Engraftment patterns for ALL- 66 in NSG (A, B) or NOD/SCID (C, D) mice inoculated via the IF (A, C) or IV (B, D) routes. Dashed black lines represent ALL-66 established in mice without treatment with VXL drugs while the solid red lines represent the VXL-treated mice. The solid black squares indicate the VXL treatment period.

99

A B 1 0 0 1 0 0

) l

8 0 )

l 8 0 a t a t o t o

t

f f o

6 0 o 6 0 % % ( (

+ + 5 5

4 4 0 4 4 0

D D C C u u h

h 2 0 2 0

0 0 r r B n u r e g y in r r r g y n e u n e B n u u e n e i P e m iv u n a P e v n a l m d r e m m i u r e e L L i B l e L L id p f f p f fe B S t t K S t K h f h ft g e g e i L i R R L

C D 1 0 0 1 0 0

) l

a 8 0 ) t

l 8 0 o

a t t

o f t

o f 6 0 o

6 0 % ( %

( +

+ 5

4 0 5 4 4 4 0 D D C C u u

h 2 0 h 2 0

0 0 r r y B n u r e g in e u v n e a r r r g y n P e m i u n r B n u u e n e i l e m d P e v n a p f e L L i B le m m i u r f e e L L id S t t K p f f B h f S t t K g e h f i L ig e R R L

Figure 3.7. Infiltration of NSG mouse organs with ALL-66 at the time of harvest (A) NSG IF (B) NSG IV (C) NSG IF VXL and (D) NOD/SCID IV VXL. PB, peripheral blood. Bars represent the median.

100

A B

1 0 0 1 0 0 )

l ) a 8 0 l

t 8 0 a t o t o

t

f f o o

6 0 6 0 % % ( (

+ + 5 5

4 0 4 4 0 4 D

D C C u h u 2 0 h 2 0

0 0 r r B n r g y n r r e u u e n e i n r g y n P iv u n a B u u e n e i le m m d r P e v n a p fe e L L i B e m m i u r f l e L L id S t t K p f fe B h f S t t K ig e h f R L ig e R L

C D 1 0 0 1 0 0 )

l 8 0 a ) t

l 8 0 o a t t

o f t

o f 6 0 o

6 0 % ( %

( +

+ 5

5 4 4 0

4 4 0 D D C C u u

h 2 0 h 2 0

0 0 r r r g y n r r B n u u e n e i n r g y n P e v n a B u u e n e i le m m i u r P e v n a e e L L id B e m m i u r p f f l e L L id S t t K p f fe B h f S t t K ig e h f R L ig e R L

Figure 3.8. Infiltration of NOD/SCID mice organs with ALL-66 at day of harvest. (A) NOD/SCID IF (B) NOD/SCID IV (C) NOD/SCID IF VXL and (D) NOD/SCID IV.VXL. PB, peripheral blood. Bars represent median.

101

Enlarged Lymph Nodes

Enlarged spleen

Figure 3.9. Image of ALL-66 established in NSG mice showing splenomegaly and lymphadenopathy. Enlarged spleen and lymph nodes marked with arrows.

102 3.4.4 ALL-67

This xenograft was derived from an ALL patient who has been in complete remission for more than 5 years (A4334). In NSG mice, ALL-67 exhibited a gradual increase with some fluctuation in the levels of human CD45+ cells over the monitoring period in all non-VXL treated mice (Figure 3.10A and B). NSG mice that received VXL treatment exhibited a delay in appearance of cells in the PB and consistently maintained the gradual increases in levels compared to the non-treated groups. Similarly, the pattern of human CD45+ cell engraftment in NOD/SCID mice was consistent with what I observed in NSG mice although more time was required for cells to appear in the PB of

NOD/SCID mice (Figure 3.10C and D). Further delay in the growth of cells in

NOD/SCID mice was observed upon VXL treatment.

Assessment of organ infiltration revealed high levels of human CD45+ cells in all organs of non-treated NSG mice except variable levels were detected in the brain

(Figure 3.11A and B). NSG mice inoculated via the IF route that received VXL treatment showed extensive infiltration in all organs except the lungs, kidneys and brain in both of the engrafted mice (Figure 3.11C), whereas high levels were detected in all organs except the brain of mice inoculated via the IV route (Figure 3.11D). ALL-67 inoculated into NOD/SCID mice via the IF route showed high levels of human CD45+ cells in the BM and spleen, moderately high levels in the PB and liver, and lower levels in the lungs, kidneys and brain (Figure 3.12A). In NOD/SCID mice inoculated with patient cells via the IV route, high levels of human CD45+ cells were detected in all organs except moderate levels in the lungs and brain (Figure 3.12B). Organ infiltration data of NOD/SCID mice inoculated via the IF route and treated with VXL showed high levels in the BM and spleen, lower levels in the PB and liver but no cells infiltrated

103 other organs (Figure 3.12C). Extensive infiltration of ALL-67 cells into organs of

NOD/SCID mice inoculated via the IV route and treated with VXL treatment were detected in the spleen and BM and moderate levels in other body organs except brain

(Figure 3.12D).

104

A B 100 100 ALL-67 NSG IF ALL-67 NSG IV ALL-67 NSG IF VXL ALL-67 NSG IV VXL 75 75 in PB in in PB in + 50 + 50

%huCD45 25 25 % huCD45

0 0 0 25 50 75 100 125 150 175 200 225 250 275 300 0 25 50 75 100 125 150 175 200 225 250 275 300 Days post inoculation Days post inoculation

C D 100 100 ALL-67 NOD/SCID IF ALL-67 NOD/SCID IV ALL-67 NOD/SCID IF VXL ALL-67 NOD/SCID IV VXL 80 80

60 60 in PB in in PB in + +

40 40 %huCD45 20 % huCD45 20

0 0

0 25 50 75 100 125 150 175 200 225 250 275 300 0 25 50 75 100 125 150 175 200 225 250 275 300 Days post inoculation Days post inoculation Figure 3.10. Engraftment of patient sample A4334 inoculated into VXL-treated and non-treated mice to establish ALL-67 xenograft. Engraftment patterns for ALL- 67 in NSG (A, B) or NOD/SCID (C, D) mice inoculated via the IF (A, C) or IV (B, D) routes. Dashed black lines represent ALL-67 established in mice without treatment with VXL drugs while the solid red lines represent the VXL-treated mice. The solid black squares indicate the VXL treatment period.

105

A B 1 0 0 1 0 0 ) l a ) l t 8 0 8 0

a o t t

o t f

f o

6 0 o 6 0 % % (

(

+

+ 5 4 0 5 4

4 4 0 D D C C u u

2 0 h h 2 0

0 0 r n r r g y n r r r g y n B u u e n e i B n u u e n e i P e v a P e v n a e m i u n r e m m i u r l e m d l e L L id p f e L L i B p f fe B f S t K S t t K h ft h f g e ig e i L R L R

C D 1 0 0 1 0 0

) l ) l 8 0 a t a 8 0 t o o t t

f f o o

6 0 6 0 % % ( (

+ + 5 5 4 4 0 4 0 4 D D C u C h 2 0 u 2 0 h

0 0 r r B n r g y n r r e u u e n e i B n u r e g y n P e m iv u n ra e u n e i l e m d P iv u n a p f e L L i B le m m d r S t f K p fe e L L i B h ft f g e S t t K i L h f R ig e R L

Figure 3.11. Infiltration of NSG mice organs with ALL-67 at day of harvest. (A) NSG IF (B) NSG IV (C) NSG IF VXL and (D) NOD/SCID IV VXL. PB, peripheral blood. Bars represent the median.

106

A

B 1 0 0 1 0 0 ) l

a 8 0 t ) l o 8 0 a t t

f o t o

6 0 f o

% 6 0 (

% ( +

5 4 0 + 4 5

4 4 0 D

D C C u u

h 2 0 h 2 0 0 0 r r B n u r e g y in r r e u n e B n r g y n P iv u n a e u u e n e i le m m r P iv u n a e e L L id B le m m d r p f f p fe e L L i B S t t K S t f K h f h ft g e g e i L i R R L

C D 1 0 0 1 0 0

)

8 0 ) l l a t a 8 0 t o t o

t f

f o

6 0 o 6 0 % ( %

( +

5 + 4

4 0 5 4 0 4 D C D u C h 2 0 u h 2 0

0 0

r r r g y n r r r g y n B n u u e n e i B n u e e i P e v n a P e u v n a e m m i u r e m i u n r l e L L id l e m d p f fe B p f e L L i B S t K f h ft S t t K g e h f i L ig e R R L

Figure 3.12. Infiltration of NOD/SCID mice organs with ALL-67 at harvest. (A) NOD/SCID IF (B) NOD/SCID IV (C) NOD/SCID IF VXL and (D) NOD/SCID IV VXL. PB, peripheral blood. Bars represent the median.

107 3.5 Comparison of engraftment efficiency between IR ALL xenografts established using various engraftment conditions

Analysis of the overall efficiency of engraftment between samples of each pair of IR

ALL patients revealed that samples of patients who relapsed early after treatment had better engraftment efficiency compared to that for their paired samples of non-relapsed cases, as detailed below. Of the mice inoculated with the first pair of patient samples, there were 13/31 (41.9%) mice engrafted with ALL-64 whereas ALL-65 was able to establish in 10/32 (28.1%) mice (P=0.4390 Fisher's exact test) (Table 3.2). The efficiency of engraftment in the second pair of IR ALL patients was much higher than that for the first pair of patients. ALL-66 established in 32/32 (100%) mice whereas

ALL-67 established in 28/32 (87.5%) mice (P=0.1132 Fisher's exact test).

I next analysed the efficiency of engraftment between different transplantation conditions used to establish the panel of IR ALL xenografts. The NSG mouse strain was found to be highly receptive for engraftment of samples from IR ALL patients compared to the NOD/SCID mice. The total number of engrafted mice inoculated via the IV route, was 15/16 (93.8 %) in NSG mice compared to 11/16 (68.8 %) engrafted in the NOD/SCID mouse strain (P=0.1719 Fisher's exact test). Similarly, the total number of engrafted mice, inoculated via the IF route, was higher in NSG mice compared to

NOD/SCID mice 14/16 (87.5%) versus 9/16 (56.3%) (P=0.1134 Fisher's exact test).

This high engraftment efficiency was also observed when the total number of engrafted mice that received VXL treatment were compared between both mouse strains (Table

3.2).

108 To assess the effect of the method of inoculation on the engraftment efficiency, the total number of engrafted mice from each mouse strain was compared between the IV and IF inoculated mice. Regardless of the mouse strain and treatment, there was higher engraftment efficiency in mice inoculated via the IV route compared to those inoculated via the IF route with the exception of similar engraftment between NOD/SCID mice that received VXL treatment (Table 3.2).

Table 3.2. Summary of the engraftment efficiency for the IR ALL xenograft panel established using different engraftment conditions

Number of Mice Engrafted/Inoculated

Engraftment First Pair Second Pair Total Condition ALL-64 ALL-65 ALL-66 ALL-67 engrafted (Rel) (CR1) (Rel) (CR1) /model (%)

NSG IV 4/4 3/4 4/4 4/4 15/16 (93.8%)

NSG IF 3/4 3/4 4/4 4/4 14/16 (87.5%)

NOD/SCID IV 0/4 3/4 4/4 4/4 11/16 (68.8%)

NOD/SCID IF 0/4 1/4 4/4 4/4 9/16 (56.3%) NSG IV VXL 4/4 0/4 4/4 4/4 12/16 (75%) NSG IF VXL 2/3* 0/4 4/4 2/4 8/15 (53.3%) NOD/SCID IV VXL 0/4 0/4 4/4 3/4 7/16 (46.6%) NOD/SCID IF VXL 0 /4 0/4 4/4 3/4 7/16 (46.6 %) Total engrafted 13 /31 10 /32 32 /32 28/32 /xenograft (%) (41.9%) (28.9%) (100%) (87.5%) Rel, relapse; CR1, first complete remission. *One mouse was excluded due to heart abnormality at an early stage of the experiment before treatment initiation.

109 3.6 Comparison of rate of engraftment between IR ALL xenografts established using various engraftment conditions

Difference in engraftment kinetics between various ALL patients could depend on the ability of leukaemia cells to engraft, proliferate and disseminate into the blood in recipient mice. Although engraftment of ALL patient samples can be relatively quicker in patients who presented with adverse prognosis, establishment and growth of xenograft cells could be influenced by heterogeneity of leukaemia populations in recipient mice which may be affected by the xenotransplantation strategies used to establish ALL xenografts.

Because the goal of this experiment was to optimise the most appropriate engraftment condition which allows better engraftment efficiency and prediction of relapse in the IR

ALL patients, I sought first to examine the difference in speed of engraftment between various transplantation conditions used in this study. The difference in engraftment kinetics was characterised based on measuring the time required for appearance and increase in the proportion of human CD45+ cells in the PB at specific levels (1% and

25%) and manifestation of TTL features as these parameters represent time points of the lag and exponential phases of ALL xenograft growth. In this and the next section, the difference in engraftment efficiency and in the ability to improve relapse prediction are characterised based on detailed analysis of engraftment at 1% human CD45+ cells in the

PB. The difference in engraftment features based on time to 25% human CD45+ cells and TTL are described in section 3.8 and 3.9 consecutively.

3.6.1 NSG versus NOD/SCID

In order to investigate the influence of mouse strain on engraftment of ALL patient samples, median times to event (defined as 1% human CD45+ cells in the PB) were 110 compared between NSG and NOD/SCID mice which were inoculated via the same route to establish each xenograft. NSG recipients were seen to have the shortest median time to engraftment in all xenografts established for this study compared to NOD/SCID mice (Figure 3.13A). When engraftment was compared between both mouse strains inoculated via the IV route, significant differences in the median time to 1% were observed in ALL-64, ALL-66 and ALL-67 (Figure 3.13A). ALL 64 reached event in 64

(range 53.3-72.5) days when inoculated into NSG mice, whereas engraftment of cells from the same source was not detected in NOD/SCID mice within 266 days of monitoring. ALL-65 reached event after 153.2 (range 112.5- >241) days in NSG mice whereas the EFS in the NOD/SCID mice was 168.5 (150.5-280) days. Similarly, rapid engraftment was observed for ALL-66 and ALL-67 in NSG [43.1 days, range (42.8-

43.1) and 71.1 days range (67.9-71.9) respectively compared to that in NOD/SCID mice

47.8 days, range (47.1-49.4) and 83.5 days, range (78.6-92.2) respectively] for both xenografts (Figure 3.13A).

The difference in speed of engraftment between xenografts inoculated into NSG and

NOD/SCID via IF inoculation was also comparable to that observed between both mouse strains when cells were inoculated via the IV route although a statistically significant difference was only achieved in ALL-64 (Figure 3.13B). These findings indicate that the NSG mouse strain allows better growth of IR ALL patient cells.

However, the critical aspect of favouring one mouse strain over the other in this study will be also based on its utility to predict patient outcomes which will be discussed in

Section 3.7.

111

A 1 0 0 N S G IV

A L L -64 8 0 A L L -65 A L L -66

l A L L -67 a v i

v 6 0 N O D /S C ID IV r u s

t A L L -64 n

e A L L -65 c r 4 0 A L L -66 e

P A L L -67 2 0

0 0 2 5 5 0 7 5 1 0 0 1 2 5 1 5 0 1 7 5 2 3 0 2 8 0

D a y s p o st in o c u la tio n Median EFS (Range) P value NSG IV NOD/SCID IV (log-rank)

ALL-64 64 (53.3-72.5) >266 0.0067

ALL-65 153.2 (112.5- >241) 168.5 (150.5->280) 0.8860

ALL-66 43.1 (42.8-43.1) 47.8 (47.1-49.4) 0.0069

ALL-67 71.1 (67.9-71.9) 83.5 (78.6-92.2) 0.0067

B 1 0 0 N S G I.F A L L -64 8 0 A L L -65 A L L -66 l a A L L -67 v i

v 6 0 r N O D /S C ID I.F u s

t A L L -64 n

e A L L -65 c

r 4 0 e A L L -66 P A L L -67 2 0

0 0 2 5 5 0 7 5 1 0 0 1 2 5 1 5 0 1 7 5 2 0 0 2 2 0 2 8 0 D a y s p o st in o c u la tio n Median EFS (Range)

P value NSG IF NOD/SCID IF (log- rank) ALL-64 63.2 (56- >280) >266 0.0401 ALL-65 133.4 (112.3- >280) >280 (194.6->280) N/A ALL-66 48.9 (44-68.4) 59.1 (51.1-89.4) 0.3420 ALL-67 83.9 (71.2-116.7) 104.1 (86-153.8) 0.3237

112 Figure 3.13. Graphs of individual engraftment conditions showing how each patient sample engrafted at 1% human CD45+ in NSG mice compared to NOD/SCID mice. Kaplan-Meier survival curves show the EFS of NSG and NOD/SCID mice inoculated with patient samples via the IV (A) or IF (B) route. The table below each graph shows the median (range) time to 1% when inoculated with patient samples using different engraftment conditions. The log-rank test was used to calculate the difference in engraftment.

113 3.6.2 IV versus IF inoculation

To assess the effect of inoculation route on engraftment of cells from IR ALL patients, the median EFS was compared between each xenograft established via the IV and IF routes in each mouse strain. In general, the growth characteristic of ALL-64 was relatively comparable between mice inoculated via both routes and most of other IR

ALL xenografts established via IV route engrafted earlier than the IF inoculated groups.

As shown in Figure 3.14A, the median EFS in NSG mice inoculated with patient cells to establish ALL-64 was not significantly different from that obtained when mice were inoculated with cells via the IF route [64 days, range (53.3-72.5)] versus 63.2 days, range (56 ->280) (P= 0.6993 by log-rank test)]. Conversely, ALL-65 reached event in

133.4 (112.3 - > 280) days when patient cells were inoculated into NSG mice via the IF route whereas longer time to event [153.2 (112.5- >241) days] was observed in the IV inoculated group although the difference was not significant (P= 0.7779 by log-rank test). Nevertheless, the IF inoculation did not accelerate the time required for leukaemia to engraft in NSG mice when patient cells were injected into mice to establish ALL-66 and ALL-67 (Figure 3.14A). Similarly, the IF inoculation route did not improve the speed of engraftment in NOD/SCID mice and conversely a clear trend toward quicker engraftment of mice inoculated via the IV route was observed (Figure 3.14B).

114

A N S G IV 1 0 0 A L L -64 A L L -65 A L L -66 8 0 A L L -67 N S G IF l a v

i A L L -64

v 6 0 r A L L -65 u s

A L L -66 t

n A L L -67 e c

r 4 0 e

P

2 0

0 0 2 5 5 0 7 5 1 0 0 1 2 5 1 5 0 1 7 5 2 0 0 2 2 0 2 8 0

D a y s p o st in o c u la tio n Median EFS (Range) P value NSG IV NSG IF (log-rank)

ALL-64 64 (53.3-72.5) 63.2 (56 - >280) 0.6993

ALL-65 153.2 (112.5- >241) 133.4 (112.3- >280) 0.7794

ALL-66 43.1 (42.8-43.1) 48.9 (44-68.4) 0.0069

ALL-67 71.1 (67.9-71.9) 83.9 (71.2-116.7) 0.0558

B 1 0 0 N O D /S C ID IV A L L -64 A L L -65 8 0 A L L -66 A L L -67 l a

v N O D /S C ID IF i

v 6 0 r A L L -64 u

s

t A L L -65 n

e A L L -66 c

r 4 0

e A L L -67

P

2 0

0 0 2 5 5 0 7 5 1 0 0 1 2 5 1 5 0 1 7 5 2 0 0 2 2 0 2 8 0

D a y s p o st in o c u la tio n Median EFS (Range)

P value NOD/SCID IV NOD/SCID IF (log-rank) ALL-64 >266 >266 N/A ALL-65 168.5 (150.5->280) >280 (194.6->280) N/A ALL-66 47.9 (47.1-49.4) 59.1 (51.1-89.4) 0.0069 ALL-67 83.5 (78.6-92.2) 104.1 (86.5- 153.8) 0.0510

115 Figure 3.14. Graphs of individual engraftment conditions showing mouse EFS when mice were inoculated via IV and IF routes. Kaplan-Meier survival curves show the EFS of NSG mice inoculated with patient samples via the IF and IV routes (A) or the EFS of NOD/SCID mice inoculated with patient samples via the IF and IV routes (B). The table below each graph shows the median (range) time to 1% when inoculated with patient samples using different engraftment conditions. The log-rank test was used to calculate the difference in engraftment.

116 3.7 Assessing the capability of different transplantation conditions to improve prediction of outcomes in IR ALL patients

A short length of complete remission post first induction treatment of ALL patients with chemotherapy is a major predictive marker in identifying relapsed paediatric ALL patients with poor overall survival (Nguyen et al., 2008). As the main focus of this study was to establish a preclinical model that differentiates between IR ALL patients who demonstrated relapse after a short remission period from those who maintained longer complete remission after induction (CR1) based on engraftment features of their samples, I have compared the extent to which the engraftment features of each transplantation condition could distinguish between xenografts established from patients with different outcomes.

The analysis was based on calculating the difference in engraftment between xenografts in each pair and then calculating the overall difference in median EFS and the difference in first mouse to reach 1% human CD45+ cells in the PB between the xenografts derived from the relapse samples (ALL-64 or ALL-66) and earliest and median rate of engraftment in xenografts derived from CR1 samples (ALL-65 or ALL-67).

Furthermore, the difference in response to in vivo VXL treatment was compared between both groups. The rational for testing the difference in EFS of the first engrafted mouse is the hypothesis that early engraftment of leukaemia cells represent growth of an aggressive leukaemia sub-clone which gets selected either due to engraftment of patient samples and/or the selective pressure from the VXL treatment in immunodeficient mice.

117 3.7.1 Stratification of IR ALL xenografts based on engraftment in NSG mice at time to 1% human CD45+

A comparison of engraftment of leukaemia cells inoculated into NSG mice via the IF route between xenografts derived from patients who relapsed and those who remain in

CR1 revealed that the median EFS was somewhat shorter for ALL-64 compared to

ALL-65 although the difference was not statistically significant [63.2 (56 - >280) versus

133.4 (112.3- >280) days, P=0.4000 log rank-test] (Figure 3.15A, Table 3.3). However, a significant difference was observed when ALL-66 was compared to ALL-67 [48.9

(44-68.4) versus 83.9 (71.2-116.7) days, P=0.0067 log-rank test]. This transplantation condition resulted in 20.7 days difference in median EFS and 15.2 days difference in the first mouse to reach event between xenografts derived from patients who relapsed and those who remain in CR1 (Figure 3.15A).

Inoculation of samples into NSG mice via the IV route also resulted in faster engraftment of ALL-64 compared to ALL-65 [64 (53.3-72.5) versus 153.2 (112.5- >

241) days, P= 0.0067 log-rank test] as well as in ALL-66 when compared to ALL-67

[43.1 (42.8-43.1) versus 71.1 (67.9-71.9) days, P=0.0069 log-rank test]. This transplantation condition allowed 7.1 days difference in median EFS and 14.6 days difference in first mouse to reach event between samples from patients who relapsed compared with those in CR1 samples (Figure 3.15B, Table 3.3).

Although the inoculation of IR ALL patient samples into NSG mice via either IF or IV routes showed a trend toward quicker engraftment of samples of early relapsed cases, these differences may not be enough to provide stringent criteria to maintain unbiased separation between the engraftment features of IR ALL patients who relapsed from those of non-relapsed patients when expanded into a larger cohort of patients. 118 Substantial improvement in stratifying IR ALL patients based on the engraftment of their samples was observed when NSG mice received VXL treatment. The median EFS was concordantly faster for ALL-64 compared to ALL-65 [61.8 (56.5- >274) versus >

273 days, P= 0.0701] and for ALL-66 compared to ALL-67 [51.3 (44.6-70.1) versus

186.8 (70.5- >245) days, P= 0.0067 log-rank test] when cells were transplanted into

NSG mice via the IF route (Figure 3.15C, Table 3.3). Interestingly, this transplantation condition allowed 125 days difference in median EFS and 14 days difference in first mouse to reach event between samples from patients who relapsed compared with those in CR1 samples (Figure 3.15C).

The ability of xenografts derived from samples of patients who relapsed to engraft earlier than xenografts derived from samples of CR1 was also observed when NSG mice were inoculated via the IV route and received VXL treatment (Figure 3.15D,

Table 3.3). The median EFS was shorter for ALL-64 compared to ALL-65 [58.9 (55.8-

60.6) versus > 273 days, P= 0.0067] and for ALL-66 compared to ALL-67 [49.3 (47.2-

54.2) versus 105.8 (95.7-108.4) days, P=0.0067 log-rank test]. Therefore, this transplantation condition allowed 46.9 days difference in median EFS and 39.9 days difference in the first mouse to reach event between xenografts derived from patients who relapsed and from those who remain in CR1 (Figure 3.15D, Table 3.3). Together, these results show that the pattern of leukaemia growth in NSG mice that received VXL treatment varied according to the patient outcomes, reflecting the ability of VXL treatment to mimic the sensitivity of xenograft cells to induction therapies in patients.

In concordance with these findings, the difference in LGD was calculated by subtracting the median EFS and first mouse to reach event between control groups and VXL-treated

119 of each xenografts. Based on the LGDs, xenografts established from relapsed patients were resistant to VXL treatment, whereas xenografts derived from patients who remain in CR1 exhibited a significant delay in their growth upon VXL treatment (Table 3.3.).

120

NSG IV A NSG IF B Relapse 100 Relapse 100 Difference in ALL-64 Difference in ALL-64 First mouse = First mouse= ALL-66 80 ALL-66 80 14.6 days 15.2 days CR1 CR1 ALL-65 60 ALL-65 60 ALL-67 ALL-67 Difference in

% survival 40 40 % Survival Median =20.7 Difference in Median days =7.1 days 20 20

0 0 0 25 50 75 100 125 150 250 280 0 25 50 75 100 125 150 175 200 225 250 Days post inoculation Days post inoculation

NSG IF VXL C D NSG IV VXL 100 Difference in First 100 Relapse Relapse mouse = 14 days ALL-64 75 ALL-66 Difference in ALL-64 75 First mouse ALL-66 CR1 = 39.9 days CR1 50 ALL-65 ALL-65 50 ALL-67 ALL-67 % Survival % Survival 25 Difference. in Median =125 days 25 Difference. in Median = 46.9 0 days 0 25 50 75 100 125 150 200 280 0 0 25 50 75 100 125 200 280 Days post inoculation Days post inoculation

Figure 3.15. Comparison of the EFS of NSG mice inoculated with IR ALL patient samples using different transplantation conditions. Each Kaplan-Meier curve shows the mouse EFS of IR ALL when established in (A) NSG IF, (B) NSG IV, (C) NSG IF VXL and (D) NSG IV VXL. Solid lines show the EFS of mice inoculated with cells from relapse cases whereas the dotted lines indicate xenografts derived from CR1 cases. Double headed arrows show the overall difference in EFS at TT1% (Median and First mouse) between xenografts of CR1 and relapse patients.

121

Table 3.3. Summary of engraftment kinetics of IR ALL panel in NSG mice at time to 1% human CD45+

Median EFS (Range) LGD LGD (First Xenograft (Median) Mouse) ID Non- treated VXL-Treated model NSG IV NSG IV VXL

ALL-64 64 (53.3-72.5) 58.9 (55.8-60.6) - 5.1 -2.5

Rel ALL-66 43.1 (42.8-43.1) 49.3 (47.2-54.2) 6.2 4.4

ALL-65 153.2 (112.5->241) > 273 > 119.8 >160.5

CR1 ALL-67 71.1 (67.9-71.9) 105.8 (95.7-108.4) 34.7 27.8

NSG IF NSG IF VXL

ALL-64 63.2 (56 - >280) 61.8 (56.5- > 274) -1.4 0.5

Rel ALL-66 48.9 (44-68.4) 51.3 (44.6-70.1) 2.4 0.6

ALL-65 >133.4 (112.3->235.3) > 273 > 139.6 160.7

CR1 ALL-67 83.9 (71.2-116.7) 186.8 (70.5- > 245) >102.9 0.7

Rel, relapse; CR1, first complete remission; LGD, leukaemia growth delay

122 3.7.2 Stratification of IR ALL xenografts based on engraftment in NOD/SCID mice at time to 1% human CD45+

The observation that ALL-64, which was derived from a patient who relapsed early, did not engraft in NOD/SCID mice over 266 days has limited the reliability of NOD/SCID mice to provide an efficient approach for stratifying IR ALL patients based on engraftment of their samples. The median EFS and the time for the first mouse to reach event for ALL-65 using various xenotransplantation conditions were much shorter than

ALL-64 (Figure 3.16, Table 3.4). Nevertheless, the difference in EFS between ALL-66 and ALL-67 established in NOD/SCID mice using different xenotransplantation conditions was more pronounced than that observed in NSG mice. As exemplified for both xenografts when established in NOD/SCID mice via the IF route, the median EFS of mice was shorter for ALL-66 compared to ALL-67 [(59.1 (51.1-89.4) versus 104.1

(86.5-153.8) days, P= 0.0266 log-rank test] and thus this transplantation condition allowed 45 days difference in median EFS and 35.4 days difference in the first mouse to reach event between ALL-66 and ALL-67 (Figure 3.16A). A similar difference between xenografts of patients who relapsed and xenografts of patients who maintain CR1 was observed when patient cells were inoculated into NOD/SCID mice via the IV route

((Figure 3.16A). As seen in NSG mice, the substantial difference in engraftment between ALL-66 and ALL-67 was accentuated by VXL treatment. For example, the median EFS for mice inoculated with ALL-66 via the IF route and treated with VXL was 76.2 (58.7-84.9) days compared to 183.3 (149.6- >247) days for ALL-67 (P=0.0067 log-rank test) (Figure 3.16C, Table 3.4). In addition, ALL-66 also exhibited a short

LGD compared to xenografts derived from CR1 cases (Table 3.4).

Although the engraftment of patient cells in NOD/SCID mice showed better stratification of ALL-66, ALL-65 and ALL-67 according to their clinical outcomes, the 123 lack of engraftment of cells of ALL-64 in this mouse strain suggests that using this strain would reduce our ability to develop a xenograft model that allows prediction of relapse in IR ALL patients.

124

A NOD/SCID IF B NOD/SCID IV 100 100 Difference in First Relapse Difference in First mouse = Relapse mouse =35.4 days 80 ALL-64 80 31.5 days ALL-64 ALL-66 ALL-66 CR1 60 CR1 60 ALL-67 Difference in ALL-65 ALL-65

40 Median = 45 ALL-67 % Survival

% Survival 40 Difference in days Median = 35.6 days 20 20

0 0 0 25 50 75 100 125 150 175 200 220 280 0 25 50 75 100 125 150 175 200 220280

Days post inoculation Days post inoculation

NOD/SCID IF VXL C D NOD/SCID IV VXL 100 100 Difference in Relapse Relapse 80 First mouse = ALL-64 ALL-64 90.9 days 75 ALL-66 Difference in ALL-66 60 CRCR1 First mouse = CR1 48.3 days ALL-65 50 ALL-65 40 % Survival ALL-67 ALL-67 % survival Difference in 20 Median = 25 Difference in Median 107.1 days =75.5 days 0 0 0 50 100 150 200 250 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Days post inoculation

Figure 3.16. Comparison of the EFS of NOD/SCID mice inoculated with IR ALL patient samples using different transplantation conditions. Solid lines show the EFS of xenografts derived from relapse cases whereas dotted lines indicate xenografts derived from CR1 cases when established in (A) NOD/SCID IF, (B) NOD/SCID IV, (C) NOD/SCID IF VXL and (D) NOD/SCID IV VXL.

125

Table 3.4. Summary of EFS of NOD/SCID mice inoculated with IR ALL patient samples and assessed at time to 1% human CD45+ cells.

Median EFS (Range)

LGD LGD (Median) (First Mouse) Xenograft Non-treated VXL -Treated ID model

NOD/SCID IV NOD/SCID IV VXL

ALL-64 >266 > 198.5 N/A N/A

Rel ALL-66 47.9 (47.1-49.4) 55.6 (53.7-56.2) 7.7 6.6

ALL-65 168.5 (150.5- >280) > 273 > 104.8 >122.5

CR1 ALL-67 83.5 (78.6-92.2) > 131.1 (102 - >224) > 47.6 23.7

NOD/SCID IF NOD/SCID IF VXL ALL-64 > 266 > 246.5 (>245->248) N/A N/A

Rel ALL-66 59.1 (51.1-89.4) 76.2 (58.7-84.9) 17.1 7.6

ALL-65 >280 (194.6->280) > 245 >35 50.4

CR1 ALL-67 104.1 (86.5- 153.8) 183.3 (149.6- >247) 79.2 63.1

Rel, relapse; CR1, first complete remission; LGD, leukaemia growth delay

126 3.8 Assessing the engraftment features of IR ALL xenografts based on time to 25% human CD45+ cells in mice PB

The results presented in the previous sections assessed the difference in engraftment based on the time to 1% human CD45+ cells. To ascertain whether or not our ability to select the most appropriate engraftment condition that allows prediction of relapse in IR

ALL patients could be depending on the end point of observation, I have further characterised the difference in engraftment features at the exponential phase of xenograft cell growth. Herein, the influence of the mouse strain and method of inoculation and ability to predict patient outcomes are discussed based on median times to event defined as 25% human CD45+ cells in the PB.

3.8.1 NSG versus NOD/SCID

When the speed of engraftment was compared between the non-VXL treated NSG and

NOD/SCID mice inoculated via the same route to establish each xenograft, the engraftment was found to be quicker in NSG compared to NOD/SCID mice. For instance, shorter time to event was observed for ALL-66 and ALL-67 in NSG mice inoculated via the IV route [65.4 days (range 62.6.4-74.2) and 136.8 days (range 132.9-

141.9) respectively] compared to that in NOD/SCID mice inoculated via the same route

[75.3 days (range 74.2-79.3) and 154.8 days (range 145.3-172.5) respectively] for both xenografts. Figure 3.17 shows the comparison of EFS of mice between each xenograft established in NSG and NOD/SCID mice.

3.8.2 IV versus IF inoculation

To assess the optimal method of inoculation at time to 25% human CD45+, the median

EFS of either NSG or NOD/SCID mice inoculated via the IF route was compared with the median EFS of those inoculated via the IV route to establish each xenograft.

Consistent with the findings observed at time to 1% human CD45+ cells, the IF 127 inoculation did not improve the speed of engraftment across all established xenografts.

As shown in Figure 3.18A, the median EFS values in most NSG mice inoculated with patient cells via the IV route were not significantly different than that for mice inoculated with patient cells via the IF route. Likewise, the time required for the

NOD/SCID mice inoculated via the IV route to reach event was not significantly than that observed when mice were inoculated via the IF route (Figure 3.18B).

128

A A

100 NSG IV 80 ALL-64 ALL-65 60 ALL-66 ALL-67 40 NOD/SCID IV

Percent survival Percent ALL-64 20 ALL-65 ALL-66 ALL-67 0 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Median EFS (Range) P value NSG IV NOD/SCID IV (log-rank) ALL-64 210.7 (204.8- 269.4) >266 0.0455 ALL-65 201.9 (170.3- >241) 209.3 (205.7- >280) 0.6003

ALL-66 65.4 (62.6.4-74.2) 75.3 (74.2-79.3) 0.0169

ALL-67 136.8 (132.9-141.9) 154.8 (145.3-172.5) 0.0067

B B 100 NSG IF ALL-64 80 ALL-65 ALL-66 ALL-67 60 NOD/SCID IF ALL-64 40 ALL-65

Percent survival Percent ALL-67 20 ALL-66

0 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Median EFS (Range) P value NSG IF NOD/SCID IF (log-rank)

ALL-64 207.8 (197.5 - >280) >266 0.0401

ALL-65 179.1 (155.1->280) >247 (213.9- >280) 0.1389

ALL-66 73 (62.7-101.6) 88 (76.2-134.3) 0.3237

ALL-67 138.2 (121.9-184.9) 167.3 (153.8->280) 0.2406

129 Figure 3.17. Graphs of individual engraftment conditions showing how each patient sample engrafted at 25% human CD45+ in NSG mice compared to NOD/SCID mice. Kaplan-Meier survival curves show the EFS of NSG and NOD/SCID mice inoculated with patient samples via the IV (A) or the IF (B) route. The table below each graph shows the median (range) time to 25% when inoculated with patient samples using different engraftment conditions. The log-rank test was used to calculate the difference in engraftment.

130 A A 100 NSG IV 80 ALL-64 ALL-65 60 ALL-66 ALL-67

40 NSG IF ALL-64 Percent survival Percent ALL-65 20 ALL-66 ALL-67 0 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation

Median EFS (Range)

P value NSG IV NSG IF (log-rank) ALL-64 210.7 (204.8- 269.4) 207.8 (197.5 - >280) 0.7818

ALL-65 201.9 (170.3- >241) 179.1 (155.1->280) 0.6729

ALL-66 65.4 (62.6.4-74.2) 73 (62.7-101.6) 0.1787

ALL-67 136.8 (132.9-141.9) 138.2 (121.9-184.9) 0.4883

B B

100 NOD/SCID IV ALL-64 80 ALL-65 ALL-66 ALL-67 60 NOD/SCID IF ALL-64 40 ALL-65

survival Percent ALL-67 20 ALL-66

0 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation

Median EFS (Range) NOD/SCID P value NOD/SCID IF IV (log-rank)

ALL-64 >266 >266 N/A

ALL-65 209.3 (205.7- >247 (213.9- >280) 0.3424

>280) ALL-66 75.3 (74.2-79.3) 88 (76.2-134.3) 0.1787

ALL-67 154.8 (145.3-172.5)167.3 (153.8->280) 0.2082

131 Figure 3.18. Graphs of individual engraftment conditions comparing the mouse EFS at time to 25% human CD45+ cells when mice were inoculated via IV and IF routes. Kaplan-Meier survival curves show the EFS of NSG mice inoculated with patient samples via the IF and IV routes (A) or the EFS of NOD/SCID mice inoculated with patient samples via the IF and IV routes (B). The table below each graph shows the median (range) time to 25% when inoculated with patient samples using different engraftment conditions. The log-rank test was used to calculate the difference in engraftment.

132 3.8.3 Stratification of IR ALL xenografts based on EFS of NSG mice at time to 25% human CD45+ cells

The engraftment of patient samples in mice inoculated using various transplantation strategies was also compared according to patient outcomes. Generally, the EFS of mice at time to 25% human CD45+ cells stratified all xenografts according to the outcomes of patients from whom they were derived with the exception of ALL-64.

The EFS of the non-VXL treated NSG mice which were inoculated with cells from a relapsed patient to establish ALL-64 demonstrated longer time to event compared to mice inoculated with cells from a patient who maintained a CR to establish ALL-65.

However, the difference in EFS between mice of ALL-66 and ALL-67 was more pronounced than that observed at time to 1% human CD45+ cells (Figure 3.19A and B).

For example, ALL-64 required 210.7 days (204.8- 269.4) to reach event in NSG mice which were inoculated via the IV route versus 201.9 (170.3- >241) days for ALL-65

(P= 0.8860 log-rank test) whereas ALL-66 required 65.4 days (62.6.4-74.2) to reach event compared to 136.8 (132.9-141.9) days for ALL-67, P= 0.0067 log-rank test]. Thus a minimum of 71.4 days difference in median EFS and 70.3 days difference in the first mouse to reach event were observed between ALL-66 and ALL-67 using this transplantation condition (Figure 3.19B).

In contrast to what I observed when the engraftment of patient samples was compared based on EFS of the VXL-treated mice at time to 1% human CD45+ cells, none of the

VXL-treated NSG mice which were used to establish ALL-64 reached event but a more appreciable difference in EFS was observed between ALL-66 and ALL-67 compared to that at time to 1% human CD45+ cells (Figure 3.19C and D). As exemplified for both xenografts when established in NSG mice via the IV route, the median EFS of mice was 133 shorter for ALL-66 compared to ALL-67 [(75.4 (71.7-95.2) versus 158.3 (149.9-166) days, P= 0.0067 log-rank test] and thus 82.9 days difference in median EFS and 78.2 days difference in the first mouse to reach event were observed between ALL-66 and

ALL-67 using this transplantation condition (Figure 3.19 D).

Based on the LGD, the pattern of sensitivity to VXL treatment of all xenografts except in ALL-64 also reflected the response of patients to their outcomes. For example, a major delay in leukaemia progression was observed between the VXL-treated and non- treated ALL-64 and ALL-65 xenografts in both the IV and IF groups. ALL-66 demonstrated very minimal delay in progression of cells in response to VXL treatment.

ALL-67 showed minor delay in progression between VXL-treated and non-treated mice which were inoculated via the IV route but a major delay in progression in those inoculated via the IF route (Table 3.5).

134

NSG IV A NSG IF B Relapse 100 100 ALL-64 Difference in Relapse Difference in ALL-66 80 First mouse ALL-64 80 First mouse =59.2 days ALL-66 =70.3 days CR1 CR1 60 60 ALL-65 ALL-65 ALL-67 ALL-67 40 40 Percent survival Percent Percent survival Percent 20 Difference in 20 Difference in Median = Median = 0 65.2 days 71.4 days 0 25 50 75 100 125 150 175 200 225 250 275 0 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Days post inoculation

C NSG IF VXL D NSG IV VXL 100 100 Difference in Relapse Difference in Relapse First mouse = 80 First mouse = 80 82.9 days ALL-64 78.2 days ALL-64 ALL-66 ALL-66 60 60 CR1 40 CR 40 Difference ALL-65 Percent survival Percent Percent survival Percent in Median = ALL-65 Difference ALL-67 20 152.2 days ALL-67 20 in Median = 82.9 days 0 0 0 25 50 75 100 125 150 175 200 225 250 275 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Days post inoculation Figure 3.19. Comparison of the EFS of NSG mice at time to 25% human CD45+ cells inoculated with IR ALL patient samples using different transplantation conditions. Each Kaplan-Meier curve shows the mouse EFS of IR ALL when established in (A) NSG IF, (B) NSG IV, (C) NSG IF VXL and (D) NSG IV VXL. Solid lines show the EFS of mice inoculated with cells from relapse cases whereas the dotted lines indicate xenografts derived from CR1 cases. Double headed arrows show the overall difference in EFS (Median and First mouse) between xenografts of CR1 and Relapse.

135

Table 3.5. Summary of EFS at TT25% of IR ALL panel in NOD/SCID mice and LGD values in response to VXL treatment

Median Time to 25 % (Range)

LGD Non-treated VXL -Treated Xenograft ID LGD (First

(Median) mouse)

NSG IV NSG IV VXL

l ALL-64 210.7 (204.8- 269.4) > 274 > 63.3 > 69.2 e R ALL-66 65.4 (62.6.4-74.2) 75.4 (71.7-95.2) 10 9.1

A1 LL-65 201.9 (170.3- >241) > 273 > 71.1 > 102.3 C 1 r CR1 AR LL-67 136.8 (132.9-141.9) 158.3 (149.9-166) 21.5 17

NSG IF NSG IF VXL

l ALL-64 207.8 (197.5 - >280) > 274 > 66.2 > 76.5 e R ALL-66 73 (62.7-101.6) 80.4 (74.2-81.9) 7.4 11.5

1 ALL-65 179.1 (155.1->280) > 273 > 93.9 > 117.9 R

C ALL-67 138.2 (121.9-184.9) >232.6 (144.6- >245) > 94.4 22.7 NOD/SCID IV NOD/SCID IV VXL

l ALL-64 > 266 >262.5 (>205->280) N/A N/A e R ALL-66 75.3 (74.2-79.3) 94.8 (84.6-100.1) 19.5 10.4

1 ALL-65 209.3 (205.7- >280) >255 (>239->262) >45.7 >33.3 R

C ALL-67 154.8 (145.3-172.5) 179.6 (140.6->224) 24.8 -4.3

NOD/SCID IF NOD/SCID IF VXL

ALL-64 >266 > 246.5 (>245->248) N/A N/A l e

R ALL-66 88 (76.2-134.3) 105.2 (97.9-121.1) 17.2 21.7 ALL-65 >247 (>176- >280) > 261.5 (>214->278) N/A N/A 1 R

C ALL-67 167.3 (153.8->280) > 247 (207->247) >79.7 >53.2 Rel, relapse; CR1, first complete remission; LGD, leukaemia growth delay

136 3.8.4 Stratification of IR ALL xenografts based on EFS of NOD/SCID mice at time to 25% human CD45+ cells

Analysis of engraftment between samples of patients with different outcomes based on the EFS of NOD/SCID mice at time to 25% human CD45+ cells revealed comparable patterns of EFS to that observed at time to 1% human CD45+ cells although a better separation was noticed between ALL-66 and ALL-67 in all transplantation conditions.

No events were recorded for all mice used to establish ALL-64 over 266 days, while the median EFS values of mice used to establish ALL-65 exceeded 200 days in the non-

VXL treated group (Figure 3.20). However, shorter median and first mouse EFS values were observed in ALL-66 and ALL-67 with a significant difference between both xenografts (Figure 3.20). As an example, ALL-66 reached event in 75.3 days (74.2-

79.3) when inoculated into mice via the IV route versus 154.8 (145.3-172.5) days for

ALL-67, P= 0.0067 log-rank test]. Therefore, 79.5 days difference in median EFS and

71.1 days difference in the first mouse to reach event were observed between ALL-66 and ALL-67 (Figure 3.20B).

In the VXL-treated group, all mice inoculated with patient cells to establish ALL-64 and ALL-65 showed no event over the monitoring period but distinct patterns of EFS were observed between ALL-66 and ALL-67. For example, ALL-66 reached event in

94.8 days (84.6-100.1) when inoculated into mice via the IV route and treated with VXL chemotherapy whereas 179.6 (140.6- >224) days were required for ALL-67, P= 0.0067 log-rank test] (Figure 3.20, Table 3.5). Therefore, 84.8 days difference in median EFS and 56 days difference in the first mouse to reach event were observed between ALL-66 and ALL-67 (Figure 3.20D).

137 Based on the LGD data, the VXL treatment led to a major delay in the progression of

ALL-64 and ALL-65 when engraftment was compared between the VXL-treated and non-VXL treated mice within each xenograft However, minor delays in leukaemia progression in all mice used to establish ALL-66 were observed (Table 3.5), and in the leukaemia progression in mice inoculated with cells via the IV route to establish ALL-

67 (Table 3.5).

138

A NOD/SCID IF B NOD/SCID IV

100 Relapse 100 Relapse Difference in ALL-64 80 First mouse ALL-66 80 ALL-64 =77.6 days Difference in ALL-66 CR1 First mouse 60 60 ALL-65 =71.1 days CR1 ALL-67 40 ALL-65 40 ALL-67 Percent survival Percent Percent survival Percent 20 Difference in Median = 20 Difference in 79.3 days Median = 0 0 79.5 days 0 25 50 75 100 125 150 175 200 225 250 275 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Days post inoculation

D NOD/SCID IV VXL C NOD/SCID IF VXL 100 100 Difference in Relapse Difference in Relapse 80 80 First mouse = First mouse = ALL-66 56 days ALL-64 109.1 days ALL-66 60 ALL-64 60 CR1 CR1 40 Difference in 40 Median = ALL-65 Difference in ALL-65 Percent survival Percent 141.8 days ALL-67 survival Percent Median = ALL-67 20 20 84.8 days

0 0 0 25 50 75 100 125 150 175 200 225 250 275 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Days post inoculation

Figure 3.20. Comparison of the EFS at time to 25% human CD45+ cells of NOD/SCID mice inoculated with IR ALL patient samples using different transplantation conditions. Solid lines show the EFS of xenografts derived from relapse cases whereas dotted lines indicate xenografts derived from CR1 cases when established in (A) NOD/SCID IF, (B) NOD/SCID IV, (C) NOD/SCID IF VXL and (D) NOD/SCID IV VXL.

139 3.9 Assessing the engraftment features of IR ALL xenografts based on time to leukaemia

To further investigate whether our ability to select the most appropriate transplantation condition could be improved by assessing the engraftment of patient samples based on a different time point of xenograft cell progression, the optimal mouse strain, route of inoculation and engraftment condition that allows prediction of relapse in IR ALL patients were characterised based on the time elapsing between transplantation of leukaemia cells and manifestation of overt leukaemia which was defined as TTL.

Consistent with the findings observed based on EFS at 1% and 25% human CD45+ cells, the difference in mouse EFS at TTL between either mouse strain and route of inoculation used to establish each xenograft emphasised the tendency of shorter time to event in NSG over the NOD/SCID mice (Figure 3.21) and in the IV inoculated mice over the IF inoculated group (Figure 3.22).

To determine whether stratification of xenografts according to outcomes of patients from whom they were derived could be influenced by assessing the mouse EFS at TTL, the time to event was compared between xenografts of patients with various outcomes when established using every transplantation condition. In general, the pattern of EFS observed between mice of different xenografts bears a resemblance to that observed based on time to 25% human CD45+ cells. In NSG mice, although both ALL-64 and

ALL-66 were derived from patients who relapsed early, short TTL was only observed in mice used to establish ALL-66 but all mice used to establish ALL-64 required longer time to event in the non-VXL treated mice and remained event free over the monitoring period in the VXL-treated group (Table 3.6). However, both ALL-65 and ALL-67 xenografts which were established from patients who remain on CR1 exhibited long

TTL in all mice with the exception of no event recorded in the VXL-treated mice of 140 ALL-65. In all NSG mice, there was more than 80 days difference in TTL between the xenograft of a relapsed patient and the xenograft of a CR patient (ALL-66 and ALL-67, respectively) (Figure 3.23). Consistent with what I observed at time to 25% human

CD45+, comparison of engraftment between each xenograft established in the VXL- treated and non-treated NSG mice revealed that the VXL treatment caused prominent delay in the progression of ALL-64 and ALL-65, minor delay in progression of cells in all mice used to establish ALL-66 and in progression of cells inoculated via the IV route to establish ALL-67 (Table 3.6). The pattern of difference in EFS of the NOD/SCID mice used to establish xenografts of patients with different outcome and the LGD values within each xenograft established in NOD/SCID mice were very consistent with that observed in NSG mice (Figure 3.24, Table 3.6).

141

A 100 NSG IV 80 ALL-64 ALL-65 60 ALL-66 ALL-67 NOD/SCID IV 40

Percent survival Percent ALL-64 20 ALL-65 ALL-66 ALL-67 0 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Median EFS (Range) P value NSG IV NOD/SCID IV (log-rank)

ALL-64 267 (248- >280) >266 0.1278

ALL-65 248 (224->280) 253 (207->280) 0.8864

ALL-66 97 120 (90-141) 0.2059

ALL-67 180 (176-194) 197.5 (181-223) 0.0601

B 100 NSG IF ALL-64 80 ALL-65 ALL-66 ALL-67 60 NOD/SCID IF ALL-64 40 ALL-65

Percent survival Percent ALL-66 20 ALL-67

0 0 25 50 75 100 125 150 175 200 225 250 275

Days post inoculation

Median EFS (Range)

P value NSG IF NOD/SCID IF (log- rank) ALL-64 233.5 (215->280) >266 0.0401 ALL-65 252 (223->280) >280 (268->280) 0.5372 ALL-66 112 (100-124) 123.5 (100-152) 0.2962 ALL-67 201.5 (166->264) 263 (194- >280) 0.3091

142 Figure 3.21. Graphs of individual engraftment conditions showing how each patient sample engrafted in NSG mice compared to NOD/SCID mice at time to leukaemia (TTL). Kaplan-Meier survival curves show the EFS of NSG and NOD/SCID mice inoculated with patient samples via the IV (A) or the IF (B) route. The table below each graph shows the median (range) time to leukaemia when inoculated with patient samples using different engraftment conditions. The log-rank test was used to calculate the difference in engraftment.

143

A 100 NSG IV ALL-64 80 ALL-65 ALL-66

60 ALL-67 NSG IF 40 ALL-64 ALL-65

survival Percent ALL-66 20 ALL-67 0 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Median EFS (Range) P value NSG IV NSG IF (log-rank) ALL-64 267 (248- >280) 233.5 (215->280) 0.7084

ALL-65 248 (224->280) 252 (223->280) 0.8103

ALL-66 97 112 (100-124) 0.0082

ALL-67 0.3107 180 (176-194) 201.5 (166->264)

B 100 NOD/SCID IV ALL-64 A 80 ALL-65 ALL-66 60 ALL-67 NOD/SCID IF 40 ALL-64 ALL-65 Percent survival Percent ALL-67 20 ALL-66

0 0 25 50 75 100 125 150 175 200 225 250 275

Days post inoculation

Median EFS (Range) P value NOD/SCID IV NOD/SCID IF (log-rank) ALL-64 >266 >266 N/A ALL-65 253 (207->280) >280 (268->280) 0.2222 ALL-66 120 (90-141) 123.5 (100-152) 0.6729 ALL-67 197.5 (181-223) 263 (194- >280) 0.0558

144 Figure 3.22. Graphs of individual engraftment conditions comparing the mouse EFS at TTL when mice were inoculated via IV and IF routes. Kaplan-Meier survival curves show the EFS of NSG mice inoculated with patient samples via the IF and IV routes (A) or the EFS of NOD/SCID mice inoculated with patient samples via the IF and IV routes (B). The table below each graph shows the median (range) time to leukaemia when inoculated with patient samples using different engraftment conditions. The log-rank test was used to calculate the difference in engraftment.

145

A NSG IF B NSG IV 100 100 Relapse Difference in Difference in First mouse = Relapse ALL-64 First mouse = 80 76 days 80 ALL-66 79 days ALL-64 ALL-66 60 CR1 60 ALL-65 CR1 40 40 ALL-67 ALL-65 Percent survival Percent Percent survival Percent Difference in Difference in 20 20 ALL-67 Median= Median= 89.5 days 83 days 0 0 0 25 50 75 100 125 150 175 200 225 250 275 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Days post inoculation

C NSG IF VXL D NSG IV VXL 100 100 Difference in Relapse Difference in Relapse 80 First mouse = 80 First mouse = 82 days ALL-64 83 days ALL-64 ALL-66 60 60 ALL-66 CR1 CR1 40 40 ALL-65 ALL-65 Percent survival Percent Percent survival Percent Difference Difference ALL-67 20 in Median = ALL-67 20 in Median = 120 days 88.5 days 0 0 0 25 50 75 100 125 150 175 200 225 250 275 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Days post inoculation Figure 3.23. Comparison of the EFS of NSG mice at time to leukaemia when IR ALL patient samples were inoculated using different transplantation conditions. Each Kaplan-Meier curve shows the mouse EFS of IR ALL when established in (A) NSG IF, (B) NSG IV, (C) NSG IF VXL and (D) NSG IV VXL. Solid lines show the EFS of mice inoculated with cells from relapse cases whereas the dotted lines indicate xenografts derived from CR1 cases. Double headed arrows show the overall difference in EFS (Median and First mouse) between xenografts of CR1 and Relapse.

146

A NOD/SCID IF B NOD/SCID IV

100 100 Difference in Relapse Difference in Relapse First mouse = First mouse = ALL-64 80 ALL-64 80 94 days 90 days ALL-66 ALL-66 60 60 CR1 CR1 ALL-65 40 ALL-65 40

Percent survival Percent ALL-67

Difference survival Percent ALL-67 20 in Median = 20 Difference 139.5 days in Median = 77.5 days 0 0 0 25 50 75 100 125 150 175 200 225 250 275 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Days post inoculation

C NOD/SCID IF VXL D NOD/SCID IV VXL 100 100 Difference in Relapse Difference in Relapse 80 First mouse = 80 First mouse = 120 days ALL-64 49 days ALL-64 ALL-66 ALL-66 60 60 CR CR1 40 ALL-65 40 ALL-65

Percent survival Percent Difference in ALL-67 Percent survival Percent Median = Difference in ALL-67 20 20 104 days Median = 116 days 0 0 25 50 75 100 125 150 175 200 225 250 275 0 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Days post inoculation

Figure 3.24 Comparison of the EFS at time to leukaemia of NOD/SCID mice inoculated with IR ALL patient samples using different transplantation conditions. Solid lines show the EFS of xenografts derived from relapse cases whereas dotted lines indicate xenografts derived from CR1 cases when established in (A) NOD/SCID IF, (B) NOD/SCID IV, (C) NOD/SCID IF VXL and (D) NOD/SCID IV VXL.

147 Table 3.6. Summary of EFS at TTL of IR ALL panel in NOD/SCID mice and LGD values in response to VXL treatment

Mouse EFS (Range )

Xenograft Non-treated VXL-Treated LGD LGD ID model model (Median) (First mouse) Days Days

NSG IV NSG IV VXL

ALL-64 267 (248- >280) > 274 > 7 >26

Rel ALL-66 97 123 (119-125) 26 22

ALL-65 233 (224-> 280) > 273 > 40 >59

CR1 ALL-67 180 (176-194) 211.5 (202-233) 31.5 26 NSG IF NSG IF VXL

ALL-64 233.5 (215->280) > 274 > 40.5 >59

Rel ALL-66 112 (100-124) 125 (121-134) 13 9

ALL-65 252 (223->280) > 273 > 28 >50

CR1 ALL-67 201.5 (166->264) >245 (203- > 246) > 65 >37

NOD/SCID IV NOD/SCID IV VXL

ALL-64 >266 >262.5 (>205->280) N/A N/A

Rel ALL-66 120 (90-141) 129 (128-139) 9 38

ALL-65 253 (207->280) >255 (>239->262) > 2 >32

CR1 ALL-67 197.5 (181-223) 245 (177 - 245) >47.5 -4

NOD/SCID IF NOD/SCID IF VXL

ALL-64 >266 > 246.5 (>245->248) N/A N/A

Rel ALL-66 123.5 (100-152) 143 (118-149) 19.5 18 ALL-65 >280 (268->280) > 261.5 (>214->278) N/A N/A

ALL-67 263 (194-

CR1 247 (238- > 252) >31 > 44 >280) Rel, relapse; CR1, first complete remission; LGD, leukaemia growth delay

148 3.10 Summary and Discussion

The potential for immunodeficient mice to engraft human ALL samples has been previously reported (Liem et al., 2004; Nijmeijer et al., 2001; Uckun et al., 1998). It has been demonstrated that using NOD/SCID mice to establish xenografts for childhood

ALL patients provides a preclinical model that recapitulates the heterogeneity of the disease and mimics the in vivo responses of ALL patients to induction chemotherapy drugs, dexamethasone, L-asparaginase and vincristine, used in the clinic to treat those patients (Liem et al., 2004). Further studies have also showed that the use of a more immunodeficient strain, the NSG mouse allows better growth of xenografts derived from ALL patients and retains the fundamental biological characteristics of the disease

(Agliano et al., 2008; Diamanti et al., 2012; Schmitz et al., 2011). Therefore, it is anticipated that utilising xenografts for translational research will improve the treatment of ALL patients, with the ultimate goal of targeting unique disease biology and identifying ALL patients who are prone to relapse at an early stage of their diseases.

The difference in response to current chemotherapy protocols among IR ALL indicates that all patients should not receive a uniform treatment protocol. Therefore, early identification of patients who need more intensive or different chemotherapy treatments would facilitate improvements in outcomes of IR ALL patients. An important step toward utilising xenografts to predict the outcomes of IR ALL patients is to increase the robustness of available xenograft models to predict the response of IR ALL patients to their treatments. Recent research suggests that achieving this goal could be influenced by different conditions to transplant patient samples, and/or the clonal architecture of patient cells in various transplantation strategies.

149 In the current study, two pairs of IR ALL patients who presented with a uniform set of clinical features between patients of each pair at diagnosis but exhibited variable response to their treatment were selected for xenotransplantation into immunodeficent mice. The Pilot Study assessed the influence of specific factors (mouse strain, route of inoculation and selection with VXL chemotherapy) on engraftment of leukaemia cells that potentially could recapitulate the outcomes of patients from whom they were derived. Although the engraftment of leukaemia cells could be influenced by the quality of samples (storage and cell viability) to be inoculated into mice, the cryopreserved samples were the only option to establish xenografts for the Pilot Study. Thus, frozen samples with more than 60% viable cells were inoculated into mice. Xenografts were successfully established from all the four patient samples and all engrafted mice were able to tolerate xenografts and most mice showed no evidence of leukaemia related morbidity until they reached a high level of engraftment.

Generally, the appearance of xenograft cells from the IR ALL patients in the PB of mice varied according to the patient outcomes. ALL-64 and ALL-66, which were derived from ALL patients who developed BM relapse at 19 months after diagnosis, manifested earlier time to 1% of human CD45+ cells in the PB compared to xenografts of CR1 patients. The patients from whom ALL-65 and ALL-67 were derived remain in clinical

CR1 and manifested late appearance of cells in the PB of mice. In addition, the pattern of increase in the proportion of human CD45+ cells in the PB and manifestation of TTL was also dependent on the outcomes of patients from whom cells were used to establish xenografts with the exception of ALL-64.

150 To compare the influence of various xenotransplantation conditions on the engraftment efficiency of samples derived from IR ALL patients, I first analysed the difference in engraftment capacity between the two mouse strains used to establish xenografts of IR

ALL patients. When the total number of engrafted mice and speed of engraftment at different time points were compared between both mouse strains, the NSG mouse strain permitted better growth of each xenograft independent of the method of inoculation and treatment. The NOD/SCID mouse strain was not receptive for engraftment of the ALL-

64 patient sample and a longer time was required for establishment of xenografts in most engrafted mice. In line with these findings, engraftment of ALL cells was previously reported to be faster in NSG compared to NOD/SCID mice (Agliano et al.,

2008). Similarly, Diamanti et al. (2012) have also found that NSG mice allow engraftment of ALL cells from paediatric patients irrespective of patient prognosis and or disease status, and NSG mice permitted engraftment of leukaemia clones that were not observed when cells from the same source were inoculated into NOD/SCID mice.

I then analysed the difference in engraftment capacity between each samples from IR

ALL patients when inoculated via the IV and IF routes. Regardless of the time point of observation, the efficiency of engraftment in mice inoculated via the IV route was higher than that achieved in the IF inoculated group. In addition, the choice of IF inoculation did not improve the speed of engraftment when time to engraftment was compared between mice inoculated with patient cells via both routes to establish each xenograft. Although many reports have found that the IF method of transplantation provides efficient and robust engraftment of haematopoietic and leukaemia cells in immunodeficent mice (Mazurier et al., 2003; Notta et al., 2011; Schmitz et al., 2011), my results showed that there was no further enhancement to the speed and efficiency of

151 engraftment by direct injection of cells into the BM compared to the IV route. In addition, two of the mice which received patient cells via IF inoculation had developed leg tumours in the region of the initial site of transplantation at autopsy but also exhibited the same pattern of engraftment observed in other mice. This localised engraftment could be due to failure of the fine 27G needle of the 0.5 ml insulin syringe in some cases to penetrate the femoral cavity as this needle can be easily bent and miss the femoral compartment and therefore some leukaemia cells could be trapped within tissues surrounding the right femur bone and might have proliferated to form a localised cell mass. These findings suggest that cells from patients who are stratified into the IR

ALL subtype seed into a suitable environment that supports leukaemia cell growth irrespective of transplantation routes.

The general conclusion from this study is that xenografts of IR ALL reflected outcomes of their corresponding patients. However, there were variable degree to which different transplantation strategies could provide a robust approach toward prediction of patient outcomes based on engraftment of their samples. Based on EFS at time to 1% human

CD45+, a comparison of engraftment between diagnosis samples of patients who relapsed and those who remain in CR1 when inoculated into NSG mice via either the IF or IV routes revealed minimal ability to stratify IR ALL patients according to their outcome. Injection of IR ALL samples into NOD/SCID mice showed prominent differences in time to engraftment between the xenograft of a relapse patient (ALL-66) and the two xenografts of ALL-CR1 patients (ALL-65 and ALL-67). However, the ability of NOD/SCID mice inoculated via either the IF or IV routes to stratify xenografts according to patient outcomes was limited by the lack of engraftment of

ALL-64 in this mouse strain.

152 The results merit comparison with other reports which showed that engraftment of samples derived from ALL patients at the time of diagnosis in NOD/SCID mice did not correlate with duration of complete remission in patient and the association was only observed when samples were collected from patients at time of relapse (Lock et al.,

2002). Similarly, Schmitz et al. (2011) showed that engraftment of samples collected at time of relapse from ALL patients who were stratified into various risk subtypes in

NSG mice was also associated with the duration of complete remission of patients from whom they were derived. However, another paper published by Meyer et al. (2011) indicated that the time to clinical manifestation of morbidity signs related to ALL disease in NOD/SCID mice correlated with the duration of clinical remission of the

ALL patients when samples collected at diagnosis were used to develop ALL xenografts.

The most clinically relevant difference between the engraftment of samples from ALL-

Relapse and ALL-CR1 patients was demonstrated when immunodeficent mice were transplanted with patient cells and then selected with VXL treatment for two weeks.

Based on 1% human CD45, the xenografts established in NSG mice were stratified into two quite disparate groups: those derived from patients who relapsed early (ALL-64 and

ALL-66), and those derived from patients who maintain CR (ALL-65 and ALL-67).

Further, the pattern of drug sensitivity of xenografts to VXL treatment at time to 1% human CD45+ mirrored the progression of disease in the patients from whom the xenografts were established. The VXL treatment allowed early engraftment of cells from ALL-Relapse while delaying engraftment of xenografts from ALL-CR1. The ability of VXL treatment to stratify xenografts according to the outcome of their respective patients was also highlighted in all xenografts except ALL-64 when

153 engraftment was characterised based on time to 25% human CD45+ and the time to leukaemia.

In agreement with what has been reported by Szymanska et al. (2012), xenografts established from relapsed cases were resistant to VXL treatment. In contrast, there was a significant difference in median EFS between the VXL-treated and non-VXL treated xenografts of the long term surviving patients. Relapse in ALL patients could arise from the growth of leukaemia populations which are intrinsically resistance to chemotherapy or after the acquisition of resistance during therapy (Szczepanek et al., 2011). The finding that applying VXL pressure led to stratification of IR ALL xenografts based on outcomes of their patients suggest that VXL may allow better reflection of the intrinsic growth activity of leukaemia initiating cells (LICs) in mice which mimic the growth in

ALL patients under such selective pressure.

In summary, proper modelling of ALL disease progression in immunodeficient mice is challenged by the intrinsic properties of the disease and the influence of the model system. The results presented in this chapter signified the capability of the NSG mouse strain and the IV route to provide higher engraftment efficiency and faster establishment of xenografts over the NOD/SCID mouse strain and the IF route, although these conditions showed limited ability to stratify xenografts according to outcomes of their patients. In addition, the results emphasised the role of VXL treatment to reflect a patient's response to chemotherapy and thus provided preliminary evidence that xenografts in immunodeficient mice could predict outcome of IR ALL patients. This experiment did not resolve whether the objective definition or the subjective clinical assessment of engraftment is the most optimal strategy for assessing the difference in

154 survival of mice used to establish xenografts. Thus, the utility of the NSG mouse strain in the presence of VXL treatment conditions to independently identify relapses in IR

ALL patients will be tested at different time points of engraftment in a larger panel of patient samples in the next chapter.

155 CHAPTER 4 MAIN STUDY: ASSESSING THE CAPABILITY OF XENOGRAFTS TO PREDICT RELAPSE IN IR BCP-ALL PATIENTS

156 4.1 Introduction

Development of an improved method to efficiently predict relapse soon after diagnosis in patients stratified into the IR subgroup has the potential to contribute to increase the cure rate for those patients who are otherwise difficult to distinguish and to reduce the toxicity from chemotherapy treatment in non-relapsed cases (Conter et al., 2007).

Among IR ALL patients, some may experience relapse at an early stage after treatment, while others may relapse after a longer remission period or may remain disease-free after treatment (Chessells et al., 2003; Gandemer et al., 2012).

The clinical heterogeneity among IR ALL patients cannot be precisely defined through available prognostic features. In general, the timing of relapse in ALL patients has been shown to be indicative of clinical outcome and is being linked with the biology of relapse. For instance, the success of reinduction with chemotherapy to achieve a second complete remission (CR2) in ALL cases with BM relapse varies depending on the time to relapse (Bhojwani et al., 2006; Raetz and Bhatla, 2012). Moreover, the rate of subsequent remission of ALL patients who experienced late BM relapse approaches

95% whereas that for early relapse patients ranges from 70%-85% but it decreases to

50% for very early relapses (Raetz and Bhatla, 2012). In addition, it has been shown that the mechanism by which relapse is encountered at an early stage can be due to the emergence of clones similar to their respective cell populations in diagnostic samples.

In contrast, late relapse seems to be mediated by acquisition of further somatic and/or epigenetic changes (Bhojwani et al., 2006).

The evidence that timing of relapse correlates with the overall survival of ALL patients provides a rationale to identify early and late relapses from non-relapsed patients for

157 better management of the disease. For this strategy to be successful, an innovative preclinical model that recapitulates the clinical heterogeneity among patients who are stratified into the IR ALL subtype with an approach toward a more refined classification of those patients may prove beneficial.

The previous chapter detailed the establishment of the optimal experimental strategy, which could be used to investigate the reliability of time to engraftment of samples derived from diagnostic IR ALL patients for stratifying patients according to their outcomes. The Pilot Study reported in that chapter was a prelude to the Main Study, and a number of findings in the Pilot Study suggest the feasibility of conducting a larger study to develop an improved relapse prediction strategy based on the engraftment of IR

ALL patient samples in immunodeficient mice. Analysis of the engraftment among the four IR ALL patient samples, which were inoculated into mice using various transplantation conditions, demonstrated that the NSG mouse strain allows greater efficiency and faster engraftment of the patient samples compared with the NOD/SCID mice. It also showed that inoculation of patient samples into mice via the IV route provides good efficiency and faster engraftment of IR ALL patient samples compared with IF inoculation. However, limited ability to stratify the IR ALL xenografts according to the outcomes of their corresponding patients was achieved using either mouse strain when inoculated via either route.

The response of IR ALL xenografts to the VXL induction chemotherapy, however, led to superior difference in engraftment, reflecting the clinical outcomes of the patients from whom xenografts were derived. On the basis of these observations, inoculation of

IR ALL patient cells into NSG mice via the IV route followed by VXL selection was suggested to be the most appropriate transplantation strategy that could allow efficient 158 and reasonable stratification of xenografts according to the outcome of their corresponding patients. Nevertheless, despite no indication for prediction of outcome using the non-VXL treated mouse models, the data from the Pilot Study represent a small cohort of IR ALL patients and indicated the importance of the LGD parameter for stratification of xenografts according to patient outcomes. Therefore, it was appropriate for the Main Study to include control mice to emphasise the finding from the Pilot

Study, explain whether or not samples, which could not engraft in the VXL-treated mice, are able to engraft in the control mice and also to assess the predictive capability of the LGD parameter to stratify xenografts according to patient outcomes. I therefore decided to carry out the Main Study using a group of NSG mice that will either receive

VXL chemotherapy or vehicle control.

The aim of the Main Study was to assess the validity of IR ALL patient derived xenografts to distinguish between early relapse, late relapse and maintained CR1 based on engraftment characteristics of diagnostic samples from a large cohort that included patients who experienced early or late relapsed and patients who remain in CR1. The objectives of the experiments described in this chapter were:

(1) To establish xenografts from diagnostic BM samples of IR ALL patients who

demonstrated different response to their treatment yet were all stratified into the

IR ALL subtype.

(2) To assess the reliability of the optimised engraftment condition to predict relapse

in IR ALL patients based on different engraftment characteristics.

159 4.2 Patient selection for the Main Study

The Tumour Bank database at CCI was searched for available diagnosis BM samples of

BCP-ALL patients who were enrolled on the ANZCHOG Study VIII protocol and the data cross-referenced to the CCI MRD group records to identify IR ALL patients with different responses to their treatments. This search revealed more than 100 samples from Study VIII BCP-ALL IR patients who are in CR1, 9 samples of IR ALL patients who had an early relapse (defined as relapsed on therapy ≤ 25 months) and 10 samples of IR ALL patients who had late relapse (defined as relapsed after 25 months). My co- supervisor and I were able to select 30 IR ALL patient samples including all available

BM samples of patients who relapsed (early and late) and 11 samples of patients who are in CR1 for the Main Study. Table 4.1 shows the clinical characteristics of the three panels of patients including early and late relapse and non-relapsed patients who presented with clinical features that stratified them into the IR subtype, with some variation in MRD response to induction treatment and eventually had different survival outcomes.

Each group contained male and female patients who were aged 2 years to less than 18 years old, most patients presented at diagnosis with a WCC of less than 50 x 109/L and with favourable cytogenetic features such as hyperdiploid chromosomes or ETV6-

RUNX fusion, or other unfavourable cytogenetic features such as “B other”, or TCF3-

PBX1 fusion. The relapses in patients were encountered as early as 17 months from diagnosis with the longest duration between treatment and relapse being 66 months.

Most of the early relapsed patients are deceased from their disease whereas the majority of those who relapsed late after initial treatment are currently in second complete remission (CR2).

160 The cohort included 3 patient samples (A5072, A1839 and A1795) that were used in the

Pilot Study (Tables 3.1 and 4.1). Since the Main Study was carried out in a blinded fashion it will be useful to compare the reproducibility of xenograft models established from A5072 (ALL-64 and ALL-202), A1839 (ALL-66 and ALL-215) and A1795

(ALL-65 and ALL-220).

161 Table 4.1. Clinical characteristics of IR ALL patient samples used for the Main Study

Xenograft Age at Dx WCC Length of MRD MRD Status at last UPN Sex Cytogenetic group site of relapse ID (Year) 9 CR1 (Month) follow up (10 /L) Day 15 Day 33 ALL-201 A2497 F 16.7 11.2 B OTHER 25 BM relapse >1x10-4 >1x10-4 Alive in CR2 ALL-202 A5072 F 1.6 3.1 Hyperdiploid >50 19 BM relapse N/A Pos DOD ALL-203 A3729 M 17.3 47.74 B OTHER 20 BM relapse Pos Neg DOD -1 -4 ALL-206 A1524 M 4.4 6.62 ETV6-RUNX1 >120 Non >1x10 >1x10 CR1 >10 years ALL-207 A1693 M 12.6 7.8 B OTHER >84 Non Pos >1x10-4 CR1 >7 years -1 -3 ALL-208 A1725 M 12.7 50.3 B OTHER >60 Non >1x10 >1x10 CR1 >5 years ALL-209 A1692 M 12.9 22.2 B OTHER 29 BM & CNS relapse >1x10-3 Pos Alive in CR2 -1 -4 ALL-210 A2455 F 13.5 4.5 Hyperdiploid >50 24 BM relapse >1x10 >1x10 TRM ALL-211 A2275 M 3.4 9.4 Hyperdiploid >50 24 CNS relapse >1x10-2 >1x10-3 DOD ALL-212 A1495 M 6.3 3.2 Hyperdiploid >50 >120 Non >1x10-3 >1x10-4 CR1 >10 years -3 ALL-213 A1702 F 3 158.4 B OTHER >108 Non >1x10 Pos CR1 >9 years ALL-214 A1696 M 6.4 45.8 B OTHER >120 Non >1x10-2 >1x10-4 CR1 >10 years ALL-215 A1839 F 2.4 135.8 ETV6-RUNX1 19 BM Relapse Pos Pos Alive in CR2 ALL-216 A1803 M 8.7 34.9 B OTHER 17 CNS relapse N/A >1x10-3 DOD ALL-217 A1799 M 12.2 4.54 TCF3-PBX1 24 BM relapse N/A Pos Alive in CR2

ALL-218 A3279 F 6.5 21.1 Hyperdiploid >50 36 BM relapse N/A >1x10-2 Alive in CR2 Table continued on the following page

162

Xenograft Age at Dx WCC Length of CR1 MRD MRD Status at last UPN Sex Cytogenetic group site of relapse ID (Year) (109/L) (Month) Day 15 Day 33 follow up ALL-219 A1512 M 15.7 9.7 B OTHER 25 CNS relapse N/A N/A DOD ALL-220 A1795 F 2 9.8 Hyperdiploid >50 >84 Non Pos Pos CR1 >7 years ALL-221 A3102 M 3.9 38.32 ETV6-RUNX1 38 BM relapse >1x10-3 >1x10-4 Alive in CR2

ALL-222 A2865 F 7.6 33.41 ETV6-RUNX1 35 BM relapse >1x10-2 >1x10-4 Alive in CR2 ALL-223 A1744 F 3.6 2.6 Unclear >108 Non >1x10-2 >1x10-4 CR1 >9 years -1 -4 ALL-224 A1746 F 8.5 22 B OTHER >120 Non >1x10 >1x10 CR1 >10 years ALL-225 A2749 F 10.8 2.6 Hyperdiploid >50 38 BM relapse >1x10-2 >1x10-3 Alive in CR2 ALL-226 A2522 F 5.2 12.4 B OTHER 30 BM Relapse N/A >1x10-4 Alive in CR2 ALL-227 A1747 M 3.1 70.9 B OTHER 28 BM/CNS/Testis relapse >1x10-2 >1x10-3 Alive in CR4 ALL-228 A1780 M 6 27.5 ETV6-RUNX1 66 BM relapse >1x10-2 >1x10-3 Alive in CR2

ALL-229 A1901 F 11.2 4.6 B OTHER 31 BM & CNS relapse N/A Neg DOD ALL-230 A2173 F 8.2 428.1 B OTHER 27 CNS relapse >1x10-2 >1x10-3 Alive in CR3 ALL-231 A1849 F 5.1 20.1 B OTHER >108 Non N/A Pos CR1 >9 years ALL-232 A1919 M 2.9 30.7 ETV6-RUNX1 >96 Non >1x10-4 Pos CR1 >8 years

UPN, unique patient number; F, female; M, male; Dx, diagnosis; CR1/2/3/4, first/second/third/fourth complete remission; WCC, white cell count; BM, bone marrow; CNS, central nervous system; MRD, minimal residual disease; N/A, not available; DOD, dead of disease; TRM, transplantation related mortality; pos, MRD positive; neg, MRD negative

163 4.3 Establishing and characterising engraftment of the IR BCP-ALL patient primary samples

Cryopreserved BM samples of all the 30 patients were obtained from the CCI Tumour

Bank and transferred into our own internal liquid nitrogen tank. Patient samples were thawed, washed, and one million cells from each patient sample were inoculated into 6 female NSG mice via the IV route as described in the Pilot Study. Mice were randomised two weeks following inoculation to receive either an induction like VXL- regimen for two weeks or normal saline. Patient cells were tested for immunophenotype and showed strong positivity for either human CD45 or HLA–DR markers and therefore leukaemia engraftment was monitored weekly based on the proportion of human CD45 or HLA-DR positive cells in the murine PB. As described in the Pilot Study, the EFS of mice was calculated based on the time required for first and median mouse of each group to reach 1%, 25% human cells (TT1% and TT25%) and TTL. The LGD was calculated between the EFS of the VXL-treated mice compared with that of the saline- treated groups. The progress of engraftment was monitored for each mouse until the end point determined by either EFS, signs of morbidity and distress, or reaching the maximum holding time approved by the UNSW Animal Care and Ethics Committee. To assess the dissemination of leukaemic cells into body organs samples were taken from bone marrow, liver, kidney, lung and brain at necropsy.

To ensure the independence of data generated from the experiments aimed to monitor the engraftment features of IR ALL patient samples and prevent any bias from the experimenter expectations or preferences of sample engraftability, I was blinded to the patient clinical information and established all xenografts without access to the clinical data until the end of the study. Therefore, every patient sample was labelled with a

164 unique identification number by an experienced staff member of the Leukaemia Biology

Program to distinguish xenografts during the process of xenograft development.

The development of ALL xenografts from the 30 patient primary samples required careful handling, monitoring of mice and analysis of leukaemia engraftment.

Inoculation of patient cells, handling of mice and running FACS were performed in collaboration with Dr. Babasaheb Yadav of the Leukaemia Biology Program. To ease establishment of ALL xenografts, the cohort was divided into a manageable three panels of 10 patients’ samples for inoculation into mice at different times. Each group of patient samples was arranged by experienced LB staff members to include samples of early and late relapsed patients and samples of patients who are in CR1. The clinical information of patients was revealed and matched with their corresponding xenografts after the end of the monitoring period in every group. The process of xenograft establishment is described in Figure 4.1.

165

9 Samples of 10 Samples of 11 Samples of early relapsed late relapsed CR1 patient patients patients Weekly One million cells NSG 1. Each sample was labelled with a unique ID monitoring of injected +/- 2 weeks of VXL number which was used to recognise patient engraftment

derived xenograft during the monitoring period treatment after two via the IV route

weeks of inoculation 2. The cohort was randomised into 3 groups of 6 mice /Patient n=10, each group processed in a separate sample experiment

Human cells

Mouse cells Purification of Cryopreservation Mice euthanised at

TTL mononuclear cells

Data analysis: 1. Assess the difference in engraftment between Data processing: mice that received cells of the three IR ALL 1. Reveal the clinical data of each patient groups based on mouse EFS at TT1% patient from which xenograft and TT25% human cells and TTL was derived at end of the study

2. Assess the probability of relapse in patients 2. Compile the engraftment data of based on the engraftment properties of their patient samples from the three samples experiments Determine the EFS of mice at different level of engraftment

Figure 4.1. Experimental scheme for the Main Study. Thirty samples of IR ALL patients were labelled with unique identification numbers and divided into three panels, each panel contained a mix of patients with various lengths of CR1. Mice were injected with primary ALL cells via the IV route and two weeks later were randomised to receive VXL treatment or normal saline for two weeks and mice were monitored for engraftment. The mouse EFS at different time points of engraftment and the LGD values were calculated. Upon signs of engraftment, mice were euthanased and organs processed for isolation of leukaemic cells. The patient clinical data was revealed after the end of the study and relapse probability was assessed based on engraftment of patient samples. CR1, first complete remission; EFS, event free survival; TT1%, time to 1% human cell engraftment; TT25%, time to 25% human cell engraftment; TTL, time to leukaemia.

166 4.3.1 Xenografts established from the first panel of primary ALL patient samples

The first panel of 10 patient primary samples included 3 samples derived from early relapsed patients, 4 samples derived from late relapsed patients and 3 samples derived from patients in CR1.

Inoculation of samples derived from the three patients who experienced early relapse led to establishment of ALL-202 and ALL-203 but not ALL-201 (Figure 4.2A and B).

ALL-202 was derived from a female patient who experienced three relapse events after complete remission and thereafter died of the disease. The sample of this patient exhibited early appearance in in the PB of all mice. However, the engraftment kinetics fluctuated for several weeks before the levels of human cells increased gradually and reached high values in the PB of two mice from the non-treated group and one mouse of the VXL-treated group. Thus progression of xenograft cells in the PB varied between mice inoculated with the same patient sample (Figure 4.2A). The engraftment pattern of

ALL-202 showed some resemblance to ALL-64 in the Pilot Study, both of which were derived from the same patient sample (compare figures 3.2B and 4.2A). ALL-203 was derived from a male patient who also manifested BM relapse and died of the disease. In non-VXL-treated mice, this patient sample engrafted quickly and reached high levels in the PB. In the VXL-treated group, two mice exhibited relatively longer time to engraft in the PB than observed in the control mice and one mouse was excluded due to the appearance of a lump on the front right paw (Figure 4.2B). The third patient sample, which was derived from a relapsed female patient who is currently in CR2 and used to establish ALL-201, showed no evidence of engraftment in murine PB or other body organs (data not shown).

167 With the exception of the patient sample used to establish ALL-218, all of the samples derived from patients who experienced late relapse were successfully engrafted in all inoculated mice (Figure 4.2C-F). As an example, ALL-209 was derived from a male patient who experienced combined a BM and CNS relapse but maintains a CR after a second course of treatment. This patient sample engrafted within 50 days in the PB of non-VXL treated mice, and the VXL treatment did not greatly impede engraftment in the PB of mice (Figure 4.2C). All mice were engrafted with a high level of xenograft cells in the PB and the pattern of increase in xenograft cells was consistent between the non-VXL treated as well as the VXL-treated mice. In contrast, ALL-218, which was derived from a female patient who also experienced BM relapse and maintains second

CR, showed poor engraftment in two out of the three control mice as one mouse was excluded from the study due to excessive body weight loss. The level of engraftment barely reached 5% in the PB of the two control mice and the levels were maintained at lower values with no evidence of signs related to leukaemia morbidity until mice had reached the maximum holding time. None of the VXL-treated mice showed any evidence of engraftment in either PB or body organs at the time of harvest (Figure

4.2D).

ALL-221 was derived from a male patient who experienced BM relapse and maintains

CR after a second course of treatment with chemotherapy. Engraftment of this patient sample in the PB of the non-drug treated mice was moderately slow with a gradual increase in human cell level over the monitoring period (Figure 4.2E). The engraftment in PB was not hampered by VXL treatment, instead two mice of the VXL-treated group had an accelerated pattern of engraftment after applying treatment. ALL-227 was derived from a male patient who experienced combined BM and CNS relapses but

168 maintains a CR after three events of relapses. The time required for appearance of xenograft cells in the PB of the non-drug treated mice was longer than that observed in

ALL-209 but shorter than that observed in ALL-221 (Figure 4.2F). The VXL treatment caused a slight delay in the engraftment but a uniform pattern of increase in engraftment was observed among all the non-drug treated and the VXL-treated groups.

169 100 100 ALL-202 A ALL-203 B ALL-202 VXL ALL-203 VXL 80 80

60 60 in PB in in PB in + +

40 40 % huCD45 % huCD45 20 20

0 0

0 25 50 75 100 125 150 175 200 225 250 275 0 25 50 75 100 125 150 175 200 225 250 Days post inoculation Days post inoculation

100 ALL-209 C 50 ALL 209 VXL ALL-218 D 80 40 ALL-218 VXL

60 30 in PB in +

40 20

% HuCD45 in PB 10 20 %huCD45 0 0 -10 0 25 50 75 100 125 150 175 200 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Days post inoculation

100 100 ALL-221 E ALL-227 F ALL-221 VXL ALL-227 VXL 80 80

60 60 in PB in in PB in + +

40 40 % huCD45 % huCD45 20 20

0 0

0 25 50 75 100 125 150 175 200 225 250 275 0 25 50 75 100 125 150 175 200 Days post inoculation Days post inoculation

Figure 4.2. Engraftment of samples derived from the relapsed patients in the first panel. Graphs show the engraftment characteristics of two samples derived from early relapse patients to establish (A) ALL-202 and (B) ALL-203 and four samples derived from late relapse patients to establish (C) ALL-209, (D) ALL-218, (E) ALL-221, and (F) ALL-227 in control and VXL-treated mice over the monitoring period. Dashed black lines represent control mice while the solid red lines represent VXL-treated mice. The solid black squares indicate the VXL treatment period.

170 Inoculation of mice with samples derived from 3 patients who maintain a CR1 resulted in establishment of only two xenografts (ALL-207 and ALL-214). All mice inoculated with cells from a male patient, who has maintained a complete remission for more than

10 years to establish ALL-212 failed to show any evidence of engraftment in the PB or any body organs.

ALL-207 and ALL-214 were derived from two male patients who are in CR1 for more than 7 and 10 years, respectively. Both xenografts established in the PB of the non-VXL treated mice as quickly as that observed in mice used to establish xenografts of two relapsed patients (ALL-202 and ALL-209) (compare Figure 4.3 and Figure 4.2A and

C). However, the engraftment was appreciably delayed in both xenografts after applying

VXL treatment. The patterns of increase in engraftment of both patient samples were remarkably consistent within the non-VXL treated as well as the VXL-treated mice. Of note all mice engrafted with ALL-207 reached TTL while the percentage of human leukaemia cells was only at moderate levels in the PB. In contrast, all mice engrafted with ALL-214 reached high levels of PB infiltration before manifestation of any sign of leukaemia-related morbidity (Figure 4.3A and B).

171

100 ALL-207 A ALL-207 VXL 80

60

40 %Hu CD45+ in PB in CD45+ %Hu 20

0

0 25 50 75 100 125 150 175 200 225 250 Days post inoculation

100 ALL-214 B ALL-214 VXL 80 60

40

20 % Hu CD45+ cells in PB 0

0 25 50 75 100 125 150 175 200 225 250

Days post inoculation

Figure 4.3. Engraftment of samples derived from patients who maintain CR1 in the first panel. Graphs show the engraftment characteristics of two samples derived from patients who maintain complete remission to establish (A) ALL-207 and (B) ALL- 214 in the VXL-treated and non-treated mice over the monitoring period. Dashed black lines represent non-VXL treated mice while the solid red lines represent VXL-treated mice. The solid black squares indicate the VXL treatment period.

172 4.3.1.1 Summary of xenograft engraftment kinetics Eight ALL xenografts derived from the first 10 patient samples were successfully established and showed leukaemia cells to certain levels in the PB of mice during the monitoring period. Overall, 6 samples representing patients with different outcomes engrafted within one to less than two months of inoculation, 2 samples required longer time to engraft and mice inoculated with 2 patient samples did not show evidence of engraftment over the monitoring period (ALL-201 and ALL-212). The engraftment of most patient samples was delayed for less than one month in the VXL-treated mice relative to the non-VXL treated mice. The majority of mice inoculated with patient cells showed consistent patterns of engraftment within each group of mice inoculated using the same condition, although one mouse inoculated to establish ALL-202 (Figure 4.2A) and two mice inoculated to establish ALL-221 (Figure 4.2E) engrafted earlier than other mice within the same group upon VXL treatment. The engraftment of patient cells continued to progress in murine PB and high engraftment levels were achieved in most of the successfully engrafted mice with the exception of those inoculated with patient cells to establish ALL-218 (Figure 4.2D) and ALL-207 (Figure 4.3A). The details of engraftment kinetics are shown in Table 4.2.

173

Table 4.2. Summary of engraftment kinetics of the first panel of patient samples

No of engrafted Median to EFS at TT1% (Range) Median to EFS at TT25% (Range) Median to EFS at TTL (Range) Mice Xeno ID C T C T T-C C T T-C C T T-C

ALL-201 0/3 0/3 >259 >259 N/A >259 >259 N/A >259 >259 N/A ALL-202 3/3 3/3 43.9 (36.1-45.7) 51.2 (25.1 -51.7) 7.2 198.4 (137.5->267) 175 (129-231) 1* >259 231 (201-267) 1* ALL-203 3/3 2/2 36.8 (36.1-37.1) 62.4 (61 -63.8 25.6 62.3 (58.8-68.5) 86 (80.6-91.5) 23.7 128 156 (142-170) 28 ALL-209 3/3 3/3 43.5 (41.7-43.8) 57.3(54.6-59.6) 13.8 60.6 (57.8-64.6) 73.7(70.6-75.7) 13.1 91 (83-91) 111 20

ALL-218 2/2 0/3 177 (176.6- 177.5) >267 >90 >267 >267 N/A >267 >267 N/A ALL-221 3/3 3/3 81.6 (81.3-86.4) 93.3 (89.7-132.8) 11.7 145 (143.8-150.8) 121.5 (112-7-237.1) 1* 218 (200-218) 155 (155-265) 1* ALL-227 3/3 3/3 57.4 (52.3-57.5) 78.7 (56.5-80.2) 21.3 87.8 (87.7-100.1) 119 (104.1-125) 31.2 128 153 (142-154) 25 ALL-207 3/3 3/3 42 (39.9-44.9) 80.4 (79.8-85.3) 38.4 80.4(78.9-84) >128 >47.6 84 128 44

ALL-214 3/3 3/3 40.3(38.5-42.5) 73.7 (48.8-77.5) 33.4 73.4(69.3-75.6) 123.2 (122-132.9) 49.7 115 177 (170-190) 62 ALL-212 0/3 0/3 >259 >259 N/A >259 >259 N/A >259 >259 N/A

Xeno, Xenograft; C, control; T, VXL-treated; T-C, VXL-treated - control; >, “more than” denotes a patient sample which never engrafted or engrafted but did not reach an event; N/A, Not applicable when the difference caused by the VXL treatment cannot be calculated due to patient sample never engrafted. 1* denotes that there was a negative difference between T-C.

174 4.3.2 Xenografts established from the second panel of primary ALL patient samples

The second panel of 10 patient primary samples included 3 samples derived from early relapsed patients, 2 samples derived from late relapsed patients and 5 samples derived from patients in CR1.

The samples derived from early relapse patients in this panel were characterised by various patterns of engraftment and responses to treatment (Figure 4.4A, B and C).

ALL-210 was derived from a female patient who developed BM relapse and died due to bone marrow transplantation related mortality. This patient sample failed to engraft in the PB of 4 mice and barely reached 3% in the PB of one of the non-VXL treated mice

(Figure 4.4A). The engraftment data for this patient sample was available for only 5 mice as one of the non–drug treated mice was excluded due to appearance of an abdominal mass after 5 months of monitoring with no evidence of engraftment in mouse organs (spleen and BM) and the PB. In contrast, ALL-211 xenograft, which was derived from a male patient who firstly relapsed in the CNS and subsequently relapsed in the BM and CNS and died of the disease, demonstrated prolonged time to appearance of human cells in the PB of the non-VXL treated mice and showed a slightly longer latency period in the VXL-treated mice (Figure 4.4B). The engraftment levels in all mice continued to increase with remarkably concordant patterns within each group of mice. ALL-216 was derived from a female patient who relapsed at the first occurrence in the CNS followed by second relapse in BM and CNS and died of the disease.

Samples of this patient consistently engrafted quickly in the PB of all the non-VXL treated mice whereas a considerably longer time was required for human cells to appear in the PB of VXL-treated mice (Figure 4.4C). Leukaemia cells continued to increase in the PB of the control mice with the same consistent pattern until high engraftment was 175 achieved. However, considerable inter-mouse variability in engraftment was observed between mice in the VXL-treated group (Figure 4.4C).

Inoculation of samples from the two late relapse patients into mice led to successful establishment of ALL-228 and ALL-229. ALL-228 was derived from a male patient who experienced BM relapse and ALL-229 was derived from a female patient who developed combined BM and CNS relapse and at last follow up both patients were in remission after further treatment. These xenografts exhibited a long latency period appearing in the PB of the control mice and showed much further delay in appearance of cells in the PB of the VXL-treated mice (Figure 4.4D and E). A high level of engraftment was achieved in the majority of mice used to establish these two xenografts

176 50 100 ALL-210 A ALL-211 B ALL-211 VXL 40 ALL-210 VXL 80

30 60 20 40 10 %Hu CD45+ in PB % HuCD45 in PB 20 0 0 -10 0 25 50 75 100 125 150 175 200 225 250 275 0 25 50 75 100 125 150 175 200 225 250 Days post inoculation Days post inoculation

100 100 ALL-216 C ALL-228 D ALL-216 VXL ALL-228 VXL 80 80

60 60 in PB in +

40 40 % huCD45 20 % Hu CD45+ in PB 20

0 0

0 25 50 75 100 125 150 175 200 225 250 0 25 50 75 100 125 150 175 200 225 250

Days post inoculation Days post inoculation

100 ALL-229 E ALL-229 VXL 80

60 40

%Hu CD45+ in PB 20 0 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation

Figure 4.4. Engraftment of samples derived from the relapsed patients in the second panel. Graphs show the engraftment characteristics of the three samples derived from early relapse patients to establish A) ALL-210, B) ALL-211 and C) ALL-216 and two samples derived from late relapse patients to establish D) ALL-228 and E) ALL- 229 in the VXL-treated and non-treated mice over the monitoring period. Dashed black lines represent control mice while the solid red lines represent VXL-treated mice. The solid black squares indicate the VXL treatment period.

177 ALL xenografts developed from patients who remain in long-term remission were established with diverse patterns of engraftment and response to VXL treatment. ALL-

206 was derived from a male patient who maintained a CR for more than 10 years. This sample engrafted gradually in the PB of the non-VXL-treated mice and the level of engraftment continued to increase in the PB of all control mice until a high level was achieved (Figure 4.5A). However, very low level of engraftment was detected in the PB of one mouse of the VXL-treated group over the monitoring period with no appreciable engraftment in the other 2 mice.

ALL-208 was derived from a female patient who remained in CR for more than five years. Samples of this patient engrafted slowly in all non-VXL treated mice and the engraftment was maintained at lower than 20% in the murine PB (Figure 4.5B). Mice that received VXL treatment exhibited a further delay in appearance of cells in the PB and consistently maintained the engraftment at lower levels over the monitoring period.

ALL-224 is another example of a poorly engrafting sample that was also derived from a female patient who remained in CR for more than 10 years. The sample from this patient engrafted poorly in all non-VXL treated and VXL-treated mice with levels hardly reaching 5% over the monitoring period (Figure 4.5C), and leukaemia cells were not detected in the PB or body organs (spleen, BM) of all the VXL-treated mice at harvest.

In contrast, ALL-213, which was derived from a female patient who remained in CR for more than 9 years, demonstrated robust engraftment in the PB of all the non-VXL treated mice compared to other xenografts (Figure 4.5D). Notably, this xenograft exhibited high and consistent levels of engraftment within each group of mice. In

178 addition, the VXL treatment exerted a moderate delay in engraftment and progression of cells in the PB of all the VXL-treated mice (Figure 4.5D). ALL-223 was derived from a female patient who remained in CR for more than 9 years. The sample from this patient exhibited prolonged time to engraftment in the PB of all non-VXL treated mice, however, the percentage of engrafted cells continued to increase and reached high levels before cell harvest (Figure 4.5E). VXL treatment in turn exerted an appreciable delay in engraftment and progression of cells in the PB of all the VXL-treated mice and only one mouse showed evidence of high engraftment (Figure 4.5E).

179

100 100 ALL-206 A ALL-208 B ALL-206 VXL 80 80 ALL-208 VXL

60 60 in PB in in PB in + +

40 40 % huCD45 20 % huCD45 20

0 0

0 25 50 75 100 125 150 175 200 225 250 275 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Days post inoculation

50 100 ALL-224 C ALL-213 D ALL-224 VXL ALL 213VXL 40 80

30 60 in PB in in PB in + + 20 40 10 %huCD45 % huCD45 20 0 0 -10 0 25 50 75 100 125 150 175 200 225 250 275 0 25 50 75 100 125 150 175 200 225 250 Days post inoculation Days post inoculation

100 ALL-223 E ALL-223 VXL 80

60 in PB in + 40

% huCD45 20

0 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Figure 4.5. Engraftment of samples derived from the CR patients in the second panel. Graphs show the engraftment characteristics of five samples derived from patients who remained in CR1 to establish (A) ALL-206, (B) ALL-213, (C) ALL-224 and (D) ALL-213, and (E) ALL-223 in control and VXL-treated mice over the monitoring period. Dashed black lines represent control mice while the solid red lines represent the VXL-treated mice. The solid black squares indicate the VXL treatment period. 180 4.3.2.1 Summary of xenograft engraftment kinetics All samples in the second panel of IR ALL patients showed evidence of engraftment in the PB of at least one mouse per xenograft. Three patterns of engraftment were seen in xenografts established from this panel: 2 samples engrafted in the non-VXL treated mice within one to less than 2 months of inoculation (ALL-216 and ALL-213), 5 patient samples engrafted within a range of more than two months to less than four months of inoculation, 2 samples showed very long latency period, but no engraftment was detected in most of the mice inoculated with one patient sample used to establish ALL-

210. A constant increase in the level of engraftment with a uniform pattern of engraftment between mice was observed for most xenografts. The VXL treatment in turn caused more than one-month delay in engraftment of mice inoculated with all patient samples except those inoculated with cells used to establish ALL-211 and ALL-

213. The engraftment kinetics at different time points of progression are summarised in

Table 4.3.

181 Table 4.3. Summary of engraftment kinetics of the second panel of patient samples

No of Median EFS at TT1% (Range) Median EFS at TT25% (Range) engrafted Median EFS at TTL (Range)

Mice Xeno ID C T C T T-C C T T-C C T T-C ALL-216 3/3 3/3 42.3(39.3-48.7) 147.5 (115.4-158.7) 105.2 59.5 (57.1-65.7) 178.4 (130.8-183.1) 118.9 81 205 (152 -226) 124 ALL-210 1/2 0/3 >265 >265 N/A >265 >265 N/A >265 >265 N/A ALL-211 3/3 3/3 99.7 (95.1-100.5) 113.7(112.3-116) 14 130.3 (125.9-137.9) 154 (151-157.8) 23.8 171 200 29

ALL-229 3/3 3/3 94.9 (81.5-95.3) 186.4 (144.7-202.3) 91.5 126.1 (125.9-130.3) 230.6 (203.2-248) 104.5 187 >265 (265- >265) >78

ALL-228 3/3 3/3 79.6 (75.5-80.7) 128.8 (90.8-141.9) 49.2 116.4 (111.5-119.2) 171.9 (159.5-196.2) 55.4 176 239 63 ALL-208 3/3 3/3 161.1(141.6-163.8) 207.6 (205.9-209.5) 46.5 >265 >265 N/A >265 >265 N/A ALL-213 3/3 3/3 29.5 (24.3-30.2) 57.6 (49.8-58.3) 28.1 39 (38.9-40.3) 70.8 (66 -75.3) 31.8 57 88 31 ALL-223 3/3 3/3 107.5 (106.6-107.8) 158.3 (156.7-158.3) 50.7 144.3 (142.9-168.6) >265 (198->265) >120.7 170 (170-205) >265 (220->265) >95 ALL-206 3/3 1/3 81.4 (80.4-86.0) >265 (123->265) >183.6 135.3 (127.6-141) >265 >129.7 220 (205-220) >265 >45

ALL-224 3/3 0/3 170.4 (169.1-187.7) >265 N/A >265 >265 N/A >265 >265 N/A

Xeno, Xenograft; C, control; T, VXL-treated; T-C, VXL-treated - control; >, “more than” denotes a patient sample which never engrafted or engrafted but did not reach an event; N/A, Not applicable when the difference caused by the VXL treatment cannot be calculated due to patient sample never engrafted

182 4.3.3 Xenografts established from the third panel of primary ALL patient samples

The third panel of 10 patient primary samples included 3 samples derived from early relapsed patients, 4 samples derived from late relapsed patients and 3 samples derived from patients who remain in CR1.

Inoculation of samples derived from patients who relapsed early led to establishment of three xenografts (Figure 4.6A-C). ALL-215 was derived from a female patient who developed BM relapse and maintained a complete remission after further treatment.

This patient sample engrafted quickly and exhibited high and uniform patterns of engraftment in the PB of all of the non-VXL treated mice (Figure 4.6A). The VXL treatment did not greatly impede the engraftment of leukaemia cells in the PB of all of the VXL-treated mice or change the pattern of growth over the monitoring period. The engraftment pattern of ALL-215 was remarkably similar to ALL-66, both of which were derived from the same patient sample (compare Figures 3.6B and 4.6A). ALL-217 was derived from a male patient who experienced BM relapse but maintained a CR after additional treatment. The time required for appearance of xenograft cells in the PB of the non-VXL treated mice was longer than that observed in ALL-215 (Figure 4.6B).

This patient sample displayed inter-mouse variability in engraftment within the non-

VXL treated mice. The VXL treatment caused a varied pattern of appearance of leukaemia cells in the PB of mice, with one rapid, one intermediate and one non- engrafting mouse (Figure 4.6B). The third xenograft, ALL-219 was established from a patient who developed CNS relapse and died of the disease. This patient sample engrafted slowly, with a gradual increase in the level of human cells in the PB of the non-VXL treated mice. No engraftment data were obtained from mice that received

VXL treatment. Of the three VXL-treated mice, one was found dead after two weeks 183 from the inoculation, one mouse had unexplained weight loss and deterioration of well being, and the last mouse succumbed to thymoma and thus these two mice were euthanased before any evidence of engraftment in the PB (Figure 4.6C).

Xenotransplantation of mice with samples from the late relapse patients resulted in establishment of four xenografts (Figure 4.6D-G). The engraftment of these samples was relatively slow in the non-VXL treated mice and was largely delayed in the VXL- treated mice. As an example, ALL-222 was established from a female patient who developed a BM relapse and maintained a CR after additional treatment. This patient sample consistently engrafted at a slow rate in the PB of all the non-VXL treated mice, but one mouse showed a faster increase compared to other non-VXL treated mice

(Figure 4.6D). Mice that received VXL treatment exhibited a large delay in appearance of cells and a gradual increase in the proportion of cells in two mice over the monitoring period. The rest of patient samples (ALL-225, ALL-226 and ALL-230) also showed slow engraftment in the PB of the non-VXL treated mice and levels increased gradually over the monitoring period (Figure 4.6E-G). This observation was made based on engraftment of two out of the three mice, as one mouse from every non-VXL treated group had to be euthanased before any evidence of engraftment due to deterioration of health. In the VXL-treated mice, a large delay in engraftment was observed in all patient samples except a moderate delay in engraftment of the patient sample used to establish

ALL-225 (Figure 4.6E). Furthermore, most mice inoculated with patient cells reached the maximum holding time without evidence of TTL features (Figure 4.6E-G).

184

100 100 ALL-215 A ALL-217 B ALL-215 VXL ALL-217 VXL 80 75

60 in PB in in PB in + + 50 40 25 % hu CD45 20 % huCD45

0 0

0 25 50 75 100 125 150 175 200 0 25 50 75 100 125 150 175 200 225 250 275

Days post inoculation Days post inoculation

100 100 ALL-222 D ALL-219 C ALL-222 VXL ALL-219 VXL 80 80 in PB in

+ 60 60 in PB in +

40 40 %huCD45 20 20 % Hu HLA-DR

0 0

0 25 50 75 100 125 150 175 200 225 250 0 25 50 75 100 125 150 175 200 225 250 275

Days post inoculation Days post inoculation

100 100 ALL-225 E ALL-226 F ALL-225 VXL ALL-226 VXL 80 80 in PB in

60 + 60

40 40 %Hu CD45+ in PB

20 % Hu HLA-DR 20

0 0

0 25 50 75 100 125 150 175 200 225 250 275 0 25 50 75 100 125 150 175 200 225 250 275

Days post inoculation Days post inoculation

185

100 ALL-230 G 80 ALL-230 VXL

in PB in 60

+

40

% huCD45 20

0 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation

Figure 4.6. Engraftment of samples derived from the relapsed patients in the third panel. Graphs show the engraftment of three patient samples derived from early relapse patients used to establish (A) ALL-215, (B) ALL-217 and (C) ALL-219 and four samples derived from late relapse patients used to establish (D) ALL-222, (E) ALL-225, (F) ALL-226 and (G) ALL-230 in the VXL-treated and non-treated mice over the monitoring period. Dashed black lines represent the VXL-treated mice while the solid red lines represent VXL-treated mice. The solid black squares indicate the VXL treatment period.

186 Inoculation of samples derived from patients who maintained a CR1 resulted in establishment of three xenografts (Figure 4.7A-C). ALL-220 was derived from a female patient who had been in remission for more than 7 years. Engraftment of this patient sample in the non-VXL treated mice was relatively slow but levels continued to increase over the monitoring period in all engrafted mice (Figure 4.7A). However, the engraftment was not apparent in the PB of any mice upon treatment with VXL chemotherapy over the monitoring period. The pattern of engraftment of ALL-220 was almost identical to the of ALL-65, and both were derived from the same patient sample

(compare Figure 3.4B and 4.7A). All patient samples used to establish other xenografts

(ALL-231 and ALL-232) in the non-VXL treated mice demonstrated a slow pattern of engraftment comparable to that observed in ALL-220 although the time to engraftment varied between mice used to establish ALL-231 (Figure 4.7B and C). However, except for one mouse in the VXL-treated group used to establish ALL-231, none of the mice that received VXL chemotherapy showed evidence of engraftment over the monitoring period (Figure 4.7A, B and C).

187

100 ALL-220 A 100 ALL-231 B ALL-220 VXL ALL-231 VXL 80 80 in PB in

60 + in PB in 60 + 40 40

% huCD45 20 20 % Hu HLA-DR 0 0

0 25 50 75 100 125 150 175 200 225 250 275 0 25 50 75 100 125 150 175 200 225 250 275

Days post inoculation Days post inoculation

100 ALL-232 C ALL-232 VXL 80

60 in PB in + 40 % huCD45 20

0

0 25 50 75 100 125 150 175 200 225 250 275

Days post inoculation

Figure 4.7. Engraftment of samples derived from patients who maintain complete remission in the third panel. Graphs show the engraftment of three patient samples derived from CR patients used to establish (A) ALL-220, (B) ALL-231 and (C) ALL- 232 in control and VXL-treated mice over the monitoring period. Dashed black lines represent control mice while the solid red lines represent VXL-treated mice. The solid black squares indicate the VXL treatment period.

188 4.3.3.1 Summary of xenograft engraftment kinetics All patient samples transplanted into mice from this panel engrafted to certain levels in all of the non-VXL treated mice with the exception of those eliminated from the study.

There was one sample from a relapsed patient that showed robust engraftment in the PB of the non-VXL treated mice and the fast pattern of engraftment was not hampered by the VXL chemotherapy in the VXL-treated mice (Figure 4.6A). The engraftment of 8 patient samples ranged from more than two to less than 4 months following inoculation in the non-VXL treated mice but longer time to engraft was required for mice inoculated with the patient sample used to establish ALL-231. The engraftment of patient samples in the VXL-treated mice was delayed for more than one month relative to the non-VXL treated mice. The engraftment of these patient samples in the VXL-treaA summary of engraftment kinetics of these patient samples is provided in Table 4.4.

189 Table 4.4 Summary of engraftment kinetics of the third panel of patient samples

No of engrafted Median EFS at TT1% (Range) Median EFS at TT25% (Range) Median EFS at TTL (Range) Mice Xeno ID C T C T T-C C T T-C C T T-C ALL-215 3/3 3/3 36.6 (31-38.1) 44.3 (42.7-44.5) 7.7 54.6 (54.1-56.8) 73.3 (73.3-77.4) 18.7 94 113 (113 -130) 19 ALL-217 3/3 3/3 72.4 (64.5-73.4) 143.1 (93.7->267) 70.4 103.8 (73.6-125.7) 173.9 (116.7->267) 70.1 130 (102-145) 211 (149->267) 81 ALL-219 3/3 0 93.4 (87.4-93.9) XX N/A 161.8 (140.8-165.8) XX N/A 211 (161-211) XX N/A ALL-230 2/2 3/3 75 (74.5-75.4) 134.3 (117-148.9) 59.3 >265 >261 (178.7->269) N/A >265 >269 (261- >269) N/A ALL-226 2/2 3/3 115.3 (111.8-118.8) 200.8 (185-245.5) 85.5 201.6 (191.1->211) >267 >65.4 236.5 (212->261) >267 >30.5

ALL-222 3/3 2/3 66.3(62.4-71.6) 159.8 (131.1->267) 93.5 99.8 (84.1-111.1) 224.6 (188.8->267) 124.8 145 (113-164) >265 >122 ALL-225 2/2 3/3 104.8 (97.4-112.3) 144.5 (128.6-148.7) 39.6 157.2 (147.2-167.1) 194.5 (164.7-197.7) 37.3 221 >268 >47 ALL-220 3/3 0/3 117.5 (112.8-118.2) >267 50.7 184.3 (176.3-210.7) >267 >82.7 225 (225-261) >267 >42 ALL-232 3/3 0/3 84.3 (83.6.4-94) >267 >182.7 126.4 (123.1-144.2) >267 >140.6 163 >267 >104 ALL-231 3/3 1/3 164.2 (106.4-196.9) >267 (149.2- >267) >102.8 >261 (175.7->269) >267 (232.8->267) N/A 261 (204 ->267) >267 (245->267) >6

Xeno, Xenograft; C, control; T, VXL-treated; T-C, VXL-treated - control; >, “more than” denotes a patient sample which never engrafted or engrafted but did not reach an event; N/A, Not applicable when the difference caused by the VXL treatment cannot be calculated due to patient sample never engrafted or due to indefinite growth difference between the VXL-treated and non-treated mice. XX indicate no engraftment data due to exclusion of all the three mice from the study

190 4.4 Assessment of the reliability of combining the engraftment data from the entire cohort for analysis of relapse prediction

The results presented in Section 4.3 described developing xenografts from the entire cohort of 30 patient samples, which was carried out in three experiments according to a similar protocol. The main goal from establishing IR ALL xenografts is to develop an improved strategy for predicting IR ALL patient outcomes based on engraftment of their samples. Thus, it would be more efficient to combine the engraftment features of patients from the three panels into one dataset to increase the statistical power of data for estimation of xenograft ability to predict patient outcomes. However, as xenografts were developed in three separate experiments, the reliability of gathering data from the three panels needs to be checked for pre and/or post-processing sources of bias that could compromise the consistency between the data generated from multiple experiments. I thus analysed whether the distribution of patient clinical characteristics and the engraftment data are significantly different between the three panels of patient samples before I combined the data from the three panels.

4.4.1 Patients’ clinical characteristics

Given the heterogeneity within IR ALL patients, it was essential to ensure that patients with diverse clinical features were equally randomised between the three panels. I thus investigated whether or not the distribution of patients according to the duration of complete remission and other baseline clinical covariates of patients were significantly different among all the three panels. Figure 4.8 illustrates the distribution of patients according to the length of CR1 between the three panels. It shows that the relapse-free survival in the first panel of 10 patients is not significantly different from that for patients in the second panel (P= 0.4162, log-rank test) and also for patients in the third panel (P=0.9917, log-rank test). Similarly, there was no statistically significant

191 difference in length of CR between patients in the second and third panels (P=0.4318, log-rank test).

100 First Panel 80 Second Panel Third Panel

60 3 ER, 2 LR and 3 CR1

40

Percent survival Percent 3 ER, 4 LR and 3 CR1

3 ER, 4 LR and 3 CR1 20

0 0 25 50 75 100 125

Length of complete remission 1 (CR1)

Groups Log-rank P value First panel versus Second panel 0.4162 First panel versus Third panel 0.9917 Second panel versus Third panel 0.4318

Figure 4.8. Distribution of patients based on duration of complete remission. Kaplan-Meier survival curves compare the remission status between the three panels of patients. The first panel versus the second panel, the first panel versus the third panel and the second panel versus the third panel. ER, early relapse, LR, late relapse and CR1, complete remission 1. The log-rank test was used to test the statistical difference in EFS of the patients.

192 A good randomisation strategy of patient samples between the three panels should also take into account that some baseline clinical characteristics of ALL patients including age, sex, WCC, cytogenetic features and MRD levels might be related to patient outcomes. As mentioned previously, the NCI risk criteria stratify ALL children aged

>10 years or with a diagnostic WCC > 50x109/L as patients at high risk of relapse.

Further, ALL patients who present with certain cytogenetic features in their leukaemia cells and low MRD values at different time points usually have a good prognosis. Thus it is required to ensure that the clinical features of patients allocated for each panel are not significantly different from those of patients in the other panels. A comparison of age between patients of the three panels showed that 60-70% of patients were aged less than 10 years old at time of diagnosis in all the three panels whereas 20-30% of patients were aged more than 10 years old (Figure 4.9A). The distribution of patients according to the WCC at diagnosis was also comparable between patients from the three panels

(80-90% of patients presented with WCC < 50 x109/L and 10-20% presented with WCC

> 50 x109/L) (Figure 4.9B).

Most ALL patients who are stratified into the IR ALL risk subtype present at diagnosis with favourable cytogenetic abnormalities include hyperdiploidy and the ETV6-RUNX1 translocation and others may present with unfavourable features including TCF-PBX1, and “B others”. In this cohort, the favourable and unfavourable prognostic features were evenly distributed among all the three panels of patients (40-50% of patients presented with favourable cytogenetic features and 44-60% presented with unfavourable cytogenetic features. One patient in the second panel presented with unclear cytogenetic features (Figure 4.9C).

193 When the distribution of patients was assessed between the three panels based on the site of relapse, both the first and third panels include 50% of patients who developed a

BM relapse and 30% of patients who maintained a CR1, whereas only 20% of patients relapsed in the BM and 50% of patients maintained CR1 in the second panel of patients.

Each of the second and third panels had 20% of patients who relapsed in the CNS and there were 3 patients in the first and second panels who presented with combined BM and CNS relapse (20% and 10% consecutively) (Figure 4.9D).

The MRD levels at day 15 and 33 were also shown to be predictors of outcome in ALL patients (Karsa et al., 2013). Four patterns of MRD levels were observed in patients of the entire cohort: patients presented with indefinable MRD levels; negative MRD levels; positive MRD with levels below or just around the quantifiable threshold

(>1x10-5); and high MRD with levels ranging from >1x10-4 to >1x10-2. In regard to

MRD at day 15, a comparison of MRD values between the three panels revealed that

40-80% of patients in all the three panels were MRD positive, while 20% of patients in the first and third panels had high MRD values (Figure 4.9E). The MRD value was not tested for 20% of patients in the first and second panels and 40% of patients in the third panel (Figure 4.9E). Analysis of the distribution of patients according to the MRD values at day 33 between the three panels showed that 70% and 80% of patients in the first and second panel, respectively, had high MRD levels whereas 40% of patients in the third panel manifested high MRD levels (Figure 4.9F). There were 20% of patients in the first panel and 10% of those in the second panel who presented with positive

MRD values whereas 50% of patients in the third panel manifested positive MRD values. However, 10% of patients in the first and second panels had negative MRD values, and a similar proportion in the third panel was not tested for MRD.

194 A B 100 100 WCC > 50X109 Age >10 yrs 80 80 WCC < 50X109 Age < 10 yrs 60 60 40 40 % patients (WCC)

% patients (Age) 20 20 0 0

First panel Second panel Third panel First panel Second panel Third panel

Group Chi Square Group Chi Square test test First panel versus second panel 0.6390 First panel versus second panel 0.3922

First panel versus third panel 0.6390 First panel versus third panel 0.3922 WCC Age Second panel versus third panel 1 Second panel versus third panel 1

C D 100 100 Favourable= ETV6-RUNX1, Hyperdiploid No relapse 80 80 BM relapse Unfavourable= B OTHER, CNS relapse TCF3-PBX1 60 BM and CNS relapse 60 Unclear 40 40

% patients (Site of relapse) 20 20

% patients (Cytogenetic feautures) 0 0

First panel Second panel Third panel First panel Second panel Third panel Group Chi Square Group Chi Square test test First panel versus second panel 0.5795 First panel versus second panel 0.3553

First panel versus third panel 0.6530 First panel versus third panel 0.5737 site

Re lapse Second panel versus third panel 0.0188

Cytogentic Second panel versus third panel 0.5737

E F 100 100 MRD negative MRD negative MRD positive 80 MRD positive 80 MRD high MRD high Not tested Not tested 60 60

40 40

20 20 % patients (MRD value at day 33) % patients (MRD value at day 15)

0 0

First panel First panel Third panel Third panel Second panel Second panel Group Chi Square Group Chi Square test test First panel versus second panel 0.3189 First panel versus second panel 0.3189 First panel versus third panel 0.5866 First panel versus third panel 0.2504 day15 day 33

MRD at Second panel versus third panel 0.1353 MRD at Second panel versus third panel 0.1116

Figure 4.9. Distribution of patients between the three panels based on patient clinical features. The distribution of patients between the three panels is shown based on six clinical factors. (A) Age, (B) WCC, (C) Cytogenetic features, (D) Site of relapse, (E) MRD level at day 15 and (F) MRD level at day 33. The tables below graphs evaluate the statistical differences in clinical features between patient groups.

195 Overall, there was no notable difference in the distribution of patients based on age,

WCC and the cytogenetic features between the three panels. Based on the site of relapse, the first and second panels equally included a similar number of patients who experienced BM relapse and patients who maintain CR but in both panels there were two patients who had different patterns of relapse sites. However, the second panel contained a lower number of patients who experienced BM relapse and more CR patients in this panel.

The analysis of patient distribution based on the available data of MRD levels at day 15 showed that most patients from every panel presented with positive MRD levels at day

15 although a greater proportion was observed in the second panel. Both the first and third panels had two patients who presented with high MRD levels at day 15. A comparison of the MRD values between patients of the three panels revealed similar levels among patients in the first and second panels as many patients in both panels had high MRD levels and fewer of whom had positive and negative MRD values. However, the third panel had more patients who presented with MRD positive levels compared to the other panels. The overall observations indicated that most of the clinical features were sufficiently randomised between the three panels. This suggested that the clinical features of the patients were unlikely to lead to significant bias in engraftment of patient samples among the three panels.

196 4.4.2 Engraftment characteristics

The analysis was then aimed to determine whether the reliability of compiling the engraftment data of patient samples for predicting patient outcomes could be biased by failure to control for unknown confounding factors introduced during the process of xenograft establishment. The integrity of the data was assessed based on testing the difference in the engraftment parameters of patient samples between the three panels established in the mice. Assessment of data normality within each dataset revealed that the engraftment data were not normally distributed and I thus applied non-parametric statistics to test the difference in engraftment between the three panels based on the EFS of first mouse and median mouse at every time to event endpoint.

A comparison of engraftment based on median EFS of the control mice at 1% human cells in the PB revealed that there was no statistical difference in engraftment of patient samples from the first panel when compared with the second panel (P=0.5281) and the third panel (P=0.4682) or between the second and the third panels (P=0.6305) (Figure

4.10A). Similarly, there was no significant difference between the EFS of first mouse to engraft with patient samples from the three panels (Figure 4.10A). In concordance, engraftment data of patient samples based on EFS of median or first mouse to reach

25% human cells in the PB and TTL exhibited comparable patterns of engraftment between the three panels of patients (Figure 4.10B and C).

197

0.6447 P= 0.6305 P= P= 0.8649 P= 0.9550 A P= 0.5281 P=0.5659 B P= 0.9885 P= 0.9247 P= 0.4682 P= 0.4693 P= 0.6949 P= 0.8532 350 350 333 300 300 200 Median EFS 250 First mouse Median EFS EFS First mouse 150 EFS 200

100

Days to 1% engraftment 1% to Days 150

Days to 25% engraftment 25% to Days 50 100

0 50 30

First Panel Third panel First Panel Third Panel Second Panel Second Panel First Panel Third Panel First Panel Third Panel Second Panel Second Panel

P= 0.9109 P= 0.8067

P= > 0.9999 P= 0.7503 C P= 0.6656 P= 0.8677 350 333.0 300 240 225 Median EFS First mouse 200 EFS 175 150 125

leukaemia to time to Days 100 75 50

First Panel Third Panel First Panel Third Panel Second Panel Second Panel Figure 4.10. Assessment of the difference in the engraftment profiles of patient samples between the three panels at various time points. The time to engraftment (median or first control mouse) were compared between the three cohorts at (A) TT1% human cells, (B) TT25% human cells and (C) time to leukaemia (TTL). Bars represent the median values.

198 4.5 Systematic analysis of modelling IR ALL patient outcomes in patient derived xenografts

The results presented in the previous section demonstrated the suitability of merging the engraftment data of the three panels to increase the statistical power of data for estimation of xenograft ability to predict patient outcomes. I thus gathered the engraftment data of the 30 IR ALL patients into one large dataset that lists the engraftment kinetics of all patient samples, generated from the VXL-treated and non- treated mice, and the LGD values at the three time to event endpoints (TT1%, TT25% human cell or TTL). The engraftment data were available for all mice except the three

VXL-treated mice which received patient cells to establish ALL-219, and another 6 mice, which were excluded from the study due to deterioration of health from factors not related to the leukaemia disease. Of the 180 mice that received cells from IR ALL patients, the engraftment data were available from 85 (94.4%) control mice, including

33 mice representing xenografts of the 11 patients in CR1, 26 mice representing xenografts of the 10 late relapsed patients and 26 mice representing xenografts of the 9 early relapsed patients. In addition, 86 (95.6%) mice were available for analysis from

VXL-treated mice, including 33 mice representing xenografts of the 11 patients in CR1,

30 mice representing xenografts of the 10 late relapsed patients and 23 mice representing 8 of the 9 early relapsed patients. Overall, data were available from 171

(95%) of mice originally inoculated with patient samples. A summary of engraftment of these mice is shown in Figure 4.11.

To take the full advantage of the data, I decided to carry out a systematic analysis of engraftment data for selecting the most appropriate criteria allowing the maximum difference in engraftment of samples from the three patient groups. With guidance from two biostatisticians at CCI, the analysis started by asking whether the engraftment of 199 patient samples in the control or the VXL-treated mice would reflect outcomes of patients from whom xenografts were derived. The second question focused on answering which time to event endpoint (TT1%, TT25% or TTL) may allow significant differences in EFS between mice inoculated with cells from early, late relapsed patients and those who remain in CR1. I then proceeded to determine whether measuring the

EFS based on the EFS of the first or the median mouse to reach event would improve my ability to stratify ALL xenografts according to patient outcomes. Subsequently, the engraftment criteria with superior ability to distinguish between xenografts of the three patient groups were prioritised for testing the capability of xenografts to predict IR ALL patient outcomes. The following section details the analysis of engraftment criteria between mice used to establish ALL xenografts from patients with different outcomes.

200

Control mice VXL-treated mice

100 100 A B

80 80

60 60

40 40

20 %Hu CD45 or HLA-DR in PB 20 %Hu CD45 or HLA-DR in PB

0 0 0 25 50 75 100 125 150 175 200 225 250 275 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Days post inoculation Xenograft of patients in CR1 (Graph A & B)

ALL-213 ALL-207 ALL-223 ALL-231 ALL-232 ALL-212 ALL-214 ALL-206 ALL-208 ALL-220 ALL-224

100 100 C D

80 80

60 60

40 40

20 20 %Hu CD45 or HLA-DR in PB %Hu CD45 or HLA-DR in PB 0 0

0 25 50 75 100 125 150 175 200 225 250 275 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Days post inoculation

Xenografts of early relapse patients (Graph C & D)

ALL-211 ALL-216 ALL-201 ALL-203 ALL-215 ALL-210 ALL-202 ALL-219 ALL-217

100 100 F E 80 80

60 60

40 40

%Hu CD45+ in PB 20 20

%Hu CD45 or HLA-DR in PB 0 0

0 25 50 75 100 125 150 175 200 225 250 275 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation Days post inoculation Xenografts of late relapse patients (Graph E and F) ALL-221 ALL-229 ALL-222 ALL-225 ALL-218 ALL-209 ALL-228 ALL-226 ALL-227 ALL-230

201 Figure 4.11. Compiled engraftment data of the three panels of patient samples. Graphs show the engraftment patterns of samples, which were inoculated into control (A, C, E) and VXL-treated (B, D, F) mice to establish xenografts of CR1 patients (A, B), xenografts of early relapse patients (C, D) and xenografts of late relapse patients (E, F).

202 4.5.1 Impact of engraftment in VXL-treated and non-treated mouse models

To assess whether engraftment of IR ALL patients’ samples in control or the VXL- treated mice allow for better stratification of xenografts according to patient outcomes, the EFS of all mice at each event endpoint (TT1%, TT25% or TTL) was compared. This comparison involved stratifying according to patient outcome in the first instance. Mice were censored if they did not reach event during the maximum holding period.

A comparison of mouse EFS at TT1% revealed a significance difference in the number of censored and uncensored mice between the VXL-treated and non-treated mice that received cells from patients in CR1 (3 censored and 30 uncensored control mice versus

16 censored and 17 uncensored VXL-treated mice, P=0.008 Fisher's exact test) (Figure

4.12A). In contrast, no significant difference was apparent in mice inoculated with samples from late relapse patients (0 censored and 26 uncensored control mice versus 4 censored and 26 uncensored mice VXL-treated mice, P=0.1153 Fisher's exact test) or those inoculated with samples from patients who experienced early relapsed (4 censored and 22 uncensored control mice versus 7 censored and 16 uncensored VXL-treated mice, P=0.3063 Fisher's exact test). When the EFS at TT1% was compared based on patient outcomes within the control or VXL-treated groups, the only statistically significant difference in the number of censored and uncensored mice was between

VXL-treated mice that were inoculated with samples of patients in CR1 or late relapse

(17 censored and 16 uncensored “CR1” mice versus 4 censored and 26 uncensored “late relapse” mice, P=0.0033 Fisher's exact test) (Figure 4.12B). These results indicate that a significant number of the VXL-treated mice, but not the control mice, that received cells from patients in CR1 did not engraft compared to mice that received cells from relapsed

203 patients. These data highlight the advantage of VXL treatment to differentiate the pattern of 1% human cell engraftment between relapsed and non-relapsed patients.

A CR1 Late relapse Early relapse 33 30 Censored 27 Uncensored 24 21 18

mice 15 Fisher's exact test Group Censored Uncensored p value 12 CR1 9 Control 3 30 0.008 VXL-treated 16 17 6 Late relapse Number of censored & uncensored & censored of Number 3 Control 0 26 0.1153 VXL-treated 4 26 0 Early relapse Control 4 22 0.3063 VXL-treated 7 16 Control Control Control

VXL-treated VXL-treated VXL-treated

B Control mice VXL-treated mice 33 30 27 Censored Mice 24 Uncensored Mice 21 Fisher's exact test 18 Group Censored Uncensored p value mice 15 CR1 3 30 0.2478 12 Late relapse 0 26 CR1 3 30 9 0.6881 Early relapse 4 22 6 Control mice Control Number of censored & uncensored & censored of Number Late relapse 0 26 0.1104 3 Early relapse 4 22

0 CR1 16 17 0.0033 Late relapse 4 26 CR1 CR1 CR1 16 17 0.2697 Early relapse 7 16 Late relapseEarly relapse Late relapseEarly relapse Late relapse 4 26 VXL-treated mice VXL-treated 0.1767 Early relapse 7 16

Figure 4.12. Plots comparing the number of censored and uncensored mice at TT1%. The engraftment of patient samples in mice is presented based on the number of censored and uncensored mice at TT1% and numbers were compared between (A) the VXL-treated and control mice used to establish xenografts of each patient group, and (B) mice in both treatment or control groups based on patient outcomes.

204

Consistent with what I observed at TT1%, a significant difference in the number of censored and uncensored mice between the VXL-treated and non-treated groups at

TT25% was only observed between mice that received cells from patients in CR1 (11 censored and 22 uncensored control mice versus 25 censored and 8 uncensored VXL- treated mice, P=0.001 Fisher's exact test) (Figure 4.13A). Interestingly, the number of censored and uncensored VXL-treated mice at TT25% that received cells from patients in CR1 was significantly different from that observed in the VXL-treated mice that received cells from both early or late relapse patients (Figure 4.13B). For example, there were 25 censored and 8 uncensored VXL-treated mice in the CR1 group, and only 7 censored and 16 uncensored VXL-treated mice in the early relapse group (P=0.001

Fisher's exact test). These data indicate that the VXL chemotherapy caused a delay in progression of leukaemia cells to 25% engraftment in a larger number of mice that received cells from patients in CR1 compared to those that received cells from the relapsed patients, providing an ability to differentiate the pattern of 25% human cell engraftment between relapsed and non-relapsed patients.

205

A CR1 Late relapse Early relapse 33 30 27 24 Censored mice Uncensored mice 21 18

mice 15 Fisher's exact test Group Censored Uncensored p value 12 CR1 Control 11 22 0.001 9 VXL -treated 25 8 Late relapse

Number of censored & uncensored & censored of Number 6 Control 5 21 0.5372 VXL-treated 9 21 3 Early relapse Control 6 20 0.7471 0 VXL-treated 7 16

Control Control Control VXL-treated VXL-treated VXL-treated

B Control mice VXL-treated mice 33 30 27 Censored Mice 24 Uncensored Mice 21

18 Fisher's exact Group Censored Uncensored test p value mice 15 CR1 11 22 0.2552 12 Late relapse 5 21 CR1 11 22 0.5635 9 Early relapse 6 20 Control mice Control 6 Late relapse 5 21

Number of censored & uncensored & censored of Number 1 Early relapse 6 20 3 CR1 25 8 0.004 0 Late relapse 9 21 CR1 25 8 0.001 Early relapse 7 16 CR1 CR1

VXL-treated mice VXL-treated Late relapse 9 21 1 Early relapse 7 16 Late relapseEarly relapse Late relapseEarly relapse

Figure 4.13. Plots comparing the number of censored and uncensored mice at TT25%. The engraftment of patient samples in mice is presented based on the number of censored and uncensored mice at TT25% and numbers were compared between (A) the VXL-treated and control mice used to establish xenografts of each patient group, and (B) mice in both treatment or control group based on patient outcomes.

206 Based on TTL, the analysis reiterated the significant differences in the number of censored and uncensored mice between the VXL-treated and non-treated groups used to establish xenografts from patients in CR1 but also showed a significant difference in the number of censored/uncensored mice between the VXL-treated and non-treated mice used to establish xenografts of the late relapse patients (Figure 4.14A). For example, there were 10 censored and 23 uncensored control mice compared to 22 censored and

11 uncensored VXL-treated mice in the CR1 group (P=0.0063 Fisher's exact test). In addition, there were 5 censored and 21 uncensored control mice compared to 16 censored and 14 uncensored VXL-treated mice in the late relapse group (P= 0.0126

Fisher's exact test). When the number of censored versus encensored mice was compared based on patient outcome, there was a significant difference only between the

VXL-treated mice in the CR1 and early relapse groups (Figure 4.14B). Specifically, there were 22 censored and 11 uncensored VXL-treated mice in the CR1 group compared to 7 censored and 16 uncensored VXL-treated mice in the early relapse group

(P=0.0137 Fisher's exact test). These data showed that the VXL chemotherapy revealed a significant difference in the number of mice showing TTL between early relapse patients and patients in CR1 which could provide a means to distinguish early relapsed patients based on the early TTL of their samples in mice.

207

A CR1 Late relapse Early relapse 33 30 27 24 Censored mice 21 Uncensored mice 18

mice 15 Fisher's exact Group Censored Uncensored 12 test p value CR1 9 Control 10 23 0.0063 6 VXL -treated 22 11 Number of censored & uncensored & censored of Number 3 Late relapse Control 5 21 0.0126 0 VXL-treated 16 14 Early relapse Control 7 19 Control Control Control 1 VXL-treated 7 16 VXL-treated VXL-treated VXL-treated

B Control mice VXL-treated mice 33 30 27 Censored Mice 24 Uncensored Mice

Fisher's exact test Group Censored Uncensored 21 p value

18 CR1 10 23 0.3812 Late relapse 5 21 mice 15 CR1 10 23 1 12 Early relapse 7 19 Control mice Control Late relapse 5 21 0.7432 9 Early relapse 7 19

CR1 22 11 6 0.3127

Number of censored & uncensored & censored of Number Late relapse 16 14

3 CR1 22 11 0.0137 Early relapse 7 16 0

VXL-treated mice VXL-treated Late relapse 16 14 0.1615 Early relapse 7 16 CR1 CR1

Late relapseEarly relapse Late relapseEarly relapse

Figure 4.14. Plots comparing the number of censored and uncensored mice at TTL. The engraftment of patient samples in mice is presented based on the number of censored and uncensored mice at TTL and numbers were compared between (A) the VXL-treated and control mice used to establish xenografts of each patient group, and (B) mice in both treatment or control groups based on patient outcomes.

208 Regardless of the time to event endpoint, there was no statistically significant difference in the proportion of censored and uncensored mice in the control groups between xenografts established from patients with different outcomes. However, the VXL- treated group were characterised by a pattern of censored/uncensored mice at each time to event endpoint different between xenografts of patients in CR1 and xenografts of early and/or late relapsed patients. These data indicate that xenografts established under the influence of VXL chemotherapy are the most appropriate mouse model for reflecting the outcomes of IR ALL patients.

4.5.2 The impact of time to event on stratification of IR ALL xenografts

As the analysis of differences in the number of censored and uncensored mice between

IR ALL xenografts demonstrated that engraftment of patient samples in the VXL- treated mice provides an indication for stratifying IR ALL xenografts according to patient outcomes, I thus decided to prioritise the engraftment data generated in the

VXL-treated mice for prediction of outcomes in IR ALL patients. The next question to address was which time to event endpoint of EFS for the VXL-treated mice would be the most appropriate indicator to distinguish between patients with different outcomes.

To investigate whether our ability to predict IR ALL patient outcomes can be improved by measuring the mouse EFS at a certain time to event endpoint, the EFS of all mice and the LGD parameter were compared between the three patient groups at different time to event endpoints.

The EFS data of the VXL-treated mice indicated that a large number of mice that received cells from patients in CR1 required a longer time to reach 1% human cells in the mice PB compared to xenografts derived from early or late relapse patients, although statistical significance was only observed when the EFS was compared

209 between mice that received cells from patients in CR1 and late relapse patients

(P=0.0061, log-rank test) (Figure 4.15). The difference between the CR1 and early relapse groups only approached significance (P=0.0702 log-rank test). In addition, when

EFS was compared between the early and late relapse groups, there was no significant difference (P=0.776 log-rank test).

Survival of the VXL treated mice at TT1%

100 CR1 Late relaps Early relapse 75

50

Percent survival Percent

25

0 0 25 50 75 100 125 150 175 200 225 250 275

Days post inoculation

Log-rank Groups P value CR1 versus Late relapse 0.0061 CR1 versus Early relapse 0.0702 Early relapse versus Late relapse 0.776

Figure 4.15 Comparison of the EFS at TT1% between mice used to establish xenografts from patients with different outcomes. The engraftment of IR ALL patient samples is presented based on the time required for each VXL-treated mouse inoculated with cells from CR1, early and late relapse patients to reach 1% human cell engraftment. The table below the graph shows the statistical evaluation in EFS of mice between patient groups. The log-rank test was used to test for statistical significance.

210 Assessment of the engraftment of patient samples in mice at TT25% revealed that the difference in time to event between the CR1 group and the relapsed group was more pronounced than that observed at TT1%. At TT25%, mice in the CR1 group exhibited significantly longer EFS compared to mice in the late relapse groups (P= 0.0012 log- rank test) as well as those in early relapse group (P=0.0016 log-rank test) (Figure 4.16).

In addition, the analysis of EFS demonstrated no significant difference in time to event between mice in the early and late relapse groups (P= 0.6021 log rank test).

Survival of the VXL treated mice at TT25%

100

CR1 Late relaps 75 Early relapse

50

survival Percent 25

0 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation

Groups Log-rank P value CR1 versus Late relapse 0.0012

CR1 versus Early relapse 0.0016 Early relapse versus Late relapse 0.6021 Figure 4.16 Comparison of the EFS at TT25% between mice used to establish xenografts from patients with different outcomes. The engraftment of IR ALL patient samples is presented based on the time required for each VXL-treated mouse inoculated with cells from CR1, early and late relapse patients to reach 25% human cell engraftment. The table below the graph shows the statistical evaluation in EFS of mice between patient groups. The log-rank test was used to test for the statistical significance.

211 Analysis of the pattern of TTL revealed that mice in the CR1 group showed comparable

EFS to that for mice in the late relapse group, but significantly longer than those in the early relapsed group (P= 0.0362 log-rank test) (Figure 4.17). Unlike what was observed at TT1% and TT25%, the mouse EFS at TTL was significantly different between mice used to establish xenografts from early and late relapse patients (P= 0.0476 log-rank test) suggesting that early TTL in mice is a characteristic feature for cells derived from patients who relapsed early after treatment.

Survival of the VXL treated mice at TTL A 100

80 CR1 Late relaps 60 Early relapse

40 Percent survival Percent 20

0 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation

Log-rank Groups P value CR1 versus Late relapse 0.4154

CR1 versus Early relapse 0.0362 Early relapse versus Late relapse 0.0476 Figure 4.17 Comparison of the EFS at TTL between mice used to establish xenografts from patients with different outcomes. The engraftment of IR ALL patient samples is presented based on the time required for each mouse inoculated with cells from CR1, early and late relapse patients to show manifestation of TTL. The table below the graph shows the statistical evaluation in EFS of mice between patient groups. The log-rank test was used to test for statistical significance.

212 The difference in engraftment characteristics between samples from the three patient groups was also analysed based on the LGD parameter, which was calculated by subtracting the median EFS between control and VXL-treated mice of each xenograft.

Regardless of the time to event endpoint, the LGD values ranged from 1 day to four months for most xenografts and there were some xenografts that showed undefined

LGD values, due to either the patient sample never engrafting or indefinite growth difference between the VXL-treated and non-treated mice. When the LGD values were compared between xenografts according to patient outcome, comparable patterns of

LGD values were observed at TT1% between xenografts derived from CR1, early and late relapsed patients (Figure 4.18A). In contrast, significant differences in LGD patterns were observed at TT25% when the LGD values of xenografts derived from patients in CR1 were compared to those for xenografts derived from late relapse patients (P=0.0256 Mann–Whitney U test) as well as mice that received cells from early relapse patients (P=0.0099 Mann–Whitney U test) (Figure 4.18B). In regard to the TTL, the significant difference in LGD patterns was only observed when the LGD values were compared between xenografts derived from patients in CR1 and those derived from early relapse patients (P=0.03 Mann–Whitney U test) (Figure 4.18C).

213

B A P=0.9654 P=0.6475 * P=0.1649 P= 0.0256 ** P= 0.0099 P= 0.1304

150 150 140 140 120 120

100 100

80 80

60 60 LGD value at TT25% TT25% at LGDvalue TT1% at LGDvalue 40 40 20 20

0 0

CR1 CR1

Late relaps Late relapse Early relapse Early relapse

C P= 0.3457 P= 0.3152 * P= 0.0300

150 140

120

100 80 60 TTL at LGDvalue 40 20 0 CR1

Late relapse Early relapse

Figure 4.18 Comparison of the LGD values between mice used to establish xenografts from patients with different outcomes. Graphs show the difference in LGD values between xenografts derived from CR1, early and late relapse patients at (A) TT1%, (B) TT25%, and (C) TTL. The LGD was set at 150 days if xenografts never engrafted or when the LGD could not be defined due to indefinite growth difference between the VXL-treated and non-treated mice. The time to event was set at 1 if the VXL-treated mice engrafted faster than the control mice. Bar represents the median. Groups were compared by the Mann-Whitney U test.

214 The comparison of the time to event between IR ALL patients indicated that the engraftment at TT25% in the VXL-treated mice was the most efficient criteria allowing for distinguished stratification of xenografts according to whether xenografts were derived from relapsed or non-relapsed patients, and showed high concordance between the clinical response of IR ALL patients to their treatments and the in vivo sensitivity of mice to VXL chemotherapy as represented by LGD values. Furthermore, the EFS of mice at the TTL parameter differentiated between xenografts of early and late relapse patients. The overall observations from this analysis suggested that for efficient prediction of IR ALL patient outcomes based on the EFS of the VXL-treated mice, relapsed samples should be first differentiated from non-relapsed cases based on the mouse EFS at TT25% human cell and the TTL can be then used to distinguish between early and late relapsed patients.

4.5.3 Median versus first mouse EFS

Using of median mouse EFS is a common strategy to characterise the engraftment of

ALL patient samples. However, owing to the heterogeneity in time to event (TT25% or

TTL) observed within some groups of mice that received cells from the same patient, it was required to ascertain whether representation of the engraftment to patient outcomes could be improved by measuring the EFS of the first mouse to reach event. Despite the fact that using median EFS provides more statistical power than using the EFS of the first mouse to reach event, it could be argued that the earliest engrafted VXL-treated mouse represents selection of leukaemia clones upon VXL chemotherapy, which could be more important for stratification of xenografts according to patient outcomes.

To assess whether measuring the EFS based on the median or first mouse to reach event is more representative of patient outcomes, I first compared the EFS of the median and

215 first mouse at TT25% between mice that received cells from the three patient groups and then I assessed the difference in EFS between the median and first mouse to reach event of the same patient group. Based on the median mouse EFS, the pattern of differences in time to event between mice that received cells from relapsed and non- relapsed patients recapitulated that observed based on the EFS of all mice, which was described in Figure 4.16, although with less significance. A significant difference in time to event was observed between mice that received cells from patients in CR1 and those received cells from late relapse patients (P=0.0374 log-rank test) as well as early relapsed patients (P=0.0332 log-rank test) (Figure 4.19A). In addition, no significant difference in time to event was observed between mice inoculated with cells from early and late relapse patients (P=0.5885 log-rank) (Figure 4.19A).

In comparison with median EFS, measuring the time to event based on the EFS of the first mouse to reach event was less valuable for stratifying IR ALL xenografts according to patient outcomes. A significant difference in time to event was observed when the

EFS of mice that received cells from CR1 patients compared to mice that received cells from early relapsed patients (P= 0.0484 log-rank test) but not when compared to that for mice of late relapsed patients (P= 0.078 log-rank test) (Figure 4.19B). Furthermore, there was no significant difference in time to event between mice that received cells from early and late relapsed patients (P= 0.6595 log-rank test) (Figure 4.19B).

When a direct comparison was performed between median and first mouse EFS representing the same patient group, there was no statistical differences in time to event between median and first mouse of the CR1, early and late relapse groups (Figure

4.19C). Overall, characterising the engraftment based on median EFS time showed

216 better stratification of mice according to patient outcomes compared to the EFS of the first mouse. This finding suggests that the median EFS at TT25% is more reliable than using first mouse data for assessing the predictive power of the engraftment of patient samples.

217

A

100

80

Median mouse EFS 60 CR1 Late relapse 40 Early relapse log-rank Percent survival Percent Median EFS at 25% human cell P value 20 CR1 versus Late relapse 0.0374 CR1 versus Early relapse 0.0322 0 0 25 50 75 100 125 150 175 200 225 250 275 300 Early relapse versus Late relapse 0.5885 Days post inoculation

B 100

First Mouse 80 CR1 Early relapse 60 Late relapse

40 Log-rank

Percent survival Percent First mouse EFS at 25% human cell P value 20 CR1 versus Late relapse 0.0484 CR1 versus Early relapse 0.078 0 0 25 50 75 100 125 150 175 200 225 250 275 Early relapse versus Late relapse 0.6595 Days post inoculation

C 100

80 Median mouse EFS CR1 60 Late relapse Log-rank TT25% human cell Early relapse P value First mouse EFS Median EFS versus First mouse EFS 40 0.376

Percent survival Percent CR1 (CR1) Late relapse Early relapse 20 Median EFS versus First mouse EFS 0.402 (Late relapse) 0 Median EFS versus First mouse EFS 0 25 50 75 100 125 150 175 200 225 250 275 300 0.5061 (Late relapse) Days post inoculation

Figure 4.19. Comparison of the engraftment of IR ALL patients based on median and first mouse EFS at TT25%. The engraftment of samples derived from CR1, early and late relapsed patients presented based on (A) median EFS to reach 25% and (B) the EFS of the first mouse to reach 25%. Graph C compares the median and first mouse EFS of xenografts established from the same patient group. The tables beside the graphs show the statistical difference in EFS of mice presented in the graphs. The log-rank test was used to test the statistical difference in EFS of mice.

218 The difference in time to event based on the median and first mouse EFS was also analysed at TTL. Unlike what I observed when the TTL was presented based on the

EFS of all mice (Figure 4.17), the analysis of TTL based on the median or the first mouse EFS demonstrated no significant difference in time to event between the CR1 and early or late relapsed groups (Figure 4.20A and B). Furthermore, there was no significant difference in EFS when the TTL was compared between the medians and first mouse EFS of the same patient group (Figure 4.20C).

219

A 100

Median EFS 80 CR1 Late relapse 60 Early relapse

40 Median mouse EFS at TTL log-rank Percent survival Percent P value CR1 versus Late relapse 0.6089 20 CR1 versus Early relapse 0.0904 Early relapse versus Late relapse 0.2109 0 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation

B 100

80 First mouse EFS CR1 FIRST 60 Late relapse Early relapse

40 First mouse EFS at TTL log-rank P value Percent survival Percent CR1 versus Late relapse 0.9150 20 CR1 versus Early relapse 0.1584 Early relapse versus Late relapse 0.1887 0 0 25 50 75 100 125 150 175 200 225 250 275 Days post inoculation

C

100

Median mouse EFS 80 CR1 Late relapse TTL log-rank 60 Early relapse P value Median EFS versus first mouse EFS 0.4171

40 First mouse EFS (CR1)

Percent survival Percent CR1 Early relapse Median EFS versus first mouse EFS 0.7010 20 Late relapse (Late relapse)

0 Median EFS versus first mouse EFS 0.5224 0 25 50 75 100 125 150 175 200 225 250 275 (Early relapse) Days post inoculation

Figure 4.20. Comparison of the engraftment of IR ALL patients based on median and first mouse EFS at TTL. The engraftment of samples derived from CR1, early and late relapsed patients presented based on (A) median EFS to show manifestation of TTL and (B) the EFS of the first mouse to show manifestation of TTL. Graph C compares the median and first mouse EFS of xenografts established from the same patient group. The tables beside the graphs show the statistical difference in EFS of mice presented in the graphs. The log-rank test was used to test the statistical difference in EFS of mice. 220 4.6 Characterising the engraftment at TT25% human cells in the VXL-treated mice as a predictor for relapse in IR ALL patients

The ultimate goal of the analysis described in the previous sections was to define an improved engraftment strategy for modelling outcomes in IR ALL patients. Having established that the median EFS of the VXL-treated mice and the LGD values at

TT25% can distinguish between xenografts derived from relapsed and non-relapsed IR

ALL patients, I proposed that evaluation of the time required for IR ALL patient cells to engraft in the VXL-treated mice at median TT25% or evaluation of the LGD values at

TT25% could provide prognostic benefits for upfront identification of the risk of relapse in IR ALL patients. According to the data presented in the previous sections, xenografts of relapsed patients are characterised by shorter TT25% and low LGD values compared to xenografts of non-relapse patients which could independently lead to prediction of IR

ALL patient outcomes and thus help us to correctly assign IR ALL patients into an appropriate risk category.

A challenge to prediction of IR ALL patient outcomes based on a continuous variable

(mouse EFS) or LGD values was to define a distinct data point which could serve as an outcome measure to reveal the significant difference in patient outcomes. As there is no evidence from the literature indicating an optimum cut-off point where the median

TT25% or LGD values can be tested to predict patient outcomes with reasonable sensitivity and specificity, it was thus required to subjectively set up multiple thresholds for separation of the cohort into 2 subgroups (short and long TT25% or low and high

LGD values) then analyse the difference in patient outcomes between the resulting subgroups according to the length of CR1. To eliminate the risk of introducing false positive results due to testing multiple thresholds for prediction of patient outcomes, I determined the potential of false positive findings among selected thresholds based on 221 adjusting the P value for multiple comparison using Benjamini and Hochberg test

(Benjamini and Hochberg, 1995).

4.6.1 Stratification of IR ALL patients based on median TT25%

Based on the engraftment data of median TT25%, the time to event ranged from 66 to

230.6 days for 15 xenografts (median 154 days) and 14 xenografts failed to show event over the monitoring period (Figure 4.19A). Therefore I decided to apply multiple threshold measures of median TT25% starting from 150 to 235 days to separate the cohort into two subgroups (early and late median TT25%) and compare the length of

CR1 between patients of the resultant subgroups. A threshold of 150 days divided the cohort into 7 patients from whom xenografts required <150 days to show median

TT25% and 22 patients from whom xenografts did not show an event within 150 days.

On the basis of this threshold there was no significant difference in the probability of relapse observed between patients from whom xenografts required short and long median TT25% in mice (adjusted P value =0.4975 log-rank test) (Table 4.5).

In contrast, assessing the probability of relapse in IR ALL patients in relation to the median TT25% based on a threshold of 180 days revealed a considerable tendency toward significant difference in outcomes between patients from whom xenografts exhibited short and long TT25% (adjusted P value =0.051 log-rank test) (Table 4.5,

Figure 4.21A). This threshold separated the cohort into 12 xenografts that reached time to event in <180 days and 17 xenografts that did not show event within 180 days. This parameter allowed true prediction of 10 relapsed cases out of the 12 patients from whom xenografts exhibited events in <180 days providing a positive predictive value of 83.3% and allowed true prediction of 9 CR patients out of the 17 patients from whom xenografts did not show event within 180 days, providing a negative predictive value of

222 52.9%. Thus, this threshold is characterised by a sensitivity of 55.5% (10/18) [95% confidence interval (CI) 30.7-78.4%] for detection of true relapsed cases based on values below the cut-off point and a specificity of 81.8% (9/11) (95% CI 48.2-97.7%) for detection of true CR1 cases based on values above the cut-off point.

The ability of median TT25% to distinguish between relapse and non-relapsed patients was also reproduced based on a threshold of 210 days. This threshold divided the cohort into 13 xenografts that manifested median TT25% in <210 days and 16 xenografts which did not show event within 210 days and revealed a considerable trend toward significant difference in outcomes between patients from whom xenografts exhibited event below or above the threshold (adjusted P value =0.051 log-rank test) (Table 4.5,

Figure 4.21B). This threshold allowed a positive predictive value of 84.6% (11/13) for detection of relapse in patients based on median TT25% of less than the cut-off point and a negative predictive value of 56.3% (9/16) for identification of CR1 patients based on median TT25% of more than the cut-off point. Using this criterion resulted in an increase in the sensitivity values [61.1% (11/18) 95% CI 35.7-82.7%] for detection of true relapsed cases but a similar specificity value for detection of CR1 cases to that observed based using the previous threshold.

The most significant criterion for prediction of outcomes in IR ALL patients was observed when the probability of relapse was analysed based on the median TT25% using a threshold of 235 days. Based on this threshold, the cohort was divided into 15 patients from whom xenografts exhibited event within 235 days median TT25% and 14 patients from whom xenografts did not show event within 235 days. Importantly, the analysis of length of CR1 based on this cut-off point demonstrated the largest

223 significant difference in probability of relapse between IR ALL patients (adjusted P value= 0.0292 log-rank test) (Table 4.5, Figure 4.21C). This threshold allowed a positive predictive value of 86.6% (13/15) for detection of relapse in patients based on median TT25% of less than the cut-off point and a negative predictive value of 64.3%

(9/14) for identification of CR1 patients based on median TT25% above the cut-off point. Furthermore, it provided a sensitivity of 72.2% (13/18) (95% CI 46.5-90.3%) for detection of true relapsed cases based on value below the cut-off point with a similar specificity [81.8% (9/11) with 95% CI 48.2-97.7%] for detection of true CR1 cases based on values above the cut-off point compared to what observed based on the previous thresholds.

224 Table 4.5. Analysis the impact of different thresholds points of engraftment data at the TT25% on identifying patients according to their outcomes

Engraftment Threshold # of # of CR1 adjusted Sensitivity (%) Specificity (%) parameter (Days) Total Relapse P-value (%) (%) (95% CI) (95% CI) <150 7 5 (71.4%) 2 ( 25.8% ) 27.7 81.8 1 0.4975 >150 22 13 (59% ) 9 (41%) (9.6 - 53.4%) (48.2- 97.7%) <180 12 10 (83.3%) 2 (16.7%) 55.5 81.8 2 0.051 >180 17 8 (47% ) 9 (53%) (30.7 -78.4%) (48.22- 97.7%)

Median EFS <210 13 11 (84.6%) 2 (15.4%) 61.1 81.8 3 0.051 >210 16 7 (43.7%) 9 (56.3%) (35.7- 82.7%) (48.2-97.7%) <235 15 13 (86.6%) 2 (15.4%) 72.2 81.8 4 0.0292 >235 14 5 (35.7%) 9 (64.3%) (46.5-9.3%) (48.2-97.7%)

Total, Total number of patients according to different thresholds; # of relapse, the number of relapsed patients according to different thresholds; # of CR1, the number of patients in CR1 according to different thresholds; CI, Confidence interval. Raw P value was calculated by log-rank test.

225

A B

100 100

80 80 Median TT25% >210 days Median TT25% >180 days N=16 patients 60 N=17 patients 60

40 40 Percent survival Percent Median TT25% < 180 days survival Percent Median TT25% < 210 days N=12 patients N=13 patients 20 20 adjusted P = 0.051 adjusted P = 0.051 0 0 0 25 50 75 100 125 0 25 50 75 100 125 Length of CR1 (months) Length of CR1 (months)

C 100

80 Median TT25% >235 days N=14 patients 60 40

survival Percent Median TT25% < 235 days N=15 patients 20 * adjusted P = 0.0292 0 0 25 50 75 100 125 Length of CR1 (months)

Figure 4.21. Evaluation of IR ALL patient outcomes in relation to the median TT25%. Kaplan-Meier graphs estimate the relapse probabilities in the IR ALL cohort based on a classifier of (A) 180 days (B) 220 days or (C) 235 days required for median group of mice to reach 25% human cell in the PB. N denotes number of patients. The log-rank test was used to analyse the statistical difference in length of CR1 between the two subgroups of patients and the P value adjusted for multiple testing with the Benjamini and Hochberg test.

226 Nevertheless, the 235 days classifier failed to predict relapse in 5 patients out of the 18 samples from relapsed patients (27.8%). Of the 5 patient samples, there was no engraftment in the VXL-treated mice used to establish ALL-201, ALL-210 and ALL-

218 and poor engraftment in mice used to establish ALL-226 and ALL-230. In order to understand why these xenografts did not exhibit time to event similar to other xenografts of relapsed patients, I asked whether the difference in median TT25% was influenced by the baseline clinical features of patients from whom the xenografts were derived. To determine the extent to which the clinical features affected the reliability of median TT25% for accurate prediction of relapse in ALL patients, I compared the clinical features between the relapsed patients from whom xenografts demonstrated median TT25% below or above 235 days. As shown in Table 4.6, there was no significant difference in the pattern of clinical features observed between relapsed patients from whom xenografts required less or more than 235 days to show median

TT25% human cell. These results indicate that there is no association between the patterns of engraftment based on the median TT25% of the samples derived from the relapsed patients and the clinical features of the patients from whom the xenografts were derived.

227

Table 4.6. Analysis of the association of clinical features of the relapsed IR ALL patients with the median TT25% in the VXL-treated mice. Characteristic Median TT25% Median TT25% P value (<235 days) (> 235 days) No of patient= 18 13 5 Time to relapse Early relapse 6 (46.2%) 2 (40%) 1 Late relapse 7 (53.8%) 3 (60%) Age <10 years 8 (61.5%) 3 (60%) 1 >10 years 5 (38.5%) 2 (40%) WCC <100X109 12 (92.3%) 4 (80%) 0.4902 >100X109 1 (7.7%) 1 (20%) Cytogenetics Favourable (SR) 7 (53.8%) 3 (60%) 1 Unfavourable (IR) 6 (46.1%) 2 (40%) Site of relapse BM/BM+ 11 (84.6%) 4 (80%) 1 CNS 2 (15.4%) 1 (20%) *MRD at day 15 Positive 4 (30.8%) 1 (20%) 1 High 5 (38.5%) 2 (40%) * MRD at day 33 Negative 2 (15.4%) 0 (0%) 1 Positive 11 (84. 6%) 4 (80%)

Data are expressed as absolute number and percentage. SR, standard risk; IR, intermediate risk. BM+, relapse in BM and other sites (CNS and/or testis). * Denotes that no MRD data were available for some patients. P value calculated using Fisher's exact test.

Cytogenetics: MRD value: Favourable: ETV-RUNX1, hyperdiploid Negative: (<1x10-5) Unfavourable: TCF3-PBX1, B other Positive: (≥1x10-5, < 1x10-2) High: (≥1x10-2 )

228 4.6.2 Stratification of IR ALL patients according to the LGD values at TT25%

The feasibility of using the engraftment features at TT25% to reflect the length of CR1 in patients was also investigated based on the LGD values which depends on the pattern of response of xenografts to VXL chemotherapy. The pattern of LGD for 14 xenografts ranged from 1 to 125 days (median 32 days) and 15 mice exhibited indefinite LGD values over the monitoring period (Figure 4.18C). To analyse this parameter for prediction of IR ALL patient outcome, I applied multiple threshold measures, starting from 40 to 130 days for splitting the cohort into two subgroups. A threshold of 40 days divided the cohort into 9 xenografts with LGD values <40 days and 20 xenografts with

LGD values of >40 days and others with indefinite LGD values. This approach identified relapse in 8 out the 9 (88.8%) patients from whom xenografts exhibited LGD

<40 days and identified CR in 10 out of the 20 (50%) patients from whom xenografts exhibited LGD of >40 days and indefinite LGD values, although the difference was close to significance (adjusted P value= 0.0774 log-rank test) (Table 4.7, Figure 4.22A).

Analysis of the accuracy measurement for this criterion revealed low sensitivity [44.4%

(8/18) 95% CI 21.5-69.2%] for identifying relapsed patients based on values below the cut of point but high specificity [90.9% (10/11) 95% CI 21.5-69.2%] for identifying CR patients based on values above the cut of point.

Applying a threshold at 60 days divided the cohort into 11 xenografts with LGD values of < 60 days and 18 xenografts which exhibited LGD of > 60 days and indefinite LGD values. Based on this strategy, there was no significant difference in length of CR1 between patients from whom xenografts exhibited LGD values of less or more than the cut-off point (adjusted P value =0.1264 log-rank test) (Table 4.7). Despite the insignificant difference observed in outcomes between the patients, this criterion 229 allowed an increase in sensitivity [50% (9/18) 95% CI 26.0-73.9%] for detection of relapsed cases based on value below the cut-off point but a decrease in specificity value

[81.8% (9/11) 95% CI 48.2-97.7%] for detection of CR patients based on value above the cut-off point. Similarly, setting a threshold at an LGD of 80 days did not result in significant difference in probability of relapse between 12 xenografts which exhibited

LGD value <80 days and 17 xenografts which exhibited LGD values >80 days or indefinite LGD values (adjusted p value= 0.1184 log-rank test) (Table 4.7). This criterion showed a further increase in the sensitivity of the LGD parameter [55.5%

(10/18) 95% CI 30.7-78.4%] for detection of true relapse but no change in the specificity when compared with what was observed based on the previous threshold.

The ability of the LGD parameter to predict IR ALL patient outcomes was more certain when a threshold was set at an LGD of 130 days. This threshold divided the cohort into

15 patients from whom xenografts exhibited LGD values of <130 days and 14 patients from whom xenografts that exhibited indefinite LGD values. Interestingly, the analysis showed that 13 out of 15 (86.6%) xenografts that exhibited LGD values of <130 were derived from relapsed patients and 9 out of 11 (64.3%) xenografts showed indefinite

LGD values were derived from non-relapsed patients (adjusted P value= 0.0292 log- rank test) (Table 4.7, Figure 4.22B). Hence, this threshold provided a sensitivity of

[72.2% (13/18) 95% CI 46.5-90.3%] for detection of true relapsed cases based on values below the cut-off point and a specificity of 81.8% (9/11) 95% CI 48.2-97.7%] for detection of true CR1 cases based on values above the cut-off point compared to that observed using the previous thresholds.

230 Table 4.7. Analysis the impact of different thresholds of LGD values at the TT25% on identifying patients according to their outcomes

Engraftment Threshold # of Relapse # of CR1 adjusted Sensitivity (%) Specificity (%) Total parameter (Days) (%) (%) P-value (95% CI) (95% CI) <40 9 8 (88.9%) 1 (11.1%) 44.4 90.9 1 0.0774 >40 20 10 (50% ) 10 (50%) (21.5- 69.2%) (58.7-99.7%)

<60 11 9 (81.8%) 2 (18.2%) 50 81.8 2 0.1264 >60 18 9 (50% ) 9 (50%) (26.0-73.9%) (48.2- 97.7%) LGD values <80 12 10 ( 83.4%) 2 (16.6%) 55.5 81.8 3 0.1184 >80 17 8 (47% ) 9 (53%) (30.7-78.4%) (48.2-97.7%) <130 15 13 (86.6%) 2 (15.4%) 72.2 81.8 4 0.0292 Indefinite 14 5 (35.7%) 9 (64.3%) (46.52- 90.3%) (48.2- 97.7%)

Total, Total number of patients according to different thresholds; # of relapse, the number of relapsed patients according to different thresholds; # of CR1, the number of patients in CR1 according to different thresholds; CI, Confidence interval. Raw P value was calculated by log-rank test.

231

A

100

80 LGD > 40 days 60 N= 20 patients

40 Percent survival Percent LGD <40 days 20 N= 9 patients

adjusted P=0.0774 0 0 25 50 75 100 125 150 Length of CR1 (months) B 100

80 Indefinite LGD N= 14 patients

60

40

Percent survival Percent LGD <130 days N= 15 patients 20

adjusted P=0.0292 0 0 25 50 75 100 125 150 Length of CR1 (months)

Figure 4.22 Evaluation of the relapse probability in IR ALL patients based on the LGD values at TT25%. Kaplan-Meier graphs estimate the relapse probabilities in the IR ALL cohort in relation to the classification ability of LGD based on (A) 40 days or (B) 130 days. N denotes number of patients. The log-rank test was used to test statistical differences in length of CR1 between the two subgroups of patients and the P value adjusted for multiple testing using the Benjamini and Hochberg test.

232 4.7 Characterising the TTL in the VXL-treated mice as a predictor for identifying early from late relapsed IR ALL patients

The overall rate of survival of ALL patients differs between those who experience early or late relapse (Chessells et al., 2003; Parker et al., 2010). This concept was also noticed in the relapsed patients selected for this study. Out of the 19 relapsed IR ALL patients, 9 patients suffered from early relapse (≤ 25 months) and 10 patients exhibited late relapse

(>25 months) (Table 4.1). Among the early relapsed patients 5 patients died of the disease whereas there was only one late relapse patient who died of the disease. The difference in the overall survival between early and late relapse IR ALL patients implies that early relapse patients should be distinguished from those with late relapse for better management of the disease.

The proposal for distinguishing early from late relapse patients was to firstly utilise the median TT25% to differentiate between relapse and non-relapsed patients and subsequently TTL can be monitored for predicting the time to relapse in those identified to have high risk of relapse based on median TT25%. Unexpectedly, the high sensitivity and specificity to differentiate between relapse and non-relapsed patients was observed based on median TT25% at 235 days and monitoring of the TTL beyond this time point would be not ideal for proper management of early relapsed patients. To test the predictive power of TTL, I asked whether early relapse could be distinguished from late relapse patients based on the threshold of 235 days so prediction of the time to relapse will be available immediately.

A challenge to this aim were the findings shown in Figure 4.20 that neither the first mouse EFS nor the median EFS at TTL reproduced the difference in TTL between

233 xenografts of early and relapsed patients observed based on EFS of all mice (Figure

4.17). This finding introduced uncertainty about whether to use the EFS of the first or median mouse for prediction of early relapse in IR ALL patients. However, I decided to evaluate both the EFS of the first and median mouse and identify which EFS measure is more reliable for prediction of patient outcomes.

Based on the first mouse EFS at TTL, the threshold of 235 days divided the cohort into

10 and 8 mice which manifested TTL in less or more than 235 days, respectively.

Assessing the time to relapse between patients of the two subgroups revealed no significant difference in length of CR1 between the two subgroups (P=0.5321 log-rank test) (Figure 4.23A). In other words, there was no significant stratification of patients into early and late relapse based on the first mouse EFS at TTL. However, a significant difference in the time to relapse was observed based on the median mouse EFS at TTL.

Using the threshold of 235 days divided the cohort into two groups of 9 mice, which manifested TTL in less or more than 235 days. Analysis of the probability of relapse between patients from whom cells required <235 days to show clear manifestation of

TTL in mice and those from whom xenografts did not show TTL cells within 235 days revealed a significant difference in length of CR1 between the two subgroups (0.0479 log-rank test) (Figure 4.23B). Interestingly, 6 out of the 9 patients (66.6%) from whom cells required < 235 days to show manifestation of TTL were derived from early relapse patients (≤ 25 months) and 7 out of the 9 patients (77.7%) from whom cells required >

235 days to show manifestation of TTL were derived from late relapse patients (>25 months). This criterion allowed a sensitivity of 75% (6/8) (95% CI 34.9-96.8%) for identifying early relapse based on the median EFS of mice at TTL below 235 days and a

234 specificity of 70% (7/10) (95% CI 34.9-96.8%) for identifying late relapse based on the median EFS of mice at TTL above the 235 days.

Despite that there was a narrow difference in length of CR1 between early and late relapsed patients used in this study, the EFS at TTL for the median mouse showed an ability to stratify patients based on their time to relapse. For instance, the TTL of <235 days predicted early relapse in 75% of patients who experienced relapse within 17 to 25 months (median length of CR1 is 24 months) and late relapse in 70% of patients who experienced relapse within 27 to 66 months (median length of CR1 of 31 months). The findings from this analysis indicate that monitoring of the TTL based on 235 days in the median mouse could serve as a prognostic feature for differentiating early from late relapse patients in those predicted to have relapse based on median TT25%.

235

A 100

80

60 First mouse EFS at TTL > 235 days

40 Percent survival Percent First mouse EFS at TTL < 235 day 20

P= 0.5321 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 Length of complete remission1 (CR1)

B

100

80

60 Median mouse EFS at TTL> 235 days N=9 40 Percent survival Percent Median mouse 20 EFS at TTL< 235 N=9 patients P= 0.0497 0 0 5 10 15 20 25 30 35 40 45 50 55 60 65 70 Length of complete remission1 (CR1)

Figure 4.23. Evaluation of IR ALL patient outcomes based on the EFS of mice at TTL in the VXL-treated mice. Kaplan-Meier graph presents the time to relapse in IR ALL cohort in relation to (A) the EFS of the first mouse to show TTL and (B) the EFS of the median mouse to show TTL based on the classifier of 235 days. Vertical red line on 25 indicates difference in CR1 between early and late relapse difference. The log- rank test was used to test the statistical difference in time to relapse of patients.

236

4.8 Summary and Discussion

The clinical heterogeneity among IR ALL patients is a well recognised yet challenging clinical problem. Current available prognostic features don’t provide an efficient strategy for prediction of relapse in the IR BCP-ALL childhood patients (Biondi et al.,

2000; Conter et al., 2010). It is thus required to develop an improved strategy for identification of IR BCP-ALL patients who are going to relapse from those who remain disease free. The evidence of a correlation between the length of CR1 and the overall survival in ALL patients suggests the possibility of upfront identifying outcomes of

ALL patients. Establishing a pre-clinical model for ALL patients that maintains a correlation with the length of CR1 may offer an opportunity to improve relapse prediction in ALL patients who are stratified into the IR subtype. Encouraged by the observations from the Pilot Study that engraftment of IR ALL patient samples in the

VXL-treated NSG mice provided an efficient approach to stratify patients according to their outcomes, I aimed to expand this finding and test the extent to which this model can predict early and late relapse from non-relapse using a larger cohort of IR ALL patients.

Thirty samples were identified from IR BCP-ALL patients, who were treated under the

ANZCHOG Study VIII protocol and experienced various lengths of CR1 after treatment to establish a xenograft model capable of improved relapse prediction in IR ALL patients. For establishing xenografts from this large number of patient samples, the cohort was divided into three panels containing a mix of patients with diverse length of

CR1. The thirty patient samples were xenotransplanted into mice and three experiments conducted at different times. On the whole, 28 samples were successfully engrafted into at least one mouse over the monitoring time. Depending on the patient sample there was diversity in the level of leukaemia cells engrafted in the murine PB. Some patient 237 samples engrafted early in the PB, others required a longer time to engraft, however two samples derived from one relapsed and one patient in CR1 did not show any evidence of engraftment in the mouse PB. The majority of patient samples maintained consistent patterns of engraftment within individual groups of mice used to establish each xenograft although some variability was observed in the VXL-treated groups.

Furthermore, the 3 patient samples (A5072, A1839 and A1795) that were used to establish ALL-202, ALL-215 and ALL-220, respectively, exhibited patterns of engraftment and responses to VXL chemotherapy that were remarkably similar to those observed when the same patient samples were used to establish ALL-64, ALL-66 and

ALL-65 in the Pilot Study. These observations highlight the reproducibility of engraftment and responses to VXL chemotherapy of IR ALL patient samples in NSG mice.

Following the completion of the experimental work, the median time to engraftment and time required for the first mouse to engraft and the LGD parameters were calculated at different levels of human cell infiltration (1% and 25% human cell engraftment in the

PB) and TTL for all patient derived xenografts. To prepare the data for analysing the difference in engraftment between xenografts of the three patient groups, I firstly assessed whether allocation of patients based on the length of CR1 in the three panels was not selecting for bias due to differences in other baseline clinical features. The analysis of distribution of the patient’s clinical features including age, sex, cytogenetic features, site of relapse and MRD levels at day 15 and 33 between the three panels revealed no considerable difference in the distribution of clinical features between patients of the three panels. To further avoid any possible bias introduced in the engraftment of samples due to prior knowledge of outcomes of the patients under

238 investigation, I decided to keep myself blinded to the outcomes of the patients from whom the xenografts were derived until the end of the experiment. I also assessed the compatibility between the engraftment data of xenografts established in multiple experiments to ensure that no bias was introduced from unknown confounding factors during the process of xenograft establishment. Comparison of the distribution of engraftment data between the three panels indicated no statistical difference in the time to engraftment parameters between xenografts of the three panels. These findings indicated the reliability of pooling the engraftment data from the three experiments for analysis of the relapse probability in all patients based on the engraftment features of their patient samples. I therefore gathered available engraftment data of samples from the three patient groups into one dataset to increase the statistical power for defining an improved xenograft model capable of prediction of outcomes in IR ALL patients.

The approach to analyse the data was to define which engraftment criteria allowed maximum differences in engraftment between samples from patients with different outcomes. The analysis started by comparing the pattern of engraftment between samples of the three patient groups based on the engraftment in VXL-treated and non- treated mice. The analysis highlighted the ability of VXL treatment to allow differences in the number of censored and uncensored mice at every engraftment endpoint relevant to IR ALL patient outcomes. This finding is consistent with what I observed in the Pilot

Study that the engraftment in the VXL-treated but not the control mice stratified the four IR ALL xenografts according to patient outcomes. Furthermore, these data are in line with findings from clinical studies which indicate the prognostic significance of response of ALL patients to chemotherapy (Gaynon et al., 1998; Miller et al., 1989;

Panzer-Grümayer et al., 2000), and another finding of a preclinical study which

239 emphasised the importance of using VXL treatment to delay progression of ALL xenografts established from patients with good prognosis (Szymanska et al., 2012). An interesting finding of this preclinical study is that the plasma drug levels of the optimised VXL chemotherapy in mice were comparable to those achievable in humans, indicating that the VXL chemotherapy regimen used in mice is a clinically relevant protocol for modelling the response of ALL patients to their treatment (Szymanska et al., 2012).

The results of both studies (Pilot and Main) indicated the feasibility of applying a short- term chemotherapy treatment soon after inoculation of patient cells into mice for the development of a clinically relevant model that reflects the responses of IR ALL patients to their treatment. The rationale that ALL disease at diagnosis is composed of highly proliferative and self-renewing cells and minor populations of slowly proliferating cells, which are mostly reported to be the source of relapse, prompted the use of VXL treatment post inoculation in mice. Such an approach might exert positive and/or negative selective pressure on leukaemia populations at an early stage of engraftment allowing for slowly dividing cells to expand and drive relapse. This could help to exploit the differences in the VXL responses and engraftment behaviour between samples from IR ALL patients with different outcomes. Although the ideal strategy for modelling response of ALL patients to their treatment is to start VXL chemotherapy upon establishment of overt leukaemia disease in the mouse, applying this approach might reduce the likelihood of VXL chemotherapy selecting leukaemia subpopulations due to systemic engraftment and high disease burden in recipient mice.

Furthermore, the latter approach will increase the turn-around time required for

240 prediction of patient outcome and is therefore impractical for the clinical translation of the findings from these PDX models.

On the basis of the observation above, the engraftment data from the VXL-treated mice were selected for prediction of outcomes in IR ALL patients and I therefore continued the analysis by asking which time to event endpoint would accentuate the difference in the EFS between the VXL-treated mice that received cells from the three patient groups.

Although analysing the engraftment based on each time to event endpoint presented statistical differences in EFS between mice which received cells from CR1 and those that received cells from early and/or late relapsed patients, the difference was more pronounced based on the mouse EFS at TT25%. While most published studies had investigated the relevance of the ALL xenograft model to reflect ALL patient outcomes based on either a predetermined endpoint or the first indication of leukaemia morbidity to stop monitoring of the engraftment (Meyer et al., 2011; Woiterski et al., 2013), there is no evidence from the literature that supports monitoring engraftment of ALL patient samples in mice to 25% for stratifying xenografts according to ALL patient outcomes.

The rational for evaluating the difference in engraftment based on the TT25% human cell parameter is that the 25% engraftment represent the exponential phase of leukaemia cell growth which could reflect the difference in the acceleration of leukaemia cell expansion between ALL xenografts.

The importance of the TTL parameter was highlighted by the difference in EFS between mice, which received cells from early and late relapse suggesting that this parameter could lead us to distinguish early from late relapsed patients. In support to this finding, a previous study reported by Meyer et al. (2011) showed that samples from early relapsed

ALL patients manifested shorter TTL in NOD/SCID mice, compared to those that 241 received cells from other ALL patients. The authors claimed that early TTL is an engraftment characteristic for samples of early relapse ALL patients. The TTL parameter reflects the ability of leukaemia cells to infiltrate mouse organs and cause deterioration in mouse health. The finding from the analysis of mouse EFS at TTL indicated that xenografts of early relapse IR ALL patients could be distinguished from xenografts of late relapsed patients based on their ability to show manifestation of the disease in mice.

Given these results, I proposed that the engraftment of IR ALL patient samples in the

VXL-treated mice at TT25% human cell could provide an approach to distinguish between relapse and non-relapsed patients, and subsequent monitoring of TTL should allow differentiation of early relapse from late relapse patients. Before testing the predictive advantage of the TT25% and TTL parameters, it was required to determine whether monitoring of the engraftment based on the EFS of the first mouse or EFS of the median mouse would be more valuable for prediction of outcomes of IR ALL patients.

The analysis of mouse EFS at TT25% suggested the reliability of the median mouse for testing the predictive power of TT25% and therefore this parameter was prioritised for testing the prognostic indication. To analyse the predictive power of the median mouse

EFS at TT25%, different thresholds were applied to separate the cohort into short and long median TT25% and the statistical difference in length of CR1 was adjusted for the possible bias which could be introduced due to the multiple comparison of different thresholds. Comparison of length of CR1 in patients between the resultant subgroups highlighted considerable sensitivity and specificity of the median TT25% at 180 and

242 210 days to stratify patients into relapse and non-relapse groups. Most importantly, a threshold of 235 days allowed higher sensitivity and specificity for differentiating relapse and non-relapsed patients.

To further ascertain whether measuring the LGD at TT25% could enhance stratification of patients according to their outcomes, the length of CR1 for IR ALL patients was tested in relation to the LGD values. The analysis indicated significant differences in

LGD values between xenografts of relapsed and non-relapsed patients particularly when the threshold of 130 days was used to divide the cohort into two subgroups. This observation highlighted the difference in sensitivity to VXL chemotherapy between xenografts of patients with different outcomes and showed the advantage of using this feature to differentiate between relapse and non-relapse patients. However, the analysis indicated that there was no further enhancement to the predictive power of the TT25% compared to what I observed based on the EFS of the median mouse. Considering that both median mouse EFS and LGD parameters at TT25% resulted in high accuracy measures to distinguish between relapse and non-relapsed patients, further analysis of the data should include testing the influence of the combination between these factors to increase the predictive power for identifying relapse in IR ALL patients.

Since I observed that the median TT25% based on a threshold of 235 days provided the most discriminative criteria with reasonably high sensitivity and specificity for identifying relapse in IR ALL patients, I asked whether early relapse can be distinguished from the late relapse patients based on the TTL patterns at 235 days. The analysis of difference in length of CR1 between patients according to median mouse

EFS at TTL predicted early relapse in 75% of patients from whom xenografts required

<235 days to show sign of TTL and late relapse in 70% of patients from whom 243 xenografts did not show TTL within 235 days. The finding from this study is relatively similar to what was reported by Meyer et al. (2011) although there is inconsistency between the two studies in the time required for mice to show TTL features. Meyer et al. (2011) assessed the relevance of TTL of ALL xenografts established in non-drug treated NOD/SCID mice to predict relapse in a cohort of 50 ALL patients. When the engraftment data were analysed for the 43 non-high risk ALL patients, the authors identified that mice inoculated with samples from all the 4 early relapsed patients of this cohort are characterised by short TTL (<70 days), those inoculated with samples of 21 patients including late relapse and CR1 patients were characterised by long TTL (>70 days) and samples of 18 CR1 patients showed no engraftment over the monitoring period. The much higher time required for appearance of TTL in my study could be attributed to the delay of progression of IR ALL xenografts by VXL treatment.

Furthermore, while this study only focused on analysis of the engraftment features as independent variables for defining IR ALL patient outcome, the predictive power of engraftment features could be assessed to find whether the combination between the engraftment characteristics and the clinical features of IR ALL patients (age, sex, WCC, cytogenetics and MRD values) could improve our ability to predict outcomes of IR

ALL patients.

The results presented herein show for the first time that ALL xenografts can differentiate IR ALL patients according to their probability of relapse. Certainly, engraftment of IR ALL patient samples in NSG VXL-treated mice at median TT25% predicted relapse, and median TTL identified early from late relapse patients, although the more accurate prediction of outcomes required monitoring of engraftment for 235

244 days. The criterion of 235 days is not an ideal outcome measure, but could be translated to the clinic for immediate adjusting of maintenance treatment according to patient needs, since the first evidence of relapse in the IR ALL patients treated on ANZCHOG study VIII was not apparent until 17 months (~517 days) from diagnosis. However, as relapse was encountered in around 20% of IR ALL patients, using the median TT25% at the 235 days cut-off with a probability of 87% to predict relapse, could identify one case in every 6 xenografted IR ALL patients. The application of these strategies for the benefit of IR ALL patients could be challenged by the cost required for evaluating the engraftment of 6 patients in 18 NSG mice for the identification of every relapse case.

This study provides the basis for further work aiming to improve the time line for monitoring ALL xenografts and to reduce the cost of establishing xenograft models.

This potentially will contribute to upfront prediction of relapse in IR ALL patients and guide adjusting chemotherapy treatment according to patient needs.

245 CHAPTER 5 CHARACTERISATION OF THE HETEROGENEITY BETWEEN IR ALL XENOGRAFTS AND THEIR RESPONSE TO CHEMOTHERAPEUTIC DRUGS

246 5.1 Introduction

The current concepts in the biology of ALL disease acknowledge the diverse biological and functional characteristics of leukaemia cell populations and their roles in disease progression (Landau et al., 2014; Notta et al., 2011; Schmitz et al., 2011). A large body of evidence highlighted by profiling the genome of ALL disease, has demonstrated the genetic heterogeneity within leukaemia cell populations (Anderson et al., 2011; Gawad et al., 2014). In efforts to understand the link between the diversity in genetic compositions of ALL disease and development of relapse, several studies investigated the difference in the clonal architectures between paired samples of ALL diseases at time of diagnosis and relapse. The analysis identified that many genetic features are common between paired samples although some relapsed cases were found to have clones which present as minor subpopulations at diagnosis or carry unique genetic features compared to that detected at time of diagnosis (Hogan et al., 2011; Ma et al.,

2015; Mullighan et al., 2008; Yang et al., 2008). Thus, relapse in some ALL patients is suggested to be due to the outgrowth of sub-clones that were already present in the primary disease at diagnosis, but at low frequency or due to evolution of ALL clonal populations after chemotherapy treatment (Jan and Majeti, 2013). However, it is uncertain what are the mechanisms by which leukaemia cells maintain or gain an enhanced ability to survive the inhibitory effects of chemotherapy.

Understanding the contribution of leukaemia heterogeneity to the biology of ALL disease could provide a rationale for personalising treatment for ALL patients. It could lead to define therapeutic interventions to target the source of chemotherapy resistance in ALL patients at high risk of relapse. Given that relapse clones are commonly more resistant to treatment, it would be of interest to characterise the biological and functional

247 properties of ALL clones at the time of diagnosis with the intention of understanding the clonal heterogeneity in ALL disease biology and its influence on disease progression.

Owing to the limited availability of leukaemia patient samples, as well as the difficulty in culturing leukaemia cells, progress in this field requires an innovative practical technique that allows capturing the disease heterogeneity and defining its impact on improving the treatment for ALL patients. As described in the introductory chapter of this thesis, ALL patient-derived xenografts in many ways represent a valuable tool for modelling disease in ALL patients. Development of ALL patient-derived xenografts using immune-compromised mice has been shown to offer preclinical testing of drugs for treatment of ALL patients. In addition, several studies along with the data presented in the two previous chapters indicate a good correlation of ALL patient outcomes with the engraftment of their BM biopsy samples in immune-compromised mice (Meyer et al., 2011; Schmitz et al., 2011; Woiterski et al., 2013). Beyond these approaches, ALL patient-derived xenografts have been shown to recapitulate the genomic diversity of primary leukaemia samples and maintain the genetic and epigenetic profiles of paediatric ALL patients (Cheung et al., 2010; Krivtsov et al., 2014; Schmitz et al.,

2011).

The high concordance in genotypes and phenotypes between ALL xenografts and ALL disease in children suggests that the clonal progression of xenograft cells in immunodeficent mice could represent the activity of clonal populations in ALL disease

(Krivtsov et al., 2014; le Viseur et al., 2008). The engraftment process imposes barriers to the growth of ALL populations, which could support selection of leukaemia cells with an enhanced ability to establish the disease in mice. This provides an opportunity

248 to study the diversity within ALL disease, better understand how leukaemia clones grow under the influence of different engraftment factors in murine models, and further ascertain the impact of an extrinsic selective pressure applied from treatment with chemotherapy drugs on ALL disease biology (Clappier et al., 2011; Nowak et al., 2015;

Samuels et al., 2014).

In the third chapter of this thesis, a panel of four ALL patient diagnostic biopsies (2 from relapsed and 2 from non-relapsed patients) were matched according to clinical factors and four ALL xenografts were established from these samples (ALL-64, ALL-

65, ALL-66 and ALL-67) by altering the engraftment conditions based on different mouse strains, site of inoculation and in the presence or absence of selection with VXL chemotherapy. A high percentage of xenograft cells infiltrated the spleens of most mice used to establish ALL-64 and ALL-65 except those treated with VXL, but more than

85% of the mice used to establish ALL-66 and ALL-67 showed evidence of engraftment. In this chapter I set out to use the ALL xenografts established in the Pilot

Study to characterise the features of ALL cells after engraftment in mouse models using samples harvested from the spleens of highly engrafted mice. The aims of this chapter were to:

1- Characterise the diversity in transcriptional profiles within an individual xenograft

established using different engraftment conditions

2- Investigate whether the ex-vivo sensitivity to induction chemotherapy could

differentiate between paired IR ALL xenografts, which were established from

patients with diverse outcomes

3- Assess whether the in vivo selection with VXL treatment could lead to the

development of chemotherapy resistance

249 5.2 Diversity in transcriptional profiles of an IR ALL xenograft and its corresponding patient sample

Xenotransplantation of diagnostic samples from ALL patients into mouse models may favour selection of clinically relevant clones in mice or that which might not fully represent the patient’s disease. Altering the conditions of engraftment such as the mouse strains and inoculation sites could apply various degrees of stresses on engraftment of

ALL patient samples which could increase the likelihood of capturing the heterogeneity in ALL clonal populations. In an approach to investigate the heterogeneity in ALL disease, I decided to characterise the pattern of transcriptional profiles within an ALL xenograft established using different transplantation conditions.

For this study, the ALL-67 xenograft was selected for global analysis of gene expression through the microarray platform as this xenograft exhibited different patterns of leukaemia cell progression in mice inoculated with patient cells using different transplantation conditions (Figure 3.10). RNA was extracted from frozen stocks of ALL-67 xenograft cells

(3 NSG IV, 3 NSG IF, 2 NOD/SCID IV and 2 NOD/SCID IF) and two samples of patient

BM cells that were used to establish ALL-67. High quality RNA samples with a 260/280 ratio of more than 1.8, a 260/230 ratio around 2.0 and a RNA Integrity Number (RIN) more than 8.5 were processed for gene expression analysis using human Illumina chip array. The pattern of transcriptional profiles was compared between samples of ALL-67 established in different mouse strains (NSG versus NOD/SCID mice) via different inoculation routes (IV versus IF routes). The expression profiles from ALL-67 samples were also compared to the original patient biopsy used to establish the xenograft (patient A4334).

To analyse the pattern of gene expression between sample groups, raw microarray data were log transformed to stabilise variance among the range of expression and 250 normalised to remove non-biological difference introduced during the process of hybridising complementary RNA (cRNA) onto microarray chips. The gene expression data were then subjected to hierarchical cluster analysis as well as assessing differentially expressed genes between the ALL-67 xenograft and the patient biopsy.

Unsupervised hierarchical clustering of the gene expression data revealed that the gene expression profiles divided the samples into two broad clusters; one cluster represents the gene expression of the two samples of primary ALL patients, and another large cluster represents samples of ALL-67 distributed within two subclusters without clear stratification according to mouse strain or site of inoculation (Figure 5.1). The observation that all samples of ALL-67 clustered separately from the patient samples, suggest that the process of engraftment enriched for cells with common gene expression patterns shared between samples established using different transplantation conditions.

251

NSG IF NSG IF NSG IF NSG IF NSG IV NSG IV NSG IV NOD/SCID IF NOD/SCID IF NOD/SCID IF NOD/SCID IV NOD/SCID IV NOD/SCID IV Patient A4334 Patient A4334 Patient

Figure 5.1. Hierarchical clustering dendrogram represents broad differences in gene expression profiles between samples of ALL-67 and primary samples of the corresponding patient. Red colour indicates samples of primary ALL patient, ALL-67 established in NSG mice via the IV and IF route are labelled with yellow and green respectively, and light lavender and grey colours highlight those established in NOD/SCID mice via the IV and IF route, respectively.

252 To address which genes were significantly differentially expressed within ALL-67 established using different transplantation conditions or between ALL-67 and samples of its corresponding primary ALL patient, the normalised gene expression data were compared between sample groups using limma module based on t-statistic calculation.

The significanct difference in gene expression values was only considered if a gene was differentially expressed between sample groups with a false discovery rate (FDR) of less than 0.05. Analysis of statistical differences in gene expression between samples of

ALL-67 established in NSG mice compared to those in NOD/SCID mice via a similar route of inoculation revealed no genes that were significant. Similarly, when gene expression profiles were compared based on the route of inoculation (IV versus IF), there were no statistically significant genes identified. This pattern of gene expression demonstrates the overall similarity of gene expression profiles between samples of

ALL-67, which suggests uniformity between ALL populations established using various engraftment strategies.

I further aimed to determine whether ALL-67 is characterised by a gene expression signature associated with the engraftment of patient samples in mouse models. I thus compared the gene expression profiles of samples from ALL-67 with those of two samples from the patient biopsy using the limma module. Interestingly, this analysis identified 2108 significantly differentially expressed genes between ALL-67 and the corresponding patient biopsy. Of these genes, 841 were significantly up-regulated and

1267 were significantly down-regulated in ALL-67, in comparison to the primary samples (Appendix B1 and B2). The top 100 differentially expressed genes between sample groups are shown in Figure 5.2.

253

Patient A4334

Figure 5.2. Heat map showing the top 100 most differentially expressed genes between ALL-67 and the two patient samples. The top 50 genes are over-expressed and the bottom 50 are under-expressed in ALL-67 in comparison to the primary ALL patient sample. Each column represents an individual sample of ALL-67 harvested from a mouse spleen or samples of the primary ALL patient biopsy. Change in colour within each row indicates expression levels relative to the average of gene expression of all samples. red indicates up-regulation, blue down-regulation.

254

To gain an insight into the influence of engrafting leukaemia cells into mouse models on biological characteristic of ALL cells, the differentially expressed genes between primary ALL samples and the related ALL-67 xenograft were further studied for the enrichment of specific biological processes by performing gene enology (GO) analysis.

The list of GO terms significantly enriched in genes down regulated in ALL-67 compared to its corresponding patient samples include immune system process, cellular defense response, oxygen binding and haemopoietic organ development (Table 5.1).

This suggests that the gene expression profiles of normal haematopoiesis and immune system process were over-represented in the BM biopsy samples of the patient and lost during the process of xenograft development.

Of the GO terms significantly enriched in genes up-regulated in ALL-67 compared to its primary ALL patient sample, a group of terms were identified that represent processes related to cell cycle phases, especially M phase, DNA repair, microtubule organisation and biogenesis, DNA replication and DNA integrity checkpoints (Table 5.2). Identification that many of the highly expressed genes in ALL-

67 are involved in regulation of cell division indicate that engraftment of primary ALL patient samples into mice enriched for ALL clonal populations with high proliferative activity.

255

Table 5.1. Gene Ontology terms down-regulated in the ALL-67 xenograft compared to the patient sample.

NAME ES NES FDR CELLULAR_DEFENSE_RESPONSE -0.77 -2.58 0 DEFENSE_RESPONSE -0.56 -2.31 0.009 LYMPHOCYTE_ACTIVATION -0.63 -2.22 0.02 GROWTH_FACTOR_BINDING -0.71 -2.21 0.01 IMMUNE_SYSTEM_PROCESS -0.52 -2.19 0.02 OXYGEN_BINDING -0.78 -2.16 0.02 CYTOSOLIC_PART -0.82 -2.16 0.02 T_CELL_ACTIVATION -0.71 -2.07 0.05 INFLAMMATORY_RESPONSE -0.54 -2.05 0.06 CARBON_OXYGEN_LYASE_ACTIVITY -0.69 -2.04 0.05 CYTOKINE_BINDING -0.62 -2.04 0.05 LEUKOCYTE_ACTIVATION -0.61 -2.03 0.05 HYDRO_LYASE_ACTIVITY -0.74 -2.01 0.06 IMMUNE_SYSTEM_DEVELOPMENT -0.56 -1.99 0.07 INTERLEUKIN_RECEPTOR_ACTIVITY -0.749 -1.97 0.08 IMMUNE_RESPONSE -0.49 -1.96 0.08 T_CELL_DIFFERENTIATION -0.73 -1.95 0.08 CELL_ACTIVATION -0.59 -1.95 0.07 HEMOPOIESIS -0.58 -1.95 0.07 HEMOPOIETIC_OR_LYMPHOID_ORGAN -0.58 -1.95 0.07 _DEVELOPMENT

ES, enrichment score; NES, normalised enrichment score; FDR, false discovery rate

256

Table 5.2. Gene Ontology terms up-regulated in the ALL-67 xenograft compared to the patient sample.

NAME ES NES FDR M_PHASE_OF_MITOTIC_CELL_CYCLE 0.65 2.74 0

M_PHASE 0.63 2.72 0 MITOTIC_CELL_CYCLE 0.60 2.71 0 CELL_CYCLE_PROCESS 0.58 2.59 0 CELL_CYCLE_PHASE 0.57 2.53 0 MITOSIS 0.65 2.46 0 RESPONSE_TO_DNA_DAMAGE_STIMULUS 0.56 2.45 0 CELL_CYCLE_GO_0007049 0.52 2.38 0 DNA_REPAIR 0.58 2.37 0 MICROTUBULE_CYTOSKELETON_ORGAN 0.69 2.37 0 IZATION_AND_BIOGENESIS CHROMOSOME 0.55 2.32 0 SPINDLE 0.69 2.29 0 CELL_CYCLE_CHECKPOINT_GO_0000075 0.63 2.28 7.69E-04 RESPONSE_TO_ENDOGENOUS_STIMULUS 0.51 2.28 7.14E-04 DNA_HELICASE_ACTIVITY 0.73 2.26 6.67E-04 HELICASE_ACTIVITY 0.64 2.26 6.25E-04 CHROMOSOMAL_PART 0.55 2.22 5.88E-04 DNA_METABOLIC_PROCESS 0.49 2.21 5.56E-04 CHROMOSOME PERICENTRIC_REGION 0.65 2.19 5.26E-04

ES, enrichment score; NES, normalised enrichment score; FDR, false discovery rate

257 Gene set enrichment analysis (GSEA) is another approach that could be used to delineate changes in gene expression between ALL samples in relation to publically available gene networks and signatures uploaded into the molecular signatures database

(Subramanian et al., 2005). In order to screen whether any set of the differentially expressed genes could provide further insight as to transcriptional networks and signatures related to the biology of ALL disease being modulated in ALL cells after engraftment into mice, the differentially expressed genes between the two groups of samples were entered into the GSEA database. GSEA revealed significant enrichment of the “Whiteford pediatric cancer marker” signature in the ALL-67 xenograft compared to its corresponding patient samples (Normalised Enrichment Score, ) (Figure

5.3). This signature represents a list of cancer specific genes that are overexpressed in a panel of xenografts established from eight common paediatric tumours including ALL disease compared to the normal human tissues (Whiteford et al., 2007).

To validate the enrichment of this signature in the samples of ALL-67, I selected candidate genes ranked at various levels of the gene signature (BIRC5, PTTG1, MCM4 and CCNB2) to confirm the difference in gene expression between ALL-67 and samples of its corresponding ALL patient using real-time quantitative PCR (RQ-PCR) (Figure

5.3B). The analysis identified significant differences in expression of all genes, except a trend toward significance observed in the expression of MCM4 gene, between ALL-67 samples and the two samples from the ALL patient (Figure 5.3C). Due to limited availability of patient samples in the CCI Tumour Bank, no validation of expression was done at the protein level.

258

Normalised enrichment score =3.2 A False Discovery Rate, FDR= 0

B

ALL-67 samples

Patient Patient A4334

C P=0.0513 P=0.0256 P= 0.0256 P=0.0303 8.0 Patient sample 7.5 ALL-67 Xenograft 7.0 6.5 050100150 6.0 5.5 5.0

(EF1A) Ct Delta Average 4.5 100150500 4.0 MCM4 PTTG1 CCNB2 BIRC5

Figure 5.3. Characterisation of Whiteford pediatric cancer marker gene set identified in ALL-67 xenograft. (A) GSEA plot shows the enrichment of Whiteford pediatric cancer marker gene set in ALL-67 versus primary ALL patient sample A4334. (B) Heat map represents the expression of genes selected from the above signature for validation using RQ-PCR. Graph (C) represents mRNA expression levels in samples of ALL-67 and primary samples of patient A4334 based on RQ-PCR analysis. The cycle thresholds (Ct) were normalised to EF1A mRNA to calculate delta Ct values. The difference in average delta Ct values from three technical repeat experiments between the two sets of samples is shown. The data were expressed as mean ± SEM from 3 independent experiments. The low average delta Ct levels indicate higher abundance of mRNA expressed in samples and vice versa. The statistical difference in delta Ct values for expression of each gene was compared between samples based on the Mann- Whitney U test.

259 The observation that ALL-67 cells are characterised by a unique gene expression signature compared to their corresponding primary patient cells raises a question to whether ALL-67 preserves the phenotype of the ALL disease after engraftment.

Previous studies from our group showed that the gene expression profiling clustered

ALL xenografts samples according to disease subtype (B-ALL, MLL-ALL and T-

ALL), indicating that ALL xenografts reflect the biology of the ALL subtype (Suryani et al., 2014).

ALL-67 was derived from an ALL patient classified by immunophenotype as BCP-ALL and characterised by the TEL/AML fusion. In order to verify whether ALL-67 retains the molecular features of ALL disease subtype, I assessed whether the gene expression profile of ALL-67 led to accurate clustering of this xenograft according to disease subtype. For this experiment, the gene expression analysis was done on two samples of

ALL-67 established in different mouse strains along with historical gene expression data available in the Leukaemia Biology Program of samples representing 8 MLL-ALL,

12 B-ALL and 9 T-ALL xenografts. The raw data of gene expression were log transformed, normalised and hierarchical clustering of ALL xenografts was then generated based on using the complete linkage model. The analysis of hierarchical clustering based on gene expression revealed that samples of ALL-67 clustered along with a group of xenografts derived from B-ALL patients demonstrating excellent representation of B-ALL phenotypes in ALL-67. Furthermore, ALL-67 samples clustered close to a group of B-ALL xenografts that included those derived from B-ALL patients harbours TEL/AML fusions (ALL-26 and ALL-53) as reported by Wong et al.

(2014) (Figure 5.4).

260

MLL T ALL B ALL

Figure 5.4. Hierarchical clustering of samples of ALL-67 xenograft with a panel of ALL xenografts representing different ALL subtypes. Unsupervised clustering dendrogram showing how the two samples of ALL-67 harvested from NSG and NOD/SCID (N/S) mice clustered along with samples from ALL xenografts. Blue line, MLL-ALL xenografts; green line, T-ALL xenografts; red line, B-ALL xenografts.

261 5.3 Ex vivo sensitivity of ALL xenografts to induction chemotherapy

The treatment for paediatric ALL patients is comprised of different chemotherapeutic drugs that are given to patients in three phases (induction, intensification and maintenance). The initial response of patients to the induction protocol, which typically includes vincristine, L-asparaginase and prednisone or dexamethasone chemotherapy is an important prognostic factor for childhood ALL patients (Den Boer et al., 2003;

Kaspers et al., 1997). It is thus important to determine which patients are likely to be sensitive to treatment and who have resistant disease and what could be the mechanisms mediating resistance to induction chemotherapy in IR ALL patients. A key credential of xenografts as a preclinical model for ALL disease is their ability to recapitulate the pattern of response of cells from ALL patients to their treatment (Lock et al., 2002;

Schmitz et al., 2011). ALL xenografts provide an approach to investigate the diversity in responses to induction chemotherapy among ALL patients and capture the acquisition of resistance to induction chemotherapy. A better modelling of response to induction treatments of ALL patients in ALL xenograft models might help to understand which

ALL patients could benefit from certain therapeutic agents and guide the development of effective chemotherapy regimens for ALL patients.

For assessment of the response to induction chemotherapy in IR ALL patients, samples from the four IR ALL xenografts established in the Pilot Study were assessed for ex vivo sensitivity to a single chemotherapy agent (vincristine, dexamethasone or L- asparaginase) based on Alamar Blue cytotoxicity assays.

262 5.3.1 Response of xenografts derived from ALL patients with different outcome (relapse vs. non-relapse) to induction chemotherapy

The first analysis focused on assessment of heterogeneity in response to induction chemotherapy among paired IR ALL xenografts derived from patients who presented with similar clinical characteristics at the time of diagnosis and consistent patterns of

MRD levels upon receiving induction therapy, yet responded differentially to their treatments. As shown in Chapter 3, the in vivo response of the four IR ALL xenografts to VXL induction chemotherapy reflected outcomes of patients from whom the xenografts were derived; the two xenografts derived from relapsed cases, ALL-64 and

ALL-66, exhibited in vivo resistance to VXL treatment in mice whereas ALL-65 and

ALL-67, which were derived from patients in CR, were sensitive to VXL chemotherapy. Since mice bearing ALL xenografts had been treated with a combination of dexamethasone, vincristine and L-asparaginase for two weeks, it is important to understand whether the pattern of in vivo response of ALL xenografts to drugs could be represented by the response to the single agents involved in the induction protocol.

To assess whether the pattern of response to a single chemotherapy agent could reflect the heterogeneity in response to induction chemotherapy between paired IR ALL xenografts, samples of ALL xenografts harvested from the spleens of the non-VXL treated mice were tested ex vivo against a single chemotherapy agent and the difference in sensitivity relative to control was measured after 48 hours of incubation with increasing concentrations of the drug. The pattern of ex vivo responses of ALL-64 to each drug was compared with that for ALL-65 and the response of ALL-66 to drugs was compared with that for ALL-67.

263 Measuring the response of cells after exposure to increasing concentrations of dexamethasone showed that both xenografts of the first pair, ALL-64 and ALL-65, exhibited resistance to dexamethasone with varying degree in sensitivity within samples from individual xenografts (Figure 5.5A). When the response to dexamethasone was assessed between samples of xenografts in the second pair, ALL-66 and ALL-67, both xenografts were very sensitive to low concentrations of dexamethasone (Figure 5.5B).

A comparison of statistical difference in sensitivity at a representative point 1 µM of dexamethasone revealed no significant difference between samples of xenografts of the first pair (P= > 0.99) and between samples of xenografts in the second pair (p= > 0.99)

(Figure 5.5C).

When the sensitivity profiles to increasing concentrations of vincristine were compared between samples of the first pair, ALL-64 exhibited resistance to the cytotoxic effects of vincristine with variable patterns of sensitivity between samples (Figure 5.6A). In contrast, vincristine had a dramatic cytotoxic effect on samples of ALL-65. The xenografts of the second pair (ALL-66&ALL-67) exhibited comparable levels of resistance to vincristine although ALL-66 was slightly more resistant especially at higher concentrations of vincristine (Figure 5.6B). Thus, significant differences in response to vincristine were observed within xenografts of the first pair (P=0.0268) and between xenografts of the second pair (P=0.0043) (Figure 5.6C).

In regard to L-asparaginase, both xenografts of the first pair exhibited a sensitive pattern in response to the drug (Figure 5.7A). However, xenografts of the second pair demonstrated different patterns of sensitivity in response to the drug. A high resistant pattern was maintained in different samples of ALL-66 compared to that in cells from

264 ALL-67 (Figure 5.7B). A comparison of response at a representative point 0.5 kU of L- asparaginase revealed no statistically significant difference in sensitivity between samples of xenografts in the first pair (P=0.4587) but a statistically significant difference in sensitivity was observed for xenografts of the second pair (p=0.0079)

(Figure 5.7C).

265

A B

120 ALL-64 100 100 ALL-65 ALL-66 80 ALL-67 80

60 60

40 40 Viability (% of control) of (% Viability Viability (% of control) of (% Viability 20 20

0 0 -10 -9 -8 -7 -6 -5 -10 -9 -8 -7 -6 -5 Dexamethasone concentration (log M) Dexamethasone concentration (log M)

C

100 P= > 0.99 P= > 0.99

80

60

40

20

Viability(% control) Viability(% 0 ALL-64 ALL-65 ALL-66 ALL-67 Dexamethasone (1µM)

Figure 5.5. Ex vivo assessment of ALL xenografts in response to dexamethasone. Samples from xenografts of the first pair (ALL-64&ALL-65) (A) and xenografts of the second pair (ALL-66&ALL-67) (B) were treated with increasing concentrations of dexamethasone, and sensitivity assessed based on Alamar Blue technique after 48 h incubation. Values expressed as a percentage of untreated control. Each data point indicates the sensitivity of ALL xenograft samples harvested from the spleens of different mice at certain drug concentrations. The data were expressed as mean ± SEM from 3 independent experiments. (C) The difference in sensitivity was calculated at a representative concentration (1µM) and compared between samples of each pair using the non-parametric Mann-Whitney U test.

266

A B 120 100 100 ALL-66 ALL-64 ALL-67 ALL-65 80 80 60 60 40 40 Viability (% of control) of (% Viability

control) of (% Viability 20 20

0 -10 -9 -8 -7 -6 -5 0 -10 -9 -8 -7 -6 -5 Vincristine concentration (log M) Vincristine concentration (log M)

C * ** 100 P= 0.0286 P= 0.0043

80

60

40 Viability(% control) Viability(%

20

0 ALL-64 ALL-65 ALL-66 AL-67 Vincristine (1 µM)

Figure 5.6. Ex vivo assessment of ALL xenografts in response to vincristine. Samples from xenografts of the first pair (ALL-64&ALL-65) (A) and xenografts of the second pair (ALL-66&ALL-67) (B) were treated with increasing concentrations of vincristine, and sensitivity assessed based on Alamar Blue technique after 48 h incubation. Values expressed as a percentage of untreated control. Each data point indicates the sensitivity of ALL xenograft samples harvested from the spleens of different mice at certain drug concentrations. The data were expressed as mean ± SEM from 3 independent experiments. (C) The difference in sensitivity was calculated at a representative concentration (1µM) and compared between samples of each pair using the non-parametric Mann-Whitney U test.

267

A 120 B ALL-64 ALL-66 ALL-67 100 ALL-65 100 80 80 60 60

40 40

Survival (% of control) of (% Survival Survival (% of control) of (% Survival 20 20

0 0 -12 -11 -10 -9 -8 -7 -6 -12 -11 -10 -9 -8 -7 -6 L-Asparaginase concentration (log kU) L-Asparaginase concentration (log kU)

** P= 0.0079 C 80

60 P= 0.4857

40

control) (% Viability

20

0 ALL-64 ALL-65 ALL-66 AL-67

L-Asparaginase ( 0.5 kU)

Figure 5.7. Ex vivo assessment of ALL xenografts response to L-asparaginase. Samples from xenografts of the first pair (ALL-64&ALL-65) (A) and xenografts of the second pair (ALL-66&ALL-67) (B) were treated with increasing concentrations of L- asparaginase, and sensitivity assessed based on Alamar Blue technique after 48 h incubation. Values expressed as a percentage of untreated control. Each data point indicates the sensitivity of ALL xenograft samples harvested from the spleens of different mice at certain drug concentrations. The data were expressed as mean ± SEM from 3 independent experiments. (C) The difference in sensitivity was calculated at a representative concentration (0.5 kU) and compared between samples of each pair using the non-parametric Mann-Whitney U test.

268 5.3.2 Assessment of the development of chemotherapy induced resistance

Although the use of chemotherapy agents in the treatment of childhood ALL has dramatically improved the survival of ALL patients, their efficacy can be limited by the development of resistance after treatment or selection and expansion of resistant sub- clones that were already present at minor quantities at the time of diagnosis.

Administration of the VXL regimen to mice may lead to induction of resistance to chemotherapy of ALL xenografts thereby mimicking the resistance to chemotherapy observed in ALL patients and thus would be of importance to ascertain whether the pattern of sensitivity to single drugs changes within ALL xenografts established in the

VXL-treated and non-treated mice.

In this section I aimed to determine whether the VXL selection led to development of resistance to a single agent of induction chemotherapy (vincristine, dexamethasone or L- asparaginase) in VXL-treated cells compared to their corresponding non-treated cells established from an individual xenograft. Samples of ALL-66 and ALL-67 were screened ex vivo for changes in response to increasing concentrations of a single chemotherapy agent incubated with cells for 48 hours and the difference in sensitivity was assessed based on Alamar Blue cytotoxicity assays.

In response to dexamethasone, samples of the ALL-66 xenograft which were established in the VXL-treated mice exhibited comparable sensitivity profiles to those observed in samples of the non-treated mice (Figure 5.8A) with no statistically significant difference in sensitivity between samples of both groups as shown at a representative concentration, 1 µM of dexamethasone (P= >0.99) (Figure 5.8C).

Likewise, a similar pattern of cytotoxic effect of dexamethasone was observed between

269 ALL-67 cells derived from the VXL-treated mice and its corresponding cells derived from the non-treated mice (Figure 5.8B). Both xenograft lines were sensitive to dexamethasone with no significant difference in cell viability between samples (P= >

0.99 at 1 µM).

Upon incubation of xenograft cells with vincristine, ALL-66 cells harvested from the

VXL-treated mice exhibited a pattern of response to the drug similar to that observed in cells derived from the non-treated mice. The results illustrated in Figure 5.9A demonstrate that both group of samples were resistant to various drug concentrations.

Comparison of the sensitivity between cells of ALL-66 at a representative drug concentration, 1 µM of vincristine, revealed no significant difference in response to the drug between samples of ALL-66 harvested from the VXL-treated and non-treated mice

(P=0.6623) (Figure 5.9C). In contrast, a noticeable change in the pattern of response to vincristine was observed within samples of ALL-67. Interestingly, ALL-67 cells exhibited a trend toward resistance in the VXL-treated cells especially upon exposure to high drug concentrations (Figure 5.9B). Therefore, significant difference in response to vincristine between samples of both groups was achieved as exemplified at a representative drug concentration, 1 µM of vincristine (P=0.0079) (Figure 5.9C).

The cytotoxic affect of L-asparaginase against ALL-66 derived from the VXL-treated mice was not different from that observed in cells that were derived from the non- treated mice (Figure 5.10A). Both groups of ALL-66 samples taken from the VXL- treated and non-treated mice were highly resistant to L-asparaginase with no significant difference in sensitivity between samples of both groups as demonstrated at 0.5 kU of the drug (P= >0.99) (Figure 5.10C). Similarly, the sensitivity of ALL-67 was not

270 changed between samples taken from the VXL-treated and non-treated mice upon incubation with L-asparaginase (Figure 5.10B). Both sample groups were relatively resistant to the cytotoxic effects of L-asparaginase with no statistically significant difference in sensitivity observed between both sample groups as shown at a representative concentration 0.5 kU of the drug (P=0.222) (Figure 5.10C).

Overall, there was no significant decrease in the sensitivity to any of the induction chemotherapy drugs observed in the VXL-treated cells of ALL-66 compared to that in non-VXL treated mice. These results indicate that ALL-66 is intrinsically resistant to vincristine and L-asparaginase. These results are supported by previous observations in this thesis. As shown in Figure 3.6, the pattern of in vivo engraftment of ALL-66 in mice PB was not changed by selection with VXL chemotherapy. Similarly, the VXL- treated cells of ALL-67 showed no significant change in sensitivity to dexamethasone and L-asparaginase. However, the observation that the VXL-treated cells of ALL-67 developed resistance to vincristine suggests that treatment of ALL-67 in mice with VXL chemotherapy led to an expansion of leukaemia cells which were able to overcome the inhibitory affects of vincristine.

271

A 120 B

100 100 ALL-66 Control ALL-67 Control ALL-66 VXL 80 80 ALL-67 VXL

60 60

40 40 Viability (% of control) Viability (% control)

20 20

0 0 -10 -9 -8 -7 -6 -5 -10 -9 -8 -7 -6 -5 Dexamethasone concentration (log M) Dexamethasone concentration (log M)

Dexamethasone (1 µM) C 80

P= > 0.99 60 P= > 0.99

40

Viability ( control) ( Viability 20

0

ALL-66 VXL ALL-66 VXL ALL-66 control ALL-66 control

Figure 5.8. Comparison of ex vivo sensitivity of ALL-66 and ALL-67 cells from VXL-treated mice and cells from the non-VXL treated mice to dexamethasone. The responses of ALL-66 (A) and ALL-67 (B) xenografts to increasing concentration of dexamethasone were compared between cells harvested from the VXL-treated and non- treated mice. Graph (C) compares the sensitivity between xenograft cells at a representative concentration (1 µM). Each data point shows xenograft cells harvested from different mice spleens. Horizontal lines indicate the SEM of each cohort.

272

A B 100 100 ALL-66 Control ALL-67 Control ALL-66 VXL 80 ALL-67 VXL 80 60 60 40 40 Viability (% of control) of (% Viability

20 control) of (% Viability 20 0 -10 -9 -8 -7 -6 -5 0 Vincristine concentration (log M) -10 -9 -8 -7 -6 -5 Vincristine concentration (log M)

Vincristine (1 M) C ** 80 P=0.6623 P= 0.0079

70

60

50 Viability (% of control) of (% Viability 40

30

ALL-66 VXL ALL-67 VXL ALL-66 Control ALL-67 Control Figure 5.9. Comparison of ex vivo sensitivity of ALL-66 and ALL-67 cells from VXL-treated mice and cells from the non-VXL treated mice to vincristine. The responses of ALL-66 (A) and ALL-67 (B) xenografts to increasing concentration of vincristine were compared between cells harvested from the VXL-treated and non- treated mice. Graph (C) compares the sensitivity between xenograft cells at a representative concentration (1 µM). Each data point shows xenograft cells harvested from different mice spleens. Horizontal lines indicate the SEM of each cohort.

273

A B

100 ALL-66 Control 100 ALL-66 VXL 90 ALL-67 Control 80 80 ALL-67 VXL

70 60 60 40 50 Viability (% of control) control) of (% Viability Viability (% of control) of (% Viability 40 20 30

20 0 -12 -11 -10 -9 -8 -7 -6 -12 -11 -10 -9 -8 -7 -6 L-Asparaginase concentration (log kU) L-Asparaginase concentration (log kU)

C L-asparaginase (0.5 kU)

100 P= > 0.99 P= 0.222

80

60

40

Viability (% of control) of (% Viability 20

0

ALL-66 VXL ALL-67 VXL ALL-66 Control ALL-67 Control

Figure 5.10. Comparison of ex vivo sensitivity of ALL-66 and ALL-67 cells from VXL-treated mice and cells from the non-VXL treated mice to L-asparaginse. The responses of ALL-66 (A) and ALL-67 (B) xenografts to increasing concentration of L- asparaginse were compared between cells harvested from the VXL-treated and non- treated mice. Graph (C) compares the cell sensitivity between xenograft cells at a representative concentration (0.5 kU). Each data point shows xenograft cells harvested from different mice spleens. Horizontal lines indicate the SEM of each cohort.

274 5.4 Identification of drug leads that could overcome vincristine resistance in VXL-treated ALL-67 cells

Given the significant change in response of the VXL-treated ALL-67 cells to vincristine compared to their corresponding non-treated samples demonstrated in Figure 5.9B, I proceeded to attempt to reverse the resistance developed to the antimicrotuble agent vincristine in the VXL-treated samples of this xenograft. In attempts to find a strategy to restore sensitivity to vincristine in the VXL-treated cells, I assessed whether in vivo treatment of ALL-67 with VXL chemotherapy had selected for a unique gene expression profile in the VXL-treated cells in comparison to that in the non-treated cells. The gene expression analysis based on microarray was performed on 10 samples from the VXL-treated ALL-67 and 11 samples of the non-treated ALL-67 cells. The raw signal intensity data were log transformed and normalised using the normalisation module within the GenePattern suite. The normalised gene expression was subjected to identification of differential gene expression between samples of the two groups using the limma module.

Assessment of differences in gene expression between samples revealed more than 3000 genes significantly changed between the VXL-treated and non-treated samples of ALL-

67. Genes with the largest expression change between the VXL-treated and non-treated samples of ALL-67 are shown in Figure 5.11. At an FDR< 0.05 and P value < 0.01, there were 1188 genes differentially expressed between the two groups (954 up- regulated and 234 down-regulated) (Appendix B3 and B4). In order to screen for a potential reversing agent to vincristine, I planned to use the list of differentially expressed genes defined by FDR< 0.05 and P-value <0.01 for identifying compounds that would be able to inhibit gene signatures associated with the development of vincristine resistance in the VXL-treated cells. This list of differentially expressed genes 275 was thus prepared for uploading into the Connectivity Map (CMap) database which allows discovery of connection between drugs and genes through matching uploaded gene list with thousands of gene signatures derived from various cell lines treated with bioactive compounds (Qu and Rajpal, 2012) As resources in the CMap database were generated using Affymetrix_U113A gene chips, it was required first to convert all differentially expressed genes from Unigene Code IDs to Affymetrix U113_A IDs as described in Section 2.4.5.

Table 5.3 summarises the top 20 ranked CMap agents which positively or negatively associate with the uploaded gene signature. The top ranked CMap agents include only two compounds, which had been known to produce anticancer activity. Interestingly these agents showed the highest negative association with the uploaded gene signature and they belong to the same class. Both tanespimycin (17 AAG) and geldanamycin are agents that inhibit the activity of the heat shock protein 90 (HSP90) which plays a key role in stabilising a variety of client proteins required for maintaining cell survival and response to cell stress (Li et al., 2009; Parker et al., 2014).

The relevance of the HSP90 inhibitors to the gene signature changed in the vincristine resistant cells was emphasised by the observed difference in level of HSP90 mRNA expression between the VXL-treated and non-treated ALL-67 (Figure 5.12). Previous studies from the PPTP showed limited activity of the HSP90 inhibitor 17 DMAG as a single agent against ALL xenografts (Smith et al., 2008). However, the purpose of this experiment was to use the HSP90 inhibitor as a mediator for reversing resistance developed against vincristine in the VXL-treated ALL-67 cells. The molecular chaperone HSP90 has been shown to bind to microtubules although this association is

276 not suggested to be related to microtubule assembly, but instead it could be as protective measure against the effect of cellular stress (Parker et al., 2014; Weis et al., 2010).

Further studies showed that microtubule associated proteins (MAPs) influence the activity of certain client proteins of HSP90 with implications to cancer cell sensitivity against chemotherapy (Lee et al., 2008).

277

ALL-67 ALL-67 VXL

Figure 5.11. Heat map showing the top 100 most differentially expressed genes between the VXL-treated and non-treated ALL-67 cells. The top 50 genes are over- expressed and the bottom 50 are under-expressed in the VXL-treated ALL-67 in comparison to the non-VXL treated ALL-67 samples. Each column represents an individual sample of ALL-67 harvested from a mouse spleen of the VXL-treated and non-treated mice. Change in colour within each row indicates expression levels relative to the average of gene expression of all samples. Red indicates up-regulation, blue down-regulation. 278 Table 5.3. CMap results of drug leads associated with differentially expressed genes between the VXL-treated and non-treated ALL-67 cells. The 20 compounds that showed high negative or positive association with the uploaded signature are shown. The two HSP90 inhibitors, which negatively associated with the signature, are highlighted in yellow.

Rank Cmap name Enrichment P 1 Tanespimycin -0.422 0

2 Amantadine 0.867 0.0004 3 Geldanamycin -0.492 0.0006

4 Fludrocortisone 0.647 0.000 5 nipecotic acid -0.842 0.001

6 Triflupromazine 0.833 0.001 7 Clorsulon 0.826 0.001

8 Amitriptyline 0.715 0.001

9 Hexetidine 0.82 0.001 10 metamizole sodium -0.695 0.001 11 Haloperidol 0.313 0.002 12 Isoflupredone 0.889 0.002 13 Furazolidone 0.804 0.002 14 Pentoxifylline 0.737 0.002 15 ursodeoxycholic acid -0.881 0.003 16 Famprofazone -0.67 0.003 17 Karakoline 0.67 0.003 18 Oxolamine 0.792 0.003 19 Timolol 0.788 0.003 20 Levomepromazine 0.781 0.004

279

ALL-67 ALL-67 VXL

Figure 5.12. mRNA expression of HSP90 gene family in the VXL-treated versus non-treated cells of ALL-67.

280 To test the potential benefit from inhibiting HSP90 for reversing resistance developed against vincristine I decided to incubate samples of the VXL-treated ALL-67 with increasing concentrations of the geldanamycin analogue 17 DMAG as a single agent and in combination with a fixed concentration (10 µM) of vincristine. 17 DMAG and 17

AAG are both derivatives of geldanamycin, which target the HSP90 molecule for inhibition. 17 DMAG was selected based on its higher stability than geldanamycin, solubility in water compared to 17 AAG and it has been already optimised for use in our mouse models (Smith et al., 2008). Based on a study comparing the inhibitory effect of

17 DMAG and 17 AAG on HSP90, the depletion of HSP90 was found to be more pronounced in cells treated with 17 DMAG compared to those that received 17 AAG

(Smith et al., 2005)

The response to 17 DMAG as a single agent or in combination with 10 µM of vincristine was tested against two selected samples of the VXL-treated ALL-67 cells and sensitivity to every drug concentration was plotted in relation to untreated control cells. The sensitivity of samples was also plotted for combined drugs relative to the control cells, and response of cells was compared against the expected additive effect calculated from the total effect of each drug alone. The interaction between any combined drugs can be defined as additive, synergistic or antagonistic. If the sensitivity curve of the combined drugs falls below the expected additive curve, the activity of combined drugs can be classified as synergistic while it is classified as antagonistic activity if it falls above the expected additive curve.

As shown in Figure 5.13, 17 DMAG showed potent anti-tumour activity as a single agent against the two samples of ALL-67 xenograft. However, the effect of the 281 combined drugs against the two samples of the VXL selected ALL-67 was antagonistic interaction as the viability of ALL-67 after exposure to combined drugs was higher than the expected response of additive effect calculated from the sum of cytotoxic effects produced by each agent separately. The reasons for failure of 17 DMAG to enhance the sensitivity of ALL-67 toward vincristine are uncertain. A potential reason could be related to the inability of ALL xenografts to proliferate ex vivo, which might affect the integrity of functional regulatory events of gene expression.

282

A 100 90 10 µM Vincristine 80 17 DMAG 10 µM VCR+17 DMAG 70 Expected additive 60 50 40 30 Viability (% of control) of (% Viability 20 10 0 -10 -9 -8 -7 -6 -5 Drug concentration (log M)

B 110 100 10 µM Vincristine 17 DMAG 90 10 µM VCR+17 DMAG 80 Expected additive 70 60 50 40

Viability (% of control) of (% Viability 30 20

10 0 -10 -9 -8 -7 -6 -5 Drug concentration (log M)

Figure 5.13. Ex-vivo assessment of the cytotoxic effect of 17 DMAG ± 10 µM of vincristine on ALL-67 VXL-treated samples. Two samples of the VXL-treated ALL- 67 (A and B) were treated with increasing concentrations of 17-DMAG ± 10 µM vincristine, and sensitivity was assessed by Alamar Blue assay after 48 hours incubation. The Predicted Effect was calculated as [(% viability of cell Drug A) x (% viability of cell Drug B)/100)] for each concentration and was plotted along with effect of drug tested. Sensitivity of treated cells was presented as a percentage of untreated control. Each data point represents the mean ± SEM from 3 independent experiments.

283 5.5 Identification of genes associated with vincristine resistance in the ALL-67 xenograft

Characterisation of the mechanisms underlying the development of vincristine resistance in the VXL-treated ALL-67 xenograft could help to understand how cells can avoid the cytotoxic effect of the drug and suggest biomarkers for resistance to the drugs.

Vincristine is a microtubule-targeting agent (MTA), which exerts its effect by binding to tubulin subunits and disrupting polymerisation of microtubule networks required for formation of mitotic spindles during cell division (Verrills and Kavallaris, 2005).

Insensitivity of cells to MTAs has been suggested to be due to either efflux of drugs by membrane pump proteins, and inability of drugs to interact with tubulins, due to alterations in drug-protein binding sites. The mechanisms by which cells develop resistance to the MTAs also include alterations in total tubulin contents, tubulin polymerisation, microtubule associated protein or tubulin isotype content and failure of cells to induce cell death in response to the disruption of the microtubule network

(Verrills and Kavallaris, 2005).

In search for the mechanisms by which ALL-67 established resistance to vincristine, I decided to make use of differentially expressed genes between the VXL-treated and non-treated ALL-67 xenograft to infer the biological pathways or gene signatures with possible relevance to vincristine resistance. All differentially expressed genes between samples of ALL-67 were subjected to analysis of gene ontology terms based on the

GSEA module. The top 20 gene ontology terms which were significantly presented in the VXL-treated versus non-treated ALL-67 included terms representing the process of

RNA regulation, translation process, protein transportation and genes involved in the structure of cell compartments such as mitochondria (Table 5.4). The finding that the

VXL-treated ALL-67 samples express elevated levels of translation-related genes 284 indicates high status of protein synthesis in these samples. Within the top 50 GO terms

(FDR<0.01) upregulated in the VXL-treated ALL-67, there was significant enrichment of microtubule cytoskeleton genes in the VXL-treated ALL-67 treated samples which exhibited resistance to vincristine (NES=2.12 and FDR=8.98E-5) (Figure 5.14A). This finding highlights a difference in genes involved in cellular response to vincristine between vincristine resistance and sensitive ALL-67 and thus suggests that abundance of microtubule genes could contribute to the vincristine resistance observed in the VXL- treated ALL-67 cells.

To specify which genes involved in microtubule networks are most highly upregulated in the VXL-treated versus non-treated cells of ALL-67, I searched the list of differentially expressed genes between samples based on FDR <0.1 and P value 0.02 for those related to the microtubule network. The search resulted in identification of a list of alpha and beta tubulin isotype genes including TUBA1, TUBA1B, TUBB6, TUBB and

TUBB3, microtubule associated proteins like MAP1S, MAPRE1, MAPRE2 and TPX2, genes encoding an enzyme involved in posttranslational modification of tubulin such as

TTL, other genes known to associate and/or bind to tubulin and genes which are involved in entry into cell cycle and cell cycle progression after DNA damage from chemotherapy such as the MSTL1 gene (Figure 5.14B). Although no validation of expression levels of these genes has been done in this thesis, the finding of high expression of many molecules involved in the microtubule system indicates the potential relevance to resistance of ALL-67 to vincristine.

285

Table 5.4. Gene Ontology terms up-regulated in the VXL-treated ALL-67 cells compared to non-treated cells.

NAME ES NES FDR STRUCTURAL_CONSTITUENT_OF_RIBOSOME 0.77 2.86 0 RIBONUCLEOPROTEIN_COMPLEX 0.63 2.48 0 RIBOSOME 0.74 2.47 0 MITOCHONDRIAL_PART 0.60 2.41 0 RNA_PROCESSING 0.60 2.39 0 RNA_BINDING 0.58 2.39 0 RNA_SPLICING 0.63 2.36 0 TRANSLATION 0.58 2.35 0 MITOCHONDRIAL_LUMEN 0.68 2.35 0 0.55 2.31 0 INTRACELLULAR_PROTEIN_TRANSPORT 0.58 2.29 0 ORGANELLE_LUMEN 0.54 2.28 0 MEMBRANE_ENCLOSED_LUMEN 0.54 2.28 0 MITOCHONDRIAL_MATRIX 0.68 2.27 0 ENVELOPE 0.56 2.26 0 ORGANELLE_ENVELOPE 0.56 2.25 0 MITOCHONDRIAL_INNER_MEMBRANE 0.62 2.25 0 RIBONUCLEOPROTEIN_COMPLEX_BIOGE NESIS_AND_ASSEMBLY 0.60 2.25 0

MITOCHONDRIAL_MEMBRANE_PART 0.65 2.24 0

286

B

A

ALL-67 ALL-67 VXL

Normalized Enrichment Score=2.12

FDR value= 8.98E-5

Figure 5.14. Representation of microtubule related genes in the ALL-67 xenograft. (A) GSEA graph showing significant enrichment of a microtubule cytoskeleton gene signature in the VXL-treated ALL-67 samples. B) Heat map of microtubule related genes differentially expressed between samples of ALL-67 that received in vivo VXL treatment and samples from non-treated ALL-67. Genes that were significantly upregulated in vincristine resistant ALL-67 xenografts with a false discovery rate (FDR)

< 0.1 and P value < 0.02 include alpha/beta tubulin isotypes and microtubule associated genes.

287

5.6 Summary and Discussion

The heterogeneity in biology of ALL disease has been linked to failure of chemotherapy treatment and difference in outcomes of ALL patients. Studying the heterogeneity within ALL clonal populations could reveal the biological and functional characteristics of ALL disease clones able to establish the disease and provide rationale for better management of ALL disease. One way to examine the heterogeneity in leukaemia disease is through characterising the biological and functional features of cells able to survive under different stress as applied from engraftment into mouse models and/ or in vivo treatment with induction chemotherapy. The experiments described in this chapter attempted to employ ALL xenografts established in immunocompromised mice to capture the heterogeneity in ALL disease at the time of diagnosis.

5.6.1 Diversity in transcriptional profiles of an IR ALL xenograft and its corresponding patient sample

The engraftment of ALL cell populations into mouse models might depend on the conditions applied to establish ALL xenografts in mice (Landau et al., 2014).

Xenotransplantation of ALL samples utilising mouse strains with different levels of immunity could favour growth of ALL clones according to their ability to establish under the influence of the murine immune system and their ability to circumvent factors that may hinder cell homing into an appropriate niche microenvironment. To explore the heterogeneity in clonal populations within an ALL-patient diagnosis sample transplanted into two mouse strains with different levels of immunity (NSG and

NOD/SCID) via two different routes of inoculation (IV and IF), I selected 10 samples of

ALL-67 established using various engraftment conditions for analysis of gene expression based on microarray technology. The result presented in Section 5.2 showed that neither mouse strain nor route of inoculation revealed significant differences in 288 gene expression profiles within samples of ALL-67 established using different engraftment conditions. Although the engraftment of the patient sample used to establish ALL-67 in mice was somewhat different between cells engrafted in NSG and

NOD/SCID inoculated via the same route of inoculation (Table 3.6), the gene expression profiles were highly consistent among ALL-67 cells established using various engraftment conditions. This high level of transcriptional concordance between samples of ALL-67 established using different engraftment conditions suggests that

ALL-67 represents common ALL clonal populations that are characterised by enhanced ability to engraft in mouse models. The selective advantages applied on inoculated ALL patient samples by mouse models are poorly understood. However, the difference in levels of immune system and site of engraftment are likely to have a role in selection of leukaemia populations (le Viseur et al., 2008; Meyer and Debatin, 2011).

When the gene expression profiles of ALL-67 were compared with those of samples from the patient from whom the xenograft was derived, the analysis showed that samples of ALL-67 clustered separately from their patient samples and xenograft samples appear to express an increased amount of molecules involved in the cell cycle and mitosis compared to the parental primary ALL cells. An interesting observation from this analysis was that there was overexpression of a gene signature previously shown to be common in samples of xenografts derived from different paediatric cancer patients including ALL disease (Whiteford et al., 2007).

Overall, these results indicate that engraftment of ALL patient samples in immunodeficent mice led to selection and expansion of leukaemia cell populations with 289 high proliferative potency compared to that in primary ALL patient. In agreement with this finding, Clappier et al. (2011) had previously reported enrichment of cell cycle and mitosis signatures in 9 T-ALL xenografts compared to their corresponding diagnosis samples of ALL patients. The authors had also identified that these signatures present in relapse samples of ALL patients, which supported their claim that xenograft cells represent clonal selection of ALL disease. The finding that ALL-67 xenograft was characterised by an increased expression of proliferation molecules has potential impact on the use of ALL xenografts in mouse models for preclinical testing of anti cell cycle and anti proliferation molecules due to overestimation of drug efficacy.

Another aspect highlighted from this study was that the primary ALL samples showed significant up-regulation of genes involved in haematopoiesis and immune system compared to ALL-67 xenograft samples. The possible explanation of this observation is that diagnostic primary ALL biopsy samples could be contaminated with normal haematopoietic cells even though the report from the hospital indicated 97% blast cells present in the BM biopsy samples.

5.6.2 Response of xenografts derived from ALL patients with different outcome (relapse vs. non-relapse) to induction chemotherapy

Although patients who are stratified into the IR ALL subtype present with similar clinical features at the time of diagnosis, patients demonstrate diverse outcomes in response to their treatment. Among the large number of chemotherapeutic agents used for the treatment of ALL patients, dexamethasone, vincristine and L-aspargaginase are the only drugs which their response could predict outcome in ALL patients (Kaspers et al., 1997). Therefore, the pattern of responses of ALL cells to these drugs were extensively studied using biopsy samples of ALL patients, cell lines and ALL 290 xenografts established in mouse models for characterising treatment failure in ALL disease and exploring critical biologic determinants of treatment outcome.

As the in vivo response of IR ALL xenografts to VXL treatment stratified xenografts according to outcomes of patients from whom xenografts were derived, I intended to test whether the discrepancy in response to chemotherapy treatment between IR ALL patients who presented at diagnosis with similar clinical features could be identified by response to individual drugs involved in the induction regimen. For this purpose, I tested the pattern of sensitivity to single agent dexamethasone, vincristine and L- aspargaginase between samples of four IR ALL xenografts established from two pairs of patients with different outcome. The ex vivo pattern of response to chemotherapy was compared between samples of ALL xenografts from each pair harvested from the non- drug treated mice (1st pair [ALL-64 and ALL-65] and 2nd pair [ALL-66 and ALL-67]).

The xenografts of the first pair were equally resistant to dexamethasone, sensitive to L- asparaginase but differentially respond to vincristine. In the second pair, both xenografts were equally sensitive to dexamethasone but variably respond to vincristine and L- asparaginase. Unexpectedly, the response of paired ALL xenografts to dexamethasone did not reflect outcomes of ALL patients whom xenografts were derived whereas the sensitivity to vincristine in the first pair and vincristine and L-asparaginase in the second pair had reflected the in vivo response of xenografts to VXL selection and outcomes of

ALL patients. These observations are not consistent with results from previous study conducted by our group. In that study Liem et al. (2004) showed that the patterns of in vivo response of ALL xenografts to dexamethasone and vincristine were stratified according to patient outcome. However, ex vivo assessment of sensitivity of these xenografts to each drug revealed that the sensitivity to dexamethasone, but not

291 vincristine, showed significant correlation with the in vivo response in mice and patient outcomes (Liem et al., 2004).

The response to glucocorticoids is known to be a major prognostic factor in childhood

ALL, and the in vitro sensitivity of primary ALL cells was found to correlate with the clinical outcome of the paediatric patients from whom the cells were obtained (Klumper et al.,1995). Data from the experiment testing the responses of IR ALL xenografts to chemotherapy highlighted a disparity between the ex vivo and in vivo responses of some of the ALL xenografts to dexamethasone. The discordant in vivo/vitro responses were not expected as previous studies from our lab showed consistency between the ex vivo and in vivo dexamethasone responses of a panel of childhood ALL biopsies that reflected the clinical outcome of the patients from whom they were derived (Liem et al.,

2004; Bachmann et al., 2005). In my study, ALL-66 xenograft cells exhibited resistance to the in vivo treatment with VXL chemotherapy, represented by the small differences in

EFS between the VXL treated and non-treated mice, whereas the ex vivo treatment of cells with dexamethasone for 48 hours resulted in high sensitivity to the drug. This discrepancy could be explained by the differences in drug exposure between in vitro and in vivo models. In vitro the cells are continuously exposed to drug, but in vivo the exposure of cells to the drug is affected by metabolism or pharmacokinetics of the drug.

A potential reason for xenograft cells evading the in vivo treatment with dexamethasone is the ability of xenograft cells to hide from the toxic insult of the drug in specific sanctuary sites and/ or by interacting with their surrounding stromal cells.

The data of this experiment highlighted possible differences in response to chemotherapy among IR ALL patients with different clinical and cytogenetic features.

292 For instance, it was noted that both xenografts of the first pair were derived from ALL patients with chromosomal hyperdiploidy and showed resistance to dexamethasone whereas xenografts of the second pair were derived from ALL patients with TEL/AML cytogenetic abnormality and were both sensitive to dexamethasone. While investigation into the biological determinants of heterogeneity of ALL xenografts in response of IR

ALL xenografts to drugs was not followed in this study, it would be of interest to validate the difference in response of paired IR ALL xenografts to vincristine and L- asparaginase in vivo and study the mechanisms underlying any difference in response between patients with similar clinical and cytogenetic features.

5.6.3 Assessment of the development of chemotherapy induced resistance

Resistance to chemotherapy can develop either due to selection of leukaemia populations, which are intrinsically resistant to chemotherapy or acquisition of drug resistance upon treatment with chemotherapy. In an effort to understand whether treatment with VXL chemotherapy can lead to changes in drug sensitivity of ALL xenografts, the ex vivo sensitivity to single agent dexamethasone, vincristine and L- aspargaginase was compared between samples taken from ALL-66 or ALL-67 established in the presence and absence of two weeks treatment with VXL chemotherapy. No change in sensitivity was observed between samples of ALL-66 in response to any of the three drugs and between samples of ALL-67 in response to dexamethasone and L-asparaginase. A possible caveat of these results is that Alamar

Blue technique estimates the cytotoxic affect based on the mitochondrial activity and may not accurately reflect moderate resistance. However, the VXL-treated ALL-67 cells exhibited increase in resistance to vincristine when compared with their non-treated cells. Interestingly, similar drug responses were observed among samples harvested from individual mice which received VXL treatment. This result indicates that the VXL

293 treatment has led to expansion of ALL clones, which are characterised by increased ability to survive the inhibitory effect of vincristine.

A previous study published by Samuels et al. (2014) supported the development of drug resistance in ALL xenografts after exposure to repeated cycles of chemotherapy. In this study, T-ALL xenografts were reinoculated into NOD/SCID mice and engrafted mice were exposed to either in vivo selection with multiple courses of a four-drug combination of vincristine, dexamethasone, L-asparaginase and daunorubicin (VXLD) or vehicle normal saline. Relapse cells after VXLD treatment were harvested from mice and then reinoculated into NOD/SCID mice. Once engraftment of ALL cells was apparent in the murine PB, all engrafted mice then received selection with VXLD or normal saline in a manner similar to that applied in the previous passages. Analysis of ex vivo sensitivity between cells harvested from mice which received VXLD1 and

VXLD2 treatment and their control mice revealed that the VXLD treated sublines were characterised by decrease in sensitivity to dexamethasone and L-asparaginase compared to their vehicle-treated control mice and a higher trend of resistance was observed in the

VXLD2 sublines. This study showed the acquisition of resistance to two of the main drugs used in induction chemotherapy for treatment of T-ALL children after applying

VXLD treatment, suggesting that drug resistance observed in mice upon VXLD selection may resemble what is observed in the clinic.

5.6.4 Identification of drug leads that could overcome vincristine resistance in the VXL-treated ALL-67 cells

Vincristine is an integral component for successful treatment of ALL disease in children. Development of resistance to vincristine or other drugs of the induction chemotherapy reduces the options for treating ALL patients and impacts the survival of

294 ALL patients. Overcoming vincristine resistance could improve outcomes of ALL patients who develop relapse due to vincristine resistance. The differences in sensitivity to vincristine patterns observed in the VXL-treated ALL-67 may provide an approach for modelling the resistance to the drug observed in the clinic. Although it was not clear whether establishment of resistance to vincristine in ALL-67 after applying VXL treatment was due to selection of vincristine resistant clones or due to acquisition of resistance mediated by chemotherapy effect, characterising the difference in biological features associated with vincristine resistance could suggest a strategy for reversing resistance to the drug.

Differential gene expression using microarray is an appropriate tool commonly used to examine the difference in biological features of chemotherapy responders and non- responders (Cleaver et al., 2010; Holleman et al., 2004). Using this approach, I sought to make use of the significantly differentially expressed genes associated with vincristine resistance to search for compounds that would be predicted to reverse the resistance phenotype. The defined gene signature that discriminates between ALL-67 control cells and the VXL-treated cells exhibiting resistance to vincristine was uploaded into the CMap database. This analysis identified two HSP90 inhibitors as potential reversing agents to the signature enriched in the vincristine resistance cells. HSP90 is a molecular chaperone that regulates and stabilises many proteins involved in growth, differentiation and survival of normal and cancerous cells. The basal level expression of

HSP90 is required for normal protein folding and maintenance of newly translated proteins whereas high expression of this molecule is induced by cellular stresses as a consequence of hyperthermia, chemotherapy, radiation and other environmental insults

(Richardson et al., 2011; Weis et al., 2010). This response is assumed to be as a protective measure of cells from potentially lethal dangers that could affect cell

295 survival. The involvement of HSP90 in these crucial functions makes it an attractive target for inhibiting the survival of cancer cells and thus the activity of HSP90 inhibitors was tested against various cancer types including haematological malignancies (Miyata et al., 2013; Richardson et al., 2011; Smith et al., 2008).

Given the fundamental roles of HSP90 in cell survival, ALL cells might use this pathway as a mechanism to sustain cell survival. I hypothesised that 17 DMAG would sensitise ALL-67 cells to killing by vincristine. The 17 DMAG demonstrated ex vivo antitumor activity against ALL-67 VXL-treated cells as a single agent. This finding is in agreement with what was observed by other studies in various cancer types, which could be attributed to addiction of cancer cells on oncogenic signals to support their survival. However, there was no synergistic interaction observed when ALL-67 cells were exposed to this drug in combination with vincristine. In contrast to my results, a few studies reported the enhancement of antimicrotubule agents by HSP90 inhibitors.

For example, the in vivo assessment of activity of the antimicrotubule agent docetaxel and the HSP90 inhibitor IPI-504 against non small-cell lung cancer tumour (NSCLC) xenograft cells showed synergistic interaction between the two drugs (O'Connell et al.,

2014). Another study published by Proia et al. (2012) showed that the HSP90 inhibitor ganetespib was able to potentiate the in vitro cytotoxic activity of the antimicrotubule agents paclitaxel, docetaxel and vincristine at a number of doses against the H1975

NSCLC cell line. Interestingly, these findings were reproduced in in vivo models when

NSCLC cell lines were xenografted into SCID mice. The authors reported significant inhibition of tumour growth in response to the combination between ganetespib and paclitaxel or docetaxel agents (Proia et al., 2012).

296 As I explained previously the lack of proliferation in ALL xenografts cells outside their hosting animal might limit the efficiency of the combined drug effect due to inadequate activity of cell signaling required for transmission of death signals. Nevertheless, it could be argued that this proposal is challenged by the observation that vincristine exhibited moderate cytotoxic activity in the non-VXL treated ALL-67 xenograft cells. A potential explanation for this observation is that vincristine-induced cell death is independent of mitotic arrest. A recent study showed that primary ALL cells in G1 phase of cell cycle are also susceptible to the cytotoxic effects of vincristine ex vivo

(Kothari et al., 2016). Although both dividing and non-dividing cells can be targeted by vincristine, the VXL-treated ALL-67 cells exhibited ex vivo resistance to vincristine alone or in combination with 17 DMAG. The lack of effect of vincristine in a limited proliferation state supports testing the efficacy of the combination between the two drugs in vivo. The therapeutic enhancement of the two drugs should be tested by comparing the EFS of mice in response to an appropriate schedule of each drug between mice that received single drugs and those that received the combination. The assessment should also include testing whether the 17 DMAG induces inhibition of HSP90 protein by treating another group of mice with different concentrations of the HSP90 inhibitor and harvesting cells at appropriate time points for comparing the changes in the levels of HSP90 protein between the 17 DMAG treated and non-treated cells.

5.6.5 Identification of genes associated with vincristine resistance in the ALL-67 xenograft

A significant body of literature has linked vincristine resistance in ALL to alterations in the microtubule system (Arai et al., 2006; Bhat and Setaluri, 2007; Cheung et al., 2010).

To examine whether any gene sets and pathways with direct relevance to the cellular

297 response to vincristine were significantly altered in the VXL-treated ALL-67 cells which exhibited vincristine resistance, gene expression based on microarray analysis was conducted on VXL-treated and non-treated samples of ALL-67. The analysis revealed enrichment of microtubule cytoskeleton gene sets, many of significantly differentially expressed genes and pathways including an up-regulation of microtubule associated proteins and tubulin isotypes (alpha and beta), and genes involved in regulation of the cell cycle. High expression of tubulin isotypes especially beta and microtubule associated proteins have been frequently reported to be related to resistance of cancer cells to microtubule-targeting agents (Arai et al., 2006; Bhat and Setaluri,

2007; Cheung et al., 2010). However, multiple factors can also control microtubule dynamics including folding of tubulin monomers by tubulin folding cofactors, posttranslational modifications of tubulins (Nogales, 2000). In my study I observed an increase in expression of genes involved in microtubule cytoskeleton system in the

VXL-treated ALL-67, which exhibited resistance to vincristine. These genes include alpha and beta tubulin isotypes, MAPs, tubulin folding cofactors and an enzyme involved in regulation of the tubulin detyrosination–tyrosination cycle. This could suggest that the observed changes in expression of these genes support the stability of microtubules in the VXL-treated ALL-67 in the presence of vincristine.

Characterisation of the functional importance of these genes in vincristine resistance using an appropriate model could provide a greater understanding of resistance of ALL cells to vincristine, which could eventually provide an approach for improving outcomes of ALL patients with resistance to vincristine.

298 CHAPTER 6 FINAL DISCUSSION AND FUTURE DIRECTIONS

Current treatment protocols stratify ALL patients at diagnosis into Standard Risk (SR),

Intermediate Risk (IR) and High Risk (HR) subgroups. Evaluation of the risk of relapse in ALL patients is based on assessing the molecular/cytogenetic characteristics of ALL and their response to initial treatment with glucocorticoid at time of diagnosis. These strategies select ALL patents for receiving multi agent chemotherapy protocols adjusted according to their predicted risk of relapse (Moricke et al., 2008; Raetz and Bhatla,

2012). Although the risk-adjusted treatments have improved the 5-year survival of childhood ALL to approximately 90%, up to 20% of ALL patients including a considerable large number of those stratified into the IR subtype suffer from a poor survival rate due to relapse of the disease (Conter et al., 2010; Nguyen et al., 2008).

There is therefore an intention to identify IR ALL patients who require more intensive chemotherapy treatments or special drug interventions to prevent relapse and distinguish IR ALL patients with risk of relapse from other patients who may respond well to standard chemotherapy. The progress in identifying prognostic factors, which could be broadly applied to ALL patients, has reached its limit with little progress over the past 20 years. There is much hope that ALL patients who share common genetic profiles could benefit from a design of treatment based on their disease biology however this purpose is challenged by the genetic heterogeneity of ALL disease.

Evidence from using preclinical mouse models has shown that ALL patients whose transplanted cells engrafted quicker in immuocompromised mice had inferior outcome with an increased risk of relapse. It was thus suggested that ALL xenografts in murine models may provide an approach for upfront prediction of ALL patient outcome (Meyer et al., 2011; Woiterski et al., 2013). Although these reports and others indicated that 299 ALL xenograft is reasonably predictive of high and standard risk of relapse, it is unknown whether the sensitivity of this model is high enough to separate between patients who have different outcomes yet were assigned to the IR ALL subtype. With the accumulation of reports highlighting the clinical relevance of ALL xenograft models, there is much confusion about the most appropriate engraftment setup up which could allow upfront identification of treatment response in ALL patients. Various engraftment strategies, including different mouse strains (NOD/SCID and NSG), different inoculation sites (IV and IF), pre-engraftment conditioning of mice with total body radiation, and post engraftment selection with chemotherapy, have been used to investigate the correlation between engraftment and outcomes of the patients from whom xenografts were derived. It was therefore of critical importance to determine which engraftment strategies could maximise the relevance of IR ALL xenografts to reflect outcomes of patients from whom they were derived.

In chapter 3, I carried out a Pilot Study using a panel of four IR ALL patients (samples from 2 relapsed and samples from 2 non-relapsed patients) treated according to the same protocol who exhibited different responses to their treatment. This study aimed to optimise the most appropriate engraftment condition, which retains high efficiency, and quicker engraftment along with its ability to allow stratification of IR ALL xenografts according to outcomes of patients. The analysis showed that the efficiency and speed of engraftment were greater for NSG compared with NOD/SCID mice, and IV compared to IF inoculation, with no stratification of ALL xenografts according to patient outcomes. Importantly, the response to VXL induction chemotherapy in xenografts reflected the clinical outcome of the patients from whom they were derived. The finding that IR ALL xenografts derived from patients in CR1 were more susceptible to VXL treatment than xenografts derived from patients who died of their disease is in 300 agreement with the observation reported by Szymanska et al. (2012). In this study, the authors showed the clinical relevance of ALL xenografts treated with VXL chemotherapy to reflect outcomes of a small and heterogeneous cohort of ALL patients treated under variable induction chemotherapy protocols (Szymanska et al., 2012). The critical concept drawn from my experiment was that the extent to which engraftment of

IR ALL patient samples into mice is able to reveal the heterogeneity in patient outcomes could not be only depending on the ability of ALL cells to establish themselves in host animals but importantly depends on their response to a selective pressure applied from chemotherapy treatment.

Encouraged by the observations from the Pilot Study, I further aimed to validate my hypothesis and study the reliability of using engraftment of samples from IR ALL patients for prediction of patient outcomes based on a larger cohort of IR ALL patients.

I decided to characterise the engraftment of 30 IR ALL samples derived from early and late relapse and CR1 patients inoculated into NSG mice via the IV route in presence or absence of 2 weeks of VXL treatment. Assessing the difference in EFS of control mice at various time points of engraftment (1% and 25% human cells and TTL) in relation to the outcomes of IR ALL patients from whom xenografts were derived demonstrated no significant difference in time to events of mice used to establish ALL xenografts. This emphasised the observation from the Pilot Study, which indicated no prognostic benefits for IR ALL patients using the engraftment properties of their samples in the non-VXL treated mice.

The most important observation of the Main Study was the reproducibility of the finding from the Pilot Study, which demonstrated that engraftment of ALL patients in the VXL-treated mice, stratifies ALL xenografts according to whether they were 301 derived from relapsed or non-relapsed patients. This strongly indicated that the in vivo responses of IR ALL xenografts to VXL treatment improved the clinical relevance of this model to the disease of patients from whom xenografts were derived. A comparison of the EFS of the VXL-treated mice that received cells from patients with different outcomes indicated significant differences in time to events between mice inoculated with samples from CR1 and late relapse patients at TT1% and TT25%, and between mice inoculated with samples from CR1 and early relapse at TT25% and TTL. The patterns of median mouse EFS at TT25% and TTL were translated into strategies for prediction of IR ALL patient response to the treatment although the most accurate classifiers for identifying patient outcomes were dependent on monitoring of engraftment for at least 235 days.

The data from this study provide the first step toward the clinical application of ALL xenografts for IR ALL patients. Perhaps the predictive power of the median TT25% and

TTL could be verified in the context of an ongoing clinical trial, which uses a treatment protocol for IR ALL patients similar to that used in the ANZCHOG Study VIII clinical trial. Ideally, a small study could be planned to establish xenografts from fresh diagnostic samples of IR ALL patients and the reliability of this approach could be evaluated to inform the clinical outcome of IR ALL patients. However, given that better stratification ability of ALL xenografts according to patient outcomes was observed based on a 235-day classifier, future work should focus on improving the method of defining the engraftment kinetics of IR ALL patient samples established in NSG mice treated with VXL chemotherapy for prediction of patient outcomes in an efficient time to allow utilising of this model in a reasonable time, which allows for clinical decisions to be made.

302 Recent unpublished data from our lab showed that a bioluminescence imaging (BLI) system provided a more efficient method for monitoring the engraftment kinetics and response of ALL xenografts to VXL treatment over the standard PB monitoring strategy. Based on this strategy, engraftment of ALL xenografts was detected much earlier than the detection of 1% human leukaemia cells in murine PB. However, this study expressed some concerns about the application of BLI in preclinical models due to possible instability of the green fluorescence protein (GFP)/luciferase reporters in some ALL xenografts and the necessity for cell sorting and expansion before achieving measurable signals of cells in mice. Therefore, this approach could not be used for measuring the difference in engraftment kinetics of ALL patients for prediction of outcome. Perhaps, other bioimaging approaches such as Positron Emission

Tomography (PET) could be considered in the future especially if there is an enhancement to the specificity of probes to detect dispersed human leukaemia cells in the mouse tissues.

In Chapter 5, the relevance of IR ALL xenografts to ALL disease in patients was tested in the context of biological and functional features of the disease. In this chapter I characterised the heterogeneity in transcriptional profiles of ALL cells within an individual ALL xenograft and its corresponding primary ALL sample and assessed the diversity in response to agents involved in the induction treatment protocol for ALL patients between ALL xenografts.

Transplantation of ALL patient samples into suitable niches for establishing disease in mouse models might be affected by cell homing, murine immune system and supportive microenvironment niches which could provide a competitive advantage for engraftment of ALL clonal populations. In an effort to explain whether ALL cells established using 303 various mouse strains and different routes of inoculation could provide an approach for capturing the heterogeneity between clonal populations of ALL disease, the transcriptional profiles of ALL-67 were compared within an individual ALL xenograft established using various engraftment conditions and also compared between ALL-67 and its corresponding ALL patient samples. The analysis revealed consistent patterns of gene expression between cells from ALL-67 established using various engraftment conditions suggesting similarities in the dominant xenograft cells infiltrating mouse spleens. Indeed, investigation of heterogeneity in this study could be limited by the microarray approach, as such a strategy did not account for clonal populations with low frequency. In addition, it would be ideal if the analysis was also performed on samples from the BM to check for BM specific engraftment mediated by inoculation of samples via the IF route over the IV route.

However, the pattern of gene expression of ALL-67 cells was significantly different from that observed in the corresponding biopsy samples from which ALL-67 was derived. ALL-67 enriched for gene expression profiles suggest high proliferative activity relative to samples of its corresponding ALL patient. This observation was in agreement with findings from a study reported by Clappier et al. (2011). In this study, the authors explored the changes in transcriptional profiles of the xenograft cells relative to their corresponding diagnosis cells of T-ALL patients and identified up-regulation of genes that differentiated ALL xenografts from their corresponding diagnosis leukaemias. Similar to the observation from my study, many of the upregulated genes in

T-ALL xenografts were significantly enriched in cell cycle and mitosis processes. The authors suggested that an increased proliferation activity in ALL xenografts might contributed to the selective advantage of the engrafted cells (Clappier et al., 2011).

304

In attempts to assess the fidelity of ALL xenograft cells to reflect the response of ALL patients to their chemotherapy treatment, the ex vivo sensitivity of ALL xenografts to single agent induction chemotherapy was compared between paired IR ALL patients who presented with similar clinical features at the time of diagnosis but responded differentially to chemotherapy in the clinic. The result of this experiment showed variable responses to vincristine and L-asparaginase but not to dexamethasone between

ALL xenografts of paired IR ALL patients, which could explain the difference in in vivo response to VXL chemotherapy between paired ALL xenografts and clinical outcomes of patients from whom the xenografts were derived. While investigation into the biological determinants of heterogeneity of ALL xenografts in response of IR ALL xenografts to drugs was not followed in this study, it would be of interest to validate the difference in response of paired IR ALL xenografts to vincristine and L-asparaginase in vivo and study the mechanisms underlying the difference in response between patients with similar clinical and cytogenetic features.

While the reasons for drug failure in the clinic are likely to be multifactorial, evidence from the literature and clinical studies indicate that drug-resistant phenotypes developing after chemotherapy treatment represent growth of leukaemia clones through treatment with chemotherapy (Choi et al., 2007; Jan and Majeti, 2013; Kunz et al.,

2015). Experimentally derived drug resistance might provide an opportunity to investigate clinically relevant mechanisms of chemotherapy-induced resistance.

Examining the propensity of ALL xenografts to exhibit ex vivo changes in sensitivity to induction chemotherapy treatment after in vivo treatment with VXL chemotherapy revealed a significant decrease in sensitivity of ALL-67 to vincristine. CMap analysis 305 was performed to determine drug leads for reversing the resistance to vincristine based on the most upregulated gene signature. CMap analysis identified HSP90 inhibitors as candidates for overcoming drug resistance. Assessing the ex vivo sensitivity of the ALL-

67 xenograft to the HSP90 inhibitor 17 DMAG revealed efficient antitumour activity as a single agent but no enhancement of the cytotoxic effects of vincristine was achieved.

Furthermore, the pattern of change in sensitivity in the VXL-treated cells concurred with consistent overexpression of genes involved in microtubule cytoskeleton system, including alpha and beta tubulin isotypes and microtubule-associated proteins. These findings could lead to the proposal that overexpression of these genes may have an implication for vincristine resistance and further investigations should be direct into assessing the link between these genes and the observed resistance to vincristine.

Furthermore, the pattern of change in sensitivity in the VXL-treated cells concurred with consistent overexpression of genes involved in microtubule cytoskeleton system, including alpha and beta tubulin isotypes and microtubule-associated proteins. These findings could lead to the proposal that overexpression of these genes may have an implication for vincristine resistance and further investigations should be direct into assessing the link between these genes and the observed resistance to vincristine.

In summary, the work presented in this thesis demonstrated a novel strategy for prediction of relapse in IR ALL patients. I established the most suitable engraftment condition for stratifying IR ALL xenografts according to patient outcomes. Validation of the optimised engraftment condition to identify relapse in IR ALL patients using a larger cohort of IR ALL patients emphasised the role of VXL treatment to discriminate between relapsed and non-relapsed IR ALL patients. In addition, assessment of the ex

306 vivo sensitivity of ALL xenografts to induction chemotherapy showed promise for reproducing the response of IR ALL patients to chemotherapy and for understanding the development of drug resistance in childhood IR ALL patients. Treatment of immunodeficient mice with an induction-type VXL regimen following inoculation of diagnosis bone marrow biopsy cells may lead to improved relapse prediction in IR paediatric ALL which could allow for tailoring the intensity of treatments according to the risk of relapse and guide the development of novel chemotherapy regimens.

307 REFERENCES

Agliano, A., Martin-Padura, I., Mancuso, P., Marighetti, P., Rabascio, C., Pruneri, G., Shultz, L.D., and Bertolini, F. (2008). Human acute leukemia cells injected in NOD/LTSz-scid/IL-2Rγnull mice generate a faster and more efficient disease compared to other NOD/SCID-related strains. International Journal of Cancer 123, 2222-2227.

Ahmed, S.A., Gogal, R.M., Jr., and Walsh, J.E. (1994). A new rapid and simple non- radioactive assay to monitor and determine the proliferation of lymphocytes: An alternative to [3h]thymidine incorporation assay. Journal of immunological methods 170, 211-224.

Alvarnas, J.C., Brown, P.A., Aoun, P., Ballen, K.K., Bellam, N., Blum, W., Boyer, M.W., Carraway, H.E., Coccia, P.F., Coutre, S.E., et al. (2012). Acute lymphoblastic leukemia. Journal of the National Comprehensive Cancer Network 10, 858-914.

Anderson, K., Lutz, C., van Delft, F.W., Bateman, C.M., Guo, Y., Colman, S.M., Kempski, H., Moorman, A.V., Titley, I., Swansbury, J., et al. (2011). Genetic variegation of clonal architecture and propagating cells in leukaemia. Nature 469, 356- 361.

Appel, I.M., van Kessel-Bakvis, C., Stigter, R., and Pieters, R. (2007). Influence of two different regimens of concomitant treatment with asparaginase and dexamethasone on hemostasis in childhood acute lymphoblastic leukemia. Leukemia 21, 2377-2380.

Arai, K., Matsumoto, Y., Nagashima, Y., and Yagasaki, K. (2006). Regulation of class II β-tubulin expression by tumor suppressor p53 protein in mouse melanoma cells in response to vinca alkaloid. Molecular Cancer Research 4, 247-255.

Arico, M., Baruchel, A., Bertrand, Y., Biondi, A., Conter, V., Eden, T., Gadner, H., Gaynon, P., Horibe, K., Hunger, S.P., et al. (2005). The seventh international childhood acute lymphoblastic leukemia workshop report: Palermo, italy, january 29-30, 2005. Leukemia 19, 1145-1152.

Armstrong, S.A., Kung, A.L., Mabon, M.E., Silverman, L.B., Stam, R.W., Den Boer, M.L., Pieters, R., Kersey, J.H., Sallan, S.E., Fletcher, J.A., et al. (2003). Inhibition of

308 FLT3 in MLL: Validation of a therapeutic target identified by gene expression based classification. Cancer Cell 3, 173-183.

Aslanian, A.M., and Kilberg, M.S. (2001). Multiple adaptive mechanisms affect asparagine synthetase substrate availability in asparaginase-resistant MOLT-4 human leukaemia cells. Biochemical Journal 358, 59-67.

Assumpção, J.G., Paula, F.D.F., Xavier, S.G., Murao, M., de Aguirre, J.C., Dutra, Á.P., Lima, E.R., de Oliveira, B.M., and Viana, M.B. (2013). Gene rearrangement study for minimal residual disease monitoring in children with acute lymphocytic leukemia. Revista Brasileira de Hematologiae Hemoterapia 35, 337-342.

Avramis, V.I. (2012). Asparaginases: Biochemical pharmacology and modes of drug resistance. Anticancer Research 32, 2423-2437.

Avramis, V.I., Sencer, S., Periclou, A.P., Sather, H., Bostrom, B.C., Cohen, L.J., Ettinger, A.G., Ettinger, L.J., Franklin, J., Gaynon, P.S., et al. (2002). A randomized comparison of native escherichia coli asparaginase and polyethylene glycol conjugated asparaginase for treatment of children with newly diagnosed standard-risk acute lymphoblastic leukemia: A children's cancer group study. Blood 99, 1986-1994.

Bachmann, P. S., Gorman, R., MacKenzie, K. L., Lutze-Mann, L and Lock, R. B. (2005). "Dexamethasone resistance in B-cell precursor childhood acute lymphoblastic leukemia occurs downstream of ligand-induced nuclear translocation of the glucocorticoid receptor." Blood 105, 2519-2526.

Bachmann, P.S., Gorman, R., Papa, R.A., Bardell, J.E., Ford, J., Kees, U.R., Marshall, G.M., and Lock, R.B. (2007). Divergent mechanisms of glucocorticoid resistance in experimental models of pediatric acute lymphoblastic leukemia. Cancer Research 67, 4482-4490.

Bain, B.J. (2007). Diagnosis and classification of acute leukaemia. In Postgraduate haematology (Blackwell Publishing Ltd), pp. 476-491.

Bassan, R., and Hoelzer, D. (2011). Modern therapy of acute lymphoblastic leukemia. Journal of Clinical Oncology 29, 532-543.

309 Bateman, C.M., Colman, S.M., Chaplin, T., Young, B.D., Eden, T.O., Bhakta, M., Gratias, E.J., van Wering, E.R., Cazzaniga, G., Harrison, C.J., et al. (2010). Acquisition of genome-wide copy number alterations in monozygotic twins with acute lymphoblastic leukemia. Blood 115, 3553-3558.

Behm, F.G. (2003). Classification of acute leukemias. In Treatment of acute leukemias: New directions for clinical research, C.-H. Pui, ed. (Totowa, NJ: Humana Press), pp. 43-58.

Benjamini, Y., and Hochberg, Y. (1995). Controlling the false discovery rate: A practical and powerful approach to multiple testing. Journal of the Royal Statistical Society Series B (Methodological) 57, 289-300.

Bennett, J.M., Catovsky, D., Daniel, M.-T., Flandrin, G., Galton, D.A.G., Gralnick, H.R., and Sultan, C. (1976). Proposals for the classification of the acute leukaemias French-American-British (FAB) co-operative group. British Journal of Haematology 33, 451-458.

Berger, N.A. (1986). Cancer chemotherapy: New strategies for success. The Journal of Clinical Investigation 78, 1131-1135.

Bernt, K.M., and Hunger, S.P. (2014). Current concepts in pediatric philadelphia chromosome-positive acute lymphoblastic leukemia. Frontiers in oncology 4, 54.

Bhat, K.M.R., and Setaluri, V. (2007). Microtubule-associated proteins as targets in cancer chemotherapy. Clinical Cancer Research 13, 2849-2854.

Bhojwani, D., Kang, H., Moskowitz, N.P., Min, D.-J., Lee, H., Potter, J.W., Davidson, G., Willman, C.L., Borowitz, M.J., Belitskaya-Levy, I., et al. (2006). Biologic pathways associated with relapse in childhood acute lymphoblastic leukemia: A children's oncology group study. Blood 108, 711-717.

Biomarkers Definitions Working, G. (2001). Biomarkers and surrogate endpoints: Preferred definitions and conceptual framework. Clinical Pharmacology & Therapeutics 69, 89-95.

310 Biondi, A., and Cazzaniga, G. (2013). Novel clinical trials for pediatric leukemias: Lessons learned from genomic analyses. ASH Education Program Book 2013, 612-619.

Biondi, A., Valsecchi, M.G., Seriu, T., D'Aniello, E., Willemse, M.J., Fasching, K., Pannunzio, A., Gadner, H., Schrappe, M., Kamps, W.A., et al. (2000). Molecular detection of minimal residual disease is a strong predictive factor of relapse in childhood B-lineage acute lymphoblastic leukemia with medium risk features. A case control study of the international bfm study group. Leukemia 14, 1939-1943.

Bjorklund, E., Mazur, J., Soderhll, S., and Porwit-MacDonald, A. (2003). Flow cytometric follow-up of minimal residual disease in bone marrow gives prognostic information in children with acute lymphoblastic leukemia. Leukemia 17, 138-148.

Bomken, S., Fiser, K., Heidenreich, O., and Vormoor, J. (2010). Understanding the cancer stem cell. British Journal of Cancer 103, 439-445.

Borgmann, A., von Stackelberg, A., Hartmann, R., Ebell, W., Klingebiel, T., Peters, C., and Henze, G. (2003). Unrelated donor stem cell transplantation compared with chemotherapy for children with acute lymphoblastic leukemia in a second remission: A matched-pair analysis. Blood 101, 3835-3839.

Borowitz, M.J., Devidas, M., Hunger, S.P., Bowman, W.P., Carroll, A.J., Carroll, W.L., Linda, S., Martin, P.L., Pullen, D.J., Viswanatha, D., et al. (2008). Clinical significance of minimal residual disease in childhood acute lymphoblastic leukemia and its relationship to other prognostic factors: A children's oncology group study. Blood 111, 5477-5485.

Bostrom, B.C., Sensel, M.R., Sather, H.N., Gaynon, P.S., La, M.K., Johnston, K., Erdmann, G.R., Gold, S., Heerema, N.A., Hutchinson, R.J., et al. (2003). Dexamethasone versus prednisone and daily oral versus weekly intravenous mercaptopurine for patients with standard-risk acute lymphoblastic leukemia: A report from the children's cancer group. Blood 101, 3809-3817.

Bruggemann, M., Schrauder, A., Raff, T., Pfeifer, H., Dworzak, M., Ottmann, O.G., Asnafi, V., Baruchel, A., Bassan, R., Benoit, Y., et al. (2010). Standardized mrd quantification in european all trials: Proceedings of the second international symposium on MRD assessment in Kiel, Germany, 18-20 september 2008. Leukemia 24, 521-535. 311 Burchenal, J.H., Murphy, M.L., Ellison, R.R., Sykes, M.P., Tan, T.C., Leone, L.A., Karnofsky, D.A., Craver, L.F., Dargeon, H.W., and Rhoads, C.P. (1953). Clinical evaluation of a new antimetabolite, 6-mercaptopurine, in the treatment of leukemia and allied diseases. Blood 8, 965-999.

Cameron, D.G., Townsend, S.R., Mills, E.S., and Mathews, W.H. (1951). ACTH and cortisone in leukæmia. Canadian Medical Association Journal 65, 552-555.

Campana, D., and Coustan-Smith, E. (1999). Detection of minimal residual disease in acute leukemia by flow cytometry. Cytometry 38, 139-152.

Campana, D., and Pui, C.H. (1995). Detection of minimal residual disease in acute leukemia: Methodologic advances and clinical significance. Blood 85, 1416-1434.

Campo, E., Swerdlow, S.H., Harris, N.L., Pileri, S., Stein, H., and Jaffe, E.S. (2011). The 2008 WHO classification of lymphoid neoplasms and beyond: Evolving concepts and practical applications. Blood 117, 5019-5032.

Castor, A., Nilsson, L., Astrand-Grundstrom, I., Buitenhuis, M., Ramirez, C., Anderson, K., Strombeck, B., Garwicz, S., Bekassy, A.N., Schmiegelow, K., et al. (2005). Distinct patterns of hematopoietic stem cell involvement in acute lymphoblastic leukemia. Nature Medicine 11, 630-637.

Chabner, B.A., and Roberts, T.G. (2005). Chemotherapy and the war on cancer. Nature Reviews Cancer 5, 65-72.

Chessells, J.M., Richards, S.M., Bailey, C.C., Lilleyman, J.S., and Eden, O.B. (1995). Gender and treatment outcome in childhood lymphoblastic leukaemia: Report from the mrc UKALL trials*. British Journal of Haematology 89, 364-372.

Chessells, J.M., Veys, P., Kempski, H., Henley, P., Leiper, A., Webb, D., and Hann, I.M. (2003). Long-term follow-up of relapsed childhood acute lymphoblastic leukaemia. British Journal of Haematology 123, 396-405.

Cheung, C.H.A., Wu, S.-Y., Lee, T.-R., Chang, C.-Y., Wu, J.-S., Hsieh, H.-P., and Chang, J.-Y. (2010). Cancer cells acquire mitotic drug resistance properties through

312 Beta I-tubulin mutations and alterations in the expression of Beta-tubulin isotypes. PLoS ONE 5, e12564.

Chiaretti, S., Zini, G., and Bassan, R. (2014). Diagnosis and subclassification of acute lymphoblastic leukemia. Mediterranean Journal of Hematology and Infectious Diseases 6, e2014073.

Choi, S., Henderson, M.J., Kwan, E., Beesley, A.H., Sutton, R., Bahar, A.Y., Giles, J., Venn, N.C., Pozza, L.D., Baker, D.L., et al. (2007). Relapse in children with acute lymphoblastic leukemia involving selection of a preexisting drug-resistant subclone. Blood 110, 632-639.

Christoph, S., Schlegel, J., Alvarez-Calderon, F., Kim, Y.M., Brandao, L.N., DeRyckere, D., and Graham, D.K. (2013). Bioluminescence imaging of leukemia cell lines in vitro and in mouse xenografts: Effects of monoclonal and polyclonal cell populations on intensity and kinetics of photon emission. Journal of Hematology & Oncology 6, 10.

Clappier, E., Gerby, B., Sigaux, F., Delord, M., Touzri, F., Hernandez, L., Ballerini, P., Baruchel, A., Pflumio, F., and Soulier, J. (2011). Clonal selection in xenografted human T cell acute lymphoblastic leukemia recapitulates gain of malignancy at relapse. Journal of Experimental Medicine 208, 653-661.

Cleaver, A.L., Beesley, A.H., Firth, M.J., Sturges, N.C., O'Leary, R.A., Hunger, S.P., Baker, D.L., and Kees, U.R. (2010). Gene-based outcome prediction in multiple cohorts of pediatric T-cell acute lymphoblastic leukemia: A children's oncology group study. Molecular Cancer 9, 1-12.

Cobaleda, C., Gutierrez-Cianca, N., Perez-Losada, J., Flores, T., Garcia-Sanz, R., Gonzalez, M., and Sanchez-Garcia, I. (2000). A primitive hematopoietic cell is the target for the leukemic transformation in human philadelphia-positive acute lymphoblastic leukemia. Blood 95, 1007-1013.

313 Colmone A., Amorim M., Pontier AL., Wang S., Jablonski E., Sipkins DA. (2008). Leukemic cells create bone marrow niches that disrupt the behavior of normal hematopoietic progenitor cells. Science. 322, 1861-1865.

Conter, V., Bartram, C.R., Valsecchi, M.G., Schrauder, A., Panzer-Grumayer, R., Moricke, A., Arico, M., Zimmermann, M., Mann, G., De Rossi, G., et al. (2010). Molecular response to treatment redefines all prognostic factors in children and adolescents with B-cell precursor acute lymphoblastic leukemia: Results in 3184 patients of the AIEOP-BFM ALL 2000 study. Blood. 115, 3206-3214.

Conter, V., Valsecchi, M.G., Silvestri, D., Campbell, M., Dibar, E., Magyarosy, E., Gadner, H., Stary, J., Benoit, Y., Zimmermann, M., et al. (2007). Pulses of vincristine and dexamethasone in addition to intensive chemotherapy for children with intermediate-risk acute lymphoblastic leukaemia: A multicentre randomised trial. The Lancet 369, 123-131.

Cox, C.V., Diamanti, P., Evely, R.S., Kearns, P.R., and Blair, A. (2009). Expression of CD133 on leukemia-initiating cells in childhood ALL. Blood 113, 3287-3296.

Cox, C.V., Evely, R.S., Oakhill, A., Pamphilon, D.H., Goulden, N.J., and Blair, A. (2004). Characterization of acute lymphoblastic leukemia progenitor cells. Blood 104, 2919-2925.

Curran, E., and Stock, W. (2015). How I treat acute lymphoblastic leukemia in older adolescents and young adults. Blood 125, 3702-3710.

Den Boer, M.L., Harms, D.O., Pieters, R., Kazemier, K.M., Göbel, U., Körholz, D., Graubner, U., Haas, R.J., Jorch, N., Spaar, H.J., et al. (2003). Patient stratification based on prednisolone-vincristine-asparaginase resistance profiles in children with acute lymphoblastic leukemia. Journal of Clinical Oncology 21, 3262-3268.

Den Boer, M.L., van Slegtenhorst, M., De Menezes, R.X., Cheok, M.H., Buijs- Gladdines, J.G., Peters, S.T., Van Zutven, L.J.C.M., Beverloo, H.B., Van der Spek, P.J., Escherich, G., et al. (2009). A subtype of childhood acute lymphoblastic leukaemia with 314 poor treatment outcome: A genome-wide classification study. The Lancet Oncology 10, 125-134.

Diamanti, P., Cox, C.V., and Blair, A. (2012). Comparison of childhood leukemia initiating cell populations in NOD/SCID and NSG mice. Leukemia 26, 376-380.

Dick, J.E. (2008). Stem cell concepts renew cancer research. Blood 112, 4793-4807.

Dorantes-Acosta, E., and Pelayo, R. (2012). Lineage switching in acute leukemias: A consequence of stem cell plasticity? Bone Marrow Research 2012, 406796.

Doulatov, S., Notta, F., Laurenti, E., and Dick, J.E. (2012). Hematopoiesis: A human perspective. Cell Stem Cell 10, 120-136.

Drexler, H.G., and MacLeod, R.A.F. (2003). Leukemia-lymphoma cell lines as model systems for hematopoietic research. Annuals of Medicine 35, 404-412.

Drexler, H.G., Matsuo, Y., and MacLeod, R.A.F. (2000). Continuous hematopoietic cell lines as model systems for leukemia–lymphoma research. Leukemia Research 24, 881- 911.

Drukman, S., and Kavallaris, M. (2002). Microtubule alterations and resistance to tubulin-binding agents (review). International Journal of Oncology 21, 621-628.

Dumontet, C., and Jordan, M.A. (2010). Microtubule-binding agents: A dynamic field of cancer therapeutics. Nature Reviews Drug Discovery 9, 790-803.

Dumontet, C., and Sikic, B.I. (1999). Mechanisms of action of and resistance to antitubulin agents: Microtubule dynamics, drug transport, and cell death. Journal of Clinical Oncology 17, 1061.

Dworzak, M.N., Schumich, A., Printz, D., Pötschger, U., Husak, Z., Attarbaschi, A., Basso, G., Gaipa, G., Ratei, R., Mann, G., et al. (2008). CD20 up-regulation in pediatric

315 B-cell precursor acute lymphoblastic leukemia during induction treatment: Setting the stage for anti-CD20 directed immunotherapy. Blood 112, 3982-3988.

Eberhart, K., Renner, K., Ritter, I., Kastenberger, M., Singer, K., Hellerbrand, C., Kreutz, M., Kofler, R., and Oefner, P.J. (2009). Low doses of 2-deoxy-glucose sensitize acute lymphoblastic leukemia cells to glucocorticoid-induced apoptosis. Leukemia 23, 2167-2170.

Eguiguren, J.M., Schell, M.J., Crist, W.M., Kunkel, K., and Rivera, G.K. (1992). Complications and outcome in childhood acute lymphoblastic leukemia with hyperleukocytosis. Blood 79, 871-875.

Einsiedel, H.G., von Stackelberg, A., Hartmann, R., Fengler, R., Schrappe, M., Janka- Schaub, G., Mann, G., Hählen, K., Göbel, U., Klingebiel, T., et al. (2005). Long-term outcome in children with relapsed ALL by risk-stratified salvage therapy: Results of trial acute lymphoblastic leukemia-relapse study of the Berlin-Frankfurt-Münster group 87. Journal of Clinical Oncology 23, 7942-7950.

Estlin, E.J., Ronghe, M., Burke, G.A.A., and Yule, S.M. (2000). The clinical and cellular pharmacology of vincristine, corticosteroids, l-asparaginase, anthracyclines and cyclophosphamide in relation to childhood acute lymphoblastic leukaemia. British Journal of Haematology 110, 780-790.

Faham, M., Zheng, J., Moorhead, M., Carlton, V.E., Stow, P., Coustan-Smith, E., Pui, C.H., and Campana, D. (2012). Deep-sequencing approach for minimal residual disease detection in acute lymphoblastic leukemia. Blood 120, 5173-5180.

Farber, S., Diamond, L.K., Mercer, R.D., Sylvester, R.F., and Wolff, J.A. (1948). Temporary remissions in acute leukemia in children produced by folic acid antagonist, 4-aminopteroyl-glutamic acid (aminopterin). New England Journal of Medicine 238, 787-793.

Fazzina, R., Lombardini, L., Mezzanotte, L., Roda, A., Hrelia, P., Pession, A., and Tonelli, R. (2012). Generation and characterization of bioluminescent xenograft mouse 316 models of MLL-related acute leukemias and in vivo evaluation of luciferase-targeting sirna nanoparticles. International Journal of Oncology 41, 621-628.

Felsher, D.W., and Bishop, J.M. (1999). Reversible tumorigenesis by MYC in hematopoietic lineages. Molecular Cell 4, 199-207.

Fleury, I., Primeau, A., Doreau, I., Costea, A., Moghrabi, D., Sinnett and Krajinovic, M. (2004). "Polymorphisms in genes involved in the corticosteroid response and the outcome of childhood acute lymphoblastic leukemia." Am J Pharmacogenomics 4, 331- 341.

Frei, E., 3rd, Karon, M., Levin, R.H., Freireich, E.J., Taylor, R.J., Hananian, J., Selawry, O., Holland, J.F., Hoogstraten, B., Wolman, I.J., et al. (1965). The effectiveness of combinations of antileukemic agents in inducing and maintaining remission in children with acute leukemia. Blood 26, 642-656.

Freireich, E.J., Wiernik, P.H., and Steensma, D.P. (2014). The leukemias: A half- century of discovery. Journal of Clinical Oncology 32, 3463-3469.

Friedmann, A.M., and Weinstein, H.J. (2000). The role of prognostic features in the treatment of childhood acute lymphoblastic leukemia. The Oncologist 5, 321-328.

Frost, B.-M., Nygren, P., Gustafsson, G., Forestier, E., Jonsson, O.G., Kanerva, J., Nygaard, R., Schmiegelow, K., Larsson, R., Lönnerholm, G., et al. (2003). Increased in vitro cellular drug resistance is related to poor outcome in high-risk childhood acute lymphoblastic leukaemia. British Journal of Haematology 122, 376-385.

Gaipa, G., Basso, G., Aliprandi, S., Migliavacca, M., Vallinoto, C., Maglia, O., Faini, A., Veltroni, M., Husak, D., Schumich, A., et al. (2008). Prednisone induces immunophenotypic modulation of CD10 and CD34 in nonapoptotic B-cell precursor acute lymphoblastic leukemia cells. Cytometry Part B: Clinical Cytometry 74B, 150- 155.

317 Gajjar, A., Ribeiro, R., Hancock, M.L., Rivera, G.K., Mahmoud, H., Sandlund, J.T., Crist, W.M., and Pui, C.H. (1995). Persistence of circulating blasts after 1 week of multiagent chemotherapy confers a poor prognosis in childhood acute lymphoblastic leukemia. Blood 86, 1292-1295.

Gandemer, V., Chevret, S., Petit, A., Vermylen, C., Leblanc, T., Michel, G., Schmitt, C., Lejars, O., Schneider, P., Demeocq, F., et al. (2012). Excellent prognosis of late relapses of ETV6/RUNX1-positive childhood acute lymphoblastic leukemia: Lessons from the FRALLE 93 protocol. Haematologica 97, 1743-1750.

Gawad, C., Koh, W., and Quake, S.R. (2014). Dissecting the clonal origins of childhood acute lymphoblastic leukemia by single-cell genomics. Proceedings of the National Academy of Sciences of the United States of America 111, 17947-17952.

Gaynon, P.S., Desai, A.A., Bostrom, B.C., Hutchinson, R.J., Lange, B.J., Nachman, J.B., Reaman, G.H., Sather, H.N., Steinherz, P.G., Trigg, M.E., et al. (1997). Early response to therapy and outcome in childhood acute lymphoblastic leukemia. Cancer 80, 1717-1726.

Gaynon, P.S., Qu, R.P., Chappell, R.J., Willoughby, M.L., Tubergen, D.G., Steinherz, P.G., and Trigg, M.E. (1998). Survival after relapse in childhood acute lymphoblastic leukemia: Impact of site and time to first relapse-the children's cancer group experience. Cancer 82, 1387-1395.

Gorlick, R., Goker, E., Trippett, T., Waltham, M., Banerjee, D., and Bertino, J.R. (1996). Intrinsic and acquired resistance to methotrexate in acute leukemia. New England Journal of Medicine 335, 1041-1048.

Greaves, M.F. (1997). Aetiology of acute leukaemia. The Lancet 349, 344-349. Greiner, D.L., Hesselton, R.A., and Shultz, L.D. (1998). Scid mouse models of human stem cell engraftment. Stem Cells 16, 166-177.

Greystoke, B.F., Huang, X., Wilks, D.P., Atkinson, S., and Somervaille, T.C.P. (2013). Very high frequencies of leukaemia-initiating cells in precursor T-acute lymphoblastic 318 leukaemia may be obscured by cryopreservation. British Journal of Haematology 163, 538-541.

Groninger, E., Meeuwsen-de Boer, T., Koopmans, P., Uges, D., Sluiter, W., Veerman, A., Kamps, W., and de Graaf, S. (2002). Pharmacokinetics of vincristine monotherapy in childhood acute lymphoblastic leukemia. Pediatric Research 52, 113-118.

Grove, C.S., and Vassiliou, G.S. (2014). Acute myeloid leukaemia: A paradigm for the clonal evolution of cancer? Disease Models & Mechanisms 7, 941-951.

Harris, Jaffe, Diebold, Flandrin, Muller, H., Vardiman, Lister, and Bloomfield (2000). The world health organization classification of neoplastic diseases of the haematopoietic and lymphoid tissues: Report of the clinical advisory committee meeting, Airlie house, Virginia, November 1997. Histopathology 36, 69-86.

Harrison, C.J. (2013). Targeting signaling pathways in acute lymphoblastic leukemia: New insights. ASH Education Program Book 2013, 118-125.

Heerema, N.A., Sather, H.N., Sensel, M.G., Zhang, T., Hutchinson, R.J., Nachman, J.B., Lange, B.J., Steinherz, P.G., Bostrom, B.C., Reaman, G.H., et al. (2000). Prognostic impact of trisomies of chromosomes 10, 17, and 5 among children with acute lymphoblastic leukemia and high hyperdiploidy (> 50 chromosomes). Journal of Clinical Oncology 18, 1876-1887.

Heidenreich, O., and Vormoor, J. (2009). Malignant stem cells in childhood all: The debate continues! Blood 113, 4476-4477.

Henderson, E.S., and Samaha, R.J. (1969). Evidence that drugs in multiple combinations have materially advanced the treatment of human malignancies. Cancer Research 29, 2272-2280.

Hodges, L.M., Markova, S.M., Chinn, L.W., Gow, J.M., Kroetz, D.L., Klein, T.E., and Altman, R.B. (2011). Very important pharmacogene summary: ABCB1 (MDR1, p- glycoprotein). Pharmacogenetics and Genomics 21, 152-161. 319

Hogan, L.E., Meyer, J.A., Yang, J., Wang, J., Wong, N., Yang, W., Condos, G., Hunger, S.P., Raetz, E., Saffery, R., et al. (2011). Integrated genomic analysis of relapsed childhood acute lymphoblastic leukemia reveals therapeutic strategies. Blood 118, 5218-5226.

Holleman, A., Cheok, M.H., den Boer, M.L., Yang, W., Veerman, A.J.P., Kazemier, K.M., Pei, D., Cheng, C., Pui, C.-H., Relling, M.V., et al. (2004). Gene-expression patterns in drug-resistant acute lymphoblastic leukemia cells and response to treatment. New England Journal of Medicine 351, 533-542.

Holmes, L., Hossain, J., desVignes-Kendrick, M., and Opara, F. (2012). Sex variability in pediatric leukemia survival: Large cohort evidence. International Scholarly Research Notices Oncology 2012, 439070.

Hong, D., Gupta, R., Ancliff, P., Atzberger, A., Brown, J., Soneji, S., Green, J., Colman, S., Piacibello, W., Buckle, V., et al. (2008). Initiating and cancer-propagating cells in TEL-AML1-associated childhood leukemia. Science 319, 336-339.

Hongo, T., Yajima, S., Sakurai, M., Horikoshi, Y., and Hanada, R. (1997). In vitro drug sensitivity testing can predict induction failure and early relapse of childhood acute lymphoblastic leukemia. Blood 89, 2959-2965.

Houghton, P.J., Morton, C.L., Tucker, C., Payne, D., Favours, E., Cole, C., Gorlick, R., Kolb, E.A., Zhang, W., Lock, R., et al. (2007). The pediatric preclinical testing program: Description of models and early testing results. Pediatric Blood Cancer 49, 928-940.

Hudson, M.M., Link, M.P., and Simone, J.V. (2014). Milestones in the curability of pediatric cancers. Journal of Clinical Oncology 32, 2391-2397.

Hulleman, E., Kazemier, K.M., Holleman, A., VanderWeele, D.J., Rudin, C.M., Broekhuis, M.J., Evans, W.E., Pieters, R., and Den Boer, M.L. (2009). Inhibition of

320 glycolysis modulates prednisolone resistance in acute lymphoblastic leukemia cells. Blood 113, 2014-2021.

Huntly, B.J.P., and Gilliland, D.G. (2005). Leukaemia stem cells and the evolution of cancer-stem-cell research. Nature Reviews Cancer 5, 311-321.

Inaba, H., and Pui, C.-H. (2010). Glucocorticoid use in acute lymphoblastic leukemia: Comparison of prednisone and dexamethasone. The Lancet Oncology 11, 1096-1106.

Iwamoto, S., Mihara, K., Downing, J.R., Pui, C.H., and Campana, D. (2007). Mesenchymal cells regulate the response of acute lymphoblastic leukemia cells to asparaginase. Journal of Clinical Investigations 117, 1049-1057.

Jacoby, E., Chien, C.D., and Fry, T.J. (2014). Murine models of acute leukemia: Important tools in current pediatric leukemia research. Frontiers in Oncology 4, 95. Jan, M., and Majeti, R. (2013). Clonal evolution of acute leukemia genomes. Oncogene 32, 135-140.

Jeha, S., Pei, D., Raimondi, S.C., Onciu, M., Campana, D., Cheng, C., Sandlund, J.T., Ribeiro, R.C., Rubnitz, J.E., Howard, S.C., et al. (2009). Increased risk for CNS relapse in pre-B cell leukemia with the t(1;19)/TCF3-PBX1. Leukemia 23, 1406-1409.

Jeha, S., and Pui, C.-H. (2009). Risk-adapted treatment of pediatric acute lymphoblastic leukemia. Hematology/Oncology Clinics of North America 23, 973-990.

Karsa, M., Dalla Pozza, L., Venn, N.C., Law, T., Shi, R., Giles, J.E., Bahar, A.Y., Cross, S., Catchpoole, D., Haber, M., et al. (2013). Improving the identification of high risk precursor B acute lymphoblastic leukemia patients with earlier quantification of minimal residual disease. PLoS ONE 8, e76455.

Kaspers, G.J.L., Veerman, A.J.P., Pieters, R., Van Zantwijk, C.H., Smets, L.A., Van Wering, E.R., and Van Der Does-Van Den Berg, A. (1997). In vitro cellular drug resistance and prognosis in newly diagnosed childhood acute lymphoblastic leukemia. Blood 90, 2723-2729. 321

Kaspers, G.J.L., Wijnands, J.J.M., Hartmann, R., Huismans, L., Loonen, A.H., Stackelberg, A., Henze, G., Pieters, R., Hählen, K., Van Wering, E.R., et al. (2005). Immunophenotypic cell lineage and in vitro cellular drug resistance in childhood relapsed acute lymphoblastic leukaemia. European Journal of Cancer 41, 1300-1303.

Kavallaris, M. (2010). Microtubules and resistance to tubulin-binding agents. Nature Reviews Cancer 10, 194-204.

Kawamata, N., Ogawa, S., Seeger, K., Kirschner-Schwabe, R., Huynh, T., Chen, J., Megrabian, N., Harbott, J., Zimmermann, M., Henze, G., et al. (2009). Molecular allelokaryotyping of relapsed pediatric acute lymphoblastic leukemia. International Journal of Oncology 34, 1603-1612.

Kebriaei, P., Anastasi, J., and Larson, R.A. (2002). Acute lymphoblastic leukaemia: Diagnosis and classification. Best Practice & Research Clinical Haematology 15, 597- 621.

Kempski, H., Chalker, J., Chessells, J., Sturt, N., Brickell, P., Webb, J., Clink, J.M., and Reeves, B. (1999). An investigation of the t(12;21) rearrangement in children with B- precursor acute lymphoblastic leukaemia using cytogenetic and molecular methods. British Journal of Haematology 105, 684-689.

Kennedy, J.A., and Barabe, F. (2008). Investigating human leukemogenesis: From cell lines to in vivo models of human leukemia. Leukemia 22, 2029-2040.

Kerbel, R.S., Kobayashi, H., and Graham, C.H. (1994). Intrinsic or acquired drug resistance and metastasis: Are they linked phenotypes? Journal of Cellular Biochemistry 56, 37-47.

Klumper, E., Pieters, R., Veerman, A.J., Huismans, D.R., Loonen, A.H., Hahlen, K., Kaspers, G.J., van Wering, E.R., Hartmann, R., and Henze, G. (1995). In vitro cellular

322 drug resistance in children with relapsed/refractory acute lymphoblastic leukemia. Blood 86, 3861-3868.

Kobayashi, H., Takemura, Y., Holland, J.F., and Ohnuma, T. (1998). Vincristine saturation of cellular binding sites and its cytotoxic activity in human lymphoblastic leukemia cells: Mechanism of inoculum effect. Biochemical Pharmacology 55, 1229- 1234.

Kofler, R., Schmidt, S., Kofler, A., and Ausserlechner, M. (2003). Resistance to glucocorticoid-induced apoptosis in lymphoblastic leukemia. Journal of Endocrinology 178, 19-27.

Kothari, A., Hittelman, W.N., and Chambers, T. C. (2016). "Cell Cycle–Dependent Mechanisms Underlie Vincristine-Induced Death of Primary Acute Lymphoblastic Leukemia Cells." Cancer Research 76, 3553-3561.

Krentz, S., Hof, J., Mendioroz, A., Vaggopoulou, R., Dorge, P., Lottaz, C., Engelmann, J.C., Groeneveld, T.W.L., Korner, G., Seeger, K., et al. (2013). Prognostic value of genetic alterations in children with first bone marrow relapse of childhood B-cell precursor acute lymphoblastic leukemia. Leukemia 27, 295-304.

Krishnan, S., Masurekar, A., and Saha, V. (2011). Identifying targets for new therapies in children with acute lymphoblastic leukemia. In New agents for the treatment of acute lymphoblastic leukemia, V. Saha, and P. Kearns, eds. (New York, NY: Springer New York), pp. 25-37.

Krivtsov, A.V., Wang, X., Farnoud, N.R., Hadler, M., Martin, M.S., Knapp, K.M., Stein, E.M., Lipson, D., Nahas, M.K., Stephens, P.J., et al. (2014). Patient derived xenograft (PDX) models recapitulate the genomic-driver composition of acute leukemia samples. Blood 124, 286-286.

Kunz, J.B., Rausch, T., Bandapalli, O.R., Eilers, J., Pechanska, P., Schuessele, S., Assenov, Y., Stütz, A.M., Kirschner-Schwabe, R., Hof, J., et al. (2015). Pediatric T-cell

323 lymphoblastic leukemia evolves into relapse by clonal selection, acquisition of mutations and hypomethylation. Haematologica 100, 1442-1450.

Labuda, M., Gahier, A., Gagné, V., Moghrabi, A., Sinnett, D., and Krajinovic, M. (2010). "Polymorphisms in glucocorticoid receptor gene and the outcome of childhood acute lymphoblastic leukemia (ALL)." Leukemia Research 34, 492-497.

Lamb, J. (2007). The connectivity map: A new tool for biomedical research. Nature Reviews Cancer 7, 54-60.

Landau, D.A., Carter, S.L., Getz, G., and Wu, C.J. (2014). Clonal evolution in hematological malignancies and therapeutic implications. Leukemia 28, 34-43.

Lane, S.W., and Gilliland, D.G. (2010). Leukemia stem cells. Seminar in Cancer Biology 20, 71-76.

Langerak, A.W., Groenen, P.J.T.A., Bruggemann, M., Beldjord, K., Bellan, C., Bonello, L., Boone, E., Carter, G.I., Catherwood, M., Davi, F., et al. (2012). Euroclonality/biomed-2 guidelines for interpretation and reporting of IG/TCR clonality testing in suspected lymphoproliferations. Leukemia 26, 2159-2171.

Lanzkowsky, P. (2011). Chapter 17 - leukemias. In Manual of pediatric hematology and oncology (fifth edition), P. Lanzkowsky, ed. (San Diego: Academic Press), pp. 518-566.

Lavoie Smith, E.M., Li, L., Hutchinson, R.J., Ho, R., Burnette, W.B., Wells, E., Bridges, C., and Renbarger, J. (2013). Measuring vincristine-induced peripheral neuropathy in children with acute lymphoblastic leukemia. Cancer nursing 36, E49-60. le Viseur, C., Hotfilder, M., Bomken, S., Wilson, K., Röttgers, S., Schrauder, A., Rosemann, A., Irving, J., Stam, R.W., Shultz, L.D., et al. (2008). In childhood acute lymphoblastic leukemia, blasts at different stages of immunophenotypic maturation have stem cell properties. Cancer Cell 14, 47-58.

LeBien, T.W. (2000). Fates of human B-cell precursors. Blood 96, 9-23.

324

LeBien, T.W., and Tedder, T.F. (2008). B lymphocytes: How they develop and function. Blood 112, 1570-1580.

Lee, S.Y., Kim, J.W., Jeong, M.H., An, J.H., Jang, S.M., Song, K.H., and Choi, K.H. (2008). Microtubule-associated protein 1B light chain (MAP1B-LC1) negatively regulates the activity of tumor suppressor p53 in neuroblastoma cells. Federation of European Biochemical Societies Letters 582, 2826-2832.

Lee-Sherick, A.B., Linger, R.M., Gore, L., Keating, A.K., and Graham, D.K. (2010). Targeting paediatric acute lymphoblastic leukaemia: Novel therapies currently in development. British Journal of Haematology 151, 295-311.

Liem, N.L.M., Papa, R.A., Milross, C.G., Schmid, M.A., Tajbakhsh, M., Choi, S., Ramirez, C.D., Rice, A.M., Haber, M., Norris, M.D., et al. (2004). Characterization of childhood acute lymphoblastic leukemia xenograft models for the preclinical evaluation of new therapies. Blood 103, 3905-3914.

Li, Y., Zhang, T., and Sun, D. (2009). New developments in hsp90 inhibitors as anti- cancer therapeutics: Mechanisms, clinical perspective and more potential. Drug resistance updates: reviews and commentaries in antimicrobial and anticancer chemotherapy 12, 17-27.

Lilleyman, J.S. (1998). Clinical importance of speed of response to therapy in childhood lymphoblastic leukaemia. Leukemia & Lymphoma 31, 501-506.

Liu, D., Ahmet, A., Ward, L., Krishnamoorthy, P., Mandelcorn, E., Leigh, R., Brown, J., Cohen, A., and Kim, H. (2013). A practical guide to the monitoring and management of the complications of systemic corticosteroid therapy. Allergy, Asthma & Clinical Immunology 9, 30.

Lobo, N.A., Shimono, Y., Qian, D., and Clarke, M.F. (2007). The biology of cancer stem cells. Annual Reviews of Cell and Developmental Biology 23, 675-699.

325 Locatelli, F., Schrappe, M., Bernardo, M.E., and Rutella, S. (2012). How I treat relapsed childhood acute lymphoblastic leukemia. Blood 120, 2807-2816.

Lock, R.B., Liem, N., Farnsworth, M.L., Milross, C.G., Xue, C., Tajbakhsh, M., Haber, M., Norris, M.D., Marshall, G.M., and Rice, A.M. (2002). The nonobese diabetic/severe combined immunodeficient (NOD/SCID) mouse model of childhood acute lymphoblastic leukemia reveals intrinsic differences in biologic characteristics at diagnosis and relapse. Blood 99, 4100-4108.

Loh, M.L., Tasian, S.K., Rabin, K.R., Brown, P., Magoon, D., Reid, J.M., Chen, X., Ahern, C.H., Weigel, B.J., and Blaney, S.M. (2015). A phase 1 dosing study of ruxolitinib in children with relapsed or refractory solid tumors, leukemias, or myeloproliferative neoplasms: A children's oncology group phase 1 consortium study (advl1011). Pediatric blood & cancer 62, 1717-1724.

Lugthart, S., Cheok, M.H., den Boer, M.L., Yang, W., Holleman, A., Cheng, C., Pui, C.-H., Relling, M.V., Janka-Schaub, G.E., Pieters, R., et al. (2005). Identification of genes associated with chemotherapy crossresistance and treatment response in childhood acute lymphoblastic leukemia. Cancer Cell 7, 375-386.

Ma, X., Edmonson, M., Yergeau, D., Muzny, D.M., Hampton, O.A., Rusch, M., Song, G., Easton, J., Harvey, R.C., Wheeler, D.A., et al. (2015). Rise and fall of subclones from diagnosis to relapse in pediatric B-acute lymphoblastic leukaemia. Nature

Communications 6, 6604.

Marshall, G.M., Dalla Pozza, L., Sutton, R., Ng, A., de Groot-Kruseman, H.A., van der Velden, V.H., Venn, N.C., van den Berg, H., de Bont, E.S., Maarten Egeler, R., et al. (2013). High-risk childhood acute lymphoblastic leukemia in first remission treated with novel intensive chemotherapy and allogeneic transplantation. Leukemia 27, 1497- 1503.

Marshall, G.M., Haber, M., Kwan, E., Zhu, L., Ferrara, D., Xue, C., Brisco, M.J., Sykes, P.J., Morley, A., Webster, B., et al. (2003). Importance of minimal residual

326 disease testing during the second year of therapy for children with acute lymphoblastic leukemia. Journal of clinical oncology 21, 704-709.

Matthias, P., and Rolink, A.G. (2005). Transcriptional networks in developing and mature B cells. Nature Reviews Immunology 5, 497-508.

Mazurier, F., Doedens, M., Gan, O.I., and Dick, J.E. (2003). Rapid myeloerythroid repopulation after intrafemoral transplantation of NOD-SCID mice reveals a new class of human stem cells. Nature Medicine 9, 959-963.

McCredie, K.B., Ho, D.H.W., and Freireich, E.J. (1973). L-asparaginase for the treatment of cancer. CA: A Cancer Journal for Clinicians 23, 220-227.

McCune, J., Namikawa, R., Kaneshima, H., Shultz, L., Lieberman, M., and Weissman, I. (1988). The scid-hu mouse: Murine model for the analysis of human hematolymphoid differentiation and function. Science 241, 1632-1639.

McDermott, M., Eustace, A.J., Busschots, S., Breen, L., Crown, J., Clynes, M., O’Donovan, N., and Stordal, B. (2014). In vitro development of chemotherapy and targeted therapy drug-resistant cancer cell lines: A practical guide with case studies. Frontiers in Oncology 4, 40.

McGregor, S., McNeer, J., and Gurbuxani, S. (2012). Beyond the 2008 world health organization classification: The role of the hematopathology laboratory in the diagnosis and management of acute lymphoblastic leukemia. Seminars in diagnostic pathology 29, 2-11.

McLean, T.W., Ringold, S., Neuberg, D., Stegmaier, K., Tantravahi, R., Ritz, J., Koeffler, H.P., Takeuchi, S., Janssen, J.W., Seriu, T., et al. (1996). TEL/AML-1 dimerizes and is associated with a favorable outcome in childhood acute lymphoblastic leukemia. Blood 88, 4252-4258.

327 McNeer, J.L., and Nachman, J.B. (2010). The optimal use of steroids in paediatric acute lymphoblastic leukaemia: No easy answers. British Journal of Haematology 149, 638- 652.

Meyer, L.H., and Debatin, K.-M. (2011). Diversity of human leukemia xenograft mouse models: Implications for disease biology. Cancer Research 71, 7141-7144.

Meyer, L.H., Eckhoff, S.M., Queudeville, M., Kraus, J.M., Giordan, M., Stursberg, J., Zangrando, A., Vendramini, E., Moricke, A., Zimmermann, M., et al. (2011). Early relapse in all is identified by time to leukemia in nod/scid mice and is characterized by a gene signature involving survival pathways. Cancer cell 19, 206-217.

Miller, D.R., Coccia, P.F., Bleyer, W.A., Lukens, J.N., Siegel, S.E., Sather, H.N., and Hammond, G.D. (1989). Early response to induction therapy as a predictor of disease- free survival and late recurrence of childhood acute lymphoblastic leukemia: A report from the childrens cancer study group. Journal of Clinical Oncology 7, 1807-1815.

Mitchell, C., Richards, S., Harrison, C.J., and Eden, T. (2010). Long-term follow-up of the united kingdom medical research council protocols for childhood acute lymphoblastic leukaemia, 1980–2001. Leukemia 24, 406-418.

Miyata, Y., Nakamoto, H., and Neckers, L. (2013). The therapeutic target hsp90 and cancer hallmarks. Current Pharmaceutical Design 19, 347-365.

Moorman, A.V., Enshaei, A., Schwab, C., Wade, R., Chilton, L., Elliott, A., Richardson, S., Hancock, J., Kinsey, S.E., Mitchell, C.D., et al. (2014). A novel integrated cytogenetic and genomic classification refines risk stratification in pediatric acute lymphoblastic leukemia. Blood 124, 1434-1444.

Moricke, A., Reiter, A., Zimmermann, M., Gadner, H., Stanulla, M., Dordelmann, M., Loning, L., Beier, R., Ludwig, W.D., and Ratei, R. (2008). Risk-adjusted therapy of acute lymphoblastic leukemia can decrease treatment burden and improve survival: Treatment results of 2169 unselected pediatric and adolescent patients enrolled in the trial ALL-BFM 95. Blood 111. 328

Moricke, A., Zimmermann, M., Reiter, A., Henze, G., Schrauder, A., Gadner, H., Ludwig, W.D., Ritter, J., Harbott, J., Mann, G., et al. (2009). Long-term results of five consecutive trials in childhood acute lymphoblastic leukemia performed by the ALL- BFM study group from 1981 to 2000. Leukemia 24, 265-284.

Moudi, M., Go, R., Yien, C.Y.S., and Nazre, M. (2013). Vinca alkaloids. International Journal of Preventive Medicine 4, 1231-1235.

Mullighan, C.G. (2012). Molecular genetics of B-precursor acute lymphoblastic leukemia. The Journal of Clinical Investigation 122, 3407-3415.

Mullighan, C.G., Goorha, S., Radtke, I., Miller, C.B., Coustan-Smith, E., Dalton, J.D., Girtman, K., Mathew, S., Ma, J., Pounds, S.B., et al. (2007). Genome-wide analysis of genetic alterations in acute lymphoblastic leukaemia. Nature 446, 758-764.

Mullighan, C.G., Phillips, L.A., Su, X., Ma, J., Miller, C.B., Shurtleff, S.A., and Downing, J.R. (2008). Genomic analysis of the clonal origins of relapsed acute lymphoblastic leukemia. Science 322, 1377-1380.

Mullighan, C.G., Su, X., Zhang, J., Radtke, I., Phillips, L.A.A., Miller, C.B., Ma, J., Liu, W., Cheng, C., Schulman, B.A., et al. (2009). Deletion of IKZF1 and prognosis in acute lymphoblastic leukemia. New England Journal of Medicine 360, 470-480.

Nagasawa, T. (2006). Microenvironmental niches in the bone marrow required for B- cell development. Nature Reviews Immunology 6, 107-116.

Neale, G., Su, X., Morton, C.L., Phelps, D., Gorlick, R., Lock, R.B., Reynolds, C.P., Maris, J.M., Friedman, H.S., Dome, J., et al. (2008). Molecular characterization of the pediatric preclinical testing panel. Clinical cancer research 14, 4572-4583.

Nersting, J., Borst, L., and Schmiegelow, K. (2011). Challenges in implementing individualized medicine illustrated by antimetabolite therapy of childhood acute lymphoblastic leukemia. Clinical proteomics 8, 8-8. 329

Nguyen, K., Devidas, M., Cheng, S.C., La, M., Raetz, E.A., Carroll, W.L., Winick, N.J., Hunger, S.P., Gaynon, P.S., and Loh, M.L. (2008). Factors influencing survival after relapse from acute lymphoblastic leukemia: A children's oncology group study. Leukemia 22, 2142-2150.

Nicolaides, N.C., Galata, Z., Kino, T., Chrousos, G.P., and Charmandari, E. (2010). The human glucocorticoid receptor: Molecular basis of biologic function. Steroids 75, 1-12.

Nijmeijer, B.A., Mollevanger, P., van Zelderen-Bhola, S.L., Kluin-Nelemans, H.C., Willemze, R., and Falkenburg, J.H. (2001). Monitoring of engraftment and progression of acute lymphoblastic leukemia in individual NOD/SCID mice. Experimental Hematology 29, 322-329.

Nijmeijer, B.A., Szuhai, K., Goselink, H.M., van Schie, M.L.J., van der Burg, M., de Jong, D., Marijt, E.W., Ottmann, O.G., Willemze, R., and Falkenburg, J.H.F. (2009). Long–term culture of primary human lymphoblastic leukemia cells in the absence of serum or hematopoietic growth factors. Experimental Hematology 37, 376-385.

Nogales, E. (2000). Structural insights into microtubule function. Annual Review of Biochemisty 69, 277-302.

Notta, F., Mullighan, C.G., Wang, J.C., Poeppl, A., Doulatov, S., Phillips, L.A., Ma, J., Minden, M.D., Downing, J.R., and Dick, J.E. (2011). Evolution of human BCR-ABL1 lymphoblastic leukaemia-initiating cells. Nature 469, 362-367.

Nowak, D., Liem, N.L., Mossner, M., Klaumunzer, M., Papa, R.A., Nowak, V., Jann, J.C., Akagi, T., Kawamata, N., Okamoto, R., et al. (2015). Variegated clonality and rapid emergence of new molecular lesions in xenografts of acute lymphoblastic leukemia are associated with drug resistance. Experimental Hematology 43, 32-43.e31- 35.

Nwajei, F., and Konopleva, M. (2013). The bone marrow microenvironment as niche retreats for hematopoietic and leukemic stem cells. Advances in Hematology 2013, 8. 330

O'Connell, B.C., O'Callaghan, K., Tillotson, B., Douglas, M., Hafeez, N., West, K.A., Stern, H., Ali, J.A., Changelian, P., Fritz, C.C., et al. (2014). HSP90 inhibition enhances antimitotic drug-induced mitotic arrest and cell death in preclinical models of non-small cell lung cancer. PLoS ONE 9, e115228.

Orkin, S.H., and Zon, L.I. (2008). Hematopoiesis: An evolving paradigm for stem cell biology. Cell 132, 631-644.

Pal, D., Blair, H. J., Elder, A., Dormon, K., Rennie, K. J., Coleman, D. J. L., Weiland, J., Rankin, K. S.,Filby, A., Heidenreich, O and Vormoor, J. (2016). "Long-term in vitro maintenance of clonal abundance and leukaemia-initiating potential in acute lymphoblastic leukaemia." Leukemia. 30, 1691–1700.

Palmi, C., Vendramini, E., Silvestri, D., Longinotti, G., Frison, D., Cario, G., Shochat, C., Stanulla, M., Rossi, V., Di Meglio, A.M., et al. (2012). Poor prognosis for P2RY8- CRLF2 fusion but not for CRLF2 over-expression in children with intermediate risk B- cell precursor acute lymphoblastic leukemia. Leukemia 26, 2245-2253.

Panzer-Grümayer, E.R., Schneider, M., Panzer, S., Fasching, K., and Gadner, H. (2000). Rapid molecular response during early induction chemotherapy predicts a good outcome in childhood acute lymphoblastic leukemia. Blood 95, 790-794.

Parker, A.L., Kavallaris, M., and McCarroll, J.A. (2014). Microtubules and their role in cellular stress in cancer. Frontiers in Oncology 4, 153.

Parker, C., Waters, R., Leighton, C., Hancock, J., Sutton, R., Moorman, A.V., Ancliff, P., Morgan, M., Masurekar, A., Goulden, N., et al. (2010). Effect of mitoxantrone on outcome of children with first relapse of acute lymphoblastic leukaemia (ALL R3): An open-label randomised trial. The Lancet 376, 2009-2017.

Pastorczak, A., W. Fendler, B. Zalewska-Szewczyk, P. Górniak, M. Lejman, J. Trelińska, J. Walenciak, J. Kowalczyk, T. Szczepanski and W. Mlynarski (2014). "Asparagine synthetase (ASNS) gene polymorphism is associated with the outcome of 331 childhood acute lymphoblastic leukemia by affecting early response to treatment." Leukemia Research 38, 180-183.

Pearson, T., Greiner, D.L., and Shultz, L.D. (2008). Humanized Scid mouse models for biomedical research. Current topics in microbiology and immunology 324, 25-51.

Perez-Vera, P., Reyes-Leon, A., and Fuentes-Panana, E.M. (2011). Signaling proteins and transcription factors in normal and malignant early B cell development. Bone Marrow Research 2011, 502751.

Pieters, R., den Boer, M.L., Durian, M., Janka, G., Schmiegelow, K., Kaspers, G.J., van Wering, E.R., and Veerman, A.J. (1998). Relation between age, immunophenotype and in vitro drug resistance in 395 children with acute lymphoblastic leukemia--implications for treatment of infants. Leukemia 12, 1344-1348.

Piller, G.J. (2001). Leukaemia – a brief historical review from ancient times to 1950. British Journal of Haematology 112, 282-292.

Pinkel, D., Simone, J., Hustu, H.O., and Aur, R.J. (1972). Nine years' experience with "total therapy" of childhood acute lymphocytic leukemia. Pediatrics 50, 246-251.

Politi, K., and Pao, W. (2011). How genetically engineered mouse tumor models provide insights into human cancers. Journal of Clinical Oncology 29, 2273-2281.

Prochazka, M., Gaskins, H.R., Shultz, L.D., and Leiter, E.H. (1992). The nonobese diabetic scid mouse: Model for spontaneous thymomagenesis associated with immunodeficiency. Proceedings of the National Academy of Sciences of the United States of America 89, 3290-3294.

Proia, D.A., Sang, J., He, S., Smith, D.L., Sequeira, M., Zhang, C., Liu, Y., Ye, S., Zhou, D., Blackman, R.K., et al. (2012). Synergistic activity of the HSP90 inhibitor ganetespib with taxanes in non-small cell lung cancer models. Investigational new drugs 30, 2201-2209. 332

Pui, C.-H., ed. (2006). Childhood leukemias (Cambridge University Press).

Pui, C.-H., and Evans, W.E. (2006). Treatment of acute lymphoblastic leukemia. New England Journal of Medicine 354, 166-178.

Pui, C.-H., and Evans, W.E. (2013). A 50-year journey to cure childhood acute lymphoblastic leukemia. Seminars in hematology 50, 185-196.

Pui, C.-H., Mullighan, C.G., Evans, W.E., and Relling, M.V. (2012). Pediatric acute lymphoblastic leukemia: Where are we going and how do we get there? Blood 120, 1165-1174.

Pui, C.H., Behm, F.G., and Crist, W.M. (1993). Clinical and biologic relevance of immunologic marker studies in childhood acute lymphoblastic leukemia. Blood 82, 343- 362.

Qu, X.A., and Rajpal, D.K. (2012). Applications of connectivity map in drug discovery and development. Drug Discovery Today 17, 1289-1298.

Raetz, E.A., and Bhatla, T. (2012). Where do we stand in the treatment of relapsed acute lymphoblastic leukemia? ASH Education Program Book 2012, 129-136.

Raimondi, S.C., Behm, F.G., Roberson, P.K., Williams, D.L., Pui, C.H., Crist, W.M., Look, A.T., and Rivera, G.K. (1990). Cytogenetics of pre-B-cell acute lymphoblastic leukemia with emphasis on prognostic implications of the t(1;19). Journal of Clinical Oncology 8, 1380-1388.

Rehe, K., Wilson, K., Bomken, S., Williamson, D., Irving, J., den Boer, M.L., Stanulla, M., Schrappe, M., Hall, A.G., Heidenreich, O., et al. (2013). Acute B lymphoblastic leukaemia-propagating cells are present at high frequency in diverse lymphoblast populations. EMBO Molecular Medicine 5, 38-51.

333 Reiter, A., Schrappe, M., Ludwig, W.D., Hiddemann, W., Sauter, S., Henze, G., Zimmermann, M., Lampert, F., Havers, W., Niethammer, D., et al. (1994). Chemotherapy in 998 unselected childhood acute lymphoblastic leukemia patients. Results and conclusions of the multicenter trial ALL-BFM 86. Blood 84, 3122-3133.

Richardson, P.G., Mitsiades, C.S., Laubach, J.P., Lonial, S., Chanan-Khan, A.A., and Anderson, K.C. (2011). Inhibition of heat shock protein 90 (HSP90) as a therapeutic strategy for the treatment of myeloma and other cancers. British Journal of Haematology 152, 367-379.

Rivera, G.K., Zhou, Y., Hancock, M.L., Gajjar, A., Rubnitz, J., Ribeiro, R.C., Sandlund, J.T., Hudson, M., Relling, M., Evans, W.E., et al. (2005). Bone marrow recurrence after initial intensive treatment for childhood acute lymphoblastic leukemia. Cancer 103, 368-376.

Roberts, A.W., Seymour, J.F., Brown, J.R., Wierda, W.G., Kipps, T.J., Khaw, S.L., Carney, D.A., He, S.Z., Huang, D.C.S., Xiong, H., et al. (2011). Substantial susceptibility of chronic lymphocytic leukemia to BCL2 inhibition: Results of a phase I study of navitoclax in patients with relapsed or refractory disease. Journal of Clinical Oncology.

Robison, L.L. (2011). Late effects of acute lymphoblastic leukemia therapy in patients diagnosed at 0-20 years of age. ASH Education Program Book 2011, 238-242.

Roganovic, J. (2013). Acute lymphoblastic leukemia in children. Jelena Roganovic (2013). Acute Lymphoblastic Leukemia in Children, Leukemia, InTech, DOI: 10.5772/55655.

Samuels, A.L., Beesley, A.H., Yadav, B.D., Papa, R.A., Sutton, R., Anderson, D., Marshall, G.M., Cole, C.H., Kees, U.R., and Lock, R.B. (2014). A pre-clinical model of resistance to induction therapy in pediatric acute lymphoblastic leukemia. Blood Cancer Journal 4, e232.

334 Schmiegelow, K., Nielsen, S.N., Frandsen, T.L., and Nersting, J. (2014). Mercaptopurine/methotrexate maintenance therapy of childhood acute lymphoblastic leukemia: Clinical facts and fiction. Journal of Pediatric Hematology/Oncology 36, 503-517.

Schmitz, M., Breithaupt, P., Scheidegger, N., Cario, G., Bonapace, L., Meissner, B., Mirkowska, P., Tchinda, J., Niggli, F.K., Stanulla, M., et al. (2011). Xenografts of highly resistant leukemia recapitulate the clonal composition of the leukemogenic compartment. Blood 118, 1854-1864.

Schrappe, M., Reiter, A., Ludwig, W.D., Harbott, J., Zimmermann, M., Hiddemann, W., Niemeyer, C., Henze, G., Feldges, A., Zintl, F., et al. (2000). Improved outcome in childhood acute lymphoblastic leukemia despite reduced use of anthracyclines and cranial radiotherapy: Results of trial ALL-BFM 90. German-austrian-swiss all-bfm study group. Blood 95, 3310-3322.

Schultz, K.R., Pullen, D.J., Sather, H.N., Shuster, J.J., Devidas, M., Borowitz, M.J., Carroll, A.J., Heerema, N.A., Rubnitz, J.E., Loh, M.L., et al. (2007). Risk- and response-based classification of childhood B-precursor acute lymphoblastic leukemia: A combined analysis of prognostic markers from the pediatric oncology group (POG) and children's cancer group (CCG). Blood 109, 926-935.

Schwab, C.J., Chilton, L., Morrison, H., Jones, L., Al-Shehhi, H., Erhorn, A., Russell, L.J., Moorman, A.V., and Harrison, C.J. (2013). Genes commonly deleted in childhood B-cell precursor acute lymphoblastic leukemia: Association with cytogenetics and clinical features. Haematologica 98, 1081-1088.

Sędek, Ł., Flores-Montero, J., Bulsa, J., Barrena, S., Almeida, J., Orfao, A., and Szczepański, T. (2012). Flow cytometric immunophenotyping as diagnostic tool of hematopoietic malignancies. In Molecular aspects of hematologic malignancies, M. Witt, M. Dawidowska, and T. Szczepanski, eds. (Springer Berlin Heidelberg), pp. 143- 160.

335 Seibel, N.L. (2008). Treatment of acute lymphoblastic leukemia in children and adolescents: Peaks and pitfalls. Hematology Am Soc Hematol Educ Program, 374-380. Shackleton, M., Quintana, E., Fearon, E.R., and Morrison, S.J. (2009). Heterogeneity in cancer: Cancer stem cells versus clonal evolution. Cell 138, 822-829.

Shinnick, S.E., Browning, M.L., and Koontz, S.E. (2013). Managing hypersensitivity to asparaginase in pediatrics, adolescents, and young adults. Journal of Pediatric Oncology Nursing 30, 63-77.

Shultz, L.D., Lyons, B.L., Burzenski, L.M., Gott, B., Chen, X., Chaleff, S., Kotb, M., Gillies, S.D., King, M., Mangada, J., et al. (2005). Human lymphoid and myeloid cell development in NOD/LTSz-scid IL2Rγnull mice engrafted with mobilized human hemopoietic stem cells. The Journal of Immunology 174, 6477-6489.

Shultz, L.D., Schweitzer, P.A., Christianson, S.W., Gott, B., Schweitzer, I.B., Tennent, B., McKenna, S., Mobraaten, L., Rajan, T.V., and Greiner, D.L. (1995). Multiple defects in innate and adaptive immunologic function in nod/ltsz-scid mice. The Journal of Immunology 154, 180-191.

Siolas, D., and Hannon, G.J. (2013). Patient-derived tumor xenografts: Transforming clinical samples into mouse models. Cancer Research 73, 5315-5319.

Smith, M., Arthur, D., Camitta, B., Carroll, A.J., Crist, W., Gaynon, P., Gelber, R., Heerema, N., Korn, E.L., Link, M., et al. (1996). Uniform approach to risk classification and treatment assignment for children with acute lymphoblastic leukemia. Journal of Clinical Oncology 14, 18-24.

Smith, M.A., Morton, C.L., Phelps, D.A., Kolb, E.A., Lock, R., Carol, H., Reynolds, C.P., Maris, J.M., Keir, S.T., Wu, J., et al. (2008). Stage 1 testing and pharmacodynamic evaluation of the HSP90 inhibitor alvespimycin (17-DMAG, KOS- 1022) by the pediatric preclinical testing program. Pediatric Blood & Cancer 51, 34-41.

Smith, V., Sausville, E.A., Camalier, R.F., Fiebig, H.H., and Burger, A.M. (2005). Comparison of 17-dimethylaminoethylamino-17-demethoxy-geldanamycin (17DMAG) 336 and 17-allylamino-17-demethoxygeldanamycin (17AAG) in vitro: Effects on HSP90 and client proteins in melanoma models. Cancer Chemotherapy Pharmacology 56, 126- 137.

Somasundaram, R., Prasad, M.A.J., Ungerbäck, J., and Sigvardsson, M. (2015). Transcription factor networks in B-cell differentiation link development to acute lymphoid leukemia. Blood 126, 144-152.

Spiegel, A., Kollet, O., Peled, A., Abel, L., Nagler, A., Bielorai, B., Rechavi, G., Vormoor, J., and Lapidot, T. (2004). Unique SDF-1-induced activation of human precursor-B ALL cells as a result of altered CXCR4 expression and signaling. Blood 103, 2900-2907.

Stanulla, M., and Schrappe, M. (2009). Treatment of childhood acute lymphoblastic leukemia. Seminars in Hematology 46, 52-63.

Stiller, C.A. (2004). Epidemiology and genetics of childhood cancer. Oncogene 23, 6429-6444.

Styczynski, J., Piatkowska, M., Jaworska-Posadzy, A., Czyzewski, K., Kubicka, M., Kolodziej, B., Kurylo-Rafinska, B., Debski, R., Pogorzala, M., and Wysocki, M. (2012). Comparison of prognostic value of in vitro drug resistance and bone marrow residual disease on day 15 of therapy in childhood acute lymphoblastic leukemia. Anticancer Research 32, 5495-5499.

Subramanian, A., Tamayo, P., Mootha, V.K., Mukherjee, S., Ebert, B.L., Gillette, M.A., Paulovich, A., Pomeroy, S.L., Golub, T.R., Lander, E.S., et al. (2005). Gene set enrichment analysis: A knowledge-based approach for interpreting genome-wide expression profiles. Proceedings of the National Academy of Sciences of the USA 102, 15545-15550.

Suryani, S., Carol, H., Chonghaile, T.N., Frismantas, V., Sarmah, C., High, L., Bornhauser, B., Cowley, M.J., Szymanska, B., Evans, K., et al. (2014). Cell and molecular determinants of in vivo efficacy of the BH3 mimetic ABT-263 against 337 pediatric acute lymphoblastic leukemia xenografts. Clinical cancer research 20, 4520- 4531.

Sutcliffe, M.J., Shuster, J.J., Sather, H.N., Camitta, B.M., Pullen, J., Schultz, K.R., Borowitz, M.J., Gaynon, P.S., Carroll, A.J., and Heerema, N.A. (2005). High concordance from independent studies by the children's cancer group (CCG) and pediatric oncology group (POG) associating favorable prognosis with combined trisomies 4, 10, and 17 in children with NCI standard-risk B-precursor acute lymphoblastic leukemia: A children's oncology group (COG) initiative. Leukemia 19, 734-740.

Szczepanek, J., Styczyński, J., Haus, O., Tretyn, A., and Wysocki, M. (2011). Relapse of acute lymphoblastic leukemia in children in the context of microarray analyses. Archivum Immunologiae et Therapiae Experimentalis 59, 61-68.

Szczepański, T., van der Velden Vincent, H.J., and van Dongen Jacques, J.M. (2006). Flow-cytometric immunophenotyping of normal and malignant lymphocytes. In Clinical Chemical Laboratory Medicine, pp. 775.

Szymanska, B., Wilczynska-Kalak, U., Kang, M.H., Liem, N.L.M., Carol, H., Boehm, I., Groepper, D., Reynolds, C.P., Stewart, C.F., and Lock, R.B. (2012). Pharmacokinetic modeling of an induction regimen for in vivo combined testing of novel drugs against pediatric acute lymphoblastic leukemia xenografts. PLoS ONE 7, e33894.

Tasian, S.K., Doral, M.Y., Borowitz, M.J., Wood, B.L., Chen, I.M., Harvey, R.C., Gastier-Foster, J.M., Willman, C.L., Hunger, S.P., Mullighan, C.G., et al. (2012). Aberrant STAT5 and PI3K/MTOR pathway signaling occurs in human CRLF2- rearranged b-precursor acute lymphoblastic leukemia. Blood 120, 833-842.

Teachey, D.T., and Hunger, S.P. (2013). Predicting relapse risk in childhood acute lymphoblastic leukaemia. British Journal of Haematology 162, 606-620.

338 Tissing, W.J.E., Meijerink, J.P.P., den Boer, M.L., and Pieters, R. (2003). Molecular determinants of glucocorticoid sensitivity and resistance in acute lymphoblastic leukemia. Leukemia 17, 17-25.

Toyoda, Y., Manabe, A., Tsuchida, M., Hanada, R., Ikuta, K., Okimoto, Y., Ohara, A., Ohkawa, Y., Mori, T., Ishimoto, K., et al. (2000). Six months of maintenance chemotherapy after intensified treatment for acute lymphoblastic leukemia of childhood. Journal of Clinical Oncology 18, 1508-1516.

Tubergen, D.G., Gilchrist, G.S., O'Brien, R.T., Coccia, P.F., Sather, H.N., Waskerwitz, M.J., and Hammond, G.D. (1993a). Improved outcome with delayed intensification for children with acute lymphoblastic leukemia and intermediate presenting features: A childrens cancer group phase III trial. Journal of Clinical Oncology 11, 527-537.

Tubergen, D.G., Gilchrist, G.S., O'Brien, R.T., Coccia, P.F., Sather, H.N., Waskerwitz, M.J., and Hammond, G.D. (1993b). Prevention of CNS disease in intermediate-risk acute lymphoblastic leukemia: Comparison of cranial radiation and intrathecal methotrexate and the importance of systemic therapy: A childrens cancer group report. Journal of Clinical Oncology 11, 520-526.

Turner, N.C., and Reis-Filho, J.S. (2012). Genetic heterogeneity and cancer drug resistance. The Lancet Oncology 13, e178-e185.

Tzifi, F., Economopoulou, C., Gourgiotis, D., Ardavanis, A., Papageorgiou, S., and Scorilas, A. (2012). The role of BCL2 family of apoptosis regulator proteins in acute and chronic leukemias. Advances in Hematology 2012, 15.

Uckun, F.M., Sather, H.N., Waurzyniak, B.J., Sensel, M.G., Chelstrom, L., Ek, O., Sarquis, M.B., Nachman, J., Bostrom, B., Reaman, G.H., et al. (1998). Prognostic significance of B-lineage leukemic cell growth in scid mice: A children's cancer group study. Leukemia and Lymphoma 30, 503-514. van der Weyden, L., Adams, D.J., and Bradley, A. (2002). Tools for targeted manipulation of the mouse genome. Physiological Genomics 11, 133-164. 339 van Dongen, J.J.M., Seriu, T., Panzer-Grümayer, E.R., Biondi, A., Pongers-Willemse, M.J., Corral, L., Stolz, F., Schrappe, M., Masera, G., Kamps, W.A., et al. (1998). Prognostic value of minimal residual disease in acute lymphoblastic leukaemia in childhood. The Lancet 352, 1731-1738.

Vangala, R.K., Heiss-Neumann, M.S., Rangatia, J.S., Singh, S.M., Schoch, C., Tenen, D.G., Hiddemann, W., and Behre, G. (2003). The myeloid master regulator transcription factor PU.1 is inactivated by AML1-ETO in t(8;21) myeloid leukemia. Blood 101, 270-277.

Verrills, N.M., and Kavallaris, M. (2005). Improving the targeting of tubulin-binding agents: Lessons from drug resistance studies. Current Pharmaceutical Design 11, 1719- 1733.

Virely, C., Moulin, S., Cobaleda, C., Lasgi, C., Alberdi, A., Soulier, J., Sigaux, F., Chan, S., Kastner, P., and Ghysdael, J. (2010). Haploinsufficiency of the IKZF1 (IKAROS) cooperates with BCR-ABL in a transgenic model of acute lymphoblastic leukemia. Leukemia 24, 1200-1204.

Walsh, D., and Avashia, J. (1992). Glucocorticoids in clinical oncology. Cleveland Clinic Journal of Medicine 59, 505-515.

Weis, F., Moullintraffort, L., Heichette, C., Chrétien, D., and Garnier, C. (2010). The 90-Kda heat shock protein HSP90 protects tubulin against thermal denaturation. The Journal of Biological Chemistry 285, 9525-9534.

Whiteford, C.C., Bilke, S., Greer, B.T., Chen, Q., Braunschweig, T.A., Cenacchi, N., Wei, J.S., Smith, M.A., Houghton, P., Morton, C., et al. (2007). Credentialing preclinical pediatric xenograft models using gene expression and tissue microarray analysis. Cancer Research 67, 32-40.

Wilding, J.L., and Bodmer, W.F. (2014). Cancer cell lines for drug discovery and development. Cancer Research 74, 2377-2384. 340

Williams, S.A., Anderson, W.C., Santaguida, M.T., and Dylla, S.J. (2013). Patient- derived xenografts, the cancer stem cell paradigm, and cancer pathobiology in the 21st century. Laboratory Investigation 93, 970-982.

Woiterski, J., Ebinger, M., Witte, K.E., Goecke, B., Heininger, V., Philippek, M., Bonin, M., Schrauder, A., Rottgers, S., Herr, W., et al. (2013). Engraftment of low numbers of pediatric acute lymphoid and myeloid leukemias into NOD/SCID/IL2Rcgammanull mice reflects individual leukemogenecity and highly correlates with clinical outcome. International Journal of Cancer 133, 1547-1556.

Wong, N.C., Bhadri, V.A., Maksimovic, J., Parkinson-Bates, M., Ng, J., Craig, J.M., Saffery, R., and Lock, R.B. (2014). Stability of gene expression and epigenetic profiles highlights the utility of patient-derived paediatric acute lymphoblastic leukaemia xenografts for investigating molecular mechanisms of drug resistance. BMC Genomics 15, 416.

Wu, C.P., Calcagno, A.M., and Ambudkar, S.V. (2008). Reversal of abc drug transporter-mediated multidrug resistance in cancer cells: Evaluation of current strategies. Current Molecular Pharmacology 1, 93-105.

Yang, J.J., Bhojwani, D., Yang, W., Cai, X., Stocco, G., Crews, K., Wang, J., Morrison, D., Devidas, M., Hunger, S.P., et al. (2008). Genome-wide copy number profiling reveals molecular evolution from diagnosis to relapse in childhood acute lymphoblastic leukemia. Blood 112, 4178-4183.

American Cancer Society. Cancer Facts & Figures 2015. Atlanta: American Cancer Society; 2015 at: http://www.cancer.org/acs/groups/content/@editorial/documents/document/acspc- 044552.pdf.

Cancer Council QLD, 2014 at: https://cancerqld.org.au/wp- content/uploads/2015/11/summary-of-childhood-cancer-in-australia.pdf.

341

Appendix A

Figure A.1. Criteria used to report mouse health and identify TTL manifestation over the monitoring period.

342

Appendix B

Table B.1. List of the 841 genes that were significantly up-regulated in the ALL-67 xenograft panel in comparison to the corresponding patient sample. The cut-off for significance was set at FDR <0.05. absFC, absolute fold change.

abs absF Gene P.Value Q Value Gene P.Value Q Value FC C RGS1 4.9 1.027E-08 0.00000300 SAAL1 1.4 0.0007817 0.0166187 LOC1001295 PDE4B 3.5 1.619E-08 0.00000438 66 1.6 0.0007873 0.0167259 ATF3 7.9 3.71E-08 0.00000910 SLC25A4 1.7 0.0007895 0.0167552 CDCA8 2.4 6.637E-08 0.00001476 SNORA67 2.7 0.0007913 0.0167681 SOD2 2.9 8.836E-08 0.00001854 C10ORF12 1.4 0.000795 0.0168326 KIF2C 2.0 9.363E-08 0.00001949 LSP1 1.6 0.0007984 0.0168783 HJURP 2.2 9.873E-08 0.00002031 GBA3 2.0 0.0008064 0.0169685 SNHG3 - ODC1 2.9 1.912E-07 0.00003385 RCC1 1.6 0.0008066 0.0169685 DHRS3 2.2 2.007E-07 0.00003529 LOC643509 1.6 0.0008113 0.0170538 GINS2 2.1 2.086E-07 0.00003619 LOC728368 1.4 0.0008162 0.0171431 ID3 4.2 2.103E-07 0.00003624 HIST1H2AG 1.5 0.0008201 0.017197 RAD54L 2.3 2.951E-07 0.00004926 LOC653610 2.2 0.0008217 0.0172173 LOC1001320 FABP5L2 3.1 2.972E-07 0.00004926 32 1.6 0.0008251 0.0172468 ZWINT 2.1 4.542E-07 0.00007093 TOB1 3.6 0.0008265 0.0172612 BAMBI 5.1 5.564E-07 0.00008226 LOC646256 1.3 0.0008329 0.0173695 ESPL1 2.1 0.000000597 0.0000868 USP3 1.3 0.0008332 0.0173695 CKS1B 1.8 8.483E-07 0.0001162 C1ORF55 1.5 0.0008347 0.0173695 CDC25A 2.0 9.685E-07 0.0001292 ALDH7A1 1.7 0.0008404 0.017467 SERPINE1 2.8 9.773E-07 0.0001298 NP 1.6 0.0008485 0.0176028 TPX2 2.3 0.000001078 0.0001402 ITPRIP 1.5 0.0008548 0.0176785 SLC30A1 2.2 0.000001103 0.0001428 RND1 2.1 0.0008573 0.0176785 BCL6 3.2 0.000001119 0.0001442 CCT3 1.4 0.000858 0.0176785 LOC653381 2.1 0.000001138 0.0001455 RELT 1.8 0.0008616 0.0177378 MYLIP 2.6 0.000001141 0.0001455 HIST1H2BM 3.9 0.0008636 0.0177378 DYRK3 2.5 0.000001164 0.0001474 PKP4 1.8 0.0008691 0.0178344 EBP 1.9 0.000001172 0.0001474 HIST1H3J 2.6 0.0008713 0.0178541 PIM1 4.0 0.000001184 0.0001481 TOE1 1.4 0.0008792 0.0179451 FSCN1 3.5 0.00000122 0.0001508 LOC645651 1.3 0.0008812 0.0179572 BTG3 2.2 0.000001339 0.0001606 KIF14 1.7 0.0008854 0.018015 INCENP 2.0 0.000001364 0.0001628 BCL2L12 1.4 0.0008934 0.0181207 DUSP10 2.8 0.000001382 0.0001642 STIL 1.7 0.0008979 0.0181833 DUSP26 2.9 0.000001414 0.0001672 CCDC34 1.6 0.0009006 0.0182069 TRIB1 4.8 0.000001672 0.0001925 ACSL3 1.3 0.0009039 0.0182347 SLC20A1 1.9 0.000001762 0.0002011 AQR 1.4 0.0009157 0.0184442 HBEGF 7.1 0.000001771 0.0002013 FAM53C 3.0 0.0009197 0.0184984 LOC1001343 ATIC 2.1 0.000001963 0.0002165 60 1.6 0.0009279 0.0186182 MCM5 2.0 0.000002199 0.0002365 ATOX1 1.7 0.0009312 0.0186351 NME1 2.3 0.000002504 0.0002628 BCLAF1 1.6 0.0009314 0.0186351 WDR33 1.8 0.000002527 0.0002641 LOC729090 1.5 0.000935 0.0186742 RRAGD 2.3 0.000002538 0.0002641 HSPA1A 7.5 0.0009387 0.0187339 PSRC1 1.8 0.000002629 0.0002684 PEG3 1.8 0.0009466 0.0188469 SOX4 2.2 0.000002848 0.000285 FAM58B 1.3 0.0009478 0.0188576 LOC345041 2.5 0.000002968 0.0002959 PRC1 1.8 0.0009489 0.0188636 BNIP1 1.6 0.000003022 0.0002979 OSM 2.2 0.0009531 0.0189298 343

PLAUR 2.3 0.000003523 0.0003383 C3ORF38 1.5 0.0009537 0.0189298 PRDX1 1.9 0.000003628 0.0003458 LOC643284 1.6 0.0009583 0.0189908 MTP18 3.1 0.000004004 0.0003721 NDUFA6 1.4 0.0009589 0.0189908 C6ORF173 2.1 0.000004037 0.0003739 XKR4 1.6 0.000961 0.0189938 FAM110A 1.8 0.000004062 0.0003748 EIF4H 1.3 0.0009616 0.0189938 C20ORF11 1 2.1 0.000004291 0.000388 NDUFB6 1.4 0.0009715 0.0191092 CHEK1 2.2 0.000004661 0.0004156 CEP170 1.6 0.0009813 0.0192591 PPP1R15A 1.8 0.000004663 0.0004156 POLR2H 1.4 0.0009878 0.019342 IL1B 2.2 0.000004741 0.0004197 C1ORF112 1.6 0.0009896 0.0193546 ETV6 1.8 0.000004896 0.0004304 BHLHB2 4.2 0.000991 0.0193546 FAM107B 2.0 0.000005225 0.0004547 LOC646197 1.9 0.0009955 0.0194204 HN1 2.0 0.00000548 0.0004738 KIF4A 1.7 0.0009973 0.0194394 RMND5A 2.2 0.000005516 0.0004743 GAS1 1.9 0.0010061 0.0195684 ELK3 2.0 0.000005564 0.0004743 TUBG1 1.6 0.0010198 0.0197736 CD83 7.4 0.000005613 0.0004743 NR4A2 3.2 0.0010212 0.0197736 BYSL 2.1 0.000005897 0.000495 RSRC2 1.4 0.0010276 0.0198519 LOC647000 1.9 0.00000672 0.0005534 HS.124358 1.2 0.0010313 0.0198687 CDKN2D 2.0 0.000006868 0.0005638 NEUROG2 1.5 0.0010314 0.0198687 RAN 2.0 0.000006953 0.0005677 PIK3C2B 1.8 0.001036 0.0199404 LOC1001330 CPXM1 1.8 0.000007002 0.0005691 12 1.6 0.00104 0.0199932 CBX1 2.4 0.000007098 0.0005737 NRM 1.6 0.0010409 0.0199932 PLEKHO1 1.9 0.000007611 0.0006033 PCDH18 2.2 0.001058 0.0202207 GATA2 2.1 0.000007627 0.0006033 ZNF593 1.7 0.0010666 0.0203049 MCM4 2.0 0.000008021 0.0006249 LOC400455 1.4 0.0010827 0.0205803 EYA3 1.7 0.000008312 0.0006419 SNORD35A 2.0 0.0010839 0.0205887 SLC5A6 1.8 0.000008449 0.0006486 CACNB2 1.5 0.0011022 0.0208761 MAPK6 1.8 0.000008683 0.0006626 COL5A1 1.6 0.0011189 0.0211011 EZH2 2.0 0.00001029 0.0007695 RPL27A 1.5 0.0011413 0.0214755 LIG1 1.4 0.00001039 0.0007745 AMY1C 1.7 0.0011486 0.0215107 LOC652128 2.2 0.00001078 0.0007996 LOC653663 1.5 0.0011574 0.0215746 TRIAP1 1.6 0.00001083 0.0008003 GCHFR 1.6 0.001162 0.0216463 RERGL 5.2 0.00001103 0.0008067 C12ORF24 1.4 0.0011631 0.0216489 CENPN 1.9 0.00001117 0.0008145 EIF4A3 1.7 0.0011717 0.0217193 UBE2C 2.3 0.00001157 0.0008389 PLK4 1.6 0.0011817 0.0218554 PDLIM7 2.4 0.00001249 0.0008944 EIF3I 1.4 0.0011862 0.0219235 C19ORF48 1.8 0.00001456 0.0010053 DDX28 1.3 0.0011987 0.0220928 ATP9A 2.5 0.00001463 0.0010072 C6ORF54 1.5 0.0012055 0.0221931 LOC730167 2.2 0.00001502 0.001026 MIR216A 2.3 0.0012066 0.0221931 PTTG1 2.0 0.00001518 0.0010316 KIF20B 1.6 0.0012067 0.0221931 SNORD21 1.7 0.00001527 0.0010346 P2RY5 1.7 0.0012077 0.022195 BTG2 2.2 0.00001602 0.0010744 C16ORF87 1.6 0.0012157 0.0222952 MGC40489 1.8 0.00001619 0.0010829 SNORD51 1.4 0.00122 0.0223424 KIFC1 1.9 0.00001703 0.0011213 TMEM39B 1.6 0.0012218 0.0223436 NET1 2.0 0.00001734 0.0011311 KIF20A 1.9 0.0012394 0.0226398 LOC647719 2.8 0.00001794 0.0011644 H1FX 1.6 0.0012464 0.0227292 ATF5 2.6 0.00001834 0.0011815 SOX11 1.6 0.0012505 0.0227557 LOC728037 1.7 0.00001873 0.0011942 MTHFD2 1.7 0.0012552 0.0228265 TTK 1.8 0.00002043 0.0012747 DDX39 1.4 0.001261 0.0228845 PTTG3P 2.2 0.00002064 0.0012816 ATXN7L2 1.8 0.0012619 0.0228845 ATL3 1.6 0.00002115 0.0013076 UBE2J1 2.1 0.0012633 0.0228845 LOC646993 2.2 0.00002121 0.0013081 RAD1 1.3 0.0012654 0.0228845 CDKN1A 6.1 0.00002229 0.0013575 GNL3 1.4 0.0012662 0.0228845 PIAS2 1.6 0.00002229 0.0013575 INPP5K 1.3 0.0012663 0.0228845 FABP5 2.6 0.00002353 0.0014172 DTL 1.6 0.0012727 0.0229634

344

MYH10 1.7 0.00002403 0.0014406 WIPF1 1.7 0.0012732 0.0229634 IRF2BP2 2.4 0.00002417 0.0014426 HIST2H2AC 1.9 0.0012733 0.0229634 CXORF64 1.8 0.00002457 0.0014599 ELOVL1 1.4 0.0012919 0.0232499 LOC729964 2.3 0.00002514 0.0014836 DCTPP1 1.5 0.0013004 0.023386 TMEM48 1.6 0.00002691 0.001564 TBC1D5 1.3 0.0013179 0.0236038 C6ORF66 1.5 0.00002693 0.001564 HLA-DMA 1.5 0.0013229 0.0236599 RCBTB2 1.8 0.00002769 0.0016046 CIRH1A 1.4 0.0013397 0.0239276 PAICS 1.9 0.00002905 0.0016614 PABPC4 1.4 0.0013418 0.0239484 LOC650557 2.1 0.00002962 0.0016827 ZCCHC7 1.6 0.0013489 0.0240488 CTCF 1.9 0.00003003 0.0016954 LOC645726 1.6 0.0013501 0.0240488 LOC652479 2.5 0.00003057 0.0017184 DDX3X 1.8 0.001351 0.0240488 NUDT15 1.8 0.00003081 0.0017277 CSF3R 2.3 0.0013533 0.0240598 CKS2 2.3 0.00003187 0.0017686 SNRPD1 1.4 0.0013536 0.0240598 KIF22 1.9 0.00003326 0.001826 EIF4G2 1.4 0.0013609 0.0241571 CDC20 2.4 0.00003675 0.0019852 RAB17 1.5 0.0013691 0.0242712 CARHSP1 1.7 0.00003685 0.0019852 POU2AF1 1.5 0.0013709 0.0242859 C17ORF53 2.2 0.00003688 0.0019852 CABLES2 1.4 0.0013853 0.0244733 VKORC1L 1 1.8 0.00003725 0.0019987 PGRMC1 1.4 0.001388 0.024505 TOP1 1.5 0.00003805 0.0020248 TMEM97 1.5 0.0014038 0.0247509 IL18BP 1.4 0.0000384 0.0020394 EBPL 1.5 0.0014081 0.0248088 LTB 2.4 0.00003902 0.002064 MSRA 1.6 0.0014176 0.0249597 HSPA1B 8.0 0.00003917 0.0020674 SLC25A25 1.5 0.0014195 0.0249765 CCNE1 1.7 0.00003942 0.0020765 DUSP5 1.9 0.001426 0.0250562 BIRC5 2.0 0.00003951 0.0020774 CLDN12 1.5 0.0014318 0.0251345 LOC731049 2.7 0.00003965 0.0020802 LOC727913 1.5 0.0014333 0.0251345 CCDC136 2.3 0.00004028 0.0021069 NUSAP1 2.0 0.0014382 0.0251858 TAGAP 1.9 0.00004038 0.0021069 TSC22D1 2.2 0.0014437 0.0252316 HS.276860 1.6 0.00004125 0.00213 ZNF844 1.7 0.0014602 0.0254507 LOC1001327 LOC642956 2.5 0.0000418 0.0021533 40 1.4 0.001462 0.0254665 TROAP 2.2 0.0000422 0.0021659 POLQ 1.5 0.0014705 0.0255799 TRIM8 1.7 0.00004272 0.00218 ETNK1 1.6 0.0014735 0.0256142 FAM54A 1.5 0.00004565 0.0022811 LOC652233 1.3 0.0014759 0.02564 LOC100134 444 3.3 0.00004568 0.0022811 HS.527671 1.6 0.0014811 0.0257121 CAD 1.5 0.00004576 0.0022811 EFTUD2 1.3 0.0014873 0.0257677 NOP56 1.7 0.00004633 0.0023006 MRPL21 1.3 0.0014909 0.0257958 ING2 2.1 0.00005213 0.0025004 GATAD2A 1.4 0.0014934 0.0258224 H2AFJ 2.0 0.00005217 0.0025004 HIST2H3D 2.2 0.0015052 0.0259925 LOC390940 1.5 0.00005227 0.0025005 DIAPH1 1.4 0.0015081 0.0260132 RGS4 1.4 0.00005349 0.0025399 NEU4 1.9 0.0015154 0.0260996 SLC2A3 6.0 0.00005367 0.0025439 KIAA1618 1.7 0.0015173 0.0261018 HIST1H2B H 3.1 0.00005393 0.0025515 STOX2 1.3 0.0015253 0.0262179 PCP4 2.6 0.00005411 0.0025556 NCKAP1L 1.4 0.0015302 0.0262838 TUBA1B 1.4 0.00005536 0.0026003 MPHOSPH9 1.4 0.0015314 0.0262866 LDHA 1.7 0.00005646 0.0026424 CHAF1B 1.9 0.0015475 0.0264763 MCM7 1.5 0.00005877 0.002726 C19ORF54 1.3 0.0015484 0.0264763 CRELD2 1.8 0.0000589 0.0027274 MGC16703 1.7 0.0015914 0.0270139 DSCC1 1.9 0.00005931 0.0027371 FARSLB 1.5 0.0016039 0.0271386 CSMD2 2.5 0.00005972 0.0027461 OBFC2B 1.3 0.0016061 0.0271584 CCNJL 1.7 0.00006333 0.0028913 SEMA3F 1.5 0.0016125 0.0272308 MTA3 1.7 0.00006451 0.0029202 FEM1B 1.6 0.0016146 0.0272308 RRP12 1.6 0.00006452 0.0029202 RANBP1 1.5 0.0016207 0.0273157 PSME4 1.6 0.00006676 0.0029955 HPS1 1.3 0.0016218 0.0273167 EFNB1 2.7 0.00006702 0.003002 RHOB 2.2 0.0016302 0.0274228 345

FBXO33 1.7 0.00007254 0.0032046 SNX5 1.4 0.0016335 0.0274598 LOC1001334 C10ORF64 1.5 0.00007325 0.0032201 20 1.6 0.0016437 0.02757 POP7 1.7 0.00007448 0.0032686 PIM2 1.7 0.0016443 0.02757 IRX5 2.3 0.00007615 0.003325 ORC1L 1.4 0.0016456 0.027575 PHLDA1 3.0 0.00007641 0.003325 PHF19 1.4 0.0016607 0.0277924 C16ORF59 1.7 0.00007765 0.0033622 ZNF827 1.5 0.0016659 0.0278609 SNORD52 1.7 0.00007822 0.0033814 DIS3L2 1.4 0.0016681 0.0278716 GTF2IRD2 B 1.9 0.00007849 0.0033815 KLF10 2.3 0.0016687 0.0278716 RRM1 1.9 0.00007849 0.0033815 CYGB 2.2 0.0016752 0.027963 XRCC6 1.6 0.0000797 0.0034111 C9ORF86 1.4 0.0016804 0.0280228 LRRC8C 2.0 0.00008066 0.0034355 RFX3 1.3 0.0016855 0.0280804 HMGB3 1.8 0.00008176 0.0034652 MICALL1 1.4 0.0016925 0.0281785 SNORD11 2.1 0.0000826 0.0034895 CHD3 1.4 0.0016998 0.0282466 LOC387703 1.6 0.00008305 0.0034991 DSEL 1.4 0.0017205 0.0284721 CCNF 1.7 0.0000831 0.0034991 SPC24 1.8 0.001721 0.0284721 HS.574671 1.6 0.00008371 0.0035135 ENC1 2.0 0.0017227 0.0284823 C16ORF80 1.6 0.00008448 0.0035346 LOC732429 1.2 0.0017257 0.0285123 SHPK 1.4 0.00008909 0.00368 ZNF350 1.3 0.0017312 0.0285693 FOXM1 1.8 0.00009196 0.0037686 LOC388796 1.5 0.0017313 0.0285693 IER2 3.0 0.00009229 0.0037762 TP53INP2 1.7 0.0017424 0.0286987 NEXN 1.5 0.00009466 0.0038491 MSH6 1.7 0.0017516 0.02881 CALU 1.5 0.00009505 0.0038589 MSI2 1.9 0.0017559 0.0288284 INHBB 2.5 0.00009601 0.0038917 ARIH1 1.3 0.0017591 0.0288454 C16ORF75 1.6 0.00009651 0.003894 RETNLB 2.6 0.0017677 0.0289278 TACC3 2.1 0.0001 0.003992 TRUB2 1.3 0.0017686 0.0289278 MED20 1.6 0.0001016 0.0040309 CHEK2 1.5 0.001778 0.0290632 CDCA5 2.0 0.0001033 0.0040855 EPHA2 1.5 0.0017835 0.0290998 POLR3K 1.5 0.0001033 0.0040855 THRAP3 1.4 0.0017869 0.0291313 TNFAIP3 2.7 0.000105 0.0041457 MAP7D1 1.7 0.0017889 0.0291313 COX5A 1.6 0.0001069 0.0041736 HS.250648 1.4 0.0017943 0.0291626 PRR8 2.0 0.0001073 0.0041736 SNORD68 1.5 0.0017953 0.0291626 IER5 2.5 0.0001073 0.0041736 NCOR1 1.3 0.0017997 0.0292035 LOC1001336 C16ORF61 1.5 0.0001097 0.0042555 97 1.6 0.0018145 0.029402 VAV1 1.7 0.0001121 0.0043314 NCAPG2 1.7 0.001818 0.0294223 SGOL1 1.6 0.0001131 0.0043459 OR5P3 1.5 0.0018198 0.0294229 ETV3 1.9 0.0001144 0.0043834 UGT3A2 3.2 0.0018251 0.0294809 VASH2 1.8 0.0001145 0.0043837 ZFAND2A 1.5 0.001841 0.0296439 CDCA3 1.9 0.0001149 0.0043905 LOC728715 1.8 0.0018483 0.0297087 GLA 1.4 0.0001167 0.0044476 TOMM22 1.3 0.001863 0.0298597 KIF15 1.9 0.000119 0.0045279 SLED1 1.4 0.0018657 0.0298597 KIF25 2.2 0.0001195 0.0045364 LOC653210 1.3 0.0018856 0.0301063 LGALS8 1.6 0.0001202 0.0045451 PVT1 1.6 0.001889 0.0301396 C7ORF40 1.8 0.0001203 0.0045451 XRN2 1.3 0.001897 0.0302303 CENPM 1.6 0.0001217 0.0045824 CENPO 1.3 0.0019152 0.0304461 ABCG1 1.9 0.0001219 0.0045824 HSPD1 1.8 0.0019174 0.0304623 ACBD7 2.2 0.0001258 0.0047228 STMN1 1.4 0.0019254 0.0305469 TNS1 2.1 0.0001272 0.0047476 COPS6 1.3 0.0019492 0.0308549 LOC441734 2.7 0.0001274 0.0047493 RGL4 1.9 0.0019622 0.0310216 TRIM24 1.7 0.0001279 0.0047634 XPNPEP3 1.4 0.0019898 0.0314017 LOC1001343 EWSR1 1.5 0.0001287 0.0047704 04 1.6 0.0019971 0.0314968 DHX9 1.6 0.0001294 0.0047832 HS.573575 1.3 0.0020019 0.0315351 DNAJB9 1.9 0.0001296 0.0047842 HIST1H2BC 1.8 0.0020078 0.0315748 DYNLL2 2.1 0.0001309 0.0048184 LOC729317 1.3 0.0020099 0.0315748 346

SLC43A3 2.0 0.000132 0.004844 GADD45B 2.6 0.0020104 0.0315748 HS.571024 2.4 0.0001328 0.0048611 CD300LF 1.6 0.0020111 0.0315748 WDR62 1.5 0.0001335 0.0048708 C21ORF96 1.5 0.0020201 0.0316307 RFTN1 1.7 0.0001338 0.004875 BZRAP1 1.6 0.0020463 0.0319441 FEN1 1.8 0.0001345 0.0048812 CKAP5 1.4 0.0020513 0.0319845 TUBB2C 2.3 0.0001345 0.0048812 YWHAH 1.5 0.0020559 0.0320367 LOC727808 1.3 0.0001372 0.0049525 WDR51A 1.7 0.0020623 0.0321175 SLC3A2 2.2 0.0001374 0.0049525 RTN4R 1.4 0.0020653 0.0321437 MELK 2.0 0.0001381 0.0049721 EXO1 1.7 0.0020757 0.0322212 RHOH 1.9 0.0001389 0.0049845 FRAG1 1.3 0.0020764 0.0322212 MRPL27 1.5 0.0001404 0.0050187 ZMYM3 1.2 0.0020777 0.0322212 SYNCRIP 1.6 0.0001412 0.0050415 PTPLAD1 1.5 0.0020797 0.0322337 TIMELESS 1.6 0.0001438 0.0051124 NAPG 1.4 0.0020859 0.0322913 SLC35A2 1.5 0.0001506 0.0052955 RNASEH2A 1.4 0.0020997 0.0324467 MYADM 1.7 0.0001508 0.0052955 HEY2 1.7 0.0021079 0.0324897 CCRN4L 1.8 0.0001542 0.0053994 LOC651845 1.6 0.0021087 0.0324897 DNAJA1 1.7 0.0001565 0.0054296 SNORA77 1.5 0.0021102 0.0324932 OTUD1 2.4 0.000157 0.0054337 HIST1H3G 2.7 0.0021166 0.0325536 LOC728873 1.9 0.0001582 0.0054663 TK1 1.7 0.00212 0.032587 LOC651816 3.0 0.0001594 0.0055017 LOC644745 1.5 0.0021448 0.0328508 CCND1 2.0 0.0001596 0.0055025 PAFAH1B3 1.6 0.0021467 0.0328605 HIST1H2A B 3.7 0.0001627 0.0055862 CHERP 1.3 0.0021519 0.0329129 LOC1001336 HS.430851 1.5 0.0001635 0.0056046 49 1.7 0.0021648 0.0330608 PIM3 2.1 0.000164 0.0056105 C15ORF42 1.5 0.0021777 0.03318 NCAPG 2.2 0.0001686 0.0057283 MIR1204 1.4 0.0022208 0.0336375 LOC442597 1.8 0.0001706 0.0057664 STIP1 1.6 0.0022288 0.0336588 HIST1H2AI 2.3 0.0001752 0.0058522 ATP1A1 1.3 0.002235 0.033698 ZC4H2 1.7 0.0001757 0.0058543 SERPINE2 1.8 0.0022368 0.0337052 LOC387934 1.7 0.0001794 0.0059386 MRTO4 1.4 0.002241 0.0337486 MCM2 1.4 0.0001796 0.0059386 SMAGP 2.2 0.0022645 0.034043 FLJ14166 1.5 0.0001817 0.0059775 MARCKSL1 1.7 0.0022686 0.0340696 COPS3 1.5 0.0001825 0.0059969 LOC644975 1.3 0.002277 0.03413 BRSK1 1.6 0.0001829 0.0060018 HS.570751 1.5 0.0023087 0.0344302 HERC2P4 1.4 0.0001847 0.0060314 GGH 1.7 0.0023116 0.034433 FAM64A 2.1 0.0001859 0.0060619 EPOR 1.7 0.0023208 0.0344912 TYMS 2.2 0.000187 0.0060839 HIST1H2AH 3.4 0.0023291 0.0345362 JMJD6 1.9 0.0001889 0.0061035 HMGA1 1.8 0.0023291 0.0345362 NASP 1.9 0.0001896 0.0061133 HSP90AA1 1.8 0.0023311 0.034546 CEP72 1.4 0.0001918 0.0061619 PKN3 1.5 0.0023337 0.0345651 LOC100133 LOC1001285 477 1.5 0.0001918 0.0061619 05 1.3 0.0023397 0.0346146 SASH3 1.5 0.0001939 0.0062074 MRPL14 1.3 0.0023437 0.0346341 TLR6 1.5 0.000194 0.0062074 DIAPH3 1.4 0.0023453 0.0346379 LOC650526 1.9 0.0001975 0.006284 MAPK13 1.9 0.0023481 0.0346592 LOC1001290 AGPAT9 1.5 0.0001994 0.0063342 34 1.5 0.0023594 0.0347636 GSS 1.6 0.0001998 0.0063359 MPDU1 1.5 0.0023635 0.0347887 HS.25892 1.7 0.0002001 0.0063359 CALR 1.9 0.0023654 0.0347962 HS.497591 1.5 0.0002001 0.0063359 TUSC3 1.4 0.0023715 0.0348671 MICB 1.8 0.0002037 0.0064331 IL12A 1.3 0.002375 0.0348987 LOC1001322 LMNB1 2.1 0.000205 0.0064583 99 1.4 0.0023816 0.0349232 SFPQ 1.5 0.0002098 0.0065871 CYB5A 1.3 0.0023907 0.0350106 FCGRT 1.8 0.0002102 0.0065876 CDH29 1.5 0.0023952 0.0350426 LOC645251 2.1 0.0002104 0.0065876 SF3B5 1.4 0.0023965 0.0350426

347

SCARNA10 1.5 0.0002197 0.0068297 SNRNP25 1.5 0.0023989 0.0350426 DPP3 1.4 0.0002231 0.0069194 SIVA 1.5 0.0023994 0.0350426 CDC25C 1.4 0.0002239 0.0069315 TCP11L1 1.3 0.002401 0.0350426 DCPS 1.4 0.0002261 0.0069699 LOC649679 1.8 0.0024026 0.0350459 C16ORF91 1.4 0.0002271 0.006993 KIF23 1.5 0.0024179 0.0352499 SLC30A3 1.3 0.0002277 0.0070027 NXT1 1.5 0.0024247 0.0353293 PAF1 1.5 0.0002287 0.0070093 FLJ11783 1.5 0.0024291 0.0353731 H2AFZ 1.7 0.000233 0.0071176 ID2 2.4 0.0024345 0.0354325 CEACAM1 2.0 0.0002352 0.0071769 UBE2I 1.3 0.0024513 0.0356095 CKAP4 2.0 0.0002383 0.0072542 TNPO2 1.2 0.0024521 0.0356095 NPY 2.4 0.000239 0.0072574 NOC4L 1.5 0.0024596 0.0356786 C12ORF52 1.5 0.0002396 0.0072574 C1ORF163 1.3 0.0024718 0.0358151 HIST1H3F 2.8 0.0002398 0.0072574 EIF4G1 1.3 0.0024779 0.0358471 CTNND1 1.5 0.0002421 0.007301 PFKM 1.6 0.002479 0.0358471 AK2 1.6 0.000243 0.007319 APP 1.6 0.0024837 0.0358471 DDIT3 3.7 0.0002449 0.007358 GRAMD1B 1.4 0.0024896 0.0358931 AFG3L2 1.6 0.000247 0.0073972 MKI67 1.5 0.0024969 0.0359382 HIST1H3H 2.7 0.0002495 0.0074452 C19ORF52 1.3 0.0025102 0.0360279 CHST10 1.5 0.0002514 0.0074852 CENPF 1.8 0.0025207 0.0361017 LOC1001327 ZNF317 1.6 0.0002546 0.007571 15 1.5 0.0025309 0.0362072 TMED1 1.7 0.0002565 0.0075981 SNORD14B 1.5 0.0025325 0.0362108 NTNG2 1.6 0.0002586 0.007629 GPR18 1.7 0.0025406 0.0363062 IKBKE 1.4 0.0002598 0.0076556 SLCO5A1 1.2 0.0025524 0.0364558 IDH2 1.6 0.0002619 0.0077096 PSMC3IP 1.3 0.0025564 0.0364768 SNORD14 A 1.6 0.0002641 0.0077389 MYC 1.8 0.0025915 0.0368033 ARRDC3 2.8 0.0002696 0.0078816 PCNA 1.8 0.0025919 0.0368033 HHEX 2.0 0.0002723 0.0079448 ODF2 1.5 0.0025923 0.0368033 GNA15 2.0 0.0002791 0.0081056 NMT2 1.5 0.002597 0.0368498 C6ORF108 1.5 0.0002802 0.008124 MRPS26 1.4 0.0026008 0.0368829 LOC100128 266 1.7 0.0002807 0.008124 IFRD1 1.9 0.0026023 0.0368848 HELLS 1.8 0.0002817 0.0081457 MRPL22 1.2 0.0026091 0.036932 ACBD6 1.5 0.0002839 0.0081832 AURKAPS1 1.2 0.0026091 0.036932 LOC653604 2.2 0.000284 0.0081832 MRPS9 1.3 0.0026099 0.036932 MEX3D 1.6 0.0002859 0.008213 EFTUD1 1.4 0.0026134 0.0369418 TP53BP2 1.3 0.0002868 0.0082292 CCDC28B 1.3 0.0026226 0.0370315 GLDC 1.6 0.0002874 0.0082365 OIP5 1.7 0.0026343 0.037081 PLK2 2.5 0.0002881 0.008248 TJP2 1.4 0.0026347 0.037081 AARSD1 1.6 0.0002886 0.0082533 QPRT 1.4 0.0026388 0.0371181 CHTF18 1.7 0.0002925 0.0083185 TUBB4Q 2.1 0.0026498 0.0372538 HSP90AB1 1.8 0.0002949 0.00836 MORF4L2 1.3 0.0026548 0.0373038 KLHL36 1.6 0.0002997 0.0084469 DUS3L 1.3 0.0026572 0.037317 MLLT11 1.9 0.0003007 0.0084469 FBXO5 1.6 0.002661 0.0373506 LOC285074 1.7 0.0003015 0.0084553 SNRPC 1.4 0.0026782 0.0375473 C7ORF68 1.5 0.0003078 0.0086027 HMGXB4 1.4 0.0026794 0.0375473 PXMP2 1.6 0.0003091 0.0086209 NOP10 1.3 0.0026905 0.0376604 CDT1 2.2 0.0003112 0.0086438 SPTY2D1 1.5 0.0026958 0.0376757 UBTF 1.4 0.0003191 0.0087504 DNMT1 1.6 0.0027048 0.0377618 MCM6 1.7 0.0003204 0.0087776 FGFR1OP 1.4 0.0027093 0.0378044 ASAP2 1.5 0.000324 0.0088279 GPT 1.5 0.0027169 0.0378691 RECQL4 1.6 0.0003261 0.0088502 HARS 1.5 0.0027203 0.037897 PDIA6 1.9 0.0003269 0.0088523 LOC652272 1.3 0.00273 0.037971 EIF4A1 1.5 0.0003297 0.0088998 HS.527241 1.3 0.0027595 0.038279 LOC440093 1.8 0.0003309 0.0089245 HDAC2 1.6 0.0027612 0.0382824

348

CCNB2 2.0 0.0003326 0.0089499 RPLP1 1.5 0.0027635 0.0382928 BCAT2 1.6 0.0003332 0.0089532 POU2F1 1.4 0.0027782 0.0384561 LOC1001280 CACYBP 1.8 0.0003395 0.0090791 07 1.5 0.0027897 0.0385531 NAP1L5 1.7 0.0003402 0.0090891 NUCKS1 1.5 0.002795 0.038606 VWCE 2.7 0.0003423 0.009126 DLGAP5 1.9 0.0028141 0.0388295 RANBP6 1.5 0.0003451 0.009154 CUL1 1.2 0.0028206 0.0388491 SGK269 1.5 0.00035 0.0092576 PHF17 1.6 0.0028215 0.0388491 B3GALTL 1.4 0.0003518 0.0092845 ISOC1 1.4 0.0028253 0.0388678 BCL3 2.8 0.0003531 0.0092862 CENPV 1.9 0.0028355 0.0389588 SPRY1 1.6 0.0003544 0.0093075 LOC647361 1.3 0.0028435 0.0390481 PRMT1 1.5 0.0003552 0.0093172 ELAC2 1.2 0.0028618 0.0392579 B4GALT6 2.3 0.0003656 0.009534 EXOSC7 1.5 0.0028653 0.0392673 CCDC21 1.4 0.0003673 0.0095678 CTPS 1.4 0.0028717 0.0393318 NNAT 3.3 0.0003694 0.0096005 DERL2 1.4 0.0028764 0.0393749 LOC1001299 HMOX1 1.8 0.0003775 0.009766 05 2.6 0.0029025 0.0396702 PARP2 1.5 0.0003797 0.0098015 UBAP1 1.4 0.0029053 0.0396878 ZNF784 1.6 0.0003798 0.0098015 LOC729040 1.4 0.002914 0.0397547 RAD51C 1.7 0.0003804 0.0098015 LOC728312 1.3 0.0029144 0.0397547 CBX4 3.7 0.0003856 0.0099066 WDR41 1.6 0.0029298 0.0399391 HIST1H3C 3.7 0.0003889 0.0099708 SUPT6H 1.2 0.0029341 0.0399758 RYR3 2.4 0.0003899 0.009987 HIST1H3D 2.5 0.0029363 0.0399856 LOC1001295 TXNRD1 2.0 0.0003938 0.0100767 41 1.6 0.0029437 0.0400651 E2F2 2.4 0.0003965 0.0101276 HIST1H1E 2.4 0.0029563 0.0402158 DDX24 1.7 0.000409 0.0103961 RNU86 1.8 0.002976 0.0404413 GOT1 1.4 0.0004108 0.0104196 RGS12 1.3 0.0029832 0.040518 LOC1001285 SNRPB 1.7 0.0004132 0.0104587 33 1.3 0.0029856 0.0405288 SRF 1.9 0.000414 0.0104678 LOC647322 1.3 0.0029893 0.0405583 SLC7A3 1.5 0.0004152 0.0104806 CDK2AP1 1.7 0.0029948 0.0406122 SLC38A5 2.0 0.0004259 0.0106987 SLC25A40 1.5 0.0030222 0.0408986 SPAG5 1.6 0.0004379 0.010947 ZNF460 1.3 0.0030278 0.0409307 CCNA2 2.1 0.0004379 0.010947 IER3 3.7 0.0030351 0.0409874 TMPO 1.7 0.0004401 0.010991 SLC16A12 1.6 0.0030613 0.0412346 KLHL15 2.3 0.0004452 0.0110866 LOC729217 1.4 0.0030684 0.0412876 TBC1D16 1.5 0.0004534 0.011237 POLD1 1.5 0.0030828 0.0414387 TSPYL1 1.4 0.0004584 0.0113292 DPH3 1.5 0.0030882 0.0414678 CDCA2 1.6 0.000472 0.0116098 SEC13 1.4 0.0030914 0.0414683 ANO1 1.8 0.0004743 0.0116524 LOC728361 1.2 0.0031062 0.0416233 FANCG 1.5 0.0004749 0.0116524 RBM28 1.4 0.0031159 0.0417091 PTS 1.5 0.0004752 0.0116524 LOC646347 1.5 0.003119 0.0417091 CASP7 1.4 0.0004755 0.0116524 LOC652400 1.3 0.0031299 0.0417692 HNRNPAB 1.7 0.0004782 0.0117061 FAM129C 1.5 0.0031431 0.0418802 TRH 3.1 0.0004793 0.0117225 TBCD 1.8 0.0031509 0.041963 SAE1 1.4 0.0004811 0.0117374 SS18L2 1.4 0.003156 0.0419777 WDR40A 1.6 0.0004815 0.0117374 ARL9 1.4 0.0031623 0.0419862 UBE2T 1.7 0.0004822 0.0117374 PSMC3 1.6 0.0031871 0.0422726 PHPT1 2.1 0.0004887 0.0118659 PM20D2 1.6 0.0031944 0.0423045 LOC1001334 KLC1 1.3 0.0004899 0.0118824 89 1.6 0.0032139 0.0425195 LRP8 1.5 0.0004966 0.0119987 DNAJB11 1.5 0.0032205 0.0425849 HS.171009 2.8 0.0004993 0.0120428 BLM 1.6 0.0032275 0.0426342 ARHGAP1 1B 1.6 0.0005077 0.0121769 SNUPN 1.5 0.003258 0.0429063 JUN 1.5 0.0005079 0.0121769 HIST2H2AB 2.4 0.0032616 0.0429278

349

AVPI1 1.5 0.0005137 0.0122535 LOC644670 1.5 0.0032723 0.0430084 TOPORS 1.5 0.0005137 0.0122535 PMM2 1.4 0.0033152 0.0433966 ATAD2 1.9 0.0005144 0.0122582 FOXD4L4 1.3 0.0033179 0.0434101 TOB2 1.9 0.0005151 0.0122643 AATF 1.3 0.0033252 0.0434616 CDC45L 2.0 0.0005201 0.0123721 HIST1H2BF 2.5 0.003373 0.0439319 C9ORF30 1.7 0.000532 0.0125782 RRM2 1.9 0.0033981 0.0441926 JOSD1 1.6 0.0005422 0.012758 IQSEC2 1.2 0.003412 0.0442629 HIST2H3C 2.4 0.000545 0.0128109 PFAS 1.4 0.0034541 0.0447413 DEF8 1.7 0.0005462 0.0128195 NME2 2.0 0.0034726 0.0449266 TRAFD1 1.6 0.0005503 0.0129024 SNORA34 1.7 0.0034906 0.0450398 PLEKHG2 1.6 0.0005555 0.0129996 GCM2 1.3 0.0034909 0.0450398 C9ORF126 1.4 0.0005562 0.0130041 PKMYT1 1.8 0.0035155 0.0452916 ZNF165 1.5 0.0005576 0.0130144 CCT2 1.3 0.0035157 0.0452916 SIK1 3.3 0.0005658 0.0131473 SPAG1 1.5 0.0035316 0.0454737 KPNA2 1.7 0.0005677 0.0131662 BTN1A1 2.2 0.0035361 0.0455093 U2AF1L4 1.5 0.000575 0.013303 PIF1 1.5 0.0035411 0.0455513 NSD1 1.4 0.0005817 0.0134308 C1ORF85 1.5 0.0035455 0.0455855 C18ORF22 1.5 0.0005877 0.013512 CNNM2 1.3 0.0035524 0.0456295 MTHFD1L 1.8 0.0005936 0.013609 CCND2 1.5 0.003567 0.0457715 TPR 1.4 0.000595 0.0136302 PRODH2 1.8 0.0035816 0.045913 LOC731314 2.0 0.0005968 0.0136532 LOC285804 1.5 0.0035939 0.0460158 POLE2 1.6 0.0005981 0.0136532 THOC6 1.5 0.0035949 0.0460158 LOC643995 2.1 0.0005984 0.0136532 LOC646575 1.3 0.0036153 0.0462083 ECE2 1.4 0.0006052 0.0137786 LUZP1 1.7 0.0036329 0.0464081 RBM38 2.3 0.0006069 0.013781 SNORA53 1.3 0.0036456 0.0465281 GNL2 1.3 0.0006143 0.0139003 HS.547277 1.8 0.0036604 0.0466712 ZFP36L1 2.1 0.0006291 0.0141291 SUCNR1 1.3 0.0036713 0.0467737 MDGA2 1.6 0.0006317 0.0141721 TRAF2 1.3 0.0036721 0.0467737 PMAIP1 1.9 0.0006336 0.0141927 HS.371609 2.6 0.0036763 0.0467872 CHN2 2.2 0.0006337 0.0141927 GOT2 1.3 0.0036817 0.0468122 MRPL54 1.5 0.0006379 0.0142613 SNORD99 1.9 0.0036896 0.0468598 SSRP1 1.6 0.0006385 0.0142627 RACGAP1 1.5 0.0037216 0.0471513 HS.369643 1.8 0.0006464 0.014414 SAT1 1.5 0.0037425 0.0473688 SNRNP40 1.3 0.000649 0.0144471 ETV2 1.5 0.0037459 0.0473893 LOC1001322 WDFY3 1.6 0.0006499 0.014456 52 1.2 0.003754 0.0474476 SEH1L 1.4 0.0006536 0.0145209 LOC644684 1.4 0.0037569 0.0474593 USP13 1.4 0.0006577 0.0145653 FAM72A 1.4 0.0037587 0.0474593 LOC729816 1.5 0.0006586 0.0145742 CEBPB 1.6 0.0037813 0.047706 LOC100128 805 1.4 0.0006618 0.0146067 LOC653526 1.3 0.0037838 0.047706 SNORA70 2.1 0.0006689 0.0147203 HS.352677 1.8 0.0037857 0.0477076 PDCD7 1.3 0.0006692 0.0147203 SUPT5H 1.3 0.0037949 0.0478001 LOC1001296 LOC440864 2.9 0.0006698 0.0147212 08 1.6 0.0038165 0.048026 LOC100131 940 1.6 0.0006767 0.014847 FAM115A 1.3 0.003856 0.0484057 AURKB 1.8 0.0006828 0.0149443 ADCY3 1.5 0.0039017 0.0488856 TUBA3D 1.7 0.0006902 0.015077 ZCCHC8 1.4 0.0039113 0.0489822 FLJ45032 1.3 0.0006906 0.015077 MAMLD1 1.8 0.0039166 0.0490008 PIGO 1.3 0.000697 0.0151697 HNRNPA1 1.5 0.0039271 0.0491088 TET1 1.5 0.0006972 0.0151697 SH2B3 1.6 0.0039373 0.0491547 C6ORF115 1.5 0.0007047 0.0152566 ARRB1 1.3 0.003938 0.0491547 GLO1 1.7 0.0007103 0.015365 HSPA2 2.1 0.0039383 0.0491547 MORF4 1.2 0.0007133 0.0154172 C19ORF59 1.7 0.003943 0.0491659 MIMT1 1.4 0.0007144 0.0154216 FOXRED2 1.3 0.0039598 0.0493046 SNORD100 1.7 0.0007239 0.0155763 SSB 1.5 0.0039714 0.0493785 350

TRERF1 1.5 0.0007394 0.0158502 DEAF1 1.4 0.003975 0.0493987 RFC5 1.5 0.000747 0.0159997 HIST1H1B 3.1 0.0039896 0.0494856 GPR83 4.1 0.0007482 0.0160112 ADM 3.2 0.0040197 0.0498055 C16ORF33 1.6 0.0007513 0.0160648 GPI 1.5 0.0040267 0.0498514 ZNF695 1.4 0.0007545 0.0161207 LOC284422 1.8 0.0040343 0.0499122 HS.334272 1.6 0.0007647 0.0163242 KCNJ2 1.5 0.0040412 0.0499596 HIST2H2A A4 1.6 0.0007704 0.0164319 LAT1-3TM 1.6 0.0040479 0.0499706 NOLC1 1.6 0.0007754 0.0165253

351

Table B.2. List of the 1267 genes that were significantly down-regulated in the ALL-67 xenograft panel in comparison to the corresponding patient sample. The cut-off for significance was set at FDR <0.05. absFC, absolute fold change.

abs abs Gene P.Value Q Value Gene P.Value Q Value FC FC HBB 259.3 < 2.2e-16 6.695E-14 GCLM 2.0 0.00031 0.0086 HBA1 198.0 < 2.2e-16 6.695E-14 LOC401640 1.6 0.00031 0.0086 HBG1 76.5 < 2.2e-16 6.695E-14 SH2D1B 1.4 0.00031 0.0086 CA1 22.1 < 2.2e-16 3.855E-13 DPH5 1.3 0.00031 0.0087 HBG2 65.8 < 2.2e-16 3.855E-13 NCRNA00085 1.9 0.00031 0.0087 GNLY 9.8 1.635E-15 7.09E-12 NFE2 2.4 0.00031 0.0087 GZMH 8.9 2.642E-15 8.847E-12 SGSM1 2.0 0.00031 0.0087 HBM 18.9 2.72E-15 8.847E-12 MT1X 1.8 0.00032 0.0087 AHSP 15.9 1.477E-14 4.271E-11 OSBPL2 1.4 0.00032 0.0087 GZMB 11.6 4.53E-14 1.179E-10 SH2D1A 2.0 0.00032 0.0087 CCL5 14.9 6.502E-14 1.538E-10 SLC25A43 1.6 0.00032 0.0088 PRF1 4.7 7.841E-14 1.7E-10 ELA1 1.3 0.00032 0.0088 CD8A 8.2 8.806E-14 1.763E-10 N6AMT1 1.4 0.00032 0.0088 CD3D 8.5 1.069E-13 1.988E-10 SSH2 1.8 0.00032 0.0088 LOC1001311 64 7.2 1.376E-13 2.387E-10 NCK1 1.5 0.00032 0.0088 HS.534427 11.1 5.564E-13 9.049E-10 ZNF790 1.3 0.00032 0.0088 HBD 49.8 5.966E-13 9.133E-10 CRK 1.5 0.00032 0.0088 KLRB1 13.0 7.449E-13 1.077E-09 ZFP3 1.5 0.00032 0.0088 IL2RB 4.3 9.581E-13 1.312E-09 AHCYL1 1.5 0.00032 0.0088 TMEM100 16.3 1.127E-12 1.467E-09 ASB1 1.9 0.00033 0.0089 SPOCK2 5.3 1.913E-12 2.252E-09 SEC14L1 1.4 0.00033 0.0089 CD7 4.0 1.951E-12 2.252E-09 FBLN2 1.5 0.00033 0.0089 SLAMF6 5.8 1.991E-12 2.252E-09 FCER1A 1.4 0.00033 0.0089 CD2 4.7 2.195E-12 2.38E-09 ZNF337 1.5 0.00033 0.0089 IGJ 6.5 3.964E-12 4.126E-09 CYP4V2 1.8 0.00033 0.0090 ZBTB16 6.6 5.291E-12 5.296E-09 LOC100129657 1.8 0.00033 0.0090 CD6 3.7 8.353E-12 8.051E-09 CX3CR1 2.3 0.00034 0.0090 ITK 5.9 1.054E-11 9.797E-09 LATS2 1.4 0.00034 0.0091 EPB42 3.4 1.096E-11 9.839E-09 MRPS31 1.5 0.00034 0.0091 SKAP1 3.8 1.204E-11 1.044E-08 HS.145049 1.6 0.00034 0.0091 ANXA1 6.5 1.326E-11 1.113E-08 FAIM3 2.5 0.00034 0.0091 TSPAN32 4.5 1.402E-11 1.14E-08 HS.534439 1.6 0.00034 0.0091 CD96 3.3 1.97E-11 1.553E-08 HMBS 1.6 0.00034 0.0092 S1PR5 2.8 2.272E-11 1.739E-08 SETBP1 2.6 0.00035 0.0093 C5ORF62 8.3 2.676E-11 1.99E-08 OMA1 1.5 0.00035 0.0093 SLC4A1 4.2 4.353E-11 3.146E-08 RAGE 1.3 0.00035 0.0093 KLRD1 4.7 4.542E-11 3.194E-08 LOC100129387 1.4 0.00035 0.0093 CCR7 6.0 5.696E-11 3.901E-08 CHURC1 1.5 0.00035 0.0093 SLC2A7 3.2 6.429E-11 4.29E-08 KRT1 1.3 0.00035 0.0093 FGFBP2 5.5 8.198E-11 5.334E-08 LOC346887 1.4 0.00036 0.0093 LOC647506 4.3 8.733E-11 5.543E-08 C2CD2 1.4 0.00036 0.0093 ZNF318 6.6 9.298E-11 5.761E-08 MBNL2 1.8 0.00036 0.0095 KLF9 5.1 1.018E-10 6.161E-08 C1ORF25 1.4 0.00036 0.0095 FLJ30428 3.2 1.326E-10 7.843E-08 FBXO6 1.5 0.00037 0.0095 GYPE 4.1 1.374E-10 7.943E-08 FLJ45244 1.4 0.00037 0.0096 CD36 3.0 1.514E-10 8.567E-08 CTLA4 1.4 0.00037 0.0096 SLA 5.8 1.775E-10 9.829E-08 ARID1A 1.5 0.00037 0.0096

352

IL32 3.6 1.849E-10 1.003E-07 HS.4988 1.6 0.00037 0.0096 CST7 2.7 1.956E-10 1.039E-07 LOC653778 2.1 0.00037 0.0096 S100A10 4.1 2.094E-10 0.0000001 CEP135 1.4 0.00038 0.0098 LOC654053 5.5 2.141E-10 1.093E-07 BTG1 1.8 0.00039 0.0099 KLRF1 3.5 2.343E-10 1.173E-07 LOC100129148 1.4 0.00039 0.0099 HOPX 3.3 2.71E-10 1.331E-07 CREB5 2.5 0.00039 0.0099 RHAG 3.5 2.809E-10 1.354E-07 CD247 3.4 0.00040 0.0101 IL10RA 2.5 3.9E-10 1.845E-07 PDDC1 1.4 0.00040 0.0102 CLN8 3.0 4.357E-10 2.025E-07 EPB41 1.5 0.00040 0.0103 RUNX3 5.8 5.13E-10 2.342E-07 USP6 1.4 0.00041 0.0104 C10ORF54 3.6 5.233E-10 2.348E-07 LOC441775 1.5 0.00041 0.0104 FCER1G 3.0 6.982E-10 0.0000003 GLS 1.8 0.00041 0.0104 CLEC4E 5.5 7.809E-10 3.387E-07 C2ORF64 1.9 0.00041 0.0105 GVIN1 5.5 8.14E-10 3.444E-07 LOC400027 1.4 0.00041 0.0105 MGC4677 3.2 8.206E-10 3.444E-07 LOC728139 1.4 0.00042 0.0105 STAT4 3.1 8.647E-10 3.572E-07 HS.155736 1.4 0.00042 0.0106 EDNRB 2.9 9.379E-10 3.813E-07 HS.536748 1.7 0.00042 0.0107 FCRL3 2.9 1.181E-09 0.0000004 SP110 1.5 0.00043 0.0107 DNAJA4 2.5 1.182E-09 0.0000004 ZNF271 1.4 0.00043 0.0108 HAS3 2.8 1.329E-09 5.162E-07 LAG3 1.6 0.00043 0.0108 LYZ 4.1 1.383E-09 5.294E-07 LOC284757 1.4 0.00044 0.0109 P2RY14 6.1 1.508E-09 5.686E-07 BTN3A3 1.5 0.00044 0.0110 GYPA 2.2 1.57E-09 5.836E-07 ANKS1A 1.3 0.00044 0.0110 GYPB 4.4 1.778E-09 6.515E-07 C1ORF172 1.7 0.00045 0.0111 TYROBP 3.7 2.436E-09 8.803E-07 UTS2D 1.4 0.00045 0.0111 ALAS2 8.1 2.816E-09 0.0000010 CTSB 2.3 0.00045 0.0111 TXK 2.1 3.832E-09 0.0000013 KCNA5 4.8 0.00045 0.0112 MTUS1 2.6 4.381E-09 0.0000015 ODZ4 2.5 0.00045 0.0112 HBQ1 2.2 5.179E-09 0.0000017 FADD 1.4 0.00045 0.0112 POLR3GL 2.1 6.069E-09 0.0000020 CNKSR3 1.5 0.00046 0.0114 CCL17 3.3 6.141E-09 0.0000020 C11ORF92 1.6 0.00046 0.0114 SRGN 8.5 6.543E-09 0.0000021 ZBTB42 1.6 0.00047 0.0115 STAG3 4.4 6.698E-09 0.0000021 CPOX 1.6 0.00047 0.0116 CCDC146 2.7 6.802E-09 0.0000021 C1ORF83 1.4 0.00048 0.0117 OSBP2 2.4 6.824E-09 0.0000021 ACP5 1.4 0.00048 0.0117 SLC31A2 3.3 6.889E-09 0.0000021 KIAA0247 1.7 0.00049 0.0118 CLIC3 2.8 7.553E-09 0.0000023 GGNBP2 1.4 0.00049 0.0119 CD160 2.2 7.782E-09 0.0000023 RHBDF2 2.4 0.00049 0.0119 TRIM10 2.4 8.281E-09 0.0000025 COMMD6 1.6 0.00049 0.0119 WDR49 2.2 9.83E-09 0.0000029 RPL14L 1.6 0.00049 0.0120 CD3G 2.1 9.941E-09 0.0000029 LOC389634 1.5 0.00050 0.0120 SESN1 7.2 1.194E-08 0.0000034 CYBA 1.7 0.00050 0.0121 FKBP5 5.3 1.241E-08 0.0000035 ZNF434 1.5 0.00051 0.0122 LOC642113 4.2 1.323E-08 0.0000037 MFNG 1.4 0.00051 0.0122 IFNG 2.3 1.36E-08 0.0000038 GCLC 1.5 0.00051 0.0122 LONRF1 4.1 1.405E-08 0.0000038 TPK1 1.3 0.00051 0.0122 LOC653337 3.8 1.413E-08 0.0000038 LOC440957 1.5 0.00051 0.0122 HS.536451 1.9 1.639E-08 0.0000043 ARV1 1.6 0.00051 0.0122 TNFSF8 2.4 1.967E-08 0.0000051 HOXB2 1.4 0.00051 0.0122 NT5E 5.8 1.977E-08 0.0000051 ACRC 1.5 0.00052 0.0124 MCOLN2 2.1 2.093E-08 0.0000054 TSPAN14 2.0 0.00052 0.0124 MYL4 2.2 2.224E-08 0.0000057 CD93 2.1 0.00053 0.0125 S100A9 2.9 2.829E-08 0.0000072 SCGB3A1 1.3 0.00053 0.0125 PIK3IP1 4.6 2.987E-08 0.0000075 LOC100128936 1.4 0.00053 0.0126

353

FGR 3.2 3.24E-08 0.0000081 PRDM2 1.5 0.00053 0.0126 DUSP23 2.2 3.341E-08 0.0000082 SPTA1 1.9 0.00053 0.0126 LOC1001343 31 1.9 4.131E-08 0.0000100 TP53INP1 3.3 0.00053 0.0126 MAL 3.4 4.177E-08 0.0000100 LOC647285 1.7 0.00053 0.0126 CD53 2.9 4.62E-08 0.0000110 C1QB 1.3 0.00054 0.0126 C7ORF41 2.0 4.679E-08 0.0000110 APPBP2 1.8 0.00055 0.0128 LRIG1 5.2 4.713E-08 0.0000110 LOC728855 1.8 0.00055 0.0129 KCNK3 3.6 4.99E-08 0.0000115 LOC200030 1.4 0.00056 0.0130 LOC652694 2.8 5.131E-08 0.0000118 BLK 2.0 0.00056 0.0130 PCTP 3.4 5.232E-08 0.0000119 ZNF154 1.3 0.00056 0.0131 CUGBP2 3.3 5.453E-08 0.0000123 SDHAP2 1.7 0.00056 0.0131 TMEM2 4.1 6.233E-08 0.0000139 CROCCL2 1.3 0.00056 0.0131 RHCE 1.8 7.192E-08 0.0000158 ACAD10 1.7 0.00057 0.0132 HS.570988 4.3 7.249E-08 0.0000158 KIAA0430 1.4 0.00057 0.0132 CA2 4.2 7.486E-08 0.0000162 LILRA3 1.4 0.00058 0.0133 FCRLA 3.3 8.281E-08 0.0000178 NPHP3 1.7 0.00058 0.0133 LOC652493 3.4 8.51E-08 0.0000181 HS.14706 1.3 0.00058 0.0134 FLJ13197 2.3 8.797E-08 0.0000185 ZNF518B 1.5 0.00059 0.0135 GALNT11 1.9 9.912E-08 0.0000203 SLC4A7 2.5 0.00059 0.0135 ISG20 3.3 1.049E-07 0.0000212 TNFRSF13B 1.4 0.00059 0.0135 CDC42EP3 2.4 1.058E-07 0.0000212 RNASE2 1.5 0.00059 0.0135 FAM171A1 5.2 1.069E-07 0.000021 SOCS2 1.9 0.00059 0.0135 SERPINI2 2.5 1.069E-07 0.0000212 CALCOCO1 1.9 0.00060 0.0137 RFESD 2.0 1.287E-07 0.0000253 C2ORF68 2.3 0.00060 0.0137 GRAMD1C 1.7 1.294E-07 0.0000253 MORC1 1.6 0.00060 0.0137 CD44 5.8 1.377E-07 0.0000267 LOC285053 1.6 0.00061 0.0138 TESC 2.1 1.426E-07 0.00075 RBM43 1.6 0.00061 0.0138 IFITM1 4.0 1.447E-07 0.0000277 TMEM60 1.4 0.00061 0.0138 GCA 2.5 1.521E-07 0.000028 CYTH1 1.5 0.00061 0.0139 LOC730024 3.6 0.000000154 0.0000288 AFTPH 1.5 0.00061 0.0139 EPHA4 2.0 1.547E-07 0.0000288 SNX25 1.6 0.00062 0.0139 S100A8 2.8 0.000000155 0.0000288 LIPT1 1.4 0.00062 0.0140 LIMCH1 1.7 1.574E-07 0.0000290 FAU 1.5 0.00062 0.0140 SYTL2 2.7 1.601E-07 0.0000293 PPM1M 1.6 0.00062 0.0140 SH3TC1 2.7 0.000000165 0.0000299 ZNF540 1.3 0.00062 0.0140 TCL6 4.5 1.659E-07 0.0000299 LOC339192 1.3 0.00063 0.0141 FAM46A 2.0 1.736E-07 0.0000311 LOC727826 1.6 0.00063 0.0141 ZNF594 3.2 1.847E-07 0.0000329 WBP4 1.6 0.00063 0.0141 SLC16A3 1.8 2.033E-07 0.0000355 ATF7IP2 1.3 0.00064 0.0142 FLJ39639 2.2 2.272E-07 0.0000389 HERC6 1.3 0.00064 0.0143 FLJ11795 1.7 2.571E-07 0.0000437 MGC13057 1.3 0.00065 0.0144 LOC642035 2.3 2.929E-07 0.0000492 FLJ20125 1.5 0.00065 0.0145 HS.131041 1.8 0.000000296 0.0000492 GBP2 2.7 0.00066 0.0145 ADARB1 2.1 3.027E-07 0.0000498 CBR4 1.8 0.00066 0.0145 LOXL4 2.4 3.113E-07 0.0000509 BACH2 2.9 0.00066 0.0146 KLRC3 1.7 3.172E-07 0.0000515 ECGF1 1.3 0.00066 0.0146 TMEM43 1.6 3.273E-07 0.0000529 C1ORF71 1.8 0.00066 0.0146 LOC391044 2.0 3.457E-07 0.0000555 HIVEP2 1.3 0.00067 0.0147 NKG7 8.1 3.564E-07 0.0000568 CNOT7 1.5 0.00068 0.0148 PURB 2.0 4.007E-07 0.0000633 TOMM7 1.7 0.00068 0.0149 LPIN1 2.4 4.017E-07 0.0000633 AOAH 1.4 0.00068 0.0149 HAL 1.9 4.552E-07 0.0000709 RBPMS2 1.3 0.00068 0.0150 ADAM8 1.7 0.000000465 0.0000720 POLR1D 1.4 0.00069 0.0151 ANKRD28 2.4 0.000000471 0.0000725 LOC347544 1.8 0.00069 0.0151 354

CTSW 2.0 4.776E-07 0.0000731 SEMA4D 1.3 0.00070 0.0152 SAMD3 2.3 4.822E-07 0.000073 LOC729355 1.3 0.00070 0.0152 LOC391833 2.0 4.914E-07 0.0000743 STARD13 1.4 0.00070 0.0152 LOC647450 2.5 5.149E-07 0.0000746 LOC286444 1.4 0.00070 0.0152 BMF 4.8 5.203E-07 0.0000778 LOC649946 1.7 0.00070 0.0153 PRDM1 3.7 5.403E-07 0.0000803 STEAP3 1.3 0.00071 0.0154 ANKRD55 1.6 5.612E-07 0.0000825 IL1RAP 1.5 0.00072 0.0156 AHNAK 6.8 5.685E-07 0.0000831 HS.7572 1.9 0.00072 0.0156 UNC5CL 2.0 6.039E-07 0.0000873 PGS1 1.5 0.00072 0.0156 CCDC6 1.9 0.00000064 0.0000920 GLIPR1 1.6 0.00073 0.0156 SP140 1.7 6.687E-07 0.0000956 BIVM 1.3 0.00073 0.0158 MBP 3.4 7.085E-07 0.0001007 PLEKHG4 1.4 0.00074 0.0159 CPVL 1.8 7.304E-07 0.0001031 PSTPIP1 1.4 0.00078 0.0165 HS.566482 1.9 0.000000733 0.0001031 KIR2DL4 1.6 0.00078 0.0166 FAM179A 1.7 8.058E-07 0.0001127 CORO1A 1.9 0.00079 0.0168 ITGB2 4.1 8.128E-07 0.0001131 UTS2 1.4 0.00080 0.0168 PATL2 1.9 8.182E-07 0.0001133 MARCH7 1.7 0.00080 0.0169 C14ORF139 2.4 8.435E-07 0.0001161 PTPN22 1.3 0.00080 0.0169 XK 1.7 8.665E-07 0.0001179 SLC7A6 1.4 0.00081 0.0170 ELANE 1.6 8.698E-07 0.0001179 EMP3 2.6 0.00081 0.0170 ITGB7 2.4 0.000000959 0.0001292 FNBP1 1.5 0.00082 0.0172 ZNF35 1.5 9.665E-07 0.0001292 C19ORF18 1.4 0.00082 0.0172 CCL4L2 3.6 0.000001006 0.0001329 LOC643531 1.5 0.00082 0.0172 C6ORF190 1.7 0.000001033 0.0001358 FLJ20444 1.3 0.00083 0.0174 SEC14L4 1.7 0.000001052 0.0001375 ZNF404 1.9 0.00084 0.0174 HS.528873 2.0 0.000001171 0.0001474 HS.365998 1.3 0.00084 0.0175 TLE1 2.6 0.000001208 0.0001504 SFRS7 1.5 0.00085 0.0176 MATK 2.0 0.000001223 0.0001508 VN1R1 1.5 0.00085 0.0176 LOC338799 2.3 0.000001236 0.0001511 MRC1L1 1.3 0.00085 0.0176 ARL4A 1.7 0.000001236 0.0001511 LYST 1.4 0.00086 0.0177 JAZF1 2.1 0.000001279 0.0001555 ZNF671 1.6 0.00086 0.0177 MT1F 1.7 0.000001299 0.0001572 ZNF181 1.6 0.00086 0.0177 GZMK 4.3 0.000001327 0.0001599 FAM120A 1.5 0.00086 0.0177 MPO 2.2 0.000001448 0.0001705 SETD6 1.3 0.00086 0.0177 CA6 1.8 0.000001476 0.0001731 SCAND2 1.3 0.00087 0.0178 ITGAX 1.5 0.00000161 0.0001878 LOC644315 1.9 0.00088 0.0179 CH25H 1.8 0.000001641 0.0001906 SLC35D2 1.5 0.00088 0.0179 CHKA 3.0 0.000001665 0.0001925 SIGLEC10 1.4 0.00088 0.0179 WDR20 2.0 0.000001705 0.0001954 ELMO2 1.6 0.00088 0.0179 LFNG 3.1 0.000001826 0.0002066 ERP29 1.3 0.00088 0.0180 KIAA1407 1.8 0.000001846 0.0002078 BET1L 1.4 0.00088 0.0180 KEL 1.6 0.000001853 0.0002078 C7ORF25 1.4 0.00089 0.0180 CNRIP1 1.6 0.000001873 0.0002092 TMEM56 1.3 0.00089 0.0181 CEBPD 7.1 0.000001922 0.0002137 MANBA 1.5 0.00089 0.0181 FHL2 1.6 0.000001934 0.0002141 FBXO38 1.5 0.00090 0.0182 PHB 2.8 0.000002028 0.0002227 LOC730704 1.3 0.00090 0.0182 CNTNAP2 1.8 0.000002076 0.0002262 CLECL1 2.4 0.00090 0.0182 LOC651751 1.6 0.000002081 0.0002262 LOC730183 1.5 0.00090 0.0182 PRG2 1.6 0.000002086 0.0002262 WASF2 1.5 0.00091 0.0184 PXMP3 1.6 0.000002133 0.0002303 CD79B 1.7 0.00092 0.0185 GYG1 3.1 0.000002217 0.0002374 ANXA2 2.3 0.00093 0.0186 PRMT2 1.8 0.000002276 0.0002428 MUC1 1.4 0.00093 0.0186 FCN1 1.8 0.000002295 0.0002438 EPHX2 1.4 0.00093 0.0186 JSRP1 1.9 0.000002377 0.0002514 PPP1R3E 1.4 0.00093 0.0186

355

PCDH9 2.4 0.000002418 0.0002548 HS.577646 1.4 0.00093 0.0187 CD5 1.7 0.00000257 0.0002664 LOC100128060 2.0 0.00094 0.0188 ACACB 1.8 0.0000026 0.0002683 LOC554207 1.3 0.00094 0.0188 CDC42EP2 2.1 0.000002608 0.0002683 DPRXP4 1.5 0.00096 0.0190 NFATC1 1.7 0.00000263 0.0002684 FAM120B 1.5 0.00096 0.0190 NCK2 2.4 0.000002649 0.0002692 TADA2B 1.5 0.00096 0.0190 NRBP2 2.2 0.000002703 0.0002737 ZNF397 1.5 0.00096 0.0190 HS1BP3 1.6 0.000002794 0.0002818 PRKAB2 1.4 0.00097 0.0190 POLA2 2.3 0.000002805 0.0002819 FGD3 1.7 0.00097 0.0191 AMN1 1.6 0.00000301 0.0002979 C9ORF72 1.3 0.00097 0.0191 OR2B6 2.2 0.000003021 0.0002979 ZFP14 1.4 0.00097 0.0191 BTLA 1.8 0.000003064 0.0003008 ATG12 1.5 0.00098 0.0192 ZNF84 2.0 0.000003074 0.0003008 LOC649839 1.5 0.00098 0.0193 FAR1 1.7 0.000003239 0.0003157 BMP8B 1.5 0.00098 0.0193 DDR1 1.8 0.000003396 0.0003293 FMO5 1.6 0.00099 0.0194 PLEKHA2 2.5 0.000003404 0.0003293 MAP3K6 1.6 0.00099 0.0194 HS.12876 2.4 0.000003477 0.0003351 PDE7B 1.3 0.00100 0.0195 LOC144481 2.6 0.000003538 0.0003385 FAM178B 1.4 0.00100 0.0195 IFIT1L 1.7 0.00000369 0.0003491 BCR 1.4 0.00102 0.0198 LOC728820 1.9 0.000003695 0.0003491 PLEKHF1 1.4 0.00102 0.0198 ZDHHC23 2.2 0.000003702 0.0003491 CPT2 1.9 0.00102 0.0198 ZNF683 1.8 0.000003757 0.0003529 NOTCH2 1.5 0.00102 0.0198 WDR37 1.9 0.000003865 0.0003618 RBM33 1.7 0.00102 0.0198 MINPP1 1.5 0.000003881 0.000362 LOC644173 1.6 0.00103 0.0198 C12ORF35 2.6 0.00000412 0.0003789 FLJ35776 1.3 0.00103 0.0199 SLC16A9 3.3 0.000004166 0.0003818 LOC728128 1.5 0.00103 0.0199 PRKX 2.4 0.000004243 0.0003874 FRAT1 1.4 0.00104 0.0200 CD9 2.6 0.000004272 0.000388 RNASEH2C 1.3 0.00105 0.0201 C8ORF44 1.6 0.000004295 0.000388 MYO1F 1.4 0.00105 0.0202 TFEB 2.0 0.000004551 0.0004098 LOC100130000 1.3 0.00105 0.0202 LOC197135 1.6 0.000004587 0.0004116 COL9A2 1.4 0.00106 0.0202 CHPT1 1.8 0.000004689 0.0004164 FLJ32810 1.7 0.00106 0.0202 FAM46C 2.4 0.00000479 0.0004225 LOC401431 1.5 0.00106 0.0202 IRGM 1.8 0.000004985 0.0004367 C14ORF132 1.8 0.00106 0.0202 LOC730278 1.8 0.000005099 0.0004453 NPC1 1.5 0.00106 0.0203 ITPRIPL2 2.4 0.000005322 0.0004616 TXLNB 1.3 0.00106 0.0203 MS4A1 1.6 0.000005575 0.0004743 ADCY4 1.4 0.00106 0.0203 KLHL9 1.7 0.00000559 0.0004743 TMEM187 1.5 0.00108 0.0206 LOC645630 2.1 0.000005601 0.0004743 ANKAR 1.3 0.00109 0.0206 SMAP2 2.7 0.000005609 0.0004743 ZNF658 1.3 0.00109 0.0208 NT5DC2 2.3 0.000005809 0.0004892 TANK 1.4 0.00110 0.0208 SLC44A1 3.3 0.000005946 0.0004975 FOXO3 1.6 0.00111 0.0209 HS.123119 1.9 0.000005983 0.0004989 HS.154948 1.4 0.00111 0.0210 ZNF514 1.8 0.000006 0.0004989 PTPN1 1.6 0.00111 0.0210 CFLAR 1.9 0.000006652 0.0005498 HTATIP2 1.6 0.00111 0.0210 C2ORF63 2.0 0.000006656 0.0005498 ZSWIM1 1.4 0.00112 0.0211 CHST2 2.5 0.000006959 0.0005677 HS.537004 1.3 0.00112 0.0211 C10ORF73 2.7 0.00000702 0.0005691 LOC136143 1.6 0.00113 0.0212 TRAK2 2.3 0.000007169 0.0005776 LOC651436 1.6 0.00114 0.0215 PRAGMIN 2.0 0.000007235 0.0005811 LOC730412 1.4 0.00115 0.0215 HS.436572 1.8 0.000007355 0.0005889 FAM58A 1.4 0.00115 0.0215 LOC1001298 78 2.4 0.000007417 0.0005921 FBXW11 1.4 0.00115 0.0215 CALCOCO2 2.1 0.000007529 0.0005992 SUCLG2 1.4 0.00115 0.0215 ASXL2 1.7 0.000007686 0.0006061 NEURL1B 1.7 0.00115 0.0215 356

TMCO3 1.7 0.000007748 0.0006092 SFRS18 1.4 0.00115 0.0215 ZNF516 1.5 0.000007847 0.0006151 MAN2A2 1.4 0.00115 0.0215 DPEP2 1.8 0.000007974 0.0006231 TMOD2 1.5 0.00115 0.0215 LOC285359 1.6 0.000008203 0.0006372 POLD4 1.4 0.00115 0.0215 LILRB2 2.4 0.000008265 0.0006401 HS.443185 1.5 0.00116 0.0216 ATM 2.5 0.000008348 0.0006427 HS.254006 1.4 0.00116 0.0216 BLVRB 2.4 0.000008474 0.0006486 TMEM188 1.4 0.00117 0.0217 LOC728835 5.2 0.000008779 0.000668 KRT72 1.4 0.00117 0.0217 IFNGR1 1.8 0.000009012 0.0006838 SPG11 1.8 0.00117 0.0217 HMHB1 2.5 0.000009408 0.0007117 COQ5 1.3 0.00117 0.0217 DARC 1.7 0.000009817 0.0007405 UNKL 1.4 0.00117 0.0217 TXNIP 3.5 0.00001015 0.0007614 ZNF550 1.4 0.00118 0.0218 SLFNL1 1.7 0.00001015 0.0007614 CBLN3 1.5 0.00119 0.0220 LGALS3 1.5 0.00001069 0.0007948 HS.189987 1.4 0.00119 0.0220 LAT 1.7 0.00001098 0.0008067 FUT4 1.5 0.00120 0.0221 UBASH3B 2.1 0.00001103 0.0008067 LOC727865 1.7 0.00121 0.0223 BEX2 1.8 0.00001104 0.0008067 C2ORF24 1.3 0.00121 0.0223 C17ORF48 2.0 0.00001141 0.0008296 ZNF319 1.5 0.00122 0.0223 MIR155HG 2.2 0.00001181 0.0008536 SNORA74B 1.3 0.00122 0.0223 IFI27 1.4 0.00001196 0.0008621 C1ORF101 1.4 0.00124 0.0226 C1ORF104 1.7 0.00001232 0.0008854 LOC339290 1.7 0.00125 0.0227 ERP27 1.9 0.00001251 0.0008944 LOC643505 1.2 0.00125 0.0227 FAM159A 1.5 0.00001256 0.0008944 C9ORF152 5.4 0.00125 0.0228 LOC132241 2.4 0.00001258 0.0008944 TNS3 1.4 0.00126 0.0229 ZNF831 2.1 0.00001278 0.0009063 TAX1BP1 1.4 0.00126 0.0229 TMOD1 1.5 0.00001306 0.0009211 LOC728758 1.5 0.00126 0.0229 TIPARP 1.8 0.00001306 0.0009211 LOC652651 1.9 0.00128 0.0231 ANK1 1.6 0.0000133 0.0009353 DUSP18 1.8 0.00129 0.0232 CAST 1.8 0.00001338 0.0009388 PEX11B 1.8 0.00130 0.0234 C1RL 1.7 0.00001351 0.0009448 ABR 1.6 0.00131 0.0236 HS.572538 1.7 0.00001371 0.0009566 ADAM17 1.5 0.00131 0.0236 ZNF471 1.5 0.00001391 0.000968 MARCH2 1.4 0.00132 0.0236 N4BP2L1 1.8 0.00001419 0.0009845 SH3PXD2A 1.5 0.00132 0.0236 LOC283663 1.7 0.00001441 0.0009974 UBE4B 1.4 0.00132 0.0237 IL6ST 1.7 0.00001469 0.0010089 ARL17P1 1.4 0.00133 0.0237 LOC646123 1.6 0.00001477 0.0010113 MYH9 1.4 0.00135 0.0240 TBP 1.5 0.00001507 0.0010265 CMYA5 1.4 0.00136 0.0241 SCD 1.6 0.00001564 0.0010573 STX7 1.5 0.00137 0.0243 CA5B 1.8 0.00001589 0.0010715 MPHOSPH8 1.4 0.00138 0.0244 CRIP2 1.5 0.00001601 0.0010744 HS.555252 1.4 0.00138 0.0244 HCG27 2.1 0.00001625 0.0010844 SLC41A1 1.3 0.00138 0.0245 ZDHHC14 1.6 0.00001647 0.0010961 IFI16 1.5 0.00139 0.0245 GUCA1B 1.6 0.0000166 0.0011019 HERC4 1.3 0.00142 0.0250 HS.126889 1.9 0.00001686 0.0011163 RAX2 1.6 0.00143 0.0251 HS.440088 1.9 0.00001704 0.0011213 NLRC5 1.6 0.00144 0.0252 GH1 1.7 0.00001706 0.0011213 HS.143408 1.5 0.00144 0.0252 LOC651957 1.6 0.00001729 0.0011311 ANPEP 1.4 0.00144 0.0252 LOC727970 1.6 0.00001734 0.0011311 KCNH2 1.3 0.00145 0.0253 ABLIM2 1.6 0.00001792 0.0011644 KRCC1 1.5 0.00145 0.0254 LGALS4 1.5 0.00001817 0.0011761 BCDIN3D 1.7 0.00146 0.0255 NGFRAP1 1.6 0.00001834 0.0011815 TMEM220 1.5 0.00147 0.0255 RCAN1 4.8 0.00001849 0.0011882 LOC389404 1.5 0.00148 0.0257 CAPN7 1.5 0.00001865 0.0011942 USP42 1.4 0.00149 0.0258 TRIM22 1.9 0.00001872 0.0011942 RAB9P1 1.3 0.00149 0.0258

357

HS.445414 1.7 0.00001877 0.0011942 HS.400256 1.5 0.00150 0.0260 LOC649639 1.6 0.00001899 0.0012032 DHX58 1.4 0.00151 0.0260 PARP4 1.8 0.000019 0.0012032 LOC654350 1.7 0.00151 0.0261 UROD 1.4 0.00001913 0.0012083 HS.405877 1.4 0.00152 0.0261 SNX3 1.5 0.00001939 0.0012217 TCTA 1.5 0.00153 0.0263 RNF157 1.6 0.00001953 0.0012257 CD40LG 1.3 0.00154 0.0263 ETS2 2.9 0.00001955 0.0012257 CNOT4 1.3 0.00155 0.0265 TRIM4 1.9 0.00002009 0.0012569 CLK4 1.7 0.00155 0.0265 ZNF419 1.5 0.00002062 0.0012816 SH2D4B 1.4 0.00155 0.0265 C9ORF164 1.7 0.00002075 0.0012859 C9ORF123 1.5 0.00155 0.0265 PFKFB2 3.2 0.00002154 0.0013231 RASGRP2 1.5 0.00155 0.0265 HS.530359 1.6 0.00002156 0.0013231 DEF6 1.6 0.00155 0.0265 VPS13D 1.4 0.00002211 0.0013536 LOC606724 1.7 0.00156 0.0265 IDI1 2.1 0.00002236 0.0013575 SLC7A9 1.3 0.00156 0.0266 SYTL3 1.6 0.00002238 0.0013575 MORN1 1.3 0.00157 0.0266 SLC44A2 1.7 0.00002278 0.0013783 SNX1 1.5 0.00157 0.0267 SOX7 4.0 0.00002343 0.0014147 FAM84B 1.3 0.00158 0.0269 INPP1 2.4 0.00002361 0.0014189 RNF144B 1.4 0.00159 0.0269 IL13RA1 3.2 0.00002414 0.0014426 NSUN2 1.4 0.00159 0.0270 CD3E 1.5 0.00002429 0.0014466 ZNFX1 1.4 0.00160 0.0271 RRAS 1.6 0.00002478 0.0014691 NFE2L2 1.7 0.00160 0.0271 PLEKHM1 1.6 0.00002511 0.0014836 EEF1G 1.8 0.00160 0.0271 LOC441268 2.3 0.00002546 0.0014985 ANXA2P1 1.7 0.00161 0.0272 KLF12 1.7 0.00002551 0.0014985 CSPP1 1.3 0.00161 0.0272 MIAT 1.5 0.00002559 0.0015001 FCRL2 2.1 0.00163 0.0274 MRI1 1.8 0.00002645 0.0015467 HCP5 1.3 0.00164 0.0276 DENND3 2.5 0.00002688 0.001564 HS.377257 1.3 0.00164 0.0276 MTMR9 1.8 0.00002776 0.0016051 USP9X 1.4 0.00166 0.0278 TNIK 1.4 0.00002791 0.0016102 FAM62B 1.4 0.00168 0.0280 LOC729264 1.9 0.00002808 0.0016166 UTP6 1.3 0.00170 0.0282 MMP7 1.5 0.00002891 0.001661 C5ORF51 1.5 0.00170 0.0282 NISCH 1.9 0.000029 0.0016614 CLINT1 2.1 0.00170 0.0283 HCG26 1.7 0.00002926 0.0016697 SLC2A5 2.5 0.00171 0.0283 PSTPIP2 1.7 0.0000294 0.0016742 LOC646093 1.6 0.00171 0.0284 LOC388969 1.5 0.00002968 0.0016827 TSPYL3 1.5 0.00172 0.0285 CYB561D1 1.9 0.00002999 0.0016954 LOC158345 1.6 0.00172 0.0285 E2F5 1.5 0.00003017 0.0016996 C3ORF19 1.3 0.00173 0.0286 LOC1001331 63 1.9 0.0000312 0.0017458 PPAPDC1B 1.4 0.00174 0.0286 RAB7L1 1.8 0.00003163 0.0017599 LOC731366 1.4 0.00175 0.0288 XPO6 1.6 0.00003165 0.0017599 HS.560343 1.3 0.00175 0.0288 NEIL2 1.6 0.00003165 0.0017599 TMEM77 1.4 0.00175 0.0288 CYFIP1 1.9 0.00003201 0.0017723 P2RX5 1.3 0.00176 0.0288 HS.121070 1.7 0.00003258 0.0018003 IKZF1 2.4 0.00177 0.0289 S100A11 1.4 0.00003285 0.0018109 GALK1 1.5 0.00177 0.0289 ALDH1A1 2.2 0.00003301 0.0018159 CTSG 1.3 0.00178 0.0291 HS.203201 1.8 0.00003389 0.0018567 LOC100131655 1.4 0.00178 0.0291 TRAF3IP3 1.8 0.0000344 0.0018805 CP 1.3 0.00179 0.0291 KLF13 1.7 0.00003451 0.0018825 LOC100131675 1.3 0.00179 0.0291 HS.279842 1.6 0.00003611 0.001966 HS.406790 1.4 0.00179 0.0292 ZNF526 1.7 0.00003653 0.0019844 SETD2 1.3 0.00180 0.0292 PWWP2B 2.1 0.00003672 0.0019852 SFRS17A 1.4 0.00181 0.0293 HBBP1 1.6 0.00003692 0.0019852 FASLG 1.2 0.00182 0.0294 APOF 1.4 0.00003736 0.0020003 ACVR2A 1.3 0.00182 0.0294 ZNF562 2.0 0.0000376 0.0020094 SPINK2 1.2 0.00183 0.0295 358

VNN2 1.5 0.0000378 0.0020155 C7ORF38 1.4 0.00184 0.0296 ZNF564 1.4 0.00003849 0.0020397 SH2B1 1.5 0.00184 0.0296 ZHX3 1.7 0.0000404 0.0021069 CACNA1H 1.2 0.00184 0.0296 ADCY7 1.5 0.00004058 0.0021119 KIF26A 1.6 0.00184 0.0296 HEMGN 2.4 0.00004072 0.0021153 RAB39B 1.3 0.00184 0.0297 TNFRSF25 2.2 0.00004085 0.0021178 AKNA 1.6 0.00185 0.0298 RALGPS1 1.8 0.00004113 0.002128 AMT 1.5 0.00186 0.0298 TP53TG3 1.8 0.00004187 0.0021533 LOC642299 1.8 0.00186 0.0298 PPP1R10 1.8 0.00004232 0.0021678 HS.534061 2.2 0.00186 0.0299 HS.407903 2.7 0.00004245 0.0021704 LAMA5 1.3 0.00187 0.0299 NBEA 1.5 0.00004292 0.0021859 C11ORF46 1.5 0.00187 0.0299 GPR68 1.7 0.00004389 0.0022308 GEMIN6 1.4 0.00187 0.0299 LOC648980 1.8 0.00004399 0.0022317 HOOK1 1.3 0.00189 0.0301 HSPC157 1.6 0.00004414 0.0022345 DBT 1.4 0.00190 0.0302 BTN3A2 1.6 0.00004429 0.0022348 UST 1.2 0.00190 0.0303 HBZ 1.5 0.00004431 0.0022348 LOC729810 1.3 0.00190 0.0303 MYF6 1.4 0.00004517 0.0022706 LOC653197 1.3 0.00191 0.0303 EHD3 1.5 0.0000452 0.0022706 SEPP1 1.3 0.00193 0.0305 HS.314414 1.4 0.00004545 0.0022789 MYLK4 1.4 0.00193 0.0305 RAB9A 1.6 0.00004612 0.0022947 ST3GAL6 1.5 0.00193 0.0305 TBC1D14 1.9 0.00004726 0.0023418 PWWP2 1.3 0.00193 0.0306 MYRIP 2.2 0.00004742 0.0023418 ZNF696 1.4 0.00196 0.0310 LY6E 1.8 0.00004742 0.0023418 LOC100134584 1.7 0.00198 0.0312 LOC644128 2.5 0.00004775 0.0023533 NARG1L 1.4 0.00199 0.0314 LOC253039 2.4 0.0000479 0.0023561 ZNF79 1.3 0.00200 0.0315 CDH1 1.4 0.00004799 0.0023561 MANBAL 1.3 0.00201 0.0316 AGPAT4 1.5 0.00004808 0.0023561 TTC15 1.3 0.00201 0.0316 G6PD 1.8 0.00004836 0.0023623 NMU 1.4 0.00201 0.0316 NRTN 2.1 0.00004842 0.0023623 ZNF274 1.3 0.00202 0.0316 LOC649923 1.5 0.00004847 0.0023623 HELQ 1.4 0.00202 0.0316 LOC648405 2.3 0.00004895 0.0023717 VCAN 1.3 0.00203 0.0317 RREB1 1.5 0.00004899 0.0023717 LEPROTL1 1.4 0.00203 0.0317 C5ORF41 1.8 0.00004902 0.0023717 ELL2 1.3 0.00204 0.0318 NLK 1.9 0.00004912 0.0023717 HNRPLL 1.3 0.00204 0.0318 HS.570229 1.5 0.00004912 0.0023717 MRPS18B 1.7 0.00205 0.0320 SLC40A1 1.4 0.0000496 0.0023903 LOC100129599 1.6 0.00207 0.0322 EPSTI1 2.0 0.00005184 0.0024937 LOC100130070 1.5 0.00207 0.0322 LOC390354 2.2 0.00005279 0.0025208 C6ORF85 2.1 0.00208 0.0322 SLAMF1 1.4 0.00005306 0.002529 SH3KBP1 1.7 0.00208 0.0322 DLGAP2 2.1 0.00005337 0.0025388 ZNF264 1.5 0.00209 0.0323 C10ORF104 1.9 0.00005437 0.0025633 CCDC121 1.5 0.00210 0.0324 YPEL4 1.5 0.00005486 0.0025818 HS.202577 1.5 0.00210 0.0325 EPB49 1.4 0.00005558 0.002606 PFAAP5 1.5 0.00211 0.0325 TUBA4A 1.9 0.0000566 0.0026444 ANKIB1 1.6 0.00211 0.0325 ANKRA2 1.9 0.00005696 0.0026562 C8ORF51 1.3 0.00212 0.0326 RORA 1.4 0.00005767 0.0026845 MEIS3P1 1.5 0.00213 0.0326 CD151 1.6 0.00005791 0.002691 C12ORF61 1.2 0.00213 0.0326 FZD8 1.7 0.00005932 0.0027371 ATP6V0A1 1.7 0.00213 0.0326 ZFC3H1 1.9 0.00005973 0.0027461 C14ORF148 1.2 0.00213 0.0327 FLJ39653 2.1 0.00006083 0.0027917 PLCL2 1.4 0.00213 0.0327 MAN1C1 1.3 0.00006249 0.0028581 LOC285900 1.7 0.00215 0.0329 STRADB 1.8 0.00006249 0.0028581 GCET2 1.5 0.00216 0.0330 PSCD1 1.5 0.00006346 0.0028921 LOC730187 1.5 0.00217 0.0331 PFKFB4 2.2 0.00006415 0.0029187 LOC654103 2.2 0.00217 0.0331

359

MTIF3 1.6 0.00006446 0.0029202 LOC648364 1.4 0.00217 0.0331 ITGAE 1.8 0.00006505 0.0029343 PIK3R5 1.4 0.00218 0.0332 NCRNA0015 2 1.3 0.00006506 0.0029343 AKAP11 1.6 0.00219 0.0333 STS-1 2.1 0.00006542 0.0029454 GSN 2.2 0.00219 0.0333 C11ORF21 1.5 0.00006562 0.0029493 ZNF774 1.3 0.00219 0.0333 GALNT4 1.5 0.00006737 0.0030125 PRX 1.5 0.00219 0.0333 HS.534680 1.8 0.00006849 0.0030572 ATG5 1.5 0.00220 0.0334 CFD 1.6 0.00007066 0.0031487 CTDSP2 1.5 0.00221 0.0335 FLJ25006 1.8 0.00007136 0.0031697 DPYSL2 1.5 0.00221 0.0336 C7ORF13 1.7 0.00007138 0.0031697 HS.158923 1.5 0.00222 0.0336 RINL 1.9 0.0000715 0.0031697 ZNF688 1.4 0.00222 0.0336 RNASE6 2.0 0.00007186 0.0031802 CUL3 1.4 0.00222 0.0336 LOC389599 1.6 0.00007272 0.0032046 ZFP92 1.4 0.00222 0.0336 DIDO1 2.1 0.00007278 0.0032046 FGD2 2.0 0.00223 0.0337 HDC 1.4 0.00007582 0.0033217 RPS9 1.4 0.00223 0.0337 TARSL2 1.7 0.00007628 0.003325 CAPRIN2 1.3 0.00223 0.0337 ATPBD1B 1.7 0.0000763 0.003325 THBS2 1.2 0.00225 0.0339 WDR5B 1.7 0.00007681 0.0033372 CATSPER2 2.0 0.00226 0.0340 HS.549729 1.6 0.00007729 0.0033521 IFI44L 1.7 0.00227 0.0341 NICN1 1.6 0.00007862 0.0033815 KLHL6 1.4 0.00228 0.0341 MRFAP1L1 1.6 0.00007902 0.0033934 POU4F1 1.9 0.00228 0.0341 BTN3A1 1.7 0.00007966 0.0034111 LOC644979 1.3 0.00228 0.0341 LILRA2 1.8 0.00008011 0.0034232 LOC644739 2.0 0.00228 0.0342 LOC729559 1.6 0.00008051 0.0034347 C20ORF108 2.0 0.00228 0.0342 CXCR5 1.4 0.00008143 0.003459 LEPR 1.2 0.00229 0.0342 MRPS21 1.9 0.00008148 0.003459 CERKL 1.3 0.00229 0.0342 FLJ20699 1.4 0.00008192 0.0034662 LOC649292 1.4 0.00229 0.0343 ZNF137 1.5 0.00008343 0.0035073 FBXO21 1.3 0.00230 0.0343 ZFP90 1.5 0.00008426 0.003531 OXR1 1.6 0.00231 0.0344 RARA 2.0 0.00008471 0.0035384 HS.107418 1.5 0.00231 0.0344 ARHGAP24 2.0 0.00008722 0.0036366 ANKZF1 1.5 0.00232 0.0345 KCNS3 2.0 0.00008749 0.0036366 MTMR3 1.4 0.00232 0.0345 SLC23A3 1.5 0.00008753 0.0036366 RPAP2 1.4 0.00232 0.0345 TMBIM4 1.4 0.00008762 0.0036366 ITPR1 1.4 0.00232 0.0345 AFF3 2.6 0.00008861 0.00367 RAPSN 1.3 0.00232 0.0345 USP47 1.6 0.00008871 0.00367 ZNF189 1.5 0.00234 0.0346 NUAK1 1.4 0.00008934 0.0036847 RAB3GAP2 1.5 0.00234 0.0346 FAM18A 1.4 0.00008979 0.0036965 HS.354359 1.6 0.00235 0.0347 SIDT2 1.8 0.00008992 0.0036965 HS.171082 1.4 0.00235 0.0347 IRF9 2.0 0.00009037 0.0037092 ZNF720 1.4 0.00236 0.0348 TBC1D10C 1.8 0.00009413 0.0038408 KRTCAP3 1.2 0.00238 0.0349 VEZF1 1.6 0.00009421 0.0038408 KIAA1430 1.3 0.00238 0.0349 HS.568741 1.6 0.00009431 0.0038408 ATP6V1E2 1.4 0.00238 0.0349 LOC651453 1.7 0.00009628 0.003894 PHF20L1 1.3 0.00238 0.0349 SAMD9L 1.9 0.00009646 0.003894 VIL2 1.4 0.00240 0.0350 BIN2 1.4 0.00009685 0.0039014 CRIP1 2.3 0.00245 0.0356 LOC646785 1.7 0.00009792 0.0039383 SCN11A 1.3 0.00245 0.0356 CBX7 2.0 0.00009878 0.0039579 SIPA1L2 1.7 0.00246 0.0356 TTC14 1.8 0.00009882 0.0039579 SLC46A3 1.8 0.00246 0.0357 IGF2R 1.6 0.00009886 0.0039579 LOC642367 1.4 0.00248 0.0358 TNFRSF17 1.5 0.00009962 0.0039824 RERE 1.3 0.00248 0.0358 SNTB1 1.5 0.0001008 0.0040165 LOC120376 1.3 0.00248 0.0358 RNPC3 1.9 0.0001013 0.0040309 PRIC285 1.5 0.00248 0.0358 REG4 1.3 0.0001016 0.0040309 B4GALT5 1.4 0.00249 0.0359 360

PAN2 1.4 0.0001051 0.0041459 SLC22A15 1.2 0.00249 0.0359 KBTBD8 1.7 0.0001054 0.004149 HS.22689 1.7 0.00249 0.0359 SYVN1 1.7 0.0001056 0.0041495 C19ORF28 1.5 0.00250 0.0360 WDR66 1.6 0.0001067 0.0041736 PASD1 1.3 0.00250 0.0360 SORBS3 2.2 0.0001069 0.0041736 RPS6P1 2.1 0.00251 0.0360 RAB11FIP4 1.4 0.000107 0.0041736 LOC643985 1.4 0.00251 0.0360 MPPED2 2.9 0.0001071 0.0041736 TMED6 2.3 0.00251 0.0360 KIR2DL3 2.3 0.0001093 0.0042441 DNMT3B 1.9 0.00252 0.0361 AOX2P 1.5 0.0001116 0.0043224 FMO4 1.3 0.00252 0.0361 ZBP1 1.4 0.0001122 0.0043314 CD48 1.7 0.00252 0.0361 UNC50 2.0 0.0001125 0.0043384 PLG 1.3 0.00256 0.0365 HS.130916 1.6 0.0001127 0.0043395 SCML4 1.3 0.00257 0.0367 PTPRC 1.8 0.0001133 0.00435 LOC100129637 1.5 0.00257 0.0367 HS.546375 2.2 0.0001151 0.0043905 DNAJC14 1.3 0.00258 0.0367 GALK2 1.4 0.0001196 0.0045364 DBI 1.5 0.00258 0.0367 AMICA1 1.4 0.0001199 0.0045399 SIRPG 1.2 0.00258 0.0368 RBM12B 2.2 0.0001216 0.0045824 TRIM56 1.3 0.00259 0.0368 NMNAT3 1.4 0.0001262 0.0047311 FZD4 1.2 0.00261 0.0369 HS.538157 1.4 0.0001268 0.0047472 OPTN 1.5 0.00262 0.0370 UBL3 1.6 0.000127 0.0047472 GON4L 1.2 0.00263 0.0371 COPS7A 1.4 0.0001281 0.0047634 HPS4 1.4 0.00263 0.0371 NAPSB 2.3 0.0001286 0.0047704 EOMES 1.8 0.00263 0.0371 LOC641518 1.4 0.0001291 0.0047787 PARS2 1.2 0.00263 0.0371 RNASET2 2.2 0.0001307 0.0048183 CXCR3 1.3 0.00266 0.0374 LOC1001295 50 1.7 0.0001317 0.0048409 LACTB2 1.7 0.00268 0.0376 LOC647349 1.5 0.0001328 0.0048611 LOC647673 1.3 0.00269 0.0377 SERPINA1 1.4 0.0001331 0.0048655 VPS33B 1.2 0.00269 0.0377 LOC400657 1.7 0.0001344 0.0048812 FLJ12334 1.3 0.00270 0.0377 LOC649864 1.8 0.0001363 0.0049396 COX19 1.7 0.00271 0.0378 C6ORF81 1.4 0.0001365 0.0049408 CLDND1 1.4 0.00273 0.0380 HS.569104 1.6 0.0001372 0.0049525 RPS14 1.3 0.00273 0.0380 SAMD9 1.8 0.0001387 0.0049845 TBC1D10A 1.5 0.00273 0.0380 TMEM156 2.1 0.0001395 0.0049953 BTBD11 1.3 0.00274 0.0380 COL24A1 1.9 0.0001396 0.0049953 FUK 1.4 0.00274 0.0381 NAPB 1.4 0.0001419 0.0050589 GPD2 1.3 0.00276 0.0383 NFIA 1.5 0.0001424 0.0050706 TPD52 1.4 0.00277 0.0384 LOC1001315 72 1.8 0.0001445 0.0051291 WDSUB1 1.5 0.00279 0.0385 CHRM3 1.6 0.000145 0.0051412 LOC100133373 1.3 0.00279 0.0386 LPHN3 1.9 0.0001455 0.0051528 PPAP2B 1.5 0.00280 0.0387 CXCR6 1.3 0.0001465 0.0051804 TMEM170B 1.4 0.00282 0.0388 METTL3 1.6 0.0001478 0.0052204 ZNF500 1.4 0.00282 0.0388 LOC440359 2.1 0.0001484 0.0052324 KIAA0495 1.4 0.00283 0.0389 PRTN3 1.3 0.0001493 0.005258 NUMB 1.4 0.00284 0.0390 ZSWIM3 1.4 0.0001518 0.0053249 GSK3A 1.3 0.00286 0.0392 LOC1001305 57 1.7 0.0001548 0.005413 KLRA1 1.4 0.00287 0.0393 PARP9 1.6 0.0001557 0.0054287 RB1CC1 1.6 0.00290 0.0396 C2ORF89 1.5 0.0001557 0.0054287 LOC648470 1.2 0.00290 0.0397 TGFBR2 1.8 0.0001558 0.0054287 HS.441076 1.3 0.00291 0.0398 LOC728543 1.3 0.0001563 0.0054296 LOC254398 2.1 0.00296 0.0403 KLHL24 2.2 0.0001566 0.0054296 ITGAD 1.2 0.00300 0.0407 RNF170 1.7 0.0001567 0.0054296 ZNF507 1.3 0.00300 0.0407 HS.162734 1.8 0.0001616 0.0055629 HMGCL 1.4 0.00302 0.0409

361

SULF2 1.6 0.0001619 0.0055652 ECHDC3 1.4 0.00302 0.0409 ZNF323 1.8 0.0001641 0.0056105 KIAA0528 1.5 0.00303 0.0410 TACSTD1 1.3 0.0001648 0.0056283 LOC646808 1.7 0.00304 0.0410 ZFYVE26 1.6 0.0001655 0.0056441 RNASEL 1.5 0.00304 0.0410 ZC3H10 1.4 0.000166 0.0056551 ZCCHC14 1.5 0.00305 0.0411 HS.572030 1.4 0.0001671 0.0056847 HS.128615 1.4 0.00305 0.0411 ZNF234 2.0 0.0001689 0.0057309 LOC390031 1.2 0.00307 0.0413 LOC646909 1.8 0.0001694 0.0057406 PEX5 1.3 0.00308 0.0414 LOC1001291 95 1.5 0.0001702 0.0057597 FGL2 1.3 0.00309 0.0415 PRSS7 1.4 0.0001713 0.0057806 C3ORF62 1.4 0.00309 0.0415 RASSF4 1.5 0.0001727 0.005818 CUL5 1.5 0.00310 0.0416 PREX1 1.6 0.0001728 0.005818 PFKP 2.0 0.00311 0.0417 ZSCAN18 1.5 0.000174 0.0058494 NELL2 1.4 0.00312 0.0417 FHDC1 1.4 0.0001746 0.0058494 KLHL20 1.5 0.00313 0.0418 NUDT16 2.0 0.0001746 0.0058494 LOC643733 1.3 0.00313 0.0418 LOC729926 1.6 0.0001747 0.0058494 CPA3 1.3 0.00313 0.0418 LOC647436 1.6 0.0001749 0.0058494 NSUN5C 1.5 0.00314 0.0419 CAPN14 1.6 0.0001757 0.0058543 BEST1 1.3 0.00314 0.0419 ARHGEF18 2.2 0.000177 0.0058858 PIWIL4 1.3 0.00315 0.0420 PTCD2 1.5 0.0001771 0.0058858 LOC644284 1.4 0.00316 0.0420 MTX2 1.4 0.000179 0.0059386 UBN2 1.3 0.00316 0.0420 ATP6V1A 2.1 0.0001796 0.0059386 RFXAP 1.4 0.00316 0.0420 INTS5 1.5 0.0001804 0.0059589 GPR162 3.1 0.00318 0.0422 TCEA3 1.4 0.0001813 0.0059775 LOC148413 1.6 0.00319 0.0423 PDCD6IP 1.9 0.0001815 0.0059775 SLC22A23 1.4 0.00319 0.0423 SCRN1 1.4 0.0001832 0.0060045 SEC62 1.4 0.00320 0.0424 OXSM 1.4 0.0001836 0.00601 TTC17 1.5 0.00323 0.0426 CHST7 2.1 0.0001845 0.0060314 SNX27 1.3 0.00323 0.0427 LOC643888 1.5 0.0001862 0.0060659 HS.493947 1.3 0.00324 0.0427 HS.19339 1.8 0.0001874 0.0060859 COBLL1 1.6 0.00325 0.0429 TMEM217 1.6 0.0001876 0.0060859 C1ORF91 1.3 0.00325 0.0429 SERTAD2 1.8 0.0001881 0.0060956 XPO7 1.5 0.00326 0.0429 LOC646588 1.4 0.0001887 0.0061035 HOXA9 1.2 0.00326 0.0429 P2RX1 2.3 0.000189 0.0061035 MT2A 1.5 0.00327 0.0430 TRK1 2.9 0.000191 0.0061528 WDR42A 1.4 0.00327 0.0430 PIGV 2.1 0.0001939 0.0062074 FAM175B 1.5 0.00330 0.0433 CD97 2.6 0.0001942 0.0062074 IFT57 1.5 0.00330 0.0434 TPST1 1.5 0.0001964 0.0062713 SIGLEC14 1.2 0.00331 0.0434 GFOD1 2.4 0.000197 0.0062821 PML 1.3 0.00331 0.0434 SLC9A5 1.6 0.0001972 0.0062821 SERINC5 1.4 0.00331 0.0434 PDE8B 1.8 0.0002006 0.0063437 TSTA3 1.4 0.00332 0.0434 LOC731898 1.3 0.0002047 0.0064564 TAL1 1.3 0.00332 0.0435 TUBB6 2.2 0.000207 0.0065132 FLJ10213 1.4 0.00333 0.0435 TTC38 1.5 0.0002089 0.0065645 PION 2.1 0.00333 0.0435 LOC402176 1.6 0.0002138 0.0066859 UTP14C 1.5 0.00334 0.0436 CTSS 1.6 0.0002149 0.0067138 HS.573264 1.4 0.00335 0.0436 WDR78 1.7 0.0002159 0.0067382 HS.211930 1.3 0.00335 0.0437 HS.154336 1.7 0.0002174 0.0067769 TCEAL5 1.3 0.00335 0.0437 MTERF 1.4 0.0002192 0.006822 TBC1D2B 1.4 0.00338 0.0441 ZFPM1 1.8 0.00022 0.0068328 FLT3LG 1.6 0.00339 0.0442 FECH 1.4 0.000224 0.0069315 ZFP161 1.5 0.00340 0.0442 ACRBP 1.4 0.0002253 0.0069642 LOC648852 2.1 0.00341 0.0443 RPL23AP53 2.2 0.0002257 0.0069681 LOC100132673 1.5 0.00341 0.0443 COX8C 1.3 0.0002285 0.0070093 ANGPT2 1.2 0.00341 0.0443 362

LSS 1.5 0.0002285 0.0070093 YOD1 1.5 0.00342 0.0443 RAB31 1.4 0.0002298 0.0070344 LOC100131139 1.4 0.00344 0.0445 CHFR 2.9 0.0002318 0.0070876 ACP1 1.3 0.00347 0.0449 TXN 2.0 0.0002359 0.0071876 FTSJ2 1.3 0.00347 0.0449 KIAA0125 1.4 0.0002397 0.0072574 C15ORF29 1.3 0.00348 0.0449 SDHAF1 1.4 0.0002397 0.0072574 LOC645094 1.4 0.00349 0.0450 FALZ 1.4 0.0002406 0.007273 SERPINB10 1.2 0.00349 0.0450 FAM53B 4.4 0.0002421 0.007301 LOC100130112 1.3 0.00350 0.0451 FLT3 1.5 0.000244 0.0073403 OSBPL8 1.5 0.00355 0.0456 LOC388339 1.5 0.0002458 0.0073733 WDR81 1.5 0.00356 0.0457 C10ORF137 1.9 0.0002459 0.0073733 PXDN 1.5 0.00358 0.0459 BBS10 1.7 0.0002477 0.0074097 LOC644029 1.3 0.00359 0.0460 PITPNM1 2.0 0.0002489 0.0074359 C10ORF32 1.5 0.00360 0.0460 ARHGAP27 1.5 0.0002507 0.0074745 LOC388588 1.7 0.00360 0.0461 MVP 1.7 0.0002562 0.0075981 FAM179B 1.5 0.00363 0.0464 WWC3 1.8 0.0002564 0.0075981 S100A4 2.5 0.00366 0.0466 IL28RA 1.3 0.0002566 0.0075981 HEATR5B 1.5 0.00368 0.0468 LOC1001319 67 2.1 0.0002571 0.0076017 DUS1L 1.3 0.00368 0.0468 PTPN12 1.9 0.0002586 0.007629 LOC151162 1.8 0.00369 0.0468 RIOK3 1.5 0.0002634 0.0077389 RGS9 1.3 0.00370 0.0470 RHD 1.4 0.0002635 0.0077389 LOC654191 1.6 0.00371 0.0471 PDZD4 1.4 0.0002641 0.0077389 ZXDC 1.4 0.00371 0.0471 SELENBP1 1.4 0.0002646 0.0077442 FYCO1 1.3 0.00372 0.0471 CRYGS 1.7 0.0002713 0.0079245 PNMT 1.3 0.00373 0.0473 FAM13AOS 1.7 0.0002727 0.0079461 LOC441481 1.4 0.00375 0.0474 MUC20 1.3 0.0002732 0.0079521 C9ORF45 1.4 0.00378 0.0477 HS.560896 1.4 0.0002756 0.0080136 PIGP 1.3 0.00380 0.0479 HS.20255 1.6 0.0002804 0.008124 SEPT1 1.3 0.00384 0.0483 AFAP1L2 2.0 0.0002823 0.0081546 TAPT1 1.6 0.00385 0.0483 ATP8B4 1.4 0.000285 0.0081969 PROSC 1.5 0.00385 0.0483 C16ORF86 1.4 0.0002851 0.0081969 LOC401321 1.3 0.00385 0.0484 MGC15763 1.6 0.0002892 0.0082618 AFF4 1.5 0.00386 0.0484 NME3 2.3 0.0002905 0.0082878 MTE 1.2 0.00388 0.0487 LOC643389 1.6 0.0002909 0.008292 TSC22D3 1.6 0.00388 0.0487 HSPA1L 1.6 0.0002917 0.0083044 ADAP2 2.0 0.00391 0.0490 NCR3 1.4 0.0002944 0.0083564 ARL17B 1.3 0.00394 0.0492 LOC729350 1.4 0.0002945 0.0083564 PSMD10 1.3 0.00394 0.0492 MTHFR 1.6 0.0002954 0.0083638 IGSF6 1.2 0.00395 0.0492 WWP1 1.5 0.0002968 0.0083906 CD72 1.7 0.00396 0.0493 LAMB2L 1.4 0.000297 0.0083906 ZNF7 1.4 0.00397 0.0493 RHOBTB3 1.8 0.0002978 0.0084056 LOC642897 1.3 0.00397 0.0494 FAM149B1 1.6 0.0003 0.0084469 ARGLU1 1.8 0.00398 0.0495 SPON2 1.4 0.0003004 0.0084469 LOC100130516 2.0 0.00399 0.0495 CIB4 1.7 0.0003009 0.0084469 MAP1A 1.8 0.00399 0.0495 ABHD10 1.4 0.0003037 0.0084978 LOC100131930 1.2 0.00400 0.0496 MGC87895 1.4 0.0003037 0.0084978 CREBBP 1.3 0.00402 0.0498 LOC158376 1.5 0.0003082 0.0086045 IFIT5 1.3 0.00404 0.0499 RNF135 1.7 0.0003095 0.0086234

363

Table B.3. List of the 954 genes that were significantly up-regulated in the VXL- treated versus non-treated samples of ALL-67 xenograft. The cut-off for significance was set at FDR <0.05. absFC, absolute fold change.

abs Q Gene P.Value Q Value Gene absFC P.Value FC Value C16ORF70 1.29 0.000001 0.019490 LOC642947 1.16 0.0010 0.0395 LOC401717 1.48 0.000002 0.019490 H2AFY 1.27 0.0011 0.0395 PLXNB1 1.42 0.000005 0.019490 LOC647074 1.27 0.0011 0.0395 LOC727826 1.54 0.000007 0.019490 CD19 1.55 0.0011 0.0395 ADI1 1.45 0.000008 0.019490 PLDN 1.39 0.0011 0.0396 PRPF3 1.48 0.000013 0.019490 TOLLIP 1.18 0.0011 0.0398 LOC728324 1.44 0.000013 0.019490 CLUAP1 1.25 0.0011 0.0398 LOC284821 1.30 0.000014 0.019490 FBN2 1.40 0.0011 0.0400 LOC400652 1.39 0.000014 0.019490 VEGFB 1.38 0.0011 0.0400 CYBA 1.69 0.000015 0.019490 RCN1 1.49 0.0011 0.0400 POLR2J2 1.48 0.000017 0.019490 PCBP1 1.38 0.0011 0.0400 STAR 1.21 0.000018 0.019490 LOC400963 1.28 0.0011 0.0400 MRPL28 1.25 0.000018 0.019490 LSM4 1.46 0.0011 0.0400 PCDHGB6 1.24 0.000019 0.019490 SPRY2 1.73 0.0011 0.0400 LOC730455 1.46 0.000019 0.019490 CRK 1.29 0.0011 0.0400 LOC100130750 1.29 0.000020 0.019490 GOT1 1.42 0.0011 0.0400 DAD1 1.36 0.000020 0.019490 TUBA1B 1.26 0.0011 0.0400 LOC100133812 1.23 0.000020 0.019490 CHD7 1.45 0.0011 0.0400 RPL39L 1.60 0.000020 0.019490 EVI2B 1.60 0.0011 0.0400 LOC729952 1.35 0.000020 0.019490 TMEM108 1.13 0.0011 0.0400 IGFL3 1.30 0.000028 0.023320 LOC646200 1.35 0.0011 0.0400 AOAH 1.21 0.000031 0.023320 LILRA3 1.28 0.0011 0.0400 POLR2J3 1.54 0.000032 0.023320 ZWINT 1.47 0.0011 0.0400 COX6A1 1.47 0.000032 0.023320 DUSP28 1.32 0.0011 0.0400 LOC442270 1.39 0.000032 0.023320 DPYSL2 1.46 0.0011 0.0400 LOC392285 1.64 0.000033 0.023320 GTF2A2 1.37 0.0011 0.0400 RPS28 1.39 0.000034 0.023320 TAF1B 1.38 0.0011 0.0400 LOC100129067 1.35 0.000037 0.024580 HAX1 1.23 0.0011 0.0400 LOC645317 1.37 0.000039 0.024580 PRDX1 1.33 0.0011 0.0400 PGD 1.43 0.000039 0.024580 CKAP5 1.43 0.0011 0.0400 LOC729255 1.50 0.000041 0.024860 EXOSC9 1.38 0.0011 0.0400 LOC100133277 1.53 0.000043 0.024860 RPL9 1.25 0.0011 0.0401 GAB3 1.54 0.000045 0.024860 LOC389101 1.24 0.0011 0.0402 LOC1001332 BMP8B 1.37 0.000045 0.024860 98 1.27 0.0011 0.0402 TMPRSS12 1.18 0.000046 0.024860 LMO2 1.49 0.0011 0.0404 LOC401640 1.41 0.000052 0.024860 TUBA1A 1.25 0.0011 0.0404 FAM58A 1.52 0.000054 0.024860 LOC646103 1.15 0.0011 0.0404 SEPW1 1.53 0.000056 0.024860 CYBASC3 1.41 0.0011 0.0404 TMEM189 - LOC647099 1.24 0.000057 0.024860 UBE2V1 1.47 0.0012 0.0404 STIM2 1.61 0.000058 0.024860 IL1RAP 1.31 0.0012 0.0404 LOC645630 1.61 0.000061 0.024860 LOC390735 1.41 0.0012 0.0407 XBP1 1.50 0.000061 0.024860 TLR10 2.20 0.0012 0.0408 PYGB 1.46 0.000067 0.024860 EIF3M 1.37 0.0012 0.0408 HS.153349 1.23 0.000067 0.024860 LIN7B 1.35 0.0012 0.0408 LOC728973 1.47 0.000068 0.024860 ATP5O 1.39 0.0012 0.0408 LOC649946 1.60 0.000069 0.024860 MAP1S 1.50 0.0012 0.0414 GALNT4 1.34 0.000071 0.024860 LOC649076 1.17 0.0012 0.0415 LOC390183 1.56 0.000075 0.024860 RABL2A 1.24 0.0012 0.0415 364

SAE1 1.43 0.000076 0.024860 C18ORF8 1.49 0.0012 0.0415 CETN2 1.64 0.000077 0.024860 MED8 1.23 0.0012 0.0415 LOC401676 1.39 0.000080 0.024860 LOC390578 1.38 0.0012 0.0415 CD63 1.39 0.000080 0.024860 TXNDC17 1.51 0.0012 0.0415 LOC643997 1.59 0.000081 0.024860 HS.561575 1.19 0.0012 0.0415 SCML1 2.00 0.000082 0.024860 SH3GLB2 2.00 0.0012 0.0415 LOC100130553 1.29 0.000082 0.024860 DCTN3 1.35 0.0012 0.0415 LOC100131572 1.47 0.000084 0.024860 ZNF32 1.31 0.0012 0.0415 HS.541235 1.15 0.000084 0.024860 LOC730740 1.27 0.0012 0.0415 RPS10P3 1.34 0.000088 0.024860 C2ORF24 1.32 0.0012 0.0415 ATP6V0B 1.46 0.000090 0.024860 PPA2 1.31 0.0012 0.0415 LOC648740 1.47 0.000091 0.024860 LOC652493 1.77 0.0012 0.0415 SDHB 1.43 0.000092 0.024860 BID 1.57 0.0012 0.0415 TUBB6 1.68 0.000093 0.024860 LOC728128 1.35 0.0012 0.0416 USP42 1.39 0.000094 0.024860 LOC649214 1.30 0.0012 0.0416 ZGPAT 1.21 0.000094 0.024860 HS.117299 1.19 0.0012 0.0416 LOC1001290 LOC728602 1.30 0.000094 0.024860 94 1.25 0.0012 0.0416 LOC645693 1.46 0.000095 0.024860 CHCHD9 1.33 0.0012 0.0416 RHEB 1.54 0.000096 0.024860 TP53RK 1.29 0.0012 0.0416 AMY1A 1.80 0.000097 0.024860 LOC391019 1.40 0.0012 0.0416 RPS11 1.26 0.000100 0.024860 LOC441896 1.32 0.0012 0.0416 EFNA4 1.43 0.000100 0.024860 TMEM147 1.33 0.0012 0.0416 LOC338870 1.51 0.000103 0.024860 ZNF770 1.32 0.0013 0.0417 LOC1001308 COX17 1.50 0.000104 0.024860 18 1.17 0.0013 0.0417 LOC730288 1.42 0.000105 0.024860 C19ORF31 1.25 0.0013 0.0420 GNB1 1.38 0.000105 0.024860 LOC729769 1.29 0.0013 0.0420 LOC100130070 1.48 0.000105 0.024860 NDUFB6 1.36 0.0013 0.0420 C17ORF60 2.10 0.000106 0.024860 MAPK9 1.31 0.0013 0.0420 LOC1001327 EVL 1.39 0.000107 0.024860 71 1.17 0.0013 0.0423 RIC8A 1.37 0.000107 0.024860 CBS 1.25 0.0013 0.0424 C6ORF26 1.31 0.000107 0.024860 RPS27A 1.21 0.0013 0.0424 HCST 1.61 0.000109 0.024860 DUS1L 1.23 0.0013 0.0425 LOC126235 1.37 0.000110 0.024860 ATP5I 1.27 0.0013 0.0425 UQCRH 1.41 0.000110 0.024860 RBM23 1.29 0.0013 0.0425 SEC61G 1.38 0.000111 0.024860 LOC651149 1.43 0.0013 0.0425 PGK1 1.27 0.000111 0.024860 TCEAL8 1.44 0.0013 0.0426 LOC402644 1.77 0.000111 0.024860 LOC647285 1.47 0.0013 0.0426 LOC391833 1.45 0.000114 0.024860 NOMO1 1.24 0.0013 0.0426 IDH2 1.44 0.000115 0.024860 SEC22C 1.34 0.0013 0.0429 CDKN2AIPN LOC649873 1.30 0.000123 0.025910 L 1.19 0.0013 0.0429 LOC158345 1.49 0.000123 0.025910 ZNF668 1.24 0.0013 0.0429 LOC644039 1.24 0.000125 0.026150 LOC391532 1.46 0.0013 0.0429 TMSB10 1.32 0.000132 0.027100 FLJ46552 1.15 0.0014 0.0429 KIAA1539 1.48 0.000135 0.027420 C9ORF80 1.22 0.0014 0.0429 LOC727970 1.26 0.000138 0.027500 HS.426162 1.11 0.0014 0.0429 PRDX5 1.50 0.000141 0.027500 MAP1LC3A 1.67 0.0014 0.0429 INPP5B 1.30 0.000149 0.027500 LOC341315 1.57 0.0014 0.0429 LOC730382 1.46 0.000150 0.027500 HS.565338 1.16 0.0014 0.0429 LOC1001309 NR1I2 1.12 0.000153 0.027500 19 1.30 0.0014 0.0429 C18ORF51 2.03 0.000155 0.027500 RPRC1 1.32 0.0014 0.0429 LOC729595 1.20 0.000155 0.027500 EPHB4 1.87 0.0014 0.0429 LOC100127893 1.26 0.000156 0.027500 G3BP2 1.48 0.0014 0.0430

365

LOC728002 1.28 0.000163 0.027500 SNTG2 1.14 0.0014 0.0430 ATP5EP2 1.38 0.000163 0.027500 TMEM14D 1.30 0.0014 0.0430 LOC678655 1.42 0.000165 0.027500 CAB39 1.43 0.0014 0.0430 LOC1001286 RPL7A 1.43 0.000167 0.027500 53 1.16 0.0014 0.0431 LOC130773 1.31 0.000167 0.027500 LOC728138 1.33 0.0014 0.0431 C16ORF33 1.60 0.000167 0.027500 SSX3 1.32 0.0014 0.0431 LOC388339 1.56 0.000168 0.027500 BUB3 1.49 0.0014 0.0431 NEDD8 1.38 0.000168 0.027500 CLTA 1.27 0.0014 0.0431 LOC100132037 1.30 0.000169 0.027500 LOC728666 1.67 0.0014 0.0431 LOC643949 1.41 0.000171 0.027500 RIOK3 1.31 0.0014 0.0431 MAD1L1 1.46 0.000173 0.027500 LOC647276 1.22 0.0014 0.0431 LOC1001288 MEIS1 1.19 0.000174 0.027500 36 1.39 0.0014 0.0431 AP4E1 1.44 0.000176 0.027500 SNRNP70 1.26 0.0014 0.0431 HIGD1A 1.45 0.000177 0.027500 MYH9 1.33 0.0014 0.0431 RPL27 1.31 0.000179 0.027500 ADIPOR2 1.45 0.0014 0.0431 NDUFA12 1.46 0.000180 0.027500 TBC1D9B 1.44 0.0014 0.0431 LOC1001301 LOC645715 1.26 0.000181 0.027500 54 1.59 0.0014 0.0431 LOC1001287 AMY1C 1.47 0.000183 0.027500 75 2.02 0.0014 0.0431 CTNNBL1 1.40 0.000183 0.027500 MCM7 1.28 0.0014 0.0431 CORO1A 2.90 0.000184 0.027500 CDK5R1 1.28 0.0014 0.0431 LOC645138 1.48 0.000184 0.027500 HS.34558 1.40 0.0014 0.0431 ACTG1 1.27 0.000190 0.028050 SLC25A39 1.37 0.0014 0.0431 HS.546375 1.99 0.000194 0.028110 REC8 1.40 0.0014 0.0431 LOC440063 1.75 0.000196 0.028170 ANGPTL2 2.80 0.0014 0.0431 CDKN2D 1.77 0.000204 0.028890 UROD 1.28 0.0014 0.0431 COMMD6 1.42 0.000205 0.028910 PPIA 1.40 0.0014 0.0431 LOC646630 1.37 0.000211 0.028910 LOC391825 1.32 0.0014 0.0431 UQCRHL 1.33 0.000212 0.028910 TMEM39A 1.18 0.0014 0.0431 LOC1001333 C1ORF41 1.43 0.000213 0.028910 29 1.41 0.0014 0.0431 GAS1 1.84 0.000213 0.028910 FBXO2 1.17 0.0014 0.0431 LOC1001314 LOC651064 1.37 0.000217 0.028910 52 1.21 0.0014 0.0431 NBEA 1.26 0.000219 0.028910 PAFAH1B3 1.41 0.0014 0.0431 LOC728774 1.20 0.000221 0.028910 FBXO21 1.50 0.0015 0.0431 LOC441481 1.38 0.000221 0.028910 CCT3 1.28 0.0015 0.0431 LOC646966 1.49 0.000222 0.028910 ARHGDIB 1.21 0.0015 0.0431 HS.524129 1.58 0.000225 0.028910 RPS18 1.45 0.0015 0.0431 PI4K2A 1.20 0.000226 0.028910 LOC728672 1.26 0.0015 0.0433 LOC1001279 LOC441550 1.48 0.000227 0.028910 18 1.43 0.0015 0.0433 LOC648622 1.38 0.000227 0.028910 OR1L8 1.16 0.0015 0.0433 LOC100132199 1.48 0.000229 0.028910 LOC728416 1.31 0.0015 0.0433 LOC134997 1.39 0.000233 0.028910 LOC651643 1.14 0.0015 0.0435 LOC1001289 TLE1 1.67 0.000235 0.028910 75 1.31 0.0015 0.0435 PQLC3 1.69 0.000235 0.028910 LOC387930 1.19 0.0015 0.0437 ANXA5 1.79 0.000235 0.028910 YBX1 1.33 0.0015 0.0437 LOC100128936 1.45 0.000236 0.028910 LOC646949 1.39 0.0015 0.0437 HS.552025 1.24 0.000237 0.028910 TMEM14B 1.47 0.0015 0.0438 WDR61 1.47 0.000240 0.028910 HS.564874 1.34 0.0015 0.0439 DOK1 1.17 0.000242 0.028910 C16ORF61 1.39 0.0015 0.0440 LOC729466 1.36 0.000242 0.028910 ATP5J 1.29 0.0015 0.0440 GSTP1 1.44 0.000242 0.028910 RPL6 1.19 0.0015 0.0442

366

LOC728820 1.60 0.000243 0.028910 SLC7A3 1.48 0.0015 0.0442 RALB 1.61 0.000245 0.028910 PARP9 1.30 0.0015 0.0442 LOC644937 1.52 0.000248 0.028910 PTGES3 1.69 0.0015 0.0442 NBEAL2 1.36 0.000248 0.028910 AP2A2 1.17 0.0015 0.0443 CAPG 1.61 0.000249 0.028910 HS.128557 1.15 0.0016 0.0446 IPP 1.21 0.000255 0.028910 SOD1 1.27 0.0016 0.0446 COX6B1 1.36 0.000255 0.028910 APIP 1.45 0.0016 0.0446 CAPZA1 1.54 0.000256 0.028910 HS.374257 1.34 0.0016 0.0446 LOC100132863 1.45 0.000259 0.028910 LOC642738 1.28 0.0016 0.0446 LOC202227 1.30 0.000259 0.028910 IL17RA 1.35 0.0016 0.0446 TMEM5 1.50 0.000260 0.028910 FAM89A 2.01 0.0016 0.0446 ATP5C1 1.45 0.000260 0.028910 ITPKB 1.22 0.0016 0.0448 HS.542632 1.14 0.000263 0.028970 SEMA4B 1.35 0.0016 0.0448 IL18 1.19 0.000265 0.028970 CYTH4 1.94 0.0016 0.0448 ETFA 1.53 0.000265 0.028970 MGC26356 1.18 0.0016 0.0448 RRAGA 1.50 0.000266 0.028970 EIF2AK2 1.36 0.0016 0.0448 LOC647673 1.51 0.000268 0.029130 BCAT1 1.79 0.0016 0.0448 IMPA2 2.04 0.000271 0.029220 HS.543875 1.11 0.0016 0.0449 GLRX5 1.50 0.000272 0.029270 STK10 1.23 0.0016 0.0449 LSMD1 1.34 0.000274 0.029270 PCDHB13 1.21 0.0016 0.0450 MARS 1.32 0.000278 0.029490 NACA2 1.17 0.0016 0.0450 LOC100129243 1.20 0.000281 0.029590 ZYX 1.88 0.0016 0.0450 RPL38 1.36 0.000282 0.029620 ICAM2 1.59 0.0016 0.0450 DEF6 1.31 0.000286 0.029830 LOC649150 1.17 0.0016 0.0450 MEGF9 1.33 0.000289 0.029860 LOC220433 1.44 0.0016 0.0450 SNRPD2 1.32 0.000291 0.029880 TXNDC12 1.27 0.0016 0.0451 LOC1001322 LOC401397 1.55 0.000292 0.029880 91 1.36 0.0016 0.0451 TIMM8B 1.21 0.000293 0.029880 MRPS6 1.45 0.0016 0.0451 ECGF1 1.38 0.000298 0.030070 NT5DC1 1.51 0.0016 0.0451 CHCHD2 1.36 0.000302 0.030070 C1ORF162 2.30 0.0016 0.0452 RPL30 1.31 0.000311 0.030430 ADAM17 1.26 0.0016 0.0452 TOMM7 1.35 0.000313 0.030430 SLC25A5 1.24 0.0016 0.0452 TIMM8A 1.36 0.000313 0.030430 ZNF296 1.75 0.0017 0.0453 LOC391370 1.40 0.000316 0.030430 LOC642567 1.36 0.0017 0.0453 NPC2 1.39 0.000317 0.030430 MRPL55 1.25 0.0017 0.0453 MIR877 1.17 0.000319 0.030430 HBXIP 1.33 0.0017 0.0454 GNAS 1.28 0.000319 0.030430 CBL 1.51 0.0017 0.0454 LOC646819 1.31 0.000325 0.030530 LOC728877 1.35 0.0017 0.0455 RPS25 1.26 0.000326 0.030530 COCH 1.67 0.0017 0.0455 LOC392501 1.24 0.000327 0.030530 DCTN5 1.31 0.0017 0.0455 RAX2 1.29 0.000329 0.030530 LOC392008 1.12 0.0017 0.0455 STRN3 1.36 0.000329 0.030530 PASK 1.51 0.0017 0.0456 LOC642771 1.14 0.000331 0.030530 C17ORF58 1.20 0.0017 0.0456 SSH1 1.27 0.000334 0.030650 LOC441032 1.20 0.0017 0.0456 HS.534061 1.54 0.000338 0.030790 SYVN1 1.31 0.0017 0.0458 PHCA 1.30 0.000339 0.030790 GTF2IRD2P 1.24 0.0017 0.0458 LOC441154 1.39 0.000343 0.031010 SH3KBP1 1.96 0.0017 0.0459 MRPL41 1.93 0.000346 0.031010 RNF40 1.21 0.0017 0.0459 RPL13L 1.46 0.000347 0.031010 FAM96A 1.43 0.0017 0.0459 SFRS14 1.37 0.000351 0.031060 CARKD 1.49 0.0017 0.0459 SPCS1 1.45 0.000353 0.031100 HS.125087 1.16 0.0017 0.0459 PDCD4 1.27 0.000356 0.031180 SLMO1 1.47 0.0017 0.0459 UXT 1.33 0.000356 0.031180 ISL2 1.15 0.0017 0.0459 CD74 1.25 0.000361 0.031280 ETV6 1.58 0.0017 0.0459 ABHD15 1.38 0.000364 0.031280 FAM84B 1.50 0.0017 0.0459

367

IMAA 1.27 0.000365 0.031280 QDPR 1.45 0.0017 0.0459 LOC727803 1.74 0.000365 0.031280 PEX5 1.28 0.0017 0.0459 LOC100129673 1.37 0.000367 0.031280 LOC399804 1.36 0.0017 0.0459 LCP1 1.55 0.000368 0.031280 GLCE 2.01 0.0018 0.0459 LOC643911 1.70 0.000368 0.031280 AKR1B1 1.71 0.0018 0.0459 QSER1 1.18 0.000370 0.031280 CPT1B 1.44 0.0018 0.0459 MYD88 1.57 0.000370 0.031280 MICAL1 1.64 0.0018 0.0459 LOC100130624 1.41 0.000374 0.031340 TAX1BP1 1.36 0.0018 0.0459 LOC644265 1.21 0.000376 0.031370 PIGX 1.21 0.0018 0.0459 KIAA0182 1.63 0.000378 0.031450 LASS5 1.23 0.0018 0.0459 AGK 1.42 0.000381 0.031450 WASF1 1.43 0.0018 0.0459 OR5D18 1.10 0.000388 0.031830 RPL7L1 1.20 0.0018 0.0459 S100A6 3.07 0.000390 0.031830 IWS1 1.28 0.0018 0.0459 CFP 1.22 0.000391 0.031830 STS-1 1.55 0.0018 0.0459 FERMT3 1.65 0.000392 0.031830 LOC651202 1.41 0.0018 0.0459 DBI 1.50 0.000394 0.031830 MVP 1.29 0.0018 0.0459 CIB1 1.57 0.000394 0.031830 DNMT1 1.38 0.0018 0.0459 LOC387934 1.44 0.000396 0.031890 SACS 1.36 0.0018 0.0459 LOC402175 1.33 0.000398 0.031890 SH3PXD2A 1.39 0.0018 0.0459 ZNF320 1.24 0.000400 0.031990 TMED9 1.37 0.0018 0.0459 ATP5G2 1.42 0.000409 0.032380 HIGD2A 1.43 0.0018 0.0459 SIVA1 1.41 0.000411 0.032380 ERCC-00117 1.10 0.0018 0.0459 MCF2 1.20 0.000412 0.032380 XYLT2 1.33 0.0018 0.0460 PSMD10 1.46 0.000414 0.032380 RPL12 1.35 0.0018 0.0460 LOC100134273 1.45 0.000415 0.032380 CISD2 1.30 0.0018 0.0462 RPAIN 1.32 0.000419 0.032410 FNTA 1.49 0.0018 0.0462 UBL5 1.30 0.000420 0.032410 OCIAD1 1.27 0.0018 0.0462 CD14 1.24 0.000428 0.032410 MGC3032 1.29 0.0018 0.0462 PTPN6 1.59 0.000428 0.032410 FAM21D 1.42 0.0018 0.0462 RPL14L 1.43 0.000431 0.032410 SULT1A1 1.25 0.0018 0.0462 MGC87895 1.38 0.000432 0.032410 MSRB2 1.40 0.0018 0.0462 GCHFR 1.45 0.000433 0.032410 GBGT1 1.33 0.0018 0.0462 PFN1 1.33 0.000433 0.032410 HS.417966 1.17 0.0018 0.0462 LOC388532 1.31 0.000434 0.032410 SFRS15 1.27 0.0018 0.0462 PISD 1.31 0.000436 0.032410 LOC649821 1.34 0.0018 0.0463 ZNHIT3 1.38 0.000442 0.032550 C17ORF47 1.15 0.0018 0.0463 ARL6IP6 1.62 0.000445 0.032550 LOC642357 1.25 0.0018 0.0463 NCRNA0008 RPS14 1.30 0.000446 0.032550 5 1.48 0.0019 0.0465 PEPD 1.41 0.000449 0.032680 CCPG1 1.28 0.0019 0.0465 LOC391126 1.42 0.000453 0.032860 NTAN1 1.34 0.0019 0.0465 RPL27A 1.23 0.000455 0.032890 LOC645289 1.17 0.0019 0.0465 LOC728244 1.39 0.000456 0.032890 HDGF 1.41 0.0019 0.0465 RAP2A 1.51 0.000460 0.033080 ATPIF1 1.26 0.0019 0.0465 LOC137107 1.28 0.000464 0.033250 AMZ2 1.27 0.0019 0.0465 HS.571599 1.28 0.000467 0.033330 LOC347544 1.50 0.0019 0.0465 SRP14P1 1.40 0.000474 0.033560 IL10 1.30 0.0019 0.0465 LOC390354 1.54 0.000477 0.033560 LOC647436 1.39 0.0019 0.0465 LOC100132673 1.82 0.000477 0.033560 FLII 1.24 0.0019 0.0465 NARG1L 1.25 0.000481 0.033560 PDLIM1 1.57 0.0019 0.0465 TMEM214 1.29 0.000483 0.033560 PDPR 1.27 0.0019 0.0466 LOC654350 1.60 0.000486 0.033560 PDLIM7 1.48 0.0019 0.0466 CYP2C8 1.21 0.000487 0.033560 SYK 1.43 0.0019 0.0466 ESRRA 1.27 0.000489 0.033560 C14ORF82 1.33 0.0019 0.0466 TMEM14A 1.43 0.000490 0.033560 C6ORF125 1.29 0.0019 0.0466 ATG10 1.44 0.000490 0.033560 LOC286208 1.24 0.0019 0.0467

368

LOC730754 1.37 0.000492 0.033560 FUCA1 1.75 0.0019 0.0467 DTX3L 1.20 0.000493 0.033560 FBXO6 1.19 0.0019 0.0467 CRMP1 1.52 0.000495 0.033640 SLC7A6 1.32 0.0019 0.0467 LOC100128062 1.37 0.000504 0.033740 ARHGEF18 1.45 0.0019 0.0467 LOC100133390 1.21 0.000505 0.033740 NAGPA 1.26 0.0019 0.0469 CD79B 1.59 0.000505 0.033740 LOC136143 1.35 0.0019 0.0470 LOC646821 1.42 0.000508 0.033740 ZMYND19 1.28 0.0019 0.0470 LOC442727 1.53 0.000508 0.033740 CLEC18C 1.18 0.0019 0.0470 PHGDH 2.74 0.000512 0.033740 LAGE3 1.45 0.0019 0.0470 SEC24C 1.27 0.000513 0.033740 LMBRD1 1.36 0.0020 0.0470 ALKBH7 1.29 0.000518 0.033740 ATP5H 1.36 0.0020 0.0470 TRMT112 1.35 0.000518 0.033740 METTL4 1.22 0.0020 0.0470 CALM1 1.46 0.000520 0.033740 ZNF408 1.25 0.0020 0.0470 RPS15A 1.23 0.000520 0.033740 HSCB 1.27 0.0020 0.0470 IRAK1 1.59 0.000524 0.033740 MMD 1.44 0.0020 0.0470 CLEC14A 1.45 0.000527 0.033740 RPS16 1.21 0.0020 0.0470 LOC644907 1.47 0.000527 0.033740 PLCB3 1.11 0.0020 0.0470 ST6GAL1 1.57 0.000527 0.033740 LOC729142 1.11 0.0020 0.0470 LOC649839 1.38 0.000528 0.033740 HNRPR 1.30 0.0020 0.0470 ATP5J2 1.35 0.000530 0.033740 HOXB5 1.11 0.0020 0.0470 CSTB 1.91 0.000531 0.033740 EIF3I 1.26 0.0020 0.0470 AP2S1 1.35 0.000534 0.033740 CECR6 1.29 0.0020 0.0470 LOC284393 1.18 0.000536 0.033740 LOC440927 1.35 0.0020 0.0471 NDUFB5 1.43 0.000538 0.033740 HS.291195 1.17 0.0020 0.0471 LOC646093 1.28 0.000539 0.033740 LOC439992 1.17 0.0020 0.0471 LOC730004 1.30 0.000540 0.033740 LOC729679 1.52 0.0020 0.0471 LOC257396 1.14 0.000541 0.033740 POTEF 1.44 0.0020 0.0473 KCNQ1OT1 1.21 0.000545 0.033810 LOC646784 1.22 0.0020 0.0473 LOC390956 1.12 0.000547 0.033810 ARRB2 1.53 0.0020 0.0473 RPS21 1.53 0.000551 0.033820 C4ORF34 1.63 0.0020 0.0473 FTHL11 1.53 0.000556 0.033820 ARID3B 1.26 0.0020 0.0473 SNRPD3 1.47 0.000557 0.033820 C13ORF27 1.59 0.0020 0.0474 LOC645018 1.43 0.000559 0.033820 WDR40A 1.51 0.0020 0.0474 VPS29 1.43 0.000560 0.033820 LOC390834 1.13 0.0020 0.0474 CFL1 1.45 0.000561 0.033820 IGSF3 1.51 0.0020 0.0474 GABARAP 1.29 0.000564 0.033820 LOC644315 1.32 0.0020 0.0474 EFTUD2 1.27 0.000566 0.033820 POGK 1.57 0.0020 0.0474 KIAA2013 1.34 0.000567 0.033820 ISCU 1.26 0.0020 0.0474 DUSP14 1.51 0.000568 0.033820 E2F1 1.22 0.0020 0.0474 VIM 2.35 0.000568 0.033820 ZNF362 1.33 0.0020 0.0474 TINP1 1.56 0.000572 0.033850 LOC644949 1.20 0.0020 0.0474 LOC641727 1.32 0.000575 0.033940 LOC606724 2.38 0.0020 0.0474 PYCARD 1.73 0.000577 0.033990 MTHFD2 1.58 0.0021 0.0474 LOC100132060 1.22 0.000584 0.034220 LOC389223 1.19 0.0021 0.0475 AHI1 1.52 0.000589 0.034410 PHKG2 1.20 0.0021 0.0475 NINJ1 1.48 0.000591 0.034420 LOC347376 1.45 0.0021 0.0475 CMPK1 1.51 0.000592 0.034420 TOMM5 1.40 0.0021 0.0475 UPF3A 1.41 0.000596 0.034440 SYAP1 1.33 0.0021 0.0475 LOC1001299 RASGRP2 1.68 0.000599 0.034480 34 1.16 0.0021 0.0476 CYB5A 1.41 0.000601 0.034480 LOC441073 1.32 0.0021 0.0476 SNORD1A 1.11 0.000603 0.034480 TM9SF1 1.26 0.0021 0.0476 LOC402694 1.31 0.000606 0.034480 SEC14L1 1.26 0.0021 0.0476 CXORF40A 1.35 0.000607 0.034480 LOC440595 1.17 0.0021 0.0476 LOC1001313 LOC100134159 1.30 0.000611 0.034480 87 1.24 0.0021 0.0476

369

C6ORF48 1.24 0.000612 0.034480 LOC728484 1.54 0.0021 0.0476 LOC729279 1.38 0.000613 0.034480 CAPN1 1.24 0.0021 0.0476 TRPV1 1.26 0.000615 0.034480 ZNF519 1.25 0.0021 0.0476 LOC390345 1.42 0.000635 0.034860 LOC389386 1.25 0.0021 0.0476 BOLA2 1.28 0.000635 0.034860 LOC728942 1.14 0.0021 0.0476 LOC643433 1.33 0.000637 0.034860 CHEK1 1.55 0.0021 0.0476 PSCD4 2.00 0.000639 0.034860 PAFAH1B1 1.45 0.0021 0.0476 HS.580058 1.21 0.000642 0.034860 RPS9 1.32 0.0021 0.0476 LOC728368 1.25 0.000645 0.034860 NDUFA4 1.29 0.0021 0.0476 LOC727865 1.43 0.000646 0.034860 C9ORF7 1.16 0.0021 0.0476 LOC646527 1.52 0.000647 0.034860 PPIAL4A 1.28 0.0021 0.0476 MAGEH1 1.33 0.000648 0.034860 SSR4 1.35 0.0022 0.0478 IL17D 1.31 0.000650 0.034860 LOC649365 1.12 0.0022 0.0478 PTPRCAP 1.55 0.000650 0.034860 ZC3HAV1L 1.13 0.0022 0.0478 FLNA 1.26 0.000651 0.034860 LOC648099 1.32 0.0022 0.0478 HS.543412 1.13 0.000652 0.034860 PATE3 1.29 0.0022 0.0478 LOC401115 1.57 0.000656 0.034860 EEF1AL7 1.22 0.0022 0.0478 PRR13 1.44 0.000657 0.034860 SASH3 1.50 0.0022 0.0478 HINT1 1.36 0.000657 0.034860 CXORF40B 1.27 0.0022 0.0478 LOC1001282 1.35 0.000657 0.034860 91 1.21 0.0022 0.0478 TMEM126B 1.51 0.000658 0.034860 CAPRIN2 1.31 0.0022 0.0478 CRCP 1.28 0.000660 0.034860 LOC650157 1.31 0.0022 0.0478 LOC100129553 1.21 0.000662 0.034860 IDH3G 1.38 0.0022 0.0480 VEZF1 1.45 0.000663 0.034860 FKSG30 1.59 0.0022 0.0480 NDUFB10 1.44 0.000663 0.034860 LOC400948 1.25 0.0022 0.0480 NQO1 1.17 0.000663 0.034860 ATP6V0D1 1.49 0.0022 0.0480 NEUROD4 1.12 0.000668 0.035020 YIF1B 1.24 0.0022 0.0480 MOBKL2A 1.56 0.000673 0.035080 ZNF626 1.24 0.0022 0.0480 RPL19 1.27 0.000675 0.035080 LRFN4 1.27 0.0022 0.0480 LOC128192 1.50 0.000675 0.035080 PMS2L4 1.26 0.0022 0.0480 EGLN2 1.35 0.000675 0.035080 LOC644863 1.33 0.0022 0.0481 PSMC3IP 1.31 0.000680 0.035260 C11ORF10 1.30 0.0022 0.0481 LOC439953 1.31 0.000686 0.035390 SNTA1 1.56 0.0022 0.0481 LOC203547 1.46 0.000687 0.035390 PIK3C2A 1.26 0.0022 0.0481 UCHL5IP 1.48 0.000690 0.035400 LOC645688 1.36 0.0022 0.0481 TWF2 1.30 0.000691 0.035400 SHPK 1.36 0.0022 0.0481 LOC388524 1.23 0.000695 0.035420 LOC731640 1.34 0.0022 0.0481 DHPS 1.35 0.000697 0.035420 LOC647349 1.39 0.0022 0.0481 SELI 1.40 0.000698 0.035420 ARL3 1.20 0.0022 0.0481 LOC729926 1.39 0.000707 0.035720 MDC1 1.30 0.0022 0.0481 TOMM40 1.36 0.000719 0.036140 C14ORF156 1.38 0.0022 0.0482 LOC100127993 1.28 0.000720 0.036140 GLRX3 1.44 0.0022 0.0482 LYN 1.63 0.000720 0.036140 LOC728755 1.39 0.0022 0.0482 CEBPG 1.39 0.000732 0.036440 CCDC72 1.34 0.0023 0.0483 STMN1 1.47 0.000734 0.036440 SLC43A3 1.44 0.0023 0.0483 LOC100133772 1.33 0.000736 0.036440 C16ORF53 1.62 0.0023 0.0483 JMJD8 1.34 0.000739 0.036540 LOC389404 1.30 0.0023 0.0483 LOC392522 1.19 0.000741 0.036550 FOLR2 1.26 0.0023 0.0485 PRMT3 1.32 0.000744 0.036550 CAP2 1.46 0.0023 0.0485 LOC649447 1.29 0.000747 0.036550 TRIM11 1.17 0.0023 0.0485 RAB24 1.69 0.000750 0.036550 FAM72B 1.22 0.0023 0.0485 LOC650152 1.29 0.000751 0.036550 CACNB3 1.31 0.0023 0.0485 TUBA3D 1.52 0.000757 0.036710 EIF5A 1.56 0.0023 0.0486 S100A4 2.19 0.000758 0.036710 NUDT7 1.49 0.0023 0.0486 LOC644328 1.14 0.000760 0.036710 BCL2L12 1.36 0.0023 0.0486

370

HS.130077 1.14 0.000766 0.036710 HS.582009 1.12 0.0023 0.0488 AMY1B 1.39 0.000768 0.036710 RBBP8 1.46 0.0023 0.0488 LEF1 1.48 0.000773 0.036710 ACAT1 1.36 0.0023 0.0488 ACAA1 1.40 0.000775 0.036710 TRIM33 1.43 0.0023 0.0488 CLIC1 1.34 0.000776 0.036710 TBCA 1.38 0.0023 0.0488 TMEM57 1.21 0.000777 0.036710 PSME1 1.30 0.0023 0.0488 LOC100129882 1.95 0.000781 0.036710 MRPL21 1.40 0.0023 0.0488 LOC100133129 1.19 0.000781 0.036710 CBX3 1.24 0.0023 0.0488 LOC1001293 C15ORF21 1.28 0.000783 0.036710 79 1.35 0.0023 0.0488 LOC100130746 1.16 0.000784 0.036710 INSC 1.36 0.0023 0.0488 ISG15 1.36 0.000784 0.036710 SMPD1 1.18 0.0023 0.0488 RPS27 1.23 0.000787 0.036760 AURKAIP1 1.29 0.0023 0.0488 CSRP2BP 1.36 0.000790 0.036810 HS.544751 1.13 0.0024 0.0488 SLC12A9 1.53 0.000794 0.036810 OSTCL 1.24 0.0024 0.0489 FAM110A 1.39 0.000794 0.036810 LOC728576 1.24 0.0024 0.0489 HS.513842 1.14 0.000796 0.036810 ST3GAL1 1.47 0.0024 0.0489 BTBD6 1.60 0.000796 0.036810 SCYL1BP1 1.35 0.0024 0.0489 HAUS8 1.33 0.000800 0.036880 NPLOC4 1.23 0.0024 0.0489 NUP62 1.36 0.000801 0.036890 UBAC1 1.32 0.0024 0.0489 LOC440027 1.21 0.000806 0.037000 NEDD4 1.11 0.0024 0.0489 RNF5P1 1.37 0.000815 0.037310 SPATS2L 1.70 0.0024 0.0489 LOC653778 1.61 0.000816 0.037310 C3ORF21 1.28 0.0024 0.0489 NONO 1.33 0.000822 0.037450 LOC554207 1.18 0.0024 0.0489 FBXW4 1.20 0.000825 0.037470 SNAP29 1.50 0.0024 0.0489 PREX1 1.53 0.000829 0.037470 PDCD5 1.43 0.0024 0.0489 HS.149495 1.15 0.000829 0.037470 CNOT1 1.35 0.0024 0.0489 TAX1BP3 1.46 0.000831 0.037470 SRP14 1.34 0.0024 0.0489 CTXN1 1.60 0.000834 0.037470 SH3BGRL3 1.66 0.0024 0.0489 C8ORF37 1.20 0.000837 0.037470 C20ORF43 1.34 0.0024 0.0489 AP1S2 1.89 0.000838 0.037470 ATP6AP1L 1.11 0.0024 0.0489 CLPTM1L 1.53 0.000838 0.037470 PRKCE 1.43 0.0024 0.0489 NFYC 1.53 0.000843 0.037470 SLC39A1 1.28 0.0024 0.0489 ARMET 1.39 0.000844 0.037470 SEC61B 1.33 0.0024 0.0489 AKAP7 1.63 0.000848 0.037470 FYN 1.36 0.0024 0.0489 PSIP1 1.68 0.000850 0.037470 SF3B14 1.37 0.0024 0.0490 LOC1001309 MPHOSPH6 1.34 0.000852 0.037470 32 1.35 0.0024 0.0490 LOC651436 1.33 0.000854 0.037470 BRI3BP 1.39 0.0024 0.0490 LDHB 1.54 0.000858 0.037470 RPS26P10 1.32 0.0024 0.0490 NUP93 1.29 0.000864 0.037590 HS.548213 1.15 0.0024 0.0490 KIAA0556 1.33 0.000876 0.037590 CCL5 1.33 0.0024 0.0490 POP7 1.38 0.000880 0.037590 ARSB 1.33 0.0024 0.0491 RPL34 1.36 0.000883 0.037590 TMED10P 1.46 0.0024 0.0491 HS.130916 1.26 0.000883 0.037590 ZNF518B 1.33 0.0025 0.0492 IMPDH1 1.43 0.000884 0.037590 CD96 1.19 0.0025 0.0494 TICAM2 1.33 0.000885 0.037590 TALDO1 1.32 0.0025 0.0494 HS.550193 1.22 0.000885 0.037590 FADD 1.41 0.0025 0.0494 RHBDF2 1.98 0.000888 0.037590 C4ORF43 1.16 0.0025 0.0494 ARPC2 1.31 0.000889 0.037590 MGC52498 1.11 0.0025 0.0494 SIDT2 1.32 0.000890 0.037590 LOC644029 1.27 0.0025 0.0494 LOC729680 1.38 0.000893 0.037590 SHFM1 1.33 0.0025 0.0494 ACSL3 1.37 0.000894 0.037590 LOC642975 1.36 0.0025 0.0494 LOC100131160 1.17 0.000895 0.037590 LOC402677 1.12 0.0025 0.0494 LOC401019 1.27 0.000895 0.037590 LOC650369 1.30 0.0025 0.0494 TMBIM4 1.38 0.000895 0.037590 COBRA1 1.25 0.0025 0.0494

371

FAM43A 2.10 0.000897 0.037590 PRMT1 1.31 0.0025 0.0494 HEXB 1.40 0.000898 0.037590 LOC642250 1.17 0.0025 0.0494 GPX1 1.34 0.000902 0.037680 LOC644464 1.20 0.0025 0.0495 NUP85 1.32 0.000905 0.037720 PRMT6 1.58 0.0025 0.0496 XPO6 1.23 0.000910 0.037850 PSMD6 1.36 0.0025 0.0496 ACP1 1.40 0.000914 0.037870 LOC729340 1.39 0.0025 0.0496 LOC100129599 1.42 0.000922 0.037870 GLB1 1.31 0.0025 0.0496 MRPS18C 1.40 0.000924 0.037870 CDC37 1.26 0.0025 0.0496 TCEB2 1.28 0.000924 0.037870 PSMD4 1.30 0.0025 0.0498 C3ORF26 1.47 0.000925 0.037870 NSMCE4A 1.36 0.0025 0.0498 SDF2L1 1.65 0.000925 0.037870 TSPYL4 1.28 0.0025 0.0498 SPATA6 1.15 0.000926 0.037870 MRVI1 1.21 0.0025 0.0498 LOC642210 1.27 0.000928 0.037870 NDUFB2 1.31 0.0026 0.0498 LOC648608 1.13 0.000929 0.037870 CYC1 1.41 0.0026 0.0498 HS.199657 1.12 0.000930 0.037870 WDR77 1.27 0.0026 0.0498 NMRAL1 1.32 0.000931 0.037870 ACYP2 1.44 0.0026 0.0498 LOC643007 1.51 0.000937 0.037950 IMP3 1.40 0.0026 0.0498 RPL39 1.33 0.000940 0.037950 RPS4X 1.34 0.0026 0.0498 LZTR1 1.28 0.000940 0.037950 TRIP6 1.23 0.0026 0.0498 LOC652698 1.14 0.000942 0.037950 TXNL2 1.44 0.0026 0.0498 KLHL29 1.23 0.000944 0.037950 SCO2 1.26 0.0026 0.0498 FKTN 1.15 0.000944 0.037950 LOC283412 1.39 0.0026 0.0498 FAM135A 1.32 0.000949 0.038050 RAC2 1.47 0.0026 0.0499 AFG3L1 1.41 0.000951 0.038070 NUP160 1.45 0.0026 0.0499 RBM8A 1.15 0.000954 0.038120 LOC440733 1.26 0.0026 0.0499 CAMLG 1.40 0.000956 0.038120 PCNT 1.20 0.0026 0.0499 ATL1 1.21 0.000967 0.038270 LOC346887 1.36 0.0026 0.0499 RABAC1 1.51 0.000969 0.038270 FLJ38482 1.36 0.0026 0.0499 IRF9 1.66 0.000969 0.038270 ASNSD1 1.51 0.0026 0.0499 LOC645762 1.24 0.000973 0.038270 ARL9 1.97 0.0026 0.0499 HS.539981 1.17 0.000973 0.038270 SUZ12 1.34 0.0026 0.0499 PNPT1 1.18 0.000974 0.038270 NME1 1.52 0.0026 0.0499 MCM4 1.50 0.000975 0.038270 UQCRB 1.30 0.0026 0.0499 LOC729903 1.31 0.000978 0.038270 ADAP2 1.78 0.0026 0.0499 GNAI2 1.48 0.000984 0.038370 LOC644790 1.28 0.0026 0.0499 LOC728517 1.24 0.000984 0.038370 SLC44A4 1.18 0.0026 0.0499 EIF4ENIF1 1.37 0.000988 0.038430 MSH3 1.24 0.0026 0.0499 H3F3A 1.15 0.000994 0.038600 ATP5L 1.25 0.0026 0.0499 TSPO 1.39 0.000996 0.038600 C19ORF30 1.15 0.0026 0.0499 NKRF 1.25 0.001006 0.038830 DUSP13 1.12 0.0026 0.0499 LOC641989 1.15 0.001007 0.038830 LST1 1.82 0.0026 0.0499 MTMR4 1.47 0.001012 0.038950 EIF4G3 1.32 0.0026 0.0499 LOC285741 1.60 0.001019 0.039150 PHACTR4 1.41 0.0026 0.0499 BVES 1.48 0.001026 0.039300 HS.147420 1.13 0.0026 0.0499 KIAA0114 1.34 0.001029 0.039310 LOC728481 1.32 0.0027 0.0499 HS.484967 1.15 0.001030 0.039310 ZNF641 1.33 0.0027 0.0499 ACER2 1.17 0.001039 0.039480 FAU 1.26 0.0027 0.0499 LOC647450 1.51 0.001039 0.039480 GOLGA7 1.36 0.0027 0.0499 AKAP11 1.46 0.001041 0.039480 OXCT1 1.25 0.0027 0.0499 PSME2 1.32 0.001042 0.039480 SNRPD1 1.30 0.0027 0.0499 KCTD5 1.41 0.001044 0.039480 FLJ34047 1.21 0.0027 0.0499 ADCY7 1.24 0.001045 0.039480 PRICKLE4 1.29 0.0027 0.0499 KIAA0664 1.20 0.0027 0.0499 LOC646956 1.22 0.0027 0.0499

372

Table B.4. List of the 234 genes that were significantly down-regulated in the VXL- treated versus non-treated samples of ALL-67 xenograft. absFC, absolute fold change.

abs Q abs Gene P.Value Gene P.Value Q Value FC Value FC ERCC-00073 1.16 0.00001371 0.01949 DCDC5 1.10 0.0012163 0.04152 C1ORF226 1.15 0.00001475 0.01949 FLJ40330 1.12 0.0012248 0.04153 DKFZP686O2416 6 1.13 0.00001558 0.01949 LOC646503 1.15 0.0012443 0.04158 LOC100134210 1.15 0.00002468 0.02276 LOC100132319 1.12 0.0012615 0.04181 LOC728388 1.14 0.00003281 0.02332 CCR8 1.13 0.0012787 0.04207 C11ORF70 1.14 0.00004752 0.02486 C12ORF71 1.15 0.0012881 0.04231 ERCC-00071 1.16 0.0000599 0.02486 HS.527241 1.13 0.0012934 0.04236 LOC729957 1.12 0.00006679 0.02486 LOC442249 1.12 0.0013071 0.04249 OFCC1 1.17 0.00007684 0.02486 LOC646869 1.10 0.0013142 0.0425 LOC100132329 1.16 0.00007767 0.02486 SCRN3 1.11 0.0013149 0.0425 MIR578 1.16 0.00008018 0.02486 LOC285634 1.10 0.0013209 0.04263 LOC647025 1.17 0.00008589 0.02486 LOC338588 1.13 0.0013474 0.04291 HS.545655 1.13 0.00009642 0.02486 FLJ45139 1.11 0.0013582 0.04291 HS.562980 1.18 0.00009671 0.02486 LOC728285 1.10 0.0013617 0.04291 MGC39584 1.17 0.0001125 0.02486 MIR643 1.11 0.0013753 0.04295 LOC653492 1.15 0.0001216 0.02591 LOC645868 1.13 0.001405 0.04313 LOC100133724 1.15 0.0001306 0.02702 MIR362 1.15 0.0014147 0.04313 GPHB5 1.11 0.0001378 0.0275 HS.121380 1.14 0.001428 0.04313 PCDHA1 1.15 0.0001411 0.0275 LOC654165 1.13 0.0014493 0.04313 LOC645908 1.14 0.0001459 0.0275 CLDN16 1.12 0.0014621 0.04326 LOC652696 1.15 0.0001473 0.0275 HS.574784 1.15 0.0014649 0.04326 C14ORF125 1.17 0.0001505 0.0275 CHGA 1.14 0.0014693 0.04326 LOC390547 1.16 0.0001505 0.0275 LOC729562 1.15 0.0014851 0.04349 LOC649947 1.13 0.0001534 0.0275 HS.453381 1.13 0.0014953 0.04367 DDX25 1.15 0.0001636 0.0275 HS.445500 1.12 0.001509 0.04378 WNT6 1.15 0.0001641 0.0275 LOC646128 1.11 0.0015107 0.04378 PCDHB14 1.16 0.0001683 0.0275 LARP6 1.10 0.001543 0.04425 LYG2 1.16 0.0001851 0.0275 GATA4 1.10 0.0015458 0.04425 HS.495112 1.15 0.0001926 0.02811 HTRA1 1.11 0.0015487 0.04425 SNX9 1.14 0.0001942 0.02811 C11ORF40 1.12 0.0015617 0.04456 HS.571292 1.18 0.0001989 0.02841 HS.495041 1.16 0.0015757 0.04458 LOC646783 1.21 0.0002202 0.02891 LOC646201 1.13 0.0015906 0.04475 LOC645033 1.16 0.0002324 0.02891 FLJ41603 1.10 0.001614 0.04495 NEUROG1 1.16 0.0002347 0.02891 TRPC3 1.11 0.0016275 0.04501 MAL 1.20 0.0002459 0.02891 HS.539161 1.16 0.001677 0.04547 SRD5A2L2 1.13 0.0002518 0.02891 LOC645718 1.09 0.0016872 0.04559 HS.124358 1.12 0.0002584 0.02891 KRT77 1.11 0.0016959 0.04562 LOC642220 1.16 0.0002784 0.02949 ZNF862 1.13 0.0017009 0.0457 C15ORF60 1.12 0.000287 0.02983 OR3A3 1.13 0.0017215 0.04587 LOC100130966 1.14 0.0002977 0.03007 HS.176498 1.12 0.0017293 0.04587 LOC649199 1.13 0.0003 0.03007 LOC731837 1.17 0.0017405 0.04591 PCOLCE2 1.11 0.0003015 0.03007 LOC643580 1.11 0.0017408 0.04591 HS.546008 1.15 0.0003095 0.03043 LOC441662 1.11 0.00176 0.04592 LOC729270 1.12 0.0003108 0.03043 HS.573575 1.12 0.0017643 0.04592 HS.541982 1.20 0.0003156 0.03043 NAP1L3 1.15 0.0017775 0.04592 LOC100129236 1.12 0.0003228 0.03053 LOC100134153 1.13 0.0017893 0.04593 LOC441347 1.11 0.0003309 0.03053 HS.557431 1.15 0.0018152 0.04623 LRRC34 1.12 0.0003365 0.03078 LOC653609 1.14 0.0018265 0.04623

373

GSX1 1.13 0.0003486 0.03101 LOC728649 1.11 0.0018583 0.0465 LOC644944 1.12 0.0003488 0.03101 LOC728660 1.14 0.0018622 0.0465 LOC100131401 1.12 0.0003718 0.0313 XKR9 1.11 0.0018684 0.0465 ACCN2 1.13 0.0003806 0.03145 FLJ42133 1.12 0.0018834 0.0465 TBX5 1.12 0.0004116 0.03238 TLR8 1.12 0.0019415 0.04701 LOC652745 1.14 0.0004136 0.03238 KRT6C 1.12 0.0019467 0.04701 KC6 1.14 0.0004201 0.03241 LOC642165 1.11 0.0019595 0.04701 SNORD5 1.14 0.0004223 0.03241 WWC2 1.12 0.0019703 0.04701 ANKRD18A 1.10 0.0004265 0.03241 RNASE13 1.14 0.0019723 0.04701 PLAC1 1.12 0.0004357 0.03241 LOC643326 1.15 0.0019886 0.04714 MIR30E 1.15 0.0004394 0.03254 OR6C70 1.11 0.0020002 0.04726 MIR299 1.13 0.0004436 0.03255 LOC642762 1.12 0.0020194 0.04736 PCDHGB1 1.17 0.0004781 0.03356 B3GALT5 1.10 0.0020293 0.04739 HS.162891 1.19 0.0004803 0.03356 LOC648470 1.10 0.0020337 0.04739 LOC100131891 1.13 0.0004836 0.03356 LOC651556 1.13 0.0020513 0.04741 RAB38 1.14 0.0005157 0.03374 XCL2 1.15 0.002075 0.04759 SUCNR1 1.14 0.0005185 0.03374 PARM1 1.13 0.0020765 0.04759 HS.332843 1.12 0.0005346 0.03374 C21ORF121 1.12 0.0020904 0.04759 LOC652568 1.11 0.0005383 0.03374 LRFN5 1.13 0.002094 0.04759 NPTXR 1.12 0.000541 0.03374 LOC100133067 1.09 0.0020944 0.04759 LOC642362 1.11 0.000545 0.03381 LOC643897 1.13 0.002095 0.04759 LOC650148 1.13 0.0005503 0.03382 LOC100128064 1.11 0.0021024 0.04759 LOC100129578 1.10 0.0005557 0.03382 FMO3 1.11 0.0021032 0.04759 LOC647080 1.13 0.0005614 0.03382 STARD9 1.10 0.00211 0.04762 LOC100129626 1.12 0.0005712 0.03385 PIKFYVE 1.11 0.0021294 0.04763 FBXL21 1.15 0.0005803 0.03408 HS.150092 1.13 0.0021416 0.04784 HS.545028 1.15 0.0005944 0.03444 LOC100134031 1.09 0.0021524 0.04784 ABCG8 1.13 0.00061 0.03448 SPARCL1 1.11 0.0021627 0.04784 OR8U9 1.14 0.0006109 0.03448 LIX1 1.14 0.0021662 0.04784 CTXN2 1.11 0.0006143 0.03448 LOC648289 1.10 0.0021722 0.04784 LOC399939 1.15 0.0006325 0.03486 LOC646863 1.14 0.0021747 0.04784 LOC729025 1.11 0.0006402 0.03486 LOC645448 1.13 0.0021751 0.04784 LOC732429 1.13 0.0006476 0.03486 LOC100132318 1.15 0.0021776 0.04784 LOC440888 1.14 0.0006851 0.03539 LOC653759 1.11 0.0022004 0.048 LGALS14 1.14 0.0006956 0.03542 LOC729866 1.13 0.0022004 0.048 LOC642351 1.12 0.0007004 0.03548 PRDM5 1.13 0.002235 0.04808 CRABP1 1.12 0.0007302 0.03644 HS.130639 1.12 0.0022604 0.0483 MIR2114 1.12 0.0007344 0.03644 C10ORF99 1.13 0.0022644 0.04832 CKMT1B 1.13 0.0007354 0.03644 ZNF205 1.11 0.0022834 0.04847 LOC100132904 1.11 0.0007477 0.03655 HRC 1.14 0.0023168 0.0488 LOC651648 1.11 0.0007492 0.03655 LOC650566 1.15 0.0023187 0.0488 LOC651603 1.10 0.0007619 0.03671 LOC653155 1.12 0.0023379 0.04882 LOC642153 1.12 0.0007753 0.03671 LOC344595 1.10 0.0023447 0.04882 LOC391073 1.16 0.0007795 0.03671 C1ORF220 1.10 0.0023465 0.04882 C4ORF39 1.14 0.0007823 0.03671 CD3D 1.12 0.0023792 0.04886 LOC645879 1.12 0.000821 0.03745 OR2C1 1.12 0.0023795 0.04886 OR10H1 1.14 0.0008418 0.03747 LOC643265 1.15 0.002393 0.04894 LOC728307 1.14 0.0008476 0.03747 MYH16 1.13 0.0024154 0.04894 TMEM151A 1.13 0.000848 0.03747 GJD4 1.15 0.0024162 0.04894 HS.548415 1.16 0.0008566 0.03747 LOC646629 1.13 0.0024179 0.04894 LOC651128 1.13 0.0008567 0.03747 LOC642672 1.11 0.002419 0.04894 FCGR1C 1.11 0.0008591 0.03747 SYT8 1.12 0.0024207 0.04894 LOC100132252 1.13 0.0008599 0.03747 HS.129334 1.12 0.0024487 0.04911 MIR548H3 1.11 0.0008662 0.03759 FAM47A 1.10 0.0024524 0.04913 TXNDC8 1.15 0.0008697 0.03759 MASP2 1.11 0.0024762 0.04937 LOC643382 1.14 0.0008728 0.03759 C14ORF23 1.11 0.00249 0.04937 374

WDR64 1.16 0.000921 0.03787 C21ORF54 1.10 0.0024988 0.04939 HS.579678 1.16 0.0009323 0.03787 SNORD83A 1.14 0.0025296 0.0496 LOC728970 1.22 0.0009603 0.03824 LOC728364 1.12 0.0025548 0.04977 LOC388248 1.17 0.000975 0.03827 ZNF829 1.10 0.0025563 0.04977 OR10H3 1.11 0.0009782 0.03827 LOC643152 1.12 0.0026089 0.04994 C6ORF154 1.14 0.0010005 0.03872 HS.562146 1.14 0.0026093 0.04994 HCN2 1.12 0.0010204 0.03915 TRPA1 1.13 0.0026176 0.04994 LOC643158 1.13 0.0010481 0.03948 LOC391556 1.10 0.0026294 0.04994 LOC653570 1.12 0.0010537 0.03948 LOC100130084 1.13 0.002638 0.04994 ENTPD5 1.14 0.0010554 0.03948 LOC652794 1.10 0.0026483 0.04994 HS.571039 1.14 0.0011109 0.03999 GPR15 1.11 0.0026614 0.04994 RAD52 1.14 0.0011205 0.04003 LOC440225 1.10 0.0026614 0.04994 EFCAB6 1.11 0.0011918 0.04137 LOC644280 1.12 0.002669 0.04994

375

376